Molecular Evolution of Sex-Biased Genes in Drosophila
http://www.100md.com
分子生物学进展 2004年第11期
Department of Biology II, Section of Evolutionary Biology, University of Munich (LMU), Munich, Germany
E-mail: parsch@zi.biologie.uni-muenchen.de.
Abstract
Studies of morphology, interspecific hybridization, protein/DNA sequences, and levels of gene expression have suggested that sex-related characters (particularly those involved in male reproduction) evolve rapidly relative to non–sex-related characters. Here we report a general comparison of evolutionary rates of sex-biased genes using data from cDNA microarray experiments and comparative genomic studies of Drosophila. Comparisons of nonsynonymous/synonymous substitution rates (dN/dS) between species of the D. melanogaster subgroup revealed that genes with male-biased expression had significantly faster rates of evolution than genes with female-biased or unbiased expression. The difference was caused primarily by a higher dN in the male-biased genes. The same pattern was observed for comparisons among more distantly related species. In comparisons between D. melanogaster and D. pseudoobscura, genes with highly biased male expression were significantly more divergent than genes with highly biased female expression. In many cases, orthologs of D. melanogaster male-biased genes could not be identified in D. pseudoobscura through a Blast search. In contrast to the male-biased genes, there was no clear evidence for accelerated rates of evolution in female-biased genes, and most comparisons indicated a reduced rate of evolution in female-biased genes relative to unbiased genes. Male-biased genes did not show an increased ratio of nonsynonymous/synonymous polymorphism within D. melanogaster, and comparisons of polymorphism/divergence ratios suggest that the rapid evolution of male-biased genes is caused by positive selection.
Key Words: codon bias ? comparative genomics ? gene expression ? sexual selection
Introduction
It has long been known that traits associated with sexual reproduction (particularly those related to male reproductive success) often show greater interspecific divergence than do nonreproductive traits. Darwin (1871) documented the frequent occurrence of extravagant secondary sexual characteristics in species of many taxa (including crustaceans, insects, fish, birds, and mammals) and observed that these traits often show large morphological differences between closely related species. Furthermore, Darwin noted that such traits were, with few exceptions, limited to the male of the species. These observations were explained by the theory of sexual selection, which posited that male reproductive traits evolved in response to male-male competition for mating opportunities and the preferential mating of females to males with "attractive" phenotypes. Morphological evidence suggests that sexual selection also acts on primary sexual characteristics. For example, sperm size and morphology are known to differ greatly among insect species (Jamieson 1987), as is the external morphology of male genitalia (Eberhard 1985; Hosken and Stockley 2004).
Studies of interspecific hybridization have also suggested the rapid evolution of male reproductive characters. Haldane (1922) noted a common pattern regarding the viability and fertility of species hybrids. His observation, known as Haldane's Rule, was that when hybrid offspring of only one sex are either inviable or infertile, it is most often the heterogametic sex. In many taxa, such as mammals and Drosophila, the males are heterogametic (XY), and, thus, hybrid male offspring are more prone to be inviable or sterile. Two major hypotheses have been proposed to explain Haldane's rule. The first hypothesis, known as the "dominance" hypothesis posits that hybrid incompatibilities are often recessive and, thus, are only observed in the sex with hemizygous sex chromosomes (Turelli and Orr 1995). The second hypothesis is known as "faster male evolution" and posits that genes involved in male reproduction evolve faster than genes involved in female reproduction or genes with nonreproductive function (Wu and Davis 1993). This hypothesis can only completely explain Haldane's rule for taxa in which the males are heterogametic and is expected to apply primarily to hybrid sterility, not inviability. However, the two hypotheses are not mutually exclusive, and it is likely that both faster male evolution and dominance play a role in hybrid breakdown (Presgraves and Orr 1998). Faster male evolution may also explain the overwhelming preponderance of male sterility factors relative to inviability factors that have been identified from Drosophila hybridizations (Wu and Davis 1993; True, Weir, and Laurie 1996; Tao et al. 2003).
Recent studies have indicated that sex-related genes show increased rates of evolution in their protein/DNA sequences, and it has been suggested that sexual selection affects the evolution of a broad range of genes with reproductive functions (see reviews by Civetta and Singh [1999], Singh and Kulathinal [2000], and Swanson and Vacquier [2002]). In Drosophila, the rapid evolution of reproductive proteins is suggested by the relatively large interspecific differences in migration pattern observed for these proteins on two-dimensional electrophoresis gels (Coulthart and Singh 1988; Civetta and Singh 1995) and by the elevated rate of amino acid substitution between D. melanogaster and D. simulans observed for a large number of male-specific accessory gland proteins (Swanson et al. 2001). In addition, a number of male-specific genes showing evidence for rapid evolution caused by positive selection have been identified, including Acp26Aa (Tsaur and Wu 1997; Aguadé 1998; Tsaur, Ting, and Wu 1998), OdsH (Ting et al. 1998), Sdic (Nurminsky et al. 1998; Nurminsky et al. 2001), Dntf-2r (Betrán and Long 2003), and jan-ocn (Parsch et al. 2001; Parsch, Meiklejohn, and Hartl 2001). Recently, observations of faster male evolution have been extended to the level of gene expression (Meiklejohn et al. 2003; Ranz et al. 2003). These studies used competitive cDNA microarray hybridizations to demonstrate that genes with male-biased expression show greater expression differences both within and between species than do either female-biased or unbiased genes.
In this article, we use data from published cDNA microarray experiments to identify Drosophila genes showing either male bias or female bias in their expression. We then investigate the rates of evolution of these genes (using the ratio of nonsynonymous/synonymous substitution rates, dN/dS) among Drosophila species and compare them with a collection of control genes that show no sex bias in their expression. Our results indicate that male-biased genes have a significantly higher rate of evolution than both female-biased and unbiased genes. A similar pattern is observed for the evolution of highly sex-biased genes between D. melanogaster and D. pseudoobscura, with male-biased genes showing significantly greater divergence between these two species than do either female-biased or unbiased genes. Polymorphism and divergence data suggest that these differences are caused by increased positive selection on male-biased genes.
Materials and Methods
Identification of Genes with Sex-Biased Expression
Two independent data sets that compared male-biased and female-biased gene expression in D. melanogaster through competitive microarray hybridizations were used to classify genes as male biased, female biased, or unbiased in their expression (Parisi et al. 2003; Ranz et al. 2003). Although these data sets were generated using similar experimental approaches, a direct comparison of the two is difficult because of methodological differences in microarray platform, gene number, RNA source material, replication scheme, and statistical analysis. For example, Parisi et al. (2003) used arrays of PCR amplicons (averaging 410 bp in length) corresponding to individual exons of approximately 75% of the predicted genes in the D. melanogaster genome, whereas Ranz et al. (2003) used arrays of full-length cDNAs corresponding to approximately 40% of all predicted genes (Rubin et al. 2000). In addition, Parisi et al. (2003) performed dissections to compare gene expression between testes and ovaries, whereas Ranz et al. (2003) performed comparisons of whole flies. Finally, Parisi et al. (2003) used an expression difference of twofold to classify genes as sex biased, whereas Ranz et al. (2003) used a Bayesian approach (Townsend and Hartl 2002) that could detect significant expression differences of less than twofold. To be conservative, we used the twofold cutoff for both data sets. That is, genes with a male/female (or testes/ovaries) ratio greater than 2.0 were considered male biased, genes with a ratio less than 0.5 were considered female biased, and genes with a ratio between 0.5 and 2.0 were considered unbiased (table 1). In rare cases (7% of genes) where the two data sets resulted in different classifications (e.g., greater than twofold male bias in one data set and less than twofold male bias in the other), the gene was considered sex biased. However, exclusion of these genes from our analyses does not affect the results (data not shown). A small number of genes (0.08%) with sex-bias conflict (e.g., male biased in one data set and female biased in the other) were excluded from further analyses. The twofold cutoff for expression bias was chosen as a conventional standard to allow comparison of microarray results generated using different array platforms and experimental designs. Reanalysis of the data using cutoff values ranging from 1.5-fold to threefold did not alter the qualitative pattern or statistical significance of our results (data not shown). Finally, it should be emphasized that all sex-bias classifications are based on expression studies of D. melanogaster. In some cases, the expression bias could be confirmed in D. simulans (Ranz et al. 2003). However, expression levels in more distantly related species, such as D. yakuba or D. pseudoobscura, have not been determined.
