ARTICLE IN PRESS

Metabolic Engineering 10 (2008) 234 245

Contents lists available at ScienceDirect

Metabolic Engineering
journal homepage: www.elsevier.com/locate/ymben

A new model for the anaerobic fermentation of glycerol in enteric bacteria:

Trunk and auxiliary pathways in Escherichia coli
Ramon Gonzalez , Abhishek Murarka 1, Yandi Dharmadi 1, Syed Shams Yazdani
Department of Chemical and Biomolecular Engineering, Rice University, P.O. Box 1892, Houston, TX 77251-1892, USA

a r t i c l e in f o

a b s t r a c t

Article history:
Received 31 July 2007
Received in revised form
3 May 2008
Accepted 13 May 2008
Available online 27 May 2008

Anaerobic fermentation of glycerol in the Enterobacteriaceae family has long been considered a unique
property of species that synthesize 1,3-propanediol (1,3-PDO). However, we have discovered that
Escherichia coli can ferment glycerol in a 1,3-PDO-independent manner. We identied 1,2-propanediol
(1,2-PDO) as a fermentation product and established the pathway that mediates its synthesis as well as
its role in the metabolism of glycerol. We also showed that the trunk pathway responsible for the
conversion of glycerol into glycolytic intermediates is composed of two enzymes: a type II glycerol
dehydrogenase (glyDH-II) and a dihydroxyacetone kinase (DHAK), the former of previously unknown
physiological role. Based on our ndings, we propose a new model for glycerol fermentation in enteric
bacteria in which: (i) the production of 1,2-PDO provides a means to consume reducing equivalents
generated in the synthesis of cell mass, thus facilitating redox balance, and (ii) the conversion of glycerol
to ethanol, through a redox-balanced pathway, fullls energy requirements by generating ATP via
substrate-level phosphorylation. The activity of the formate hydrogen-lyase and F0F1-ATPase systems
were also found to facilitate the fermentative metabolism of glycerol, and along with the ethanol and
1,2-PDO pathways, were considered auxiliary or enabling. We demonstrated that glycerol fermentation
in E. coli was not previously observed due to the use of medium formulations and culture conditions
that impair the aforementioned pathways. These include high concentrations of potassium and
phosphate, low concentrations of glycerol, alkaline pH, and closed cultivation systems that promote the
accumulation of hydrogen gas.
& 2008 Elsevier Inc. All rights reserved.

Keywords:
Glycerol fermentation
Escherichia coli
Enteric bacteria

1. Introduction
Glycerol has become an inexpensive and abundant carbon
source due to its generation as inevitable by-product of biodiesel
fuel production. Worldwide surplus of glycerol has prompted the
shutdown of facilities dedicated to its production or rening and
the economic viability of the biodiesel industry has been greatly
affected (Yazdani and Gonzalez, 2007 and references therein).
Given the highly reduced state of carbon in glycerol, its conversion
to fuels or reduced products could result in yields higher than
those obtained with the use of common sugars. Realizing this
potential, however, would require the anaerobic fermentation of
glycerol in the absence of external electron acceptors.
The ability to conduct fermentative metabolism of glycerol in
the Enterobacteriaceae family is shared by only a few members
such as Citrobacter freundii and Klebsiella pneumoniae (Booth,
2005; Bouvet et al., 1995). This metabolic process is mediated by a

 Corresponding author. Fax: +1713 348 5478.

E-mail address: Ramon.Gonzalez@rice.edu (R. Gonzalez).

These authors contributed equally to this work.

1096-7176/$ - see front matter & 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.ymben.2008.05.001

two-branch pathway (Fig. 1A), which results in the synthesis of

glycolytic intermediate dihydroxyacetone-phosphate (DHAP: see
Footnote2 for all acronyms) and fermentation product 1,3propanediol (1,3-PDO) (Booth, 2005). In the oxidative branch,
glycerol is dehydrogenated to dihydroxyacetone (DHA) by a
type I, NAD-linked glycerol dehydrogenase (glyDH-I). DHA is then
phosphorylated by ATP- or PEP-dependent DHA kinases (DHAK) to
generate DHAP. Through the parallel reductive branch, glycerol is
dehydrated by the coenzyme B12-dependent glycerol dehydratase
to form 3-hydroxypropionaldehyde (3-HPA). 3-HPA is then
reduced to the major fermentation product 1,3-PDO by the

2
ADH, alcohol dehydrogenase; AKR, aldo-keto reductase; COSY, COrrelation
SpectroscopY; DHA, dihydroxyacetone; DHAK, DHA kinase; DHAP, DHA phosphate;
FHL, formate-hydrogen lyase complex; F0F1-ATPase, F0F1-H+-translocating ATP
(hydrol-/synthet-)ase; GLYC, glycerol; glyDH, glycerol dehydrogenase; G3P,
sn-glycerol-3-phosphate; G3PDH, G3P dehydrogenase; a-G3PDH, aerobic G3PDH;
an-G3PDH, anaerobic G3PDH; HA, hydroxyacetone; LAL, lactaldehyde; MG,
methylglyoxal; MGR, MG reductase; MGS, MG synthase; MM, minimal medium;
NMR, nuclear magnetic resonance; PEP, phosphoenolpyruvate; PYR, pyruvate;
1,2-PDO, 1,2-propanediol; 1,2-PDOR, 1,2-PDO reductase; 1,3-PDO, 1,3-propanediol;
1,3-PDODH, 1,3-PDO dehydrogenase; 3HPA, 3-hydroxypropionaldehyde.

ARTICLE IN PRESS
R. Gonzalez et al. / Metabolic Engineering 10 (2008) 234245

GLYC
H2O
3HPA

ATP ADP
glyD

GLYC

glyDH-I

H
1,3-PDODH
1, 3-PDO

G3P

GK

DHAK

DHA
ATP/PEP

G3PDH

DHAP

ADP/PYR

PYR
Glycolysis

DHAP

MGS
(mgsA)

AKR ( yeaE, yghZ, yafB )

MG
NAD(P)H

HA
NAD(P)H

NA(P)D +
NADH

MGR
NA(P)D +

NADH

(fucO)

LAL

glyDH
(gldA )

NAD +
+
NAD
1,2-PDO

1,2-PDOR
Fig. 1. Glycerol dissimilation and 1,3- and 1,2-propanediol synthesis. (A) Glycerolfermenting species in the Enterobacteriaceae dissimilate glycerol via a two-branch
pathway (shaded): the reductive, 1,3-PDO-producing branch acts as a sink for the
reducing equivalents generated in the oxidative branch. This metabolic process
represents the established model for glycerol fermentation in enteric bacteria.
Metabolism of glycerol in species unable to synthesize 1,3-PDO, such as E. coli,
takes place through a respiratory pathway that requires an external electron
acceptor (enclosed in an oval). (B) Metabolic pathways leading to the synthesis of
1,2-propanediol (1,2-PDO) from dihydroxyacetone-phosphate in E. coli. Reactions
catalyzed by MGS, MGR, AKR, glyDH, and 1,2-PDOR are as previously reported
(Altaras and Cameron, 1999; Boronat and Aguilar, 1979; Cooper, 1984; Ko et al.,
2005; Misra et al., 1996; Saikusa et al., 1987; Truniger and Boos, 1994). Thick lines
indicate the route used by E. coli MG1655 for 1,2-PDO synthesis during glycerol
fermentation, as inferred from the experimental evidence summarized in Figs. 3
and 4. Abbreviations: AKR, aldo-keto reductases; DHA, dihydroxyacetone; DHAK,
DHA kinase; DHAP, DHA phosphate; GK, glycerol kinase; GLYC, glycerol; glyD,
glycerol dehydratase; glyDH-I, glycerol dehydrogenase, type I; G3P, glycerol-3phosphate; G3PDH, G3P dehydrogenase; H, reducing equivalents (H NADH/
NADPH/FADH2); HA, hydroxyacetone; LAL, lactaldehyde; MG, methylglyoxal; MGR,
methylglyoxal reductase; MGS, methylglyoxal synthase; PEP, phosphoenolpyruvate; PYR, pyruvate; 1,2-PDOR, 1,2-propanediol reductase; 1,3-PDO, 1,3-propanediol; 1,3-PDODH, 1,3-PDO dehydrogenase; 3HPA, 3-hydroxypropionaldehyde.
Metabolites shown in bold are extracellular.

NADH-linked 1,3-PDO dehydrogenase (1,3-PDODH), thereby regenerating NAD+ (Fig. 1A). Only eight taxa of the Enterobacteriaceae grow fermentatively on glycerol and in all cases they
produce 1,3-PDO and possess the enzymes glyDH-I and 1,3PDODH (Bouvet et al., 1995). Although several types of glyDHs
have been found in species unable to ferment glycerol (including
type II, glyDH-II, in Escherichia coli), their role remains unknown
(Bouvet et al., 1995).
As no 1,3-PDO-producing capacity has been identied in wildtype E. coli strains, it is believed that the metabolism of glycerol in
this organism requires the presence of electron acceptors (Booth,
2005; Bouvet et al., 1994, 1995; Lin, 1976; Quastel et al., 1925;
Quastel and Stephenson, 1925). The respiratory pathway mediating glycerol utilization involves a glycerol transporter, a glycerol
kinase, and two respiratory glycerol-3-P dehydrogenases
(G3PDHs) (Booth, 2005; Borgnia and Agre, 2001; Lin, 1976;
Pettigrew et al., 1990; Schryvers and Weiner, 1982; Walz et al.,

235

2002) (Fig. 1A). Under aerobic conditions a homodimeric enzyme

associated with the cytoplasmic membrane (aerobic G3P dehydrogenase: a-G3PDH, encoded by the glpD gene) is known to be
essential for the metabolism of glycerol. In the absence of oxygen
and presence of other electron acceptors such as fumarate, a
three-subunit enzyme (anaerobic G3P dehydrogenase: an-G3PDH,
encoded by the glpABC operon) converts glycerol-3-phosphate
(G3P) into DHAP. The lack of a G3PDH activity in the absence of
electron acceptors is thought to result in the accumulation of G3P
at levels that become inhibitory for cell growth, a condition that is
relieved by the presence of fumarate (Booth, 2005). Therefore, the
inability of E. coli to metabolize G3P under fermentative conditions has been suggested as the reason for this organisms
inability to ferment glycerol (Booth, 2005).
Despite the belief that the metabolism of glycerol in E. coli is
restricted to respiratory conditions, recent studies in our laboratory
showed that this organism can ferment glycerol in the absence of
external electron acceptors (Dharmadi et al., 2006). Since glycerol
fermentation was dependent on the supplementation of the
medium with tryptone or other nutrients, we conducted further
studies to demonstrate the fermentative nature of this metabolic
process and the use of glycerol in the synthesis of cell mass
(Murarka et al., 2008). Our results also indicate that wild-type
E. coli strains might have the ability to produce 1,2-propanediol
(1,2-PDO) during glycerol fermentation. Fig. 1B shows the pathways
that have been reported to mediate the synthesis of 1,2-PDO from
glycolytic intermediate DHAP in E. coli (Altaras and Cameron, 1999;
Boronat and Aguilar, 1979; Cooper, 1984; Ko et al., 2005; Misra
et al., 1996; Saikusa et al., 1987; Truniger and Boos, 1994).
In the studies described here we report the conditions,
pathways, and mechanisms mediating the fermentative metabolism of glycerol in E. coli. Based on our ndings, a new model for
the 1,3-PDO-independent fermentation of glycerol in enteric
bacteria is proposed. Genetic and environmental determinants
of the fermentative metabolism of glycerol are discussed in the
context of the proposed model.

