• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of geneticsGeneticsCurrent IssueInformation for AuthorsEditorial BoardSubscribeSubmit a Manuscript
Genetics. Jan 2007; 175(1): 199–206.
PMCID: PMC1775009

Impact of Transcriptional Properties on Essentiality and Evolutionary Rate

Abstract

We characterized general transcriptional activity and variability of eukaryotic genes from global expression profiles of human, mouse, rat, fly, plants, and yeast. The variability shows a higher degree of divergence between distant species, implying that it is more closely related to phenotypic evolution, than the activity. More specifically, we show that transcriptional variability should be a true indicator of evolutionary rate. If we rule out the effect of translational selection, which seems to operate only in yeast, the apparent slow evolution of highly expressed genes should be attributed to their low variability. Meanwhile, rapidly evolving genes may acquire a high level of transcriptional variability and contribute to phenotypic variations. Essentiality also seems to be correlated with the variability, not the activity. We show that indispensable or highly interactive proteins tend to be present in high abundance to maintain a low variability. Our results challenge the current theory that highly expressed genes are essential and evolve slowly. Transcriptional variability, rather than transcriptional activity, might be a common indicator of essentiality and evolutionary rate, contributing to the correlation between the two variables.

EVOLUTION of gene expression, which has long been a subject of great interest (King and Wilson 1975), is now being studied on a genomic scale with the help of rapidly growing microarray and genome sequence data (Enard et al. 2002; Oleksiak et al. 2002; Meiklejohn et al. 2003; Ranz et al. 2003; Rifkin et al. 2003; Khaitovich et al. 2004; Denver et al. 2005). Of particular importance, expression level has been believed to be the best indicator of the evolutionary rate of encoded proteins. Highly expressed genes were found to evolve slowly from bacteria to mammals (Sharp 1991; Duret and Mouchiroud 2000; Pal et al. 2001; Herbeck et al. 2003; Urrutia and Hurst 2003; Subramanian and Kumar 2004; Drummond et al. 2005). In addition, it has recently emerged as a governing factor behind the apparent relationships between evolutionary rate and other important genomic features. Specifically, the influences of protein–protein interactions and dispensability on evolutionary rate have been disputed on the grounds that their effects may be confounded with gene expression level (Hirsh and Fraser 2001; Fraser et al. 2002; Bloom and Adami 2003; Pal et al. 2003; Bloom and Adami 2004; Fraser and Hirsh 2004; Wall et al. 2005). In other words, when expression level was statistically controlled, the effects decreased or disappeared. Pal et al. (2003) argued that essential proteins evolve more slowly only because they are highly expressed. To rule out the direct effect of essentiality on evolutionary rate, they argued on the basis of the following two hypothetical relations:

  1. Essentiality and transcriptional activity: Each protein molecule may have the same amount of phenotypic contribution to an organism's fitness. Under this hypothesis, proteins that have more phenotypic contribution should have higher levels of active molecules in the cell. However, the validity of this hypothesis is highly questionable as genes involved in functions such as transcriptional regulation, ligand binding, and signal transduction are required only in small quantities even though they are vital for the organism. High abundance does not necessarily mean high fitness effect.
  2. Transcriptional activity and selective pressure: Highly expressed genes may prefer translationally efficient codons, which leads to a slow rate of nucleotide sequence changes (Akashi 2001, 2003; Akashi and Gojobori 2002). Recently, using Saccharomyces as a model organism, Drummond et al. (2005) argued that selection may act on codon preference (for translational accuracy) and on amino acid sequence (for translational robustness) to minimize the detrimental effects of protein misfolding. Since they experience more translation events, highly expressed genes should be subject to stronger selective pressure. However, this is hardly applicable to higher eukaryotes. First, the lack of translational selection on codon preference in larger genomes is a well-known phenomenon (Shields et al. 1988; Sharp et al. 1995; Akashi 1997, 2001; dos Reis et al. 2004). Second, in a recent study (Wright et al. 2004), it has been shown that the number of tissues in which a gene is expressed, not the total amount of translation events across tissues, is an important determinant of evolutionary rate in Arabidopsis. We expect that this should be the case with other multicellular organisms.

