Anda di halaman 1dari 384

Volume 105, Number 2

Introduction: Antibiotic Resistance

Christopher T. Walsh is the Hamilton Kuhn Professor of Biological


Chemistry and Molecular Pharmacology (BCMP) at Harvard Medical
School. For his biography, see p 426.

Gerry Wright is Professor and Chair of the Department of Biochemistry


and Biomedical Sciences at McMaster University in Hamilton, Ontario.
For his biography, see p 529.

One of the central themes of success in human


therapeutics in the 20th century was the discovery
and development of antibiotics and antibacterial
agents, for the treatment of bacterial infections. The
introduction of antibiotics helped drop the death rates
from infectious disease from 797 per hundred thousand in 1900 to 36 per hundred thousand in 1980, a
20-fold improvement. Two lines of chemical investigation proved fruitful: the isolation of natural products with antibiotic activity from microbial sources
and the purposeful synthesis of antibacterial agents
by medicinal chemists.
The first line of discovery, the isolation of microbial
metabolites from Nature, was initiated by Flemmings discovery of a penicillin-producing fungus and
was closely followed by systematic search of antibacterial producing microorganisms by pioneers such as
Dubos and Waksman. This strategy produced many
of the famous classes of antibiotics. These include
both the cephalosporin and penicillin branches of the
-lactams, the aromatic polyketides of the tetracycline class, the aminoglycosides represented by
streptomycin, the polyketide macrolactones exemplified by erythromycin, and the glycopeptides of the
vancomycin and teicoplanin family. The search by
medicinal chemists for antibacterial magic bullets by
synthetic efforts has produced the sulfa drugs, the
dihydrofolate reductase inhibitors, the fluoroquinolones, and most recently the oxazolidinones.

In the course of these discovery efforts, the active


natural products and synthetic antibacterials have
proven to be valuable probes for deciphering the
identity of targets in pathogenic bacteria. Historically, this has turned out to be a target poor therapeutic area with only four robust targets for widely
used groups of antibiotics: bacterial cell wall biosynthesis; bacterial protein biosynthesis; DNA replication and repair; and folate coenzyme biosynthesis.
The golden age of antibiotic discovery in the 20th
century was actually quite short (Table 1). The two
decades from 1940 to 1960 saw isolation of most of
the major classes of natural antibiotics. The sulfa
drugs were introduced in the 1930s and have been
in continuous use for 70 years. The first versions of
the quinolone synthetic drugs were introduced in
1962.
Much subsequent activity has involved refinement
of existing antibiotic structures to deal with the onset
of resistance. For example, the first generation
penicillin V was replaced by second generation versions, methicillin and ampicillin, which in turn were
superseded by such extended spectrum molecules as
piperacillin. A parallel effort at semisynthetic modification of cephalosporins has led from first generation (cephazolin) to second (cefoxatin) to third (ceftriaxone) and now fourth (cefipime) generation versions
where the bicyclic lactam core scaffold is maintained
and the periphery tailored to deal with succeeding

10.1021/cr030100y CCC: $53.50 2005 American Chemical Society


Published on Web 02/09/2005

392 Chemical Reviews, 2005, Vol. 105, No. 2

Editorial

Table 1. History of Antibiotic Classes in Clinical Use


year
1929 (activity)
1940 (purification)
1932
1944
1945
1947
1948
1950
1955
1955
1955
1955
1959
1962
1969
2000
2003

antibiotic

class

penicillin

-lactam

sufapyridine
streptomycin
cephalosporin
chloramphenicol
chlortetracycline
erythromycin
vancomycin
virginiamycin
amphotericin
lincomycin
rifamycin
nalidixic acid
fosfomycin
linezolid
daptomycin

sulfonamide
aminoglycoside
-lactam
phenypropanoid
tetracycline
macrolide
glycopeptide
streptogramin
polyene (antifungal)
lincosamide
ansamycin
quinolone
phosphonate
oxazolidinone
lipopeptide

waves of resistant bacteria. In the macrolide lineage,


the fully natural product erythromycin was tailored
to give the second generation semisynthetic azithromycin and clarithromycin and now the third generation ketolides such as telithromycin.
There was a 38-year interval between introduction
of the fluoroquinolone class of antibiotics in 1962 and
the next new structural class, the oxazolidinones, in
2000.
This innovation gap is directly relevant to the
current situation where the antibiotic cupboard is
rather bare for meeting the challenges of new outbreaks of resistant bacteria.
The pattern of introduction of successive generations of -lactam antibiotics and of macrolides, tetracyclines, and aminoglycosides is strong testimony to
the almost inescapable correlation that introduction
of a new antibiotic into widespread clinical use
induces the rise of resistant bacteria. Given the vast
numbers of bacteria, their short generation times,
and typical gene mutation frequencies of 1 in 107
bacteria, resistance is inevitable. Antibiotics select
for those very rare bacteria in a population that are
less susceptible and allow them to become dominant
in the populations as susceptible bacteria die off. For
all the major classes of antibiotics noted above, both
natural and synthetic, clinically significant antibiotic
resistance has ensued after introduction into human
therapeutic use. The time frame has been as short
as one year for penicillin V and as long as 30 years
for vancomycin. The resistance kinetics reflect a
range of intrinsic and acquired molecular mechanisms and the extent of dissemination of the antibiotic.
Three major mechanisms of antibiotic resistance
reveal a few common themes used by bacteria to fend
off antibiotics. One mechanism is destruction of the
antibiotic by bacterial enzymes, and this is the
quantitatively significant route for disabling -lactams by hydrolysis of the drug lactam warhead. A
second mechanism is bacterial reprogramming of the
antibiotic target to lowered susceptibility, and this
is the path vancomycin resistant enteroccci (VRE)
take to escape vancomycin action. The third major
route, especially prevalent in pseudomonads but
common to all bacteria, is to pump out antibiotics via

natural product

synthetic

x
x
x
x
x
x
x
x
x
x
x
x
x

x
x

transmembrane efflux pumps, keeping antibiotic


concentration within the bacterial cell below toxic
threshold concentrations.
Thus, the expectation in the first decade of the 21st
century is that as a new antibiotic is introduced to
combat pathogenic bacteria, resistant organisms will
be selected and will vitiate the utility of that antibiotic. New antibiotics will therefore be required periodically as waves of resistance follow. The anticipated
cyclical need for new antibiotics raises key questions
of where new antibiotics will come from and whether
monotherapy will continue to be an effective practice
for many life threatening infections.
To optimize the discovery and development time
for new antibiotics requires both sources of new
molecules and understanding of the molecular mechanisms of resistance. This thematic issue of Chemical
Reviews collects in one place reviews by experts both
on the mechanisms of action of the major classes of
antibacterial drugs and on the major resistance
mechanisms.
Articles 1-3 deal with cell wall biosynthetic targets, starting with the -lactam antibiotics and then
discussing the glycopeptides of the vancomcyin and
teicoplanin class. The glycopeptides are substrate
binders for un-cross-linked D-Ala-D-Ala termini of
peptidoglycan biosynthetic intermediates. The ramoplanin class of lipodepsipeptide antibiotics also
functions by substrate binding but targets the Lipid
II molecules that act as carriers for the nascent
peptidoglycan units.
Articles 4-7 deal with antibiotics that target
different facets of bacterial protein biosynthesis. The
first two articles in this cluster take up aminoglycosides, which bind at sites on 16S ribosomal RNA on
the small subunit of bacterial ribosomes, and macrolides, which bind to 23S ribosomal RNA near the
peptidyl exit tunnel. Article 6 evaluates the newer
streptogramins and oxazolidinones for their mechanism of ribosome inhibition. Article 7 reviews the
knowledge base for polyketide biosynthesis and combinatorial biosynthesis approaches to novel structural
and functional activities for third generation antibiotics in this polyketide class.
Articles 8-10 address antibiotics that act to interdict information flow from DNA to RNA in bacterial

Editorial

cells, discussing fluoroquinolones that target type II


DNA topoisomerases, antifolates that block provision
of monomers for DNA synthesis, and rifamycins that
block DNA-dependent RNA polymerases.
Articles 11 and 12 take up two variants of peptide
antibiotic classes, the ribosomally generated lantibiotics and the nonribosomal thiopeptide antibiotics.
Article 13 takes a broader look at the biosynthetic
logic for nonribosomal peptide assembly line machinery responsible not only for thiopeptides but also the
glycopeptides and the depsipeptide classes. The
antibiotic class most recently approved by the FDA
is the nonribosomal lipodepsipeptide daptomycin.
Article 14 examines the biosynthesis and mechanism of additional nonribosomal peptide and polyketide natural products, originally identified as antibiotics, but with special potency and utility as
antitumor agents. These include bleomycin, the enediynes, and mitomycin, all of which target DNA.
The last article addresses the pressing and continuing need for new antibiotics by reviewing how

Chemical Reviews, 2005, Vol. 105, No. 2 393

new targets for antibiotics are discovered in bacteria


by genetic and genomic approaches. It also summarizes contemporary approaches for screening of
leads in antimicrobial drug discovery.
This comprehensive collection of reviews on the
major classes of antibiotics in human clinical use
should be a central resource for scientists to grapple
with the questions of how antibiotics work, why they
stop working, and how the molecular insights into
molecules and their targets condition strategies for
much needed new antibiotics.
Christopher Walsh
Harvard Medical School
Gerard Wright
McMaster University
CR030100Y

Chem. Rev. 2005, 105, 395424

395

Bacterial Resistance to -Lactam Antibiotics: Compelling Opportunism,


Compelling Opportunity
Jed F. Fisher, Samy O. Meroueh, and Shahriar Mobashery*
Department of Chemistry and Biochemistry, University of Notre Dame, Notre Dame, Indiana 46556
Received June 3, 2004

Contents
1. Introduction
2. Penicillin-Binding Proteins
2.1. Enzymes of Cell Wall Biosynthesis
2.2. -Lactam-Sensing Proteins
3. -Lactamases
3.1. Overview and Classification
3.2. Class A -Lactamases
3.3. Class C -Lactamases
3.4. Class D -Lactamases
3.5. Class B Metallo--lactamases
4. Other Resistance Mechanisms
4.1. Porin Deletion
4.2. Transporter Expression
5. Envoi
6. Acknowledgments
7. Abbreviations
8. Note Added in Proof
9. References

395
398
398
402
404
404
406
411
412
413
416
416
417
417
419
419
419
419

1. Introduction
The simplistic image of the bacterium as an isolated, planktonic, self-cloning automaton is refuted.
We now recognize bacteria as microorganisms of
enormous diversity and adaptability. They can thrive
under conditions that we regard as extremesin the
absence of oxygen and at high temperatures, to
choose but two examplessand they can adjust with
surprising alacrity to their environment, and to their
circumstance, so as to improve their fitness for
survival.
The focus of this review is that of bacterial biochemical adaptation to a particular circumstance of
profound concern to the human species: that of
bacterial tolerance and resistance to the -lactam
antibiotics.1 The -lactam antibiotic family originally
was limited to the penicillin (sulfur-containing penams) and then the cephalosporin (sulfur-containing
This review is dedicated to our Ph.D. mentors: To Chris Walsh,
on the occasion of his 60th birthday and with appreciation for his
unique ability to inspire and guide self-discovery. To Bill Hase,
for his guidance and creating an environment that expected
perfection in every aspect of scientific investigation. To Michael
Johnston, for his creativity, passionate engagement, and singleminded ability to contribute to the many things that he has tackled
in life.
* To whom correspondence should be addressed. E-mail: mobashery
@nd.edu.

cephems) -lactams but now includes natural and


synthetic monocyclic -lactams, carbapenems, oxapenams, carbacephems, and oxacephems (Scheme 1).
The -lactams are one of the three largest antibiotic
classes (the others are the macrolides and fluoroquinolones).2,3 Mere words cannot properly emphasize
the role that these antibiotics have to the preservation of human health. Nor do words adequately
emphasize the disquieting reality that at the same
moment we profit from the use of antibiotics, that
there is cost and that this cost is inexorable bacterial
resistance.4-8
Bacterial resistance is not a new phenomenon. We
now recognize that resistance is the inevitable consequence of organisms competing for the same ecological niche. Yet it is only during the past 60 years
that resistance has been transformed by man (as the
driving evolutionary force)9 from what might be
reasonably described as stasissbacteria competing
against bacteriasto that of a disequilibrium of chemical warfare2,10-14 where bacteria additionally compete
directly with us. Assuredly, this is a competition with
uncertain outcome. While the phenomenon of bacterial resistance is evolutionarily ancient,15,16 the consequence of this (so very recent) warfare is that of
accelerated dispersion of the mechanisms for resistance across the bacterial kingdom, increasing selection for bacteria that have acquired these mechanisms, and devaluation of our antibiotic armamentarium.
Bacterial resistance mechanisms with respect to
the -lactam antibiotics are divided between those
that occur at the level of primary metabolism (altered
and acquired proteins and enzymes) and those that
occur at the level of secondary metabolism (the
biosynthesis of modified -lactams that are better
antagonists of the altered proteins). While the realization of the outstanding importance of secondary
metabolites as drug templates dates to the moment
of their very discovery by man,17,18 the recognition
that these secondary metabolites occupy a logical
place in the evolution of bacterial resistance is a more
recent consensus, as discussed in several lucid
reviews13,19-23 The parallel medicinal chemistry development of -lactam structure is presented by
Dalhoff and Thomson24 and is not further discussed
here. Our focus is the bacterial response to the
selection pressure exerted by the -lactam antibiotics.
Among the changes that are accomplished are alterations (mutations) in the molecular target of the

10.1021/cr030102i CCC: $53.50 2005 American Chemical Society


Published on Web 02/09/2005

396 Chemical Reviews, 2005, Vol. 105, No. 2

Jed F. Fisher is an alumnus of S.U.N.Y. Stony Brook (B.S., under Bill


Fowler) and M.I.T. (Ph.D., under Chris Walsh). Following a postdoctoral
stay at Harvard (with Jeremy Knowles), a faculty position at the University
of Minnesota, and nearly two decades in the pharmaceutical industry (at
Upjohn and Pharmacia), he joined the Mobashery group at the University
of Notre Dame in 2004. His scientific accomplishments include the
collaborative discovery of the utility of coenzyme mimetics to discern
flavoenzyme mechanism, the acylenzyme pathway for -lactamase
substrate hydrolysis and inhibitor inactivation, the autocatalytic pathway
for anthracycline and mitomycin reductive nucleophilic substitution, the
first rational design of an inhibitor of proteinprotein association, and the
virtue of glycosylation to alter peptidomimetic pharmacodynamics.

Samy Meroueh studied chemistry at Wayne State University, where he


received his Ph.D. degree in Physical Chemistry, under the supervision
of William L. Hase. He then joined the group of Shahriar Mobashery,
where he is currently a Walther Cancer Institute Fellow at the University
of Notre Dame in South Bend, IN. He is using a variety of tools, including
computational, calorimetry, and X-ray crystallography, toward understanding
the catalytic machinery of enzymes that are involved in antimicrobial
resistance and the design and discovery of molecules that would target
those enzymes as potential antimicrobial agents.

-lactams, the (ancient) transformation of these


enzymes into families of -lactam hydrolytic deactivating enzymes (the -lactamases), the expression of
protein inhibitors of the -lactamases, the deletion
of porin proteins in the membrane, the acquisition
and activation of efflux exporter proteins, and the
modification of the cell wall to minimize -lactam
antibiotic access to their targets. While the alteration
in the cell wall biosynthesizing targets and the
expression of -lactamases may be regarded as the
primary bacterial defensive measures, none of these
defensive measures is unimportant. Depending on
the bacterium and the particular circumstance of the
-lactam challenge, the bacterium that devises a
successful combination of these responses is the
bacterium that survives. This review covers the

Fisher et al.

Shahriar Mobashery received his undergraduate (1981) and doctoral (1985)


degrees from the University of Southern California and University of
Chicago, respectively. After postdoctoral research (19861988) at Rockefeller University, he joined the faculty at Wayne State University in 1989.
Mobashery and his research group relocated to the University of Notre
Dame in 2003, where he holds the position of Navari Family Professor of
Life Sciences in the Department of Chemistry and Biochemistry.

primary literature of the past 4 years with citation


to the preceding literature via the many outstanding
concurrent reviews on this topic.
Notwithstanding the likely familiarity of the reader
with the -lactam antibiotics as mechanism-based
enzyme inhibitors of cell wall biosynthesis (inhibiting
the -lactam binding protein enzymes) and enzyme
substrates (of the hydrolytic antibiotic-destroying
enzymes, the -lactamases),16 an overview of the
chemistry of the -lactams is essential context. The
point of reference is the eponymous -lactam (that
is, a 2-azetidinone) four-membered cyclic amide ring
of the penicillins and cephalosporins (Scheme 1). The
neighboring carboxylate (on the second ring of these
bicyclic structures) and the acylamino substituent
upon the -lactam immediately (from our contemporary perspective) define these structures as conformationally constrained tripeptides, having an exposed C-terminus and imbued with the capacity to
acylate (with opening of the -lactam) susceptible
nucleophiles. The antibiotic property implies that
these peptidomimetics mimic an essential peptide
motif possessed by the bacteria and engage this
mimicry to the purpose of confounding acylation of a
critical enzymatic target. Indeed, this hypothesis
coincides to the guiding -lactam presumptionsthe
Tipper-Strominger hypothesissthat dates from the
discovery by Tipper and Strominger (and simultaneously and independently by Wise and Park) that
a penicillin-derived entity is irreversibly incorporated
into the transpeptidase and carboxypeptidase enzymes of bacterial cell wall biosynthesis.25,26 Among
the cell wall biosynthetic enzymes an obvious candidate for such interference is the peptidoglycan
cross-linking transpeptidase wherein the peptidoglycan D-Ala-D-Ala terminus is cleaved in the serine
acylation half-reaction (with loss of the terminal
D-Ala) and the cross-link is formed in the deacylation
half-reaction (by acyl transfer to an amine substituent of the neighboring peptidoglycan strand). Should
this enzyme be presented with a substrate mimetic
wherein the initial acylation reaction remains enabled, but the capacity for deacylation is abolished,

Bacterial Resistance to -Lactam Antibiotics


Scheme 1

then the enzyme will fail to complete its catalytic


cycle.27,28 The loss of these enzymatic activities
disrupts the homeostasis of cell wall integrity, leading (through a poorly understood process that ultimately involves activation of cell wall degradative
enzymes, termed autolysins) to lethal cell wall
defects.1,29-31 A bacterium unable to maintain the
integrity of its cell wall will be unable to reproduce
(a bacteriostatic antibiotic) or survive (a bactericidal
antibiotic) wherein the impaired cell wall no longer
contains the osmotic pressure of the cytoplasm.32,33
Within this mechanistic perspective is found the
two important strategies for bacterial acquisition of
resistance to the -lactam antibiotics. If a cell wall
transpeptidase is deceived by a peptidomimetic in the
acylating half-reaction, then resistance can be achieved
by alteration of the transpeptidase such that acylation by the peptidomimetic does not happen. Second,
if a cell wall transpeptidase is unable to complete
deacylation of the peptidomimetic, then resistance

Chemical Reviews, 2005, Vol. 105, No. 2 397

may be achieved by alteration of the transpeptidase


so as to enable this reaction. The first answer is that
of mutated (resistant) -lactam (penicillin) binding
proteins (PBPs). The second answer is that of the
transformation of the PBPs to new, fully capacitateds
often operational at the substrate diffusional rate
constantshydrolytic (wherein water is the acyl acceptor) enzymes, the -lactamases. As these alterations must be accomplished without loss of bacterial
fitness, it may be expected that specific circumstance
will make one the more effective strategy than the
other. This has proven to be so. Both contribute.
However, whereas the altered PBP is generally
regarded as a relatively recent development in Grampositive bacteria resistance, the advent of -lactamases is understood as an ancient evolutionary event
in bacteria resistance. With respect to the last 60
years, it is far less the evolutionary development of
these resistant enzymes as it is their broadened
distribution, which is of immediate concern.
We have posited a relationship between the mechanisms by which -lactams serve as antibiotics and
the primary mechanisms by which bacteria acquire
resistance to these mechanisms in terms of enzymatic
acylation and deacylation half-reactions. Further
development of these concepts requires, however, a
clearer description of these events as chemical reactions. Two reactions are pertinent. Under enzymatic
control each comprises an acylation event (wherein
an enzyme active site serine is involved for both) and
a deacylation event (wherein for the transpeptidase
the incoming nucleophile is usually a lysine, ornithine, or diaminopimelate amino group and wherein
for the -lactamase the acyl acceptor is water). For
the transpeptidase, the net reaction is the crosslinking of the peptidoglycan NAM-pentapeptide, the
major constituent of the cell wall. For the -lactamase, the net reaction is -lactam hydrolysis to the
biologically harmless penicilloate.
An objective of the two following sectionssthe
relationship of the PBPs and -lactamases to bacterial -lactam resistancesis to communicate at the
simplest possible level the operation of these acylation/deacylation reactions at the protein level. This
understanding remains a challenging task. The first
aspect of this task is the comparative transition-state
energies for ordinary (acyclic) amide (or peptide)
cleavage and -lactam opening. As Page discussed,34
the historical presumption that -lactam antibiotic
ability correlates to the release of significant strain
energy upon -lactam ring opening is overvalued.
Rather, the strain energy is modest and such greater
-lactam reactivity as does exist among the -lactam
family members (for example, the Sweet-Woodward
correlation)35 is unlikely to be expressed in the critical
(rate-limiting) enzymatic acylation step.36 Moreover,
the spatial requirements for general acid catalysis
of transient tetrahedral species collapse to the acylenzyme are rather different for a peptide compared
to a -lactam.34 The answers to the critical questions
as to why -lactams successfully inhibit the PBPs and
have become favorable substrates for the PBPderived -lactamases are found in the experiments
that address these questions: How is the -lactam

398 Chemical Reviews, 2005, Vol. 105, No. 2

recognized as a peptidomimetic by both enzyme


classes? How does the -lactam exploit for acylation
the PBP catalytic machinery, for which it is not
intended? How does the resulting PBP acyl-enzyme
resist the catalytic machinery for deacylation (or
transacylation)? How has the -lactamase acylenzyme become fully competent for acyl transfer to
water? The extraction of answers to these questions
requires challenging experimental design. This design would be more straightforward if the mechanistic basis for enzyme catalysis was evident from
protein structure. It is not!34,37-39 Enzyme catalysis
is the subtle orchestration of a panoply of electrostatic
forces, often significantly influenced by distal (nonactive site) changes in protein structure. While we
can visualize atomssthe structures of several PBPs
and of numerous -lactamases are knownswe must
intuit forces and conjecture the transition states that
they stabilize. The relationship of such conjectures
(and transition states) to bacterial -lactam resistance is the objective of the remainder of this review.

Fisher et al.
Scheme 2

2. Penicillin-Binding Proteins
2.1. Enzymes of Cell Wall Biosynthesis
To have survived means to have been opportunistic. Among the survivors, across the eons of time, are
the single-cell microorganisms of the domain bacteria. The seminal observation that a crystal violet
stain is retained by some bacteria, but not by others,
is known now to signify two different exoskeleton
constructs. The positive staining bacteria have a
single multilayered polymericsthat of a cross-linked
peptidoglycansexoskeleton, while the nonstaining
bacteria have a thinner (two and in places three
layers) of a polymeric (and also a cross-linked peptidoglycan) exoskeleton, further surrounded by a
gellike periplasmic layer,32 itself enclosed by a complex (outer) membrane bilayer. Despite this substantial difference, there are at the functional level and
at the molecular level remarkable similarities between the two. The peptidoglycan exoskeleton (termed
the murein sacculus) is durable and elastic, strong
enough to contain the osmotic turgor of the living bacterium yet permitting nutrient access to the porins
and transporter proteins.40 The cell wall componentss
synthesized in the cytoplasm and transported across
the cytoplasmic membrane for polymerizationsof
both are remarkably similar.41 For Gram-negative
bacteria (and for many Gram-positive ones), in the
final cell wall assembly step the D-Ala terminus of
the pentapeptide-functionalized N-acetylglucosamine
(termed N-acetylmuramic acid or NAM, and assembled with N-acetylglucosamine or NAG, as a
NAG-NAM disaccharide repeat) is removed. The
resulting acyl species is then transferred (crosslinked) to an amino group of a neighboring chain,
thereby unifying the peptidoglycan sacculus as a
single polymeric macromolecule (Scheme 2).26 Given
the sophistication of this process (which is also
intimate to the resealing of the sacculus during cell
division), cooperative multienzyme catalysis (that
includes the transpeptidase just described) is implicated. The study of this enzyme ensemble (termed

the divisome when as part of cell division) to bacterial


shape, integrity, and function is of outstanding
scientific importance.26,32,42-45 With regard to antibiotics (not only the -lactams, although these are our
focus), this ensemble is the story of compelling
opportunity and compelling opportunism.
The events that lead to the cross-linking reaction
have been elucidated with the 1.2 resolution X-ray
structure of the acyl-enzyme formed from the Dalanyl-D-alanine carboxypeptidase/transpeptidase from
Streptomyces sp. strain R61 and a unique cephalosporin.33 This cephalosporin (compound 1, Scheme
3) was designed to incorporate components of the cell
Scheme 3

wall into its own structure. As anticipated, compound


1 modified the active site serine, binding between the
all R-helix and R/ domain. The portions of compound
1 that mimic strands 1 and 2 from the peptidoglycan
(see the green- and red-colored portions in Figure 1
and Scheme 3, for example) were found to be oriented
within the active site. The left portion of compound
1 mimics the acyl-D-Ala-D-Ala portion of the first
strand of the peptidoglycan (portion of 1 in green),
while the right portion mimics the approach of the
nucleophilesthe diaminopimelatesfrom the second

Bacterial Resistance to -Lactam Antibiotics

Figure 1. (A) Stereoview of the three-dimensional structures of two strands of peptidoglycan bound to the active
site of the D,D-transpeptidase/D,D-carboxypeptidase from
Streptomyces R61 PBP, constructed computationally from
the 1.2 resolution structure for the acyl-enzyme species
with compound 1 (the extension reaches the NAG-NAM
units on the peptidoglycan). The protein is shown in yellow
ribbon representation, while the bound computational
model representing the two strands of the peptidoglycan
are shown in green and red capped-sticks representation.
The blue van der Waals surface defines the active site. (B)
Schematic representing the peptidoglycan from A, showing
the various hydrogen-bonding interactions and color-coded
according to the three-dimensional model.

strand of the peptidoglycan (red portions in structure


2). The acylation is proposed to occur after activation
of the active site Ser62 (located at the N-terminus of
an R-helix, which is expected to modulate the serine
pKa as proposed by Moews et al.46) by the general
base Lys65swhich is in direct contact with Ser62 at
a distance of 3.0 . This mechanism is consistent
with previously proposed mechanisms of the acylation of PBPs, where the universally conserved lysine
acts as the general base, abstracting a proton from
serine, followed by the back-donation of the proton
(from lysine) to the peptide or -lactam nitrogen.47
To provide a more detailed picture of the cross-linking
event, a computational model was constructed from
the high-resolution structure by extending compound
2 to include the full pentapeptide and a NAG-NAM
extension. The resulting model was fully solvated and

Chemical Reviews, 2005, Vol. 105, No. 2 399

energy minimized (shown in Figure 1A). The peptidoglycan strands were found to form a network of
electrostatic interactions (shown in Figure 1B). These
interactions should play important roles in properly
positioning the peptidoglycan strands and for other
important events such as deprotonation of diaminopimelate in the cross-linking event. It is of interest
to note that the three-dimensional model differed
from a previous model48 that used only the apo
PBP2x structure with respect to the location of the
saccharide-binding grooves.
The description of behavior as either moral or
immoral is (primarily) a human characteristic. For
other organisms this distinction is irrelevant, and the
focus of their behavior is survival to the point of
reproduction. Among the bacteria (and fungi, for
which bacteria are a food source) survivalsthat is,
preservation within an ecological nichesrequires
exploitation of vulnerability. In addition, the biosynthetic enzymes of bacterial cell wall biosynthesis are
vulnerable. The basis of their vulnerability (which
is one and the same with that of the bacterium) is
the combination of essentiality and exposure. These
enzymes are located underneath the very cell wall
that they assemble. For the Gram-negative bacteria
these enzymes are either within the periplasmic
space orsfor the most essential of these enzymess
with active sites exposed to the periplasmic space and
a transmembrane domain (with small cytoplasmic
anchor) within the cytoplasmic membrane. Hence,
bimolecular encounter with an inhibitor of these
enzymes requires only the successful passage of
the inhibitorsintermingling with solute nutrientss
through the lipopolysaccharide of the outer membrane into contact with the peptidoglycan surface of
the periplasmic space. While this simple requirement
cannot be underestimated (especially insofar as
antibacterial design and for resistance development)
for the penicillin and cephalosporin -lactams secreted by the biosynthesizing bacteria and fungi
within the niche, the passage and encounter with
these biosynthetic enzymes is straightforward. Astonishingly, each enzyme of the ensemble is capable
of inactivation (via the same mechanism of irreversible acylation) by an appropriately substituted -lactam. The inactivation is facile for susceptible Grampositive and -negative strains and less so for resistant
bacteria. This truly remarkable event is commemorated by historical nomenclature: these enzymes are
collectively the penicillin-binding proteins (or PBPs)
of the bacteria.
The chemically intriguing aspect of this event is
the recognition by each enzyme regardless of the
specific cell wall biosynthetic role. The inescapable
conclusion is a fundamental of homology of structure
and of alignment with the active site. However,
despite this homology, all -lactams do not inhibit
all PBPs, likely due to subtle differences in the active
site and distal regions in the protein. Regardless of
cell morphology (Gram positive or negative) and
regardless of individual specific synthetic function,
these enzymes must possess such similarity as to
implicate a mere handful of, if not a single, ancestral
progenitor(s). As the intact -lactam antibiotic was

400 Chemical Reviews, 2005, Vol. 105, No. 2

recognized by this ancestral enzyme as a mimic of


the peptidic terminus of the peptidoglycan, so too
there is continued recognition of these antibiotics by
the offspring of this enzyme. Once the -lactam
antibiotic is recognized by the PBP, the recognition
culminates in the formation of a stable acyl-enzyme
species. Hydrolysis of the acyl-enzyme ester bond
is slow in PBPs, with half-lifes that substantially
exceed the doubling time of the organism. To appreciate the degree of inefficiency of this step in
PBPs, comparison of the deacylation rate constant
with that of -lactamasessresistance enzymes that
are believed to have descended from the PBPss
reveals that the rate of deacylation in PBPs is up to
6 orders of magnitude slower.49 The irreversibility of
the deacylation step in PBPs is at the root of the
antimicrobial action of -lactam antibiotics.
The basis for these differences must clearly derive
from the differences in the targetsPBP and -lactamasessstructures. The first X-ray diffraction structures of two low molecular weight PBPs were solved
nearly 25 years ago.50-52 Since then only a handful
of additional structures have been solved, despite the
very large number of PBPs that are now known.
Those with solved structures include the low Mr PBPs
from Streptomyces R6153 and Streptomyces K15 (PDB
code 1SKF),54,55 a zinc-dependent PBP from Streptomyces albus G (PDB code 1LBU),50 the PBP5 from
Escherichia coli (PDB code 1NZO),56 and the high Mr
(and -lactam resistant) PBP2x (PDB code 1PMD)48
from Streptococcus pneumoniae and PBP2a from
Staphylococcous aureus.57 A comparison of these
structures reveals similarities but also differences.
The fold of the transpeptidase/carboxypeptidase appears to be conserved among PBPssthe exception
would appear to be the abundant penicillin-binding
protein from Treponema pallidum, Tp47, which has
a unique multidomain fold.58 This unit consists of two
regions: one is an all R-helix, and the second is a
mixed R/ structure consisting of a -sheet that is
flanked on both sides with R-helices. An indication
of their similarities can be gleaned from a superimposition along the common backbone atoms of the E.
coli PBP5 and the Streptomyces K15 PBP that results
in a 1.2 rms deviation. Whereas the function of the
transpeptidase/carboxypeptidase domain among these
PBPs is known, the function of additional domains
that are found in the structures of both low- and highMr PBPs remain unknown. One example, in PBP5
from E. coli the additional unique domain is found
in a nearly perpendicular orientation relative to the
transpeptidase-like domain (that is shown in Figure
2A).59 This domain is composed of two- and threestranded antiparallel -sheets with noticeable hydrophobic properties.56,59 Other examples include two
high molecular weight class B PBPs: PBP2a from
S. aureus and PBP2x from S. pneumoniae. The threedimensional structure of PBP2a (Figure 2B) shows
a non-penicillin-binding domain and an N-terminal
domain whose functions remain unknown.
The similarity in the three-dimensional structure
of the carboxypeptidase/transpeptidase domains of
PBPs is also matched by a high degree of similarity
in the relative position of residues from three highly

Fisher et al.

Figure 2. (A) Stereoview of the active site of the PBP2x


from S. pneumoniae (cyan), the PBP2a from S. aureus
(magenta), the PBP from Streptomyces K15 (orange), the
D,D-transpeptidase/D,D-carboxypeptidase from Streptomyces
R61 (red), and the PBP5 from E. coli (yellow) superimposed
along the R-carbons of the conserved residues (shown in
capped-sticks representation). The cyan, yellow, green, and
white arrows point to the conserved lysine from the SXXK
motif, the serine from the SXXK motif, the lysine/histidine
from the KTS/KTG motif, and the serine/tyrosine from the
SDN/SGN/TXN motif, respectively. (B) Stereoview of the
three-dimensional structure of PBP2a from S. aureus
shown in ribbon representation. The non-penicillin-binding
domain is shown in yellow and green representation; the
yellow domain corresponds to the N-terminal domain. The
structure that is red corresponds to the D,D-transpeptidase
domain. (C) Stereoview of the three-dimensional structure
of PBP5 from E. coli shown in ribbon representation. The
arrow points to the flexible -like loop.

conserved motifs (Figure 2C). The first motif is the


strictly conserved (both PBPs and -lactamases)
SXXK tetrad. The serine corresponds to the amino
acid that is activated to undergo acylation in both
peptidase and -lactamase. It is located at the Nterminus of a helix, which likely modulates its pKa,

Bacterial Resistance to -Lactam Antibiotics

thus facilitating its activation as a nucleophile and


nucleofugacity as a leaving group.46 The lysine in this
motif is the general base that activates the serine for
acylation.60 The second conserved motif is the (S/Y)XN tripeptide sequence. This triad is SDN in Streptomyces R61, SGN in the Streptomyces K15 D,Dtranspeptidase, and YSN in the Streptomyces R61
D,D-peptidase. The S/Y residue in this motif is thought
to be required for back-donation of a proton to the
nitrogen atom of the -lactam ring after formation
of the tetrahedral intermediate. From the superimposition of the amino acids within the active sites of
these PBPs (as shown in Figure 2C) it appears that
the position of the hydroxyl (whether serine or
tyrosine) is conserved (comparing the SDN and YXN
motifs). The third conserved motif in PBPs is a KTS
or KTG motif (except in the case of Streptomyces R61
where it is an HT(S/G)). In -lactamases the role of
this lysine (or histidine) is important in modulating
the pKa of the universally conserved lysine that is
two residues downstream of the active site serine.
The overall similarity of the transpeptidase/carboxypeptidase region and the conserved relative
positions of highly conserved residues in PBPs do not
translate into similar catalytic (or functional) behavior. PBP5 of E. coli is a case in point. While most
PBPs have low deacylation rate constants, PBP5 has
an unusually high deacylation rate constant with a
half-life t1/2 < 10 min with penicillin G; this is to be
contrasted to the more typical deacylation rate
constant of 8.3 10-6 s-1 for the acyl-enzyme species
of PBP2a with the same substrate. What makes this
PBP so unusual? Earlier studies had shown that a
G105D mutation reduced the deacylation rate by 30fold.61 The X-ray structure of this mutant did not
reveal the basis for the reduced deacylation59 since
the mutation does not occur in the active site.
However, the X-ray structure of the wild-type enzyme56 revealed disorder at a loop located near the
active site serine. The position of this loop is reminiscent of the -loop (vide infra) in class A -lactamases, based on the superimposition of the TEM-1
-lactamase and PBP5.59 The different conformation
of the loop adopted by the mutant likely contributes
to the deacylation rate difference. This understanding
takes us to the curious event (and a theme of this
review). What are the molecular events that result
in failed recognition of -lactam antibiotics by the
PBPs? Alteration of the PBPs is a dominant mechanism of Gram-positive resistance. For this compelling reason the focus of Gram-positive PBP biochemistry has changed from the historical (the enzymes
of the nonpathogenic Bacillus subtilis)62 to those of
the resistant and pathogenic S. pneumoniae and S.
aureus. How are the PBPs of these resistant pathogens different? Two limiting possibilities exist. The
PBPs are the same but exist in increased copy
number or the PBPs are altered by selection of
mutant variants so as to diminish recognition of the
-lactam without compromise of the peptidoglycan
biosynthetic role. Both processes exist, but it is the
latter that has proven to be the most versatilesand
expandingsmechanism of Gram-positive bacterial
resistance.

Chemical Reviews, 2005, Vol. 105, No. 2 401

The expansion of resistant PBPs is a medical


problem with microbiological (what is the ecological
circumstance where resistance is acquired from one
bacterium by another), molecular biological (what are
the mutations, and what is the genetic basis for their
transfer), biochemical (how do these mutations defeat
recognition of the -lactam as a mechanism-based
PBP inhibitor), and chemical (what will be the design
of new generation -lactam antibiotics effective against
resistant pathogens) manifestations. A decrease in
the spread of antimicrobial drug resistance will
require societal change and scientific discovery in
response to each of these manifestations. Of these
four we will briefly address the microbiological and
molecular biological, and focus on the biochemical.
The topic of the chemical is reviewed elsewhere.24
An excellent point of biochemical entry to the
reality of S. aureus -lactam resistance is Pucci and
Doughertys analysis, by saturating penicillin inactivation, of the PBP distribution and stoichiometry
in susceptible and resistant S. aureus.63 While it is
long known that substituent changes made to the
penicillin and cephalosporin periphery influence relative affinity (specificity) among the penicillin-binding
proteins, by judicious substituent choice and high
concentration it is nonetheless possible to saturate
(to titrate to the point of complete inactivation) the
entire ensemble and so obtain their relative abundance (copies per bacterium). Susceptible S. aureus
contains four PBPs, three of which are high molecular mass enzymes (70-80 kDa) and one of which is
a low molecular mass enzyme (46 kDa). The high Mr
enzymes are PBP1 (approximately 185 enzymes per
cell), PBP2 (460 enzymes), and PBP3 (150 enzymes).
The low Mr enzyme is PBP4 (285 enzymes per cell).
While this is a smaller number of enzymes than is
found in other bacteria species (B. subtilis has 12;
E. coli has 16) the critical biosynthetic enzymes (with
respect to -lactam lethality) for all are the (vide
infra) high Mr enzymes, which are usually bifunctional (one activity is the -lactam-sensitive transpeptidase activity, and the second is a non--lactamsensitive transglycosylase activity). For S. aureus the
bifunctional enzyme target of the -lactams (wherein
the transpeptidase but not the transglycosylase
activity is inhibited) is the PBP2 enzyme.
What is the PBP composition of the -lactamresistant S. aureus? A comparison of the PBP composition of a resistant strain, by saturating penicillin
inactivation, shows that the resistant bacterium
contains the same four PBPs as the susceptible
bacterium (and in the same quantity per bacterium
as the susceptible bacterium) and an additional fifth
enzyme (termed PBP2a).63,64 The PBP2a is present
in substantial quantity (approximately 800 copies per
cell with some variability) and is a low affinity
enzyme with respect to -lactam binding and inactivation. When resistant S. aureus is challenged by
a -lactam antibiotic, the transpeptidase activity of
its PBP2 enzyme is inactivated but the transpeptidase activity of its PBP2a enzyme is unaffected. The
PBP2a performs the transpeptidase function of the
now inactivated PBP2, and the bacterium survives.
Simply stated, the -lactam concentration attained

402 Chemical Reviews, 2005, Vol. 105, No. 2

during chemotherapy is insufficient to inactivate the


transpeptidase activity of this new enzyme. (The
PBP2a contains a second domain that is presumed
to possess a catalytic function which is not transglycosylase activity. The role of this second PBP2a
domain remains unknown.) In the presence of -lactam antibiotics the functioning transglycosylase domain of the PBP2 (its transpeptidase having been
inactivated) works cooperatively with the active
transpeptidase of the PBP2a to maintain the cell wall
integrity of the resistant S. aureus.65,66 Therefore, the
salient issues to the understanding of Gram-positive
-lactam resistance are the circumstance of PBP2a
acquisition, the genetic origin of PBP2a, and the
molecular alteration(s) within the transpeptidase
PBP2a domain that result in a change from high to
low susceptibility to -lactam inactivation.
The appearance of methicillin resistance soon followed (within a year) the introduction of methicillin
in the clinic.67 The mechanism by which the mecA
gene, carried by mobile genetic elements known as
the staphylococcal cassette chromosome mec (SCCmec),68 was acquired by S. aureus remains unknown.
It is suggested that this gene was acquired from
Staphylococcus sciuri (a bacterium found in the gut
of animals) which possesses a close mecA gene
homologue.69 Upon activation of the mecA gene, the
PBP2a protein is expressed. It was shown in the mid1980s that the presence of PBP2a in S. aureus conferred resistance to the clinically used -lactams,70-72
as evidenced by a 500-fold increase in the MIC
(minimal inhibitory concentration) for penicillins.
These bacteria came to be known as methicillinresistant S. aureus (or MRSA). Kinetic characterization of the reaction of PBP2a with -lactams provided
valuable mechanistic information and revealed that
the resistance to -lactams was not merely due to a
large Kd value (a commonly held belief) but to the
acylation rate constant as well.73 The dissociation
constant of the PBP2a-benzylpenicillin complex was
13 mM,74 similar to what is found for other (susceptible) PBPs. A much clearer difference is the apparent
second-order rate constant for acylation of the active
site serine. For the reaction of PBP2a with benzylpenicillin this rate constant is 2-3 orders of magnitude smaller than those found for other high molecular weight PBPs (such as the S. aureus PBP2 and
S. pneumoniae PBP2x).74 The structural features of
PBP2a that are responsible for the poor acylation
were elucidated recently with X-ray structures for the
apo and acyl-enzyme complex of PBP2a with benzyl-penicillin, nitrocefin, and methicillin.57 Comparison of the apo-PBP and acyl-PBP structures revealed noticeable differences. In the acyl-enzyme
structure the CR, C, and O of Ser403sthe acylated
serinesare 1.1, 1.4, and 1.8 away from the same
atoms in the apo-structure, suggesting that the
R-helix that holds Ser403 must undergo a conformational change for acylation to occur.57
Structural modifications in PBPs that result in
increased resistance are not confined to S. aureus.
The three-dimensional structures of two additional
PBPs have been solved by X-ray crystallography. One
is PBP2x from S. pneumoniae. PBP2x is a class B

Fisher et al.

high molecular weight PBP, and its structure was


solved nearly a decade ago.48 The X-ray diffraction
structure of a mutant PBP2x provided the first
glimpse into the structural bases for resistance by
mutation of PBPs.75,76 Two drug-resistant PBP2x
mutants have been characterized. The first X-ray
structure describes the effects of mutations Thr338Ala
and Met339Phe, which along with other mutations
alter the acylation efficiency by 20-fold. Both of these
mutations occur close to the active site Ser-337, the
residue that is acylated by -lactam antibiotics. The
effect of these mutations is attributed to the disruption of hydrogen-bonding interactions between Thr388
and a conserved water molecule. Also, a conformational change that occurs for the 3 strand is attributed to the collective effects of the larger Phe339
side chain and smaller Ala338 chain. These conformational changes enable an alternative conformation
for Ser337 that might be less prone to activation. The
most likely candidate for the base that promotes
Ser337 acylation is Lys340, whose amine is in
hydrogen-bonding contact (3 ) with the serine
hydroxyl. The structure of another mutant of PBP2x
has been recently solved.75 It reveals similar conformational changes as the PBP2x above except that in
this case the effects are attributed to the change in
polarity introduced by Gln552Glu and to a narrower
active site.
PBP5, a transpeptidase from Enterococcus faeciums
a bacterium which incidentally does not produce
-lactamasesshas also been implicated in resistance
to -lactams through either modification or overexpression of the enzyme.77,78 The enterococci are less
virulent than S. aureus and S. pneumoniae but have
become prominent in the clinic due to their increased
levels of resistance to a variety of antibiotics.79
PBP5fm has low affinity for -lactam antibiotics. Two
reasons for this decreased affinity are suggested from
the structure of this protein.80 A glutamate (Glu622)
near the active site may present a steric barrier to
-lactam binding. This interpretation is consistent
with the reduced affinity for benzylpenicillin that
results when the equivalent site in PBP2x is changed
to glutamate.81 Second, an arginine (Arg464) may
interact with its neighbors in the conserved loop
spanning residues 461-465, resulting in a more rigid
cleft that would lead to the reduced affinity of
PBP5fm.80

2.2. -Lactam-Sensing Proteins


Resistance to antibiotics in Gram-positive and
Gram-negative bacteria has manifested itself through
various mechanisms, including the production of
-lactamases or PBPs that are insensitive to the
action of -lactams, such as the case of PBP2a from
S. aureus. In the mid-1980s Lampen and co-workers
observed that the synthesis of the -lactamase from
Bacillus licheniformis 749/I gradually peaks at 1-1.5
h after exposure to a -lactam and decreases slowly
in the following 1-2 h82 (induction in S. aureus is
far more rapid, complete within 11 min83). This
experiment confirmed that -lactamase production is
inducible and implied the presence of a transduction

Bacterial Resistance to -Lactam Antibiotics

mechanism. The identification of a membrane-spanning protein, BlaR, from B. licheniformis soon followed. That this enzyme contained a sensor/transduscer domain that is highly similar to the class D
-lactamases,84 a membrane domain consisting of a
four-helix bundle85 and an intracellular domain
containing a zinc ion,84 made it a strong candidate
to carry out the signal transmission events in a
transduction mechanism. The BlaR protein is the
product of the blaR1 gene, which is a member of a
triad of genes from the bla divergon; blaP and blaI
are the remaining genes that encode the effector
protein (-lactamase) and repressor protein BlaI.86
The BlaI proteinsa DNA-binding repressor protein
that is located immediately upstream of the genes
blaP and blaRsblocks expression of both structural
and regulatory genes, including itself.87
The regulation of the production of resistance
enzymes has also been recently studied in S. aureus,
an organism that, in addition to -lactamases, produces a low-affinity PBP, namely, PBP2a, which is
regulated by a similar mechanism that involves a
sensor-transducer (mecR), a DNA-binding repressor
(mecI), and a structural gene (mecA, Figure 3A). It
is worth noting that the mere presence of the mecA
gene is insufficient for expression of resistance in S.
aureus as yet other (and yet unknown) genetic
changes are also necessary.88,89 The bla or mec
regulatory genes regulate production of PBP2a and
-lactamase due to a high degree of homology between the two systems.90 However, the inability of
-lactams to induce PBP2a in S. aureus and the fact
that the blaI/blaR system interacts with the mecA
promoter indicate that this system could also be
responsible for the induction of mecA transcription.91
The sequence of events that lead to expression of the
blaZ gene (for -lactamase) in S. aureus is similar to
that of B. licheniformis: following the binding of a
-lactam to the sensor domain of BlaR, a signal is
transmitted across the membrane and leads to activation and autocatalytic cleavage of the intracellular
zinc-ion-dependent domain of blaR; the activated
metalloprotease either directly or with the aid of
cofactors cleaves the DNA-bound repressor protein
BlaI,92 which is left unable to dimerize and efficiently
bind to its operator for blockage of expression of the
structural genes.90 Autocleavage of BlaR leads to
incapacitation of the protein, which has to be regenerated continuously. The protein presumably expressed by the blaR2 gene is proposed to play a role
in the induction mechanism, but its exact role
remains unelucidated.
Transcription of the genes encoding -lactamases
is set in motion by the binding of a -lactam antibiotic
to the extracellular sensor domain of BlaR. The
kinetics of this process have been characterized for
B. licheniformis and S. aureus. Kinetics of BlaR from
S. aureus with various -lactams shows that a single
acylation event occurs over the lifetime of the organism, making BlaR like a PBP.93 The acylation step
is efficient (second-order rate constant k2/Ks of
104-106 M-1 s-1, while the first-order deacylation rate
constant has an exceedingly slow value of 10-5 s-1.93
The three-dimensional structure of BlaR from B.

Chemical Reviews, 2005, Vol. 105, No. 2 403

Figure 3. (A) Schematic of the various proteins that are


involved in regulation of -lactamase and PBP2a. All
proteins are color-coded based on their secondary structures and shown in ribbon representation. The lipid bilayer
of the inner membrane is shown with green spheres
representing the phospholipid head, and the lines represent
the lipid tail. The sensor/transducer BlaR is anchored to
the surface through a helix bundle, shown in red ribbon
representation, which culminates in the C-terminal metalloproteinase domain, located in the cytoplasm. PBP2a is
shown anchored to the membrane, and a class A -lactamase (BlaZ) is shown unanchored in the periplasm. Shown
in the cytoplasm is the MecI dimer (ribbon) in complex with
its operator DNA (capped-sticks). (B) Close-up depiction
of the complex of MecI dimer with its operator DNA. The
MecI dimer is shown in ribbon representation with the
monomers colored in gray and yellow, while the DNA
oligonucleotide is shown in blue capped-sticks representation with a ribbon along the duplex backbone. The red
arrows point to the cleavage sites in the two monomers.

licheniformis has been recently solved by X-ray


crystallography94 and confirms the postulated high
degree of similarity between the C-terminal domain
of BlaR and the class D -lactamases,84 While the
three-dimensional structure of BlaR is nearly identical to that of the class D -lactamases, the X-ray
structure of this enzyme does not reveal N-carboxylation at the active site lysine,94 an event that is
widely believed to occur in the class D -lactamases.95-97 Previous kinetic studies of the BlaR
protein from S. aureus had found that BlaR from S.
aureus contains an active site lysine (Lys392) that
reacts with CO2 to form a carbamate.93 The authors
of the X-ray structure suggest that the presence of a
threonine residuestwo residues downstream of the

404 Chemical Reviews, 2005, Vol. 105, No. 2

active serine residue where a valine residue would


be usually located in class D -lactamasesis likely
the reason behind the lack of carboxylation of that
lysine residue.94 However, the signature 13C NMR
signal for N-carboxylated lysine in BlaR of S. aureus
has been documented in solution.93
More recently, the structures of the acyl-enzyme
complex of BlaR from S. aureus with benzylpenicillin
and ceftazidime have been independently solved by
X-ray crystallography.98,99 These confirm the similarities in the structures of BlaR from S. aureus to
BlaR of B. licheniformis and the class D -lactamases.
Interestingly, both of these studies found that BlaR
was not carboxylated in the acyl-enzyme complex,
in contrast to the 13C NMR, fluorescence, and mutagenesis studies identifying the carboxylated lysine
in the resting BlaR protein.93 QM/MM calculations
carried out by Birck et al.99 reveal that upon protonation of the carbamate nitrogen in the class D
-lactamases a barrierless decarboxylation occurs.
The authors postulate that the same N-decarboxylation event occurs in BlaR, but unlike the class D
-lactamases decarboxylation results in an inactive
enzyme unable to recarboxylate due to a hydrogen
bond between Lys392 and Asn439. This conclusion
is consistent with the fact that BlaR undergoes a
single acylation event over the lifetime of the organism.
The nature of the signaling event across the
membrane that transpires as a result of -lactam
binding to the sensor domain of BlaR is not well
understood. Golemi-Kotra and co-workers have shown
that binding of the antibiotic to BlaR of S. aureus is
accompanied by significant conformational changes
that likely have a role in the signal transduction
mechanism.93 A recent study of the transduction
mechanism in B. licheniformis did not identify a
conformational change in the C-terminal domain.100
An alternative mechanism was noted based on an
interdomain conformational change in the membrane
protein consisting of the loss of interaction between
the C-terminal domain and an L2 loop of BlaR that
connects two R-helices of the four-helix bundle. It was
shown that in the absence of the antibiotic the sensor
domain and the L2 loop form an interaction. Recent
mutations of highly conserved residues in the L2 loop
appear to be lending credence to this mechanism, as
the organisms exhibited no -lactamase activity,
since the level of antibiotic remained at similar levels
to that of a -lactamase-negative control.98 Whereas
the mechanism that leads to the transduction of the
signal appears to be contentious, the event that
follows is accepted to consist of autocleavage of the
zinc-containing intracellular domain,87 which in the
case of S. aureus is followed by inactivation of the
repressor proteins MecI/BlaI, while in B. licheniformis this process is thought to occur through an
intermediary coactivator.101
The structures of BlaI102 and MecI103,104 have been
recently solved by NMR and X-ray crystallography,
respectively. The BlaI and MecI proteins from S.
aureus share 60% identity and 31 to 41% identity to
BlaI from B. licheniformis.102 The structure of MecI
consists of a dimer in the shape of a triangle (shown

Fisher et al.

in Figure 3B).104 Dimerization occurs at the C-terminal domain, while the DNA-binding domain is located
at the N-terminus (see Figure 3B for structure of
MecI-DNA complex). The topology of MecI follows a
winged-helix architecture103,104 with a helix-turnhelix DNA-binding motif; the second helix of this
motif binds to the major groove of DNA with up to
16 hydrogen bonds and salt bridges (see Figure
3B).104 The high level of conservation of the residues
that form contact between the repressor protein and
the operator DNA in S. aureus and B. licheniformis
suggests that this complexation is likely similar in
BlaI and MecI.104

3. -Lactamases
3.1. Overview and Classification
The -lactamases predate the antibiotic era. The
evolution of these enzymes is presumed to have taken
place in parallel to the biosynthetic steps leading to
-lactam antibiotics.105 Indeed, the first -lactamase
was identified in the early 1940s prior to the first
large-scale use of penicillins in Boston 2 years
subsequent.16 However, extensive clinical use of these
antibioticss-lactams comprise approximately 55%
of all antibiotics used currentlyshas accelerated the
selection process for the emergence of once rare genes
for these antibiotic-resistant enzymes. The rare
bacterium that harbored the gene for a -lactamase
would have had the opportunity to grow unencumbered once the susceptible organisms in a heterogeneous population of bacteria were eliminated in the
course of an antibiotic treatment regimen. In essence,
the less fit bacterium is eliminated by the -lactam
challenge and the resistant organism experiences
unlimited nutritional resources to propagate. The
once rare gene for the -lactamase is amplified.
Sharing of genetic materials among microbial
populations is relatively facile. Genes, often residing
on inherently mobile genetic elements such as plasmids and transposons, are shared not merely members of a given species of bacteria but also among
unrelated genera. The facility of genetic sharing is
underscored by the observation that some organisms,
such as the Streptoccoci, are able to acquire freestanding stretches of nucleic acids (containing entire
genes) directly from the environment. All these
processes have contributed by clinical selection to the
amplification of once rare antibiotic-resistant genes
and their liberal sharing among various bacterial
populations.
The account given above outlines the plausible
events that have given rise to the emergence of the
parental genes for -lactamases. As a consequence
of the inappropriate use of -lactam antibiotics for
the past half a century, especially in the community,
many variants of the parental enzymes have emerged.
This accelerated evolution of the antibiotic-resistant
genes has been abetted by the creative molecular
tinkering of medicinal chemists in the past decades,
the fruits of which are an ensemble of -lactam
antibiotic structures. The dynamics between the
discovery, the creation of new -lactam antibiotics,
and the clinical responses by microbial population to

Bacterial Resistance to -Lactam Antibiotics

these developments have been outlined by Bush and


Mobashery.106 In light of the different properties of
these various types of -lactam antibioticssfor example, antibiotics that target different sets of PBPs
or those that have been imparted with resistance to
the action of -lactamasessthe clinical response by
bacteria has been the selection of mutant variants
of -lactamases that often have broadened the catalytic ability of the enzymes. For example, the TEM-1
-lactamase, a plasmid-borne enzyme described first
in 1963, has given rise to 133 variants (as of February
2004).15 While this variety does reflect the successful
therapeutic use of -lactams, it also may be taken
as presaging the obsolescence of these versatile
antibiotics sometime in the future.
Prompted by a critical biochemical requirement,
nature deftly develops catalysts to meet the need.
Despite the outstanding stability of the peptide bond
(half-life of approximately 500 years when uncatalyzed for hydrolysis),107,108 multiple classes of proteases have evolved to hydrolyze this bond, in light
of the central importance of proteolysis to many
biological processes. The same has been true for
-lactamases in the face of the life or death options
to the organisms presented by the antibiotics. There
are four known classes of bona fide -lactamases,
each of which operates by a distinct reaction mechanism.15,105,109,110
While a handful of -lactamases were known in the
early 1970s, the number at the present exceeds 470
(communication by Dr. Karen Bush). Two general
types of -lactamases are known: those that require
the zinc ion for their function, and those that pursue
a transient serine acylation/deacylation strategy (if
the unique -lactamase activity of the T. pallidum
PBP is also found in other organisms, this will yet
be another type of -lactamase as it does not require
a zinc ion nor does it pursue the covalent catalytic
strategy of the serine enzymes).58,111 The widely
accepted molecular classification places -lactamases
into four classes: three serine-dependent enzyme
classes (classes A, C, and D) and one metal-dependent (class B). This classification is not to be confused
with that of Ghuysen for the PBPs in which the two
groups of low molecular weight and high molecular
weight PBPs are divided among classes A, B, and C
(for a total of six PBP classes).
Both PBPs and -lactamases are present in the
periplasmic space of Gram-negative bacteria. In
Gram-positive organisms (which lack the outer membrane) the PBPs are located on the outer surface of
the cytoplasmic membrane and the -lactamases are
either excreted or bound to the cytoplasmic membrane.112 All -lactamases are expected to have
divested completely their ability to bind the peptidoglycan substrate of their ancestral PBPs.14,113 If
not, the opportunity to function as a vanguard
against the incoming antibiotics would be lost (the
structural means to this end was revealed recently
for the class C -lactamases).113 Moreover, this
same conclusion may be intuited from the ability of
many of the class A and C -lactamases to act at
the diffusion-limited rate for their preferred substrates.114,115

Chemical Reviews, 2005, Vol. 105, No. 2 405

A summary of the mechanistic expectations for


-lactamase catalysis is a useful prelude to the
discussion of the relationship between -lactamase
structure and the evolution of function (and mechanism) to confer -lactam resistance. Entry of the
-lactam substrate is guided by relatively long-range
electrostatic attractions between the cationic side
chain of an active site amino acid and the carboxylate
of the -lactam. Positioning of the -lactam then
occurs by shorter range attractive electrostatic interaction involving hydrogen bonding from the protein to the -lactam carbonyl (the oxyanion hole,
formed in the class A -lactamases by the hydrogen
bonding to the -lactam carbonyl by the backbone
amide nitrogens of Ser-70 and Ala-237).116 Appropriately functionalized -lactams (here with special
reference to the bicyclic structure of nearly all of the
antibacterial -lactams) sequester in the active site.
This complex formation results in localized (ground
state) destabilizing electrostatic interactions at the
locus that enable catalysis, compensated by stabilizing hydrophobic interactions elsewhere in the enzyme-substrate complex. The higher carbonyl IR
stretching frequency of the -lactam when it is in the
oxyanion hole, indicative of enhanced reactivity
toward nucleophile addition, exemplifies localized
ground-state destabilization.117 Moreover, within the
complex the total ensemble of the remaining amino
acids are now predisposed to initiate turnover. The
outcome of this positioning is a decrease in the
transition-state energy by the simultaneous operation
of Lewis (clearly exemplified by the zinc atom of the
metalloproteases but other through-space electrostatic interactions as well) and Bronsted catalysis.
The rate-limiting step of nonenzymatic -lactam
hydrolysis is oxyanion addition to the -lactam carbonyl, and there is little doubt that this event is also
rate limiting for hydrolysis by class A -lactamases.34,116
For the serine (classes A, C, and D) -lactamases
(the metallo--lactamases are discussed subsequently)
the critical result of Henri-Michaelis complex formation is to attain such stabilization of the incipient
tetrahedral species so as to enable general base
catalysis for serine addition to the -lactam. Upon
serine oxygen addition considerable basicity develops
on the nitrogen. At the point in the transition state
of (substantial) negative charge transfer (from the
general base, through the serine) to the tetrahedral
species, a concomitant increase in the -lactam
nitrogen basicity is attained as to accept proton
donation to this nitrogen. The source of this proton
is debated. Nonetheless, nitrogen protonation is an
obligatory event for productive tetrahedral collapse.
As noted by Page and Laws,34 oxyanion addition to
the less hindered re face of the -lactam and the
Bronsted general acid N-protonation both must occur
on this same face. This imposes a spatial (stereoelectronic) constraint on the positioning of these two
catalytic groups within the active site (if they are
allowed contact with each other they will simply selfannihilate by proton transfer). An obvious possibility
is the use of the protonated amino acid that was used
as the general base for oxyanion addition (as it is

406 Chemical Reviews, 2005, Vol. 105, No. 2

already on this face and makes the self-annihilation


issue moot). Hydrolysis of the serine acyl-enzyme,
the deacylation step, is energetically less demanding
(a more reactive carbonyl and thus having a diminished need for general acid catalysis in tetrahedral
collapse).
Within the active sites of the serine -lactamases
are, therefore, two ensembles of amino acids. The
first is the catalytic ensemble comprised of a minimum of five amino acids: the serine, the general base
for the serine, the oxyanion hole (two amino acids),
and the cationic recognition site for the carboxylate.
These amino acids are expected to be invariant (or
very highly conserved). The second is the recognition
ensemble (in one form or another, all of the remaining amino acids) that complements the hydrophobic
and hydrophilic segments presented by the remaining structure of the -lactam. These recognition
amino acids will be variable and correspond to the
amino acids that will mutate under selection pressure. For a given -lactamase and given substrate
the complementarity to the rate-limiting transition
state and to overall recognition of the -lactam will
vary. This variability is, of course, quantified as the
unique kcat/Km for the substrate. While this may seem
obvious, there is a less appreciated corollary. The
relative individual role played by any member of
these ensembles during catalysis (measured, say, in
terms of an amino acid pKa) is therefore substrate
dependent. Deletion of an essential amino acidswhile
an essential tool of mechanistic enzymologysalters
the energetic landscape of the active site in such
fashion as to make all subsequent interpretation
cautionary. The criteria that determine which amino
acids of the recognition ensemble transform under
selection pressure are straightforward. The integrity
of the catalytic ensemble must be preserved, and the
integrity of the protein as the whole (for example, as
measured in terms of thermal stability or expression/
folding capability) can only be lightly varied.118-121
Likewise, mutations that require only a single nucleotide change and that preserve common codon usage
are more probable than those that do not.120 This
implies that certain -lactamase families are anticipated to be more plastic than others, and indeed,
the serine -lactamases provide such a contrast with
the great phenotypic diversity of the serine class A
TEM enzymes as compared to the class A SHV
enzymes, which have diversified but without substantial phenotypic evolution.15 Given these boundary
conditions and superlative kinetic and structural
data for these enzymes, one might presume that the
assignment of function within the catalytic ensemble
and within the recognition ensemble as these develop
under selection pressure would be straightforward.
One would be wrong.

3.2. Class A -Lactamases


This is the largest and best mechanistically characterized serine -lactamase class. Historically, these
-lactamases were described as penicillinases as
their ability to catalyze penicillin hydrolysis was
greater than that for cephalosporins. They have
become so efficient at their function that they are

Fisher et al.

diffusion controlled, where the apparent second-order


rate constant kcat/Km has reached the (upper) diffusion limit estimated from collision theory. As a result,
class A -lactamases may be described as having
reached catalytic perfection for their preferred substrates.114 Variants with much broader substrate
preferences are now known, including enzymes imparting clinical resistance to late generation -lactams.15,109,122-124 The class A -lactamases are closely
related in sequence to low molecular weight class
C PBPs such as PBP4 of E. coli, H. influenza, and
M. tuberculosis.60 As judged by the comparison of
crystal structures, the catalytic domain of the larger
E. coli PBP5 (low Mr PBP class A) shows high
similarity as well.56,125 In terms of bacterial resistance, three class A -lactamases subclasses dominate: the (historically Gram-negative plasmid penicillinase) TEM/SHV, the P. aeruginosa PER/OXA/
TOHO cephalosporinases, and the CTX-M (NMC-A)
carbapenemase subclasses.126 As of October 2004, 135
TEM and 57 SHV -lactamase variants are known
(http://www.lahey.org/studies/webt.htm). While the
sequence homology among the three is easily recognizable and the fundamental catalytic mechanism for
each is the same, the differences render broad structural and mechanistic generalizationssespecially as
they relate to resistance developmentsunwarranted.
Several class A variants resist inactivation by the
mechanism-based inhibitors (clavulanate and sulbactam) used in clinical formulations with otherwise
-lactamase-susceptible penicillins. Until very recently the occurrence of these inactivation-resistant
class A -lactamases was limited to the aforementioned TEM subclass, and hence, the term inhibitorresistant TEM (IRT) was coined. However, in light
of the discovery of inhibitor resistance in the SHVtype enzymes, this group is better referred to as
inhibitor-resistant -lactamases. Furthermore, new
class A -lactamases that are active against the more
recent cephalosporins (ceftazidime and cefotaxime
and the monobactam aztreonam) and others that are
active against the carbapenems are known collectively (also with other class C and D enzymes) as
expanded-spectrum -lactamases (ESBL).124
The crystal structure of several Gram-positive
(including enzymes from B. licheniformis and S.
aureus) and Gram-negative (from E. coli) class A
-lactamases were solved in the late 1980s and early
1990s.127-131 The TEM-1 -lactamase structure has
two domains: an R/ domain consisting of fivestranded -sheet and three R-helices and an R
domain consisting of eight R-helices.128,131 Together,
these two domains sandwich the core of the active
site. The class A -lactamases reveal striking similarities to the PBP structures132 (compare Figure 4A
and B, D, and F). Further similarities can be found
in the relative three-dimensional position of several
highly conserved residues, including Ser70, Lys73,
Lys234, and Ser130. While some of these residues
are not invariant, their substitutes contain similar
functional groups capable of carrying out the same
functions (for example, Ser130 is typically replaced
by tyrosine in the class C -lactamases, and Lys234
is sometimes replaced by a histidine as in the

Bacterial Resistance to -Lactam Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 407

Figure 4. Stereoviews of the three-dimensional structures of (A) a class A -lactamase (TEM-1; PDB code 1TEM), (C) a
class C -lactamase (AmpC; PDB code 1FCO), (E) and a class D -lactamase (OXA-10; PDB code 1K57). Close-up stereoviews
of the active sites of the acyl-enzyme complex are shown as (B) TEM-1 with 6R-hydroxymethylpenicillate, (D) AmpC with
moxalactam, and (F) OXA-10 with 6-(1-hydroxy-1-methylethyl)penicillanic acid. The enzymes are in yellow ribbon
representation with a van der Waals surface in blue for the active site. The important active site residues are depicted in
capped-sticks representation (color-coded according to atom type: S, yellow; O, red; N, blue; C, white). The hydrolytic
water molecule is shown as the red sphere. Hydrophobic residues in the active site and other residues that are close to
important residues but are not directly involved in the catalytic process are shown in orange capped-sticks representation.

structure of the Streptomyces R61 D,D-peptidase/


transpeptidase).
The remaining invariant amino acid is Glu166.
This glutamate is located on a loop (termed the
-loop) that sequesters, by hydrogen bonding, a
single water molecule as a bridge to the Ser70
adjacent to the re face of the -lactam carbonyl. This
glutamate has been proposed by some to have a role
in the rate-limiting acylation step.133-137 This mechanism envisions that this glutamate, acting through
the bridging water as a proton shuttle, activates
Ser70 for nucleophilic addition to the -lactam.136,138
The alternative possibility for the general base, the
invariant Lys73,129,131 must then be electrostatic
stabilization (such as to increase the Ser70 acidity
and hence nucleophilicity). The evidence in favor of
Glu166 as the serine-activating general base is
summarized. A compelling argument in favor of a
Lewis acid, rather than a Bronsted base, role for
Lys73 was the 15N NMR determination of its pKa as
greater than 10.139 This determination was followed
by Poisson-Boltzmann electrostatic calculations by
Lamotte-Brasseur et al. supporting the pKa > 10
assignment,140,141 although others (using the same
calculation method) estimated a value of 8.142 More
recently each of three methodssenzyme kinetics, 15N
NMR, and free-energy calculations using the ther-

modynamic integrationssupports a pKa value for


Lys73 of 8.0-8.5.143 Ultrahigh-resolution crystal
structures of the TEM-1133 and SHV-2136 enzymes
show a spatial arrangement of Glu166 and the
invariant water, in the presence of a bound transition
state mimic, to be consistent with serine activation
via the proton shuttle.133 The TEM-1 structure resolves a hydrogen atom on Glu166, while the SHV-2
structure shows the hydrogen of the Ser70 hydroxyl
pointing to the conserved water molecule. Following
serine addition the Ser130 hydroxyl is positioned
ideally to shuttle the proton on Lys73 to the -lactam
nitrogen of the tetrahedral species (Figure 4B),
driving ring opening to the acyl-enzyme species. It
is widely accepted that Glu166 is the general base
for hydrolysis of the acyl-enzyme ester. In fact, sitedirected mutagenesis of Glu166 (E166A) abolishes
deacylation while impairing (but not abolishing) the
acylation process by a factor of 103.127,137
Over the past three decades several strategies have
emerged, in the guise of new -lactams, to incapacitate the class A -lactamases. The first strategy is
exemplified by clavulanate and the penam sulfones
(sulbactam and tazobactam), which are poor PBP
inactivators but excellent -lactamase inactivators.144
The key mechanistic event for both is a quite similar
fragmentation reaction of the respective serine acyl-

408 Chemical Reviews, 2005, Vol. 105, No. 2

enzyme intermediates that is competitive with hydrolytic deacylation and gives a new acyl-enzyme
intermediate improperly positioned for catalytic
deacylation.145-150 As noted previously, these -lactams are formulated with other -lactam PBP inactivators to target -lactamase-expressing pathogens.
The second strategy is exemplified by the carbapenems (such as imipenem) and the cephamycins (exemplified by cefoxitin), which resist -lactamase
hydrolysis by diminished ability at acylation and/or
(especially) deacylation events of -lactamase catalysis. Both of these -lactams have unusual -lactam
substituents that are believed to interfere with the
proper positioning of their -lactam segments when
in complex with the -lactamase. As these structural
features do not interfere with PBP affinity, these are
used therapeutically as single agents. The third
strategy is that of the third- (and fourth) generation
cephalosporins (exemplified by cefotaxime, ceftazidime, and cefepime), which are highly functionalized
cephalosporins that are poorly recognized by the class
A -lactamases. These too are used as single-agent
therapies, although this may change. With respect
to -lactam antibacterial design, the structural and
mechanistic basis for the evolution of -lactamases
that have overcome these barriers and now recognize
these -lactams as substrates is a topic of more than
idle curiosity.
In the event the acquisition of clavulanate and
penam sulfone inhibitor-resistant TEM -lactamases
is accomplished by single-point mutations at one of
several key amino acids,146-148,151-156 a compensatory
second point mutation, unrelated to resistance development but rather to restore enzyme stability,119,154 is also seen in some clinical isolates. The
relative ease of this transformation may be understood in terms of the required clavulanate (or penam
sulfone) acyl-enzyme fragmentation as an offpathway event, unrelated to normal catalysis, and
hence easily disposed.155 Moreover, it is evident from
the kinetic properties of the enzymes that incremental adjustment of the kinetic parameters suffice to
impart resistance to these inactivators. For example,
N276D mutation of TEM-1 is representative of common clavulanate-resistant IRT variants wherein the
clavulate Ki increases from 0.4 (TEM-1) to 17 M
(N276D TEM-1) and kcat increases from 0.02 to 0.16
s-1.151 The other kinetic parameters (kinact, krec) are
less altered. The critical fragmentation event in
clavulanate inactivation of the TEM -lactamase is
known to require a protonation event wherein the
proton is provided by a conserved structural water.146
Replacement of the neutral Asn276 with the charged
Asp276 results in substantial movement of the Asp
side chain so as to engage the Arg244 guanidinium
that is ordinarily involved in substrate carboxylate
recognition. The resulting electrostatic modulation
manifests in dissociation (and thus loss) of this
conserved water. Very similar kinetic changes are
seen with respect to clavulanate inactivation of the
M69L TEM-33 variant.152 This methionine, while
clearly not a conserved TEM residue, is nonetheless
located in a region of strong structural constraint (at
the beginning of the TEM H2 R-helix and in contact

Fisher et al.

with the B3 and B4 -strands and thus while


removed from the active site influences the active site
geometry).152,154 Replacement of the methionine with
leucine gives a -lactamase (the TEM-33 enzyme)
having an identical ability to hydrolyze penicillin G.
Clavulanate, however, binds more poorly (by 1.9 (
0.2 kcal mol-1 for the pre-acylation complex).152 An
explanation for this difference is not evident from the
protein structure but is suggested by computational
analyses that indicate less favorable van der Waals
and electrostatic energies for clavulanate binding to
the M69L mutant. Conversely, improved clavulanate
inactivation is seen for the (clinically not observed)
M69G TEM-1 mutant.154 Yet another mechanism by
which the TEM enzymes evade clavulanate has been
suggested to involve disruption of the active site
interactions of the Ser130 and whose oxygen engages
customarily in postfragmentation cross-linking to the
clavulanate (and penam sulfone) remnant acylenzyme.154 For both the TEM-32 (M69I/M182T) and
TEM-34 (M69V) variants the local environment of
this serine is perturbed such that the cross-linking,
leading to a long-lived acyl-enzyme, does not occur.154 An evaluation of Ser130 SHV mutations
identified only S130G as conferring clavulanate
resistance, again resulting from destabilization of the
clavulanate pre-acylation complex.149 It is evident
that the basis for the evasion is not loss of the ability
to cross-link157 but rather the simple result of an
overall diminished affinity for these variants to bind
these inactivators (even when this is accomplished
at the cost of loss of catalytic function, as measured
by kcat/Km, toward -lactam substrates).121,158 Last,
the ESBL -lactamases generally retain susceptibility toward clavulanate and penam sulfone inactivation (in contrast to the IRT enzymes), and thus, the
recent generation cephalosporins may eventually be
combined with these inactivators in clinical practice.159 This likelihood is also reflected in the continuing interest in other -lactam templates, such as the
6-methylidene penam sulfones and penems,160,161 that
have a broad-based ability to inactivate -lactamases
by acyl-enzyme fragmentations and nucleophile
additions.
Among the most common of the IRT TEM variants
are those with replacement of arginine-244 (alone
and in cooperation with other mutations).158 Arg244
is a conserved residue of the 4 strand of the parent
TEM -lactamase. Its replacement by serine gives the
TEM-30 () TEM-41)/IRT-2 enzyme, by cysteine the
TEM-31/IRT-1 enzyme, and by histidine the TEM51/IRT-15 enzyme. As this arginine is an active site
residue and as its replacement dramatically alters
the substrate specificity of the enzyme (exemplified
by a greater than 10-fold decrease in kcat/Km for
penicillin substrates), considerable effort has been
made to identify its role in catalysis. This effort is
further driven by the unique properties acquired
upon replacement of this arginine by these three
amino acids (the TEM-30, -31, and -51 -lactamases
are virtually identical).162 The most important of
these properties is resistance to the TEM inactivators
clavulanate, sulbactam, and tazobactam, accomplished by perturbation of the partitioning of the

Bacterial Resistance to -Lactam Antibiotics

acyl-enzyme between hydrolysis (where there is


little change in normal turnover) and the slower
fragmentation and cross-linking (where this inactivation event is even further suppressed).145,154,163 Two
possible roles for Arg244 in normal substrate turnover have been proposed; both are consistent with
the alteration in the steady-state kinetics (decreased
kcat/Km). The first possibility is that the arginine side
chain participates, with a highly ordered proximal
water, in -lactam substrate recognition and active
site orientation via electrostatic pairing with the
carboxylate substituent, which is found in all bicyclic
-lactam antibiotics.155,164 The likelihood of such an
interaction is well substantiated both by active site
simulation and by crystallography.118,119,154 The second proposal is that Arg244 facilitates turnover by
assisting, now via electrostatic repulsion, in product
dissociation.165 These two possibilities are not mutually exclusive. With respect to resistance, the paramount question is clearly that of the structural
consequence of arginine replacement and the correlation of that consequence to the mechanism of
clavulanate and penam sulfone TEM -lactamase
inactivation. The answer, it appears, is the pivotal
role of the proximal (to the arginine guanidinium)
water that is lost to the active site when this arginine
is replaced.146,154 This water molecule provides the
critical proton catalyst necessary to the fragmentation event of the clavulanate acyl-enzyme. Upon
arginine replacement this water molecule is lost and
these inactivators of the parent TEM become ordinary substrates for these IRT TEM variants. Other
IRT variants (such as occur at Met69) that retain this
arginine are nonetheless also able to resist these
inactivators by perturbing the second residue, Ser130,
critical to this fragmentation (and its sequelae).
With respect to resistance development, a particular objective is the understanding of the relationship
between the mutations securing the IRT class A
variants and the mutations (such as the R164S) that
secure to the class A -lactamases the ability to
hydrolyze late generation cephalosporins (the class
A ESBL).166-170 The consensussfor the momentsis
that these two phenotypes are mutually exclusive.
For example, the TEM -lactamase double mutant
(R164S, R244S) retains clavulanate resistance but is
no longer capable of ceftazidime hydrolysis.166,168 The
obvious possibility identified by this observation that
inactivator/late generation cephalosporin combination therapy might prove clinically advantageous is
now in the process of preliminary evaluation.159
The outstanding features of the carbapenem (e.g.,
imipenem) and cephamycin (e.g., cefoxitin) classes of
-lactamase inhibitors are the respective 6R-hydroxyethyl and 7R-methoxy substitution of these -lactams.
The potential for these substituents to interfere with
nucleophile approach to the -lactam carbonyl is
immediately evident, and this hypothesis is proven
especially with respect to catalytic deacylation.171 Of
the two, cefoxitin is the more straightforward. It has
a standard cephalosporin scaffold but with an unusual 7R-methoxy substituent in a position occupied
customarily by a hydrogen in the cephalosporins.
Thus, cefoxitin engages many of the standard recog-

Chemical Reviews, 2005, Vol. 105, No. 2 409

nition features (albeit abnormally) for class A -lactamase substrates. The critical mechanistic event
occurs upon serine acylation. In the cefoxitin acylenzyme intermediate the 7R-methoxy not only displaces the catalytic water172,173 but also interferes
with the Asn132 side chain. This side chain is
compelled to move from its customary location (where
with normal substrates it is engaged in a hydrogen
bond with the substrate 7-amide).173 This movement
alters the cefoxitin 7-side chain in such a way as to
induce further active site distortion, especially of the
-loop. The cumulative effect is a remarkably stable
acyl-enzyme. Subtler (but no less complex nor less
intriguing) mechanisms operate for the class A carbapenemases. As the dominant mechanism for carbapenem resistance in pathogens involves the acquisition of class B or metallo--lactamases, however,
the enzymatic basis for resistance due to expression
of a modified class A -lactamase has been less well
studied. Yet there is no doubt that the new mechanism coincides, in part, with a stunning structural
transformation of the TEM peptide, the introduction
of a second disulfide bond (Cys69-Cys238). The
presence of this cystine is intimately related to the
acquisition of the carbapenemase activity.174-176 The
role(s) that this insertion has with respect to the
mechanism has only begun to be understood. To
begin with, for the E. cloacae NMC-A enzyme a
nearly 100-fold diminution in the ability to hydrolyze
penicillins but a 100-fold improvement in the ability
to hydrolyze imipenem is seen.177 Crystallographic
inspection of a stable acyl-enzyme species shows
that the positioning of the acyl-enzyme is very
similar to that seen for TEM acyl-enzymes (notably
normal oxyanion hole hydrogen bonding) but importantly a repositioning of Asn132 so as to open the
active site for efficient catalytic delivery of the
deacylation water. That even further structural and
mechanistic accommodation may be anticipated is
suggested by the structure of the related (70%
sequence identity to NMC-A) class A SME-1 carbapenemase (and which also shows impaired penicillin
hydrolysis). In the resting enzyme the presumptive
acylation/deacylation general base Glu166 (on the
-loop) hydrogen bonds directly to Ser70 (without a
water bridge).175 This suggests that the role of the
second cystine is to enable an alternative approach
(evading the steric barrier of the 6R-hydroxyethyl
substituent) of the serine in the acylation halfreaction.
The remaining aspect of class A -lactamase evolution as it relates to the acquisition of -lactam
resistance by bacterial pathogens is the ESBL enzymes.15 Following the clinical appearance of the
third-generation oxyimino-cephalosporins (ceftriaxone, cefotaxime) some two decades ago, resistant
bacteria appeared. The basis for the resistance was
the acquisition and dissemination of class A, C, and
D -lactamases capable (often with a narrow substrate spectrum) of hydrolysis of these oxyiminocephalosporins. Three related class A groups are
pertinent: the TEM and SHV ESBL variants initially
among the Enterobacteriaceae but increasingly among
the Pseudomonas;122 the VEB and PER variants

410 Chemical Reviews, 2005, Vol. 105, No. 2

among the Pseudomonas; and the CTX-M variants.123


Of these three the CTX-M are of greatest concern.
These enzymes, formerly most commonly found in
nosocomial pathogens, are now found in community
strains (Vibrio, nontyphoid Salmonella, and Shigella). Moreover, the recent D240G CTX-M variants
show improved ability to hydrolyze ceftazidime.123
For the moment these enzymes remain capable of
inactivation by clavulanate, penam sulfones, cefoxitin, and imipenem, but there is no reason to believe
that this will not change. For this reason evaluation
of the structural basis for the acquisition of the
oxyiminocephalosporins as substrates by all of these
enzymes is a major current focus of -lactamase
enzymology.
The efforts concerning the simplest class A transformation into an ESBLsthat of the SHV (and TEM)
single G238S (SHV-2) point mutationsare instructive as to the difficulties inherent to such an understanding. This single change results in an MIC
increase (E. coli) from 2 to 8 g mL-1 for ceftazidime
and from 0.125 to 16 g mL-1 for cefotaxime (comparing to E. coli with the parent SHV-1 -lactamase,
for which neither oxyiminocephalosporin is a substrate).178 The G238S change gives a 4-fold improvement in kcat/Km for a typical cephalosporin (cephaloridine). Moreover, this mutation is exceptionally well
expressed, suggesting a basis (also with the optimal
kinetics) for the clinical selection of Ser238 as acquisition of (albeit weaker) ESBL activity occurs for
other G238 mutants (including G238A, G238N,
G238M, G238C, and G238I).120 An analogous transformation (appearance of cefotaxime and ceftazidime
as substrates) is seen for the G238S TEM mutant.179
The structural basis for this transformation (after
speculation, as summarized by Hujer et al.120) was
resolved by crystallography.136,180 Glycine 238 is
located on the b3 -strand, in close proximity to the
Asn170, Met69, and Ala237 residues of the active
site. Replacement by serine results in a significant
conformational alteration spanning the 238-242
positions but with overall preservation of R-carbon
positions elsewhere in the enzyme (especially those
of the -loop). The deformation that results from the
conformational alteration is borne fully by the b3
-strand, opening the distance between the lower
(active site) portion of this strand and the -loop by
nearly 3 .136,180 A hydrogen bond from the Ser238
hydroxyl to the main-chain carbonyl of the -loop
Asn170 is seen, which may be presumed (from its
orientation and distance and relatively poor solvent
accessibility) to be of sufficient stability as to be
unlikely to engage bound substrate. The most consistent explanation for the acquisition by this enzyme
of the oxyiminocephalosporins as substrates is that
the serine opens the active site just enough to
accommodate the larger mass of the oxyimino side
chain while preserving the orientations and roles of
all of the catalytic residues.
It is probable that a similar process (shape-selective expansion of the active site) is operative for the
other class A ESBL enzymes, although the process
(in terms of protein adjustment) differs for each. The
P. aeruginosa PER-1 enzyme, for example, is char-

Fisher et al.

acterized inter alia by an altered -loop,126 and the


P. vulgaris K1 enzyme lacks the Arg244 customarily
thought to be involved in substrate recognition and
has atypical residue replacements (notably a Ser237)
that may have specific roles in substrate recognition
or protein adjustment.181 Extensive evaluation of the
CTX-M Toho-1 structure by Shimamura et al.182
(acyl-enzymes with the E166A defective mutant)
and Ibuka et al.183 (apo-enzyme) demonstrates that
similar accommodations operate to embrace the oxyimino cephalosporins as substrates in this class A
enzyme.
These observations emphasize the remarkable
ability of bacteria to alter protein structure, under
selection pressure, to acquire new function. While the
perfect -lactamasesone that hydrolyzes all -lactam structures regardless of their substitutionshas
yet to be encountered, it is not necessary for such an
enzyme to be created in order to attain this high level
of resistance. As will become evident from the following sections, it is only necessary for the bacterium
to acquire a selection of complementary resistance
mechanisms. The CTX-M Toho-1 enzyme, for example, is quite susceptible to inactivation by cefoxitin. The bacterium does not need for this enzyme to
evolve to this function as the alternativesacquiring
a second -lactamase with this functionsis operationally more facile. Hence, the value of these exhaustive efforts to understand the relationship between structure and function of these -lactamases
is strategic: as the characteristics of the enzyme and
the design limits for its evolutionary adaptation are
better understood, the design of improved drug
structure (or the use of a new drug regimen) is made
possible.
A new and encouraging event (where there are not
many), with respect to our understanding of this
structure-function relationship, is the ability to
quickly assess the capability of -lactamases (not just
class A) to acquire new -lactams as substrates. In
the years following the classic experiment of Hall and
Knowles184 with the TEM -lactamases, which proved
that such a capability existed, new methods have
been developed to critically assess the structural
outcome on the -lactamase from the selection pressure exerted by a particular -lactam. Using DNA
shuffling (and E. coli hypermutator expression) Stemmer et al. isolated the triple-point mutation (E104K/
M182T/G238S) TEM-52 variant with ESBL cefotaxime activity.180,185 By the use of a highly error-prone
DNA polymerase, Camps et al.186 isolated three TEM
mutants (E104K/R164S, E104K/R164H/G267R, and
E104K/R164S/G267R) conferring a >50-fold resistance phenotype to aztreonam, a monobactam that
is a very poor substrate of the parent TEM-1 -lactamase. Two of these point mutations are known, but
the third (G267R) is new. Barlow and Hall developed
an in-vitro error-prone PCR method (see also Vakulenko et al.187 for the application of PCR to evaluate
the -lactamase response to clavulanate/sulbactam
with ampicillin) that is argued as predictive of
-lactamase evolution under clinical -lactam pressure. Using this method TEM -lactamase variants
resistant to the entire ensemble of third-generation

Bacterial Resistance to -Lactam Antibiotics

cephalosporins (cefotaxime, cefuroxime, ceftazidime,


and cefepime) as well as aztreonam were identified.188-190 Notwithstanding these examples, this
method also has identified examples where resistance
phentotypes have not emerged (as exemplified by
metallo--lactamases with imipenem, as discussed
later), strongly arguing against the presumption that
these enzymes have an unconstrained ability to adapt
to all variations in antibiotic structure.190 The value
of these methods to drug design and to the design of
clinical drug regimens is self-evident.189,190

3.3. Class C -Lactamases


Class C -lactamases share with the class A -lactamases a similar mechanismsactive site acylation
and hydrolytic deacylationsfor -lactam hydrolysis.
This ability was inherited from their respective
ancestral PBPs. Nonetheless, at the catalytic level
there is a significant difference for deacylation. As
first documented by Knox and colleagues,191 the two
classes use opposite faces of the acyl-enzyme species
for the approach of the hydrolytic water. In the class
C enzymes this water approaches from the -direction. This distinction refutes any possibility of a direct
evolutionary link between the two classes. Indeed,
class C -lactamases are evolutionarily closer to low
Mr class B PBPs.15,60 Furthermore, the residue responsible for activation of the hydrolytic water in the
deacylation has been proposed to be Tyr150,191-193 a
process that has been suggested to be assisted by the
amine of the acyl-enzyme species that was previously the -lactam nitrogen of the antibiotic.194,195
Therefore, the deacylation mechanism of the class C
enzymes is entirely distinct from that of class A
-lactamases.
The class C -lactamases originally were termed
cephalosporinases due to a substrate preference for
cephalosporins. They are found, with few exceptions,
in most Gram-negative bacteria and are chromosomally encoded in several organisms (including
Citrobacter freundii, Enterobacter aerogenes, and
Enterobacter cloacae).196 An increased incidence of
plasmid-encoded class C -lactamases15,196 was observed 15 years after their first discovery.197 Plasmidencoded class C enzymes have been found in E. coli,
K. pneumoniae, Salmonella spp., C. freundii, E.
aerogenes, and Proteus mirabilis.198-200 Most worrisome is that the rate of incidences of these enzymes
is highest in K. pneumoniae and E. coli, organisms
common to the hospital and community settings.196
Class C -lactamases have reached catalytic perfection for their preferred substrates, the cephalosporin-based -lactams.115 However, the efficiency
of turnover of the penicillins by the class C -lactamases also remains high. This is attributed to low
Km values as a result of deacylation as the ratelimiting step for the class C -lactamases (unlike
class A -lactamases), resulting in high kcat/Km values
(105-108 M-1 s-1).201 The structure of these enzymes
was first revealed for the class C -lactamase from
C. freundii (and E. cloacae strain P99).191,192 The C.
freundii enzyme has also been determined with
aztreonam bound, at 2.5 resolution.192 The structures reveal an overall similarity to the class A

Chemical Reviews, 2005, Vol. 105, No. 2 411

-lactamases (see Figure 4A and C). Superimposition


of the class C -lactamase (from E. cloacae) and a
representative class A -lactamase reveals a handful
of active site residues that occupy similar positions.
In the class C -lactamase these are Ser64, Lys67,
Lys315, and Tyr150 that correspond, respectively, to
the class A residues Ser70, Lys73, Lys315, and
Ser130.191 On the basis of the structures released to
date, this correspondence is common among the
classes A, C, and D -lactamases and the PBPs.
The first crystal structure for a class C -lactamases prompted the proposal that the role of general
base, assisting Ser64 acylation, is carried out by the
conserved residue Tyr150.192 It is important to note
that the side chain functions of both Lys67 and
Tyr150 are in hydrogen-bonding contact with the
serine hydroxyl. After accepting a proton from the
serine in the formation of the tetrahedral intermediate, this tyrosine then donates the proton back to the
-lactam nitrogen to drive forward the collapse of the
tetrahedral intermediate.192 The same Tyr150 then
promotes a water molecule to achieve deacylation of
the acyl-enzyme species to complete the catalytic
cycle. This hypothesis is based in part on the structural superposition of the active sites of the class C
-lactamase with that of chymotrypsin (a serine
protease). In this superimposition the -lactamase
Tyr150 occupies a similar position to the histidine
general base in chymotrypsin.192 The viability of this
proposal is diminished by recent NMR and sitedirected mutagenesis studies. 13C NMR evaluation
shows that the chemical shifts of Tyr150 are invariant up to pH 11,202 implying a neutral Tyr150 in the
substrate-free enzyme. This result challenges earlier
calculations (using Poisson-Boltzmann methodology)
that predicted an unusually low pKa value (of 8.3)
for the Tyr150 phenol.203 The second line of evidence
arguing against Tyr150 as the general base is the
site-directed mutagenesis study by Dubus et al.204
The steady-state kinetics were not substantially
altered by replacement of the Tyr150 with a phenylalanine. This result is inconsistent with a role for
Tyr150 as a direct participant in the turnover events.
Tyr150 is suggested to contribute to the acylation
process indirectly, perhaps as a proton shuttle to the
-lactam ring nitrogen (with water serving this role
in the Phe mutant). It is of interest to note that the
effects on the kinetic parameters were akin to those
obtained with the D,D-transpeptidase/D,D-carboxypeptidase from Streptomyces species R61 at Tyr159, a
residue equivalent to Tyr150.
In light of the recent data by Kato-Toma et al.
indicating a normal pKa value for Tyr150,202 the
proposed mechanism for the second half of the
reaction of the class C -lactamases has to be
reevaluated as well. Oefner et al. indicated that the
deprotonated Tyr150 serves as the general base for
the second step of the reaction in activation of a water
molecule for the deacylation step, a proposal that has
been widely accepted.192 A protonated Tyr150 cannot
perform this function.
These observations invoke the involvement of
Lys67 (as a free-base) in activation of the active site
serine for the acylation event in class C enzymes or

412 Chemical Reviews, 2005, Vol. 105, No. 2

that it may abstract a proton from Tyr150, which in


turn activates the serine. Furthermore, for the deacylation step, in the absence of a residue equivalent to
Glu166 in class A -lactamases, crystal structures
from E. cloacae P99191 and C. freundii192 argue for
the approach of the hydrolytic water from the -face
of the -lactam antibiotic. Bulychev et al. propose
that the nitrogen of the thiazoline (from -lactam
opening) is ideally positioned to promote hydrolytic
water addition to the acyl-enzyme to accomplish
deacylation.193 This would be an example of a substrate-assisted catalysis. Two non--lactam synthetic
molecules were used to test this concept. These
compounds were chemically predisposed to acylate
the active site serine in the E. cloacae class C
-lactamase, resulting in acyl-enzyme species. One
compound lacked the amine of the acyl-enzyme
species, whereas the other possessed it. The one with
the amine experienced turnover, whereas the one
without it served merely as an irreversible inhibitor
of the enzyme.194 A recent X-ray crystallographic
structure of an acyl-enzyme species of AmpC -lactamase and moxalactam lends further evidence to
this proposal. The authors note that a water molecule
is organized to take advantage of the ring nitrogen
in the hydrolytic step.195
Third-generation cephasloporins have been effectively used as a strategy against class C -lactamases for over a decade. However, class C -lactamases capable of hydrolyzing most third-generation
cephalosporins were first isolated in the 1980s205 and
yet more recently from a virulent strain of E. cloacae.206 These enzymes, termed extended-spectrum
-lactamases (ESBLs),124 mediate resistance to (extended-spectrum) third-generation cephalosporins
(exemplified by ceftazidime, cefotaxime, cefepime,
and ceftriaxone) and monobactams (aztreonam) but
do not affect cephamycins (cefoxitin and cefotetan)
or carbapenems (Meropenem or imipenem). The
structural basis for resistance mediated by these
ESBLs was revealed by the structure of the E. cloacae
GC1 enzyme, solved by Crichlow et al.207 These
authors suggest that conformational flexibility of the
expanded -loop facilitates hydrolysis of thirdgeneration cephalosporins by enabling greater mobility of the acyl moiety. Further evidence implicating
the -loop modification with resistance to these
third-generation cephalosporins came from the structures of the AmpC -lactamase in complex with
various third-generation cephalosporins.208 A more
recent structure of a phosphonate transition-state
mimetic bound to the E. cloacae GC1 and the wildtype C. freundii GN346 class C enzymes afforded
further insight to the basis for resistance. It was
found that the designer molecule adopted different
conformations in both enzymes, with the mutated
-loop of the GC1 enzyme able to accommodate the
cefotaxime side chain in a different conformation,
enabling it to allow the approach of the hydrolytic
water to the acyl-enzyme species.209

3.4. Class D -Lactamases


The class D -lactamases are increasingly encountered among the defensive -lactamase ensemble of

Fisher et al.

certain Gram-negative pathogens.15,109,124,210 These


-lactamases were first termed as oxacillinases (and
for this reason are still described as OXA -lactamase
variants) for their ability to hydrolyze the 5-methyl3-phenylisoxazole-4-carboxy side chain penicillin class,
exemplified by oxacillin and cloxacillin. Over 50 class
D OXA variants are now known,211 including variants
that have dispensed with their oxacillinase activity
in the process of acquiring ESBL activity against
carbapenem and third-generation cephalosporins.212
In the course of these transformations the class D
enzymes have expanded from their historical P.
aeruginosa niche 213,214 into other Gram-negative
pathogens including E. coli,215,216 P. mirabilis,217
Salmonella sp.,218 K. pneumoniae,219,220 and especially
Acinetobacter baumannii.221-228 While at present the
clinical impact of the OXA -lactamases is associated
with infections by P. aeruginosa and A. baumannii,
the widening Gram-negative distribution provides
powerful support for the concern that the clinical
value of carbapenems (and third-generation cephalosporins) may quickly diminish.211,220,222,225,229
These enzymes only recently have become the
subject of detailed structural and mechanistic studies. This is due in some measure to their very recent
appearance in clinically relevant pathogens (as distinct from the other three -lactamase classes) but
in greater measure to their heterogeneous properties
(not withstanding their interrelatedness) and the
historical difficulty of their in vitro assay. The
literature is replete with descriptions of poorly reproducible biphasic (burst-type) progress curves for
some variants, corresponding (in the apparent steadystate) to unusually low kcat/Km values (as exemplified
by the data of Danel et al.,230 Franceschini et al.,231
Pernot et al.,232 and Heritier et al.226). Among the
enzyme properties explored as possible explanations
for these difficulties were a monomer-dimer equilibrium with possible additional divalent metal dependency.230,231,233,234 No credible explanation was
found. For example, some OXA variants are monomeric,97 and the dimer Kd in any case (typically
micromolar) is relevant only to the in-vivo and not
in-vitro kinetics, while other variants show little
capacity for divalent metal binding.224 As a consequence of this dilemma, the possibility that other
mechanisms (in addition to the OXA -lactamase)
contribute to the -lactam resistance phenotype
has been discussed.221 Also, while the importance of
porin deletion211,216,219 and peptidoglycan remodeling
(in addition to the assembly of mixed class -lactamase ensembles or single -lactamase hyperexpression) to Gram-negative resistance cannot be underestimated, the identity of the likely critical and
confounding variable contributing to the in vitro
assay variabilitysCO2swas identified by Golemi et
al.95,96
The structural basis by which CO2 activates the
class D enzymes for -lactam hydrolysis was elucidated concurrent with several independent crystallographic studies of the class D -lactamases. The
first studies on a class D enzymes (the P. aeruginosa
OXA-10 -lactamase) showed that the class D domain
folding was similar to the other two serine -lacta-

Bacterial Resistance to -Lactam Antibiotics

mase classes.235,236 However, subsequent crystallographic analysis revealed that the active site lysine
was N-carboxylated as a result of addition of the
lysine side chain amine (as the free-base amine) to
CO2.96,237 The resulting carbamic acid ionizes to give
a carbamate functional group in hydrogen-bonding
contact with the active site serine. As the formation
of this carbamate is reversible, the earlier reports of
the absence of lysine carboxylation are explained. In
light of the high physiological concentration of CO2
(low millimolar) this lysine is expected to be fully
carboxylated in vivo.95 Kinetic analysis of a mutant
enzyme where this lysine is replaced shows the total
loss of catalytic activity, indicating a direct involvement in catalysis.95 The role for this unusual lysinederived carbamate is general base activation for both
acylation (activation of the serine) and deacylation
(activation of water) steps of catalysis.95,237 Moreover,
the assignment to this CO2-derived lysine carbamate
of this role as general base catalyst is consistent with
the absence of alternative possibilities. Not only does
the class D -loop not contain a counterpart for the
class A Glu166, but the tyrosine of the conserved
parental class D Y144-G145-N146 motif on this loop
is replaced by phenylalanine in ESBL (both carbapenem and cephalosporin) class D variants, obviating
direct participation of this tyrosine in catalysis (as
is more fully discussed elsewhere).97,222,226,238 The
OXA-1 crystal also shows the lysine carbamate97
whereas the OXA-13 enzyme232 does not (almost
certainly an artifact of crystallization). The relationship of the CO2 to the complicated kinetics extends
beyond the relative portion of the lysines that are
activated for catalysis. During catalysis the lysine
carbamate is prone to spontaneous decarboxylation
in the middle of the turnover process, thus arresting
catalysis at the acyl-enzyme stage. This must,
however, be regarded as an artifact of in vitro assay
since supplementation of the medium with bicarbonate (as a CO2 source) restores the kinetic profile.95 It
has been argued that the more complicated biphasic
turnover profile for these enzymes with some substrates is due to a branching mechanism. As the
enzyme experiences decarboxylation midcatalysis, it
becomes inactivated (the branching species), pending
the availability of a CO2 molecule to restore the lysine
to its active carboxylated form. The enzyme is then
able to complete the second step of catalysis.95,236
With a few substrates, however, it was shown that
supplementation of the medium with bicarbonate
does not simplify the turnover process. For these few
cases a branching mechanism (as might occur by a
conformational change at the acyl-enzyme state) has
been invoked.
[Dr. Roger Labia, a pioneer of studies of -lactamases, kindly communicated that his early investigations of the class D -lactamases in 1970s were
frustrating because of the complicated and erratic
kinetic behavior of the enzymes. He opted to abandon
studies of the class D enzymes. He now attributes
the erratic behavior of the enzymes to the seasonal
variations in the quality of the water, which has high
carbonation in the summers and low carbonation in
the winters.]