2. Materials and methods

2.1. Strains, plasmids, and genetic methods
Wild-type E. coli strains MC4100 (ATCC 35695), W3110 (ATCC
27325), and E. coli B (ATCC 11303) and enteric bacteria
Enterobacter cloacae subsp. cloacae NCDC 279-56 (ATCC 13047),
Buttiauxella agrestis (ATCC 33994), Serratia plymuthica (ATCC
15928), and Leminorella richardii (ATCC 33998) were obtained
from the American Type Culture Collection (ATCC, Mannassas,
VA). K12 strain MG1655 (F-l-ilvG-rfb-50 rph-1) along with the
following otherwise isogenic derivatives (mutation details in
parentheses) were obtained from the University of Wisconsin
E. coli Genome Project (www.genome.wisc.edu): FB21196
(glpDHTn5KAN-I-SceI), FB20724 (glpAHTn5KAN-I-SceI), FB22899
(gldAHTn5KAN-I-SceI), FB21569 (dhaKHTn5KAN-I-SceI), FB21424
(atpDHTn5KAN-I-SceI), FB21425 (atpFHTn5KAN-I-SceI), FB21975
(mgsAHTn5KAN-I-SceI), FB20935 (fucOHTn5KAN-I-SceI), FB20044
(yafBHTn5KAN-I-SceI), FB22576 (yeaEHTn5KAN-I-SceI), and
FB23263 (yghZHTn5KAN-I-SceI). These strains carry a transposon
(Tn5) insertion mutation in the specied gene (Kang et al., 2004).
Disruption of genes/operons fdhF, gldA, dhaKLM, and glpABC were
created in both MG1655 and W3110 backgrounds using the
method described by Datsenko and Wanner (2000). Information
about plasmids (pKD4, pKD46, and pCP20) and primers (d-fdhF,
v-fdhF, d-gldA, v-gldA, d-dhaKLM, v-dhaKLM, d-glpABC, and
v-glpABC) used in their construction is provided in Table 1. All
mutants were veried through genomic PCR after construction to

ARTICLE IN PRESS
236

R. Gonzalez et al. / Metabolic Engineering 10 (2008) 234245

Table 1
Plasmids and primers used in this study
Plasmid/primer

Structure or description

Source

Plasmids
pKD4
pKD46
pCP20
pZSKLM
pZSKLcf
pZSgldA
pZSKLMgldA
pZSmgsAgldA
pCA24NgldA
pCA24NfdhF
pCA24NatpF

repR6KgApR FRT KmR FRT

R
+
repts
pSC101 Ap ParaBAD g b exo
R
R
repts
pSC101 Ap Cm cI857l PR p+
E. coli dhaKLM under control of PLtetO-1 (tetR, oriR SC101*, cat)
C. freundii dhaKL under control of PLtetO-1 (tetR, oriR SC101*, cat)
E. coli gldA under control of PLtetO-1 (tetR, oriR SC101*, cat)
E. coli dhaKLM and gldA under control of PLtetO-1 (tetR, oriR SC101*, cat)
E. coli mgsA and gldA under control of PLtetO-1 (tetR, oriR SC101*, cat)
E. coli gldA under control of IPTG-inducible promoter pT5/lac (oriR SC101*, cat)
E. coli fdhF under control of IPTG-inducible promoter pT5/lac (oriR SC101*, cat)
E. coli atpF under control of IPTG-inducible promoter pT5/lac (oriR SC101*, cat)

Datsenko and Wanner (2000)

Datsenko and Wanner (2000)
Datsenko and Wanner (2000)
Gutknecht et al. (2001)
Gutknecht et al. (2001)
This study
This study
This study
Kitagawa et al. (2005)
Kitagawa et al. (2005)
Kitagawa et al. (2005)

gaagggttattatggctgggacttcattaacgatacgtgtaggctggagctgcttc
gcagtatttgtactccggcgttttcgtaatccatatgaatatcctccttag
cgtaatatcagggaatgaccc
gggcaaagaatgtcaaaaacaa
gcagtgtggcgcaattctcggtatcggtggcggaaaaacgtgtaggctggagctgcttc
gacatcttctttaatatccagttgagcgagagttattggcaaacctacatatgaatatcctccttag
atggaccgcattattcaatcac
gcctacaaaagcacgcaaattc
gctggagcaaaataatgaaaaaattgatcaatgatgtgcaagacggtgtaggctggagctgcttc
actgcgggagttcttctttcgtttgggtcaggtggcatatgaatatcctccttag
tatcccgcatcccttatgac
aaaccattagtgctgagtaaatt
agtcacggtaccatggaccgcattattcaatcac
gatcgtctgcagttattcccactcttgcaggaaac
gacactgcagaggagcaattatggaccgca
caagctacgcgtttattcccactcttgcagga
atactgggtaccatggaactgacgactcgc
gatcgtctgcagttacttcagacggtccgc

This study

Primersa
d-fdhF
v-fdhF
d-gldA
v-gldA
d-dhaKLM
v-dhaKLM
c1-gldA
c2-gldA
c3-gldA

This study
This study
This study
This study
This study
This study
This study
This study

a
Priming sequences for plasmid pKD4 are underlined. d and v indicate that the primer sequences (50 to 30 ) were used for deletion (d) and verication (v)
purposes during the creation of disruption mutants as previously described (Datsenko and Wanner, 2000). A c indicates that the primer was used for cloning purposes,
c1 to clone gldA alone (pZSgldA), c2 to clone gldA along with dhaKLM (pZSKLMgldA), and c3 to clone gldA along with mgsA (pZSmgsAgldA). The forward sequence
follows the reverse sequence in each case. Genes or operons deleted or cloned are apparent from primer names.

ensure that the gene of interest had been disrupted. By using

plasmid pKD4 as template, we ensured the creation of in-frame
gene deletions thus minimizing polarity effects on the expression
of downstream genes (Datsenko and Wanner, 2000). Throughout
the paper, gene disruption mutants are referred to using the
following nomenclature: Dgene_name(s), where gene_name(s)
indicates the disrupted gene(s).
Table 1 describes other plasmids used in this study, which
include: (i) pZSKLM, expressing E. coli DHA kinase subunits
DhaKLM; (ii) pZSKLMgldA, expressing both E. coli DHA kinase
subunits DhaKLM and E. coli glycerol dehydrogenase (GldA); (iii)
pZSgldA, expressing E. coli glycerol dehydrogenase (GldA); (iv)
pZSmgsAgldA, expressing E. coli methylglyoxal synthase (MgsA)
and glycerol dehydrogenase (GldA); and (v) derivatives of plasmid
pCA24N carrying the genes fdhF, atpF, and gldA. Plasmid pZSKLM
was kindly provided by Dr. B. Erni, Universitat Bern, Switzerland
(Gutknecht et al., 2001). The expression vector pZSgldA was
constructed as follows. The coding region of the gldA gene was
PCR amplied using genomic DNA of E. coli MG1655 as template
and primers described in Table 1 (c1-gldA primers). The
restriction enzyme sites KpnI and PstI were introduced through
the forward and reverse primers, respectively, to facilitate cloning
of PCR product in the expression vector pZSKLM (Bachler et al.,
2005). The PCR was performed using Pfu turbo DNA polymerase
(Stratagene, CA, USA) under standard conditions as described by
the supplier. The amplied product (1.1 kB) was digested with
KpnI and PstI and used for ligation at the corresponding sites of
the pZSKLM plasmid. The ligated product was used to transform
E. coli DH5aT1 (Invitrogen, CA, USA). Positive clones were
screened by plasmid isolation and restriction digestion. The

expression vectors pZSKLMgldA and pZSmgsAgldA were constructed in a similar fashion. For plasmid pZSKLMgldA, the gldA
gene was amplied along with its ribosome-binding site from the
genomic DNA of E. coli MG1655 using the primers described in
Table 1 (c2-gldA primers). The amplied product was digested
with PstI and MluI and cloned at the corresponding sites of
pZSKLM, downstream of the dhaKLM gene (Bachler et al., 2005).
For plasmid pZSmgsAgldA, the coding region of mgsA was PCR
amplied using genomic DNA of E. coli MG1655 as template and
the primers described in Table 1 (c3-gldA primers). The
amplied product was digested with KpnI and PstI and cloned
at the corresponding sites of expression vector pZSKLMgldA
(replacing the dhaKLM genes). pCA24N derivatives were obtained
from the Genome Analysis Project Japan (http://ecoli.aist-nara.
ac.jp/), which provides clones of each ORF predicted from the
genome sequence of E. coli W3110. Every ORF has been cloned into
plasmid pCA24N, which contains the IPTG-inducible promoter
pT5/lac (Kitagawa et al., 2005). Derivatives of pCA24N expressing
formate dehydrogenase, the b subunit of the F0 complex of
F0F1-ATPase, and glycerol dehydrogenase are referred to here as
pCA24NfdhF, pCA24NatpF, and pCA24NgldA, respectively. Plasmids were transformed into E. coli strains and selected on Luria
Bertani (LB) plates containing appropriate antibiotics. Gene
expression in the aforementioned plasmids was induced with
100 ng/mL anhydrotetracycline (pZS series plasmids) or 10 mM
IPTG (isopropyl-b-D-thiogalactopyranoside) (pCA24N derivatives).
Standard recombinant DNA procedures were used for plasmid
isolation, electroporation, and polymerase chain reaction (Sambrook et al., 1989). The strains were kept in 32.5% glycerol stocks
at 80 1C. Plates were prepared using LB medium containing 1.5%