Here we propose two alternative relations:

  1. Essentiality and transcriptional variability: One can anticipate that essential proteins should have low genetic and physiological variation. To perform core functions in the cell, they are constitutively required in different individuals and physiological conditions. This leads to the expectation that transcriptional variability, which can be expressed as genetic, physiological, or random variation at different levels, may be a better indicator of essentiality than transcriptional activity. To test this hypothesis, we compared the transcriptional variability of genes associated with essential gene ontology (GO) categories to that of the other genes. For experimental validations, protein dispensability and interaction data for yeast proteins were used for a measure of essentiality.
  2. Transcriptional variability and selective pressure: The presence of high variation among individuals in a population may indicate the action of weak purifying selection on that gene. One can envision that those genes have evolved to possess a high transcriptional variability to be expressed in specific conditions. In contrast, the genes under strong purifying selection are likely to exhibit a constant level of expression in various conditions, maintaining a limited level of variability. This explains the observation of a correlation between evolutionary rate and tissue specificity (Wright et al. 2004). Duplicates could be used as a good source for the quantitative measure of a correlation between expression divergence and sequence divergence (Gu 2004; Gu et al. 2004, 2005). Drummond et al. (2005) demonstrated that divergence in transcriptional activity correlated with sequence divergence between duplicates in yeast. However, we speculated that divergence in transcriptional variability would correlate better with sequence divergence in eukaryotes. We tested the second hypothesis with a genomewide analysis followed by the duplicates studies.

To provide experimental support for these hypotheses, we made use of a microarray database to extract general transcriptional properties of each gene. An average expression level across a wide range of biological conditions, including different individuals, times, tissues, disease states, environmental conditions, and so on, defined general transcriptional activity. Coefficient of variation (CV), the ratio of standard deviation (SD) over mean, was adopted as a measure of general transcriptional variability. CV has been used as a measure of stochastic fluctuation or “noise” in gene expression (Elowitz et al. 2002; Ozbudak et al. 2002; Blake et al. 2003; Raser and O'Shea 2004). According to the models of stochastic gene expression, noise should increase as the amount of transcript decreases (Hasty and Collins 2002; Swain et al. 2002). Low expression level may be coupled with high physiological or genetic variation as well as stochastic variation. Small changes in the amount of proteins that are normally expressed at a low level may have a greater impact on the cell or organism than large changes in the amount of proteins whose normal expression level is high. Therefore, it may be that there is a close correspondence between transcriptional activity and variability.

MATERIALS AND METHODS

Measuring transcriptional activity and variability:

Supplemental Table 1 at http://www.genetics.org/supplemental/ shows a compendium of the multispecies microarray data obtained from the GEO database (http://www.ncbi.nlm.nih.gov/geo/). The raw Affymetrix data were used without log transformation. A flooring of the lowest 1% and ceiling of the top 1% was applied. From the microarray data sets with at least five arrays, we computed the mean and CV of expression values for each gene. To integrate different scales of microarray data sets, we scaled the mean values to equation M1, where equation M2 is the mean expression level of ith gene in the jth data set, while equation M3 and equation M4 indicate the average and standard deviation of all mean values in the jth data set. Finally, global transcriptional properties of gene i were obtained as equation M5 and equation M6, namely, the mean z-score and mean CV across n data sets. These final estimates of transcriptional activity and variability can be downloaded at our supplemental website, http://centi.kribb.re.kr, along with all of the raw microarray data.

Evolutionary conservation index:

To provide a measure of evolutionary rate on a macroevolutionary timescale, we used the HomoloGene database (http://www.ncbi.nlm.nih.gov/HomoloGene). We assigned each gene a conservation score defined as the number of its orthologs among 18 eukaryotic species. Thus, the score ranges from 1, meaning the gene is specific to one species, to 18, meaning the gene is conserved among all the 18 species. Rapidly evolving genes, including recently emerged genes (Long et al. 2003), will probably be conserved only in a few organisms and thus have a low conservation score. On the contrary, slowly evolving genes, such as ancient genes, will have a high conservation score. This score is expected to reflect selective pressure differently than substitution rates. As such, we used the conservation score as an inverse estimate of evolutionary rate. We refer to it as the evolutionary conservation index (ECI).