Table 1 Summary of Gene Expression and Comparative Genomic Data Sets
Analysis of Comparative Genomic Data Sets
Comparative sequence data for 371 adult cDNAs from D. yakuba (Domazet-Loso and Tautz 2003) were provided by T. Domazet-Loso. Of the 371 cDNAs, 237 could be classified as male biased, female biased, or unbiased based on microarray expression data (table 1). Evolutionary rates of these genes (dN, dS, and dN/dS) were calculated from pairwise comparisons of D. yakuba and D. melanogaster using the codeml program of the PAML software package (Yang 1997). Levels of codon bias were calculated as either effective number of codons (ENC [Wright 1990]) or frequency of optimal codons (Fop [Ikemura 1981]) based on the full-length transcript from D. melanogaster using the codonW program (http://bioweb.pasteur.fr/seqanal/interfaces/codonw.html). In cases where multiple transcripts were predicted for a gene, the longest transcript was used. For calculating Fop, codon frequency data from D. melanogaster were used.
Comparative sequence data for 81 orthologous genes from D. melanogaster, D. erecta, D. pseudoobscura, D. willistoni, and D. virilis (Bergman et al. 2002) were downloaded from the Berkeley Drosophila Genome Project Web site (http://www.fruitfly.org/comparative/index.html). These data come from sequenced fosmid clones (40 kb each) corresponding to the D. melanogaster genomic regions containing the apterous, even-skipped, fushi tarazu, twist, and Rhodopsin 1, 2, 3, and 4 genes. Of the 81 genes, 50 could be classified as male biased, female biased, or unbiased based on the microarray data (table 1). For these genes, values of dN, dS, and dN/dS were calculated for all pairwise comparisons of species using PAML (codeml runmode–2). For genes with sequences available from three or more species, dN/dS was also calculated for each gene using all available sequences and assuming a constant dN/dS over all branches of the phylogenetic tree (codeml runmode 0, model 0). In addition, we applied the "free-ratio" model (Yang 1998) that allows dN/dS to vary over all branches of the tree (codeml runmode 0, model 1) and compared the likelihood ratio of the two models using a 2 test with the degrees of freedom equal to the difference in parameter number (i.e., the number of branches in the tree minus 1). For these analyses, the phylogenetic relationship of Drosophila species given in Bergman et al. (2002) was used.
Comparison of Highly Sex-Biased Genes Between Drosophila Genomes
To investigate the evolutionary rates of genes showing the strongest sex bias in expression, we selected the 50 genes with the highest and lowest male/female (or testes/ovaries) ratios from both the Parisi et al. (2003) and Ranz et al. (2003) data sets. As a control, we selected the 50 genes showing a male/female ratio closest to 1 from each data set. Only genes corresponding to predicted transcripts in the D. melanogaster genome release 3.0 (Celniker et al. 2002) were included. Of the 100 genes selected for each expression class, 93 male-biased, 92 female-biased, and 99 unbiased genes remained after removing redundancies (table 1). The paucity of overlapping genes between the two data sets was primarily the result of differences in array composition; that is, genes present in one data set but absent in the other (50% of the genes). For another 47% of the genes, the difference was only in the level of sex bias, that is, the gene was in the top 50 of its expression class in one data set but not the other. For 3% of the genes, there was a conflict such that a gene was sex biased in one data set but unbiased in the other. There were no cases of conflicting male/female sex-bias classification among these genes. The coding sequences of these genes from D. melanogaster were used for a BlastN version 2.0 (Altschul et al. 1999) search of the D. pseudoobscura genome sequence using the Baylor College of Medicine Drosophila Genome Project Web site (http://www.hgsc.bcm.tmc.edu/projects/Drosophila/). Because the coding sequences of many of the genes could not be aligned between the two species, we report divergence as either Blast e-values or Blast scores. For score calculation, the default values of –5 and –2 were used for the gap creation and gap extension penalties, respectively. Levels of codon bias were calculated as described above using the full-length coding sequences from D. melanogaster.
Comparison of Polymorphism and Divergence
Because the largest number of DNA sequence polymorphism surveys have been conducted in D. melanogaster, we chose this species to investigate levels of polymorphism in sex-biased genes. We began with a database of 101 protein-encoding genes extracted from the literature and GenBank (provided by D. Presgraves) for which multiple (at least six) D. melanogaster alleles had been sequenced and at least one D. simulans allele was available for divergence analysis. A total of 55 genes could be classified as male biased, female biased, or unbiased based on the microarray expression data (table 1). A list of the individual genes and their associated references is provided as Supplementary Material online. One gene (Dntf-2r) was classified as unbiased based on the Parisi et al. (2003) data set, although it had been shown to be testis-specific by RT-PCR (Betrán and Long 2003). In this case, we classified the gene as male biased. The conflicting microarray result may be caused by cross-hybridization between Dntf-2r and its close paralog Dntf-2, which is expressed in both sexes (Betrán and Long 2003). DnaSP version 4 (Rozas et al. 2003) was used to calculate levels of polymorphism and divergence and to perform the MK test (McDonald and Kreitman 1991). All available D. melanogaster sequences were used for polymorphism calculations, and a single D. simulans allele was used for divergence.
Results and Discussion
Comparison of D. yakuba cDNAs to D. melanogaster
In an analysis of orphan gene evolution, Domazet-Loso and Tautz (2003) cloned and sequenced cDNAs from D. yakuba and estimated rates of evolution (as dN, dS, and dN/dS) by comparing the cDNA sequences to the D. melanogaster genome sequence (Celniker et al. 2002). We used the microarray results of Parisi et al. (2003) and Ranz et al. (2003) to classify the adult cDNAs analyzed by Domazet-Loso and Tautz (2003) as male biased, female biased, or unbiased, and then compared evolutionary rates among genes of the three expression classes (table 2). These comparisons clearly indicated an increased rate of evolution in male-biased genes relative to female-biased genes. The average value of dN/dS was threefold higher for male-biased genes than for female-biased genes, and that of dN was over fivefold higher. Male-biased genes also showed faster evolutionary rates than unbiased genes, with average values of dN/dS and dN both differing by a factor of 2 between the two expression classes. In contrast, female-biased genes had lower values of dN/dS and dN than did unbiased genes. The differences in dN/dS and dN were significant for all comparisons and were highly significant for comparisons between male-biased and female-biased genes (table 2).
Table 2 Evolutionary Rates and Levels of Codon Bias for Genes with Sex-Biased Expression Compared Between D. melanogaster and D. yakuba
We also observed significant differences in synonymous substitution rates among the three classes of genes. Male-biased genes had significantly higher dS than did unbiased genes, whereas female-biased genes had significantly lower dS than did unbiased genes (table 2). To investigate whether or not these differences could be explained by differing selective constraints on synonymous codon usage among genes of the three expression classes, we determined levels of codon bias for all genes by two measures (table 2; see Materials and Methods). For both ENC (where lower values indicate greater bias) and Fop (where higher values indicate greater bias), female-biased genes showed the greatest codon usage bias, whereas male-biased genes showed the least. The differences in codon bias were significant for all comparisons (table 2) and indicated an inverse relationship between level of codon bias and dS for the three classes of genes. Furthermore, there was a significant negative correlation between Fop and dS within the female-biased genes (fig. 1A), suggesting that selection for optimal codon usage might be responsible for the reduced synonymous substitution rate observed for this class of genes. It is also possible that the fixation of strongly selected amino acid replacements results in the fixation of linked, weakly deleterious, nonsynonymous substitutions (Betancourt and Presgraves 2002; Kim 2004). Consistent with this, we observed a significant negative correlation between Fop and dn within the female-biased genes (fig. 1B) and the unbiased genes (not shown). There was also a negative correlation between Fop and dn within the male-biased genes, although this correlation was not significant because of the small sample size and the absence of male-biased genes with high levels of codon bias.
FIG. 1.— Correlation between substitution rates and levels of codon bias for 78 female-biased genes compared between D. melanogaster and D. yakuba. (A) Plot of Fopversus dS (Spearman rank correlation test, R = –0.31, P < 0.01). (B) Plot of Fop versus dN (R = –0.36, P < 0.01). Fop (frequency of optimal codons; Ikemura 1981) was calculated based on D. melanogaster codon usage.
Previous studies in Drosophila demonstrated a significant paucity of male-biased genes and a significant excess of female-biased genes on the X chromosome (Swanson et al. 2001; Parisi et al. 2003; Ranz et al. 2003). In our comparison, 6% (2/33) of the male-biased genes were on the X chromosome, whereas 15% (12/78) of the female-biased genes were on the X chromosome. Thus, a general tendency for slower evolution of X chromosomal genes (Orr and Betancourt 2001) could potentially explain our observations. Consistent with previous observations (Betancourt, Presgraves, and Swanson 2002), we observed slightly lower (although not significantly so) evolutionary rates for X-linked genes (not shown). However, this pattern cannot explain the observed evolutionary rate differences between male-biased and female-biased genes. If we consider only autosomal genes, the average values of dN/dS, dN, and dS, for male-biased genes are 0.13, 0.05, and 0.35, whereas the corresponding values for female-biased genes are 0.05, 0.01, and 0.20. All of these differences are significant (Mann-Whitney test, P < 0.001). The small number of X-linked male-biased genes (two) precluded testing for evolutionary rate differences between male-biased and female-biased genes located on the X chromosome.