ARTICLE IN PRESS
R. Gonzalez et al. / Metabolic Engineering 10 (2008) 234245

agar. Antibiotics were included as needed at the following

concentrations: 100 mg/mL ampicillin, 34 mg/mL chloramphenicol,
50 mg/mL kanamycin, and 12.5 mg/mL tetracycline.
2.2. Culture medium and cultivation conditions
The minimal medium (MM) designed by Neidhardt et al.
(1974) supplemented with 2 g/L tryptone (Difco, USA), 110 mM
glycerol, 5 mM sodium selenite, and 1.32 mM Na2HPO4 in place of
K2HPO4 was used, unless otherwise specied. MOPS (morpholinopropanesulfonic acid) was used only in the inoculum preparation
phase, which was conducted in tubes with no external control of
pH. When indicated, the medium was supplemented with
specied concentrations of the following compounds: monobasic
and dibasic sodium and potassium phosphates, monobasic and
dibasic ammonium phosphates, sodium chloride, sodium sulfate,
potassium chloride, and potassium sulfate. Chemicals were
obtained from Fisher Scientic (Pittsburgh, PA) and Sigma-Aldrich
Co. (St. Louis, MO).
Fermentations were conducted in a SixFors multi-fermentation
system (Infors HT, Bottmingen, Switzerland) with six 500 mL
working volume vessels and independent control of temperature
(37 1C), pH (externally controlled with NaOH and H2SO4), and
stirrer speed (200 rpm) (Dharmadi et al., 2006). Anaerobic
conditions were maintained by initially sparging the medium
with ultrahigh purity argon (Matheson, Tri-Gas, Houston, TX) and
thereafter ushing the headspace with the same gas at 0.01 LPM.
An oxygen trap (Alltech Associates, Inc., Deereld, IL) was used to
eliminate traces of oxygen from the gas stream. In some
experiments (specied in each case), carbon dioxide or hydrogen
were also included in the gas phase.
Prior to use, the cultures (stored as glycerol stocks at 80 1C)
were streaked onto LB plates (with appropriate antibiotics if
required) and incubated overnight at 37 1C in an Oxoid anaerobic
jar with the CO2 gas generating kit (Oxoid Ltd., Basingstoke,
Hampshire, UK). A single colony was used to inoculate 17.5-mL
Hungate tubes (Bellco Glass, Inc., Vineland, NJ) completely lled
with MM supplemented with 10 g/L tryptone, 5 g/L yeast extract,
and 110 mM glycerol. The tubes were incubated at 37 1C until an
OD550 of 0.4 was reached. An appropriate volume of this actively
growing pre-culture was centrifuged and the pellet washed and
used to inoculate 350 mL of medium in each fermenter, with the
target starting optical density of 0.05 at 550 nm.
2.3. Analytical methods
Optical density was measured at 550 nm and used as an
estimate of cell mass (1 O.D. 0.34 g dry weight/L). After
centrifugation, the supernatant was stored at 20 1C for HPLC
(High-Performance Liquid Chromatography) and NMR (nuclear
magnetic resonance) analysis (see next section for NMR experiments). To quantify concentration of glycerol, organic acids, and
ethanol, samples were analyzed with ion-exclusion HPLC using
a Shimadzu Prominence SIL 20 system (Shimadzu Scientic
Instruments Inc., Columbia, MD) equipped with an HPX-87H
organic acid column (Bio-Rad, Hercules, CA). Operating conditions to optimize peak separation (30 mM H2SO4 in mobile phase
at 0.3 mL/min, column temperature 42 1C) were determined
using a previously described approach (Dharmadi and Gonzalez,
2005).
2.4. NMR experiments
The identity of the fermentation products was determined
through NMR experiments. Sample preparation and initial

237

characterization through 1D 1H NMR spectroscopy was conducted

in a Varian 500 MHz Inova spectrometer equipped with a Penta
probe (Varian, Inc., Palo Alto, CA), as previously described
(Dharmadi et al., 2006) (Murarka et al., 2008). Further characterization of the samples was achieved through a 2D 1H-1H COSY
(COrrelation SpectroscopY) NMR experiment. A double-quantum
ltered COSY spectrum with watergate solvent suppression was
obtained using the wgdqfcosy pulse sequence that is part of the
BioPack suite of pulse sequences (Varian, Inc., Palo Alto, CA). The
following parameters were used: 6000 Hz sweep width; 0.5 s
acquisition time; 600 complex points in t1 dimension; 32
transients; 5.5 ms pulse width; and 1 s relaxation delay. The
resulting spectra were analyzed using FELIX 2001 software
(Accelrys Software Inc., Burlington, MA).

2.5. Enzyme activities

Cells from anaerobic cultures (OD550 of 0.7) were harvested
by centrifugation (2 min, 10,000  g), washed twice with 9 g/L
NaCl, and stored as cell pellets at 20 1C. For enzyme assays, cells
were resuspended in 0.2 mL of the buffer used in the specic assay
and permeabilized by vortex mixing with chloroform (Osman
et al., 1987; Tao et al., 2001). Oxidative glycerol dehydrogenase
activity (i.e., toward glycerol) was assayed by measuring the
change in absorbance at 340 nm and 25 1C in a 1 mL reaction
mixture containing 2 mM MgCl2, 500 mM NAD+, 100 mM glycerol,
30 mL crude cell extract, and 100 mM of the appropriate buffer
according to the pH of the assay (see below) (Truniger and
Boos, 1994). Reductive glycerol dehydrogenase activity (i.e.,
toward hydroxyacetone, HA) was measured in a similar mixture,
but with HA and NADH replacing glycerol and NAD+, respectively.
Since strain DgldA did not grow in a medium supplemented
with 2 g/L tryptone, a 10 g/L tryptone and 5 g/L yeast extract
supplementation was used for the purpose of measuring
enzyme activities in this strain. Methylglyoxal (MG)-reducing
activities (e.g., MG reductase and aldo-keto reductases) were
measured in a 1 mL reaction mixture containing 10 mM MG,
0.1 mM NAD(P)H, and appropriate buffer (see below) (Ko et al.,
2005). 1D proton NMR spectroscopy was used to identify the
products of MG- and HA-reducing reactions. NMR measurements
were carried out after cell debris was removed from a reaction
mixture containing cell extract (30 mL), MG/HA (10 mM), coenzymes (1 mM NADH or 1 mM NADPH), buffer (100 mM potassium
phosphate, pH 7.0), and D2O. The NMR data were collected 1.5 h
after incubation at 25 1C. To study the effect of pH on enzyme
activities, the following buffers were used (pH in parenthesis):
sodium phosphate (pH 68), potassium carbonate (pH 9.5), and
sodium citrate (pH 56). Linearity of reactions (protein concentration and time) was established for all preparations. The
nonenzymatic rates were subtracted from the observed initial
reaction rates. Results are expressed as mmoles of substrate/
minute/mg of cell protein and represent averages for at least three
cell preparations.

2.6. Calculation of fermentation parameters

Maximum specic growth rates (mM, h1) were estimated by
plotting total cell concentration versus the integral of cell
concentration, and tting these plots to polynomial functions.
The slope of the curves thus obtained (a straight line during
exponential growth) was used as the average specic rate. Growth
yield (YX/S, mg cell/g glycerol) was calculated as the increase in
cell mass per glycerol consumed once the cultures reached
stationary phase.

ARTICLE IN PRESS
238

R. Gonzalez et al. / Metabolic Engineering 10 (2008) 234245

3. Results
3.1. 1,2-Propanediol as a product of glycerol fermentation: pathways
mediating its synthesis and signicance for the fermentative
utilization of glycerol
During the analysis of fermentation samples in a previous study
of glycerol fermentation by E. coli MG1655, we noted two NMR
peaks at the same chemical shifts as that of methyl protons of 1,2PDO (doublet at 1.15 ppm) (Murarka et al., 2008). These peaks were
observed in the spectra of late fermentation samples (Fig. 2A) but
were absent in the initial samples (Fig. 2B). Since the peaks for
other 1,2-PDO protons were masked by tryptone and glycerol
signals, and only part of those multiplet patterns were discernable,
we conducted a 2D 1H-1H COSY (COrrelation SpectroscopY) NMR
experiment to further investigate the identity of the molecule
generating the doublet at 1.15 ppm. The resulting spectra had the
cross peaks corresponding to the C2 and C3 protons of 1,2-PDO,
thereby conrming its synthesis (mutiplets at 1.15 and 3.88 ppm;
Fig. 2C). Our ndings represent the rst report of 1,2-PDO
production during the metabolism of glycerol in E. coli. 1,2-PDO
synthesis by wild-type E. coli was thought to be restricted to the
fermentation of 6-deoxyhexose sugars fucose and rhamnose
(Boronat and Aguilar, 1979; Hacking and Lin, 1976). We found that
the concentration of 1,2-PDO in stationary phase cultures was
about 0.570.15 mM. As we previously reported, ethanol was the

Ethanol

1,2-PDO
1.24 1.22 1.20 1.18 1.16 1.14
ppm

1.24 1.22 1.20 1.18 1.16 1.14

ppm

3.86

3.90

ppm

3.88

3.92

3.94
1.153

1.149

1.146
ppm

1.143

1.140

Fig. 2. Identication of 1,2-propanediol (1,2-PDO) as a product of glycerol

fermentation. (A) 1D-1H NMR spectrum of a late fermentation sample showing
two peaks at the same chemical shifts as those of methyl protons of 1,2-PDO
(doublet at 1.15 ppm). (B) 1D-1H NMR spectrum of a time zero sample, which
shows no evidence of these peaks. (C) 2D 1H-1H COSY (COrrelation SpectroscopY)
NMR spectrum of a late fermentation sample conrming the presence of 1,2-PDO:
COSY cross peaks of the C2 and C3 protons of 1,2-PDO at 1.15 and 3.88 ppm are
shown.