Measuring substitution rates:

Orthologous relationships among human, mouse, and rat proteins were obtained from the HomoloGene database. Synonymous (dS) and nonsynonymous (dN) substitution rates were calculated from the alignments of coding sequences using PAML's codonml (Yang 1997). PAML's baseml was then used to estimate Jukes–Cantor distances between orthologous nucleotide sequences. A total of 12,671 human–mouse and 7154 mouse–rat ortholog pairs with observed expression pattern produced appropriate protein and nucleotide sequence alignments for PAML input. Paralogs were also identified from the HomoloGene database. We performed all-against-all comparisons between paralogs to estimate dN, dS, the transcriptional activity difference, and transcriptional variability difference. Transcriptional-activity divergence was measured as |z1z2|, where z1 and z2 mean the transcriptional activity of two paralogous genes. Transcriptional-variability divergence was measured as |CV1 − CV2|, where CV1 and CV2 mean the transcriptional variability of two paralogous genes.

Principal-component analysis-based regression:

Although similar in principle, the two representative principal-component analysis (PCA)-based regression methods have different purposes. Principal-component regression (PCR) determines the linear combinations of the predictors that explain most of the variation in these predictors, ignoring the response variable. Partial least-squares regression (PLS), in contrast, finds the linear combinations that best explain the response, yielding different results according to the type of the response (in this case, ECI, dN, dS, the fitness effect, and protein interaction number). Although practically it produces similar results to PLS, PCR is not designed to determine the influence of the predictors on the response. In fact, PCR failed to distinguish the contribution of the predictors (transcriptional activity and variability) in many cases in our analysis. The R package “pls” was used to perform PLS. We scaled the predictors to zero mean and unit variance before conducting PLS.

Functional category analysis:

For each GO category with >10 genes, we computed a normalized transcriptional variability (z-score) from the CVs of all members of that category. The z-score was defined as equation M7, where equation M8 is the average CV for the genes in the GO category, equation M9 is the average CV for all genes in the species, equation M10 is the standard deviation for the genes in the category, and equation M11 is the number of the genes in the category. The z-scores are listed for each species in the supporting information available at http://centi.kribb.re.kr. For each category, we obtained a specieswise z-score as equation M12, where equation M13 indicates the z-score in the kth species when equation M14. The specieswise z-scores are given in supplemental Table 4 at http://www.genetics.org/supplemental/. The significant categories with more than a total of 100 genes across the species were selected for presentation in supplemental Tables 2 and 3 at http://www.genetics.org/supplemental/.

RESULTS AND DISCUSSION

Evolution of general transcriptional properties:

We collected Affymetrix microarray data for the species for which a considerable amount of expression profiles have been produced (supplemental Table 1 at http://www.genetics.org/supplemental/). For example, 177 different data sets containing 16,446 HomoloGene entries in 4136 arrays were gathered as a human expression profile. The mean and CV of expression values were obtained for each gene. The mean value was scaled to a z-score. The average z-score and CV were computed across the data sets within a species to represent general transcriptional activity and variability, respectively.

We needed to check if the expression profiles used in the analysis contained a sufficient amount of data to convey information on the universal transcriptional properties of each gene. To test this, we made comparisons of the z-score and CV among the species on the assumption that the transcriptional properties should be evolutionarily conserved. Since human, mouse, and rat had much more microarray data sets than the other species (supplemental Table 1 at http://www.genetics.org/supplemental/), we randomly selected 20 data sets for each species to remove the sample-size effect. For HomoloGene entries whose transcriptional properties were observed in more than one species, we calculated the correlation coefficient for every species pair and found striking correlations in most cases (Table 1). The transcriptional activity is well conserved from yeast to human. Whereas maintaining a high level of similarity among human, mouse, and rat, the variability shows some degree of divergence between distant species. Using the whole data for human, mouse, and rat resulted in the same patterns. These findings suggest that our measures of transcriptional activity and variability reflect inherent biological features that are highly conserved among orthologs, while remaining free of sampling problems. Moreover, the higher level of divergence in the variability over that measured in the activity implies that transcriptional variability might play a more important role in the evolution of phenotypic variations.

TABLE 1
Correlation of transcriptional properties among orthologs

Transcriptional activity vs. transcriptional variability:

To address our main question, we essentially needed to measure the correlations of these transcriptional properties with various response variables, namely estimates of essentiality and evolutionary rate. Spearman's rank correlation coefficient was used since it is robust to outliers and able to properly handle different scales of various estimates. We denote the correlation of transcriptional activity as Ra and that of transcriptional variability as Rv. A primary concern during the analysis was the association between transcriptional activity and variability. As mentioned in the Introduction, there seems to be a correspondence between them. We observed the pattern as expected in all the species (data not shown). This led to the use of a partial correlation with which we can measure the correlation of transcriptional activity (or variability) with the response variables when controlling for transcriptional variability (or activity). We denote it as Ra|v (or Rv|a). We also employed multivariate regression to compare the influences of the two predictors on the response variables, as estimated by the regression coefficient or slope (Ca and Cv).