Comparison of Orthologous Genomic Regions Among Drosophila Species
To assess the impact of comparative sequence data on genome annotation, Bergman et al. (2002) sequenced fosmid clones (40 kb each) corresponding to the D. melanogaster genomic regions containing the apterous, even-skipped, fushi tarazu, twist, and Rhodopsin 1, 2, 3, and 4 genes in four diverse Drosophila species: D. erecta, D. pseudoobscura, D. willistoni, and D. virilis. These autosomal regions contain 81 known or predicted genes in D. melanogaster. However, orthologous sequences from all of these genes were not obtained from every species because of either incomplete overlap of fosmid clones or genomic rearrangements between species. Based on microarray data, we were able to classify 50 of the genes as male biased, female biased, or unbiased in their expression and compare evolutionary rates among the three classes (fig. 2A). Using all available sequences and assuming a constant dN/dS over all branches of the phylogenetic tree, we calculated average dN/dS values of 0.11, 0.02, and 0.07 for male-biased, female-biased, and unbiased genes, respectively. The difference between male-biased and female-biased genes was significant (Mann-Whitney test, P = 0.04), although comparisons among other classes were not significant (P > 0.05) because of the limited sample size. For 30 genes (eight male biased, four female biased, and 18 unbiased) for which sequences from three or more species were available, we applied a "free-ratio" model (Yang 1998) that allowed dN/dS to vary over all branches of the phylogenetic tree. This model did not provide a significantly better fit to the data for any of the female-biased genes, but it did provide a significantly better fit for four (22%) of the unbiased genes and five (63%) of the male-biased genes. This indicates significant evolutionary rate heterogeneity among lineages, particularly for male-biased genes. If dN/dS is calculated as an average over all branches (scaled by their length), then the average dN/dS values for male-biased, female-biased, and unbiased genes are 0.14, 0.02, and 0.10, respectively. The difference between male-biased and female-biased genes is significant (Mann-Whitney test, P = 0.03), although other comparisons are not significant.
FIG. 2.— Evolutionary rates of sex-biased genes in the apterous, even-skipped, fushi tarazu, twist, and Rhodopsin 1, 2, 3, and 4 genomic regions (Bergman et al. 2002). (A) Mean dN/dS assuming a constant evolutionary rate over all branches of the phylogenetic tree for all available sequences (D. melanogaster, D. erecta, D. pseudoobscura, D. willistoni, and D. virilis) and for pairwise comparisons of D. melanogaster (mel) and D. erecta (ere). (B) Mean dN for pairwise comparisons of D. melanogaster versus D. erecta (ere), D. pseudoobscura (pse), and D. willistoni (wil). The number of genes in each expression class is given above the bar.
We also calculated dN/dS for all pairwise comparisons of species. However, these values were not informative for most comparisons because of saturation of dS. For example, comparisons of D. melanogaster to D. erecta, D. pseudoobscura, and D. willistoni produced average dS values of 0.31, 4.9, 16.0, respectively. The only pairwise comparison that did not show saturation at synonymous sites was between D. melanogaster and D. erecta; this comparison also indicated an increased evolutionary rate in male-biased genes relative to female-biased genes (Mann-Whitney test, P = 0.02 [fig. 2A]). Female-biased genes had a lower average dN/dS than unbiased genes, although this difference was not significant. Because of the saturation of dS, we considered only dN for the other pairwise species comparisons. The average dN values for comparisons of D. melanogaster versus D. erecta, D. pseudoobscura, and D. willistoni are shown in figure 2B (D. virilis was not included because it had only one female-biased gene in common with D. melanogaster). In all cases, male-biased genes had higher nonsynonymous substitution rates than female-biased and unbiased genes, whereas female-biased genes tended to have nonsynonymous substitution rates lower than unbiased genes. However, these differences were not significant because of the limited number of genes in each comparison.
Comparison of Highly Sex-Biased Genes Between D. melanogaster and D. pseudoobscura
To investigate the evolutionary rates of genes showing the most extreme levels of sex-biased expression, we extracted the 50 genes with the highest and lowest male/female expression ratios from both the Parisi et al. (2003) and Ranz et al. (2003) data sets. As a control, the 50 genes showing male/female ratios closest to 1 were extracted from each data set. After removing redundancies, the final list contained 93 male-biased, 92 female-biased, and 99 unbiased genes (table 1). The coding sequences from these genes were used for a Blast search of the recently completed D. pseudoobscura genome. The three sex-bias classes showed significant differences in the number of genes with Blast matches over a wide range of e-value cutoffs (fig. 3). In all cases, male-biased genes showed the least conservation between the two species. For example, using a conservative e-value cutoff of 10–9, 46% of the male-biased genes did not have a significant Blast match. The corresponding numbers for female-biased and unbiased genes were 20% and 4%, respectively. Female-biased genes were less conserved than were unbiased genes for all cutoff values (fig. 3), but even in the most extreme case (e-value of cutoff of 10–6), the difference between female-biased and unbiased genes was not significant (2 = 1.51, P = 0.22).
FIG. 3.— Conservation of sex-biased genes between D. melanogaster and D. pseudoobscura. Coding sequences of D. melanogaster genes showing the strongest male-biased or female-bias in expression were used for a Blast search of the D. pseudoobscura genome. The 50 genes showing male/female expression ratios closest to one in the same data sets were used as unbiased controls. The distribution of genes among expression classes differs significantly from the random expectation (2 test; P < 0.02) for all e-values shown except 10–1 (2 = 2.59; P = 0.27).
Because of the high divergence of sex-biased genes (particularly those with male bias) between D. melanogaster and D. pseudoobscura, we were unable to align open reading frames for many of the genes and, thus, could not quantify divergence in terms of dN or dS. However, we could use the Blast score (Altschul et al. 1999) to get an estimate of combined synonymous and nonsynonymous divergence in exon sequences between the two species. Male-biased genes had significantly lower scores than both female-biased and unbiased genes (table 3), indicating a greater divergence of male-biased genes relative to genes of the other two expression classes. Female-biased genes had slightly higher scores than unbiased genes, although this difference was not significant (table 3). Levels of codon bias in genes of the three classes were inversely related to divergence, with male-biased genes having significantly less codon bias than female-biased and unbiased genes (table 3). Female-biased and unbiased genes were nearly identical in their levels of codon bias (table 3).
Table 3 Sequence Conservation and Levels of Codon Bias for Highly Sex-Biased Genes Compared Between D. melanogaster and D. pseudoobscura
The X/autosome distribution of the highly sex-biased genes was even more extreme than for the D. yakuba cDNA comparisons (see above). One percent (1/93) of the highly male-biased genes were on the X chromosome, whereas 26% (24/92) of the highly female-biased genes were on the X chromosome. However, these differences in chromosomal distribution cannot explain the observed differences in evolutionary rates. If only autosomal genes are considered, male-biased genes still show significantly fewer Blast matches between species (2 = 6.1, P = 0.01) and have significantly lower Blast scores (Mann-Whitney test, P < 0.001) than female-biased genes (not shown).
Polymorphism and Divergence in Sex-Biased Genes
The most plausible explanation for the observed evolutionary rate differences among male-biased, female-biased, and unbiased genes is variation in the strength and/or type of natural selection acting on genes of the three expression classes. One possibility is that male-biased genes are under relaxed selective constraints relative to genes of the other two classes and, thus, accumulate a larger fraction of neutral amino acid replacements between species. Alternatively, male-biased genes could be subject to increased positive selection because of male-male or male-female interactions/conflicts and, thus, accumulate more adaptive amino acid substitutions between species. To distinguish between these two possibilities, we examined DNA sequence polymorphism in 55 D. melanogaster protein-encoding genes that could be classified as male-biased, female-biased, or unbiased in their expression based on the microarray data (table 1). The divergence of these genes between D. melanogaster and D. simulans showed the same pattern observed for the other comparative genomic data sets, with male-biased genes showing the greatest divergence and female-biased genes showing the least (table 4). If the increased dN/dS ratio observed for male-biased genes were the result of relaxed selective constraints, then one would expect male-biased genes to show a corresponding increase in their ratio of nonsynoymous/synonymous polymorphism (N/S) relative to female-biased and unbiased genes. However, the opposite pattern was observed. Male-biased genes had lower average N/S than both female-biased and unbiased genes (table 4). Thus, the polymorphism data do not support a general reduction of selective constraint on male-biased genes.