main product of glycerol fermentation along with minor amounts

of succinic, acetic, and formic acids (Murarka et al., 2008). No other
product was found in the NMR spectra of fermentation samples.
The pathways that could mediate the synthesis of 1,2-PDO
from DHAP, along with relevant enzymes and corresponding
genes, are summarized in Fig. 1B (Altaras and Cameron, 1999;
Boronat and Aguilar, 1979; Cooper, 1984; Ko et al., 2005; Misra
et al., 1996; Saikusa et al., 1987; Truniger and Boos, 1994). It is
important to note that experimental evidence for many of the
reactions is limited and a signicant portion of it originates from
the study of strains that were engineered to overproduce 1,2-PDO
from sugars. Moreover, in many cases, genes encoding the
proposed activities have not been identied and in none of the
studies was glycerol used as carbon source. Finally, note that
DHAP is a glycolytic intermediate generated during the utilization
of glycerol (see Fig. 1A and next section). In order to identify the
specic route mediating the synthesis of 1,2-PDO during the
fermentative metabolism of glycerol, we used several genetic and
biochemical approaches and the results are described in the
following.
Since the synthesis of methylglyoxal (MG) from DHAP is a
common step in the 1,2-PDO pathway, regardless of the branch
used for the conversion of MG to 1,2-PDO (Fig. 1B), we evaluated a
mutant devoid of the gene encoding MG synthase (MGS) and
demonstrated that 1,2-PDO synthesis was almost eliminated
(Fig. 3A). Residual levels of 1,2-PDO in the MGS mutant could be
due to the spontaneous conversion of DHA to MG, as previously
reported (Riddle and Lorenz, 1968). Since the main branch point in
this pathway is the reduction of MG (Fig. 1B), we also evaluated
mutants devoid of genes involved in each branch (Fig. 3A).
A decrease in 1,2-PDO levels was observed in disruption mutants
of genes encoding aldo-keto reductase activities (AKR: DyghZ,
DyeaE, and DyafB), which catalyze the conversion of MG to HA
(Fig. 3A). However, wild-type levels of 1,2-PDO were produced by
a DfucO mutant (Fig. 3A): fucO encodes a 1,2-PDO reductase
activity, which converts lactaldehyde (LAL) to 1,2-PDO (Fig. 1B).
These results suggest that the MG-HA-1,2-PDO branch is responsible for 1,2-PDO synthesis. To corroborate this hypothesis we
assayed the cells for MG-reducing enzymes (MGR) and observed
signicant activity in the presence of either NADH or NADPH
(Fig. 3B, panel I). Since these activities could be involved in the
conversion of MG to either HA or LAL, we used 1D 1H NMR
spectroscopy to characterize the reaction(s). Fig. 3B, panel II,
shows that the product of MG-reducing activities is indeed HA
(no LAL was detected). Similar results were obtained with mutant
strain DfucO (data not shown), which demonstrates that the
absence of LAL in the reaction mixture is not due to its fast
conversion to 1,2-PDO. These results agree with the analysis of
genetic mutants, which showed no changes in 1,2-PDO synthesis
upon elimination of the activity that converts LAL to 1,2-PDO
(Fig. 3A). We also demonstrated that cells fermenting glycerol
exhibit signicant HA-reducing activity (0.09570.002 mmol/min/
mg protein). In a study similar to the one discussed above, we
identied the HA-reducing activity as being NADH-dependent and
responsible for the conversion of HA to 1,2-PDO (data not shown).
We also demonstrated that the gldA gene encodes this activity, as
no HA reduction was observed with cell extracts of a DgldA
mutant. Given the central role of MG- and HA-reducing activities
on 1,2-PDO synthesis, and the signicant effect of pH on glycerol
fermentation (Dharmadi et al., 2006; Murarka et al., 2008), we
assessed the effect of pH on these activities (Fig. 3C). In all cases,
neutral to slightly alkaline conditions were found to be optimum.
The above results clearly show that the synthesis of 1,2-PDO
during glycerol fermentation occurs through the conversion
of DHAP to MG to HA to 1,2-PDO (represented by thick lines in
Fig. 1B). We further investigated whether 1,2-PDO synthesis could

ARTICLE IN PRESS

0.6

120

0.5

80

0.4

40

0.3

Cell growth (OD550)

MG1655

1,2-PDO (% of wild type)

239

1,2-PDO Mutants

0.4

0.3

4
3

0.2
2
0.1

0.0

Strain

MG1655

MG1655

Hydroxyacetone

No

Yes

Glycerol fermented (g/L)

Cell growth (OD550)

R. Gonzalez et al. / Metabolic Engineering 10 (2008) 234245

MG1655
MG1655
(pZSblank) (pZSmgsAgldA)
No

No

Fig. 4. Changes in cell growth and glycerol fermentation in response to

perturbations in 1,2-propanediol (1,2-PDO) synthesis. (A) Cell growth (white bars)
and 1,2-PDO synthesis (gray bars) in MG1655 and mutants of aldo-keto reductases
(AKR: DyeaE, DyghZ, and DyafB) and methylglyoxal synthase (MGS: DmgsA)
enzymes. Average performance for all mutants (referred to as 1,2-PDO mutants)
is shown. See Fig. 1B for role of these genes. (B) Effect of amplication of the 1,2PDO pathway on cell growth (white bars) and fermented glycerol (gray bars). All
cultures were conducted in MOPS minimal medium with no tryptone supplementation and hydroxyacetone (20 mM) was included when indicated. Plasmid
pZSmgsAgldA expresses methylglyoxal synthase (mgsA) and glycerol dehydrogenase (gldA), two key enzymes mediating the synthesis of 1,2-PDO (see Figs. 1B
and 3). Plasmid pZSblank was used as the control plasmid and contains the
backbone of plasmid pZSmgsAgldA without genes mgsA and gldA.

Fig. 3. Synthesis of 1,2-propanediol (1,2-PDO) during glycerol fermentation takes

place through the conversion of dihydroxyacetone-P to methylglyoxal (MG), and
further reduction of MG to hydroxyacetone (HA) and HA to 1,2-PDO. (A) Changes in
1,2-PDO levels in response to disruption of genes potentially involved in its
synthesis: see Fig. 1B for role of these genes. (B) Quantication of NAD(P)Hdependent MG-reducing activities in cell extracts (panel I) and identication of HA
as the product of MG reduction (panel II). Spectra for initial (upper) and nal
(lower) samples in the enzyme assays are shown. Symbols indicate MG (*), HA (.),
and acetate (#, impurity in MG). No lactaldehyde was detected. (C) Effect of pH on
MG- and HA-reducing activities. The use of NADPH or NADH as cofactor by MG
reductases (MGR) and HA as substrate by glycerol dehydrogenase (glyDH) is
indicated.

play a role on glycerol fermentation. The analysis of mutants in

which the production of 1,2-PDO had been reduced (Fig. 3A)
revealed that lower 1,2-PDO levels correlated with decreased cell
growth (Fig. 4A). Two additional experiments, in which amplication of the 1,2-PDO pathway allowed glycerol fermentation in the
absence of tryptone supplementation, provided conclusive evidence of the critical role of 1,2-PDO synthesis. In the rst of these
experiments, we supplemented the growth medium of strain
MG1655 with HA and observed that glycerol fermentation took
place in the complete absence of tryptone (an OD550 of 0.35 and
3.8 g/L of glycerol fermented; Fig. 4B). Externally provided HA was
completely converted to 1,2-PDO. In another experiment, we

increased 1,2-PDO synthesis by genetic means: that is, by

overexpressing MGS (mgsA) and glycerol dehydrogenase (gldA),
two of the enzymes we had identied as key members of the 1,2PDO pathway. Strain MG1655 (pZSmgsAgldA) grew to an OD550 of
0.14 and fermented 1.6 g/L of glycerol in the absence of any
supplement (Fig. 4B). These results clearly demonstrate the
enabling role the 1,2-PDO pathway plays on the ability of E. coli
to ferment glycerol.

3.2. Glycerol dehydrogenase and dihydroxyacetone kinase are

required for the anaerobic fermentation of glycerol in E. coli
Although the current model for glycerol metabolism in E. coli
assumes an absolute requirement of respiratory G3PDHs (Fig. 1),
we have found that G3PDH-decient strains DglpD and DglpA were
able to ferment glycerol (Murarka et al., 2008). These results not
only supported the fermentative nature of glycerol dissimilation
but, more importantly, indicated the existence of alternative
pathways for the metabolism of this compound in the absence of
electron acceptors.
Glycerol could be directly oxidized by the enzyme glycerol
dehydrogenase (GldA, a type II glycerol dehydrogenase encoded
by gldA) generating DHA. However, GldA is thought to be cryptic
in wild-type strains, not essential to any function, and its
physiological role is uncertain (Liyanage et al., 2001; Truniger
and Boos, 1994). GldA activation was reported to require the
disruption of the respiratory route, followed by mutagenesis and

ARTICLE IN PRESS
240

R. Gonzalez et al. / Metabolic Engineering 10 (2008) 234245

selection of mutants that recovered the ability to aerobically

metabolize glycerol (Jin et al., 1983; Tang et al., 1982a, b). Even in
such mutants, fermentative metabolism of glycerol was not
feasible. Interestingly, we found that GldA deciency rendered
the cells unable to anaerobically grow on or ferment glycerol
(Fig. 5, DgldA). Glycerol dehydrogenase activity was 0.2237
0.023 mmol/min/mg protein in MG1655 but undetectable in strain
DgldA. Moreover, strain DgldA recovered its ability to ferment
glycerol upon complementation with a plasmid encoded GldA
activity (Fig. 5: strain (DgldA(gldA+)). Since GldA is also involved in
the conversion of HA to 1,2-PDO (Fig. 1B), and this activity was
detected in MG1655 and abolished in mutant strain DgldA (see
previous section), the effect of the gldA mutation could be related
to blocking either the conversion of glycerol to DHA or the
synthesis of 1,2-PDO. In the rst case, the DHA produced by GldA
would need to be phosphorylated before entering the glycolytic
pathway. A PEP-dependent, nontransporting enzyme II complex
that phosphorylates DHA (DHA kinase, DHAK) has been char-

10

0.26

0.8

0.13

0.6

0.00

GldA activity
(moles/mg
protein/min.)
5

10

pH

0.4

0.2

0.0
MG1655

gldA

Glycerol fermented (g/L)

Cell growth (OD550)

1.0

0
gldA dhaKLM dhaKLM gldA
(dhaKLM+) (gldA+,dhaKLM+)
(gldA+)

Fig. 5. Identication of enzymes responsible for the fermentative dissimilation of

glycerol to glycolytic intermediate dihydroxyacetone-P (DHAP). Average performance is shown for DgldA and DdhaKLM mutants created in MG1655 and W3110
backgrounds. Strains DgldA(gldA+), DdhaKLM(dhaKLM+), and DgldA(gldA+,dhaKLM+)
carry plasmids pCA24NgldA (or pZSgldA), pZSKLM, and pZSKLMgldA, respectively.
These plasmids express E. colis glycerol dehydrogenase (GldA) and dihydroxyacetone kinase (DhaKLM) enzymes. The data for DgldA(gldA+) represent the average
of independent experiments in which strain DgldA was transformed with either
pCA24NgldA or pZSgldA. Inset shows the effect of pH on the oxidative activity of
glycerol dehydrogenase (glyDH, encoded by gldA), which converts glycerol to
dihydroxyacetone. Glycerol fermentation (white bars) and cell growth (gray bars)
are shown: values represent the means and bars standard deviations for three
samples taken once the cultures reached stationary phase.

acterized in E. coli (Gutknecht et al., 2001; Paulsen et al., 2000).