According to Drummond et al. (2006), noisy variables may result in spurious partial correlations. In this regard, they suggest the use of a PCA-based regression approach. However, if one of the original correlations, Ra or Rv, is found to be much higher than the other, the noise problem will not change the conclusion based on the partial correlations. Therefore, we can simply ask if |Ra| > |Rv| (or |Rv| > |Ra|) when Ra|v (or Rv|a) is significant. However, we also made use of a PCA-based regression method as suggested. We present the percentage of contributions of transcriptional activity and variability to the first principal component as PCa and PCv in the following sections.

First hypothesis—essentiality and transcriptional variability:

To estimate essentiality, we adopted data from yeast and mouse deletion experiments. Growth rates of yeast deletion strains were measured by an array-based method (Giaever et al. 2004). Using this data set, Wall et al. (2005) recently reported a positive relationship between dispensability and evolutionary rate. As an inverse measure of essentiality, we used the fitness effect, f(i), as 1 − g(i)/g(max), where g(i) is the growth rate of the strain with gene i deleted, and g(max) is the maximal growth rate (Hirsh and Fraser 2001). Mouse genes subject to deletion experiments were obtained from the Mouse Genome Informatics (MGI) database (http://www.informatics.jax.org). We selected genes with the knockout phenotype of lethality (MP:0005373, “lethality postnatal”; MP:0005374, “lethality embryonic/perinatal”). The phenotype database included 1427 essential genes (lethality = 1) and 1956 nonessential genes (lethality = 0) according to our criteria. Ra and Rv for the fitness effect and lethality were estimated (Table 2). We observed significant partial correlations of transcriptional variability (Rv|a = −0.244 and −0.115) while |Rv| > |Ra|. On the contrary, transcriptional activity showed weak correlations (Ra|v = 0.057 and 0.015). The regression methods confirmed the disproportionate contribution of transcriptional variability to fitness effect and lethality. Hubs in protein networks are known to be essential, which prompted us to use the number of protein interactions as an estimate of essentiality (He and Zhang 2006). The same trend was found in our data (Table 2), which is in good agreement with the previously reported negative relationship between genetic variation in gene expression and the number of protein interactions (Lemos et al. 2004). It was also shown that this relationship was not confounded by gene expression level.

TABLE 2
Comparative analysis of the influences of transcriptional activity and variability on essentiality in yeast and mouse

To extend this conclusion to other taxa, a functional category analysis was carried out. For each GO category, the average CV of the genes in the category was obtained, normalized to a z-score in each species, and converted to a specieswise z-score (see materials and methods). Significantly negative z-scores indicate that the genes in the category exhibit a low variability relative to the average of all genes in the species. Supplemental Table 2 at http://www.genetics.org/supplemental/ lists the GO terms with significantly low z-scores while supplemental Table 3 at http://www.genetics.org/supplemental/ shows those with significantly high z-scores (the full list of the specieswise z-scores is given in supplemental Table 4, and the species-specific z-scores are available at our supplemental website, http://centi.kribb.re.kr). We find that the GO categories in supplemental Table 2 are biased toward essential cellular processes such as transcriptional and translational regulation, protein folding and transport, protein catabolism, protein complexation, RNA processing, etc. In terms of the cellular component, the genes located in the nucleus, cytoplasm, mitochondrion, and Golgi apparatus show low transcriptional variability. In contrast, supplemental Table 3 is enriched for the GO terms associated with extracellular communication such as immune response, cell–cell signaling, cell adhesion, surface receptors, growth factor activity, hormone activity, chemotaxis, etc. In the same context, the genes located in the extracellular space and plasma membrane show high variability. Notably, these findings are strikingly consistent with the results of a recent study by Chuang and Li (2004). Most of the GO categories reported to have low substitution rates are found in our list of the GO categories with less variable genes, and vice versa. Likewise, there is a remarkable overlap between the GO categories with high substitution rates and those with highly variable genes. These findings are suggestive of correspondence between transcriptional variability and substitution rate.

