Analytica Chimica Acta 556 (2006) 69–79

Review

Advanced analytical tools in proteomics

Resmi C. Panicker a , Souvik Chattopadhaya b , Shao Q. Yao a,b,∗
a Department of Chemistry, National University of Singapore, 3 Science Drive 3, Singapore 117543, Singapore
b Department of Biological Science, National University of Singapore, 3 Science Drive 3, Singapore 117543, Singapore

Received 27 March 2005; received in revised form 16 May 2005; accepted 26 May 2005
Available online 1 July 2005

Abstract

Proteomics deals with the study of proteins, their structures, localizations, posttranslational modifications, functions and interactions with
other proteins. The mapping of protein structure–function holds the key to a better understanding of cellular functions under both normal and
disease states, which is critical for modern drug discovery. However, the study of human proteome presents scientists with a task much more
daunting than the human genome project. In fact, the estimated >100,000 different proteins expressed from 30,000 to 40,000 human genes make
it extremely challenging, if not impossible with existing protein analysis techniques, to map the entire cellular functions at the translational
level. Consequently, there have been rapid advances in the techniques and methods capable of large-scale proteomic studies. Among them, the
recently developed high-throughput screening methods have enabled scientists to analyze proteins quickly and efficiently at an organism-wide
scale. Herein, we overview some of these emerging tools for high-throughput protein analysis. In particular, we focus on recent advances in the
bioassay development, which has provided sensitive and selective tools for high-throughput identification and characterizations of enzymes.
Finally, the recently developed bioimaging techniques to visualize and quantify proteins in living cells are also discussed.
© 2005 Elsevier B.V. All rights reserved.

Keywords: Microarray; Peptide; Protein; Enzyme; Proteomics; 2D gel electrophoresis; Protein labeling; Bioimaging

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
2. Analytical techniques in proteomics—an overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3. Enzyme-profiling assays on chips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.1. Enzyme assays using protein arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.2. Enzyme assays using peptide/small molecule substrate arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.3. Enzyme assays using other types of arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4. Bioimaging—an emerging technique for protein analysis in living cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5. Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

1. Introduction functions and interactions with other proteins. The mapping

of protein structure–function holds the key to a better under-
Proteomics deals with the study of proteins, their standing of cellular functions under both normal and disease
structures, localizations, posttranslational modifications, states, which is critical for modern drug discovery. However,
the study of human proteome presents scientists with a task
∗ Corresponding author. Tel.: +65 68741683; fax: +65 67791691. much more daunting than the human genome project [1].
E-mail address: chmyaosq@nus.edu.sg (S.Q. Yao). In fact, the estimated >100,000 different proteins expressed

0003-2670/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.aca.2005.05.060
70 R.C. Panicker et al. / Analytica Chimica Acta 556 (2006) 69–79
from 30,000 to 40,000 human genes make it extremely
challenging, if not impossible with existing protein analysis
techniques, to map the entire cellular functions at the transla-
tional level. Consequently, there have been rapid advances in
the techniques and methods capable of large-scale proteomic
studies. Among them, the recently developed high-
throughput screening methods have enabled scientists to
analyze proteins quickly and efficiently at an organism-wide
scale. Herein, we overview some of these emerging tools for
high-throughput protein analysis. In particular, we focus on
recent advances in the bioassay development, which has pro-
vided sensitive and selective tools for high-throughput iden-
tification and characterizations of enzymes [2–29]. Finally,
the recently developed bioimaging techniques to visualize
and quantify proteins in living cells are also discussed.
2. Analytical techniques in proteomics—an overview
The conventional two-dimensional gel electrophoresis
(2D-GE), in combination with advanced mass spectrometric
techniques, has facilitated the rapid characterization of thou-
sands of proteins in a single polyacrylamide gel. The ability of
this technique to separate thousands of proteins in a specific
cell or tissue, including their posttranslationally modified
forms, has enabled it to become a major separation technique
in protein analysis and, thus, is well suited for the global
analysis of protein expression in an organism. This, together
with the newly developed activity-based profiling approaches
that target different classes of proteins, has now allowed the
study of protein functions based on their intrinsic enzymatic
activities [5,9]. However, 2D-GE suffers from a number of
long-standing problems, including low throughput, a limited
dynamic detection range, poor reproducibility, low sensitiv-
ity, as well as its difficulties in analyzing hydrophobic, small,
and very basic or acidic proteins. Incremental improvements
in the 2D-GE technology, including the use of sensitive stain-
ing methods and higher-resolving gels, and sample fractiona-
tion prior to 2D-GE, have alleviated, but not eliminated, some
of these problems [30,31]. Differential gel electrophoresis
(DIGE) [32] and Multiplexed proteomics Approach (MP
approach) [33] are some other recent developments related to
2D-GE. Yates and co-workers introduced a multidimensional
protein identification technology (MudPIT) [34]. In this tech-
nique, multiple types of columns are coupled together to
separate proteins using different physiochemical properties in
addition to molecular weights and charges, thereby extending
the analytical range to proteins having high or low molecular
weights, as well as low-abundance and insoluble proteins.
MudPIT also allows the flexibility of incorporating a suit-
able affinity column to enable selective analysis of protein
complexes and protein–protein interactions. Among differ-
ent isotope-based proteomics techniques, the isotope-coded
affinity tagging (ICATTM ) approach utilizes stable isotope
labeling to perform quantitative analysis of paired protein
samples, followed by separation and identification of proteins
within these complex mixtures with liquid chromatography
and mass spectrometry [35]. More recently, the gel-based
ICAT has also been reported [36]. The ICAT approach is
able to simplify the analysis of the proteome mixture by two
orders of magnitude based on the occurrence of cysteine in
proteins (1.7%). However, the strategy is limited to probing
known proteins containing cysteine residues and depends on
non-specific binding to CAT reagents.
From the very beginning of the proteomic era, mass spec-
trometry (MS) has played a major role in the high-throughput
identification of proteins following separation techniques
such as 2D-GE, ICAT, etc. With the advent of the ‘soft’ ioniza-
tion techniques such as ESI and MALDI coupled with various
mass analyzers including the ion trap, time-of-flight (TOF),
quadrupole and Fourier transform ion cyclotron (FT-MS),
the analysis of large intact protein complexes is now possible
[37], thus providing complementary data to that obtained by
well-established methods in structural biology such as elec-
tron microscopy, X-ray crystallography and NMR. Another
MS-based technology for quantitative analysis of protein
mixtures is known as the surface-enhanced laser desorption
ionization-time-of-flight (SELDI-TOF) [38]. This technique
utilizes stainless steel or aluminum-based supports, or chips,
engineered with chemical or biological bait surfaces that
allow differential capture of proteins (based on the intrinsic
properties of proteins). After the removal of non-specifically
adhered proteins, the bound proteins are laser-desorbed and
ionized for MS analysis. This field is currently developed
as a prominent technique when combined with advances in
protein chip technology.
While DNA microarray has enabled the high-throughput
analysis of genetic materials in a miniaturized format
[39], the development of similar devices for the analysis
of proteins has largely been hindered till the end of 20th
century, primarily due to the more delicate nature of protein
structures. Early attempts had focused mainly on the use
of peptide libraries generated in situ using methods such
as SPOTTM synthesis [40], light-directed parallel synthesis
using photolithographic methods [41] and so forth. In
2000, MacBeath and Schreiber published their landmark
paper and put forward, for the first time, the feasibility of
using microarray-based technologies for high-throughput
protein analysis [42]. Ever since, many more examples of
array-based proteomic tools have been demonstrated, which
allow the elucidation of multiple proteins by their different
biological properties, i.e. expression, activity and protein
interactions, etc. [2,6,10]. Like DNA microarray, the beauty
of protein microarray lies in the fact that it allows the simul-
taneous determination, in a single experiment, of thousands
of proteins under a variety of parameters with only minute
amounts of samples. Most of the microarray-based screening
methods developed so far have relied on the detection of
protein functions by virtue of the non-covalent binding of
a protein with its ligand(s) thereby limiting the screening to
protein–protein, protein–ligand or antigen–antibody interac-
tions. As a result, these approaches essentially preclude the
 R.C. Panicker et al. / Analytica Chimica Acta 556 (2006) 69–79 71