We found that this enzyme is also required for glycerol
fermentation, as a DHAK-decient strain (DdhaKLM) did not grow
fermentatively on glycerol (Fig. 5). When strain DdhaKLM was
transformed with plasmid pZSKLM (Bachler et al., 2005), expressing E. coli DHA kinase subunits DhaKLM, it recovered the ability
to grow and ferment glycerol (Fig. 5). Since E. coli DHAK is a nontransporting enzyme and DHA was not found in the fermentation
broth, we inferred that the DHA produced by the action of GldA
is not secreted into the medium but phosphorylated to DHAP
by DHAK.
Although complementation of the gldA mutation with a plasmidencoded GldA activity was weak (plasmid pZSgldA), the combined
expression of GldA and DhaKLM from plasmid pZSKLMgldA
restored glycerol fermentation in strain DgldA to those levels
observed in MG1655 (Fig. 5). Given the central role of GldA on
both dissimilation of glycerol and 1,2-PDO synthesis, the signicant
effect of pH on glycerol utilization (Dharmadi et al., 2006; Murarka
et al., 2008), and considering the unusual pH dependence reported
for GldA activity (Truniger and Boos, 1994), we assessed the effect of
pH on the GldA-dependent oxidation of glycerol (Fig. 5, inset). As
previously reported (Truniger and Boos, 1994), the highest activity
was observed at the most alkaline pH tested.
The ability to ferment glycerol among enteric bacteria is
thought to be limited to those species possessing type I glycerol
dehydrogenase (glyDH-I) and 1,3-PDODH, the latter being an
enzyme that facilitates the synthesis of 1,3-PDO from glycerol
(Fig. 1A). However, as shown in this section, a type II glycerol
dehydrogenase, glyDH-II (i.e., GldA) mediates the anaerobic
fermentation of glycerol in E. coli, a metabolic process that does
not involve a 1,3-PDO pathway. Since two other glyDHs of
unknown role have been identied in members of the Enterobacteriaceae (glyDH-III and glyDH-IV) (Bouvet et al., 1995), we
investigated their potential involvement in the anaerobic fermentation of glycerol. To this end, we tested several enteric bacteria
reported to possess glyDH types II (E. cloacae NCDC 279-56), III
(L. richardii), and IV (B. agrestis), along with an organism reported
not to possess any glyDH activity (S. plymuthica) (Bouvet et al.,
1995). Only strain E. cloacae NCDC 279-56, which like E. coli
possesses a glyDH-II, was able to ferment glycerol (Table 2). We
have also demonstrated that the fermentative metabolism of
glycerol is a general characteristic of E. coli species as other tested
strains (W3110, MC4100, and E. coli B) were also able to conduct
this metabolic process (Table 2). In all tested strains that were

Table 2
Relationship between the ability to ferment glycerol in the absence of external electron acceptors, the synthesis of 1,2-propanediol (1,2-PDO), and the presence and
inducibility of a type II glycerol dehydrogenase (glyDH-II) activity in E. coli strains and other members of the Enterobacteriaceae family
Organism

Parameter

Property

mM7SD

YX/S7SD

1,2-PDO synthesis

glyDH type

glyDH inducer

E. coli strains
MG1655
E. coli B
W3110
MC4100

0.04070.003
0.03670.002
0.03170.002
0.02970.004

32.972.9
34.172.7
32.273.1
54.978.8

Yes
Yes
Yes
Yes

II
II
II
II

GLYC/HA
GLYC/HA
GLYC/HA
GLYC/HA

Enteric bacteria
E. cloacae
L. richardii
B. agrestis
S. plymuthica

0.02270.002
NG
NG
NG

30.972.8
NG
NG
NG

Yes
No
No
No

II
III
IV
None

GLYC/HA
GLYC
HA
None

Experiments were conducted in minimal medium supplemented with 2 g/L tryptone and 110 mM glycerol. mM: Maximum specic growth rate (h1) calculated during
exponential growth. YX/S: growth yield (mg cell/g glycerol) calculated as the increase in cell mass per glycerol consumed once the cultures reached stationary phase.
Synthesis of 1,2-PDO was identied through NMR as described in Section 2. glyDH type and inducibility is as described elsewhere (Bouvet et al., 1995). GLYC, glycerol; HA,
hydroxyacetone; NG, no growth observed; SD, standard deviation.

ARTICLE IN PRESS
R. Gonzalez et al. / Metabolic Engineering 10 (2008) 234245

20% CO2
80% Argon

9
pH 7.5

0.8

0.4

0.0

Glycerol fermented (g/L)

Cell growth (OD550)

1.2

5
5
F
F
F
F
F + ) pF
+)
t
65 dh
F
F
dh atp 165 fdh atp

a (atp
G
G1 f (fdh
M
M
F
F
tp
dh
a
f
Fig. 6. Effect of perturbations in the formate-hydrogen lyase (FHL) and F0F1-H+translocating ATP (hydrol-/synthet-)ase (F0F1-ATPase) systems on glycerol
fermentation. Strains DfdhF(fdhF+) and DatpF(atpF+) carry plasmids pCA24NfdhF
and pCA24NatpF, respectively. These plasmids express E. colis formate dehydrogenase (pCA24NfdhF) and b subunit of the F0 complex of F0F1-ATPase (pCA24NatpF). Cell growth (white bars) and glycerol fermentation (gray bars) are shown.
Unless otherwise indicated, experiments were conducted at pH 6.3 and under an
argon atmosphere. Values represent the means and bars standard deviations for
three samples taken once the cultures reached stationary phase.

able to ferment glycerol, the dissimilation of this carbon source

was accompanied by the synthesis of 1,2-PDO.

3.3. Formate-hydrogen lyase and F0F1-H+-translocating ATP

(hydrol-/synthet-)ase systems facilitate the fermentative metabolism
of glycerol
We previously reported that the anaerobic metabolism of
glycerol in a rich medium required an active formate hydrogenlyase (FHL) system (Dharmadi et al., 2006). In a low-supplemented medium (this study), cell growth and glycerol fermentation at
acidic conditions were also impaired in FHL-decient strain DfdhF
(Fig. 6) (FdhF is required for FHL activity; Bagramyan and
Trchounian, 2003; Bagramyan et al., 2002; Self et al., 2004).
Strain DfdhF recovered the ability to ferment glycerol upon
complementation with a plasmid-borne fdhF gene (Fig. 6). As in
the case of rich medium (Dharmadi et al., 2006), the FHL-decient
strain partially recovered its ability to ferment glycerol when CO2,
a product of the FHL-mediated oxidation of formate, was included
in the gas atmosphere (Fig. 6). Hydrogen, the other product of
formate oxidation, has a negative effect on glycerol fermentation,
which we have demonstrated is due to its recycling and creation
of a redox imbalance (Murarka et al., 2008).
Since coupling between FHL, the F0F1-H+-translocating ATP
(hydrol-/synthet-)ase (F0F1-ATPase), and the low afnity K+
transport system TrkA has been suggested (Hakobyan et al.,
2005; Bagramyan and Trchounian, 2003; Bagramyan et al., 2002),
we investigated the potential involvement of F0F1-ATPase and
TrkA on glycerol fermentation. While mutations on both subunits
of F0F1-ATPase (DatpF and DatpD, F0 and F1 subunits, respectively)
impaired cell growth and glycerol fermentation (Fig. 6: only DatpF
mutant shown), disruption of the TrkA system had no effect (data
not shown). Strain DatpF recovered the ability to ferment glycerol
upon complementation with a plasmid-borne atpF gene (Fig. 6).
Unlike the FHL-decient mutant, a CO2-enriched atmosphere did
not allow the F0F1-ATPase-decient strain to recover its ability to
ferment glycerol (Fig. 6). It is noteworthy to mention that FHLand F0F1-ATPase-decient mutants are able to conduct fermentative metabolism of other carbon sources (Hakobyan et al., 2005;
Self et al., 2004; Bagramyan and Trchounian, 2003; Bagramyan
et al., 2002).

241

In contrast to glycerol fermentation in rich medium (Dharmadi

et al., 2006), the use of low supplementation allowed this process to
proceed at alkaline pHs, albeit less efciently (Fig. 6) (Murarka et al.,
2008). FHL-mediated cleavage of formate to CO2 and H2 has also
been shown to proceed at alkaline conditions and formate
dehydrogenase H (fdhF) is known to be required for the activity of
the FHL complex (Bagramyan and Trchounian, 2003; Bagramyan et
al., 2002). Our results also indicate that FHL is active at pH 7.5 as the
amount of formate accumulated in the culture medium (56 mM) is
only 80% of that of ethanol (70 mM): considering that pyruvate
dissimilation under fermentative conditions primarily takes place
through the action of the enzyme pyruvate-formate lyase (which
converts pyruvate to acetyl-CoA and formate), the molar concentration of formic acid should equal that of ethanol (plus acetate) in the
absence of FHL activity. We then investigated the potential role of
FHL on glycerol fermentation at alkaline conditions. Unlike acidic pH,
the fdhF mutation had no effect on cell growth or glycerol
fermentation at pH 7.5 (Fig. 6). The lack of FHL activity in the fdhF
mutant is evident in the increase in formate accumulation observed:
about 78 mM formate accumulated in the culture medium of strain
DfdhF compared to 56 mM in wild-type MG1655. Also unexpected
was the nding that disruption of the F0F1-ATPase system
completely impaired glycerol fermentation at pH 7.5 (Fig. 6). These
results indicate that at alkaline conditions the roles of FHL and F0F1ATPase systems on glycerol fermentation appear to be unrelated.

