The field of proteomics is taking on increased significance as the relevance of investigating and understanding protein expression in disease and drug development is appreciated. Recent advances in proteomics have been driven by the availability of numerous annotated whole-genome sequences and a broad range of technological and bioinformatic developments that underscore the complexity of the proteome. This review briefly addresses some of the various technologies that comprise Expression Proteomics and Functional Proteomics, citing examples where these emerging approaches have been applied to pharmacology, toxicology, and the development of drugs.
Over the last several years, the genomes of more than a hundred organisms have been sequenced, and more are being completed each month. Genomics has provided DNA sequences for a tremendous number of bacteria, viruses, and yeasts, as well as humans and a number of model higher organisms (eg, worm, fly, fish, mouse, rat, etc). This DNA sequence information provides the foundation for Functional Genomics. This multidisciplinary field, described in detail below, utilizes various measures of mRNA and protein expression and function to better understand the basis of metabolism and disease processes.
Transcriptomics, a branch of Functional Genomics, is an approach that enables the analysis of gene expression through the detection and relative quantitation of individual messenger RNAs (mRNAs). By comparing transcription profiles of untreated vs treated or normal vs diseased cells or organisms, much can be learned about the biology of the system being studied. Transcriptomic methods usually are based on some type of microarray technology (eg, cDNA or oligonucleotide arrays). Using these methods, quantitative information can be obtained quickly on thousands or tens of thousands of transcripts at a time. Bioinformatic approaches are being developed in an attempt to make sense of this mountain of information, deriving pathway maps and inferring protein function. This, in turn, provides new insight into the gene products associated with cellular processes and disease.
Transcriptomics, though only a few years old, is already providing very useful information about mRNA expression. Ultimately, however, it is not mRNA that regulates and mediates cellular function. Instead, function is directed by the proteins translated from these mRNAs. Studies comparing mRNA and protein levels typically have shown a weak or poor correlation between them.1,2,3,4,5 To better determine what actually is happening in a cell, one must know two things: (1) the quantities and activities of the various proteins present at a given time, and (2) the level of critical post-translational modifications that regulate these protein activities. The relative quantitation of mRNA, even on a global scale, cannot provide this information, nor can it describe the dynamics of protein–protein interactions that may have dramatic effects on protein function.
Proteomics essentially is a subdiscipline of Functional Genomics that measures the qualitative and quantitative changes in protein content of a cell or tissue in response to treatment or disease and determines protein–protein and protein–ligand interactions. As a complementary approach to mRNA expression technologies, proteomics increasingly is being applied to drug discovery efforts. The pharmaceutical industry has expressed significant interest in proteomics, with the expectation that this technology will lead to the identification and validation of protein targets and, ultimately, to the discovery and development of viable drug candidates.6
Most proteomic tools are new and rapidly developing. Since each expressed protein is unique and the final protein product usually is the result of multiple post-translational processing events, wide-ranging approaches and techniques are necessary to detect, quantify, and identify individual proteins and to elucidate protein activities, protein–protein interactions, etc. Unlike the situation for DNA/RNA where the polymerase chain reaction (PCR) can be used to amplify very small amounts of material, there is no comparable methodology to amplify proteins. Consequently, protein analysis often is sample-limited, and advances in proteomic research are driven by new analytical technologies and/or increases in the sensitivities of existing measurement capabilities. This review will briefly address several of the various technologies that comprise the proteomics ‘arsenal’ and provide examples where each has been applied to pharmacology, toxicology, and/or drug development. Numerous excellent reviews covering the field have been published.7,8,9 In this article, we will not attempt to cover the earlier literature but instead will concentrate on reviewing significant developments and applications published within the last few years.