Table 4 Comparison of Divergence (D. melanogaster Versus D. simulans) and Polymorphism (D. melanogaster) in Sex-Biased Genes
The type of selection affecting a particular gene can be inferred by the MK test (McDonald and Kreitman 1991), which compares ratios of polymorphism to divergence for synonymous and nonsynonymous sites. An excess of nonsynonymous divergence relative to nonsynonymous polymorphism is indicative of positive selection, whereas the opposite is indicative of balancing selection. Six of 13 (46%) male-biased genes showed a significant departure from neutrality by the MK test, including four (31%) with departures consistent with positive selection (table 4). This is a higher fraction than observed for the female-biased or unbiased genes (table 4) and suggests that the increased rate of amino acid replacement observed for male-biased genes may be caused by increased positive selection on these genes.
The above results should be interpreted cautiously for several reasons. First, they are based on published surveys of DNA polymorphism that used different sample sizes and population sampling schemes, including African and non-African samples. Thus, the results may be affected by demographic factors, such as bottlenecks or population subdivision. Second, polymorphism is known to be much more sensitive to chromosomal environment (e.g., local recombination rate) than is divergence (Begun and Aquadro 1992), and because of the limited available data, we were unable to partition genes based on chromosomal location. By considering only the ratio of nonsynonymous to synonymous polymorphism (and not the separate values), we could partially control for the above two factors. However, it is possible that these factors also influence the N/S ratio. Finally, there is likely an ascertainment bias in the polymorphism data present in the literature. Some of the genes may have been surveyed with an a priori expectation of positive (or balancing) selection based on functional or divergence data. In addition, there may be a publication bias towards genes that depart from neutrality rather than those that fit the neutral model. These limitations can be addressed in future studies that use common population samples and select genes based only on expression class without a priori expectations of selection.
Conclusions
We used results from two male versus female competitive cDNA microarray hybridization experiments and four comparative genomic data sets to investigate evolutionary rates of genes with sex-biased expression in Drosophila. The results consistently indicated an accelerated rate of evolution in male-biased genes relative to female-biased and unbiased genes. Furthermore, the available polymorphism data suggested that male-biased genes are more often targets of positive selection than genes of the other two expression classes. Taken together, these observations suggest that the rapid evolution of male-biased genes is driven more by male-male competition than by antagonistic coevolution of male-biased and female-biased genes. However, because our classification of male-biased and female-biased genes was based solely on relative expression levels, it is possible that the female counterparts of sexually antagonistic gene interactions were systematically underrepresented. For example, male-biased genes that influence female reproduction and behavior (such as accessory protein genes [Wolfner 1997]) may be expressed at high levels in male reproductive tissues, whereas their female counterparts may show less sex specificity or may be expressed in nonreproductive tissues. Additional functional studies are needed to determine whether such expression asymmetries are common among genes with sexually antagonistic interactions.
Supplementary Material
A table listing the 55 genes (and their associated references) used for our analysis of D. melanogaster polymorphism is provided as Supplementary Material online linked to this article on the MBE homepage (http://www.mbe.oupjournals.org).
Acknowledgements
We thank T. Domazet-Loso for providing D. yakuba cDNA data and D. Presgraves for providing D. melanogaster polymorphism data. The manuscript was improved thanks to comments from C. Meiklejohn, D. Presgraves, J. Ranz, L. Rose, and two anonymous reviewers. This research was supported by funds from the University of Munich (LMU) and Deutsche Forschungsgemeinschaft grant PA 903/2-1.
References
Aguadé, M. 1998. Different forces drive the evolution of the Acp26Aa and Acp26Ab accessory gland genes in the Drosophila melanogaster species complex. Genetics 150:1079–1089.
———. 1999. Positive selection drives the evolution of the Acp29AB accessory gland protein in Drosophila. Genetics 152:543–551.
Altschul, S. F., L. T. L. Madden, A. A. Sch?ffer, J. Zhang, Z. Zhang, W. Miller, and D. J. Lipman. 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25:3389–3402.
Begun, D. J., and C. F. Aquadro. 1992. Levels of naturally occurring DNA polymorphism correlate with recombination rates in D. melanogaster. Nature 356:519–520.
Bergman, C. M., B. D. Pfeiffer, D. E. Rincon-Limas et al. (14 co-authors). 2002. Assessing the impact of comparative genomic sequence data on the functional annotation of the Drosophila genome. Genome Biol. 3:research0086.1–0086.20.
Betancourt, A. J., and D. C. Presgraves. 2002. Linkage limits the power of natural selection in Drosophila. Proc. Natl. Acad. Sci. USA 99:13616–13620.
Betancourt, A. J., D. C. Presgraves, and W. J. Swanson. 2002. A test for faster X evolution in Drosophila. Mol. Biol. Evol. 19:1816–1819.
Betrán, E., and M. Long. 2003. Dntf-2r, a young Drosophila retroposed gene with specific male expression under positive Darwinian selection. Genetics 164:977–988.
Celniker, S. E., D. A. Wheeler, B. Kronmiller et al. (29 Co-authors). 2002. Finishing a whole-genome shotgun: release 3 of the Drosophila melanogaster euchromatic genome sequence. Genome Biol. 3:research0079. 1–0079.
Civetta, A., and R. S. Singh. 1995. High divergence of reproductive tract proteins and their association with postzygotic reproductive isolation in Drosophila melanogaster and Drosophila virilis group species. J. Mol. Evol. 41:1085–1095.
———. 1999. Broad-sense sexual selection, sex gene pool evolution, and speciation. Genome 42:1033–1041.
Coulthart, M. B., and R. S Singh. 1988. High level of divergence of male-reproductive-tract proteins, between Drosophila melanogaster and its sibling species, D. simulans. Mol. Biol. Evol. 5:182–191.
Darwin, C. 1871. The descent of man, and selection in relation to sex. John Murray, London.
Domazet-Loso, T., and D. Tautz. 2003. An evolutionary analysis of orphan genes in Drosophila. Genome Res. 13:2213–2219.
Eberhard, W. G. 1985. Sexual selection and animal genitalia. Harvard University Press, Cambridge, Mass.
Haldane, J. B. S. 1922. Sex-ratio and unisexual sterility in hybrid animals. J. Genetics 12:101–109.
Hosken, D. J., and P. Stockley. 2004. Sexual selection and genital evolution. Trends Ecol. Evol. 19:87–93.
Ikemura, T. 1981. Correlation between the abundance of Escherichia coli transfer RNAs and the occurrence of the respective codons in its protein genes: a proposal for a synonymous codon choice that is optimal for the E. coli translational system. J. Mol. Biol. 151:389–409.
Jamieson, B. G. M. 1987. The ultrastructure and phylogeny of insect spermatozoa. Cambridge University Press, Cambridge, UK.
Kim, Y. 2004. Effect of strong directional selection on weakly selected mutations at linked sites: implications for synonymous codon usage. Mol. Biol. Evol. 21:286–294.
McDonald, J. H., and M. Kreitman. 1991. Adaptive protein evolution at the Adh locus in Drosophila. Nature 351:652–654.
Meiklejohn, C. M., J. Parsch, J. M. Ranz, and D. L. Hartl. 2003. Rapid evolution of male-biased gene expression in Drosophila. Proc. Natl. Acad. Sci. USA 100:9894–9899.
Nurminsky, D., D. De Aguiar, C. D. Bustamante, and D. L. Hartl. 2001. Chromosomal effects of rapid gene evolution in Drosophila melanogaster. Science 291:128–130.
Nurminsky, D. I., M. V. Nurminskaya, D. De Aguiar, and D. L. Hartl. 1998. Selective sweep of a newly evolved sperm-specific gene in Drosophila. Nature 396:572–575.
Orr, H. A., and A. J. Betancourt. 2001. Haldane's sieve and adaptation from the standing genetic variation. Genetics 157:875–884.
Parisi, M., R. Nuttal, D. Naiman, G. Bouffard, J. Malley, J. Andrews, S. Eastman, and B. Oliver. 2003. Paucity of genes on the Drosophila X chromosome showing male-biased expression. Science 299:697–700.
Parsch, J., C. D. Meiklejohn, and D. L. Hartl. 2001. Patterns of DNA sequence variation suggest the recent action of positive selection in the janus-ocnus region of Drosophila simulans. Genetics 159:647–657.
