Whether as drug targets or new drugs themselves, proteins are key to the future of therapeutics. For this reason, the study of the protein complement of an organism, proteomics, appears to be the next wave in biomedical science. According to Blackstock and Weir of Glaxo Wellcome Research and Development (Stevenage, U.K.), “Proteomics can be divided into expression proteomics, the study of global changes in protein expression, and cell-map proteomics, the systematic study of protein–protein interactions” (1). Obviously the proteome, unlike the fixed genome, is a dynamic entity because it is the product of both gene expression and posttranslational alterations; the proteome differs from cell type to cell type and from embryo to adult, and it is greatly affected by the environment. The first step on the pathway to the proteome thus lies in optimizing the methods used to analyze and identify the individual proteins themselves.

This past decade, the evolution of genomics from basic research in the laboratory to a real-world technology depended on the scaling up and speeding up that only automation could achieve. If proteomics is ever to develop into a mature and useful discipline, a similar focus on automation is required. Historically, protein analysis has relied on the time-honored technology of two-dimensional electrophoresis (2DE) to isolate proteins of unique interest. However, the need for speed and sensitivity is pushing this workhorse to the limit, encouraging researchers to not only improve 2DE but also develop and use HPLC alternatives.

“Classical” proteomics
Protein chemistry has been a foundation of modern biotechnology, evolving from the middle of the last century into the mature and powerful analytical science it is today. In the 1950s, the key breakthrough was the development of the “Edman degradation”, in which “phenylisothiocyanate reacts with the N-terminus of the protein to form a cyclic intermediate that facilitates hydrolysis of the adjacent amide bond. The effect of this reaction is to remove the N-terminal amino acid, [generating] a cleaved derivative of that amino acid and a protein that is shortened by one amino acid and has a new N-terminus” (2). As the reaction runs, amino acids are cleaved off in order, one by one, and can be purified and identified using standard techniques. By the 1960s, the power of this method was revealed as it was adopted as the basis for the first automated amino acid sequencers.

But perhaps the most physiologically significant breakthrough in protein analysis has been the development and refinement of 1DE and 2DE, which separate proteins from one another so that they can be studied individually (see “Proteomic Image Analysis” for a description of 2DE methodology).

Finding the unique proteins that vary in differential 2DE displays has led to a host of breakthroughs in understanding animal and plant physiology. The true glory of this technique is the ability to subsequently excise the proteins of interest from their visualized spots (labeled radioactively or with Coomassie blue, fluorescent markers, or silver stains) in order to study them more closely.

Typically, the first step in analyzing such a spot has been digestion of the protein into peptide fragments using trypsin. “An ‘average’ protein such as bovine serum albumin with a molecular weight of 69.3 kDa has 607 amino acids in its sequence; with the proteolytic enzyme trypsin, it produces 87 different peptides” (2). Classical proteomics also typically used automated amino acid sequencers on whole small proteins or peptide fragments (separated chromatographically) to determine amino acid sequences.

Using tandem MS to analyze proteins

Mass spectrometers analyze ions in the gaseous state. Therefore, several methods are used to gasify and ionize compounds from an LC fraction before the MS analysis can be performed. These include electrospray ionization (ESI) and matrix-assisted laser desorption ionization (MALDI) techniques. MS then separates the entering ions according to their m/z (the mass/charge ratio). Tandem mass spectrometers operate by using this separation as a first fractionation step. Before entering the second mass spectrometer, ion fractions from the first are collisionally dissociated by passage through a neutral gas to induce fragmentation. These fragments exist as a family of subset ions of the original parent ion. Computerized analysis of the m/z spectrum of these subset ions can be used to determine the structure of the parent ion. In protein analysis, the parent ions are whole proteins or peptides, and the subsets consist of amino acid chains of varying length that have characteristic m/z fingerprints. Figure 1 shows how this approach can be used to identify variations between the primary amino acid sequences of human hemoglobin (in this case, detecting the presence of a valine substitution where normally there is a glutamate).