GENERAL PROTEOMICS METHODS AND STRATEGIES
Proteomics can be divided arbitrarily into two main areas, ‘Expression Proteomics’ and ‘Functional Proteomics’. In essence, Expression Proteomics involves the analysis of differential protein expression via protein quantitation and identification. Differential protein analysis compares the expression profiles of proteins (the proteomes) of cells, tissues, or organisms in one condition (eg, disease, injury, drug therapy, or intoxication) to a standard or control proteome. Differences in the proteomes provide insight into those molecular processes that are involved in a biological response, hence the utility of this approach in the development of drugs. In contrast, Functional Proteomics concerns the manner in which proteins interact and, in turn, how these interactions determine function, both normal and abnormal.
EXPRESSION PROTEOMICS: TOOLS AND APPLICATIONS
Two-Dimensional Gel Electrophoresis (2DE) and Mass Spectrometry (MS)
The combination of 2DE and MS is currently the most widely used proteomic approach and might even be called ‘classic’ or ‘blue-collar’ proteomics.10 Proteins first are separated by charge and by molecular weight using isoelectric focusing and SDS-polyacrylamide gel electrophoresis, respectively, and then stained for image analysis or visual inspection. Since the dynamic range of protein expression in most whole cell or tissue lysates is huge (estimated to be up to nine orders of magnitude, in extreme cases) and only the most abundant proteins from 2D gels can be analyzed (despite excellent mass spectrometer sensitivity), far too many proteins at low expression levels are overlooked. Typically, only 2000 or so of the most abundant proteins in a particular cell or tissue can be reliably separated and identified. Therefore, to detect the less abundant proteins in a sample, prefractionation of the starting material is required and/or very narrow pI range separations must be performed. To that end, novel sample preparation steps aimed at increasing protein recovery and resolution have been reported.11,12,13 Despite their shortcomings, 2D gel-based experiments remain an exceptional approach for assessing differential protein expression, particularly in studies with pharmacological considerations.
Mass spectrometry is the method of choice for identifying 2D gel-separated proteins. Separated protein spots can be identified after digestion with trypsin and analysis by one of several MS methods. A tryptic digest (eg, peptide mixture) can be analyzed by matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry.14 The measured and optimized monisotopic mass data then are compared with theoretically derived peptide mass databases, generated by applying specific enzymatic cleavage rules to predicted/known protein sequences. High-throughput MALDI-based peptide mass fingerprinting (PMF) is accurate and sensitive. For unambiguous identification of 2D gel-separated proteins, spectral data derived from MS/MS experiments can be compared to spectra predicted from sequence databases using search programs such as Mascot15 or SEQUEST.16 In some cases, peptide sequence tags17 from either MALDI or electrospray ionization ion sources can also be used for protein identification (provided collision-induced fragmentation and fragment ion detection is carried out). Large stretches (≥20 amino acids) of protein sequence also can be determined using chemically assisted fragmentation MALDI-PSD MS.18 This is especially useful when trying to identify proteins isolated from organisms having unsequenced genomes. Finally, with the advent of the TOF/TOF mass spectrometer, protein identification is entering a new era in which all of the above methods will become faster and more accurate.19
Since its inception, 2DE has been a useful tool in drug development, especially in the analysis of adverse drug effects. For instance, methapyrilene, a putative antihistamine and sleep aid withdrawn from the market after it was discovered to be a potent liver carcinogen in rodents, was shown by 2DE to target the hepatic mitochondrion by forming protein adducts specifically in this organelle.20 The nature of the protein modification indicated that it was covalent and resulted from binding of a negatively charged adduct. This study raised the possibility that the putative reactive metabolite, by attacking mitochondrial but not nuclear DNA, could exert its genotoxicity in an unconventional way. Detection of the protein modification by 2DE demonstrates the distinctiveness of this proteomic method in detecting and quantifying drug-related processes that result in the formation of reactive metabolites.
