Sci. Aging Knowl. Environ., 10 November 2004
Vol. 2004, Issue 45, p. pe41
[DOI: 10.1126/sageke.2004.45.pe41]


Proteomics in Aging-Related Research

Ari Gafni

The author is at the Institute of Gerontology, the Department of Biological Chemistry and Biophysics Research Division, University of Michigan, Ann Arbor MI 48109, USA. E-mail: arigafni{at}

Key Words: proteomics • proteome • high-throughput • bioinformatics • protein complement

Aging as a Multigene Phenomenon

Even a quick review of recent published work in the field of biogerontology, for example using one's favorite Web search engine, reveals the great progress that has been made in identifying specific genes (and gene products) that affect longevity in a variety of organisms (see also the SAGE KE Genes/Interventions database and "(Data)base Desires"). A recent article in Chemical & Engineering News (1) nicely summarizes the state of the art in this field and presents a number of models, and their associated candidate proteins, that might feature in extending the life spans of yeast, nematodes, fruit flies, and mice. Although the evidence connecting a specific gene (or a small number of genes) to the extended longevity in each case is compelling, it also appears clear that there still is no unifying explanation for these life-extending effects, let alone for the fundamental process of aging. Indeed, work in different laboratories has demonstrated that significant life-span extension can be affected by altering the expression (or activity) of any of a long list of individual proteins. A partial list includes proteins involved in (i) telomere repair (2), (ii) stress response (3), (iii) anti-oxidant defense (4), (iv) nicotinamide deamination (5), (v) insulin/insulin-like growth factor-1 signaling (6), and (vi) histone deacetylation (7), as well as (vii) regulation of the transcription of specific proteins, such as those involved in pituitary development (8). These observations are highly intriguing and potentially of great importance, because they indicate that it should be possible to extend the life span by using drugs that regulate (either enhance or inhibit) the activities of individual longevity-associated proteins. Indeed, several startup companies have recently sprouted looking to do exactly this (see "'Gero-Tech' Sprouts, But Will It Bloom?").

Although this recent advancement is exciting, the fact that so many different single-gene interventions lead to a similar outcome clearly reveals that no one protein has a monopoly on aging and that this process involves interactions among the products of many genes. It therefore is likely that the successful protocols for prolonging life span by the manipulation of a single gene work by altering complex networks of interactions in which the protein of choice is involved. Stated from a different perspective, whereas the large difference in longevity between mice and humans clearly originates in differences between their genomes, it can be argued that if longevity were determined by the products of one or two genes, we probably would have by now seen the rare 90-year-old mutant mouse appear among the millions of mice bred for research. Even though we cannot yet quote a precise number of proteins that are involved in aging, it appears that "many" is a good starting guess. Conceptually, we could hope to find all the gene products that play a role in aging and construct a (more) comprehensive model of cellular aging, if we could identify and quantify all the gene products in the cell and follow their concentrations, activities, localization, and interactions continuously across the life span. Although this task is yet to be accomplished, recent methodological advancements hold promise that this may become a feasible goal in the not-too-far future.

Beyond Genomics

We are already entering the post-genome era, a time when the sequences of all the genes that define a given organism are known. Although our ability to derive this information represents a major breakthrough in the biological sciences, it is becoming widely recognized that this achievement is only the beginning; mapping the genome still does not allow us to explain how the corresponding organism develops, ages, fights disease, and interacts with its environment. This is so because, whereas the genome encodes the blueprints for all the proteins that a given organism can produce, this information is largely static and does not change with time nor does it respond to external stimuli (although exceptions, such as DNA methylation, do exist and are currently the focus of numerous studies). The complement of all proteins in the cell (termed the proteome), on the other hand, does represent the cellular identity (it differs greatly between liver and muscle cells in the same organism, for example) and vigorously responds to changes in external conditions, thus being different, in the same cell, at any given moment. For example, caloric restriction in rodents and a number of other organisms, or mild heat shock in young nematodes, has little effect on the respective genomes but significantly extend life span by altering the expression of specific genes; that is, the proteome. It is still not possible, by inspecting the blueprint represented by the genome, to ascertain which genes will be expressed in a given cell at a given time and to what extent (let alone to understand how the gene products will interact and become postsynthetically modified, for example). It is thus reasonable to assume that in order to understand a living cell more fully, we need detailed knowledge about dynamic processes: information regarding the status of gene products rather than gene composition.