Chemical Reviews, 2005, Vol. 105, No. 2 413

While there are numerous reports evaluating the


-lactam substrate profile for the class D OXA
enzymes, essentially all of these predate the discovery of requirement for in-vitro CO2 (but most unlikely
in vivo) activation. These earlier data may not be
reliable. Nonetheless, all these data suggest that the
substrate profile for these enzymes, individually and
as a class, is not broad.211 The value of the class D
enzymes to bacteria is the ability of this enzyme to
adapt, under selection pressure, to the specific -lactam. For example, while most of the ESBL OXA
variants hydrolyze ceftazidime better than cefepime,
the reverse is true for the OXA-30 variant (derived
from OXA-1).215 Full discussions of the orientation
(and contacts) within the class D active site for
(inhibitor) acyl-enzymes232,237 and possible orientations for substrates97 are presented elsewhere.
A final issue is the origin of the class D enzymes
in comparison to the two other serine -lactamase
classes. All three classes are now encountered as both
chromosomal- and plasmid-borne genes. However,
the evolutionary history of each class is distinct.15 In
contrast to the class C AmpC family, where the
mobilization to plasmids is a modern (antibiotic era)
event, plasmid mobilization of the class A TEM and
class D OXA are ancient events.239 The primary
distinction between the class A and C -lactamases
is the much more rapid and extensive diversifications
at this point in timesof the former within the Gramnegative bacteria. Also of particular interest is the
complete absence of this diversification within the
Gram-positive bacteria given the estimate by Hall
and Barlow that the horizontal transfer of the class
C gene from the Gram-negative to the Bacillus Grampositive bacteria is an ancient event of some 320-575
Myr (but is after the divergence of B. subtilis from
S. aureus).15 A homology between the hydrophilic
carboxy domain of the BlaR/MecR -lactam-signaling
receptors and the class D -lactamases was noted
previously.84,240 The recent discovery within B. subtilis of a gene encoding an enzyme with weak class
D -lactamase activity, yet also resembling a penicillin-binding protein,241 suggests that study of the class
D gene within those Gram-positive bacteria that
possess it may further refine our understanding of
the structural and functional relationships between
the PBP and -lactamase enzymes.

3.5. Class B Metallo--lactamases


The secondsand in many respects no less forebodingsvision of the -lactamase future is that of
the disseminated metallo--lactamases (MBLs).242
First observed in 1967 by Kawabata and Abraham
as chromosomal enzymes of the innocuous Grampositive B. cereus, these enzymes occupy a position
of concern (in terms of breadth of distribution and
breadth of -lactam catalytic activity) with respect
to the inexorable expansion of -lactam resistance.109,229,243,244 These metal-dependent (almost always divalent zinc) -lactamases have a broad -lactam substrate tolerance that encompasses many of
the newer generation cephalosporins, carbapenems,
and other -lactamase inhibitory (clavulanate and
penam sulfones) -lactams important to the treat-

414 Chemical Reviews, 2005, Vol. 105, No. 2

ment of Gram-negative infection.210,229 As a different


chemical mechanism (compared to the serine -lactamases) is used by the metallo--lactamases for
-lactam hydrolysissnotably, a mechanism without
a covalent enzyme intermediatesan entirely different
strategy for their inhibition (or inactivation) will be
required should (or is it when) these enzymes expand
beyond their present niche (that of minor, and
opportunistic, Gram-negative pathogens). While several inhibition strategies have been identified, none
has yielded anything resembling that of a clinically
effective inhibitor. Also, although there exist as yet
chemical reasons (in terms of kcat/Km these are not
optimized enzymes) and biochemical reasons (zinc as
a limiting nutrient) that ultimately may limit the role
the metallo--lactamases will have as a resistance
mechanism (vide infra), the proposal discussed by
Fast et al.,245 Fabiane et al.,246 and Wommer et al.247
that these are young enzymes, only now in the
process of maturation under evolutionary pressure,
is both credible and worrisome.
The metallo--lactamases (also termed Ambler
class B and Bush-Jacoby-Medeiros Group 3 -lactamases)105,243 are small enzymes sharing a common
four-layer RR motif with a central -sandwich and
two R-helices on either side.246,248 This motif, arising
possibly by a gene duplication event,60,245 is found also
in other proteins (glyoxalase and certain flavoenzymes). The motif has an intrinsic metal-binding site
located at an edge of the -sandwich.249,250 For the
metallo--lactamases this site is occupied by a divalent zinc ion having a tetrahedral array of three
histidines and water. The importance of the zinc ion
to -lactam substrate binding is unquestionable.251,252
The role of the zinc in the hydrolytic mechanism,
beyond that of Lewis acid catalysis, is less certain.
Nonetheless, there is a consensus that the water
ligand of the zinc ion is the -lactam ring-opening
nucleophile (via a mechanism likely with parallel to
that of the zinc metalloproteases), but there are no
direct data establishing this role.
Beyond these structural commonalities the metallo--lactamases possess a surprising breadth of
primary structure that most notably includes the
creation (in some enzymes) of a second zinc binding
site.243 In these binuclear enzymes the two zincs are
proximal (approximately a 3.5 separation) with
both participating in the water coordination. The
ligand environment of the second zinc ion is very
different from that for the first both in array (trigonal
bipyramid) and amino acid ligands (variable among
the binuclear class). The purpose of the second zinc
ion is regarded as catalytic augmentationsthat is,
accomplishing an incremental increase in the -lactam substrate kcat/Kmsand not a role of catalytic
necessity.245,247,253-255 Nonetheless, the specific environment of the zinc ensemble determines the individual enzyme catalytic behavior toward substrates
and the detail of the rate-limiting (highest energy)
catalytic step.245,255-259 The relative affinity of the
enzyme for the two zinc ions is unequal, and in-vitro
enzyme kinetic analysis requires added Zn2+.247,255,260
Wommer et al.247 discussed in detail the role of the
-lactam substrate in the recruitment of zinc to, and

Fisher et al.

hence activation of, the -lactamase. This process is


argued as one of physiological necessity wherein the
-lactamase exists as an apoenzyme and the available zinc is reserved to other enzymes until the
-lactam antibiotic is encountered. The likelihood of
a fully zinc complemented metallo--lactamase (binuclear site fully occupied) in vivo is regarded as
small. It is therefore understandable why the bacterium possessing a metallo--lactamases is often found
with a serine -lactamase as well.261,262 The demonstrable fact that these enzymes are disseminating is
proof of evolutionary pressure for resistance development within human clinical practice.
While the details245,255,263 of the hydrolytic mechanism used by these enzymes is beyond the scope of
this review, the mechanistic fundamental is assuredly not. This fundamental is delivery of the zinccoordinated water, possibly as a hydroxide ion (pKa
) 4.9 to 5.6),245,260,264,265 to the -lactam carbonyl. (The
metalloprotease carboxypeptidase A has a similar
catalytic pKa that is also assigned to a zinc hydroxide.
However, the recent 67Zn NMR study by Lipton et
al. forcefully argues against this assumption.266 Accordingly, a wholesale mechanistic revision for metallo--lactamase catalysis may be necessary.) The
presumed involvement of a zinc hydroxide intermediate for the metallo--lactamases has stimulated
exquisite studies on the in-vitro mechanism of metalcatalyzed -lactam solvolysis.34,267-269 These studies
suggest the rate-limiting step to be either metalcoordinated hydroxide addition or (following the
hydroxide addition) the collapse of the tetrahedral
species. The mechanism is dependent on -lactam
structure in such fashion as to strongly implicate
zinc coordination of the -lactam in the transition
state.34,252,253,256,270-272 A similar mechanismsthat is,
hydrolysis without a covalent enzyme intermediates
is posited for the enzymatic reaction. Solvent kinetic
isotope effects for the enzymatic reaction implicate
additional transition-state stabilization by proton
flight.255,260 The identity of the enzymatic proton
donor/acceptor, however, remains a particular focus
of mechanistic study. A notable commonality of the
structure-kinetics analysis of the nonenzymatic solvolysis and the comparative enzymatic structurekinetics is the particular susceptibility of the carbapenems to hydrolysis.34,229,269,270
This realitysthe susceptibility toward metallo-lactamase hydrolysis of nearly all (vide infra) of the
serine -lactamase inhibitorssis a growing concern.
There are three reasons for this. The first is the
burgeoning presence of metallo--lactamases (first
the IMP and now the VIM and SMP class B1
variants) as mobile plasmid-encoded (and often also
as integron, or cassette) genes by the enterobacteria.273,274 Second, these enzyme variants have a
remarkable ability to alter their substrate capability
(as can be accomplished by simple-point mutation,
even remote to the active site), raising the possibility
of rapid adjustment to encompass new substrates
(and to thwart new inhibitors). Last, the metallo-lactamase plasmids often encode additional (multisubstrate) resistance mechanisms.

Bacterial Resistance to -Lactam Antibiotics

The class B1 metallo--lactamases are monomeric,


binuclear zinc enzymes constituting the largest metallo--lactamase subclass. Within this subclass the
dominant subtypes are the IMP and VIM enzymes.
While these subtype enzymes are defined by sequence, substantial diversity is found within each.
For example, the recently observed IMP-12 variant
has 89% sequence identity to its closest (IMP-8)
relative but includes 10 amino acid changes at
positions that are otherwise invariant among the 11
other subtypes.275 The IMP enzymes are originally
Asian (and the VIM enzymes originally European),
but it is quite evident that geographical characterization of both of these enzymes is irrelevant. First
observed in 1988, the IMP enzymes have a broad
substrate (primarily cephalosporins and carbapenems, less so penicillins) acceptance, although at the
level of specific -lactam structure the kcat/Km variation can be substantial. Replacement of glycine-196
(a noncatalytic residue adjacent to His197, a zinc
ligand) in the IMP-3 and IMP-6 variants with serine
(as is found in IMP-1) results in a significant kcat/Km
improvement toward certain cephalosporins and
toward imipenem (IMP-1 kcat/Km is 10-50-fold greater
compared to IMP-3).272,276 Nonetheless, a comprehensive in vitro substrate evaluation of IMP-6 and
microbiological evaluation of E. coli possessing the
TEM-1/IMP-6 plasmid indicate the IMP-6 to be the
better enzyme in terms of extended carbapenemase
activity.277 Whereas Meropenem and imipenem are
essentially equivalent (as measured by kcat/Km) IMP-1
substrates, the IMP-6 glycine replacement results in
a 6-fold improvement in the Meropenem kcat/Km (and
a 2-fold loss for imipenem). This improved kcat/Km
contributes to the significant difference in MIC (E.
coli with the IMP-1/TEM-1 plasmid, 64 g mL-1 for
Meropenem and 2-8 g mL-1 for imipenem; without,
0.25 g mL-1).277 Efforts to correlate IMP sequence
with altered substrate acceptance have been made
by Oelschlaeger et al. (comparing G196 IMP-6 to
S196 IMP-1)272 and Moali et al. (evaluating the role
of the distal 60-66 loop).271 In the former study a
favorable serine-196-lysine-33 interaction improves
packing and rigidifies histidine-197; this rigidity
propagates throughout the active site. A calculated
enzyme-substrate stability index was found to correlate well to the experimental kcat/Km. In the latter
study the loop was confirmed as nonessential for
catalysis but contributed (especially tryptophan-64)
to hydrophobic substrate binding. A mutagenesis
study by Materon and Palzkill of IMP-1 active site
proximal amino acids identified 52% of the IMP-1
amino acids as intolerant to substitution (by comparison, the TEM-1 value is 33%).278 Materon and
Palzkill278 suggest that the IMP metallo--lactamases
may have a relatively more limited ability to adjust
to extended spectrum -lactam structure compared
to the TEM enzymes (which have, of course, already
done so). A relatedsand no less pertinentsquestion
is whether the metallo--lactamases have the capability to develop their catalytic apparatus to function
at the substrate diffusion limit (that is, at full
catalytic competence), as is already the case for the
class A and C serine -lactamases.115 The extant IMP

Chemical Reviews, 2005, Vol. 105, No. 2 415

kinetic data indicate that while certain cephalosporins have kcat/Km values that approach the diffusion
limit of (107-108 M-1 s-1), most substrates have lower
values (typical carbapenem kcat/Km values are approximately 106 M-1 s-1). Halls full mutagenesis invitro evaluation of the IMP-1 structure, which failed
to identify a mutant enzyme more capable of imipenem hydrolysis, is consistent with one (or both) of
these possibilities.279 As the diversity of known carbapenem structure is not nearly that of the penicillins
and cephalosporins, cautious optimism may be entertained that newer generation -lactams poorly
capable of metallo--lactamase hydrolysis may yet be
made.
Two additional aspects may temper this conclusion.
It is clearly not possible to determine, by evaluation
of enzyme sequence or enzyme kinetics, a direction
for metallo--lactamase variant evolution (which
variant arose from which variant). Hence, the apparent evolutionary limitation of IMP-1 with imipenem has no predictive value with respect to other
metallo--lactamase variants. As bluntly stated by
Hall, in order to understand the risks posed by
metallo--lactamases, it will be necessary to conduct
similar studies on representative members of each
of the three metallo--lactamase subfamilies and to
include all clinically relevant carbapenems.279 Second, it is evident that incremental changes in -lactam fitnesssin terms of PBP inactivation and competence as a substrate for -lactamase hydrolysiss
suffice to determine whether a bacterium is susceptible
or resistant. An effect need not be dramatic to be
important.
The VIM B1 subclass is newer (first observed in
1997) and biochemically less well studied.280 A VIM
sequence homology with IMP is recognizable (approximately 30-40%), and the overall pattern of
-lactam substrate recognition by the two enzymes
is similar.229 Seven VIM variants are extant.281 In less
than 7 years the VIM metallo--lactamases have
transitioned from chromosomal expression by nonfermenting Gram-negative bacteria (where it contributes to high-level antibiotic resistance in P.
aeruginosa pathogenic strains)282,283 to transferable
plasmid expression in Gram-negative enterobacteria.262,273 The presumptive circumstances leading to
this change (evolutionary pressure for plasmid dissemination is not necessarily carbapenem but rather
multidrug driven) and probable consequence of this
change (likelihood of carbapenem clinical failure)
underscore the caution expressed in the previous
paragraph.
A consistent observation from the in vitro evaluation of the metallo--lactamase substrate spectrum
is the inability of these enzymes to hydrolyze the
monocyclic N-sulfonyl -lactam antibiotic aztreonam.
Consequently, bacteria that have these metallo-lactamases can retain aztreonam susceptibility (although moderate to substantial increases in the
aztreonam MIC values, due to other plasmid-conferred resistance mechanisms or to the presence of
aztreonam-capable serine -lactamases, is common).
As the intrinsic reactivity toward solvolysis of the
aztreonam -lactam is identical to that of the penicil-

416 Chemical Reviews, 2005, Vol. 105, No. 2

lins,34,116 the inability of the metallo--lactamases to


hydrolyze aztreonam must correspond to a failure
either to bind to the enzyme or to bind but in a
nonproductive orientation. Quite surprisingly, there
do not appear to be enzymatic kinetic data on this
matter. Diaz et al. predict, on the basis of computational study, the latter answer as correct.269 Accordingly, aztreonam may represent an unusual example
of what has emerged as a general strategy for
metallo--lactamase inhibition. Following the proven
basis for zinc protease inhibitor design (that of inhibitors providing sulfur coordination to the thiophilic
zinc), thiol-substituted carboxylic acids represent a
general metallo--lactamase inhibitor motif.248,259,284-286
The carboxylate of these inhibitors occupies the
identical site used in -lactam carboxylate recognition
and the thiol (as the thiolate) displaces the nucleophilic hydroxide from the zinc pair. When the juxtaposition is optimal (typically corresponding to Ki
values of 0.1-1 M) a reorganization of the active
site is seen that is believed to be similar to what
occurs in substrate binding.248,259,287,288 Given the
exemplary quality of these inhibitor-enzyme crystallographic structures and the increasing sophistication
with which proven -lactam synthetic methodology
is being applied to thiocarboxylate design, yet more
potent inhibitors (approaching what will likely be
required for clinical efficacy) are anticipated. Whether
these will also possess the appropriate pharmacodynamics to effectively synergize with a -lactam
antibiotic remains to be seen.

4. Other Resistance Mechanisms


4.1. Porin Deletion
The temporal response of bacteria to antibiotics is
both immediate (abrupt gene repression and gene
activation) and evolutionary (empirical gene mutation and gene acquisition). Not surprisingly the
increased sophistication with which these changes
may be assessed and the dramatic breadth of change
particularly in the immediate response289,290 have led
to justifiable optimism that understanding these
responses will identify new targets for antibiotic
design.291,292 At a simpler level the comparison of
bacterial protein expression before (susceptible) and
after (resistant) antibiotic exposure has been the
mainstay to the understanding of antibiotic resistance development, and it is these studies (as have
been just described) that support the generalization
that PBP alteration is a principle mechanism of
Gram-positive resistance and -lactamase expression
is a principle mechanism of Gram-negative resistance. Nonetheless, these same studies indicate that
other resistance mechanisms exist. In this section an
overview of two of these other mechanismssdecreased
antibiotic permeability and increased antibiotic effluxs
is given.
It is axiomatic that a successful antibiotic has
potency (the ability to incapacitate an essential
target) and access (to its target). Indeed, PBP alteration is a strategy to render impotence and -lactamase expression compromises access. A second method
for controlling access is by changes within the outer

Fisher et al.

membrane (in Gram-negative bacteria) and cell wall


(in Gram-positive bacteria). While the evaluation of
these changes is among the most difficult tasks to
accomplish at the microbiological level, the recent
observations concerning several resistant K. pneumoniae speciessthose of an important Gram-negative
pathogensare revealing. The report by Bradford et
al. is regrettably now typical.293 An examination of
12 highly -lactam-resistant K. pneumoniae strains
and 6 E. coli strains from a single hospital showed
that 17 possessed multiple -lactamases. The three
most resistant K. pneumoniae strains (as defined by
resistance to imipenem, the prototypical carbapenem)
achieved their resistance by the combined expression
of an AmpC extended-spectrum -lactamase and
deletion of a major 42 kDa outer membrane protein
(omp). A second report294 likewise describes this same
combination of the AmpC -lactamase and the omp
protein deletion giving an imipenem-resistant K.
pneumoniae. The surmise that this omp proteinsa
porin, or nonspecific solute poresis an important
point of ingress for the -lactam to the periplasmic
space is supported by the subsequent studies of
Nelson et al.295 and Domenech-Sanchez et al.296
Resistance derives from the synergistic combination
of reduced permeability and the -lactamase; the
degree of resistance of these two mechanisms in
concert exceeds that of each mechanism alone. Similar observations have been made recently (citing
representative examples) with respect to -lactamresistant E. coli,216,297 Salmonella enterica,298 Heliobacter pylori,299 Acinetobacter baumannii,300 E. aerogenes,301,302 K. pneumoniae,303 and P. aeruginosa.304,305
A simple conclusion as to the importance of porin
deletion (or modification) remains, however, elusive.
It most certainly contributes for certain bacteria
when selected for by certain antibiotics. The reasons
why a broader generalization is not possible are
straightforward. The variation in intrinsic antibiotic
permeability among the bacteria is substantial. For
example, Lakaye et al.306 estimate that E. coli is
approximately 20-1000-fold more permeable for a
given -lactam than E. cloacae. In addition, the
variation in relative permeability as a function of
-lactam structure is no less variable: as assessed
by Matsumura et al.,307 the relative E. coli permeability of imipenem (most permeable) is approximately 60-fold greater than ceftazidime (the least
permeable of six -lactams evaluated). An appreciation for the basis of this variability is provided by
Nikaidos superlative account45 of the utter complexity of the Gram-negative outer membrane dynamics: an extraordinarily asymmetric bilayer dominated
on the outside by the lipopolysacharide surface,
which itself exerts significant permeability selection
particularly against hydrophobic solutes,308,309 and
punctuated on both surfaces by an array of nonspecific protein pores (porins) and transporters. Relative
solute permeability is influenced by a nearly limitless
number of variables. Porin deletion in a -lactamresistant E. coli is accompanied by (uncharacterized)
changes in the outer membrane (and perhaps peptidoglycan) that could also contribute to resistance.297

Bacterial Resistance to -Lactam Antibiotics

The presumption that compensatory responses


such as porin deletion exact a fitness cost is almost
certainly correct.5,91,310 In this regard bacteria are
little different from other organisms: to the extent
that a choice is possible between death and discomfort, the latter is chosen. An example of accommodation between survival and vastly decreased solute
permeability is provided by the mycobacteria. The
mycobacteria are notable (and also opportunistic)
human pathogens that are suggested evolutionarily
to bridge the Gram-negative and -positive bacteria.
Their high intrinsic resistance to chemotherapys
including complete -lactam resistancesis believed
to result from a combination of restricted porin
ingress (either by limited abundance or by pore
character) and an impenetrable exoskeleton (consisting of a thick peptidoglycan and an outer membrane
having on its outer leaflet an additional barrier of
long mycolic fatty acids attached an arabinogalactan
chain).45,311 It may therefore be surmised that exoskeleton adjustment, so as to limit -lactam exposure,
is also a strategy exploited by the Gram-positive
bacteria (the more so because they are regarded as
more solute permeable than either the Gram-negative or mycobacteria). This surmise is correct. However, the astonishing complexity of these adjustments
is just now being appreciated. While it has been
known for some time that the cell wall composition
of both Gram-negative and -positive bacteria changes
in response to -lactam exposure,26,312,313 the presumption with respect to the Gram positives is that
these changes reflected the direct (for want of a better
term) resistance response: the peptidoglycan is
altered by compensatory overexpression of a PBP, or
as a result of differential inactivation by the -lactam
of a PBP from among the ensemble, or the differential
recognition of the selected low-affinity PBP for the
biosynthetic cell wall precursors results in the modified cell wall. That a more complex stratagem was
in play was discovered by Filipe and Tomasz314 from
the observation that inactivation of the murMN
operon, encoding the murM and murN cell wall crossbridge biosynthesizing enzymes, abolished -lactam
resistance in low-affinity PBP containing (and thus
previously) highly -lactam-resistant S. pneumoniae.315 The indisputable correlation of murM enzymatic activity (murN deletion has little effect) with
robust penicillin resistance is discussed by Fiser et
al.316 and Rohrer and Berger-Bachi.317 The facile
conclusionsthat the murM cross-link-enabling reaction, the acylation by serine of the lysine -amino of
Lipid II, is a necessary event in the construction of a
-lactam-impermeable peptidoglycansmay indeed
prove correct. Fiser et al. suggest a more intriguing
possibility based on homology modeling of the murM
sequence with FemA and myristoyl transferase: a
winged coiled-coil helical DNA-binding domain structure similar to the bacterial transcription factors
known to control multidrug exporter expression.
Their conjecture that murM additionally serves to
regulate bacterial gene expression following -lactam
exposure, such as by exporter expression, is consistent with the appearance of high-level -lactam
resistance. We know this conjecture to be credible

Chemical Reviews, 2005, Vol. 105, No. 2 417

since the operation of a drug transporter is already


known to confer high-level -lactam resistance to
another insidious bacterial pathogen, P. aeruginosa.

4.2. Transporter Expression


The P. aruginosa species is a particular contributor
to highly drug-resistant biofilm infections in cystic
fibrosis patients for which carbapenem therapy is
often the only recourse. The breadth of its resistance
is believed to result from the combination of overall
membrane impermeability (especially porin deletion
and thickened peptidoglycan) and the action of active
transporter-catalyzed drug efflux (typically the RND
MexAB-OprM transporter).304,305,318,319 The bacterial
transporters (of which five families are now recognized, abbreviated as MFS, SMR, MATE, RND, and
ABC) have become a central issue to the understanding of overall bacterial vitality. Simply put, the
volume of regulated (such as by transporters) and
unregulated (such as by porins) molecular traffic in
and out of the bacterium is astonishing; it is now
believed that up to 20% of the E. coli genome encodes
transporters of one variety or the other.320 The
importance of this phenomenon, especially in relation
to drug resistance and virulence factor release,321,322
is attested to by the volume of exemplary reviews
that describe the rapidly changing status of this
field.320,323-328 What is particularly disconcerting is
the appearance of highly resistant bacteria wherein
the operation of these transporters, as part of an
ensemble of resistance mechanisms including porin
loss and -lactamase acquisition, suffice to compromise the carbapenems as an effective therapy. Moreover, bacteria that operate these transporters coincide to a multidrug-resistant phenotype. Also, while
imipenem is understood not to be an efficient substrate of the P. aeruginosa transporter, the increasing
appearance of imipenem-resistant P. aeruginosa
indicates that refinement of the panoply of P. aeruginosa resistance mechanisms to embrace imipenem
is in progress. Whether antibacterial design of new
-lactam structures that evade these structures is
possible remains to be seen. There is decreasing room
for optimism: imipenem is an exceptionally small
and permeable antibacterial. The strategy in medicinal chemistry design is invariably that of increased
functional-group complexity and concomitant increased molecular mass, entirely in the wrong direction for transporter evasion. Perhaps more promising
is the concept of synergism via co-administration of
transporter inhibitors, which has emerged as an
active area of drug design.329-332 Even more intense
interest in these approaches is likely as the role of
these transporters in Gram-negative333 and Grampositive316 drug resistance is clarified.

5. Envoi
That the thoughtless use of antibiotics is reckless
is an opinion that will fail to provoke dispute from
any reader of these words. Indeed, the issue is no
longer whether a clinical problem exists (the statistical data would be deemed indisputable even by
Samuel Clemens in his retelling of Disraelis quote

418 Chemical Reviews, 2005, Vol. 105, No. 2

about statistics)6,334-338 nor largely what could (or


should) be done to forestall the inevitable crisis but
rather how to invest today in the scientific and
medical strategies that will provide preparedness
tomorrow. Jeffersons insightful connection between
vigilance and liberty is no less appropriate to parlous
geopolitical circumstances as it is the inexorable
progression of resistant human pathogens. Two questions address the relationship of scientific vigilance
to the future role of the -lactam antibacterials in
the treatment of infections. Do the -lactams remain
a viable template for drug discovery? Do their targetss
the enzymes of cell wall biosynthesissremain a
viable target for antibacterial design?
The former question is unquestionably the more
contentious, especially given the precipitous decline
in pharma investment in antibacterial discovery.2,339-343 The choice of template, always paramount
to ultimate success in drug discovery, is yet more so
when resources are limited. Notwithstanding the
historical dominance of the -lactams, which continues to this very day, the trend toward greater mass
and functional-group complexity in successive -lactam generations strongly suggests that design limits
are being approached. In the past these conceptual
limits have been breached by revelation from Natures
a vastly more imaginative engineer of chemical
structure than mansbut here as well pharma investment is in decline.344,345 It is evident that the era of
medicinal chemistry manipulation of the -lactam,
guided by the paradigm of iterative optimization of
MIC values, is coming to an end. This is not to say,
however, that future medicinal chemistry efforts
toward -lactam optimization are exercises in futility.
Rather, future antibacterial discovery (not just -lactams) will follow the much more complicated process
wherein chemical structure is evaluated in terms of
the interrelationships among bacterial genomics and
proteomics relating to several antibacterial targets.290-292,346,347 As has been noted on several occasions within this review, bacterial resistance as a
phenotype is the result of multiple compensatory
adjustments, many of which are incremental. An
example of the possible fragility of this accommodation (and as well the depth of our ignorance as to
these compensatory adjustments) is the synergistic
and compensatory relationship among the bacterial
genetic background, -lactamase expression, and the
maintenance and expression of plasmid-carried mecA
in -lactam-resistant S. aureus, as discussed by
Katayama et al.348 This observation would immediately suggest the beneficial combination of a -lactam
targeting the PBPs with a -lactam targeting the
-lactamase, which has been of course a mainstay of
clinical therapy for some two decades. The conclusion
that this combination therapy would prove successful
is arguably, in retrospect, facile. Whether future
antibacterial therapy of multidrug-resistant microorganisms will also require complex drug combinations (as is already the case for anti-HIV therapy) is
not nearly as obvious. This is, nonetheless, a likely
course of events. Fortunately, the experimental resources needed to accomplish this task are coming
into place. Single-cell microscopy, for example, may

Fisher et al.

allow one to identify the role of the target in cell wall


biosynthesis as the cell is seen to respond to drug
exposure.62,349-352 This response can be correlated to
proteomic analysis.289,291,353 The understanding of the
structural basis for -lactam induction of the bacterial SOS response354 and for creation of persister
bacteria populations290,355,356 are but two examples of
how future -lactam SAR may be guided.357 Screening to identify drug combinations that synergize with
-lactams (as is currently being done with efflux
pump inhibitors), with respect to target pairings, can
be done on a high-throughput scale. Old -lactams,
with proven safety and performance, can be given
continued clinical relevance.358 Nonetheless, the likelihood that future antibacterial chemotherapy will be
a multidrug regimen is real.359
The task of identifying and then optimizing multidrug safety and efficacy is daunting. Drug discovery
is already the zenith of the collaboration between
human scientific and engineering ingenuity, and the
emerging anticipation that this may now need to be
done on a multifactorial scale is part of the reason
that the economics for future antimicrobial drug
discovery are so dismal. While these economics may
change after the crisis (as only a crisis galvanizes
consensus of opinion), the prudent and sobering
reminder is that successful drug discovery is emphatically not instantaneous. Should our vigilance in
antibacterial drug discovery falter, the length and
depth of the crisis may be unlike anything modern
man has experienced.
There remain, of course, the companion questions
as to what should be done now to preserve -lactam
efficacy for future generations. There is no other antibacterial class that can substitute for -lactam antibiotics in the foreseeable future, and none are in the
pipeline.4,360 These are questions that involve the entire breadth of clinical practice, including minimizing
global environmental antibiotic exposure (the reduction of antibiotics in animal feeds is likely a step in
the right direction), re-appreciating the value of hospital quarantine, and re-emphasizing the incredible
importance of proper hygiene to minimizing infection.
At the level of drug therapy, additional possibilities
are emerging. We now appreciate that the gut is, to
use the delectable phrasing of Courvalin and Davies,4
a veritable microbial bordello with extensive capacity for genetic exchange.361,362 Exploratory therapies
that include -lactamases (to destroy nonabsorbed
-lactams, either in situ or post facto) have shown
promise.363-366 The interrelationship of communityand hospital-resistant microorganism reservoirs now
is recognized.367 Judicious use of early generation
-lactam therapy can mitigate resistance development against later generation -lactams.358,368-371
The central importance of the cell wall biosynthetic
enzymes as antibacterial targets is irrefutably validated by the -lactams themselves. Not only are
these enzymes accessible and essential, but also these
are enzymes with a demonstrated commonality for
inhibitor (and substrate) recognition at their active
sites. The proposal that the -lactams constitute the
only motif for inhibitor design is not merely unproven
but largely untested.372 Antibacterial screening is

Bacterial Resistance to -Lactam Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 419

only now transitioning from classical broth MIC


identification to discrete enzyme (target) screening
evaluation. As this is accomplished and as we more
clearly understand the relative importance of (say)
one specific PBP over another, opportunities for
structure-based -lactam design (a remarkably open
frontier!)14,28,373-376 and for new template identification will be created. The role and identity of each
enzyme contributing to the assembly of the cell wall
are only now emerging. As each falls into place, yet
another discrete target for antibacterial discovery is
acquired.
Nonetheless, the long-term future of the -lactams
is uncertain. While their widespread clinical use is
certain to continue for the foreseeable future, as with
all classes of anti-infective drugs continued efficacy
is the difference between chemical innovation and
clinical erosion by resistance development. The latter
is not merely certain but is both irreversible and
progressive. Also, while one should underestimate
neither human resilience nor human innovation, the
confluence of these attributes to accomplish drug
discovery has always required the context of need
and reward. While the imperfections of this system
have long been evident (look no further than the
obliviousness of major pharma to third-world disease), until now a demonstrable connection between
anti-infective need and reward has existed. This is
no longer true. Until society understands the difference between chemicals that are commodities and
chemicals that are creations, the investment of
human intellect in the -lactams may soon extinguish. Sheehans description of the -lactams as the
enchanted rings was a tribute to their intricacy,
safety, and efficacy. There is no reason to believe that
the burgeoning microbial resistance to their efficacy
is anything other than opportunity for further enchantment, unless we choose otherwise.

bacterial intervention. Two counterpoints on the


structure of the murein polymer are presented.383,384
Arbeloa et al. make the significant discovery that S.
aureus PBP2a can confer -lactam resistance to other
Gram-positive bacteria, with the synthesis of mosaic
peptidoglycan cross-bridges.385 Gardete et al. provide
further insight to the role of murE in S. aureus
resistance by control of PBP2 and PBP2a expression.386 The Class B (monofunctional) PBP2b transpeptidase from resistant S. pneumoniae has a T446A
mutation that reduces penicillin affinity by 99%.387
The crystal structures are disclosed of the R61 D,Dpeptidase (inactivated with a peptidoglycan-mimetic
penicillin),388 of a truncated S. pneumoniae Class A
cell division PBP1b enzyme (inactivated with nitrocefin and cefotaxime),389 and of the S. pneumoniae
PBP3 peptidoglycan synthesis regulatory factor.390
Labia reviews the structural evolution of the TEM
and SHV Class A -lactamases.391 Computational
modeling of the Class A -lactamase acylation supports Glu166 as the general base activating Ser70.392
An engineered cystine in the Toho-1 ESBL alters the
active site, reducing activity toward third-generation
cephalosporins.393 The sequence requirements of the
IMP-1 and FEZ-1 metallo--lactamases,394,395 evidence for direct -lactam-metal contact,396 and the
crystal structure of the CphA carbapenemase metallo--lactamase (complexed with biapenem)397 are
discussed. Two reports evaluate the structural basis
for the high affinity -lactamase-BLIP (-lactamase
inhibitory protein) protein-protein complex.398,399
Freiberg et al. discuss the impact of transcriptome
and proteome analysis on antibacterial drug discovery.400 The inactivation mechanism of broad-spectrum methylidene penem -lactamase inhibitors401 is
revealed by the structure of the inactivated -lactamase.402 A series of reviews update recent progress
in -lactam medicinal chemistry.403-409

6. Acknowledgments

9. References

The constructive comments of the referees are


appreciated. This work was supported by Grant
AI33170 from the National Institute of Allergy &
Infectious Disease and by Grant GM61629 from the
National Institute of General Medical Science.

7. Abbreviations
ESBL
IRT
MBL
MIC
MRSA
PBP
rms

extended spectrum -lactamase


inhibitor-resistant TEM (class A) -lactamase
metallo--lactamase
minimal inhibitory concentration of an antibacterial
methicillin-resistant S. aureus
penicillin-binding protein
root-mean square

8. Note Added in Proof


Levy and Marshall377 and Payne and Tomasz378
offer perspectives on the phenomenon of bacterial
resistance, whereas Poole379 provides a complimentary review of bacterial -lactam resistance. Mallorqui-Fernandez et al.,380 Walsh and Amyes,381 and
Gotz382 review aspects of the molecular basis for
MRSA/VRSA and possible new strategies for anti-

(1) Charpentier, E.; Tuomanen, E. Microbes Infect. 2000, 2, 1855.


(2) Walsh, C. T. Nat. Rev. Microbiol. 2003, 1, 65.
(3) Raja, A.; Lebbos, J.; Kirkpatrick, P. Nat. Rev. Drug Discovery
2004, 3, 733.
(4) Courvalin, P.; Davies, J. Curr. Opin. Microbiol. 2003, 6, 425.
(5) Andersson, D. I. Curr. Opin. Microbiol. 2003, 6, 452.
(6) Sutcliffe, J. A. Bioorg. Med. Chem. Lett. 2003, 13, 4159.
(7) Walsh, C. T. Nature 2000, 406, 775.
(8) Wise, R. J. Antimicrob. Chemother. 2004, 54, 306.
(9) Palumbi, S. Science 2001, 293, 1786.
(10) Czaran, T. L.; Hoekstra, R. F.; Pagie, L. Proc. Natl. Acad. Sci.
U.S.A. 2002, 99, 786.
(11) Enright, M. C. Curr. Opin. Pharmacol. 2003, 3, 474.
(12) Lenski, R. E.; Riley, M. A. Proc. Natl. Acad. Sci. U.S.A. 2002,
99, 556.
(13) Challis, G. L.; Hopwood, D. A. Proc. Natl. Acad. Sci. U.S.A. 2003,
100 (Suppl. 2), 14555.
(14) Pratt, R. F. J. Chem. Soc., Perkin Trans. 2 2002, 851.
(15) Hall, B. G.; Barlow, M. Drug Resist. Updates 2004, 7, 111.
(16) Abraham, E. BioEssays 1990, 12, 601.
(17) Bentley, R. Perspect. Biol. Med. 1997, 40, 197.
(18) Bentley, R. Perspect. Biol. Med. 1997, 40, 364.
(19) Williams, D. H.; Stone, M. J.; Hauck, P. R.; Rahman, S. K. J.
Nat. Prod. 1989, 52, 1189.
(20) Maplestone, R. A.; Stone, M. J.; Williams, D. H. Gene 1992, 115,
151.
(21) Stone, M. J.; Williams, D. H. Mol. Microbiol. 1992, 6, 29.
(22) Firn, R. D.; Jones, C. G. Mol. Microbiol. 2000, 37, 989.
(23) Firn, R. D.; Jones, C. G. Nat. Prod. Rep. 2003, 20, 382.
(24) Dalhoff, A.; Thomson, C. J. Chemotherapy 2003, 49, 105.
(25) Waxman, D. J.; Yocum, R. R.; Strominger, J. L. Philos. Trans.
R. Soc. London, Ser. B 1980, 289, 257.
(26) Koch, A. L. Clin. Microbiol. Rev. 2003, 16, 673.

420 Chemical Reviews, 2005, Vol. 105, No. 2


(27) Silvaggi, N. R.; Kaur, K.; Adediran, S. A.; Pratt, R. F.; Kelly, J.
A. Biochemistry 2004, 43, 7046.
(28) Josephine, H. R.; Kumar, I.; Pratt, R. F. J. Am. Chem. Soc. 2004,
126, 8122.
(29) Giesbrecht, P.; Kerstein, T.; Maidhof, H.; Wecke, J. Microbiol.
Mol. Biol. Rev. 1998, 62, 1371.
(30) Sugai, M.; Yamada, S.; Nakashima, S.; Komatsuzawa, H.;
Matsumoto, A.; Oshida, T.; Suginaka, H. J. Bacteriol. 1997, 179,
2958.
(31) Severin, A.; Severina, E.; Tomasz, A. Antimicrob. Agents Chemother. 1997, 41, 504.
(32) Beveridge, T. J. J. Bacteriol. 1999, 181, 4725.
(33) Lee, W.; McDonough, M. A.; Kotra, L. P.; Li, Z. H.; Silvaggi, N.
R.; Takeda, Y.; Kelly, J. A.; Mobashery, S. Proc. Natl. Acad. Sci.
U.S.A. 2001, 98, 1427.
(34) Page, M. I.; Laws, A. P. Chem. Commun. 1998, 1611.
(35) Woodward, R. B. Philos. Trans. R. Soc. London, Ser. B 1980,
289, 239.
(36) Sykes, N. O.; Macdonald, S. J. F.; Page, M. I. J. Med. Chem.
2002, 45, 2850.
(37) Benkovic, S. J.; Hammes-Schiffer, S. Science 2003, 301, 1196.
(38) Kraut, D. A.; Carroll, K. S.; Herschlag, D. Annu. Rev. Biochem.
2003, 72, 517.
(39) Williams, D. H.; Stephens, E.; Zhou, M. Chem. Commun. 2003,
1973.
(40) Rogers, H. J.; Perkins, H. R.; Ward, J. B. Microbial cell walls
and membranes; Chapman and Hall: London, New York, 1980.
(41) Bugg, T. D. H. In Comprehensive Natural Products Chemistry;
Elservier: New York, 1999.
(42) Young, K. D. Mol. Microbiol. 2003, 49, 571.
(43) Popham, D. L.; Young, K. D. Curr. Opin. Microbiol. 2003, 6, 594.
(44) Nanninga, N. Microbiol. Mol. Biol. Rev. 1998, 62, 110.
(45) Nikaido, H. Microbiol. Mol. Biol. Rev. 2003, 67, 593.
(46) Moews, P. C.; Knox, J. R.; Dideberg, O.; Charlier, P.; Frere, J.
M. Proteins: Struct., Funct., Genet. 1990, 7, 156.
(47) Kuzin, A. P.; Liu, H.; Kelly, J. A.; Knox, J. R. Biochemistry 1995,
34, 9532.
(48) Pares, S.; Mouz, N.; Petillot, Y.; Hakenbeck, R.; Dideberg, O.
Nat. Struct. Biol. 1996, 3, 284.
(49) Knox, J. R.; Moews, P. C.; Frere, J. M. Chem. Biol. 1996, 3, 937.
(50) Dideberg, O.; Charlier, P.; Dupont, L.; Vermeire, M.; Frere, J.
M.; Ghuysen, J. M. FEBS Lett. 1980, 117, 212.
(51) Kelly, J. A.; Moews, P. C.; Knox, J. R.; Frere, J. M.; Ghuysen, J.
M. Science 1982, 218, 479.
(52) Knox, J. R.; DeLucia, M. L.; Murthy, N. S.; Kelly, J. A.; Moews,
P. C.; Frere, J. M.; Ghuysen, J. M. J. Mol. Biol. 1979, 127, 217.
(53) Kelly, J. A.; Kuzin, A. P. J. Mol. Biol. 1995, 254, 223.
(54) Rhazi, N.; Charlier, P.; Dehareng, D.; Engher, D.; Vermeire, M.;
Fre`re, J. M.; Nguyen-Diste`che, M.; Fonze, E. Biochemistry 2003,
42, 2895.
(55) Fonze, E.; Vermeire, M.; Nguyen-Disteche, M.; Brasseur, R.;
Charlier, P. J. Biol. Chem. 1999, 274, 21853.
(56) Nicholas, R. A.; Krings, S.; Tomberg, J.; Nicola, G.; Davies, C.
J. Biol. Chem. 2003, 278, 52826.
(57) Lim, D.; Strynadka, N. C. Nat. Struct. Biol. 2002, 9, 870.
(58) Deka, R. K.; Machius, M.; Norgard, M. V.; Tomchick, D. R. J.
Biol. Chem. 2002, 277, 41857.
(59) Davies, C.; White, S. W.; Nicholas, R. A. J. Biol. Chem. 2001,
276, 616.
(60) Massova, I.; Mobashery, S. Antimicrob. Agents Chemother. 1998,
42, 1.
(61) Amanuma, H.; Strominger, J. L. J. Biol. Chem. 1984, 259, 1294.
(62) Scheffers, D. J.; Jones, L. J. F.; Errington, J. Mol. Microbiol.
2004, 51, 749.
(63) Pucci, M. J.; Dougherty, T. J. J. Bacteriol. 2002, 184, 588.
(64) Hartman, B. J.; Tomasz, A. J. Bacteriol. 1984, 158, 513.
(65) Pinho, M. G.; Filipe, S. R.; de Lencastre, H.; Tomasz, A. J.
Bacteriol. 2001, 183, 6525.
(66) Couto, I.; Wu, S. W.; Tomasz, A.; de Lencastre, H. J. Bacteriol.
2003, 185, 645.
(67) Jevons, M. P.; Rolinson, G. N.; Knox, R. Br. Med. J. 1961, 1,
124.
(68) Ito, T.; Katayama, Y.; Asada, K.; Mori, N.; Tsutsumimoto, K.;
Tiensasitorn, C.; Hiramatsu, K. Antimicrob. Agents Chemother.
2001, 45, 1323.
(69) Wu, S. W.; de Lencastre, H.; Tomasz, A. J. Bacteriol. 2001, 183,
2417.
(70) Hartman, B. J.; Tomasz, A. Antimicrob. Agents Chemother. 1986,
29, 85.
(71) Reynolds, P. E.; Brown, D. F. FEBS Lett. 1985, 192, 28.
(72) Utsui, Y.; Yokota, T. Antimicrob. Agents Chemother. 1985, 28,
397.
(73) Fuda, C.; Suvorov, M.; Vakulenko, S. B.; Mobashery, S. J. Biol.
Chem. 2004, 279, 40802.
(74) Lu, W. P.; Kincaid, E.; Sun, Y.; Bauer, M. D. J. Biol. Chem. 2001,
276, 31494.
(75) Pernot, L.; Chesnel, L.; Legouellec, A.; Croize, J.; Vernet, T.;
Dideberg, O.; Dessen, A. J. Biol. Chem. 2004, 279, 16463.

Fisher et al.
(76) Chesnel, L.; Pernot, L.; Lemaire, D.; Champelovier, D.; Croize,
J.; Dideberg, O.; Vernet, T.; Zapun, A. J. Biol. Chem. 2003, 278,
44448.
(77) Rice, L. B.; Bellais, S.; Carias, L. L.; Hutton-Thomas, R.; Bonomo,
R. A.; Caspers, P.; Page, M. G.; Gutmann, L. Antimicrob. Agents
Chemother. 2004, 48, 3028.
(78) Sifaoui, F.; Arthur, M.; Rice, L.; Gutmann, L. Antimicrob. Agents
Chemother. 2001, 45, 2594.
(79) Jones, R. N.; Marshall, S. A.; Pfaller, M. A.; Wilke, W. W.; Hollis,
R. J.; Erwin, M. E.; Edmond, M. B.; Wenzel, R. P. Diagn.
Microbiol. Infect. Dis. 1997, 29, 95.
(80) Sauvage, E.; Kerff, F.; Fonze, E.; Herman, R.; Schoot, B.;
Marquette, J. P.; Taburet, Y.; Prevost, D.; Dumas, J.; Leonard,
G.; Stefanic, P.; Coyette, J.; Charlier, P. Cell Mol. Life Sci. 2002,
59, 1223.
(81) Mouz, N.; Di Guilmi, A. M.; Gordon, E.; Hakenbeck, R.; Dideberg,
O.; Vernet, T. J. Biol. Chem. 1999, 274, 19175.
(82) Salerno, A. J.; Lampen, J. O. J. Bacteriol. 1986, 166, 769.
(83) McKinney, T. K.; Sharma, V. K.; Craig, W. A.; Archer, G. L. J.
Bacteriol. 2001, 183, 6862.
(84) Zhu, Y. F.; Curran, I. H.; Joris, B.; Ghuysen, J. M.; Lampen, J.
O. J. Bacteriol. 1990, 172, 1137.
(85) Hardt, K.; Joris, B.; Lepage, S.; Brasseur, R.; Lampen, J. O.;
Frere, J. M.; Fink, A. L.; Ghuysen, J. M. Mol. Microbiol. 1997,
23, 935.
(86) Kobayashi, T.; Zhu, Y. F.; Nicholls, N. J.; Lampen, J. O. J.
Bacteriol. 1987, 169, 3873.
(87) Clarke, S. R.; Dyke, K. G. Microbiology 2001, 147, 803.
(88) Finan, J. E.; Archer, G. L.; Pucci, M. J.; Climo, M. W. Antimicrob.
Agents Chemother. 2001, 45, 3070.
(89) Niemeyer, D. M.; Pucci, M. J.; Thanassi, J. A.; Sharma, V. K.;
Archer, G. L. J. Bacteriol. 1996, 178, 5464.
(90) Zhang, H. Z.; Hackbarth, C. J.; Chansky, K. M.; Chambers, H.
F. Science 2001, 291, 1962.
(91) Ender, M.; McCallum, N.; Adhikari, R.; Berger-Bachi, B. Antimicrob. Agents Chemother. 2004, 48, 2295.
(92) Gregory, P. D.; Lewis, R. A.; Curnock, S. P.; Dyke, K. G. Mol.
Microbiol. 1997, 24, 1025.
(93) Golemi-Kotra, D.; Cha, J. Y.; Meroueh, S. O.; Vakulenko, S. B.;
Mobashery, S. J. Biol. Chem. 2003, 278, 18419.
(94) Kerff, F.; Charlier, P.; Colombo, M. L.; Sauvage, E.; Brans, A.;
Frere, J. M.; Joris, B.; Fonze, E. Biochemistry 2003, 42, 12835.
(95) Golemi, D.; Maveyraud, L.; Vakulenko, S.; Samama, J. P.;
Mobashery, S. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 14280.
(96) Maveyraud, L.; Golemi, D.; Kotra, L. P.; Tranier, S.; Vakulenko,
S.; Mobashery, S.; Samama, J. P. Struct. Fold. Des. 2000, 8,
1289.
(97) Sun, T.; Nukaga, M.; Mayama, K.; Braswell, E. H.; Knox, J. R.
Protein Sci. 2003, 12, 82.
(98) Wilke, M. S.; Hills, T. L.; Zhang, H. Z.; Chambers, H. F.;
Strynadka, N. C. J. Biol. Chem. 2004, 47278.
(99) Birck, C.; Cha, J. Y.; Cross, J.; Schulze-Briese, C.; Meroueh, S.
O.; Schlegel, H. B.; Mobashery, S.; Samama, J. P. J. Am. Chem.
Soc. 2004, 126, 13945.
(100) Hanique, S.; Colombo, M. L.; Goormaghtigh, E.; Soumillion, P.;
Frere, J. M.; Joris, B. J. Biol. Chem. 2004, 279, 14264.
(101) Filee, P.; Benlafya, K.; Delmarcelle, M.; Moutzourelis, G.; Frere,
J. M.; Brans, A.; Joris, B. Mol. Microbiol. 2002, 44, 685.
(102) Melckebeke, H. V.; Vreuls, C.; Gans, P.; Filee, P.; Llabres, G.;
Joris, B.; Simorre, J. P. J. Mol. Biol. 2003, 333, 711.
(103) Garcia-Castellanos, R.; Marrero, A.; Mallorqui-Fernandez, G.;
Potempa, J.; Coll, M.; Gomis-Ruth, F. X. J. Biol. Chem. 2003,
278, 39897.
(104) Garcia-Castellanos, R.; Mallorqui-Fernandez, G.; Marrero, A.;
Potempa, J.; Coll, M.; Gomis-Ruth, F. X. J. Biol. Chem. 2004,
279, 17888.
(105) Bush, K.; Jacoby, G. A.; Medeiros, A. A. Antimicrob. Agents
Chemother. 1995, 39, 1211.
(106) Bush, K.; Mobashery, S. In Resolving the Antibiotic Paradox;
Rosen, B. P., Mobashery, S., Eds.; Kluwer Academic/Plenum
Publishers: New York, 1998; Vol. 456.
(107) Bryant, R. A. R.; Hansen, D. E. J. Am. Chem. Soc. 1996, 118,
5498.
(108) Radzicka, A.; Wolfenden, R. J. Am. Chem. Soc. 1996, 118, 6105.
(109) Thomson, K. S.; Moland, E. S. Microbes Infect. 2000, 2, 1225.
(110) Kotra, L.; Samama, J. P.; Mobashery, S. In Bacterial Resistance
to Antimicrobials, Mechanisms, Genetics, Medical Practice and
Public Health; Lewis, K., Salyers, A. A., Haber, H. W., Wax, R.
G., Eds.; Marcel Dekkar, Inc.: New York, 2002.
(111) Cha, J. Y.; Ishiwata, A.; Mobashery, S. J. Biol. Chem. 2004, 279,
14917.
(112) Nielsen, B. K.; Lampen, J. O. J. Biol. Chem. 1982, 257, 490.
(113) Meroueh, S. O.; Minasov, G.; Lee, W.; Shoichet, B. K.; Mobashery, S. J. Am. Chem. Soc. 2003, 125, 9612.
(114) Hardy, L. W.; Kirsch, J. F. Biochemistry 1984, 23, 1275.
(115) Bulychev, A.; Mobashery, S. Antimicrob. Agents Chemother.
1999, 43, 1743.
(116) Page, M. I.; Laws, A. P. Tetrahedron 2000, 56, 5631.

Bacterial Resistance to -Lactam Antibiotics


(117) Hokenson, M. J.; Cope, G. A.; Lewis, E. R.; Oberg, K. A.; Fink,
A. L. Biochemistry 2000, 39, 6538.
(118) Wang, X.; Minasov, G.; Shoichet, B. K. Proteins: Struct., Funct.,
Genet. 2002, 47, 86.
(119) Wang, X.; Minasov, G.; Shoichet, B. K. J. Mol. Biol. 2002, 320,
85.
(120) Hujer, A. M.; Hujer, K. M.; Helfand, M. S.; Anderson, V. E.;
Bonomo, R. A. Antimicrob. Agents Chemother. 2002, 46, 3971.
(121) Helfand, M. S.; Bethel, C. R.; Hujer, A. M.; Hujer, K. M.;
Anderson, V. E.; Bonomo, R. A. J. Biol. Chem. 2003, 278, 52724.
(122) Weldhagen, G. F.; Poirel, L.; Nordmann, P. Antimicrob. Agents
Chemother. 2003, 47, 2385.
(123) Bonnet, R. Antimicrob. Agents Chemother. 2004, 48, 1.
(124) Bradford, P. A. Clin. Rev. Microbiol. 2001, 14, 933.
(125) Davies, C.; White, S. W.; Nicholas, R. A. J. Biol. Chem. 2000,
276, 616.
(126) Tranier, S.; Bouthors, A. T.; Maveyraud, L.; Guillet, V.; Sougakoff, W.; Samama, J. P. J. Biol. Chem. 2000, 275, 28075.
(127) Moult, J.; Sawyer, L.; Herzberg, O.; Jones, C. L.; Coulson, A. F.;
Green, D. W.; Harding, M. M.; Ambler, R. P. Biochem. J. 1985,
225, 167.
(128) Jelsch, C.; Lenfant, F.; Masson, J. M.; Samama, J. P. FEBS Lett.
1992, 299, 135.
(129) Jelsch, C.; Mourey, L.; Masson, J. M.; Samama, J. P. Proteins
1993, 16, 364.
(130) Knox, J. R.; Moews, P. C. J. Mol. Biol. 1991, 220, 435.
(131) Strynadka, N. C.; Adachi, H.; Jensen, S. E.; Johns, K.; Sielecki,
A.; Betzel, C.; Sutoh, K.; James, M. N. Nature 1992, 359, 700.
(132) Tipper, D. J.; Strominger, J. L. Proc. Natl. Acad. Sci. U.S.A.
1965, 54, 1133.
(133) Minasov, G.; Wang, X.; Shoichet, B. K. J. Am. Chem. Soc. 2002,
124, 5333.
(134) Fujii, Y.; Okimoto, N.; Hata, M.; Narumi, T.; Yasuoka, K.;
Susukita, R.; Suenaga, A.; Futatsugi, N.; Koishi, T.; Furusawa,
H.; Kawai, A.; Ebisuzaki, T.; Neya, S.; Hoshino, T. J. Phys.
Chem. B 2003, 107, 10274.
(135) Hermann, J. C.; Ridder, L.; Mulholland, A. J.; Holtje, H. D. J.
Am. Chem. Soc. 2003, 125, 9590.
(136) Nukaga, M.; Mayama, K.; Hujer, A. M.; Bonomo, R. A.; Knox,
J. R. J. Mol. Biol. 2003, 328, 289.
(137) Guillaume, G.; Vanhove, M.; Lamotte-Brasseur, J.; Ledent, P.;
Jamin, M.; Joris, B.; Frere, J. M. J. Biol. Chem. 1997, 272, 5438.
(138) Lamotte-Brasseur, J.; Dive, G.; Dideberg, O.; Charlier, P.; Frere,
J. M.; Ghuysen, J. M. Biochem. J. 1991, 279, 213.
(139) Damblon, C.; Raquet, X.; Lian, L. Y.; Lamotte-Brasseur, J.;
Fonze, E.; Charlier, P.; Roberts, G. C. K.; Frere, J. M. Proc. Natl.
Acad. Sci. U.S.A. 1996, 93, 1747.
(140) Lamotte-Brasseur, J.; Lounnas, V.; Raquet, X.; Wade, R. C.
Protein Sci. 1999, 8, 404.
(141) Raquet, X.; Lounnas, V.; Lamotte-Brasseur, J.; Frere, J. M.;
Wade, R. C. Biophys. J. 1997, 73, 2416.
(142) Swaren, P.; Maveyraud, L.; Guillet, V.; Masson, J. M.; Mourey,
L.; Samama, J. P. Structure 1995, 3, 603.
(143) Golemi-Kotra, D.; Meroueh, S. O.; Kim, C.; Vakulenko, S. B.;
Bulychev, A.; Stemmler, A. J.; Stemmler, T. L.; Mobashery, S.
J. Biol. Chem. 2004, 34665.
(144) Finlay, J.; Miller, L.; Poupard, J. A. J. Antimicrob. Chemother.
2003, 52, 18.
(145) Imtiaz, U.; Billings, E. M.; Knox, J. R.; Mobashery, S. Biochemistry 1994, 33, 5728.
(146) Imtiaz, U.; Billings, E.; Knox, J. R.; Manavathu, E. K.; Lerner,
S. A.; Mobashery, S. J. Am. Chem. Soc. 1993, 115, 4435.
(147) Yang, Y.; Janota, K.; Tabei, K.; Huang, N.; Siegel, M. M.; Lin,
Y. I.; Rasmussen, B. A.; Shlaes, D. M. J. Biol. Chem. 2000, 275,
26674.
(148) Kuzin, A. P.; Nukaga, M.; Nukaga, Y.; Hujer, A.; Bonomo, R.
A.; Knox, J. R. Biochemistry 2001, 40, 1861.
(149) Helfand, M. S.; Totir, M. A.; Carey, P.; Hujer, A. M.; Bonomo,
R. A.; Carey, P. R. Biochemistry 2003, 42, 13386.
(150) Padayatti, P. S.; Helfand, M. S.; Totir, M. A.; Carey, M. P.; Hujer,
A. M.; Carey, P. R.; Bonomo, R. A.; van den Akker, F. Biochemistry 2004, 43, 843.
(151) Swaren, P.; Golemi, D.; Cabantous, S.; Bulychev, A.; Maveyraud,
L.; Mobashery, S.; Samama, J. P. Biochemistry 1999, 38, 9570.
(152) Meroueh, S. O.; Roblin, P.; Golemi, D.; Maveyraud, L.; Vakulenko, S. B.; Zhang, Y.; Samama, J. P.; Mobashery, S. J. Am.
Chem. Soc. 2002, 124, 9422.
(153) Madec, S.; Blin, C.; Krishnamoorthy, R.; Picard, B.; Chaibi, E.
B.; Fouchereau-Peron, M.; Labia, R. FEMS Microbiol. Lett. 2002,
211, 13.
(154) Wang, X.; Minasov, G.; Shoichet, B. K. J. Biol. Chem. 2002, 277,
24744.
(155) Wang, X.; Minasov, G.; Blazquez, J.; Caselli, E.; Prati, F.;
Shoichet, B. K. Biochemistry 2003, 42, 8434.
(156) Doucet, N.; De Wals, P. Y.; Pelletier, J. N. J. Biol. Chem. 2004,
279, 46295.
(157) Pagan-Rodriguez, D.; Zhou, X.; Simmons, R.; Bethel, C. R.; Hujer,
A. M.; Helfand, M. S.; Jin, Z.; Guo, B.; Anderson, V. E.; Lily,
M.; Ng, L. M.; Bonomo, R. A. J. Biol. Chem. 2004, 279, 19494.

Chemical Reviews, 2005, Vol. 105, No. 2 421


(158) Chaibi, E. B.; Sirot, D.; Paul, G.; Labia, R. J. Antimicrob.
Chemother. 1999, 43, 447.
(159) Lavigne, J. P.; Bonnet, R.; Michaux-Charachon, S.; Jourdan, J.;
Caillon, J.; Sotto, A. J. Antimicrob. Chemother. 2004, 53, 616.
(160) Nukaga, M.; Abe, T.; Venkatesan, A. M.; Mansour, T. S.; Bonomo,
R. A.; Knox, J. R. Biochemistry 2003, 42, 13152.
(161) Beharry, Z.; Chen, H.; Gadhachanda, V. R.; Buynak, J. D.;
Palzkill, T. Biochem. Biophys. Res. Commun. 2004, 313, 541.
(162) Bret, L.; Chaibi, E. B.; Chanal-Claris, C.; Sirot, D.; Labia, R.;
Sirot, J. Antimicrob. Agents Chemother. 1997, 41, 2547.
(163) Delaire, M.; Labia, R.; Samama, J. P.; Masson, J. M. J. Biol.
Chem. 1992, 267, 20600.
(164) Zafaralla, G.; Manavathu, E. K.; Lerner, S. A.; Mobashery, S.
Biochemistry 1992, 31, 3847.
(165) Delaire, M.; Labia, R.; Samama, J. P.; Masson, J. M. J. Biol.
Chem. 1992, 267, 20600.
(166) Schroeder, W. A.; Locke, T. R.; Jensen, S. E. Antimicrob. Agents
Chemother. 2002, 46, 3568.
(167) Randegger, C. C.; Hachler, H. J. Antimicrob. Chemother. 2001,
47, 547.
(168) Stapleton, P. D.; Shannon, K. P.; French, G. L. Antimicrob.
Agents Chemother. 1999, 43, 1881.
(169) Vakulenko, S. B.; Taibi-Tronche, P.; Toth, M.; Massova, I.;
Lerner, S. A.; Mobashery, S. J. Biol. Chem. 1999, 274, 23052.
(170) Vakulenko, S.; Golemi, D. Antimicrob. Agents Chemother. 2002,
46, 646.
(171) Maveyraud, L.; Mourey, L.; Kotra, L. P.; Pedelacq, J. D.; Guillet,
V.; Mobashery, S.; Samama, J. P. J. Am. Chem. Soc. 1998, 120,
9748.
(172) Vilanova, B.; Donoso, J.; Frau, J.; Munoz, F. Helv. Chim. Acta
1999, 82, 1274.
(173) Fonze, E.; Vanhove, M.; Dive, G.; Sauvage, E.; Fre`re, J. M.;
Charlier, P. Biochemistry 2002, 41, 1877.
(174) Sougakoff, W.; Naas, T.; Nordmann, P.; Collatz, E.; Jarlier, V.
Biochim. Biophys. Acta 1999, 1433, 153.
(175) Sougakoff, W.; LHermite, G.; Pernot, L.; Naas, T.; Guillet, V.;
Nordmann, P.; Jarlier, V.; Delettre, J. Acta Crystallogr., Sect.
D 2002, 58, 267.
(176) Majiduddin, F. K.; Palzkill, T. Antimicrob. Agents Chemother.
2003, 47, 1062.
(177) Mourey, L.; Miyashita, K.; Swaren, P.; Bulychev, A.; Samama,
J. P.; Mobashery, S. J. Am. Chem. Soc. 1998, 120, 9382.
(178) Hujer, A. M.; Hujer, K. M.; Bonomo, R. A. Biochim. Biophys.
Acta 2001, 1547, 37.
(179) Cantu, C., III; Palzkill, T. J. Biol. Chem. 1998, 273, 26603.
(180) Orencia, M. C.; Yoon, J. S.; Ness, J. E.; Stemmer, W. P. C.;
Stevens, R. C. Nat. Struct. Biol. 2001, 8, 238.
(181) Nukaga, K.; Mayama, K.; Crichlow, G. V.; Knox, J. R. J. Mol.
Biol. 2002, 317, 109.
(182) Shimamura, T.; Ibuka, A. S.; Fushinobu, S.; Wakagi, T.; Ishiguro,
M.; Ishii, Y.; Matsuzawa, H. J. Biol. Chem. 2002, 277, 46601.
(183) Ibuka, A. S.; Ishii, Y.; Galleni, M.; Ishiguro, M.; Yamaguchi, K.;
Fre`re, J. M.; Matsuzawa, H.; Sakai, H. Biochemistry 2003, 42,
10634.
(184) Hall, A.; Knowles, J. R. Nature 1976, 264, 803.
(185) Stemmer, W. P. C. J. Mol. Catal. B 2002, 19, 3.
(186) Camps, M.; Naukkarinen, J.; Johnson, B. P.; Loeb, L. A. Proc.
Natl. Acad. Sci. U.S.A. 2003, 100, 9727.
(187) Vakulenko, S. B.; Geryk, B.; Kotra, L. P.; Mobashery, S.; Lerner,
S. A. Antimicrob. Agents Chemother. 1998, 42, 1542.
(188) Barlow, M.; Hall, B. G. Genetics 2003, 163, 1237.
(189) Barlow, M.; Hall, B. G. Genetics 2003, 164, 23.
(190) Hall, B. G. Nat. Rev. Microbiol. 2004, 2, 430.
(191) Lobkovsky, E.; Moews, P. C.; Liu, H.; Zhao, H.; Frere, J. M.;
Knox, J. R. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 11257.
(192) Oefner, C.; DArcy, A.; Daly, J. J.; Gubernator, K.; Charnas, R.
L.; Heinze, I.; Hubschwerlen, C.; Winkler, F. K. Nature 1990,
343, 284.
(193) Gherman, B. F.; Goldberg, S. D.; Cornish, V. W.; Friesner, R. A.
J. Am. Chem. Soc. 2004, 126, 7652.
(194) Bulychev, A.; Massova, I.; Miyashita, K.; Mobashery, S. J. Am.
Chem. Soc. 1997, 119, 7619.
(195) Patera, A.; Blaszczak, L. C.; Shoichet, B. K. J. Am. Chem. Soc.
2000, 122, 10504.
(196) Rice, L. B.; Bonomo, R. A. Drug Resist. Update 2000, 3, 178.
(197) Bauernfeind, A.; Chong, Y.; Schweighart, S. Infection 1989, 17,
316.
(198) Bauernfeind, A.; Chong, Y.; Lee, K. Yonsei Med. J. 1998, 39,
520.
(199) Livermore, D. M. Clin. Microbiol. Rev. 1995, 8, 557.
(200) Philippon, A.; Arlet, G.; Lagrange, P. H. Eur. J. Clin. Microbiol.
Infect. Dis. 1994, 13 (Suppl 1), S17.
(201) Galleni, M.; Amicosante, G.; Frere, J. M. Biochem. J. 1988, 255,
123.
(202) Kato-Toma, Y.; Iwashita, T.; Masuda, K.; Oyama, Y.; Ishiguro,
M. Biochem. J. 2003, 371, 175.
(203) Lamotte-Brasseur, J.; Dubus, A.; Wade, R. C. Proteins 2000, 40,
23.

422 Chemical Reviews, 2005, Vol. 105, No. 2


(204) Dubus, A.; Normark, S.; Kania, M.; Page, M. G. Biochemistry
1994, 33, 8577.
(205) Knothe, H.; Shah, P.; Krcmery, V.; Antal, M.; Mitsuhashi, S.
Infection 1983, 11, 315.
(206) Nukaga, M.; Haruta, S.; Tanimoto, K.; Kogure, K.; Taniguchi,
K.; Tamaki, M.; Sawai, T. J. Biol. Chem. 1995, 270, 5729.
(207) Crichlow, G. V.; Kuzin, A. P.; Nukaga, M.; Mayama, K.; Sawai,
T.; Knox, J. R. Biochemistry 1999, 38, 10256.
(208) Powers, R. A.; Caselli, E.; Focia, P. J.; Prati, F.; Shoichet, B. K.
Biochemistry 2001, 40, 9207.
(209) Nukaga, M.; Kumar, S.; Nukaga, K.; Pratt, R. F.; Knox, J. R. J.
Biol. Chem. 2004, 279, 9344.
(210) Nordmann, P.; Poirel, L. Clin. Microbiol. Infect. 2002, 8, 321.
(211) Heritier, C.; Poirel, L.; Nordmann, P. Antimicrob. Agents
Chemother. 2004, 48, 1670.
(212) Bou, G.; Martnez-Beltran, J. Antimicrob. Agents Chemother.
2000, 44, 428.
(213) Poirel, L.; Girlich, D.; Naas, T.; Nordmann, P. Antimicrob. Agents
Chemother. 2001, 45, 447.
(214) Toleman, M. A.; Rolston, K.; Jones, R. N.; Walsh, T. R. Antimicrob. Agents Chemother. 2003, 47, 2859.
(215) Dubois, V.; Arpin, C.; Quentin, C.; Texier-Maugein, J.; Poirel,
L.; Nordmann, P. Antimicrob. Agents Chemother. 2003, 47, 2380.
(216) Oliver, A.; Weigel, L. M.; Rasheed, J. K.; McGowan, J. E.; Raney,
P.; Tenover, F. C. Antimicrob. Agents Chemother. 2002, 46, 3829.
(217) Bonnet, R.; Marchandin, H.; Chanal, C.; Sirot, D.; Labia, R.; De
Champs, C.; Jumas-Bilak, E.; Sirot, D. Antimicrob. Agents
Chemother. 2002, 46, 2004.
(218) Orman, B. E.; Pineiro, S. A.; Arduino, S.; Galas, M.; Melano, R.;
Caffer, M. I.; Sordelli, D. O.; Centron, D. Antimicrob. Agents
Chemother. 2002, 46, 3963.
(219) Melano, R.; Corso, A.; Petroni, A.; Centron, D.; Orman, B.;
Pereyra, A.; Moreno, N.; Galas, M. J. Antimicrob. Chemother.
2003, 52, 36.
(220) Poirel, L.; Heritier, C.; Nordmann, P. Antimicrob. Agents
Chemother. 2004, 48, 348.
(221) Bou, G.; Cervero, G.; Dominguez, M. A.; Quereda, C.; MartinezBeltran, J. J. Clin. Microbiol. 2000, 38, 3299.
(222) Afzal-Shah, M.; Woodford, N.; Livemore, D. M. Antimicrob.
Agents Chemother. 2001, 45, 583.
(223) Danes, C.; Navia, M. N.; Ruiz, J.; Marco, F.; Jurado, A.; Jimenez
de Anta, M. T.; Vila, J. J. Antimicrob. Chemother. 2002, 50, 261.
(224) Lopez-Otsoa, F.; Gallego, L.; Towner, K. J.; Tysall, L.; Woodford,
N.; Livermore, D. M. J. Clin. Microbiol. 2002, 40, 4741.
(225) Dalla-Costa, L. M.; Coelho, J. M.; Souza, H. A. P.; Castro, M. E.
S.; Stier, C. J. N.; Bragagnolo, K. L.; Rea-Neto, A.; PenteadoFilho, S. R.; Livermore, D. M.; Woodford, N. J. Clin. Microbiol.
2003, 41, 3403.
(226) Heritier, C.; Poirel, L.; Aubert, D.; Nordmann, P. Antimicrob.
Agents Chemother. 2003, 47, 268.
(227) Mushtaq, S.; Ge, Y.; Livermore, D. Antimicrob. Agents Chemother. 2004, 48, 1313.
(228) Zarrilli, R.; Crispino, M.; Bagattini, M.; Barretta, E.; Di Popolo,
A.; Triassi, A. M.; Villari, P. J. Clin. Microbiol. 2004, 42, 946.
(229) Livermore, D. M.; Woodford, N. Curr. Opin. Microbiol. 2000, 3,
489.
(230) Danel, F.; Frere, J. M.; Livemore, D. M. Biochim. Biophys. Acta
2001, 1546, 132.
(231) Franceschini, N.; Segatore, B.; Perilli, M.; Vessilier, S.; Franchino, L.; Amicosante, G. J. Antimicrob. Chemother. 2002, 49, 395.
(232) Pernot, L.; Frenois, F.; Rybkine, T.; LHermite, G.; Petrella, S.;
Delettre, J.; Jarlier, V.; Collatz, E.; Sougakoff, W. J. Mol. Biol.
2001, 310, 859.
(233) Danel, F.; Paetzel, M.; Strynadka, N. C. J.; Page, M. G. P.
Biochemistry 2001, 40, 9412.
(234) Franceschini, N.; Boschi, L.; Pollini, S.; Herman, R.; Perilli, M.;
Galleni, M.; Frere, J. M.; Amicosante, G.; Rossolini, G. M.
Antimicrob. Agents Chemother. 2001, 45, 3509.
(235) Paetzel, M.; Danel, F.; de Castro, L.; Mosimann, S. C.; Page, M.
G. P.; Strynadka, N. C. J. Nat. Struct. Biol. 2000, 7, 918.
(236) Golemi, D.; Maveyraud, L.; Vakulenko, S.; Tranier, S.; Ishiwata,
A.; Kotra, L. P.; Samama, J. P.; Mobashery, S. J. Am. Chem.
Soc. 2000, 122, 6132.
(237) Maveyraud, L.; Golemi-Kotra, D.; Ishiwata, A.; Meroueh, O.;
Mobashery, S.; Samama, J. P. J. Am. Chem. Soc. 2002, 124,
2461.
(238) Donald, H. M.; Scaife, W.; Amyes, S. G. B.; Young, H. K.
Antimicrob. Agents Chemother. 2000, 44, 196.
(239) Barlow, M.; Hall, B. G. J. Mol. Evol. 2002, 55, 314.
(240) Zhu, Y.; Englebert, S.; Joris, B.; Ghuysen, J. M.; Kobayashi, T.;
Lampen, J. O. J. Bacteriol. 1992, 174, 6171.
(241) Colombo, M. L.; Hanique, S.; Baurin, S. L.; Bauvois, C.; De
Vriendt, K.; Van Beeumen, J. J.; Fre`re, J. M.; Joris, B. Antimicrob. Agents Chemother. 2004, 48, 484.
(242) Garau, G.; Garcia-Saez, I.; Bebrone, C.; Anne, C.; Mercuri, P.;
Galleni, M.; Fre`re, J. M.; Dideberg, O. Antimicrob. Agents
Chemother. 2004, 48, 2347.

Fisher et al.
(243) Galleni, M.; Lamotte-Brasseur, J.; Rossolini, G. M.; Spencer, J.;
Dideberg, O.; Fre`re, J. M. Antimicrob. Agents. Chemother. 2001,
45, 660.
(244) Gniadkowski, M. Clin. Microbiol. Infect. 2001, 7, 597.
(245) Fast, W.; Wang, Z.; Benkovic, S. J. Biochemistry 2001, 40, 1640.
(246) Fabiane, S. M.; Sohi, M. K.; Wan, T.; Payne, D. J.; Bateson, J.
H.; Mitchell, T.; Sutton, B. J. Biochemistry 1998, 37, 12404.
(247) Wommer, S.; Rival, S.; Hein, U.; Galleni, M.; Fre`re, J. M.;
Franceschini, N.; Amicosante, G.; Rasmussen, B.; Bauer, R.;
Adolph, H. W. J. Biol. Chem. 2002, 277, 24142.
(248) Concha, N. O.; Janson, C. A.; Rowling, P.; Pearson, S.; Cheever,
C. A.; Clarke, B. P.; Lewis, C.; Galleni, M.; Fre`re, J. M.; Payne,
D. J.; Bateson, J. H.; Abdel-Meguid, S. S. Biochemistry 2000,
39, 4288.
(249) Gomes, C. M.; Frazao, C.; Xavier, A. V.; Legall, J.; Teixeira, M.
Protein Sci. 2002, 11, 707.
(250) Schilling, O.; Wenzel, N.; Naylor, M.; Vogel, A.; Crowder, M. W.;
Makaroff, C.; Meyer-Klaucke, W. Biochemistry 2003, 42, 11777.
(251) Spencer, J.; Clarke, A. R.; Walsh, T. R. J. Biol. Chem. 2001, 276,
33638.
(252) Rasia, R. M.; Vila, A. J. J. Biol. Chem. 2004, 279, 26046.
(253) Ullah, J. H.; Walsh, T. R.; Taylor, I. A.; Emery, D. C.; Verma,
C. S.; Gamblin, S. J.; Spencer, J. J. Mol. Biol. 1998, 284, 125.
(254) Rasia, R. M.; Ceoln, M.; Vila, A. J. Protein Sci. 2003, 12, 1538.
(255) Garrity, J. D.; Carenbauer, A. L.; Herron, L. R.; Crowder, M.
W. J. Biol. Chem. 2004, 279, 920.
(256) Dal Peraro, M.; Vila, A. J.; Carloni, P. Proteins: Struct., Funct.,
Bioinform. 2004, 54, 412.
(257) Dal Peraro, M.; Vila, A. J.; Carloni, P. Inorg. Chem. 2003, 42,
4245.
(258) Garceia-Saez, I.; Mercuri, P. S.; Papamicael, C.; Kahn, R.; Fre`re,
J. M.; Galleni, M.; Rossolini, G. M.; Dideberg, O. J. Mol. Biol.
2003, 325, 651.
(259) Garceia-Saez, I.; Hopkins, J.; Papamicael, C.; Franceschini, N.;
Amicosante, G.; Rossolini, G. M.; Galleni, M.; Fre`re, J. M.;
Dideberg, O. J. Biol. Chem. 2003, 278, 23868.
(260) Bounaga, S.; Laws, A. P.; Galleni, M.; Page, M. I. Biochem. J.
1998, 331, 703.
(261) Bush, K. Clin. Infect. Dis. 2001, 32, 1085.
(262) Miriagou, V.; Tzelepi, E.; Gianneli, D.; Tzouvelekis, L. S.
Antimicrob. Agents Chemother. 2003, 47, 395.
(263) Yanchak, M. P.; Taylor, R. A.; Crowder, M. W. Biochemistry
2000, 39, 11330.
(264) Krauss, M.; Gilson, H. S. R.; Gresh, N. J. Phys. Chem. B 2001,
105, 8040.
(265) Rasia, R. M.; Vila, A. J. Biochemistry 2002, 41, 1853.
(266) Lipton, A. S.; Heck, R. W.; Ellis, P. D. J. Am. Chem. Soc. 2004,
126, 4735.
(267) Montoya-Pelaez, P. J.; Brown, R. S. Inorg. Chem. 2002, 41, 309.
(268) Montoya-Pelaez, P. J.; Gibson, G. T. T.; Neverov, A. A.; Brown,
R. S. Inorg. Chem. 2003, 42, 8624.
(269) Daz, N.; Sordo, T. L.; Suarez, D.; Mendez, R.; Martn-Villacorta,
J. New J. Chem. 2004, 28, 15.
(270) Diaz, N.; Suarez, D.; Merz, K. M., Jr. J. Am. Chem. Soc. 2001,
123, 9867.
(271) Moali, C.; Anne, C.; Lamotte-Brasseur, J.; Groslambert, S.;
Devreese, B.; Van Beeumen, J.; Galleni, M.; Fre`re, J. M. Chem.
Biol. 2003, 10, 319.
(272) Oelschlaeger, P.; Schmid, R. D.; Pleiss, J. Biochemistry 2003,
42, 8945.
(273) Luzzaro, F.; Docquier, J. D.; Colinon, C.; Endimiani, A.; Lombardi, G.; Amicosante, G.; Rossolini, G.; Toniolol, A. Antimicrob.
Agents Chemother. 2004, 48, 648.
(274) Weldhagen, G. F. Int. J. Antimicrob. Agents 2004, 23, 556.
(275) Docquier, J. D.; Riccio, M. L.; Mugnaioli, C.; Luzzaro, F.;
Endimiani, A.; Toniolo, A.; Amicosante, G.; Rossolini, G. M.
Antimicrob. Agents Chemother. 2003, 47, 1522.
(276) Iyobe, S.; Kusadokoro, H.; Ozaki, J.; Matsumura, N.; Minami,
S.; Haruata, S.; Sawai, T.; OHara, K. Antimicrob. Agents
Chemother. 2000, 44, 2023.
(277) Yano, H.; Kuga, A.; Okamoto, R.; Kitasato, H.; Kobayashi, T.;
Inoue, M. Antimicrob. Agents Chemother. 2001, 45, 1343.
(278) Materon, I. C.; Palzkill, T. Protein Sci. 2001, 10, 2556.
(279) Hall, B. G. Antimicrob. Agents Chemother. 2004, 48, 1032.
(280) Docquier, J. D.; Lamotte-Brasseur, J.; Galleni, M.; Amicosante,
G.; Frere, J. M.; Rossolini, G. M. J. Antimicrob. Chemother. 2003,
51, 257.
(281) Toleman, M. A.; Rolston, K.; Jones, R. N.; Walsh, T. R. Antimicrob. Agents Chemother. 2004, 48, 329.
(282) Cornaglia, G.; Mazzariol, A.; Lauretti, L.; Rossolini, M.; Fontana,
R. Clin. Infect. Dis. 2000, 31, 1119.
(283) Yatsuyanagi, J.; Saito, S.; Harata, S.; Suzuki, N.; Ito, Y.; Amano,
K.; Enomoto, K. Antimicrob. Agents Chemother. 2004, 48, 626.
(284) Siemann, S.; Clarke, A. J.; Viswanatha, T.; Dmitrienko, G. I.
Biochemistry 2003, 42, 1673.
(285) Tsang, W. Y.; Dhanda, A.; Schofield, C. J.; Fre`re, J. M.; Galleni,
M.; Page, M. I. Bioorg. Med. Chem. Lett. 2004, 2004, 1737.

Bacterial Resistance to -Lactam Antibiotics


(286) Buynak, J. D.; Chen, H.; Vogeti, L.; Gadhachanda, V. R.;
Buchanan, C. A.; Palzkill, T.; Shaw, R. W.; Spencer, J.; Walsh,
T. R. Bioorg. Med. Chem. Lett. 2004, 14, 1299.
(287) Heinz, U.; Bauer, R.; Wommer, S.; Meyer-Klaucke, W.; Papamichaels, C.; Bateson, J.; Adolph, H. W. J. Biol. Chem. 2003, 278,
20659.
(288) Mollard, C.; Moali, C.; Papamicael, C.; Damblon, C.; Vessilier,
S.; Amicosante, G.; Schofield, C. J.; Galleni, N.; Fre`re, J. M.;
Roberts, G. C. K. J. Biol. Chem. 2001, 276, 45015.
(289) Goh, E. B.; Yim, G.; Tsui, W.; McClure, J.; Surette, M. G.; Davies,
J. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 17025.
(290) Kaldalu, N.; Mei, R.; Lewis, K. Antimicrob. Agents Chemother.
2004, 48, 890.
(291) Bandow, J. E.; Brotz, H.; Leichert, L. I. O.; Labischinski, H.;
Hecker, M. Antimicrob. Agents Chemother. 2003, 47, 948.
(292) Mills, S. D. J. Antimicrob. Chemother. 2003, 51, 749.
(293) Bradford, P. A.; Urban, C.; Mariano, N.; Projan, S. J.; Rahal, J.
J.; Bush, K. Antimicrob. Agents Chemother. 1997, 41, 563.
(294) Cao, V. T. B.; Arlet, G.; Ericsson, B. M.; Tammelin, A.; Courvalin,
P.; Lambert, T. J. Antimicrob. Chemother. 2000, 46, 895.
(295) Nelson, E. C.; Segal, H.; Elisha, B. G. J. Antimicrob. Chemother.
2003, 52, 899.
(296) Domenech-Sanchez, A.; Martnez-Martnez, L.; Hernandez-Alles,
S.; del Carmen Conejo, M.; Pascual, A
.; Tomas, J. M.; Albert,
S.; Bened, V. J. Antimicrob. Agents Chemother. 2003, 48, 3332.
(297) Martinez, M. B.; Flickinger, M.; Higgins, L.; Krick, T.; Nelsestuen, G. L. Biochemistry 2001, 40, 11965.
(298) Armand-Lefvre, L.; Leflon-Guibout, V.; Bredin, J.; Barguelli, F.;
Amor, A.; Pages, J. M.; Nicolas-Chanoine, M. H. Antimicrob.
Agents Chemother. 2003, 47, 1165.
(299) Kwon, D. H.; Dore, M. P.; Kim, J. J.; Kato, M.; Lee, M.; Wu, J.
Y.; Graham, D. Y. Antimicrob. Agents Chemother. 2003, 47, 2169.
(300) Fernandez-Cuencal, F.; Martnez-Martnez, L.; Conejo, M. C.;
Ayala, J. A.; Perea, E. J.; Pascual, A. J. Antimicrob. Chemother.
2003, 51, 565.
(301) Gayet, S.; Chollet, R.; Molle, G.; Page`s, J. M.; Chevalier, J.
Antimicrob. Agents Chemother. 2003, 47, 1555.
(302) Bornet, C.; Saint, N.; Fetnaci, L.; Dupont, M.; Davin-Regli, A.;
Bollet, C.; Page`s, J. M. Antimicrob. Agents Chemother. 2004, 48,
2153.
(303) Jacoby, G. A.; Mills, D. M.; Chow, N. Antimicrob. Agents
Chemother. 2004, 48, 3203.
(304) Okamoto, K.; Gotoh, N.; Nishino, T. Antimicrob. Agents Chemother. 2001, 45, 1964.
(305) Pai, H.; Kim, J. W.; Kim, J.; Lee, J.; Choe, K. W.; Gotoh, N.
Antimicrob. Agents Chemother. 2001, 45, 480.
(306) Lakaye, B.; Dubus, A.; Joris, B.; Fre`re, J. M. Antimicrob. Agents
Chemother. 2002, 46, 2901.
(307) Matsumura, N.; Minami, S.; Watanabe, Y.; Iyobe, S.; Mitsuhashi,
S. Antimicrob. Agents Chemother. 1999, 43, 2084.
(308) Kotra, L. P.; Golemi, D.; Amro, N. A.; Liu, G. Y.; Mobashery, S.
J. Am. Chem. Soc. 1999, 121, 8707.
(309) Snyder, D. S.; McIntosh, T. J. Biochemistry 2000, 39, 11777.
(310) Gillespie, S. H.; McHugh, T. D. Trends Microbiol. 1997, 5, 337.
(311) Faller, M.; Niederweis, M.; Schulz, G. E. Science 2004, 303, 1189.
(312) Antignac, A.; Boneca, I. G.; Rousselle, J. C.; Namane, A.; Carlier,
J. P.; Vazquez, J. A.; Fox, A.; Alonso, J. M.; Taha, M. K. J. Biol.
Chem. 2003, 278, 31529.
(313) Antignac, A.; Rousselle, J. C.; Namane, A.; Labigne, A.; Taha,
M. K.; Boneca, I. G. J. Biol. Chem. 2003, 278, 31521.
(314) Filipe, S. R.; Tomasz, A. Proc. Natl. Acad. Sci. U.S.A. 2000, 99,
4891.
(315) Smith, A. M.; Klugman, K. P. Antimicrob. Agents Chemother.
2001, 45, 2393.
(316) Fiser, A.; Filipe, S. R.; Tomasz, A. Trends Microbiol. 2003, 11,
547.
(317) Rohrer, S.; Berger-Bachi, B. Antimicrob. Agents Chemother.
2003, 47, 837.
(318) Okamoto, K.; Gotoh, N.; Nishino, T. Antimicrob. Agents Chemother. 2002, 46, 2696.
(319) Middlemiss, J. K.; Poole, K. J. Bacteriol. 2004, 186, 1258.
(320) Van Bambekel, F.; Glupczynski, Y.; Plesiat, P.; Peche`re, J. C.;
Tulkens, P. M. J. Antimicrob. Chemother. 2003, 51, 1055.
(321) Burns, D. L. Curr. Opin. Microbiol. 2003, 6, 29.
(322) Cascales, E.; Christie, P. J. Nat. Rev. Microbiol. 2003, 1, 137.
(323) Borges-Walmsley, M. I.; McKeegan, K. S.; Walmsley, A. R.
Biochem. J. 2003, 376, 313.
(324) Grkovic, S.; Brown, M. H.; Skurray, R. A. Microbiol. Mol. Biol.
2002, 66, 671.
(325) McKeegan, K. S.; Borges-Walmsley, M. I.; Walmsley, A. R.
Trends Microbiol. 2003, 11, 21.
(326) Murakami, S.; Yamaguchi, A. Curr. Opin. Struct. Biol. 2003,
13, 443.
(327) Paulsen, I. T. Curr. Opin. Microbiol. 2003, 6, 446.
(328) Schmacher, M. A.; Brennan, R. G. Mol. Microbiol. 2002, 45, 885.
(329) Chevalier, J.; Bredin, J.; Mahamoud, A.; Mallea, M.; Barbe, J.;
Page`s, J. M. Antimicrob. Agents Chemother. 2004, 48, 1043.
(330) Nakayama, K.; Ishida, Y.; Ohtsuka, M.; Kawato, H.; Yoshida,
K.; Yokomizo, Y.; Hosono, S.; Ohta, T.; Hoshino, K.; Ishida, H.;

Chemical Reviews, 2005, Vol. 105, No. 2 423

(331)
(332)

(333)
(334)
(335)

(336)
(337)
(338)
(339)
(340)
(341)
(342)
(343)
(344)
(345)
(346)
(347)
(348)
(349)
(350)
(351)
(352)
(353)
(354)
(355)
(356)
(357)
(358)
(359)
(360)
(361)
(362)
(363)
(364)

(365)
(366)
(367)
(368)
(369)
(370)

(371)
(372)
(373)
(374)
(375)

Renau, T. E.; Leger, R.; Zhang, J. Z.; Lee, V. J.; Watkins, W. J.


Bioorg. Med. Chem. Lett. 2003, 13, 4201.
Mallea, M.; Mahamoud, A.; Chevalier, J.; Alibert-Franco, S.;
Brouant, P.; Barbe, J.; Pages, J. M. Biochem. J. 2003, 376, 801.
Renau, T.; Leger, R.; Filonova, L.; Flamme, E.; Wang, M.; Yen,
R.; Madsen, D.; Griffith, D.; Chamberland, S.; Dudley, M.; Lee,
V.; Lomovskaya, O.; Watkins, W.; Ohta, T.; Nakayama, K.;
Ishida, Y. Bioorg. Med. Chem. Lett. 2003, 13, 2755.
Zgurskaya, H.; Nikaido, H. Mol. Microbiol. 2000, 37, 219.
McGeer, A.; Low, D. E. Nat. Med. 2003, 9, 390.
McCormick, A. W.; Whitney, C. G.; Farley, M. M.; Lynfield, R.;
Harrison, L. H.; Bennett, N. M.; Schaffner, W.; Reingold, A.;
Hadler, J.; Cieslak, P.; Samore, M. H.; Lipsitch, M. Nat. Med.
2003, 9, 424.
Service, R. F. Science 2004, 303, 1798.
Sheldon, T. Nat. Med. 2004, 10, 6.
Oliveira, D. C.; Tomasz, A.; de Lencastre, H. Lancet Infect. Dis.
2002, 2, 180.
Projan, S. J. Curr. Opin. Microbiol. 2003, 6, 427.
Projan, S. J. Curr. Opin. Pharmacol. 2003, 3, 457.
Shlaes, D. M. Curr. Opin. Pharmacol. 2003, 3, 470.
Projan, S. J. Curr. Opin. Pharmacol. 2002, 2, 513.
Livermore, D. Nat. Rev. Microbiol. 2004, 2, 73.
Eisner, T. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 14517.
Meinwald, J.; Eisner, T. Helv. Chim. Acta 2003, 86, 3633.
Coates, A.; Hu, Y.; Bax, R.; Page, C. Nat. Rev. Drug Discovery
2002, 1, 895.
Hutter, B.; Schaab, C.; Albrecht, S.; Borgmann, M.; Brunner,
N. A.; Freiberg, C.; Ziegelbauer, K.; Rock, C. O.; Ivanov, I.;
Loferer, H. Antimicrob. Agents Chemother. 2004, 48, 2838.
Katayama, Y.; Zhang, H. Z.; Hong, D.; Chambers, H. F. J.
Bacteriol. 2003, 185, 5465.
Weiss, D. S.; Chen, J. C.; Ghigo, J. M.; Boyd, D.; Beckwith, J. J.
Bacteriol. 1999, 181, 508.
Zhao, G.; Meier, T. I.; Kahl, S. D.; Gee, K. R.; Blaszczak, L. C.
Antimicrob. Agents Chemother. 1999, 43, 1124.
Kyriacou, S. V.; Brownlow, W. J.; Xu, X. H. N. Biochemistry 2004,
43, 140.
Steel, C.; Wan, Q.; Xu, X. H. N. Biochemistry 2004, 43, 175.
Campbell, D. A.; Szardenings, A. K. Curr. Opin. Chem. Biol.
2003, 7, 296.
Miller, C.; Thomsen, L. E.; Gaggero, C.; Mosseri, R.; Ingmer,
H.; Cohen, S. N. Science 2004, 305, 1629.
Spoering, A. L.; Lewis, K. J. Bacteriol. 2001, 183, 6746.
Balaban, N. Q.; Merrin, J.; Chait, R.; Kowalik, L.; Leibler, S.
Science 2004, 305, 1622.
Hastings, P. J.; Rosenberg, S. M.; Slack, A. Trends Microbiol.
2004, 12, 401.
Rice, L. B. Curr. Opin. Pharmacol. 2003, 3, 459.
Rand, K. H.; Houck, H. J. Antimicrob. Agents Chemother. 2004,
48, 2871.
Barrett, C. T.; Barrett, J. F. Curr. Opin. Biotechnol. 2003, 14,
621.
Gustafsson, I.; Sjolund, M.; Torell, E.; Johannesson, M.; Engstrandl, L.; Cars, O.; Andersson, D. I. J. Antimicrob. Chemother.
2003, 52, 645.
Salyers, A. A.; Gupta, A.; Wang, Y. Trends Microbiol. 2004, 12,
412.
Kummerer, K.; Henniger, A. Clin. Microbiol. Infect. 2003, 9,
1203.
Harmoinen, J.; Mentula, S.; Heikkila, M.; van der Rest, M.;
Rajala-Schultz, P. J.; Donskey, C. J.; Frias, R.; Koski, P.;
Wickstrand, N.; Jousimies-Somer, H.; Westermarck, E.; Lindevall, K. Antimicrob. Agents Chemother. 2004, 48, 75.
Harmoinen, J.; Vaali, K.; Koski, P.; Syrjanen, K.; Laitinen, O.;
Lindevall, K.; Westermarck, E. J. Antimicrob. Chemother. 2003,
51, 361.
Stiefel, U.; Pultz, N. J.; Harmoinen, J.; Koski, P.; Lindevall, K.;
Helfand, M. S.; Donskey, C. J. J. Infect. Dis. 2003, 188, 1605.
Cooper, B. S.; Medley, G. F.; Stone, S. P.; Kibbler, C. C.; Cookson,
B. D.; Roberts, J. A.; Duckworth, G.; Lai, R.; Ebrahim, S. Proc.
Natl. Acad. Sci. U.S.A. 2004, 101, 10223.
Livermore, D. M.; Dudley, M. N. Curr. Opin. Microbiol. 2000,
3, 487.
Monroe, S.; Polk, R. Curr. Opin. Microbiol. 2000, 3, 496.
Bantar, C.; Vesco, E.; Heft, C.; Salamone, F.; Krayeski, M.;
Gomez, H.; Coassolo, M. A.; Fiorillo, A.; Franco, D.; Arango, C.;
Duret, F.; Oliva, M. E. Antimicrob. Agents Chemother. 2004, 48,
392.
Livemore, D. M.; Brown, D. F. J.; Quinn, J. P.; Carmeli, Y.;
Paterson, D. L.; Yu, V. L. Clin. Microbiol. Infect. 2004, 10, 84.
Chandrakala, B.; Shandil, R. K.; Mehra, U.; Ravishankar, S.;
Kaur, P.; Usha, V.; Joe, B.; deSousa, S. M. Antimicrob. Agents
Chemother. 2004, 48, 30.
Tondi, D.; Powers, R. A.; Caselli, E.; Negri, M. C.; Blazquez, J.;
Costi, M. P.; Shoichet, B. K. Chem. Biol. 2001, 8, 593.
Powers, R. A.; Shoichet, B. K. J. Med. Chem. 2002, 45, 3222.
Powers, R. A.; Morandi, F.; Shoichet, B. K. Structure 2002, 10,
1013.

424 Chemical Reviews, 2005, Vol. 105, No. 2


(376) Trehan, I.; Morandi, F.; Blaszczak, L. C.; Shoichet, B. K. Chem.
Biol. 2002, 9, 971.
(377) Levy, S. B.; Marshall, B. Nat. Med. Suppl. 2004, 10, S122.
(378) Payne, D.; Tomasz, A. Curr. Opin. Microbiol. 2004, 7, 435.
(379) Poole, K. Cell. Mol. Life Sci. 2004, 61, 2200.
(380) Mallorqui-Fernandez, G.; Marrero, A.; Garcia-Pique, S.; GarciaCastellanos, R.; Gomis-Ruth, F. X. FEMS Microbiol. Lett. 2004,
235, 1.
(381) Walsh, F. M.; Amyes, S. G. B. Curr. Opin. Microbiol. 2004, 7,
439.
(382) Gotz, F. Curr. Opin. Microbiol. 2004, 7, 477.
(383) Dmitriev, B. A.; Toukach, F. V.; Holst, O.; Rietschel, E. T.; Ehlers,
S. J. Bacteriol. 2004, 186, 7141.
(384) Vollmer, W.; Holtje, J. V. J. Bacteriol. 2004, 186, 5978.
(385) Arbeloa, A.; Hugonnet, J.-E.; Sentilhes, A.-C.; Josseaume, N.;
Dubost, L.; Monsempes, C.; Blanot, D.; Brouard, J.-P.; Arthur,
M. J. Biol. Chem. 2004, 279, 41546.
(386) Gardete, S.; Ludovice, A. M.; Sobral, R. G.; Filipe, S. R.; De
Lencastre, H.; Tomasz, A. J. Bacteriol. 2004, 186, 1705.
(387) Pagliero, E.; Chesnel, L.; Hopkins, J.; Croize, J.; Dideberg, O.;
Vernet, T.; Di Guilmi, A. M. Antimicrob. Agents Chemother.
2004, 48, 1848.
(388) Silvaggi, N. R.; Josephine, H. R.; Kuzin, A. P.; Nagarajan, R.;
Pratt, R. F.; Kelly, J. A. J. Mol. Biol. 2005, 345, 251.
(389) Macheboeuf, P.; Di Guilmi, A. M.; Job, V.; Vernet, T.; Dideberg,
O.; Dessen, A. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 577.
(390) Morlot, C.; Pernot, L.; Le Gouellec, A.; Di Guilmi, A. M.; Vernet,
T.; Dideberg, O.; Dessen, A. J. Biol. Chem. 2005, in press.
(391) Labia, R. Curr. Med. Chem. Anti-Infect. Agents 2004, 3, 251.
(392) Ke, Y.-Y.; Lin, T.-H. Biophys. Chem. 2005, 114, 103.
(393) Shimizu-Ibuka, A.; Matsuzawa, H.; Sakai, H. Biochemistry 2004,
43, 15737.
(394) Materon, I. C.; Beharry, Z.; Huang, W.; Perez, C.; Palzkill, T. J.
Mol. Biol. 2004, 344, 653.

Fisher et al.
(395) Mercuri, P. S.; Garcia-Saez, I.; De Vriendt, K.; Thamm, I.;
Devreese, B.; Van Beeumen, J.; Dideberg, O.; Rossolini, G. M.;
Frere, J.-M.; Galleni, M. J. Biol. Chem. 2004, 279, 33630.
(396) Garrity, J. D.; Bennett, B.; Crowder, M. W. Biochemistry 2005,
44, 1078.
(397) Garau, G.; Bebrone, C.; Anne, C.; Galleni, M.; Frere, J.-M.;
Dideberg, O. J. Mol. Biol. 2005, 345, 785.
(398) Zhang, Z.; Palzkill, T. J. Biol. Chem. 2004, 279, 42860.
(399) Reichmann, D.; Rahat, O.; Albeck, S.; Meged, R.; Dym, O.;
Schreiber, G. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 57.
(400) Freiberg, C.; Brotz-Oesterhelt, H.; Labischinski, H. Curr. Opin.
Microbiol. 2004, 7, 451.
(401) Venkatesan, A. M.; Agarwal, A.; Abe, T.; Ushirogochi, H.;
Yamamura, I.; Kumagai, T.; Petersen, P. J.; Weiss, W. J.; Lenoy,
E.; Yang, Y.; Schlaes, D. M.; Ryan, J. L.; Mansour, T. S. Biorg.
Med. Chem. 2004, 12, 5807.
(402) Venkatesan, A. M.; Gu, Y.; Dos Santos, O.; Abe, T.; Agarwal,
A.; Yang, Y.; Petersen, P. J.; Weiss, W. J.; Mansour, T. S.;
Nukaga, M.; Hujer, A. M.; Bonomo, R. A.; Knox, J. R. J. Med.
Chem. 2004, 47, 6556.
(403) Bush, K.; Macielag, M.; Weidner-Wells, M. Curr. Opin. Microbiol.
2004, 7, 466.
(404) Buynak, J. D. Curr. Med. Chem. 2004, 11, 1951.
(405) Georgopapadakou, N., H. Exp. Opin. Invest. Drugs 2004, 13,
1307.
(406) Long, T. E. IDrugs 2003, 6, 351.
(407) Singh, G. S. Mini-Rev. Med. Chem. 2004, 4, 69.
(408) Singh, G. S. Mini-Rev. Med. Chem. 2004, 4, 93.
(409) Thomson, C. J.; Power, E.; Ruebsamen-Waigmann, H.; Labischinski, H. Curr. Opin. Microbiol. 2004, 7, 445.