Other examples of studies in which 2DE has been used in drug development include the characterization of Cyclosporin A toxicity in rat kidney21 and brain,22 and the protein-level characterization of histopathologies observed in the liver during preclinical assessment of drug candidates (substituted pyrimidine derivatives).23 The hepatic effects of cholesterol-lowering statin drugs (HMG-CoA reductase inhibitors) such as lovastatin (eg, Mevacor®, a natural product) and fluvastatin (eg, Lescol®, a totally synthetic product), also have been investigated.24,25 The 2DE-based results of these studies demonstrated that inhibition of HMG-CoA reductase was associated with a statin dose-specific regulatory response in the cholesterol synthesis pathway that included the induction of cytosolic HMG-CoA synthase and isopentenyl-diphosphate delta-isomerase (potential alternative drug targets), alterations in key enzymes of carbohydrate metabolism, and toxicity as reflected by changes in a heterogeneous set of proteins involved in functions such as cytoskeletal structure, calcium homeostasis, protease inhibition, cell signaling, or apoptosis. These studies also lead to the conclusion that this approach may be useful in high-throughput assay format to compare the therapeutic windows of various members of a drug family.
More recently, a similar approach was used successfully to investigate ovarian cancer cell responses to manumycin, a farnesyl transferase inhibitor for treating ras-mutated human tumors.26 Using 2DE analysis of total proteins from treated and untreated cell lines followed by MS identification of differentially expressed proteins, heat shock protein 70 (HSP70) was shown to increase following treatment. Combined with data from other studies, this suggests that HSP70 may enhance resistance and account for the altered sensitivity of certain cancers to farnesyl transferase inhibitors. 2DE analysis also was used to investigate the protein molecular basis for resistance to daunorubicin developed by pancreatic cancer cells.27
Another of the utilities of 2D gel-based proteomics lies in biomarker development, for example, generating sets of proteins that can be used as indicators28,29 and even predictors of chemical effects on cells and tissues,30 as in candidate screening in drug or chemical safety evaluation and efficacy studies. Recognition and validation of sets of appropriate protein biomarkers must first be accomplished before truly functional, high-throughput drug toxicity screening systems such as protein chips containing these proteins can be developed. 2D gel-based proteomics can serve as a way to identify such a pool of biomarkers, as a recent application of this approach to characterize the nephrotoxicant gentamicin demonstrates.31 Proteins identified in that study fell into various metabolic or functional categories, for example, gluconeogenesis and glycolysis, fatty acid transport and utilization, citric acid cycle enzymes, and the cellular stress response. Their collective alterations were deemed consistent with known mechanisms implicated in gentamicin-induced toxicity.
Liquid Chromatography Separations and MS
Despite the history and excellent track record of 2DE for protein separation, 2D gels have a number of limitations. Among these are difficulties in focusing very acidic (pI<3.5) and basic (pI>9) proteins, difficulties in recovering hydrophobic proteins (eg, membrane proteins), and a limited dynamic range for accurate protein quantitation (usually, one to two orders of magnitude). Attempts to complement 2D gel-based methods for protein separation and identification with various LC separations coupled to MS identification have been successful in a number of laboratories. Perhaps the most popular of these techniques so far is multidimensional protein identification technology, often referred to as MudPIT.32 Protein mixtures from partially purified protein complexes or from crude cell lysates are digested with trypsin. The mixture of resulting peptides then is loaded onto a biphasic microcapillary column containing a strong cation exchange (SCX) resin upstream of a reversed phase (RP) resin directly coupled to a tandem mass spectrometer. Peptides are displaced from the SCX resin using a step gradient of salt and bind to the RP resin. Elution from the RP resin is accomplished using an acetonitrile gradient, and peptides are analyzed online by MS/MS. Repeated rounds of step and gradient elutions result in large numbers of peptides being analyzed and identified in a single run.
These multidimensional purification procedures are especially effective when working with partially purified protein complexes. Sanders et al33 used a slight modification of the original MudPIT procedure to study the general yeast transcription factor TFIID. This complex of proteins was purified from whole cell extracts using several different polyclonal antibodies to known members of the complex. After digestion, peptides were applied to a SCX column, and fractions were collected. Individual fractions were then analyzed by RP-LC/MS/MS. Using this approach several novel protein components of the TFIID complex were identified, and numerous protein–protein interactions could be inferred from the data. Ostrowski et al34 studied the protein complement of human cilia using a combination of 2DE, 1DE, LC/MS/MS, and LC/LC/MS/MS. In total, more than 200 proteins were identified. Furthermore, all of the techniques used were complementary to each other: some proteins were identified by more than one method, whereas others were identified uniquely by a particular method.