3.4. Why was glycerol fermentation not previously observed in

E. coli?
In an attempt to identify environmental determinants of
glycerol fermentation, we investigated culture conditions previously used in the study of glycerol metabolism in E. coli. Unlike
our medium, the media used in previous studies contained high
levels of potassium and sodium phosphates, presumably used to
maintain neutral to slightly alkaline conditions (pH in the 77.5
range). This medium was rst reported by Tanaka et al. (1967) and
subsequently used by other investigators (Bouvet et al., 1995;
Freedberg et al., 1971; Richey and Lin, 1972; Sprenger et al., 1989;
St. Martin et al., 1977; Tang et al., 1982a, b; Zwaig et al., 1970). We
found that supplementing our medium with 34 mM NaH2PO4 and
64 mM K2HPO4 (as in the above media) severely impaired glycerol
fermentation (Fig. 7). To distinguish whether sodium, potassium,
or phosphate contributed to the observed behavior, we added
them individually to the culture medium at the concentrations
mentioned above. Addition of sodium, as either chloride or sulfate
salt, did not have any effect (Fig. 7). Addition of either potassium
or phosphate, however, greatly limited glycerol fermentation
(Fig. 7), regardless of the counterion used. Previous medium
formulations also used concentrations of glycerol ranging from 20
to 30 mM (Bouvet et al., 1995; Freedberg et al., 1971; Richey and
Lin, 1972; Sprenger et al., 1989; St. Martin et al., 1977; Tanaka
et al., 1967; Tang et al., 1982a, b; Zwaig et al., 1970). The use of a
medium containing 20 mM glycerol, along with high levels of
phosphate and potassium, completely impaired cell growth and
glycerol fermentation at pH 7.5 (Fig. 7).
Another difference in the conditions used in our studies
relates to the cultivation system. Our experiments were conducted in fully controlled fermenters in which anaerobic conditions are maintained by circulating oxygen-free argon through the
headspace. Most studies reported in the literature, however, were
conducted in closed tubes/asks, often completely lled with
medium. Such cultivation systems promote the accumulation of
fermentation gas hydrogen, which we have shown to be detrimental for glycerol fermentation (Murarka et al., 2008). In an
experiment in which a Hungate tube was completely lled with

ARTICLE IN PRESS
R. Gonzalez et al. / Metabolic Engineering 10 (2008) 234245

0.6
20 mM glycerol

Cell growth (OD550)

0.8

0.4
0.2

8
6
4
2

Glycerol fermented (g/L)

10

pH 6.3
Closed tube, 100 mM glycerol

1.0

Yield (g succinic acid / g glycerol)

242

0.0
None

None

All

Na+

K+

PO43-

Al l

0.45
0.36
0.27
0.18
0.09
0.00

II

III

IV

All

Sodium, potassium or phosphate added

Fig. 7. Environmental determinants of glycerol fermentation in E. coli. Effect of
+
concentration of sodium (Na+), potassium (K+), phosphate (PO3
4 ), and glycerol. Na
(34 mM NaCl or 17 mM Na2SO4), K+ (128 mM KCl), and PO43 (98 mM NH4H2PO4)
were added to the medium to assess the individual impact of these ions on glycerol
fermentation. The combined effect of sodium, potassium, and phosphate was
evaluated by adding 34 mM NaH2PO4 and 64 mM K2HPO4 indicated. Unless
otherwise noted, experiments were conducted at pH 7.5, 37 1C, 110 mM glycerol,
2 g/L tryptone, and ushing the headspace with argon. Cell growth (white bars)
and glycerol fermentation are shown. Values represent the means and bars
represent the standard deviations for three samples taken once the cultures
reached stationary phase. The effect of Na+, K+, and PO3
4 was independent of the
counterion: data represent average behavior for cases in which different counterions were used.

minimal medium supplemented with the aforementioned levels

of sodium, potassium, and phosphate, and in the presence of
110 mM glycerol and 2 g/L of tryptone, no signicant cell growth
or glycerol fermentation was observed (Fig. 7). Taken together,
these results clearly demonstrate that previous attempts to
anaerobically ferment glycerol using E. coli were not successful
because the experiments were conducted at conditions that
negatively affect glycerol fermentation.
3.5. Implications of our ndings for metabolic engineering
Since the feasibility of engineering E. coli for the production of
chemicals and fuels has been extensively documented, our ndings
could enable the use of metabolic engineering strategies to develop
E. coli-based platforms for the anaerobic production of reduced
chemicals from glycerol at yields higher than those obtained from
common sugars. An example is succinic acid, whose production
from sugars is limited by the availability of reducing equivalents.
Fortunately, its synthesis from glycerol is feasible through a redoxbalanced pathway (Table 3). However, very low production of
succinate from glycerol was observed in our experiments with
MG1655 (Fig. 8), a consequence of glycerol dissimilation through
the PEP-dependent GldA-DHAK pathway (i.e., low PEP availability).
Using a combination of appropriate pH and concentration of CO2,
along with the replacement of E. coli PEP-dependent DHAK (DdhaK
mutation) with C. freundii ATP-dependent DHAK (expression from
plasmid pZSKLcf), we achieved an almost 10-fold increase in
succinate yield (Fig. 8). These results clearly demonstrate the
feasibility of developing a metabolic engineering platform for the
production of fuels and reduced chemicals from glycerol.

4. Discussion
4.1. A new model for the fermentative metabolism of glycerol in the
Enterobacteriaceae: trunk and auxiliary pathways in E. coli
The anaerobic fermentation of glycerol in enteric bacteria has
long been considered a privilege of species that have an active 1,3-

Fig. 8. Anaerobic fermentation of glycerol as a platform for the production of fuels

and reduced chemicals. Production of succinic acid from glycerol(I) MG1655: pH
6.3, argon; (II) MG1655: pH 6.3, 10% CO2, (III) MG1655: pH 7.5, 20% CO2; and (IV)
DdhaKLM(pZSKLcf): pH 7.5, 20% CO2. Plasmid pZSKLcf expresses C. freundii DHA
kinase subunits DhaKL.

Table 3
Analysis of redox balance for the conversion of glycerol into cell mass and selected
fermentation products
Pathway

Stoichiometrya (kb)

DKc (Hd)

Glycerol-cell mass
Glycerol-ethanol+formatef
Glycerol-succinate
Glycerol-1,2/1,3-PDO

C3H8O3(14/3)-3CH1.9O0.5N0.2(4.3)e
C3H8O3(14/3)-C2H6O(6)+CH2O2(2)
C3H8O3(14/3)+CO2(0)-C4H6O4(14/4)
C3H8O3(14/3)-C3H8O2(16/3)

1.1
0
0
2

(0.55H)
(0H)
(0H)
(1H)

a
Pathway stoichiometry accounts only for carbon balance between reactants
and products.
b
The degree of reduction per carbon, k, was estimated as described elsewhere
(Nielsen et al., 2003).
P
c
Degree of reduction balance (DK) is estimated as
over i reactants ni ci ki 
P
n
c
k
,
where
n
and
c
are
the
stoichiometric
coefcient and the
over j products j i j
number of carbon atoms for each compound, respectively.
d
Net redox units, H, are expressed per mole of glycerol (HNAD(P)H
FADH2 H2).
e
Cell mass formula is the average reported for E. coli (Nielsen et al., 2003).
Conversion of glycerol into cell mass neglects carbon losses as 1-C metabolites. In
consequence, the degree of reduction balance in this case represents the minimum
amount of redox units generated.
f
Similar results are obtained by considering the conversion of glycerol to
ethanol and H2-CO2.

PDO pathway (Fig. 1). The synthesis of 1,3-PDO allows the cells to
attain redox balance by consuming reducing equivalents generated during the incorporation of glycerol into cell mass (Table 3:
glycerol-1,3-PDO and glycerol-cell mass pathways). We
have demonstrated, however, that E. coli can fermentatively
metabolize glycerol in a 1,3-PDO-independent manner. The ability
to ferment glycerol was correlated to the cells capacity to
synthesize 1,2-PDO, a compound we identied as a product of
glycerol fermentation. Since the conversion of glycerol to 1,2-PDO
consumes 1 mole of reducing equivalents per mole of 1,2-PDO
synthesized (Table 3), it follows that the amount of 1,2-PDO found
in the fermentation broth (0.5 mM) would be sufcient to
provide a sink for the reducing equivalents generated in the
synthesis of cell mass (0.6 mM reducing equivalents). The latter
calculation assumes that about 20% of the cell mass originates
from glycerol (Murarka et al., 2008), and makes use of the degree
of reduction analysis shown in Table 3 for the conversion of
glycerol to cell mass. Clearly, the incorporation of glycerol into cell
mass generates excess reducing equivalents that E. coli can only
dispose off by the synthesis of 1,2-PDO. Given the low activity of
the 1,2-PDO pathway, the utilization of building blocks contained
in the tryptone reduces the use of glycerol in the synthesis of cell
mass and the redox imbalance associated with it. In agreement
with this hypothesis, we found that stimulation of the 1,2-PDO

ARTICLE IN PRESS
R. Gonzalez et al. / Metabolic Engineering 10 (2008) 234245

243

OH

xATP
HO

Cell Mass

OH

OH

Ethanol

Glycerol
yNADH
GldA
(gldA)

2NADH
O

NADH

OH

CoA

OH
HO

1, 2-propanediol

FHL ( fdhF,
hycB-I)

Acetyl-CoA

OH

Dihydroxyacetone

CO2 + H2

AdhE
(adhE )

Formate OH

PFL
( pflB)

O
GldA
( gldA)

NADH

DHAK
(dhaKLM )

Pyruvate

NAD(P)H

Hydroxyacetone

Dihydroxyacetone-P

O
HO

OH

O
HO

OH
P OH
O

Glycolysis

NADH
ATP

Phosphoenolpyruvate
O
HO O
P
O

O
OH

Fig. 9. A new paradigm for glycerol fermentation in E. coli and other enteric bacteria possessing a type II glycerol dehydrogenase (glyDH-II). The proposed trunk pathway for
the conversion of glycerol to glycolytic intermediate dihydroxyacetone-P in E. coli is composed of enzymes glycerol dehydrogenase (GldA, a glyDH-II) and dihydroxyacetone
kinase (DHAK). The pathways involved in the synthesis of 1,2-propanediol and ethanol are considered auxiliary as they enable glycerol fermentation by ensuring redox
balance and ATP generation, respectively. GldA is a key enzyme in this model and, as other type II glyDHs, it is induced by both glycerol and hydroxyacetone. The latter is an
intermediate in the synthesis of 1,2-propanediol. Coefcients x and y were used to represent the moles of ATP and NADH consumed and generated, respectively, in the
synthesis of cell mass from glycerol.