screening of enzymes, which play vital roles in all cellular

processes and metabolic transformations. Traditionally,
enzymatic assays are carried out in a microplate format,
i.e. in an ELISA, where enzymes are mixed separately with
their substrates in individual wells, and their activities read
out by the measurement of substrate turnovers [10]. The
wells function as individual compartments to confine each
enzymatic assay, thus preventing its cross contaminations
from neighboring reactions. However, with the emergence
of protein microarray and related technologies, recent efforts
have been focused on the further miniaturization of enzyme
assays which may analyze multiple enzyme activities in a
microarray format, both reproducibly and quantitatively.
This will be discussed in the following paragraphs.

3. Enzyme-profiling assays on chips

Enzyme assays not only detect enzymes but also can be

used to profile enzymes in terms of their substrate preference
to generate “fingerprint” profiles that are unique for each
enzyme [43]. Such analysis reveals the type of chemical
entities accepted by the enzyme as its potential substrates,
and thereby helps in a better understanding of its catalytic
mechanism and properties. Similarly, the unique pattern
generated for an unknown enzyme using a set of known
substrates may possibly be used to delineate its identity. Con-
ceptually, enzyme-profiling assays on chips may be carried
out with any of the three existing microarray-based platform
technologies—namely the protein array, the peptide array
and the small molecule array [2,4,6]. Immobilization of these
biomolecules to solid supports is a crucial step to retain the
specific activity and has been reviewed extensively [3]. As a
viable and global assay format, the major detection schemes
to track enzyme activities and on-chip assays for enzyme
identification, profiling and the kinetic measurements are key
developments. Different types of enzyme assays have been
developed which are primarily based on well-known detec-
tion methods, i.e. fluorescence, radioactivity and chemilumi-
nescence [43]. Among them, fluorescence-based methods are
most suitable because of their operational simplicity, reliabil-
ity and greater sensitivity for low enzyme turnover reactions.
Fig. 1. (a) Principle of activity-based detection of enzymes in a microarray;
3.1. Enzyme assays using protein arrays (b) mechanism-based probes used for detection of phosphatases (PT-Cy3),
cysteine (VS-Cy3) and serine (FP-Cy3) proteases; (c) enzymatic screening
With the introduction of protein microarray as the plat- using PT-Cy3 (Panel A), VS-Cy3 (Panel B) and FP-Cy3 (Panel C) of 12
form technology for high-throughput enzyme screenings, the commercial enzymes on a microarray.
critical issue is to develop strategies that allow the simulta-
neous assessment of the activity of individual proteins while enzyme assays, including enzyme substrates, from freely dif-
minimizing cross contaminations from neighboring proteins fusing across the whole array surface thus makes it difficult,
immobilized on the same surface. As mentioned earlier, if not impossible, for microplate-based enzymatic assays to
enzyme assay in a microplate format is straightforward, as be transferred directly to a microarray format. To address
each well serves as an individual compartment to confine this problem, we recently developed a novel activity-based
each enzymatic reaction. In a protein microarray, thousands strategy that allows sensitive detection of enzymes immobi-
of protein spots are immobilized on a single flat surface. The lized in a protein microarray (Fig. 1a) [15]. Small molecule
lack of any physical barrier which prevents reagents used in probes derived from mechanism-based suicide inhibitors
72 R.C. Panicker et al. / Analytica Chimica Acta 556 (2006) 69–79
(Fig. 1b) had previously been reported only in gel-based
methods for the global analysis of enzyme expression and
functions [5]. By covalent modifications of target enzymes
in a highly selective manner, these probes facilitate the easy
identification and/or purification of proteins from complex
proteomes. Several types of probes have been developed
that target different classes of proteins on the basis of their
enzymatic activities [23–29]. More recently, affinity-based
approaches have also been developed to profile other classes
of enzymes previously inaccessible by activity-based profil-
ing approaches [22]. As depicted in Fig. 1a, we successfully
extended this activity-based enzyme profiling approach to
protein microarray. Our studies showed that these probes are
suitable for array-based detection of enzyme activities and
inhibitions [15]. Incubation of these probes with an enzyme
array leads to its reaction with the immobilized enzymes
by virtue of their activity against the inhibitors. Because the
probes are pre-labeled with a fluorescent dye, the formation
of covalent enzyme-inhibitor adducts renders the enzymes
detectable. Our strategy was successfully carried out with
a protein microarray immobilized with different classes of