MudPIT also has been applied to the analysis of much more complicated samples. Koller et al35 recently identified a combined total of 2528 unique proteins in the leaf, root, and seed tissue of rice using a combination of MudPIT and 2DE. A total of 1972 proteins were identified only by MudPIT, whereas 165 proteins were identified only by the 2DE approach. This clearly demonstrates the complementarity of these two approaches and emphasizes that both are required for maximizing the total number of proteins detected and identified. Florens et al36 used MudPIT to identify over 2400 proteins associated with various stages of the life cycle of the causative agent of malaria, Plasmodium falciparum. This information should be useful in designing drugs and vaccines. Multidimensional protein separation/MS technologies will undoubtedly continue to play a role in the characterization of other complex proteomes, such as the human plasma proteome that may contain up to 10 million different proteins.37 Finally, it should be noted that, while impressive, these types of multidimensional analyses are not yet routine in most labs. One current significant obstacle is that these types of experiments generate huge data files, many of them requiring significant computing power for analysis and interpretation.38
LC/MS Methods and ICAT
A novel technique termed isotope coded affinity tagging (ICAT) has been used to improve the quantitative capability of the MS methods. Using light or heavy isotope-labeled peptide ‘tagging’ reagents that differ only in molecular mass, proteins derived from normal/diseased or untreated/treated samples can be quantified and identified using LC-MS/MS.39 The tagging reagents contain a cysteine-reactive group at one end and a biotin tag at the other. After mixing the differently labeled proteins together, they are digested with trypsin, and the cysteine-containing peptides are purified over an avidin column. These peptides are further separated on a C18 reversed-phase column directly coupled to an MS/MS instrument. The relative amounts of peptides in the original sample are determined from the ratio of the isotope-labeled ion pairs, and proteins are identified from the fragmentation pattern.
Many variations of the method exist. To reduce both the complexity of the sample and the computing resources necessary for data analysis, intact ICAT-labeled proteins have been fractionated initially on a 2D gel, and then digested, quantitated, and identified by MS.40 Similarly, MS methods using microcapillary electrospray ionization time-of-flight mass spectrometry to determine those peptides that differ in abundance followed by their identification using tandem mass spectrometry recently have been developed by Griffin et al.41 Proteins secreted by human prostate epithelial cells were quantitated and identified, and the expression of several of these was found to correlate with the tumorigenic potential of the cell lines. Arnott et al42 used a modified ICAT reagent to verify the presence and measure the amount of membrane proteins that were predicted to be present from gene expression studies. Finally, second-generation ICAT reagents that overcome some of the technical limitations of the original reagent recently have become available.43 For instance, the addition of an acid-cleavable linker in the ICAT molecule allows removal of the biotin affinity tag before MS and MS/MS analysis. This improves MS/MS performance and significantly increases the number of proteins identified and quantified. Use of 13C as the heavy ICAT isotope (rather than deuterium) improves coelution of both heavy and light isotopes in reversed-phase chromatography and increases accuracy of comparative quantification. Additionally, different chemistries and approaches to differentially tagged protein samples have been developed,44,45 and stable isotope incorporation methods for cells grown in culture have been devised.46
Protein Profiling by SELDI
A slightly different method for proteomic analysis using MS makes use of an instrument marketed by Ciphergen (Fremont, CA, USA). Their ProteinChip® technology is centered around using specifically modified slides having various surface chemistries (eg, ion exchange, hydrophobic interaction, metal chelation, etc) to bind and selectively purify proteins from a complex biological sample. For a given ProteinChip®, various buffer and elution conditions can be tried to further fractionate the sample. The slide then can be analyzed using a surface-enhanced laser desorption-ionization (SELDI) mass spectrometer (essentially a MALDI-TOF instrument). The ease and speed of screening samples have made this a popular method for biomarker detection for clinical and toxicological samples.47,48 It should be noted that the current SELDI MS instrumentation usually is capable of detecting proteins with molecular weights less than 25 000 and that most proteins cannot be identified using this technique alone. Next-generation instruments using ProteinChip® tandem MS techniques are being developed and should greatly improve the ability to identify proteins directly.49
Petricoin et al50 recently published the results of a landmark study in which they identified a proteomic pattern in serum that is diagnostic for ovarian cancer. Using the ProteinChip®/SELDI technology and a custom algorithm to recognize protein patterns in the spectra, they first used sera from 50 unaffected women and 50 patients with ovarian cancer to identify a proteomic pattern that completely discriminated cancer from noncancer. This pattern consisted of a cluster of five distinguishing ions representing yet-to-be identified proteins. Using this cluster as the ‘biomarker’ pattern for a set of 116 masked serum samples, Petricoin et al. correctly identified 50/50 ovarian cancer cases and correctly identified 63/66 noncancerous samples as noncancerous. These data are very encouraging for the establishment of a diagnostic method for a cancer normally extremely difficult to detect early.
