After 10 years of extensive proteomic research, it has become increasingly apparent that new technologies are sorely needed for detecting the low-abundance proteome—those proteins (up to 50% in any proteome) whose concentration in tissues or cells falls below the detection limits of currently available methodologies. Here we survey one such method: a combinatorial ligand library (called ProteoMiner), comprising dozens of millions of hexapeptides capable of interacting with most, if not all, proteins in any given proteome. They act by drastically reducing the signal of high-abundance species while increasing the level of the low-abundance components to bring their signal within the detection limit of present-day tools. Such a library has been tested against a number of human biological fluids, such as sera, urine, cerebrospinal fluid as well as against cell lysates (e.g., platelets, red blood cells) with interesting results.

Peptide Libraries and Proteomics Applications

Proteomics helps to distinguish useful information from the masses of DNA-based microarray data, bringing us closer to understanding the functions of the corresponding protein products. There are at least three main research avenues in modern proteomic studies: (i) bold cataloging initiatives that are still largely focused on protein identifications (such initiatives are coordinated by the Human Proteome Organization and include cataloging all human plasma, liver, and brain proteins); (ii) diagnostic projects aimed at finding biomarkers in body fluids and/or tissue biopsies for monitoring diseases, especially at their early onset (e.g., in the case of cancer, identification of a panel of specific markers or a pattern of changes that would reflect the insurgence of the disease, its progress, and its regression upon specific treatment); and (iii) use of proteomics as the basis for development of new drugs. Almost all drug targets are proteins, thus their profiling upon drug treatment is one of the main aims of the pharmaceutical industry.

All the projects outlined above face a formidable challenge: about 50% of all proteins in any proteome are low-abundance species that fall below the detection limit of current technologies. In principle, two-dimensional (2-D) maps with very high (10-mg) protein loads on large-format 2-D gels may detect 1000 copies per cell of a protein via silver staining. Such amounts probably would not be analyzable by mass spectrometry (MS). In practice, most researchers load <1 mg of protein on a single 2-D gel, so the chances of observing low-abundance proteins with a one-extract, one-gel approach are practically nil. In order to overcome that, a number of pre-fractionation tools have been described, as reviewed in Herbert et al. (1). Such tools comprise not only all possible chromatographic and electrophoretic methods (especially those based on isoelectric focusing) but also organelle pre-fractionation and immunosubtraction of most abundant species via antibody columns (up to 12 different antibodies mixed in a single column). Other tools much in vogue today regard the capture of classes of proteomes based on their structure or function, such as the specific seizure of the glycoproteome via lectin columns (2), the capture of the phosphoproteome via metal chelator resins or affinity for titanium oxide (3), or via antiphosphoamino acid antibodies (4). Other techniques focus on charting protein-protein interactions via two main methods: (i) the yeast two-hybrid system, which allows mapping of binary or pairwise associations (5), or (ii) affinity capturing methods, coupled to MS identification (6).

All of the above techniques allow a deeper insight on proteomes, albeit on a narrower scale, by selecting and amplifying some classes or families of proteins. For example, we can discern functions of proteins localized in subcellular organelles or performing given functions within a cell, such as coordinating action in molecular assemblies or acting in pathways for achieving a particular task. Observing the protein expression of genomes in a holistic manner would require methodologies able to achieve substantial audits of protein expression in any cell, tissue, or biological fluid. Two techniques fulfilling these requirements seem to be emerging in recent times. One is the so-called Human Protein Atlas (HPA), recently designed and constructed by Uhlén et al. (7), and the other is the combinatorial ligand library, known as ProteoMiner (Bio-Rad Laboratories, Hercules, CA, USA), which is the object of the present survey (8,9,10). In the case of HPA, the expression of normal and pathological tissues is studied with antibody-based proteomics. HPA aims to systematically generate high-quality antibodies to all nonredundant human proteins and use these to localize all proteins in human tissues. The nonredundant set of human proteins is defined as one product from each gene locus and is obtained by using epitopes that are common to various protein isoforms. These antibodies are produced by high-throughput methods that involve the cloning and expression of protein-epitope signature tags. In principle, if one were able to produce antibodies against all human proteins, one should be able to map all of them in any tissue or cell population. In practice, release 2.0 of HPA (30 October 2006) contains over 1500 antibodies representing over 1300 different proteins. The goal of obtaining a library comprising all possible antibodies for all possible human proteins is still quite far from being achieved, considering that the entire human genome is comprised of ∼23,000 different genes. In the case of ProteoMiner, the performance and properties of this ligand library will be reported below.