2DE—From peptides to proteins
Classical proteomics began to be transformed with the introduction of MS as a routine detector system. For typical analysis, protein spots from 1DE or 2DE are dissected out; then the individual proteins are subjected to the tryptic digests described above, passed through capillary HPLC (or affinity-HPLC for fragments of posttranslationally modified proteins), and injected into a mass spectrometer (see box, “Using tandem MS to analyze proteins”).

Fragments can be sequenced by automated inspection and comparison of the mass spectra obtained to the limited number of available m/z bands capable of being produced by relatively short combinations of the 20 amino acids. Although not all are unique combinations, cross comparisons with other breakdown products can narrow possibilities, especially when trying to identify a protein with a known sequence.

Sequence analysis and reconstruction of these tryptic fragments lead to a picture of the complete protein. In some cases, a specific peptide fragment (or a few fragments in rare combination) proves unique to a particular type or class of protein in a species being studied. This fragment or combination of fragments can be used as “signature peptides” to instantly identify the presence of the protein in a trypsin digest of multiple unknown proteins, such as occurs in whole cell analysis (3).

Alternatively, nondigested whole proteins can be run through a tandem mass spectrometer system, which finds the weights of the molecular ions in the first mass spectrometer and shatters the protein into peptide fragments for analysis in the second MS, as in the case above.

Recently, such technology has even been shown capable of dealing with glycosylation, which many proteins have as part of their posttranslational architecture. However, the more that tandem MS was used, the less acceptable the limitations of 1-D and 2-D gels became.

2DE—From proteins to proteomes
Currently, whether to choose 2DE depends very much on the level of analysis needed. If it is desirable to visualize whole-protein differentials between treated and untreated cells or healthy and diseased tissues, and the proteins are not in extremely low abundance or in membranes, then 2DE is the better choice. The technique still manages to capture a significant part of the available proteome, more so if techniques such as whole gel slicing rather than the use of only the stained bands for analysis are employed.

Equally important, the separation and imaging systems associated with this technology are currently among the most advanced, and the selection of available protocols in the literature is vast.

Quantum proteomics?
If anything has promised to completely transform protein analysis, it is the development and use of multidimensional HPLC as a replacement for classical 2DE. The potential benefits of using HPLC rather than 2DE are immense. The most prominent are

• no long gel runs,
• no gel removal steps or gel interactions,
• fewer size limitations (HPLC can accommodate molecules ranging from small peptides to large proteins),
• no visualization problems with regard to the need for staining,
• greater sensitivity,
• the ability to act as an immediate portal to subsequent HPLC runs,
• multiplicity of kinds of HPLC separation possible,
• the ability to interface directly with a variety of MS systems, and
• ease of automation using standard technologies.

Because of its ability to separate and identify molecules with much lower molecular weights, HPLC is the technology of choice for protein identification through peptide analysis, often after a 2DE run isolates the particular protein for tryptic digestion. Where HPLC systems have traditionally lagged, however, is in this initial separation of proteins from complex mixtures, especially because they lack the visualization step that a gel display provides.

HPLC systems, however, are catching up to 2DE even in the area of complex protein mixtures. Multidimensional HPLC for protein analysis primarily relies on tandem or multiple LC combined with tandem MS. Given the sensitivity of the initial LC steps and the fact that everything is done in liquid in-line, such systems are not only easily automated but also capable of separating even the low-abundance and membrane proteins. These proteins cause difficulties in 2DE analysis because of poor yields, difficult visualization, or interference with the gel runs. And, with new informatics tools, analysis of the chromatography data can be presented in more user-friendly displays of multidimensional runs.