Rosty et al51 used the ProteinChip® / SELDI technology to screen for differentially expressed proteins in pancreatic juice from patients with and without pancreatic adenocarcinoma. After finding a protein with a molecular weight of ∼16 500 that was elevated significantly in the affected individuals, they used an immunoassay to identify this protein as hepatocarcinoma-intestine–pancreas/pancreatitis-associated protein-I (HIP/PAP-I). In similar fashion, Li et al52 recently screened serum samples from a total of 169 patients and found that a panel of three biomarker ions derived from SELDI analysis enabled a breast cancer detection sensitivity of 93% for all cancer patients and a detection specificity of 91% for all controls.
To isolate a certain class of proteins within a complex mixture, one can take advantage of unique biochemical properties that are characteristic of that class of proteins. All of these approaches exploit small molecule/protein binding specificity to tag proteins of interest so that they can be readily isolated, identified, and further studied. For example, a series of serine hydrolase inhibitors that covalently react with active proteases, lipases, and esterases have been tagged with biotin or fluorescent labels.53,54 These activity-based probes have been used by Jessani et al55 to profile the active hydrolases in a number of human breast and melanoma cancer cell lines. Distinctive proteomic signatures of many of these cell lines suggested novel targets for the diagnosis and treatment of human cancer. Similarly, activity-based probes that specifically tag active cysteine proteases have been developed.56 Binding partners that interact in a noncovalent manner can be studied by modification of these methods as well. For example, specific noncovalent complexes formed between small molecule ligands and the proteins of interest can be converted into stable covalent complexes by photocrosslinking (eg, with an azido group57).
This technology, whose approach is analogous to creating a genomic microarray, is very promising but still in the early stages of its development.58,59 For protein microarrays, each feature on the array must be a binding partner for a protein that might be in the sample.60 Most commonly, antibodies to the proteins of interest are prepared in high-throughput or semi-high-throughput fashion, and then spotted onto a specially treated surface (eg, glass, silica, etc). The rate-limiting step often is obtaining enough useful antibodies or binding proteins to produce a truly functional array. Furthermore, the preparation of the surface is crucially important to maintain protein structure and binding properties.61 Protein detection and quantitation can be accomplished using several methods, but fluorescence is the most common.
Haab et al62 were some of the first to demonstrate the feasibility of using protein microarrays to simultaneously detect multiple proteins in complex samples. A total of 115 antibody/antigen pairs were characterized for selectivity and sensitivity of detection in the presence of serum. By spotting binding proteins other than antibodies, protein–protein interactions could be measured. Recently, Kukar et al63 immobilized proteins onto derivatized glass or nitrocellulose-coated slides and then detected these proteins in a quantitative manner by probing with fluorescent proteins known to be binding partners. Cahill64 have produced and purified more than 10 000 unique human proteins and arranged them onto PVDF filters and glass slides. These high-density arrays are being used to screen sera from patients with autoimmune diseases in order to identify the proteins to which the body is making autoantibodies.