CR030102I

Chem. Rev. 2005, 105, 425448

425

Glycopeptide and Lipoglycopeptide Antibiotics


Dan Kahne,*, Catherine Leimkuhler, Wei Lu, and Christopher Walsh*,
The Department of Chemistry, Princeton University, Princeton, New Jersey 08544, and the Department of Biological Chemistry and
Molecular Pharmacology, Harvard Medical School, Boston, Massachusetts 02115
Received May 12, 2004

Contents

1. Introduction

1. Introduction
2. Structures of Vancomycin and Teicoplanin
2.1. Variations in Vancomycin and Teicoplanin
Natural Analogues
3. Biosynthetic StrategiessEnzymatic Assembly
Lines and Tailoring Enzymes
4. Mechanism of Action of Glycopeptide Antibiotics
4.1. Cellular Targets of Antibiotics
4.2. Three Stages of Peptidoglycan Biosynthesis
4.3. Transglycosylase and Transpeptidase
Isoforms
4.4. Peptidoglycan Biosynthesis as an Antibiotic
Target
4.5. Early Studies on the Mechanism of Action of
Glycopeptide Antibiotics
5. Mechanisms of Resistance
5.1. Vancomycin Resistant Enterococci (VRE)
5.2. VRE Genotypes: VanB Resistance and How
To Overcome It
5.3. VRE Genotypes: VanA Resistance and the
Compounds That Overcome It
5.4. Models for How Vancomycin Analogues
Overcome Vancomycin Resistance
5.4.1. Dimerization and Membrane Anchoring
5.4.2. A Second Mechanism of ActionsDirect
Interaction with the Transglycosylase
5.5. Vancomycin-Resistant S. aureus (VRSA)
6. New Directions for Treating VRE and VRSA
6.1. Recently Approved Drugs for VRE
6.2. Second-Generation Semisynthetic
Lipoglycopeptides in Clinical Development
6.2.1. Oritavancin
6.2.2. Dalbavancin
6.2.3. TD-6424
6.3. In Discovery: Chemoenzymatic Routes To
Modify Sugars and Acyl Groups on
Heptapeptide Scaffolds
7. Acknowledgments
8. Note Added after ASAP Publication
9. References

425
426
428
428
431
431
432
433
435
436
437
437
437
439
441
441
441
442
442
443
443
443
444
444
444
446
446
446

* To whom correspondence should be addressed. C.W.: phone,


617-432-1715; fax, 617-432-0438; e-mail, Christopher_walsh@
hms.harvard.edu. D.K.: phone, 617-496-0208; fax, 617-496-0215;
e-mail, kahne@chemistry.harvard.edu.
Current address: The Department of Chemistry and Chemical
Biology, Harvard University, Cambridge, MA 02138.
Harvard Medical School.

Vancomycin and teicoplanin are the two glycopeptide antibiotics that are used clinically, and the
emergence of resistance to them poses a serious
threat to human health. Some microorganisms are
resistant to both vancomycin and teicoplanin, but
some resistant strains remain sensitive to teicoplanin when resistance to vancomycin develops.
Given the apparent structural and mechanistic similarities of these two drugs, one longstanding question is why they have different effects on some
microorganisms. This review will summarize what
is known about the structure and function of the
glycopeptide and lipoglycopeptide antibiotics, how
glycopeptide resistance develops, and how natural
and semisynthetic lipoglycopeptide derivatives are
being used to learn how to overcome glycopeptide
resistance and for development as novel therapeutic
agents.
Vancomycin and teicoplanin are used to treat
serious Gram-positive bacterial infections that are
resistant to other antibiotics, e.g. -lactams. The
frequency of resistance to the glycopeptide antibiotics
has increased significantly over the past decade and
now represents a serious threat to public health.1
Moreover, multiple genera, including Staphylococcus,
have developed resistance to these drugs.2 It is hoped
that research into the molecular basis for glycopeptide resistance will lead to the ability to design new
glycopeptide antibiotics with activity against both
sensitive and resistant bacterial strains. In this
review we will describe the structures of a set of
important natural glycopeptide antibiotics and outline their biosynthetic pathways. We will then discuss
what is known about the mode of action of the
glycopeptides and how structural differences influence biological activity. Next, we will analyze how
resistance to the glycopeptides develops and summarize the efforts to develop glycopeptides that can
overcome resistance. Finally, we will discuss how
access to unnatural glycopeptides has provided tools
that have been used to identify new targets of the
glycopeptide class and activity differences observed
for natural glycopeptides against different bacterial
pathogens.
Antibiotics can be classified on several axes. One
is by the nature of the targets in susceptible bacteria,
for example blockade of bacterial cell wall biosynthesis or bacterial protein synthesis, or DNA and
RNA replication.3 A second axis is whether the
antibiotics derive from natural product scaffolds or

10.1021/cr030103a CCC: $53.50 2005 American Chemical Society


Published on Web 01/22/2005

426 Chemical Reviews, 2005, Vol. 105, No. 2

Daniel Kahne recently moved to Harvard University from Princeton University, where he was on the faculty for 16 years. He holds appointments
in the departments of Chemistry and Chemical Biology (CCB) and
Biological Chemistry and Molecular Pharmacology (BCMP). He trained
as a synthetic organic chemist with Gilbert Stork and continued postdoctoral
training at Columbia with Clark Still. He has longstanding interests in the
chemistry and biology of natural products, and in recent years become
interested in how natural products can be used to probe cellular pathways.
Professor Kahnes research group is divided into students who develop
synthetic methods to make and modify complex natural products and
students who combine some chemistry with molecular and cellular biology
to address questions relating to how various natural products function. In
the last five years, the Kahne group has become interested in antibiotic
resistance and has made significant contributions to understanding the
mechanisms of action of glycopeptide antibiotics and derivatives that
overcome glycopeptide resistance. The Kahne group has also been using
glycopeptide derivatives and other antibiotics in conjunction with genetics
to probe pathways involving cell wall biosynthesis and outer membrane
biogenesis.

Catherine E. Leimkuhler attended Villanova University, where she received


her B.S. in Chemistry in 2001. Upon graduation, she attended Princeton
University, where she received her M.A. under the direction of Daniel
Kahne. She has worked on a chemoenyzmatic synthesis of novel
glycopeptide antibiotics. She is currently pursuing her Ph.D. with Daniel
Kahne at Harvard University.

are synthetic antibacterial drugs.4 The glycopeptide


antibiotics of the vancomycin (1) and teicoplanin (2)
(Figure 1) class block steps in the biosynthesis of the
peptidoglycan layer of bacterial cell walls. They are
natural products elaborated by actinomycete soil
bacteria, with vancomycin being isolated in the 1950s
from Amycolatopsis orientalis and teicoplanin around
1980 from Actinoplanes teichomyceticus.5 Each of
these antibiotics is effective against Gram-positive
bacteria but not Gram-negative ones, due to the
permeability barrier of the intact outer membrane

Kahne et al.

Wei Lu received his B.E. degree from East China University of Chemical
Technology in 1992. He did his Ph.D. in Philip Coles lab at Johns Hopkins
University. His Ph.D. research focused on the regulation of protein tyrosine
phosphatases. Since 2002, he has been pursuing his postdoctoral research
in Christopher Walshs group at Harvard Medical School. His current
research interest lies in functional analysis of glycosyltransferases involved
in natural product biosynthesis. He is currently a Ruth L. Kirschstein Nation
Institute of Health postdoctoral fellow.

Christopher T. Walsh is the Hamilton Kuhn Professor of Biological


Chemistry and Molecular Pharmacology (BCMP) at Harvard Medical
School. He has had extensive experience in academic administration,
including Chairmanship of the MIT Chemistry Dept (19821987) and the
HMS Biological Chemistry & Molecular Pharmacology Department (1987
1995) as well as serving as President and CEO of the Dana Farber Cancer
Institute (19921995). His research has focused on enzymes and enzyme
inhibitors, with recent specialization on antibiotics. He and his group have
authored over 590 research papers; a text, Enzymatic Reaction Mechanisms; and a book, Antibiotics: Origins, Actions, Resistance. He is a
member of the National Academy of Sciences, the Institute of Medicine,
and the American Philosophical Society.

in Gram-negative bacteria that keeps the glycopeptides from reaching their targets at the periplasmic
face of the cytoplasmic membrane.
Vancomycin has been approved for human use in
many countries, while teicoplanin, widely prescribed
in Europe, was never been brought successfully
through the FDA approval process in the US. The
drugs have been front line agents for treating endocarditis caused by enterococci, opportunistic pathogens, and have become mainstays in the treatment
of life-threatening infections due to methicillin resistant Staphylococcus aureus (MRSA).1

2. Structures of Vancomycin and Teicoplanin


In this review, the term lipopeptide is used for
acylated peptide natural products (such as daptomy-

Glycopeptide and Lipoglycopeptide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 427

Figure 1. Structures of vancomycin (1) and teicoplanin (2), glycopeptide and lipoglycopeptide antibiotics approved for
human therapy.

Figure 2. Oxidative cross-linking by hemeprotein enzymes of the nascent heptapeptides to yield the oxidized, crosslinked heptapeptide scaffolds: three cross-links at residues 2-4, 4-6, and 5-7 in the vancomycin scaffold and a fourth
cross-link at residues 1-3 in the teicoplanin scaffold.

cin); the term glycopeptide is used for molecules such


as vancomycin. Since teicoplanin has a peptide scaffold and both covalently attached glycosyl and long
chain acyl groups, the term lipoglycopoeptide is used
for such natural products.
Vancomycin and teicoplanin are closely related
structures, containing a homologous heptapeptide
scaffold that has undergone extensive oxidative crosslinking.6 Five of the seven residues in vancomycin
are aromatic, while all seven are in teicoplanin: Leu1
and Asn3 of vancomycin are replaced by hydroxyphenylglycine regioisomers in teicoplanin. There are five
such nonproteinogenic phenylglycines in teicoplanin
and three in vancomycin. The phenylglycines are
either 4-hydroxyphenylglycine (HPG) (residues 4 and
5 in 1 and 2 and also residue 1 in teicoplanin) or 3,5dihydroxyphenylglycine (DPG) (residue 7 in 1 and
residues 3 and 7 in 2). In vancomycin, the remaining
two residues are modified tyrosines, with chlorine at
the meta position of the aromatic ring and an OH
substituent at the benzylic carbon of the side chain.
In teicoplanin, only Tyr6 has been converted to the
-OH-Tyr.
The electron-rich side chains of these aromatic
amino acid residues facilitate the oxidative crosslinking (see below) that sets the rigid architecture of
the heptapeptide scaffolds of the two drugs. In

vancomycin, three cross-links are effected, joining the


aromatic rings of 2-4 and 4-6 in aryl ether linkages
and of 5-7 in a direct C-C coupling (Figure 2). In
teicoplanin, residues 1 and 3 are now hydroxyphenylglycines in the un-cross-linked precursor heptapeptide, and a fourth cross link, 1-3, is produced by
the teicoplanin tailoring enzymes (Figure 2). The four
cross-links of the teicoplanin scaffold involve all seven
side chains to produce the rigid, cup-shaped aglycon.
These are glycopeptide antibiotics, so subsequently,
the peptide framework is glycosylated, by a disaccharide chain on residue 4 in vancomycin or by three
monosaccharides at residues 4, 6, and 7 in teicoplanin. The disaccharide in vancomycin is a D-glucosyl2,1-D-vancosamine, linked through C1 of glucose to
the phenolic oxygen of OH-Phegly4 of the crosslinked peptide. The vancosamine sugar is a 2,3,6trideoxy-L-hexose, with methyl and amino substituents at C3. Such deoxyhexoses are typical in antibiotics
and mediate some of the recognition of target
molecules.7-11 In teicoplanin, there is a glucosamine
at the corresponding OH-Phegly4 residue, a GlcNAc
at the -hydroxyl substituent of -OH-Tyr6, and a
mannose on the cross-linked Dpg7. The N-decanoyl
moiety on the glucosamine installed at residue 4
blocks the elongation to the disaccharide chain
observed in the vancomycin family.

428 Chemical Reviews, 2005, Vol. 105, No. 2

Kahne et al.

Teicoplanin, but not vancomycin, is a lipoglycopeptide with a C10 fatty acyl chain in an amide linkage
to the amino group of the glucosamine sugar moiety
(a series of fatty acyl variants are found in the
producer organism, but control of the feedstock
during fermentation yields the C10 acylated form of
teicoplanin). The hydrophobic acyl chain alters the
physical properties and presumably the partitioning
of lipoglycopeptide vs glycopeptide. This is likely a
determining factor not only in the differential activity
of 2 vs 1 against some forms of vancomycin-resistant
enterococci (VRE) noted below in section 5 but also
in the second-generation semisynthetic lipoglycopeptides oritavancin, dalbavancin, and TD-6424 discussed in section 6.

additional epivancosamine residue is found at the


benzylic oxygen of -OH-Tyr6. The biosynthetic gene
clusters for 3 and 412-14 have been reported and have
revealed much about the molecular logic for assembly
of these antibiotics.
In the teicoplanin subfamily, we note two congeners, A47934 (5), a sulfated aglycon form produced
by Streptomyces toyocaensis, and A40926 (6), from a
Nonomuraea species. In section 6, we shall note that
chloroeremomycin is the starting point for oritavancin and A40926 the starting point for dalbavancin,
the two most clinically advanced second-generation,
semisynthetic lipoglycopeptides. The DNA sequence
for the biosynthetic clusters of 5 and 6 have also been
reported.15-17

2.1. Variations in Vancomycin and Teicoplanin


Natural Analogues

3. Biosynthetic StrategiessEnzymatic Assembly


Lines and Tailoring Enzymes

After the clinical success of vancomycin and teicoplanin, dozens of additional glycopeptide congeners
have been isolated from strains of actinomycetes.6
Congeners with the vancomycin heptapeptide scaffold
include balhimycin (3) and chloroeremomycin (4)
(Figure 3, structures 3-6). Both have the identical
heptapeptide scaffold of vancomycin. Balhimycin
differs in having a vancosamine derivative, 4-oxovancosamine, at residue 6. Chloroeremomycin has
almost the same disaccharide appended to OHPhegly4. The difference is in the terminal 2,3,6trideoxy-3-amino-3-methyl hexose. In 4, the trideoxyhexose-4-OH is equatorial rather than the 4-axial
OH of vancosamine. This is epivancosamine, and an

Although the total synthesis of the vancomycin and


teicoplanin family of glycopeptides and lipoglycopeptides have been accomplished 6,18,19 with substantial
invention of new chemistry, the complexity of these
natural products makes fermentative routes the only
viable route to bulk production. Semisynthetic chemical tailoring of the scaffolds has been the predominant route to structure activity variation to produce
second-generation clinical development candidates,
as will be noted in section 6.20
Thus, understanding the biosynthetic logic for
construction of the heptapeptide core and its subsequent oxidations, glycosylations, and acylations is of
both basic and applied interest. It had been clear

Figure 3. Balhimycin (3) and chloroeremomycin (4) in the vancomycin family and A47934 (5) and A40926 (6) in the
teicoplanin family.

Glycopeptide and Lipoglycopeptide Antibiotics

Figure 4. Core domains: C-A-T, in the elongation modules


of the NRPS assembly line for vancomcyin and teicoplanin
family members: C ) condensation domain, A ) adenylation domain, and T ) thiolation domain. Amino acids get
selected and activated by the A domain and installed as
thioester on the T domain. The C domain catalyzes peptide
bond formation.

from the prevalence of the nonproteinogenic phenylglycines in the peptide backbones that vancomycin and
teicoplanin family members must be assembled by
nonribosomal peptide synthetase (NRPS) enzymatic
machinery.5 The DNA sequence for the chloroeremomycin gene cluster12 revealed 30 adjacent genes
attributable to the pathway and validated the existence of seven NRPS modules, distributed over three
subunits, Cep ABC, in a 3/3/1 distribution.21 Each of
the seven modules selects and activates one amino
acid, and the order of the modules mandates the
heptapeptide sequence (Leu-Tyr-Asn-Hpg-Hpg-TyrDpg). In the teicoplanin subfamily, the catalytic
domains in modules one and three of the NRPS
assembly lines instead select and activate Hpg and
Dpg, respectively.16,22 In addition to a catalytic domain for amino acid selection and activation as the
aminoacyl-AMP, each module has a thiolation domain modified with phosphopantetheine23 to provide
a thiol for covalent aminoacyl-S-enzyme formation.
The third domain in each module is the condensation
domain, responsible for amide bond formation and
directional peptide chain growth and translocation
from N-terminal to C-terminal modules (Figure 4).
The source of the nonproteinogenic amino acids in
the microbial cell at the time antibiotic production
is switched on is relevant for yields and coordination.
Among the 30 clustered Cep biosynthetic genes are
four that encode enzymes that generate L-Hpg from
the common bacterial metabolic intermediate chorismate.21,24 There are another four encoded enzymes
in the cluster that channel four molecules of malonyl
CoA to the eight-carbon intermediate dihydroxyphenylacetyl CoA on the way to L-Dpg.25-27 Residues 2
and 6 in 1 and 4 differ from the proteinogenic amino
acid L-tyrosine by chlorine at the meta position of the
aromatic ring and the -OH that is the site of
glycosylation in 2-4. The Cep cluster contains a gene
encoding a flavoprotein halogenase12 as well as a pair
of enzymes28 that hydroxylate tyrosine while installed
as a tyrosyl-S-enzyme. Thus, 11 enzymes are coordinately induced to enable generation of residues 2,
4, 5, 6, and 7 required for the vancomycin-type NRPS
assembly lines and for generation of six of the seven

Chemical Reviews, 2005, Vol. 105, No. 2 429

residues in teicoplanin NRPS multimodular enzymes.


One additional feature of these NRPS assembly
lines is worth note. Although the biosynthetic pathways generate the L-forms of modified tyrosines,
L-Hpg, and L-Dpg, the heptapeptide released by the
NRPS assembly line has the configuration D-D-L-LD-D-L. Inspection of the CepABC subunit domain
composition predicts epimerase (E) domains in module 2, 4, and 5 to convert L-aminoacyl-/peptidyl-Senzyme intermediates to the D-isomers. The first
module lacks an epimerase domain but can select and
activate D-Leu directly.29 All told there are 24 domains in the seven-module CepABC assembly line,
each with a well-defined function. Note that condensation domains downstream of E domains must be
chiral peptide synthase catalysts that recognize
upstream D-aminoacyl-/peptidyl donors. Depending
on the timing of the epimerization, the downstream
acceptors can be L- or D-amino acids.26,30
Following release from the last module of CepC in
the NRPS assembly line, the nascent heptapeptide
7 undergoes three kinds of postassembly tailoring
reactions to yield the glycopeptides 1 and 4 (Figure
5). The first reaction is N-methylation of the amino
terminal D-Leu1 residue carried out by orf 16 in the
Cep biosynthetic cluster.31
Then the oxidative cross-links are introduced. In
both the Cep and balhimycin gene clusters there are
three tandem heme proteins, termed OxyABC,32,33
that act ad seriatim to generate the fully cross-linked
aglycon 8. Genetic analysis in the balhimycin system
has defined the regiospecificity and timing of crosslinking, one bond each introduced by catalytic action
of the hemeproteins33 as determined by gene knockouts and structure determination of the accumulating
intermediates (OxyB > OxyA > OxyC) (Figure 6).
The coupling chemistry is probably via one-electron
pathways and regioselectivity and atropisomer selectivity may be imposed by the folding and orientation of acyclic and partially cross-linked substrates
in each hemeprotein active site. Although there are
X-ray structures of two of the three bahlimycin Oxy
proteins, OxyB and OxyC,34,35 in vitro activity of the
purified enzymes has not yet been reconstituted so
the possible promiscuity toward other substrates is
not yet known.
In teicoplanin heptapeptide scaffolds, a fourth
cross-link, connecting Hpg1 and Dpg3, needs to be
introduced. In the A47934, A40926, and teicoplanin
clusters,15,16,22 there is indeed a fourth tandem hemeprotein (OxyD) that is the obvious catalytic candidate. The timing of the 1-3 cross-link compared to
2-4, 4-6, and 5-7 is not yet known.
The last phase of tailoring involves the glycosyl
transferases (Gtfs) (Figure 7). There are three Gtfs,
GtfABC, embedded in the Cep cluster12 and corresponding two Gtfs, GtfDE, in the vancomycin biosynthetic gene cluster.36 GtfB ()GtfE) transfers
D-glucose from the nucleoside diphosphosugar dTDPD-glucose to the phenolic-OH at PheGly4 in the crosslinked aglycon peptide. This yields the monoglycosylated heptapeptide scaffold, 9, known historically as
desvancosaminyl vancomycin (DVV). The other two
Gtfs in the Cep cluster, GtfA and GtfC, transfer

430 Chemical Reviews, 2005, Vol. 105, No. 2

Kahne et al.

Figure 5. Post-NRPS assembly line tailoring enzymatic modifications to convert the heptapeptides to vancomycin and
teicoplanin: N-methylation, oxidative cross-linking, and glycosylations.

Figure 6. Sequential action of OxyB, OxyA, OxyC to introduce the 2-4, 4-6, and 5-7 cross-links in the bahlimycin
scaffold.
L-epivancosamine to the -OH-Tyr6 and the C2 of the
glucosyl moiety of DVV, respectively37,38 accounting
for seven post-assembly line tailoring enzymes in the
Cep cluster.
The donor substrate for GtfA and GtfC is dTDPL--epivancosamine 10. It is not a standard primary
metabolite and is also made with just-in-time inventory control logic like the nonproteinogenic amino

acids. In particular, there are five genes, EvaA-E, in


the Cep cluster that channel the common NDP sugar
dTDP-D-glucose to dTDP-L-epivancosamine.39 These
enzymes effect remarkable changes of the hexose
attached to dTDP, deoxygenating C6 and then C2 and
reductively aminating a C3 keto intermediate and
then C-methylating it. This yields the dTDP-4oxovancosamine, the sugar donor to residue 6 in

Glycopeptide and Lipoglycopeptide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 431

Figure 7. Action of three glycosyltransferases, GtfABC, to add the three sugars in chloroeremoycin maturation and of
two Gtfs, GtfDE, in vancomycin maturation.

bahlimycin tailoring. The C4 ketone in dTDP-4oxovancosamine can be reduced with chirality control
to yield the 4-equatorial OH to finish the biosynthesis
of dTDP-L-epivancosamine for GtfC.39 Alternatively,

stereospecific enzymatic reduction to the other carbonyl face yields the 4-axial-OH and dTDP-L-vancosamine in vancomycin-producing actinomycetes,
presumably used by GtfD to build the D-glucosyl-2,1l-vancosamine disaccharide chain as the last step in
formation of 1.
Including the EvaA-E enzymes in the list of dedicated monomer generation and tailoring enzymes
yields a total of 24 proteins, over and above the three
NRPS assembly line subunits needed to make chloroeremomycin (one Gtf less to make vancomycin)
(Figure 8). The biosynthetic gene clusters for A40926
and for teicoplanin reveal the anticipated comparable
logic with a few variations, including a putative
mannosyl transferase for glycosylation of Dpg7 and
an acyl transferase that is the catalyst for the C10
fatty acyl amide formation to the glucosamine moiety
in 2 and 6.17

4. Mechanism of Action of Glycopeptide


Antibiotics
4.1. Cellular Targets of Antibiotics

Figure 8. Twenty-four open reading frames (Orfs) in the


chloroeremomycin biosynthetic cluster.

The majority of antibacterial agents inhibit the


synthesis of DNA, RNA, proteins, or peptidoglycan.
These are all macromolecules containing different
monomer building blocks. Therefore, it is possible to

432 Chemical Reviews, 2005, Vol. 105, No. 2

Kahne et al.

Figure 9. The peptidoglycan layer (PG) surrounding bacterial cells is a giant macromolecular meshwork with peptide
cross-bridges connecting glycan strands.

ascertain generally how an antibiotic functions by


examining whether it affects the incorporation of
radiolabeled monomer units into one of these four
different types of macromolecules. Vancomycin, the
first glycopeptide antibiotic discovered, was shown
to inhibit the incoporation of [14C]GlcNAc from UDP[14C]GlcNAc into bacterial peptidoglycan and was
therefore classified as a cell wall active antibiotic.40,41
An overview of the structure and biosynthesis of
bacterial peptidoglycan is provided in the following
section as a prelude to a detailed discussion of the
mechanism of action of vancomycin and other glycopeptide antibiotics.

4.2. Three Stages of Peptidoglycan Biosynthesis


The internal osmotic pressure inside a typical
bacterial cell fluctuates significantly, depending on
environmental conditions. For bacteria to survive,
their cell membranes must be able to withstand
osmotic pressures in excess of 5-15 atm without
rupturing. Bacterial cells are surrounded by layers
of peptidoglycan, a mesh-like carbohydrate polymer
that provides the mechanical support necessary to
prevent the cells from lysing as the osmotic pressure
fluctuates.42 Peptidoglycan is composed of linear
chains of -(1,4)-N-acetyl hexosamine units joined by
peptide cross-links (Figure 9).
The biosynthesis of peptidoglycan takes place in
three distinct stages (for a review on peptidoglycan
biosynthesis, see Bugg et al.43) (Figure 10). The first
stage takes place in the cytoplasm and involves the
conversion of UDP-GlcNAc to UDP-N-acetylmuramylpentapepide.4 The muramyl group is a D-2-N-acetyl3-O-lactylglucose, assembled from UDP-GlcNAc and
PEP by MurA and MurB catalysis. Mur C, D, E
sequentially add L-Ala-D--Gln and Lys (Gram positives) or meso diaminopimelate (DAP in Gram negatives), respectively, in ATP-dependent amide-forming

steps to create the UDP muramyl-tripeptide. MurF


adds preformed D-Ala-D-Ala in the fourth amideforming step to create the UDP-muramyl-L-Ala--DGln-L-Lys-D-Ala-D-Ala (pentapeptide) and complete
stage I.
Stage II involves the transfer of the muramylpentapeptide from UDP to a C55 isoprenol-P (bactoprenol) carrier. The first reaction (MraY) moves the
muramyl pentapeptide to the membrane interface
and creates lipid I. The second reaction, catalyzed by
MurG, adds GlcNAc in a -1,4-linkage to the muramyl moiety, generating the disaccharyl-pentapeptide (Figure 10) attached to bactoprenol-PP.44-47 This
is lipid II, which is the donor for the peptidoglycan
elongation reactions that occur on the external surface of the bacterial membrane. Once lipid II is
formed, the disaccharide-pentapeptide is somehow
flipped from the cytoplasmic phase of the membrane
to the external face (Figure 11). That flipping, still
mysterious in whether an enzyme accelerates the
facial equilibration, completes phase II of PG assembly.
The third stage of PG synthesis involves lipid II
molecules presenting disaccharyl-pentapeptide as
donor, on the outer face of membranes, to the 4-OH
of the GlcNAc termini of existing PG glycan strands
as acceptors. This is a net disaccharide transfer in
each elongation step, catalyzed by enzymes known
as transglycosylases (TGases) (Figure 12). These
newly elongated glycan strands are immature, or
nascent, in the sense that the pentapeptide units are
not cross-linked, so this portion of the PG layer will
be mechanically weak until transpeptidation between
adjacent peptide strands has occurred. This is the
task of a family of transpeptidases (TPases), using
the -NH2 of Lys3 on one strand to attack D-Ala4 on
an adjacent strand, liberating D-Ala5 as the free
amino acid (Figure 12). The Lys3-D-Ala4 interstrand

Glycopeptide and Lipoglycopeptide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 433

Figure 10. Cytoplasmic phases of peptidoglycan assembly. Phase I culminates in UDP murmamyl pentapeptide assembly
in the bacterial cytoplasm; phase II involves enzymatic transfer of muramyl pentapeptide to C55 bactoprnol-P to make
lipid I, followed by MurG-mediated GlcNAc transfer to make the C55-P-P-disaccharly pentapeptide, lipid II.

the fact that the active site serine side chains that
function as catalytic nucleophiles in the transpeptidases are covalently acylated by penicillins (and
cephalosporins).3
The first two stages of peptidoglycan biosynthesis,
leading to the production of lipid II, are wellunderstood at this point. Most of the enzymes involved in this part of the pathway have been studied
extensively, and crystal structures are available for
all of the essential enzymes in stage I and stage II,
except MraY.47-55 The final stage of peptidoglycan
biosynthesis, involving glycan polymerization and
cross-linking, is not nearly as well-understood. From
a chemical perspective, there are only two different
reactions involved in this final stage of peptidoglycan
synthesissa glycosidic bond-forming reaction and a
transpeptidation reaction; however, the resulting
product is a complex polymer, and there is presumed
to be considerable structural heterogeneity at the
molecular level in both the final product and the
various intermediates. Furthermore, there are several different transglycosylase domains and an even
larger number of transpeptidase domains involved
in the biosynthesis of this polymer, as noted in the
section below.

4.3. Transglycosylase and Transpeptidase


Isoforms
Figure 11. Translocation of lipid II. Lipid II is flipped
between the inner (cytoplasmic) and the external face of
the bacterial membranes via an unknown mechanism.

isopeptide bond thus generated is a mechanically


strengthening covalent cross-link. The crucial role of
transpeptidation in PG maturation is borne out by

There are multiple genes encoding TGases and


TPases in bacterial cells.42,56 These genes play different roles in cell growth and division.
Historically, transpeptidases were classified into
high and low molecular weight forms, collectively
known as penicillin-binding proteins (PBPs). Some

434 Chemical Reviews, 2005, Vol. 105, No. 2

Kahne et al.

Figure 12. External phases of peptidoglycan assembly. Phase III involves lipid II as disaccharyl pentapeptide donor to
the 4-OH of GlcNAc at the ends of PG glycan chains undergoing elongation (transglycosylases) and then isopeptide bond
formation between Lys3 on one peptide strand and D-Ala4 on a neighboring peptide strand (transpeptidases).

Figure 13. Schematic of bifunctional TGase/Tpase domains in high molecular weight PBPs.

of the high molecular weight PBPs contain transglycosylase domains in addition to transpeptidase
domains. In Escherichia coli, for example, PBP1A,
-1B, and -1C are bifunctional TGase/TPase enzymes
(Figure 13).57-59 E. coli PBP1B has been proposed to
carry out 85% of PG elongation in that organism.60-62
The remaining 15% of peptidoglycan is made by some
combination of other transglycosylases and transpeptidases. Other high molecular weight PBPs, such as

PBP2 and PBP3 in E. coli, have domains that appear


to be vestigial transglycosylase domains.56 Genetics
experiments have shown that PBP2 is involved in cell
elongation/maintaining cell shape and PBP3 in cell
division.63-65 In addition to the high molecular weight
PBPs, there are several low molecular weight PBPs
containing only transpeptidase or carboxypeptidase
activities. These enzymes are thought to be important
in maintaining the shape of the cell.56,65

Glycopeptide and Lipoglycopeptide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 435

Table 1. List of Predicted Bifunctional TGase/TPases


in Gram-Positive Bacterial Pathogens
bifunctional
PBPs
bacteria
E. faecalis
E. faecium
S. aureus
S. pneumoniae
B. subtilis

enzyme

gene

unnamed
unnamed
unnamed
unnamed
unnamed
unnamed
PBP2
PBP1a
PBP2a
PBP1b
PBP1
PBP2c
PBP4
PBP2d

ponA
pbpF
pbpZ
ponA
pbpF
pbpZ
pbpB
pbp1a
pbp2a
pbp1b
ponA
pbpF
pbpd
ywhE

Several high molecular weight TGase/TPase enzyme forms have been identified in other bacterial
strains, including Gram-positive organisms such as
S. aureus, Streptococcus pneumoniae, and Enterococcus faecalis (Table 1).59,66 The putative functions
of most of these enzymes are based on what has been
learned about the corresponding enzymes in E. coli
from a combination of genetic and biochemical studies.
Although genetics has provided considerable insight into the general roles of different PBPs in cell
growth and division, only limited biochemical work
on either the transglycosylase or transpeptidase
domains of the high molecular weight PBPs has been
carried out. In large part, this is because assays to
monitor the activity of the Tgase and Tpase domains
of high molecular weight PBPs were not available
until recently. The lack of good assays for the high
molecular weight PBPs was related to difficulties in
obtaining adequate quantities of lipid II for study.
Both chemical and enzymatic methods to produce
quantities of lipid II have recently enabled the study
of high molecular weight PBPs, and kinetic characterizations of E. coli PBP1b and S. pneumoniae
PBP2a have been reported.10,67-71,136
The historical lack of methods to study and deconvolute the roles of individual transglycosylase and
transpeptidase domains in vitro and in bacterial cells
has made it difficult to understand the detailed
mechanisms of antibiotics that inhibit the final steps
of peptidoglycan biosynthesis. Nevertheless, it is clear
from studying the glycopeptides that some of the
differences in activity between antibiotics with ostensibly similar mechanisms of action are related to
differential inhibition of related targets (i.e., PBPs).
Now that better biochemical and genetic tools to
probe the function of different PBPs have become
available, it should be possible to learn more about
why different glycopeptides have different effects on
cells. This knowledge, in turn, may lead to the
development of better antibiotics.

4.4. Peptidoglycan Biosynthesis as an Antibiotic


Target
Peptidoglycan biosynthesis is a good pathway to
target for antimicrobial chemotherapy, because it is

essential for survival and highly conserved even in


disparate microorganisms. In fact, many antibiotics
that inhibit the biosynthesis of bacterial peptidoglycan are derived from natural products produced by
microorganisms themselves.3 For reasons that are
not clear, a disproportionate number of natural
product-based, cell wall-active antibiotics inhibit the
final steps of peptidoglycan biosynthesis of peptidoglycan biosynthesis. It is possible that these steps
are preferentially targeted because the enzymes
involved are extracellular and are thus accessible to
compounds that would not penetrate the cell membranes. It has also been argued, however, that it is
advantageous to inhibit metabolic processes that
involve multiple, related enzymes because spontaneous mutations in single enzymes do not result in
resistance.72
Antibiotics that inhibit the final stages of peptidoglycan biosynthesis fall into three major classes
with respect to mechanism of action (Figure 14).73
The first class, which includes the -lactams, carbapenems, and cephalosporins, comprises those antibiotics that prevent glycan cross-linking by inhibiting the active sites of enzymes catalyzing transpeptidation. The second class includes those antibiotics that inhibit glycan polymerization by binding to
bacterial transglycosylases. Moenomycin, which is
used commercially in animal feed, is the prototypical
member of this class. It is notable for its outstanding
potency, but is not used in humans because it has
poor physical properties. The third class includes
antibiotics that can inhibit glycan polymerization
and/or cross-linking by binding to the substrates of
transglycosylases and transpeptidases. The glycopeptides are the best known of these substratebinding antibiotics, but there are many others,
including ramoplanin and members of the lantibiotic
family of antibiotics, both of which are reviewed in
this issue in the reviews by Walker and Boger and
by Van der Donk et al. The glycopeptides present
special challenges for mechanistic analysis, because
they are potentially able to inhibit multiple enzymes
involved in two distinct processessi.e., glycan polymerization and cross-linkingsat the same time.
Vancomycin is the prototypical member of the
glycopeptide family of antibiotics and has served as
the model system for many mechanistic investigations of glycopeptides. It is commonly assumed that
teicoplanin and other glycopeptides kill bacterial cells
by the same mechanism of action as vancomycin,
because they recognize the same dipeptide motif on
peptidoglycan precursors (see below). However, there
is increasing evidence that different glycopeptides
preferentially target different enzymatic steps. A
major goal of this review is to draw attention to the
differences between vancomycin, teicoplanin, and
other glycopeptide antibiotics and to address possible
mechanistic explanations for these differences. In the
following sections, we provide an overview of the
general mechanism of action and the molecular basis
for resistance of the glycopeptide class of antibiotics
as a prelude to discussing the differences between
compounds.

436 Chemical Reviews, 2005, Vol. 105, No. 2

Kahne et al.

Figure 14. Three classes of antibiotics inhibiting stage III of peptidoglycan assembly: -lactams, moenomycin, and
glycopeptides.

4.5. Early Studies on the Mechanism of Action of


Glycopeptide Antibiotics
Vancomycin was discovered in 1956 and its general
mechanism of action was elucidated in the 1960s.41,74
In 1965, Strominger and co-workers reported the first
studies on the mechanism of action of vancomycin
and the related glycopeptide ristocetin.40 These authors carried out a series of studies using crude
bacterial membranes containing peptidoglycan-synthesizing enzymes supplemented with radiolabeled
UDP-MurNAc pentapeptide and UDP-GlcNAc in
which they showed that the glycopeptide antibiotics
block peptidoglycan biosynthesis at either transglyoslyation or transpeptidation. This work represented
a tour de force at the time, because key steps in the
biosynthetic pathway to peptidoglycan were not
understood. Subsequent work by Perkins showed
that vancomycin actually binds to peptidoglycan
precursors, specifically to the D-Ala-D-Ala terminus,
leading to the hypothesis that enzyme inhibition is
related to this phenomenon.75,76 Stepwise degradation
of UDP-MurNAc-pentapeptide indicated that vancomycin interacts with the D-Ala-D-Ala terminus of this
precursor. In addition to UDP-MurNAc-pentapeptide,
other peptidoglycan intermediates that contain the
D-Ala-D-Ala dipeptide include (inward facing) lipid I,
(inward and outward facing) lipid II, and nascent (uncross-linked) peptidoglycan (Figures 10 and 12). Since
experiments with radioactive vancomycin derivatives
showed that vancomycin does not enter cells, it was
concluded that vancomycin and other glycopeptides
affect the extracellular enzymes that utilize intermediates downstream of lipid I, such as lipid II.77
Many of the structural features of vancomycin were
determined in 1978, when Sheldrick and Williams
determined the structure of a degradation product,
CDP-1, through X-ray analysis.78 However, the structure of vancomycin was damaged during the degradation process and the correct structure was finally

Figure 15. Complexation of vancomycin with N-acyl-DAla4-D-Ala5 termini: five hydrogen bonds between the
underside of the glycopeptide and the acyl-D,D-dipeptide
moiety.

put forth in 1982, based on important contributions


from the Harris and Williams labs.79,80 Shortly after
the structure of vancomycin was solved, the Williams
lab used NMR to show the binding interaction
between vancomycin and the D-Ala-D-Ala dipeptide.81
Binding was shown to occur through a set of backbone contacts between the D-Ala-D-Ala dipeptide and
the amides that line a cleft formed by the cross-linked
heptapeptide of the glycopeptide. (Figure 15) All
glycopeptides have a similar binding pocket and are
believed to make the same contacts.82 Through binding to D-Ala-D-Ala, the bound glycopeptide acts as a
steric impediment that prevents lipid II and/or the

Glycopeptide and Lipoglycopeptide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 437

nascent glycan chain from being processed further.83


The net result is inhibition of the transglycosylation
and/or transpeptidation steps of peptidoglycan synthesis, which overall weakens the peptidoglycan
layers and leaves the cell susceptible to lysis due to
changes in osmotic pressure.

5. Mechanisms of Resistance
Scientists have been intrigued by the unusual
mechanism of action and behavior of vancomycin
since its discovery. Researchers noticed early on that
it is difficult to induce resistance to vancomycin.
Ziegler et al., for example, compared penicillin and
vancomycin, both of which inhibit stage III of peptidoglycan synthesis, and found that the MICs of
penicillin against a range of S. aureus strains increased by more than 100 000-fold after 25 serial
passages in antibiotic-containing media.84 In contrast,
the MICs of vancomycin increased by only 8-fold.
Furthermore, whereas resistance to the -lactams
appeared almost immediately upon the introduction
of penicillin into clinical use, glycopeptide resistance
was not observed for a very long time. Resistance to
antibiotics typically develops when either an antibiotic itself or its target is altered in some way.11 Unlike
the -lactams, the glycopeptides did not appear to be
susceptible to modifications that rendered them
inactive. Furthermore, it was widely believed that
microorganisms could not readily alter the target of
the glycopeptides, the D-Ala-D-Ala peptide terminus,
because that would entail simultaneous, coordinated
changes to multiple enzymes in the pathway to
peptidoglycan synthesis. In 1986, Cooper and Given
noted that during the three decades in which vancomycin has been in clinical use, there has been no
trend toward resistance among organisms usually
susceptible, and speculated that the mode of action
of glycopeptide antibiotics made the development of
high-level resistance virtually impossible.85 One year
later, vancomycin-resistant enterococcal strains began to appear in hospitals, and 15 years after that
the incidence of VRE in hospitalized patients with
enterococcal infections in the US had spread to 30%.1
In retrospect, the appearance of significant glycopeptide resistance is not surprising, because the
widespread use of an antibiotic virtually guarantees
the emergence of resistance.86 However, high-level
resistance to the glycopeptides in enterococci is not
the result of spontaneous mutations in clinically
relevant microorganisms. Instead, the genes conferring glycopeptide resistance in the organisms producing glycopeptide antibiotics appear to have been
transferred to pathogenic microorganisms. The mechanism of glycopeptide resistance in enterococci was
elucidated by Courvalin, Walsh, and their co-workers
in the 1990s. Subsequent work on glycopeptide
resistance in producer organisms has revealed that
they contain the same sets of resistance genes as the
resistant enterococcal strains. The mechanism of
glycopeptide resistance in enterococci is described
below.

Figure 16. Schematic of the Five Gene Van RSHAX


cassette in VanA and VanB phenotypes of VRE.

including additional variants of these classes89). The


VanA and VanB phenotypes were initially distinguished by differential susceptibility to vancomycin
vs teicoplanin. The VanA phenotype shows 1000-fold
increased resistance to both drugs, while VanB VRE
isolates have equivalent resistance to vancomycin but
remain susceptible to teicoplanin.4 Additionally, a
VanC VRE phenotype has been observed: it confers
about 1 log increase in resistance to vancomycin and
has not been a widespread problem in humans.88
VanA and VanB isolates of VRE contain five van
genes, VanRSHAX, necessary and sufficient to cause
high-level resistance (Figure 16).
The encoded five proteins sort into two categories.
The VanR and VanS pair up to function as a twocomponent regulatory system.90 VanS is a transmembrane receptor histidine kinase. The extramembrane domain, directly or indirectly, senses vancomycin on the outside and transmits that information
to the cytoplasmic domain, which autophosphorylates
on the His side chain. The phospho-VanS then
transfers the PO3 group to an aspartyl side chain in
the N-terminal domain of VanR in the enterococcal
cytoplasm. The C-terminal domain of phospho-VanR
now acts as a transcriptional regulator91,92 to activate
the transcription of the VanHAX genes.
The VanHAX proteins comprise the second category. All three are enzymes, coordinately acting to
reprogram the peptidoglycan termini from N-acyl-DAla-D-Ala, a high-affinity target of vancomycin and
teicoplanin, as noted above in section 4, to N-acyl-DAla-D-lactate. VanH reduces the common metabolite
pyruvate to D-lactate. VanA is a D,D-depsipeptide
ligase, making D-Ala-D-lactate. VanX is a D-Ala-D-Ala
dipeptidase, removing the normal D-D-dipeptide intermediate hydrolytically while sparing D-Ala-Dlactate. The D-Ala-D-Lac accumulates and gets added
by the MurF ligase to UDP muramyltripeptide to
generate the UDP muramyl-L-Ala-D--Glu-L-Lys-DAla-D-Lac that can serve as cross-linking substrate
for transpeptidase action (Figure 17).
The 1000-fold resistance in VanA and -B VRE
phenotypes results from the reprogramming of the
peptidoglycan termini from D-Ala-D-Ala to D-Ala-Dlactate.90,93 The absence of the amide NH bond and
the ground-state repulsion of the oxygen lone pair
cause a 1000-fold loss of binding affinity for vancomycin and teicoplanin to N-acyl-D-Ala-D-lactate, correlating genotype with phenotype (Figure 18).93,94

5.1. Vancomycin Reistant Enterococci (VRE)

5.2. VRE Genotypes: VanB Resistance and How


To Overcome It

VRE has been categorized into distinct clinical


phenotypes, originally VanA and VanB87,88 (but now

One of the interesting aspects of glycopeptide


resistance is that the VanB genotype remains sus-

438 Chemical Reviews, 2005, Vol. 105, No. 2

Figure 17. Action of VanHAX to make D-Ala-D-Lac and


destroy competing D-Ala-D-Ala as a substrate for MurF.

Figure 18. Loss of the one H-bond in the vancomycin


complex with D-Ala-D-Lac vs D-Ala-D-Ala and ground-state
repulsion by the ester oxygen accounts for a 1000-fold drop
in affinity.

ceptible to teicoplanin. VanB strains of VRE have


vanRSHAX genes and can reprogram PG termini to
produce D-Ala-D-Lac. However, only vancomycin induces these strains to make altered peptidoglycan
precursors.87 The difference in behavior between
vancomycin and teicoplanin with respect to VanB
strains is puzzling, because vancomycin and teicoplanin have remarkably similar structures (Figure
1). There are many hypotheses that attempt to
explain these observations.
Williams and co-workers found that lipidated glycopeptides are anchored to the bacterial cell membrane, whereas nonlipidated glycopeptides are distributed more broadly in the peptidoglycan layers.95,96
They reasoned that teicoplanin inserts itself into the
membrane, where the peptidoglycan precursors terminating in D-Ala-D-Lac are located. Thus, the two
binding partners are located in close proximity to
each other and the interaction between teicoplanin
and D-Ala-D-Lac becomes intramolecular, which is
more favorable than an intermolecular interac-

Kahne et al.

tion.97,98 Williams and co-workers concluded that it


is the restriction of motion of intermediates and the
intramolecular bond formation that allow teicoplanin
to overcome VanB resistance.31
In 1996, Courvalin and co-workers found that
VanB VRE remain susceptible to teicoplanin, because
this antibiotic does not induce the cell wall reprogramming machinery. They showed that vancomycin,
but not teicoplanin, activates the VanSb sensor kinase.99 Furthermore, they found that mutations in
the VanSb sensor kinase render cells resistant to
teicoplanin.100 From these findings, they reasoned
that activation of VanSb required direct interaction
with the glycopeptide.100,101
To determine which structural features of teicoplanin and vancomycin correlate with induction of
resistance, Dong et al. undertook a direct comparison
of pairs of teicoplanin and vancomycin analogues.
The compounds examined included the teicoplanin
and vancomycin aglycons, the monoglucosylated scaffolds bearing the same disaccharyl chain, and the two
scaffolds bearing a C10-acyl glucosamine group (Figure 19).102
The teicoplanin and vancomycin aglycons and the
monoglucosylated scaffolds induced resistance, but
the lipoglycopeptides did not (as judged by susceptibility of VanB VRE strains to the test compounds).
Thus, adding a simple C10 acyl chain to vancomycin
abrogated its ability to activate the resistance genes.
More recently, Boger and co-workers have investigated the activity of methyl ester derivatives of
vancomycin and teicoplanin, replacing the carbohydrate portion of the glycopeptide with a methyl
group.103,104 They found that these compounds were
still active against VanB VRE strains and attributed
their activity to the hydorophobic nature of these
compounds. Thus, they reasoned that it might not
be necessary to add a lipid chain to the glycopeptide
to overcome VanB resistance; rather, a simple methyl
group will prove effective.
Two hypotheses to explain the differences between
lipoglycopeptides and the corresponding glycopeptides vis-a` -vis the induction of resistance have been
put forth.102 Both hypotheses attribute the differences
between lipidated/hydrophobic and nonlipidated glycopeptides to differential partitioning among available targets on bacterial cell surfaces. Williams and
co-workers showed that the lipid chain of teicoplanin
localizes the molecule to the membrane. Therefore,
it has been suggested that membrane localization
renders teicoplanin and other lipidated glycopeptides
inaccessible to the sensor kinase, so that induction
does not occur. This hypothesis presumes that induction of resistance involves a direct interaction between the sensor kinase and the antibiotic.100,101
Alternatively, it has been suggested that the sensor
kinase interacts not with the glycopeptide itself but
with peptidoglycan intermediates or degradation
products produced by the metabolic blockade.102
Teicoplanin does not induce resistance according to
this hypothesis, because it blocks a different enzymatic step than vancomycin, which means that
different intermediates and degradation products
buildup.

Glycopeptide and Lipoglycopeptide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 439

Figure 19. Teicoplanin and vancomycin analogues. A set of matched pairs of vancomycin and teicoplanin derivatives
(a-f) were used to probe what triggers the sensor kinase. It was found that C and D did not induce resistance, whereas
all other compounds did.

There is no direct evidence about how the sensor


kinase is activated. However, using an E. coli model
system, Kahne and co-workers have shown that
lipidated glycopeptides preferentially inhibit the
transglycosylation step of peptidoglycan biosynthesis,
whereas nonlipidated glycopeptides inhibit the
transpeptidation step.8,105 Membrane-anchored glycopeptides evidently inhibit transglycosylation, because they bind preferentially to lipid II, the membrane-anchored substrate for the transglycosylases,
rather than to nascent peptidoglycan (Figure 20).
Thus, relatively small structural changes in glycopeptides can lead to differential partitioning among
available binding sites and thus, in turn, can have
functional consequences in terms of the metabolic

steps that are affected. While it is still not known


how the VanSB sensor kinases are activated, mechanistic investigations aimed at understanding the
molecular basis for glycopeptide resistance have
already led to a clear prescription for how to avoid
reprogramming the bacterial cell wall, namely, use
lipoglycopeptides.74,102

5.3. VRE Genotypes: VanA Resistance and the


Compounds That Overcome It
VanA-resistant enterococcal strains contain similar
sets of genes for sensing glycopeptide activity and for
reprogramming peptidoglycan precursors as VanB
strains. Unlike the latter, however, VanA strains are

440 Chemical Reviews, 2005, Vol. 105, No. 2

Kahne et al.

Figure 20. Two types of TPase substrates containing -D-Ala-D-Ala termini (in red) that can be complexed with vancomycin.
lipid II are molecules embedded in the bacterial membrane via the C55 anchor. Immature PG chains that have been elongated
by TGase action and are now ready for TPase action.

resistant to all natural glycopeptides that have been


examined, including teicoplanin. Thus, irrespective
of differences in how they are distributed among the
possible binding sites or which steps they preferentially block, natural glycopeptides induce VanA strains
to make altered cell wall precursors. Nevertheless,
some semisynthetic glycopeptide derivatives overcome VanA resistance, and studies on these compounds are beginning to provide clues as to how this
form of resistance can be overcome.
Researchers at Eli Lilly reported the first glycopeptide derivatives capable of overcoming both VanA
and VanB resistance. These compounds, which were
vancomycin derivatives containing lipid substituents
attached to the nitrogen of the vancosamino sugar
(Figure 21), showed good activity against a broad
panel of vancomycin-resistant strains as well as
outstanding activity against sensitive strains.106,107
The vancosamino nitrogen of vancomycin can be
readily modified without extensive protecting group
manipulations and was, therefore, a logical position
to explore. The motivation for attaching hydrophobic
substituents was not made clear in the reports on
these compounds, but it is possible that teicoplanins
activity against VanB strains provided the inspiration. It would have been reasonable to speculate that
activity against VanB strains could be recovered by
attaching a hydrophobic substituent to the vanco-

Figure 21. Lipoglycopeptide derivatives found at Eli Lilly


to reverse VanA phenotypic resistance.

mycin carbohydrate, because the teicoplanin carbohydrate, which is attached to the same amino acid,
contains a hydrophobic substituent. Remarkably,

Glycopeptide and Lipoglycopeptide Antibiotics

some of the glycolipid derivatives of vancomycin


proved to be active not only against VanB strains but
also against VanA strains, and the Lilly group
ultimately took a lipid derivative of chloroeremomycin, now known as oritavancin, into the clinic (vide
infra). Understanding the mechanism by which VanA
resistance is overcome is crucial to developing better
second-generation glycolipid derivatives. Two different models for how compounds such as oritavancin
overcome VanA resistance have been proposed and
are discussed below.

Chemical Reviews, 2005, Vol. 105, No. 2 441


Table 2. MICs for a Series of Glycopeptides
Derivatives against Sensitive and Resistant Strainsa

5.4. Models for How Vancomycin Analogues


Overcome Vancomycin Resistance
5.4.1. Dimerization and Membrane Anchoring
Williams and co-workers proposed the first model
which holds that vancomycin analogues kill resistant
bacterial strains by the same mechanism that they
kill sensitive strains, i.e., these analogues still bind
to peptide termini of peptidoglycan precursors such
as lipid II, but they have acquired the ability to bind
to D-Ala-D-Lac better than vancomycin itself can (for
a review, see Williams et al.).31 Williams has proposed that the lipid substituent on vancomycin
facilitates dimerization and membrane localization.
It is these characteristics that promote better binding
to D-Ala-D-Lac peptides that are presented in multiple
copies in close proximity on the cell surface.108 Williams argues that the second binding event is entropically favored to the extent that it compensates
for the loss of the hydrogen bond between the
depsipeptide ligand and the glycopeptide. There is
considerable evidence to support the proposal that
dimerization can enhance binding avidity to ligands
that are presented in multiple copies on a surface of
vesicles.109 Moreover, it has been shown that covalent
dimers and covalent trimers of vancomycin bind with
high avidity to polyvalent peptide ligands.110,111 However, there is no evidence that improved biological
activity is related to the ability of glycopeptide
analogues to dimerize. Furthermore, there is no
evidence that these glycopeptides kill resistant bacteria by binding to D-Ala-D-Lac.
Membrane anchoring has also been suggested as
a factor contributing to enhanced activity against
resistant strains. It has been proposed that localizing
the glycopeptide near the membrane partially compensates for the weak binding to D-Ala-D-Lac. Again,
there is both theoretical and experimental support
for the hypothesis that anchoring to membranes
should increase avidity of glycopeptide analogues for
membrane-anchored precursors presenting D-Ala-DAla or D-Ala-D-Lac.95,108 Williams argues that it is a
combination of membrane anchoring and dimerization that allows lipidated derivatives of vancomycin
to bind to D-Ala-D-lactate. To show the effectiveness
of the dimerization/membrane localization combination, Allen and co-workers damaged the lipidated
derivative of vancomycin so that it could no longer
bind D-Ala-D-Ala.112 It has been shown that if the
N-terminal amino acid is removed from vancomycin,
it can no longer bind to peptidoglycan precursors.113
Allen and co-workers showed that the damaged

MIC (g/mL)
glycopeptide
vancomycin
chlorobiphenyl vancomycin
damaged vancomycin
damaged chlorobiphenyl
vancomycin

sensitive
E. faecium

resistant
E. faecium

1
0.03
no activity
10

2048
16
no activity
40

a Compounds lacking the N-terminal methylleucine amino


acid were used to evaluate which component of the activity
derives from a peptide-binding-independent mechanism.

compounds still formed dimers and that these dimers


were able to bind D-Ala-D-Ala, albeit very weakly.112
Allen and co-workers attributed the biological activity
of these damaged compounds to their ability to
dimerize, which allows them to weakly bind to
precursors. However, it is interesting to note that
Ellman and co-workers have published data showing
that covalent dimers of damaged vancomycin derivatives, nevertheless, have biological activity.114,115 The
following section describes a second mechanism of
action that explains how these lipidated compounds
kill resistant bacteria, despite being unable to bind
peptidoglycan precursors.

5.4.2. A Second Mechanism of ActionsDirect Interaction


with the Transglycosylase
Biophysical studies of binding using model systems
are at best suggestive with respect to mechanism of
action. Kahne and co-workers, therefore, designed
experiments to test the hypothesis that activity
against resistant bacterial strains depends on binding
to D-Ala-D-Lac. To address this issue, vancomycin
analogues in which the peptide binding pocket was
damaged were prepared (Table 2). These analogues
lack the N-methylleucine moiety of the aglycon,
which makes both hydrogen bonding and electrostatic
contacts to D-Ala-D-Ala. It has been established that
these damaged vancomycin analogues are unable to
bind N-acyl-D-Ala-D-Ala.113 The damaged vancomycin
analogues were tested for activity against both sensitive and vancomycin-resistant strains. The chlorobiphenyl derivative lost considerable activity against
sensitive strains, as expected; however, its activity
against resistant strains did not change significantly.
While it is conceivable that membrane-anchoring
or dimerization could compensate for the loss of a
single hydrogen bond, it would not be reasonable to

442 Chemical Reviews, 2005, Vol. 105, No. 2

Figure 22. Two mechanisms of action for glycopeptides:


(a) inhibition of the transpeptidase step by binding to the
D-Ala-D-Ala terminus and (b) direct inhibition of the
transglycosylases.

conclude that these phenomena compensate for the


loss of two hydrogen bonds and an electrostatic
contact. Kahne and co-workers therefore concluded
that depsipeptide binding does not play a role in the
mechanism of action by which chlorobiphenyl vancomycin analogues kill VanA resistant bacteria. They
proposed, instead, that there must be a second mechanism of action that explains the activity against resistant strains. Studies aimed at probing the site of
inhibition of the damaged chlorobiphenyl vancomycin
analogues established that the damaged compound,
like the intact parent, blocks peptidoglycan biosynthesis8 at the transglycosylation step (Figure 22).
On the basis of these experiments, they suggested
that chlorobiphenyl vancomycin analogues might
interact directly with a key component of the transglycosylation machinery.
Goldman and co-workers provided additional evidence for a second mechanism of action for chlorobiphenyl derivatives of vancomycin when they showed
that both chlorobiphenyl vancomycin and damaged
chlorobiphenyl vancomycin are able to inhibit formation of nascent peptidoglycan when UDP-MurNactetrapeptide (L-Ala--D-Gln-L-Lys-D-Ala) is provided
to bacterial cells as a substrate. Vancomycin, in
contrast, does not inhibit the incorporation of this
substrate into peptidoglycan.116 The following year,
Chapman and co-workers showed through affinity
chromatography of bacterial cell extracts that chlorobiphenyl derivatives of vancomycin retained multiple PBPs,117 including E. coli PBP1b, whereas
vancomycin was unable to retain PBPs. These experiments provided support for the hypothesis that
chlorobiphenyl derivatives of vancomycin interact
directly with PBPs involved in peptidoglycan biosynthesis. Chen et al. subsequently developed an assay
to monitor the activity of E. coli PBP1b and found
that chlorobiphenyl vancomycin analogues can inhibit this enzyme without binding substrate.10 Inhibition by vancomycin, in contrast, depends on peptide
binding. Taken together, these studies provide strong
evidence for a second mechanism of action. At this
point, however, it is essential to develop methods to
study Gram-positive transglycosylases found in relevant microorganisms in order to determine whether
chlorobiphenyl vancomycin analogues directly inhibit
these enzymes.137 If, in fact, vancomycin analogues
with activity against resistant microorganisms func-

Kahne et al.

Figure 23. Proposal for recombination in a patient


infected with VRE and MRSA allowing TN1546 to move
to MRSA and create VRSA.

tion because the derivatized carbohydrate moiety


facilitates a direct and inhibitory interaction with
transglycosylases, then it should be possible to design
improved antibiotics by attaching to the vancomycin
aglycon structural elements that have better activity
against relevant lipid II-utilizing transglycosylases.8,20,46,118 The design of better antibiotics is important not only for overcoming VanA resistance but also
for tackling the problem of the newly emerging
vancomycin-resistant staphylococcus.

5.5. Vancomycin-Resistant S. aureus (VRSA)


For some years the levels of vancomcyin resistance
among clinical isolates of Staphylococci were low. S.
aureus with such diminished susceptibility to vancomycin were termed VISA (vancomycin intermediate S. aureus).119,120 The genotypes of these isolates
were not of the VanHAX type noted above for VRE.
Rather, they typically generated multilayered, thickened cell walls as though more sites for stoichiometric
binding of drug were the cause of lessened susceptibility. But in 2003, VRSA were isolated from a
dialysis patient who also had a chronic infection with
VRE.121 Genotyping of the VRSA showed the same
five VanRSHAX genes found also on transposons in
the VRE (Figure 23). Every indication is that the
transposon hopped from the enterococcal host to the
S. aureus.122 It remains to be seen what the rate of
spread of the VanRSHAX genes will be into MRSA
strains, but there is no doubt that this only hastens
the need for second- and third-generation forms of
glycopeptides and lipoglycopeptides that can combat
various phenotypes of VRE and VRSA.

6. New Directions for Treating VRE and VRSA


The long time frame and progressively broader
clinical use of the first-generation natural product
glycopeptides vancomycin and teicoplanin as front
line agents has eventually led to widespread incidence of VRE (up to 30% of enterocccal infections in
US hospitals in 2002) and the recent emergence of
VRSA. Over the past decade there has been keen
interest in the discovery and development of new
treatment regimes for these life-threatening bacterial

Glycopeptide and Lipoglycopeptide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 443

Figure 24. Three nonglycopeptide classes of antibiotics recently approved for treatment of VRE infections: Synercid (11)
combination of Quinupristin and Dalfopristin blocks protein synthesis; the oxazolidinone Linezolid (12) blocks the first
peptide-bond-forming step at bacterial ribosomes; the lipopeptide daptomycin (13) damages membranes and causes ion
leaks in bacteria.

Figure 25. Lipoglycopeptide antibiotics in clinical development: oritavancin (14) and TD-624 (15) are N-aryl and N-alkyl
hydrophobic derivatives modified on the vancomycin-type heptapeptide scaffold; dalbavancin (16) is a lipoglycopeptide
modified in the amide linkage on a teicoplanin-type scaffold.

pathogens. Two parallel lines of discovery and development have ensued. One has been the search for

a replacement to vancomycin and teicoplanin and has


yielded newly approved drugs. The other line began

444 Chemical Reviews, 2005, Vol. 105, No. 2

Kahne et al.

Figure 26. (a) Reconstitution of chloroermomycin from the aglycon scaffold by consecutive action of Gtfs B, A, and C. (b)
Decoration of the teicoplanin aglycon with GlcNAc at residue 6, with glucosamine at residue 4 by tGtfA and tGtfB, and
N-acylation of the glucosamine residue by acyl transferase action.

with semisynthetic improvements to the vancomcyin


and teicoplanin subfamily scaffolds.

6.1. Recently Approved Drugs for VRE


Over the past 5 years three new classes of antibiotics have been approved in the US with efficacy
against enterococcal infections (Figure 24).3 None of
the drugs are glycopeptides, so they do not induce
the VanRSHAX enzymes and therefore are active
against all VRE strains. Synercid (11), approved in
1998, is a mixture of two natural product cyclic
lactones, pristinamycins I and II. They block bacterial
protein synthesis in the elongation phase.4 Linezolid
(12), a synthetic antibacterial agent with an oxazolidinone backbone approved in 2000, blocks the first
peptide-bond-forming step at the peptidyl transferase
center of bacterial ribosomes. The third molecule,
daptomycin (13), approved in 2003 under the trade
name Cubicin, is also a natural product. It is a
lipopeptidolactone, made nonribosomally,4 that dis-

rupts bacterial membrane integrity, causing ions to


leak out and thus bacterial death. All three of these
drugs then act by mechanisms independent of those
used by vancomycin and teicoplanin and will not be
subject to cross-resistance with the glycopeptides.

6.2. Second-Generation Semisynthetic


Lipoglycopeptides in Clinical Development
Given the structural complexity of the glycopeptide
antibiotic class, in particular, the oxidatively crosslinked heptapeptide scaffolds that are variant in
vancomcyin and teicoplanin family members, efforts
have been undertaken to create semisynthetic drugs
by chemical modification of the natural products.
One successful route has involved the single-step
reductive alkylation of the amino group of the vancosamine sugar in the vancomycin/chloroeremomycin
scaffold and led to oritavancin (14). Use of an N-alkyl
rather than N-aryl group has generated TD-624 (15)
(Figure 25). A comparable single-step chemical modi-

Glycopeptide and Lipoglycopeptide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 445

Figure 27. Enzymatic synthesis of a variant glycopeptide, starting from the teicoplanin aglycon, UDP-4-deoxyglucose
and GtfB, and TDP-l-epivancosamine and GtfC.

fication by amidation of the carboxy terminus of the


heptapeptide scaffold of the teicoplanin family member A40926 has generated dalbavancin (16).

6.2.1. Oritavancin
Screening of the natural and semisynthetic glycopeptide collection at Lilly led to the choice of the
chloroeremomycin scaffold and a series of N-alkyl or
N-acyl derivatives of the terminal aminosugar on the
glucosyl-epivancosamine chain.123 The oritavancin
molecule, originally known as LY333328, is a simple
N-aryl derivative of the natural product chloroeremomycin 4. The epivancosamine moiety in the
disaccharide chain could be selectively derivatized at
the amino group by imine formation with aryl aldehydes, followed by reduction to the stable secondary
amine. The chlorobiphenyl group was one of several
lipophilic groups that conferred activity against VRE,
allowing a regain of about 2 of the 3 logs of activity
lost in VanA type VRE.123 Arylamines have been used
elsewhere to decorate natural product scaffolds,

notably in the antifungal echinocandins as replacements for fatty acyl amide groups,124 and they may
mimic the N-acyl substituents found in the teicoplanin series of natural lipoglycopeptides. Oritavancin is active against both VanA and VanB phenotypes
of VRE and also MRSA and VRSA at MIC values of
<1 g/mL.123 Once-a-day administration is proposed.
Oritavancin shows strong bacteriocidal properties
under conditions where vancomycin is bacteriostatic.
Oritavancin has shown clinical effectiveness in complicated skin and soft tissue infections caused by
Gram-positive bacteria, including MRSA.125

6.2.2. Dalbavancin
Dalbavancin is a semisynthetic variant of the
teicoplanin family member A40926.22,126,137 Like
A40926 and teicoplanin, it has a long fatty acyl
moiety, in this case a C12 terminally branched chain,
in amide linkage to the glucosamine. The mannose
at residue 7 is present but not the GlcNAc at residue
6. The synthetic modification is amidation of the

446 Chemical Reviews, 2005, Vol. 105, No. 2

C-terminal carboxylic acid of the cross-linked heptapeptide scaffold by a N,N-dimethyl propylamide


group.
This lipoglycopeptide is active against VanB type
VRE at 0.1 g/mL but is ineffective against VanA
phenotypes. It has enhanced potency over vancomycin and teicoplanin against susceptible enterococci
and against MRSA.126 Dalbavancin also has bacteriocidal activity. A once weekly administration is
being planned, given the long lifetime.127 Dalbavancin
has also shown efficacy in complicated skin and skin
structure infections.128

6.2.3. TD-6424
This is the third of the second-generation, semisynthetic lipoglycopeptide variants to enter clinical
development. In some analogy to oritavnacin, TD6424 utilizes a vancomycin-type cross-linked heptapeptide scaffold. There is one modification in that
scaffold at residue 7. The Dpg7 has been replaced
with a 4-aminoethyl side chain, in turn bearing a
CH2PO32- substituent, to enhance solubility of the
scaffold. The other modification maintains the lipoglycopeptide character of these second-generation
compounds. The vancosamine sugar is alkylated by
a long alkyl chain, with an NH at the two positions.
The reductive alkylation is again a presumptive
mimic of the normal fatty acyl chain attached to the
single sugar of teicoplanin. TD-6424 is rapdly bacteriocidal129 and more potent than teicoplanin or
vancomycin against MSSA and MRSA.

6.3. In Discovery: Chemoenzymatic Routes To


Modify Sugars and Acyl Groups on Heptapeptide
Scaffolds
The second-generation lipoglycopeptides in development noted above represent different one-step
chemical manipulations of chloroeremomycin or
A40926, and multiple-step chemical manipulation of
both scaffold and aminosugar N-alkylation in TD-624
on the natural products derived from fermentation.
The current knowledge of the biosynthetic gene
clusters for both vancomycin and teicoplanin family
members noted in section 3 and the importance of
the sugars and acyl groups decorating the heptapeptide scaffold noted in section 5 have focused attention
on using the tailoring enzymes to embellish the
scaffolds in new combinations.
For example, three glycosyltransferases, GtfA, -B,
and -C, convert the vancomycin aglycon to chloroeremomycin, a pathway that has been reconstituted in
vitro with the purified enzymes (Figure 26a)38 Analogously, the two homologous Gtfs from the teicoplanin
cluster, tGtfA and tGtfB, have been overproduced
and shown to transfer GlcNAc to residues 6 and 4 of
the teicoplanin aglycon.16,17 Further, tGtfB will transfer glucosamine, which can be N-acylated by an acyl
transferase also encoded in the biosynthetic cluster
(Figure 26b).17 Different sugars and scaffolds can be
used by some of the Gtfs such that novel glycopeptides can be assembled. Thus, GtfB will take several
UDP-glucose derivatives,130,131 some of which can be
elongated by GtfC or by GtfD from the vancomycin
biosynthetic cluster. It is possible, for example, to

Kahne et al.

build a variant disaccharide, e.g. 4-deoxy-D-glucose2,1-L-epivancosamine, on an altered scaffold, the


teicoplanin aglycon, introducing three variations via
the action of GtfB and -D (Figure 27).132 X-ray
structures of GtfA, -B, and -D133-135 have been
determined to allow for molecular insights into Gtf
specificity for NDP sugar donor and scaffold acceptor
and aid in structure-based Gtf evolution.
Finally, it is possible to N-alkylate and N-arylate
variant aminosugars introduced by Gtfs on the heptapeptide scaffolds to effect chemoenzymatic manipulations that enable multistep variations of structure
in glycopeptide and lipoglycopeptide antibiotics.102
This approach should allow construction of focused
libraries of novel lipoglycopeptide structures to optimize properties desired in a third-generation glycopeptide antibiotic, including broad spectrum, rapid
cidality, oral activity, high potency, good pharmacokinetic, and therapeutic ratio.

7. Acknowledgments
The authors are indebted to many colleagues in the
Kahne and Walsh research groups for contributions
to glycopeptide and lipoglycopeptide projects, some
of which are acknowledged in the references cited.
Research in the Walsh group was supported in part
by NIH Grant GM49338 and in the Kahne group by
NIH Grant 66174. C.W. is a member of the board of
directors of Vicuron Pharmaceuticals.

8. Note Added after ASAP Publication


This review was posted ASAP on January 22, 2005.
The spelling of transglycosylase was corrected in
Figures 12 and 13, and refs 136 and 137 were added.
The review was reposted January 27, 2005.

9. References
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)

Murray, B. E. N. Engl. J. Med. 2000, 342, 710.


Ferber, D. Science 2003, 302, 1488.
Walsh, C. T. Nat. Rev. Microbiol. 2003, 1, 65.
Walsh, C. T. Antibiotics: Actions, Origins, Resistance; ASM
Press: 2003.
Williams, D. H. Nat. Prod. Rep. 1996, 13, 469.
Nicolaou, K. C.; Boddy, C. N.; Brase, S.; Winssinger, N. Angew.
Chem., Int. Ed. 1999, 38, 2096.
Thorson, J. S., Hosted, T. J., Jiang, J. Q., Biggins, J. B., Ahlert,
J. Curr. Org. Chem. 2001, 5, 139.
Ge, M.; Chen, Z.; Onishi, H. R.; Kohler, J.; Silver, L. L.; Kerns,
R.; Fukuzawa, S.; Thompson, C.; Kahne, D. Science 1999, 284,
507.
Kahne, D.; Silva, D.; Walker, S. In Bioorganic Chemistry:
Carbohydrates; Hecht, S. M., Ed.; Oxford University Press: New
York, 1999.
Chen, L.; Walker, D.; Sun, B.; Hu, Y.; Walker, S.; Kahne, D.
Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 5658.
Walsh, C.; Freel Meyers, C. L.; Losey, H. C. J. Med. Chem. 2003,
46, 3425.
van Wageningen, A. M.; Kirkpatrick, P. N.; Williams, D. H.;
Harris, B. R.; Kershaw, J. K.; Lennard, N. J.; Jones, M.; Jones,
S. J.; Solenberg, P. J. Chem. Biol. 1998, 5, 155.
Pelzer, S.; Sussmuth, R.; Heckmann, D.; Recktenwald, J.; Huber,
P.; Jung, G.; Wohlleben, W. Antimicrob. Agents Chemother. 1999,
43, 1565.
Recktenwald, J.; Shawky, R.; Puk, O.; Pfennig, F.; Keller, U.;
Wohlleben, W.; Pelzer, S. Microbiology 2002, 148, 1105.
Pootoolal, J.; Thomas, M. G.; Marshall, C. G.; Neu, J. M.;
Hubbard, B. K.; Walsh, C. T.; Wright, G. D. Proc. Natl. Acad.
Sci. U.S.A. 2002, 99, 8962.
Sosio, M.; Kloosterman, H.; Bianchi, A.; de Vreugd, P.; Dijkhuizen, L.; Donadio, S. Microbiology 2004, 150, 95.

Glycopeptide and Lipoglycopeptide Antibiotics


(17) Li, T. L.; Huang, F.; Haydock, S. F.; Mironenko, T.; Leadlay, P.
F.; Spencer, J. B. Chem. Biol. 2004, 11, 107.
(18) Evans, D.; Wood, M.; Trotter, B.; Richardson, T.; Barrow, J.;
Katz, J. Angew. Chem., Int. Ed. 1998, 37, 2700.
(19) Boger, D. L. Med. Res. Rev. 2001, 21, 356.
(20) Thompson, C.; Ge, M.; Kahne, D. J. Am. Chem. Soc. 1999, 121,
1237.
(21) Hubbard, B. K.; Walsh, C. T. Angew. Chem., Int. Ed. 2003, 42,
730.
(22) Sosio, M.; Stinchi, S.; Beltrametti, F.; Lazzarini, A.; Donadio,
S. Chem. Biol. 2003, 10, 541.
(23) Lambalot, R. H.; Gehring, A. M.; Flugel, R. S.; Zuber, P.; LaCelle,
M.; Marahiel, M. A.; Reid, R.; Khosla, C.; Walsh, C. T. Chem.
Biol. 1996, 3, 923.
(24) Hubbard, B. K.; Thomas, M. G.; Walsh, C. T. Chem. Biol. 2000,
7, 931.
(25) Pfeifer, V.; Nicholson, G. J.; Ries, J.; Recktenwald, J.; Schefer,
A. B.; Shawky, R. M.; Schroder, J.; Wohlleben, W.; Pelzer, S. J.
Biol. Chem. 2001, 276, 38370.
(26) Luo, L.; Kohli, R. M.; Onishi, M.; Linne, U.; Marahiel, M. A.;
Walsh, C. T. Biochemistry 2002, 41, 9184.
(27) Chen, H.; Tseng, C. C.; Hubbard, B. K.; Walsh, C. T. Proc. Natl.
Acad. Sci. U.S.A. 2001, 98, 14901.
(28) Chen, H.; Walsh, C. T. Chem. Biol. 2001, 8, 301.
(29) Trauger, J. W.; Walsh, C. T. Proc. Natl. Acad. Sci. U.S.A. 2000,
97, 3112.
(30) Luo, L.; Walsh, C. T. Biochemistry 2001, 40, 5329.
(31) Williams, D. H., Bardsley, B. Angew. Chem., Int. Ed. 1999, 38,
1172.
(32) Bischoff, D.; Pelzer, S.; Bister, B.; Nicholson, G. J.; Stockert, S.;
Schirle, M.; Wohlleben, W.; Jung, G.; Sussmuth, R. D. Angew.
Chem., Int. Ed. 2001, 40, 4688.
(33) Bischoff, D.; Pelzer, S.; Holtzel, A.; Nicholson, G. J.; Stockert,
S.; Wohlleben, W.; Jung, G.; Sussmuth, R. D. Angew. Chem.,
Int. Ed. 2001, 40, 1693.
(34) Zerbe, K.; Pylypenko, O.; Vitali, F.; Zhang, W.; Rouset, S.; Heck,
M.; Vrijbloed, J. W.; Bischoff, D.; Bister, B.; Sussmuth, R. D.;
Pelzer, S.; Wohlleben, W.; Robinson, J. A.; Schlichting, I. J. Biol.
Chem. 2002, 277, 47476.
(35) Pylypenko, O.; Vitali, F.; Zerbe, K.; Robinson, J. A.; Schlichting,
I. J. Biol. Chem. 2003, 278, 46727.
(36) Solenberg, P. J.; Matsushima, P.; Stack, D. R.; Wilkie, S. C.;
Thompson, R. C.; Baltz, R. H. Chem. Biol. 1997, 4, 195.
(37) Losey, H. C.; Peczuh, M. W.; Chen, Z.; Eggert, U. S.; Dong, S.
D.; Pelczer, I.; Kahne, D.; Walsh, C. T. Biochemistry 2001, 40,
4745.
(38) Lu, W.; Oberthur, M.; Leimkuhler, C.; Tao, J.; Kahne, D.; Walsh,
C. T. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 4390.
(39) Chen, H.; Thomas, M. G.; Hubbard, B. K.; Losey, H. C.; Walsh,
C. T.; Burkart, M. D. Proc. Natl. Acad. Sci. U.S.A. 2000, 97,
11942.
(40) Anderson, J. S.; Matsuhashi, M.; Haskin, M. A.; Strominger, J.
L. Proc. Natl. Acad. Sci. U.S.A. 1965, 53, 881.
(41) Gale, E. F.; Cundliffe, E.; Reynold, P. E.; Richmond, M. H.;
Waring, M. J. In Molecular Basis of Antibiotic Actions; Wiley
and Sons Ltd.: London, 1981.
(42) Holtje, J. V. Microbiol. Mol. Biol. Rev. 1998, 62, 181.
(43) Bugg, T. D. In Comprehensive Natural Products Chemistry;
Pinto, B. M., Ed.; Elsevier: Oxford, 1999; Vol. 3.
(44) Men, H.; Park, P.; Ge, M.; Walker, S. J. Am. Chem. Soc. 1998,
120, 2484.
(45) Bupp, K.; van Heijenoort, J. J. Bacteriol. 1993, 175, 1841.
(46) Chen, Z.; Eggert, U. S.; Dong, S. D.; Shaw, S. J.; Sun, B.; LaTour,
J. V.; Kahne, D. Tetrahedron 2002, 58, 6585.
(47) Ha, S., Walker, D., Shi, Y., Walker, S. Protein Sci. 2000, 9, 1045.
(48) Yan, Y.; Munshi, S.; Leiting, B.; Anderson, M. S.; Chrzas, J.;
Chen, Z. J. Mol. Biol. 2000, 304, 435.
(49) Schonbrunn, E.; Sack, S.; Eschenburg, S.; Perrakis, A.; Krekel,
F.; Amrhein, N.; Mandelkow, E. Structure 1996, 4, 1065.
(50) Benson, T. E.; Harris, M. S.; Choi, G. H.; Cialdella, J. I.; Herberg,
J. T.; Martin, J. P., Jr.; Baldwin, E. T. Biochemistry 2001, 40,
2340.
(51) Mol, C. D.; Brooun, A.; Dougan, D. R.; Hilgers, M. T.; Tari, L.
W.; Wijnands, R. A.; Knuth, M. W.; McRee, D. E.; Swanson, R.
V. J. Bacteriol. 2003, 185, 4152.
(52) Bertrand, J.; Auger, G.; Fanchon, E.; Marin, L.; Blanot, D.; van
Heijenoort, J.; Dideberg, O. EMBO J. 2000, 16, 3416.
(53) Gordon, E.; Flouret, B.; Chantalat, L.; van Heijenoort, J.;
Mengin-Lecreulx, D.; Dideberg, O. J. Biol. Chem. 2001, 276,
10999.
(54) Fan, C.; Park, I. S.; Walsh, C. T.; Knox, J. R. Biochemistry 1997,
36, 6, 2531.
(55) Yamaguchi, H.; Kato, H.; Hata, Y.; Nishioka, T.; Kimura, A.;
Oda, J.; Katsube, Y. J. Mol. Biol. 1993, 229, 1083.
(56) Goffin, C.; Ghuysen, J. Microbiol. Mol. Biol. Rev. 1998, 62, 1079.
(57) Schiffer, G.; Holtje, J.-V. J. Biol. Chem. 1999, 274, 32031.
(58) Broome-Smith, J. K.; Edelman, A.; Yousif, S.; Spratt, B. G. Eur.
J. Biochem. 1985, 147, 437.

Chemical Reviews, 2005, Vol. 105, No. 2 447


(59) DiGuilmi, A. M.; Dessen, A.; Dideberg, O.; Vernet, T. Curr.
Pharam. Biotechnol. 2002, 3, 63.
(60) Suzuki, H.; Nishimura, Y.; Hirota, Y. Proc. Natl. Acad. Sci.
U.S.A. 1978, 75, 664.
(61) Tamaki, S.; Nakajima, S.; Matsuhashi, M. Proc. Natl. Acad. Sci.
U.S.A. 1977, 74, 5472.
(62) Nakagawa, J.; Tamaki, S.; Tomioka, S.; Matsuhashi, M. J. Biol.
Chem. 1984, 259, 13937.
(63) Nanninga, N. Microbiol. Mol. Biol. Rev. 1998, 62, 110.
(64) Buddelmeijer, N.; Beckwith, J. Curr. Opin. Microbiol. 2002, 5,
553.
(65) Popham, D. L.; Young, K. D. Curr. Opin. Microbiol. 2003, 6, 594.
(66) Arbeloa, A.; Segal, H.; Hugonnet, J.-E.; Josseaume, N.; Dubost,
L.; Brouard, J.-P.; Gutmann, L.; Mengin-Lecreulx, D.; Arthur,
M. J. Bacteriol. 2004, 186, 1221.
(67) DiGuilmi, A. M.; Dessen, A.; Dideberg, O.; Vernet, T. J. Bacteriol.
2003, 185, 4418.
(68) Ye, X. Y.; Lo, M. C.; Brunner, L.; Walker, D.; Kahne, D.; Walker,
S. J. Am. Chem. Soc. 2001, 123, 3155.
(69) VanNieuwenhze, M. S.; Mauldin, S. C.; Zia-Ebrahimi, M.;
Winger, B. E.; Hornback, W. J.; Saha, S. L.; Aikins, J. A.;
Blaszczak, L. C. J. Am. Chem. Soc. 2002, 124, 3656.
(70) Schwartz, B.; Markwalder, J. A.; Wang, Y. J. Am. Chem. Soc.
2001, 123, 11638.
(71) Schwartz, B.; Markwalder, J. A.; Seitz, S. P.; Wang, Y.; Stein,
R. L. Biochemistry 2002, 41, 12552.
(72) Silver, L. L.; Bostian, K. A. Antimicrob. Agents Chemother. 1993,
37, 377.
(73) Silver, L. L. Curr. Opin. Microbiol. 2003, 6, 431.
(74) McCormick, M. H.; McGuire, J. M.; Pittenger, G. E.; Pittenger,
R. C.; Stark, W. M. Antibiot. Annu. 1956, 3, 606.
(75) Perkins, H. R. Biochem. J. 1969, 111, 195.
(76) Chatterjee, A. N.; Perkins, H. R. Biochem. Biophys. Res. Commun. 1966, 24, 489.
(77) Bordet, C.; Perkins, H. R. Biochem. J. 1970, 119, 877.
(78) Sheldrick, G. M.; Jones, P. G.; Kennard, O.; Williams, D. H.;
Smith, G. A. Nature 1978, 271, 223.
(79) Williamson, M. P.; Williams, D. H. J. Am. Chem. Soc. 1981, 103,
6580.
(80) Harris, C. M.; Harris, T. M. J. Am. Chem. Soc. 1982, 104, 4293.
(81) Williams, D. H.; Williamson, M. P.; Butcher, D. W.; Hammond,
S. J. J. Am. Chem. Soc. 1983, 105, 2.
(82) Barna, J. C.; Williams, D. H. Annu. Rev. Microbiol. 1984, 38,
339.
(83) Reynolds, P. Eur. J. Clin. Microbiol. Infect. Dis. 1989, 943.
(84) Ziegler, D. W.; Wolfe, R. N.; McGuire, J. M. Antibiot. Annu. 1956,
612.
(85) Cooper, G. L.; Given, D. B. Vancomycin, A Comprehensive Review
of 30 Years of Clinical Experience; Park Row Publishers: 1986.
(86) Levy, S. B. Sci. Am. 1998, 278, 46.
(87) Arthur, M.; Courvalin, P. Antimicrob. Agents Chemother. 1993,
37, 1563.
(88) Evers, S.; Quintiliani, R., Jr.; Courvalin, P. Microb. Drug Resist.
1996, 2, 219.
(89) Cetinkaya, Y.; Falk, P.; Mayhall, C. G. Clin. Microbiol. Rev. 2000,
13, 686.
(90) Walsh, C. T.; Fisher, S. L.; Park, I. S.; Prahalad, M.; Wu, Z.
Chem. Biol. 1996, 3, 21.
(91) Wright, G. D.; Holman, T. R.; Walsh, C. T. Biochemistry 1993,
32, 5057.
(92) Holman, T. R.; Wu, Z.; Wanner, B. L.; Walsh, C. T. Biochemistry
1994, 33, 4625.
(93) Bugg, T. D.; Wright, G. D.; Dutka-Malen, S.; Arthur, M.;
Courvalin, P.; Walsh, C. T. Biochemistry 1991, 30, 10408.
(94) McComas, C. C.; Crowley, B. M.; Boger, D. L. J. Am. Chem. Soc.
2003, 125, 9314.
(95) Beauregard, D. A.; Williams, D. H.; Gwynn, M. N.; Knowles, D.
J. Antimicrob. Agents Chemother. 1995, 39, 781.
(96) Westwell, M. S.; Gerhard, U.; Williams, D. H. J. Antibiot. 1995,
48, 1292.
(97) Cooper, M. A.; Williams, D. H. Chem. Biol. 1999, 6, 891.
(98) Westwell, M. S.; Bardsley, B.; Dancer, R. J.; Try, A. C.; Williams,
D. H. Chem. Commun. 1996, 6, 589.
(99) Evers, S.; Courvalin, P. J. Bacteriol. 1996, 178, 1302.
(100) Baptista, M.; Rodrigues, P.; Depardieu, F.; Courvalin, P.; Arthur,
M. Mol. Microbiol. 1999, 32, 17.
(101) Arthur, M.; Depardieu, F.; Courvalin, P. Microbiology 1999, 145
(Pt 8), 1849.
(102) Dong, S. D.; Oberthur, M.; Losey, H. C.; Anderson, J. W.; Eggert,
U. S.; Peczuh, M. W.; Walsh, C. T.; Kahne, D. J. Am. Chem.
Soc. 2002, 124, 9064.
(103) McAtee, J. J.; Castle, S. L.; Jin, Q.; Boger, D. L. Bioorg. Med.
Chem. Lett. 2002, 12, 1319.
(104) McComas, C. C.; Crowley, B. M.; Hwang, I.; Boger, D. L. Bioorg.
Med. Chem. Lett. 2003, 13, 2933.
(105) Kerns, R.; Dong, S. D.; Fukuzawa, S.; Carbeck, J.; Kohler, J.;
Silver, L. L.; Kahne, D. J. Am. Chem. Soc. 2000, 122, 12608.
(106) Nagarajan, R.; Schabel, A. A.; Occolowitz, J. L.; Counter, F. T.;
Ott, J. L.; Felty-Duckworth, A. M. J. Antibiot. 1989, 42, 63.

448 Chemical Reviews, 2005, Vol. 105, No. 2


(107) Nagarajan, R. J. Antibiot. 1993, 46, 118.
(108) Sharman, G. J.; Try, A. C.; Dancer, R. J.; Cho, Y. R.; Staroske,
T.; Bardsley, B.; Maguire, A. J.; Coper, M. A.; OBrien, D. P.;
Williams, D. H. J. Am. Chem. Soc. 1997, 119, 12041.
(109) OBrien, D. P.; Entress, R. M. H.; Cooper, M. A.; OBrien, S. W.;
Hopkinson, A.; Williams, D. H. J. Am. Chem. Soc. 1999, 121,
5259.
(110) Rao, J.; Lahiri, J.; Isaacs, L.; Weis, R. M.; Whitesides, G. M.
Science 1998, 280, 708.
(111) Rao, J.; Yan, L.; Lahiri, J.; Whitesides, G. M.; Weis, R. M.;
Warren, H. S. Chem. Biol. 1999, 6, 353.
(112) Allen, N. E.; LeTourneau, D. L.; Hobbs, J. N., Jr.; Thompson,
R. C. Antimicrob. Agents Chemother. 2002, 46, 2344.
(113) Booth, P. M.; Williams, D. H. J. Chem. Soc., Perkin Trans 1 1989,
2335.
(114) Ahrendt, K. A.; Olsen, J. A.; Wakao, M.; Trias, J.; Ellman, J. A.
Bioorg. Med. Chem. Lett. 2003, 13, 1683.
(115) Jain, R. K.; Trias, J.; Ellman, J. A. J. Am. Chem. Soc. 2003,
125, 8740.
(116) Goldman, R. C.; Baizman, E. R.; Longley, C. B.; Branstrom, A.
A. FEMS Microbiol. Lett. 2000, 183, 209.
(117) Roy, R. S.; Yang, P.; Kodali, S.; Xiong, Y.; Kim, R. M.; Griffin,
P. R.; Onishi, R.; Kohler, J.; Silver, L. L.; Chapman, K. Chem.
Biol. 2001, 8, 1095.
(118) Sun, B.; Chen, Z.; Eggert, U. S.; Shaw, S. J.; LaTour, J. V.;
Kahne, D. J. Am. Chem. Soc. 2001, 123, 12722.
(119) Smith, T. L.; Pearson, M. L.; Wilcox, K. R.; Cruz, C.; Lancaster,
M. V.; Robinson-Dunn, B.; Tenover, F. C.; Zervos, M. J.; Band,
J. D.; White, E.; Jarvis, W. R. N. Engl. J. Med. 1999, 340, 493.
(120) Sieradzki, K.; Tomasz, A. J. Bacteriol. 1999, 181, 7566.
(121) Chang, S.; Sievert, D. M.; Hageman, J. C.; Boulton, M. L.;
Tenover, F. C.; Downes, F. P.; Shah, S.; Rudrik, J. T.; Pupp, G.
R.; Brown, W. J.; Cardo, D.; Fridkin, S. K. N. Engl. J. Med. 2003,
348, 1342.
(122) Weigel, L. M.; Clewell, D. B.; Gill, S. R.; Clark, N. C.; McDougal,
L. K.; Flannagan, S. E.; Kolonay, J. F.; Shetty, J.; Killgore, G.
E.; Tenover, F. C. Science 2003, 302, 1569.
(123) Allen, N. E.; Nicas, T. I. FEMS Microbiol. Rev. 2003, 26, 511.

Kahne et al.
(124) Pfaller, M. A.; Marco, F.; Messer, S. A.; Jones, R. N. Diagn.
Microbiol. Infect. Dis. 1998, 30, 251.
(125) Wasilewski, M. 41st ICAAC, late breaking poster, 2001.
(126) Candiani, G.; Abbondi, M.; Borgonovi, M.; Romano, G.; Parenti,
F. J. Antimicrob. Chemother. 1999, 44, 179.
(127) Steiert, S. Cuur. Opin. Investig. Drugs 2002, 3, 229.
(128) Seltzer, E.; Dorr, M. B.; Goldstein, B. P.; Perry, M.; Dowell, J.
A.; Henkel, T. Clin. Infect. Dis. 2003, 37, 1298.
(129) Pace, J. L.; Krause, K.; Johnston, D.; Debabov, D.; Wu, T.;
Farrington, L.; Lane, C.; Higgins, D. L.; Christensen, B.; Judice,
J. K.; Kaniga, K. Antimicrob. Agents Chemother. 2003, 47, 3602.
(130) Losey, H. C.; Jiang, J.; Biggins, J. B.; Oberthur, M.; Ye, X. Y.;
Dong, S. D.; Kahne, D.; Thorson, J. S.; Walsh, C. T. Chem. Biol.
2002, 9, 1305.
(131) Fu, X.; Albermann, C.; Jiang, J.; Liao, J.; Zhang, C.; Thorson, J.
S. Nat. Biotechnol. 2003, 21, 1467.
(132) Losey, H. C.; Jiang, J.; Biggins, J. B.; Oberthur, M.; Ye, X. Y.;
Dong, S. D.; Kahne, D.; Thorson, J. S.; Walsh, C. T. Chem. Biol.
2002, 9, 1305.
(133) Mulichak, A. M.; Losey, H. C.; Walsh, C. T.; Garavito, R. M.
Structure (Camb) 2001, 9, 547.
(134) Mulichak, A. M.; Losey, H. C.; Lu, W.; Wawrzak, Z.; Walsh, C.
T.; Garavito, R. M. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 9238.
(135) Mulichak, A. M.; Lu, W.; Losey, H. C.; Walsh, C. T.; Garavito,
R. M. Biochemistry 2004, 43, 5170.
(136) S. aureus PBP2, the major transglycosylase in Staph, was
recently overexpressed, purified, and characterized. Barrett, D.;
Leimkuhler, C.; Chen, L.; Walker, D.; Kahne, D.; Walker, S. J.
Bacteriol., in press.
(137) An assay to monitor inhibition of the major transglycosylase in
S. aureus has been developed. Various glycopeptides, including
damaged chlorobiphenyl vancomycin and damaged dalbavancin,
were shown to inhibit S. aureus PBP2. Leimkuhler, C.; Chen,
L.; Barrett, D.; Panzone, G.; Sun, B.; Falcone, B.; Oberthur, M.;
Donadio, S.; Walker, S.; Kahne, D. J. Am. Chem. Soc., in press.

CR030103A

Chem. Rev. 2005, 105, 449475

449

Chemistry and Biology of Ramoplanin: A Lipoglycodepsipeptide with Potent


Antibiotic Activity
Suzanne Walker,*, Lan Chen, Yanan Hu, Yosup Rew, Dongwoo Shin, and Dale L. Boger*,
The Department of Microbiology & Molecular Genetics, Harvard Medical School, Boston, Massachusetts 02115, The Department of Chemistry and
The Skaggs Institute for Chemical Biology, The Scripps Research Institute, La Jolla, California 92037, and The Department of Chemistry,
Princeton University, Princeton, New Jersey 08544
Received July 12, 2004

Contents
1. Introduction
2. Ramoplanin Basics
2.1. Structural Overview
2.2. Antimicrobial Activity
2.3. Clinical Status
3. Mechanism of Action of RamoplaninsEarly Work
3.1. Cellular Targets of Antibiotics
3.2. Peptidoglycan Structure and Biosynthesis
3.3. MurG Is Proposed as the Target of
Ramoplanin
3.4. Ramoplanin Is Shown To Bind to an
Intermediate in Peptidoglycan Biosynthesis
3.5. Ramoplanin Is Proposed To Block the
Transglycosylation Step of Peptidoglycan
Biosynthesis
4. Mechanism of Action of RamoplaninsRecent
Work
4.1. Technical Advances in the Study of
Peptidoglycan-Synthesizing Enzymes
4.2. Expected Inhibition Kinetics for Substrate
Binders
4.3. Inhibition Kinetics of Ramoplanin
4.3.1. Transglycosylase Inhibition
4.3.2. MurG Inhibition
4.4. Evaluating the Proposed Cellular Targets of
Ramoplanin
5. Molecular Recognition by Ramoplanin
5.1. Problem of Fibril Formation
5.2. Comparison of Substrate-Binding Affinities
5.3. Structural Studies on Ramoplanin and
Ramoplanin Complexes
6. Total Synthesis of Ramoplanin and Key
Analogues
6.1. Preparation of Key Amino Acids
6.1.1. aThr (allo-Threonines)
6.1.2 HAsn (L-threo--Hydroxyasparagine)
6.2. Total Synthesis of the Ramoplanin A2 and
Ramoplanose Aglycon
6.3. Total Synthesis and Structure of the
Ramoplanin A1 and A3 Aglycons

449
452
452
453
454
454
454
454
455
456
456
457
457
457
458
458
459
459
460
460
461
462
463
463
463
463
464
467

* To whom correspondence should be addressed. E-mail:


suzanne_walker@hms.harvard.edu and boger@scripps.edu.
Harvard Medical School.
Princeton University.
The Scripps Research Institute.

6.4. [N-Acetyl-Asn1]ramoplanin Aglycon


6.5. Total Synthesis of Ramoplanin Amide
Analogues
6.6. Solid-Phase Synthesis of a Simplified
Analogue
7. Degradation and Semisynthetic Studies
7.1. Ramoplanin Aglycons
7.2. Depsipeptide Hydrolysis
7.3. Lipid Side-Chain Reduction
7.4. Orn4 and Orn10 Derivatization
7.5. Lipid Side-Chain Replacement
7.6. Summary
8. StructureActivity Studies on Ramoplanin and Its
Synthetic and Semisynthetic Analogues
8.1. Antimicrobial Activity of Key Derivatives and
Analogues
8.2. Mechanistic Analysis of Ramoplanin
AnaloguessThe Path Forward
9. Conclusion
10. Abbreviations
11. References

467
468
469
470
470
470
470
470
471
471
471
471
472
473
473
474

1. Introduction
Before the introduction of antibiotics in the 1940s
and 1950s, patients with bacteremiasbacteria in the
blood streamshad almost no chance of survival.1,2
Antibiotics (Figure 1) were hailed as miracle drugs
because they rapidly cured infections that would
otherwise have proven fatal. In the belief that
antibiotics, vaccination, and modern sanitation methods had defeated infectious disease, the Surgeon
General declared in 1970 that the United States was
ready to close the book on infectious disease as a
major health threat. Twenty-five years later, hospitalacquired (nosocomial) infections cost several billion
dollars and contribute to 100 000 deaths annually.3
According to one estimate, 20% of patients admitted
to hospitals have or will develop an infection and 70%
of the bacteria that give rise to these infections are
resistant to at least one of the main antimicrobial
agents used to fight infection.4
More than one-third of nosocomial infections are
caused by three Gram-positive pathogenssStaphylococcus aureus, coagulase-negative staphylococci, and
enterococci. Acquired antibiotic resistance in these
organisms is a major concern. Clinical isolates of

10.1021/cr030106n CCC: $53.50 2005 American Chemical Society


Published on Web 01/15/2005

450 Chemical Reviews, 2005, Vol. 105, No. 2

Suzanne Walker (above right) received her Ph.D. degree from Princeton
University in 1992. She joined the Department of Chemistry at Princeton
as a faculty member in 1995, where she initiated a program to study
peptidoglycan biosynthesis and its inhibition. In 2004 she moved to the
Department of Microbiology & Molecular Genetics at Harvard Medical
School. She is interested in bacterial cell growth and division as well as
the mechanism of action of unusual antibiotics.
Lan Chen (above left), born in Wuhan, China, obtained her M.S. degree
in Chemistry from Wuhan University in 1993. She received her Ph.D.
degree in 2000 from Rutgers University, where she studied mechanistic
enzymology under the supervision of Professor W. Phillip Huskey. In 2000
she joined the laboratory of Professor Suzanne Walker as a postdoctoral
associate, where she studied the kinetic mechanism of E. coli MurG,
developed methods to study bacterial transglycosylases, and helped
characterize the mechanism of action of ramoplanin. She joined Cubist
(Lexington, MA) as a Research Scientist in 2004.

Yanan Hu was born in Dalian, China, and obtained her M.S. degree in
Chemistry from Tsinghua University in Beijing in 2000. She joined the
laboratory of Professor Suzanne Walker as a graduate research associate
at Princeton University in 2000. She solved the X-ray structure of a cocomplex of E. coli MurG with UDP-GlcNAc bound, developed a highthroughput screen for MurG inhibitors, and helped characterize the
mechanism of action of ramoplanin. After receiving her Ph.D. degree in
2004, she joined Colgate-Palmolive (Piscataway, NJ) as a Research
Scientist.

S. aureus, which infects burns, skin, and surgical


wounds, are typically resistant to a range of antibiotics, including methicillin.5 The glycopeptide antibiotic
vancomycin is used to treat infections caused by drugresistant S. aureus strains and other Gram-positive
pathogens.6 Unfortunately, vancomycin resistance
has become increasingly common in enterococci and
is now beginning to spread to other organisms.7,8 For
example, a methicillin-resistant S. aureus strain
displaying high-level resistance to vancomycin was
isolated recently from a dialysis patient in Michigan.9,10 This isolate had apparently acquired the

Walker et al.