Evolutionary conservation on a macroevolutionary timescale:

As this study spans a long evolutionary time from yeast to human, we needed to estimate evolutionary rates differently from substitution rates between paired nucleotide sequences. One solution was to count the number of genomes in which the gene is present. This approach was successfully used in the previous studies about the relationships between evolutionary rate and protein–protein interactions or protein dispensability (Jordan et al. 2002, 2003; Wuchty et al. 2003).

The ECI produced results that are consistent with conventional substitution rates. Substitution rates between orthologs cannot be reliably used for distantly related species. Therefore, we calculated them between mammalian orthologs and compared them to the ECI data. Strong inverse correlations were found: Spearman's R = −0.362–~−0.420 (P-values [double less-than sign]10−16). Moreover, the correlations of ECI with protein–protein interactions (Spearman's R = 0.224) and dispensability (Spearman's R = −0.270) were compatible with the reported correlations of substitution rates with protein–protein interactions (Spearman's R = −0.21, Fraser et al. 2002) and dispensability (Spearman's R = 0.230, Wall et al. 2005). However, using different reference species for the computation of substitution rates has fueled arguments about the association of protein–protein interactions and dispensability with evolutionary rate (Jordan et al. 2003; Pal et al. 2003; Wall et al. 2005). In contrast, the ECI is expected to offer a global and unified measure of evolutionary rate for diverse organisms spanning a long evolutionary time period.

Second hypothesis—transcriptional variability and selective pressure:

Tables 355 show the statistical analysis results regarding the relationship between evolutionary rate and transcriptional properties. First, as a genomewide analysis, we investigated the influence of transcriptional activity and variability on ECI. As shown in Table 3, the correlation, partial correlation, multivariate regression, and PCA-based regression analyses all indicate that the influence of variability is much stronger than that of activity. This tendency holds true when we use traditional substitution rates between mammalian orthologs (Table 4). The only exception occurred in yeast and is discussed later. Using ECI as the response variable, all the different statistical techniques showed the same pattern, in favor of transcriptional activity in yeast and variability in the other species. Moreover, the levels of the statistics are comparable among various species (e.g., −0.1 < Rv|a < −0.25, −21 < Cv < −33).

TABLE 3
Comparative analysis of the influences of transcriptional activity and variability on the evolutionary conservation index
TABLE 4
Comparative analysis of the influences of transcriptional activity and variability on substitution rates between mammalian orthologs
TABLE 5
Comparative analysis of the influences of transcriptional-activity divergence and transcriptional-variability divergence on sequence divergence between paralogous genes

Next, we extracted paralog information from the HomoloGene database. Accelerated evolution of duplicate gene expression has been shown in yeast and fruit fly (Gu et al. 2004). Here, we compared the transcriptional variability of duplicates with that of singletons in all the species for which we obtained data. As shown in supplemental Figure 1 at http://www.genetics.org/supplemental/, the distributions shifted toward high CV values with gene duplications, indicating increased transcriptional variability among duplicates. We also observed that overall CV values are relatively lower in plants and in yeast as compared to higher eukaryotes. We next asked whether sequence divergence, measured by dN and dS, is better explained by transcriptional-activity divergence (|z1z2|, where z1 and z2 mean the transcriptional activity of two paralogous genes) or by transcriptional-variability divergence (|CV1 − CV2|, where CV1 and CV2 mean the transcriptional variability of two paralogous genes). For human, mouse, and fly, all the results are supportive of our hypothesis (Table 5). In the case of rat, although the correlation and partial correlation analysis failed to detect a specific trend, the regression analysis results strongly support our hypothesis. The signals are relatively weak in plants and yeast; nonetheless, in agreement with the ECI results, the partial correlations indicate that the effect of transcriptional activity is stronger in yeast.

Do highly expressed genes evolve slowly?

We have shown that contrary to the consensus that transcriptionally active genes evolve slowly, evolution shapes transcriptional variability rather than transcriptional activity. Selective pressure seems to act primarily on transcriptional variability. If this is so, how can we explain the apparent correlation between transcriptional activity and evolutionary rate observed in the previous studies?

First, as described in the Introduction, translational selection has mainly explained the correlation (Akashi 2001, 2003; Akashi and Gojobori 2002; Drummond et al. 2005). Tables 3 and and55 show that evolutionary rate is still correlated with transcriptional activity in yeast, even after controlling for transcriptional variability. Therefore, if we rule out the effect of translational selection such as in higher eukaryotes (Shields et al. 1988; Sharp et al. 1995; Akashi 1997, 2001; dos Reis et al. 2004; Wright et al. 2004), we can conclude that invariable genes evolve slowly.