enzymes, including phosphatases, cysteine proteases, serine
hydrolases etc. using a panel of different mechanism-based
probes as given in Fig. 1c.
3.2. Enzyme assays using peptide/small molecule
substrate arrays
Peptide/small molecule substrate arrays were used to
identify the enzymatic activity of proteases and other
hydrolytic enzymes, kinases and Carbohydrate-modifying
enzymes. Proteases are highly discriminating for substrate
cleavage sequences. Though a number of approaches, such
as the use of FRET peptides [44], substrate phage libraries
[45], positional-scanning libraries [46,47], mixture-based
and oriented peptide libraries [48,49] have been described
to determine the substrate specificity of proteases, they do
not possess the degree of throughput normally provided
for by microarray-based technologies. Consequently, the
complete specificity profile of a protease, as well as the
kinetic evaluation of all of its potential substrates, is not
easily attainable. To address these shortcomings, a number
of microarray-based protease profiling methods have been
developed [50–53]. By using a solid support for peptide
immobilization, Kiyonaka et al. studied protease activities
using fluorogenic peptides trapped in a 3D supra-molecular
hydrogel [50]. Gosalia and Diamond recently described
the use of a liquid-phase microarray that utilizes nanoliter
sample volumes for protease screenings [51]. Such fluid
phase reactions are advantageous as it is possible to tailor
optimized reaction conditions at each individual position on
the array. A chromogenic assay for detecting hydrolases was
described by Park and Clark [52]. In this method, hydrolases
were encapsulated in sol–gel microstructures to create the
‘solzyme’ array and hydrolysis was monitored by the color
change of a generic indicator, bromothymol blue. The array
was also used to probe the enzyme specificity and inhibition
using different peptide substrates. Ellman et al. recently
developed an elegant approach for potential high-throughput
profiling of proteases in a microarray format [53]. They made
use of combinatorial libraries of peptidyl coumarin deriva-
tives immobilized site-specifically in a peptide microarray.
7-amino-4-carbamoylmethyl coumarin (ACC) was used
as the fluorescence reporter. We independently developed
a similar approach for high-throughput identifications of
hydrolytic enzymes including esterases, lipases, proteases
and epoxide hydrolases [21]. In our method, a fluorogenic
coumarin derivative was suitably modified to generate a
series of substrates that were immobilized on a glass slide to
yield a small molecule array. Upon enzymatic cleavage of the
head/substrate, the fluorescent coumarin was “unmasked”
thereby allowing for enzymatic activity to be translated to
fluorescent readouts (Fig. 2). Besides these studies, the use
of fluorescently tagged affinity ligand as a means to quantita-
tively measure on-chip bindings of reversible and irreversible
inhibitors to proteases has been reported by Miyake and
coworkers [54].
For profiling kinases, peptide phage display [55], on-
bead peptide libraries [56], or SPOTTM technology [57]
are the early-developed methods. The earliest functional
kinase assay in a microarray format was in the example of
Kemptide phosphorylation catalyzed by cAMP-dependent
protein kinase A (PKA) [42]. In a more pervasive study, Zhu
et al. tested 119 of the 122 putative yeast kinases against
a panel of 17 different proteins immobilized in a nanowell
format. Peptides as kinase substrates were at first used by
Falsey et al. who showed the incorporation of 33 P into
an immobilized peptide by p60c-src tyrosine kinase [58].
Alternatively, Houseman et al. developed self-assembled
monolayer (SAM)-based peptide microarrays for the rapid
and quantitative evaluation of kinase activity [59–61].
A recent addition to the identification of peptidic kinase
substrates has been the use of “phospho-site” collection
arrays, which consist of libraries of peptides derived from
annotated phosphorylation sites in human proteins [62,63].
In order to avert the drawbacks of radioactivity-based detec-
tion methods used in the aforementioned studies without
the loss in sensitivity, our group developed an alternative
detection scheme based on the use of fluorescently labeled
anti-phosphoamino antibodies as depicted in Fig. 3a [19].
In recent years, another fluorescence-based method for
the detection of phosphorylated peptides/proteins has also
become commercially available, which makes use of the
dye ProQ DiamondTM [64,65]. To further expand the utility
of microarray-based techniques in the profiling of kinase
substrate specificities, we developed a combinatorial method
to characterize kinase substrates (Fig. 3b) [20]. Traditional
microarrays employ a ‘one-spot, one-compound’ approach
for the direct identification of positive hits. This, however,
limits the number of compounds that can be studied in an
array to what may be synthesized, thus severely limiting
throughput. In order to study a greater diversity of compounds
 R.C. Panicker et al. / Analytica Chimica Acta 556 (2006) 69–79 73