Yosup Rew, born in 1968 in Andong, South Korea, received his B.Sc.
and M.Sc. degrees in Chemistry from the Seoul National University, South
Korea, in 1990 and 1992, respectively. After working as a research scientist
of the new fungicide discovery team in the agrochemical division of LG
Chemical Ltd., Daejeon, South Korea. for 5 years, he continued his studies
in chemistry at the University of California, San Diego. where he received
his Ph.D. degree in 2002 under the supervision of the late Professor
Murray Goodman in the field of peptide chemistry. Following this he was
a postdoctoral fellow at The Scripps Research Institute working with
Professor Boger and involved in the total synthesis of key analogues of
ramoplanin. He joined Amgen (San Francisco) in 2004.

Dongwoo Shin (above left), born in 1971 in Seoul, South Korea, obtained
his B.Sc. degree in Chemistry from Hanyang University, Seoul, South
Korea in 1996. He received his Ph.D. degree (2002) at The University of
Illinois at Chicago, where he worked on the design and synthesis of
protease inhibitors under supervision of Professor Arun K. Ghosh. In 2002
he joined the laboratory of Professor Dale L. Boger as a postdoctoral
associate. His current interests cover the total synthesis and structure
activity studies of bioactive molecules.
Dale L. Boger (above right) received his Ph.D. degree from Harvard
University in 1980. He joined the Department of Medicinal Chemistry of
the University of Kansas, moved to Purdue University in 1986, and joined
the newly founded Department of Chemistry at The Scripps Research
Institute in 1990. His research interests include the total synthesis and
structural exploration of biologically active natural products.

genes that confer vancomycin resistance from a


coexisting E. faecalis strain and was now harboring
these genes on a large multiresistance conjugative
plasmid. Additional genes on the plasmid encoded
resistance to trimethoprim, aminoglycosides, -lactams, and disinfectants.10
As the foregoing discussion emphasizes, there is a
clear need for new antibiotics to treat drug-resistant
bacterial infections. Antibiotics that operate by new
mechanisms or belong to new structural classes are
particularly attractive because they are less likely to
show cross-resistance with existing antibiotics. Un-

Chemistry and Biology of Ramoplanin

Chemical Reviews, 2005, Vol. 105, No. 2 451

Figure 1. Structures of selected natural products.

fortunately, between 1962 and 2000 only one such


antibiotic, linezolid, was introduced clinically, although others have since been introduced. All of the
other agents that entered the market between 1962
and 2000 were analogues of existing drugs.4,11 Collaborations between industrial and academic scientists may be required to meet the challenges involved
in discovering, understanding, and developing new

antimicrobial agents. Here we review an antibiotic


that belongs to a new structural class of compounds
and has not shown any cross-resistance to other
antibacterial agents. This antibiotic, ramoplanin
(Figure 2), is currently in Phase III clinical trials for
the prevention of vancomycin-resistant enterococcal
infections in hospitalized patients, and it possesses
some potentially important advantages over many

452 Chemical Reviews, 2005, Vol. 105, No. 2

Walker et al.

either in vitro or in vivo. However, ramoplanin has


not yet been developed for systemic use, and its
potential applications, albeit important, are currently
limited.14 A better understanding of the mechanism
of action of ramoplanin, combined with the ability to
prepare analogues, could lead to new derivatives with
broader utility.
This review will present our understanding of the
mechanism of action of ramoplanin. An introduction
to the primary structure of ramoplanin, its spectrum
of activity, and current clinical indications and
progress will be followed by an overview of the early
mechanistic investigations of this antibiotic. Technical limitations restricted the studies that were initially conducted on ramoplanin, and its mechanism
of action continues to be refined and revised since
the early investigations. Recent chemical and biochemical advances that have made it possible to
probe ramoplanins mechanism of action in greater
detail will be described, and the current mechanistic
understanding will be presented. This will be followed by a description of the synthetic studies that
have been conducted on ramoplanin. The ability to
synthesize ramoplanin and analogues is critical to
addressing the relationship between structure and
activity, which in turn is central to developing better
antibiotics. The review will conclude with an assessment of where we are with respect to understanding
ramoplanin and where we need to go.

2. Ramoplanin Basics
2.1. Structural Overview

Figure 2. (a) Structures of the ramoplanins. (b) Structures


of ramoplanin A2 and synthetic analogues described in the
text.

other antibiotics.12,13 Chief among them is that


resistance to ramoplanin does not develop readily

The discovery, structure elucidation, and biosynthesis of ramoplanin have been reviewed recently by
McCafferty et al., and only a few salient features will
be repeated here.15 Ramoplanin factors A1 (1), A2 (2),
and A3 (3) were discovered in 1984 from a fermentation broth of Actinoplanes and identified as cell wall
active antibiotics by Biosearch Italia (then Gruppo
LePetit).16,17 Ramoplanins A1-A3 consist of a 49membered macrocyclic depsipeptide containing 17
amino acids joined by a lactone linkage between a
beta hydroxy group on amino acid 2 (HAsn2) and the
carboxyl terminus of amino acid 17 (Chp17). Ramoplanin is produced by nonribosomal peptide synthesis
and, like many such natural products, contains a
mixture of L and D amino acids (nine L, seven D) as
well as several nonproteinogenic side chains, including ornithine (Orn), hydroxyphenylglycine (Hpg), and
chlorohydroxyphenylglycine (Chp) in addition to
-hydroxyasparagine (-OH-Asn, HAsn). Ten of
the 16 residues in the ramoplanin macrocycle are
-branched: L--HAsn2, D-Hpg3, D-allo-Thr5, L-Hpg6,
D-Hpg7, L-allo-Thr8, L-Hpg11, D-allo-Thr12, L-Hpg13,
and L-Chp17. The first amino acid in the depsipeptide
is acylated at the amino terminus with a lipid
substituent. Ramoplanins A1-A3 differ in minor
ways in the length and structure of this appended
lipid substituent; however, all three have R,,,unsaturated chains. The lipid chains of ramoplanins
A1-A3 were originally assigned the (2Z,4Z) stereochemistry around these double bonds,18 but all three
have been reassigned as (2Z,4E) based on work by

Chemistry and Biology of Ramoplanin

Chemical Reviews, 2005, Vol. 105, No. 2 453

Kurz and Guba (A2)19 and Boger and co-workers20


(see section 6.3). Ramoplanin A2 is the compound
now in clinical trials because it can be produced in
much larger quantities than factors A1 and A3;
however, the antimicrobial activities of the lipid
variants are virtually indistinguishable. The ramoplanins also contain a saccharide moiety attached by
an R-glycosidic linkage to the phenol of amino acid
11. In ramoplanins A1-A3 the saccharide moiety is
an R-1,2-dimannosyl group; however, ramoplanin
variants in which the terminal mannose has been
removed or the saccharide consists of a branched
trimannosyl group (ramoplanose, 4) have been reported.21 The activities of the saccharide variants
appear to be similar to those of ramoplanins A1-A3.

nonpolar surfaces such as the plastic plates that are


usually used for MIC measurements. BSA reduces
this nonspecific binding, thereby increasing the
amount of ramoplanin in solution.
Ramoplanin has been compared to vancomycin in
a number of different in vitro studies and consistently
has been found to be a more potent antibiotic on a
molar basis regardless of the method of testing
used.15,30 Ramoplanin is also bactericidal at concentrations close to its MIC, which may provide it some
advantages over vancomycin, which is bacteriostatic
at concentrations near its MIC. Finally, ramoplanin
retains full activity against vancomycin-resistant
enterococcal strains and methicillin-resistant staphylococcal strains,27,28 and significant ramoplanin
resistance has not been observed to date, despite
efforts to elicit it in the laboratory. The mechanism
of action of ramoplanin makes the spontaneous
development of high-level resistance improbable.
However, as with other natural product antibiotics,
resistance in clinically relevant pathogens could
develop to ramoplanin by the horizontal transfer of
genetic information from the producing organism to
other microorganisms. The likelihood of this occurring rapidly depends on usage patterns of ramoplanin
and related antibiotics, the biology of the producing
microorganisms, and the mechanism of resistance in
the producer organisms. It should be noted that the
mechanism by which the ramoplanin producer resists
the effects of ramoplanin has not yet been elucidated.
It should also be noted that only one confirmed
structurally related compound has been reported,
enduracidin (Figure 3), and this compound is not
used clinically or commercially.31 Another potentially
related antibiotic, janiemycin, has also been reported
but not yet structurally characterized.32

2.2. Antimicrobial Activity


Ramoplanin is active against a wide range of
Gram-positive organisms, including many different
species of Staphylococcus, Enterococcus, and Bacillus
(Table 1). Like vancomycin and many other antibiotics, ramoplanin displays no activity against Gramnegative bacteria,22,23 presumably because it cannot penetrate the outer membrane. In general, ramoplanin has MICs (MIC ) minimum inhibitory
concentration, defined as the lowest concentration at
which bacterial growth ceases) below 1 g/mL against
most Gram-positive strains, although the measured
MICs can vary widely based upon the method used
to test growth inhibition.24-28 It has been reported
in two separate studies that the addition of BSA to
the antibiotic stock solution or to plate wells containing bacteria that are used to carry out the measurements results in values up to 30-fold lower (more
active) than without BSA.29 This is consistent with
the observation that ramoplanin tends to stick to

Table 1. Minimum Inhibitory Concentrations (MICs) of Ramoplanin and Its Analogues Against Different
Bacterial Strains (g/mL; compound structures are shown in Figure 2b)
compound

S. aureusa

S. aureusb

2
5b
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28

0.5-1.56
0.25-0.78
>128
>128
0.39
6.25-15
>35-50
>100

0.125

S. aureusc

E. faeciume

E. faecalisf

B. subtilisd

0.1

0.1

0.03
0.03
64

0.3
30
25

0.1
15
15

15

20

0.8
50

0.8
50

2
12.5-80

>128
4

8
4
0.125
0.25
0.25
1
0.06
0.06

a
S. aureus ATCC25923, refs 109 and 110. b S. aureus Smith SA819, ref 111. c S. aureus Tour, ref 112.
ref 30. e E. faecium L19624, refs 93 and 113. f E. faecalis Z9212, refs 36 and 113.

0.25
4
16
d

B. subtilis ATCC 8037,

454 Chemical Reviews, 2005, Vol. 105, No. 2

Figure 3. Structures of the enduracidins.

2.3. Clinical Status


Ramoplanin has not yet been developed for parenteral use due, inter alia, to problems associated with
its stability in the blood stream.33 However, it is
currently in phase III clinical trials for the prevention
of systemic nosocomial infections caused by vancomycin-resistant enterococci (VRE). Certain groups of
patients, including those undergoing abdominal surgery and those receiving chemotherapy, are at high
risk of developing systemic VRE infections in the
hospital. It is believed that the GI tracts of many of
these patients are colonized with VRE (possibly
acquired in the hospital from other patients or
contaminated instruments, etc.) and that damage to
the intestinal mucosa caused by surgery or certain
drugs can lead to systemic VRE infections. Therefore,
eradicating VRE from the GI tracts of these patients
prior to treatment may reduce the incidence of
systemic VRE infections. Early clinical results are
promising, and because the incidence of nosocomial
VRE infections is increasing, the FDA has assigned
ramoplanin Fast Track status for this indication.14
In addition to its use for the prevention of blood
stream VRE infections, ramoplanin is also in clinical
trials for the treatment of Clostridium difficileassociated diarrhea (CDAD).3 C. difficile-associated
diarrhea affects over 400 000 patients a year, and
strains resistant to many commonly used therapeutic
agents have appeared. As a result, the FDA recently
designated this application of ramoplanin for Fast
Track status as well.

3. Mechanism of Action of RamoplaninsEarly


Work
3.1. Cellular Targets of Antibiotics
The majority of antibiotics act by inhibiting DNA
synthesis, RNA synthesis, protein synthesis, or peptidoglycan synthesis.34 DNA, RNA, and proteins are
found in eukaryotic cells as well as prokaryotes, but
peptidoglycan is unique to bacteria.35 Peptidoglycan
is a polymeric mesh that forms layers around bacterial cell membranes. One of its primary functions is

Walker et al.

to stabilize bacterial membranes so that they do not


rupture as the osmotic pressure within the cells
fluctuates.36 Both the chemical structure of peptidoglycan and the process by which it is made are highly
conserved in bacteria, even across genera that occupy
very different evolutionary niches.35 Microorganisms
produce a range of natural products that kill other
microorganisms by blocking peptidoglycan biosynthesis.34 Because there are no analogous metabolic
pathways in mammalian cells, many of these natural
products have proven to be safe as well as efficacious
antibiotics. Because peptidoglycan biosynthesis is
regarded as an attractive pathway with which to
interfere, many research groups in academia and
industry have established programs to identify inhibitors of this pathway.37,38 Ramoplanin was discovered in one such program. An overview of peptidoglycan structure and biosynthesis is provided in the
following section to provide a context for discussing
the current understanding of the mechanism of
action of ramoplanin.

3.2. Peptidoglycan Structure and Biosynthesis


Peptidoglycan is comprised of linear chains of a
repeating disaccharide held together by cross-links
between peptides appended to one of the sugars of
the disaccharide.39-42 The repeating disaccharide unit
consists of N-acetyl glucosamine attached via a
-glycosidic linkage to the C4 alcohol of an N-acetyl
muramic acid derivative. Muramic acid, a sugar
found only in bacteria, is identical to glucose except
that the C3 hydroxyl is replaced by a lactic acid unit.
The peptide chain that is involved in cross-linking
the glycan strands is attached via its N-terminus to
the lactate moiety. Prior to cross-linking, the peptide
typically consists of a pentapeptide terminating in
D-Ala-D-Ala. There are strain-dependent variations
in the composition of the peptides, but all contain a
nucleophilic group that attacks the D-Ala-D-Ala peptide bond to form cross-links.39-42
Peptidoglycan is synthesized in three stages (Figure 4), the first of which takes place in the cytoplasm
and involves the conversion of UDP-N-acetyl glucosamine to UDP-N-acetyl muramic acid pentapeptide by a series of seven enzymes.43-54 The first two,
MurA and MurB, convert UDP-GlcNAc to UDPMurNAc; the rest are ligases involved in the assembly of the peptide chain. The second stage of
peptidoglycan biosynthesis, which takes place on the
cytoplasmic surface of the bacterial membrane, commences when MraY, an enzyme containing several
membrane-spanning regions, catalyzes a pyrophosphate exchange reaction in which membrane-bound
undecaprenyl phosphate attacks the R-phosphate of
UDP-MurNAc pentapeptide to form a lipid-anchored
MurNAc pentapeptide known as Lipid I with release
of UMP.43,55,56 A membrane-associated glycosyltransferase, MurG, then catalyzes the transfer GlcNAc
from UDP to the C4 hydroxyl of the MurNAc sugar
of Lipid I to form Lipid II.40,57-59 Once Lipid II is
made, the disaccharide is somehow transported from
the cytoplasmic surface of the bacterial membrane
to the external surface where it is polymerized and
cross-linked in the final stage of peptidoglycan bio-

Chemistry and Biology of Ramoplanin

Chemical Reviews, 2005, Vol. 105, No. 2 455

Figure 4. Three stages of peptidoglycan biosynthesis.

synthesis.39,60 The enzymes responsible for polymerizing the GlcNAc-MurNAc disaccharide to form the
glycan chains of peptidoglycan are known as transglycosylases; the enzymes involved in cross-linking
the glycan chains are known as transpeptidases.39
Bacteria contain several transglycosylases and transpeptidases that play different roles in cell growth and
division.61 Most, but not all, of the transglycosylases
are found as domains in bifunctional enzymes that
also contain transpeptidase domains. It is believed
that the bulk of peptidoglycan biosynthesis is carried
out by a subset of the transglycosylases.62,63 In E. coli,
for example, the major transglycosylases are contained in the bifunctional enzymes PBP1a and PBP1b,
with PBP1b believed to be responsible for 85% of
glycan strand synthesis.64,65 Homologous enzymes in
other bacterial strains can be readily identified based
on sequence similarities and are assumed to play
comparable roles.62
Several antibiotics in clinical use block key steps
in peptidoglycan biosynthesis.34 For example, fosfomycin, used to treat urinary track infections, inhibits
MurA, which catalyzes the first committed step in
the biosynthetic pathway.66 Vancomycin and other
glycopeptide antibiotics block glycan polymerization
and cross-linking by binding to the D-Ala-D-Ala
dipeptide terminus of Lipid II and nascent peptidoglycan.6 Penicillin and other -lactams covalently
modify the active site of the enzymes involved in
transpeptidation.67,68

3.3. MurG Is Proposed as the Target of


Ramoplanin
Ramoplanin was discovered by scientists at Biosearch Italia in the course of screening fermentation
products for compounds that killed a methicillinresistant S. aureus strain without affecting a variant
lacking a cell wall and grown under hypotonic conditions.16,17 This screen was designed to identify com-

pounds that block peptidoglycan biosynthesis. Experiments probing inhibition of macromolecular


synthesis showed that ramoplanin blocks the incorporation of radiolabeled precursors into peptidoglycan
without affecting the incorporation of radiolabeled
building blocks into DNA, RNA, or protein, thereby
confirming that ramoplanin inhibits cell wall synthesis. The challenge then became identifying the
step that ramoplanin affects. At that time, identifying
the site of inhibition of a compound that blocks
peptidoglycan synthesis was a daunting challenge.
While the structures of all the intermediates in the
pathway were known, most of the enzymes involved
in the individual chemical transformations had not
been isolated or characterized, and, with few exceptions, direct assays for the enzymes did not exist.
In 1990, Somner and Reynolds reported detailed
investigations of the mechanism of action of ramoplanin.69,70 They first showed that bacteria treated with
ramoplanin accumulate UDP-MurNAc-pentapeptide,
implying that ramoplanin blocks one of the membraneassociated steps that takes place after the formation
of the final cytoplasmic intermediate. They then tried
to determine which membrane-associated step ramoplanin affects. In 1990, two related assays existed
to probe the membrane-associated steps of peptidoglycan synthesis. One assay involved monitoring the
conversion of UDP-MurNAc pentapeptide to nascent
peptidoglycan using washed, particulate bacterial
membranes as a source of the required membranebound enzymes, including MraY, MurG, and the
transglycosylases/transpeptidases.71,72 The second
assay involved monitoring the conversion of UDPMurNAc pentapeptide to mature peptidoglycan in
bacterial cells rendered membrane permeable with
an organic solvent such as toluene.73 Methods to
isolate UDP-MurNAc pentapeptide from bacterial
cells, combined with techniques to separate Lipid I/II
from peptidoglycan and to distinguish cross-linked

456 Chemical Reviews, 2005, Vol. 105, No. 2

Figure 5. Schematic of the membrane assay used to


determine the site of inhibition of ramoplanin.

from un-cross-linked peptidoglycan, had enabled the


development of these assays.71,72 By using isotopically
labeled UDP-MurNAc pentapeptide and/or UDPGlcNAc, Somner and Reynolds were able to follow
the formation of various intermediates along the
pathway and monitor changes in product formation
upon addition of ramoplanin and other antibiotics.
Somner and Reynolds noted that addition of ramoplanin to permeabilized bacterial cells supplemented
with radiolabeled UDP-MurNAc-pentapeptide prevented incorporation of radiolabel into both un-crosslinked and cross-linked peptidoglycan. Un-crosslinked peptidoglycan was formed in the presence of
an antibiotic known to block only the cross-linking
step of peptidoglycan biosynthesis. Therefore, Somner
and Reynolds concluded that ramoplanin acts before
the transpeptidation step. They then showed that
ramoplanin blocks the formation of Lipid II but does
not prevent the formation of Lipid I. Thus, Somner
and Reynolds proposed that ramoplanin acts by
blocking peptidoglycan biosynthesis at the step catalyzed by the glycosyltransferase, MurG (Figure 5).
On the basis of preliminary experiments that were
mentioned but not described in the paper, Somner
and Reynolds also proposed that ramoplanin inhibits
its target by binding to and sequestering Lipid I.
At the time of their experiments Somner and
Reynolds were missing an important piece of information about MurG, namely, that it is located on the
intracellular surface of the cytoplasmic membrane. 74
Therefore, to reach MurG and/or its substrate, ramoplanin would have to pass through the cell membrane. Somner and Reynolds observed in 1990 that
ramoplanin is unlikely to penetrate the cytoplasmic
membrane because of its size and polar nature, but
they did not consider alternative mechanisms of
action because the cellular location of MurG had not
been established. Moreover, the biochemical results
were clear.
The hypothesis that ramoplanin inhibits peptidoglycan biosynthesis at the MurG step became widely
accepted over the next decade. Although Bupp et al.
reported in 1993 that MurG is located on the intracellular surface of the cytoplasmic membrane,74 this
finding did not initially prompt a reexamination of
ramoplanins mechanism. In fact, tools to test the
hypothesis that ramoplanin inhibits MurG by binding
to Lipid I were still not available. Lipid I, the MurG
substrate that ramoplanin was proposed to bind, is
present in miniscule quantities in bacterial cells,
making isolation from natural sources all but impos-

Walker et al.

sible.75,76 Because Lipid I could not be obtained, there


was no direct assay to monitor MurG activity and
no straightforward way to assess whether ramoplanin binds to Lipid I. Although Lipid II is found in
somewhat larger amounts than Lipid I, it also
presented major challenges for isolation. Furthermore, natural Lipid I and Lipid II both contain a 55carbon undecaprenyl chain that renders them insoluble in water and thus difficult to manipulate.
Until these issues of substrate availability and physical properties were addressed, it was not feasible to
carry out detailed investigations of how ramoplanin
inhibits individual membrane-associated enzymes of
peptidoglycan synthesis.

3.4. Ramoplanin Is Shown To Bind to an


Intermediate in Peptidoglycan Biosynthesis
The first direct evidence that ramoplanin is a
substrate binder was reported by Brotz et al. in
1998.77,78 Brotz, Sahl, and co-workers were studying
several lantibiotics, including nisin and epidermin
(Figure 1). Lantibiotics are ribosomally synthesized
peptides that are posttranslationally dehydrated at
serine and threonine side chains.79,80 They then
undergo a series of enzyme-catalyzed cyclizations in
which cysteine residues attack the dehydroamino
acids to form thioether-linked macrocycles. Lantibiotics display a range of mechanisms of action, but
many appear to target Lipid II and other molecules
presented on bacterial cell surfaces (the lantibiotics
are reviewed by van der Donk in this issue). Brotz
et al. showed that ramoplanin as well as the lantibiotics nisin and epidermin comigrate with Lipid II
upon thin-layer chromatography, indicating that all
three molecules interact with this peptidoglycan
intermediate.78 They also reported that nisin and
epidermin block the formation of Lipid II in a
particulate membrane assay similar to that employed
by Somner and Reynolds in their study of ramoplanin. Thus, like ramoplanin, both nisin and epidermin
can inhibit the MurG step of peptidoglycan biosynthesis. Brotz et al. did not conclude, however, that
nisin and epidermin function in cells by inhibiting
MurG because other information pointed to a different mechanism of action for these lantibiotics. By
showing that nisin and epidermin behave similarly
to ramoplanin with respect to MurG inhibition in a
particulate membrane assay and yet are presumed
not to inhibit MurG in cells, Brotz et al. thus provided
the basis for questioning the interpretation that
ramoplanin inhibits the MurG in a cellular context.

3.5. Ramoplanin Is Proposed To Block the


Transglycosylation Step of Peptidoglycan
Biosynthesis
The possibility that ramoplanin functions by inhibiting a different cellular target than MurG was
addressed in 2000 by Lo et al., who noted that the
original mechanistic investigations on the antibiotic
were not conclusive.81 Assays involving the use of a
precursor substrate to monitor flux through several
steps in a pathway can provide inaccurate information about cellular targets if more than one step in

Chemistry and Biology of Ramoplanin

the pathway is inhibited by a particular compound.


Under these circumstances, such vectorial assays
report only the first step to be inhibited, which may
not be the relevant step in a cellular context. In the
case of a proposed substrate binder, there is special
cause for concern because the substrates along a
particular pathway often bear a close resemblance
to one another. Furthermore, in the case of peptidoglycan biosynthesis, it is necessary to carry out these
vectorial assays under conditions in which the polar
precursor substrates can access enzymes that are
normally found on the internal surface of the bacterial membrane. Thus, the integrity of the membrane
must be disrupted by organic solvents, detergents,
or mechanical means for the assays to proceed, and
so a major constraint on the mechanism of actions
the membrane barriersis removed. Lo et al. speculated that in the presence of an intact membrane
barrier, ramoplanin might block the polymerization
of Lipid II, which is catalyzed by bacterial transglycosylases located on the external surface of the
bacterial membrane, instead of the synthesis of Lipid
II, which is catalyzed by an enzyme located on the
internal surface of the bacterial membrane. Inhibition of the transglycosylation step would not have
been observed under assay conditions in which the
formation of Lipid II is blocked.
The hypothesis that ramoplanin can block transglycosylation was tested by Lo et al. using a modified
version of a particulate membrane assay in which
UDP-MurNAc pentapeptide and radiolabeled UDPGlcNAc were added as precursor substrates.81,82 In
this modified assay, Triton X-100 was added to the
bacterial membranes, causing inhibition of the bacterial transglycosylases and allowing radiolabeled Lipid
II to accumulate. The addition of a Triton X-100
scavenger led to reactivation of the transglycosylases
and concomitant formation of peptidoglycan. This
modified assay permits one to add inhibitors after
Lipid II is formed and thus evaluate steps after the
formation of this intermediate.82 Ramoplanin was
found to block transglycosylation in this assay,
implicating a second target for this antibiotic. Thus,
depending on the assay conditions, ramoplanin is
capable of inhibiting either MurG or the bacterial
transglycosylases. Lo et al. suggested that the bacterial transglycosylases were likely to be the primary
targets of ramoplanin because they are extracellular
and would be encountered first by ramoplanin. The
validity of this argument depends on issues such as
whether transglycosylase inhibition is comparable to
or better than MurG inhibition and whether ramoplanin is able to penetrate cell membranes. The
former issue is addressed in the sections below.

4. Mechanism of Action of RamoplaninsRecent


Work
4.1. Technical Advances in the Study of
Peptidoglycan-Synthesizing Enzymes
The development of better methods to obtain
natural Lipid I and II substrates and analogues has
made it possible to assay MurG and the bacterial
transglycosylases directly, enabling in turn detailed

Chemical Reviews, 2005, Vol. 105, No. 2 457

studies of ramoplanin inhibition. For example, in


1997 Auger et al. reported a method to convert UDPMurNAc pentapeptide, which can be isolated in large
quantities from natural sources, to a heptaprenyl
Lipid I analogue, potentially enabling the study of
MurG.83 Shortly thereafter Men et al. reported the
chemical synthesis of a Lipid I analogue (29, Table
2) and its use in a biotin capture assay to monitor
MurG activity.84 Two total syntheses of natural Lipid
I containing the 55-carbon undecaprenyl chain were
reported in 2001.85,86 E. coli MurG was purified and
characterized using synthetic Lipid I analogues,87,88
and several different fluorescence-based and radiometric assays to monitor enzymatic activity using
synthetic substrate analogues have been reported.84,87-94 Moreover, access to purified MurG has
made possible the enzymatic conversion of synthetic
Lipid I to Lipid II. The first synthesis of a Lipid II
analogue containing a 10-carbon lipid chain (30,
Table 2) was reported in 2000.81 This accomplishment
was followed in 2001 by the conversion of synthetic
Lipid I containing the intact 55-carbon chain to Lipid
II using MurG to form the glycosidic linkage.85 This
synthesis of Lipid II was followed by two alternative
routes to the compound in which the glycosidic bond
was formed by chemical methods.95,96 Finally, in 2003
an efficient enzymatic strategy to produce quantities
of Lipid II and analogues from UDP-MurNAc pentapeptide using bacterial membranes was reported.97
Methods to study purified transglycosylases have
followed developments in Lipid II synthesis, enabling
the kinetic analysis of a variety of antibiotics proposed to block transglycosylation.95,98,99

4.2. Expected Inhibition Kinetics for Substrate


Binders
The availability of substrates and assays to monitor
MurG and the bacterial transglycosylases has made
it possible to examine how ramoplanin inhibits these
enzymes. Before detailing the results of recent studies on ramoplanin, it is worth describing the inhibition curves that would be expected for a substrate
binder.100 Figure 6a shows a calculated curve for
inhibition of enzymatic activity by a substrate binder
that forms a 1:1 complex and has a dissociation
constant that is low relative to the Km of the substrate. The reaction rate is negligible at low substrate
concentrations because the substrate is fully bound
and unable to react. When the substrate concentration increases to the level where free substrate
becomes available, the reaction rate jumps rapidly.
At high substrate concentrations inhibition is overcome. Thus, the plot of reaction rate as a function of
substrate concentration has a pronounced sigmoidal
shape. Compounds that bind the substrate less
tightly still yield sigmoidal inhibition curves, but the
behavior is not as pronounced. Figure 6b shows
inhibition curves for three hypothetical substrate
binders having Kds ranging from 0.01 to 1 M. Hence,
the shape of the inhibition curves can provide information both on the mechanism of inhibition and on
the affinity of substrate binding.

458 Chemical Reviews, 2005, Vol. 105, No. 2

Walker et al.

Table 2. Substrate Recognition of Ramoplanina

R1 ) L-Ala--D-Glu-L-Lys-D-Ala-D-Ala. R2 ) L-Ala--D-Glu-L-Dap-D-Ala-D-Ala. R3 ) L-Ala--D-Glu-CONHCH3. R4 ) L-Ala--D-Gln.

4.3. Inhibition Kinetics of Ramoplanin


4.3.1. Transglycosylase Inhibition
Plots of reaction rate versus substrate concentration for inhibition of E. coli PBP1b, a prototypical
bacterial transglycosylase, as well as E. coli MurG
are shown in Figure 7a and b.93,101 The plots for
inhibition of the transglycosylase are sigmoidal, and
inhibition is overcome at high substrate concentrations, consistent with a substrate-binding mechanism
for inhibition of this enzyme. The absence of enzymatic activity in the first part of the curve implies
that ramoplanin binds to Lipid II with a dissociation
constant that is low relative to the substrate concen-

trations used. In addition, the substrate concentration at which the reaction rate increases suggests
that ramoplanin binds to Lipid II with a stoichiometry of 2:1. For example, when the ramoplanin
concentration is 6 M, the reaction rate jumps when
the substrate concentration exceeds 3 M; when the
ramoplanin concentration is 8 M, the reaction rate
jumps when the substrate concentration exceeds 4
M. Thus, the inhibition kinetics reveal that ramoplanin binds with submicromolar affinity and a 2:1
ramoplanin:Lipid II stoichiometry. Assuming that
ramoplanin acts by a pure equilibrium binding mechanism, a dissociation constant of 50 nM can be
calculated from the kinetic data.101

Chemistry and Biology of Ramoplanin

Chemical Reviews, 2005, Vol. 105, No. 2 459

Figure 6. Calculated inhibition curves for hypothetical substrate binders. (a) Curves generated at four different inhibitor
concentrations (0, 5, 10, and 15 mM) for a substrate binder that forms a 1:1 complex and has a Kd of 10 nM. (b) Curves
generated at a single inhibitor concentration for three substrate binders that form 1:1 complexes and have Kds of 1, 0.1,
and 0.01 mM.

Figure 7. Inhibition kinetics of ramoplanin against transglycosylase PBP1b and MurG.

4.3.2. MurG Inhibition


The curves for MurG inhibition are different from
those for transglycosylase inhibition in two respects
(Figure 7).93,101 First, inhibition cannot be overcome
by increasing the substrate concentration. Second,
the shapes of the curves are different. Whereas the
curves for transglycosylase inhibition are strongly
sigmoidal, those for MurG inhibition are not. In fact,
only at the highest ramoplanin concentration is there
any kinetic evidence that substrate binding may
occur. This observation, combined with the fact that
inhibition cannot be overcome, reveals that ramoplanin does not inhibit MurG simply by sequestering
Lipid I. The kinetic data do not, however, rule out
the possibility that ramoplanin forms a complex with
Lipid I and that this complex inhibits MurG in a
noncompetitive manner.
To determine whether Lipid I binding is required
for inhibition of MurG by ramoplanin, alanine was
attached to the Orn4 and Orn10 amines to produce
derivatives 24 and 25 (Figure 16). The Orn10 derivative was found to be incapable of binding to Lipid I,
while the Orn4 derivative retained the ability to bind
Lipid I.93 The compounds were tested for MurG
inhibition, and both had IC50s comparable to the
parent compound, indicating that Lipid I binding is

not required for MurG inhibition. On the basis of


these results, it was proposed that ramoplanin inhibits E. coli MurG not by binding to Lipid I, as
originally suggested by Somner and Reynolds,69 but
by binding directly to the enzyme. This hypothesis
is supported by studies showing that fluorescent
ramoplanin derivatives bind directly to E. coli MurG
at concentrations comparable to the concentrations
required to inhibit enzymatic activity.93

4.4. Evaluating the Proposed Cellular Targets of


Ramoplanin
The aforementioned studies have revealed that
ramoplanin inhibits MurG and the bacterial transglycosylase, PBP1b, by different mechanisms. These
differences in the mode of inhibition are unexpected
and, therefore, interesting; however, the results do
not by themselves provide insight into which target
is preferred in a cellular context. To draw conclusions
as to the cellular target from the biochemical experiments, it is necessary to consider other issues as well.
One issue that has already been raised is that of
accessibility. Lipid II and the bacterial transglycosylases are located on the external surface of the
bacterial membrane, where they are accessible to
ramoplanin. MurG, however, is located on the cyto-

460 Chemical Reviews, 2005, Vol. 105, No. 2

plasmic surface of the bacterial membrane, and


ramoplanin would have to penetrate the membrane
to reach this enzyme. Ramoplanin has a molecular
weight of more than 2500 and an estimated water
solubility of greater than 100 mg/mL. In the absence
of a transport system, a molecule having these
properties is unlikely to penetrate bacterial membranes efficiently. No experiments addressing whether
ramoplanin accumulates inside bacterial cells have
yet been reported, and the existence of a transport
mechanism cannot be discounted.
Another issue that bears on the mechanism of
action is that of the correspondence between the
concentrations of antibiotic needed for substrate
binding or enzyme inhibition relative to those needed
to inhibit bacterial growth. As noted earlier, ramoplanin is a potent antibiotic that inhibits a range of
Gram-positive bacterial strains at submicromolar
concentrations. Ramoplanin binds to and inhibits E.
coli MurG at concentrations at least 10-fold higher
than these MICs. Unless ramoplanin inhibits MurG
homologues from clinically relevant Gram-positive
microorganisms at significantly different concentrations (or somehow becomes concentrated at the site
of action), these low MICs cannot be explained by
inhibition of MurG. In contrast, the kinetics for
transglycosylase inhibition indicate that ramoplanin
binds to Lipid II very tightly, with an estimated
dissociation constant below 10-7 M.101 Moreover, a
qualitative comparison of ramoplanin binding to
Lipid I and Lipid II analogues shows that the
molecule binds to Lipid II better than to Lipid I (see
below). Thus, a mechanism of action involving inhibition of enzymes that utilize Lipid II would be
consistent with the experimental evidence. On the
basis of both target accessibility and the correspondence between the results of biochemical and cellbased assays, a strong argument can now be made
that ramoplanin inhibits bacterial transglycosylases
rather than MurG in bacterial cells.
Recently, another set of experiments addressing
the cellular target of ramoplanin was reported.
Scientists at Genome Therapeutics constructed a S.
aureus strain in which the expression of MurG is
increased. Compounds that inhibit MurG directly
should show an increase in MICs against this strain
relative to typical S. aureus strains. The MIC of
ramoplanin against this strain was found to be
comparable to its MICs against other strains, implying that the MurG enzyme itself is not the cellular
target of the antibiotic.102

Walker et al.

challenging to study because ramoplanin does not


form a discrete complex with Lipid II in vitro. Rather,
it associates to form ordered fibrils that do not yield
readily to structural analysis.81

5.1. Problem of Fibril Formation


The first attempts to characterize the interaction
between ramoplanin and Lipid II were reported by
Lo et al., who prepared a water-soluble analogue of
Lipid II containing a citronellyl chain in order to
probe the interaction of this molecule with ramoplanin.81 Natural Lipid II contains a 55-carbon lipid
chain, which causes it to aggregate extensively in
water, and it was expected that the water-soluble
analogue would facilitate structural analysis of the
complex. To address this possibility, Lo et al. carried
out an NMR titration of ramoplanin with the Lipid
II analogue (30) (Figure 8).81 Unexpectedly, the NMR
resonances of ramoplanin disappeared upon addition
of approximately 0.5 equiv of Lipid II analogue. A
CD titration of ramoplanin with the Lipid II analogue
revealed significant changes in the peptide backbone
region of the antibiotic along with the appearance of
a new band in the far UV region of the spectrum,
implying a nonrandom orientation of the aromatic
side chains in the complex (Figure 9). The presence
of isosbestic points in spectra acquired at different
Lipid II concentrations indicated a two state transi-

Figure 8. NMR titrations of ramoplanin with Lipid I and


Lipid II analogues. Spectra show downfield region of
ramoplanin in D2O upon addition of increasing amounts
of Lipid I/II analogues.

5. Molecular Recognition by Ramoplanin


Brotz et al. provided the first direct evidence for
an interaction between ramoplanin and a substrate
involved in peptidoglycan biosynthesis when they
showed that ramoplanin comigrates with Lipid II.77
Subsequent work from members of the Walker lab
has shown that ramoplanin inhibits the transglycosylation step of peptidoglycan biosynthesis by binding
to Lipid II.81,101 Thus, understanding how ramoplanin
recognizes Lipid II is essential for understanding its
mechanism of action. As described below, the recognition of Lipid II by ramoplanin has proven extremely

Figure 9. CD spectra of ramoplanin upon titration with


Lipid II analogue.

Chemistry and Biology of Ramoplanin

Chemical Reviews, 2005, Vol. 105, No. 2 461

lactate in 33 disfavors binding. In any event, the


peptide chain does not appear to be essential for
binding of lipid intermediates to ramoplanin. There
is evidence, however, that the intact peptide chain
plays a role in the affinity of UDP-MurNAc-pentapeptide for ramoplanin.104

5.2. Comparison of Substrate-Binding Affinities

Figure 10. Transmission electron micrograph of ramoplanin with Lipid II analogue.

tion between free ramoplanin and the complexed


species. These data, combined with the NMR results,
suggested that ramoplanin forms an ordered assembly in the presence of the Lipid II analogue.
Electron-microscopic analysis of the mixture of ramoplanin and the substrate analogue revealed, in
fact, that the ramoplanin complexes assemble to form
fibrils (Figure 10).81,103 The formation of these fibrils
explained the disappearing NMR resonances but
raised questions about whether the ability of ramoplanin to self-assemble in the presence of substrate
is in some way biologically relevant. This is a question that has not yet been answered.
The propensity of ramoplanin complexes to form
fibrils in the presence of Lipid II greatly complicates
structural analysis and makes it difficult to evaluate
differences in the binding affinities of different
substrates. Therefore, the relative contributions of
the different parts of Lipid II to binding are not well
understood. Lo and Cudic et al. evaluated the behavior of ramoplanin in the presence of different
Lipid I analogues or fragments thereof using CD
changes and fibril formation as a measure of binding.
From these studies it is known that the diphosphate
portion of the molecule is essential because the
MurNAc monophosphate derivatives 31 and 39 do
not interact with ramoplanin nor does the lipid
monophosphate derivative 32 (Table 2).30,81,103,104
However, lipid diphosphates such as farnesyl pyrophosphate 34 do interact, promoting the formation
of ramoplanin fibrils.103 It is also known that the
GlcNAc sugar plays a role in binding of Lipid II
because NMR titrations with Lipid I and Lipid II
analogues show that ramoplanin binds Lipid II more
tightly than it binds Lipid I (Figure 8). A comparison
of the inhibition kinetics of transglycosylase and
MurG also suggests that Lipid II is a better ligand
for ramoplanin than Lipid I.93,101 Finally, Cudic et al.
also suggested that the peptide chain on the MurNAc
sugar is essential for recognition based on studies
showing that compound 33 (Table 2) does not interact
with ramoplanin.30 Lo reported, however, that compounds 34, 35, and 36 (Table 2) do interact with
ramoplanin, with the binding of 35 and 36 qualitatively indistinguishable from 29, which contains the
intact peptide chain.103 It is possible that the free

A quantitative binding assay is required to understand the structural features that are important for
substrate recognition by ramoplanin. In an effort to
develop such an assay, Cudic et al. explored the
behavior of a variety of ramoplanin complexes (i.e.,
ramoplanin with UDP-MurNAc pentapeptide and
several Lipid I analogues) under different solution
conditions and found that some complexes remain in
solution even at relatively high concentrations (10-4
M) provided that 20% DMSO is added.104 Therefore,
it is possible to measure dissociation constants for
the compounds that form these soluble complexes
using NMR methods.104 Cudic et al. measured dissociation constants for UDP-MurNAc-pentapeptide
37 and the corresponding dipeptide analogue 38 and
found a 10-fold difference in affinity, indicating that
the terminal L-Dap-D-Ala-D-Ala tripeptide plays a role
in the binding of these substrate analogues to ramoplanin. Whether this is also true for Lipid II is
not clear because there may be significant differences
in how the UDP analogues and Lipid II bind to
ramoplanin (see below).
To characterize the binding interface of the DMSOsolubilized ramoplanin:UDP-MurNAc pentapeptide
complex, Cudic et al. analyzed NMR chemical shift
changes and intermolecular NOEs. The data showed
that ramoplanin binds to UDP-MurNAc-pentapeptide
in a 1:1 ratio and that the majority of the contacts
from ramoplanin to UDP-MurNAc-pentapeptide involve residues 2-8. Cudic et al. suggested that both
ornithine 4 and ornithine 10 make stabilizing electrostatic contacts to the ligand, helping to orient it.
They proposed that a contact from Orn4 to the
carboxylate of the terminal D-Ala-D-Ala dipeptide
plays a role in discriminating the pentapeptide from
the dipeptide. This NMR study was carried out based
on the assumption that structural information on
weakly binding peptidoglycan intermediates, which
are not as susceptible to fibril formation as Lipid II,
would be relevant to understanding how ramoplanin
recognizes Lipid II, which is presumed to be the
biologically relevant substrate. The validity of this
assumption has been called into question by the
kinetic results of Hu et al., showing that ramoplanin
binds to Lipid II with a stoichiometry of 2:1.101 A Job
titration has confirmed ramoplanin binds to Lipid II
in a 2:1 mode rather than the 1:1 mode reported for
the UDP-MurNAc-pentapeptide complex (Figure 11).
Therefore, UDP-MurNAc-pentapeptide may not be a
good model system for Lipid II. There is clearly still
a need for direct structural information on ramoplanin bound to Lipid II or a suitable analogue. It
may be necessary to produce isotopically labeled
ramoplanin and Lipid II analogues in order to obtain
the required structural information.

462 Chemical Reviews, 2005, Vol. 105, No. 2

Walker et al.

Figure 11. (a) Titration of a fluorescently labeled ramoplanin analogue (Orn4-fluorescein) with Lipid II. (b) Job titration
of Orn4-fluorescein ramoplanin with Lipid II.

5.3. Structural Studies on Ramoplanin and


Ramoplanin Complexes
In the absence of direct structural information on
the ramoplanin complex, models for how ramoplanin
interacts with Lipid II must be pieced together based
on the solution structure of free ramoplanin combined
with available data on substrate recognition. The first
solution structure of a ramoplanin factor was determined more than a decade ago by Williams and coworkers, who were studying an analogue of ramoplanin A2 called ramoplanose.21 Ramoplanose is identical
to ramoplanin A2 except that it contains a branched
trisaccharide rather than a disaccharide moiety at
position Hpg11. In water, ramoplanose was found to
have an antiparallel -sheet structure with a reverse
turn at L-Thr8/L-Phe9 and another turn in the vicinity
of Gly14-D-Ala16. The chirality of the amino acids
flanking L-Thr8/L-Phe9 alternates in a D,L,D pattern,
with the result that the R groups on both sides of
this turn are located on the same surface of the
-sheet. Williams and co-workers noted that the
-sheet curves upward slightly in order to alleviate
the steric congestion between side chains on the same
side of the molecule.
The first solution structure for ramoplanin A2 itself
was reported by Kurz and Guba in 1996.19 The
structure of ramoplanin A2 was found to be similar
in many respects to that reported for ramoplanose,
as expected based on the structural similarities
between the molecules. For example, like Williams
and co-workers, Kurz and Guba observed that ramoplanin adopts an antiparallel -stranded structure
with turns in the vicinity of residues 8-9 and 1416.19,21 However, there were also significant differences between the two NMR structures, and the
overall topologies were different. For example, whereas
ramoplanose was reported to have a slight curvature,
ramoplanin A2 was found be distinctly U-shaped,
with amino acid side chains located at the two ends
of the antiparallel sheet structure coming into contact. There were minor differences in the solution
conditions under which the NMR data were acquired.
Ramoplanose was studied in water, and ramoplanin
was studied in a 90:10 water:DMSO mixture. In
addition, the carbohydrate moieties on the two compounds were slightly different. However, the differences in the structures are probably related to the

Figure 12. NMR structure of ramoplanin dimer in methanol.

fact that Kurz and Guba used many more NOE


constraints in their calculations, including constraints between amino acids 9 and 17 at opposite
ends of the molecule. Williams and co-workers noted
that they were unable to achieve convergence due to
limitations in the modeling packages unless the
structural calculations were performed with a reduced set of distance constraints and atoms.21 Subsequent NMR studies on ramoplanin A2 by Lo et al.
and by Cudic et al. support the U-shaped topology
observed by Kurz and Guba.104,105
The third NMR study of ramoplanin in the absence
of substrate was reported by Lo et al., who observed
that ramoplanin A2 exists as a mixture of monomer
and dimer forms in slow exchange in methanol.105
The structure of the dimer was solved in the hope
that it might shed light on how ramoplanin binds to
Lipid II or on how it self-associates or both. The
ramoplanin dimer in methanol was found to consist
of a C2-symmetric antiparallel structure in which the
two monomer units are held together by interactions
between amino acids 10-14 (Figure 12). The structures of the monomer units themselves are very
similar to the monomer structure reported by Kurz
and Guba. The chirality of the residues in the
-strands spanning residues 10-14 alternates in a
D,L,D,L,Gly pattern. Cyclic peptides consisting of
alternating D,L residues are known to self-associate
via a combination of hydrogen bonds and interactions
between side chains aligned on the same side of the
peptide backbone, and the interface between ramoplanin monomers is reminiscent of the interface

Chemistry and Biology of Ramoplanin

between these self-associating peptides.106-108 The


ramoplanin dimer contains a cleft that provides a
possible binding site for Lipid II. An Orn10 residue
from each monomer flanks this cleft. The Orn4
residues are presented on opposite side of the dimer.
Derivatives of ramoplanin containing alanine at
either Orn4 or Orn10 have been prepared and tested
for biological activity and the ability to bind peptidoglycan intermediates.93 The Orn4Ala derivative
retained biological activity and the ability to bind
Lipid I substrate analogues, but the Orn10Ala derivative was inactive and unable to bind Lipid I analogues. These results suggest that Orn10 makes
critical contacts to the substrate but Orn4 does not.
The results are consistent with a model in which
Lipid II binds in a dimer cleft flanked by the two
Orn10 residues (Figure 12). However, confidence in
the model will require additional experiments, particularly since results that might not be consistent
with this model have also been reported (see section
8). Work on various synthetic and semi-synthetic
derivatives of ramoplanin that shed more light on the
Lipid II recognition event and on the structural
requirements for biological activity is discussed in the
following section.
In summary, three different solution structures of
ramoplanin factors have now been reported.19,21,105
Two of the structures are of monomers that differ in
overall topology,19,21 and one is of a dimer comprised
of two molecules that closely resemble one of the
reported monomers.105 The two monomer structures
were determined in largely aqueous media, while the
dimer structure was determined in methanol. Ramoplanin is believed to act at a membrane interface, and
Lo et al. suggested that methanol may mimic the
conditions under which ramoplanin acts better than
water.105 Whether this is so remains to be established. However, it is worth noting that the dimer
structure is consistent with other evidence showing
that ramoplanin complexes Lipid II with a stoichiometry of 2:1. Structural studies of ramoplanin in the
presence of mono- or bilayers could provide more
insight into the bioactive conformation of ramoplanin.
At this point, however, it is evident that ramoplanin
can exist as a monomer, dimer, and, in the presence
of Lipid II, higher order assembly with a 2:1 stoichiometry.101 The changeable nature of ramoplanin is
remarkable.

6. Total Synthesis of Ramoplanin and Key


Analogues
6.1. Preparation of Key Amino Acids
Ramoplanin consists of 17 amino acids of which
seven possess the nonproteinogenic D-configuration
and 12 possess nonstandard side chains. Of these,
the -hydroxy R-amino acids (D/L-allo-threonines and
L-threo--hydroxyasparagine) are the most abundant components of the ramoplanins whose preparations constitute significant elements of any projected
total synthesis. Since -hydroxy R-amino acids are
important components of many complex natural
products, a number of approaches have been reported
for their synthesis. These can be classified as (a)

Chemical Reviews, 2005, Vol. 105, No. 2 463

aldol-type additions of anions or enolates possessing


chiral auxiliaries including bis-lactams,114 oxazolidinones,115,116 oxazinones,117 and imidazolidinones,118
(b) the use of glycine Schiff bases,119 (c) asymmetric
epoxidation,120,121 dihydroxylation,122,123 and aminohydroxylation of olefins,124,125 (d) enantioselective
hydrogenation,126-128 (e) preparation from naturally
occurring amino acids,129-133 and (f) enzymatic
synthesis.134-137 Summarized below are the approaches implemented in efforts leading to the total
synthesis of ramoplanin.

6.1.1. aThr (allo-Threonines)


Despite their biological significance, the availability
of allo-threonines, nonproteinogenic amino acids, is
limited. Consequently, various methods for their
preparation have been developed. Of these, derivatization of a readily available natural amino acid is
the simplest. For efforts directed at ramoplanin, the
approach described by Beaulieu where D- or L-serine
is converted to L- or D-,,-trichloro-allo-threonine,
followed by catalytic hydrogenolysis to provide diasteromerically pure L- or D-allo-threonine, was employed (Scheme 1).132
Scheme 1

6.1.2. HAsn (L-threo--Hydroxyasparagine)


In the ramoplanins the hydroxyl group of L-threo-HAsn2 forms a labile ester bond with the Cterminus of 3-chloro-4-hydroxyphenylglycine (Chp17),
and the preparation of this unnatural amino acid,
L-threo--HAsn, constitutes one of the key elements
in the total synthesis of the ramoplanin aglycons. The
isomers of -HAsn were originally isolated from
human urine138 and had been synthesized as a
racemic mixture that was separated by resolution.139,140 Although there are reports of the enantioselective synthesis of the isomers of L--hydroxyaspartic acid (L--HAsp)141-144 and L-erythro--HAsn,145
an asymmetric synthesis of L-threo--HAsn had not
been disclosed. In efforts leading to the total synthesis of the ramoplanin aglycons, Boger et al.
disclosed the preparation of L-threo--HAsn enlisting
the Sharpless asymmetric aminohydroxylation (AA)
of methyl 4-methoxycinnamate to provide the corresponding amino alcohol in 71% yield and >99% ee
(Scheme 2).109,146 Sequential protection of the alcohol
(TBDMSOTf), one-pot N-Cbz/Boc exchange, and direct aminolysis of the methyl ester afforded the
corresponding amide in 94% overall yield with no

464 Chemical Reviews, 2005, Vol. 105, No. 2

Walker et al.

Scheme 2

evidence of epimerization. The aryl ring was oxidatively cleaved with RuO4 to release the carboxylic
acid, which was subsequently protected to provide the
benzyl ester in 65% and 84% yields, respectively.
Treatment of Boc-L-threo-HAsn(OTBS)-OBn with 4
N HCl-EtOAc led to the single-step removal of the
Boc and TBS protecting groups, and the resulting
amine was Fmoc protected (92%, 2 steps). A final
treatment with trityl alcohol and acetic anhydride
under acidic conditions provided Fmoc-L-threo-HAsn(Trt)-OBn in 71% yield and suitably protected
for direct incorporation into the following synthetic
efforts.

Figure 13. Key disconnections for synthesis of the ramoplanins.

6.2. Total Synthesis of the Ramoplanin A2 and


Ramoplanose Aglycon
The first, and to date only, total synthesis of the
ramoplanin A2 and ramoplanose aglycon was disclosed in 2001.109,147,148 Three key subunits composed
of residues 3-9 (heptapeptide 53), the pentadepsipeptide 64 (residues 1, 2, and 15-17), and
pentapeptide 72 (residues 10-14) were sequentially
coupled and cyclized in a solution-phase, convergent
approach to the 49-membered depsipeptide core of
1-3 (Figure 13).
The indicated coupling sites were chosen to maximize the convergency of the synthesis including that
of the three subunits, to minimize the use of protecting groups, to prevent late-stage opportunities for
racemization of carboxylate-activated phenylglycinederived residues, and to enlist -sheet preorganization of an acyclic macrocyclization substrate for ring
closure. Macrocyclization at the Phe9-D-Orn10 site
proved unusually successful and represents a site
found at the corner of a -turn at the end of the
H-bonded antiparallel -strands of ramoplanin (Figure 14). Consequently, closure at this site may benefit
from both -sheet preorganization of the substrate149
as well as closure at a D-amine terminus.150,151 A
second alternative cyclization site at Gly14-Leu15 site
was also successfully examined and lies within a
small flexible loop at the other end of the H-bonded
antiparallel -strands. In addition to potentially
benefiting from -sheet preorganization of the macrocyclization substrate, it represents closure at a nonhindered site incapable of racemization. Also key to
the implementation of the approach was the judicious
choice of the Orn4/Orn10 SES protection and the Asn1

Figure 14. Macrolactamization sites.

Fmoc protection providing orthogonal protecting


groups stable to Boc, Cbz, and benzyl ester deprotections yet capable of sequential and selective removal
in the presence of the labile depsipeptide ester.

Chemistry and Biology of Ramoplanin

Chemical Reviews, 2005, Vol. 105, No. 2 465

Scheme 3

Scheme 4

Heptapeptide 53, the first of the three key subunits


which contains all but Orn10 of the putative Hpg3Orn10 recognition sequence,30 was assembled as shown
in Scheme 3 from the tripeptide 45 and tetrapeptide
51 followed by Boc deprotection. This coupling cleanly
provided 52 without deliberate protection of the free
alcohols and with no evidence of competitive -elimination or trace racemization. Tripeptide 45 was
prepared by benzyl ester deprotection of 44 obtained
by coupling D-aThr-OBn132,152 with the dipeptide 43.
In turn, 43 was obtained by benzyl ester deprotection
of 42 derived from coupling Boc-D-Hpg-OH and
D-Orn(SES)-OBn.153,154 Tetrapeptide 51 was prepared
from the dipeptides 46 and 48 secured by the
coupling of Boc-L-Hpg-OH with D-Hpg-OBn and BocL-aThr-OH132,152 with L-Phe-OBn, respectively. Notably, benzyl ester hydrogenolysis of 46 followed by
coupling with 49 activated by DEPBT155,156 was
accomplished with no detectable racemization of the
sensitive D-Hpg residue (>99% de) or -elimination
of the hindered L-aThr, whereas alternative coupling
reagents gave lower conversions accompanied by
substantial epimerization.
Synthesis of the second key subunit, depsipentapeptide 64 which contains the labile backbone ester,
is summarized in Scheme 4. Fmoc deprotection of
Fmoc-L-threo--HAsn(Trt)-OBn (54, synthesis described in section 6.1.2) and coupling of 55 with
Fmoc-L-Asn(Trt)-OH provided 56. The key esterification of 56 with 57157 required activation with EDCI

conducted in the presence of DMAP158,159 catalyst. A


wide range of alternative esterification protocols were
examined, but these and EDCI- or DCC-promoted
couplings in the absence of DMAP provided no or
lower conversions and lower diastereomeric ratios or
suffered competitive -elimination. Even the EDCIDMAP, or a less satisfactory DCC-DMAP,158,159
reaction required carefully controlled reaction conditions with higher reaction temperatures (25 vs 0 C),
longer reaction times, or more DMAP (0.5-2 equiv
vs 0.15-0.2 equiv) offering less satisfactory conversions. Boc removal,160 coupling with 61, buffered TBS
deprotection,161,162 and benzyl ester hydrogenolysis
provided 64. Efforts to shorten this sequence by first
coupling 61 with L-Chp(OTBS)-OBn followed by
benzyl ester deprotection and esterification with 57
were not successful, and an unbuffered Bu4NF deprotection of 62 (no HOAc) led to competitive epimerization of the Chp R-aminoester center.
The preparation of the final subunit 72 is detailed
in Scheme 5. D-aThr-OBn132,152 was coupled with BocL-Hpg-OH to give dipeptide 65 devoid of diastereomeric contaminants. Boc deprotection and coupling
of 66 with Boc-D-Orn(SES)-OH provided tripeptide
67. Benzyl ester deprotection and coupling of 68 with
70 gave pentapeptide 71. In turn, the dipeptide 70
was obtained in two steps by coupling Boc-L-Hpg-OH
and Gly-OBn hydrochloride to give dipeptide 69,
followed by Boc deprotection (HCl-EtOAc). Notably,
C-terminus protection of D-aThr as its benzyl ester
in 67 and elsewhere throughout the synthesis permitted hydrogenolysis deprotection, avoiding basecatalyzed -elimination. Similarly, coupling each
sensitive Hpg-OH residue in 72, like that throughout the synthesis, conducted upon activation with
DEPBT156 minimized epimerization especially when
coupled with a hindered and sensitive aThr amine
terminus.
Assembly of three key fragments and macrocyclization at the Phe9-Orn10 site are detailed in

466 Chemical Reviews, 2005, Vol. 105, No. 2

Walker et al.

Scheme 5

Scheme 6

Scheme 6. The coupling of subunits 53 and 64 proved


to be the most challenging step of the synthesis.
Carboxylate activation of 64 typically resulted in
preferential -elimination of the acyloxy substituent.
This can be attributed to the combination of a superb
leaving group, the hindered nature of the activated
carboxylate resulting from its R,,-trisubstitution
and the large protecting groups (trityl and Fmoc), and
the enhanced R-carbon acidity of the activated carboxylate derived from 64 resulting from the use of a
N-acyl versus carbamate derivative. Only DEPBT156
promoted the coupling to provide 74 in superb yields
with no competitive -elimination, whereas all other
alternative coupling reagents and conditions surveyed provided predominantly -elimination products. Boc removal was accomplished under mild
conditions160 that preserved the trityl protecting
groups. Coupling of the crude free amine with 72
provided the key acyclic depsipeptide 75. Successive Boc removal, benzyl ester hydrogenolysis, and
macrocyclization afforded the cyclic depsipeptide core
76. Presumably, -sheet preorganization of the cyclization substrate149 and closure at a D-amine terminus150,151 at the corner of a -turn contributes to
the superb conversions for closure of the 49-membered ring (89% when conducted with purified amino
acid).
The alternative Gly14-Leu15 site was also examined
for the closure of the linear peptide (Scheme 7). This
approach would be beneficial for the preparation of
analogues of the sensitive depsipeptide region of the
natural products enlisting an alternative coupling
order of first assembling Hpg3-Gly14 as a common
precursor, followed by coupling with modified depsipentapeptide subunits, and a final key macrocyclization at the Gly14-Leu15 site. It constitutes closure at

the other end of the H-bonded antiparallel -strands


at a site within a small flexible loop and should
benefit from -sheet preorganization of the cyclization
substrate like closure at the Phe9-Orn10 site. Unlike
the Phe9-Orn10 site it does not benefit from closure
at a D-amine terminus, although the carboxylate
terminus now lacks a L-amino acid side chain that
may decelerate such ring closures. It also precludes
competitive racemization of the activated carboxylate.
As such, and based on these criteria alone, the two
closures might be expected to perform comparably.
The distinction between the two sites rests with Phe9Orn10 lying at the corner of a conformationally
defined -turn adjacent to the H-bonded antiparallel
-strands whereas the Gly14-Leu15 site is embedded
in what appears to be a conformationally more
flexible region of the molecule. Benzyl ester deprotection of 74 followed by coupling with 73 provided
77. Boc deprotection, benzyl ester deprotection, followed by macrocyclization provided 76 in good con-

Chemistry and Biology of Ramoplanin


Scheme 7

Scheme 8

versions (40-50%). This preliminary observation,


which was not subjected to optimization efforts,
suggests that closure at the Gly14-Leu15 site, while
perfectly acceptable, may not be as facile as closure
at the Phe9-Orn10 site.
Fmoc removal of 76 under specially developed
conditions (8 equiv of Bu4NF, 10 equiv of i-PrOH,
DMF, 25 C, 1 h) that do not promote competitive
Asn2 -elimination followed by side-chain acylation
of crude free amine 78 with anhydride 79 provided
80 in excellent overall conversions, Scheme 8. A
single-step HF deprotection of the trityl and SES
groups163,164 or sequential trityl and SES deprotection
provided the ramoplanin A2 and ramoplanose aglycon 5b. Notably, the Fmoc deprotection required the
development of a new method of Fmoc cleavage that
preserves the sensitive depsipeptide ester and the
Orn4 and Orn10 SES deprotections upon treatment
with HF enlisted conditions first introduced and
developed to avoid a competitive depsipeptide cleavage under typical strongly basic conditions.163,164

Chemical Reviews, 2005, Vol. 105, No. 2 467

6.3. Total Synthesis and Structure of the


Ramoplanin A1 and A3 Aglycons
The structure of ramoplanins (1-3) differs only
in the acyl group attached to the Asn1 N-terminus.15,17,18,165 Initially the stereochemistry of the two
double bonds in the three different acyl groups was
assigned as cis-cis.18 Three years after its structural
elucidation, Williams disclosed the structure of ramoplanose (4) whose composition was identical to
ramoplanin A2 with the exception of the branched
mannose trisaccharide attached at Hpg11 and the
stereochemistry of the lipid side chain (cis,trans- vs
cis,cis-7-methyloctadi-2,4-enoic acid).21 Soon thereafter, the stereochemistry of the 7-methyloctadi-2,4enoic acid side chain of ramoplanin A2 was also
revised to cis-trans by Kurz et al. in studies that
also served to provide a solution-phase conformation
of the natural product.19 The synthesis of the ramoplanin A2 aglycon detailed above confirmed this
revised structure of ramoplanin A2.109,147,148 Key to
the strategic planning of the approach was the
introduction of the lipid side chain onto the fully
functionalized cyclic depsipeptide core, thereby potentially providing direct access to all natural aglycons from a common, late-stage intermediate. In
recent studies this has been implemented for the total
synthesis of the ramoplanin A1 and A3 aglycons,
which confirmed an expected analogous structural
revision of the lipid side-chain stereochemistry of
ramoplanins A1 and A3.20
Removal of the Fmoc protecting group from 76
(Bu4NF, i-PrOH) afforded the free amine, the advanced synthetic intermediate prepared en route to
the ramoplanin A2 aglycon (Scheme 9).20,109,148 Subsequent treatment of the resulting free amine with
the anhydrides 81/83 provided the protected aglycons
82/84. Global deprotection using HF furnished the
ramoplanin A1 and A3 aglycons (5a/5c).
For comparison, the authentic ramoplanin A1 and
A3 aglycons were obtained from the natural ramoplanin complex by deglycosidation (HF) followed by
HPLC purification.166 The 1H NMR spectroscopic data
of the synthetic ramoplanin A1 and A3 aglycons (5a/
5c) were in complete agreement with the authentic
compounds. Although the H and H proton signals
of the lipid side chains are not clearly resolved in the
1H NMR spectra due to partial coincidence with other
resonances, JR and J can be measured directly from
the remaining two olefin proton signals, HR and H,
in the spectra. Not only are their values very similar
in ramoplanin A1-A3 (Figure 15), but the coupling
constants of 11.3-11.8 Hz and 14.9-15.4 Hz for JR
and J, respectively, define a cis stereochemistry for
the CR-C double bonds and a revised trans stereochemistry for C-C double bonds.

6.4. [N-Acetyl-Asn1]ramoplanin Aglycon


[N-Acetyl-Asn1]ramoplanin aglycon (13) was prepared and assessed in which the variable Asn1 lipid
N-acyl group was replaced with a minimal N-acetyl
group. Semisynthetic modifications of the natural
products and their aglycons have established that
removal of the side-chain unsaturation (hydro-

468 Chemical Reviews, 2005, Vol. 105, No. 2

Walker et al.

Scheme 9

Figure 15. Diagnostic coupling constants.


Scheme 10

6.5. Total Synthesis of Ramoplanin Amide


Analogues

genation) appears to have a minimal effect,167 but


the impact of the lipid side chain presence and its
potential role was not known. This was addressed
with the total synthesis of 13 as shown in Scheme
10.110 Removal of the Fmoc protecting group from
76109,110,148 (Bu4NF, i-PrOH), the advanced synthetic
intermediate we prepared en route to the ramoplanin
aglycon, followed by treatment with Ac2O provided
the protected N-acetyl derivative 85. Global deprotection using HF furnished 13.

Recently, Boger et al. described the total synthesis


of two key analogues of ramoplanin which entailed
the replacement of HAsn2 with L-2,3-diaminopropionic acid (Dap) and L-2,4-diaminobutyric acid (Dab).110
These deep-seated modifications provide a new generation of stable ramoplanin analogues incorporating
a hydrolytically stable amide bond (vs ester) between
residues 17 and 2.
Not only is the HAsn2 residue the most lengthy to
secure,146 but its incorporation into and maintenance
throughout the late stages of synthesis constitutes
the most difficult challenge limiting the ease of a total
synthesis of ramoplanin.109,147,148 More significantly,
the depsipeptide ester is the most fragile linkage in
the molecule and is central to the instability of the
natural products. Williams described the mild hydrolysis of ramoplanose providing the inactive linear
peptide.149 McCafferty and co-workers reported the
time-dependent depsipeptide cleavage of ramoplanin
A2 aglycon in acidic solutions,104 and Boger and coworkers described depsipeptide cleavage of ramoplanin A2 or its aglycon that occurs rapidly even under
mild basic conditions (1% Et3N-H2O, 25 C, 40%
hydrolysis in 2 min).109,147,148 Since it lies at the end
of the antiparallel -sheet with the ester adopting a
transoid carbonyl-eclipsed conformation analogous to
that of an amide and forms the link to the short
flexible loop (residue 15-17, see Figure 14),19 ester
replacement with an amide with incorporation of
L-Dap or L-Dab to provide 8 and 10 was expected to
have a minimal impact on the ramoplanin conformation but would preclude hydrolysis (8 and 10) and
significantly reduce (8) or preclude (10) -eliminationderived cleavage. Both would simplify the synthetic
challenges associated with assembling a library of
ramoplanin analogues required to probe each residue
and would be expected to improve in vivo stability.
The modified pentapeptide subunits 92a and 92b,
incorporating the Dap2 and Dab2 residues in place

Chemistry and Biology of Ramoplanin


Scheme 11

Chemical Reviews, 2005, Vol. 105, No. 2 469

of HAsn2, were prepared from tripeptide 86 (Scheme


11), which in turn was obtained by coupling Boc-LeuD-Ala-OH109,148 with Chp-OBn109,148 followed by deprotection of the benzyl ester. Fmoc-Dap-OBn (89a) and
Fmoc-Dab-OBn (89b), obtained from Fmoc-Dap(Boc)OH/Fmoc-Dab(Boc)-OH, were coupled with 87 using
DEPBT to give a single diastereomer of each tetrapeptide 90a and 90b. Fmoc removal upon treatment
with the Bu4NF and i-PrOH,109,148 coupling with
Fmoc-Asn(Trt)-OH, and benzyl ester hydrogenolysis
provided 92a and 92b.
The convergent assembly of the modified pentapeptide subunits into the linear precursors for macrocyclization highlights a strategic advantage of the
solution-phase modular synthesis developed for the
ramoplanin synthesis. DEPBT coupling of 92a and
92b with heptapeptide 53109,148 provided 93a/93b. Boc
removal was accomplished under conditions that
preserve the trityl protecting group, and the crude
amines 94a and 94b were coupled with pentapeptide
72109,148 to yield the macrocyclization substrates 95a
and 95b. Successive Boc removal, benzyl ester hydrogenolysis, and macrocyclization afforded the cyclic
peptide cores 96a and 96b. Fmoc removal (Bu4NF
and i-PrOH),109,148 Asn1 acyl side-chain introduction,
and global deprotection of the trityl and SES groups
upon HF treatment provided the key amide analogues 8 and 10.
The amide analogues lacking the Asn1 side chain,
9 and 11, were also prepared from 96a and 96b by
successive Fmoc removal and HF deprotection of the
trityl and SES groups.

6.6. Solid-Phase Synthesis of a Simplified


Analogue
Although a total synthesis of ramoplanin using
solid-phase chemistry has not yet been reported, a
simplified analogue has been assembled. McCafferty
and co-workers30 synthesized a cyclic peptide, derived
from the putative ramoplanin peptidoglycan recognition sequence, NR-octanoyl-Asn1-Asn2-c[Cys3-D-Orn4D-aThr5-Hpg6-D-Hpg7-aThr8-Cys9]-D-Orn10-NH2, by
solid-phase peptide synthesis of the linear peptide,
Scheme 12

470 Chemical Reviews, 2005, Vol. 105, No. 2

Walker et al.

Figure 16. Degradation and derivatization of the ramoplanins.

Asp1 N-acylation, resin cleavage, and oxidative intramolecular cyclization (Scheme 12).

7. Degradation and Semisynthetic Studies


In efforts to examine the role of structural features
and functional groups in ramoplanin, accessible
moieties have been chemically deleted, masked, or
modified in the naturally occurring structures by
several groups. Figure 16 summarizes the semisynthetic analogues examined to date and conditions
used for their synthesis.

7.1. Ramoplanin Aglycons


In initial studies treatment of the ramoplanin
complex with either (1) TMSI or TMSCl/NaI followed
by hydrolysis or (2) HCl in anhydrous BuOH provided
the deglycosylated derivatives of ramoplanin in 2030% yield.167 A recent improved procedure relied on
reaction of the ramoplanin complex with anhydrous
HF.166 Reverse-phase HPLC purification of the crude
mixture which serves to separate the A1-A3 aglycons gave pure A1 (3%, 5a), A2 (46%, 5b), and A3
(3%, 5c).

7.2. Depsipeptide Hydrolysis


The most fragile linkage in the ramoplanins, the
depsipeptide ester formed between HAsn2 and Chp17,
can be hydrolyzed under mildly basic conditions to
afford acyclic ramoplanin derivatives.109,149 Williams
first reported this hydrolysis with ramoplanose (1%
Et3N-H2O, 25 C, 1 h), and more recently the ease
of hydrolysis was examined by Boger and co-workers
in the course of synthetic studies. Treatment of either
ramoplanin A2 or the ramoplanin A2 aglycon under
these conditions resulted in rapid hydrolysis of the
depsipeptide ester to cleanly provide the linear acyclic
derivatives 6 and 7 (>90% isolated yield, 20 min; 80%
conversion, 5 min; 30% conversion, 1 min) with little
or no difference in the rates of reaction. Significantly,

Williams and co-workers observed that the corresponding acyclic ramoplanose partially retained the
secondary -sheet structure of the parent molecule
(1H NMR NOEs). On the basis of this presumed
preorganized -sheet conformation of a linear peptide
derivative, the Phe9-D-Orn10 macrocyclization site
found at the corner of the -turn and the end of the
H-bonded antiparallel -sheet proved unusually successful in the total synthesis of the ramoplanin
aglycons.

7.3. Lipid Side-Chain Reduction


Tetrahydroramoplanin (e.g., tetrahydroramoplanin
A2, 12) and ramoplanin aglycon derivatives167,112 can
be easily accessed by catalytic hydrogenation (5% Pd/
C) of the natural products or their aglycons.

7.4. Orn4 and Orn10 Derivatization


In efforts to probe the biological importance of
the Orn4 and Orn10 -amino groups, several groups
have described methods for their derivatization.
McCafferty and co-workers converted the terminal
amines of both Orn4 and Orn10 to the diguanidylated,
diisovaleryl, and diacetyl derivatives.30 Thus, treatment of ramoplanin A2 with either 1H-pyrazole-1carboxamidine or acetic anhydride in the presence
of pyridine provided the [Orn4, Orn10]-diguanidylated
(26) or -diacetylated (28) products. Reductive amination of ramoplanin A2 with isovaleraldehyde and
NaBH3CN afforded [Orn4, Orn10]-diisovaleryl ramoplanin A2 (27). In elegant studies designed to address
the relative role of each Orn residue, Walker and coworkers disclosed the reaction of ramoplanin A2 with
Boc-Ala-OSu followed by removal of Boc groups with
TFA to provide the [Orn4-Ala] (24) and [Orn10-Ala]
(25) derivatives.93 Both retain the charged (protonated) amine of the parent molecule but displace its
positioning by three atoms. Finally, treatment of the
ramoplanin A2 aglycon with SESCl (Et3N, DMF, -20
C) afforded a mixture of the [Orn4,Orn10]-diSES

Chemistry and Biology of Ramoplanin

derivative 22 and the [Orn10]-monoSES derivative 23


which were separated by reverse-phase HPLC.109

7.5. Lipid Side-Chain Replacement


Complementary to the totally synthetic preparation
of the [N-acetyl-Asn1]ramoplanin aglycon, Ciabatti
recently detailed the preparation of an extensive
series of semisynthetic side-chain-modified ramoplanin analogues.111 The Orn4 and Orn10 -amino groups
of ramoplanin A2 were first masked with Fmoc or
Cbz protecting groups. Ozonolysis of the ramoplanin
derivatives 98 and subsequent reductive amination
of the resulting N-acyl glyoxal with benzylamine
afforded a precursor suitable for Edman degradation
with phenylisothiocyanate in pyridine-H2O to provide the ramoplanin-free amine. The resulting amine
was then acylated with an extensive series of lipid
side-chain replacements and the Orn protecting
groups removed to give over 130 analogues of the
natural products (Scheme 13).
Scheme 13

7.6. Summary
Notably, degradative and semisynthetic modifications of the ramoplanins have been limited to date
and probed only small regions of the natural product.
A renewed effort in such studies could shed insight
into additional roles of undefined peripheral functional groups complementary to more deep-seated

Chemical Reviews, 2005, Vol. 105, No. 2 471

structural modifications and amino acid side-chain


substitutions that are now accessible by total synthesis.

8. StructureActivity Studies on Ramoplanin and


Its Synthetic and Semisynthetic Analogues
8.1. Antimicrobial Activity of Key Derivatives and
Analogues
Ramoplanin A2 is the most abundant component
of the ramoplanin complex. Consequently, synthetic
and semisynthetic analogues of ramoplanin A2 have
been the subject of most investigations seeking to
define the key structural requirements for activity.
The antimicrobial activities of such key derivatives
are summarized in Table 1.
The in vitro antimicrobial activity of ramoplanin
A2 (2) and its aglycon (5b) are not distinguishable109,110,167 with the latter typically displaying slightly
greater in vitro potency.109,110 Consequently, glycosidation does not contribute to the intrinsic in vitro
antimicrobial activity and the role of the natural
product di- or trisaccharides remains to be established. Acycloramoplanin A2 (6) and its aglycon (7),
in which the macrocyclic lactone was hydrolyzed,
have been reported to be inactive (MIC >128 g/mL)
and >250- to 500-fold less potent than the natural
product109 or roughly 2000-fold less active than
ramoplanin A2 when tested against a more sensitive
organism.30 Although it is difficult to establish whether
such residual activity (e0.05%) can be attributed to
contaminant natural product in such samples, it is
clear that the intact macrocyclic ring of the ramoplanin is key to their properties.
Two depsipeptide amide analogues were recently
synthesized, one of which was found to be slightly
more potent than the natural aglycon in antimicrobial assays providing a new lead structure with an
improved profile and a more stable and accessible
macrocyclic template on which to conduct structurefunction studies.110 The depsipeptide amine analogue
8 of the ramoplanin A2 aglycon, in which the backbone ester was replaced with an amide, retained or
exhibited a slightly increased antimicrobial potency
(MIC ) 0.39 g/mL) relative to the ramoplanin A2
aglycon (MIC ) 0.78 g/mL). Not only did this
demonstrate that the depsipeptide ester may be
replaced with a more stable amide without impacting
the in vitro antimicrobial activity, but also that the
HAsn2 carboxamide found in ramoplanin (but is
absent in 8) does not appear to contribute directly to
its properties. Notably, the ramoplanin ester adopts
a typical trans (carbonyl eclipsed) conformation
within the flexible loop capping one end of the
antiparallel -sheet, and it might not be surprising
that its replacement with an amide was well tolerated. Just as importantly, 8 proved to be completely
stable to mildly basic conditions (1% Et3N-H2O, 25
C, 24 h, 0% disappearance)110 that rapidly consume
the ramoplanin aglycon (80% within 5 min at 25
C).109 In sharp contrast, the depsipeptide amide
analogue 10 containing the single additional methylene in the macrocycle was inactive (MIC g 35-50
g/mL), exhibiting a >100-fold loss in activity relative

472 Chemical Reviews, 2005, Vol. 105, No. 2

to the ramoplanin A2 aglycon and 8.110 Thus, even a


simple methylene insertion with incorporation of the
Dab2 residue into the cyclic peptide abolished activity
completely, suggesting not only that the cyclic structure is critical to ramoplanins properties, but that
subtle perturbations to even the flexible loop can
have a pronounced impact. The impact of this modification was subsequently explored in greater detail
as described in the next section.
Recent studies have also defined the impact of the
lipid side chain. The tetrahydroramoplanin A1-A3,
which are under investigation for use as topical
antibiotics for treatment of wound infections, exhibited slightly reduced activities (e.g., 12, Table 1).112
The active ramoplanin amide analogue (9), lacking
the Asn1 lipid side chain, and [N-acetyl-Asn1]ramoplanin aglycon (13) both were approximately 16-110
or 100-fold113 less potent than the corresponding
compounds containing the natural A2 side chain.
These results indicate that the lipid side chain of
ramoplanin A2 contributes significantly to its properties but is not essential.110 These observations are in
accord with those of Ciabatti and co-workers, who
prepared an extensive series of over 130 semisynthetic ramoplanin analogues with modifications to
the lipid side chain. When the length and composition
of the aliphatic side chain is similar to that of the
naturally occurring lipid side chains, the antimicrobial activities of the corresponding analogues are
retained, whereas the activities are significantly
reduced with longer and shorter lengths (e.g., 1419 in Table 1 and Scheme 13). Interestingly, some
analogues which possess aromatic side chains including naphthyl or biphenyl were found to possess the
same or more potent antimicrobial activity than
ramoplanin (e.g., 20 and 21).111 Further insight into
the role of the lipid side chain was obtained from
studies detailed in the following section.
The importance of the free amines of the two
ornithine residues (Orn4 and Orn10) has been studied
extensively. McCafferty and co-workers found that
[Orn4,Orn10-diacetyl]ramoplanin A2 (28), in which
both amines are acetylated, was 500-fold less potent
than ramoplanin A2, highlighting the importance of
these two basic amines which are conserved among
all members of this class of natural products. They
also found that preservation of cationic charge by
guanidylation (26) or transformation to a secondary
amine (27) by reductive amination resulted in only
small to modest alterations in the antimicrobial
activity.30 Similarly, the [Orn4,Orn10-diSES]ramoplanin A2 aglycon (22), in which both amines were
converted to sulfonamide derivatives, was found to
be inactive (MIC > 128 g/mL, >500-fold loss in
activity).109 Much more interestingly, [Orn10-SES]ramoplanin A2 aglycon (23), in which only the Orn10
-amino group is protected, was 16-fold less potent
than the free aglycon but >32-fold more potent than
the diSES derivative.109 Clearly, both the Orn4 and
Orn10 amines contribute to the antimicrobial activity,
and the latter result suggested that the Orn4 free
amine might be more important than the Orn10
amine. However, more recent work of Walkers,93
enlisting alanine derivatives of the Orn4 and Orn10

Walker et al.

-amines, which maintain but move or extend the


position of the free amines, found that [Orn4-Ala]ramoplanin A2 (24) was an active antimicrobial agent
(MIC ) 0.8 g/mL) whereas [Orn10-Ala]ramoplanin
A2 (25) was nearly inactive (MIC ) 50 g/mL),
implicating Orn10 versus Orn4 as the more important
of the two residues. Notably, the positively charged
amine of Orn4 flanks the one side of the ramoplanin
solution structure, whereas that of Orn10 flanks the
other, and they lie at opposite ends of the conserved
putative peptidoglycan-binding domain Hpg3-Hpg.10
Further studies on these and additional synthetic and
semisynthetic derivatives should serve to clarify the
role of Orn4 and Orn10 amines and the putative Hpg3Orn10 recognition domain and offer insight into the
mechanism of action of ramoplanin.

8.2. Mechanistic Analysis of Ramoplanin


AnaloguessThe Path Forward
Most recently, several of the key ramoplanin
analogues have been examined for transglycosylation
inhibition enlisting the assay introduced by Walker
and co-workers113 to quantitate the direct conversion
of Lipid II to peptidoglycan. Combined with the
antimicrobial activity of the analogues, the results
provide insight into the roles of the key structural
features required for Lipid II binding, inhibition of
the transglycosylation reaction, and biological activity. Just as importantly, the work defines a protocol
of assays from which key insights can be garnered
about the functional role of ramoplanin structural
features that can be used to design new, related, and
improved analogues or derivatives. Thus, the active
amide analogue 8 of the ramoplanin A2 aglycon, its
free amine derivative 9 lacking the lipid side chain,
[N-acetyl-Asn1]ramoplanin aglycon (13), and the
inactive ring-expanded amide analogue 10 were
examined in the kinetic assay for inhibition of the
transglycosylase PBP1b. Notably, the kinetics of
inhibition in this assay is a sensitive measure of
substrate binding and provides insight into both the
affinity and stoichiometry of Lipid II binding. As
shown in Figure 17 each analogue bound to Lipid II
with comparable affinity and the same 2:1 stoichiometry as ramoplanin itself. Thus, replacing the Asn2
ester with an amide does not affect Lipid II binding,
even when an additional methylene unit is inserted.
Furthermore, truncating or removing the lipid side
chain does not affect the inhibition kinetics, indicating that it does not play a direct role in substrate
binding or transglycosylase inhibition.
Nonetheless, and with the exception of 8, which
was equipotent with 2, each of these analogues was
at least 2 orders of magnitude less potent than
ramoplanin A2 in antimicrobial assays. Thus, although the lipid side chain does not play a role in
substrate binding or in vitro inhibition of the transglycosylase reaction, it plays a key role in the
biological activity. Most likely this arises by membrane targeting and localization by the lipid side
chain, positioning the antibiotic near its target, Lipid
II, which is located on the outer surface of the
bacterial membrane.
Similarly, 10, which contains an extra methylene
in the macrocyclic ring, is also 100-fold less potent

Chemistry and Biology of Ramoplanin

Figure 17. Inhibition of E. coli PBP1b by ramoplanin and


its key analogues: 2 (O), 8 (0), 10 (4), and 13 (2) at 6 M
concentration and no inhibitor (b).

than 2 or 8. NMR analysis revealed that 10, unlike


ramoplanin and 8, aggregates extensively in aqueous
buffer. Williams and co-workers previously reported
that the ring-opened form of ramoplanin maintains
the same general conformation as the cyclized parent
compound but shows a pronounced tendency to
aggregate, presumably because the increased flexibility permits the -strands in the molecule to
associate in an intermolecular fashion.149 The extra
methylene in 10 may increase macrocycle flexibility,
promoting self-association and perhaps enabling
undesirable interactions with other molecules as well.
Unfavorable partitioning of the more flexible ramoplanin analogue 10 in cell-based assays would explain its higher MICs.
Thus, substituting an amide linkage lacking the
HAsn2 carboxamide side chain for the more labile and
more complex ester linkage does not affect substrate
binding, in vitro activity, or in vivo antimicrobial
activity provided that the ring size is maintained and
the lipid chain is not removed. The amide-linked
macrocycle is considerably more stable than the ester
and may have significant advantages as a therapeutic agent. Increasing the ring size does not appear to
affect substrate binding or inhibition of the transglycosylation reaction but greatly increases the tendency of the molecule to associate, which likely leads
to the decrease in biological activity. Any modifications that increase the flexibility of the molecule may,
therefore, have a generally deleterious effect on
biological activity. Most importantly, the lipid chain
plays a key role in biological activity without directly
influencing binding to Lipid II or in vitro inhibition
of transglycosylation. It is proposed that the lipid
helps target ramoplanin to bacterial membranes. If
so, substitution of the lipid chain with other groups
that also facilitate localization may lead to analogues
with improved activity.

9. Conclusion
Ramoplanin is an antibiotic with a novel structure
and distinct mechanism of action. As such, it repre-

Chemical Reviews, 2005, Vol. 105, No. 2 473

sents an excellent target for clinical development to


treat bacteria resistant to the existing panel of
clinically used antibiotics. Ramoplanin is currently
in Phase III trials for the treatment of VRE; however,
further development of this promising antibiotic
requires a deeper understanding of its mechanism
and the development of structural analogues that
may address its present limitations (e.g., stability).
Though it was originally believed to inhibit MurG, a
series of experiments has shown that the transglycosylation step of peptidoglycan biosynthesis is the
likely target for this molecule. The results of the
mechanistic work also show that ramoplanain may
bind to its target, Lipid II, with a 2:1 stoichiometry.
The reevaluation of the enzymatic target and mechanism of action for ramoplanin has important implications for the design and synthesis of analogues with
potentially improved potency, stability, and solubility.
Concurrent with the mechanistic work, a series of
elegant synthetic studies resulted in the first total
synthesis of the ramoplanin aglycon. The total synthesis enabled the construction of ramoplanin analogues that probed the contribution of specific aspects
of the ramoplanin structure to biological activity. In
particular, the role of the lipid tail, the size of the
macrocycle, and the contribution of the ornithine
residues to activity have been examined to date. Most
significantly, a synthetic stable amide analogue of the
ramoplanin aglycon has been identified (8, [L-Dap2]ramoplanin A2 aglycon) that can now serve as an
accessible macrocyclic template upon which systematic modifications to the structure can be conducted.
Using this stable and accessible template, the key
structural elements of the putative recognition domain and the dimerization domain can be probed
with subsequent synthetic analogues to provide
insight into the mechanism of action and potentially
provide compounds that address issues limiting the
clinical use of ramoplanin itself (e.g., stability). Thus,
development of synthetic methodology and a synthesis of ramoplanin analogues have provided a means
to further probe the mechanism of action.
The combination of synthetic and mechanistic work
has furthered our understanding of ramoplanin.
However, a number of questions remain about the
drug, in particular, the role of ligand-induced polymerization in antibiotic activity and the structure of
the ramoplanin:Lipid II complex. The ability to
synthesize ramoplanin analogues now makes it possible to prepare specific derivatives for such structural and mechanistic analysis.

10. Abbreviations
BCB
Chp
Dab
Dap
DCC
DEPBT
DMAP
EDCI
HOAt

B-bromocatecholborane
L-chlorohydroxyphenylglycine
L-2,3-diaminobutyric acid
L-2,3-diaminopropionic acid

1,3-dicyclohexylcarbodiimide
3-(diethoxyphosphoryloxy)-1,2,3-benzotriazin4(3H)-one
4-(dimethylamino)pyridine
1-(3-dimethylaminopropyl)-3-ethylcarbodiimide
hydrochloride
1-hydroxy-7-azabenzotriazole

474 Chemical Reviews, 2005, Vol. 105, No. 2


HOBt
Hpg
SES

1-hydroxybenzotriazole
L-hydroxyphenylglycine

2-trimethylsilylethanesulfonyl

11. References
(1) Austrian, R.; Gold, J. Ann. Intern. Med. 1964, 60, 759.
(2) Dineen, P.; Homan, W. P.; Grafe, W. R. Ann. Surg. 1976, 184,
717.
(3) U.S. Congress, Office of Technology Assessment; Impacts of
Antibiotic-Resistant Bacteria; OTA-H-629; U.S. Government
Printing Office: Washington, DC, Sept 1995.
(4) Coates, A.; Hu, Y.; Bax, R.; Page, C. Nat. Rev. Drug Discov. 2002,
1, 895.
(5) Sievert, D. M.; Boulton, M. L.; Stoltman, G.; Johnson, D.;
Stobierski, M. G.; Downes, F. P.; Somsel, P. A.; Rudrik, J. T.;
Brown, W.; Hafeez, W.; Lundstrom, T.; Flanagan, E.; Johnson,
R.; Mitchell, J.; Chang, S. CDC Morbid. Mortal. Weekly Rep.
2002, 51, 565.
(6) Williams, D. H.; Bardsley, B. Angew. Chem., Int. Ed. Engl. 1999,
38, 1172.
(7) Derlot, E.; Courvalin, P. Am. J. Med. 1991, 91, S82.
(8) Walsh, F. M.; Amyes, S. G. Curr. Opin. Microbiol. 2004, 7, 439.
(9) Smith, T. L.; Pearson, M. L.; Wilcox, K. R.; Cruz, C.; Lancaster,
M. V.; Robinson-Dunn, B.; Tenover, F. C.; Zervos, M. J.; Band,
J. D.; White, E.; Jarvis, W. R. N. Engl. J. Med. 1999, 340, 493.
(10) Weigel, L. M.; Clewell, D. B.; Gill, S. R.; Clark, N. C.; McDougal,
L. K.; Flannagan, S. E.; Kolonay, J. F.; Shetty, J.; Killgore, G.
E.; Tenover, F. C. Science 2003, 302, 1569.
(11) Walsh, C. Nat. Rev. Microbiol. 2003, 1, 65.
(12) Woodford, N. Expert Opin. Inv. Drug 2003, 12, 117.
(13) Patel, R. J. Antimicrob. Chemother. 2003, 51, 13.
(14) Montecalvo, M. A. J. Antimicrob. Chemother. 2003, 51, 31.
(15) McCafferty, D. G.; Cudic, P.; Frankel, B. A.; Barkallah, S.;
Kruger, R. G.; Li, W. Biopolymers 2002, 66, 261.
(16) Cavalleri, B.; Pagani, H.; Volpe, G.; Selva, E.; Parenti, F. J.
Antibiot. 1984, 37, 309.
(17) Parenti, F.; Ciabatti, R.; Cavalleri, B.; Kettenring, J. Drugs Exp.
Clin. Res. 1990, 16, 451.
(18) Ciabatti, R.; Kettenring, J. K.; Winters, G.; Tuan, G.; Zerilli, L.;
Cavalleri, B. J. Antibiot. 1989, 42, 254.
(19) Kurz, M.; Guba, W. Biochemistry 1996, 35, 12570.
(20) Shin, D.; Rew, Y.; Boger, D. L. Proc. Natl. Acad. Sci. U.S.A. 2004,
101, 11977.
(21) Skelton, N. J.; Harding, M. M.; Mortishire-Smith, R. J.; Rahman,
S. K.; Williams, D. H.; Rance, M. J.; Ruddock, J. C. J. Am. Chem.
Soc. 1991, 113, 7522.
(22) Jones, R. N.; Barry, A. L. Diagn. Microbiol. Infect. Dis. 1989,
12, 279.
(23) Maple, P. A. C.; Hamiltonmiller, J. M. T.; Brumfitt, W. J.
Antimicrob. Chemother. 1989, 23, 517.
(24) Francis, J.; Webster, H.; Newsom, S. W. B. Drugs Exp. Clin. Res.
1990, 16, 457.
(25) Ohare, M. D.; Felmingham, D.; Gruneberg, R. N. Drug. Exp.
Clin. Res. 1988, 14, 617.
(26) Ohare, M. D.; Ghosh, G.; Felmingham, D.; Gruneberg, R. N. J.
Antimicrob. Chemother. 1990, 25, 217.
(27) Ristow, T. A.; Noskin, G. A.; Warren, J. R.; Peterson, L. R.
Microb. Drug Resist. 1995, 1, 335.
(28) Mobarakai, N.; Quale, J. M.; Landman, D. Antimicrob. Agents
Chemother. 1994, 38, 385.
(29) Scotti, R.; Dulworth, J. K.; Kenny, M. T.; Goldstein, B. P. Diagn.
Microbiol. Infect. Dis. 1993, 17, 209.
(30) Cudic, P.; Behenna, D. C.; Kranz, J. K.; Kruger, R. G.; Wand,
A. J.; Veklich, Y. I.; Weisel, J. W.; McCafferty, D. G. Chem. Biol.
2002, 9, 897.
(31) Higashide, E.; Hatano, K.; Shibata, M.; Nakazawa, K. J. Antibiot.
1968, 21, 126.
(32) Meyers, E.; Weisenborn, F. L.; Pansy, F. E.; Slusarchyk, D. S.;
Von Saltza, M. H.; Rathnum, M. L.; Parker, W. L. J. Antibiot.
1970, 23, 502.
(33) Espersen, F. Curr. Opin. Anti-Infect. Invest. Drugs 1999, 1, 78.
(34) Walsh, C. T. Antibiotics: Actions, Origins, Resistance; ASM
Press: Washington, D.C., 2003.
(35) Gale, E. F.; Cundliffe, E.; Reynolds, P. E.; Richmond, M. H.;
Waring, M. J. The Molecular Basis of Antibiotic Action; WileyInterscience: New York, 1981.
(36) Rogers, H. J.; Perkins, H. R.; Ward, J. B. Microbial Cell Walls
and Membranes; Chapman & Hall: London, 1980.
(37) Wong, K. K.; Pompliano, D. L. Adv. Exp. Med. Biol. 1998, 456,
197.
(38) Silver, L. L. Curr. Opin. Microbiol. 2003, 6, 431.
(39) van Heijenoort, J. Glycobiology 2001, 11, 25R.
(40) van Heijenoort, J. In Escherichia coli and Salmonella Cellular
and Molecular Biology; Neidhardt, F. C., Ed.; ASM Press:
Washington, D.C., 1996.

Walker et al.
(41) Park, J. T. In Escherichia coli and Salmonella Cellular Biology;
Neidhardt, F. C., Ed.; ASM Press: Washington, D.C., 1996.
(42) Ghuysen, J.-M. New Comprehensive Biochemistry, Bacterial Cell
Wall; Elsevier: Amsterdam, 1994.
(43) Bugg, T. D. H.; Walsh, C. T. Nat. Prod. Rep. 1992, 199.
(44) Mengin-Lecreulx, D.; Flouret, B.; van Heijenoort, J. J. Bacteriol.
1982, 151, 1109.
(45) Mengin-Lecreulx, D.; Parquet, C.; Desviat, L. R.; Pla, J.; Flouret,
B.; Ayala, J. A.; van Heijenoort, J. J. Bacteriol. 1989, 171, 6126.
(46) Marquardt, J. L.; Siegele, D. A.; Kolter, R.; Walsh, C. T. J.
Bacteriol. 1992, 174, 5748.
(47) Benson, T. E.; Marquardt, J. L.; Marquardt, A. C.; Etzkorn, F.
A.; Walsh, C. T. Biochemistry 1993, 32, 2024.
(48) Betina, V. The Chemistry and Biology of Antibiotics; Elsevier
Scientific Publishing Co.: New York, 1983.
(49) Benson, T. E.; Harris, M. S.; Choi, G. H.; Cialdella, J. I.; Herberg,
J. T.; Martin, J. P., Jr.; Baldwin, E. T. Biochemistry 2001, 40,
2340.
(50) Bertrand, J. A.; Auger, G.; Fanchon, E.; Martin, L.; Blanot, D.;
van Heijenoort, J.; Dideberg, O. EMBO J 1997, 16, 3416.
(51) Gordon, E.; Flouret, B.; Chantalat, L.; van Heijenoort, J.;
Mengin-Lecreulx, D.; Dideberg, O. J. Biol. Chem. 2001, 276,
10999.
(52) Mol, C. D.; Brooun, A.; Dougan, D. R.; Hilgers, M. T.; Tari, L.
W.; Wijnands, R. A.; Knuth, M. W.; McRee, D. E.; Swanson, R.
V. J. Bacteriol. 2003, 185, 4152.
(53) Schonbrunn, E.; Sack, S.; Eschenburg, S.; Perrakis, A.; Krekel,
F.; Amrhein, N.; Mandelkow, E. Structure 1996, 4, 1065.
(54) Yan, Y.; Munshi, S.; Leiting, B.; Anderson, M. S.; Chrzas, J.;
Chen, Z. J. Mol. Biol. 2000, 304, 435.
(55) Ikeda, M.; Wachi, M.; Jung, H. K.; Ishino, F.; Matsuhashi, M.
J. Bacteriol. 1991, 173, 1021.
(56) Bouhss, A.; Crouvoisier, M.; Blanot, D.; Mengin-Lecreulx, D. J.
Biol. Chem. 2004, 279, 29974.
(57) Taku, A.; Fan, D. P. J. Biol. Chem. 1976, 251, 6154.
(58) Ha, S.; Walker, D.; Shi, Y.; Walker, S. Protein Sci. 2000, 9, 1045.
(59) Hu, Y.; Chen, L.; Ha, S.; Gross, B.; Falcone, B.; Walker, D.;
Mokhtarzadeh, M.; Walker, S. Proc. Natl. Acad. Sci. U.S.A. 2003,
100, 845.
(60) Holtje, J. V. Microbiol. Mol. Biol. Rev. 1998, 62, 181.
(61) Popham, D. L.; Young, K. D. Curr. Opin. Microbiol. 2003, 6, 594.
(62) Goffin, C.; Ghuysen, J. M. Microbiol. Mol. Biol. Rev. 1998, 62,
1079.
(63) Young, K. D. Biochimie 2001, 83, 99.
(64) Suzuki, H.; Nishimura, Y.; Hirota, Y. Proc. Natl. Acad. Sci.
U.S.A. 1978, 75, 664.
(65) Tamaki, S.; Nakajima, S.; Matsuhashi, M. Proc. Natl. Acad. Sci.
U.S.A. 1977, 74, 5472.
(66) Kahan, F. M.; Kahan, J. S.; Cassidy, P. J.; Kropp, H. Ann. N. Y.
Acad. Sci. 1974, 235, 364.
(67) Ghuysen, J.-M. Annu. Rev. Microbiol. 1991, 45, 37.
(68) Ghuysen, J.-M.; Dive, G. In Bacterial Cell Wall; Ghuysen, J.M., Hakenbeck, R., Eds.; Elsevier: Amsterdam, 1994.
(69) Somner, E. A.; Reynolds, P. E. Antimicrob. Agents Chemother.
1990, 34, 413.
(70) Reynolds, P. E.; Somner, E. A. Drugs Exp. Clin. Res. 1990, 16,
385.
(71) Anderson, J. S.; Matsuhashi, M.; Haskin, M. A.; Strominger, J.
L. Proc. Natl. Acad. Sci. U.S.A. 1965, 53, 881.
(72) Anderson, J. S.; Strominger, J. L. Biochem. Biophys. Res.
Commun. 1965, 21, 516.
(73) Schrader, W. P.; Fan, D. P. J. Biol. Chem. 1974, 249, 4815.
(74) Bupp, K.; van Heijenoort, J. J. Bacteriol. 1993, 175, 1841.
(75) Pless, D. D.; Neuhaus, F. C. J. Biol. Chem. 1973, 248, 1568.
(76) van Heijenoort, Y.; Gomez, M.; Derrien, M.; Ayala, J.; van
Heijenoort, J. J. Bacteriol. 1992, 174, 3549.
(77) Brotz, H.; Bierbaum, G.; Leopold, K.; Reynolds, P. E.; Sahl, H.
G. Antimicrob. Agents Chemother. 1998, 42, 154.
(78) Brotz, H.; Josten, M.; Wiedemann, I.; Schneider, U.; Gotz, F.;
Bierbaum, G.; Sahl, H. G. Mol. Microbiol. 1998, 30, 317.
(79) Hechard, Y.; Sahl, H. G. Biochimie 2002, 84, 545.
(80) Hoffmann, A.; Pag, U.; Wiedemann, I.; Sahl, H. G. Farmaco
2002, 57, 685.
(81) Lo, M.-C.; Men, H.; Branstrom, A.; Helm, J.; Yao, N.; Goldman,
R.; Walker, S. J. Am. Chem. Soc. 2000, 122, 3540.
(82) Branstrom, A. A.; Midha, S.; Goldman, R. C. FEMS Microbiol.
Lett. 2000, 191, 187.
(83) Auger, G.; Crouvoisier, M.; Caroff, M.; van Heijenoort, J.; Blanot,
D. Lett. Pept. Sci. 1997, 4, 371.
(84) Men, H. B.; Park, P.; Ge, M.; Walker, S. J. Am. Chem. Soc. 1998,
120, 2484.
(85) Ye, X.-Y.; Lo, M.-C.; Brunner, L.; Walker, D.; Kahne, D.; Walker,
S. J. Am. Chem. Soc. 2001, 123, 3155.
(86) VanNieuwenhze, M. S.; Mauldin, S. C.; Zia-Ebrahimi, M.; Aikins,
J. A.; Blaszczak, L. C. J. Am. Chem. Soc. 2001, 123, 6983.
(87) Ha, S.; Chang, E.; Lo, M. C.; Men, H.; Park, P.; Ge, M.; Walker,
S. J. Am. Chem. Soc. 1999, 121, 8415.
(88) Chen, L.; Men, H.; Ha, S.; Ye, X. Y.; Brunner, L.; Hu, Y.; Walker,
S. Biochemistry 2002, 41, 6824.