Second, from the inverse relationship between the mean and CV, we can speculate that as transcriptional activity increases, relative transcriptional variability will tend to decrease. In biological terms, expression changes of abundant proteins are likely to have smaller effects on the cell than those of scanty proteins. High expression levels may confer tolerance to a fluctuation in the amount, while low expression levels may enable delicate transcriptional regulation. From a stochastic perspective, a high level of transcriptional activity leads to the reduction of random noise, which means a reduced variability. This explanation can be also applied to the correlation of expression level and essentiality. Indeed, the production of essential proteins was shown to involve lower levels of noise (Fraser et al. 2004). These speculations suggest that a high abundance of slowly evolving or essential proteins may be evolutionarily favored to maintain a low variability in expression.

Taken together, the two aspects, the action of translational selection in yeast and the correspondence between transcriptional activity and variability, may explain the apparent correlation between transcriptional activity and evolutionary rate in the previous studies.

Conclusion:

In this study, we characterized general transcriptional activity and variability of eukaryotic genes from global expression profiles of various species spanning a long evolutionary time period. While the transcriptional properties were shown to be remarkably conserved during the evolutionary processes, the variability showed a higher degree of divergence between distant species. Transcriptional variability might be related to phenotypic variations and thus be more subject to selective pressure. Indeed, we showed that transcriptional variability should be a true indicator of evolutionary rate. If we rule out the effect of translational selection, which seems to operate only in yeast, the apparent slow evolution of highly expressed genes should be attributed to their low variability. Selective forces may enable phenotypic variations to evolve mainly by shaping transcriptional variability. Furthermore, we suggest that a high abundance of essential proteins may be favored to maintain a low variability in their amount. Transcriptional variability, rather than transcriptional activity, might be a common indicator of essentiality and evolutionary rate, contributing to the correlation between the two variables.

Acknowledgments

We are grateful to the members of the Integrative Evolutionary Analysis Team, ArrayPort Team, and Core Team at Korean BioInformation Center for producing the data and maintaining the computer systems we used in the analysis. This work was supported by the Korean Ministry of Science and Technology under grant no. M10407010001-04N0701-00110.