Protein chips certainly have the potential to be more sensitive, reproducible, and quantitative compared to current methods. It is possible that aptamer or even small-molecule chemical arrays will eventually take the place of antibody arrays to probe the proteome. Aptamers are single-stranded DNA molecules that bind target proteins with high specificity and affinity.65 In this respect, they are analogous to antibodies, but aptamers can be selected from combinatorial libraries and can be produced synthetically and fairly inexpensively. Furthermore, these DNA aptamers are less sensitive to denaturation than antibodies or other proteins spotted onto a chip. By substituting a bromodeoxyuridine for a thymidine, proteins can be crosslinked by ultraviolet light to the photoaptamer at specific sites, further increasing the specificity and selectivity of these molecules. Aptamers have been successfully generated against many proteins, a recent example being aptamers against the hepatitis C virus RNA polymerase.66 Automated selection techniques and analytical methods soon should make it possible to select aptamers against total organismal proteomes.67
FUNCTIONAL PROTEOMICS: TOOLS AND APPLICATIONS
Protein–Protein Interactions: Isolation of Protein Complexes
Most major tasks within the cell, from gene transcription and mRNA splicing to protein degradation, signal transduction, and cell cycle regulation, tend to be carried out by multiprotein complexes rather than individual proteins.68 Thus, examining protein complexes often will yield information about protein function. One way to approach this problem is to isolate large multiprotein complexes, such as ribosomes. Using multidimensional LC and tandem MS, Link et al69 showed that yeast ribosomes contain over 80 components, many of which were not previously known to be associated with ribosomes. Several of these proteins provide a link of the translational apparatus to other processes in the cell such as transcription, RNA processing, metabolic signaling, etc.
Essentially all proteins in the cell interact with other proteins as members of multiprotein complexes. However, most of these complexes cannot easily be isolated for study. Antibodies against one member of a complex sometimes are useful for purification, but often the epitope is not accessible when the protein is in a complex. In general, the use of molecular biology techniques to insert a tag (eg, His-tag, Flag-tag, GST, etc) into the gene coding for the protein of interest has been more successful. This approach has worked well, especially as demonstrated by two recent landmark studies. Gavin et al70 used a novel tandem-affinity purification (TAP) tagging technique to isolate over 200 distinct multiprotein complexes from yeast. The size of the complexes varied from 2 to 83 different proteins, with an average of 5–10 proteins per complex. Most proteins were members of (on average) about five distinct complexes. Over 300 proteins were assigned new roles based on their association with proteins of known function. When done on a large scale, such work begins to unravel the connectedness of proteins and the interconnected biochemical networks in which they participate. Ho et al,71 using high-throughput mass spectrometric protein complex identification (HMS-PCI) very analogous in many ways to the Gavin et al approach, elucidated similar protein complexes and networks. Many of the identified yeast proteins and protein complexes have human homologs, so the implications for extension to humans and drug discovery research are clear. Furthermore, the systems are being adapted to directly work in human cells.70
Protein–Protein Interactions: Yeast 2-Hybrid and Other Methods
The yeast two-hybrid system is a unique genetic-based approach for analyzing protein–protein, protein–nucleic acid, and protein–ligand interactions.72,73 Basically, a DNA binding domain and an activation domain that normally reside together in a single transcription factor are cloned into separate vectors. The bait, or the protein of interest, usually is cloned as a fusion protein into the binding domain vector. The prey consists of all members of a cDNA library that are cloned as fusion proteins into the activation domain vector. When the bait and prey proteins interact, the binding and activation domains of the original transcription factor are brought close enough together to promote transcription, and yeast clones containing the interacting proteins are selected. A detailed description of the yeast two-hybrid system can be found elsewhere.74
Using the yeast two-hybrid system, Uetz et al75 and Ito et al76 determined several thousand interactions between proteins in yeast. These were direct interactions between protein pairs, making this information complementary to the protein interaction data that are inferred from having several proteins present in the same complex (see previous section). Adam et al77 recently utilized the yeast two-hybrid system as part of a comprehensive proteomic analysis of human breast cancer cell membranes. Tumor cell-derived membrane proteins were isolated, separated on 1D gels, digested with trypsin, and identified by MS. Nearly one-third of the 501 proteins identified were previously uncharacterized (eg, ESTs of unknown function). Three of these uncharacterized proteins were analyzed further for interacting protein partners using the yeast two-hybrid system. All three were shown to interact with known (ie, well-characterized) proteins, thus leading to a clearer understanding of their EST-encoded partners and their potential roles in cancer.