Chemistry and Biology of Ramoplanin


(89) Cudic, P.; Behenna, D. C.; Yu, M. K.; Kruger, R. G.; Szewczuk,
L. M.; McCafferty, D. G. Bioorg. Med. Chem. Lett. 2001, 11, 3107.
(90) Auger, G.; van Heijenoort, J.; Mengin-Lecreulx, D.; Blanot, D.
FEMS Microbiol. Lett. 2003, 219, 115.
(91) Liu, H.; Ritter, T. K.; Sadamoto, R.; Sears, P. S.; Wu, M.; Wong,
C.-H. ChemBiochem 2003, 4, 603.
(92) Li, J. J.; Bugg, T. D. Chem. Commun. 2004, 182.
(93) Helm, J. S.; Chen, L.; Walker, S. J. Am. Chem. Soc. 2002, 124,
13970.
(94) Hu, Y.; Helm, J. S.; Chen, L.; Ginsberg, C.; Gross, B.; Kraybill,
B.; Tiyanont, K.; Fang, X.; Wu, T.; Walker, S. Chem. Biol. 2004,
11, 703.
(95) Schwartz, B.; Markwalder, J. A.; Wang, Y. J. Am. Chem. Soc.
2001, 123, 11638.
(96) VanNieuwenhze, M. S.; Mauldin, S. C.; Zia-Ebrahimi, M.;
Winger, B. E.; Hornback, W. J.; Saha, S. L.; Aikins, J. A.;
Blaszczak, L. C. J. Am. Chem. Soc. 2002, 124, 3656.
(97) Breukink, E.; van Heusden, H. E.; Vollmerhaus, P. J.;
Swiezewska, E.; Brunner, L.; Walker, S.; Heck, A. J.; de Kruijff,
B. J. Biol. Chem. 2003, 278, 19898.
(98) Chen, L.; Walker, D.; Sun, B.; Hu, Y.; Walker, S.; Kahne, D.
Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 5658.
(99) Barrett, D. S.; Chen, L.; Litterman, N. K.; Walker, S. Biochemistry 2004, 43, 12375.
(100) Segel, I. H. Enzyme Kinetics; John Wiley & Sons: New York,
1975.
(101) Hu, Y. N.; Helm, J. S.; Chen, L.; Ye, X. Y.; Walker, S. J. Am.
Chem. Soc. 2003, 125, 8736.
(102) Puyang, X.; Fan, J.; Schumacher, T.; Borsari, B.; Davey, M.; Ling,
L. 43rd Interscience Conference on Antimicrobial Agents and
Chemotheraphy (ICAAC), Chicago, IL, 2003.
(103) Lo, M.-C. Ph.D. Thesis, Princeton University, 2000.
(104) Cudic, P.; Kranz, J. K.; Behenna, D. C.; Kruger, R. G.; Tadesse,
H.; Wand, A. J.; Veklich, Y. I.; Weisel, J. W.; McCafferty, D. G.
Proc. Nat. Acad. Sci. U.S.A. 2002, 99, 7384.
(105) Lo, M.-C.; Helm, J. S.; Sarngadharan, G.; Pelczer, I.; Walker, S.
J. Am. Chem. Soc. 2001, 123, 8640.
(106) Ghadiri, M. R.; Granja, J. R.; Milligan, R. A.; McRee, D. E.;
Khazanovich, N. Nature 1993, 366, 324.
(107) Bong, D. T.; Ghadiri, M. R. Angew. Chem., Int. Ed. 2001, 40,
2163.
(108) Karle, I.; Gopi, H. N.; Balaram, P. Proc. Natl. Acad. Sci. U.S.A.
2002, 99, 5160.
(109) Jiang, W.; Wanner, J.; Lee, R. J.; Bounaud, P.-Y.; Boger, D. L.
J. Am. Chem. Soc. 2003, 125, 1877.
(110) Rew, Y.; Shin, D.; Hwang, I.; Boger, D. L. J. Am. Chem. Soc.
2004, 126, 1041.
(111) Ciabatti, R.; Maffioli, S.; Checchia, A.; Romano, G.; Candiani,
G.; Panzone, G. WO 2003076460, 2003; Chem. Abstr. 2003, 139,
261566.
(112) Ciabatti, R.; Cavalleri, B. Eur. Patent 321696, 1989; Chem. Abstr.
1989, 112, 77952.
(113) Chen, L.; Yuan, Y.; Helm, J. S.; Hu, Y.; Rew, Y.; Shin, D.; Boger,
D. L.; Walker, S. J. Am. Chem. Soc. 2004, 126, 7462.
(114) Schoellkopf, U. Tetrahedron 1983, 39, 2085.
(115) Evans, D. A.; Sjogren, E. B.; Weber, A. E.; Conn, R. E.
Tetrahedron Lett. 1987, 28, 39.
(116) Evans, D. A.; Weber, A. E. J. Am. Chem. Soc. 1986, 108, 6757.
(117) Reno, D. S.; Lotz, B. T.; Miller, M. J. Tetrahedron Lett. 1990,
31, 827.
(118) Seebach, D.; Juaristi, E.; Miller, D. D.; Schickli, C.; Weber, T.
Helv. Chim. Acta 1987, 70, 237.
(119) Belokon, Y. N.; Bulychev, A. G.; Vitt, S. V.; Struchkov, Y. T.;
Batsanov, A. S.; Timofeeva, T. V.; Tsyryapkin, V. A.; Ryzhov,
M. G.; Lysova, L. A.; Bakhmutov, V. I.; Belikov, V. M. J. Am.
Chem. Soc. 1985, 107, 4252.
(120) Jung, M. E.; Jung, Y. H. Tetrahedron Lett. 1989, 30, 6637.
(121) Pons, D.; Savignac, M.; Genet, J. P. Tetrahedron Lett. 1990, 31,
5023.
(122) Woltering, T. J.; Weitz-Schmidt, G.; Wong, C.-H. Tetrahedron
Lett. 1996, 37, 9033.
(123) Shao, H.; Goodman, M. J. Org. Chem. 1996, 61, 2582.
(124) Li, G.; Chang, H.-T.; Sharpless, K. B. Angew. Chem., Int. Ed.
Engl. 1996, 35, 451.
(125) Li, G.; Angert, H. H.; Sharpless, K. B. Angew. Chem., Int. Ed.
Engl. 1997, 35, 22813.

Chemical Reviews, 2005, Vol. 105, No. 2 475


(126) Noyori, R.; Ikeda, T.; Ohkuma, T.; Widhalm, M.; Kitamura, M.;
Takaya, H.; Akutagawa, S.; Sayo, N.; Saito, T.; Taketomi, T.;
Kumobayashis, H. J. Am. Chem. Soc. 1989, 111, 9134.
(127) Genet, J. P.; Pinel, C.; Mallart, S.; Juge, S.; Thorimbert, S.;
Laffitte, J. A. Tetrahedron: Asymmetry 1991, 2, 555.
(128) Kuwano, R.; Okuda, S.; Ito, Y. J. Org. Chem. 1998, 63, 3499.
(129) Morell, J. L.; Fleckenstein, P.; Gross, E. J. Org. Chem. 1977,
42, 355.
(130) Maurer, P. J.; Takahata, H.; Rapoport, H. J. Am. Chem. Soc.
1984, 106, 1095.
(131) Roemmele, R. C.; Rapoport, H. J. Org. Chem. 1989, 54, 1866.
(132) Beaulieu, P. L. Tetrahedron Lett. 1991, 32, 1031.
(133) Blaskovich, M. A.; Lajoie, G. A. Tetrahedron Lett. 1993, 34, 3837.
(134) Saeed, A.; Young, D. W. Tetrahedron 1992, 48, 2507.
(135) Lotz, B. T.; Gasparski, C. M.; Peterson, K.; Miller, M. J. J. Am.
Chem. Soc. 1990, 112, 7.
(136) Vassilev, V. P.; Uchiyama, T.; Kajimoto, T.; Wong, C.-H.
Tetrahedron Lett. 1995, 36, 4081.
(137) Kimura, T.; Vassilev, V. P.; Shen, G.-J.; Wong, C.-H. J. Am.
Chem. Soc. 1997, 119, 11734.
(138) Tominaga, F.; Hiwaki, C.; Maekawa, T.; Voshida, H. J. Biochem.
Tokyo 1963, 53, 227.
(139) Singerman, A.; Liwschitz, Y. Tetrahedron Lett. 1968, 4733.
(140) Okai, H.; Izumiya, N. Bull. Chem. Soc. Jpn. 1969, 42, 3550.
(141) Hanessian, S.; Vanasse, B. Can. J. Chem. 1993, 71, 1401.
(142) Cho, G. Y.; Ko, S. Y. J. Org. Chem. 1999, 64, 8745.
(143) Cardillo, G.; Gentilucci, L.; Tolomelli, A.; Tomasini, C. Synlett
1999, 1727.
(144) Deng, J.; Hamada, Y.; Shioiri, T. J. Am. Chem. Soc. 1995, 117,
7824.
(145) Tohdo, K.; Hamada, Y.; Shioiri, T. Synlett 1994, 247.
(146) Boger, D. L.; Lee, R. J.; Bounaud, P.-Y.; Meier, P. J. Org. Chem.
2000, 65, 6770.
(147) Boger, D. L. Med. Res. Rev. 2001, 21, 356.
(148) Jiang, W.; Wanner, J.; Lee, R. J.; Bounaud, P.-Y.; Boger, D. L.
J. Am. Chem. Soc. 2002, 124, 5288.
(149) Maplestone, R. A.; Cox, J. P. L.; Williams, D. H. FEBS Lett. 1993,
326, 95.
(150) Brady, S. F.; Varga, S. L.; Freidinger, R. M.; Schwenk, D. A.;
Mendlowski, M.; Holly, F. W.; Veber, D. F. J. Org. Chem. 1979,
44, 3101.
(151) Rich, D. H.; Bhatnagar, P.; Mathiaparanam, P.; Grant, J. A.;
Tam, J. P. J. Org. Chem. 1978, 43, 296.
(152) Beaulieu, P. L.; Schiller, P. W. Tetrahedron Lett. 1988, 29, 2019.
(153) Weinreb, S. M.; Demko, D. M.; Lessen, T. A.; Demers, J. P.
Tetrahedron Lett. 1986, 27, 2099.
(154) Huang, J.; Widlanski, T. S. Tetrahedron Lett. 1992, 33, 2657.
(155) Fan, C.-X.; Hao, X.-L.; Ye, Y.-H. Synth. Commun. 1996, 26, 1455.
(156) Li, H.; Jiang, X.; Ye, Y.-H.; Fan, C.; Romoff, T.; Goodman, M.
Org. Lett. 1999, 1, 91.
(157) Holdrege, C. T. U.S. Patent 3646024, 1972; Chem. Abstr. 1972,
77, 34543.
(158) Boger, D. L.; Chen, J.-H. J. Am. Chem. Soc. 1993, 115, 11624.
(159) Boger, D. L.; Chen, J.-H.; Saionz, K. W. J. Am. Chem. Soc. 1996,
118, 1629.
(160) Boeckman, R. K., Jr.; Potenza, J. C. Tetrahedron Lett. 1985, 26,
1411.
(161) Boger, D. L.; Miyazaki, S.; Kim, S. H.; Wu, J. H.; Loiseleur, O.
R.; Castle, S. L. J. Am. Chem. Soc. 1999, 121, 3226.
(162) Boger, D. L.; Miyazaki, S.; Kim, S. H.; Wu, J. H.; Castle, S. L.;
Loiseleur, O.; Jin, Q. J. Am. Chem. Soc. 1999, 121, 10004.
(163) Boger, D. L.; Ledeboer, M. W.; Kume, M. J. Am. Chem. Soc. 1999,
121, 1098.
(164) Boger, D. L.; Ledeboer, M. W.; Kume, M.; Searcey, M.; Jin, Q.
J. Am. Chem. Soc. 1999, 121, 11375.
(165) Kettenring, J. K.; Ciabatti, R.; Winters, G.; Tamborini, G.;
Cavalleri, B. J. Antibiot. 1989, 42, 268.
(166) Wanner, J.; Tang, D.; McComas, C. C.; Crowley, B. M.; Jiang,
W.; Moss, J.; Boger, D. L. Bioorg. Med. Chem. Lett. 2003, 13,
1169.
(167) Ciabatti, R.; Cavalleri, B. Eur. Patent 337203, 1989; Chem. Abstr.
1990, 112, 179893.

CR030106N

Chem. Rev. 2005, 105, 477497

477

Molecular Insights into Aminoglycoside Action and Resistance


Sophie Magnet and John S. Blanchard*
Department of Biochemistry, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, New York 10461
Received April 20, 2004

Contents
1. Introduction
2. Aminoglycosides
2.1. Chemical Structures
2.2. Ribosome Binding Site and Translation
Effects
2.3. Secondary and Bactericidal Effects
2.4. Properties and Clinical Use
3. Aminoglycoside Resistance
3.1. Decreased Intracellular Concentration of the
Drug
3.2. Target Modification
3.2.1. 16S rRNA Methylation
3.2.2. Ribosomal Mutations
3.3. Enzymatic Drug Modification
3.3.1. Aminoglycoside Adenylyltransferases
3.3.2. Aminoglycoside Phosphotransferases
3.3.3. Aminoglycoside Acetyltransferases
3.4. Origin and Prevalence
4. Resisting Resistance
5. Acklnowledgment
6. References

477
478
478
478
482
482
482
482
485
485
485
486
486
487
489
493
494
495
495

1. Introduction
Following the discovery of penicillin, which was
inactive against Mycobacterium tuberculosis, Waksman discovered the first antituberculosis agent,
streptomycin, by systematic screening of bacterial
culture supernatants for the presence of M. tuberculosis inhibitory activity.1 Several years after the
introduction of this aminoglycoside in human antibacterial chemotherapy, organisms resistant to streptomycin began to appear. Subsequently, other antituberculosis agents, including isoniazid, rifampicin,
and ethambutol, were discovered and replaced streptomycin in the treatment of tuberculosis. However,
resistance to these drugs also appeared rapidly. To
reduce the emergence of resistant organisms, a sixmonth-long, multidrug chemotherapy regimen was
adopted. More recently, streptomycin has regained
interest and significance for the treatment of multidrug-resistant (isoniazid- and rifampicin-resistant)
M. tuberculosis infections. In this interval, a large
number of other aminoglycosides were isolated from
various Actinomycetes producers. Among them, gen* Corresponding author. Phone: (718) 430-3096. Fax: (718) 4308565. E-mail: blanchar@aecom.yu.edu.

tamicin, isolated from Micromonospora in 1963,


constituted a significant advance in the treatment of
Gram-negative bacterial infections. Several other
molecules, such as dibekacin and amikacin, were
semisynthesized by modification of natural compounds with the aims of extending their antibacterial
spectrum and potency, reducing their nephro- and
ototoxicity, and evading the resistance mechanisms.
The most recent aminoglycoside introduced into
human antibacterial therapy was arbekacin, a kanamycin B derivative used in Japan since 1990. Today,
the family of aminoglycosides includes a large number of compounds. However, the number of resistance
mechanisms developed by microorganisms has increased in parallel with the number of drugs available and the frequency of their use.
The mechanisms of bacterial resistance to aminoglycosides have been the subject of numerous
genetic and biochemical studies and have been the
focus of many recent reviews.2-4 In some species,
broad spectrum and low level resistance can be
achieved by decreasing the intracellular drug concentration. However, decreased affinity of the drug
for its target, the bacterial ribosome, by modification
of the drug or ribosome is the major cause of aminoglycoside resistance, and among these mechanisms
the enzymatic inactivation of the drug is by far the
most clinically significant. There are three classes of
aminoglycoside-modifying enzymes: the aminoglycoside nucleotidyltransferases (ANTs), the aminoglycoside phosphotransferases (APHs), and the aminoglycoside acetyltransferases (AACs). The reactions
catalyzed by these enzymes are usually regioselective, and the site of modification is indicated in
parentheses. A roman numeral and a letter are then
added to the nomenclature to distinguish the enzymes according to the pattern of aminoglycoside
resistance that they confer and to their primary
sequence, respectively. Nevertheless, in some bacterial species other mechanisms of resistance that
involve unique features of the bacterium are predominant. This is the case for aminoglycoside resistance due to impermeability in Pseudomonas aeruginosa or due to ribosomal modification in M. tuberculosis.
In the past decade, several high-resolution structural studies have been performed to identify the
molecular nature of the interactions between aminoglycosides and the ribosome or the proteins involved in their inactivation via enzymatic modification. This work has brought considerable new insight

10.1021/cr0301088 CCC: $53.50 2005 American Chemical Society


Published on Web 12/08/2004

478 Chemical Reviews, 2005, Vol. 105, No. 2

Magnet and Blanchard

in antibiotic use. Finally, we will present the different


strategies used to design inhibitors of the resistant
determinants or new compounds that can escape
from known mechanisms of resistance.

2. Aminoglycosides
2.1. Chemical Structures

Sophie Magnet was born in Montelimar, France, in 1973. She studied


Biology at the Pierre et Marie Curie University in Paris and received her
Masters degree in Microbiology in 1997. She then did her graduate
research on aminoglycoside resistance in the Unite des Agents Antibacteriens at the Institut Pasteur in Paris. After receiving her Ph.D. in 2001,
she joined the laboratory of Dr. John S. Blanchard at the Albert Einstein
College of Medicine, New York, for a Postdoctoral fellowship in
Biochemistry, during which time she performed mechanistic and kinetic
studies of aminoglycoside acetyltransferases and DNA ligases. She is
returning to France to work on enzymes involved in cell wall synthesis in
Gram-positive bacteria at INSERM.

Aminoglycosides are hydrophilic molecules, consisting of a characteristic, central aminocyclitol linked


to one or more amino sugars by pseudoglycosidic
bond(s). For the majority of clinically useful compounds, referred to in this review as typical aminoglycosides, the aminocyclitol is the 2-deoxystreptamine ring, and it can be monosubstituted in
position 4, as is the case for neamine, or disubstituted
in positions 4 and 5, or 4 and 6. The accepted
nomenclature usually used refers to ring I as the
primed ring and corresponds to the amino sugar at
position 4 of the deoxystreptamine ring. Ring II is
unnumbered and corresponds to the central aminocyclitol, while ring III is referred to as the doubly
primed ring and has the substituent in position 5 or
6 of the deoxystreptamine ring. Ring IV (triply
primed numbering) corresponds to any additional
ring attached to ring III (Figure 1). However, a
number of active aminoglycosides are structurally
atypical according to the above description. For
instance, streptomycin possesses a streptidine ring
as the central aminocyclitol, and spectinomycin consists of three fused rings. The structures of the more
common atypical aminoglycosides are shown in Figure 2.

2.2. Ribosome Binding Site and Translation


Effects

John S. Blanchard was born in Waterbury, CN. He received his B.S. in


Chemistry from Lake Forest College and did his graduate research in
biochemistry in the laboratory of Dr. W. W. Cleland at the University of
Wisconsin. After a three-year Postdoctoral NIH-supported Fellowship, he
was appointed Assistant Professor of Biochemistry at the Albert Einstein
College of Medicine, New York, in 1983 and became the Dan Dancinger
Professor of Biochemistry in 1998. His early interests in enzyme kinetics
and isotope effects focused on flavoenzyme catalyzed reactions. His
research efforts into the mechanism of action of isoniazid in the human
bacterial pathogen Mycobacterium tuberculosis led to his current interest
in antibiotic resistance. He was supported in 2000 by a Fogarty
International fellowship in the laboratory of Dr. Patrice Courvalin at the
Institut Pasteur in Paris, where he studied aminoglycoside N-acetyltransferases and met Dr. Magnet. He is the author of over 100 peer-reviewed
manuscripts and 20 reviews and has been awarded six United States
patents. His work has been generously supported by the National Institutes
of Health for the last 21 years.

into the mechanism of action of various compounds


and into the mechanisms by which bacteria become
resistant to them.
In this review we will describe the current molecular understanding of aminoglycoside action and
resistance, focusing on recent structural advances.
The emergence of new mechanisms of resistance and
the evolution of the distribution of resistance determinants will be presented in parallel with changes

The primary target of aminoglycosides is the bacterial small ribosomal subunit. Aminoglycoside binding
to the 16S rRNA, at the tRNA acceptor A site
(Aminoacyl site), inhibits the translation process by
causing misreading and/or hindering the translocation step. Two crystal structures of the 30S ribosomal
subunit of Thermus thermophilus were solved by
X-ray diffraction methods in 2000.5,6 These studies
brought considerable insights into the components
and the function of the ribosomal A site. It is believed
that the fidelity of translation depends on two steps,
an initial recognition between the codon of the mRNA
and the anticodon of a charged tRNA, and subsequent proofreading. The A site includes portions of
the 530 loop, helix 34, and the base of helix 44. The
tRNA anticodons bind within a cleft formed between
the individual domains, and relative movements of
these domains are likely to be involved in both the
decoding and translocation processes.
The earliest structural studies of complexes containing aminoglycosides were performed using a 27nucleotide-long RNA stem loop which mimicked the
conserved helix 44 moiety of the 16S rRNA A site
that was shown to bind the 2-deoxystreptaminecontaining aminoglycosides. A stoichiometric 1:1
complex was generated with the 4,5-disubstituted
2-deoxystreptamine, paromomycin, and its three-

Molecular Insights into Aminoglycoside Action and Resistance

Chemical Reviews, 2005, Vol. 105, No. 2 479

Figure 1. Structures of clinically useful typical aminoglycosides: A, 4,5-disubstituted deoxystreptamine aminoglycosides;


B, 4,6-disubstituted deoxystreptamine aminoglycosides. The deoxystreptamine ring is shown in red.

dimensional structure was solved using transfer


nuclear Overhauser effect-derived distance constraints and simulated annealing methods.7 The data
revealed that the antibiotic binds in the major groove
of the RNA, where it adopts an L-shaped conformation. The primed ring was bound in a pocket formed
by the non-Watson-Crick A1408-A1493 base pair
and the unpaired A1492, which generates a bulge in
the helical structure. The majority of the intermolecular contacts were made with the central 2-deoxystreptamine ring and the primed 2,6-dideoxy-2,6diamino-glucose ring substituents. Comparison of the
unliganded RNA oligonucleotide with the drug-RNA
complex structure showed that the most dramatic
change occurred at the totally conserved triplet

adenine pocket (A1408, A1492, A1493), which is


displaced toward the minor groove upon paromomycin binding.8 In another report, the same authors
described the structures of three additional aminoglycoside-RNA complexes,9 one containing the
structurally simplest aminoglycoside, neamine, and
two others containing the 4,5-disubstituted 2-deoxystreptamine ribostamycin or neomycin. The conformation of the bound drug, the binding site, and the
intermolecular contacts made with the RNA were
similar for ribostamycin and neomycin to those
observed in the paromomycin complex. In the case
of neamine, two possible orientations were proposed
involving opposite interactions of the 1- and 3-amino
groups of the deoxystreptamine ring. In 2000, Carter

480 Chemical Reviews, 2005, Vol. 105, No. 2

Magnet and Blanchard

Figure 2. Structures of clinically useful atypical aminoglycosides. The aminocycitol ring is shown in red.

et al. described the crystal structures of various


complexes of aminoglycoside bound to the intact 30S
ribosomal subunit.10 The structure of the paromomycin complex revealed, as expected, that the drug
binds at the A site in the major groove of helix 44.
The interactions reported in the NMR structures
were largely confirmed. In addition to a stacking
interaction between the primed ring and G1491, the
hydroxyl groups in positions 4 and 6 make hydrogen
bonds with the N1 of A1408 and the phosphate
oxygen of A1493, respectively (Figure 3, top). The two
amino groups of the deoxystreptamine ring in positions 1 and 3 interact directly with N7 of G1494 and
O4 of U1495, respectively. Besides an intramolecular
hydrogen bond observed between the 2-amino group
and the 5-hydroxyl group, the double primed ribose
ring makes a single interaction with the N7 group of
G1491. Finally, the triply primed 2,6-dideoxy-2,6diamino-glucose ring, which was disordered in the
NMR structure, was clearly observed in the 30S
subunit structure interacting with backbone phosphates from both sides of helix 44, including C1490
and G1405. In the 30S-paromomycin complex, A1492
and A1493 are flipped out compared to the free 30S
subunit and to a greater extent than previously
observed in the NMR structure. The structures of two
4,6-disubstituted aminoglycosides, gentamicin C1a11
and tobramycin,12 bound to a single or a dimeric
synthetic A site RNA, respectively, were also solved
by X-ray crystallography. Both complexes were very
similar, and only small differences were observed in
the position of the double primed ring. The interactions with specific nucleotides as well as with phosphate backbone oxygen atoms and the central and
primed rings were, as expected, very similar to those
observed for the 4,5-disubstituted deoxystreptamines.
The 2-hydroxyl group and 3-amino group of ring
III make additional specific contacts with O6 and N7
of G1405, respectively, compared to the 4,5-disubstituted aminoglycosides (Figure 3, bottom). Two additional intramolecular hydrogen bonds were also

Figure 3. Interactions between typical aminoglycosides


and the 30S ribosome. Top: Interactions between paromomycin and the 30S ribosome (Reprinted with permission
from Nature (http://wwww.nature.com), ref 10. Copyright
2000 Nature Publishing Group.). Bottom: Interactions
between tobramycin and the 30S ribosome (Reprinted with
permission from ref 12. Copyright 2002 Elsevier). The
superscripts refer to the nucleotide functional group that
interacts with the bound aminoglycoside. The nucleotide
numbering is for the homologous E. coli ribosome.

revealed between the 5-hydroxyl group and the ring


III oxygen and between the 1-amino and the 2hydroxyl groups of the drug.
The selection of aminoacyl-tRNA during translation occurs by formation of a minihelix between the

Molecular Insights into Aminoglycoside Action and Resistance

codon of the mRNA and the anticodon of the cognate


tRNA. When a tRNA-mRNA complex is formed,
A1492 and A1493 form a hydrogen-bonding network
with the ribose 2-hydroxyl groups of the two first
bases of both the codon and the anticodon, allowing
discrimination between different base pairing geometries. When a cognate mRNA-tRNA complex is so
recognized, the two adenines flip out from the helix
with the net effect of increasing the affinity for the
cognate tRNA and so stabilizing the complex. The
structural data presented above showed that the
binding of a disubstituted 2-deoxystreptamine in the
decoding center causes a similar conformational
change to the formation of a cognate mRNA-tRNA
complex. These data confirm biochemical results that
show that this class of aminoglycosides both decreases the rate of A site tRNA dissociation13 and
increases the tRNA binding affinity.14 As a consequence, the affinity of the A site for a noncognate
mRNA-tRNA complex is increased upon drug binding, preventing the ribosome from efficiently discriminating between noncognate and cognate complexes. This provides a dramatic atomic level rationalization of the well-known miscoding effect of the
disusbtituted 2-deoxystreptamines antibiotics.
The structures of three structurally atypical aminoglycosides, streptomycin, spectinomycin, and hygromycin B, bound to the 30S subunit, have also been
reported. The structure of the streptomycin-30S
complex revealed that the drug binds tightly to the
A site, but its binding site is adjacent to the one of
the disubstituted deoxystreptamines.10 In contrast to
paromomycin, which interacts only with residues
contained within helix 44, streptomycin makes interactions with the backbone phosphates and ribose
hydroxyl groups of residues from four different
domains of the 16S rRNA molecule, including U14
in helix 1, C526 and G527 of the 530 loop in helix
18, A913 and A914 from helix 27 and 28, respectively,
and C1490 and G1491 from helix 44 (Figure 4, top).
In addition, streptomycin is the only aminoglycoside
that interacts with ribosomal protein side chains.
Indeed, the structure of the complex showed that the
-amino group of K45 of the S12 ribosomal protein
contacts both the 4- and 5-hydroxyl groups of the
streptamine ring. This structure has provided considerable insights into the dynamic functions that
occur during aminoacyl-tRNA binding to the A site.
It was proposed that H27, which interacts with H44,
can have two alternative base pairing schemes during
translationsone which leads to a ribosomal ambiguity (ram) conformation of the decoding center, with
high affinity for tRNA which results in increased
miscoding, and a second that leads at the opposite
to a restrictive state with low tRNA affinitysand the
balance of theses two states could be involved in the
proofreading process.15 The binding site of streptomycin revealed by the structural data suggests that
the antibiotic affects the dynamic equilibrium of the
two ribosomal conformations by stabilizing the ram
state, providing an explanation for the error prone
effect of this drug.10 However, a confirmation of this
model would require an atomic resolution structure
of the restrictive form of the ribosome.

Chemical Reviews, 2005, Vol. 105, No. 2 481

Figure 4. Interactions between atypical aminoglycosides


and the 30S ribosome. Top: Interactions between streptomycin and the 30S ribosome (Reprinted with permission
from Nature (http://wwww.nature.com), ref 10. Copyright
2000 Nature Publishing Group.). Bottom: Interactions
between hygromycin B and the 30S ribosome (Reprinted
with permission from ref 16. Copyright 2000 Elsevier). The
superscripts refer to the nucleotide functional group that
interacts with the bound aminoglycoside. The nucleotide
numbering is for the homologous E. coli ribosome.

The structure of spectinomycin bound to the 30S


subunit was also described in the same paper.10 The
rigid molecule binds in the minor groove of the 16S
rRNA, at the end of helix 34 near to the pivot point
of the head region. The drug makes four hydrogen
bonds with the bases G1064, C1066, and C1192 of
helix 34 in addition to the interaction with the 2ribose hydroxyl group of G1068 contained within
helix 36. These observations lead the authors to
propose that the binding of spectinomycin, known to
inhibit the EF-G catalyzed GTP-dependent translocation of the peptidyl-tRNA from the A site to the P
site (Peptidyl site), prevents the movement of the
head region and helix rearrangements necessary for
the translocation step.
Finally, the structure of the 30S-hygromycin B
complex has been reported by the same group in a
separate paper.16 Hygromycin B is active against both
prokaryotic and eucaryotic ribosomes. It acts primarily by inhibiting the translocation step but also, to
a lesser extent than streptomycin or the disubstituted
2-deoxystreptamines, causes miscoding. Hygromycin
B binds just above the binding site of paromomycin
at the very top of helix 44. The ring I substituents
interact with both the backbone phosphate oxygen
atoms of G1494 and the bases G1405 G1494, U1495,
and C1496 (Figure 4, bottom). The rest of the
molecule makes only weak base specific hydrogen
bonds with C1403 and U1498. Ring IV is positioned

482 Chemical Reviews, 2005, Vol. 105, No. 2

very close to the P site. The binding of hygromycin


B between the A site and the P site explains its dual
effect on translation accuracy and translocation.
Moreover, the absence of specific contact between
hygromycin B and A1408, a residue totally conserved
among eubacteria, can certainly account for the
affinity of this molecule for eucaryotic rRNA, which
possesses a G at the corresponding position.17

2.3. Secondary and Bactericidal Effects


Whereas spectinomycin and kasugamycin, two
inhibitors of the translocation step, act bacteriostatically, streptomycin and the disubstituted 2-deoxystreptamines are bactericidal antibiotics. Although
the mechanism of action of aminoglycosides at the
translational level was substantially clarified by the
structural data detailed above, the origin of the
bactericidal effect of most of these compounds remains enigmatic. Comparison of the effect of aminoglycosides on translation and on cell viability
suggests that lethality is correlated with the production of mistranslated proteins, which could have fatal
secondary effects. It was shown that aminoglycosides
that cause misreading, like streptomycin, increase
the passive permeability of the cell membranes to
several small molecules, as originally probed by the
release of K+ ions.18 This permeabilization presumably occurs as a result of the incorporation of truncated or incorrectly folded proteins in the membranes, since it has not been observed in mutant
strains in which protein synthesis is insensitive to
streptomycin by mutation in the rpsL gene.19-21 This
phenomenon results in an increased aminoglycoside
uptake that leads to an irreversible accumulation of
the drug inside the cell. It was proposed that following the rapid degradation of the membrane-associated mistranslated proteins the drug becomes trapped
inside the cell.18 The resulting high intracellular drug
concentration is believed to play a critical role in the
bactericidal effect.21 The saturation of the ribosomes
with aminoglycosides, potentially coupled to the
inhibition of new ribosome synthesis and assembly,22
could result in the complete inhibition of protein
synthesis, leading to bacterial death. Alternatively,
the production of mistranslated proteins might affect
another vital cellular processes. For instance, it has
been reported that aminoglycosides that cause misreading inhibit the initiation of DNA replication,20
but this finding requires more detailed investigations
that have not been pursued.

2.4. Properties and Clinical Use


Because of their poor oral absorption, aminoglycosides are most often administered parenterally in
order to achieve therapeutically adequate serum
concentrations. As a consequence of their polar
structure, the drugs inefficiently cross biological
membranes and thus their intracellular tissue concentration is low except in the proximal renal tubule,
where they are concentrated.23,24 Whereas the conventional dosing for gentamicin, tobramycin, or netilmicin is around 5 (mg/kg)/day, 3-fold higher doses are
used for other compounds such as streptomycin or

Magnet and Blanchard

amikacin.23,25 For patients with normal renal activity,


a single daily dosing was shown to be more favorable
than the traditionally recommended multiple daily
dosing regimens. Because of the concentration-dependent activity and postantibiotic effect of aminoglycosides, a similar efficacy can be obtained by
single daily dosing with a lower cost, reduced toxicity,
and reduced emergence of an adaptive resistant
population.25-31 In addition, aerosolized or liposomeencapsulated aminoglycosides have been shown to be
advantageous for the treatment of respiratory tract
infections32-34 or against intracellular bacteria like
Mycobacterium avium,35-38 respectively. Finally,
therapy strategies such as alternating between different classes of antibiotics, switch therapy, or the
rotational use of different aminoglycosides are increasingly encouraged to prevent therapeutic failure
due to resistance.31,39,40
Aminoglycosides are active against a wide range
of aerobic Gram-negative bacilli, many staphylococci,
mycobacteria, and some streptococci.4,23,24,41,42 They
are particularly useful for the treatment of neutropenic patients and serious infections caused by
aerobic Gram-negative bacteria. Gentamicin continues to be the aminoglycoside of choice to treat
hospital acquired enterobacteriaceae and P. aeruginosa infections. Most often it is used in combination
with a -lactam, which results in a synergistic
bactericidal effect due to an enhanced uptake of
aminoglycosides.43 Other inhibitors of bacterial cell
wall synthesis can also be coadministered with aminoglycosides to treat infections due to bacterial
species that are naturally resistant to aminoglycosides because of impaired uptake, including the
enterococci.41,42 For the treatment of urinary tract
infections, aminoglycosides can be used alone in
monotherapy. More recently introduced aminoglycosides, such as amikacin or arbekacin, which are not
substrates for a number of aminoglycoside-modifying
enzymes23,44 or which retain their antibacterial activity after modification,45 are used to treat infections
due to gentamicin-resistant organisms, including
methicillin-resistant Staphylococcus aureus.46-50 Finally, streptomycin is still used in multidrug chemotherapy to treat multidrug-resistant M. tuberculosis.23,39

3. Aminoglycoside Resistance
3.1. Decreased Intracellular Concentration of the
Drug
Decreased aminoglycoside concentration inside a
target cell, by reduction of drug uptake, activation
of drug efflux, or both, will affect the susceptibility
of the strain to the whole class of aminoglycoside
compounds and can be the cause of intrinsic or
acquired resistance. Although the exact mechanism
of aminoglycoside uptake remains unknown, it is
accepted that the process consists of three consecutive steps.51,52 The first step is the adsorption of the
cationic compounds to the surface of bacteria by
electrostatic interactions with the negatively charged
lipopolysaccharides of the outer membrane of Gramnegative bacteria. The next two steps are dependent

Molecular Insights into Aminoglycoside Action and Resistance

on the transmembrane potential generated by the


respiratory chain, with the second being characterized by a faster rate of uptake. As a result, anaerobic
bacteria are intrinsically resistant to aminoglycosides
due to impermeability.53 Similarly, respiratory chain
mutants or strains containing functional mutations
in their ATP synthases were shown to exhibit decreased susceptibility to aminoglycosides.51,54 Such
mutants have been isolated from clinical or experimental endocarditis caused by infection with Escherichia coli, S. aureus, or P. aeruginosa.55 Changes
in membrane components involved in the initial
electrostatic binding of aminoglycosides have also
been associated with increased levels of resistance,
especially in the case of P. aeruginosa.56 Clinical
strains exhibiting low level resistance to gentamicin
were shown to have a modified, less negatively
charged, lipopolysaccharide that exhibits a lower
affinity for gentamicin.57 In addition, the extracellular alginate produced by mucoid strains of P. aeruginosa, which both inhibits phagocytosis by monocytes
and neutrophils and enhances bacterial adherence
to the respiratory epithelia, was shown to decrease
the uptake and early bactericidal effect of aminoglycosides. It was proposed that the viscous polyanionic
alginate gel acts as a physical and ionic trap for the
drug.58
Energy-dependent bacterial efflux is now recognized as a major cause of antibiotic resistance. This
is particularly true for the multidrug-resistant opportunist pathogens responsible for nosocomial infections, which have to counter the environmental
pressure exerted by the constant presence of antibiotics. Bacterial species constitutively expressing such
transporters are intrinsically resistant to low levels
of various antibiotics. Moreover, mutations in the
regulatory genes of the pumps, or induction of
expression in the presence of substrate, can lead to
the overexpression of the originally constitutive, or
silent, pump genes. It had been thought that multidrug transporters were specific for hydrophobic or
amphiphilic compounds and, thus, that aminoglycosides would not be affected by this mechanism of
resistance. However, in the last several years, aminoglycosides have been shown to be substrates for a
number of multidrug efflux pumps, including members of the five superfamilies of bacterial transporters.59
Recently, crystallographic structural studies on
different components of the tripartite transporters of
the RND (resistance nodulation cell division) superfamily, which play a particularly important role in
Gram-negative bacteria, have been revealing.60,61 The
transporters of the RND superfamily use the membrane proton-motive force as energy source. They are
localized in the cytoplasmic membrane, and in Gramnegative bacteria they interact with a membrane
fusion protein (MFP), located in the periplasmic
space, and an outer membrane protein (OMP) to form
a continuous, tripartite channel able to export substrates directly out of the cell. E. coli AcrB, which
exhibits broad substrate specificity but does not efflux
aminoglycosides, has served as the structural prototype of a bacterial RND protein involved in antibiotic

Chemical Reviews, 2005, Vol. 105, No. 2 483

resistance. AcrB interacts with the MFP, AcrA, and


the OMP, TolC. The structures of the AcrB apoprotein,60 AcrB in complex with four substrates (rhodamine 6G, ethidium bromide, dequalinium, and ciprofloxacin),62 and TolC63 have been solved. Both
membrane proteins exist as homotrimers. Trimeric
TolC forms a barrel composed of 12 -strands that
span the outer membrane and 12 R-helices that
extend into the periplasmic space over 100 . The
internal cavity is open to the external medium and
provides solvent access. Each monomer of AcrB
contains 12 transmembrane domains and two large
periplasmic domains. The transmembrane domains
of the three protomers are arranged in a ringlike
manner, creating a 30 diameter cavity, where
substrates bind. This cavity is connected to the
periplasmic funnel by a very narrow pore, and the
dimensions of the funnel are compatible with a direct
interaction with TolC. Between the AcrB monomers,
an opening formed by residues of the periplasmic
domaims is observed that links the central channel
to the periplasm. It was proposed that these openings
(vestibules) provide a way for substrates selected
from the outer leaflet of the cell membrane or from
the periplasmic space to gain access to the channel
and be exported60 (Figure 5). Such a model explains
the structurally broad substrate specificity of this
type of transporter, since both amphiphilic compounds, which can partially penetrate the lipid
bilayer, and polycationic molecules such as aminoglycosides, which interact electrostatically with the
phospholipids of the membrane, can be captured at
the entrance of the vestibule.61 The selection of efflux
substrates would then be determined by the nature
of the residues at the entrance of the openings. This
is supported by domain-swapping experiments between two RND proteins of P. aeruginosa, MexY and
MexB,64 and two RND proteins of E. coli, AcrB and
AcrD,65 indicating that the periplasmic domains are
responsible for efflux specificity. The presence of
many more acidic residues at the entrance of the
vestibules of AcrD, which does export aminoglycosides, compared to AcrB, which does not, also supports this model.61 The structures of AcrB in complex
with various substrates have shown than once they
reach the central cavity, the structurally distinct
ligands bind to different sites, with a stoichiometry
of one per protomer.62 Three conserved charged
residues, located in transmembrane helices 4 and 10,
might constitute the transmembrane proton translocation site.60 Protonation of these residues would
disrupt the ion pair and trigger a conformational
change leading to the pore opening. A preliminary
structure of AcrA by electron crystallography reveals
that the protein has an elongated shape of 100-200
overall length,66 but the nature of the contacts
made with AcrB and TolC, as well as its role in the
system, is still very poorly understood.
Several RND proteins have been shown to be
involved in intrinsic and/or acquired, proton motive
force-dependent, aminoglycoside resistance in various
Gram-negative pathogens, including P. aeruginosa,
Burkholderia pseudomallei, Acinetobacter baumannii, and E. coli. The disruption or deletion of the

484 Chemical Reviews, 2005, Vol. 105, No. 2

Magnet and Blanchard

Figure 5. Schematic representation of a model proposed for drug capture at the outer leaflet of the cytoplasmic membrane
by a trimeric RND protein. Hydrophilic and positively charged compounds such as aminoglycosides may bind to the acidic
outer surface of the membrane, and amphiphilic compounds such as fluoroquinolones or chloramphenicol may partially
diffuse into the lipid bilayer before being recognized by specific interactions at the periplasmic vestibules of the RND and
drawn into the central cavity: OMP, outer membrane protein; RND, resistance nodulation cell division.

genes encoding the RND proteins MexY from P.


aeruginosa,67,68 AmrB from B. pseudomallei,69 and
AcrD from E. coli70 resulted in the hypersusceptibility
of the strains to aminoglycosides, indicating that
these proteins contribute to the intrinsic resistance
of these species to aminoglycosides. Whereas the
AmrAB system seems to be specific for aminoglycosides and macrolides, others exhibit a very broad
spectrum of antibacterial substrate specificity. Although no clear correlation has been made between
the level of expression of MexXY and the MIC values
of aminoglycosides in clinical isolates of P. aeruginosa, overexpression of the mexXY structural genes
has been shown to be responsible for high-level
acquired resistance in some cases.71 The MexXY
system also appears to be involved, in combination
with membrane impermeability, in the ability of P.
aeruginosa to develop adaptative resistance in response to exposure to inhibitory concentrations of
aminoglycosides.72 The AdeABC system of A. baumannii is another example of an RND pump mediating acquired aminoglycoside resistance. It was shown
that the inactivation of the adeB gene from a multidrug-resistant clinical isolate of A. baumannii
restored the susceptibility of the strain to various
drugs, including aminoglycosides (from 2- to 32-fold),
fluoroquinolones, cefotaxime, erythromycin, tetracycline, chloramphenicol, and trimethoprim.73
Gene(s) encoding putative transcriptional regulator(s) are frequently present in the immediate environment of the structural genes of the efflux system.
The genes amrR69 and mexZ,68 and the two component regulatory gene homologues adeRS,73 have been
identified upstream from the corresponding efflux
genes. While the regulator proteins of some efflux
systems belonging to distinct superfamilies of transporters, such as BmR from Bacillus subtilis or QacR

from S. aureus, have been the subject of detailed


studies showing that they modulate the transcription
of the adjacent genes following drug binding,74,75 very
little is known about the regulators of the RND
family transporters. Although mutations in mexZ
have been associated with increased expression of
mexXY, other results suggest that the regulation of
the system is a more complex process involving
various components, including yet uncharacterized
factors.71 On another hand, the expression of the
OMP component could play an important role in the
functional expression of these efflux systems. In the
case of P. aeruginosa, no gene encoding a cognate
OMP was found in the closed environment of mexXY,
suggesting that the pump can use independently
encoded protein(s) to form an efficient tripartite
system. Although it has been shown that MexXY can
act together with OprM, a more recent report identified two other OMPs, OpmG and OpmI, that can
contribute to aminoglycoside resistance in this species.76
Members of the major facilitator superfamily (MFS)
of transporters have also been shown to decrease
aminoglycoside susceptibility in strains harboring the
structural gene on a multicopy plasmid. The MFS
proteins are implicated in the active transport of both
sugars and antibiotics. They contain 12 or 14 transmembrane segments and use the proton motive force
as energy source. MdfA from E. coli was the first
putative MFS protein reported to have an effect on
aminoglycoside resistance.77 Althought this effect,
observed in a strain of E. coli harboring the gene on
a plasmid, was very modest and not clinically relevant (2- to 3-fold increases in the MIC values of
kanamycin and neomycin), this finding led to other
investigations on the role of MFS proteins in other
species. The tap and P55 genes, isolated from Myco-

Molecular Insights into Aminoglycoside Action and Resistance

bacterium fortuitum78 and Mycobacterium bovis,79


respectively, encode putative MFS proteins and
conferred resistance to aminoglycosides, including
streptomycin, when cloned in Mycobacterium smegmatis. Genes and proteins homologous to P55/P55
were detected by polymerase chain reaction (PCR)
or by western blot analysis, in many Mycobacterium
species, including M. tuberculosis (Rv1410c). These
observations, together with the identification of a
total of 16 open reading frames that encode putative
MFS proteins in the genome of M. tuberculosis,80 may
account for streptomycin-resistant clinical isolates of
M. tuberculosis that cannot be assigned to mutations
in rpsL or rrs genes or enzymatic modification of the
drug.

3.2. Target Modification


3.2.1. 16S rRNA Methylation
The lack of post-transcriptional methylation of
A1518 and A1519 16S rRNA nucleotides, as a result
of mutations in the ksgA gene, which encodes an
S-adenosylmethionine (SAM)-dependent RNA methylase, was associated with resistance to kasugamycin,
an inhibitor of the initiation step of translation, in
E. coli and Bacillus stearothermophilus.81-83 Although this example is not clinically relevant, it
illustrates the influence of the 16S rRNA methylation
pattern on the interactions with aminoglycosides.
Members of the Actinomycetes produce inactive
aminoglycosides, including acetylated- or phosphorylated-forms, which are cleaved during or after their
export out from the cell by specific enzymes, to form
the active antibiotics.84-86 To further resist the
secreted active compounds, many aminoglycosideproducing organisms also express rRNA methylases
capable of modifying the 16S rRNA molecule at
specific positions critical for the tight binding of the
drug.87 A number of genes encoding such enzymes
have been identified from several aminoglycoside
producers.88-95 The corresponding rRNA methyltransferases form the Agr family of methyltransferases (for aminoglycoside resistance).96 Some of
these enzymes have been characterized. KamA from
Streptomyces tenjimariensis and KamB from Streptomyces tenebrarius catalyze the modification of
A1408 at the N1 position and confer high-level
resistance to kanamycin, tobramycin, sisomicin, and
apramycin but not gentamicin.87,88 GmrA from the
gentamicin producer Micromonospora purpurea and
KgmB from S. tenebrarius catalyze the modification
of G1405 at the N7 position, conferring high-level
resistance only to the 4,6-disubstituted deoxystreptamines, including gentamicin.87,89 Methylation of
these nucleotides presumably abolishes the intermolecular contacts that they make with the drug
(discussed previously in section 2.2 and Figure 3).
The specific interaction observed between ring III of
the 4,6-disubstituted deoxystreptamines and G1405
is consistent with the resistance pattern conferred
by G1405 methylation. The presence of a methyl
group at the 6 position of some compounds of the
gentamicin mixture could modify the interaction with
A1408 compared to those observed with tobramycin,

Chemical Reviews, 2005, Vol. 105, No. 2 485

explaining why modification of this residue does not


confer resistance to gentamicin.
Until recently, genes encoding a 16S rRNA methyltransferase had been restricted to the aminoglycoside producers, but three reports in 2003 and 2004
described the characterization of similar genes in
clinical isolates of human Gram-negative pathogens.
The rmtA and rmtB genes, located on plasmid-borne
transposons, were found in clinical isolates of P.
aeruginosa and Serratia marcescens, respectively.97,98
These strains were all isolated in Japan, where
arbekacin has been used extensively since 1990. The
two genes share 82% sequence identity, and the
encoded Rmt enzymes confer high-level resistance
(MICs > 1024 g/mL) to almost all clinically useful
aminoglycosides, including arbekacin, but not streptomycin. The considerable primary sequence similarity observed between the Rmt proteins and the 16S
rRNA methylases of Actinomycetes, as well as the
high G-C content of the gene (55%), suggests a
possible gene transfer from the producing organisms
to Gram-negative pathogens. Another 16S rRNA
methylase was characterized from Klebsiella pneumoniae. The structural gene, armA, was located on
a plasmid containing several other resistant genes,
including those conferring resistance to -lactams,
trimethoprim, sulfonamides, and other aminoglycoside resistance determinants. In contrast to rmtA and
rmtB, the low G-C content of armA (30%) does not
suggest a direct or recent acquisition from the Actinomycetes. However, the ArmA methylase is able to
confer high-level resistance to essentially all aminoglycosides except streptomycin.99 The site specificity of the modification catalyzed by these two enzymes was not determined, but the pattern of
resistance associated with ArmA, which includes all
4,6-disubstituted deoxystreptamines but not apramycin, led the authors to propose that the N7 of
G1405 is the locus of modification. The armA gene
has been detected by PCR in several other Enterobacteriaceae isolated from different European countries.99 Because these methylases can modify all
copies of the 16S rRNA and lead to high-level
resistance to an extremely wide range of compounds,
the emergence of this resistance mechanism in human pathogens is of concern for the future, especially
considering that the structural genes can apparently
be easily disseminated.

3.2.2. Ribosomal Mutations


Resistance to aminoglycosides by mutation of the
ribosomal target is clinically relevant only for streptomycin in M. tuberculosis. Mycobacterium is the only
genus of eubacteria with species that contain a single
copy of the ribosomal operon, which implies that a
single mutation can lead to the production of a
homogeneous population of mutant ribosomes and
thus can result in resistance regardless of the recessive nature of the mutation involved. The mutations
in the rrs gene, encoding the 16S rRNA and associated with streptomycin resistance in M. tuberculosis,
affect two highly conserved regions, the 530 loop and
the region around nucleotide 912, according to E. coli
numbering,100-105 and result in decreased affinity for

486 Chemical Reviews, 2005, Vol. 105, No. 2

Magnet and Blanchard

Figure 6. Reaction catalyzed by aminoglycoside O-adenylyltransferases. The reaction shown is that catalyzed by ANT(4)
catalyzing the MgATP-dependent 4-O-adenylylation of kanamycin A.

streptomycin.106 The structure of the streptomycin30S ribosomal subunit complex revealed specific
interactions between the drug and the backbone
phosphate or ribose hydroxyl groups of nucleotides
C526 and G527 of helix 18 and A913 and A914 of
helixes 27 and 28, respectively,10 providing a rationale for the location of mutations previously identified and their effect on streptomycin binding. Apart
from clinically significant streptomycin resistance in
M. tuberculosis, a few reports have described 16S
rRNA mutations associated with aminoglycoside
resistance in clinical isolates of microorganisms
containing a single or low copy number of rrs genes.
Mutations at positions 1400 and 1401 were found in
kanamycin-resistant isolates of M. tuberculosis,107 the
mutation A1408G (corresponding to the eucaryotic
allele of this position) was identified in the unique
rRNA operon of Mycobacterium abscessus and Mycobacterium chelonae isolates resistant to 2-deoxystreptamine-containing aminoglycosides, and mutations affecting the base pair G1064-C1192 of helix
34 were found in the three copies of the 16S rRNA of
spectinomycin-resistant isolates of Neisseria.108
The introduction of a single rrs gene on a multicopy
plasmid has been used for many years to study the
effect of 16S rRNA mutants on the activity of aminoglycosides in a heterogeneous population of ribosomes or homogeneous purified mutant ribosomes.
These studies have shown that at least half of the
ribosomes must be in the mutant form to confer
aminoglycoside resistance.106,109,110 More recently, to
circumvent the problem of the recessive nature of
ribosomal mutations, strains containing a single copy
of the rrs gene have been genetically engineered.111-113
Synthetic oligonucleotides mimicking the A site have
also been used for similar in vitro studies.114 These
studies have confirmed the effect of the naturally
occurring mutations previously shown to affect streptomycin activity,115 and they revealed other mutations associated with spectinomycin,116,117 hygromycin
B,118,119 or 4,6-disubstitued 2-deoxystreptamine113,120-122
resistance. These studies have shown that mutations
leading to a steric or allosteric change in the drugbinding pocket can be more deleterious for antibiotic
activity than mutations abolishing direct contacts
between the drug and the 16S rRNA.
Mutations in genes encoding ribosomal proteins
can also alter the activity of aminoglycosides. Notably, mutations in protein S12 are the other major
cause of streptomycin resistance in M. tuberculosis100,101,105,123 and other species.124,125 Mutations occurred in two regions, located around residues 42 and
87 (E. coli numbering), that contact helixes 18, 27,
and 44 of the 16S rRNA. The most frequent substitu-

tions were found to occur at residues P41, K42, K87,


and P90. Although structural data showed that S12
makes direct contact with streptomycin,10 the effect
of such mutations appears likely to be conformational
changes in the rRNA that prevent drug binding. This
conclusion, highlighted by structural data and chemical protection experiments, was first predicted by the
observation that different mutations in S12 lead to
different phenotypes: streptomycin resistance or
streptomycin dependence.115,126-129 In addition, S12
mutations can be compensated for by other mutations
in the rRNA or in other ribosomal proteins.128-131
With the exception of K42R, mutations in S12 are
associated with a hyperaccurate phenotype that can
lead to dependence on streptomycin, which stabilizes
the ram state of the A site. The streptomycindependent phenotype can be relieved by ram mutations in proteins S4 and S5 and located at the
interface of the two proteins.127-129,132 Other mutations altering rRNA, EF-Tu, or S12 itself can also
compensate for streptomycin dependence,128 indicating that all three components are involved in the
conformational stability of the A site. Restrictive
mutations located in 50S subunit components, including truncation of the C terminus of L6 or substitution of G2661, have also been associated with
resistance to various aminoglycosides.126,131 Such
mutations do not affect the binding of aminoglycosides but likely compensate for the ram phenotypic
effect of these drugs by increasing the affinity of the
EF-Tu for 50S. Finally, mutations in the N-terminal
half of S5, which contact helix 34, can confer resistance to spectinomycin by destabilizing the network
of interactions in the 30S subunit, allowing the head
region to move even in the presence of the drug.10,133

3.3. Enzymatic Drug Modification


3.3.1. Aminoglycoside Adenylyltransferases
Aminoglycoside adenylyltransferases, with only 10
enzymes identified to date, include examples of both
chromosomally encoded and plasmid-encoded enzymes. In Gram-negative organisms, the ant(2) and
ant(3) genes encoding adenylyltransferases are often
identified on mobile genetic elements in resistant
organisms. In Gram-positive organisms, the ant(4),
ant(6), and ant(9) genes are also found on plasmids
or integrated into transposons.134 These enzymes
catalyze the reaction between Mg-ATP and aminoglycoside to form the O-adenylylated aminoglycoside and the magnesium chelate of inorganic pyrophosphate (Figure 6). Enzymes that regioselectively
adenylylate the 6 and 3 positions and the 9 and 3
positions, of the atypical aminoglycosides streptomy-

Molecular Insights into Aminoglycoside Action and Resistance

cin and spectinomycin, respectively, have been identified. From a clinical perspective, the reactions
catalyzed by the ANT(2) and ANT(4) are of most
significance and have been the most thoroughly
mechanistically and structurally studied.
The earliest mechanistic studies of the adenylyltransferases were those of Lombardini135 and
Northrop136-139 on the E. coli ANT(2). Both found
the kinetic mechanism to be sequential, requiring
both substrates to be present before catalysis could
take place, and Lombardini suggested an ordered
mechanism of substrate binding, with nucleotide
(ATP) binding before aminoglycoside. In a detailed
series of kinetic investigations, Northrop argued for
this same order of substrate addition but added the
ordered release of inorganic pyrophosphate followed
by the rate-limiting release of the adenylylated
aminoglycoside. The slow release of this final product
makes the kinetic mechanism Theorell-Chance. The
enzyme exhibits comparable activity with all nucleotide triphosphates and even their deoxy derivatives.
It also shows activity with a broad array of 4,6disubstituted substrates, but the relative V/K values
for these substrates vary by a factor of 4000. Interestingly, there is a significant positive correlation
between the enzymatic V/K values for aminoglycoside
substrates and the MIC values for these aminoglycosides in strains expressing ANT(2).140
The S. aureus ANT(4) has also been studied in
significant detail. The enzyme was initially identified
from a kanamycin- and gentamicin-resistant clinical
strain141,142 and subsequently shown to have activity
with a large number of aminoglycosides possessing
either 4- or 4-hydroxyl substituents.143 The kinetic
mechanism is sequential, and the stereochemistry at
the R-phosphorus atom of ATP has been shown to
undergo inversion during turnover,144 suggesting that
adenylyl transfer occurred via a direct displacement
of the leaving group, inorganic pyrophosphate, by the
nucleophilic hydroxyl group of the aminoglycoside.
In 1993, Holden reported the three-dimensional
crystal structure of the enzyme145 and subsequently
reported the structure of the ANT(4)-kanamycinAMg-AMPCPP (R,-methylene-ATP) ternary complex.146 These structures were determined using a
thermostable mutant of the wild-type enzyme, and
this was both the first structure of an aminoglycosidemodifying enzyme and the first structure of an
aminoglycoside in complex with a modifying enzyme.
The structure of the complex revealed an unusual
dimeric arrangement of monomers with obvious Nand C-terminal domains of approximately equal size
(Figure 7). In general, residues from the N-terminal
domain interact with the nucleotide, while those of
the C-terminal domain interact with the aminoglycoside. The two bound nucleotides are far apart, and
the majority of interactions are between the triphosphate moiety and the enzyme, suggesting a
structural basis for the lack of nucleotide triphosphate specificity. The two bound kanamycin A molecules are as close as 3.5 . There are at least four
negatively charged side chains of aspartate and
glutamate residues that interact with the aminoglycoside, and one of these, glutamate 145, appears

Chemical Reviews, 2005, Vol. 105, No. 2 487

Figure 7. Structure of the Staphylococcus aureus


ANT(4)-AMP-PNP-kanamycin A complex. The monomers making up the active dimer are shown as blue and
yellow ribbons. AMP-PNP is shown in stick representation
and colored by atom type (C, gray; N, blue; O, red; P, pink).
Kanamycin is shown in stick representation and colored
by a different atom type (C, green; N, blue; O, red).
Coordinates were obtained from the Protein Data Bank
(1KNY).

positioned to act as a general base to deprotonate the


4-hydroxyl group. The distance from the 4-hydroxyl
to the R-phosphorus atom of AMPCPP is 5.0 ,
suggesting to these authors that a precatalytic conformational change must occur to allow for the
nucleophilic attack on the nucleotide. Alternatively,
the use of the nonreactive nucleotide might not allow
for the appropriate positioning of the nucleotide, or
contacts in the crystal do not allow this positioning
to be obtained. The 4-hydroxyl group and pyrophosphate moiety are positioned for a direct, in-line
attack at the R-phosphorus atom of AMPCPP. Due
to the symmetry of kanamycin A and the resolution
at which the structure was determined (2.5 ), it was
not possible to unambiguously distinguish between
two conformations in which either the 4- or 4hydroxyl groups would be adenylylated.
This last problem was recently solved by the
assignment of the regioselectivity of the reaction
using NMR.147 In this report, the exclusive monoadenylylation of kanamycin A was demonstrated, as
was the exclusive 4-regiospecificity. A priori, both the
4- and 4-hydroxyl groups in this symmetric molecule could have been adenylylated, and there are
examples of ANT(4,4) isozymes that, when presented with a substrate lacking the 4-hydroxyl group
(dibekacin), adenylylate at the 4 position.148 Finally,
using sensitive 18O kinetic isotope effect methods and
a poor substrate, m-nitrobenzyl triphosphate, the
transition state for the enzymatic 4-adenylylation of
kanamycin A was shown to be associative.149

3.3.2. Aminoglycoside Phosphotransferases


Aminoglycoside phosphotransferases catalyze the
regiospecific transfer of the -phosphoryl group of
ATP to one of the hydroxyl substituents present on
aminoglycoside (Figure 8). They represent a large

488 Chemical Reviews, 2005, Vol. 105, No. 2

Magnet and Blanchard

Figure 8. Reaction catalyzed by aminoglycoside O-phosphotransferases. The reaction shown is that catalyzed by APH(3)
catalyzing the MgATP-dependent 4-O-phosphorylation of kanamycin A.

class of aminoglycoside-modifying enzymes and are


particularly relevant to clinical resistance to aminoglycosides in Enterococcal and Staphylococcal species. The vast majority of the biochemical and structural work on this family has focused on the plasmidencoded APH(3)-IIIa enzyme from Enterococcus
faecalis, as will this discussion. The 264 amino acid
protein was overexpressed, purified, and shown to
catalyze phosphorylation of a broad spectrum of
aminoglycosides,150 as expected by the broad range
of aminoglycosides to which the original clinical
strain was resistant (tobramycin, a 4,6-disubstituted
aminoglycoside lacking a 3-hydroxyl substituent, is
not a substrate for the enzyme, but rather a potent
inhibitor). The enzyme binds substrates tightly, with
steady-state affinities in the low micromolar range,
and the kinetic parameter for aminoglycosides which
correlated with the MIC values was V/K. Enzymecatalyzed generation of phosphorylated kanamycin
A followed by one- and two-dimensional NMR analysis revealed that the proposed 3-regioselectivity,
based on the resistance phenotype, was correct.
Several extensions of these regioselectivity studies
have been performed. On the basis of the resistance
phenotypes observed for a battery of aminoglycosides,
it was proposed that for aminoglycosides lacking a
3-hydroxyl substituent, such as the 4,5-substituted
lividomycin A, phosphorylation could occur on the 5hydroxyl substituents of the ribose ring.151 Using ,bidentate Cr-ATP as a paramagnetic probe, the
conformations of amikacin, a 4,6-disubtituted aminoglycoside, and butirosin A, a 4,5-disubstituted
aminoglycoside, bound to the APH(3)-IIIa were
probed.152 The conformations of the two types of
aminoglycosides were quite different, and the modeled conformations allowed only the 3-hydroxyl substituents of amikacin to approach the -phosphate
of ATP, whereas the conformation of enzyme-bound
butirosin allowed both the 3- and 5-hydroxyl substituents of butirosin to approach the -phosphate of
ATP. These results were confirmed and extended,
and the exclusive monophosphorylation of the 3hydroxyl substituents of 4,6-disubstituted aminoglycosides was demonstrated. Those 4,5-disubstituted
aminoglycosides containing both 3- and 5-hydroxyl
substituents were shown to be rapidly monophosphorylated and subsequently bis-phosphorylated.153
These results reveal that the APH(3)-IIIa has a
remarkable ability to bind a large number of structurally distinct aminoglycosides and catalyze their
phosphorylation with a range of regioselectivities.
The steady-state kinetic mechanism was shown to
be sequential on the basis of the intersecting pattern

of lines observed in the reciprocal plot, arguing


against mechanisms invoking a phosphoenzyme intermediate.154 ATP binds first to the enzyme followed
by aminoglycoside, since tobramycin, a potent deadend inhibitor, exhibits linear uncompetitive inhibition
versus ATP and linear competitive inhibition versus
kanamycin A. On the basis of product inhibition
studies and the uncompetitive substrate inhibition
exhibited by aminoglycosides versus ATP, the ordered release of phosphorylated drug followed by the
slow rate-limiting release of ADP was proposed. This
corresponds to a Theorell-Chance-type kinetic mechanism, where the binding of the aminoglycoside to
the E-ADP complex results in the observed uncompetitive substrate inhibition. Strong support for this
Theorell-Chance kinetic mechanism came from a
subsequent study, employing viscosity variation,
solvent kinetic isotope effect measurements, and the
use of -thio-ATP.155
The three-dimensional structure of the APH(3)IIIa-ADP complex was reported in 1997.156 Although
the selenomethionine substituted enzyme had been
crystallized in the presence of Mg-ATP, the electron
density due to the bound nucleotide was only compatible with a bis-magnesium-chelated-ADP complex.
This suggests that, during the month-long crystallization, hydrolysis of ATP occurred and that the
more tightly bound ADP product complex was observed. The enzyme existed as a doubly disulfidebonded dimer in the crystal, confirming earlier
results showing the enzyme to be active as both a
monomer and dimer, and with this equilibrium being
affected by the presence of disulfide reducing agents.150
Each 263 amino acid monomer, lacking only the
N-terminal methionine residue, folded into a 94residue N-terminal domain and a 157 residue Cterminal domain connected by a flexible 12-residue
linker (Figure 9). The bound ADP molecule was
observed in the cleft between the two domains. The
overall architecture of the APH(3)-IIIa, and in
particular the predominantly -strand N-terminal
domain, was similar to the structures previously
reported for serine/threonine protein kinases, including the catalytic subunit of the cAMP-dependent
protein kinase. Residues at the active site included
two conserved lysine residues, K33 and K44, E60,
D190, N195, and D208. K33 and D190 make no
interactions with bound ADP, but all others make
direct or water-mediated interactions with the pyrophosphate moiety of the nucleotide. These results
drove a series of site-directed mutagenesis studies
that confirmed an important role for K44 in nucleotide binding and an essential role for D190, which

Molecular Insights into Aminoglycoside Action and Resistance

Figure 9. Structure of the Enterococcus faecalis APH(3)Mg2-ADP complex. The monomer is shown composed of red
R-helices and yellow -strands. ADP is shown in stick
representation and colored by atom type (C, gray; N, blue;
O, red; P, pink), and the two magnesium atoms are shown
as purple spheres. Kanamycin is shown in stick representation and colored by a different atom type (C, green; N,
blue; O, red). Coordinates were obtained from the Protein
Data Bank (1L87).

was proposed to function as the general base in the


deprotonation of the 3- (or 5-) hydroxyl group of
bound aminoglycoside.
The three-dimensional structure of the APH(3)IIIa in complex with ADP and either kanamycin A
or neomycin B157 provided a molecular explanation
for many of the above-referenced studies and others
which probed the importance of residues in the
carboxyl terminal domain using site-directed mutagenesis.158 As expected, both aminoglycosides make
the majority of interactions with residues of the
C-terminal domain and the overwhelming majority
of these interactions are electrostatic. Thus, the
positively charged aminoglycosides interact with
E157, E160, D190, E230, D262, E262, and the
carboxyl terminus of F264. Aspartate 190 interacts
with the 3-hydroxyl, supporting its role as the
general base responsible for initiating catalysis by
deprotonating the 3-hydroxyl substituent. In the
neomycin B complex, the ribose 5-hydroxyl is pointed
toward D190, but the distance between the two is not
consistent with D190 acting as a base, in catalysis.
The majority of the interactions are between the
enzyme and the central deoxystreptamine and the
primed, 6-deoxy-6-aminoglucose ring. Two additional
subsites were identified that provide interactions to
the doubly primed rings of 4,6-substituted and 4,5substituted aminoglycosides. Although there are no
large domain movements accompanying aminoglycoside binding, residues 147-170 close over the
bound aminoglycoside and constitute a binding loop
that firmly fixes the conformation of the aminoglycoside. The conformations of the bound aminoglycosides were compared to those observed for structurally related aminoglycosides bound to the A site and
to the entire 30S ribosomal subunit. Remarkably for
these conformationally flexible tricyclic compounds,
the structures were quite similar, with root mean
squared (rms) deviations of 1.7 between neomycin

Chemical Reviews, 2005, Vol. 105, No. 2 489

B bound to APH(3)-IIIa and paromomycin bound to


the 16S rRNA of the 30S subunit. The authors
suggest that these are low energy conformations and
that, while of obviously different character, the
kinase active site has evolved to mimic the rRNA
binding site and tempt the drug into this binding
pocket before it can find its true target-binding site.
While eukaryotic protein kinases and APH(3)-IIIa
do not exhibit high degrees of overall sequence
homology, the presence of highly conserved residues
at the nucleotide binding site and the structural
similarity of the proteins suggested that the two
proteins were evolutionarily related. Whether these
similarities extended to the functional ability of the
bacterial APH(3)-IIIa to phosphorylate proteins was
demonstrated in 1999.159 APH(3)-IIIa was shown to
be capable of phosphorylating several, but not all,
basic peptides, including the MARCKS (myristolated
alanine-rich C-kinase substrates) K and R peptides
as well as protamine and myelin basic protein at
rates that were 10-20% that of kanamycin A phosphorylation. The specificity of the activity was limited, and the enzyme was incapable of phosphorylating kemptide, histone 1, and peptide substrates for
tyrosine protein kinases. Phosphoamino acid analysis
of the product phosphopeptides revealed that the
phosphorylation occurred on the serine residue of the
MARCKS K peptide, although this peptide contains
neither a threonine nor tyrosine residue (Ac-FKKSFKL-NH2).
Although this section has focused on the E. faecalis
APH(3)-IIIa, in S. aureus a bifunctional enzyme
containing both a kinase domain and an acetyltransferase domain (see below) is of major clinical significance. This enzyme is termed AAC(6)-Ie-APH(2)Ia and, because of its dual kinase-acetyltransferase
activities, can provide resistance to the majority of
clinically useful aminoglycosides. The kinetic mechanism of the kinase reaction of the bifunctional
enzyme differs from that of APH(3)-IIIA, being rapid
equilibrium random.160 APH(2) is inactivated by the
lipid kinase inhibitor, wortmannin, but contrary to
APH(3)-IIIa is not inactivated by the ATP analogue,
5-[p-(fluorosulfonyl)benzoyl]adenosine.161,162 However, like APH(3)-IIIa, the enzyme can use protein
kinase substrates to perform phosphorylation on
serine residues.159

3.3.3. Aminoglycoside Acetyltransferases


Well over four dozen unique sequences exist for
aminoglycoside acetyltransferases. These enzymes
catalyze the acetyl-CoA-dependent N-acetylation of
one of the four amino groups of typical aminoglycosides (Figure 10). They include enzymes that acetylate the 1- and 3-amino groups of the central deoxystreptamine ring and enzymes that acetylate the 2and 6-amino groups of the primed, 6-deoxy-6-aminoglucose ring. Two distinct AAC(1) activities have been
identified in E. coli and Actinomycete strains,163,164
but their importance is minor, because the E. coli
enzyme does not modify the clinically useful aminoglycosides and the other, which exhibits a broader
substrate specificity, is not found in human pathogens.

490 Chemical Reviews, 2005, Vol. 105, No. 2

Magnet and Blanchard

Figure 10. Reaction catalyzed by aminoglycoside N-acetyltransferases. The reaction shown is that catalyzed by AAC(6)
catalyzing the acetyl-CoA-dependent 6-N-acetylation of kanamycin A.

There are five aminoglycoside acetyltransferases


that catalyze regioselective acetylation of the 2amino group, and they all are chromosomally encoded
and species specific. The first to be identified was the
178 amino acid AAC(2)-Ia from Providencia stuartii.165 In wild-type strains, the low level of expression
of the aac(2)-Ia gene is not sufficient to confer
aminoglycoside resistance, but the transcription of
the structural gene was shown to be subject to a
complex regulation.166 Mutations in the aac(2)-Ia
gene resulted in altered levels of peptidoglycan
O-acetylation and cell morphology,167 suggesting that
peptidoglycan acetylation might be a physiologic
function of the enzyme. The expressed and purified
enzyme was shown biochemically to exhibit broad
specificity for aminoglycosides containing 2-amino
substituents. The relative V/K values for aminoglycosides did not correlate with in vivo MIC values,168
supporting the idea that the enzyme functions physiologically in other reactions, possibly peptidoglycan
acetylation. All other members of the AAC(2) family
have been identified in mycobacterial species. They
were originally identified in rapidly growing species
of mycobacteria.169 The aac(2)-Ib gene identified in
M. fortuitum appeared to be transcriptionally silent
in its natural host, but its expression in M. smegmatis
resulted in increased MIC values for all 2-aminosubstituted aminoglycosides.170 Subsequently, other
aac(2) genes were identified in both M. smegmatis
and M. tuberculosis. The corresponding enzymes
were shown to bear 60-70% sequence identity with
the M. fortuitum enzyme but only 30-40% identity
with the P. stuartii AAC(2)-Ia. None of the encoded
proteins bore significant sequence homology (<10%)
to other proteins, including other aminoglycoside
acetyltransferases.
The M. tuberculosis AAC(2)-Ic, chromosomally
encoded by the Rv0262 gene, was expressed and
purified and shown to catalyze the acetyl-CoAdependent acetylation of a broad range of aminoglycoside substrates. In contrast to several other aminoglycoside acetyltransferases, the AAC(2)-Ic activity,
even though highest with aminoglycosides containing
a 2-amino substitutent, could also be detected with
kanamycin A and amikacin, both of which contain a
2-hydroxyl substituent,171 suggesting that this enzyme can catalyze O-acetylation. Both standard
steady-state and alternative substrate kinetics supported the ordered addition of acetyl-CoA followed
by aminoglycoside to generate a ternary complex
from which acetyltransferase chemistry ensued. Very
modest solvent kinetic isotope effects were observed,
suggesting that chemistry was not rate limiting. The

Figure 11. Structure of the Mycobacterium tuberculosis


AAC(2)-CoA-ribostamycin complex. The monomers making up the active dimer are shown as ribbons that are
colored by sequence position (N-terminal, green > yellow
> red > blue, C-terminal). CoA is shown in stick representation and colored by atom type (C, gray; N, blue;
O, red; P, pink). Ribostamycin is shown in stick representation and colored by atom type (C, green; N, blue; O, red).
Coordinates were obtained from the Protein Data Bank
(1M4G).

appearance of substrate activation at high aminoglycoside concentrations suggested that acetylated


aminoglycoside dissociated first, followed by the slow
release of CoA, and that binding of aminoglycoside
to the CoA binary complex could increase this products rate of dissociation.
The three-dimensional structures of the full-length
181 amino acid apo form of the enzyme, plus three
different ternary complexes containing bound CoA
and tobramycin, kanamycin A, or ribostamycin, were
recently reported at resolutions between 1.5 and 1.8
.172 Although lacking any sequence homology to
other aminoglycoside N-acetyltransferases, the monomer fold was nearly identical to those of the S.
marcescens AAC(3) and E. faecalis AAC(6)-Ii, identifying AAC(2)-Ic as a member of the Gcn5-related
N-acetyltransferase (GNAT) superfamily.173 The conserved features of this fold are an N-terminal R-helix,
a central, antiparallel three-stranded -sheet, and a
helix-sheet-helix at the C-terminus of the fold
(Figure 11). This fold serves to bind and orient the
phosphopantothenoyl arm of acetyl-CoA, while very
few interactions are made between the enzyme and
the adenine ring. The majority of interactions between the enzyme and aminoglycoside occur between
acidic residues in the active site (D35, D40, E82,

Molecular Insights into Aminoglycoside Action and Resistance

D152, D179, and the carboxyl terminus of W181) and


the hydroxyl and amino substituents of the central
deoxystreptamine and primed rings, either directly
or via intervening water molecules. The 2-amino
group of the bound substrate is positioned 3.8 from
the sulfur atom of bound CoA, confirming the regioselective nature of the acetylation. The C-terminal
carboxylate of W181 is hydrogen-bonded through a
series of water molecules to the 2-amino group of
bound ribostamycin, suggesting that this residue
functions as the general base responsible for activating the amine and deprotonating the zwitterionic
tetrahedral intermediate. V84 and G83 are positioned
in the beta bulge that is conserved in GNAT
superfamily members, but a model of the complex
containing acetyl-CoA suggests that a single hydrogen bond between the thioester carbonyl of the
substrate and the backbone amide nitrogen of V84
can be accommodated. The phenolic hydroxyl group
of Y126 is appropriately positioned to function as a
general acid, protonating the leaving thiolate of CoA.
Together, these structures suggest that the enzyme
binds the two substrates tightly, in an orientation
that facilitates catalysis, and provides enzyme side
chains that can function as both general base (W131)
and general acid (Y126). Thus, in spite of the strongly
favorable thermodynamics of the reaction, the enzyme assists in carbonyl polarization, base-assisted
amine protonation, and general acid-assisted breakdown of the tetrahedral intermediate.
The AAC(3) family of aminoglycoside acetyltransferases is one of the largest and includes four major
types, I-IV, based on the pattern of aminoglycoside
resistance that they confer. The first aminoglycosidemodifying enzyme to be purified to homogeneity was
the E. coli R-plasmid-encoded gentamicin acetyltransferase.174 This allowed for the first studies of the
substrate specificity of these enzymes and the establishment of a correlation between measured MIC
values and the kinetic parameter V/K, indicative of
the efficiency of substrate utilization. These early
studies revealed that the kinetic mechanism was
sequential but was dependent on the identity of the
substrate: with good substrates such as sisomicin,
the mechanism was random, while, with poor substrates such as tobramycin, the kinetic mechanism
was rapid equilibrium random, with chemistry being
rate-limiting for the overall reaction.175 The demonstration of uncompetitive substrate inhibition suggested that aminoglycosides could bind to the product
binary E-CoA complex. A subsequent detailed structure-activity relationship allowed important interactions involving the 2-, 6-, 3-, and 3-amino groups
to be defined.140 Finally, these studies made it clear
that the kinetic parameter V/K for the aminoglycoside was positively correlated with in vivo activity
against strains expressing the enzyme.
The AAC(3)-I from S. marcescens, originally identified in 1991,176 was the first aminoglycoside acetyltransferase whose three-dimensional structure was
determined.177 The enzyme-CoA complex structure
was determined at a resolution of 2.3 , allowing the
interactions between the enzyme and the product to
be identified. The monomer fold was typical of the

Chemical Reviews, 2005, Vol. 105, No. 2 491

Figure 12. Structure of the Serratia marcescens AAC(3)CoA complex. The monomers making up the active dimer
are shown as ribbons that are colored by sequence position
(N-terminal, green > yellow > red > blue, C-terminal). CoA
is shown in stick representation and colored by atom type
(C, gray; N, blue; O, red; P, pink). Coordinates were
obtained from the Protein Data Bank (1BO4).

GNAT superfamily, with the characteristic central


antiparallel -sheet covered by the two amino terminal helices on one side and the two carboxy
terminal helices on the other (Figure 12). While the
oligomeric property of the enzyme in solution was not
discussed, the enzyme crystallizes with a dimer in
the asymmetric unit, suggesting that, like most
GNAT superfamily members, the dimer is functionally active. Unfortunately, no structures containing
bound aminoglycosides were reported, leaving the
discussion of how these enzymes regioselectively
catalyze acetylation open. The recent studies of the
AAC(6) subgroup of acetyltransferases have allowed
this issue to be resolved.
As probed by the structural data of bound aminoglycosides to the 30S ribosomal subunit,5,7,9,10 the
6-amino group plays an important role in target
binding and the subsequent antibacterial activity of
the drug. This substituent is thus not surprisingly
the target of one of the major classes of aminoglycoside-modifying enzymes, the AAC(6) class, which
includes more than 25 members. The most common
pattern of resistance associated with the production
of these enzymes (type I) includes the majority of the
useful aminoglycosides except the mixture of gentamicins. Among this subclass, three enzymes have
been extensively studied, the two species specific and
chromosomally encoded AAC(6)-Ii and AAC(6)-Iy
from E. faecium and Salmonella enterica, respectively, and the widespread, plasmid-encoded, bifunctional AAC(6)-Ie-APH(2).
In E. faecium the chromosomally encoded AAC(6)Ii is, in part, responsible for the intrinsic low-level
resistance to aminoglycosides of the species. The first
structural studies of aminoglycoside binding to
AAC(6)-Ii involved the use of NMR spectroscopy and
molecular modeling to determine the conformations
of two aminoglycosides, the 4,6-disubstituted deoxystreptamine isepamicin and the 4,5-disubstituted
butirosin A, in the active site of the enzyme.178 These
data showed that, in the ternary AAC(6)-Ii-CoAisepamicin complex, the drug can adopt two distinct
conformations, which are able to interconvert. The
structure of AAC(6)-Ii with either the substrate
acetyl-CoA179 or the product CoA180 bound at the
active site was subsequently reported. The overall

492 Chemical Reviews, 2005, Vol. 105, No. 2

Figure 13. Structure of the Enterococcus feacium


AAC(6)-CoA complex. The monomers making up the
active dimer are shown as ribbons that are colored by
sequence position (N-terminal, green > yellow > red >
blue, C-terminal). CoA is shown in stick representation and
colored by atom type (C, gray; N, blue; O, red; P, pink).
Coordinates were obtained from the Protein Data Bank
(1N71).

fold of the AAC(6)-Ii monomer revealed that the


enzyme was a member of the GNAT superfamily
characterized by a structurally conserved core (see
above). The shape of the monomer can be likened to
a V, with the acetyl-CoA binding site positioned
between the two arms. Most of the interactions
formed between the enzyme and acetyl-CoA involved
the main chain atoms, and only the side chains of
K149 and T89 interact with the substrate. Although
the monomer fold was quite similar to those of other
GNAT superfamily members, the structure of the
AAC(6)-Ii dimer revealed a significant diversity in
the dimer assembly (Figure 13). The dimeric enzyme
exhibits broad substrate specificity for acyl donors
and aminoglycosides containing a 6-amino group.181
Initial velocity and inhibition studies performed with
desulfo-CoA are consistent with an ordered sequential kinetic mechanism with acetyl-CoA binding first
and CoA released last.182 This result is consistent
with the location of the acetyl-CoA binding site in
the bottom of the active site. The rate-limiting steps
of the reaction were explored by solvent viscosity and
solvent isotope effects. The results showed that
diffusion-controlled events (substrate binding and/
or product release) were the rate-limiting steps of the
reaction rather than chemistry.182 The potential
catalytic roles of several residues located in the
AAC(6)-Ii active site were investigated kinetically
using different mutant forms of the enzyme.183 These
studies showed that none of the residues mutagenized (Q72, H74, L76, and Y147) perturb the
structural integrity of the enzyme and none are
involved in general base or acid catalysis. However,
the results suggest that Q72 may play a role in
aminoglycoside recognition and orientation in the
active site and that the amide group of L76 is
involved in transition state stabilization. Another
mutant form of AAC(6)-Ii, W104A, which does not
form a dimer in solution, was also produced to
investigate subunit cooperativity in the AAC(6)-Ii
dimer that was suggested by the partial and mixed
inhibition kinetic patterns.182 The unusual inhibition

Magnet and Blanchard

patterns were alleviated when the mutant monomeric form of the enzyme was used, and isothermal
calorimetry (ITC) analysis of aminoglycoside binding
to WT AAC(6)-Ii revealed that two nonequivalent
binding sites exist in the dimer, supporting a subunit
cooperativity. The structural homology between eucaryotic histone acetyltransferases and AAC(6)-Ii, as
well as the relatively inefficient aminoglycoside
acetyltransferase activity displayed by AAC(6)-Ii
(kcat/Km 104 M-1 s-1) and the lack of correlation
between V/K values for aminoglycosides and
MIC values, has led the authors to investigate the
ability of the enzyme to acetylate other substrates.
AAC(6)-Ii is capable of acetylating small basic proteins such as calf histones, myelin basic protein, or
ribonuclease A.179
The S. enterica AAC(6)-Iy has been shown to confer
broad aminoglycoside resistance to a clinical isolate
in which a chromosomal deletion lead to the expression of the usually silent structural gene.184 The
purified recombinant AAC(6)-Iy was expressed in E.
coli and shown to exist as a dimer in solution. The
enzyme exhibits a high preference for acetyl-CoA as
the acyl donor (Km ) 10 M) and a strict specificity
for the aminoglycosides containing a 6-amino group;
lividomycin A, which possess a 6- hydroxyl substituent, is a powerful inhibitor of the reaction. On the
basis of their kinetic behavior, the aminoglycoside
substrates can be divided into two classes, one that
exhibits Michaelis-Menten kinetics and a second
that displays substrate activation.185 The thermodynamic parameters of substrate binding, obtained
from both fluorescence spectroscopy and ITC, showed
that both aminoglycosides and acyl-CoAs can bind
to the free enzyme and that aminoglycoside binding
to AAC(6)-Iy is strongly enthalpically driven.186
Kinetic and thermodynamic studies performed using
the wild type or mutant forms of the enzyme suggest
that C70 is essential for drug binding at the active
site. Steady-state kinetics and alternative antibiotic
kinetics indicated that the enzyme displays a sequential kinetic mechanism. Dead-end inhibition performed with desulfo-CoA and lividomycin A together
with the dependence of V and V/Kacetyl-CoA on the
identity of the aminoglycoside used argued for the
random order of substrate binding to the enzyme. The
inequality of the solvent isotope effects on D2OV and
D2O
V/K suggests that release of CoA is rate-limiting.185
The three-dimensional structure of the enzyme,
solved at 2.4 resolution by multiwavelength anomalous diffraction methods, placed AAC(6)-Iy in the
Gcn5-related N-acetyltransferase (GNAT) superfamily.187 While the tertiary structure of the monomer
is very similar to those observed for other members
of the superfamily, the structure of the active dimer
consists of a continuous 12-stranded -sheet characterized by a carboxyl terminal strand exchange
(Figure 14). This exchange has not been previously
observed with aminoglycoside N-acetyltransferases
but has been observed in the dimeric yeast histone
acetyltransferase Hpa2, which is the closest structural homologue of AAC(6)-Iy. Although not added
in the crystallization solution, CoA was found in both
active sites of the enzyme formed by a negatively

Molecular Insights into Aminoglycoside Action and Resistance

Figure 14. Structure of the Salmonella enterica AAC(6)CoA-ribostamycin complex. The monomers making up the
active dimer are shown as ribbons that are colored by
sequence position (N-terminal, green > yellow > red >
blue, C-terminal). CoA is shown in stick representation and
colored by atom type (C, gray; N, blue; O, red; P, pink).
Kanamycin is shown in stick representation and colored
by atom type (C, green; N, blue; O, red). Coordinates were
obtained from the Protein Data Bank (1S3Z).

charged channel at the dimer interface. In the


AAC(6)-Iy-CoA-ribostamycin ternary complex, described in the same paper, ribostamycin is stacked
between the aromatic rings of W22 and Y66, and the
primed and central rings make contacts with the side
chains of E79, D115, and E136. The 6-amine of the
antibiotic is 3.4 away from the sulfur atom of CoA,
consistent with a direct nucleophilic attack of the
amine on the thioester. D115 was proposed to function as the general base via an intervening water
molecule between D115 and the 6-amino group.
Similarly, it was proposed that the protonation of the
thiolate after collapse of the tetrahedral intermediate
occurs using a water molecule accessible to bulk
solvent. The comparison of the two ternary complexes
of CoA and ribostamycin bound to AAC(6)-Iy or
AAC(2)-Ic provided a molecular basis for the regioselective activity of these enzymes. In the AAC(6)Iy-ribostamycin complex, the central and primed
rings make contacts with N-terminal elements of the
enzyme and the 2-amine makes an intramolecular
hydrogen bond with the 5-hydroxyl group of the
ribose ring orientating the 6-amino group toward
acetyl-CoA. In the AAC(2)-Ic complex, the primed
and central rings make contacts with C-terminal
elements of the enzyme and the 6-amino group
makes an intramolecular hydrogen bond with the 2hydroxyl group of the ribose ring, which is rotated
almost 180 relative to that of the AAC(6)-Iy complex. This change results in the positioning of the 2amino group toward acetyl-CoA. Finally, intimate
contacts between two dimers of recombinant AAC(6)-Iy were seen in the crystal structure, such that
the extended affinity tag-containing N-terminal extension of one dimer is bound into the active site of
an adjacent dimer. This observation led to the
demonstration that AAC(6)-Iy is also capable of
autoacetylation, acetylation of eucaryotic histones
and the human histone H3 N-terminal peptide in a
regioselective manner.187
The bifunctional AAC(6)-Ie-APH(2)-Ia, which
confers broad spectrum, high-level aminoglycoside
resistance in Enterococci and Staphylococci, differs
from the two AAC(6)s described above in its genetic
localization and catalytic mechanism.188 The struc-

Chemical Reviews, 2005, Vol. 105, No. 2 493

tural gene of the enzyme is generally found on


transposable elements, frequently carried on R plasmids. These mobile supports account for the intergenus transfer of the resistant determinant, originally isolated from E. faecalis. The enzyme is
monomeric in solution, and the acetyltransferase
activity exhibits exceptionally broad substrate specificity for aminoglycosides including fortimcin A and
aminoglycosides possessing a hydroxyl group at the
6-position. The O-acetylated paromomycin product
was identified on the basis of its lability under basic
conditions and infrared spectrometry analysis.188 The
ability of AAC(6)-Ie to catalyze both N- and Oacetylation was attributed to the presence of a
general base that would assist in hydroxyl deprotonation. The solvent kinetic isotope effect, pH studies,
and site-directed mutagenesis identified D99 as the
active site base required for aminoglycoside O-acetylation and N-acetylation of fortimicin A.189 Finally,
a mutant form of this enzyme, exhibiting an arbekacin 4-N-acetyltransferase activity, was identified in
a methicillin-resistant strain of S. aureus in Japan.
It was proposed that the G80D mutation affects the
conformation of the protein and leads to a change in
its enzymatic regiospecificity from 6- to 4-acetylation.190 The bifunctional AAC(6)-APH(2) enzyme
has been proposed to arise by gene fusion to confer
resistance to a wider spectrum of aminoglycosides to
its bacterial hosts than either enzyme alone, and it
can develop an AAC(4) activity to more efficiently
modify arbekacin, which is widely used in Japan.
This example illustrates the remarkable ability of
bacteria to rapidly adapt to changes in antibiotic use
and selective pressure.

3.4. Origin and Prevalence


Acquired aminoglycoside resistance is primarily
due to the expression of enzymes that catalyze the
chemical modification of the drug, thus preventing,
or substantially weakening, their interaction with the
ribosome. Whereas some enzymes are encoded by
chromosomal genes specific for a bacterial genus or
species, the majority of the structural genes encoding
inactivating enzymes, both in Gram-negative and
Gram-positive bacteria, are located on transferable
genetic elements including plasmids, transposons,
and integrons, which allowed their dissemination.191-195 For these modifying enzymes, a positive
correlation between the V/K values for aminoglycosides and the MIC values obtained in vivo for the
same aminoglycosides in strains expressing one of
these enzyme is observed. Northrop has argued that,
for the greatest efficiency, an aminoglycoside-modifying enzyme would be expected to exhibit such a
correlation.140 On the contrary, there is usually no
correlation between the V/K values for aminoglycosides and the MIC values for strains expressing a
originally chromosomally encoded modifying enzyme.
This observation suggests that enzymes carried on
mobile genetic elements have evolved to be efficient
aminoglycoside-modifying enzymes and have been
disseminated for this purpose. Their presence may
thus be the result of the dissemination of the resistant determinants of producer microorganisms. On

494 Chemical Reviews, 2005, Vol. 105, No. 2

another hand, chromosomally encoded and species


specific enzymes likely have other physiological functions in their bacterial host and are selectively
recruited to counter antibacterial agents. The possible physiological functions of some AAC(2) and
AAC(6) acetyltransferases have been discussed in
the literature;167,172,179,187 however, the physiological
substrates for such bacterial GNAT proteins have not
been unambiguously identified.
As a consequence of the influence of selective
pressure on the acquisition of transferable resistance
genes, the distribution of specific inactivating enzymes varies depending on the geographic area and
on specific aminoglycoside use.196,197 For example, the
incidence of resistance due to expression of AAC(6)-I
enzymes capable of inactivating amikacin is significantly higher in countries where this antibiotic is
used extensively, including France, Belgium, and
Greece, but is less frequent in other European
countries or in the United States.196 Because mobile
genetic supports usually harbor several antibioticresistant genes, their acquisition often results in a
multidrug resistance phenotype. Besides this genetic
linkage, the simultaneous selective pressure of various antibiotics can also be the origin of multidrug
resistance acquisition, illustrated by the higher frequency of aminoglycoside resistance in organisms
simultaneously resistant to another class of drugs.
For example, a study performed under the European
SENTRY program indicated that while the percentage of gentamicin resistance was only 7% in strains
of methicillin-susceptible S. aureus, it reached 80%
in methicillin-resistant strains.198 The emergence of
multidrug-resistant organisms expressing efflux
pumps and responsible for nosocomial infections is
another consequence of the multidrug pressure.
Finally, mechanisms of resistance such as permeability alteration and target mutation are not horizontally transferable, but they remain important and
can account for a high percentage of the resistant
population of a particular species.197 The recent
characterization of genes encoding methyltransferases that catalyze the modification of the 16S rRNA
and are located on transferable elements portends a
wide and rapid dissemination of these enzymes,
responsible for broad spectrum and high-level aminoglycoside resistance, soon.

4. Resisting Resistance
On the basis of the therapeutic revival of the
-lactam class of antibacterials upon the introduction
of formulations containing a -lactamase inhibitor
and the parent -lactam, interest in the development
of inhibitors of aminoglycoside-modifying enzymes
has increased sharply. The first report of such an
inhibitor was by Northrop, who semisynthetically
prepared the bisubstrate analogue of kanamycin B
and CoA (Figure 15).199 This exerted powerful inhibition versus the aminoglycoside acetyltransferase,
exhibiting a Ki value of 9 nM, and did not restore
sensitivity to aminoglycosides in strains expressing
the N-acetyltransferase, undoubtedly because the
compound was incapable of penetrating the bacterial
envelope. The ability of the natural product 7-hy-

Magnet and Blanchard

Figure 15. Structure of the bisubstrate analogue described by Williams and Northrop.

Figure 16. Structure of the bis-neamine dimer described


by Wong et al.

droxytropolone to inhibit the ANT(2) enzyme and


to restore sensitivity to aminoglycosides in E. coli
strains harboring the ant(2) gene was reported in
1982.200 A more recent report has attempted to
synthesize compounds that have high affinity for the
bacterial ribosomal A site but are only poorly recognized and acetylated by aminoglycoside-modifying
enzymes. An example of a successful approach used
variably cross-linked dimers of the bicyclic aminoglycoside neamine.201 This series of compounds were
shown to bind to an A site rRNA oligoribonucleotide
with high affinity (40 nM for the diaminobutane
cross-linked compound, Figure 16) and displayed
antibacterial activity (MIC ) 6.25 M). This compound was a poor substrate for both acetyltransferases AAC(6)-Ii and AAC(6)-APH(2), and in fact,
it was a powerful inhibitor of the kinase activity
(Ki ) 0.7 M) of the bifunctional enzyme. Another
approach that found some success was based on the
highly anionic nature of the aminoglycoside binding
site in the structure of the AAC(6)-Ii. Reasoning that
cationic peptides that by themselves exhibited antibacterial properties might bind to this enzyme, a
series of relatively short (12-24 residues) and highly
cationic (charge +4 to +9) peptides were synthesized
and tested as inhibitors. These peptides did indeed
inhibit the AAC(6)-Ii and both the APH (3)-IIa and
APH(2)-Ia with micromolar affinity.189
Mobashery and colleagues have synthesized a
number of aminoglycoside analogues with the potential to be poor substrates for aminoglycoside phosphotransferases, to inactivate the modifying enzyme,

Molecular Insights into Aminoglycoside Action and Resistance

Chemical Reviews, 2005, Vol. 105, No. 2 495

modified, but yet exhibit good antibacterial properties, either alone or in combination with extant
aminoglycosides, is a good one. Another example
illustrating this concept are the neamine derivatives,
substituted in positions 1 and 6, that exhibit a higher
antibacterial activity against both neamine-sensitive
and neamine-resistant strains than the parental
compounds.205

5. Acknowledgment
Figure 17. Proposed reactions of 2-nitrokanamycin with
the APH(3) phosphotransferase. (Reprinted with permission from ref 202. Copyright 1995 American Chemical
Society).

The authors would like to thank Dr. Matt W.


Vetting for assistance in preparing Figures 7, 9, 11,
12, 13, and 14. This work has been supported by the
Unites States National Institutes of Health.

6. References

Figure 18. Proposed reaction of 3-ketokanamycin with the


APH(3) phosphotransferase. (Reprinted with permission
from ref 203. Copyright 1999 American Chemical Society.)

or to regenerate themselves after enzymatic modification. The first of these was a 2-nitro-substituted
aminoglycoside. Upon phosphorylation by APH(3),
the adjacent nitro group reduces the pK of the 2proton sufficiently that elimination of the 3-phosphate occurs. The vinylogous product is a Michael
acceptor that can be captured by an enzymic nucleophile, resulting in a novel mechanism-based inhibition of the kinase (Figure 17).202 A second strategy
involved the synthesis of a 3-keto aminoglycoside
derivative.203 In solution, the ketone is hydrated and
mimics the 3-equatorial hydroxyl group that is the
locus of phosphorylation. In fact, the compound is
phosphorylated by APH(3) but decomposes with the
elimination of phosphate to regenerate the 3-keto
group (Figure 18). The compound itself exhibits weak
antibiotic activity, as determined by its MIC value
(250 M), and does modestly sensitize bacteria harboring APH(3) to other aminoglycosides (4-8-fold
decreases in MIC values). Finally, the synthesis and
evaluation of 4,4-difluorokanamycin A and neamine
derivatives have recently been reported.204 Because
of the presence of the highly electron withdrawing
fluorine substituent adjacent to the 3-hydroxy group,
the nucleophilicity of the hydroxy group is significantly diminished. The turnover numbers for the 3phosphorylation of the difluoroaminoglycosides are
decreased by almost 3 orders of magnitude. While the
MIC values of these difluoroaminoglycosides are not
impressive, the compounds are as effective against
strains expressing APH(3) as against wild-type
strains. This is a clear demonstration that the
concept of synthesizing aminoglycoside derivatives
that are slowly modified, or incapable of being

(1) Schatz, A.; Bugie, E.; Waksman, S. A. Proc. Soc. Exp. Biol. Med.
1944, 55, 66-69.
(2) Kotra, L. P.; Haddad, J.; Mobashery, S. Antimicrob. Agents
Chemother. 2000, 44, 3249-56.
(3) Vicens, Q.; Westhof, E. Biopolymers 2003, 70, 42-57.
(4) Vakulenko, S. B.; Mobashery, S. Clin. Microbiol. Rev. 2003, 16,
430-50.
(5) Wimberly, B. T.; Brodersen, D. E.; Clemons, W. M., Jr.; MorganWarren, R. J.; Carter, A. P.; Vonrhein, C.; Hartsch, T.; Ramakrishnan, V. Nature 2000, 407, 327-39.
(6) Schluenzen, F.; Tocilj, A.; Zarivach, R.; Harms, J.; Gluehmann,
M.; Janell, D.; Bashan, A.; Bartels, H.; Agmon, I.; Franceschi,
F.; Yonath, A. Cell 2000, 102, 615-23.
(7) Fourmy, D.; Recht, M. I.; Blanchard, S. C.; Puglisi, J. D. Science
1996, 274, 1367-71.
(8) Fourmy, D.; Yoshizawa, S.; Puglisi, J. D. J. Mol. Biol. 1998, 277,
333-45.
(9) Fourmy, D.; Recht, M. I.; Puglisi, J. D. J. Mol. Biol. 1998, 277,
347-62.
(10) Carter, A. P.; Clemons, W. M.; Brodersen, D. E.; Morgan-Warren,
R. J.; Wimberly, B. T.; Ramakrishnan, V. Nature 2000, 407,
340-8.
(11) Yoshizawa, S.; Fourmy, D.; Puglisi, J. D. EMBO J. 1998, 17,
6437-48.
(12) Vicens, Q.; Westhof, E. Chem. Biol. 2002, 9, 747-55.
(13) Karimi, R.; Ehrenberg, M. Eur. J. Biochem. 1994, 226, 355-60.
(14) Pape, T.; Wintermeyer, W.; Rodnina, M. V. Nat. Struct. Biol.
2000, 7, 104-7.
(15) Puglisi, J. D.; Blanchard, S. C.; Green, R. Nat. Struct. Biol. 2000,
7, 855-61.
(16) Brodersen, D. E.; Clemons, W. M., Jr.; Carter, A. P.; MorganWarren, R. J.; Wimberly, B. T.; Ramakrishnan, V. Cell 2000,
103, 1143-54.
(17) Lynch, S. R.; Puglisi, J. D. J. Mol. Biol. 2001, 306, 1037-58.
(18) Busse, H. J.; Wostmann, C.; Bakker, E. P. J. Gen. Microbiol.
1992, 138 (Pt 3), 551-61.
(19) Yoneyama, H.; Sato, K.; Nakae, T. Chemotherapy 1991, 37, 23945.
(20) Bakker, E. P. J. Gen. Microbiol. 1992, 138 (Pt 3), 563-9.
(21) Davis, B. D. Microbiol. Rev. 1987, 51, 341-50.
(22) Mehta, R.; Champney, W. S. Antimicrob. Agents Chemother.
2002, 46, 1546-9.
(23) Edson, R. S.; Terrell, C. L. Mayo Clin. Proc. 1999, 74, 519-28.
(24) Gonzalez, L. S., 3rd.; Spencer, J. P. Am. Fam. Physician 1998,
58, 1811-20.
(25) Freeman, C. D.; Nicolau, D. P.; Belliveau, P. P.; Nightingale, C.
H. J. Antimicrob. Chemother. 1997, 39, 677-86.
(26) Craig, W. A.; Vogelman, B. Ann. Intern. Med. 1987, 106, 9002.
(27) Zhanel, G. G.; Hoban, D. J.; Harding, G. K. DICP, Ann.
Pharmocother. 1991, 25, 153-63.
(28) Gilbert, D. N. Antimicrob. Agents Chemother. 1991, 35, 399405.
(29) Karlowsky, J. A.; Zhanel, G. G.; Davidson, R. J.; Hoban, D. J. J.
Antimicrob. Chemother. 1994, 33, 937-47.
(30) Preston, S. L.; Briceland, L. L. Pharmacotherapy 1995, 15, 297316.
(31) Santucci, R. A.; Krieger, J. N. J. Urol. 2000, 163, 1076-84.
(32) Ratjen, F.; Doring, G.; Nikolaizik, W. H. Lancet 2001, 358, 9834.
(33) Moss, R. B. Chest 2002, 121, 55-63.
(34) Couch, L. A. Chest 2001, 120, 114S-117S.
(35) Schiffelers, R.; Storm, G.; Bakker-Woudenberg, I. J. Antimicrob.
Chemother. 2001, 48, 333-44.