This need has already led to the development of functional genomics, a research field that aims to move beyond the "one gene at a time" approach by applying high-throughput technologies to determine the expression of multiple genes. Functional genomics uses technologies that measure gene transcript (mRNA) abundance in cells, such as DNA microarrays, to derive information about the global patterns of gene expression. It thus provides dynamic information about how the expression levels change with time in response to cellular needs or environmental challenges. Rapid development of the relevant technologies in recent years now enables researchers to follow expression patterns of complete genomes, a feat that only a few years ago looked far-fetched.

With all the excitement about functional genomics, however, it is also widely recognized that the correlation between mRNA concentrations and the identities and biological activities of the corresponding proteins is often not very strong. Events critical to the function of gene products are frequently determined posttranscriptionally through alternate splicing, posttranslational protein modifications, protein-protein interactions, or protein trafficking; and these posttranscriptional events can hardly be gleaned from gene expression patterns alone. Because of posttranscriptional processing, a typical cell often harbors many more functional gene products than genes. Moreover, the cellular location of a protein can dramatically affect its function. For example, glyceraldehyde-3-phosphate dehydrogenase, traditionally known as a cytoplasmic glycolytic enzyme, has been shown to alter its cellular localization depending on external conditions and to play key roles in unrelated processes such as membrane fusion, DNA maintenance, and even apoptosis, activities that might be linked to its localization (9). Finally, protein-protein interactions and networking, which do not show up in the genomic information, are extremely important, and their alteration can feature in development, aging, and disease.

From the discussion above, it is clear that in order to understand complex, multigene-controlled, biological phenomena such as aging more completely, we need to be able to determine, in real time, the concentrations of all gene products in the cell, the nature of any modifications in them, their cellular localization and interactions, and how the latter parameters affect their functional properties (including biological activity, interactions with other molecules, and stability). Moreover, we need to be able to determine how these variables change in response to external stimuli or to other changes in the environment (for example, as gerontologists we need to know how aging affects the nature of these changes).

Proteomics, a Not-So-New Idea Whose Time Has Finally Come

This realization, coupled with recent technological advancements, has triggered intense activity in the field of proteomics, the scientific discipline that focuses on the determination and analysis of the entire protein complement of a cell and its distribution within the cell, and on the identification of both posttranscriptional modifications and specific protein complexes (see Gibson Perspective). It should be realized that the concepts behind proteomics, as well as the basic experimental approaches involved, are not new and have been around for at least 20 years. However, the recent surge of activity in this area is a result of both a great improvement of experimental capabilities and, more important, the appreciation of the potential of this scientific approach to provide fundamentally new knowledge. Already proteomics research is affecting many areas of biological investigation, because researchers believe that once we possess the ability to identify all the changes in protein concentrations, processing, and interactions that accompany a given life process, we will be well on our way to describing its molecular basis. This expectation is already attracting the attention of researchers as well as funding agencies, including the National Institute on Aging.

So, Are We There Yet?

As mentioned above, the idea behind proteomics is by no means a new one. Quite effective separation and simultaneous display of hundreds of cellular proteins by two-dimensional gel electrophoresis was demonstrated more than 20 years ago, and changes in the concentrations of specific proteins in response to external stimuli (such as heat shock), to disease, or during aging have been amply documented. These early studies, however, could not address the complexity of the cellular response. Modern proteomics holds the promise of accomplishing this goal in the foreseeable future by enabling us to identify all cellular proteins and to assess their status at any time and infer their functional roles. Inference of function using proteomics is approached via the global analysis of proteins in the context of specific physiological states. The experimental approaches include (i) protein expression profiling between normal and perturbed states, (ii) spatial analysis to identify cellular localization, and (iii) the generation of interaction maps using either high-throughput two-hybrid methods or the pull-down of protein complexes (a technique in which complexes are purified via an affinity tag on one member of the complex). As the field of proteomics evolves, there is a need for a shift from the simple functional analyses of global protein concentrations to the identification of the roles of proteins in cellular pathways.