Fig. 2. (a) Strategies for detection of hydrolytic enzymes on microarrays (route I—for proteases and phosphatases, route II—for epoxide hydrolases and
esterases); (b) chemical structure of the fluorogenic substrates (1–4) used for the detection of the different hydrolases; (c) fluorescence profiles generated on
microarray following treatment with (A) epoxide hydrolase from Rhodococcus rhodochrous, (B) acetylcholine esterase from Electrophorus electricus, (C)
trypsin (D) alkaline phosphatase. 7-hydroxy-4-carboxymethylcoumarin was used as control for all the experiments.

more rapidly and efficiently, it is imperative that alternative have shown that, by spotting libraries with various combina-
strategies like combinatorial peptide synthesis be suitably tions of peptide sequences, it is possible to draw conclusions
employed together with array-based methods to facilitate about positive hits or substrate specificity without generating
the development of next-generation peptide microarrays. We large numbers of peptides sequences individually.
74 R.C. Panicker et al. / Analytica Chimica Acta 556 (2006) 69–79

Fig. 3. (a) Overall strategy for on-chip detection of kinase substrates using anti-phosphoamino antibodies; (b) intensity profiles and image obtained with the
different combinatorial libraries following treatment with p60c-src kinase: (1–6) deletion libraries, (7–13) alanine-scanning libraries, (16–25) positional-scanning
libraries, (14) full combinatorial mixture peptide libraries, (15) original putative substrate. All peptides were synthesized with a CGG linker at the N-terminus
and immobilized on PEGylated thioester slides.

A major impediment in glycobiology has been the lack of
convenient and effective tools for the synthesis and analysis
of complex and structurally diverse carbohydrates [66]. Sig-
nificant advances in chemical and enzymatic methods for the
solid-phase synthesis of complex carbohydrates in the past
few years have eased the development of tools to explore new
vistas in carbohydrate biology [67]. On the technology front,
carbohydrate arrays have been investigated in a series of
reports [68,69]. The usefulness of these arrays for functional
enzymatic studies has also been demonstrated. Houseman
and Mrksich reported a carbohydrate microarray in which
Diels–Alder reaction was used for carbohydrate immobiliza-
tion [59]. They subsequently used the “carbochip” to profile
the binding specificities between rhodamine-labeled lectins
and different monosaccharides. They also characterized
the substrate specificity of ␤-1,4-galactosyltransferase
and quantitatively estimated the inhibitory concentration of
␣-methyl mannose for concanavalin A in competitive binding
studies. In a related study, the same group profiled the binding
specificity of lectins against an array of monosaccharide-thiol
conjugates immobilized onto a self-assembled monolayer
(SAM)-derivatized with maleimide groups [60]. Park et
al. reported the fabrication of carbohydrate arrays by
immobilization of maleimide-terminated carbohydrates on
thiol-coated glass slides [70]. Using these arrays, the group
demonstrated both qualitative and quantitative bindings of
lectins to ␣-, ␤- and N-linked carbohydrates.
3.3. Enzyme assays using other types of arrays
Schultz and colleagues reported the development of a pep-
tide nucleic acid (PNA)-tagged, small molecule array that, in
conjunction with the standard DNA microarray technology,
can be used to monitor the level and activities of enzymes in
a proteomic scale [71,72]. Using the split-pool combinatorial
method, a small molecule library based on mechanism-based
inhibitors of cysteine proteases was synthesized, with each
member of the library having a PNA tag. The PNA tags did not
 R.C. Panicker et al. / Analytica Chimica Acta 556 (2006) 69–79 75

Fig. 4. (a) Schematic representation of expression display; (b) structure of the activity-based probe specific for identification of protein tyrosine phosphatases
(PTPs); (c) hybridization of Cy3-labeled reverse transcripts following in vitro selection using activity-based probe onto a “decoding” DNA microarray containing
384 yeast ORFs (in duplicates). The spotting pattern is indicated by white circles. Zoomed regions represent positive hits.

