SEARCH

SEARCH BY CITATION

Keywords:

  • fitness;
  • genomic arrays;
  • mRNA abundance;
  • stress response;
  • transcriptomics;
  • transcriptional profiling

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein activity (and not mRNA abundance) is typically consequential for fitness
  5. How often are the genes that transcriptional profiling ‘discovers’ consequential for fitness?
  6. Can transcriptional profiling detect stress where other methods fail?
  7. Conclusions
  8. Acknowledgments
  9. References

Global analysis of mRNA abundance via genomic arrays (i.e. transcriptomics or transcriptional profiling) is one approach to finding the genes that matter to organisms undergoing environmental stress. In evolutionary analyses of stress, mRNA abundance is often invoked as a proxy for the protein activity that may underlie variation in fitness. To provoke discussion of the utility and sensible application of this valuable approach, this manuscript examines the adequacy of mRNA abundance as a proxy for protein activity, fitness and stress. Published work to date suggests that mRNA abundance typically provides little information on protein activity and fitness and cannot substitute for detailed functional and ecological analyses of candidate genes. While the transcriptional profile can be an exquisitely sensitive indicator of stress, simpler indicators will often suffice. In view of this outcome, transcriptomics should undergo careful cost-benefit analysis before investigators deploy it in studies of stress responses and their evolution.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein activity (and not mRNA abundance) is typically consequential for fitness
  5. How often are the genes that transcriptional profiling ‘discovers’ consequential for fitness?
  6. Can transcriptional profiling detect stress where other methods fail?
  7. Conclusions
  8. Acknowledgments
  9. References

Stress has been a bellwether for ecology and evolutionary biology (Bijlsma & Loeschcke, 1997). In terms of natural processes, life under stress exhibits stark impacts and responses that biologists may readily overlook in more benign regimes. In terms of biological discovery, study of stress has yielded a disproportionate number of conceptual and technical advances as well as an impressive phenomenology (Hoffmann & Parsons, 1997). Nowhere has this impact been so evident as for the ‘molecular’ mechanisms of stress resistance. Before such mechanistic knowledge (Hochachka & Somero, 2002), studies of stress could describe organismal failures and their ecological/evolutionary consequences under extreme environmental conditions or at geographic range limits, for example, but seldom how such failures do or do not occur for some individuals, species, or higher taxa. Before such mechanistic knowledge, stress susceptibility and tolerance phenotypes could be assigned to specific genes and quantitative trait loci (QTLs), but the links between gene/QTL and phenotype were often missing. In turn, mechanistic molecular studies have been rechanneled to advance fundamental concepts. [An especially dramatic example is the heat-shock response, which led first to the recognition of molecular chaperone proteins (Feder & Hofmann, 1999) and then to a potential unique role of these proteins in the evolution of evolvability (Rutherford, 2003).] But such molecular approaches have their limitations too; the variation they examine and create sometimes has no counterpart in nature and, if so, their outcomes may not be clearly relevant to our understanding of natural phenomena. For that reason they are best viewed as complementary to studies of natural processes and variation, and vice versa (Feder & Krebs, 1997).

Some would argue that the next stage in this advancement is the application of ‘omic’ technologies, in which the entirety of an organism's genes (i.e. the genome) or traits (i.e. the phenome) undergoes simultaneous characterization. Indeed, a list of named omic technologies and approaches (proteomics, lipidomics, kinomics, metabolomics and many others) would already occupy several pages. Importantly, unlike the foregoing studies of molecular mechanisms, omic techniques make no assumption about which genes and other molecules are worthy of study; omic techniques simply characterize them all. The first and still most notorious of the omic technologies is transcriptomics or transcriptional profiling, the simultaneous assessment of mRNA or transcript abundance for much or all of the genome (Gracey & Cossins, 2003). Little could be simpler in principle: sample organisms in contrasting regimes, react their mRNA with a genomic array, and reveal ‘the genes that matter’ (Feder & Mitchell-Olds, 2003). Importantly, this procedure potentially discovers genes not previously known to play a role in responding to conditions under study. For such reasons and others, a growing number of research programmes are using transcriptomics, and still more are proposing to do so.

Transcriptomics resembles its predecessor stages in several ways with respect to stress. Its proof of concept, initial successes (DeRisi et al., 1997; Chu et al., 1998), and many of its most spectacular successes to date have characterized the responses of cells and organisms to stress, which is not unexpected due to the anticipated overwhelming impact of stress on the transcriptome. Transcriptomics has clearly informed our understanding of stress by confirming changes in expression of numerous expected genes and highlighting numerous unexpected genes for future study. It has since enjoyed broad application to diverse areas of the life and health sciences outside stress; developmental biology and molecular diagnosis are good examples. Indeed, although here we focus on the use of transcriptomics in understanding stress responses and their evolution, the issues we raise are equally pertinent to the use of transcriptomics throughout ecology, evolutionary biology, and evolutionary and ecological physiology. In this sense, ‘stress’ should be viewed as a specific case of a general context. Finally, the limitations of transcriptomics have become apparent along with its advantages. Concerns have arisen about artifact due to inappropriate methodology and statistical mis-analysis of data. Fortunately, awareness of these concerns has had a large and beneficial impact, which is reviewed elsewhere (Gracey & Cossins, 2003). Here instead we focus on biological issues relevant to the ability of transcriptomics to accomplish one purpose: revealing the genes that matter in stress responses and their evolution. Limitations in accomplishing this purpose are the principal topic here, and thus we will emphasize them, admittedly without devoting equal time to the advantages and accomplishments of transcriptional profiling. This emphasis on limitations, however, should not undermine the appropriate and reasoned application of transcriptomics. As with candidate gene vs. natural variation approaches (Feder & Krebs, 1997), transcriptomic analyses of stress usually require that other approaches be pursued in tandem and may be unrevealing in isolation.