References

  • Akashi, H., 1997. Codon bias evolution in Drosophila. Population genetics of mutation-selection drift. Gene 205: 269–278. [PubMed]
  • Akashi, H., 2001. Gene expression and molecular evolution. Curr. Opin. Genet. Dev. 11: 660–666. [PubMed]
  • Akashi, H., 2003. Translational selection and yeast proteome evolution. Genetics 164: 1291–1303. [PMC free article] [PubMed]
  • Akashi, H., and T. Gojobori, 2002. Metabolic efficiency and amino acid composition in the proteomes of Escherichia coli and Bacillus subtilis. Proc. Natl. Acad. Sci. USA 99: 3695–3700. [PMC free article] [PubMed]
  • Blake, W. J., M. Kaern, C. R. Cantor and J. J. Collins, 2003. Noise in gene expression. Nature 422: 633–637. [PubMed]
  • Bloom, J. D., and C. Adami, 2003. Apparent dependence of protein evolutionary rate on number of interactions is linked to biases in protein-protein interactions data sets. BMC Evol. Biol. 3: 21. [PMC free article] [PubMed]
  • Bloom, J. D., and C. Adami, 2004. Evolutionary rate depends on number of protein-protein interactions independently of gene expression level: response. BMC Evol. Biol. 4: 14. [PMC free article] [PubMed]
  • Chuang, J. H., and H. Li, 2004. Functional bias and spatial organization of genes in mutational hot and cold regions in the human genome. PLoS Biol. 2: e29. [PMC free article] [PubMed]
  • Denver, D. R., K. Morris, J. T. Streelman, S. K. Kim, M. Lynch et al., 2005. The transcriptional consequences of mutation and natural selection in Caenohabditis elegans. Nat. Genet. 37: 544–548. [PubMed]
  • dos Reis, M., R. Savva and L. Wernisch, 2004. Solving the riddle of codon usage preferences: a test for translational selection. Nucleic Acids Res. 32: 5036–5044. [PMC free article] [PubMed]
  • Drummond, D. A., J. D. Bloom, C. Adami, C. O. Wilke and F. H. Arnold, 2005. Why highly expressed proteins evolve slowly. Proc. Natl. Acad. Sci. USA 102: 14338–14343. [PMC free article] [PubMed]
  • Drummond, D. A., A. Raval and C. O. Wilke, 2006. A single determinant dominates the rate of yeast protein evolution. Mol. Biol. Evol. 23: 327–337. [PubMed]
  • Duret, L., and D. Mouchiroud, 2000. Determinants of substitution rates in mammalian genes: expression pattern affects selection intensity but not mutation rate. Mol. Biol. Evol. 17: 68–74. [PubMed]
  • Elowitz, M. B., A. J. Levine, E. D. Siggia and P. S. Swain, 2002. Stochastic gene expression in a single cell. Science 297: 1183–1186. [PubMed]
  • Enard, W., P. Khaitovich, J. Klose, S. Zollner, F. Heissig et al., 2002. Intra- and interspecific variation in primate gene expression patterns. Science 296: 340–343. [PubMed]
  • Fraser, H. B., and A. E. Hirsh, 2004. Evolutionary rate depends on number of protein-protein interactions independently of gene expression level. BMC Evol. Biol. 4: 13. [PMC free article] [PubMed]
  • Fraser, H. B., A. E. Hirsh, L. M. Steinmetz, C. Scharfe and M. W. Feldman, 2002. Evolutionary rate in the protein interaction network. Science 296: 750–752. [PubMed]
  • Fraser, H. B., A. E. Hirsh, G. Giaever, J. Kumm and M. B. Eisen, 2004. Noise minimization in eukaryotic gene expression. PLoS Biol. 2: e137. [PMC free article] [PubMed]
  • Giaever, G., P. Flaherty, J. Kumm, M. Proctor, C. Nislow et al., 2004. Chemogenomic profiling: identifying the functional interactions of small molecules in yeast. Proc. Natl. Acad. Sci. USA 101: 793–798. [PMC free article] [PubMed]
  • Gu, X., 2004. Statistical framework for phylogenomic analysis of gene family expression profiles. Genetics 167: 531–542. [PMC free article] [PubMed]
  • Gu, X., Z. Zhang and W. Huang, 2005. Rapid evolution of expression and regulatory divergences after yeast gene duplication. Proc. Natl. Acad. Sci. USA 102: 707–712. [PMC free article] [PubMed]
  • Gu, Z., S. A. Rifkin, K. P. White and W.-H. Li, 2004. Duplicate genes increase gene expression diversity within and between species. Nat. Genet. 36: 577–579. [PubMed]
  • Hasty, J., and J. J. Collins, 2002. Translating the noise. Nat. Genet. 31: 13–14. [PubMed]
  • He, X., and J. Zhang, 2006. Why do hubs tend to be essential in protein networks? PLoS Genet. 2: e88. [PMC free article] [PubMed]
  • Herbeck, J. T., D. P. Wall and J. J. Wernegreen, 2003. Gene expression level influences amino acid usage, but not codon usage, in the tsetse fly endosymbiont Wigglesworthia. Microbiology 149: 2585–2596. [PubMed]
  • Hirsh, A. E., and H. B. Fraser, 2001. Protein dispensability and rate of evolution. Nature 411: 1046–1048. [PubMed]