alter the activity/selectivity of these small molecules and, at
the same time served as molecular “barcodes” which could be
used to decode the library by hybridization to a suitable DNA
microarray. Alternatively, the utility of this method to monitor
enzymatic activities in a proteomic scale was demonstrated
by the detection and characterization of the caspase-3 activity
in crude extracts of cells undergoing apoptosis triggered by
Granzyme B. Since the enzyme activity governs its binding
to the small molecule, the approach may potentially be used
to guide the design of drugs against diseased phenotypes.
Our own inputs in the area of genome-wide identification
of enzyme activities using the DNA microarray technology
have been through the development of a technique termed
“expression display” as shown in Fig. 4a which combines the
advantages of three different techniques—ribosome display,
DNA microarray and activity-based enzyme profiling [11].
76 R.C. Panicker et al. / Analytica Chimica Acta 556 (2006) 69–79
In this method, a mixture of proteins was expressed, in a
single vessel, from their cDNA library by ribosome display,
which ensured that each protein member was linked and
thus encoded by its own mRNA sequence, via the formation
of a stable mRNA–ribosome–protein ternary complex.
Subsequent functional selections using an activity-based
small molecule probe enriched only complexes that contain
proteins having the desired enzymatic activity as given in
Fig. 4b. The identity of these enzymes was then decoded
in a high-throughput manner, following hybridization of
their tethered mRNA to a DNA microarray. So far we have
demonstrated that this approach could be used to isolate and
identify, in a single experiment, representative yeast tyrosine
phosphatases from a large pool of unrelated proteins as
depicted in Fig. 4c. We envision that, in future, this work
might be extended to the high-throughput screening of other
enzyme classes.
4. Bioimaging—an emerging technique for protein
analysis in living cells
While most enzyme assays are performed with proteins
either purified to homogeneity or present in a crude cell
lysate, it is much more meaningful if one can assess a
target enzyme directly under its own cellular environments.
In a broader sense, analytical techniques that allow the
visualization and quantitation of proteins, including
enzymes, in living cells is essential to study their structure
and functions within the biological context. With the advent
of fluorescent proteins and other molecular labels, cell
biologists and protein chemists have witnessed a better
understanding of intracellular proteins and subcellular events.
Together with the advancement in fluorescence and confocal
laser scanning microscopies (CLSM), 3D imaging of the
cellular events and protein dynamics can now be detected in
real time within living cells [73]. However, the main draw-
backs of GFP-like proteins include their large sizes, obligate
oligomerization, and sometimes slow or incomplete matura-
tion. Ideally, a good protein labeling strategy should satisfy
the following criteria: (1) possesses a high signal-to-noise
ratio (i.e. high specificity for target protein); (2) upholds the
integrity of the labeled protein; (3) does not interfere with the
biochemical functions or cellular localization of the labeled
protein; (4) has minimal perturbation to the normal cellular
processes. Many of these could be met by small molecule-
based labeling strategies, which in recent years have become
increasingly feasible for bioimaging experiments, as a result
of advances in the synthesis of novel fluorescent dyes, as
well as the development of new bioconjugation strategies
which allow for highly specific and efficient incorporation of
molecular probes to proteins expressed in live cells [74,75].
Since probes used for protein labeling in these strategies are
typically small, they likely do not perturb the function of the
labeled protein, thus making them attractive over FP-based
strategies. In the following paragraphs, we will summarize
some of the small molecule-based protein labeling tech-
niques developed in recent years, which holds great promises
in future high-throughput protein analysis in live cells.
A few groups, including our own, have developed novel
strategies for specific labeling of proteins with small molecu-
lar probes in live cells. Different reported strategies are based
on unique chemistries and bio-interactions. Of different
small molecule probes used, a number of criteria should
be carefully considered, including their cell permeability,
cytotoxicity, specific reactivity, good fluorescent properties,
etc. The first innovative method for the site-specific labeling
of recombinant protein with small organic molecules within
live cells was developed by Tsien et al. [76]. Their method
exploits the well-known specificity of organoarsenicals with
pairs of thiols. Here a short peptide sequence CCXXCC (in
which X is a non-cysteine amino acid) was genetically fused
to the protein of interest, which was subsequently recognized
and labeled by a cell permeable fluorescein derivative called
FlAsH as shown in Fig. 5a. The free FlAsH label is virtually
non-fluorescent, but acquiring bright fluorescence only upon
binding to CCXXCC motif present in the target protein [76].
Tsien et al. chose ECFP (enhanced cyan fluorescent protein)
as a model protein and genetically fused a CCXXCC motif
to its C-terminus. Upon transient expression in HeLa cells
and treatments of the cells with FlAsH, the formation of
a specifically labeled protein–small molecule complex
was successfully detected by monitoring the fluorescence
resonance energy transfer (FRET) from ECFP to FlAsH.
The age and life cycle of protein molecules could be studied
by sequential labeling with differently colored biarsenical
labels. Gaietta et al. applied this strategy to monitor the
translocation of connexin in and out of gap junctions [77].
Another protein-labeling method was recently developed by
Kepplar et al., which covalently labels a target protein fused
to the human DNA repair protein O6 -alkylguanine-DNA
alkyltransferase, or hAGT [78]. Specific covalent labeling
of the target protein with a small molecule was achieved
through the highly specific enzymatic reaction between
hAGT and the small molecule. The cellular role of hAGT is to
irreversibly transfer the alkyl group from O6 -alkylguanine-
DNA to one of its reactive cysteine residues, which results
in the repair of the alkylated DNA. The authors brilliantly
exploited this chemistry to label the hAGT fusion protein
expressed in live mammalian cells as depicted in Fig. 5b.
Our group has developed a novel strategy for site-specific
covalent labeling of proteins in vivo. In our strategy, a pro-
tein of interest bearing an N-terminal cysteine is expressed
inside a live cell by intein-mediated protein splicing. We
selected the 17 kDa Ssp DnaB mini intein, whose splicing
activity was genetically engineered to occur efficiently under
physiological conditions, to generate an N-terminal cysteine
in the target protein. Chemoselective native chemical lig-
ation reaction between the thioester containing probe and
the N-terminal cysteine of the protein under physiological
cellular conditions results in the specific protein labeling
as given in Fig. 5c. Our strategy provides a simple way of
 R.C. Panicker et al. / Analytica Chimica Acta 556 (2006) 69–79 77

Fig. 5. (a) Site-specific labeling of tetracysteine motif by FlAsH; (b) mechanism of labeling of hAGT fusion proteins with BG derivative; (c) labeling mechanism
of thioester probe with the N-terminal cysteine proteins in live cells; (d) NTA probe coordinating to Ni (II) ion; (e) labeling of fusion proteins based on interactions.