Protein activity (and not mRNA abundance) is typically consequential for fitness

  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein activity (and not mRNA abundance) is typically consequential for fitness
  5. How often are the genes that transcriptional profiling ‘discovers’ consequential for fitness?
  6. Can transcriptional profiling detect stress where other methods fail?
  7. Conclusions
  8. Acknowledgments
  9. References

In the final analysis, what matters most to organisms’ responses to abiotic stress and any consequent evolution is organisms’ ability to survive and reproduce; i.e. fitness. Fitness, in turn, at the biochemical level is primarily a result of the ability of proteins to function in their intra- and extracellular milieu, to integrate the diverse functions of the individual cells and organelles, to determine the levels of nonproteinaceous components, etc. Clearly, all proteins are encoded by genes, and the transcription of the encoding gene is necessary for protein expression. But to what extent can the abundance of a mRNA serve as an adequate proxy for the activity of the corresponding protein? If it cannot, then transcriptomics might provide limited insight into stress responses and their evolution.

Protein activity is not necessarily synonymous with protein abundance, although it can be. Provided that a protein is present in the cell, its activity (e.g. picomoles of substrate processed per unit time) can vary dramatically depending on the activity of its principle regulators (themselves often proteins), the abundance of its substrates and products, and other aspects of reaction conditions. Protein regulators not uncommonly interact with the proteins they regulate through multi-step networks of great complexity involving numerous other proteins. Thus, knowing how much of a protein is present in a cell, tissue, or organism – which might be inferred from mRNA abundance – prospectively tells little about that protein's activity. Indeed, understanding when, how often, and in what circumstances protein abundance can be equated with protein activity is among the goals of systems biology, but not among its extant accomplishments. Nonetheless, if no corresponding mRNA is present and has never been transcribed, no protein can be present except that inherited directly from the parent(s). Transcriptomics can clearly establish at least the possibility of a protein's presence at the time the transcriptome is sampled, and simultaneously so for the entire proteome. And, insofar as a protein's abundance might in some circumstances predict its activity, can transcriptomics find justification in predicting protein abundance?

Our current understanding of the biology of protein abundance suggests that mRNA levels sometimes can predict protein abundance but often cannot, for two reasons: First, many regulatory steps are between mRNA abundance and protein abundance (Fig. 1). These include pre, co- and post-translational modification (Lodish, 2003; Watson, 2004), each a multi-step process with substantial variation. While each can be important, alternative splicing is particularly noteworthy (Jaffe, 2003), with a single mRNA potentially yielding >23 000 different proteins (Schmucker et al., 2000; Schmucker, 2004)! Even though this may be a special case, alternative splicing seems to be a relatively common phenomenon in many organisms. Given equal quantities of mRNAs in a cell, differences in the quantity and variety of mature proteins can result. Second, even if the synthesis of mature protein is closely linked to the abundance of its corresponding mRNA, the concentration of mature protein is the net of its synthesis and degradation. Degradation mechanisms and rates can vary substantially and lead to corresponding variation in protein abundance. Admittedly these points are textbook knowledge and widely understood by established transcriptional profilers, but may not be obvious to those contemplating use of this technique.

image

Figure 1. Processes downstream of mRNA synthesis that may affect the total amount and/or activity of protein in the cell. Modified in part from (Watson, 2004).

Download figure to PowerPoint

Thus from first principles one might expect protein abundance frequently to be unpredictable from the mRNA abundance that transcriptomic studies report (Mehra et al., 2003), and a growing number of transcriptomic-proteomic comparisons bear this out (Table 1).

Table 1.  Outcome of studies relating mRNA abundance to protein abundance. Note that some of these studies were performed before current standards of experimental design and statistical rigor were in place.
ReferenceType of analysis, nature of sampleOutcome
ADirect correlation of mRNA and protein abundancer2 : mRNA vs. protein
Anderson & Seilhamer (1997)Human liver, 19 mRNAs0.23
Futcher et al. (1999)Yeast (Saccharyomyces cerevisiae), 148 proteins0.58
Gygi et al. (1999)Yeast (S. cerevisiae), 73 ‘less-expressed’ proteins0.13
Gygi et al. (1999)Yeast (S. cerevisiae), 106 ‘less-’ and ‘highly-expressed’ proteins0.87
Chen et al. (2002)Lung adenocarcinoma, 165 proteins0.00
Cardozo et al. (2003)60 proteins from mouse pancreatic islet cells and corresponding mRNAs from rat beta cells0.95
Ghaemmaghami et al. (2003)Yeast (S. cerevisiae), 4251 proteins0.44
Greenbaum et al. (2003)Yeast (S. cerevisiae), 2000 open reading frames0.44
Nishizuka et al. (2003)60 cancer cell lines, 52 proteins: nonstructural proteinsca. 0.20
Nishizuka et al. (2003)60 cancer cell lines, 52 proteins: structural proteinsca. 0.50
BCategorical analysis of mRNA and protein changes, increase vs. decrease vs. no changemRNA and protein co-vary (%)
Lian et al. (2001)Human, myeloid differentiation, 18 genes22
Fessler et al. (2002)Human cell line, ± lipopolysaccharide, 12 proteins42
Baliga et al. (2002)Halobacterium, 50 proteins34
Lorenz et al. (2003)Human, rheumatoid arthritis vs. osteoarthritis, 58 proteins changing28
CCorrelation of change in mRNA level and change in protein level between contrasting conditionsr2 : Δ mRNA vs. Δ protein
Ideker et al. (2001)Yeast (S. cerevisiae) ± galactose in various mutant backgrounds, 258 proteins0.37
Baliga et al. (2002)Halobacterium, 50 proteins0.34
Griffin et al. (2002)Yeast (S. cerevisiae) – 245 loci Galactose vs. ethanol0.04
Orntoft et al. (2002)Different human tumours, 40 proteinsUnstated but high
Lee et al. (2003)Escherichia coliUnstated but low
Washburn et al. (2003)Yeast (S. cerevisiae) – 678 loci YPDvs. nitrogen minimal medium0.20
White et al. (2004)Cell lines: serum-starved C3.6 cells, relative to HB4a cells; 43 genes0.656

These comparisons have used several approaches. First, investigators have measured mRNA and corresponding protein abundances, and simply correlated them (Table 1a). In such studies, the fraction of variation in protein abundance that is explained by variation in mRNA abundance (i.e. r2) is typically less than 0.5. Second, investigators have categorized mRNA and corresponding protein levels as increasing, decreasing, or unchanging, and asked how often the trend is identical for mRNA and protein (Table 1b). This frequency of covariation ranges from 22 to 42%. Third, investigators have compared mRNA and corresponding protein abundances after two or more experimental treatments, and correlated the change in mRNA with the change in protein (Table 1c). In such studies, r2 is also typically less than 0.5.