496 Chemical Reviews, 2005, Vol. 105, No. 2


(36) Ashtekar, D.; Duzgunes, N.; Gangadharam, P. R. J. Antimicrob.
Chemother. 1991, 28, 615-7.
(37) Gangadharam, P. R.; Ashtekar, D. R.; Flasher, D. L.; Duzgunes,
N. Antimicrob. Agents Chemother. 1995, 39, 725-30.
(38) Majumdar, S.; Flasher, D.; Friend, D. S.; Nassos, P.; Yajko, D.;
Hadley, W. K.; Duzgunes, N. Antimicrob. Agents Chemother.
1992, 36, 2808-15.
(39) Gillespie, S. H. Antimicrob. Agents Chemother. 2002, 46, 26774.
(40) Gerding, D. N. Infect. Control Hosp. Epidemiol. 2000, 21, S127.
(41) Graham, J. C.; Gould, F. K. J. Antimicrob. Chemother. 2002,
49, 437-44.
(42) Shanson, D. C. J. Antimicrob. Chemother. 1998, 42, 292-6.
(43) Nakamura, A.; Hosoda, M.; Kato, T.; Yamada, Y.; Itoh, M.;
Kanazawa, K.; Nouda, H. J. Antimicrob. Chemother. 2000, 46,
901-4.
(44) Inoue, M.; Nonoyama, M.; Okamoto, R.; Ida, T. Drugs Exp. Clin.
Res. 1994, 20, 233-9.
(45) Hotta, K.; Sunada, A.; Ikeda, Y.; Kondo, S. J. Antibiot. (Tokyo)
2000, 53, 1168-74.
(46) Cordeiro, J. C.; Reis, A. O.; Miranda, E. A.; Sader, H. S. Braz.
J. Infect. Dis. 2001, 5, 130-5.
(47) Kono, K.; Takeda, S.; Tatara, I.; Arakawa, K. Jpn. J. Antibiot.
1994, 47, 710-9.
(48) Suzuki, K. Pediatr. Int. 2003, 45, 301-6.
(49) Hamilton-Miller, J. M.; Shah, S. J. Antimicrob. Chemother. 1995,
35, 865-8.
(50) Ida, T.; Okamoto, R.; Shimauchi, C.; Okubo, T.; Kuga, A.; Inoue,
M. J. Clin. Microbiol. 2001, 39, 3115-21.
(51) Taber, H. W.; Mueller, J. P.; Miller, P. F.; Arrow, A. S. Microbiol.
Rev. 1987, 51, 439-57.
(52) Hancock, R. E. Annu. Rev. Microbiol. 1984, 38, 237-64.
(53) Bryan, L. E.; Kowand, S. K.; Van Den Elzen, H. M. Antimicrob.
Agents Chemother. 1979, 15, 7-13.
(54) Miller, M. H.; Edberg, S. C.; Mandel, L. J.; Behar, C. F.;
Steigbigel, N. H. Antimicrob. Agents Chemother. 1980, 18, 7229.
(55) Balwit, J. M.; van Langevelde, P.; Vann, J. M.; Proctor, R. A. J.
Infect. Dis. 1994, 170, 1033-7.
(56) Bryan, L. E.; OHara, K.; Wong, S. Antimicrob. Agents Chemother. 1984, 26, 250-5.
(57) Kadurugamuwa, J. L.; Lam, J. S.; Beveridge, T. J. Antimicrob.
Agents Chemother. 1993, 37, 715-21.
(58) Hatch, R. A.; Schiller, N. L. Antimicrob. Agents Chemother. 1998,
42, 974-7.
(59) Putman, M.; van Veen, H. W.; Konings, W. N. Microbiol. Mol.
Biol. Rev. 2000, 64, 672-93.
(60) Murakami, S.; Nakashima, R.; Yamashita, E.; Yamaguchi, A.
Nature 2002, 419, 587-93.
(61) Yu, E. W.; Aires, J. R.; Nikaido, H. J. Bacteriol. 2003, 185, 565764.
(62) Yu, E. W.; McDermott, G.; Zgurskaya, H. I.; Nikaido, H.;
Koshland, D. E., Jr. Science 2003, 300, 976-80.
(63) Koronakis, V.; Sharff, A.; Koronakis, E.; Luisi, B.; Hughes, C.
Nature 2000, 405, 914-9.
(64) Eda, S.; Maseda, H.; Nakae, T. J. Biol. Chem. 2003, 278, 20858.
(65) Elkins, C. A.; Nikaido, H. J. Bacteriol. 2002, 184, 6490-8.
(66) Avila-Sakar, A. J.; Misaghi, S.; Wilson-Kubalek, E. M.; Downing,
K. H.; Zgurskaya, H.; Nikaido, H.; Nogales, E. J. Struct. Biol.
2001, 136, 81-8.
(67) Westbrock-Wadman, S.; Sherman, D. R.; Hickey, M. J.; Coulter,
S. N.; Zhu, Y. Q.; Warrener, P.; Nguyen, L. Y.; Shawar, R. M.;
Folger, K. R.; Stover, C. K. Antimicrob. Agents Chemother. 1999,
43, 2975-83.
(68) Aires, J. R.; Kohler, T.; Nikaido, H.; Plesiat, P. Antimicrob.
Agents Chemother. 1999, 43, 2624-8.
(69) Moore, R. A.; DeShazer, D.; Reckseidler, S.; Weissman, A.;
Woods, D. E. Antimicrob. Agents Chemother. 1999, 43, 465-70.
(70) Rosenberg, E. Y.; Ma, D.; Nikaido, H. J. Bacteriol. 2000, 182,
1754-6.
(71) Sobel, M. L.; McKay, G. A.; Poole, K. Antimicrob. Agents
Chemother. 2003, 47, 3202-7.
(72) Hocquet, D.; Vogne, C.; El Garch, F.; Vejux, A.; Gotoh, N.; Lee,
A.; Lomovskaya, O.; Plesiat, P. Antimicrob. Agents Chemother.
2003, 47, 1371-5.
(73) Magnet, S.; Courvalin, P.; Lambert, T. Antimicrob. Agents
Chemother. 2001, 45, 3375-80.
(74) Grkovic, S.; Brown, M. H.; Skurray, R. A. Microbiol. Mol. Biol.
Rev. 2002, 66, 671-701, table of contents.
(75) Schumacher, M. A.; Brennan, R. G. Mol. Microbiol. 2002, 45,
885-93.
(76) Jo, J. T.; Brinkman, F. S.; Hancock, R. E. Antimicrob. Agents
Chemother. 2003, 47, 1101-11.
(77) Edgar, R.; Bibi, E. J. Bacteriol. 1997, 179, 2274-80.
(78) Silva, P. E.; Bigi, F.; de la Paz Santangelo, M.; Romano, M. I.;
Martin, C.; Cataldi, A.; Ainsa, J. A. Antimicrob. Agents Chemother. 2001, 45, 800-4.

Magnet and Blanchard


(79) Ainsa, J. A.; Blokpoel, M. C.; Otal, I.; Young, D. B.; De Smet, K.
A.; Martin, C. J. Bacteriol. 1998, 180, 5836-43.
(80) De Rossi, E.; Arrigo, P.; Bellinzoni, M.; Silva, P. A.; Martin, C.;
Ainsa, J. A.; Guglierame, P.; Riccardi, G. Mol. Med. 2002, 8,
714-24.
(81) Van Buul, C. P.; Damm, J. B.; Van Knippenberg, P. H. Mol. Gen.
Genet. 1983, 189, 475-8.
(82) van Buul, C. P.; van Knippenberg, P. H. Gene 1985, 38, 65-72.
(83) Vila-Sanjurjo, A.; Squires, C. L.; Dahlberg, A. E. J. Mol. Biol.
1999, 293, 1-8.
(84) Walsh, C. Antibiotics: actions, origins, resistance; ASM Press:
Washington, DC, 2003.
(85) Lacalle, R. A.; Tercero, J. A.; Vara, J.; Jimenez, A. J. Bacteriol.
1993, 175, 7474-8.
(86) Tercero, J. A.; Espinosa, J. C.; Lacalle, R. A.; Jimenez, A. J. Biol.
Chem. 1996, 271, 1579-90.
(87) Cundliffe, E. Annu. Rev. Microbiol. 1989, 43, 207-33.
(88) Skeggs, P. A.; Thompson, J.; Cundliffe, E. Mol. Gen. Genet. 1985,
200, 415-21.
(89) Thompson, J.; Skeggs, P. A.; Cundliffe, E. Mol. Gen. Genet. 1985,
201, 168-73.
(90) Kelemen, G. H.; Cundliffe, E.; Financsek, I. Gene 1991, 98, 5360.
(91) Holmes, D. J.; Cundliffe, E. Mol. Gen. Genet. 1991, 229, 22937.
(92) Demydchuk, J.; Oliynyk, Z.; Fedorenko, V. J. Basic Microbiol.
1998, 38, 231-9.
(93) Kojic, M.; Topisirovic, L.; Vasiljevic, B. J. Bacteriol. 1992, 174,
7868-72.
(94) Holmes, D. J.; Drocourt, D.; Tiraby, G.; Cundliffe, E. Gene 1991,
102, 19-26.
(95) Ohta, T.; Hasegawa, M. Gene 1993, 127, 63-9.
(96) Bujnicki, J. M.; Rychlewski, L. Acta Microbiol. Pol. 2001, 50,
7-17.
(97) Yokoyama, K.; Doi, Y.; Yamane, K.; Kurokawa, H.; Shibata, N.;
Shibayama, K.; Yagi, T.; Kato, H.; Arakawa, Y. Lancet 2003,
362, 1888-93.
(98) Doi, Y.; Yokoyama, K.; Yamane, K.; Wachino, J.; Shibata, N.;
Yagi, T.; Shibayama, K.; Kato, H.; Arakawa, Y. Antimicrob.
Agents Chemother. 2004, 48, 491-6.
(99) Galimand, M.; Courvalin, P.; Lambert, T. Antimicrob. Agents
Chemother. 2003, 47, 2565-71.
(100) Finken, M.; Kirschner, P.; Meier, A.; Wrede, A.; Bottger, E. C.
Mol. Microbiol. 1993, 9, 1239-46.
(101) Honore, N.; Cole, S. T. Antimicrob. Agents Chemother. 1994, 38,
238-42.
(102) Honore, N.; Marchal, G.; Cole, S. T. Antimicrob. Agents Chemother. 1995, 39, 769-70.
(103) Bottger, E. C. Trends Microbiol. 1994, 2, 416-21.
(104) Musser, J. M. Clin. Microbiol. Rev. 1995, 8, 496-514.
(105) Meier, A.; Kirschner, P.; Bange, F. C.; Vogel, U.; Bottger, E. C.
Antimicrob. Agents Chemother. 1994, 38, 228-33.
(106) Powers, T.; Noller, H. F. EMBO J. 1991, 10, 2203-14.
(107) Suzuki, Y.; Katsukawa, C.; Tamaru, A.; Abe, C.; Makino, M.;
Mizuguchi, Y.; Taniguchi, H. J. Clin. Microbiol. 1998, 36, 12205.
(108) Galimand, M.; Gerbaud, G.; Courvalin, P. Antimicrob. Agents
Chemother. 2000, 44, 1365-6.
(109) Sigmund, C. D.; Ettayebi, M.; Borden, A.; Morgan, E. A. Methods
Enzymol. 1988, 164, 673-90.
(110) Sparling, P. F.; Davis, B. D. Antimicrob. Agents Chemother. 1972,
1, 252-8.
(111) Asai, T.; Zaporojets, D.; Squires, C.; Squires, C. L. Proc. Natl.
Acad. Sci. U.S.A. 1999, 96, 1971-6.
(112) Pfister, P.; Hobbie, S.; Vicens, Q.; Bottger, E. C.; Westhof, E.
ChemBioChem 2003, 4, 1078-88.
(113) Sander, P.; Prammananan, T.; Bottger, E. C. Mol. Microbiol.
1996, 22, 841-8.
(114) Recht, M. I.; Douthwaite, S.; Puglisi, J. D. EMBO J. 1999, 18,
3133-8.
(115) Springer, B.; Kidan, Y. G.; Prammananan, T.; Ellrott, K.;
Bottger, E. C.; Sander, P. Antimicrob. Agents Chemother. 2001,
45, 2877-84.
(116) Brink, M. F.; Brink, G.; Verbeet, M. P.; de Boer, H. A. Nucleic
Acids Res. 1994, 22, 325-31.
(117) Johanson, U.; Hughes, D. Nucleic Acids Res. 1995, 23, 464-6.
(118) Pfister, P.; Risch, M.; Brodersen, D. E.; Bottger, E. C. Antimicrob.
Agents Chemother. 2003, 47, 1496-502.
(119) De Stasio, E. A.; Dahlberg, A. E. J. Mol. Biol. 1990, 212, 12733.
(120) Recht, M. I.; Douthwaite, S.; Dahlquist, K. D.; Puglisi, J. D. J.
Mol. Biol. 1999, 286, 33-43.
(121) Recht, M. I.; Puglisi, J. D. Antimicrob. Agents Chemother. 2001,
45, 2414-9.
(122) Prammananan, T.; Sander, P.; Brown, B. A.; Frischkorn, K.;
Onyi, G. O.; Zhang, Y.; Bottger, E. C.; Wallace, R. J., Jr. J. Infect.
Dis. 1998, 177, 1573-81.
(123) Dobner, P.; Bretzel, G.; Rusch-Gerdes, S.; Feldmann, K.; Rifai,
M.; Loscher, T.; Rinder, H. Mol. Cell Probes 1997, 11, 123-6.

Molecular Insights into Aminoglycoside Action and Resistance


(124) Stuy, J. H.; Walter, R. B. J. Bacteriol. 1992, 174, 5604-8.
(125) Bonny, C.; Montandon, P. E.; Marc-Martin, S.; Stutz, E. Biochim.
Biophys. Acta 1991, 1089, 213-9.
(126) Toivonen, J. M.; Boocock, M. R.; Jacobs, H. T. Mol. Microbiol.
1999, 31, 1735-46.
(127) Allen, P. N.; Noller, H. F. J. Mol. Biol. 1989, 208, 457-68.
(128) Gregory, S. T.; Cate, J. H.; Dahlberg, A. E. J. Mol. Biol. 2001,
309, 333-8.
(129) Bjorkman, J.; Samuelsson, P.; Andersson, D. I.; Hughes, D. Mol.
Microbiol. 1999, 31, 53-8.
(130) Ramakrishnan, V.; White, S. W. Nature 1992, 358, 768-71.
(131) Melancon, P.; Tapprich, W. E.; Brakier-Gingras, L. J. Bacteriol.
1992, 174, 7896-901.
(132) Zuurmond, A. M.; Zeef, L. A.; Kraal, B. Microbiology 1998, 144
(Pt 12), 3309-16.
(133) Davies, C.; Bussiere, D. E.; Golden, B. L.; Porter, S. J.; Ramakrishnan, V.; White, S. W. J. Mol. Biol. 1998, 279, 873-88.
(134) Shaw, K. J.; Rather, P. N.; Hare, R. S.; Miller, G. H. Microbiol.
Rev. 1993, 57, 138-63.
(135) Lombardini, J. B.; Cheng-Chu, M. Int. J. Biochem. 1980, 12,
427-31.
(136) Van Pelt, J. E.; Northrop, D. B. Arch. Biochem. Biophys. 1984,
230, 250-63.
(137) Gates, C. A.; Northrop, D. B. Biochemistry 1988, 27, 3834-42.
(138) Gates, C. A.; Northrop, D. B. Biochemistry 1988, 27, 3826-33.
(139) Gates, C. A.; Northrop, D. B. Biochemistry 1988, 27, 3820-5.
(140) Williams, J. W.; Northrop, D. B. J. Biol. Chem. 1978, 253, 590814.
(141) Le Goffic, F.; Martel, A.; Capmau, M. L.; Baca, B.; Goebel, P.;
Chardon, H.; Soussy, C. J.; Duval, J.; Bouanchaud, D. H.
Antimicrob. Agents Chemother. 1976, 10, 258-64.
(142) Le Goffic, F.; Baca, B.; Soussy, C. J.; Dublanchet, A.; Duval, J.
Ann. Microbiol. (Paris) 1976, 127, 391-9.
(143) Davies, J.; Smith, D. I. Annu. Rev. Microbiol. 1978, 32, 469518.
(144) Van Pelt, J. E.; Iyengar, R.; Frey, P. A. J. Biol. Chem. 1986,
261, 15995-9.
(145) Sakon, J.; Liao, H. H.; Kanikula, A. M.; Benning, M. M.;
Rayment, I.; Holden, H. M. Biochemistry 1993, 32, 11977-84.
(146) Pedersen, L. C.; Benning, M. M.; Holden, H. M. Biochemistry
1995, 34, 13305-11.
(147) Gerratana, B.; Cleland, W. W.; Reinhardt, L. A. Biochemistry
2001, 40, 2964-71.
(148) Schwotzer, U.; Kayser, F. H.; Schwotzer, W. FEMS Microbiol.
Lett. 1978, 3, 29-33.
(149) Gerratana, B.; Frey, P. A.; Cleland, W. W. Biochemistry 2001,
40, 2972-7.
(150) McKay, G. A.; Thompson, P. R.; Wright, G. D. Biochemistry 1994,
33, 6936-44.
(151) Trieu-Cuot, P.; Courvalin, P. Gene 1983, 23, 331-41.
(152) Cox, J. R.; McKay, G. A.; Wright, G. D.; Serpersu, E. H. J. Am.
Chem. Soc. 1996, 118, 1295-1301.
(153) Thompson, P. R.; Hughes, D. W.; Wright, G. D. Biochemistry
1996, 35, 8686-95.
(154) McKay, G. A.; Wright, G. D. J. Biol. Chem. 1995, 270, 2468692.
(155) McKay, G. A.; Wright, G. D. Biochemistry 1996, 35, 8680-5.
(156) Hon, W. C.; McKay, G. A.; Thompson, P. R.; Sweet, R. M.; Yang,
D. S.; Wright, G. D.; Berghuis, A. M. Cell 1997, 89, 887-95.
(157) Fong, D. H.; Berghuis, A. M. EMBO J. 2002, 21, 2323-31.
(158) Thompson, P. R.; Schwartzenhauer, J.; Hughes, D. W.; Berghuis,
A. M.; Wright, G. D. J. Biol. Chem. 1999, 274, 30697-706.
(159) Daigle, D. M.; McKay, G. A.; Thompson, P. R.; Wright, G. D.
Chem. Biol. 1999, 6, 11-8.
(160) Martel, A.; Masson, M.; Moreau, N.; Le Goffic, F. Eur. J.
Biochem. 1983, 133, 515-21.
(161) McKay, G. A.; Robinson, R. A.; Lane, W. S.; Wright, G. D.
Biochemistry 1994, 33, 14115-20.
(162) Boehr, D. D.; Lane, W. S.; Wright, G. D. Chem. Biol. 2001, 8,
791-800.
(163) Lovering, A. M.; White, L. O.; Reeves, D. S. J. Antimicrob.
Chemother. 1987, 20, 803-13.
(164) Sunada, A.; Nakajima, M.; Ikeda, Y.; Kondo, S.; Hotta, K. J.
Antibiot. (Tokyo) 1999, 52, 809-14.
(165) Rather, P. N.; Orosz, E.; Shaw, K. J.; Hare, R.; Miller, G. J.
Bacteriol. 1993, 175, 6492-8.
(166) Macinga, D. R.; Rather, P. N. Front. Biosci. 1999, 4, D132-40.
(167) Payie, K. G.; Rather, P. N.; Clarke, A. J. J. Bacteriol. 1995, 177,
4303-10.

Chemical Reviews, 2005, Vol. 105, No. 2 497


(168) Franklin, K.; Clarke, A. J. Antimicrob. Agents Chemother. 2001,
45, 2238-44.
(169) Udou, T.; Mizuguchi, Y.; Wallace, R. J., Jr. Am. Rev. Respir. Dis.
1987, 136, 338-43.
(170) Ainsa, J. A.; Martin, C.; Gicquel, B.; Gomez-Lus, R. Antimicrob.
Agents Chemother. 1996, 40, 2350-5.
(171) Hegde, S. S.; Javid-Majd, F.; Blanchard, J. S. J. Biol. Chem.
2001, 276, 45876-81.
(172) Vetting, M. W.; Hegde, S. S.; Javid-Majd, F.; Blanchard, J. S.;
Roderick, S. L. Nat. Struct. Biol. 2002, 9, 653-8.
(173) Dyda, F.; Klein, D. C.; Hickman, A. B. Annu. Rev. Biophys.
Biomol. Struct. 2000, 29, 81-103.
(174) Williams, J. W.; Northrop, D. B. Biochemistry 1976, 15, 12531.
(175) Williams, J. W.; Northrop, D. B. J. Biol. Chem. 1978, 253, 59027.
(176) Javier Teran, F.; Alvarez, M.; Suarez, J. E.; Mendoza, M. C. J.
Antimicrob. Chemother. 1991, 28, 333.
(177) Wolf, E.; Vassilev, A.; Makino, Y.; Sali, A.; Nakatani, Y.; Burley,
S. K. Cell 1998, 94, 439-49.
(178) DiGiammarino, E. L.; Draker, K. A.; Wright, G. D.; Serpersu,
E. H. Biochemistry 1998, 37, 3638-44.
(179) Wybenga-Groot, L. E.; Draker, K.; Wright, G. D.; Berghuis, A.
M. Structure Fold Des. 1999, 7, 497-507.
(180) Burk, D. L.; Ghuman, N.; Wybenga-Groot, L. E.; Berghuis, A.
M. Protein Sci. 2003, 12, 426-37.
(181) Wright, G. D.; Ladak, P. Antimicrob. Agents Chemother. 1997,
41, 956-60.
(182) Draker, K. A.; Northrop, D. B.; Wright, G. D. Biochemistry 2003,
42, 6565-74.
(183) Draker, K. A.; Wright, G. D. Biochemistry 2004, 43, 446-54.
(184) Magnet, S.; Courvalin, P.; Lambert, T. J. Bacteriol. 1999, 181,
6650-5.
(185) Magnet, S.; Lambert, T.; Courvalin, P.; Blanchard, J. S. Biochemistry 2001, 40, 3700-9.
(186) Hegde, S. S.; Dam, T. K.; Brewer, C. F.; Blanchard, J. S.
Biochemistry 2002, 41, 7519-27.
(187) Vetting, M. W.; Magnet, S.; Nieves, E.; Roderick, S. L.; Blanchard, J. S. Chem. Biol., in press.
(188) Daigle, D. M.; Hughes, D. W.; Wright, G. D. Chem. Biol. 1999,
6, 99-110.
(189) Boehr, D. D.; Jenkins, S. I.; Wright, G. D. J. Biol. Chem. 2003,
278, 12873-80.
(190) Fujimura, S.; Tokue, Y.; Takahashi, H.; Kobayashi, T.; Gomi,
K.; Abe, T.; Nukiwa, T.; Watanabe, A. FEMS Microbiol. Lett.
2000, 190, 299-303.
(191) Deverstein-van Hall, M. A.; Blok, H. E.; Donders, A. R.; Paauw,
A.; Fluit, A. C.; Verhoef, J. J. Infect. Dis. 2003, 187, 251-9.
(192) White, P. A.; McIver, C. J.; Rawlinson, W. D. Antimicrob. Agents
Chemother. 2001, 45, 2658-61.
(193) Culebras, E.; Martinez, J. L. Front. Biosci. 1999, 4, D1-8.
(194) Derbise, A.; Dyke, K. G.; el Solh, N. Plasmid 1996, 35, 174-88.
(195) Simjee, S.; Gill, M. J. J. Hosp. Infect. 1997, 36, 249-59.
(196) Miller, G. H.; Sabatelli, F. J.; Hare, R. S.; Glupczynski, Y.;
Mackey, P.; Shlaes, D.; Shimizu, K.; Shaw, K. J. Clin. Infect.
Dis. 1997, 24 Suppl 1, S46-62.
(197) Schmitz, F. J.; Verhoef, J.; Fluit, A. C. Eur. J. Clin. Microbiol.
Infect. Dis. 1999, 18, 414-21.
(198) Schmitz, F. J.; Fluit, A. C.; Gondolf, M.; Beyrau, R.; Lindenlauf,
E.; Verhoef, J.; Heinz, H. P.; Jones, M. E. J. Antimicrob.
Chemother. 1999, 43, 253-9.
(199) Williams, J. W.; Northrop, D. B. J. Antibiot. (Tokyo) 1979, 32,
1147-54.
(200) Allen, N. E.; Alborn, W. E., Jr.; Hobbs, J. N., Jr.; Kirst, H. A.
Antimicrob. Agents Chemother. 1982, 22, 824-31.
(201) Sucheck, S. J.; Wong, A. L.; Koeller, K. M.; Boehr, D. D.; Draker,
K.; Sears, P.; Wright, G. D. J. Am. Chem. Soc. 2000, 122, 52305231.
(202) Roestamadji, J.; Grapsas, I. L.; Mobashery, S. J. Am. Chem. Soc.
1995, 117, 80-84.
(203) Haddad, J.; Vakulenko, S.; Mobashery, S. J. Am. Chem. Soc.
1999, 121, 11922-11923.
(204) Kim, C.; Haddad, J.; Vakulenko, S. B.; Meroueh, S. O.; Wu, Y.;
Yan, H.; Mobashery, S. Biochemistry 2004, 43, 2373-83.
(205) Haddad, J.; Kotra, L. P.; Llano-Sotelo, B.; Kim, C.; Azucena, E.
F.; Liu, M.; Vakulenko, S. B.; Chow, C. S.; Mobashery, S. J. Am.
Chem. Soc. 2002, 124, 3229-37.

CR0301088

Chem. Rev. 2005, 105, 499527

499

Translation and Protein Synthesis: Macrolides


Leonard Katz* and Gary W. Ashley
Kosan Biosciences, Incorporated, 3832 Bay Center Place, Hayward, California 94545
Received July 15, 2004

Contents
1. Introduction
2. Classes of Macrolides
2.1. Twelve-Membered Macrolides
2.2. Fourteen-Membered Macrolides
2.3. Sixteen-Membered Macrolides
2.3.1. Tylactone Group
2.3.2. Platenolide Group
2.3.3. Mycinamicin
2.3.4. ChalcomycinNeutramycin Group
3. Clinical Uses of Macrolides
3.1. First-Generation Macrolides
3.2. Second-Generation Macrolides
3.3. Third-Generation Macrolides: Ketolides
3.4. Side Effects of Macrolides and Ketolides
4. Mode of Action
4.1. Inhibition of Translation
4.2. MacrolideRibosome Structural Studies
4.3. Inhibition of Ribosome Assembly
5. Macrolide Resistance
5.1. MLSB Resistance
5.1.1. Inducible Resistance
5.1.2. Constitutive Resistance
5.1.3. Inducible vs Constitutive
5.2. Efflux
5.3. Mutations in Ribosomal RNA
5.4. Mutations in Ribosomal Proteins
5.5. Enzymatic Inactivation of Macrolides
5.5.1. Hydrolysis of the Macrolactone
5.5.2. Phosphorylation
5.5.3. Glucosylation
6. Biosynthesis of Macrolides
6.1. Biosynthesis of the Aglycone: Modular
Polyketide Synthases
6.1.1. Erythromycin
6.1.2. Lankamycin and Oleandomycin
6.1.3. Methymycin and Pikromycin
6.1.4. Tylosin
6.1.5. Platenolide
6.1.6. Chalcomycin
6.2. Biosynthesis of Deoxysugars
6.3. Post-Polyketide Modification
6.3.1. Erythromycin and Megalomicin
6.3.2. Methymycin and Pikromycin

499
500
500
500
501
501
501
501
501
501
501
504
506
507
507
507
507
509
509
510
510
511
511
512
512
513
513
513
513
513
514
514
514
516
516
517
517
517
518
519
519
520

* To whom correspondence should be addressed. Phone: 510-7315264. Fax: 510-732-8400. E-mail: katz@kosan.com.

6.3.3. Tylosin
6.3.4. Other Macrolides
6.4. Regulation of Macrolide Biosynthesis
7. New Macrolides and Ketolides
7.1. Chemistry
7.2. Genetic Engineering
8. Conclusions
9. Acknowledgments
10. References

520
521
522
522
522
522
523
524
524

1. Introduction
The term macrolide was originally proposed by
R. B. Woodward in 1957 as an abbreviation of
macrolactone glycoside antibiotics, a class of natural
products composed of macrocyclic lactones to which
were attached one or more deoxysugar residues.1,2
Macrolides are produced as secondary metabolites
largely from the actinomycete family of bacteria,
organisms that inhabit the soil. The first macrolide
discovered was pikromycin in 1950, followed shortly
thereafter by erythromycin, the first macrolide introduced for clinical use in human medicine.3,4 Macrolide antibiotics have been used to treat infections in
humans and animals for more than 50 years. Interest
in derivatization of erythromycin to improve its
properties started in the 1960s and has continued to
the present time. A recent chemical derivative of
erythromycin, telithromycin, was approved for clinical use in the United States in 2004.
Macrolides can be classified in a number of ways.
From a chemical viewpoint they are divided into
groups based on the number of atoms in the macrocyclic rings: 12, 14, 16, or larger, as outlined in
section 2. Each group is subdivided further on the
basis of the general structure of the lactone moiety
or sugar substitutions. From a clinical point of view
the compounds are described as first-, second-, or
third-generation macrolides, as discussed in section
3. The first-generation molecules are the natural
products that were introduced as drugs in the 1950s,
followed by the semisynthetic second-generation
compounds in the 1990s, and the semisynthetic thirdgeneration molecules in the early 2000s.
Macrolides act as antibiotics by binding to ribosomes and consequently blocking protein synthesis.
The high affinity to bacterial ribosomes, together
with the highly conserved structure of ribosomes
across virtually all of the bacterial families, gives
macrolides broad-spectrum activity. The mode of

10.1021/cr030107f CCC: $53.50 2005 American Chemical Society


Published on Web 01/11/2005

500 Chemical Reviews, 2005, Vol. 105, No. 2

Katz and Ashley

of molecules. At the genetic level, and corresponding


biochemical level, biosynthesis of the polyketide and
deoxysugar components of macrolides is now understood well enough to account not only for the structure of macrolides, but also for the structural diversity seen among this family of compounds. In section
6 we will describe our level of understanding of
biosynthesis and discuss briefly the changes to the
structure of erythromycin and other macrolides produced from manipulation of the genes responsible for
their syntheses.

2. Classes of Macrolides
Leonard Katz received his B.Sc. degree at McGill University in Microbiology
and his Ph.D. degree in Molecular Genetics at the University of California
at Santa Barbara. After a postdoctoral fellowship at the University of
California at La Jolla studying plasmid replication, he did a brief stint as
a faculty member of the Biology Department at New York University. He
entered industry in 1977 at Schering-Plough in New Jersey. In 1979 he
went to Abbott Laboratories in Illinois, where he stayed for 19 years. For
the past 6 years Dr. Katz has been at Kosan Biosciences in Hayward,
CA, where he currently holds the title Vice President of Biological Sciences.
His interest in macrolides began in 1979. At Abbott he lead the group
that isolated and sequenced the biosynthesis genes for the antibiotic
erythromycin in S. erythraea and produced the first rationally determined
genetically engineered erythromycin analogues.

This section is limited to macrolides that have been


isolated as natural products. Some are congeners of
the parent compound. In general, the congeners are
late-pathway intermediates to the final product.

2.1. Twelve-Membered Macrolides


Only two macrolide antibiotics have been identified
that contain 12-membered rings: methymycin [1]
and neomethymycin [2]. They differ in the position
of a single OH group: C-12 in methymycin vs C-14
in neomethymycin. Both contain the deoxyaminosugar D-desosamine.

2.2. Fourteen-Membered Macrolides

Gary W. Ashley obtained his S.B. degree in Chemistry from the


Massachusetts Institute of Technology. He obtained his Ph.D. degree in
Chemistry from the University of California, Berkeley, under the guidance
of Paul A. Bartlett. After a postdoctoral fellowship in biochemistry with
JoAnne Stubbe at the University of Wisconsin, Madison, he taught
chemistry at Northwestern University. In 1993 he returned to California,
where he established the chemistry program at Kosan Biosciences, Inc.
His interests include bioorganic and natural products chemistry as well
as patent law. He was licensed to practice before the U.S. Patent and
Trademark Office in 2002 and currently divides his time at Kosan between
scientific and legal duties.

action of macrolides will be discussed in section 4


within the framework of recent structural information on macrolide-ribosome interaction. Clinical
resistance to macrolides in bacterial pathogens and
self-resistance to macrolides in the macrolide-producing actinomycetes have been well characterized and
found to share many common mechanisms. Resistance will be discussed in section 5 as a basis for the
discovery of novel, more potent compounds.
The aglycone components of macrolides are complex polyketides, partially reduced acyl chains formed
from the condensation of thioester-containing precursors in a manner common to all members of this class

Five compound families have been identified in this


class: erythromycin A [3] and its B, C, and D
congeners [4-6], pikromycin [7] and its 12-deoxy
congener narbomycin [8], megalomicin A [9] and its
congeners, oleandomycin [10], and lankamycin [11].
Erythromycin A is commonly referred to simply as
erythromycin. Megalomicin and erythromycin share
a common aglycone, 6-deoxyerythronolide B (6-dEB).
The aglycone of oleandomycin, 8,8a-deoxyoleandolide,
differs by the absence of the methyl group that is
present at C-15 in 6-dEB. All 14-membered macrolides except lankamycin contain desosamine at C-5;
lankamycin contains the neutral sugar chalcose at
that position. The neutral sugar in erythromycin A
at C-3 is L-cladinose, the 3-O-methyl derivative of
L-mycarose present at C-3 in megalomicin and in
erythromycins D and C. Megalomicin also contains
a second aminosugar, megosamine, at the 6-OH
position. Megalomicin is less potent as an antibiotic
than erythromycin A but has antiparasitic activity
through its inhibition of vesicular transport between
the medial- and trans-Golgi, resulting in the undersialylation of cellular proteins.5 The C-3 sugars in
oleandomycin and lankamycin are L-oleandrose and
L-arcanose, respectively. Pikromycin and narbomycin
contain only desosamine. The oxygen atom present
at C-3 is in the form of the ketone. Pikromycin,
discovered in 1950, therefore, is a natural ketolide,
a term first applied in the mid-1990s to describe the
new 3-descladinosyl-3-oxo derivatives of clarithromycin (6-O-methylerythromycin) that were found to
have increased antibacterial potency over erythromycin. Pikromycin, however, has weak antibacterial
activity. Erythromycin A has hydroxyl groups at both
C-6 and C-12 that are introduced by cytochrome
P450-type hydroxylases. Erythromycin congeners

Translation and Protein Synthesis

lacking the 6-OH group are weaker antibiotics.


Pikromycin lacks the 6-OH group. Oleandomycin and
lankamycin also lack the 6-OH group but are hydroxylated (lankamycin) or epoxidated (oleandomycin) at C-8. The presence of the 6- and 12-OH groups
in erythromycin A is a major source of instability
(Scheme 1). In protic solvents erythromycin A exists
as a mixture of the 9-keto form [3], the 9,12hemiketal form [3a], and the 6,9-hemiketal form [3b].
Under acidic conditions the hemiketal forms dehydrate to form enol ether derivatives [3c] and [3d],
respectively, which further degrade by reaction to
form the spiroketal derivative [3e]. Further degradation involves acid-catalyzed hydrolysis of the cladinose residue from [3e] to form erythralosamine. The
keto form [3] is the only species to have significant
antibacterial activity, whereas enol ether [3d] is a
potent agonist of the motilin receptor and is the main
cause of the gastrointestinal distress associated with
erythromycin therapy. The 12-membered macrolide
methymycin and the 14-membered macrolide pikromycin are made in the same host, Streptomyces
venezuelae.6 With the exception of the additional two
carbons in the aglycone component of pikromycin, the
two compounds are identical in structure.

2.3. Sixteen-Membered Macrolides


Sixteen-membered macrolides represent the largest
group of macrolides. We have subdivided this group
into four subgroups on the basis of the structure of
the polyketide backbone that forms the macrolactone
after release from the corresponding polyketide synthase and before any further modification, e.g.,
glycosylation, hydroxylation, etc., takes place. Some
of these aglycones are inferred from our current
understanding of the biochemistry of complex polyketide synthesis, which is described in detail below.

2.3.1. Tylactone Group


The most commercially important member of this
group is tylosin [12], produced from the bacterium
Streptomyces fradiae and used in veterinary medicine. Tylosin contains the disaccharide D-mycaminosyl-L-mycarose at C-5 and the monosaccharide D-mycinose at C-23. Tylosin carries the C-20 aldehyde
group: oxidation of the 6(S)-ethyl side chain of the
aglycone of tylosin takes place after macrolactone is
formed. Similarly, hydroxylation of the 14(R)-methyl
side chain (to enable subsequent glycosylation) is a
post-polyketide processing step. S. fradiae also produces tylosin D [13] (formerly named relomycin) in
which the aldehyde is reduced to the alcohol. Tylosin
D is much less potent than tylosin. Conversion of
tylosin to tylosin D is carried out by an adventitious
reductase that is not associated with the tylosin
biosynthesis gene cluster.7 Tylosin has undergone
extensive chemical derivatization, and the genes for
its biosynthesis have been characterized. Another
compound in this group includes rosamicin [14],
which carries only a single sugar, desosamine, at C-5,
and the 12,13-epoxide was in human clinical trials
but not further developed into a drug. Additional
members include cirramycin and the juvenimicins.

Chemical Reviews, 2005, Vol. 105, No. 2 501

2.3.2. Platenolide Group


This represents the largest group of 16-membered
macrolides. All members carry the 6(S)-CH3CH2CHO
side chain that is essential for antibiotic potency and
the mycaminosyl-mycarose disaccharide at C-5.
Platenolide does not have a C-14 methyl side chain
and thus offers no possibility of glycosylation on the
left side of the macrolactone. Fully elaborated compounds in this group may also contain various
acylations at C-3 and at the 4-hydroxyl of mycarose;
hence, families rather than single species of molecules are often produced from a single organism. We
have subdivided the platenolide-based group on the
basis of additional modification to the aglycone
moiety. The carbomycin B series contains no further
modifications. An example of a compound of this
subgroup is niddamycin [15]. The carbomycin A
series contains the 12,13-epoxide. The leucomycin
series is characterized by reduction of the C-9 keto
group and includes midecamycin A1 [16] and the
spiramycins (a series of three congeners) [17-19].
Midecamycin A1 and spiramycin were commercialized for human use. The spiramycins carry the
aminosugar forosamine at C-9 and various acyl
groups at C-3 or C4. Other members of the leucomycin series include the maridomycins, which carry
the 12,13-epoxide.

2.3.3. Mycinamicin
This group consists of one series of molecules, the
mycinamicins, produced by Micromonospora griseorubida. The aglycone contains a 2,3-trans double
bond, 4(R)-Me, 6(S)-Me, 14(R)-Me, 15(S)-Et. The
mycinamicins all contain the sugars desosamine at
C-5 and D-mycinose at C-21. The mycinamicins differ
from each other in the presence or absence of the 12,13-epoxide and 14(S)-OH group. An example is mycinamicin I [20]. Mycinamicins were not developed
for human use.

2.3.4. ChalcomycinNeutramycin Group


The aglycones of chalcomycin and mycinamicin
differ by the presence of a methyl group at C16 in
the latter compound. Neutramycin differs from chalcomycin [21] by the substitution of the C6-methyl
group in chalcomycin for a hydrogen atom. All
compounds in the group contain D-mycinose at C-20.
Chalcomycin has the neutral sugar D-chalcose at C-5,
the 12,13-epoxide, and an 8-hydroxyl group.

3. Clinical Uses of Macrolides


3.1. First-Generation Macrolides
The first-generation macrolides developed for clinical use were the natural products erythromycin A,
spiramycin, midecamycin A1, leucomycin, and carbomycin. These were isolated as fermentation products and required purification. Specifications of the
drugs allowed for small amounts of congeners; spiramycin was a mixture of 17-19. In general, the
compounds displayed excellent activity against Grampositive bacteria and were used initially to treat skin
caused by Staphylococcus aureus and Staphylococcus

502 Chemical Reviews, 2005, Vol. 105, No. 2

Katz and Ashley

Translation and Protein Synthesis

Chemical Reviews, 2005, Vol. 105, No. 2 503

Figure 1. Structures of macrolides and ketolides.

epidermidis and soft tissue infections caused by S.


aureus. In the 1960s these compounds began to be
used to treat upper and lower respiratory infections
caused by Streptococcus pneumoniae or Streptococcus
pyogenes and found lesser use against staphylococci.

The enterococci are much less susceptible to macrolides. These compounds have also been used for the
treatment of Legionnaires Disease (Legionella pneumophila), Lyme Disease (Borrelia burgdorferi), syphilis (Treponema pallidum), diphtheria (Corynebacte-

504 Chemical Reviews, 2005, Vol. 105, No. 2

Katz and Ashley

Scheme 1

rium diphtheriae), pertussis (Bordatella pertussis),


and respiratory infections caused by Moraxella cattarhalis and Mycoplasma pneumoniae. Chlamydia
pneumoniae is also susceptible to macrolides. Erythromycin has only modest activity against Gramnegative enterobacteria (e.g., Escherichia coli, Klebsiella) and no activity against Pseudomonas strains.
Sixteen-membered macrolides are somewhat more
potent against Gram negatives. Tylosin was developed to treat respiratory infections in animals, largely
caused by the Gram negatives Pasteurella multocida,
Mannheimia haemolytica, and various species of
Haemophilus. Haemophilus influenzae, a common
intracellular respiratory pathogen in children, is
treatable with macrolides.
The first-generation macrolides proved to be effective and fairly well tolerated. The most prominent
side effects of erythromycin were bitter taste and
stomach cramps, which was later found to be due to
the ability of the 8,9-anhydro-6,9-hemiketal form
([3e], Scheme 1) to mimic the effects of the hormone
motilin and stimulate gastrointestinal contractions.8
The most important drawbacks to the use of first-

generation macrolides were their short half-life and


poor oral bioavailability, prompting the requirement
for dosing three or four times a day. Despite these
weaknesses, these compounds were used successfully
for more than 25 years and were important first-line
agents for individuals with respiratory infections who
were hypersensitive to penicillin and its derivatives.
Because of their relatively low cost of production, they
are still used in Latin America, Africa, and some
parts of Asia.

3.2. Second-Generation Macrolides


The generally poor bioavailability, acid instability,
and unpredictable pharmacokinetics of the firstgeneration macrolides prompted the search for new
derivatives with improved properties. Five derivatives of erythromycin were developed and commercialized: clarithromycin (Biaxin; Abbott) [22],
dirithromycin (Dynebac; Sanofi) [23], roxithromycin
(Rulide; Aventis) [24], flurithromycin (Pierrel) [25],
and azithromycin (Zithromax; Pfizer) [26]. Miokamycin (Meiji) [27] and rokitamycin [28] were the only
16-membered second-generation compounds devel-

Translation and Protein Synthesis

Chemical Reviews, 2005, Vol. 105, No. 2 505

Scheme 2

oped for human use. Tilmicosin (Elanco) [29], a


semisynthetic derivative of tylosin, was developed for
veterinary use. Clarithromycin and azithromycin are
marketed worldwide; dirithromycin, flurithromycin,
and roxithromycin have much more limited distribution.
Clarithromycin is prepared from erythromycin A
in a short sequence of chemical transformations. The
propensity of the 6- and 12-OH groups to form
hemiketal derivatives with the 9-carbonyl together
with the higher reactivity of the 2- and 4-OH groups
on the glycosyl residues precludes efficient direct
alkylation of the 6-OH group. In a typical synthetic
sequence, erythromycin A is converted into the
9-oxime, which is then protected as an oxime ether.
The use of acetal groups to protect the oxime has
been found to be particularly convenient. Subsequent
blocking of the glycosyl hydroxyls, most simply as
trimethylsilyl ethers, provides protected derivatives
that can be efficiently methylated on the 6-OH under
basic conditions. The selectivity for alkylation of the
tertiary 6-OH group over the secondary 11-OH or
tertiary 12-OH groups is not entirely understood but
appears to be related to the unusually high acidity
of the 6-OH in erythromycin oxime derivatives.
Subsequent hydrolysis of the oxime acetal and trimethylsilyl ethers and deoximation provides clarithromycin. This six-step sequence produces clarithromycin in high yields yet significantly increases the cost
of the drug relative to erythromycin.
Azithromycin is prepared from erythromycin A
oxime by Beckmann rearrangement, for example, by
treatment with a sulfonyl chloride buffered with

aqueous sodium bicarbonate. This reaction is dependent upon trapping of the reactive Beckmann
intermediate by the 6-OH group rather than solvent
water to provide an isolable isoamide, which is
subsequently reduced to provide an intermediate
ring-expanded azalide. N-Methylation completes the
synthesis of azithromycin.
The second-generation erythromycin derivatives all
contain modifications at C6 or C9, preventing formation of the enol ether [3e] and thereby imparting
greater resistance to acid-catalyzed inactivation.
Clarithromycin is still degraded under acidic conditions to form derivatives analogous to [3d] and
descladinosyl derivatives, albeit at reduced rates
relative to erythromycin A.9-11 The five analogues
each had improved oral bioavailability and extended
half-life in plasma, enabling them to be taken orally
once (azithromycin) or twice (clarithromycin) a day.12
These compounds also exhibited enhanced tissue
penetration due to their increased lipophilicities over
the parent compound erythromycin A and hence were
effective for treatment of intracellular pathogens such
as H. influenzae.13,14 Although the search for secondgeneration macrolides was predicated on the desire
to discover compounds with expanded spectra and
improved activity, the compounds selected did not
exhibit improved activity against Gram-positive bacteria, and some, in fact, such as azithromycin, had
reduced potency.15,16 Nevertheless, they were selected
for development mainly because of their enhanced
pharmacokinetic profiles, in particular the ability to
accumulate to high levels in lung tissue. Clarithromycin is also used, generally in combination with

506 Chemical Reviews, 2005, Vol. 105, No. 2

other antibiotics, for the treatment of gastric ulcers


caused by Helicobacter pylori and for AIDS-related
respiratory infections caused by Mycobacterium avium complex.
The second-generation 16-membered macrolides
miokamycin and rokitamycin did not show enhanced
potencies in vitro over their parent compounds,
midecamycin A1 or leucomycin A5, but did show
improved in vivo potencies in experimental animals.
These compounds are not marketed for use in the
United States.

3.3. Third-Generation Macrolides: Ketolides


Whereas the search for second-generation macrolides in the 1970s and 1980s was driven by the need
for improved stability and pharmacokinetics, the
basis for the search for third-generation compounds
shifted to macrolide resistance that had arisen suddenly and rapidly in the 1980s and 1990s. A 2001
report indicated that 23% of the S. pneumoniae
strains in the United States were resistant to macrolides.17 Macrolide resistance is described in some
detail below. The only third-generation macrolide in
clinical use as of 2004 is telithromycin (Ketek;
Aventis) [30], a 14-membered ketolide that employs
clarithromycin as the starting material. The term
ketolide is used to indicate the presence of the
3-keto group in place of the L-cladinose present in
clarithromycin and other second-generation compounds and was adopted in the early 1990s to
describe new, semisynthetic series.18,19 Removal of
the cladinosyl group from erythromycin could be
accomplished by acid treatment (after protection of
the 9-keto group), but the resulting 3-OH derivative
was found to have lost much of its potency. Moreover,
oxidation of the 3-OH to the ketone was not practical
because of subsequent 3,6-cyclization. Pikromycin,
the first natural ketolide, exhibited weak potency.
Hence, generation of the potent semisynthetic ketolides awaited the creation of clarithromycin, which
occurred in the mid-1980s. Replacement of the Lcladinose moiety with the 3-keto group in clarithromycin rendered the resulting compound a noninducer
of MLSB resistance (described below), but it also
exhibited decreased potency, likely through loss of
binding interactions to the cladinose group and/or
increased flexibility of the macrolactone. Addition of
the fused 11,12-cyclic carbamate made the macrolactone more rigid, adding potency against some
strains, and addition of the aryl alkyl side chain to
the N-11a position compensated for loss of binding
interactions to cladinose and imparted 2-10-fold
enhanced in-vitro activity against macrolide-susceptible streptococci and staphylococci over clarithromycin. Telithromycin is at least as potent in vitro as
clarithromycin against H. influenzae and the atypical
respiratory pathogens M. cattharalis, L. pneumophila, M. pneumoniae, and C. pneumoniae.20 Telithromycin is not as potent as azithromycin against H.
influenzae but accumulates in lung tissue well enough
to be clinically useful against this organism.
Telithromycin is prepared from clarithromycin
using a sequence of eight chemical steps (Scheme 2).21
Acid hydrolysis of clarithromycin provides the 3-des-

Katz and Ashley

cladinosyl derivative, which is protected at the 2OH by acetylation with acetic anhydride in the
absence of added base. Under such conditions the 2OH is unusually reactive toward acylating (but not
alkylating or silylating) reagents due to the adjacent
dimethylamino functionality. Presumably the amine
reacts with the anhydride to form an acylammonium
salt, which transfers the acyl group to the 2-OH. This
is suggested by the observation that use of acid
halides rather than anhydrides results in formation
of N-acyl-N-demethyl derivatives rather than O-acyl
derivatives. Subsequent oxidation of the 3-OH to a
ketone is followed by introduction of the 11,12-cyclic
carbamate according to the method of Baker, using
an amine prepared in several steps from 4-(3-pyridyl)imidazole and 4-bromobutylphthalimide. Treatment of the product with methanol results in removal
of the 2-acetate group and production of telithromycin. This rather lengthy sequence starts from clarithromycin, and so is 14 steps removed from erythromycin
A. This adds substantially to the cost of the drug,
and indeed, telithromycin may represent the economic limit of what is feasible in the antibacterial
market.
The most important benefit of telithromycin is its
unprecedented in-vitro potency against macrolideresistant S. pneumoniae.22,23 Resistance to telithromycin in S. pneumoniae has not yet been reported
over the 2 years that the drug has been in clinical
use in Europe. As will be described in more detail
below, the two most prominent mechanisms of acquired macrolide resistance are efflux and ribosome
methylation. Unlike azithromycin and clarithromycin, telithromycin evades the efflux pumps found in
S. pneumoniae and S. pyogenes and does not induce
ribosomal methylation associated with inducible
MLSB resistance in streptococci and staphylococci.
However, staphylococcal and S. pyogenes strains that
carry methylated ribosomes are not susceptible to
telithromycin. The differences between S. pyogenes
and S. pneumoniae with regard to ribosomal methylation and telithromycin susceptibility are further
discussed below.
A second ketolide, cethromycin (ABT-773; Abbott)
[31], also designated ABT-773 (developed by Abbott
Laboratories and not yet FDA approved), carries the
11,12-cyclic carbamate and the 3-keto group present
in telithromycin, but the aryl alkyl side chain is
attached in an ether linkage to the 6-hydroxyl group.
Synthesis of cethromycin and other 6-O-arylalkyl
ketolides has been described previously.24 As with
telithromycin, cethromycin is prepared through a
lengthy series of chemical transformations. Erythromycin A is converted into the 9-oxime and protected
as the 9,2,4-tribenzoate. This derivative is allylated
on the 6-OH using the tert-butyl carbonate of 1-(3quinolyl)-2-propenol with palladium catalysis. Subsequent deblocking of the oxime and deoximation
provides the 9-ketone, which is subjected to 11,12cyclic carbamate formation in a one-pot, four-step
sequence. Subsequent cladinose hydrolysis requires
rather forcing conditions, as 4-O-acylated cladinose
is rather resistant toward hydrolysis. Oxidation of
the resulting 3-OH group provides the 3-ketone. Final

Translation and Protein Synthesis

debenzoylation of the 2-OH provides cethromycin.


Cethromycin has similar potencies as telithromycin
against the macrolide-susceptible organisms, does not
induce MLSB resistance, and is active against effluxmediated resistant and constitutive MLSB-resistant
S. pneumoniae and, unlike telithromycin, S. pyogenes.
In addition to their activity against macrolideresistant streptococci, the ketolides also have the
unexpected and unprecedented property of bactericidal activity against S. pneumoniae. All macrolides
exhibit time-dependent (12-24 h after administration), concentration-independent killing of bacteria
and are classified as bacteriostatic rather than
bactericidal agents. Ketolides, on the other hand,
exhibit concentration-dependent killing of S. pneumoniae but not S. pyogenes, S. aureus, or H. influenzae.25 Thus, although the basis is not understood,
the ketolides are considered bactericidal for S. pneumoniae only. This desirable property may forestall
the development of resistance to ketolides in these
organisms.
Although some differences between cethromycin
and telithromycin in individual pharmacokinetic
parameters have been demonstrated, the two compounds are quite comparable overall in efficacy in
experimental animals and according to initial reports
in humans as well. Telithromycin is administered
once per day, albeit at 800 mg dosing; the dosing of
cethromycin to humans has not yet been reported.
At present, the only significant reported difference
between the two compounds is the lack of efficacy in
vitro of telithromycin against constitutive MLSBbased macrolide-resistant S. pyogenes.26

3.4. Side Effects of Macrolides and Ketolides


Until recently the most significant side effects of
macrolides reported have been their ability to induce
stomach cramps in some individuals and the bitter
aftertaste of some of the compounds. High-level
interest is now focused on the occurrence of torsades
de pointes upon treatment with macrolides. Torsades
de pointes is a rare but potentially fatal ventricular
arrhythmia associated with delayed repolarization
and prolongation of the QT interval. Interactions
between macrolide antibiotics and other drugs that
prolong the QT interval have been known to cause
torsades de pointes, but recent studies have demonstrated that clarithromycin itself may induce prolongation of the QT interval and may lead directly to
ventricular arrhythmia. Azithromycin alone does not
appear to have any effect on the QT interval in rats,
but reports of QT prolongation associated with azithromycin in combination with other drugs have appeared recently.27-29 Telithromycin also induces a
modest increase in the QT interval, although smaller
than that induced by erythromycin or clarithromycin.
Subsequent studies and clinical use have also suggested unexpectedly frequent cases of temporary
visual disturbances.30 Concerns were voiced over both
potential side effects at the FDA Anti-infective Drugs
Advisory Committee hearing on telithromycin in
2001.31

Chemical Reviews, 2005, Vol. 105, No. 2 507

4. Mode of Action
4.1. Inhibition of Translation
It has been known since their discovery that
macrolides block protein synthesis, but the molecular
details of how they arrest translation has been
uncovered only very recently. Footprinting experiments and detailed studies of macrolide resistance
over a period of more than 20 years indicated that
these compounds bind to the 50S component of
bacterial ribosomes and make specific interactions
with the 23S RNA. Early studies employing biochemical assays of the individual activities associated with
the translation processsinitiation, peptide bond formation, and translocationsled to the following conclusions: all macrolides bind in the region of domain
V of the ribosome in the peptidyltransferase center;
carbomycin and other 16-membered macrolides that
carried acyl extensions on the mycarose moiety were
found to block peptidyltransferase activity (peptide
bond formation) by binding the A site and blocking
the binding of aminoacyl tRNA; erythromycin and
other 14-membered macrolides were found to have
no effect on peptidyltransferase activity. Treatment
of bacterial cells with macrolides were found to cause
accumulation of peptidyl-tRNA, prompting workers
to suggest that the primary mechanism of action
common to all macrolides was premature ejection of
peptidyl tRNA from the ribosomes.32
Determination of the nucleotide sequences of ribosomal RNA and proteins enabled identification of the
sites on the ribosome with which macrolides interacted. Footprinting experiments (protection of nucleotides in ribosomal RNA from chemical modification
due to binding of added compounds to purified
ribosomes) demonstrated direct interaction between
macrolides and 23S rRNA.33 All macrolides, ketolides,
lincosamides, and streptogramin B protected nucleotides 2058-2062 (in domain V), but tylosin also
protected nucleotide A752 (in domain II).34 Telithromycin and cethromycin also protected A752.35,36
Erythromycin, on the other hand, protected the
domain V region but made A752 more susceptible to
chemical modification.34 These experiments, along
with determinations of the sites in the 23S ribosomal
RNA that conferred resistance by mutation or enzymatic modification, identified the precise locations on
the ribosome where macrolides were bound. Less was
known about the atoms on the macrolides themselves
that interacted with the ribosomal RNA.

4.2. MacrolideRibosome Structural Studies


Solution of the ribosomal structure at atomic
resolution with macrolides bound has clarified some
of the enigmas that have arisen associated with
macrolide action yet has raised new issues as well.
X-ray crystal structures of the 50S subunits of
ribosomes from both Haloarcula morismortui and
Deinococcus radiodurans in the presence of macrolides, ketolides, or the streptogramins were determined in the laboratories of Tom Steitz and Ada
Yonath, respectively.37-41 The structures of the Haloarcula ribosomal subunit with bound macrolides

508 Chemical Reviews, 2005, Vol. 105, No. 2

Figure 2. View of erythromycin A bound to the 50S


subunit of the D. radiodurans ribosome, looking down the
peptide exit tunnel toward the peptidyl transferase center.
The macrolide binding site is composed of a purine-rich
pocket formed by residues from domain V (blue) with
contributions from domains II (red) and IV (magenta).
Binding of erythromycin blocks peptide formation by closing the peptide exit tunnel some distance from the peptidyl
transferase center. Residue A2058 (blue) is critical to
binding the desosamine sugar and is the site of methylation
in erm-based resistance. Residue 752 (red) is protected by
ketolide binding.

were obtained even though the Haloarcula ribosome


is not expected to be macrolide susceptible due to the
presence of G rather than A at position 2058 (E. coli
numbering). As revealed by the structure of erythromycin A bound to the Deinococcus subunit (Figure
2), the macrolide binding pocket consists of RNA from
domains II, IV, and V, with the majority of the pocket
being composed of residues from domain V. There are
some stabilizing contributions to the binding pocket
from ribosomal proteins L3, L4, L22, and L34, but
there appear to be no direct contacts between the
macrolide and these ribosomal proteins. The more
highly conserved structural region of the macrolides,
from C1 to C8, lies against a wall of mostly purine
residues from domain V. This interaction with domain V includes tight, specific interactions between
the desosamine residue and a binding pocket containing A2058 (Figure 3). The remaining region of
erythromycin interacts rather loosely with a pyrimidine-rich side of the tunnel composed of residues from
domains II and IV. The dearth of specific contacts
(seven H-bonds) between the macrolide and the
ribosome make it difficult to rationalize the very high
binding affinities observed. Nonetheless, RNA (Figure 4) and ribosomal protein mutations previously
known to affect macrolide susceptibility lie within or
near this binding pocket, thus adding confidence in
the relevance of these crystal structures. The binding
pocket lies in the peptide exit tunnel 10-15 distal
from the peptidyltransferase site; macrolide binding
appears to block progression of peptide chain upon
contact between the growing peptide chain and the
macrolide, which occurs after a small number of
elongation steps. This is in agreement with biochemical data showing the formation of very short peptides
in the presence of erythromycin. The cladinose residue of erythromycin points along the tunnel toward

Katz and Ashley

Figure 3. View of the erythromycin binding pocket on the


50S subunit of the D. radiodurans ribosome, showing the
close interactions with the desosamine residue. Regions of
high negative charge are colored red; a primary interaction
appears between the phosphate of G2505 (far right) and
the desosamine amino group. A2058 lies at the bottom of
the pocket in this view.

Figure 4. Position of 23S RNA residues at the macrolide


binding site (red) where mutation is known to lead to
erythromycin (yellow) resistance.

the peptidyltransferase site, in agreement with biochemical experiments, indicating that derivatives of
erythromycin acylated on the 4-OH of the cladinose
residue and thus extending further toward the peptidyltransferase site may interfere with peptidyltransferase activity.
While much of the macrolide binding pocket appears loose and rather devoid of specific contacts,
quite specific contacts are observed between the
desosamine sugar and the RNA in the region of
A2058, including a probably crucial charge interaction with the phosphate of G2505 (Figure 3). Not
surprisingly, quinupristin, a streptogramin B compound, also makes specific interactions with A2058.
As described in more detail in section 5, alterations
to A2058 result in macrolide resistance; methylation
at N6 of A2058 is a common mode of bacterial
resistance to macrolides, lincosamides, and the streptogramin B compounds as is mutation of A2058 to
G. Both alterations to A2058 result in loss of specific
contacts between A2058 and the 2-hydroxyl and 3dimethylamino groups of desosamine. Similarly,
chemical modifications to either the 2-hydroxyl or
the 3-dimethylamino group of macrolides has been
found to greatly reduce or destroy antibacterial
activity.

Translation and Protein Synthesis

Crystal structures of ketolides bound to these 50S


subunits have also been revealing. Removal of the
3-O-cladinosyl group in the ketolides results in a
dramatic loss in potency that is compensated for by
the addition of heteroaryl groups either at the 11position (telithromycin) or the 6-position (cethromycin). In agreement with ribosomal footprinting experiments, which indicated protection of residue
A752 in domain II, the heteroaryl groups of both
telithromycin and cethromycin were found to bind to
a region of domain II adjacent to the ribosomal
binding pocket.42 In both cases, the position of the
macrolactone portion of the ketolides was observed
to be slightly shifted relative to that seen for erythromycin A, leading to the suggestion that the ketolides may tolerate some perturbations to the macrolide binding site while compensating for lost
interactions by picking up new binding from the
heteroaryl groups. However, the observed shift is
roughly within the resolution of the structures, and
such findings should not be overinterpreted at this
stage of refinement. More specific interactions were
observed between the telithromycin heteroaryl group
and the ribosome than for the cethromycin heteroaryl
group; as cethromycin has generally better in-vitro
activity against a wide range of organisms, it is clear
that such apparently improved ribosomal binding
does not necessarily translate to improved antibacterial activity.
The current X-ray crystal structures of macrolides
bound to 50S ribosomal subunits offer a snapshot of
macrolide action at the ribosome, and it is important
to remember that ribosomes are dynamic machines
and that the complete picture of macrolide activity
is likely to be significantly more complex. Macrolides
are known to act during translation, for example,
with the actual inhibited complex consisting of a
macrolide bound to a complete ribosome having a
partially completed peptide in the exit tunnel. There
may well be more specific interactions between the
macrolide and the complete ribosome-peptide complex than are observed in the current X-ray crystal
structures.
The structural studies have led to the conclusion
that binding of the macrolide to the ribosome is
sufficient to block the progression of peptide synthesis beyond the di- to hexapeptide stage. Hence,
binding alone may be sufficient for the antibiotic
action of these compounds, and the additional effects
of macrolides observed in vitro may not be required
for efficacy. On the other hand, the information
obtained from the structural work on two ribosomes
that are from clinically nonrelevant organisms does
not, at this point, provide answers to all effects seen
by macrolides on different pathogenic strains or by
different macrolides on individual strains. Azithromycin and claithromycin appear to bind in a fashion
similar to the ribosome, but azithromycin has better
potency against H. influenzae and is less potent
against S. pneumoniae and S. pyogenes.43,44 Telithromycin binds E. coli and S. pneumoniae ribosomes
with Kd ) 2-10 nM; the Kd of clarithromycin is 3050 nM, yet the two compounds have equal potencies
against S. aureus and S. pneumoniae in vitro.45

Chemical Reviews, 2005, Vol. 105, No. 2 509

Finally, the structural studies do not themselves


provide any clues as to why the ketolides are bactericidal to S. pneumoniae but only bacteriostatic to S.
pyogenes, S. aureus, and H. influenzae. Are the
differences among the ribosomes from these different
organisms sufficient to account for the different
effects of these compounds? Is the mode of action of
these compounds entirely explained by their binding?
It is likely that differences in intracellular accumulation among the various bacteria, rather than differences in ribosomal structure, may account for all or
most of the observed differences. Nonetheless, it
would be interesting to see the molecular details of
interaction of macrolides and ketolides with ribosomes from E. coli, H. influenzae, and Gram-positive
pathogens.

4.3. Inhibition of Ribosome Assembly


Champney and co-workers have shown that
macrolides and ketolides inhibit the assembly of
the 50S ribosome unit in a number of organisms
including S. aureus, S. pneumoniae, E. coli, and H.
influenzae.46-51 Assembly of the 30S ribosome was
unaffected by these compounds. Using sucrose density gradient sedimentation analysis of ribosomes
prepared from cells pulse-labeled with 3H-uridine and
chased with an excess of the unlabeled nucleoside,
they showed that the addition of macrolides and
ketolides promotes accumulation of a 32S particle
which degrades upon continued exposure to the drug.
Using pulse labeling to analyze translation, they
determined that the IC50s for ketolides for arresting
translation and inhibiting ribosome assembly in S.
aureus were the same, ca. 10 nM, the approximate
Kd of ketolide-ribosome interactions in vitro. Not
surprisingly, inhibition of 50S subunit assembly
requires macrolide/ketolide binding. Assembly of the
50S subunit in bacteria takes place unobstructed in
cells that carry MLSB resistance in the presence of
macrolides/ketolides, suggesting that these compounds can interact with subribosomal particles.
These data also indicate that in the assembly of 50S
ribosomes, if methylation of A2058 does take place,
it must occur before the assembly of a ribonucleoprotein particle that can interact with macrolides.
Champney proposed that macrolide binding to such
a particle directly prevents the addition of one or
more ribosomal proteins to the maturing particle and
leaves segments of the rRNA in the particle exposed
to the action of cellular RNases. Whether cessation
of ribosome assembly is sufficient to explain the
bactericidal effect of ketolides in S. pneumoniae
remains to be seen.

5. Macrolide Resistance
Resistance to erythromycin was first reported in
1952, the same year erythromycin was introduced
into clinical practice, in two strains of S. aureus, the
first organism targeted by the drug.52 Resistance also
developed in most of the other organisms against
which erythromycin and other macrolides were used,
but accurate estimates of macrolide resistance in
different countries and different locations within

510 Chemical Reviews, 2005, Vol. 105, No. 2

Katz and Ashley

countries have been difficult to determine accurately.


Though there is uncertainty about the exact extent
of resistance, there is no doubt, however, that that
resistance to macrolides is an important basis for
clinical failure of macrolide therapy.
Genes associated with macrolide resistance have
been found in all the Gram-positive pathogens for
which erythromycin and other macrolides were prescribed as well as in strains that were not targeted
by macrolides. Resistance genes are present in the
microorganisms that produce macrolides. The two
most common resistance mechanisms in the bacterial
pathogens are (1) reduced binding of the drug due to
modification of the bacterial ribosome, either through
the acquisition of a resistance gene or through
mutation in the host, and (2) efflux of macrolides
from the bacterial cell, through acquisition of a
resistance gene. Less common mechanisms include
direct inactivation of the antibiotic itself. Clinical
strains have been uncovered that carry more than a
single type of resistance. Most of the genes that
confer self-resistance in the macrolide-producing
actinomyctes are counterparts of the resistance genes
found in clinical isolates.

S-adenosyl-methionine as methyl donor and all enzymes have signature sequences characteristic of
S-AdoMet binding sites. Structures of ErmAM and
ErmC have been solved.57-59 The erm genes have
been found on high and low copy plasmids and within
transposons, often in association with other antibioticresistance genes. They are also found in the chromosomes of macrolide-producing organisms, clustered
among the genes for macrolide biosynthesis. The
ermE gene from the erythromycin-producer Saccharopolyspora erythraea has been found in commercial
preparations of the drug, causing one to wonder
whether resistance in clinical isolates originated from
the producing strain and whether it was spread
directly from use of the drug.60-62
Erm-mediated resistance exists in two forms: inducible and constitutive. In the inducible form resistance and hence ribosome methylation develop only
after the macrolide is administered to the cells. In
hosts that are constitutively resistant to macrolides,
Erm-catalyzed methylation of the ribosomes does not
require the presence of macrolides. Both inducible
and constitutive MLSB resistance require an intact
coding sequence of the erm gene.

5.1. MLSB Resistance

5.1.1. Inducible Resistance

Gram-positive cells (and E. coli) can acquire a gene


that confers high-level resistance to macrolides, lincosamides (e.g., clindamycin), and members of the
streptogramin B class of antibiotics (e.g., pristinamycin I).53 The basis for this type of resistance is
either N6-mono- or N6,N6-dimethylation of nucleotide A2058 (E. coli numbering) in 23S ribosomal
RNA. Genetic, biochemical, and structural data have
shown that the MLSB phenotype is conferred from
the overlapping binding of these molecules to domain
V making contact with A2058. It is believed, but has
yet to be proven conclusively, that either methylation
of A2058 changes the structure of the site sufficiently
so that macrolides no longer bind or the bulky methyl
groups interfere directly with the binding of the drug.
The enzyme class was named Erm for erythromycin
resistance methylase, and the genes that determine
these enzymes were designated ermA, ermB, ermC,
etc. At present, 21 classes of erm genes, some
containing six or more members, have been identified.54 These proteins are approximately 29 KDa and
show very high degrees of sequence conservation. In
vitro, Erm-mediated methylation uses 23S rRNA as
substrate and does not take place on intact ribosomes
or the 50S subunit.55 The actual substrate for methylation in bacteria has not been determined conclusively. The Erm enzymes do not appear to be
specific for their cognate substrates: all Erm enzymes tested use 23S rRNA obtained from many
species of bacteria as well as 23S RNA generated by
in-vitro transcription. Some of the enzymes, such as
ErmN, catalyze only monomethylation, whereas others, such as ErmE and ErmC, catalyze dimethylation,
but it is not known whether dimethylation takes
place through a concerted two-step process. These
latter enzymes can use monomethylated RNA as a
substrate.56 ErmAM (also called ErmB) catalyzes
either mono- or dimethylation. Methylation employs

The best-studied mechanism of inducible MLSB


resistance involves the ermC gene found in S. aureus
and was based on the initial observations that cells
resistant to erythromycin and susceptible to 16membered macrolides, lincomycins, and pristinamycin I could become resistant to the latter three classes
if treated first with small doses of erythromycin.63
The basis of inducible MLSB resistance has emerged
over the past 30 years and is summarized here.53 The
ermC-coding region is preceded by a sequence which
encodes a 19-amino acid leader peptide and the two
genes, separated by a segment consisting of 81
nucleotides, form an operon. Each gene has its own
ribosome binding site (RBS). The mRNA segment
corresponding to the leader peptide contains several
overlapping inverted repeats and, theoretically, can
assume a number of secondary structures, including
one in which the ribosome binding site of ermC is
sequestered, resulting in the inability of the ribosomes to enter the site and translate the mRNA
corresponding to the ermC gene. Under such conditions the ribosomes would not be methylated and the
cells would be susceptible to macrolides. The gene
for the leader peptide, however, whose RBS is exposed, is expressed in these cells. In the presence of
erythromycin, the model proposes that the mRNA
corresponding to the leader peptide undergoes reorganization wherein the RBS of the ermC gene is
exposed so that it can be translated, producing the
methylase that acts to generate methylated ribosomes and thereby conferring resistance to erythromycin and other MLSB antibiotics. A fascinating
model of the induction process has been developed
and is reviewed in detail by Weisblum.53 Briefly
summarized, molecules of erythromycin enter the
cells, bind to ribosomes engaged in synthesis of the
leader peptide, and cause the translation process to
stall after the ninth amino acid is introduced into the

Translation and Protein Synthesis

nascent peptide, generating the peptide MGIFSIFVI


attached to tRNA. The induction model proposed an
association among erythromycin, the stalled leader
peptide on the ribosome, and the mRNA into a yet
to be understood complex that results in the change
in the secondary structure of the leader region to
permit ribosomes to bind to the RBS and translate
the ermC gene. A requirement for the stalled leader
peptide in the induction process was based on the
findings that mutations affecting the leader sequence
after Ileu-9 had no effect on inducibility, but mutations resulting in translation termination of the
leader before Ileu-9 resulted in noninducibility (failure of erythromycin to confer resistance). Mutations
further into leader segment in the region surrounding
the RBS, which themselves would destabilize the
secondary structure of the mRNA in that region,
resulted in constitutive resistance, i.e., expression of
ermC in the absence of erythromycin. Separation of
the leader peptide from the ermC gene, or introduction of a nonsense codon at a position corresponding
to residue 10, also resulted in noninducibility. Finally, within the first nine residues of the leader
peptide, amino acid substitutions of some of the
residues did not affect inducibility, but substitutions
at other residues resulted in noninducibility. These
findings demonstrated that induction required the
first nine amino acids of the leader peptide, that the
structure of the peptide was important, and that the
leader peptide must located cis to the erm gene and
be interrupted in its translation. Moreover, and most
importantly, induction depended upon the presence
of the antibiotic with the correct structuresa 14- or
15-membered macrolide that contained the neutral
sugar at C-3; 16-membered macrolides and (14membered)-ketolides are not inducers. The lincosamide celesticetin was later determined to be an
inducer.64 Derivatives of erythromycin that are devoid of antibiotic activity are also not inducers.
Within the current framework of ribosome structure and macrolide binding, it is difficult imagine the
role of the macrolide in the induction process. Erythromycin, binding in the polypeptide exit tunnel, could
allow the stalling of translation to generate the
9-residue leader peptidyl-tRNA, but other than the
tRNA component of the leader peptidyl-tRNA, neither the peptide itself nor erythromycin is in contact
with the mRNA, in particular the segment 70 nucleotides downstream that contain the RBS. If erythromycin does make direct contact with the mRNA,
it must employ different atoms than those used for
binding to rRNA in domain V. The cladinosyl moiety
is a likely candidate for such interactions since it is
required for induction. On the other hand, it has not
been ruled out that the noninducers, such as the
ketolides and 16-membered macrolides, cause the
ribosome to stall in the leader at a site different from
that caused by erythromycin so that the correct
inducer peptide is not produced. A structure of the
induction complex at atomic resolution is needed
to enable fuller understanding of inducible resistance.
Other examples of inducible MLSB resistance have
been reported. TlrA (also named ErmSF and ErmS)
in Streptomyces fradiae, the tylosin producer, is an

Chemical Reviews, 2005, Vol. 105, No. 2 511

A2058-dimethyltransferase that is induced by tylosin


(or a precursor in the biosynthesis pathway) not
erythromycin.64 ErmSV in Streptomyces viridochromogenes NRRL 2860 is induced by either tylosin or
erythromycin.65 Interestingly, the S. fradiae host also
contains two additional 23S rRNA methyltransferases, TlrD (ErmN), an A2058 monomethyltransferase that is induced by tylosin but not erythromycin, and TlrB, a constitutive methyltransferase that
acts on G748 in domain II. Methylation by either
TlrB or TlrD alone does not confer tylosin resistance;
resistance is conferred by the two methylations acting
synergistically.66 Induction of each of these A2058
methyltransferases is believed to occur through a
translational attenuation process analogous to that
described for ErmC with different structural requirements for the leader peptide and macrolide.
An interesting variation on the mechanism for
inducible ErmK-mediated MLSB resistance in Bacillus lichenoformis has been reported. In addition to
translational attenuation observed for ErmC production, in the absence of inducer, transcription is halted
in the leader region through a rho-independent
transcription terminator. In the presence of inducer,
transcription proceeds through ermK.67

5.1.2. Constitutive Resistance


In this class the erm genes are constitutively
expressed in their hosts and thus confer resistance
to all MLSB antibiotics without the need for prior
exposure to one or another macrolide. Both MLSBinducible (resistant to erythromycin but susceptible
to tylosin) and MLSB-constitutive (resistant to erythromycin and tylosin) strains have been found in
clinical isolates of S. aureus harboring ermC. Most
of the isolates in the latter class carry mutations,
deletions, or duplications in the leader region that
are thought to destabilize the secondary structure
and allow expression of the ermC gene in the absence
of inducer. Mutation from MLSB inducible to MLSB
constitutive can also be accomplished in the laboratory by simply plating inducible cells in the presence
of tylosin and selecting for survivors.68

5.1.3. Inducible vs Constitutive


S. aureus cells that carry erm genes exhibit either
fully MLSB-inducible or MLSB-constitutive phenotypes. In inducible strains methylation of ribosomal
RNA could not be detected prior to exposure of the
cells to erythromycin.69 Hence, MLSB-inducible S.
aureus strains are almost fully susceptible to noninducers such as 16-membered macrolides and ketolides. In clinical isolates of S. pneumoniae carrying
ermAM, a wide range of susceptibility to noninducers
has been observed. The degree of resistance (minimum inhibitory concentration) to the noninducing
macrolide and ketolides has been correlated with the
degree of dimethylation of A2058 in these strains
determined before exposure to the drug.69 Addition
of erythromycin to all strains promoted increased
resistance to tylosin, resulting from additional dimethylation of A2058. Thus, the high-level resistance
of all clinical isolates of S. pneumoniae containing
ermAM to clarithromycin and azithromycin is most

512 Chemical Reviews, 2005, Vol. 105, No. 2

likely due to full induction of the methyltransferase


by the drugs resulting in production of fully dimethylated ribosomes. The differential response of the
same strains to the noninducers tylosin and telithromycin can also be rationalized. Resistance to tylosin,
where seen, is likely due to the constitutive presence
of sufficient numbers of dimethylated ribosomes to
allow a level of protein synthesis necessary for
survival. Susceptibility to telithromycin, on the other
hand, may be due to the unique bactericidal effects
that take place upon binding of the drug to the
unmethylated (or monomethylated) ribosomes present
in these hosts.
S. pyogenes strains carrying ermAM are much more
resistant to telithromycin than their ermAM-containing S. pneumoniae counterparts but are still very
susceptible to cethromycin. It is not yet known
whether the difference between the two species is due
to the fact that telithromycin (but not cethromycin)
can induce ermAM-mediated resistance in S. pyogenes but not in S. pneumoniae. A more compelling
explanation would be that the levels of dimethylation
are greater in S. pyogenes than in S. pneumoniae and
that dimethylated S. pyogenes ribosomes, while refractory to telithromycin binding, can still bind
cethromycin with clinical efficacy.

5.2. Efflux
Decreased accumulation due to efflux in a macrolide-resistant isolate was first reported in the 1980s
in S. epidermidis and in the 1990s in S. pyogenes and
S. pneumoniae and presently accounts for a significant proportion of the macrolide-resistant S. pneumoniae strains identified.70-75 Efflux-mediated resistance is still relatively rare in S. aureus. In streptococci, macrolide efflux is mediated by the gene
product encoded by mefA, the name denoting a group
of genes encoding proteins that share >90% identity.
MefA confers resistance to 14- and 15-membered
macrolides but not 16-membered macrolides, lincosamides, or streptogramin B. Furthermore, Mef-mediated resistance is induced by the presence of clarithromycin and azithromycin but not by 16-membered
macrolides. Ketolides are poor inducers of MefA and
hence are still very potent antibacterials against
streptococci carrying this gene. Since the Mef proteins do not contain recognizable ATP binding sites
and resistance to macrolides in mefA-containing hosts
takes place in the presence of ATP-associated energy
uncouplers, Mef-mediated transport of the macrolide
is believed to be driven by a proton motive force.
Efflux in staphylococci is mediated by MsrA, a
member of the ABC superfamily that employs ATP
as the energy source for transport and is thought to
work in concert with a membrane-associated host
protein to confer resistance. MsrA confers high-level
resistance to 14- and 15-membered macrolides and
streptogramin B and weak resistance to ketolides,
and it does not confer resistance to 16-membered
macrolides and lincosamides. MsrA is induced by
clarithromycin, azithromycin, and telithromycin but
not by streptogramin B, even though the latter
compound is a substrate for MsrA-mediated transport. Nucleotide sequencing of the region upstream

Katz and Ashley

of msrA revealed a leader sequence reminiscent of


the leader upstream of ermC, suggesting that MsrA
is induced by a translational attenuation process.76
ABC transporters have also been found in some
Streptomyces hosts that produce macrolides. These
genes are located at the edges of their cognate
biosynthesis gene cluster and confer resistance to 16or 14-membered macrolides when expressed in heterologous hosts. Their roles in conferring selfresistance or in export of the macrolide during
production are not yet known.
A number of transport systems not specific for
macrolides have been identified in Gram-negative
bacteria. These tripartite pumps are members of the
RND family and are composed of an inner membrane
component, which extrudes the macrolide in exchange for a proton, a protein in the outer membrane
that may form a gated channel (pore), and a periplasmic protein that links the two membrane-associated efflux proteins. Examples include the MexABOprM system in Pseudomonas aeruginosa, the
AcrAB-TolC system in E. coli, and the acrAB-Omp2
system in H. influenzae. These systems, encoded in
the chromosomes of Gram negatives, are the primary
bases for intrinsic resistance to membered macrolides
as well as many other compounds including antibiotics such as rifampicin, novobiocin, and tetracycline.
In some hosts the RND pump genes are expressed
constitutively; in others, a mutation is required RNDmediated resistance.

5.3. Mutations in Ribosomal RNA


Bacteria contain between one and seven copies of
the operons that encode the genes for ribosomal RNA.
Mutations in domain V encoding resistance to
clarithromycin have been reported in clinical isolates
of a number of organisms, including H. pylori, S.
aureus, S. pneumoniae, H. influenzae, Neisseria gonorrheae, Mycobacterium tuberculosis, Mycobacterium
avium, and Treponenum pallidum.77-87 Several patterns of resistance were seen. Deletion of A2058 in
S. pneumoniae conferred high-level resistance to
macrolides and increased the MIC to 4 g/mL of
telithromycin, but it is not clear whether the increase
in MIC translates to resistance to the ketolide in a
clinical setting. A2058G or A2058T mutations conferred high-level resistance to all three classes of
MLSB antibiotics. A2059G mutations conferred intermediate-level resistance to macrolides and lincosamides but did not confer resistance to streptogramin B. In N. gonorrheae, a C2611T mutation (in
domain V) was identified. In H. pylori, the particular
mutation was found in each of the two copies of the
rrl gene (23S rRNA) present in the chromosome. In
S. aureus, the mutation was present in a minimum
of four of the six rrl genes present in the host and in
N. gonorrheae three of the four rrl genes. In T.
pallidum the A2058G mutation was present in both
copies of rrl. In no cases did a resistant strain carry
more than a single type of mutation, suggesting that
each mutation was introduced into a single copy of
the rrl genes, and through selection in the presence
of the drug, the mutant allele replaced all or most of
the wild-type copies of the gene in the host, likely
via a process involving recombination.

Translation and Protein Synthesis

Chemical Reviews, 2005, Vol. 105, No. 2 513

siella, and Enterobacter species) as well as in clinical


isolates of S. aureus.99,100 Currently, esterase-mediated resistance to erythromycin is rare in S. aureus
and has yet to be detected in streptococci. These
enzymes are specific for 14-membered macrolide
substrates. Two streptogramin B hydrolases, VgbA
and VgbB, have recently been identified in S. aureus.101 These enzymes do not employ macrolides as
substrates.

Mutations in domain V in E. coli strains selected


for resistance to macrolides have been mapped to
nucleotides, 2058, 2059, and 2612.88 A U2609C mutation in domain V was found in an E. coli strain
selected for resistance to telithromycin or cethromycin.89 Interestingly, this mutation increased slightly
the susceptibility of the host to erythromycin and
azithromycin.
Mutations in domain II in the vicinity of A748
conferring resistance to macrolides or ketolides have
not been reported in clinical isolates, but the U754A
mutation in the hairpin 35 segment of domain II was
found in an E. coli host selected for resistance to
telithromycin.90 A different class of laboratoryselected strains of E. coli that exhibited increased
resistance to erythromycin was found to carry mutations within a hairpin structure between nucleotides
1198 and 1247 in domain II of the 23S rRNA, close
to a segment of the RNA that encodes a pentapeptide
(E-peptide) that confers resistance to erythromycin.91
It is believed that mutation in this region of domain
II increases expression of the segment encoding the
E-peptide. Interestingly, it was found that the peptide acted only cis on ribosomes carrying the 23S
RNA harboring the domain II mutation and conferred
resistance only to erythromycin and not ketolides or
16-membered macrolides.92 By site-directed mutagenesis of the E-peptide coding region, the sequence
could be changed to permit the production of different
peptides that conferred resistance to ketolides or 16membered macrolides.93 Although this mechanism of
macrolide resistance in E. coli is not clinically relevant at the present time, these findings raise the
possibility that short peptides produced from the
rRNA as well as segments of the ribosomal RNA itself
may play a role in the binding of macrolides to
ribosomes to stop translation or, perhaps, to promote
expression of an erm gene.

Enzymes that transfer phosphate from ATP to the


2-OH of erythromycin were originally discovered in
E. coli. Members of the MphA group employ 14- and
15-membered macrolides as substrates exclusively.102,103 The MphB enzyme can phosphorylate
both 14- and 16-membered macrolides.104 Macrolide
2-phosphotransferase activity, related to MphA, was
recently detected in two clinical isolates of P. aeruginosa from hospital patients in Japan, where macrolides are used for long-term chemotherapy of P.
aeruginosa panbronchiolitis.105 A related enzyme,
MphC, has also been identified in a clinical isolate
of S. aureus.106 Expression of the mphA gene in E.
coli is regulated by an adjacent gene, mphR, whose
gene product binds to the operator-promoter region
of mphA and represses transcription. Transcription
of mphA takes place in the presence of erythromycin,
which enters the cell, binds to MphR, and removes
it from the operator-promoter.107 In this system
erythromycin is the inducer of (self-)resistance. Although the MphA-MphR resistance system has thus
far been found only in E. coli, it is reasonable to
suggest that it originated in a macrolide-producing
bacterium and that a 2-phosphatase, which would
restore antibacterial activity to 2-phosphoerythromycin, would also be uncovered in a macrolideproducing host.

5.4. Mutations in Ribosomal Proteins

5.5.3. Glucosylation

A number of clinical isolates of H. influenzae, S.


aureus, and S. pneumoniae resistant to macrolides
have been characterized to carry mutations in genes
for 50S ribosomal proteins L4 or L22.81,82,87 As
described above, these two proteins border the
polypeptide exit tunnel. Mutations in E. coli conferring resistance to erythromycin were also determined
to reside in the genes for L4 and L22 proteins.
Ribosomes from L4 mutants exhibited reduced binding of erythromycin, but the L22 mutant ribosomes
could still bind drug, indicating that the mutation
affected the structure of the tunnel such that binding
of macrolide did not block translation.94-96

Macrolide resistance mediated through 2-glucosylation has not been reported in a bacterial pathogen
but has been found in Streptomyces antibioticus, the
producer of oleandomycin.108 Extracts of several other
streptomycetes were found to contain activities that
transferred the glucose moiety from UDP-glucose to
a number of 12-, 14-, and some 16-membered macrolides, suggesting that the resistance gene spread
from a macrolide producer.109-111 In their natural
locations in the chromosome the mgt genes confer
weak resistance to macrolides on their hosts. In S.
antibioticus, the MGT gene, oleI, is accompanied by
the gene oleR, which encodes an enzyme that removes the glucose residue from 2-glucosyloleandomycin, restoring the antibacterial activity to the
compound.108 Both oleI and oleR are located in the
oleandomycin biosynthesis cluster. OleI is thought
to confer self-resistance to the host while the compound is produced intracellularly, and OleR restores
its activity during or prior to transport from the host.
It is interesting to note that the oleandomycin biosynthesis cluster does not contain an erm gene;
hence, the host makes oleandomycin-employing ribosomes that are susceptible to the drug.

5.5. Enzymatic Inactivation of Macrolides


5.5.1. Hydrolysis of the Macrolactone
Two unrelated genes, ereA and ereB, each conferring resistance to erythromycin in E. coli, were
identified on separate plasmids and shown to encode
esterases that opened the macrolactones of erythromycin and oleandomycin.97,98 These genes were subsequently identified in a number of members of other
Gram-negative bacteria (Citrobacter, Proteus, Kleb-

5.5.2. Phosphorylation

514 Chemical Reviews, 2005, Vol. 105, No. 2

6. Biosynthesis of Macrolides
Biosynthesis of macrolides follows discrete biochemical pathways but can be viewed as taking place
in three stages: synthesis of the aglycone, synthesis
of the sugars and attachment to the aglycone, tailoring steps to produce the completed product. The
genes for the biosynthesis of the aglycone and deoxysugars and the genes for the tailoring steps are
generally clustered. Genes that confer self-resistance
are located within the biosynthesis cluster. Much of
our current understanding of macrolide biosynthesis
has come from the determination of the nucleotide
sequence of the biosynthesis genes in the 1990s.
However, the biochemical pathways for erythromycin
and tylosin were largely understood well before this
period from the analysis of compounds produced in
fermentations of mutants blocked at different steps
of the synthesis.112-114 In addition, early feeding
experiments indicated that macrolides were produced
from acetate, propionate, and butyrate, but the key
experiments demonstrating the bioconversion of compounds 5-9 carbons in length with structures representing intermediates in aglycone biosynthesis into
the aglycones of erythromycin and tylosin indicated
that biosynthesis of the macrolactone takes place
through a stepwise process.115,116

6.1. Biosynthesis of the Aglycone: Modular


Polyketide Synthases
The aglycones of macrolides are complex polyketides
that are assembled through successive decarboxylative condensations of small carboxyacyl thioesters
(e.g., malonyl CoA, methylmalonyl CoA) in a manner
resembling fatty acid biosynthesis. Each step of the
synthesis is programmed to determine the acyl unit
incorporated into the growing chain (e.g., malonyl
CoA, methylmalonyl CoA, etc.) and the degree to
which the resulting -carbonyl generated from the
condensation is reduced. In addition, the stereochemistry of the R-side chain (if present) is also programmed. Programming is carried out by the polyketide synthase (PKS) that catalyzes all the steps in
assembly of the aglycone. In general, each enzymatic
step is conducted by a discrete component of the PKS,
and as in fatty acid biosynthesis, all steps take place
with the growing acyl chain tethered to the enzyme
in a thioester linkage. Macrolide PKSs are large,
multifunctional polypeptides that can contain more
than 30 enzymatic functions, but the functions associated with a single condensation and -carbonyl
reduction cycle are present in an uninterrupted linear
sequence, commonly referred to as a module, hence
the term modular PKS. Each module is similar in
overall organization to type I fatty acid synthases.
The enzymatic functions within each module are
called domains. The domains are arranged in a linear
sequence and separated by interdomain spacer regions. The KS domain, approximately 550 amino
acids in length, encodes the -ketoacyl ACP synthase
that catalyzes the condensation between the growing
acyl chain (attached in thioester linkage to the Cys173 residue of the KS) and the extender unit tethered
to the ACP domain (acyl carrier protein) through a

Katz and Ashley

thioester linkage with the 4-phosphopantotheine


prosthetic group.117 The AT domain, ca. 300 amino
acids, encodes the acyltransferase, the component
that binds the extender acyl-CoA unit via an ester
linkage with the Ser residue in the GHSxG active
site, and transfers it to the ACP for condensation
with the nascent acyl chain. Each AT domain is
selective for the extender unit it binds and transfers
to its cognate ACP. Comparisons of the sequences of
AT domains showed that malonyl- and methylmalonyl-transferring domains each clustered with members of the same group strongly, indicating structuredetermined selectivity.118
All modules in macrolide PKSs contain KS, AT, and
ACP domains. The remaining domains determine the
extent to which the -carbonyl produced through
condensation is reduced. If the KR domain (-ketoreductase) is absent or mutated, the -keto group will
not be processed further. If the KR is present, the
-keto group is reduced to the hydroxyl. The stereochemistry of the hydroxyl group is determined by
the KR domain.119,120 The KR domains have the
GxGxxAxxxA motif for NADPH binding.121 The DH
(dehydratase) domain removes the -OH group and
a proton from the R-carbon to leave an R,-double
bond. It is not known if the DH domains remove 3(R)OH and pro-2(S) hydrogen in syn eliminations as
observed in fatty acid synthase.122 All double bonds
found in macrolides are trans. The ER (enoylreductase) domain reduces a trans double bond to the
-methylene center. ER domains contain a NADPH
binding motif.
All macrolide PKSs contain a TE (thioesterase)
domain at the C terminus of the last module that acts
to release the polyketide chain from the PKS and
cyclize it. These are referred to as TE-I domains. The
TE-I domains of the erythromycin and pikromycin
PKSs have been crystallized.123,124 Macrolide biosynthesis clusters also contain a discrete gene encoding
a short-chain thioesterase (TE-II) that play a role in
macrolide production by removing aberrant intermediates produced from improper decarboxylation of the
extender molecule.125-127

6.1.1. Erythromycin
The erythromycin PKS, 6-dEB synthase, or DEBS,
was the first modular PKS identified through sequencing of the corresponding genes.128,129 DEBS
consists of three proteins though each is thought to
exist as a head-to-head dimer in the holoenzyme.130
6-dEB is made from the successive condensations of
one propionate molecule and six molecules of methylmalonate.
The predicted domain organization of DEBS and
biosynthetic intermediates at the end of each cycle
of condensation and -carbonyl reduction is shown
in Figure 5. DEBS1 contains the loading module and
modules 1 and 2. The AT domain of the loading
domain binds propionyl CoA and transfers it to the
adjacent ACP [a]. All ACP domains of DEBS are
phosphopantetheinylated by the phosphopantetheinyltransferase SePptII, whose gene is not found in
the erythromycin biosynthesis cluster.131 The propionyl residue is then transferred to the KS domain of

Translation and Protein Synthesis

Chemical Reviews, 2005, Vol. 105, No. 2 515

Figure 5. Domain organization of DEBS and structures of proposed intermediates at the end of each condensation cycle.
Linear sequences of polypetides are shown as open arrows. Domains are shown as spheres. Color-coding indicates
components of the nascent polyketide chain programmed by particular modules. Abbreviations: ACP, acyl carrier protein;
AT, acyltransferase; DH, dehydratase; ER, enoylreducase; KR, -ketoreductase; KS, -keto acyl-CoA synthase; TE,
thioesterase.

module 1. It has been shown that propionyl CoA, and


not methylmalonyl CoA, binds directly to the loading
domain and that propionyl CoA can bind directly to
the KS1 domain in the absence of a loading module,
albeit very inefficiently.132-134 All AT domains of
modules 1-6 bind and transfer the 2(S) enantiomer
of methylmalonyl CoA to their cognate ACPs; hence,
epimerization of the R-methyl group produced after
the second, fifth, and sixth condensations must take
place, but it is not yet known how these epimerizations are controlled by the PKS.135 After the first
condensation, reduction of the -carbonyl is catalyzed
by the KR domain of module 1 to generate the
diketide intermediate b. As seen in Figure 5 the
carbon atoms of the propionyl starter and first
extender ultimately become C 11-15 of the completed aglycone. The acyl chain of b is transferred to
the KS of module 2, and condensation with the
methylmalonyl CoA extender on ACP2 generates a
triketide whose -carbonyl is reduced by the KR2
domain [c]. Direct evidence for the activities of
associated with DEBS 1 comes from the production
of the predicted triketide lactone both in vivo and in
vitro from a DEBS construct in which the TE domain
was moved from the end of module 6 to the C
terminus of DEBS 1.136-139
The next step requires interpolypeptide transfer of
the nacent acyl chain from the ACP2 of DEBS1 to
KS3 of DEBS2. Recognition sequences (linker regions) at the ends and beginnings of PKS subunits
ensure proper associations to prevent aberrant nascent chain passage.140,141 Module 3 contains a sequence that resembles a KR domain, but the conserved NADP(H) binding site is not present and,
hence, is not functional. The -carbonyl of the formed

tetraketide [d] is not further processed and becomes


the C-9 keto group in 6-dEB. After the fourth
condensation the KR, DH, and ER domains process
the -carbonyl to the methylene [e] found at C7 in
6-dEB. After the fifth and sixth condensations only
ketoreductions are programmed to take place to
produce the OH groups at C-5 and C-3 of 6-dEB.
After reduction of the -carbonyl of the heptaketide,
the TE domain acts to release f from the PKS and
promotes the nucleophilic attack of the C-13 hydroxyl
on the C-1 carbanion formed, resulting in the production of the macrolactone. How the PKS is programmed to avoid premature release of the chain
prior to the last -ketoreduction is not yet understood.
The genes that determine DEBS have been expressed in a number of heterologous hosts, including
Streptomyces coelicolor, Streptomyces lividans, and
E. coli.142-145 The DEBS proteins have been purified
and used to make 6-dEB, intermediates, or derivatives in vitro.138,139,146,147 The specificities of the various KS domains of DEBS have been examined using
N-acetylcysteamine thioesters of the syn or anti
diastereomers of 2-methyl-3-hydroxyl-containing acyl
chains for direct loading onto the KS2, KS5, or KS6
domains for single or multiple chain extensions in
vivo or in vitro.117,148-151 It was found that all three
domains utilized only the syn diastereomers and that
whereas KS2 and KS5 could use either enantiomer
KS6 showed high preference for the (2S,3R) enantiomer. It should be pointed out that KS5 normally
does not utilize a 2-methyl-3-hydroxy-containing
substrate for elongation; its substrate is fully reduced
at C3.

516 Chemical Reviews, 2005, Vol. 105, No. 2

Katz and Ashley

Figure 6. Domain organization of the Pik PKS, and structures of proposed intermediates at the end of the condensation
cycle. Polypeptides and domains as in Figure 5. Abbreviations: KSQ, KS domain carrying Cys173Ala mutation; all others
as in Figure 5.

6.1.2. Lankamycin and Oleandomycin


The lankamycin PKS is identical in both module
and domain organization the DEBS but differs in
amino acid sequence.152 The only structural difference
between the untailored aglycones is the replacement
of the 13-ethyl side chain in 6-dEB with the 1(S)methyl-2(S)-hydroxypropyl group in the aglycone of
lankamycin. The discovery that the lankamycin PKS
contained only six modules (and a loading module)
suggests that the starter is either 2-methylbutyryl
CoA (which is hydroxylated at C-2 after polyketide
synthesis) or 2-methyl-3-hydroxybutyryl CoA.
The aglycone of oleandomycin is built from an
acetate starter and six molecules of methylmalonyl
CoA. The Ole PKS is organized identically to that
seen for DEBS with a single difference, a KSQ domain
in the loading module, which is discussed immediately below.