site-specifically labeling proteins in live cells, with little mod-
ification to the original protein apart from the addition of at
most a few extra amino acids. We have successfully shown its
utility in specific covalent labeling of proteins in bacteria [79],
as well as in mammalian cells [80]. More recently, our group
developed a proteomic method that allows identification of
caspases and their associating proteins from live apoptotic
cells [29]. We found that efficient in vivo labeling of caspases
expressed inside apoptotic HeLa cells could be achieved
using fluoromethylketone (fmk)-containing, activity-based
small molecule probes. We anticipate that, given the increas-
ing availability of activity-based probes that target other
enzymes [22,23], our approach may provide a viable tool for
future high-throughput studies of enzyme-associating pro-
teins.
Recently, Vogel et al. reported a generic method for the
site-selective and reversible labeling of membrane proteins
containing a polyhistidine sequence in live cells with small
molecular probes [81]. This probe consists of a chromophores
and a metal ion chelating to the nitrilotriacetate (NTA) moi-
ety. The labeling is based on the well-known interaction
between the oligohistidine sequence and the NTA moiety.
The chemical structure of the NTA probe used for the experi-
ments is shown in Fig. 5d. By applying this approach, the
authors were able to investigate the structure and plasma
membrane distribution of the ionotropic 5HT3 serotonin
receptor (5HT3 R). A separate method developed by Tsien’s
group takes advantage of the very high affinity of avidin for its
small molecule ligand, biotin (10−15 M) as shown in Fig. 5e.
In their approach, avidin functions as the genetically encoded
tag and the derivatized biotin acts as the label. The luminal
pH of specific organelles expressing an avidin-fused pro-
tein was investigated using a cell-permeable, pH-sensitive
fluorescein-biotin derivative [82]. Another strategy involv-
ing the selective binding of a chemical ligand to a receptor
protein is that based on the hapten–antibody interaction as
given in Fig. 5e. A membrane-permeant phenyloxazolinone
hapten conjugated to a pH-dependant fluorescent dye was
78 R.C. Panicker et al. / Analytica Chimica Acta 556 (2006) 69–79
trapped by a single chain antibody that had been expressed in
the lumen of the organelles and used to study pH regulation
in different compartments along the cellular secretory path-
way [83]. The noncovalent interaction between dihydrofolate
reductase (DHFR) and methotrexate (Mtx)-containing small
molecule derivates is a good alternative to avidin–biotin and
hapten–antibody interactions as depicted in Fig. 5e. The Cor-
nish group very recently exploited this binding as a generic
method for in vivo labeling of DHFR-fused proteins using
fluorescent, small molecule probes conjugated to Mtx [84].
They were able to successfully label DHFR-fused proteins
targeted to the nucleus and plasma membrane in DHFR-
deficient CHO mammalian cells.
Inteins are internal polypeptide sequences that are
posttranslationally excised from a protein precursor by
a self-catalyzed protein-splicing reaction. The result is
the concurrent ligation of the two flanking polypeptides
sequences to generate a mature, functional protein. In
recent years, intein-based labeling methods have become an
important technique in protein engineering [85]. Recently,
Muir’s group developed a novel strategy to specifically
incorporate a synthetic molecule into a protein in vivo by
utilizing trans-splicing inteins, in which the intein is split
into two pieces and its activity is subsequently regained
upon reconstitution of these fragments [86]. By exploiting
this technique, his group was able to successfully label GFP
with a FLAG epitope inside live mammalian CHO cells.
Pioneered by the Schultz group, the suppressor tRNA
technology was an ingenious and promising technique that
enables the incorporation of unnatural amino acids into pro-
teins during in vitro transcription/translation reactions [87].
Recently, Zhang et al. designed and utilized an orthogonal
tRNA–TyrRS pair that selectively and efficiently incorpo-
rated the unnatural amino acid, m-acetyl-(l)-phenylalanine,
into a protein overexpressed in E. coli, and successfully intro-
duced a “ketone handle” into the target protein [88]. This pro-
tein was subsequently modified, in living cells, by hydrazide-