Given these outcomes, the probability of predicting whether a particular protein's concentration increases or decreases under stress would seem to be greater for a flip of a coin (50%) than for transcriptomics (typically <50%). These odds can be improved by confining analysis to certain classes of genes or proteins (Ideker et al., 2001; Greenbaum et al., 2003; Nishizuka et al., 2003) but only at the risk of substantial ascertainment basis.

Finally, how predictive is mRNA concentration of protein activity, the feature that is most closely linked to fitness? To date, a single study (Glanemann et al., 2003) has addressed this question. In this study of enzymes of amino acid biosynthesis in the bacterium Corynebacterium, protein activity sometimes exceeded mRNA abundance, and vice versa. Thus, as the authors stated, ‘it is difficult to generally predict protein activity from quantitative transcriptome data.’

How often are the genes that transcriptional profiling ‘discovers’ consequential for fitness?

  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein activity (and not mRNA abundance) is typically consequential for fitness
  5. How often are the genes that transcriptional profiling ‘discovers’ consequential for fitness?
  6. Can transcriptional profiling detect stress where other methods fail?
  7. Conclusions
  8. Acknowledgments
  9. References

Even though transcriptomics might infrequently reveal novel proteins that matter for stress resistance and its evolution, are these rare successes a sufficient justification?

As even the most avid advocates of transcriptomics admit, transcriptomics should not be an end in itself, but a means for revealing candidate mRNA species whose abundance and phenotypes as proteins require subsequent study. If this promise is borne out, then thus-revealed proteins should have phenotypes relevant to stress, their experimental manipulations should be consequential for stress resistance, and their encoding genes should bear the imprint of selection for stress resistance. Should they and do they? That is, are they genes and proteins ‘of large effect’ for stress resistance?

In the pregenomic era, Fisher and his successors contended that genes of large effect are rare in natural populations undergoing natural selection because selection would rapidly purge alleles of unfavoured morphs. Considerable debate continues on this viewpoint, with the opposing view sustained by findings from evolutionary developmental biology and studies of stress. This debate, moreover, has its counterpart at the protein level. The entire field of medical genetics is essentially a catalogue of examples wherein an existing allele of a single gene, if it encodes an aberrant protein, is sufficient for a major phenotype, typically a disease. But, to return to Fisher, the issue is not whether genes/proteins of large effect exist, but their frequency.

The combination of post-genomic technology and the advent of network thinking has enabled us to address this question in novel ways. One such technical advance has been genome-wide mutagenesis that systematically knocks out each gene in turn and screens for phenotypes (Carpenter & Sabatini, 2004). Surprisingly, the vast majority of such mutations produce no discernable phenotypes (Table 2). Even when phenotypes are evident, their impact on fitness is typically nonfatal (Table 2). Patterns of nucleotide variation in the corresponding genes corroborate these findings (Hirsh & Fraser, 2001,2003). Additionally, the correlation between mRNA abundance (or change in abundance) and fitness is negligible (Giaever et al., 2002; Warringer et al., 2003). Thus, most genes are remarkable for not being essential.

Table 2.  Numbers of dispensable or nonessential genes in complete and partial genomic genes. Dispensability is based on several criteria, most commonly the ability to survive if a gene's expression is prevented.
SourceOrganismNo. of genes screenedNo. of dispensablesDispensable (%)
Goebl & Petes (1986)Yeast (S. cerevisiae)  80
Brandon et al. (1995)Mice (embryonic lethality)  About 75
Miklos & Rubin (1996)Drosophila13 600360026
Winzeler et al. (1999)Yeast (S. cerevisiae), YPD medium2026162080
Birrell et al. (2001)Yeast (S. cerevisiae), UV radiation sensitivity4267423699
Bouche & Bouchez (2001)Arabidopsis morphologyAbout 200 >98
Maeda et al. (2001)Caenorhabditis elegansAbout 10 000 73
Giaever et al. (2002)Yeast (S. cerevisiae), rich glucose medium5916481181
Steinmetz et al. (2002)Yeast (S. cerevisiae), absence of fermentable substrate5916545092
Decottignies et al., 2003)Fission yeast (Schizosaccharomyces pombe)801483
Gerdes et al. (2003)Escherichia coli37463126 
May & Martienssen (2003)Arabidopsis25 50024 50096
Stephens & Laub (2003)Bacillus subtilis4100382993
Stephens & Laub (2003)Haemophilus influenzae1725124772
Stephens & Laub (2003)Mycoplasma genitalium517167–25232–49
Stephens & Laub (2003)Staphylococcus aureus2600194275
Warringer et al. (2003)Yeast (S. cerevisiae), NaCl5916About 541691
Conant & Wagner (2004)Caenorhabditis elegans single-copy genes8861787288
Conant & Wagner (2004)C. elegans multi-copy genes4704451596

Several explanations are that these nonessential or dispensable genes (a) truly are dispensable; (b) are redundant because of gene duplication (Gu et al., 2003) or alternative metabolic pathways (Wagner, 2000); (c) have conditional phenotypes only under conditions not applied by the experimentalist (‘Contingent function’ hypothesis, Thatcher et al., 1998); and/or (d) make small contributions to fitness rather than being essential (‘marginal benefit’ hypothesis, Thatcher et al., 1998). To the extent that the first two alternatives are true, the vast majority of genes implicated by transcriptomics can be expected to have no phenotype (Giaever et al., 2002; see Fig. 2).

image

Figure 2. Numbers of nonessential genes of yeast (S. cerevisiae) that have some phenotype under the stated culture conditions, out of 4738 ‘dispensable’ genes (see Giaever et al., 2002 for details).

Download figure to PowerPoint

To what extent are the first two alternatives true? Clearly, gene duplication is a ubiquitous feature of genomes and, until the duplicates have degenerated or differentiated, they are redundant (Lynch & Conery, 2000). Gene duplications aside, organisms comprise networks of duplicate and nonduplicate genes that confer substantial robustness to stress and other perturbations (Barabasi & Oltvai, 2004). Just as the knock-out of any single server is unlikely to compromise the Internet because information packets can be routed alternatively, knock-out of any single gene/protein is unlikely to compromise a cell because mass, energy, and/or information can be re-routed via alternative proteins. The principal exceptions – just as hubs in the air transport network – are hub proteins, those proteins interacting with far more than the average number of partner proteins, often via positioning at key intermediary positions. Hub proteins are far more likely to have phenotypes if knocked out than edge or nonhub proteins (Jeong et al., 2001; Han et al., 2004). By implication, a more rational strategy than transcriptomics for finding genes that matter may be to think deeply about the interactome and the physiology it enables, and to choose for study those genes likely to have an impact on stress resistance. This strategy, however, requires that the strategist first acquire a detailed understanding of functional biology, which many evolutionists avoid or view as irrelevant (Watt, 2000).