  • Jordan, I. K., I. B. Rogozin, Y. I. Wolf and E. V. Koonin, 2002. Essential genes are more evolutionarily conserved than are nonessential genes in bacteria. Genome Res. 12: 962–968. [PMC free article] [PubMed]
  • Jordan, I. K., Y. I. Wolf and E. V. Koonin, 2003. No simple dependence between protein evolution rate and the number of protein-protein interactions: only the most prolific interactors tend to evolve slowly. BMC Evol. Biol. 3: 1. [PMC free article] [PubMed]
  • Khaitovich, P., G. Weiss, M. Lachmann, I. Hellmann, W. Enard et al., 2004. A neutral model of transcriptome evolution. PLoS Biol. 2: e132. [PMC free article] [PubMed]
  • King, M. C., and A. C. Wilson, 1975. Evolution at two levels: molecular similarities and biological differences between humans and chimpanzees. Science 188: 107–116. [PubMed]
  • Lemos, B., C. D. Meiklejohn and D. L. Hartl, 2004. Regulatory evolution across the protein interaction network. Nat. Genet. 36: 1059–1060. [PubMed]
  • Long, M., E. Betran, K. Thornton and W. Wang, 2003. The origin of new genes: glimpses from the young and old. Nat. Rev. Genet. 4: 865–875. [PubMed]
  • Meiklejohn, C. D., J. Parsch, J. M. Ranz and D. L. Hartl, 2003. Rapid evolution of male-based gene expression in Drosophila. Proc. Natl. Acad. Sci. USA 100: 9894–9899. [PMC free article] [PubMed]
  • Oleksiak, M. F., G. A. Churchill and D. L. Crawford, 2002. Variation in gene expression within and among natural populations. Nat. Genet. 32: 261–266. [PubMed]
  • Ozbudak, E. M., M. Thattai, I. Kurtser, A. D. Grossman and A. van Oudenaarden, 2002. Regulation of noise in the expression of a single gene. Nat. Genet. 31: 69–73. [PubMed]
  • Pal, C., B. Papp and L. D. Hurst, 2001. Highly expressed genes in yeast evolve slowly. Genetics 158: 927–931. [PMC free article] [PubMed]
  • Pal, C., B. Papp and L. D. Hurst, 2003. Rate of evolution and gene dispensability. Nature 421: 496–497. [PubMed]
  • Ranz, J. M., C. I. Castillo-Davis, C. D. Meiklejohn and D. L. Hartl, 2003. Sex-dependent gene expression and evolution of the Drosophila transcriptome. Science 300: 1742–1745. [PubMed]
  • Raser, J. M., and E. K. O'Shea, 2004. Control of stochasticity in eukaryotic gene expression. Science 304: 1811–1814. [PMC free article] [PubMed]
  • Rifkin, S. A., J. Kim and K. P. White, 2003. Evolution of gene expression in the Drosophila melanogaster subgroup. Nat. Genet. 33: 138–144. [PubMed]
  • Sharp, P. M., 1991. Determinants of DNA sequence divergence between Escherichia coli and Salmonella typhimurium: codon usage, map position, and concerted evolution. J. Mol. Evol. 33: 23–33. [PubMed]
  • Sharp, P. M., M. Averof, A. T. Lloyd, G. Matassi and J. F. Peden, 1995. DNA sequence evolution: the sounds of silence. Philos. Trans. R. Soc. Lond. Ser. B 349: 241–247. [PubMed]
  • Shields, D. C., P. M. Sharp, D. G. Higgins and F. Wright, 1988. “Silent” sites in Drosophila genes are not neutral: evidence of selection among synonymous codons. Mol. Biol. Evol. 5: 704–716. [PubMed]
  • Subramanian, S., and S. Kumar, 2004. Gene expression intensity shapes evolutionary rates of the proteins encoded by the vertebrate genome. Genetics 168: 373–381. [PMC free article] [PubMed]
  • Swain, P. S., M. B. Elowitz and E. D. Siggia, 2002. Intrinsic and extrinsic contributions to stochasticity in gene expression. Proc. Natl. Acad. Sci. USA 99: 12795–12800. [PMC free article] [PubMed]
  • Urrutia, A. O., and L. D. Hurst, 2003. The signature of selection mediated by expression on human genes. Genome Res. 13: 2260–2264. [PMC free article] [PubMed]
  • Wall, D. P., A. E. Hirsh, H. B. Fraser, J. Kumm, G. Giaever et al., 2005. Functional genomic analysis of the rates of protein evolution. Proc. Natl. Acad. Sci. USA 102: 5483–5488. [PMC free article] [PubMed]
  • Wright, S. I., C. B. K. Yau, M. Looseley and B. C. Meyers, 2004. Effects of gene expression on molecular evolution in Arabidopsis thaliana and Arabidopsis lyrata. Mol. Biol. Evol. 21: 1719–1726. [PubMed]
  • Wuchty, S., Z. N. Oltvai and A.-L. Barabasi, 2003. Evolutionary conservation of motif constituents in the yeast protein interaction network. Nat. Genet. 35: 176–179. [PubMed]
  • Yang, Z., 1997. PAML: a program package for phylogenetic analysis by maximum likelihood. Comput. Appl. Biosci. 13: 555–556. [PubMed]

Articles from Genetics are provided here courtesy of Genetics Society of America
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...