Parsch, J., C. D. Meiklejohn, E. Hauschteck-Jungen, P. Hunziker, and D. L. Hartl. 2001. Molecular evolution of the ocnus and janus genes in the Drosophila melanogaster species subgroup. Mol. Biol. Evol. 18:801–811.
Presgraves, D. C., and H. A. Orr. 1998. Haldane's rule in taxa lacking a hemizygous X. Science 282:952–954.
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.
Rozas, J., J. C. Sanchez-DelBarrio, X. Messeguer, and R. Rozas. 2003. DnaSP, DNA polymorphism analyses by the coalescent and other methods. Bioinformatics 19:2496–2497.
Rubin, G. M., L. Hong, P. Brokstein, M. Evans-Holm, E. Frise, M. Stapleton, and D. A. Harvey. 2000. A Drosophila complementary DNA resource. Science 287:2222–2224.
Singh, R. S., and R. J. Kulathinal. 2000. Sex gene pool evolution and speciation: a new paradigm. Genes Genet. Syst. 75:119–130.
Swanson, W. J., A. G. Clark, H. M. Waldrip-Dail, M. F. Wolfner, and C. F. Aquadro. 2001. Evolutionary EST analysis identifies rapidly evolving male reproductive proteins in Drosophila. Proc. Natl. Acad. Sci. USA 98:7375–7379.
Swanson, W. J., and V. D. Vacquier. 2002. The rapid evolution of reproductive proteins. Nat. Rev. Genet. 3:137–144.
Tao, Y., S. Chen, D. L. Hartl, and C. C. Laurie. 2003. Genetic dissection of hybrid incompatibilities between Drosophila simulans and D. mauritiana. I. Differential accumulation of hybrid male sterility effects on the X and autosomes. Genetics 164:1383–1397.
Ting, C. T., S. C. Tsaur, M. L. Wu, and C. I. Wu. 1998. A rapidly evolving homeobox at the site of a hybrid sterility gene. Science 282:1501–1504.
Townsend, J. P., and D. L. Hartl. 2002. Bayesian analysis of gene expression levels: statistical quantification of relative mRNA level across multiple strains or treatments. Genome Biol. 3:research0071.1–0071.16.
True, J. R., B. S. Weir, and C. C. Laurie. 1996. A genome-wide survey of hybrid incompatibility factors by the introgression of marked segments of Drosophila mauritiana chromosomes into Drosophila simulans. Genetics 142:819–837.
Tsaur, S. C., C. T. Ting, and C. I. Wu. 1998. Positive selection driving the evolution of a gene of male reproduction, Acp26Aa, of Drosophila: II. Divergence versus polymorphism. Mol. Biol. Evol. 15:1040–1046.
Tsaur, S. C., and C. I. Wu. 1997. Positive selection and the molecular evolution of a gene of male reproduction, Acp26Aa of Drosophila. Mol. Biol. Evol. 14:544–549.
Turelli, M., and H. A. Orr. 1995. The dominance theory of Haldane's rule. Genetics 140:389–402.
Wolfner, M. F. 1997. Tokens of love: functions and regulation of Drosophila male accessory gland products. Insect Biochem. Mol. Biol. 27:179–192.
Wright, F. 1990. The ‘effective number of codons’ used in a gene. Gene 87:23–29.
Wu, C. I., and A. W. Davis. 1993. Evolution of postmating reproductive isolation: the composite nature of Haldane's rule and its genetic bases. Am. Nat. 142:187–212.
Yang, Z. 1997. PAML: a program package for phylogenetic analysis by maximum likelihood. CABIOS 13:555–556.
———. 1998. Likelihood ratio tests for detecting positive selection and application to primate lysozyme evolution. Mol. Biol. Evol. 15:568–573.(Zhi Zhang, Tina M. Hambuc)
E-mail: parsch@zi.biologie.uni-muenchen.de.
Abstract
Studies of morphology, interspecific hybridization, protein/DNA sequences, and levels of gene expression have suggested that sex-related characters (particularly those involved in male reproduction) evolve rapidly relative to non–sex-related characters. Here we report a general comparison of evolutionary rates of sex-biased genes using data from cDNA microarray experiments and comparative genomic studies of Drosophila. Comparisons of nonsynonymous/synonymous substitution rates (dN/dS) between species of the D. melanogaster subgroup revealed that genes with male-biased expression had significantly faster rates of evolution than genes with female-biased or unbiased expression. The difference was caused primarily by a higher dN in the male-biased genes. The same pattern was observed for comparisons among more distantly related species. In comparisons between D. melanogaster and D. pseudoobscura, genes with highly biased male expression were significantly more divergent than genes with highly biased female expression. In many cases, orthologs of D. melanogaster male-biased genes could not be identified in D. pseudoobscura through a Blast search. In contrast to the male-biased genes, there was no clear evidence for accelerated rates of evolution in female-biased genes, and most comparisons indicated a reduced rate of evolution in female-biased genes relative to unbiased genes. Male-biased genes did not show an increased ratio of nonsynonymous/synonymous polymorphism within D. melanogaster, and comparisons of polymorphism/divergence ratios suggest that the rapid evolution of male-biased genes is caused by positive selection.
Key Words: codon bias ? comparative genomics ? gene expression ? sexual selection
Introduction
It has long been known that traits associated with sexual reproduction (particularly those related to male reproductive success) often show greater interspecific divergence than do nonreproductive traits. Darwin (1871) documented the frequent occurrence of extravagant secondary sexual characteristics in species of many taxa (including crustaceans, insects, fish, birds, and mammals) and observed that these traits often show large morphological differences between closely related species. Furthermore, Darwin noted that such traits were, with few exceptions, limited to the male of the species. These observations were explained by the theory of sexual selection, which posited that male reproductive traits evolved in response to male-male competition for mating opportunities and the preferential mating of females to males with "attractive" phenotypes. Morphological evidence suggests that sexual selection also acts on primary sexual characteristics. For example, sperm size and morphology are known to differ greatly among insect species (Jamieson 1987), as is the external morphology of male genitalia (Eberhard 1985; Hosken and Stockley 2004).
Studies of interspecific hybridization have also suggested the rapid evolution of male reproductive characters. Haldane (1922) noted a common pattern regarding the viability and fertility of species hybrids. His observation, known as Haldane's Rule, was that when hybrid offspring of only one sex are either inviable or infertile, it is most often the heterogametic sex. In many taxa, such as mammals and Drosophila, the males are heterogametic (XY), and, thus, hybrid male offspring are more prone to be inviable or sterile. Two major hypotheses have been proposed to explain Haldane's rule. The first hypothesis, known as the "dominance" hypothesis posits that hybrid incompatibilities are often recessive and, thus, are only observed in the sex with hemizygous sex chromosomes (Turelli and Orr 1995). The second hypothesis is known as "faster male evolution" and posits that genes involved in male reproduction evolve faster than genes involved in female reproduction or genes with nonreproductive function (Wu and Davis 1993). This hypothesis can only completely explain Haldane's rule for taxa in which the males are heterogametic and is expected to apply primarily to hybrid sterility, not inviability. However, the two hypotheses are not mutually exclusive, and it is likely that both faster male evolution and dominance play a role in hybrid breakdown (Presgraves and Orr 1998). Faster male evolution may also explain the overwhelming preponderance of male sterility factors relative to inviability factors that have been identified from Drosophila hybridizations (Wu and Davis 1993; True, Weir, and Laurie 1996; Tao et al. 2003).
Recent studies have indicated that sex-related genes show increased rates of evolution in their protein/DNA sequences, and it has been suggested that sexual selection affects the evolution of a broad range of genes with reproductive functions (see reviews by Civetta and Singh [1999], Singh and Kulathinal [2000], and Swanson and Vacquier [2002]). In Drosophila, the rapid evolution of reproductive proteins is suggested by the relatively large interspecific differences in migration pattern observed for these proteins on two-dimensional electrophoresis gels (Coulthart and Singh 1988; Civetta and Singh 1995) and by the elevated rate of amino acid substitution between D. melanogaster and D. simulans observed for a large number of male-specific accessory gland proteins (Swanson et al. 2001). In addition, a number of male-specific genes showing evidence for rapid evolution caused by positive selection have been identified, including Acp26Aa (Tsaur and Wu 1997; Aguadé 1998; Tsaur, Ting, and Wu 1998), OdsH (Ting et al. 1998), Sdic (Nurminsky et al. 1998; Nurminsky et al. 2001), Dntf-2r (Betrán and Long 2003), and jan-ocn (Parsch et al. 2001; Parsch, Meiklejohn, and Hartl 2001). Recently, observations of faster male evolution have been extended to the level of gene expression (Meiklejohn et al. 2003; Ranz et al. 2003). These studies used competitive cDNA microarray hybridizations to demonstrate that genes with male-biased expression show greater expression differences both within and between species than do either female-biased or unbiased genes.