containing small molecules using the ketone/hydrazide
chemistry, demonstrating for the first time the potential use
of tRNA technology for live cell bioimaging purposes. The
labeling reaction was selective, and in general proceeded with
relatively good yields (e.g. >75% ligation product).
5. Concluding remarks
As new roles of enzymes emerge, many more novel appli-
cations of enzyme assays can be envisioned. Rapid develop-
ments in enzymatic assays will continue to be fueled by the
identification of new enzymes stemming from the growth in
genome information which has provided a sequence-based
framework for data mining. With new analytical techniques
for high-throughput protein analysis becoming increasing
available, it is time now time to look ahead that in the coming
years, it will be entirely possible that one may be able track
enzymes en masse in their own cellular environments.
Acknowledgements
Funding support was provided by the National University
of Singapore (NUS) and the Agency for Science, Technology
and Research (A*STAR) of Singapore. S.Q.Y. is the recip-
ient of the 2002 Young Investigator Award (YIA) from the
Biomedical Research Council (BMRC).
References
[1] C. Ezzell, Sci. Am. 283 (2000) 64.
[2] M. Uttamchandani, D.P. Walsh, S.Q. Yao, Y.T. Chang, Curr. Opin.
Chem. Biol. 9 (2005) 4–13.
[3] D.S.Y. Yeo, R.C. Panicker, L.P. Tan, S.Q. Yao, Comb. Chem. High
Throughput Screen. 7 (2004) 213.
[4] R.C. Panicker, X. Huang, S.Q. Yao, Comb. Chem. High Throughput
Screen. 7 (2004) 547.
[5] Y. Hu, X. Huan, G.Y.J. Chen, S.Q. Yao, Mol. Biotechnol. 28 (2004)
63.
[6] M. Uttamchandani, X. Huang, G.Y.J. Chen, L.P. Tan, S.Q. Yao, Mol.
Biotechnol. 28 (2004) 227.
[7] R.Y.P. Lue, G.Y.J. Chen, Q. Zhu, M.L. Lesaicherre, S.Q. Yao, Meth.
Mol. Biol. 264 (2004) 85.
[8] M. Uttanchandani, G.Y.J. Chen, M.L. Lesaicherre, S.Q. Yao, Meth.
Mol. Biol. 264 (2004) 191.
[9] X. Huang, E.L.P. Tan, G.Y.J. Chen, S.Q. Yao, Appl. Genomics Pro-
teomics 2 (2003) 225.
[10] G.Y.J. Chen, M. Uttamchandani, R.Y.P. Lue, M.L. Lesaicherre, S.Q.
Yao, Curr. Top. Med. Chem. 3 (2003) 705.
[11] Y. Hu, G.Y.J. Chen, S.Q. Yao, Angew. Chem. Int. Ed. 44 (2005)
1048.
[12] L.P. Tan, R.Y.P. Lue, G.Y.J. Chen, S.Q. Yao, Bioorg. Med. Chem.
Lett. 14 (2004) 6067.
[13] M. Uttamchandani, D.P. Walsh, S.M. Khersonsky, X. Huang, S.Q.
Yao, Y.T. Chang, J. Comb. Chem. 6 (2004) 862.
[14] R.Y.P. Lue, G.Y.J. Chen, Y. Hu, Q. Zhu, S.Q. Yao, J. Am. Chem.
Soc. 126 (2004) 1055.
[15] G.Y.J. Chen, M. Uttamchandani, Q. Zhu, G. Wang, S.Q. Yao, Chem.
Biochem. 4 (2003) 336.
[16] Y. Hu, G. Wang, G.Y.J. Chen, X. Fu, S.Q. Yao, Electrophoresis 24
(2003) 1458.
[17] M.L. Lesaicherre, R.Y.P. Lue, G.Y.J. Chen, Q. Zhu, S.Q. Yao, J.
Am. Chem. Soc. 124 (2002) 8768.
[18] M.L. Lesaicherre, M. Uttamchandani, G.Y.J. Chen, S.Q. Yao, Bioorg.
Med. Chem. Lett. 12 (2002) 2079.
[19] M.L. Lesaicherre, M. Uttamchandani, G.Y.J. Chen, S.Q. Yao, Bioorg.
Med. Chem. Lett. 12 (2002) 2085.
[20] M. Uttamchandani, E.W.S. Chan, G.Y.J. Chen, S.Q. Yao, Bioorg.
Med. Chem. Lett. 13 (2003) 2997.
[21] Q. Zhu, M. Uttamchandani, D.B. Li, M.L. Lesaicherre, S.Q. Yao,
Org. Lett. 5 (2003) 1257.
[22] E.W.S. Chan, S. Chattopadhaya, R.C. Panicker, X. Huang, S.Q. Yao,
J. Am. Chem. Soc. 126 (2004) 14435.
[23] Q. Zhu, A. Girish, S. Chattopadhaya, S.Q. Yao, Chem. Commun.
(2004) 1512.
[24] Q. Zhu, X. Huang, G.Y.J. Chen, S.Q. Yao, Tetrahedron Lett. 45
(2004) 707.
[25] Q. Zhu, X. Huang, G.Y.J. Chen, S.Q. Yao, Tetrahedron Lett. 44
(2003) 2669.
[26] G. Wang, M. Uttamchandani, G.Y.J. Chen, S.Q. Yao, Org. Lett. 5
(2003) 737.
[27] G. Wang, M. Uttamchandani, G.Y.J. Chen, S.Q. Yao, Org. Lett. 5
(2003) 737.
[28] M.L. Liau, R.C. Panicker, S.Q. Yao, Tetrahedron Lett. 44 (2003)
1043.
 R.C. Panicker et al. / Analytica Chimica Acta 556 (2006) 69–79
[29] E.L.P. Tan, R.C. Panicker, G.Y.J. Chen, S.Q. Yao, Chem. Commun.
(2005) 596.
[30] T. Rabilloud, Proteomics 2 (2002) 3.
[31] C. Gauss, M. Kalkum, M. Lowe, M.H. Lehrach, J. Klose, Elec-
trophoresis 20 (1999) 575.
[32] M. Unlu, M.E. Morgan, J.S. Minden, Electrophoresis 18 (1997)
2071.
[33] T.H. Steinberg, K. Pretty, K.N. Berggren, C. Kemper, L. Jones, Z.
Diwu, R.P. Haugland, W.F. Pattonet, Proteomics 1 (2001) 841.
[34] M.P. Washburn, D. Wolters, J.R. Yates, Nat. Biotechnol. 19 (2001)
242.
[35] S.P. Gygi, B. Rist, S.A. Gerber, F. Turecek, M.H. Gelb, R. Aebersold,
Nat. Biotechnol. 17 (1999) 994.
[36] M. Smolka, H. Zhou, R. Aebersold, Mol. Cell. Proteomics 1 (2002)
19.