pathway led to cell growth and glycerol fermentation in the

absence of tryptone supplementation (Fig. 4B).
Based on our ndings, we propose a new model for the
fermentative metabolism of glycerol in enteric bacteria in which:
(i) the synthesis of 1,2-PDO provides a means to consume
reducing equivalents generated during synthesis of cell mass,
thus enabling redox-balanced conditions (Fig. 9; Table 3) and
(ii) the conversion of glycerol to ethanol, through a redoxbalanced pathway, fullls energy requirements by generating
ATP via substrate-level phosphorylation (Fig. 9; Table 3). The
proposed model includes an oxidative, GldA-DHAK-mediated
pathway that works in partnership with a reductive 1,2-PDOproducing pathway (Fig. 9). We propose that inducibility of GldA
(a glyDH-II) by both glycerol and HA (Truniger and Boos, 1994)
represents a metabolic footprint of the involvement of this
enzyme in both glycerol dissimilation and 1,2-PDO synthesis
(Fig. 9).
Trunk and auxiliary pathways can be readily identied in the
proposed model. Trunk pathways are those directly involved in
the generation of glycolytic intermediates. Auxiliary or enabling
pathways, on the other hand, are those enabling glycerol
fermentation by facilitating functions such as redox balance,
ATP synthesis, CO2 generation, pH homeostasis, generation of
proton motive force, etc. In this work we identied GldA-DHAK as
the trunk pathway responsible for the conversion of glycerol into
glycolytic intermediate DHAP. Ethanol and 1,2-PDO pathways
were considered auxiliary as they ensure ATP generation and
redox-balanced conditions, respectively. Additional auxiliary
pathways include the FHL-mediated conversion of formic acid to
CO2 and H2 and the F0F1-ATPase system (Fig. 10). FHL might be
important to maintain CO2 supply for cell growth (Dharmadi et al.,
2006) (Fig. 6) or in establishing a proton-motive force (PMF)
(Bagramyan and Trchounian, 2003; Hakobyan et al., 2005), the
latter required for cell growth and viability. We hypothesize that
F0F1-ATPase could take advantage of the PMF generated by FHL at
acidic conditions (Fig. 10) to perform function(s) essential to
glycerol fermentation. At alkaline conditions, on the other hand,
the lower intracellular pH (pHi), compared to extracellular pH

(pHe), permits the generation of PMF independent of FHL (Fig. 10).

For example, the diffusion of undissociated acids, such as formic
acid, across the membrane and their dissociation at the higher
pHe would result in the generation of PMF (Konings et al., 1995)
(Fig. 10). This hypothesis would explain our ndings that glycerol
fermentation at alkaline conditions is independent of FHL but still
requires F0F1-ATPase (Fig. 6). Given the fact that pHi is higher than
pHe at acidic conditions, generation of PMF would require an
active FHL, thus explaining why both systems are required (Fig. 6).
The use of the aforementioned PMF by the F0F1-ATPase system in
the generation of energy, metabolite transport, or other functions
appears to be a key factor in the fermentative metabolism of
glycerol.

4.2. Relationship between proposed pathways and environmental

determinants of glycerol fermentation
The above pathways provide a framework to explain the
observed changes in cell growth and glycerol fermentation as a
function of pH, concentrations of potassium, phosphate, and
glycerol, as well as the effect of cultivation systems that promote
or prevent the accumulation of fermentation gases hydrogen and
carbon dioxide (Fig. 7) (Dharmadi et al., 2006; Murarka et al.,
2008).
High levels of phosphate promote the decomposition of both
DHA and HA (two key intermediates in the aforementioned
pathways) and negatively affect GldA activity and its inducibility
by HA (Truniger and Boos, 1994). It is noteworthy that GldA
is the most important enzyme in the proposed model for
the fermentative metabolism of glycerol, being required in
both trunk and auxiliary pathways (Fig. 9). In addition, MG
synthase, a key enzyme responsible for 1,2-PDO synthesis
(Fig. 1B), is inhibited by high phosphate levels (Cooper, 1984;
Hopper and Cooper, 1971; Zhu et al., 2001). High concentrations
of potassium, on the other hand, increase the toxicity of MG
(Booth, 2005), a key intermediate in the synthesis of 1,2-PDO
(Fig. 1B). The low-afnity of GldA for glycerol (Km is 340 mM;

ARTICLE IN PRESS
244

R. Gonzalez et al. / Metabolic Engineering 10 (2008) 234245

H++HCO3-

H2 CO3

H2O

H2 +CO2

H+

HCOOH
H++HCOO- H2 2H+

H+

Periplasm

Hyd
F0

FocA/B

H+
QH2
Q+2H+

FRD

FHL
F1

Cytoplasm

H2 +CO2
ADP
+Pi

H+ ATP

Formate
CA

Fumarate +2H

Succinate

AcCoA

H2O

H2 CO3

Pyruvate/PEP

H++HCO3-

Glycerol

Q - Quinone pool

Fig. 10. Relationship between formate-hydrogen lyase (FHL) and F0F1-H+-translocating ATP (hydrol-/synthet-)ase (F0F1-ATPase) systems and the metabolism of formic acid,
carbon dioxide, and hydrogen. FHL and F0F1-ATPase are auxiliary systems required for glycerol fermentation in E. coli. Equilibrium reactions for formic acid (pKa 3.74) and
CO2 in water (CO2/HCO
3 pKa 6.3) are shown. Enzyme pyruvate formate lyase was assumed to mediate the conversion of pyruvate into AcCoA and formic acid (HCOOH).
Abbreviations/nomenclature: AcCoA, acetyl coenzyme A; CA, carbonic anhydrase; FHL, formate hydrogen-lyase; FocA/B, formate transporters; FRD, fumarate reductase;
Hyd, hydrogenases 1 and 2; and F0 and F1, subunits of the F0F1-ATPase.

Truniger and Boos, 1994) would explain both the requirement

of high concentrations of glycerol for its fermentative metabolism (Fig. 7) and our observation that 1030 mM glycerol
remained unmetabolized in the medium of cultures that have
reached stationary phase (Dharmadi et al., 2006; Murarka et al.,
2008).
The effect of pH on glycerol fermentation (Dharmadi et al.,
2006; Murarka et al., 2008) can also be related to its impact on the
aforementioned pathways. GldA exhibits strong pH dependence
with higher oxidative activity at very alkaline pHs (Fig. 5) and
reductive activity at neutral to alkaline conditions (Fig. 3C). Acidic
conditions reduce not only the oxidative activity of glyDH, but
also MG-reducing activities (Fig. 3C), which are required for the
synthesis of 1,2-PDO (Fig. 1B). Alkaline conditions, on the other
hand, could increase both the oxidizing activity of GldA (Fig. 5)
and the toxicity of MG (Booth, 2005), the latter a key intermediate
in the synthesis of 1,2-PDO. Clearly, the intracellular pH needs to
be carefully controlled to avoid low activities of key enzymes and
MG toxicity. For example, while an extracellular pH of 6.3
apparently could prevent MG toxicity, the cells would still require
a system to prevent the pH from falling below the levels
permissible for glycerol-oxidizing and MG- and HA-reducing
activities (Figs. 4C and 6). By converting formic acid to CO2 and
H2, the FHL system is known to prevent cytoplasmic acidication
(Sawers and Clark, 2004), which could be a reason why FHL is
required for glycerol fermentation at acidic conditions (Fig. 6).
Interestingly, despite the requirement of FHL for glycerol
fermentation under acidic conditions, its activity can generate
excess hydrogen that, if accumulated, would greatly impair this
metabolic process (see below).
The nature of the gas atmosphere determines, to a large extent,
the feasibility of glycerol fermentation (Dharmadi et al., 2006;
Murarka et al., 2008). This effect is also linked to the proposed
pathways. Fig. 10 illustrates the interrelation between formic acid,
carbon dioxide, and hydrogen metabolism and the FHL and F0F1ATPase systems during glycerol fermentation. In a recent study we
have demonstrated that the negative effect of hydrogen, a product
of formate oxidation via FHL, is due to its metabolic recycling,
which in turn generates an unfavorable internal redox state
(Murarka et al., 2008). The low activity of the 1,2-PDO pathway,
which represents the only active, redox-consuming route would

explain the detrimental effect of the redox imbalance created by

hydrogen recycling. The effect of CO2 could be related to the
proposed roles of FHL and F0F1-ATPase, namely the generation
(FHL) and utilization (F0F1-ATPase) of PMF (Fig. 10). If externally
provided, diffusion of CO2 into the cells and its capture as
bicarbonate through the action of carbonic anhydrase (Merlin
et al., 2003) would result in a decrease of pHi (Fig. 10). Under
conditions in which the pHe is controlled, like in our experiments,
CO2 supplementation could then result in an increase in the
PMF. Therefore, CO2 supplementation could at least partially
substitute for FHL in generating PMF. This would explain why the
FHL mutant partially recovered its ability to ferment glycerol in a
CO2-enriched atmosphere while no effect was seen on the F0F1ATPase mutant (Fig. 6). CO2-enrichment also had a positive impact
on glycerol fermentation by MG1655 at acidic conditions
(Murarka et al., 2008), an effect very unlikely related to a CO2limitation since oxidation of formate provides large amounts of
intracellular CO2.
Acknowledgments
This work was supported by grants from the National Research
Initiative of the U.S. Department of Agriculture Cooperative State
Research, Education and Extension Service (2005-35504-16698)
and the U.S. National Science Foundation (CBET-0645188).
We thank B.L. Wanner, F.R. Blattner, B. Erni, and H. Mori for
providing research materials, S. Moran for assistance with NMR
experiments, and K. Smith and Y. Moon for assistance with some
genetic constructs.
References
Altaras, N.E., Cameron, D.C., 1999. Metabolic engineering of a 1,2-propanediol
pathway in Escherichia coli. Appl. Environ. Microbiol. 65, 3118031185.
Bachler, C., Schneider, P., Bahler, P., Lustig, A., Erni, B., 2005. Escherichia coli
dihydroxyacetone kinase controls gene expression by binding to transcription
factor DhaR. EMBO J. 24, 283293.
Bagramyan, K., Trchounian, A., 2003. Structural and functional features of formate
hydrogen lyase, an enzyme of mixed-acid fermentation from Escherichia coli.
Biochemistry (Moscow) 68, 11591170.
Bagramyan, K., Mnatsakanyan, N., Vassilian, A., Trchounian, A., 2002. The roles of
hydrogenases 3 and 4, and the F0F1-ATPase, in H2 production by Escherichia coli
at alkaline and acidic pH. FEBS Lett. 516, 172178.