Other high-throughput methods for inferring protein interactions exist, as well. Two distinct examples include the correlation of mRNA expression profiles to function, as in the characterization of pharmacological perturbations by identifying a novel target of dyclonine,78 and through in silico predictions of structural or pathway interactions from whole genome comparative analyses.79 von Mering et al80 evaluated the roughly 80 000 interactions among yeast proteins currently available in various databases by comparing data from yeast two-hybrid systems, mass spectrometric protein complex purification techniques, mRNA expression profiles, genetic interaction data, as well as in silico interaction predictions from gene fusion, neighborhood, and co-occurrence analysis. Surprisingly, only ∼2400 of these interactions are supported by more than one method. Each method appears to have its own strengths and weaknesses. For example, the biochemical purification of complexes allows interactions among three or more proteins to be detected, whereas yeast two-hybrid detects interactions only between pairs of proteins. However, yeast two-hybrid is capable of determining weak, transient interactions that might not be stable during biochemical isolation of protein complexes. All things considered, to determine protein interactions comprehensively, as many complementary methods as possible should be used.
Subcellular Proteomes: Organelles
Owing to the complexity of eucaryotic cells, neither 2D gels nor multidimensional LC methods will ever be able to separate the tens (or hundreds) of thousands of proteins and their isoforms that are present in a cell. One way around this obstacle is to take advantage of normal cellular compartmentalization. Animal cells are divided into compartments called organelles that include the mitochondrion, lysozome, phagosome, endoplasmic reticulum, Golgi, nucleus, nucleolus, and many others. Using various biochemical, centrifugation, and electrophoretic methods, these organelles can be purified or greatly enriched, generally giving at least a 10-fold purification. The proteomes of many of these organelles from several species have been analyzed.81
Gagnon et al82 examined the protein composition of phagosomes and phagosomal membranes using 2D gels and MS. After compiling identifications for over 500 proteins, they noted that numerous proteins previously thought to be localized to the endoplasmic reticulum also were found in phagosomes. After ruling out cross-contamination, they proposed a detailed model for phagolysozome biogenesis in which the endoplasmic reticulum is the source of membrane during phagocytosis by macrophages. Andersen et al83 studied the human nucleolus, an organelle with putative roles in RNA transport and modification, as well as cell cycle signaling. Over 270 proteins were identified, many with unknown functions or not associated previously with the nucleolus. Changes in the protein content as a function of the transcriptional state of the cell indicated that the nucleolus is extremely metabolically active.
As outlined in this review, proteomics comprises various technologies that are rapidly developing. The ‘older’, more mature areas, such as 2D gel electrophoresis and many of the MS techniques, already have contributed substantially to the pharmacoproteomics field. Applications of new technologies, such as SELDI, protein microarrays, and protein–protein interaction methods are appearing in the literature and currently are being utilized in both the public and private sector to advance drug discovery and development. Finally, it is clear that no one proteomic technique or approach is universally applicable in pharmacoproteomic studies. In the future, successful proteomic approaches to understanding physiological function, disease, or efficacious and adverse drug effects will require the careful implementation of various combinations of these methods.
We gratefully acknowledge the critical reading of the manuscript by Drs Mu Wang, Junyu Li, and Bob LeBoeuf.
DUALITY OF INTEREST