To achieve the above-stated goal, however, we need to be able to compare proteomes obtained at different time points in a reliable, accurate, and reproducible way and to develop detailed protein interaction maps. These goals have not yet been achieved for a number of reasons: (i) We still are not able to resolve all proteins on two-dimentional gels or by a combination of chromatography and mass spectrometry, hence many of the protein spots that are observed contain mixtures of proteins. (ii) Some cellular proteins are very abundant, whereas others are very scarce, and the range of concentrations that can be simultaneously detected (the contrast ratio) by current technology is not broad enough to allow all cellular proteins to be recognized. (iii) Some proteins (membrane-bound and highly acidic or basic proteins, for example) are difficult to analyze by gel electrophoresis or chromatography and are not detected. (iv) Some proteins (or their modifications) are very unstable and might fall apart, become oxidized, or undergo other changes while being extracted and separated. (v) Conformational alterations, which may have profound effects on biological activity and protein interactions, often are not resolved by electrophoresis or chromatographic techniques. All of these technical difficulties will have to be resolved before we can determine the whole cellular proteome and follow its changes across the life span.

Protein interaction maps represent one necessary step in the process of building detailed predictive models (see Pletcher Perspective). Ideally, these maps should include information on (i) direct protein-protein interactions (both strong and weak), (ii) cellular localization, (iii) protein-DNA interactions, and (iv) interactions between proteins and small molecules (shared substrates, cofactors, and other molecules). Protein-protein interaction maps based on yeast two-hybrid technology (10, 11) or, more recently, on large-scale pull-down experiments (12) are becoming available; however, the maps produced to date are incomplete and, disconcertingly, different studies produce maps with surprisingly little overlap in the interactions revealed. It appears that none of the current proteome technologies can provide an accurate interaction map, because the data sets are still incomplete, and existing methods generate both false positives and false negatives that are difficult to sort out.

Although the data sets generated by multiple independent methods can provide corroboration for each other, it is expected that a more robust validation of interaction maps will be provided by the new science of bioinformatics, pending the development of new experimental approaches including (i) algorithms for generating interaction maps, (ii) the assignment of statistical significance values to the determined interactions, (iii) the generation of pathway models from interaction maps and other data, (iv) pathway visualization tools, and (v) integration with other genome and functional genomics data. A number of bioinformatics approaches have already been explored to this end, including the identification of functional linkages between proteins in microorganisms through analysis of gene relationships across multiple genomes (13), as well as interpretation of existing data sets via correlation with expression-profiling data and examination of yeast two-hybrid data for the identification of yeast paralogs (genes related by duplication within a genome, which evolve to possess new functions sometimes related to the original function) (14).

In light of the great potential of proteomics (coupled with bioinformatics) to help understand complex cellular processes, it is not surprising that researchers in the field of aging have begun to implement these approaches in their research. A quick Google search using the phrase "proteomics and aging" reveals a growing number of collaborations between academic researchers in specific areas of biogerontology and companies that specialize in proteomic techniques and analysis. The National Institute on Aging, being cognizant of the great potential offered to understanding the biology of aging by the new capability, has recently issued a Request for Application (RFA) soliciting proposals to apply proteomic approaches to study age-related changes in protein structure, function, and interactions. The stated purpose of this RFA is to encourage the development of projects that advance research to identify and quantify protein expression patterns, posttranslational modification of proteins, and protein-protein interactions that may change in cells or tissues as a direct result of the aging process or age-related pathology. The response to this solicitation was overwhelming, with over 100 proposals submitted, a reflection of the great interest in the scientific community in the potential of incorporating proteomics into aging-related research. The Ellison Medical Foundation, which supports innovative research in the field of aging, has also recently awarded funding to support proteomic studies of oxidatively damaged proteins in old tissues.