In this article, we use data from published cDNA microarray experiments to identify Drosophila genes showing either male bias or female bias in their expression. We then investigate the rates of evolution of these genes (using the ratio of nonsynonymous/synonymous substitution rates, dN/dS) among Drosophila species and compare them with a collection of control genes that show no sex bias in their expression. Our results indicate that male-biased genes have a significantly higher rate of evolution than both female-biased and unbiased genes. A similar pattern is observed for the evolution of highly sex-biased genes between D. melanogaster and D. pseudoobscura, with male-biased genes showing significantly greater divergence between these two species than do either female-biased or unbiased genes. Polymorphism and divergence data suggest that these differences are caused by increased positive selection on male-biased genes.
Materials and Methods
Identification of Genes with Sex-Biased Expression
Two independent data sets that compared male-biased and female-biased gene expression in D. melanogaster through competitive microarray hybridizations were used to classify genes as male biased, female biased, or unbiased in their expression (Parisi et al. 2003; Ranz et al. 2003). Although these data sets were generated using similar experimental approaches, a direct comparison of the two is difficult because of methodological differences in microarray platform, gene number, RNA source material, replication scheme, and statistical analysis. For example, Parisi et al. (2003) used arrays of PCR amplicons (averaging 410 bp in length) corresponding to individual exons of approximately 75% of the predicted genes in the D. melanogaster genome, whereas Ranz et al. (2003) used arrays of full-length cDNAs corresponding to approximately 40% of all predicted genes (Rubin et al. 2000). In addition, Parisi et al. (2003) performed dissections to compare gene expression between testes and ovaries, whereas Ranz et al. (2003) performed comparisons of whole flies. Finally, Parisi et al. (2003) used an expression difference of twofold to classify genes as sex biased, whereas Ranz et al. (2003) used a Bayesian approach (Townsend and Hartl 2002) that could detect significant expression differences of less than twofold. To be conservative, we used the twofold cutoff for both data sets. That is, genes with a male/female (or testes/ovaries) ratio greater than 2.0 were considered male biased, genes with a ratio less than 0.5 were considered female biased, and genes with a ratio between 0.5 and 2.0 were considered unbiased (table 1). In rare cases (7% of genes) where the two data sets resulted in different classifications (e.g., greater than twofold male bias in one data set and less than twofold male bias in the other), the gene was considered sex biased. However, exclusion of these genes from our analyses does not affect the results (data not shown). A small number of genes (0.08%) with sex-bias conflict (e.g., male biased in one data set and female biased in the other) were excluded from further analyses. The twofold cutoff for expression bias was chosen as a conventional standard to allow comparison of microarray results generated using different array platforms and experimental designs. Reanalysis of the data using cutoff values ranging from 1.5-fold to threefold did not alter the qualitative pattern or statistical significance of our results (data not shown). Finally, it should be emphasized that all sex-bias classifications are based on expression studies of D. melanogaster. In some cases, the expression bias could be confirmed in D. simulans (Ranz et al. 2003). However, expression levels in more distantly related species, such as D. yakuba or D. pseudoobscura, have not been determined.
Table 1 Summary of Gene Expression and Comparative Genomic Data Sets
Analysis of Comparative Genomic Data Sets
Comparative sequence data for 371 adult cDNAs from D. yakuba (Domazet-Loso and Tautz 2003) were provided by T. Domazet-Loso. Of the 371 cDNAs, 237 could be classified as male biased, female biased, or unbiased based on microarray expression data (table 1). Evolutionary rates of these genes (dN, dS, and dN/dS) were calculated from pairwise comparisons of D. yakuba and D. melanogaster using the codeml program of the PAML software package (Yang 1997). Levels of codon bias were calculated as either effective number of codons (ENC [Wright 1990]) or frequency of optimal codons (Fop [Ikemura 1981]) based on the full-length transcript from D. melanogaster using the codonW program (http://bioweb.pasteur.fr/seqanal/interfaces/codonw.html). In cases where multiple transcripts were predicted for a gene, the longest transcript was used. For calculating Fop, codon frequency data from D. melanogaster were used.
Comparative sequence data for 81 orthologous genes from D. melanogaster, D. erecta, D. pseudoobscura, D. willistoni, and D. virilis (Bergman et al. 2002) were downloaded from the Berkeley Drosophila Genome Project Web site (http://www.fruitfly.org/comparative/index.html). These data come from sequenced fosmid clones (40 kb each) corresponding to the D. melanogaster genomic regions containing the apterous, even-skipped, fushi tarazu, twist, and Rhodopsin 1, 2, 3, and 4 genes. Of the 81 genes, 50 could be classified as male biased, female biased, or unbiased based on the microarray data (table 1). For these genes, values of dN, dS, and dN/dS were calculated for all pairwise comparisons of species using PAML (codeml runmode–2). For genes with sequences available from three or more species, dN/dS was also calculated for each gene using all available sequences and assuming a constant dN/dS over all branches of the phylogenetic tree (codeml runmode 0, model 0). In addition, we applied the "free-ratio" model (Yang 1998) that allows dN/dS to vary over all branches of the tree (codeml runmode 0, model 1) and compared the likelihood ratio of the two models using a 2 test with the degrees of freedom equal to the difference in parameter number (i.e., the number of branches in the tree minus 1). For these analyses, the phylogenetic relationship of Drosophila species given in Bergman et al. (2002) was used.
Comparison of Highly Sex-Biased Genes Between Drosophila Genomes
To investigate the evolutionary rates of genes showing the strongest sex bias in expression, we selected the 50 genes with the highest and lowest male/female (or testes/ovaries) ratios from both the Parisi et al. (2003) and Ranz et al. (2003) data sets. As a control, we selected the 50 genes showing a male/female ratio closest to 1 from each data set. Only genes corresponding to predicted transcripts in the D. melanogaster genome release 3.0 (Celniker et al. 2002) were included. Of the 100 genes selected for each expression class, 93 male-biased, 92 female-biased, and 99 unbiased genes remained after removing redundancies (table 1). The paucity of overlapping genes between the two data sets was primarily the result of differences in array composition; that is, genes present in one data set but absent in the other (50% of the genes). For another 47% of the genes, the difference was only in the level of sex bias, that is, the gene was in the top 50 of its expression class in one data set but not the other. For 3% of the genes, there was a conflict such that a gene was sex biased in one data set but unbiased in the other. There were no cases of conflicting male/female sex-bias classification among these genes. The coding sequences of these genes from D. melanogaster were used for a BlastN version 2.0 (Altschul et al. 1999) search of the D. pseudoobscura genome sequence using the Baylor College of Medicine Drosophila Genome Project Web site (http://www.hgsc.bcm.tmc.edu/projects/Drosophila/). Because the coding sequences of many of the genes could not be aligned between the two species, we report divergence as either Blast e-values or Blast scores. For score calculation, the default values of –5 and –2 were used for the gap creation and gap extension penalties, respectively. Levels of codon bias were calculated as described above using the full-length coding sequences from D. melanogaster.
Comparison of Polymorphism and Divergence
Because the largest number of DNA sequence polymorphism surveys have been conducted in D. melanogaster, we chose this species to investigate levels of polymorphism in sex-biased genes. We began with a database of 101 protein-encoding genes extracted from the literature and GenBank (provided by D. Presgraves) for which multiple (at least six) D. melanogaster alleles had been sequenced and at least one D. simulans allele was available for divergence analysis. A total of 55 genes could be classified as male biased, female biased, or unbiased based on the microarray expression data (table 1). A list of the individual genes and their associated references is provided as Supplementary Material online. One gene (Dntf-2r) was classified as unbiased based on the Parisi et al. (2003) data set, although it had been shown to be testis-specific by RT-PCR (Betrán and Long 2003). In this case, we classified the gene as male biased. The conflicting microarray result may be caused by cross-hybridization between Dntf-2r and its close paralog Dntf-2, which is expressed in both sexes (Betrán and Long 2003). DnaSP version 4 (Rozas et al. 2003) was used to calculate levels of polymorphism and divergence and to perform the MK test (McDonald and Kreitman 1991). All available D. melanogaster sequences were used for polymorphism calculations, and a single D. simulans allele was used for divergence.