A case in point concerns agricultural biotechnology, in which a major goal is to engineer plants with enhanced stress resistance. Transcriptomics have revealed numerous candidate ‘transgenic intervention points’, genes whose manipulation ought to affect stress resistance. Experimentation, however, has failed to support many of these ‘transgenic intervention points’, the explanation being that duplicate or similar genes and pathways are able to compensate for the deletion or over-expression of candidate genes that transcriptomics implicates (Gutterson & Zhang, 2004). In Arabidopsis, for example, 21 genes encode variants of HSF, a major transcription factor in control of the heat-shock response (Nover et al., 2001). In knock-out studies of these, how many HSF-encoding genes are deleted may potentially be more important than which genes are deleted; i.e. many different combinations of HSF-encoding genes may suffice for a heat-shock response. Pragmatically this poses a logistical nightmare for confirming the genes that matter for stress.

Alternative (c), that some genes have no apparent phenotype because they have been studied under inappropriate or un-natural conditions, receives strong support from the work of Potts and colleagues on mouse genes (Meagher et al., 2000; Carroll et al., 2004). The genes they study have no discernable phenotypes in the laboratory. When mutant mice are placed in semi-natural enclosures, however, the same mutations have considerable effects. The problem such findings pose for the stress biologist is that in principle infinite combinations of stresses, stress levels, and stress kinetics that must be applied before concluding that gene actually has no phenotype, and the process is open-ended. Again, a more rational strategy than transcriptomics may be detailed natural historical study; i.e. asking the organism in its environment to reveal the stresses that matter and, by extension, the genes that are involved in responding to them.

In summary, several nonexclusive strategies may be preferable to transcriptomics in identifying the genes that matter for stress resistance and its evolution: (a) Nucleotide profiling; i.e. searching for evidence of selection in patterns of nucleotide variation; (b) Functional profiling; i.e. deducing from the known functions of genes which might be worthy of detailed scrutiny; (c) Genomic mutant screens; and (d) Environmental profiling; i.e. deducing from fitness variation in nature what combination of circumstances have the greatest impact on fitness. Each of these strategies, however, runs the risk of ascertainment bias; i.e. focusing attention on the most discoverable genes that matter rather typical genes. Indeed, our own research programme suffers from this bias. Elimination of this bias must await the development of robust quantitative models of all interactions among gene products in organisms, in which the contribution of each individual gene is apparent. [Interestingly, transcript abundance will clearly be a necessary component of these models (Hatzimanikatis & Lee, 1999).] First steps to this end are now available (Giot et al., 2003; Li et al., 2004), but only for the established model organisms and under limited sets of environmental conditions.

Can transcriptional profiling detect stress where other methods fail?

  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein activity (and not mRNA abundance) is typically consequential for fitness
  5. How often are the genes that transcriptional profiling ‘discovers’ consequential for fitness?
  6. Can transcriptional profiling detect stress where other methods fail?
  7. Conclusions
  8. Acknowledgments
  9. References

One oft-stated use of transcriptomics is as an exquisitely sensitive detector of stress (Gracey & Cossins, 2003). Expression of stress-inducible genes is indeed one of the very first indicators that an organism is undergoing stress, and screening the entire transcriptome maximizes the probability that any stress-inducible response will be detected. This thinking finds its counterpart in the use of transcriptional profiles as disease (e.g. cancer) biomarkers (Umar et al., 2004). Indeed, the transcriptomic profile of a tumour can reveal to which therapies it will be responsive, and whether these therapies are having their intended effect. Such insights may be unavailable by any other means. With respect to detection of environmental stress, transcriptional profiling may be valuable in some respects but less so in others. Transcriptomics can detect toxic effects on organisms in nature before whole-organism effects are evident and with no knowledge of the toxic substance(s). It may also detect rare or exotic toxic substances by their distinctive impact on gene expression (Custodia et al., 2001), and reveal which individuals in a population may be most (or least) susceptible to stress (Oleksiak et al., 2005). These capabilities may be especially important in environmental toxicology and endangered species preservation. For the bulk of studies of stress, by contrast, we suggest that many alternative indicators will suffice. Stress is often so obvious that it is detectable by classical physical methods, particularly with the advent of increasingly miniaturized and inexpensive detectors (Fitzhenry et al., 2004). In organisms amenable to transgenic manipulation, stress-responsive reporter genes can be inserted (Feder et al., 2000). Stress has a characteristic impact on microbial indicator species, which can be detected readily via nontranscriptomic microarrays (Letowski et al., 2003). Given these possibilities, does the expense and technical challenge of transcriptomics warrant its use as a stress detector? We suggest its use as a last resort rather than as a methodology of choice or convenience.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein activity (and not mRNA abundance) is typically consequential for fitness
  5. How often are the genes that transcriptional profiling ‘discovers’ consequential for fitness?
  6. Can transcriptional profiling detect stress where other methods fail?
  7. Conclusions
  8. Acknowledgments
  9. References

Genomic cDNA or oligonucleotide arrays have considerable value in evolutionary studies of stress. They are ideal, for example, for detecting gene duplication (Riehle et al., 2001), fine mapping of genes (Borevitz et al., 2003), polymorphism detection (Borevitz & Chory, 2004), gene association (Borevitz & Chory, 2004), sentinel species detection (Letowski et al., 2003), marking developmental events (Arbeitman et al., 2002), and whenever mRNA abundance sensu stricto is at issue. As an example of the last point, transcriptional profiling has provided compelling evidence that interspecific differences in the schedule of transcript expression may be a significant contributor to the reproductive isolation of species (Michalak & Noor, 2003,2004). Transcriptomics can be the method of choice if, for example, the objective is to identify those genes whose transcription might be linked or co-regulated under stress, or to deduce the nucleotide sequence responsible for high or low levels of gene expression. It may remind busy investigators about already well-known genes that have inadvertently been overlooked in designing candidate gene studies (J. Schultz, personal communication). It has demonstrated enormous complexity in the responses of wild organisms to natural stress (Gracey et al., 2004; Podrabsky & Somero, 2004).