Whole cell destruction?
Even though whole cell analysis still relies primarily on the use of 2DE as its primary separation technique, research into its replacement with multidimensional HPLC and tandem MS is proceeding. At the 49th American Society for Mass Spectrometry (ASMS) Conference on Mass Spectrometry this year, Stephen A. Cohen and co-workers at the Waters Corp. gave presentations on the benefits of multidimensional HPLC for the separation and identification of peptides and proteins (www.waters.com/waters_website/Events/ASMS_2001.htm). And Dayin Lin and John R. Yates, III, of the Scripps Research Institute (La Jolla, CA) reported in their abstract “Analysis of the Saccharomyces cerevisiae Proteome” for the Proteomics—Fundamental Studies session of the meeting how 2DE visualized approximately 1000 proteins, compared with the 2000 unique yeast proteins they determined using 2-D LC–tandem MS.

Chong and colleagues recently reviewed some of the work in analyzing the protein compositions of whole cell lysates of animal and bacterial cells using LC methods. They described their own research using nonporous reversed-phase HPLC coupled to ESI-TOF MS to provide a “virtual 1-D gel”, which easily identifies proteins that are expressed differentially between premalignant and cancer cell lines (5). All of these researchers cite a variety of what they see as significant advantages of HPLC systems over 2DE, including the ability to analyze low copy number, membrane, highly basic or acidic, or posttranslationally modified proteins.

Not down for the count?
Despite the benefits of HPLC systems, 1DE and 2DE for protein analysis are not likely to disappear anytime soon. Both methods have their strengths as historically robust technologies and are inexpensive alternatives to complex instrumentation. (At the 2001 ASMS conference, both 1DE and 2DE were still extremely well represented—as was proteomics as a whole.) Indeed, 2DE remains the best way to visualize protein differentials between biological states, and much research is under way to improve imaging analysis techniques even more. Numerous companies are also working to make significant improvements to classical 2DE to eliminate some of its weaknesses. For example, the Proteomeworks System developed through an alliance between Micromass, Ltd. (U.K.) and Bio-Rad Laboratories, Inc. (U.S.) has been designed to automate various steps in the processes of making, running, and analyzing the results of 2DE. Automated image analysis packages for 2DE are discussed elsewhere in this issue (see “Proteomic Image Analysis”).

For the futurists among us, however, the next wave appears to be increasing reliance on various multidimensional HPLC systems as an alternative to standard 1DE and 2DE gels.

Physiomics futures
Ultimately, the automation of analytical technologies for simultaneous multiple protein analysis, whether through improved 2DE or variations on HPLC-MS, is the only hope for making proteomics a living science. And such a science is the only hope for biological researchers to be able to move into the next and perhaps ultimate “-omics”: physiomics. The predicted physiomics will be the systemwide informatics science capable of modeling and understanding the biology of the complex temporal and environmental states involved in the life of an individual organism. But if such models of living cells and organisms are ever to become authentic tools for drug discovery, the complexities of the proteome first must be revealed.

References

1. Blackstock, W. P.; Weir, M. P. Trends in Biotechnology, March 1999, 17, 121–127.
2. Kinter, M.; Sherman N. E. Protein Sequencing and Identification Using Tandem Mass Spectrometry; Wiley-Interscience: New York, 2000.
3. Geng, M.; Li, J.; Regnier, F. J. Chromatogr., A 2000, 8, 295–313.
4. Gatlin, C. L.; et al. Anal. Chem. 2000, 72 (4), 757–763.
5. Chong, B. E.; et al. Anal. Chem. 2001, 73 (6), 1219–1227.

Further reading

• Snyder, A. P. Interpreting Protein Mass Spectra: A Comprehensive Resource; Oxford University Press: Washington, DC, 2000.
• Cunico, R. L.; Gooding, K. M.; Wehr, T. Basic HPLC and CE of Biomolecules; Bay Bioanalytical Laboratory: Richmond, CA, 1998.

Mark S. Lesney is a senior editor of Modern Drug Discovery. Send your comments or questions regarding this article to mdd@acs.org or the Editorial Office by fax at 202-776-8166 or by post at 1155 16th Street, NW; Washington, DC 20036.