Results and Discussion
Comparison of D. yakuba cDNAs to D. melanogaster
In an analysis of orphan gene evolution, Domazet-Loso and Tautz (2003) cloned and sequenced cDNAs from D. yakuba and estimated rates of evolution (as dN, dS, and dN/dS) by comparing the cDNA sequences to the D. melanogaster genome sequence (Celniker et al. 2002). We used the microarray results of Parisi et al. (2003) and Ranz et al. (2003) to classify the adult cDNAs analyzed by Domazet-Loso and Tautz (2003) as male biased, female biased, or unbiased, and then compared evolutionary rates among genes of the three expression classes (table 2). These comparisons clearly indicated an increased rate of evolution in male-biased genes relative to female-biased genes. The average value of dN/dS was threefold higher for male-biased genes than for female-biased genes, and that of dN was over fivefold higher. Male-biased genes also showed faster evolutionary rates than unbiased genes, with average values of dN/dS and dN both differing by a factor of 2 between the two expression classes. In contrast, female-biased genes had lower values of dN/dS and dN than did unbiased genes. The differences in dN/dS and dN were significant for all comparisons and were highly significant for comparisons between male-biased and female-biased genes (table 2).
Table 2 Evolutionary Rates and Levels of Codon Bias for Genes with Sex-Biased Expression Compared Between D. melanogaster and D. yakuba
We also observed significant differences in synonymous substitution rates among the three classes of genes. Male-biased genes had significantly higher dS than did unbiased genes, whereas female-biased genes had significantly lower dS than did unbiased genes (table 2). To investigate whether or not these differences could be explained by differing selective constraints on synonymous codon usage among genes of the three expression classes, we determined levels of codon bias for all genes by two measures (table 2; see Materials and Methods). For both ENC (where lower values indicate greater bias) and Fop (where higher values indicate greater bias), female-biased genes showed the greatest codon usage bias, whereas male-biased genes showed the least. The differences in codon bias were significant for all comparisons (table 2) and indicated an inverse relationship between level of codon bias and dS for the three classes of genes. Furthermore, there was a significant negative correlation between Fop and dS within the female-biased genes (fig. 1A), suggesting that selection for optimal codon usage might be responsible for the reduced synonymous substitution rate observed for this class of genes. It is also possible that the fixation of strongly selected amino acid replacements results in the fixation of linked, weakly deleterious, nonsynonymous substitutions (Betancourt and Presgraves 2002; Kim 2004). Consistent with this, we observed a significant negative correlation between Fop and dn within the female-biased genes (fig. 1B) and the unbiased genes (not shown). There was also a negative correlation between Fop and dn within the male-biased genes, although this correlation was not significant because of the small sample size and the absence of male-biased genes with high levels of codon bias.
FIG. 1.— Correlation between substitution rates and levels of codon bias for 78 female-biased genes compared between D. melanogaster and D. yakuba. (A) Plot of Fopversus dS (Spearman rank correlation test, R = –0.31, P < 0.01). (B) Plot of Fop versus dN (R = –0.36, P < 0.01). Fop (frequency of optimal codons; Ikemura 1981) was calculated based on D. melanogaster codon usage.
Previous studies in Drosophila demonstrated a significant paucity of male-biased genes and a significant excess of female-biased genes on the X chromosome (Swanson et al. 2001; Parisi et al. 2003; Ranz et al. 2003). In our comparison, 6% (2/33) of the male-biased genes were on the X chromosome, whereas 15% (12/78) of the female-biased genes were on the X chromosome. Thus, a general tendency for slower evolution of X chromosomal genes (Orr and Betancourt 2001) could potentially explain our observations. Consistent with previous observations (Betancourt, Presgraves, and Swanson 2002), we observed slightly lower (although not significantly so) evolutionary rates for X-linked genes (not shown). However, this pattern cannot explain the observed evolutionary rate differences between male-biased and female-biased genes. If we consider only autosomal genes, the average values of dN/dS, dN, and dS, for male-biased genes are 0.13, 0.05, and 0.35, whereas the corresponding values for female-biased genes are 0.05, 0.01, and 0.20. All of these differences are significant (Mann-Whitney test, P < 0.001). The small number of X-linked male-biased genes (two) precluded testing for evolutionary rate differences between male-biased and female-biased genes located on the X chromosome.
Comparison of Orthologous Genomic Regions Among Drosophila Species
To assess the impact of comparative sequence data on genome annotation, Bergman et al. (2002) sequenced fosmid clones (40 kb each) corresponding to the D. melanogaster genomic regions containing the apterous, even-skipped, fushi tarazu, twist, and Rhodopsin 1, 2, 3, and 4 genes in four diverse Drosophila species: D. erecta, D. pseudoobscura, D. willistoni, and D. virilis. These autosomal regions contain 81 known or predicted genes in D. melanogaster. However, orthologous sequences from all of these genes were not obtained from every species because of either incomplete overlap of fosmid clones or genomic rearrangements between species. Based on microarray data, we were able to classify 50 of the genes as male biased, female biased, or unbiased in their expression and compare evolutionary rates among the three classes (fig. 2A). Using all available sequences and assuming a constant dN/dS over all branches of the phylogenetic tree, we calculated average dN/dS values of 0.11, 0.02, and 0.07 for male-biased, female-biased, and unbiased genes, respectively. The difference between male-biased and female-biased genes was significant (Mann-Whitney test, P = 0.04), although comparisons among other classes were not significant (P > 0.05) because of the limited sample size. For 30 genes (eight male biased, four female biased, and 18 unbiased) for which sequences from three or more species were available, we applied a "free-ratio" model (Yang 1998) that allowed dN/dS to vary over all branches of the phylogenetic tree. This model did not provide a significantly better fit to the data for any of the female-biased genes, but it did provide a significantly better fit for four (22%) of the unbiased genes and five (63%) of the male-biased genes. This indicates significant evolutionary rate heterogeneity among lineages, particularly for male-biased genes. If dN/dS is calculated as an average over all branches (scaled by their length), then the average dN/dS values for male-biased, female-biased, and unbiased genes are 0.14, 0.02, and 0.10, respectively. The difference between male-biased and female-biased genes is significant (Mann-Whitney test, P = 0.03), although other comparisons are not significant.
FIG. 2.— Evolutionary rates of sex-biased genes in the apterous, even-skipped, fushi tarazu, twist, and Rhodopsin 1, 2, 3, and 4 genomic regions (Bergman et al. 2002). (A) Mean dN/dS assuming a constant evolutionary rate over all branches of the phylogenetic tree for all available sequences (D. melanogaster, D. erecta, D. pseudoobscura, D. willistoni, and D. virilis) and for pairwise comparisons of D. melanogaster (mel) and D. erecta (ere). (B) Mean dN for pairwise comparisons of D. melanogaster versus D. erecta (ere), D. pseudoobscura (pse), and D. willistoni (wil). The number of genes in each expression class is given above the bar.
We also calculated dN/dS for all pairwise comparisons of species. However, these values were not informative for most comparisons because of saturation of dS. For example, comparisons of D. melanogaster to D. erecta, D. pseudoobscura, and D. willistoni produced average dS values of 0.31, 4.9, 16.0, respectively. The only pairwise comparison that did not show saturation at synonymous sites was between D. melanogaster and D. erecta; this comparison also indicated an increased evolutionary rate in male-biased genes relative to female-biased genes (Mann-Whitney test, P = 0.02 [fig. 2A]). Female-biased genes had a lower average dN/dS than unbiased genes, although this difference was not significant. Because of the saturation of dS, we considered only dN for the other pairwise species comparisons. The average dN values for comparisons of D. melanogaster versus D. erecta, D. pseudoobscura, and D. willistoni are shown in figure 2B (D. virilis was not included because it had only one female-biased gene in common with D. melanogaster). In all cases, male-biased genes had higher nonsynonymous substitution rates than female-biased and unbiased genes, whereas female-biased genes tended to have nonsynonymous substitution rates lower than unbiased genes. However, these differences were not significant because of the limited number of genes in each comparison.
Comparison of Highly Sex-Biased Genes Between D. melanogaster and D. pseudoobscura
To investigate the evolutionary rates of genes showing the most extreme levels of sex-biased expression, we extracted the 50 genes with the highest and lowest male/female expression ratios from both the Parisi et al. (2003) and Ranz et al. (2003) data sets. As a control, the 50 genes showing male/female ratios closest to 1 were extracted from each data set. After removing redundancies, the final list contained 93 male-biased, 92 female-biased, and 99 unbiased genes (table 1). The coding sequences from these genes were used for a Blast search of the recently completed D. pseudoobscura genome. The three sex-bias classes showed significant differences in the number of genes with Blast matches over a wide range of e-value cutoffs (fig. 3). In all cases, male-biased genes showed the least conservation between the two species. For example, using a conservative e-value cutoff of 10–9, 46% of the male-biased genes did not have a significant Blast match. The corresponding numbers for female-biased and unbiased genes were 20% and 4%, respectively. Female-biased genes were less conserved than were unbiased genes for all cutoff values (fig. 3), but even in the most extreme case (e-value of cutoff of 10–6), the difference between female-biased and unbiased genes was not significant (2 = 1.51, P = 0.22).