An initial promise of transcriptional profiling, however, was that it would reveal previously unrecognized genes that matter for stress resistance (and/or for other traits of interest to ecologists and evolutionary biologists), and that these genes’ phenotypes could then be established. In our view, this specific promise is largely unfulfilled. Transcriptomics has largely confirmed the expression of already recognized genes. As yet it has seldom if ever led to the elucidation of the fitness consequences of novel or unforeseen genes that matter to ecologists and evolutionary biologists. Moreover, as soon as fitness of organisms under stress is at issue, all of the reservations expressed in the foregoing sections come into play. As stated, several approaches may be preferable to transcriptomics in this circumstance. Also, a growing number of proteomic techniques and systems biology approaches are becoming available for bypassing a principal limitation of transcriptomics, that mRNA abundance does not equal protein function. [Doubtless, however, proteomics and other ‘omics’ approaches will exhibit their own limitations and pitfalls as these approaches mature (Lubec & Pollak, 2004), and some are already apparent (Newton et al., 2004).] All such approaches, however, must likely cope with the reality of extensive redundancy in the phenome – that one-by-one manipulation of genes, proteins, and higher-level traits under a single specified stress regime may seldom reveal that genes matter for stress resistance, even when such genes actually do. Clearly this expectation has exceptions, but initial phenomic screens and models suggest that these are exceptions rather than the rule. If so, understanding adaptation to stress may be a vastly more challenging – but rewarding – enterprise than the first applications of transcriptomics suggested.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein activity (and not mRNA abundance) is typically consequential for fitness
  5. How often are the genes that transcriptional profiling ‘discovers’ consequential for fitness?
  6. Can transcriptional profiling detect stress where other methods fail?
  7. Conclusions
  8. Acknowledgments
  9. References

