Like the transcriptome, the proteome varies widely between different cell types in an organism. Human cells typically contain several thousand proteins differing in abundance over many orders of magnitude. Like nucleic acids, proteins can be detected and identified by specific molecular interactions, in most cases using antibodies or other ligands as probes. However, unlike nucleic acids, there is no procedure for cloning or amplifying rare proteins. Furthermore, the physical and chemical properties of proteins are so diverse that no single, universal methodology analogous to hybridization can be used to study the entire proteome in a single experiment.
As detailed below, proteome profiling essentially involves four steps. First, proteins within the starting protein extract (from cell lysates, tissues, and so on) are fractionated, typically using some form of gel electrophoresis or liquid chromatography. Then, separated proteins are digested with a protease, typically trypsin, to produce a series of peptides. The resulting peptide mixes are analyzed using mass spectrometry, which determines the precise molecular masses (see Box 7.8). Finally, the molecular masses are referenced against known molecular weights of amino acids, and the predicted com positions of all the peptides in a starting sample are then compared against translations of the starting genome sequence to identify coding sequences that could give rise to the predicted peptides.
Proteomes can be analyzed for different purposes, and combining proteome data with data from the corresponding genome can help to elucidate fundamental aspects of our biology, and to permit greater understanding of disease processes that may also be helpful in treating disease (Figure 1). Of the different approaches used to study proteomes, we are concerned here with expression proteomics. That may involve simply collecting quantitative data for different proteins and protein isoforms (which can help in protein and gene annotation). But a powerful additional application is to compare protein expression between related cell samples (differential proteomics). That may be done simply to understand the biology of cells, and will be important in defining sub sets of cell types. It can also be done as a way of comparing different cellular states, and here an important application is in dissecting the molecular basis of pathogenesis. We can compare, for example, cells from specific tumor types with the pre-cancerous cell states.
Fig1. Potential applications of proteomics in basic biology and clinical research. In differential proteomics, the proteomes of two different cell types or subtypes are compared, such as two subpopulations of a recognized cell type, or the same cell type under normal and disease states, and so on. (Adapted from Lippolis R & De Angelis [2016] J Proteomics Bioinform 9:63–74. With permission from OMICS International. Published under CC BY license.)
Protein separation
The most widely used methods are gel electrophoresis and liquid chromatography. Two-dimensional (2D) gel electrophoresis uses denaturing polyacrylamide gel electrophoresis (PAGE) for separating proteins. In 2D-PAGE, separation of proteins occurs in the first dimension according to the electrical charge of the protein (isoelectric focusing); thereafter separation occurs according to protein mass in a second dimension, at right angles to the first. 2D-PAGE has the power to resolve up to 10,000 proteins on a single gel, and has been widely used in protein separation. It has its limitations, however. Several classes of protein—strongly basic proteins, membrane proteins, and so on—are underrepresented on standard gels, and the sensitivity is dependent on the detection limit for very scarce proteins (but SYPRO dyes permit detection of protein spots in the nanogram range). Another major limitation of 2D-PAGE is that it is not highly suited for automation, making it difficult to carry out high-throughput analyses of many samples.
The alternative is to use liquid chromatography. High-pressure liquid chromatography can separate a starting protein extract into hundreds of fractions that can then be conveniently processed by easily automated mass spectrometry.
Protein annotation The masses of peptide fragments can be used to identify the proteins of origin by correlating the experimentally determined masses with those predicted from database sequences, including EST databases and translated nucleotide sequences. Different ways of annotating a protein by mass spectrometry are listed below, and illustrated in Figure 2.
• Peptide mass fingerprinting (PMF). A simple protein mixture (such as a single spot from a 2D gel) is digested with trypsin. The resulting tryptic peptides are subjected to MALDI-TOF MS (see Box 7.8), which returns a set of mass spectra. The spectra are used as a search query against protein sequence databases. The search algorithm carries out virtual trypsin digests of all the proteins in the database and calculates the masses of the predicted tryptic peptides. It then attempts to match these predicted masses against the experimentally determined ones. This method is best suited for simple genomes.
• Fragment ion searching. This method is more suited to the analysis of complex proteomes, and the algorithm can be modified to take into account the masses of known post-translational modifications. The tryptic peptide fragments are analyzed by tandem mass spectrometry (MS/MS; see Box 7.8) during which the pep tides are broken into random fragments. The mass spectra from these fragments can be used to search against EST databases. Any EST hits can then be used in a BLAST search to identify putative full-length homologs. A dedicated algorithm called MS-BLAST is useful for handling the short sequence signatures obtained from peptide fragment ions.
• De-novo sequencing of peptide ladders. This method is also carried out because it is impossible to account for all variants, either at the sequence level (poly morphisms) or at the protein modification level (for example, complex glycans). Sequencing of peptide ladders may provide sequence signatures that can be used as search queries to identify homologous sequences in the databases. In this technique, the peptide fragments generated by MS/MS are arranged into a nested set differing in length by a single amino acid. By comparing the masses of these fragments to standard tables of amino acids, it is possible to deduce the sequence of the peptide fragment de novo, even where a precise sequence match is not available in the database. (In practice, this approach is complicated by the presence of two fragment series, one nested at the N-terminus and one nested at the C-terminus; the two series can, however, be distinguished by attaching diagnostic mass tags to either end of the protein.)
Fig2. Protein annotation by mass spectrometry. Individual protein samples (such as spots from 2D gels) are digested with trypsin, which cleaves on the C-terminal side of lysine (K) or arginine (R) residues (as long as the next residue is not proline). The tryptic peptides can be analyzed as intact molecules by MALDI-TOF (see Box 7.8), and the masses used as search queries against protein databases and translated nucleotide databases. Algorithms are used that take protein sequences, cut them with the same cleavage specificity as trypsin, and compare the theoretical masses of these peptides to the experimental masses obtained by MS. Ideally, the masses of several peptides should identify the same parent protein (human lysozyme, in this example). If no hits are recorded (for example, because the protein has been subject to post-translational modification or artifactual modification during the experiment), ESI-tandem mass spectrometry (ESI-MS/ MS) can be used to fragment the ions. The fragment ion masses can be used to search sequence databases and obtain partial matches, which may lead eventually to the correct annotation. Alternatively, the masses of peptide ladders can be used to determine protein sequences de novo. ESI, electrospray ionization; EST, expressed sequence tag.