ARTICLE IN PRESS
R. Gonzalez et al. / Metabolic Engineering 10 (2008) 234245

Booth, I.R., 2005. Glycerol and methylglyoxal metabolism. In: Curtis, III, R., et al.
(Eds.), EcoSalEscherichia coli and Salmonella: Cellular and Molecular Biology.
ASM Press, Washington, DC (Chapter 3.4.3). /http://www.ecosal.orgS.
Borgnia, M.J., Agre, P., 2001. Reconstitution and functional comparison of puried
GlpF and AqpZ, the glycerol and water channels from Escherichia coli. Proc.
Natl. Acad. Sci. USA 98, 28882893.
Boronat, A., Aguilar, J., 1979. Rhamnose-induced propanediol oxidoreductase in
Escherichia coli: purication, properties, and comparison with the fucoseinduced enzyme. J. Bacteriol. 140, 320326.
Bouvet, O.M., Lenormand, P., Carlier, P., Grimont, P.A., Bouvet, O.M., 1994.
Phenotypic diversity of anaerobic glycerol dissimilation shown by seven
enterobacterial species. Res. Microbiol. 145, 129139.
Bouvet, O.M., Lenormand, P., Ageron, E., Grimont, P.A., 1995. Taxonomic diversity of
anaerobic glycerol dissimilation in the Enterobacteriaceae. Res. Microbiol. 146,
279290.
Cooper, R.A., 1984. Metabolism of methylglyoxal in microorganisms. Annu. Rev.
Microbiol. 38, 4968.
Datsenko, K.A., Wanner, B.L., 2000. One-step inactivation of chromosomal genes in
Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. USA 97,
66406645.
Dharmadi, Y., Gonzalez, R., 2005. A better global resolution function and a novel
iterative stochastic search method for optimization of high-performance liquid
chromatographic separation. J. Chromatogr. A 1070, 89101.
Dharmadi, Y., Murarka, A., Gonzalez, R., 2006. Anaerobic fermentation of glycerol
by Escherichia coli: a new platform for metabolic engineering. Biotechnol.
Bioeng. 94, 821829.
Freedberg, W.B., Kistler, W.S., Lin, E.C.C., 1971. Lethal synthesis of methylglyoxal by
Escherichia coli during unregulated glycerol metabolism. J. Bacteriol. 108,
137144.
Gutknecht, R., Beutler, R., Garcia-Alles, L.F., Baumann, U., Erni, B., 2001. The
dihydroxyacetone kinase of Escherichia coli utilizes a phosphoprotein instead of
ATP as phosphoryl donor. EMBO J. 20, 24802486.
Hacking, A.J., Lin, E.C.C., 1976. Disruption of the fucose pathway as a consequence
of a genetic adaptation to propanediol as a carbon source in Escherichia coli.
J. Bacteriol. 126, 11661172.
Hakobyan, M., Sargsyan, H., Bagramyan, K., 2005. Proton translocation coupled to
formate oxidation in anaerobically grown fermenting Escherichia coli. Biophys.
Chem. 115, 5561.
Hopper, D.J., Cooper, R.A., 1971. The regulation of Escherichia coli methylglyoxal
synthase: a new control site in glycolysis? FEBS Lett. 13, 213216.
Jin, R.Z., Tang, J.C., Lin, E.C., 1983. Experimental evolution of a novel pathway for
glycerol dissimilation in Escherichia coli. J. Mol. Evol. 19, 429436.
Kang, Y., Durfee, T., Glasner, J.D., Qiu, Y., Frisch, D., Winterberg, K.M., Blattner, F.R.,
2004. Systematic mutagenesis of the Escherichia coli genome. J. Bacteriol. 186,
49214930.
Kitagawa, M., Ara, T., Arifuzzaman, M., Ioka-Nakamichi, T., Inamoto, E., Toyonaga,
H., Mori, H., 2005. Complete set of ORF clones of Escherichia coli ASKA library (a
complete set of E. coli K-12 ORF archive): unique resources for biological
research. DNA Res. 12, 291299.
Ko, J., Kim, I., Yoo, S., Min, B., Kim, K., Park, C., 2005. Conversion of methylglyoxal to
acetol by Escherichia coli aldo-keto reductases. J. Bacteriol. 187, 57825789.
Konings, W.N., Lolkema, J.S., Poolman, B., 1995. The generation of metabolic energy
by solute transport. Arch. Microbiol. 164, 235242.
Lin, E.C., 1976. Glycerol dissimilation and its regulation in bacteria. Annu. Rev.
Microbiol. 30, 535578.
Liyanage, H., Kashket, S., Young, M., Kashket, E.R., 2001. Clostridium beijerinckii and
Clostridium difcile detoxify methylglyoxal by a novel mechanism involving
glycerol dehydrogenase. Appl. Environ. Microbiol. 67, 20042010.
Merlin, C., Masters, M., McAteer, S., Coulson, A., 2003. Why is carbonic anhydrase
essential to Escherichia coli? J. Bacteriol. 185, 64156424.
Misra, K., Banerjee, A.B., Ray, S., Ray, M., 1996. Reduction of methylglyoxal in
Escherichia coli K12 by an aldehyde reductase and alcohol dehydrogenase. Mol.
Cell Biochem. 156, 117124.
Murarka, A., Dharmadi, Y., Yazdani, S.S., Gonzalez, R., 2008. Fermentative
utilization of glycerol in Escherichia coli and its implications for the production
of fuels and chemicals. App. Environ. Microbiol. 74, 11241135.
Neidhardt, F.C., Bloch, P.L., Smith, D.F., 1974. Culture medium for enterobacteria.
J. Bacteriol. 119, 736747.

245

Nielsen, J., Villadsen, J., Liden, G., 2003. Bioreaction Engineering Principles. Kluwer
Academic/Plenum Publishers, New York, pp. 6073.
Osman, Y.A., Conway, T., Bonetti, S.J., Ingram, L.O., 1987. Glycolytic ux in
Zymomonas mobilis: enzyme and metabolite levels during batch fermentation.
J. Bacteriol. 169, 37263736.
Paulsen, I.T., Reizer, J., Jin, R.Z., Lin, E.C.C., Saier Jr., M.H., 2000. Functional genomic
studies of dihydroxyacetone utilization in Escherichia coli. Microbiology 146,
23432344.
Pettigrew, D.W., Yu, G.J., Liu, Y., 1990. Nucleotide regulation of Escherichia coli
glycerol kinase: initial-velocity and substrate binding studies. Biochemistry 29,
86208627.
Quastel, J.H., Stephenson, M., 1925. Further observations on the anaerobic growth
of bacteria. Biochem. J. 19, 660666.
Quastel, J.H., Stephenson, M., Whetham, M.D., 1925. Some reactions of resting
bacteria in relation to anaerobic growth. Biochem. J. 14, 304316.
Richey, D.P., Lin, E.C.C., 1972. Importance of facilitated diffusion for effective
utilization of glycerol by Escherichia coli. J. Bacteriol. 112, 784790.
Riddle, V., Lorenz, F.W., 1968. Nonenzymic, polyvalent anion-catalyzed formation
of methylglyoxal as an explanation of its presence in physiological systems.
J. Biol. Chem. 243, 27182724.
Saikusa, T., Rhee, H., Watanabe, K., Murata, K., Kimura, A., 1987. Metabolism of
2-oxoaldehydes in bacteria: purication and characterization of methylglyoxal
reductase from E. coli. Agric. Biol. Chem. 51, 18931899.
Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular Cloning: A Laboratory
Manual, second ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor,
NY.
Sawers, R.G., Clark, D.P., 2004. Fermentative pyruvate and acetyl-coenzyme A
metabolism. In: Curtis, III, R., et al. (Eds.), EcoSalEscherichia coli and
Salmonella: Cellular and Molecular Biology. ASM Press, Washington, DC
(Chapter 3.5.3). /http://www.ecosal.orgS.
Schryvers, A., Weiner, J.H., 1982. The anaerobic sn-glycerol-3-phosphate dehydrogenase: cloning and expression of the glpA gene of Escherichia coli and
identication of the glpA products. Can. J. Biochem. 60, 224231.
Self, W.T., Hasona, A., Shanmugam, K.T., 2004. Expression and regulation of a silent
operon, hyf, coding for hydrogenase 4 isoenzyme in Escherichia coli. J. Bacteriol.
186, 580587.
Sprenger, G.A., Hammer, B.A., Johnson, E.A., Lin, E.C.C., 1989. Anaerobic growth of
Escherichia coli on glycerol by importing genes of the dha regulon from
Klebsiella pneumoniae. J. Gen. Microbiol. 135, 12551262.
St. Martin, E.J., Freedberg, W.B., Lin, E.C.C., 1977. Kinase replacement by a
dehydrogenase for Escherichia coli glycerol utilization. J. Bacteriol. 131,
10261028.
Tanaka, S., Lerner, S.A., Lin, E.C.C., 1967. Replacement of a phosphoenolpyruvatedependent phosphotransferase by a nicotinamide adenine dinucleotide-linked
dehydrogenase for the utilization of mannitol. J. Bacteriol. 93, 642648.
Tang, J.C., Forage, R.G., Lin, E.C.C., 1982a. Immunochemical properties of NAD+
linked glycerol dehydrogenases from Escherichia coli and Klebsiella pneumoniae.
J. Bacteriol. 152, 11691174.
Tang, J.C., St. Martin, E.J., Lin, E.C.C., 1982b. Derepression of an NAD-linked
dehydrogenase that serves an Escherichia coli mutant for growth on glycerol.
J. Bacteriol. 152, 10011007.
Tao, H., Gonzalez, R., Martinez, A., Rodriguez, M., Ingram, L.O., Preston, J.F.,
Shanmugam, K.T., 2001. Engineering a homo-ethanol pathway in Escherichia
coli: increased glycolytic ux and levels of expression of glycolytic genes
during xylose fermentation. J. Bacteriol. 183, 29792988.
Truniger, V., Boos, W., 1994. Mapping and cloning of gldA, the structural gene of the
Escherichia coli glycerol dehydrogenase. J. Bacteriol. 176, 17961800.
Walz, A.C., Demel, R.A., de Kruijff, B., Mutzel, R., 2002. Aerobic sn-glycerol-3phosphate dehydrogenase from Escherichia coli binds to the cytoplasmic
membrane through an amphipathic a-helix. Biochem. J. 365, 471479.
Yazdani, S.S., Gonzalez, R., 2007. Anaerobic fermentation of glycerol: a path to
economic viability for the biofuels industry. Curr. Opin. Biotechnol. 18,
213219.
Zhu, M.M., Skraly, F.A., Cameron, D.C., 2001. Accumulation of methylglyoxal in
anaerobically grown Escherichia coli and its detoxication by expression of the
Pseudomonas putida glyoxalase I gene. Metab. Eng. 3, 218225.
Zwaig, N., Kistler, W.S., Lin, E.C.C., 1970. Glycerol kinase, the pacemaker for the
dissimilation of glycerol in Escherichia coli. J. Bacteriol. 102, 753759.