A key goal of the biological sciences is to generate predictive organismal models. In the specific case of aging, we would like to construct a model that can predict all the changes that occur in a given organism along its life span, leading to senescence. In our nascent post-genome era, we already have the potential to address higher levels of molecular organization than has previously been possible. Clearly, a deeper understanding of the physiology of an organism can be attained through analysis of its native pathways than can be had by studying the reactions of individual proteins or by the mere identification of the genes present in this organism. A primary role of proteomics is to generate the data and the tools (both analytical and computational) needed to build these models. At present, we clearly are at the very beginning of the proteomics era. We can visualize the immense potential of this tool to provide new knowledge about biological systems at a level of detail that was impossible to envision just a few years ago; however, it is also clear that major technological and computational developments need to take place before the full potential of proteomics can be tapped. So, although it may not yet be the right time for the average biogerontologist to practice proteomics, it definitely looks like a good time to begin to pay attention.

November 10, 2004
  1. S. L. Rovner, The secrets of aging. Chem. Engin. News 82, 30-35 (2004).
  2. M. T. Hemann, M. Strong, L.-Y. Hao, C. W. Greider, The shortest telomere, not average telomere length, is critical for cell viability and chromosome stability. Cell 107, 67-77 (2001).[CrossRef][Medline]
  3. J. Song, M. Takeda, R. I. Morimoto, Hsp70-Bag1 complex mediates a physiological stress signaling pathway that regulates Raf-1/ERK and cell growth. Nat. Cell Biol. 3, 276-282 (2001).[CrossRef][Medline]
  4. S. Melov, Mitochondrial oxidative stress; physiological consequences and potential for a role in aging. Ann. N.Y. Acad. Sci. 908, 219-225 (2000).[Medline]
  5. R. Anderson, K. Bitterman, J. Wood, O. Medvedik, D. A. Sinclair, Nicotinamide and Pnc1 govern lifespan extension by calorie restriction in S. cerevisiae. Nature 423, 181-185 (2003).[CrossRef][Medline]
  6. N. Arantes-Oliveira, J. R. Berman, C. Kenyon, Healthy animals with extreme longevity. Science 302, 611-612 (2003).[Free Full Text]
  7. H. A. Tissenbaum, L. Guarente, Increased dosage of a sir-2 gene extends lifespan in Caenorhabditis elegans. Nature 410, 227-230 (2001).[CrossRef][Medline]
  8. K. Flurkey, J. Papaconstantinou, R. A. Miller, D. E. Harrison, Lifespan extension and delayed immune and collagen aging in mutant mice with defects in growth hormone production. Proc. Nat. Acad. Sci. U.S.A. 98, 6736-6741 (2001).[Abstract/Free Full Text]
  9. M. A. Sirover, New insights into an old protein: The functional diversity of mammalian glyceraldehyde-3-phosphate dehydrogenase. Biochim. Biophys. Acta 1432, 159-184 (1999).[CrossRef][Medline]
  10. M. Fromont-Racine, A. E. Mayes, A. Brunet-Simon, J. C. Rain, A. Colley, I. Dix, L. Decourty, N. Joly, F. Richard, J. D. Beggs, et al., Genome-wide protein interaction screens reveal functional networks involving Sm-like proteins. Yeast 17, 95-110 (2000).[CrossRef][Medline]
  11. A. J. Walhout, S. J. Bolton, M. Vidal, Yeast two-hybrid systems and protein interaction mapping projects for yeast and worm. Yeast 17, 88-94 (2000).[CrossRef][Medline]
  12. A. C. Gavin, M. Bosche, R. Krause, P. Grandi, M. Marzioch, A. Bauer, J. Schultz, J. M. Rick, A. M. Michon, C. M. Cruciat, et al., Functional organization of the yeast proteome by systematic analysis of protein complexes. Nature 415, 141-147 (2002).[CrossRef][Medline]
  13. D. Eisenberg, E. M. Marcotte, I. Xenarios, T. Yeates. Protein function in the post-genomic era. Nature 405, 823-826 (2000).[CrossRef][Medline]
  14. C. M. Deane, L. Salwinski, I. Xenarios, D. Eisenberg, Protein Interactions: Two methods for assessment of the reliability of high throughput observations. Mol. Cell Proteomics 1, 349-356 (2002).[Abstract/Free Full Text]
Citation: A. Gafni, Proteomics in Aging-Related Research. Sci. Aging Knowl. Environ. 2004 (45), pe41 (2004).

Science of Aging Knowledge Environment. ISSN 1539-6150