We thank Volker Loeschcke and Kuke Bijlsma for the invitation to the meeting for which this manuscript was prepared, and Rebecca Peterson Brown for advice. We acknowledge with special thanks the comments of Andrew Cossins and Douglas Crawford, who disagreed strenuously with some of our major points and whose opinions we continue to respect. Research was supported by NSF IBN03-16627.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Protein activity (and not mRNA abundance) is typically consequential for fitness
  5. How often are the genes that transcriptional profiling ‘discovers’ consequential for fitness?
  6. Can transcriptional profiling detect stress where other methods fail?
  7. Conclusions
  8. Acknowledgments
  9. References
  • Anderson, L. & Seilhamer, J. 1997. A comparison of selected mRNA and protein abundances in human liver. Electrophoresis 18: 533537.
  • Arbeitman, M.N., Furlong, E.E.M., Imam, F., Johnson, E., Null, B.H., Baker, B.S., Krasnow, M.A., Scott, M.P., Davis, R.W. & White, K.P. 2002. Gene expression during the life cycle of Drosophila melanogaster. Science 297: 22702275.
  • Baliga, N.S., Pan, M., Goo, Y.A., Yi, E.C., Goodlett, D.R., Dimitrov, K., Shannon, P., Aebersold, R., Ng, W.V. & Hood, L. 2002. Coordinate regulation of energy transduction modules in Halobacterium sp analyzed by a global systems approach. PNAS 99: 1491314918.
  • Barabasi, A.L. & Oltvai, Z.N. 2004. Network biology: understanding the cell's functional organization. Nat. Rev. Genet. 5: 101113.
  • Bijlsma, R. & Loeschcke, V. (eds) 1997. Environmental Stress, Adaptation, and Evolution. Birkhäuser Verlag, Basel.
  • Birrell, G.W., Giaever, G., Chu, A.M., Davis, R.W. & Brown, J.M. 2001. A genome-wide screen in Saccharomyces cerevisiae for genes affecting UV radiation sensitivity. PNAS 98: 1260812613.
  • Borevitz, J.O. & Chory, J. 2004. Genomics tools for QTL analysis and gene discovery. Curr. Opin. Plant Biol. 7: 132136.
  • Borevitz, J.O., Liang, D., Plouffe, D., Chang, H.S., Zhu, T., Weigel, D., Berry, C.C., Winzeler, E. & Chory, J. 2003. Large-scale identification of single-feature polymorphisms in complex genomes. Genome Res. 13: 513523.
  • Bouche, N. & Bouchez, D. 2001. Arabidopsis gene knockout: phenotypes wanted. Curr. Opin. Plant Biol. 4: 111117.
  • Brandon, E.P., Idzerda, R.L. & McKnight, G.S. 1995. Knockouts – targeting the mouse genome.1. A compendium of knockouts. Curr. Biol. 5: 625634.
  • Cardozo, A.K., Berthou, L., Kruhoffer, M., Orntoft, T., Nicolls, M.R. & Eizirik, D.L. 2003. Gene microarray study corroborates proteomic findings in rodent islet cells. J. Proteome Res. 2: 553555.
  • Carpenter, A.E. & Sabatini, D.M. 2004. Systematic genome-wide screens of gene function. Nat. Rev. Genet. 5: 1122.
  • Carroll, L.S., Meagher, S., Morrison, L., Penn, D.J. & Potts, W.K. 2004. Fitness effects of a selfish gene (the Mus T complex) are revealed in an ecological context. Evolution 58: 13181328.
  • Chen, G.A., Gharib, T.G., Huang, C.C., Taylor, J.M.G., Misek, D.E., Kardia, S.L.R., Giordano, T.J., Iannettoni, M.D., Orringer, M.B., Hanash, S.M. & Beer, D.G. 2002. Discordant protein and mRNA expression in lung adenocarcinomas. Mol. Cell. Proteomics 1: 304313.
  • Chu, S., DeRisi, J., Eisen, M., Mulholland, J., Botstein, D., Brown, P.O. & Herskowitz, I. 1998. The transcriptional program of sporulation in budding yeast. Science 282: 699705.
  • Conant, G.C. & Wagner, A. 2004. Duplicate genes and robustness to transient gene knock-downs in Caenorhabditis elegans. Proc. R. Soc. Lond., B, Biol. Sci. 271: 8996.
  • Custodia, N., Won, S.J., Novillo, A., Wieland, M., Li, C. & Callard, I.P. 2001. XCaenorhabditiselegans as an environmental monitor using DNA microarray analysis. In: Environmental Hormones The Scientific Basis of Endocrine Disruption, Annals of the New York Academy of Sciences 948: 3242.
  • Decottignies, A., Sanchez-Perez, I. & Nurse, P. 2003. Schizosaccharomyces pombe essential genes: a pilot study. Genome Res. 13: 399406.
  • DeRisi, J.L., Iyer, V.R. & Brown, P.O. 1997. Exploring the metabolic and genetic control of gene expression on a genomic scale. Science 278: 680686.
  • Feder, M.E. & Hofmann, G.E. 1999. Heat-shock proteins, molecular chaperones, and the stress response: evolutionary and ecological physiology. Annu. Rev. Physiol. 61: 24382.
  • Feder, M.E. & Krebs, R.A. 1997. Ecological and evolutionary physiology of heat-shock proteins and the stress response in Drosophila: complementary insights from genetic engineering and natural variation. In: Stress, Adaptation, and Evolution (R.Bijlsma & V.Loeschcke, eds), pp. 155173. Birkhäuser Verlag, Basel.
  • Feder, M.E. & Mitchell-Olds, T. 2003. Evolutionary and ecological functional genomics. Nat. Rev. Genet. 4: 651657.
  • Feder, M.E., Roberts, S.P. & Bordelon, A.C. 2000. Molecular thermal telemetry of free-ranging adult Drosophila melanogaster. Oecologia 123: 460465.
  • Fessler, M.B., Malcolm, K.C., Duncan, M.W. & Worthen, G.S. 2002. A genomic and proteomic analysis of activation of the human neutrophil by lipopolysaccharide and its mediation by p38 mitogen-activated protein kinase. J. Biol. Chem. 277: 3129131302.
  • Fitzhenry, T., Halpin, P.M. & Helmuth, B. 2004. Testing the effects of wave exposure, site, and behavior on intertidal mussel body temperatures: applications and limits of temperature logger design. Marine Biol. 145: 339349.
  • Futcher, B., Latter, G.I., Monardo, P., McLaughlin, C.S. & Garrels, J.I. 1999. A sampling of the yeast proteome. Mol. Cell. Biol. 19: 73577368.
  • Gerdes, S.Y., Scholle, M.D., Campbell, J.W., Balazsi, G., Ravasz, E., Daugherty, M.D., Somera, A.L., Kyrpides, N.C., Anderson, I., Gelfand, M.S., Bhattacharya, A., Kapatral, V., D'Souza, M., Baev, M.V., Grechkin, Y., Mseeh, F., Fonstein, M.Y., Overbeek, R., Barabasi, A.L., Oltvai, Z.N. & Osterman, A.L. 2003. Experimental determination and system level analysis of essential genes in Escherichia coli MG1655. J. Bacteriol. 185: 56735684.
  • Ghaemmaghami, S., Huh, W., Bower, K., Howson, R.W., Belle, A., Dephoure, N., O'Shea, E.K. & Weissman, J.S. 2003. Global analysis of protein expression in yeast. Nature 425: 737741.