6.1.3. Methymycin and Pikromycin


The aglycones of methymycin and pikromycin differ
in structure only with respect to the additional two
carbons in the ring of pikromycin. Methymycin is
produced from one propionate, one malonate, and
four methylmalonate residues. Pikromycin requires
an additional methylmalonate. Both compounds are
produced in S. venezuelae from a single PKS (Figure
6); thus, the nascent chain to the end of the fifth
module is the same for both compounds.153 10Deoxymethynolide is released after the fifth condensation and narbonolide after the sixth. The Pik PKS
is similar to DEBS in overall organization, with a
number of interesting differences. At the N-terminus

of the loading module is a domain labeled KSQ in


which the Cys173 residue at the active site is
replaced by Gln.154 This domain, therefore, cannot
make a thioester linkage with an acyl chain and
hence cannot participate in a condensation reaction.
The domain is still capable of the decarboxylation
event that is required for chain elongation. Hence,
loading modules that carry KSQ domains use starters
that require decarboxylation such as malonyl CoA or
methylmalonyl CoA to yield the required acetyl or
propionyl moieties found in the side chains of the
completed aglycones.155,156 Reduction of the resulting
carbanion is likely conducted by the KSQ as well. The
Pik PKS, therefore, uses methylmalonyl CoA as the
starter and decarboxylates it to propionyl-ACP. The
Ole PKS uses malonyl CoA as the starter and
decarboxylates it to acetyl-CoA. The first, third, and
fourth condensations and -carbonyl-processing events
resemble those seen for 6-dEB. The second condensation employs malonyl CoA rather than methylmalonyl CoA as the extender unit, and the -carbonyl of
the triketide is reduced and then dehydrated by the
KR and DH domains in module 2. The 2,3-double
bond of c thus becomes the 8,9- or 10,11-trans double
bond of methymycin or pikromycin, respectively. The
most interesting differences from DEBS are the
events that take place after the fifth condensation.
Modules 5 and 6 in the Pik PKS are split into
separate polypeptides, PikAIII and PikAIV, respectively. Under conditions that favor the production of
methymycin, nascent chain growth terminates after
the fifth condensation event to release and cyclizes
the acyl chain to produce 10-deoxymethynolide. It has

Translation and Protein Synthesis

Chemical Reviews, 2005, Vol. 105, No. 2 517

Figure 7. Domain organization of the Tyl PKS, and structures of proposed intermediates at the end of the condensation
cycle. Polypeptides, domains, and abbreviations as in Figure 5.

been proposed that this is accomplished by the


transfer of intermediate f to the ACP of module 6
without chain elongation, a process referred to as
skipping.157-159 Once attached to ACP6, the adjacent TE domain can release and cyclize the acyl
chain. Under conditions that favor 14-membered ring
production, normal transfer of intermediate f to the
KS of module 6 would take place. How the organism
regulates the production of one macrolide over another is still not fully understood. The pik PKS genes
have been expressed, in whole or in part, in heterologous hosts.144,160 Employing various 2-methyl-3hydroxypentanoyl-S-NACs, the specificities of KS5
and KS6 of the Pik PKS were shown to be similar to
those found for the corresponding DEBS KS domains,
although KS5 was found to have high preference for
the syn (2S,3R)-enantiomer.161,162

6.1.4. Tylosin
The PKS-encoding genes from at least one member
of each of the four groups of 16-membered macrolides
have been sequenced. All contain seven modules and
are organized as shown in Figure 7 for the tylosin
PKS.163 The PKS is composed of five polypeptides:
polypeptide Isload, modules 1 and 2; IIsmodule 3;
IIIsmodules 4 and 5; IVsmodule 6; Vsmodule 7.
The aglycone tylactone is made from the precursors
malonyl CoA, methylmalonyl CoA, and ethylmalonyl
CoA. The presence of the KSQ domain in the loading
module suggests that the starter is methylmalonyl
CoA, which is decarboxylated to propionyl-S-ACP.
The first, second, fourth, and sixth condensations
employ methylmalonate extenders; the third and
seventh use malonyl CoA. The fourth extension uses
ethylmalonyl CoA, which is produced in the cell
through the 2-carboxylation of butyryl CoA. Butyryl
CoA may be produced from the degradation of fatty

acids or through a single round of fatty acid synthesis


from acetyl-CoA. A gene for crotonyl CoA reductase,
which catalyzes conversion of crotonyl CoA to butyryl
CoA, is present in the tylosin biosynthesis cluster.164
The specificities of the KS domains of the Tyl PKS
have not been examined; thus, it remains to be
determined whether the KS2 domain, which is normally presented with the anti-2-methyl-3-hydroxypentanoyl-S-ACP, has preference for one enantiomer
over the other or whether the syn diastereomer can
also be extended.

6.1.5. Platenolide
The predicted domain organization and biosynthetic intermediates of platenolide synthase, which
has been sequenced from the spiramycin and niddamycin producers, is shown in Figure 8.165,166 The
domains are identical to that of the tylosin PKS with
two exceptions: the ATs of the loading module and
module 2 transfer malonyl CoA rather than methylmalonyl CoA; the AT of module 6 transfers methoxylmalonate-thioester rather than methylmalonyl CoA.
In the platenolide cases it is not known whether the
thioester moiety of methoxymalonate is CoA, but it
is thought that methoxymalonyl-ACP is the precursor employed for biosynthesis of the complex polyketides ansimitocin and ascomycin.167,168

6.1.6. Chalcomycin
The PKS of chalcomycin is shown in Figure 9.
Although chalcomycin contains a 2,3-trans double
bond, the Chm PKS does not contain the required
KR and DH domains in module 7 to catalyze its
formation.169 A gene that could encode a ketoreductase was identified 3 kb downstream of the PKS, but
a DH gene was not found. Expression of the chm PKS

518 Chemical Reviews, 2005, Vol. 105, No. 2

Katz and Ashley

Figure 8. Domain organization of the platenolide PKS, and structures of proposed intermediates at the end of the
condensation cycle. Polypeptides, domains, and abbreviations as in Figure 5.

Figure 9. Domain organization of the Chm PKS, and structures of proposed intermediates at the end of the condensation
cycle. Polypeptides, domains, and abbreviations as in Figure 5.

genes in an S. fradiae host that had been deleted of


the tyl PKS genes resulted in the production of the
predicted macrolactone containing a 3-keto group
(chalconolide) but which contained mycaminose at
C-5, indicating that the mycaminosyltransferase used
for tylosin production could utilize chalconolide as
well.169,170 The basis for the introduction of the 2,3double bond in chalcomycin is not yet understood. In
contrast, the seventh module of the mycinamicin PKS
contains the KR and DH domains, which indicates

formation of the double bond on the nascent polyketide.152

6.2. Biosynthesis of Deoxysugars


Genes for the biosynthesis of the deoxysugar
components of macrolides have been identified in the
erythromycin, pikromycin, tylosin, megalomicin,
chalcomycin, oleandomycin, and lankamycin
clusters.152,153,164,169,171-178 Verification of the pathways
have come from (a) transfer of the genes to a

Translation and Protein Synthesis

Chemical Reviews, 2005, Vol. 105, No. 2 519

Figure 10. Composite biochemical pathways of deoxysugar biosynthesis in macrolide-producing strains. Proposed enzymes
for given steps are shown.

heterologous host and production of a macrolide


containing the corresponding deoxysugar, in some
cases a novel macrolactone-sugar combination, or (b)
loss of synthesis of the sugar component or change
of the structure of the sugar through the introduction
of a mutation in the corresponding genes.172,174,179-183
A compilation of the proposed pathways of seven
deoxysugars present in macrolides is shown in Figure
10 along with the proposed genes involved in the
particular steps from the corresponding antibioticproducing strains. The proposed pathway for the
synthesis of L-arcanose, the neutral sugar of lankamycin, is not shown. Genes from different organisms
involved in a particular step of a pathway, e.g.,
eryBIV, tylM, show highest similarity scores to each
other of all matches in the sequence databases and
are assigned the given step on the basis of proposed
function. Because they have not been determined
experimentally, the absolute order of reactions for
pathways involving more than two steps are not
certain. The nucleotide carrier thymidine diphosphate has been identified only for the deoxysugars
of tylosin, erythromycin, and oleandomycin; hence,

Figure 10 shows the generic NDP as the carrier. All


deoxysugars are made from the common intermediate 4-keto-6-deoxyglucose, which is itself made in two
steps from glucose-1-phosphate. Genes believed to
determine the enzymes for these steps have been
found in all of the macrolide biosynthesis gene
clusters examined except erythromycin, which uses
the enzymes involved in the synthesis of the deoxysugars of the cell wall.184

6.3. Post-Polyketide Modification


Following their synthesis, the aglycones are modified through glycosylation, oxidation, reduction, and
acylation. The deoxysugars also may be modified.
Each macrolide has an order sequence of reactions
to assemble the final compound, but it is often the
case that various steps may be substituted or bypassed.

6.3.1. Erythromycin and Megalomicin


Pathways for the formation of erythromycin and
megalomicin from the aglycone 6-dEB in S. erythraea

520 Chemical Reviews, 2005, Vol. 105, No. 2

Katz and Ashley

Figure 11. Biochemical pathways of erythromycin and megalomicin biosyntheses. Proposed enzymes for given steps are
shown.

and Micromonospora megalomicea, respectively, are


shown in Figure 11.177 The erythromycin pathway
was determined from the identification of compounds
produced in mutants blocked in different steps of the
pathway.185 The aglycone is hydroxylated at C-6 by
the product of the eryF or megF gene to produce
erythronolide B (EB), which is then glycosylated at
the C-3 OH with NDP-L-mycarose to produce 3-Rmycarosyl EB (MEB) by the mycarosyltransferases
EryBV or MegBV. MEB is glycosylated at the C-5
OH with NDP-desosamine by the desosaminyltransferases EryCIII or MegCIII to yield erythromycin
D.186 In S. erythraea the 6-hydroxylation step can be
bypassed in strains defective in EryF and 6-deoxyerythromycin is formed.187 Hydroxylation of erythromycin D at C12 by EryK or MegK produces
erythromycin C, the last common intermediate in the
pathways of erythromycin A and megolamicin. In S.
erythraea the 3-OH of the mycarosyl residue is
methylated by EryG, converting the residue to Lcladinose and the compound to erythromycin A.
EryG-mediated methylation of erythromycin D can
also take place to produce the side product erythromycin, but this compound is only poorly converted
to erythromycin A by EryK.188-190 In M. megalomicea
erythromycin C is glycosylated at the C-6 OH with
NDP-megosamine by MegDI to produce megalomicin
A. The 3- and 4-OH groups of the mycarose
residue can be acylated with acetate or propionate
in various combinations to produce megalomicins B,
C1, and C2. Acylations are thought to be catalyzed

by MegY.177 The cytochrome P450 hydroxylase EryF


has been crystallized.191,192

6.3.2. Methymycin and Pikromycin


The pathways from 10-deoxymethynolide and narbonolide to methymycin and pikromycin, respectively,
are shown in Figure 12. Each is converted to its final
product in two steps: glycosylation at C-5 or C-3 with
desosamine catalyzed by DesII followed by hydroxylation by PikC (also called PicK) to produce the final
compound.174,193-195 It should be noted that PikC
utilizes both YC-17 and narbomycin, two different
size macrolides as substrates, and produces two
different products from YC-17.196

6.3.3. Tylosin
The pathway for the formation of tylosin is shown
in Figure 13 and has been formulated from identification of the compounds produced in mutants blocked
at various steps.113 Unlike erythromycin, glycosylation at C-5 precedes the first oxidation step that
produces the C20 aldehyde. This is followed by a
second oxidation to add the hydroxyl at C-23 for
subsequent glycosylation. The next step is the addition of D-allose to the 23-OH to produce the diglycoside, followed by addition of L-mycarose to the
mycaminose moiety. Glycosylation of OMT by D-allose
can be bypassed in tylD, tylJ, or tylN mutants to
produce the compound desmycinosyltylosin (DMT),

Translation and Protein Synthesis

Chemical Reviews, 2005, Vol. 105, No. 2 521

Figure 12. Biochemical pathways of pikromycin and methymycin biosyntheses. Proposed enzymes for given steps are
shown.

Figure 13. Biochemical pathways of tylosin biosynthesis. Proposed enzymes for given steps are shown.

OMT containing the mycarose residue, as the end


product of the pathway.197,198

6.3.4. Other Macrolides


The biosynthesis of oleandomycin [10] follows a
pathway similar to that described for erythromycin.199-201 Three subsequent post-polyketide modi-

fications take place after biosynthesis of the aglycone


in the following order: oxidation mediated by the
P450-enzyme OleP to produce the aglycone containing the 8,8a-epoxide, attachment of the neutral sugar
L-oleandose at C-3, attachment of desosamine to C-5.
Because the biosynthetic products have not been
identified, the pathway to lankamycin is less clear.
The pathway from the aglycone requires hydroxylations at C8 and C12, glycosylations employing an

522 Chemical Reviews, 2005, Vol. 105, No. 2

NDP-neutral sugar at C3 and NDP-chalcose at C5,


and acetylation of the 11- and 4- hydroxyls. The
order of these reactions has not been established. It
is not known if the neutral sugar that is attached at
C-3 is L-arcanose or L-olivose, which is converted to
L-arcanose through 4-O-methylation after the sugar(s) is attached to the aglycone, as in the case of
erythromycin. Furthermore, as described in section
6.1.2, if the starter for the synthesis of the aglycone
is 2-methylbutyryl CoA, hydroxylation of C15 would
also be required to complete the synthesis of lankamycin.
The full complement of genes for the biosynthesis
of any platenolide-based macrolide has not been
reported; thus, little is known about the pathways
of synthesis of these compounds beyond the point of
the aglycones. In cases where reduction of the C-9
carbonyl takes place, a post-polyketide reductase is
thought to be involved. Complete sets of genes for
the biosynthesis of chalcomycin and mycinamicin
have been reported.169,202 For chalcomycin, the order
of reactions (8-hydroxylation, 12,13-epoxidation, and
glycosylations at C-20, following C-20 hydroxylation,
and at C-5) has not been established. The order of
reactions from the aglycone of mycinamicin to final
products is not known, but it has been established
that a cytochrome P450 enzyme catalyzes both the
12,13-epoxidation and 14-hydroxylation steps as the
final steps in the synthesis of mycinamicins.203

6.4. Regulation of Macrolide Biosynthesis


Though macrolides are considered to be secondary
metabolites, little is known of how their biosynthesis
is controlled to initiate toward the end of the logarithmic phase of growth and to stay on through the
stationary phase. Specific regulatory genes that
regulate expression of the PKS and other macrolide
biosynthesis genes have been identified and studied
in the pikromycin and tylosin biosynthesis clusters.
The gene pikD, present in the pik cluster, encodes a
DNA binding protein that is required for the expression of the pik PKS and desosamine genes but not
for expression of pikC.204 Inactivation of PikD leads
to loss of pikromycin and methymycin production.
Tylosin biosynthesis appears to be regulated in a
cascade fashion.205 The gene tylP appears to encode
a repressor that represses expression of tyQ, a
transcriptional activator.206 Repression is relieved by
the presence of a yet to be identified -butyrolactone,
similar to the A factor that regulates production of
streptomycin in Streptomyces griseus.207 TylQ is a
transcriptional repressor of tylR, global regulator
required for tylosin biosynthesis, and a transcriptional activator of tylS, which encodes a tylosin
pathway specific activator and is classified as a
member of the SARP (Streptomyces antibiotic regulatory proteins) family.208,209 TylS also appears to
regulate tylR.210 In addition, it has been found that
intermediates in the pathway beyond tylactone which
contain the deoxysugar mycaminose stimulate production of tylactone, but the mechanism of this
regulation is not yet understood.211

Katz and Ashley

7. New Macrolides and Ketolides


7.1. Chemistry
Erythromycin derivatives wherein the 3-O-cladinosyl moiety has been replaced with an acyl functionality, termed acylides, have been reported.212 Of
particular interest is the 3-O-(4-nitrophenyl)acetyl
derivative of clarithromycin, TEA0777, which shows
potent activity not only against macrolide-susceptible
and efflux-resistant S. pneumoniae, but also against
inducible-MLSB-resistant S. aureus as well. Recent
efforts have led to TEA0929 [32], which shows good
in-vitro activity against macrolide-susceptible and
MLSB-inducible S. aureus and S. pneumoniae and
against H. influenzae and also shows in-vivo activity
equivalent to clarithromycin.213 Ketolides bridged
across the 6-O and 11-O positions, such as EP-13417
[33], have been found to possess high in-vitro and invivo activity against typical respiratory pathogens.214

7.2. Genetic Engineering


Following the discovery of modular macrolide
PKSs, efforts commenced to alter the specificities and
activities of the domains for the purpose of changing
the structure of the corresponding aglycone. This was
enabled by the development of genetic tools for
streptomycetes that permitted DNA to be introduced
into the macrolide producers and recombination to
be selected. Hence, to create desired changes to the
structure of aglycones, the following has been accomplished: to reduce the extent of -carbonyl reduction, KR, DH, or ER domains have been inactivated
through mutation (or deletion); to increase the extent
of reduction, these domains have been introduced
into modules where not present originally; to change
the extender unit incorporated into the nascent
polyketide chain, AT domains have been exchanged.
These exchanges have been performed in the macrolide producers or in hosts into which the PKS genes
were introduced, such as E. coli or S. coelicolor.
Replacement of DEBS AT1 or AT2, AT3, AT5, and
AT6 with a malonyl-transferring AT domain in S.
erythraea or in strains of S. coelicolor or S. lividans
that carried the DEBS genes resulted in the production of the expected erythromycin analogues: AT112-desmethylerythromycin B [34]; AT2-10-desmethylerythromycin A [35] and 10-desmethylerythromycin B, AT3-8-desmethylerythromycin A
[36]; AT5-4-desmethylerythromycin A [37]; AT62-desmethylerythromycin A [38].170,215-217 In the AT1
exchange 12-desmethylerythromycin A was not produced, indicating that the EryK hydroxylase could
not use 12-desmethylerythromycin D as a substrate,
but the EryG methyltransferase utilized the intermediate to some extent (see Figure 11). In the other
cases the A congener was found but was not the most
predominant form of the product. In addition, 38 was
detected after an uncharacterized (and unrecovered)
segment of DNA from the oleandomycin producer was
introduced into a strain of S. erythraea that carried
a mutation in the DEBS PKS.218 Although the basis
of the production of 38 has not been determined,

Translation and Protein Synthesis

sequencing of the host used for introduction of the


DNA revealed an in-frame deletion of AT6 (Katz, L.
et al. Unpublished results). It is likely that the
incoming DNA carried a malonyl-transferring AT
domain that acted trans to provide malonyl-AT
function to module 6. trans AT domains have recently
been discovered in nonmacrolide modular PKSs.219,220
Replacement of the AT4 domain of DEBS with an
ethylmalonyl-transferring domain resulted in production of 6-desmethyl-6-ethylerythromycin A [39],
but the host required the addition of a ccr gene
encoding crotonyl CoA reductase.221 In the absence
of ccr, the host containing the exchanged AT domain
produced a small amount of erythromycin. Replacement of AT4 in DEBS in S. erythraea with a malonyltransferring AT was done at Biotica Technology, Ltd.,
and Kosan Biosciences, Inc., with different results.
Using the soil isolate NRRL 2338 and a malonyltransferring AT from the rapamycin PKS, the Biotica
group found that the engineered strain produced
6-desmethylerythromycin D [40], indicating that both
EryK and EryG could not utilize 40 as substrate.222
The Kosan approach employed introducing two sitespecific mutations into DEBS AT4 to alter the
specificity of the domain. In the S. lividans host
carrying the altered DEBS, the expected 6-desmethylerythronolide B was produced.223 When the same
mutations were introduced into an industrially optimized S. erythraea host, the strain produced a small
amount of 6-desmethyl-6-deoxy-7-hydroxyerythromycin D [41].224 The production of a D congener in
the Kosan strain coincides with the findings at
Biotica. The finding of a 7-OH group in 40 is difficult
to explain. Either the EryF hydroxylase had changed
its specificity to hydroxylate the substrate at C7
rather than C6 in the Kosan host or the host contains
an adventitious enzyme that hydroxylates 6-dEB at
C7 to produce a compound that cannot be hydroxylated by EryF.
Exchanges of the loading module have also been
reported. Exchange of the loading AT domain of
DEBS with the loading AT domain from the avermectin PKS in S. erythraea resulted in the production
of 14-methylerythromycin A [42] and 14-ethylerythromycin A [43] along with their B and D congeners.225
Changes were also introduced into the reductive
domains of DEBS to produce novel compounds. Two
examples of such changes in S. erythraea that produced fully elaborated molecules include the inactivation of the ER4 domain to produce 6,7-anhydroerythromycin C [44] and the replacement of the KR2
domain with a DH4/ER4/KR4 domain from the rapamycin PKS to produce 11-deoxyerythromycin
[45].170,226 Multiple changes in DEBS have been done
employing DEBS genes that had been engineered to
contain unique restriction sites at the edges of the
various domains.217 These compounds were produced
in S. lividans that carried the modified DEBS genes;
hence, the compounds were not elaborated beyond
the aglycone.
Hybrid PKSs carrying at least one module of two
different PKSs have also been made. The loading
domain of the spiramycin PKS (Figure 8) was re-

Chemical Reviews, 2005, Vol. 105, No. 2 523

placed with the loading domain of the Tyl PKS in a


Streptomyces ambofaciens host that carried a deletion
of the spiramycin sugar biosynthesis genes. The
resulting compound was the expected 15-methylplatenolide.227 DEBS-Pik, DEBS-Ole, and Tyl-Pik
PKS hybrids yielding predicted compounds have also
been reported.144,228
Of the dozen or so fully elaborated novel macrolides
produced by PKS genetic engineering, most retained
some measure of bioactivity but none showed enhanced potencies over their parent compounds. The
only example of an engineered compound that showed
improved properties was 6-deoxyerythromycin [46],
produced by targeted disruption of the eryF gene in
S. erythraea.187 The compound was less potent than
erythromycin in vitro but showed improved in-vivo
activity in experimental infections due to enhanced
acid stability.187
The most promising new molecules originate from
a combination of genetics, chemistry, and fermentation development. Jacobsen et al. demonstrated that
an S. coelicolor strain carrying DEBS that contained
a C173A replacement (KS1null) could be fed SNAC
diketides in which the C5 methyl group could be
replaced with a number of substitutions (Figure 14:
[47]) including H atoms and phenyl rings to produce
6-dEB analogues that contained the corresponding
substitutions at C13 [48].117,148,149,229 These novel
aglycones could be converted into erythromycin analogues [49] after purification and feeding to an S.
erythraea strain carrying a mutation in the one of
the DEBS genes (e.g., KS1 null host). This technology
was employed by Kosan in collaboration with J & J
Pharmaceutical Research Institute to produce a
number of novel 6-O-arylalkyl ketolides [50-55].
Preliminary studies reported that a number of these
compounds displayed in-vitro and in-vivo activities
comparable to telithromycin or cethromycin.230,231

8. Conclusions
The advancements in the isolation and crystallization of ribosomes have allowed a fuller understanding
of how macrolides and ketolides exert their antibiotic
effects. Whereas it was formerly thought that these
compounds block a specific event during the initiation
or elongation cycle of protein synthesis, it is currently
believed that their binding in the exit tunnel is
sufficient to prevent elongation of the nascent polypeptide chain. It is not yet known if the efficacy of a
compound is directly related to its strength of binding. The ketolides, which bind to domains V and II
of 23S rRNA and so may bind more tightly to
ribosomes, may become preferred as antibiotics over
macrolides, which only bind in domain V. The current
limitations of telithromycin, the only currently approved ketolide, is its modest activity against H.
influenzae, prompting the need for administration of
800 mg/day and the lack of efficacy against MLSBresistant S. pyogenes and constitutive MLSB-resistant
S. aureus. Ribosome binding studies have shed light
on the basis of macrolide resistance, but they do not
as yet enable an understanding of why, in cases of

524 Chemical Reviews, 2005, Vol. 105, No. 2

Katz and Ashley

Figure 14. Schemes showing production of 15-R erythromycin analogues. (A) Pathway to produce 15-R erythromycin. 47
is fed to S. coelicolor DEBS (KS1null) to produce 48, which is fed to S. erythraea KS1null to produce 49. (B) Production of
15-methyl ketolides. 50 is produced using scheme A and converted to 51-55 as described in the text.

macrolide-resistant strains, the ketolides are effective


as antibiotics against some but not effective against
others. The contribution of the effect on 50S ribosome
assembly by macrolides to the overall bacteriostatic
or bactericidal activities of these molecules also
requires further clarification. The ability to manipulate PKSs provided great promise initially that novel
macrolides could be made, including ones that could
not be obtained by conventional chemical modification, and which would contain enhanced properties.
Other than the small number of compounds made by
the feeding of short-chain thioesters to a genetically
engineered host, as only a first step in a three-part
process, the few fully elaborated novel macrolides
produced by genetic engineering have not yet fulfilled
the original promise. It is still too early to tell
whether this avenue of discovery will prove effective.
The findings that many of engineered PKSs either
do not produce the expected compounds or do so at
levels too low to be useful indicate that greater
understanding of the biochemical details of polyketide
biosynthesis is required before full exploitation of
their chemical potential can be realized.

9. Acknowledgments
We thank Scott Blanchard for generating figures
of the ribosome showing bound macrolide. We are
grateful to our former and present colleagues for their
dedication and effort over the many years that we
have engaged in this research.

10. References
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)

Woodward, R. B. Angew. Chem. 1957, 57, 50.


Woodward, R. B. Angew. Chem. 1957, 69, 585.
Brockmann, H.; Henckel, W. Chem. Ber. 1951, 84, 284.
McGuire, J. M.; Bunch, R. L.; Anderson, R. C.; Boaz, H. E.;
Flynn, E. H.; Powell, H. M.; Smith, J. W. Antibiot. Chemother.
1952, 2, 281.
Bonay, P.; Munro, S.; Fresno, M.; Alarcon, B. J. Biol. Chem.
1996, 271, 3719.
Maezawa, I.; Kinumaki, A.; Suzuki, M. J. Antibiot. (Tokyo) 1974,
27, 84.
Huang, S. L.; Hassell, T. C.; Yeh, W. K. J. Biol. Chem. 1993,
268, 18987.
Itoh, Z.; Nakaya, M.; Suzuki, T.; Arai, H.; Wakabayashi, K. Am.
J. Physiol. 1984, 247, G688.
Erah, P. O.; Goddard, A. F.; Barrett, D. A.; Shaw, P. N.; Spiller,
R. C. J. Antimicrob. Chemother. 1997, 39, 5.
Gill, C. J.; Abruzzo, G. K.; Flattery, A. M.; Smith, J. G.; Jackson,
J.; Kong, L.; Wilkening, R.; Shankaran, K.; Kropp, H.; Bartizal,
K. J. Antibiot. (Tokyo) 1995, 48, 1141.

Translation and Protein Synthesis


(11) Mordi, M. N.; Pelta, M. D.; Boote, V.; Morris, G. A.; Barber, J.
J. Med. Chem. 2000, 43, 467.
(12) Rodvold, K. A. Clin. Pharmacokinet. 1999, 37, 385.
(13) Nakagawa, Y.; Itai, S.; Yoshida, T.; Nagai, T. Chem. Pharm. Bull.
(Tokyo) 1992, 40, 725.
(14) Doucet-Populaire, F.; Capobianco, J. O.; Zakula, D.; Jarlier, V.;
Goldman, R. C. J. Antimicrob. Chemother. 1998, 41, 179.
(15) Barry, A. L.; Fuchs, P. C.; Brown, S. D. Eur. J. Clin. Microbiol.
Infect. Dis. 2001, 20, 494.
(16) Barry, A. L.; Jones, R. N.; Thornsberry, C. Antimicrob. Agents
Chemother. 1988, 32, 752.
(17) Sahm, D. F.; Karlowsky, J. A.; Kelly, L. J.; Critchley, I. A.; Jones,
M. E.; Thornsberry, C.; Mauriz, Y.; Kahn, J. Antimicrob. Agents
Chemother. 2001, 45, 1037.
(18) Denis, A.; Agouridas, C. Bioorg. Med. Chem. Lett. 1998, 8, 2427.
(19) Agouridas, C.; Denis, A.; Auger, J. M.; Benedetti, Y.; Bonnefoy,
A.; Bretin, F.; Chantot, J. F.; Dussarat, A.; Fromentin, C.;
DAmbrieres, S. G.; Lachaud, S.; Laurin, P.; Le Martret, O.;
Loyau, V.; Tessot, N. J. Med. Chem. 1998, 41, 4080.
(20) Zhanel, G. G.; Walters, M.; Noreddin, A.; Vercaigne, L. M.;
Wierzbowski, A.; Embil, J. M.; Gin, A. S.; Douthwaite, S.; Hoban,
D. J. Drugs 2002, 62, 1771.
(21) Denis, A.; Agouridas, C.; Auger, J. M.; Benedetti, Y.; Bonnefoy,
A.; Bretin, F.; Chantot, J. F.; Dussarat, A.; Fromentin, C.;
DAmbrieres, S. G.; Lachaud, S.; Laurin, P.; Le Martret, O.;
Loyau, V.; Tessot, N.; Pejac, J. M.; Perron, S. Bioorg. Med. Chem.
Lett. 1999, 9, 3075.
(22) Bonnefoy, A.; Guitton, M.; Delachaume, C.; Le Priol, P.; Girard,
A. M. Antimicrob. Agents Chemother. 2001, 45, 1688.
(23) Bonnefoy, A.; Le Priol, P. J. Antimicrob. Chemother. 2001, 47,
471.
(24) Ma, Z.; Clark, R. F.; Brazzale, A.; Wang, S.; Rupp, M. J.; Li, L.;
Griesgraber, G.; Zhang, S.; Yong, H.; Phan, L. T.; Nemoto, P.
A.; Chu, D. T.; Plattner, J. J.; Zhang, X.; Zhong, P.; Cao, Z.;
Nilius, A. M.; Shortridge, V. D.; Flamm, R.; Mitten, M.; Meulbroek, J.; Ewing, P.; Alder, J.; Or, Y. S. J. Med. Chem. 2001, 44,
4137.
(25) Yassin, H. M.; Dever, L. L. Expert Opin. Investig. Drugs 2001,
10, 353.
(26) Shortridge, V. D.; Zhong, P.; Cao, Z.; Beyer, J. M.; Almer, L. S.;
Ramer, N. C.; Doktor, S. Z.; Flamm, R. K. Antimicrob. Agents
Chemother. 2002, 46, 783.
(27) Ohtani, H.; Taninaka, C.; Hanada, E.; Kotaki, H.; Sato, H.;
Sawada, Y.; Iga, T. Antimicrob. Agents Chemother. 2000, 44,
2630.
(28) Rubinstein, E. Int. J. Antimicrob. Agents 2001, 18 (Suppl 1), S71.
(29) Samarenda, P.; Kumari, S.; Evans, S. J.; Sacchi, T. J.; Navarro,
V. Pacing Clin. Electrophysiol. 2001, 24, 1572.
(30) Prescrire Int. 2003, 12, 8.
(31) Johnson, A. P. Curr. Opin. Investig. Drugs 2001, 2, 1691.
(32) Menninger, J. R.; Otto, D. P. Antimicrob. Agents Chemother.
1982, 21, 811.
(33) Moazed, D.; Noller, H. F. Biochimie 1987, 69, 879.
(34) Poulsen, S. M.; Kofoed, C.; Vester, B. J. Mol. Biol. 2000, 304,
471.
(35) Hansen, L. H.; Mauvais, P.; Douthwaite, S. Mol. Microbiol. 1999,
31, 623.
(36) Douthwaite, S.; Hansen, L. H.; Mauvais, P. Mol. Microbiol. 2000,
36, 183.
(37) Ban, N.; Nissen, P.; Hansen, J.; Moore, P. B.; Steitz, T. A. Science
2000, 289, 905.
(38) Hansen, J. L.; Ippolito, J. A.; Ban, N.; Nissen, P.; Moore, P. B.;
Steitz, T. A. Mol. Cell 2002, 10, 117.
(39) Schlunzen, F.; Zarivach, R.; Harms, J.; Bashan, A.; Tocilj, A.;
Albrecht, R.; Yonath, A.; Franceschi, F. Nature 2001, 413, 814.
(40) Auerbach, T.; Bashan, A.; Harms, J.; Schluenzen, F.; Zarivach,
R.; Bartels, H.; Agmon, I.; Kessler, M.; Pioletti, M.; Franceschi,
F.; Yonath, A. Curr. Drug Targets Infect. Disord. 2002, 2, 169.
(41) Harms, J. M.; Schlunzen, F.; Fucini, P.; Bartels, H.; Yonath, A.
BMC Biol. 2004, 2, 4.
(42) Berisio, R.; Harms, J.; Schluenzen, F.; Zarivach, R.; Hansen, H.
A.; Fucini, P.; Yonath, A. J. Bacteriol. 2003, 185, 4276.
(43) Amsden, G. W. Clin. Ther. 1996, 18, 56.
(44) Jacobs, M. R.; Bajaksouzian, S.; Zilles, A.; Lin, G.; Pankuch, G.
A.; Appelbaum, P. C. Antimicrob. Agents Chemother. 1999, 43,
1901.
(45) Douthwaite, S. Clin. Microbiol. Infect. 2001, 7 (Suppl 3), 11.
(46) Champney, W. S.; Burdine, R. Curr. Microbiol. 1998, 36, 119.
(47) Champney, W. S. Curr. Top Med. Chem. 2003, 3, 929.
(48) Champney, W. S.; Tober, C. L. Curr. Microbiol. 2001, 42, 203.
(49) Champney, W. S. Curr. Drug Targets Infect. Disord. 2001, 1,
19.
(50) Champney, W. S.; Pelt, J. Curr. Microbiol. 2002, 45, 328.
(51) Champney, W. S.; Chittum, H. S.; Tober, C. L. Curr. Microbiol.
2003, 46, 453.
(52) Haight, T. H.; Finland, M. Proc. Soc. Exp. Biol. Med. 1952, 81,
183.
(53) Weisblum, B. Antimicrob. Agents Chemother. 1995, 39, 797.

Chemical Reviews, 2005, Vol. 105, No. 2 525


(54) Roberts, M. C.; Sutcliffe, J.; Courvalin, P.; Jensen, L. B.; Rood,
J.; Seppala, H. Antimicrob. Agents Chemother. 1999, 43, 2823.
(55) Skinner, R.; Cundliffe, E.; Schmidt, F. J. J. Biol. Chem. 1983,
258, 12702.
(56) Katz, L.; Brown, D.; Boris, K.; Tuan, J. Gene 1987, 55, 319.
(57) Bussiere, D. E.; Muchmore, S. W.; Dealwis, C. G.; Schluckebier,
G.; Nienaber, V. L.; Edalji, R. P.; Walter, K. A.; Ladror, U. S.;
Holzman, T. F.; Abad-Zapatero, C. Biochemistry 1998, 37, 7103.
(58) Schluckebier, G.; Zhong, P.; Stewart, K. D.; Kavanaugh, T. J.;
Abad-Zapatero, C. J. Mol. Biol. 1999, 289, 277.
(59) Yu, L.; Petros, A. M.; Schnuchel, A.; Zhong, P.; Severin, J. M.;
Walter, K.; Holzman, T. F.; Fesik, S. W. Nat. Struct. Biol. 1997,
4, 483.
(60) Webb, V.; Davies, J. Trends Biotechnol. 1994, 12, 74.
(61) Webb, V.; Davies, J. Antimicrob. Agents Chemother. 1993, 37,
2379.
(62) Davies, J. Science 1994, 264, 375.
(63) Weaver, J. R.; Patee, P. A. J. Bacteriol. 1964, 88, 574.
(64) Kamimiya, S.; Weisblum, B. J. Bacteriol. 1988, 170, 1800.
(65) Kamimiya, S.; Weisblum, B. Antimicrob. Agents Chemother.
1997, 41, 530.
(66) Liu, M.; Douthwaite, S. Proc. Natl. Acad. Sci. U.S.A. 2002, 99,
14658.
(67) Choi, S. S.; Kim, S. K.; Oh, T. G.; Choi, E. C. J. Bacteriol. 1997,
179, 2065.
(68) Hahn, J.; Grandi, G.; Gryczan, T. J.; Dubnau, D. Mol. Gen. Genet.
1982, 186, 204.
(69) Capobianco, J. O.; Cao, Z.; Shortridge, V. D.; Ma, Z.; Flamm, R.
K.; Zhong, P. Antimicrob. Agents Chemother. 2000, 44, 1562.
(70) Lampson, B. C.; von David, W.; Parisi, J. T. Antimicrob. Agents
Chemother. 1986, 30, 653.
(71) Jenssen, W. D.; Thakker-Varia, S.; Dubin, D. T.; Weinstein, M.
P. Antimicrob. Agents Chemother. 1987, 31, 883.
(72) Ross, J. I.; Farrell, A. M.; Eady, E. A.; Cove, J. H.; Cunliffe, W.
J. J. Antimicrob. Chemother. 1989, 24, 851.
(73) Ross, J. I.; Eady, E. A.; Cove, J. H.; Cunliffe, W. J.; Baumberg,
S.; Wootton, J. C. Mol. Microbiol. 1990, 4, 1207.
(74) Seppala, H.; Nissinen, A.; Yu, Q.; Huovinen, P. J Antimicrob.
Chemother. 1993, 32, 885.
(75) Sutcliffe, J.; Tait-Kamradt, A.; Wondrack, L. Antimicrob. Agents
Chemother. 1996, 40, 1817.
(76) Matsuoka, M.; Janosi, L.; Endou, K.; Nakajima, Y. FEMS
Microbiol. Lett. 1999, 181, 91.
(77) Versalovic, J.; Osato, M. S.; Spakovsky, K.; Dore, M. P.; Reddy,
R.; Stone, G. G.; Shortridge, D.; Flamm, R. K.; Tanaka, S. K.;
Graham, D. Y. J. Antimicrob. Chemother. 1997, 40, 283.
(78) Versalovic, J.; Shortridge, D.; Kibler, K.; Griffy, M. V.; Beyer,
J.; Flamm, R. K.; Tanaka, S. K.; Graham, D. Y.; Go, M. F.
Antimicrob. Agents Chemother. 1996, 40, 477.
(79) Wang, G.; Taylor, D. E. Antimicrob. Agents Chemother. 1998,
42, 1952.
(80) Prunier, A. L.; Malbruny, B.; Tande, D.; Picard, B.; Leclercq, R.
Antimicrob. Agents Chemother. 2002, 46, 3054.
(81) Tait-Kamradt, A.; Davies, T.; Cronan, M.; Jacobs, M. R.; Appelbaum, P. C.; Sutcliffe, J. Antimicrob. Agents Chemother. 2000,
44, 2118.
(82) Peric, M.; Bozdogan, B.; Jacobs, M. R.; Appelbaum, P. C.
Antimicrob. Agents Chemother. 2003, 47, 1017.
(83) Ng, L. K.; Martin, I.; Liu, G.; Bryden, L. Antimicrob. Agents
Chemother. 2002, 46, 3020.
(84) Lucier, T. S.; Heitzman, K.; Liu, S. K.; Hu, P. C. Antimicrob.
Agents Chemother. 1995, 39, 2770.
(85) Nash, K. A.; Inderlied, C. B. Antimicrob. Agents Chemother.
1995, 39, 2625.
(86) Lukehart, S. A.; Godornes, C.; Molini, B. J.; Sonnett, B. S.;
Hopkins, S.; Mulcahy, F.; Engelman, J.; Mitchell, S. J.; Rompala,
A. M.; Marra, C. M.; Klausner, J. D. N. Engl. J. Med. 2004, 351,
154.
(87) Canu, A.; Malbruny, B.; Coquemont, M.; Davies, T. A.; Appelbaum, P. C.; Leclercq, R. Antimicrob. Agents Chemother. 2002,
46, 125.
(88) Nakajima, Y. J. Infect. Chemother. 1999, 5, 61.
(89) Garza-Ramos, G.; Xiong, L.; Zhong, P.; Mankin, A. J. Bacteriol.
2001, 183, 6898.
(90) Xiong, L.; Shah, S.; Mauvais, P.; Mankin, A. S. Mol. Microbiol.
1999, 31, 633.
(91) Dam, M.; Douthwaite, S.; Tenson, T.; Mankin, A. S. J. Mol. Biol.
1996, 259, 1.
(92) Tenson, T.; DeBlasio, A.; Mankin, A. Proc. Natl. Acad. Sci. U.S.A.
1996, 93, 5641.
(93) Tenson, T.; Xiong, L.; Kloss, P.; Mankin, A. S. J. Biol. Chem.
1997, 272, 17425.
(94) Pardo, D.; Rosset, R. Mol. Gen. Genet. 1977, 153, 199.
(95) Wittmann, H. G.; Stoffler, G.; Apirion, D.; Rosen, L.; Tanaka,
K.; Tamaki, M.; Takata, R.; Dekio, S.; Otaka, E. Mol. Gen. Genet.
1973, 127, 175.
(96) Davydova, N.; Streltsov, V.; Wilce, M.; Liljas, A.; Garber, M. J.
Mol. Biol. 2002, 322, 635.
(97) Ounissi, H.; Courvalin, P. Gene 1985, 35, 271.

526 Chemical Reviews, 2005, Vol. 105, No. 2


(98) Arthur, M.; Autissier, D.; Courvalin, P. Nucleic Acids Res. 1986,
14, 4987.
(99) Arthur, M.; Andremont, A.; Courvalin, P. Antimicrob. Agents
Chemother. 1987, 31, 404.
(100) Wondrack, L.; Massa, M.; Yang, B. V.; Sutcliffe, J. Antimicrob.
Agents Chemother. 1996, 40, 992.
(101) Alignet, J.; Liassine, N.; el Solh, N. Antimicrob. Agents Chemother. 1998, 42, 1794.
(102) OHara, K.; Kanda, T.; Ohmiya, K.; Ebisu, T.; Kono, M. Antimicrob. Agents Chemother. 1989, 33, 1354.
(103) Noguchi, N.; Emura, A.; Matsuyama, H.; OHara, K.; Sasatsu,
M.; Kono, M. Antimicrob. Agents Chemother. 1995, 39, 2359.
(104) Noguchi, N.; Katayama, J.; OHara, K. FEMS Microbiol. Lett.
1996, 144, 197.
(105) Nakamura, A.; Miyakozawa, I.; Nakazawa, K.; OHara, K.;
Sawai, T. Antimicrob. Agents Chemother. 2000, 44, 3241.
(106) Matsuoka, M.; Endou, K.; Kobayashi, H.; Inoue, M.; Nakajima,
Y. FEMS Microbiol. Lett. 1998, 167, 221.
(107) Noguchi, N.; Takada, K.; Katayama, J.; Emura, A.; Sasatsu, M.
J. Bacteriol. 2000, 182, 5052.
(108) Quiros, L. M.; Aguirrezabalaga, I.; Olano, C.; Mendez, C.; Salas,
J. A. Mol. Microbiol. 1998, 28, 1177.
(109) Cundliffe, E. Antimicrob. Agents Chemother. 1992, 36, 348.
(110) Sasaki, J.; Mizoue, K.; Morimoto, S.; Omura, S. J. Antibiot.
(Tokyo) 1996, 49, 1110.
(111) Morisaki, N.; Hashimoto, Y.; Furihata, K.; Yazawa, K.; Tamura,
M.; Mikami, Y. J. Antibiot. (Tokyo) 2001, 54, 157.
(112) Queener, S. W.; Sebek, O. K.; Vezina, C. Annu. Rev. Microbiol.
1978, 32, 593.
(113) Baltz, R. H.; Seno, E. T. Ann. Rev. Microbiol. 1988, 42, 547.
(114) Weber, J. M.; Wierman, C. K.; Hutchinson, C. R. J. Bacteriol.
1985, 164, 425.
(115) Yue, S.; Duncan, J. S.; Yamamoto, Y.; Hutchinson, C. R. J. Am.
Chem. Soc. 1987, 109, 1253.
(116) Cane, D. E.; Yang, C. J. Am. Chem. Soc. 1987, 109, 1255.
(117) Jacobsen, J. R.; Hutchinson, C. R.; Cane, D. E.; Khosla, C.
Science 1997, 277, 367.
(118) Haydock, S. F.; Aparicio, J. F.; Molnar, I.; Schwecke, T.; Khaw,
L. E.; Konig, A.; Marsden, A. F.; Galloway, I. S.; Staunton, J.;
Leadlay, P. F. FEBS Lett. 1995, 374, 246.
(119) Reid, R.; Piagentini, M.; Rodriguez, E.; Ashley, G.; Viswanathan,
N.; Carney, J.; Santi, D. V.; Hutchinson, C. R.; McDaniel, R.
Biochemistry 2003, 42, 72.
(120) Caffrey, P. Chembiochem 2003, 4, 654.
(121) Scrutton, N. S.; Berry, A.; Perham, R. N. Nature 1990, 343, 38.
(122) Anderson, V. E.; Hammes, G. G. Biochemistry 1984, 23, 2088.
(123) Tsai, S. C.; Lu, H.; Cane, D. E.; Khosla, C.; Stroud, R. M.
Biochemistry 2002, 41, 12598.
(124) Tsai, S. C.; Miercke, L. J.; Krucinski, J.; Gokhale, R.; Chen, J.
C.; Foster, P. G.; Cane, D. E.; Khosla, C.; Stroud, R. M. Proc.
Natl. Acad. Sci. U.S.A. 2001, 98, 14808.
(125) Heathcote, M. L.; Staunton, J.; Leadlay, P. F. Chem. Biol. 2001,
8, 207.
(126) Hu, Z.; Pfeifer, B. A.; Chao, E.; Murli, S.; Kealey, J.; Carney, J.
R.; Ashley, G.; Khosla, C.; Hutchinson, C. R. Microbiology 2003,
149, 2213.
(127) Kim, B. S.; Cropp, T. A.; Beck, B. J.; Sherman, D. H.; Reynolds,
K. A. J. Biol. Chem. 2002, 277, 48028.
(128) Donadio, S.; Staver, M. J.; McAlpine, J. B.; Swanson, S. J.; Katz,
L. Science 1991, 252, 675.
(129) Cortes, J.; Haydock, S. F.; Roberts, G. A.; Bevitt, D. J.; Leadlay,
P. F. Nature 1990, 348, 176.
(130) Staunton, J.; Caffrey, P.; Aparicio, J. F.; Roberts, G. A.; Bethell,
S. S.; Leadlay, P. F. Nat. Struct. Biol. 1996, 3, 188.
(131) Weissman, K. J.; Hong, H.; Oliynyk, M.; Siskos, A. P.; Leadlay,
P. F. Chembiochem 2004, 5, 116.
(132) Aparicio, J. F.; Caffrey, P.; Marsden, A. F.; Staunton, J.; Leadlay,
P. F. J. Biol. Chem. 1994, 269, 8524.
(133) Weissman, K. J.; Bycroft, M.; Staunton, J.; Leadlay, P. F.
Biochemistry 1998, 37, 11012.
(134) Pereda, A.; Summers, R. G.; Stassi, D. L.; Ruan, X.; Katz, L.
Microbiology 1998, 144 (Pt 2), 543.
(135) Marsden, A. F.; Caffrey, P.; Aparicio, J. F.; Loughran, M. S.;
Staunton, J.; Leadlay, P. F. Science 1994, 263, 378.
(136) Kao, C. M.; Luo, G.; Katz, L.; Cane, D. E.; Khosla, C. J. Am.
Chem. Soc. 1994, 116, 11612.
(137) Cortes, J.; Wiesmann, K. E.; Roberts, G. A.; Brown, M. J.;
Staunton, J.; Leadlay, P. F. Science 1995, 268, 1487.
(138) Wiesmann, K. E.; Cortes, J.; Brown, M. J.; Cutter, A. L.;
Staunton, J.; Leadlay, P. F. Chem. Biol. 1995, 2, 583.
(139) Pieper, R.; Gokhale, R. S.; Luo, G.; Cane, D. E.; Khosla, C.
Biochemistry 1997, 36, 1846.
(140) Gokhale, R. S.; Tsuji, S. Y.; Cane, D. E.; Khosla, C. Science 1999,
284, 482.
(141) Wu, N.; Cane, D. E.; Khosla, C. Biochemistry 2002, 41, 5056.
(142) Kao, C. M.; Katz, L.; Khosla, C. Science 1994, 265, 509.
(143) Xue, Q.; Ashley, G.; Hutchinson, C. R.; Santi, D. V. Proc. Natl.
Acad. Sci. U.SA. 1999, 96, 11740.
(144) Tang, L.; Fu, H.; McDaniel, R. Chem. Biol. 2000, 7, 77.

Katz and Ashley


(145) Pfeifer, B. A.; Admiraal, S. J.; Gramajo, H.; Cane, D. E.; Khosla,
C. Science 2001, 291, 1790.
(146) Pieper, R.; Luo, G.; Cane, D. E.; Khosla, C. Nature 1995, 378,
263.
(147) Pieper, R.; Ebert-Khosla, S.; Cane, D.; Khosla, C. Biochemistry
1996, 35, 2054.
(148) Jacobsen, J. R.; Keatinge-Clay, A. T.; Cane, D. E.; Khosla, C.
Bioorg. Med. Chem. 1998, 6, 1171.
(149) Jacobsen, J. R.; Cane, D. E.; Khosla, C. Biochemistry 1998, 37,
4928.
(150) Weissman, K. J.; Bycroft, M.; Cutter, A. L.; Hanefeld, U.; Frost,
E. J.; Timoney, M. C.; Harris, R.; Handa, S.; Roddis, M.;
Staunton, J.; Leadlay, P. F. Chem. Biol. 1998, 5, 743.
(151) Cane, D. E.; Kudo, F.; Kinoshita, K.; Khosla, C. Chem. Biol. 2002,
9, 131.
(152) Mochizuki, S.; Hiratsu, K.; Suwa, M.; Ishii, T.; Sugino, F.;
Yamada, K.; Kinashi, H. Mol. Microbiol. 2003, 48, 1501.
(153) Xue, Y.; Zhao, L.; Liu, H. W.; Sherman, D. H. Proc. Natl. Acad.
Sci. U.S.A. 1998, 95, 12111.
(154) Shah, S.; Xue, Q.; Tang, L.; Carney, J. R.; Betlach, M.; McDaniel,
R. J. Antibiot. (Tokyo) 2000, 53, 502.
(155) Witkowski, A.; Joshi, A. K.; Lindqvist, Y.; Smith, S. Biochemistry
1999, 38, 11643.
(156) Bisang, C.; Long, P. F.; Cortes, J.; Westcott, J.; Crosby, J.;
Matharu, A. L.; Cox, R. J.; Simpson, T. J.; Staunton, J.; Leadlay,
P. F. Nature 1999, 401, 502.
(157) Xue, Y.; Sherman, D. H. Nature 2000, 403, 571.
(158) Beck, B. J.; Yoon, Y. J.; Reynolds, K. A.; Sherman, D. H. Chem.
Biol. 2002, 9, 575.
(159) Thomas, I.; Martin, C. J.; Wilkinson, C. J.; Staunton, J.; Leadlay,
P. F. Chem. Biol. 2002, 9, 781.
(160) Tang, L.; Fu, H.; Betlach, M. C.; McDaniel, R. Chem. Biol. 1999,
6, 553.
(161) Yin, Y.; Lu, H.; Khosla, C.; Cane, D. E. J. Am. Chem. Soc. 2003,
125, 5671.
(162) Beck, B. J.; Aldrich, C. C.; Fecik, R. A.; Reynolds, K. A.; Sherman,
D. H. J. Am. Chem. Soc. 2003, 125, 12551.
(163) DeHoff, B. S.; Sutton, K. L.; Rosteck, P. R., Jr. GenBank
Accession No. U78289, 1996.
(164) Gandecha, A. R.; Large, S. L.; Cundliffe, E. Gene 1997, 184, 197.
(165) Kakavas, S. J.; Katz, L.; Stassi, D. J. Bacteriol. 1997, 179, 7515.
(166) Burgett, S. G.; Kuhstoss, S. A.; Rao, R. N.; Richardson, M. A.;
Rosteck, P. R., Jr. US Patent No. 5,945,320, 1999.
(167) Carroll, B. J.; Moss, S. J.; Bai, L.; Kato, Y.; Toelzer, S.; Yu, T.
W.; Floss, H. G. J. Am. Chem. Soc. 2002, 124, 4176.
(168) Reeves, C. D.; Chung, L. M.; Liu, Y.; Xue, Q.; Carney, J. R.;
Revill, W. P.; Katz, L. J. Biol. Chem. 2002, 277, 9155.
(169) Ward, S. L.; Hu, Z.; Schirmer, A.; Reid, R.; Revill, W. P.; Reeves,
C. D.; Petrakovsky, O. V.; Dong, S. D.; Katz, L. Antimicrob.
Agents Chemother. 2004, 78, 4703.
(170) Rodriguez, E.; Hu, Z.; Ou, S.; Volchegursky, Y.; Hutchinson, C.
R.; McDaniel, R. J. Ind. Microbiol. Biotechnol. 2003, 30, 480.
(171) Haydock, S. F.; Dowson, J. A.; Dhillon, N.; Roberts, G. A.; Cortes,
J.; Leadlay, P. F. Mol. Gen. Genet. 1991, 230, 120.
(172) Gaisser, S.; Bohm, G. A.; Cortes, J.; Leadlay, P. F. Mol. Gen.
Genet. 1997, 256, 239.
(173) Summers, R. G.; Donadio, S.; Staver, M. J.; Wendt-Pienkowski,
E.; Hutchinson, C. R.; Katz, L. Microbiology 1997, 143 (Pt 10),
3251.
(174) Tang, L.; McDaniel, R. Chem. Biol. 2001, 8, 547.
(175) Bate, N.; Butler, A. R.; Smith, I. P.; Cundliffe, E. Microbiology
2000, 146 (Pt 1), 139.
(176) Bate, N.; Cundliffe, E. J. Ind. Microbiol. Biotechnol. 1999, 23,
118.
(177) Volchegursky, Y.; Hu, Z.; Katz, L.; McDaniel, R. Mol. Microbiol.
2000, 37, 752.
(178) Aguirrezabalaga, I.; Olano, C.; Allende, N.; Rodriguez, L.; Brana,
A. F.; Mendez, C.; Salas, J. A. Antimicrob. Agents Chemother.
2000, 44, 1266.
(179) Summers, R. G.; Donadio, S.; Staver, M. J.; Wendt-Pienkowski,
E.; Hutchinson, C. R.; Katz, L. Microbiology 1997, 143, 3251.
(180) Borisova, S. A.; Zhao, L.; Sherman, D. H.; Liu, H. W. Org. Lett.
1999, 1, 133.
(181) Trefzer, A.; Blanco, G.; Remsing, L.; Kunzel, E.; Rix, U.; Lipata,
F.; Brana, A. F.; Mendez, C.; Rohr, J.; Bechthold, A.; Salas, J.
A. J. Am. Chem. Soc. 2002, 124, 6056.
(182) Trefzer, A.; Salas, J. A.; Bechthold, A. Nat. Prod. Rep. 1999, 16,
283.
(183) Rodriguez, L.; Aguirrezabalaga, I.; Allende, N.; Brana, A. F.;
Mendez, C.; Salas, J. A. Chem. Biol. 2002, 9, 721.
(184) Linton, K. J.; Jarvis, B. W.; Hutchinson, C. R. Gene 1995, 153,
33.
(185) Katz, L.; Donadio, S. Annu. Rev. Microbiol. 1993, 47, 875.
(186) Majer, J.; Martin, J. R.; Egan, R. S.; Corcoran, J. W. J. Am.
Chem. Soc. 1977, 99, 1620.
(187) Weber, J. M.; Leung, J. O.; Swanson, S. J.; Idler, K. B.; McAlpine,
J. B. Science 1991, 252, 114.
(188) Paulus, T. J.; Tuan, J. S.; Luebke, V. E.; Maine, G. T.; DeWitt,
J. P.; Katz, L. J. Bacteriol. 1990, 172, 2541.

Translation and Protein Synthesis


(189) Weber, J. M.; Schoner, B.; Losick, R. Gene 1989, 75, 235.
(190) Lambalot, R. H.; Cane, D. E.; Aparicio, J. J.; Katz, L. Biochemistry 1995, 34, 1858.
(191) Cupp-Vickery, J. R.; Poulos, T. L. Nat. Struct. Biol. 1995, 2, 144.
(192) Cupp-Vickery, J. R.; Poulos, T. L. Steroids 1997, 62, 112.
(193) Xue, Y.; Wilson, D.; Zhao, L.; Liu, H.; Sherman, D. H. Chem.
Biol. 1998, 5, 661.
(194) Xue, Y.; Wilson, D.; Sherman, D. H. Gene 2000, 245, 203.
(195) Lambalot, R. H.; Cane, D. E. J. Antibiot. (Tokyo) 1992, 45, 1981.
(196) Graziani, E. I.; Cane, D. E.; Betlach, M. C.; Kealey, J. T.;
McDaniel, R. Bioorg. Med. Chem. Lett. 1998, 8, 3117.
(197) Baltz, R. H.; Seno, E. T. Antimicrob. Agents Chemother. 1981,
20, 214.
(198) Okamoto, R.; Kiyoshima, K.; Yamamoto, M.; Takada, K.; Ohnuki,
T.; Ishikura, T.; Naganawa, H.; Tatsuta, K.; Takeuchi, T.;
Umezawa, H. J. Antibiot. (Tokyo) 1982, 35, 921.
(199) Rodriguez, A. M.; Olano, C.; Mendez, C.; Hutchinson, C. R.;
Salas, J. A. FEMS Microbiol. Lett. 1995, 127, 117.
(200) Olano, C.; Rodriguez, A. M.; Michel, J. M.; Mendez, C.; Raynal,
M. C.; Salas, J. A. Mol. Gen. Genet. 1998, 259, 299.
(201) Rodriguez, L.; Rodriguez, D.; Olano, C.; Brana, A. F.; Mendez,
C.; Salas, J. A. J. Bacteriol. 2001, 183, 5358.
(202) Anzai, Y.; Saito, N.; Tanaka, M.; Kinoshita, K.; Koyama, Y.; Kato,
F. FEMS Microbiol. Lett. 2003, 218, 135.
(203) Inouye, M.; Takada, Y.; Muto, N.; Beppu, T.; Horinouchi, S. Mol.
Gen. Genet. 1994, 245, 456.
(204) Wilson, D. J.; Xue, Y.; Reynolds, K. A.; Sherman, D. H. J.
Bacteriol. 2001, 183, 3468.
(205) Bate, N.; Butler, A. R.; Gandecha, A. R.; Cundliffe, E. Chem.
Biol. 1999, 6, 617.
(206) Stratigopoulos, G.; Gandecha, A. R.; Cundliffe, E. Mol. Microbiol.
2002, 45, 735.
(207) Horinouchi, S.; Beppu, T. Mol. Microbiol. 1994, 12, 859.
(208) Stratigopoulos, G.; Cundliffe, E. Chem. Biol. 2002, 9, 71.
(209) Wietzorrek, A.; Bibb, M. Mol. Microbiol. 1997, 25, 1181.
(210) Bate, N.; Stratigopoulos, G.; Cundliffe, E. Mol. Microbiol. 2002,
43, 449.
(211) Butler, A. R.; Flint, S. A.; Cundliffe, E. Microbiology 2001, 147,
795.
(212) Tanikawa, T.; Asaka, T.; Kashimura, M.; Misawa, Y.; Suzuki,
K.; Sato, M.; Kameo, K.; Morimoto, S.; Nishida, A. J. Med. Chem.
2001, 44, 4027.
(213) Tanikawa, T.; Asaka, T.; Kashimura, M.; Suzuki, K.; Sugiyama,
H.; Sato, M.; Kameo, K.; Morimoto, S.; Nishida, A. J. Med. Chem.
2003, 46, 2706.
(214) Arya, A.; Scorneau, B.; Polemeropoulous, A.; Lillard, M.; Han,
F.; Amsler, K.; Wang, G.; Wang, Y.; Peng, Y.; Phan, L. T.; Or,

Chemical Reviews, 2005, Vol. 105, No. 2 527

(215)
(216)
(217)
(218)
(219)
(220)
(221)
(222)
(223)
(224)
(225)
(226)
(227)
(228)
(229)
(230)

(231)

Y. S. 43rd Interscience Conference on Antimicrobial Agents and


Chemotherapy, 2003; Abstract F-1190, Chicago, IL.
Oliynyk, M.; Brown, M. J.; Cortes, J.; Staunton, J.; Leadlay, P.
F. Chem. Biol. 1996, 3, 833.
Ruan, X.; Pereda, A.; Stassi, D. L.; Zeidner, D.; Summers, R.
G.; Jackson, M.; Shivakumar, A.; Kakavas, S.; Staver, M. J.;
Donadio, S.; Katz, L. J. Bacteriol. 1997, 179, 6416.
McDaniel, R.; Thamchaipenet, A.; Gustafsson, C.; Fu, H.;
Betlach, M.; Ashley, G. Proc. Natl. Acad. Sci. U.S.A. 1999, 96,
1846.
McAlpine, J. B.; Tuan, J. S.; Brown, D. P.; Grebner, K. D.;
Whittern, D. N.; Buko, A.; Katz, L. J. Antibiot. (Tokyo) 1987,
40, 1115.
Piel, J. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 14002.
Cheng, Y. Q.; Tang, G. L.; Shen, B. Proc. Natl. Acad. Sci. U.S.A.
2003, 100, 3149.
Stassi, D. L.; Kakavas, S. J.; Reynolds, K. A.; Gunawardana,
G.; Swanson, S.; Zeidner, D.; Jackson, M.; Liu, H.; Buko, A.;
Katz, L. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 7305.
Petkovic, H.; Lill, R. E.; Sheridan, R. M.; Wilkinson, B.; McCormick, E. L.; McArthur, H. A.; Staunton, J.; Leadlay, P. F.;
Kendrew, S. G. J. Antibiot. (Tokyo) 2003, 56, 543.
Reeves, C. D.; Murli, S.; Ashley, G. W.; Piagentini, M.; Hutchinson, C. R.; McDaniel, R. Biochemistry 2001, 40, 15464.
Starks, C. M.; Rodriguez, E.; Carney, J. R.; Desai, R. P.; Carreras,
C.; McDaniel, R.; Hutchinson, R.; Galazzo, J. L.; Licari, P. J. J.
Antibiot. (Tokyo) 2004, 57, 64.
Marsden, A. F.; Wilkinson, B.; Cortes, J.; Dunster, N. J.;
Staunton, J.; Leadlay, P. F. Science 1998, 279, 199.
Donadio, S.; McAlpine, J. B.; Sheldon, P. J.; Jackson, M.; Katz,
L. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 7119.
Kuhstoss, S.; Huber, M.; Turner, J. R.; Paschal, J. W.; Rao, R.
N. Gene 1996, 183, 231.
Yoon, Y. J.; Beck, B. J.; Kim, B. S.; Kang, H. Y.; Reynolds, K.
A.; Sherman, D. H. Chem. Biol. 2002, 9, 203.
Jacobsen, J. R.; Khosla, C. Curr. Opin. Chem. Biol. 1998, 2, 133.
Macielag, M.; Abbanat, D.; Ashley, G.; Foleno, B.; Fu, H.; Li,
Y.; Wira, E.; Bush, K. 42nd Interscience Conference on Antimicrobial Agents and Chemotherapy, 2002; Abstract F-1662, San
Diego, CA.
Abbanat, D. R.; Ashley, G.; Foleno, B.; Fu, H.; Hilliard, J.; Li,
Y.; Licari, P.; Macielag, M.; Melton, J.; Stryker, S.; Wira, E.;
Bush, K. 43rd Interscience Conference on Antimicrobial Agents
and Chemotherapy, 2003; Abstract F-1203, Chicago, IL.

CR030107F

Chem. Rev. 2005, 105, 529542

529

Streptogramins, Oxazolidinones, and Other Inhibitors of Bacterial Protein


Synthesis
Tariq A. Mukhtar and Gerard D. Wright*
Antimicrobial Research Centre, Department of Biochemistry and Biomedical Sciences, McMaster University, 1200 Main Street West,
Hamilton, Ontario, Canada L8N 3Z5
Received May 17, 2004

Contents
1. Introduction
2. Molecular Basis of Protein Translation
3. Streptogramin Antibiotics
3.1. Structures and Biosynthesis
3.2. Synergy and Mode of Action
3.3. Clinical Utility
3.4. Streptogramin Resistance
3.5. New Streptogramins
4. Oxazolidinone Antibiotics
4.1. Discovery and StructureActivity
Relationships
4.2. Mode of Action
4.3. Resistance to Oxazolidinones
4.4. Second-Generation Agents
5. Other Antibacterial Inhibitors of Bacterial
Translation
5.1. Chloramphenicol
5.2. Lincosamides
5.3. Pleuromutilins
6. Conclusions
7. Acknowledgment
8. References

529
531
532
532
534
535
536
537
538
538
538
539
539
539
540
540
540
540
541
541

Tariq Mukhtar was born in Toronto in 1975. He obtained his B.Sc. (Hons)
degree in Biochemistry from McMaster University in 1998, where he
subsequently started postgraduate studies. He is currently a Ph.D. student
in Biochemistry and Biomedical Sciences studying the molecular mechanisms of resistance to type B streptogramins and is supervised by G.
D. Wright.

1. Introduction
The process of protein translation and, in particular, the macromolecular ribozyme that is the ribosome were among the first recognized molecular
targets for antibiotics. A great number of these
antibiotics have since found clinical use or therapeutic promise over the past 50 years (Table 1). Translation and the ribosome remain outstanding drug
targets with numerous efforts directed toward further
mining the potential of this ribozyme and associated
activities in drug design. The recently available highresolution structures of virtually all components of
translation and the intact ribosome now make structure-based drug design across the components of the
whole process a reality.1-4 The chemical diversity of
compounds that can productively interfere with ribosome action is astonishing and includes cationic
aminoglycosides and neutral carbohydrates, mac-

Gerry Wright is Professor and Chair of the Department of Biochemistry


and Biomedical Sciences at McMaster University in Hamilton, Ontario.
He received his Ph.D. degree in Chemistry in 1990 for work on antifungal
targets and spent 2 years as a Natural Sciences and Engineering Research
Council of Canada postdoctoral fellow in Professor Christopher Walshs
laboratory at the Harvard Medical School researching the molecular
mechanisms of vancomycin resistance. He joined the Department of
Biochemistry at McMaster in 1993 and received a Medical Research
Council of Canada Scholarship. Gerrys research program includes
antibiotic-resistance mechanisms, antibiotic biosynthesis, and discovering
new targets for antibacterial and antifungal agents. He presently holds a
Canada Research Chair in Antibiotic Biochemistry.

* To whom correspondence should be addressed. Phone: (905) 5259140, ext. 22454. Fax: (905) 522-9033. E-mail: wrightge@
mcmaster.ca.

rolides, peptides, and diverse small molecules (Figure


1). These act by exploiting numerous binding sites

10.1021/cr030110z CCC: $53.50 2005 American Chemical Society


Published on Web 01/08/2005

530 Chemical Reviews, 2005, Vol. 105, No. 2

Mukhtar and Wright

Table 1. Antibiotics that Target Bacterial Translation


antibiotic

molecular target

aminoglycosides
tetracyclines
macrolides
streptogramins
oxazolidinones
lincosamides
chloramphenicol
edeine
thiostrepton
everninomycins
pleuromutilins
fusidic acid
mupirocin

16S rRNA
16S rRNA
23S rRNA
23S rRNA
23 RNA
23S rRNA
23S rRNA
16S rRNA
23S rRNA
23S rRNA
23S rRNA
EF-G
Ile t-RNA synthetase

binding site,
subunit
A, 30S
A, 30S
P, 50S
P, 50S
P, 50S
P, 50S
P, 50S
P/E, 30S
P, 50S
P, 50S
P, 50S

on the ribosome and ancillary proteins necessary for


translation fidelity.
The availability of 3D structures of the various
protein and rRNA components required for translation represents the beginning of a new era in antibiotic biochemistry.5 Co-structures of ribosometargeted antibiotics with the intact ribosomal subunits

are now available for a growing number of antibiotics,


permitting for the first time examination of structurefunction analysis of this complex structure.1 Already
this work has resulted in a new understanding of
antibiotic-ribosome interaction that has served to
rationalize decades of painstaking biochemical research on antibiotic mode of action and resistance.
Furthermore, the availability of these remarkable
structures, even at the relatively low resolution
presently on hand, has permitted the first forays into
structure-based drug design that no doubt will launch
a new generation of ribosome-directed antibiotics.
The importance of the ribosome as a drug target
can readily be seen in other reviews in this special
issue of Chemical Reviews. This review will address
the structure, mode of action, and resistance to the
streptogramin and oxazolidinone antibiotics, two
distinct classes of antibiotics that have recently been
brought to market for treatment of bacterial infections. In addition, a brief discussion of some new and
old classes of antibiotics that block translation will
be presented.

Figure 1. Structural diversity in inhibitors of bacterial translation.

Streptogramins, Oxazolidinones, and Other Inhibitors

2. Molecular Basis of Protein Translation


Bacterial protein synthesis is an iterative process
consisting of initiation, peptide elongation, and termination events (Figure 2). This process is carried
out by a number of cytoplasmic factors associated
with ribosomes, which are large ribonucleoprotein
assemblies made up of two unequal subunits (30S
and 50S) that associate at the onset of translation
initiation events. The ribosomal architecture is best
viewed in context of the three tRNA binding sites
that span both subunits: the amino-acyl tRNA
binding/decoding A site, the peptidyl-tRNA binding
P site, and the E site, where the uncharged tRNA
exits. The active site of the ribosome, the peptidyltransferase complex (PTC), lies at the interface of the
A and P sites on the 50S subunit, and the growing
peptides exit the ribosome via a 100 hydrophobic
tunnel that opens at the back of the PTC.6
Initiation of translation begins with the formation
of a ternary complex consisting of the 30S subunit,
mRNA, and the initiating tRNA charged with formyl
methionine (fMet).7 The fMet-tRNA binds to the P
site and is the only tRNA to do so as all subsequent
amino acyl-tRNAs (aa-tRNAs) must enter through
the decoding A site (Figure 2, Step 1). The initiation
step is dependent on three initiation factors: IF-1,
IF-2, and IF-3. IF-1 is approximately 70 amino acids
in length, possessing a large S1 RNA binding domain,
and binds in the A site.8 IF-2 promotes GTP-depend-

Figure 2. Overview of bacterial translation. Step 1.


Initiation. Association of initiation factors (IF-1, IF-2, IF3), mRNA, and fMet tRNA (GTP-dependent binding to P
site) with 30S ribosomal subunit. Step 2. Association of
50S ribosomal subunit to the ternary complex, completing
initiation complex. Step 3. Elongation. Ef-TuGTP delivers aa-tRNA into the A site followed by GTP hydrolysis
and accommodation into the A site. Step 4. Peptidyl
transfer occurs, transferring the amino acid from the tRNA
in the P site to the aa-tRNA in the A site. EF-G and GTP
aid in the translocation of the tRNA from the A and P sites
to the P and E sites. Step 5. The free tRNA in the E site
exits the ribosome along with the release of GDP+Pi and
EF-G. This is an iterative process in which the A site is
now unoccupied and prepared to receive an incoming aatRNA. Step 6. Termination. Termination occurs upon
encountering a stop codon. Release factors (RF-1, RF-2, RF3, RRF) dissociate the complex, releasing the polypeptide,
mRNA, and ribosomal subunits, preparing them for recycling.

Chemical Reviews, 2005, Vol. 105, No. 2 531

ent binding of the fMet-tRNA to the 30S subunit,


whereas IF-3 is believed to act as a fidelity factor
during the assembly of this ternary initiation complex and as a means of preventing association of the
two subunits prior to initiation.9 Upon formation of
the ternary initiator complex the 50S subunit associates with the 30S subunit followed by release of the
initiation factors (Figure 2, Step 2).
The next step in protein synthesis is elongation.
This consists of a series of codon-anticodon decoding,
peptide synthesis, and translocation events that have
been the focus of intense research over the past
decade. An elongation factor, EF-Tu bound with GTP,
is responsible for delivering the aa-tRNA to the A site
of the ribosome, where the decoding takes place
(Figure 2, Step 3). Decoding is an elegant molecular
process in which the ribosome ensures the accuracy
of the incoming aa-tRNA into the A site.10 Although
the aa-tRNA contains the anticodon for the respective
codon on the mRNA, this does not solely account for
the fidelity of protein synthesis. In fact, it has been
demonstrated that the energy difference between a
cognate codon-anticodon interaction and a near
cognate interaction is not large enough to account for
the high accuracy with which the ribosomes carry out
their function.11,12 The additional accuracy is contributed by a process of dynamic conformational
changes in the ribosome decoding center that detect
orientation and geometry, resulting in incorporation
of correct amino acids onto the nascent chain.13,14
The decoding process consists of a series of molecular checkpoints that ensure the correct incorporation of amino acids.10 Ensuring that the correct amino
acid is incorporated occurs in two stages: prior to
GTP hydrolysis and after GTP hydrolysis. Decoding
begins with the reversible binding of the anticodon
of an EF-Tuaa-tRNA complex to the codon at the A
site of the ribosome. If the interaction is noncognate
in nature, the complex will be released without
hydrolysis of GTP.15 Although the dissociation constants for cognate and near cognate codon-anticodon
interactions do differ, the initial selection process
does not readily distinguish between the two on the
basis of this difference. In principle, any EF-TuaatRNA complex resulting in near-cognate interactions
can bind the A site. Discrimination between cognate
and near-cognate codon-anticodon pairing then occurs at the second stage of proofreading, in which
geometry and overall fit of the aa-tRNA plays a
significant role in its stability and ultimately the rate
at which it is accommodated into the A site.14,16
Accommodation thus becomes a necessary factor for
a specific amino acid to be incorporated in the
growing peptide. Upon accommodation into the A
site, conformational changes in EF-Tu and the ribosome induce GTP hydrolysis, and EF-Tu release16,17
and peptidyl transferase reaction between the peptidyl-tRNA and aa-tRNA can take place (Figure 2,
Step 4).
A fundamental process in the elongation step is
translocation. Translocation is the ratchet-like movement of the ribosome along the mRNA by one codon,
resulting in the tRNA shifting from the A site to the
P site to the E site. The rate of this particular step

532 Chemical Reviews, 2005, Vol. 105, No. 2

is greatly increased by EF-G, which aids in translocation in a GTP-dependent manner. The use of tRNA
mimics has recently shown this process to be a
complex series of interactions and events in which
the 3 end of the aa-tRNA in the A site undergoes a
180 rotation around a local 2-fold rotation axis in
the PTC, and shift of the rest of the tRNA molecule,
to enter the P site. This event occurs concurrently
with peptidyl transfer.18 Various residues within the
PTC cavity have also been suggested to guide and
rotate the tRNA as it passes from A site to P site
and subsequently guide the nascent peptide through
the exit tunnel.18
Peptidyl transfer occurs in the PTC, which is
comprised only of RNA, and thus the ribosome is
clearly a ribozyme. The details of the chemical
mechanism of acyl transfer remain controversial, and
in particular, the contributions of acid-base catalysis
vs substrate proximity have been reviewed.5,19
The final step of protein synthesis, termination,
occurs when the ribosome encounters a stop codon
in the mRNA at the A site (Figure 2, Step 6). Various
cytoplasmic factors, known as release factors, release
the polypeptide and prepare the ribosome for recycling (RF1, RF2, RF3). RRF in conjunction with EF-G
is responsible for separating the subunits and the
associated components.

3. Streptogramin Antibiotics
3.1. Structures and Biosynthesis
Streptogramin antibiotics are natural products
produced by various members of the Streptomyces
genus. This family of antibiotics consists of two
subgroups, type A and type B, which are simultaneously produced by the same bacterial species in a
ratio of roughly 70:30.20,21 Group A streptogramins
are cyclic polyunsaturated macrolactones that are
comprised of a hybrid peptide/polyketide structure
and are cyclized through an internal ester bond
between the carboxyl of the C-terminal amino acid
(generally Pro) and an internal hydroxyl group
(Figure 3). Structural variations in type A streptogramins can arise from desaturation of the Pro
residue and by its substitution for Ala or Cys.22
Examples of group A streptogramins are pristina-

Figure 3. Structures of type A streptogramin antibiotics.

Mukhtar and Wright

Figure 4. Biosynthetic origins of various components of


group A streptogramin antibiotics.

mycin IIA (same as virginiamycin M1), madumycin


II, and the semisynthetic derivative dalfopristin
(Figure 3).
The biosynthetic origins of the components of the
group A streptogramin virginiamycin M1 have been
investigated through traditional precursor fermentation experiments (Figure 4).23 The C-terminal dehydroproline predictably arises from Pro.23 A twoenzyme system comprised of an FMN-dependent
monooxygenase and an FMN reductase that catalyzes Pro oxidation has been purified and characterized from the pristinamycin IIA producer Streptomyces pristinaespiralis.24,25 The oxazole ring is derived
from Ser,23,26 which cyclizes in a fashion reminiscent
of other natural product antibiotics such as microcin
(reviewed in ref 27). The rest of the molecule is
largely comprised of acetate units derived from a
polyketide synthesis scaffold with the exception of
positions 9 and 10 (and likely the amide nitrogen),
which are derived from Gly, and the isopropyl group
that originates in Val (via crotonyl CoA). Methyl
groups 32 and 33 (Figure 4) originate from different
sources, respectively, Met and the methyl group of
acetate, derived from an uncharacterized decarboxylation mechanism.23
Group B streptogramins are cyclic hepta- or hexadepsipeptides, e.g., pristinamycin IA, virginiamycin
S, the semisynthetic quinupristin (Figure 5). The
nomenclature in this field is highly redundant with
several molecules reported in the literature having
an identical structure but different names; for example, pristinamycin IA t streptogramin B t vernamycin BR t mikamycin IA t ostreogrycin B t
PA114B1.28 The general composition of group B
streptogramins is 3-hydroxypicolinic acid-L-Thr-Daminobutyric acid (or D-Ala)-L-Pro-L-Phe (or 4-N,N(dimethylamino)-L-Phe)-X-L-phenylglycine. Residue
X is most commonly L-4-oxo or 4-hydroxypipecolic
acid but can also be Asp or Pro. The invariant
N-terminal Thr residue is N-acetylated with 3-hydroxypicolinic acid and forms a cyclizing ester linkage
with the C-terminal carboxyl group of the peptide via
its secondary hydroxyl group.
The biosynthesis of the unusual amino acids that
comprise type B streptogramins has not been extensively studied. Labeled precursor feeding experiments have shown that phenylalanine is the source
of phenylglycine29,30 and both 4-oxopipecolic acid and
3-hydroxypicolinic acid are derived from Lys30,31

Streptogramins, Oxazolidinones, and Other Inhibitors

Chemical Reviews, 2005, Vol. 105, No. 2 533

Figure 5. Structures of type B streptogramin antibiotics.

Figure 6. Biosynthesis of 4-N,N-(dimethylamino)-L-Phe


required for type B streptogramin antibiotics.

during the biosynthesis of virginiamycin S1 by Streptomyces virginiae. A four-gene system for the production of 4-N,N-(dimethylamino)-L-Phe has been cloned
and partially characterized from the pristinamycin
producer Streptomyces pristinaespiralis.32 Three of
the genes are reminiscent of bacterial p-aminobenzoic
acid formation with papA acting as a chorismate
aminotransferase, papB as a mutase, and papC as a
predicted dehydrogenase that catalyzes formation of
4-aminphenylpyruvic acid (Figure 6). An unidentified
transaminase then converts the ketone to 4-aminoPhe. The final steps in the biosynthesis of 4-N,N(dimethylamino)-L-Phe are two successive methylations of the 4-amino group catalyzed by the enzyme
PapM, which has been partially characterized and
confirmed to be S-adenosylmethionine (SAM) dependent (Figure 6).32
The other components of type B streptogramins for
which some biochemical information is known are the
Lys-derived 3-hydroxypicolinic acid and 4-oxopipe-

colic acid. Sequencing of DNA downstream from


known regulatory genes of virginiamycin S biosynthesis in S. virginiae identified four genes, visA-D,
predicted to be involved in amino acid biosynthesis.33
Both visA and visC are predicted to encode enzymes
that generate 1-piperidine 2-carboxylic acid. VisA has
been purified and shown to be a pyridoxal-phosphatedependent Lys 2-aminotransferase capable of generating 1-piperidine 2-carboxylate.33 Inactivation of the
visA gene in S. virginiae completely blocked biosynthesis of virginiamycin S1, and addition of 3-hydroxypicolinic acid but not pipecolic acid to the culture
medium rescued this null phenotype, suggesting that
VisA is required for formation of the former (Figure
7).33 Paradoxically, the intracellular levels of 1-piperidine 2-carboxylic acid in the visA null mutant were
comparable to the wild-type organism. The visC and
visD genes show homology to 1-piperidine 2-carboxylic acid, generating Lys cyclodeaminase and P-450
oxidase, respectively. These genes and their products
have not been well studied, but they are predicted
to be involved in the synthesis of 4-oxopipecolic acid
(Figure 7).34
The presence of nonprotein amino acids in type B
streptogramins predicts the requirement of nonribosomal peptide synthetases (NRPSs) for antibiotic
assembly (see review by Sieber and Marahiel in this
issue). The NRPS-encoding genes responsible for
biosynthesis of pristinamycin I from S. prinstinaespiralis have been purified35 and cloned.36,37 Three
NRPS genes, snbA, snbC, and snbDE, are organized
in a 1,2, and 3 module arrangement (Figure 8). The
first gene encodes SnbA, which activates 3-hydroxypicolinic acid;36 this gene has been cloned and designated visB in S. virginiae.34 An epimerization
domain in the aminobutyric acid module of snbC is
consistent with the D-stereochemistry of this amino
acid in the final product. Similarly, the presence of

534 Chemical Reviews, 2005, Vol. 105, No. 2

Mukhtar and Wright

Figure 7. Biosynthesis of 3-hydroxypicolinic acid and 4-oxopipecolic acid components of type B streptogramin antibiotics.

Figure 8. Organization of the nonribosomal assembly line for the biosynthesis of pristinamycin I.

a methyltransferase domain flanking the 4-N,N(dimethylamino)-L-Phe activation domain of SnbDE


is consistent with N-methylation of the amide bond
(Figure 8).
The biosynthesis of streptogramins is highly regulated by -butyrolactone autoregulators that affect
gene expression at nanomolar levels.38 These diffusible small molecules bind to highly specific transcription factors, e.g., BarX of S. virginiae.39 These quorumdependent autoregulators, analogous to the homoserine lactones of some Gram-negative bacteria,40
have been implicated in regulation of secondary
metabolism and cellular differentiation in a number
of actinomycetes.41,42

3.2. Synergy and Mode of Action


Both the A and B streptogramins bind the P site
of the 50S ribosome, a property that is shared with
such structurally diverse antibiotics as the macrolides, lincosamides, and thiopeptides (Table 1).
Streptogramins are unique in that the two compo-

nents (A and B) are separately bacteriostatic, yet


when working in combination, they act synergistically on inhibition of bacterial growth and can become
bactericidal.43 Group A streptogramins bind to the
PTC only in the absence of aminoacyl-tRNAs and
block substrate attachment to the donor and acceptor
sites, preventing the early events in elongation. In
addition, the binding of type A streptogramins causes
a conformational change in the 50S ribosome that
increases the activity of the type B streptogramins
by 100-fold.43 Type B streptogramins, on the other
hand, prevent the extension of protein chains, cause
the release of incomplete peptides, and can bind to
the ribosomes at any step of protein synthesis.43
The structure of virginiamycin M bound to the
Haloarcula marismortui 50S subunit supports the
inhibition of the PTC activity as the antibiotic bridges
both the A and P sites.44 The structure of the 50S
ribosomal subunit from Deinococcus radiodurans in
complex with both dalfopristin and quinupristin has
also been reported (Figure 9).45 The structure con-

Streptogramins, Oxazolidinones, and Other Inhibitors

Chemical Reviews, 2005, Vol. 105, No. 2 535

Figure 9. Structure of dalfopristin and quinupristin bound to the large subunit of the bacterial ribosome. (A) Electron
density of the dalfopristin and quinupristin bound to the ribosome. (B) Space-filling model showing the positioning of
dalfopristin, quinupristin, and the peptidyl t-RNA. The peptide exit tunnel is shown to demonstrate the impact of
quinupristin binding. (Reprinted with permission from ref 45. Copyright 2004 BioMed Central.)

firmed many of the biochemical studies performed


with streptogramins and the ribosome and also sheds
new light on the basis of synergy for these antibiotics.
The structure demonstrated that quinupristin binds
at the entrance to the exit tunnel through which the
nascent peptide travels. The antibiotic-ribosome
interaction appears to be dominated by hydrophobic
interactions and also via hydrogen bonds to residues
A2062 and C2586 (Figure 5). Interestingly, the quinuclidinylthio moiety that imparts good solubility
and pharmacological properties to quinupristin occupies an empty space in the subunit. This is
consistent with the fact that various modifications
in this position do not interfere with antibiotic
binding.
Like virginiamycin M, dalfopristin is positioned in
a pocket in the PTC.45 It also appears to be bound by
a network of hydrophobic interactions in addition to
hydrogen bonds with residues G2505 and G2061.
These studies indicate that dalfopristin interferes
with the correct positioning of the substrates for the
A and P sites. Therefore, once the P site is occupied
by the substrate, then binding of dalfopristin can be
hindered. This is consistent with studies demonstrating that translating ribosomes are not susceptible to
dalfopristin.46,47
The synergistic nature of these antibiotics is assisted by hydrophobic interactions between the streptogramins. The structure indicates that both antibiotics contact A2062 through both hydrophobic interactions and hydrogen bonds. This is also consistent
with biochemical data, suggesting that A2062 undergoes conformational changes upon binding of
streptogramins.48,49 Perhaps the most interesting
suggestion of this study was the description of
conformational changes in the PTC upon dalfopristin
binding. These changes are shown to be in residue
U2585. This residue appears to be rotated by 180
in the streptogramin-bound state as compared to the
native, making hydrogen bonds with residues C2606
and G2588. Harms et al. performed simulations that
indicated that if there are no long-range conformational changes upon binding of dalfopristin, then

U2585 in the native and rotated conformations have


nearly identical energy states. The binding of dalfopristin therefore provides sufficient energy to move
the residue through the energy barrier to the rotated
conformation, thereby making new contacts via hydrogen bonds. The spontaneous reversal is predicted
to be slow even after removal of dalfopristin due to
similarity in energy state to the native form. This
observation is consistent with studies on the postantibiotic effect of the drug (retention of antibiotic
activity even after circulating levels of antibiotic have
dropped below the minimal inhibitory concentration).

3.3. Clinical Utility


Despite the fact that the streptogramins were first
discovered in the 1950s, these antibiotics only found
marginal use in Europe in the following decades as
antibacterial agents in the clinic. They did however,
and still do, find use as feed additives in agriculture.
The rise in antibiotic resistance in the 1980s and
1990s, largely as a result of the emergence of vancomycin-resistant enterococci (VRE), spurred renewed interest in antibiotic discovery in pharmaceutical companies. Medicinal chemists at RhonePoulenc worked to improve the drug-like properties
of pristinamycin, resulting in the synthesis and
characterization of the semisynthetic streptogramins
dalfopristin (type A) and quinupristin (type B). Their
combination in a 7:3 ratio resulted in the antibiotic
Synercid, which received FDA approval in 1999 for
the treatment of bacteremia caused by vancomycinresistant Enterococcus faecium and skin and skin
structure infections caused by Staphylococcus aureus
and Streptococcus pyogenes.
A number of studies have been performed that
examine the in-vitro and in-vivo antibacterial activity
of quinupristin and dalfopristin. Bouanchaud reviewed the susceptibility of the quinupristin/dalfopristin combination, illustrating its effectiveness
against a wide range of organisms.50 Synercids
spectrum of activity includes most Gram-positive
pathogens such as species of Staphylococci, Streptococci, and Enterococci, including multi-antibiotic-

536 Chemical Reviews, 2005, Vol. 105, No. 2

Mukhtar and Wright

resistant strains. Enterococcus faecalis, however, is


intrinsically resistant to the effects of Synercid, likely
due to the presence of Lsa, a predicted dalfopristinclindamycin efflux pump.51 Furthermore, important
aerobic Gram negatives including Moraxella catarrhalis, Legionella spp., Mycoplasma spp., Neisseria
spp., and to a lesser degree Haemophilus influenzae
are also susceptible to Synercid. Its antibacterial
activity has also been confirmed against a number
of anaerobic organisms such as Bacteroides spp.,
Lactobacillus spp., Clostridium perfringens, and
Clostridium difficile. The wide spectrum of antibacterial activity for quinupristin/dalfopristin therefore
provides a therapeutic alternative for life-threatening
bacterial infections.

3.4. Streptogramin Resistance


Despite the relatively recent development and
clinical entry of Synercid, multiple mechanisms of
resistance to streptogramins are known. This may
be a reflection of the significant agricultural use of
this class of antibiotic over several decades. There
are three major acquired streptogramin-resistance
mechanisms: active efflux, target modification, and
antibiotic inactivation.
Efflux of type B streptogramin antibiotics has been
found to be associated with ATP-binding transport
pumps that are specific for the 14- and 15-membered
macrolide antibiotics as well as streptogramin B, for
example, MsrA, from S. aureus RN4220.52 Efflux of
type A streptogramins has also been associated with
several proteins including Lsa, which is intrinsic to
E. faecalis,51 and Vga53 and VgaB,54 which have been
found on plasmids in S. aureus.
Another mechanism of streptogramin resistance is
target modification. Covalent modification of the 23S
rRNA by an rRNA methylase encoded by the erm
gene is a highly prevalent mechanism of streptogramin B resistance. Erm-mediated ribosomal monoand dimethylation of the N6 of A2058 (E. coli numbering) of the 23S RNA confers resistance not only
to type B streptogramins but also to macrolides such
as erythromycin and to lincosamides such as clindamycin.55 This gives rise to the so-called MLSBresistance phenotype (M ) macrolide, L ) lincosamide, SB ) type B streptogramin) that is predominantly found in Gram-positive bacteria. There
are a number of different erm genes that have been
identified in a variety of organisms, and resistance
is likely a result of a conformational change occurring
in the ribosome.56 The type A streptogramins are not
affected by Erm-mediated resistance, and as a result,
synergy between the two types of streptogramins is
maintained.
An important aspect of this type of resistance is
the regulation of the respective erm genes. Expression of the MLS resistance in staphylococci may be
inducible or constitutive. The inducible nature of the
resistance determinant depends on regions adjacent
to the gene itself. Specifically, inducible erm gene
transcripts contain an upstream regulatory element
that contain four inverted repeats.55 These give rise
to two stem loop structures that in the absence of
erythromycin sequester both the ribosome binding

Figure 10. VatD as a representative inactivator of type


A streptogramin antibiotics. (A) 3D structure of VatD
cocrystallized with Virginiamycin M1 and CoA.58 (B) Predicted molecular mechanism of acetyltransfer of VatD.

site and initiation codon for the erm gene. When


erythromycin is present, the antibiotic binds the
ribosome, resulting in stalling of translation. This
stalling event likely kinetically displaces the stem
loop structures and allows translation of the methylase. Therefore, gene regulation is maintained by a
feedback mechanism; when all the ribosomes are
methylated, erythromycin cannot bind, in turn stalling does not occur, and the mRNA adopt the stem
loop structures again.56
Recently, Tait-kamradt et al. reported clinical
isolates of S. pneumoniae conferring the MSB
phenotype that did not harbor ermB or mutations in
the 23S rRNA.57 These isolates were found to have
modified L4 ribosomal proteins with either a three
amino acid substitution (G69TG to T69PS) or in one
case a six amino acid insertion in this region.
The final type of acquired resistance to streptogramins include a cadre of antibiotic inactivation
enzymes. These consist of acetyltransferases, which
modify type A streptogramins, and lyases, which
inactivate type B streptogramins. The acetyltransferases transfer an acetyl group from acetyl-CoA on

Streptogramins, Oxazolidinones, and Other Inhibitors

Chemical Reviews, 2005, Vol. 105, No. 2 537

Figure 11. Predicted molecular mechanism of Vgb-catalyzed inactivation of type B streptogramin antibiotics.

to the secondary hydroxyl of the type A streptogramin. This hydroxyl makes multiple contacts with
G2505 (E. coli numbering) of the 23S rRNA in the
PTC as evidenced by the 3D structure of the antibiotic bound to the ribosome,45 and therefore, acetylation of this hydroxyl results in a steric block of drugtarget interaction. On the other hand, the lyases
cause the cleavage of the ester linkage on type B
streptogramins, resulting in linearization of the
peptide and loss of the bioactive conformation necessary for ribosome binding.
A number of streptogramin A acetyltransferases
have been identified in many pathogenic organisms.
The crystal structure for VatD, an acetyltransferase
from E. faecium, has been solved in the apo form,58,59
bound to substrate acetyl-CoA,58,59 product CoA,58
and the streptogramins virginiamycin M158 and dalfopristin59 (Figure 10A). The enzyme is a homotrimer
with each subunit folded into three domains. These
domains consist of a large coiled LH, an extended
loop domain, and C-terminal domain. Recent mutagenesis studies have supported a predicted mechanism where His82 acts as the general base and is a
major determinant of catalytic rate enhancement by
VatD (Figure 10B).59
The other streptogramin inactivation mechanism
associated with pathogenic bacteria is conferred by
Vgb lyase, which inactivates type B antibiotics. Early
studies on type B streptogramin-inactivating enzymes suggested that this enzyme was a lactonase
in which hydrolysis was responsible for linearizing
the cyclic peptide.60-63 However, in 1996 Bateman et
al. reported type B streptogramin-inactivating activity in crude cell-free extracts of Streptomyces lividans.
The extract inactivated the antibiotic etamycin
through an elimination mechanism as opposed to
hydrolysis.20 Recent studies on Vgb lyase, using
purified recombinant enzyme from S. aureus, and

orthologues encoded on the chromosomes of Bordetella pertussis and Streptomyces coelicolor have demonstrated that this enzyme also inactivates type B
streptogramins through an elimination mechanism
as opposed to hydrolysis (Figure 11).
In 1998, Suzuki et al. reported an enzyme from the
streptogramin-producing organism S. virginiae that
was capable of inactivating type A streptogramins by
a previously unidentified mechanism.64 This inactivation occurred through reduction of the 16-carbonyl
group of virginiamycin M, resulting in 16R-dihydroVM. Although this inactivation mechanism has
not yet been reported in the clinics, the possibility
for its appearance is an event that both researchers
and health practitioners should be aware of.

3.5. New Streptogramins


Synercid is not orally available and is administered
by intravenous routes. Efforts have therefore been
made to generate new orally active streptogramins.
For example, RPR 106972 is a 2:1 (molar) mixture
of pristinamycin IB (RPR 112808) and pristinamycin
IIB (RPR 106950) (Figure 12).65 Pristinamycin IB
differs from a closely related molecule pristinamycin
IA at the N-methyl-4-(dimethylamino)phenylalanine
residue. In this particular case pristinamycin IB
contains a N-methyl-4-(methylamino)phenylalanine
at this position. In addition, pristinamycin IIB differs
from pristinamycin IIA in the dehydroproline residue
in which it is replaced with proline.
Rhone Poulenc Rorer (subsequently Aventis Pharma) also developed additional type B streptogramins
through chemical modification of the 4-oxopipecolic
acid moiety and type A streptogramins through
modification of the dehydroproline residue. In particular, a new oral streptogramin designated XRP
2868, a combination of RPR132552A and RPR 202868

538 Chemical Reviews, 2005, Vol. 105, No. 2

Figure 12. New semisynthetic streptogramin antibiotics: (A) type A streptogramins; (B) type B streptogramins.

(Figure 12), has been shown to be very effective


against a number of Gram-positive and Gram-negative organisms and particularly effective against
pneumococcal and Haemophilus strains.66 A number
of patents have been issued to Aventis Pharma for
the development of novel streptogramin derivatives
which have shown to have activity against Grampositive organisms, particularly streptococci, enterococci, and staphylococci.67 Some examples of these
structures have been summarized in Figure 12.
The growing problem of streptogramin resistance
as well as a desire to improve pharmacological
properties will drive the development of new agents
in the future. The availability of 3D strucutres of the
ribosome with bound antibiotics and of resistance
enzymes will greatly facilitate these efforts.

4. Oxazolidinone Antibiotics
4.1. Discovery and StructureActivity
Relationships
The oxazolidinones are the only new chemical class
of antibiotic that have been discovered and success-

Figure 13. Structures of oxazolidinone antibiotics.

Mukhtar and Wright

fully implemented in the clinic over the past 40 years.


These bacteriostatic molecules find no congeners in
natural product compounds and were first synthesized by chemists at E.I. du Pont de Nemours & Co.
in the mid-1980s (e.g., DuP-721, Figure 13).68,69 While
these compounds showed good antibacterial activity
against Gram-positive bacteria such as Staphylococci,
Streptococci, and Enterococci,68 the compound class
was not pursued as a result of toxicity issues in
animals studies.70 Scientists at Pharmacia subsequently found that a structure-toxicity relationship
could be established for the oxazolidinones and
initiated a campaign to develop these compounds as
novel antibacterial agents (historically reviewed in
ref 70).
Initial structure-activity relationship studies revealed a core structure requirement for the 5-S
configuration at position 5 of the oxizolidinone ring,
an acylaminomethyl group linked to C5 and N-aryl
ring substitution (Figure 13). This work was expanded to explore functionalization of the aryl ring
that resulted in improved activity or expanded antibacterial spectrum. For example, substitution with
azole groups added increased activity toward the
Gram-negative pathogens H. influenzae and Moraxella catarrhalis,71 and compounds incorporating
thiomorpholine resulted in activity against mycobacteria.72 Substitution by piperazine,73 morpholine,74 as
well as fluorination75 was also explored. The details
of this SAR program have recently been reviewed.70
The outcome of this drug discovery program is the
antibiotic linezolid (Figure 13), which received FDA
approval in 2000 and is marketed under the trade
name Zyvox.76 Linezolid is active both in oral and
injectable forms and is indicated for the treatment
of a variety of infections caused by Gram-positive
bacteria.

4.2. Mode of Action


Oxazolidinones bind to the P site of the 50S subunit
with micromolar affinity77,78 and inhibit bacterial
translation.79 NMR studies indicate that the solution
structures and the ribosome-bound structures of the
antibiotics are very similar.78 Despite binding to the
P region of the ribosome in a region that overlaps
with chloramphenicol and lincosamide binding sites,
oxazolidinones have no effect on the PTC activity.77
Instead, inhibition of translation appears to be at the
level of translation initiation.80 In particular, binding
of oxazolidinones at the P site interferes with binding
of the initiator fMet-tRNA to this site during the
formation of the initiating complex (Figure 2).81 A
recent study indicated that oxazolidinones also promote frame shifting and stop codon read through in
an E. coli system.82

Streptogramins, Oxazolidinones, and Other Inhibitors

Mapping of 23 rRNA mutations that result in


oxazolidinone resistance (see below) have further
localized the binding site to the P site, and kinetic
studies of a number of inhibitors of translation with
oxazolidinone-resistance ribosomes reveal some crossresistance with chloramphenicol but not the A-sitespecific antibiotic sparsomycin.83 Careful work by
Pomplianos group at Bristol-Myers Squibb suggests
that oxazolidinones induce a conformational shift in
the peptidyl transferase region that is specific to the
structure of the individual antibiotic.83 Therefore, one
could imagine a suite of similar compounds with
different resistance patterns; high-resolution crystal
structures of various oxazolidinones in complex with
the ribosome will greatly facilitate interpretation of
the available data and its use in structure-guided
drug design.
Taken together, the available research points to a
mechanism of action where oxazolidinones bind to the
P site of the 50S subunit, in particular adjacent to
the 23S rRNA, preventing fMet-tRNA from binding
in a fashion that permits the formation of the first
peptide bond that begins mRNA translation. Additional effects on fidelity (e.g., frame shifting) may
also contribute to the mechanism of action.