[37] J.M. Daniel, S.D. Friess, S. Rajagopalan, S. Wendt, R. Zenobi, Int.
J. Mass Spectrom. 216 (2002) 1.
[38] H.J. Issaq, T.P. Conrads, D.A. Prieto, R. Tirumalai, T.D. Veenstra,
Anal. Chem. 75 (2003) 148A.
[39] D.J. Lockhart, E.A. Winzeler, Nature 405 (2000) 827.
[40] R. Frank, Tetrahedron 48 (1992) 9217.
[41] S.P. Fodor, J.L. Read, M.C. Pirrung, L. Stryer, A.T. Lu, D. Solas,
Science 251 (1991) 767.
[42] G. MacBeath, S.L. Schreiber, Science 289 (2000) 1760.
[43] J.P. Goddard, J.L. Reymond, Trends Biotechnol. 22 (2004) 363.
[44] S. Mizukami, K. Kikuchi, T. Higuchi, Y. Urano, T. Mashima, T.
Tsuruo, T. Nagano, FEBS Lett. 453 (1999) 356.
[45] D.J. Matthews, J.A. Wells, Science 260 (1993) 1113.
[46] B.J. Backes, J.L. Harris, F. Leonetti, C.S. Craik, J.A. Ellman, Nat.
Biotechnol. 18 (2000) 187.
[47] J.L. Harris, P.B. Alper, J. Li, M. Rechsteiner, B.J. Backes, Chem.
Biol. 8 (2001) 1131.
[48] B.E. Turk, L.C. Cantley, Methods 32 (2004) 398.
[49] B.E. Turk, L.L. Huang, E.T. Piro, L.C. Cantley, Nat. Biotechnol. 19
(2001) 661.
[50] S. Kiyonaka, K. Sada, I. Yoshimura, S. Shinkai, N. Kato, I. Hamachi,
Nat. Mater. 3 (2004) 58.
[51] D.N. Gosalia, S.L. Diamond, Proc. Natl. Acad. Sci. U.S.A. 100
(2003) 8721.
[52] C.B. Park, D.S. Clark, Biotechnol. Bioeng. 78 (2002) 229.
[53] C.M. Salisbury, D.J. Maly, J.A. Ellman, J. Am Chem. Soc. 124
(2002) 14868.
[54] J. Eppinger, D.P. Funeriu, M. Miyake, L. Denizot, J. Miyake, Angew.
Chem. Int. Ed. 43 (2004) 3806.
[55] G. Cassirini, FEBS Lett. 307 (1992) 66.
[56] K.S. Lam, M. Renil, Curr. Opin. Chem. Biol. 6 (2002) 353.
[57] W.R.G. Dostmann, M.S. Taylor, C.K. Nickl, J.E. Brayden, R. Frank,
W.J. Tegge, Proc. Natl. Acad. Sci. U.S.A. 97 (2000) 14772.
[58] J.R. Falsey, M. Renil, S. Park, S. Li, K.S. Lam, Bioconj. Chem. 12
(2001) 346.
[59] B.T. Houseman, J.H. Huh, S.J. Kron, M. Mrksich, Nat. Biotechnol.
20 (2002) 270.
79
[60] B.T. Houseman, M. Mrksich, Chem. Biol. 9 (2002) 443.
[61] B.T. Houseman, E.S. Gawalt, M. Mrksich, Langmuir 19 (2003)
1522.
[62] U. Schutkowski, S. Reimer, L. Panse, J.M. Dong, D.R. Lizcano, J.
Alessi, M. Schneider, Angew. Chem. Int. Ed. 43 (2004) 2671.
[63] J.M. Lizcano, M. Deak, N. Morrice, A. Kieloch, C.J. Hastie, L.
Dong, M. Schutkowski, U. Reimer, D.R. Alessi, J. Biol. Chem. 271
(2002) 27839.
[64] S. Fukui, T. Feizi, C. Galustian, A.M. Lawson, W. Chai, Nat.
Biotechnol. 20 (2002) 1011.
[65] D. Wang, S. Liu, B.J. Trummer, C. Deng, A. Weng, Nat. Biotechnol.
20 (2002) 275.
[66] K. Martin, T.H. Steinberg, L.A. Cooley, K.R. Gee, J.M. Beechem,
W.F. Patton, Proteomics 3 (2003) 1244.
[67] P. Sears, C.-H. Wong, Science 291 (2001) 2344.
[68] O.J. Plante, E.R. Palmacci, P.H. Seeberger, Science 291 (2001)
1523.
[69] D.M. Ratner, E.W. Adams, M.D. Disney, P.H. Seeberger, Chem.
Biochem. 5 (2004) 1375.
[70] S. Park, M.-R. Lee, S.-J. Pyo, I. Shin, J. Am. Chem. Soc. 126 (2004)
4812.
[71] N. Winssinger, J.L. Harris, B.J. Backes, P.G. Schultz, Angew. Chem.
Int. Ed. 40 (2001) 3152.
[72] N. Winssinger, S. Ficarro, P.G. Schultz, J.L. Harris, Proc. Natl. Acad.
Sci. U.S.A. 99 (2002) 11139.
[73] G.T. Hermanson, Bioconjugate Techniques, Academic Press, San
Diego, CA, 1996.
[74] R. Srinivasan, S.Q. Yao, D.S.Y. Yeo, Comb. Chem. High Throughput
Screen. 7 (2004) 597.
[75] D.S.Y. Yeo, R. Srinivasan, G.Y.J. Chen, S.Q. Yao, Chem. Eur. J. 10
(2004) 4664.
[76] B.A. Griffin, S.R. Adams, R.Y. Tsein, Science 28 (1998) 269.
[77] G. Gaietta, T.J. Deerinck, S.R. Adams, J. Bouwer, O. Tour, D.W.
Laird, G.E. Sosinsky, R.Y. Tsien, M.H. Ellisman, Science 296 (2002)
503.
[78] A. Keppler, S. Gendreizig, T. Gronemeyer, H. Pick, H. Vogel, K.
Johnsson, Nat. Biotechnol. 21 (2003) 86.
[79] D.S.Y. Yeo, R. Srinivasan, M. Uttamchandani, G.Y.J. Chen, Q. Zhu,
S.Q. Yao, Chem. Commun. (2003) 2870.
[80] D.S.Y. Yeo, R. Srinivasan, G.Y.J. Chen, S.Q. Yao, unpublished
results.
[81] E.G. Guignet, R. Hovius, H. Vogel, Nat. Biotechnol. 22 (2004) 440.
[82] J. Llopis, J.M. McCaffery, A. Miyawaki, M.G. Farquhar, R.Y. Tsien,
Proc. Natl. Acad. Sci. U.S.A. 95 (1998) 6803.
[83] J. Farinas, A.S. Verkman, J. Biol. Chem. 274 (1999) 7603.
[84] L.W. Miller, J. Sable, P. Goelet, M.P. Sheetz, V.W. Cornish, Angew.
Chem. Int. Ed. 43 (2004) 1672.
[85] T.M. Muir, Annu. Rev. Biochem. 72 (2003) 249.
[86] I. Giriat, T.W. Muir, J. Am. Chem. Soc. 125 (2003) 7180.
[87] J.A. Ellman, D. Mendal, P.G. Schultz, Science 255 (1992) 197.
[88] Z. Zhang, B.A. Smith, L. Wang, A. Brock, C. Cho, P.G. Schultz,
Biochemistry 4 (2003) 6735.