  • Giaever, G., Chu, A.M., Ni, L., Connelly, C., Riles, L., Veronneau, S., Dow, S., Lucau-Danila, A., Anderson, K., Andre, B., Arkin, A.P., Astromoff, A., El Bakkoury, M., Bangham, R., Benito, R., Brachat, S., Campanaro, S., Curtiss, M., Davis, K., Deutschbauer, A., Entian, K.D., Flaherty, P., Foury, F., Garfinkel, D.J., Gerstein, M., Gotte, D., Guldener, U., Hegemann, J.H., Hempel, S., Herman, Z., Jaramillo, D.F., Kelly, D.E., Kelly, S.L., Kotter, P., LaBonte, D., Lamb, D.C., Lan, N., Liang, H., Liao, H., Liu, L., Luo, C.Y., Lussier, M., Mao, R., Menard, P., Ooi, S.L., Revuelta, J.L., Roberts, C.J., Rose, M., Ross-Macdonald, P., Scherens, B., Schimmack, G., Shafer, B., Shoemaker, D.D., Sookhai-Mahadeo, S., Storms, R.K., Strathern, J.N., Valle, G., Voet, M., Volckaert, G., Wang, C.Y., Ward, T.R., Wilhelmy, J., Winzeler, E.A., Yang, Y.H., Yen, G., Youngman, E., Yu, K.X., Bussey, H., Boeke, J.D., Snyder, M., Philippsen, P., Davis, R.W. & Johnston, M. 2002. Functional profiling of the Saccharomyces cerevisiae genome. Nature 418: 387391.
  • Giot, L., Bader, J.S., Brouwer, C., Chaudhuri, A., Kuang, B., Li, Y., Hao, Y.L., Ooi, C.E., Godwin, B., Vitols, E., Vijayadamodar, G., Pochart, P., Machineni, H., Welsh, M., Kong, Y., Zerhusen, B., Malcolm, R., Varrone, Z., Collis, A., Minto, M., Burgess, S., McDaniel, L., Stimpson, E., Spriggs, F., Williams, J., Neurath, K., Ioime, N., Agee, M., Voss, E., Furtak, K., Renzulli, R., Aanensen, N., Carrolla, S., Bickelhaupt, E., Lazovatsky, Y., DaSilva, A., Zhong, J., Stanyon, C.A., Finley, R.L., White, K.P., Braverman, M., Jarvie, T., Gold, S., Leach, M., Knight, J., Shimkets, R.A., McKenna, M.P., Chant, J. & Rothberg, J.M. 2003. A protein interaction map of Drosophila melanogaster. Science 302: 17271736.
  • Glanemann, C., Loos, A., Gorret, N., Willis, L.B., O'Brien, X.M., Lessard, P.A. & Sinskey, A.J. 2003. Disparity between changes in mRNA abundance and enzyme activity in Corynebacterium glutamicum: implications for DNA microarray analysis. Appl. Microbiol. Biotechnol. 61: 6168.
  • Goebl, M.G. & Petes, T.D. 1986. Most of the yeast genomic sequences are not essential for cell-growth and division. Cell 46: 983992.
  • Gracey, A.Y. & Cossins, A.R. 2003. Application of microarray technology in environmental and comparative physiology. Annu. Rev. Physiol. 65: 231259.
  • Gracey, A.Y., Fraser, E.J., Li, W., Fang, Y., Taylor, R.R., Rogers, J., Brass, A. & Cossins, A.R. 2004. Coping with cold: an integrative, multitissue analysis of the transcriptome of a poikilothermic vertebrate. PNAS 101: 1697016975.
  • Greenbaum, D., Colangelo, C., Williams, K. & Gerstein, M. 2003. Comparing protein abundance and mRNA expression levels on a genomic scale. Genome Biol. 4.
  • Griffin, T.J., Gygi, S.P., Ideker, T., Rist, B., Eng, J., Hood, L. & Aebersold, R. 2002. Complementary profiling of gene expression at the transcriptome and proteome levels in Saccharomyces cerevisiae. Mol. Cell. Proteomics 1: 323333.
  • Gu, Z.L., Steinmetz, L.M., Gu, X., Scharfe, C., Davis, R.W. & Li, W.H. 2003. Role of duplicate genes in genetic robustness against null mutations. Nature 421: 6366.
  • Gutterson, N. & Zhang, J.Z. 2004. Genomics applications to biotech traits: a revolution in progress? Curr. Opin. Plant Biol. 7: 226230.
  • Gygi, S.P., Rochon, Y., Franza, B.R. & Aebersold, R. 1999. Correlation between protein and mRNA abundance in yeast. Mol. Cell. Biol. 19: 17201730.
  • Han, J.D.J., Bertin, N., Hao, T., Goldberg, D.S., Berriz, G.F., Zhang, L.V., Dupuy, D., Walhout, A.J.M., Cusick, M.E., Roth, F.P. & Vidal, M. 2004. Evidence for dynamically organized modularity in the yeast protein-protein interaction network. Nature 430: 8893.
  • Hatzimanikatis, V. & Lee, K. 1999. Dynamical analysis of gene networks requires both mRNA and protein expression information. Metab. Eng. 1: 275281.
  • Hirsh, A.E. & Fraser, H.B. 2001. Protein dispensability and rate of evolution. Nature 411: 10461049.
  • Hirsh, A.E. & Fraser, H.B. 2003. Rate of evolution and gene dispensability – Reply. Nature 421: 497498.
  • Hochachka, P.W. & Somero, G.N. 2002. Biochemical Adaptation: Mechanism and Process in Physiological Evolution. Oxford University Press, New York.
  • Hoffmann, A.A. & Parsons, P.A. 1997. Extreme Environmental Change and Evolution. Cambridge University Press, Cambridge.
  • Ideker, T., Thorsson, V., Ranish, J.A., Christmas, R., Buhler, J., Eng, J.K., Bumgarner, R., Goodlett, D.R., Aebersold, R. & Hood, L. 2001. Integrated genomic and proteomic analyses of a systematically perturbed metabolic network. Science 292: 929934.
  • Jaffe, S. 2003. Alternative splicing goes mainstream. Scientist 17: 2830.
  • Jeong, H., Mason, S.P., Barabasi, A.L. & Oltvai, Z.N. 2001. Lethality and centrality in protein networks. Nature 411: 4142.
  • Lee, P.S., Shaw, L.B., Choe, L.H., Mehra, A., Hatzimanikatis, V. & Lee, K.H. 2003. Insights into the relation between mRNA and protein expression patterns: II. Experimental observations in Escherichia coli. Biotechnol. Bioeng. 84: 834841.
  • Letowski, J., Brousseau, R. & Masson, L. 2003. DNA microarray applications in environmental microbiology. Anal. Lett. 36: 31653184.
  • Li, S.M., Armstrong, C.M., Bertin, N., Ge, H., Milstein, S., Boxem, M., Vidalain, P.O., Han, J.D.J., Chesneau, A., Hao, T., Goldberg, D.S., Li, N., Martinez, M., Rual, J.F., Lamesch, P., Xu, L., Tewari, M., Wong, S.L., Zhang, L.V., Berriz, G.F., Jacotot, L., Vaglio, P., Reboul, J., Hirozane-Kishikawa, T., Li, Q.R., Gabel, H.W., Elewa, A., Baumgartner, B., Rose, D.J., Yu, H.Y., Bosak, S., Sequerra, R., Fraser, A., Mango, S.E., Saxton, W.M., Strome, S., den Heuvel, S., Piano, F., Vandenhaute, J., Sardet, C., Gerstein, M., Doucette-Stamm, L., Gunsalus, K.C., Harper, J.W., Cusick, M.E., Roth, F.P., Hill, D.E. & Vidal, M. 2004. A map of the interactome network of the metazoan C. elegans. Science 303: 540543.
  • Lian, Z., Wang, L., Yamaga, S., Bonds, W., Beazer-Barclay, Y., Kluger, Y., Gerstein, M., Newburger, P.E., Berliner, N. & Weissman, S.M. 2001. Genomic and proteomic analysis of the myeloid differentiation program. Blood 98: 513524.