4.3. Resistance to Oxazolidinones


Widespread resistance to linezolid in the clinic is
thus far rare. In a survey of over 40 000 isolates of
Gram-positive cocci between 1998 and 2000, only
eight strains were resistant (MIC g 8 g/mL). These
included species of the genera Enterococcus, Staphylococcus, and Streptococcus and two strains of vancomycin-resistant enterococci. Prolonged use in patients has selected for resistance in methicillinresistant Staphylococcus aureus84,85 and enterococci86
and has been shown to emerge in enterococci87 and
others during in-vitro serial dilution or mutagenesis
studies.88,89 Importantly, resistant bacteria in
patients who have not had prior treatment with
linezolid has been reported as well.90,91
Resistance occurs by site mutations in the domain
V region of the 23S rRNA, consistent with the mode
of action (see above). Until recently, all clinically
derived linezolid resistance was associated with a G
f U mutation at position 2576 of the 23S rRNA.
These strains retain resistance to linezolid even in
the presence of the ribosomal methylation enzyme
Erm(C).92 However, Meka et al. recently reported a
T2500A mutation in S. aureus in addition to loss of
a copy of the 23 RRNA gene in some isolates.93 Most
bacterial species carry multiple copies of the rDNA
genes, and in S. aureus there are 5-6 23S rDNA
genes. Heterozygous mutations in these genes result
in a gradient of linezolid susceptibility. For example,
strains with two of six possible G2576T mutations
gave linezolid MIC of 8 g/mL, while mutation of five
of six gave MIC of 32 g/mL compared to the wildtype, linezolid-susceptible, bacterium with MIC of 2
g/mL.85 Reversion of this mechanism has been
reported in a strain of S. aureus with four of five
copies bearing the G2576T mutation following repeated serial passage in antibiotic-free medium.94

Chemical Reviews, 2005, Vol. 105, No. 2 539

Other ribosomal mutations have been reported


during in vitro selection experiments,82,87-89 and the
addition of mutations in DNA repair mechanisms
increases the frequency of linezolid-resistant mutations95 Sander et al. used Mycobacterium smegmatis,
which only carries one copy of the 23S rDNA, to
explore the mechanisms of linezolid resistance identifiable after serial passage experiments. As expected,
23S rRNA mutations were identified but also some
nonribosomal mutants emerged in the screen, possibly the result of altered uptake or efflux.96 Resistance to all antimicrobial agents is predictable;
however, thus far widespread resistance remains rare
for the oxazolidinones likely for two reasons. First,
the fact that these agents are not natural products,
and therefore, the microbial community has not seen
this chemical scaffold in the past. Second, ribosomal
mutations must be hardwired into the chromosome
and the presence of multiple 23S rDNA genes makes
homozygocity, which is associated with high-level
resistance, rare. This argues well for the longevity
of this class of agent and presents an opportunity to
initiate program development in new members of the
class to be ready if (or rather when) resistance
becomes more of an issue.

4.4. Second-Generation Agents


Since the launch of linezolid there has been great
interest in additional SAR studies directed at the
oxazolidinone class of antibiotics to identify new
agents with improved properties such as bacterial
spectrum and to counter emerging resistance. For
example, combinatorial synthesis of S-oxide, fluoroacetamido analogues of linezolid, such as compounds
1 and 2 in Figure 14, provide increased activity
against H. influenzae and M. catarrhalis.97 Similarly,
indolinyl derivatives such as compounds of general
structure 3 also showed improved activity against
these Gram-negative pathogens. A series of pyrrol
derivatives (4) was shown to have activity against
various mycobacterial species including multidrugresistant strains.98
In a creative approach two groups recently reported
a series of compounds linking oxazolidinone and
fluoroquinolone (ciprofloxacin) moieties, e.g., compound 7 (Figure 14).99-101 The best of these analogues
exhibited a broad spectrum of antibacterial activity,
e.g., including bacterial targets not covered by linezolid such as E. coli, and activity against linezolidand fluoroquinolone-resistant bacteria. This clever
combination approach has the potential to be further
optimized to achieve improved drug-like properties.

5. Other Antibacterial Inhibitors of Bacterial


Translation
The structural diversity of compounds shown in
Figure 1 and Table 1 graphically demonstrate that
inhibitors of bacterial translation encompass a very
broad chemical space that has great potential to be
exploited in the development of new antimicrobial
agents. Furthermore, as the lesson of the clinical
development of the streptogramins demonstrates,
reexamination of old agents also has great potential

540 Chemical Reviews, 2005, Vol. 105, No. 2

Mukhtar and Wright

Figure 14. Sample of some newer oxazolidinone antibiot504ics.

in the development of clinically important drugs. In


this section a brief discussion of some old and new
inhibitors of translation will be presented.

5.1. Chloramphenicol
The antibiotic chloramphenicol (Figure 1) was
discovered in 1947 from extracts of Streptomyces
venezuelae.102 It has excellent broad-spectrum antibacterial activity against Gram-positive and Gramnegative bacteria and was widely used clinically
following its discovery. This antibiotic however is no
longer in significant use as a result of potential fatal,
irreversible aplastic anemia which has been associated with it.103 The basis for this rare side effect is
unknown but is likely genetic. Therefore, a sensitive
screen of genotype could reinvigorate the use of this
agent in a safe fashion.
Chloramphenicol binds to the P site of the ribosome
in an area of the PTC that largely overlaps with
dalfopristin.45,104 A second binding site at the entrance of the peptide exit tunnel has also been
suggested.44,105 The crystal structure of chloramphenicol bound to the ribosome reveals the importance of intermediary Mg2+ ions in antibiotic binding
to the PTC,104 a fact that has the potential to be
exploited in the development of new analogues. One
of the key interactions is between the primary alcohol
at C3 and a Mg2+ ion. The most prevalent mechanism
of resistance to chloramphenicol is via a series of
acetyltransferases (many of which are structurally
similar to the Vat proteins described above) that
modify this important OH group.106

5.2. Lincosamides
The lincosamide antibiotics also bind the ribosome
P site in a region that partially overlaps the chloramphenicol/dalfopristin and erythromycin binding sites.104
The first member of the class, lincomycin, was
discovered as a product of Streptomyces lincolnensis
in 1962,107 and the semisynthetic derivative, clindamycin (Figure 1), continues to find some clinical use,
especially in the treatment of infections caused by
anaerobic bacteria.108 Resistance to clindamycin occurs primarily via the Erm-mediated methylation of
the 23S rRNA as described above as part of the MLSB
phentotype (L ) lincosamide).

5.3. Pleuromutilins
The pleuromutilins (Figure 1) are fungal natural
products discovered over 50 years ago.109 They have
found use in agriculture and veterinary medicine but
as a result of poor pharmacological properties have
not been used to treat infections in humans (see
references in ref 110). The pleuromutilins bind to the
PTC as assessed by chemical footprinting studies.111
While at present these antibiotics are unsuitable for
clinical use, recent medicinal chemistry efforts to
improve stability to human cytochrome P-450s, which
results in rapid degradation of the compounds, has
been reported.110

6. Conclusions
Bacterial translation is reemerging as a front-line
target for the modern discovery of antibiotics as
evidenced by the fact that of the very few new
antibacterial agents to receive regulatory approval
in the past 5 years, three of these are translation
inhibitors: Synercid, linezolid, and the ketolide,
telithromycin. Furthermore, the availability of highresolution crystal structures of the bacterial ribosome
and the ability to solve co-structures with inhibitors
of translation now provides the opportunity to launch
structure-based initiatives to develop new agents that
block translation. At the same time, highly important
biochemical and molecular biological tools such as a
strain of E. coli with all of the rRNA genes inactivated112 will greatly facilitate characterization of
mode of action, dose dependency, and resistance. The
potential of using the ribosome as a platform for new
drug discovery is in fact being pursued by a number
of drug discovery companies. Investigation of the
biochemistry, structure, and clinical application of
inhibitors of bacterial translation is therefore poised
to enter a new golden era, mirroring the discovery
phase during the 1950s. The challenge of resistance
however will remain significant in these efforts.
Overcoming existing and eventual resistance will
require the ongoing and concerted efforts of medicinal
chemists working together with microbiologists, pharmacologists, structural biologists, and biochemists to
generate the next generation of translation inhibitors.

Streptogramins, Oxazolidinones, and Other Inhibitors

7. Acknowledgments
We thank Arif Mukhtar for assistance in preparing
Figure 2. Research in our lab on inhibitors of bacterial translation and associated resistance has been
supported by the Canadian Institutes for Health
Research. T. Mukhtar is support by an Ontario
Graduate Student Scholarship, and G. Wright is
supported by a Canada Research Chair in Antibiotic
Biochemistry.

8. References
(1) Knowles, D. J.; Foloppe, N.; Matassova, N. B.; Murchie, A. I.
Curr. Opin. Pharmacol. 2002, 2, 501.
(2) Bower, J.; Drysdale, M.; Hebdon, R.; Jordan, A.; Lentzen, G.;
Matassova, N.; Murchie, A.; Powles, J.; Roughley, S. Bioorg. Med.
Chem. Lett. 2003, 13, 2455.
(3) Barluenga, S.; Simonsen, K. B.; Littlefield, E. S.; Ayida, B. K.;
Vourloumis, D.; Winters, G. C.; Takahashi, M.; Shandrick, S.;
Zhao, Q.; Han, Q.; Hermann, T. Bioorg. Med. Chem. Lett. 2004,
14, 713.
(4) Foloppe, N.; Chen, I. J.; Davis, B.; Hold, A.; Morley, D.; Howes,
R. Bioorg. Med. Chem. 2004, 12, 935.
(5) Wilson, D. N.; Nierhaus, K. H. Angew. Chem., Int Ed. Engl. 2003,
42, 3464.
(6) Nissen, P.; Hansen, J.; Ban, N.; Moore, P. B.; Steitz, T. A. Science
2000, 289, 920.
(7) Gualerzi, C. O.; Brnadi, L.; Caserta, E.; La Teana, A.; Spurio,
R.; Tomsic, J.; Pon, C. L. In The Ribosome; Garrett, R. A.,
Douthwaite, S. R., Lijas, A., Metheson, A. T., Moore, P. B.,
Noller, H. F., Eds.; ASM Press: Washington, DC, 2000.
(8) Carter, A. P.; Clemons, W. M., Jr.; Brodersen, D. E.; MorganWarren, R. J.; Hartsch, T.; Wimberly, B. T.; Ramakrishnan, V.
Science 2001, 291, 498.
(9) Pioletti, M.; Schlunzen, F.; Harms, J.; Zarivach, R.; Gluhmann,
M.; Avila, H.; Bashan, A.; Bartels, H.; Auerbach, T.; Jacobi, C.;
Hartsch, T.; Yonath, A.; Franceschi, F. EMBO J. 2001, 20, 1829.
(10) Ogle, J. M.; Carter, A. P.; Ramakrishnan, V. Trends Biochem.
Sci. 2003, 28, 259.
(11) Rodnina, M. V.; Pape, T.; Fricke, R.; Kuhn, L.; Wintermeyer,
W. J. Biol. Chem. 1996, 271, 646.
(12) Pape, T.; Wintermeyer, W.; Rodnina, M. V. EMBO J. 1998, 17,
7490.
(13) OConnor, M.; Dahlberg, A. E. J. Mol. Biol. 1995, 254, 838.
(14) Stark, H.; Rodnina, M. V.; Rinke-Appel, J.; Brimacombe, R.;
Wintermeyer, W.; van Heel, M. Nature 1997, 389, 403.
(15) Pape, T.; Wintermeyer, W.; Rodnina, M. EMBO J. 1999, 18,
3800.
(16) Rodnina, M. V.; Daviter, T.; Gromadski, K.; Wintermeyer, W.
Biochimie 2002, 84, 745.
(17) Valle, M.; Sengupta, J.; Swami, N. K.; Grassucci, R. A.; Burkhardt,
N.; Nierhaus, K. H.; Agrawal, R. K.; Frank, J. EMBO J. 2002,
21, 3557.
(18) Agmon, I.; Auerbach, T.; Baram, D.; Bartels, H.; Bashan, A.;
Berisio, R.; Fucini, P.; Hansen, H. A.; Harms, J.; Kessler, M.;
Peretz, M.; Schluenzen, F.; Yonath, A.; Zarivach, R. Eur. J.
Biochem. 2003, 270, 2543.
(19) Green, R.; Lorsch, J. R. Cell 2002, 110, 665.
(20) Bateman, K. P.; Yang, K.; Thibault, P.; White, R. L.; Vining, L.
C. J. Am. Chem. Soc. 1996, 118, 5335.
(21) Pechere, J. C. Drugs 1996, 51, 13.
(22) Johnston, N. J.; Mukhtar, T. A.; Wright, G. D. Curr. Drug
Targets 2002, 3, 335.
(23) Kingston, D. G. I.; Kolpak, M. X.; LeFevre, J. W.; BorupGrochtmann, I. J. Am. Chem. Soc. 1983, 105, 5106.
(24) Thibaut, D.; Ratet, N.; Bisch, D.; Faucher, D.; Debussche, L.;
Blanche, F. J. Bacteriol. 1995, 177, 5199.
(25) Blanc, V.; Lagneaux, D.; Didier, P.; Gil, P.; Lacroix, P.; Crouzet,
J. J. Bacteriol. 1995, 177, 5206.
(26) Purvis, M. B.; Kingston, D. G. I.; Fujii, N.; Floss, H. G. J. Chem.
Soc., Chem. Commun. 1987, 302.
(27) Roy, R. S.; Gehring, A. M.; Milne, J. C.; Belshaw, P. J.; Walsh,
C. T. Nat. Prod. Rep. 1999, 16, 249.
(28) Cocito, C. Microbiol. Rev. 1979, 43, 145.
(29) Reed, J. W.; Kingston, D. G. J. Nat. Prod. 1986, 49, 626.
(30) Molinero, A. A.; Kingston, D. G.; Reed, G. H. J. Nat. Prod. 1989,
52, 99.
(31) Reed, J. W.; Purvis, M. B.; Kingston, D. G.; Biot, A.; Grossele,
F. J. Org. Chem. 1989, 54, 1161.
(32) Blanc, V.; Gil, P.; Bamas-Jacques, N.; Lorenzon, S.; Zagorec, M.;
Schleuniger, J.; Bisch, D.; Blanche, F.; Debussche, L.; Crouzet,
J.; Thibaut, D. Mol. Microbiol. 1997, 23, 191.

Chemical Reviews, 2005, Vol. 105, No. 2 541


(33) Namwat, W.; Kinoshita, H.; Nihira, T. J. Bacteriol. 2002, 184,
4811.
(34) Namwat, W.; Kamioka, Y.; Kinoshita, H.; Yamada, Y.; Nihira,
T. Gene 2002, 286, 283.
(35) Thibaut, D.; Bisch, D.; Ratet, N.; Maton, L.; Couder, M.;
Debussche, L.; Blanche, F. J. Bacteriol. 1997, 179, 697.
(36) de Crecy-Lagard, V.; Blanc, V.; Gil, P.; Naudin, L.; Lorenzon,
S.; Famechon, A.; Bamas-Jacques, N.; Crouzet, J.; Thibaut, D.
J. Bacteriol. 1997, 179, 705.
(37) de Crecy-Lagard, V.; Saurin, W.; Thibaut, D.; Gil, P.; Naudin,
L.; Crouzet, J.; Blanc, V. Antimicrob. Agents Chemother. 1997,
41, 1904.
(38) Yamada, Y.; Sugamura, K.; Kondo, K.; Yanagimoto, M.; Okada,
H. J. Antibiot. (Tokyo) 1987, 40, 496.
(39) Kawachi, R.; Akashi, T.; Kamitani, Y.; Sy, A.; Wangchaisoonthorn, U.; Nihira, T.; Yamada, Y. Mol. Microbiol. 2000, 36,
302.
(40) Miller, M. B.; Bassler, B. L. Annu. Rev. Microbiol. 2001, 55, 165.
(41) Horinouchi, S. Front. Biosci. 2002, 7, d2045.
(42) Choi, S. U.; Lee, C. K.; Hwang, Y. I.; Kinosita, H.; Nihira, T.
Arch. Microbiol. 2003, 180, 303.
(43) Vannuffel, P.; Cocito, C. Drugs 1996, 51, 20.
(44) Hansen, J. L.; Moore, P. B.; Steitz, T. A. J. Mol. Biol. 2003, 330,
1061.
(45) Harms, J. M.; Schluenzen, F.; Fucini, P.; Bartels, H.; Yonath,
A. E. BMC Biol. 2004, 2, 4.
(46) Vannuffel, P.; Di Giambattista, M.; Cocito, C. Nucleic Acids Res
1994, 22, 4449.
(47) Porse, B. T.; Kirillov, S. V.; Awayez, M. J.; Garrett, R. A. RNA
1999, 5, 585.
(48) Chinali, G.; Di Giambattista, M.; Cocito, C. Biochemistry 1987,
26, 1592.
(49) Cocito, C.; Di Giambattista, M.; Nyssen, E.; Vannuffel, P. J.
Antimicrob. Chemother. 1997, 39 Suppl A, 7.
(50) Bouanchaud, D. H. J. Antimicrob. Chemother. 1997, 39 (Suppl.
A), 15.
(51) Singh, K. V.; Weinstock, G. M.; Murray, B. E. Antimicrob. Agents
Chemother. 2002, 46, 1845.
(52) Ross, J. I.; Eady, E. A.; Cove, J. H.; Cunliffe, W. J.; Baumberg,
S.; Wootton, J. C. Mol. Microbiol. 1990, 4, 1207.
(53) Alignet, J.; Loncle, V.; el Sohl, N. Gene 1992, 117, 45.
(54) Alignet, J.; El Solh, N. Gene 1997, 202, 133.
(55) Weisblum, B. Antimicrob. Agents Chemother. 1995, 39, 577.
(56) Leclercq, R.; Courvalin, P. Antimicrob. Agents Chemother. 1991,
35, 1267.
(57) Tait-Kamradt, A.; Davies, T.; Appelbaum, P. C.; Depardieu, F.;
Courvalin, P.; Petitpas, J.; Wondrack, L.; Walker, A.; Jacobs,
M. R.; Sutcliffe, J. Antimicrob. Agents Chemother. 2000, 44, 3395.
(58) Sugantino, M.; Roderick, S. L. Biochemistry 2002, 41, 2209.
(59) Kehoe, L. E.; Snidwongse, J.; Courvalin, P.; Rafferty, J. B.;
Murray, I. A. J. Biol. Chem. 2003, 278, 29963.
(60) Hou, C. T.; Perlman, D.; Schallock, M. R. J. Antibiot. 1970, 23,
35.
(61) Kim, C. H.; Otake, N.; Yonehara, H. J. Antibiot. 1974, 27, 903.
(62) Le Goffic, F.; Capmau, M. L.; Abbe, J.; Cerceau, C.; Dublanchet,
A.; Duval, J. Ann. Microbiol. (Paris) 1977, 128B, 471.
(63) Allignet, J.; Loncle, V.; Mazodier, P.; el Solh, N. Plasmid 1988,
20, 271.
(64) Suzuki, N.; Lee, C. K.; Nihira, T.; Yamada, Y. Antimicrob. Agents
Chemother. 1998, 42, 2985.
(65) Spangler, S. K.; Jacobs, M. R.; Appelbaum, P. C. Antimicrob.
Agents Chemother. 1996, 40, 481.
(66) Pankuch, G. A.; Kelly, L. M.; Lin, G.; Bryskier, A.; Couturier,
C.; Jacobs, M. R.; Appelbaum, P. C. Antimicrob. Agents Chemother. 2003, 47, 3270.
(67) Bonfiglio, G.; Furneri, P. M. Expert Opin. Investig. Drugs 2001,
10, 185.
(68) Slee, A. M.; Wuonola, M. A.; McRipley, R. J.; Zajac, I.; Zawada,
M. J.; Bartholomew, P. T.; Gregory, W. A.; Forbes, M. Antimicrob. Agents Chemother. 1987, 31, 1791.
(69) Gregory, W. A.; Brittelli, D. R.; Wang, C. L.; Wuonola, M. A.;
McRipley, R. J.; Eustice, D. C.; Eberly, V. S.; Bartholomew, P.
T.; Slee, A. M.; Forbes, M. J. Med. Chem. 1989, 32, 1673.
(70) Barbachyn, M. R.; Ford, C. W. Angew. Chem., Int. Ed. 2003, 42,
2010.
(71) Genin, M. J.; Allwine, D. A.; Anderson, D. J.; Barbachyn, M. R.;
Emmert, D. E.; Garmon, S. A.; Graber, D. R.; Grega, K. C.;
Hester, J. B.; Hutchinson, D. K.; Morris, J.; Reischer, R. J.; Ford,
C. W.; Zurenko, G. E.; Hamel, J. C.; Schaadt, R. D.; Stapert, D.;
Yagi, B. H. J. Med. Chem. 2000, 43, 953.
(72) Barbachyn, M. R.; Hutchinson, D. K.; Brickner, S. J.; Cynamon,
M. H.; Kilburn, J. O.; Klemens, S. P.; Glickman, S. E.; Grega,
K. C.; Hendges, S. K.; Toops, D. S.; Ford, C. W.; Zurenko, G. E.
J. Med. Chem. 1996, 39, 680.
(73) Tucker, J. A.; Allwine, D. A.; Grega, K. C.; Barbachyn, M. R.;
Klock, J. L.; Adamski, J. L.; Brickner, S. J.; Hutchinson, D. K.;
Ford, C. W.; Zurenko, G. E.; Conradi, R. A.; Burton, P. S.; Jensen,
R. M. J. Med. Chem. 1998, 41, 3727.

542 Chemical Reviews, 2005, Vol. 105, No. 2


(74) Brickner, S. J.; Hutchinson, D. K.; Barbachyn, M. R.; Manninen,
P. R.; Ulanowicz, D. A.; Garmon, S. A.; Grega, K. C.; Hendges,
S. K.; Toops, D. S.; Ford, C. W.; Zurenko, G. E. J. Med. Chem.
1996, 39, 673.
(75) Barbachyn, M. R.; Toops, D. S.; Grega, K. C.; Hendges, S. K.;
Ford, C. W.; Zurenko, G. E.; Hamel, J. C.; Schaadt, J. D.; Stapert,
D.; Yagi, B. H.; Buysse, J. M.; Demyan, W. F.; Kilburn, J. O.;
Glickman, S. E. Bioorg. Med. Chem. Lett. 1996, 6, 1009.
(76) Barrett, J. F. Curr. Opin. Investig. Drugs 2000, 1, 181.
(77) Lin, A. H.; Murray, R. W.; Vidmar, T. J.; Marotti, K. R.
Antimicrob. Agents Chemother. 1997, 41, 2127.
(78) Zhou, C. C.; Swaney, S. M.; Shinabarger, D. L.; Stockman, B. J.
Antimicrob. Agents Chemother. 2002, 46, 625.
(79) Shinabarger, D. L.; Marotti, K. R.; Murray, R. W.; Lin, A. H.;
Melchior, E. P.; Swaney, S. M.; Dunyak, D. S.; Demyan, W. F.;
Buysse, J. M. Antimicrob. Agents Chemother. 1997, 41, 2132.
(80) Swaney, S. M.; Aoki, H.; Ganoza, M. C.; Shinabarger, D. L.
Antimicrob. Agents Chemother. 1998, 42, 3251.
(81) Patel, U.; Yan, Y. P.; Hobbs, F. W., Jr.; Kaczmarczyk, J.; Slee,
A. M.; Pompliano, D. L.; Kurilla, M. G.; Bobkova, E. V. J. Biol.
Chem. 2001, 276, 37199.
(82) Thompson, J.; OConnor, M.; Mills, J. A.; Dahlberg, A. E. J. Mol.
Biol. 2002, 322, 273.
(83) Bobkova, E. V.; Yan, Y. P.; Jordan, D. B.; Kurilla, M. G.;
Pompliano, D. L. J. Biol. Chem. 2003, 278, 9802.
(84) Tsiodras, S.; Gold, H. S.; Sakoulas, G.; Eliopoulos, G. M.;
Wennersten, C.; Venkataraman, L.; Moellering, R. C.; Ferraro,
M. J. Lancet 2001, 358, 207.
(85) Wilson, P.; Andrews, J. A.; Charlesworth, R.; Walesby, R.; Singer,
M.; Farrell, D. J.; Robbins, M. J. Antimicrob. Chemother. 2003,
51, 186.
(86) Johnson, A. P.; Tysall, L.; Stockdale, M. V.; Woodford, N.;
Kaufmann, M. E.; Warner, M.; Livermore, D. M.; Asboth, F.;
Allerberger, F. J. Eur. J. Clin. Microbiol. Infect. Dis. 2002, 21,
751.
(87) Prystowsky, J.; Siddiqui, F.; Chosay, J.; Shinabarger, D. L.;
Millichap, J.; Peterson, L. R.; Noskin, G. A. Antimicrob. Agents
Chemother. 2001, 45, 2154.
(88) Kloss, P.; Xiong, L.; Shinabarger, D. L.; Mankin, A. S. J. Mol.
Biol. 1999, 294, 93.
(89) Xiong, L.; Kloss, P.; Douthwaite, S.; Andersen, N. M.; Swaney,
S.; Shinabarger, D. L.; Mankin, A. S. J. Bacteriol. 2000, 182,
5325.
(90) Jones, R. N.; Della-Latta, P.; Lee, L. V.; Biedenbach, D. J. Diagn.
Microbiol. Infect. Dis. 2002, 42, 137.
(91) Rahim, S.; Pillai, S. K.; Gold, H. S.; Venkataraman, L.; Inglima,
K.; Press, R. A. Clin. Infect. Dis. 2003, 36, E146.
(92) Sakoulas, G.; Gold, H. S.; Venkataraman, L.; Moellering, R. C.,
Jr.; Ferraro, M. J.; Eliopoulos, G. M. J. Antimicrob. Chemother.
2003, 51, 1039.

Mukhtar and Wright


(93) Meka, V. G.; Pillai, S. K.; Sakoulas, G.; Wennersten, C.;
Venkataraman, L.; DeGirolami, P. C.; Eliopoulos, G. M.; Moellering, R. C., Jr.; Gold, H. S. J. Infect. Dis. 2004, 190, 311.
(94) Meka, V. G.; Gold, H. S.; Cooke, A.; Venkataraman, L.; Eliopoulos, G. M.; Moellering, R. C., Jr.; Jenkins, S. G. J. Antimicrob.
Chemother. 2004.
(95) Willems, R. J.; Top, J.; Smith, D. J.; Roper, D. I.; North, S. E.;
Woodford, N. Antimicrob. Agents Chemother. 2003, 47, 3061.
(96) Sander, P.; Belova, L.; Kidan, Y. G.; Pfister, P.; Mankin, A. S.;
Bottger, E. C. Mol. Microbiol. 2002, 46, 1295.
(97) Singh, U.; Raju, B.; Lam, S.; Zhou, J.; Gadwood, R. C.; Ford, C.
W.; Zurenko, G. E.; Schaadt, R. D.; Morin, S. E.; Adams, W. J.;
Friis, J. M.; Courtney, M.; Palandra, J.; Hackbarth, C. J.; Lopez,
S.; Wu, C.; Mortell, K. H.; Trias, J.; Yuan, Z.; Patel, D. V.;
Gordeev, M. F. Bioorg. Med. Chem. Lett. 2003, 13, 4209.
(98) Sbardella, G.; Mai, A.; Artico, M.; Loddo, R.; Setzu, M. G.; La
Colla, P. Bioorg. Med. Chem. Lett. 2004, 14, 1537.
(99) Gordeev, M. F.; Hackbarth, C.; Barbachyn, M. R.; Banitt, L. S.;
Gage, J. R.; Luehr, G. W.; Gomez, M.; Trias, J.; Morin, S. E.;
Zurenko, G. E.; Parker, C. N.; Evans, J. M.; White, R. J.; Patel,
D. V. Bioorg. Med. Chem. Lett. 2003, 13, 4213.
(100) Hubschwerlen, C.; Specklin, J. L.; Sigwalt, C.; Schroeder, S.;
Locher, H. H. Bioorg. Med. Chem. 2003, 11, 2313.
(101) Hubschwerlen, C.; Specklin, J. L.; Baeschlin, D. K.; Borer, Y.;
Haefeli, S.; Sigwalt, C.; Schroeder, S.; Locher, H. H. Bioorg. Med.
Chem. Lett. 2003, 13, 4229.
(102) Ehrlich, J.; Bartz, Q. R.; Smith, R. M.; Joslyn, D. A.; Burkholder,
P. R. Science 1947, 106, 417.
(103) Yunis, A. A. Am. J. Med. 1989, 87, 44N.
(104) Schlunzen, F.; Zarivach, R.; Harms, J.; Bashan, A.; Tocilj, A.;
Albrecht, R.; Yonath, A.; Franceschi, F. Nature 2001, 413, 814.
(105) Long, K. S.; Porse, B. T. Nucleic Acids Res. 2003, 31, 7208.
(106) Murray, I. A.; Shaw, W. V. Antimicrob. Agents Chemother. 1997,
41, 1.
(107) Mason, D. J.; Deietz, A.; DeBoer, C. In Antimicrobial Agents and
Chemotherapy-1962; Sylvester, J. C., Ed.; American Society for
Microbiology: Ann Arbor, MI, 1963.
(108) Kasten, M. J. Mayo Clin. Proc. 1999, 74, 825.
(109) Kavanigh, F.; Hervey, A.; Robbins, W. J. Proc. Natl. Acad. Sci.
U.S.A. 1951, 37.
(110) Springer, D. M.; Sorenson, M. E.; Huang, S.; Connolly, T. P.;
Bronson, J. J.; Matson, J. A.; Hanson, R. L.; Brzozowski, D. B.;
LaPorte, T. L.; Patel, R. N. Bioorg. Med. Chem. Lett. 2003, 13,
1751.
(111) Poulsen, S. M.; Karlsson, M.; Johansson, L. B.; Vester, B. Mol.
Microbiol. 2001, 41, 1091.
(112) Asai, T.; Zaporojets, D.; Squires, C.; Squires, C. L. Proc. Natl.
Acad. Sci. U.S.A. 1999, 96, 1971.

CR030110Z

Chem. Rev. 2005, 105, 543558

543

Genetic Approaches to Polyketide Antibiotics. 1


Robert McDaniel,* Mark Welch, and C. Richard Hutchinson
Kosan Biosciences, 3832 Bay Center Place, Hayward, California 94545
Received June 18, 2004

Contents
1. Introduction
2. Biochemistry and Genetics of Polyketide
Biosynthesis
3. Polyketide Antibiotic Gene Clusters
3.1. Fourteen-Membered Macrolides
3.2. Sixteen-Membered Macrolides
3.3. Ansamycins
4. Overview of Methodologies
4.1. Modular PKSs
4.2. Aromatic PKSs
5. New Developments in Modular PKS Manipulation
5.1. Modeling and Engineering of PKS Domains
5.1.1. Acyltransferases
5.1.2. Ketoreductases
5.2. Type II Thioesterase
5.3. Cyclization/Termination
5.4. Intermodular and Intramodular
Communication
5.5. Precursor Engineering
5.6. Tailoring Pathways
6. Creating and Improving Microbial Production
Systems
7. Conclusions
8. Acknowledgments
9. References

543
544
546
546
547
549
552
552
557
553
553
553
553
554
554
554
555
555
556
556
557
0

1. Introduction
Microbial metabolites have for decades been a rich
source of natural product drug leads and therapeutically important drugs.1,2 The terms engineered biosynthesis and combinatorial biosynthesis encompass techniques aimed at increasing chemical diversity
of natural products by altering the function of the
genes and enzymes that govern the production of
these metabolites. The classes of compounds most
frequently associated with this are polyketides and
nonribosomal peptides.3 In particular, the predictable
relationship between the structure and function of
the modular type of microbial polyketide synthases
(PKSs) has enabled genetic manipulation of the
biosynthetic pathways for production of novel variants of classes of naturally occurring compounds,
such as macrolide antibiotics4,5 and antitumor compounds.6 The goals of the approach resemble those
* To whom correspondence should be addressed. E-mail:
robert.mcdaniel@codexis.com.

of medicinal chemists who synthesize analogues and


derivatives of lead compounds in an attempt to
improve upon existing drugs or find new ones.
Expression of native or engineered PKS genes, as
well as those that govern precursor supply and postPKS modification of the metabolite, in heterologous
hosts is also an important aspect of developing
commercial systems for drug production. The following review presents an overview of the initial technology enabling studies pertaining to polyketide
manipulation (which have been covered in several
previous reviews4-11) and a comprehensive review of
the more recent advances that have set the stage for
broader use of the technology. Although this review
focuses on antibacterial polyketides, engineered or
combinatorial biosynthesis extends to other classes
of polyketides of therapeutic interest including antitumor, immunosuppressive, antifungal, and other
compounds.
Polyketides are a notable class of natural products
with a number of well-established successes in clinical and agricultural applications. Examples include
the antibiotics erythromycin, tylosin, rifamycin, and
the tetracyclines, immunosuppressants FK506 and
rapamycin, and antitumor agents doxorubicin and
mithramycin. Two early seminal works led to the
promise that the genes for secondary metabolites
could be used to create novel structures. The first was
by Hopwood and co-workers12,13 in which genes from
two different antibiotic gene clusters, medermycin or
granaticin with actinorhodin, were used in combination for the first time to produce a hybrid aromatic
polyketide. The second was work performed by Leadlay and co-workers at Cambridge14 and by Katz and
co-workers at Abbot Laboratories,15 who uncovered
the first genes for a modular (type I) PKS in the
erythromycin gene cluster. These megasynthases
have since been the primary focus of polyketide
engineering, and many of the proof of concept experiments have been performed using the erythromycin
PKS.
Modular PKSs make many large and complex
natural products that are difficult to synthesize or
modify by chemistry. Since the discovery of the
erythromycin PKS, on the order of a few hundred
new compounds have been created through modification or heterologous expression of polyketide gene
clusters. These numbers suggest that, in its current
state, polyketide engineering is best suited for optimization of existing lead compounds by specifically
chosen structural modifications rather than random

10.1021/cr0301189 CCC: $53.50 2005 American Chemical Society


Published on Web 01/25/2005

544 Chemical Reviews, 2005, Vol. 105, No. 2

Robert McDaniel is a Senior Scientist at Kosan Biosciences, Inc. and


has worked extensively in the field of polyketide biosynthesis and
engineering for more than 12 years. He received his B.S. degree in
Chemical Engineering from the University of Colorado at Boulder and
M.S. and Ph.D. degrees in Chemical Engineering at Stanford University.
At Stanford he began his career in polyketide research in the laboratory
of Professor Chaitan Khosla, collaborating with Professor Sir David
Hopwood at the John Innes Institute. He then joined Kosan during its
genesis in order to transfer and further develop polyketide engineering
technology that had been developed at Stanford. He has worked at Kosan
for the past 8 years on several projects, including anti-infectives, and has
published several research articles on the subject. His current interests
are focused on ways to improve polyketide production from natural and
engineered polyketide producing microorganisms.

Mark Welch has over 10 years experience in RNA and protein engineering
with particular emphasis on directed evolution methods to create novel
biocatalysts. Mark received his B.A. degree in Biology from the University
of California at Santa Cruz with College Honors and Highest Honors in
the Major. He received his doctoral degree in Molecular, Cellular, and
Developmental Biology from the University of Colorado at Boulder in 1996
for studies of the role of 23SrRNA in the peptidyl transferase activity of
the ribosome in the laboratory of Dr. Michael Yarus. In 1998 Mark joined
Maxygen, Inc., a biotechnology company specializing in the application
of DNA shuffling methods for biomolecule-directed evolution. As a scientist
in the New Technology group at Maxygen, Mark managed an interdisciplinary team responsible for development and application of novel highthroughput assays to screen for a variety of target enzyme activities. He
joined the New Technology group at Kosan in April 2003 and is currently
leading a group focused of novel PKS engineering.

generation of large libraries for screening against


targets. Some of the applications for which this
technology is used at Kosan include the following: (i)
production of analogues of natural products to improve activity or pharmacodynamics; (ii) introducing
reactive groups into compounds for further chemical
modification; (iii) generating intermediates for chemical synthesis of natural products; and (iv) increasing

McDaniel et al.

C. Richard (Dick) Hutchinson is a Research Fellow at Kosan Biosciences,


Hayward, CA, following 3 years as the Vice President of New Technologies
after joining the company in March 2000. He spent 26 years at the
University of Wisconsin, Madison, as a Professor of Medicinal Chemistry
and Bacteriology and is now a Professor Emeritus. Prior to this he was
on the faculty of the School of Pharmacy, University of Connecticut (1970
74), conducted postdoctoral research with Sir Alan Battersby (1971), and
received his Ph.D. degree in Organic Chemistry from the University of
Minnesota (1970), working with Edward Leete, and B.S. degree in
Pharmacy from Ohio State University (1966), where he carried out
undergraduate research with Raymond Doskotch and Jack Beal. His
research has involved studies of the biosynthesis of natural products with
special emphasis on the molecular genetics and biochemistry of antibiotic
production in microorganisms. He has published over 235 scientific papers
and patents and received several prestigious awards, including Guggenheim and Fullbright fellowships, the Charles Thom Research Achievement
Award of the Society for Industrial Microbiology, the AACP Paul Dawson
Biotechnology Award, the Research Achievement Award of the American
Society for Pharmacognosy, and a Distinguished Alumni Award from Ohio
State University.

the availability or reducing the cost of natural


products (overproduction). Although the architecture
of PKSs suggests that very large libraries could be
theoretically generated, the challenges to doing this
in the laboratory require a further understanding of
PKS structure, specificity, and protein interactions
as well as technologies to perform genetic manipulation more efficiently.
Other approaches to discovering new drugs from
natural products are not discussed here but typically
rely on high-throughput DNA sequencing. For example, novel sets of secondary metabolism genes are
found by either whole genome sequencing or DNA
libraries enriched for secondary metabolite genes and
subsequently expressed through manipulation of
growth conditions or in heterologous hosts. DNA
libraries generated from unculturable organisms or
environmental samples can also be screened in this
manner or cloned directly into heterologous hosts and
screened for biological activity. Finally, the structural
modification of aromatic polyketides, made by socalled type II PKSs, is less amenable to gene manipulation and therefore is not reviewed here, except
to note recent advances in understanding this area.

2. Biochemistry and Genetics of Polyketide


Biosynthesis
Polyketide synthases evolved the capability of
making a vast number of compounds from the same
classes of substrates used by fatty acid synthases and

Genetic Approaches to Polyketide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 545

Figure 1. Illustration of the mechanism of the type II PKS involved in the biosynthesis of tetracenomycin F1 and C. The
PKS consists of individual protein subunits that act in concert to assemble the acetate starter unit (produced by
decarboxylation of enzyme-bound malonate) and nine chain extender units into a TcmM-bound decaketide by an iterative
process involving a malonyl-CoA:ACP acyltransferase (MCAT, which is shared with fatty acid biosynthesis) and the proteins
TcmJ, TcmK, TcmL, and TcmM. The decaketide is cyclized to Tcm F2 by the TcmN enzyme, with assistance by TcmJ;
then Tcm F2 is cyclized once more by TcmI to form Tcm F1. The latter intermediate is converted to tetracenomycin C by
tailoring enzymes.

Figure 2. Illustration of the mechanism of the type I modular PKS involved in the biosynthesis of 6dEB. Each of the
DEBS subunits is represented by a broad, open arrow containing the relevant domains in each module. Key enzymebound intermediates of carbon-chain assembly are shown bound to the ACP domains. Assembly begins at the loading
didomain of the first DEBS subunit upon attachment of propionate which then reacts with ACP-bound 2-methylmalonate,
obtained from its CoA ester. Further equivalents of 2-methylmalonyl-CoA are used by DEBS to produce 6dEB, as explained
in the text. 6dEB is then converted to the erythromycin A-D glycosides by tailoring enzymes.

largely by the same type of biochemistry. Examples


of two types of PKSs and their associated polyketides
are illustrated in Figures 1 and 2. Type II, or
aromatic, PKSs (Figure 1) consist of a collection of
largely monofunctional proteins that catalyze the
formation of typically polycyclic aromatic compounds,
usually from acetate and malonate only. Type I, or
modular, PKSs (Figure 2), in contrast, use large
multifunctional proteins to make polyoxygenated,

aliphatic compounds from several different kinds of


acyl-Coenzyme A substrates.
The so-called modular PKSs15-17 have led to the
most fruitful genetic engineering route to structural
variants18,19 of polyketides that are important therapeutic drugs, like the antibacterial erythromycin A
(Figure 2) or experimental agents such as 17-allylamino-17-demethoxygeldanamycin (17-AAG) that currently is undergoing clinical trials as an antitumor

546 Chemical Reviews, 2005, Vol. 105, No. 2

drug. [Two other types of PKSs, the fungal nonmodular type I and plant chalcone type III, are not
discussed here; for reviews, cf. refs 20 and 21.] A
modular PKS is a massive complex of large, multifunction proteins. Within each protein are one or
more modules, each with different combinations of
domains that function like the constituent biochemical activities of fatty acid synthases to catalyze a
single cycle of polyketide chain elongation and modification. 6-Deoxyerythronolide B synthase (DEBS) is
the PKS that forms the backbone of the erythromycins and is encoded by the three genes, eryAIIII14,15,22 (Figure 2). DEBS catalyzes formation of
6-deoxyerythronolide B (6dEB) by the successive
condensation of one propionyl and six 2-methylmalonyl molecules in their activated Coenzyme A (CoA)
thioester form. Each of the three subunits of DEBS
have two extender modules, containing the activities
needed for one cycle of polyketide chain elongation,
as illustrated by the structures of the six enzymebound intermediates in Figure 2. In addition, the first
module is preceded by a loading didomain for the
starter unit, and the last is followed by a thioesterase
domain for product release and cyclization. Every
extender module contains a ketosynthase (KS), an
acyltransferase (AT), and an acyl carrier protein
(ACP) domain that together catalyze a two-carbon
extension of the chain. In DEBS, the AT domains of
extender modules are specific for 2-methylmalonylCoA, while the AT in the loading module uses
propionyl-CoA. After each two-carbon unit condensation, the oxidation state of the -carbon is either
retained as a ketone (module 3) or modified to a
hydroxyl, methenyl, or methylene group by the
presence of a ketoreductase (KR) (module 2), a KR
+ a dehydratase (DH), or a KR + DH + an enoyl
reductase (ER) (module 4), respectively.
In effect, the AT specificity and the types of
catalytic domains within a module serve as codes for
the structure of each two-carbon unit; the order of
the modules in a PKS specifies the sequence of the
distinct two-carbon units, and the number of modules
determines the length of the polyketide chain. Variations in the acyl-CoA substrates used by a modular
PKS, the number of domains within a module, and
the number of modules in the PKS are responsible
for establishing the first set of structural characteristics of the polyketide, including the chirality of
hydroxyl- and alkyl-bearing carbon centers. After
this, the kinds of biochemical transformations the
compound produced by the PKS undergoes, such as
glycosylation or oxidation, are dictated by the tailoring enzymes that establish the final structure.
Consequently, engineering a microorganism to produce novel polyketides can involve altering only the
PKS genes or the tailoring genes as well.

3. Polyketide Antibiotic Gene Clusters


3.1. Fourteen-Membered Macrolides
The 14-membered macrolides are exemplified by
erythromycin, for which the complete gene cluster
has been sequenced.14,15,22-28 The complete or partial
gene clusters for oleandomycin,29-34 megalomicin,35

McDaniel et al.

and picromycin36 are also known (Figure 3). The


overall DEBS-like architecture of the PKS genes is
conserved with the exception of the picromycin PKS
(PicPKS) (Figure 4), in which the last two modules
are contained on two separate proteins. Both DEBS
and the megalomicin PKS (MegPKS) produce 6dEB
and have high sequence similarity.30 The oleandomycin PKS (OlePKS) produces 8,8a-deoxyoleandolide, which is equivalent to 6dEB derived from a
two-carbon starter unit rather than a three-carbon
starter unit. An important distinction between DEBS
(and MegPKS) and OlePKS is the composition and
mechanism of their loading domains. In DEBS, an
AT loads propionyl-CoA (as well as other acyl CoAs
with lower efficiency) whereas the OlePKS loading
domain contains an AT specific for malonyl-CoA and
a KSq domain, an inactive condensation domain that
serves to decarboxylate the ACP-bound substrate to
acetyl-ACP for use as the starter.30,37
The PicPKS differs from the other three PKSs in
several ways. The loading domain of the PicPKS also
contains a KSq domain but an AT that is specific for
methylmalonyl-CoA, which is then decarboxylated to
propionyl-CoA for use as the starter. Module 2 of the
PicPKS extends using a malonyl-CoA rather than a
methylmalonyl-CoA unit and also contains a DH
domain in addition to the KR -keto modifying
activity leading to the C-10/C-11 alkene. Finally,
module 6 lacks a KR, resulting in the 3-ketone of
narbonolide. The PicPKS also produces approximately stoichiometric amounts of the 12-membered
lactone precursor of methymycin, methynolide, via
a mechanism referred to as domain skipping.10,36,38,39
All four PKSs have been expressed heterologously
in Streptomyces coelicolor and/or Streptomyces lividans and their corresponding aglycones (1-3) produced in good yield.30,35,38,40 The MegPKS contains
two regions longer than 0.5 kb with 100% sequence
identity in modules 2 and 6, leading to plasmid
instability caused by homologous recombination during plasmid-borne heterologous expression.41 Similarly, OlePKS contains over 1.2 kb of identical
sequence in modules 2 and 5. Neither DEBS nor
PicPKS possess such repeats.
The polyketide macrolactones are further modified
by a series of glycosylation(s) and oxidation(s). Synthesis of 6dEB is followed by C6-hydroxylation by the
product of eryF to yield erythronolide B. Addition of
the sugar L-mycarose (via thymidine diphospho(TDP)-mycarose) yields 3-O-R-mycarosylerythronolide B, and the addition of D-desosamine (via TDPdesosamine) yields erythromycin D. The two sugars
are produced by independent sets of genes designated
eryB (mycarose) and eryC (desosamine), which flank
the PKS genes in the cluster. The final steps of
erythromycin biosynthesis are hydroxylation of erythromycin D to yield erythromycin C by a second P450
enzyme encoded by eryK and O-methylation of the
mycarosyl residue by the eryG product to yield the
cladinosyl moiety of erythromycin A and B. Megalomicin tailoring resembles that of erythromycin except
that a third deoxysugar, L-megosamine, is added to
the C-6 hydroxyl and mycarose is acylated at C-3
and C-4. The set of genes for both TDP-L-mego-

Genetic Approaches to Polyketide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 547

Figure 3. Fourteen-membered macrolide antibiotics and gene clusters.

samine formation (megD) and the O-acyl transferase


(megY) have been identified in the meg cluster and
expressed in the erythromycin-producing strain, Saccharopolyspora erythraea, to generate megalomicins
in that strain.35
Oleandomycin contains an epoxide at C-8,C8a,
introduced by OleP, a P-450 oxidase. The oleP gene
has been coexpressed with DEBS to generate 8-hydroxy derivatives of 6dEB and 3-O-glycosylated 6dEB
derivatives.30,42 The two deoxysugars added to oleandolide are D-desosamine and L-oleandrose (3-Omethyl-L-olivose). The genes encoding TDP-L-olivose
biosynthesis (oleW, oleV, oleL, oleS, oleE, and oleU)
and its conversion to L-oleandrose (oleY) flank the
OlePKS genes. These genes were used to generate
3-O-olivosyl and 3-O-oleandrosyl erythronolide B
analogues34 (see below). The genes encoding D-desosamine biosynthesis are also located within the ole
cluster and homologous to those in the ery, meg, and
pic clusters.
Picromycin and methymycin contain only the single
deoxysugar, desosamine, encoded by the des genes.
Gene knockouts of desVI,43 encoding the N-methyltransferase, desV,44 encoding a transaminase, and
desI,45 encoding a putative C-4 dehydrase, resulted
in production of methymycin analogues with altered

glycoside moieties. Attachment of desosamine to both


the 12- and 14-membered lactones is performed by
the same glycosyl transferase, encoded by desVII.
After glycosylation, PicK (also called PicC), a P-450
oxidase, hydroxylates C-12 (picromycin) or C-10
(methymycin).46,47

3.2. Sixteen-Membered Macrolides


The 16-membered macrolide antibiotics characterized thus far fall into three different classes of
polyketide backbones. These are represented by
tylactone (4), platenolide (5), and mycinamicin lactone (6)/chalcolactone (7) (Figure 4), the polyketide
components of tylosin, niddamycin/spiramycin, and
mycinamicin/chalcomycin, respectively (Figure 5).
The PKSs encoding each have been sequenced.48-51
The organization of modules within subunits is
conserved across all of the PKSs (Figure 4), comprised of the loading domain, module 1 and module
2 on the first subunit, module 3 on the second
subunit, modules 4 and 5 on the third subunit,
module 6 on the fourth subunit, and module 7 on the
fifth subunit.
The chief difference among the 16-membered macrolide PKSs is the composition of extender units

548 Chemical Reviews, 2005, Vol. 105, No. 2

McDaniel et al.

Figure 4. Organization of 14- and 16-membered macrolide PKSs and polyketide products.

utilized. Mycinamicin lactone and chalcolactone are


derived exclusively from malonyl-CoA and methylmalonyl-CoA and differ only in the starter unit and
C-3 functionality. Modules 5 of the tylactone and the
platenolide PKSs utilize a 2-ethylmalonyl-CoA extender unit. A crotonyl-CoA reductase gene, ccr, is
clustered with the tylosin biosynthetic genes (Figure
5) and presumably contributes to the flux of ethyl-

malonyl-CoA synthesis from four-carbon acyl-CoA


pools during tylosin production.50 Homologues of ccr
are present in other PKS gene clusters requiring
ethylmalonyl-CoA, and a homologue from Streptomyces collinus was expressed in S. erythraea containing a modified DEBS construct, which then incorporated ethylmalonyl-CoA at module 4.52 In addition
to the other three precursors, the platenolide PKS

Genetic Approaches to Polyketide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 549

Figure 5. Sixteen-membered macrolide antibiotics and gene clusters.

also incorporates a 2-methoxymalonyl extender unit


at module 6. Genes encoding the biosynthetic pathway of methoxymalonyl-ACP have been found in the
FK520,53 ansamitocin,53 and geldanamycin gene clusters54 (see below) and have been used to produce this
precursor in other hosts.55 Presumably, these sets of
genes exist in the unsequenced regions of the 16membered macrolide gene clusters that produce
platenolide.
The genes for the biosynthetic pathways for all
three deoxysugars attached to tylosin, D-mycaminose,
L-mycarose, and D-mycinose (added as D-allose and
modified to mycinose after attachment), have been
sequenced from S. fradiae56 (Figure 5). The tylI and
tylH genes, both encoding P-450 oxidases, are responsible for the oxidations that occur on the C-6
ethyl and C-14 methyl branches of tylosin, respectively. Chalcomycin is one of the few macrolides that
does not contain an amino-sugar at C-5 but rather

has the neutral deoxysugar, D-chalcose, at that position (Figure 5). The putative genes for TDP-Dchalcose formation have been identified in the chalcomycin51 and lankamycin57 gene clusters. The genes
for the remaining chalcomycin modifications, including TDP-mycinose biosynthesis/attachment and the
two P-450s which hydroxylate C-8 and oxygenate
C-12/C-13, have also been sequenced in the chalcomycin gene cluster.51

3.3. Ansamycins
The ansamycins are related to the macrolides
biosynthetically but differ in the choice of starter unit
(3-amino-5-hydroxybenzoic acid (AHBA)) and the lack
of glycosylation. Formation of a macrolactam between
the terminal carboxyl and 3-amino group of AHBA,
instead of a macrolactone as above, results in a
characteristic basket with handle molecular conformation.58 Rifamycin, geldanamycin, herbimycin,

550 Chemical Reviews, 2005, Vol. 105, No. 2

McDaniel et al.

Figure 6. Ansamycin family of polyketides and gene clusters.

and the ansamitocins (Figure 6) are ansamycins for


which the biosynthetic genes have been cloned and
characterized. Derivatives of rifamycin are widely
used in the treatment of tuberculosis, whereas geldanamycin analogues are undergoing clinical trials as
antitumor drugs. The ansamitocins and closely related maytansine (found in plant extracts) are among
the most cytotoxic polyketides known and are usually
referred to as maytansinoids.59
Investigations of the rifamycin genes began with
studies of the biosynthesis of AHBA at the enzymatic
level, following the proof through isotopic labeling
experiments that AHBA is the precursor of the
studied ansamycins. The rifK gene was cloned by
reverse genetics using information about the amino
acid sequence of the AHBA synthase, RifK,60 and
then used to obtain the remainder of the AHBA
biosynthesis, PKS and tailoring enzyme genes, by
gene cloning and sequencing experiments.61 Some of

these genes were cloned independently by researchers at Novartis.62 A notable feature of the modular
rifamycin PKS (RifPKS) is its tendency to shed the
enzyme-bound PKS intermediates spontaneously,63,64
quite unlike the macrolide PKSs. This directly demonstrates the processivity of the RifPKS and shows
that the 9,10-dihydronaphthoquinone ring of rifamycin is formed during this process to produce proansamycin X (8, Figure 7). Largely oxidative tailoring
reactions remodel proansamycin X via rifamycin W
(not shown) into the characteristic ansamycin framework of rifamycin B.65 Initial attempts to modify the
function of the RifPKS by the same type of domain
inactivation experiments used for DEBS and other
macrolactone PKSs (see below) have been greatly
complicated by the spontaneous shedding of truncated PKS assembly intermediates (Y. Doi-Katayama, Y. J. Yoon, C. S. Park, and C. R. Hutchinson,
unpublished work).

Genetic Approaches to Polyketide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 551

Figure 7. Organization of ansamycin PKSs and polyketide products.

In the ansamitocin gene cluster three of the AHBA


biosynthesis genes are separated from the rest of the
asm genes by at least 30 kb of intervening DNA.59,66
Disruption of many of the tailoring enzyme genes has
been used to develop a picture of the sequence of
biosynthetic steps,67 and the substrate specificity of
the Asm19 O-acyltransferase has been defined through
studies of the purified enzyme.68 This enzyme distinguishes the ansamitocins (3-O-acylesters) from
maytansine (3-O-(N-acetyl-N-methyl-L-alanine ester).
As noted above, the asm cluster also contains a set
of five genes for formation of 2-methoxymalonate, one
of the substrates used by the asm PKS for chain
elongation. In fact, studies of these genes along with

their orthologs from the FK520-producing streptomycete have elucidated what is known about the
biosynthesis of this atypical substrate.55,69
Geldanamycin and herbimycin are made by distinct Streptomyces hygroscopicus strains, but their
biosynthesis and gene clusters are nearly identical.
Rascher et al.54 described a major portion of the
geldanamycin genes, which, like those of ansamitocin
biosynthesis, occur as two separate clusters. One
cluster (Figure 6) contains the PKS and tailoring
enzyme genes plus one AHBA biosynthesis gene,
gdmO; the rest of the AHBA biosynthesis genes lie
somewhere else in the chromosome yet are essential
for geldanamycin biosynthesis, but not primary me-

552 Chemical Reviews, 2005, Vol. 105, No. 2

tabolism, because their disruption abolishes geldanamycin production (A. Rascher and C. R. Hutchinson,
unpublished results). The corresponding herbimycin
PKS gene cluster is nearly identical, lacking only the
gdmF (amide synthase) and gdmM (monooxygenase)
orthologs (A. Rascher, Z. Hu, and C. R. Hutchinson,
unpublished results). Of the gdm tailoring genes,
gdmM and gdmN have been characterized functionally by gene knockout experiments54 (A. Rascher and
C. R. Hutchinson, unpublished results). Extensive
engineering of the geldanamycin PKS (GdmPKS)
genes (Figure 7), aided by development of convenient
methods for their modification and expression from
integrative vectors, has provided several geldanamycin analogues.70 Some of these are undergoing evaluation as antitumor drugs at Kosan Biosciences in the
quest for analogues with reduced hepatotoxicity and
greater water solubility.

4. Overview of Methodologies
4.1. Modular PKSs
A number of strategies for reprogramming modular
PKSs at the genetic level have emerged over the past
decade, ranging from single point mutations to
multiple module replacements, all resulting in
polyketides with targeted structural modifications.
Most of these strategies have been discussed in
previous reviews,4-6,8,9,11,18,20 and an outline is presented here. Briefly, these strategies include active
site inactivation or replacement, module substitution,
subunit complementation, and precursor-directed
biosynthesis.
The most common method of engineering thus far
is AT substitution, in which the native AT domain
is replaced with an AT encoding a different starter
or extender unit specificity. The AT used for substitution is usually derived from a heterologous modular
PKS. All six of the methyl-malonyl-specific AT domains in DEBS have been successfully replaced by
malonyl-specific AT domains to create 6dEB or
erythromycin analogues lacking a corresponding
methyl branch.4,71-73 Similar replacements have also
been accomplished with the spiramycin,50 FK520,74
rapamycin (J. Kennedy, personal communication),
and geldanamycin PKSs (unpublished data). The
loading AT of DEBS has been replaced by loading
ATs or AT/KSq domains from the avermectin,75,76
tylosin, and oleandomycin and rapamycin PKSs77 for
production of C-13-substituted derivates of erythromycin. Replacing an AT domain in DEBS with an
ethylmalonyl-specific AT52 and a 2-methoxymalonylspecific AT55 domain to generate ethyl and methoxy
branches, respectively, in either erythromycin or
6dEB has also been performed successfully. In each
case, introduction of exogenous genes involved in
precursor metabolism (ethylmalonyl-CoA and 2-methoxymalonyl-ACP) from other PKS gene clusters was
required for production in the engineered host.
Manipulation of -keto processing activities has
been accomplished by inactivation of domains, deletion of domains, and substitution of domains. Nearly
all of these examples have been performed with
DEBS. Examples of inactivation include the ER

McDaniel et al.

domain in module 4 by site-directed mutagenesis to


generate a 6,7-anhydro erythromycin derivative78
and deletion of KR domains in modules 5 and 6 to
generate C-3 and C-5 keto derivatives of 6-dEB,
respectively,15,79 and replacement of the KR domains
in modules 2, 5, and 6 to produce C10-11, C4-5, and
C-2-3 anhydro derivatives of 6dEB.79 Similarly,
substitution of KR domains in modules 2 and 6 with
a DH+ER+KR domain resulted in C-10 and C-3
deoxy 6dEB analogues.79
There is at least one report of a single whole
module substitution, DEBS module 2 with rifamycin
module 5.80 Although the activities of the two modules are identical and production of 6dEB was
reproduced, it suggests a useful alternative route in
situations where standard domain engineering falls
short. Construction of hybrid modules, subunits, and
subunit complementation within families of related
PKSs has been demonstrated in a number of cases.
Examples include DEBS/PicPKS,30,81 DEBS/rapamycin PKS,10,82 DEBS/OlePKS,41 PicPKS/OlePKS,83
PicPKS/TylPKS,81 and chalcomycin PKS/spiramycin
PKS.84 In many of these experiments the production
levels of the polyketide remain relatively high, demonstrating effective communication between noncognate subunits.
Precursor-directed biosynthesis combines the advantages of modern synthetic chemistry with complex
polyketide biochemistry to create a powerful approach to engineering polyketide chains with unique
features. This is achieved by constructing strains
which are blocked at a particular step in the pathway
(e.g., a KS domain) or in the ability to produce a
necessary precursor and supplying synthetically derived precursors in the form of N-acylcysteamine
thioesters.85,86 Several novel 14- and 16-membered
lactones have been produced using a KS1 derivative
of DEBS and supplying synthetic diketide or triketide
precursors.87-89 Side chains harboring halogens and
reactive groups for further modification by synthetic
chemistry have been incorporated into 6dEB. To date,
this has been the most successful approach to making
macrolides with potency equal to or better than
erythromycin.6

4.2. Aromatic PKSs


Although the field of polyketide gene engineering
was catalyzed by work with aromatic PKSs, this class
has not been exploited as intensely as modular PKSs.
Most successful attempts at engineering aromatic
polyketides result from mixing and matching of
separate individual enzyme subunits. For example,
pairs of ketosynthase and chain length factors (KS/
CLF), which together synthesize a length-specific
polyketide backbone, ketoreductases, cyclases, and
aromatases from different aromatic gene clusters
have been combined to manipulate chain length,
hydroxylation pattern, cyclization regiospecificity,
and aromaticity.90-95 A recent advance is the ability
to manipulate the choice of starter unit incorporation
through the use of a separate initiation module,
following identification of such a unit by the study
of the gene clusters for some of the rare aromatic
polyketides that use nonacetate starters.96 Cycliza-

Genetic Approaches to Polyketide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 553

alter the specificity of an AT domain102,103 and


inactivation of an AT, followed by complemention
with a separate trans-acting AT subunit104 (Figure
8). Both methods hold the advantage that structural
modification to a domain is minimized and potentially detrimental perturbations are avoided. In the
former case, the crystal structure of the E. coli fatty
acid malonyl transferase and sequence alignments
were used to identify residues putatively involved in
substrate specificity. These mutations were introduced into the AT4 domain of DEBS to produce
6-desmethyl-6dEB (11),102 although the mutations led
to relaxed specificitys6dEB (1) was also produceds
rather than a complete change in specificity. Actual
structures of modular PKS AT domains may help
refine the residues that are involved in substrate
selection and permit engineering of ATs with more
stringent specificity. In the second example, an AT6null mutant of DEBS was created by mutation of the
active site and the PKS complemented with a type
II fatty acid malonyltransferase (MAT) from S. coelicolor to produce 2-desmethyl-6dEB (12) in E. coli
with yields similar to that of the wild-type PKS.104

5.1.2. Ketoreductases
Figure 8. Examples of recent strategies for AT and KR
engineering. See sections 5.1.1 and 5.1.2 for details.

tion of the polyketide chain can be controlled with


appropriate choice of cyclases and aromatization
enzymes, provided one can be found with the desired
specificity. In the absence of such enzymes, however,
the final product structure(s) is determined by spontaneous cyclization, which may be difficult to predict.
Post-PKS steps such as oxidation, O-methylation, or
glycosylation are more amenable to alteration, as
illustrated by the work of Salas and co-workers with
mithramycinandrelatedchromanequinoneantibiotics.97-100

5. New Developments in Modular PKS


Manipulation
5.1. Modeling and Engineering of PKS Domains
5.1.1. Acyltransferases
Not every attempt at modular PKS AT domain
replacement is successful when a conserved set of
boundaries is used across different PKS modules. At
least one major reason for such unproductive AT
swaps is an apparent disruption in protein structure
so that chain elongation catalyzed by the KS and
ACP domains is severely attenuated.101 Recent experiments suggest that the choice of the domain
boundaries used to create the hybrid enzyme can be
a critical determinant of success. For example, our
group failed to engineer a productive malonyl-AT
domain replacement in module 4 of DEBS using
several different domain junctions.102 However, a set
of alternative junctions used by Leadlay and coworkers did result in a productive AT replacement73
(Figure 8).
Two alternative approaches to wholesale AT swaps
recently described are site-directed mutagenesis to

KR domains in modular PKSs catalyze stereospecific reduction and may be classified into two groups
according to the stereochemical outcome relative to
the polyketide backbone. Many examples of both
types of KRs exist. Inversion of an alcohol stereocenter is possible by replacing a KR of one class with
a KR from the other,79,105 but relatively few examples
have been reported, and all were performed at the
terminal module. Alteration of KR specificity in a
module preceding another module may require concomitant modification of the downstream KS so that
the altered stereocenter is recognized and processed.
Recent studies with sequence alignments of the two
KR types have identified differences in amino acid
residues that correlate with stereospecificity.106,107
The perfect correlation of these residues allows one
to predict stereochemical outcome in cases where a
gene sequence is known but the final product or
absolute stereochemistry is not known. Two models
of substrate binding relative to the NADPH cofactor
have been proposed to explain the relative outcomes
of ketoreduction,106,108 and Caffrey107 has proposed
mechanisms by which these residues may dictate
specificity in either model.
Homology modeling to the short-chain dehydrogenase/reductase family suggested a putative catalytic
triad in modular KRs.106 Point mutation of the
catalytic serine resulted in complete inactivation of
the KR6 domain in DEBS, producing 3-deoxy-3-oxo6dEB (13). Modification of the KR by this method
resulted in only the targeted analogue, whereas
deletion of the entire KR6 domain in DEBS affected
the specificity of the adjacent AT domain and led to
unexpected products (14) as well (Figure 8). This
particular example stresses the benefit of engineering
strategies which seek to minimize structural disturbances. The same mutation has been used to inactivate KR domains in the epothilone109 and geldana-

554 Chemical Reviews, 2005, Vol. 105, No. 2

mycin PKSs (manuscript submitted) and produce


corresponding analogues.

5.2. Type II Thioesterase


Modular PKS gene clusters commonly contain
genes encoding a type II thioesterase (TEII), defined
initially through studies in E. coli as hydrolytic
enzymes acting on long-chain fatty acid thioesters.110
These enzymes are not involved in the terminal event
of PKS assembly like the thioester domains discussed
below and, to some degree, are dispensable because
inactivation of their genes does not always abolish
polyketide formation, although product titers usually
are drastically lowered, as in the case of tylosin111
and picromycin.36 In erythromycin biosynthesis by S.
erythraea, genetic and biochemical studies have
shown that the ery-ORF5 TEII enzyme favors hydrolysis of acetyl groups bound to the loading ACP
domain to ensure formation of 6dEB from a propionate instead of an acetate starter unit.112 Thus, the
most likely role of such enzymes is to edit the process
of PKS assembly by selective hydrolysis of misprimed
or -acylated active site cysteines, as directly shown
by Schwarzer et al.113 in the case of TEII enzymes of
nonribosomal peptide antibiotic biosynthesis. Understanding the exact contributions made by TEIIs will
be critical for optimizing productivity of engineered
modular PKSs.

5.3. Cyclization/Termination
The TE domains attached to the terminal modules
of PKSs are generally tolerant toward polyketide
chain length as well as substitutions at the C-2 and
C-3 positions of the lactone, although with varying
efficiencies. The TE domains from DEBS and PicPKS
have been studied in vitro89,114 and can cyclize lactone
ring sizes in the range of 6-16 carbons. The PicPKS
TE has a much higher preference for 3-keto versus
3-hydroxy substrates as compared to DEBS TE, and
fusion of the PicPKS TE to DEBS module 3 resulted
in an enzyme with greater efficiency than DEBS
module 3 with DEBS TE, indicating that the PicPKS
TE is a better catalyst for cyclizing 3-keto acyl
intermediates.114 Crystal structures of both TE domains from DEBS115 and PicPKS116 have now been
obtained and, in addition to providing clues about the
overall tertiary architecture of modular PKSs, provide a structural basis for potentially altering substrate specificity. In addition to cyclizing lactones
from a variety of macrolide PKSs, DEBS TE was used
recently to generate a novel pentaketide lactone from
the spinosyn PKS.117
The ansamycin PKSs, unlike macrolide PKSs,
utilize a separate protein for cyclization of the linear
polyketide to the corresponding macrolactam. These
amide synthases, RifF (rifamycin), GdmF (geldanamycin), and Asm9 (ansamitocin), are homologous to
N-acetyl CoA transferases. The degree of substrate
flexibility among this class of enzymes is not currently known but if similar to the TEs could be useful
for engineering novel lactams.

McDaniel et al.

5.4. Intermodular and Intramodular


Communication
Accurate programmed polyketide extension depends on several protein-protein interactions to
correctly orient catalytic domains relative to ACPlinked substrates as well as facilitate specific intermodular transfer of the growing polyketide chain. In
addition to such structural communication, proper
extension requires compatibility of enzyme specificities such that each intermediate is accepted through
the reaction sequence. Our ability to reorganize PKS
structure for the synthesis of novel compounds will
depend on our understanding the rules for functional
inter- and intramodular communication and the
flexibility with which we can modify communication
independent of catalytic activity.
Recent work suggests that one reason domain
exchange has generally yielded poorly active PKSs
could be extra-domain intramodular structural perturbation. An engineered DEBS module 6+TE, in
which the AT was replaced with that from RAPS
module 2, was shown to retain nonlimiting malonyl
transferase and KS loading activities but was impaired in substrate condensation.101 These results
suggest that the AT insertion may disrupt correct
interaction between ACP and KS domains necessary
for the C-C bond-forming extension reaction.
Much work of late has focused on the determinants
of intermodular interaction that facilitate the specific
channeling of the growing polyketide chain. There is
clear evidence that a significant role is played by the
intervening protein sequence between the ACP domain and the KS of the downstream module80,104,118-122
(Figure 9). These sequences, which have been termed
intermodular linkers, facilitate both intra- and
interpolypeptide substrate channeling. Intrapolypeptide linkers bridge modules within the same protein.
Interpolypeptide linkers are terminal sequences that
bridge C- and N-terminal modules on separate proteins.
There is considerable evidence that modulemodule interaction can be engineered independently
of module catalytic function. Linkers can be altered
or exchanged without loss of intrinsic module activity,
and several hybrid PKSs have been successfully
constructed using compatible interpolypeptide linkers
to bridge normally incommunicative modules.80,119-122
Thus, engineering of linkers may allow flexibility in
the physical reorganization of multimodular PKSs.
Recent evidence suggests that acyl chain substrate
specificity can be a significant barrier to intermodular
channeling. Specificity in the transfer of the acyl
chain from the upstream ACP to the next module KS,
in the KS-catalyzed condensation reaction, or in
eventual product release by a TE domain may restrict
elongation of novel compounds on hybrid PKSs. A
thorough study of the substrate selectivities of several PKS modules has given interesting insights into
some specificity determinants.123 In particular, it was
observed that modules unable to extend substrates
were blocked not at acyl chain transfer to the KS but
at subsequent condensation, implying that somehow
transition-state stabilization in the rate-limiting
condensation reaction is more stringent than that in

Genetic Approaches to Polyketide Antibiotics

Chemical Reviews, 2005, Vol. 105, No. 2 555

Figure 9. Intra- and intermodular communication in polyketide extension. (A) Some interactions critical to substrate
channeling are shown: 1 and 3, substrate acceptance by KS; 2 and 4, condensation of KS- and ACP-bound intermediates;
5, product hydrolysis by TE. Intermodular transfer via specific interpolypeptide linkers is illustrated in a comparison of
panels A-D. Matched linker pairs (A and D) from DEBS modules 2 and 3 (L2 and L3) or modules 4 and 5 (L4 and L5)
greatly facilitate intermodular transfer of the growing polyketide chain. Data shown are taken from Tsuji et al.119

acyl transfer. Understanding the basis for this discrimination may allow us to change or broaden
specificity, opening new routes to functional hybrid
PKSs and their diverse potential products.

5.5. Precursor Engineering


PKSs occasionally require substrates for polyketide
biosynthesis that are not commonly found in bacterial
cells, for instance, 2-methoxymalonate and AHBA.
Unlike the ubiquitous small branched-chain fatty
acids used as starter units, which are believed to
come from the catabolism of valine, leucine, and
isoleucine,124 formation of 2-methoxymalonate for the
biosynthesis of FK520, ansamitocins, and geldanamycin requires a dedicated set of five genes to convert
some glycolytic intermediate to the ACP-bound form
of 2-methoxymalonate (15), as currently believed
(Figure 10). The acyl ACP dehydrogenase gene
product has been studied in vitro,125 and heterologous
expression of the genes from the ansamitocin and
FK520 producers in S. coelicolor or S. fradiae has
been used to produce, respectively, an erythromycin55
and a midecamycin126 analogue.
E. coli is being developed as a host for polyketide
production,127,128 which also requires the introduction
of metabolic pathways to make PKS substrates.
Pathways to form (2S)-methylmalonyl-CoA (16) from
propionate via propionyl-CoA carboxylase128 or from
succinyl-CoA via methylmalonyl-CoA mutase129 have
been engineered to support polyketide production of
6dEB (1) and analogues in E. coli (Figure 10). A 15methyl-6dEB analogue was produced in E. coli by
overexpressing the acetoacetyl-CoA:acetyl-CoA trans-

Figure 10. Precursor pathways that have been engineered


in S. coelicolor and S. fradiae (15) and E. coli (16 and 17).

ferase (atoAD) to generate butyryl-CoA (17), which


is utilized by the loading AT of DEBS, from fed
butyric acid or butanol.130

5.6. Tailoring Pathways


Desosamine is present on all of the 14-membered
macrolides and is essential for antibiotic activity.

556 Chemical Reviews, 2005, Vol. 105, No. 2

Recent crystallography studies with erythromycin


bound to ribosomes reveal the essential role that
desosamine plays in binding to the ribosome.131
Fourteen-membered lactones lacking desosamine are
completely inactive, and no synthetic substitute or
modification to desosamine has yet been found which
maintains the same level of activity. The entire set
of des genes, including the glycosyl transferase from
the picromycin cluster, were coexpressed with modified DEBS genes which produce analogues of 6dEB
to produce a small library of 14-membered macrolides
containing only desosamine.132 Antibiotic activity was
detected in nearly every case, suggesting desosamine
is sufficient to confer antibiotic activity to macrolactones.
Several experiments have been conducted to introduce modified or unnatural deoxysugars to macrolide
backbones (reviewed in Mendez and Salas133). Two
approaches to producing desosaminylated tylosin
derivatives have been described. In one example, the
TylMII mycaminosyltransferase was expressed in an
engineered S. erythraea strain which does not make
erythromycin and used to bioconvert tylactone to 5-Odesosaminyl-tylactone.134 In the other case, two genes
involved in desosamine biosynthesis from narbomycin were imported into the tylosin producing strain,
Streptomyces fradiae, to convert TDP-D-mycaminose
(TDP-D-4-deoxydesosamine) to TDP-D-desosamine
and produce desosaminylated tylonolide derivatives.135 The genes encoding L-olivose and L-oleandrose biosynthesis were expressed from plasmids in
Streptomyces albus and used to bioconvert erythronolide B to 3-O-olivosyl and 3-O-oleandrosyl erythronolide B analogues.136 These and other experiments
indicate some degree of deoxysugar substrate flexibility and the ability to engineer novel glycosides of
macrolide antibiotics. However, to date, such modifications have not led to compounds with activity
superior to the natural glycosides.

6. Creating and Improving Microbial Production


Systems
In addition to genetically engineering PKSs and
manipulating tailoring enzymes, another genetic
approach to novel metabolites has involved the
development of heterologous bacterial hosts. This
work has had two goals: to expedite structurefunction investigations of PKSs and the exploration
of metabolic pathways and to improve polyketide
titers by faster means than lengthy strain improvement carried out by random mutation and screening.
E. coli is becoming a versatile host for the expression of modular PKS genes, as demonstrated by the
recent work at Stanford University and Kosan.
Pfeiffer et al.128 produced 6dEB and novel analogues
made from aromatic acid starter units, and Kosan
scientists made a wide range of 6dEB analogues with
novel substituents at C-13 derived from either butyric
acid130 or fed N-acetylcysteamine thioesters (J.
Kennedy et al., manuscript in preparation). Recent
work with the rifamycin137,138 and epothilone PKS
genes139 (S. Mutka et al., submitted for publication)
and certain deoxysugar biosynthesis genes (H. Gramajo et al., manuscript in preparation) have shown

McDaniel et al.

Figure 11. Erythromycin analogues produced by engineered DEBS genes expressed in an overproducing S.
erythraea host.

that E. coli is likely to be a quite versatile host for


polyketide pathway engineering.
Another useful approach is to use an industrial
strain whose titer has already been improved as the
host for expression of PKS genes to gain the benefit
of the presumably enhanced levels of gene expression,
substrates, or tailoring enzymes that could be present.
Two examples of this strategy have been described
in which the native PKS genes were deleted from the
high-producing strain and replaced with the genes
from a wild-type producer. In both cases, use of the
wild-type genes did not lower the polyketide titer,
demonstrating that increased titers in the industrial
strains were not due to PKS mutations. Engineered
DEBS genes were then expressed in the high-producing S. erythraea strain to generate erythromycin
analogues at substantially higher levels than those
obtained in S. coelicolor140,141 or the wild-type S.
erythraea strain (Figure 11). More recently, a highproducing S. fradiae strain has been used to produce
a number of novel 16-membered macrolides in high
yield.84,126

7. Conclusions
Since the initial cloning of genes for polyketide
biosynthesis in the late 1980s and early 1990s, a
number of strategies and tools for engineering the
genes and pathways to create new polyketide compounds have been developed. These have been used
to create analogues of structurally complex compounds that would be difficult to obtain through
conventional organic synthesis or semi-synthesis.
Though many of the strategies are empirically based,
they are fairly robust when compared to the success
rate of a typical synthesis reaction as applied to
different substrates and so can be viewed as a
practical and complementary tool in the design and
production of new therapeutic entities. These techniques should be refined through continued practice,

Genetic Approaches to Polyketide Antibiotics

and additional methods will likely emerge to provide


increased versatility. Current and future information
derived from ongoing biochemical and structural
studies of PKSs will certainly help guide these
approaches.

8. Acknowledgments
We thank David Hopwood, Hugo Gramajo, and
Bryan Julien for their helpful comments on the
manuscript.

9. References
(1) Newman, D. J.; Cragg, G. M.; Snader, K. M. J. Nat. Prod. 2003,
66, 1022.
(2) Cragg, G. M.; Newman, D. J.; Snader, K. M. J. Nat. Prod. 1997,
60, 52.
(3) Cane, D. E.; Walsh, C. T.; Khosla, C. Science 1998, 282, 63.
(4) Katz, L.; McDaniel, R. Med. Res. Rev. 1999, 19, 543.
(5) McDaniel, R.; Katz, L. In Development of Novel Antomicrobial
Agents: Emerging Strategies; Lohner, K., Ed.; Horizon Scientific
Press: Wymondham, England, 2000; Vol. 19.
(6) Hutchinson, C. R.; McDaniel, R. Curr. Opin. Investig. Drugs
2001, 2, 1681.
(7) Baltz, R. H. Trends Microbiol. 1998, 6, 76.
(8) McDaniel, R.; Khosla, C. In Enzyme technologies for pharmaceutical and biotechnological applications; Kirst, H. A., Yeh, W.K., Zmijewski, M. J., Jr., Eds.; Marcel Dekker: New York, 2001.
(9) Weissman, K. J.; Staunton, J. In Enzyme technologies for
pharmaceutical and biotechnological applications; Kirst, H. A.,
Yeh, W.-K., Zmijewski, M. J., Jr., Eds.; Marcel Dekker: New
York, 2001.
(10) Rowe, C. J.; Bohm, I. U.; Thomas, I. P.; Wilkinson, B.; M., R. B.
A.; Foster, G.; Blackaby, A. P.; Sidebottom, P. J.; Roddis, Y.;
Buss, A. D.; Staunton, J.; Leadlay, P. F. Chem. Biol. 2001, 89,
1.
(11) Reeves, C. D. Crit. Rev. Biotechnol. 2003, 23, 95.
(12) Hopwood, D. A.; Malpartida, F.; Kieser, H. M.; Ikeda, H.;
Duncan, J.; Fujii, I.; Rudd, B. A.; Floss, H. G.; Omura, S. Nature
1985, 314, 642.
(13) Omura, S.; Ikeda, H.; Malpartida, F.; Kieser, H. M.; Hopwood,
D. A. Antimicrob. Agents Chemother. 1986, 29, 13.
(14) Cortes, J.; Haydock, S. F.; Roberts, G. A.; Bevitt, D. J.; Leadlay,
P. F. Nature 1990, 348, 176.
(15) Donadio, S.; Staver, M. J.; McAlpine, J. B.; Swanson, S. J.; Katz,
L. Science 1991, 252, 675.
(16) Rawlings, B. J. Nat. Prod. Rep. 2001, 18, 190.
(17) Rawlings, B. J. Nat. Prod. Rep. 2001, 18, 231.
(18) Khosla, C.; Gokhale, R. S.; Jacobsen, J. R.; Cane, D. E. Annu.
Rev. Biochem. 1999, 68, 219.
(19) Liou, G. F.; Khosla, C. Curr. Opin. Chem. Biol. 2003, 7, 279.
(20) Staunton, J.; Weissman, K. J. Nat. Prod. Rep. 2001, 18, 380.
(21) Austin, M. B.; Noel, J. P. Nat. Prod. Rep. 2003, 20, 79.
(22) Bevitt, D. J.; Cortes, J.; Haydock, S. F.; Leadlay, P. F. Eur. J.
Biochem. 1992, 204, 39.
(23) Summers, R. G.; Donadio, S.; Staver, M. J.; Wendt-Pienkowski,
E.; Hutchinson, C. R.; Katz, L. Microbiol. 1997, 143, 3251.
(24) Haydock, S. F.; Dowson, J. A.; Dhillon, N.; Roberts, G. A.; Cortes,
J.; Leadlay, P. F. Mol. Gen. Genet. 1991, 230, 120.
(25) Stassi, D.; Donadio, S.; Staver, M. J.; Katz, L. J. Bacteriol. 1993,
175, 182.
(26) Gaisser, S.; Bohm, G. A.; Cortes, J.; Leadlay, P. F. Mol. Gen.
Genet. 1997, 256, 239.
(27) Gaisser, S.; Bohm, G. A.; Doumith, M.; Raynal, M. C.; Dhillon,
N.; Cortes, J.; Leadlay, P. F. Mol. Gen. Genet. 1998, 258, 78.
(28) Weber, J. M.; Leung, J. O.; Swanson, S. J.; Idler, K. B.; McAlpine,
J. B. Science 1991, 252, 114.
(29) Swan, D. G.; Rodriguez, A. M.; Vilches, C.; Mendez, C.; Salas,
J. A. Mol. Gen. Genet. 1994, 242, 358.
(30) Shah, S.; Xue, Q.; Tang, L.; Carney, J. R.; Betlach, M.; McDaniel,
R. J. Antibiotics 2000, 53, 502.
(31) Rodriguez, A. M.; Olano, C.; Mendez, C.; Hutchinson, C. R.
FEMS Microbiol. Lett. 1995, 127, 117.
(32) Quiros, L. M.; Aguirrezabalaga, I.; Olano, C.; Mendez, C.; Salas,
J. A. Mol. Microbiol. 1998, 28, 1177.
(33) Olano, C.; Rodriguez, A. M.; Michel, J. M.; Mendez, C.; Raynal,
M. C.; Salas, J. A. Mol. Gen. Genet. 1998, 259, 299.
(34) Aguirrezabalaga, I.; Olano, C.; Allende, N.; Rodriguez, L.; Brana,
A. F.; Mendez, C.; Salas, J. A. Antimicrob. Agents Chemother.
2000, 44, 1266.
(35) Volchegursky, Y.; Hu, Z.; Katz, L.; McDaniel, R. Mol. Microbiol.
2000, 37, 752.

Chemical Reviews, 2005, Vol. 105, No. 2 557


(36) Xue, Y.; Zhao, L.; Liu, H.-w.; Sherman, D. H. Proc. Natl. Acad.
Sci. U.S.A. 1998, 95, 12111.
(37) Bisang, C.; Long, P. F.; Cortes, J.; Westcott, J.; Crosby, J.;
Matharu, A.-L.; Cox, R. J.; Simpson, T. J.; Staunton, J.; Leadlay,
P. F. Nature 1999, 401, 502.
(38) Tang, L.; Fu, H.; Betlach, M. C.; McDaniel, R. Chem. Biol. 1999,
6, 553.
(39) Beck, B. J.; Yoon, Y. J.; Reynolds, K. A.; Sherman, D. H. Chem.
Biol. 2002, 9, 575.
(40) Kao, C. M.; Katz, L.; Khosla, C. Science 1994, 265, 509.
(41) Hu, Z.; Desai, R. P.; Volchegursky, Y.; Leaf, T.; Woo, E.; Licari,
P.; Santi, D. V.; Hutchinson, C. R.; McDaniel, R. J. Ind.
Microbiol. Biotechnol. 2003, 30, 161.
(42) Gaisser, S.; Lill, R.; Staunton, J.; Mendez, C.; Salas, J.; Leadlay,
P. F. Mol. Microbiol. 2002, 44, 771.
(43) Zhao, L.; Sherman, D. H.; Liu, H.-w. J. Am. Chem. Soc. 1998,
120, 10256.
(44) Zhao, L.; Que, N. L. S.; Xue, Y.; Sherman, D. H.; Liu, H.-w. J.
Am. Chem. Soc. 1998, 120, 12159.
(45) Borisova, S. A.; Zhao, L.; Sherman, D. H.; Liu, H. W. Org. Lett.
1999, 1, 133.
(46) Betlach, M. C.; Kealey, J. T.; Betlach, M. C.; Ashley, G. A.;
McDaniel, R. Biochemistry 1998, 37, 14937.
(47) Xue, Y.; Wilson, D.; Zhao, L.; Liu, H.-w.; Sherman, D. H. Chem.
Biol. 1998, 5, 661.
(48) DeHoff, B. S.; Sutton, K. L.; Rosteck, P. R. GenBank accession
#U78289, 1996.
(49) Kakavas, S. J.; Katz, L.; Stassi, D. J. Bacteriol. 1997, 179, 7515.
(50) Kuhstoss, S.; Huber, M.; Turner, J. R.; Paschal, J. W.; Rao, R.
N. Gene 1996, 183, 231.
(51) Ward, S. L.; Hu, Z.; Schirmer, A.; Reid, R.; Revill, P. W.; Reeves,
C. D.; Petrakovsky, O. V.; Dong, S. D.; Katz, L. Antimicrob.
Agents Chemother. 2004, in press.
(52) Stassi, D. L.; Kakavas, S. J.; Reynolds, K. A.; Gunawardana,
G.; Swanson, S.; Zeidner, D.; Jackson, M.; Liu, H.; Buko, A.;
Katz, L. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 7305.
(53) Wu, K.; Chung, L.; Revill, P. W.; Katz, L.; Reeves, C. D. Gene
2000, 251, 81.
(54) Rascher, A.; Hu, Z.; Viswanathan, N.; Schirmer, A.; Reid, R.;
Nierman, W. C.; Lewis, M.; Hutchinson, C. R. FEMS Microbiol.
Lett. 2003, 218, 223.
(55) Kato, Y.; Bai, L.; Xue, Q.; Revill, W. P.; Yu, T. W.; Floss, H. G.
J. Am. Chem. Soc. 2002, 124, 5268.
(56) Cundliffe, E.; Bate, N.; Butler, A.; Fish, S.; Gandecha, A.;
Merson-Davies, L. Antonie Van Leeuwenhoek 2001, 79, 229.
(57) Mochizuki, S.; Hiratsu, K.; Suwa, M.; Ishii, T.; Sugino, F.;
Yamada, K.; Kinashi, H. Mol. Microbiol. 2003, 48, 1501.
(58) Rinehart, K. L. J.; Shield, L. S. Fortschr. Chem. Org. Naturst.
1976, 33, 231.
(59) Cassady, J. M.; Chan, K. K.; Floss, H. G.; Leistner, E. Chem.
Pharm. Bull. 2004, 52, 1.
(60) Kim, C. G.; Yu, T. W.; Fryhle, C. B.; Handa, S.; Floss, H. G. J.
Biol. Chem. 1998, 273, 6030.
(61) August, P. R.; Tang, L.; Yoon, Y. J.; Ning, S.; Muller, R.; Yu, T.
W.; Taylor, M.; Hoffmann, D.; Kim, C. G.; Zhang, X.; Hutchinson,
C. R.; Floss, H. G. Chem. Biol. 1998, 5, 69.
(62) Schupp, T.; Toupet, C.; Engel, N.; Goff, S. FEMS Microbiol. Lett.
1998, 159, 201.
(63) Stratmann, A.; Toupet, C.; Schilling, W.; Traber, R.; Oberer, L.;
Schupp, T. Microbiol. 1999, 145 (Pt 12), 3365.
(64) Yu, T. W.; Shen, Y.; Doi-Katayama, Y.; Tang, L.; Park, C.; Moore,
B. S.; Richard Hutchinson, C.; Floss, H. G. Proc. Natl. Acad. Sci.
U.S.A. 1999, 96, 9051.
(65) Stratmann, A.; Schupp, T.; Toupet, C.; Schilling, W.; Oberer, L.;
Traber, R. J. Antibiot. 2002, 55, 396.
(66) Yu, T. W.; Bai, L.; Clade, D.; Hoffmann, D.; Toelzer, S.; Trinh,
K. Q.; Xu, J.; Moss, S. J.; Leistner, E.; Floss, H. G. Proc. Natl.
Acad. Sci. U.S.A. 2002, 99, 7968.
(67) Spiteller, P.; Bai, L.; Shang, G.; Carroll, B. J.; Yu, T. W.; Floss,
H. G. J. Am. Chem. Soc. 2003, 125, 14236.
(68) Moss, S. J.; Bai, L.; Toelzer, S.; Carroll, B. J.; Mahmud, T.; Yu,
T. W.; Floss, H. G. J. Am. Chem. Soc. 2002, 124, 6544.
(69) Carroll, B. J.; Moss, S. J.; Bai, L.; Kato, Y.; Toelzer, S.; Yu, T.
W.; Floss, H. G. J. Am. Chem. Soc. 2002, 124, 4176.
(70) Patel, K.; Piagentini, M.; Rascher, A.; Tian, Z.-Q.; Buchanan,
G. O.; Regentin, R.; Hu, Z.; Hutchinson, C. R.; McDaniel, R.
Chem. Biol. 2004, 11, 1625.
(71) Ruan, X. R.; Pereda, A.; Stassi, D. L.; Zeidner, D.; Summers, R.
G.; Jackson, M.; Shivakumar, A.; Kakavas, S.; Staver, M. J.;
Donadio, S.; Katz, L. J. Bacteriol. 1997, 179, 6416.
(72) Liu, L.; Thamchaipenet, A.; Fu, H.; Betlach, M.; Ashley, G. J.
Am. Chem. Soc. 1997, 119, 10553.
(73) Petkovic, H.; Lill, R. E.; Sheridan, R. M.; Wilkinson, B.; McCormick, E. L.; McArthur, H. A.; Staunton, J.; Leadlay, P. F.;
Kendrew, S. G. J. Antibiot. 2003, 56, 543.
(74) Revill, W. P.; Voda, J.; Reeves, C. R.; Chung, L.; Schirmer, A.;
Ashley, G.; Carney, J. R.; Fardis, M.; Carreras, C. W.; Zhou, Y.;
Feng, L.; Tucker, E.; Robinson, D.; Gold, B. G. J. Pharmacol.
Exp. Ther. 2002, 302, 1278.

558 Chemical Reviews, 2005, Vol. 105, No. 2


(75) Marsden, A. F.; Wilkinson, B.; Cortes, J.; Dunster, N. J.;
Staunton, J.; Leadlay, P. F. Science 1998, 279, 199.
(76) Pacey, M. S.; Dirlam, J. P.; Geldart, R. W.; Leadlay, P. F.;
McArthur, H. A.; McCormick, E. L.; Monday, R. A.; OConnell,
T. N.; Staunton, J.; Winchester, T. J. J. Antibiot. 1998, 51,
1029.
(77) Long, P. F.; Wilkinson, C. J.; Bisang, C. P.; Cortes, J.; Dunster,
N.; Oliynyk, M.; McCormick, E.; McArthur, H.; Mendez, C.;
Salas, J. A.; Staunton, J.; Leadlay, P. F. Mol. Microbiol. 2002,
43, 1215.
(78) Donadio, S.; McAlpine, J. B.; Sheldon, P. J.; Jackson, M.; Katz,
L. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 7119.
(79) McDaniel, R.; Thamchaipenet, A.; Gustafsson, C.; Fu, H.;
Betlach, M.; Betlach, M.; Ashley, G. Proc. Natl. Acad. Sci. U.S.A.
1999, 96, 1846.
(80) Gokhale, R. S.; Tsuji, S. Y.; Cane, D. E.; Khosla, C. Science 1999,
284, 482.
(81) Yoon, Y. J.; Beck, B. J.; Kim, B. S.; Kang, K. Y.; Reynolds, K.
A.; Sherman, D. H. Chem. Biol. 2002, 9, 203.
(82) Ranganathan, A.; Timoney, M.; Bycroft, M.; Cortes, J.; Thomas,
I. P.; Wilkinson, B.; Kellenberger, L.; Hanefeld, U.; Galloway, I.
S.; Staunton, J.; Leadlay, P. F. Chem. Biol. 1999, 6, 731.
(83) Tang, L.; Fu, H.; McDaniel, R. Chem. Biol. 2000, 7, 77.
(84) Reeves, C. D.; Ward, S. L.; Revill, P. W.; Suzuki, H.; Marcus,
M.; Petrakovsky, O. V.; Marquez, S.; Fu, H.; Dong, S. D.; Katz,
L. Chem. Biol. 2004, 11, 1465.
(85) Jacobsen, J. R.; Hutchinson, C. R.; Cane, D. E.; Khosla, C.
Science 1997, 277, 367.
(86) Kinoshita, K.; Pfeifer, B. A.; Khosla, C.; Cane, D. E. Bioorg. Med.
Chem. Lett. 2003, 13, 3701.
(87) Jacobsen, J. R.; Cane, D. E.; Khosla, C. J. Am. Chem. Soc. 1998,
120, 9096.
(88) Jacobsen, J. R.; Keatinge-Clay, A. T.; Cane, D. E.; Khosla, C.
Bioorg. Med. Chem. 1998, 6, 1171.
(89) Gokhale, R. S.; Hunziker, D.; Cane, D.; Khosla, C. Chem. Biol.
1999, 6, 117.
(90) McDaniel, R.; Ebert-Khosla, S.; Hopwood, D. A.; Khosla, C.
Nature 1995, 375, 549.
(91) Tang, Y.; Tsai, S. C.; Khosla, C. J. Am. Chem. Soc. 2003, 125,
12708.
(92) Tang, Y.; Lee, T. S.; Kobayashi, S.; Khosla, C. Biochemistry 2003,
42, 6588.
(93) Keatinge-Clay, A. T.; Shelat, A. A.; Savage, D. F.; Tsai, S. C.;
Miercke, L. J.; OConnell, J. D., III; Khosla, C.; Stroud, R. M.
Structure (Cambridge) 2003, 11, 147.
(94) Pan, H.; Tsai, S.; Meadows, E. S.; Miercke, L. J.; Keatinge-Clay,
A. T.; OConnell, J.; Khosla, C.; Stroud, R. M. Structure (Cambridge) 2002, 10, 1559.
(95) Meadows, E. S.; Khosla, C. Biochemistry 2001, 40, 14855.
(96) Tang, Y.; Lee, T. S.; Khosla, C. PLoS Biol. 2004, 2, E31.
(97) Remsing, L. L.; Gonzalez, A. M.; Nur-e-Alam, M.; FernandezLozano, M. J.; Brana, A. F.; Rix, U.; Oliveira, M. A.; Mendez,
C.; Salas, J. A.; Rohr, J. J. Am. Chem. Soc. 2003, 125, 5745.
(98) Trefzer, A.; Blanco, G.; Remsing, L.; Kunzel, E.; Rix, U.; Lipata,
F.; Brana, A. F.; Mendez, C.; Rohr, J.; Bechthold, A.; Salas, J.
A. J. Am. Chem. Soc. 2002, 124, 6056.
(99) Remsing, L. L.; Garcia-Bernardo, J.; Gonzalez, A.; Kunzel, E.;
Rix, U.; Brana, A. F.; Bearden, D. W.; Mendez, C.; Salas, J. A.;
Rohr, J. J. Am. Chem. Soc. 2002, 124, 1606.
(100) Lombo, F.; Kunzel, E.; Prado, L.; Brana, A. F.; Bindseil, K. U.;
Frevert, J.; Bearden, D.; Mendez, C.; Salas, J. A.; Rohr, J. Angew.
Chem., Int. Ed. Engl. 2000, 39, 796.
(101) Hans, M.; Hornung, A.; Dziarnowski, A.; Cane, D. E.; Khosla,
C. J. Am. Chem. Soc. 2003, 125, 5366.
(102) Reeves, C. D.; Murli, S.; Ashley, G. A.; Piagentini, M.; Hutchinson, C. R.; McDaniel, R. Biochemistry 2001, 40, 15464.
(103) Del Vecchio, F.; Petkovic, H.; Kendrew, S. G.; Low, L.; Wilkinson,
B.; Lill, R.; Cortes, J.; Rudd, B. A.; Staunton, J.; Leadlay, P. F.
J. Ind. Microbiol. Biotechnol. 2003, 30, 489.
(104) Kumar, P.; Koppisch, A. T.; Cane, D. E.; Khosla, C. J. Am. Chem.
Soc. 2003, 125, 14307.
(105) Kao, C. M.; McPherson, M.; McDaniel, R.; Fu, H.; Cane, D.;
Khosla, C. J. Am. Chem. Soc. 1998, 120, 2478.
(106) Reid, R.; Piagentini, M.; Rodriguez, E.; Ashley, G.; Viswanathan,
N.; Carney, J.; Santi, D. V.; Hutchinson, C. R.; McDaniel, R.
Biochemistry 2003, 42, 72.

McDaniel et al.
(107) Caffrey, P. Chembiochem. 2003, 4, 654.
(108) Yin, Y.; Gokhale, R.; Khosla, C.; Cane, D. E. Bioorg. Med. Chem.
Lett. 2001, 11, 1477.
(109) Tang, L.; Ward, S.; Chung, L.; Carney, J. R.; Li, Y.; Reid, R.;
Katz, L. J. Am. Chem. Soc. 2004, 126, 46.
(110) Cronan, J. E., Jr.; Rock, C. O. In Escherichia coli and Salmonella
Typhimurium: Cellular and Molecular Biology; Lin, C. C., Low,
K. B., Magasanik, B., Reznikoff, W. S., Riley, M., Schaechter,
M., Umbarger, H. E., Eds.; American Society for Microbiology:
Washington, DC, 1996.
(111) Butler, A. R.; Bate, N.; Cundliffe, E. Chem. Biol. 1999, 6,
287.
(112) Hu, Z.; Pfeifer, B. A.; Chao, E.; Murli, S.; Kealey, J.; Carney, J.
R.; Ashley, G.; Khosla, C.; Hutchinson, C. R. Microbiology 2003,
149, 2213.
(113) Schwarzer, D.; Mootz, H. D.; Linne, U.; Marahiel, M. A. Proc.
Natl. Acad. Sci. U.S.A. 2002, 99, 14083.
(114) Lu, H.; Tsai, S. C.; Khosla, C.; Cane, D. E. Biochemistry 2002,
41, 12590.
(115) Tsai, S. C.; Miercke, L. J.; Krucinski, J.; Gokhale, R.; Chen, J.
C.; Foster, P. G.; Cane, D. E.; Khosla, C.; Stroud, R. M. Proc.
Natl. Acad. Sci. U.S.A. 2001, 98, 14808.
(116) Tsai, S. C.; Lu, H.; Cane, D. E.; Khosla, C.; Stroud, R. M.
Biochemistry 2002, 41, 12598.
(117) Martin, C. J.; Timoney, M. C.; Sheridan, R. M.; Kendrew, S. G.;
Wilkinson, B.; Staunton, J. C.; Leadlay, P. F. Org. Biomol. Chem.
2003, 1, 4144.
(118) Broadhurst, R. W.; Nietlispach, D.; Wheatcroft, M. P.; Leadlay,
P. F.; Weissman, K. J. Chem. Biol. 2003, 10, 723.
(119) Tsuji, S. Y.; Cane, D. E.; Khosla, C. Biochemistry 2001, 40,
2326.
(120) Tsuji, S. Y.; Wu, N.; Khosla, C. Biochemistry 2001, 40,
2317.
(121) Wu, N.; Cane, D. E.; Khosla, C. Biochemistry 2002, 41,
5056.
(122) Wu, N.; Tsuji, S. Y.; Cane, D. E.; Khosla, C. J. Am. Chem. Soc.
2001, 123, 6465.
(123) Watanabe, K.; Wang, C. C.; Boddy, C. N.; Cane, D. E.; Khosla,
C. J. Biol. Chem. 2003, 278, 42020.
(124) Tang, L.; Zhang, Y. X.; Hutchinson, C. R. J. Bacteriol. 1994, 176,
6107.
(125) Watanabe, K.; Khosla, C.; Stroud, R. M.; Tsai, S. C. J. Mol. Biol.
2003, 334, 435.
(126) Rodriguez, E.; Ward, S.; Fu, H.; Revill, W. P.; McDaniel, R.; Katz,
L. Appl. Microbiol. Biotechnol. 2004, 66, 85.
(127) Pfeifer, B. A.; Khosla, C. Microbiol. Mol. Biol. Rev. 2001, 65,
106.
(128) Pfeifer, B. A.; Admiraal, S. J.; Gramajo, H.; Cane, D. E.; Khosla,
C. Science 2001, 291, 1790.
(129) Dayem, L. C.; Carney, J. R.; Santi, D. V.; Pfeifer, B. A.; Khosla,
C.; Kealey, J. T. Biochemistry 2002, 41, 5193.
(130) Murli, S.; Kennedy, J.; Dayem, L. C.; Carney, J. R.; Kealey, J.
T. J. Ind. Microbiol. Biotechnol. 2003, 30, 500.
(131) Schlunzen, F.; Zarivach, R.; Harms, J.; Bashan, A.; Tocilj, A.;
Albrecht, R.; Yonath, A.; Franceschi, F. Nature 2001, 413, 814.
(132) Tang, L