  • Lodish, H.F. 2003. Molecular Cell Biology, 5th edn. W.H. Freeman and Company, New York.
  • Lorenz, P., Ruschpler, P., Koczan, D., Stiehl, P. & Thiesen, H.J. 2003. From transcriptome to proteome: Differentially expressed proteins identified in synovial tissue of patients suffering from rheumatoid arthritis and osteoarthritis by an initial screen with a panel of 791 antibodies. Proteomics 3: 9911002.
  • Lubec, G. & Pollak, D. 2004. Pitfalls in proteomics 1. Amino Acids 26: 215215.
  • Lynch, M. & Conery, J.S. 2000. The evolutionary fate and consequences of duplicate genes. Science 290: 11511155.
  • Maeda, I., Kohara, Y., Yamamoto, M. & Sugimoto, A. 2001. Large-scale analysis of gene function in Caenorhabditis elegans by high-throughput RNAi. Curr. Biol. 11: 171176.
  • May, B.P. & Martienssen, R.A. 2003. Transposon mutagenesis in the study of plant development. Crit. Rev. Plant Sci. 22: 135.
  • Meagher, S., Penn, D.J. & Potts, W.K. 2000. Male-male competition magnifies inbreeding depression in wild house mice. PNAS 97: 33243329.
  • Mehra, A., Lee, K.H. & Hatzimanikatis, V. 2003. Insights into the relation between mRNA and protein expression patterns: I. Theoretical considerations. Biotechnol. Bioeng. 84: 822833.
  • Michalak, P. & Noor, M.A.F. 2003. Genome-wide patterns of expression in Drosophila pure species and hybrid males. Mol. Biol. Evol. 20: 10701076.
  • Michalak, P. & Noor, M.A.F. 2004. Association of misexpression with sterility in hybrids of Drosophila simulans and D. mauritiana. J. Mol. Evol. 59: 277282.
  • Miklos, G.L.G. & Rubin, G.M. 1996. The role of the genome project in determining gene function: insights from model organisms. Cell 86: 521529.
  • Newton, R.P., Brenton, A.G., Smith, C.J. & Dudley, E. 2004. Plant proteome analysis by mass spectrometry: principles, problems, pitfalls and recent developments. Phytochemistry 65: 14491485.
  • Nishizuka, S., Charboneau, L., Young, L., Major, S., Reinhold, W.C., Waltham, M., Kouros-Mehr, H., Bussey, K.J., Lee, J.K., Espina, V., Munson, P.J., Petricoin, E., Liotta, L.A. & Weinstein, J.N. 2003. Proteomic profiling of the NCI-60 cancer cell lines using new high-density reverse-phase lysate microarrays. PNAS 100: 1422914234.
  • Nover, L., Bharti, K., Doring, P., Mishra, S.K., Ganguli, A. & Scharf, K.D. 2001. Arabidopsis and the heat stress transcription factor world: how many heat stress transcription factors do we need? Cell Stress Chaperones 6: 177189.
  • Oleksiak, M.F., Roach, J.L. & Crawford, D.L. 2005. Natural variation in cardiac metabolism and gene expression in Fundulus heteroclitus. Nat. Genet. 37: 6772.
  • Orntoft, T.F., Thykjaer, T., Waldman, F.M., Wolf, H. & Celis, J.E. 2002. Genome-wide study of gene copy numbers, transcripts, and protein levels in pairs of non-invasive and invasive human transitional cell carcinomas. Mol. Cell. Proteomics 1: 3745.
  • Podrabsky, J.E. & Somero, G.N. 2004. Changes in gene expression associated with acclimation to constant temperatures and fluctuating daily temperatures in an annual killifish Austrofundulus limnaeus. J. Exp. Biol. 207: 22372254.
  • Riehle, M.M., Bennett, A.F. & Long, A.D. 2001. Genetic architecture of thermal adaptation in Escherichia coli. PNAS 98: 525530.
  • Rutherford, S.L. 2003. Between genotype and phenotype: protein chaperones and evolvability. Nat. Rev. Genet. 4: 263274.
  • Schmucker, D. 2004. Dscam reveals its bountiful plan. Scientist 18: 14.
  • Schmucker, D., Clemens, J.C., Shu, H., Worby, C.A., Xiao, J., Muda, M., Dixon, J.E. & Zipursky, S.L. 2000. Drosophila Dscam is an axon guidance receptor exhibiting extraordinary molecular diversity. Cell 101: 671684.
  • Steinmetz, L.M., Scharfe, C., Deutschbauer, A.M., Mokranjac, D., Herman, Z.S., Jones, T., Chu, A.M., Giaever, G., Prokisch, H., Oefner, P.J. & Davis, R.W. 2002. Systematic screen for human disease genes in yeast. Nat. Genet. 31: 400404.
  • Stephens, C.M. & Laub, M.T. 2003. Microbial genomics: all that you can't leave behind. Curr. Biol. 13: R571R573.
  • Thatcher, J.W., Shaw, J.M. & Dickinson, W.J. 1998. Marginal fitness contributions of nonessential genes in yeast. PNAS 95: 253257.
  • Umar, A., Kapetanovic, I. & Khan, J. 2004. The Applications of Bioinformatics in Cancer Detection. New York Academy of Sciences, New York.
  • Wagner, A. 2000. Robustness against mutations in genetic networks of yeast. Nat. Genet. 24: 355361.
  • Warringer, J., Ericson, E., Fernandez, L., Nerman, O. & Blomberg, A. 2003. High-resolution yeast phenomics resolves different physiological features in the saline response. PNAS 100: 1572415729.
  • Washburn, M.P., Koller, A., Oshiro, G., Ulaszek, R.R., Plouffe, D., Deciu, C., Winzeler, E. & Yates, J.R. 2003. Protein pathway and complex clustering of correlated mRNA and protein expression analyses in Saccharomyces cerevisiae. PNAS 100: 31073112.
  • Watson, J.D. 2004. Molecular Biology of the Gene, 5th edn. Pearson/Benjamin Cummings, San Francisco.
  • Watt, W.B. 2000. Avoiding paradigm-based limits to knowledge of evolution. Evol. Biol. 32: 7396.
  • White, S.L., Gharbi, S., Bertani, M.F., Chan, H.L., Waterfield, M.D. & Timms, J.F. 2004. Cellular responses to ErbB-2 overexpression in human mammary luminal epithelial cells: comparison of mRNA and protein expression. Brit. J. Cancer 90: 173181.
  • Winzeler, E.A., Shoemaker, D.D., Astromoff, A., Liang, H., Anderson, K., Andre, B., Bangham, R., Benito, R., Boeke, J.D., Bussey, H., Chu, A.M., Connelly, C., Davis, K., Dietrich, F., Dow, S.W., El Bakkoury, M., Foury, F., Friend, S.H., Gentalen, E., Giaever, G., Hegemann, J.H., Jones, T., Laub, M., Liao, H., Liebundguth, N., Lockhart, D.J., Lucau-Danila, A., Lussier, M., M'Rabet, N., Menard, P., Mittmann, M., Pai, C., Rebischung, C., Revuelta, J.L., Riles, L., Roberts, C.J., Ross-MacDonald, P., Scherens, B., Snyder, M., Sookhai-Mahadeo, S., Storms, R.K., Veronneau, S., Voet, M., Volckaert, G., Ward, T.R., Wysocki, R., Yen, G.S., Yu, K.X., Zimmermann, K., Philippsen, P., Johnston, M. & Davis, R.W. 1999. Functional characterization of the S. cerevisiae genome by gene deletion and parallel analysis. Science 285: 901906.