FIG. 3.— Conservation of sex-biased genes between D. melanogaster and D. pseudoobscura. Coding sequences of D. melanogaster genes showing the strongest male-biased or female-bias in expression were used for a Blast search of the D. pseudoobscura genome. The 50 genes showing male/female expression ratios closest to one in the same data sets were used as unbiased controls. The distribution of genes among expression classes differs significantly from the random expectation (2 test; P < 0.02) for all e-values shown except 10–1 (2 = 2.59; P = 0.27).
Because of the high divergence of sex-biased genes (particularly those with male bias) between D. melanogaster and D. pseudoobscura, we were unable to align open reading frames for many of the genes and, thus, could not quantify divergence in terms of dN or dS. However, we could use the Blast score (Altschul et al. 1999) to get an estimate of combined synonymous and nonsynonymous divergence in exon sequences between the two species. Male-biased genes had significantly lower scores than both female-biased and unbiased genes (table 3), indicating a greater divergence of male-biased genes relative to genes of the other two expression classes. Female-biased genes had slightly higher scores than unbiased genes, although this difference was not significant (table 3). Levels of codon bias in genes of the three classes were inversely related to divergence, with male-biased genes having significantly less codon bias than female-biased and unbiased genes (table 3). Female-biased and unbiased genes were nearly identical in their levels of codon bias (table 3).
Table 3 Sequence Conservation and Levels of Codon Bias for Highly Sex-Biased Genes Compared Between D. melanogaster and D. pseudoobscura
The X/autosome distribution of the highly sex-biased genes was even more extreme than for the D. yakuba cDNA comparisons (see above). One percent (1/93) of the highly male-biased genes were on the X chromosome, whereas 26% (24/92) of the highly female-biased genes were on the X chromosome. However, these differences in chromosomal distribution cannot explain the observed differences in evolutionary rates. If only autosomal genes are considered, male-biased genes still show significantly fewer Blast matches between species (2 = 6.1, P = 0.01) and have significantly lower Blast scores (Mann-Whitney test, P < 0.001) than female-biased genes (not shown).
Polymorphism and Divergence in Sex-Biased Genes
The most plausible explanation for the observed evolutionary rate differences among male-biased, female-biased, and unbiased genes is variation in the strength and/or type of natural selection acting on genes of the three expression classes. One possibility is that male-biased genes are under relaxed selective constraints relative to genes of the other two classes and, thus, accumulate a larger fraction of neutral amino acid replacements between species. Alternatively, male-biased genes could be subject to increased positive selection because of male-male or male-female interactions/conflicts and, thus, accumulate more adaptive amino acid substitutions between species. To distinguish between these two possibilities, we examined DNA sequence polymorphism in 55 D. melanogaster protein-encoding genes that could be classified as male-biased, female-biased, or unbiased in their expression based on the microarray data (table 1). The divergence of these genes between D. melanogaster and D. simulans showed the same pattern observed for the other comparative genomic data sets, with male-biased genes showing the greatest divergence and female-biased genes showing the least (table 4). If the increased dN/dS ratio observed for male-biased genes were the result of relaxed selective constraints, then one would expect male-biased genes to show a corresponding increase in their ratio of nonsynoymous/synonymous polymorphism (N/S) relative to female-biased and unbiased genes. However, the opposite pattern was observed. Male-biased genes had lower average N/S than both female-biased and unbiased genes (table 4). Thus, the polymorphism data do not support a general reduction of selective constraint on male-biased genes.
Table 4 Comparison of Divergence (D. melanogaster Versus D. simulans) and Polymorphism (D. melanogaster) in Sex-Biased Genes
The type of selection affecting a particular gene can be inferred by the MK test (McDonald and Kreitman 1991), which compares ratios of polymorphism to divergence for synonymous and nonsynonymous sites. An excess of nonsynonymous divergence relative to nonsynonymous polymorphism is indicative of positive selection, whereas the opposite is indicative of balancing selection. Six of 13 (46%) male-biased genes showed a significant departure from neutrality by the MK test, including four (31%) with departures consistent with positive selection (table 4). This is a higher fraction than observed for the female-biased or unbiased genes (table 4) and suggests that the increased rate of amino acid replacement observed for male-biased genes may be caused by increased positive selection on these genes.
The above results should be interpreted cautiously for several reasons. First, they are based on published surveys of DNA polymorphism that used different sample sizes and population sampling schemes, including African and non-African samples. Thus, the results may be affected by demographic factors, such as bottlenecks or population subdivision. Second, polymorphism is known to be much more sensitive to chromosomal environment (e.g., local recombination rate) than is divergence (Begun and Aquadro 1992), and because of the limited available data, we were unable to partition genes based on chromosomal location. By considering only the ratio of nonsynonymous to synonymous polymorphism (and not the separate values), we could partially control for the above two factors. However, it is possible that these factors also influence the N/S ratio. Finally, there is likely an ascertainment bias in the polymorphism data present in the literature. Some of the genes may have been surveyed with an a priori expectation of positive (or balancing) selection based on functional or divergence data. In addition, there may be a publication bias towards genes that depart from neutrality rather than those that fit the neutral model. These limitations can be addressed in future studies that use common population samples and select genes based only on expression class without a priori expectations of selection.
Conclusions
We used results from two male versus female competitive cDNA microarray hybridization experiments and four comparative genomic data sets to investigate evolutionary rates of genes with sex-biased expression in Drosophila. The results consistently indicated an accelerated rate of evolution in male-biased genes relative to female-biased and unbiased genes. Furthermore, the available polymorphism data suggested that male-biased genes are more often targets of positive selection than genes of the other two expression classes. Taken together, these observations suggest that the rapid evolution of male-biased genes is driven more by male-male competition than by antagonistic coevolution of male-biased and female-biased genes. However, because our classification of male-biased and female-biased genes was based solely on relative expression levels, it is possible that the female counterparts of sexually antagonistic gene interactions were systematically underrepresented. For example, male-biased genes that influence female reproduction and behavior (such as accessory protein genes [Wolfner 1997]) may be expressed at high levels in male reproductive tissues, whereas their female counterparts may show less sex specificity or may be expressed in nonreproductive tissues. Additional functional studies are needed to determine whether such expression asymmetries are common among genes with sexually antagonistic interactions.
Supplementary Material
A table listing the 55 genes (and their associated references) used for our analysis of D. melanogaster polymorphism is provided as Supplementary Material online linked to this article on the MBE homepage (http://www.mbe.oupjournals.org).
Acknowledgements
We thank T. Domazet-Loso for providing D. yakuba cDNA data and D. Presgraves for providing D. melanogaster polymorphism data. The manuscript was improved thanks to comments from C. Meiklejohn, D. Presgraves, J. Ranz, L. Rose, and two anonymous reviewers. This research was supported by funds from the University of Munich (LMU) and Deutsche Forschungsgemeinschaft grant PA 903/2-1.
References
Aguadé, M. 1998. Different forces drive the evolution of the Acp26Aa and Acp26Ab accessory gland genes in the Drosophila melanogaster species complex. Genetics 150:1079–1089.
———. 1999. Positive selection drives the evolution of the Acp29AB accessory gland protein in Drosophila. Genetics 152:543–551.
Altschul, S. F., L. T. L. Madden, A. A. Sch?ffer, J. Zhang, Z. Zhang, W. Miller, and D. J. Lipman. 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25:3389–3402.
Begun, D. J., and C. F. Aquadro. 1992. Levels of naturally occurring DNA polymorphism correlate with recombination rates in D. melanogaster. Nature 356:519–520.
Bergman, C. M., B. D. Pfeiffer, D. E. Rincon-Limas et al. (14 co-authors). 2002. Assessing the impact of comparative genomic sequence data on the functional annotation of the Drosophila genome. Genome Biol. 3:research0086.1–0086.20.
Betancourt, A. J., and D. C. Presgraves. 2002. Linkage limits the power of natural selection in Drosophila. Proc. Natl. Acad. Sci. USA 99:13616–13620.
Betancourt, A. J., D. C. Presgraves, and W. J. Swanson. 2002. A test for faster X evolution in Drosophila. Mol. Biol. Evol. 19:1816–1819.
Betrán, E., and M. Long. 2003. Dntf-2r, a young Drosophila retroposed gene with specific male expression under positive Darwinian selection. Genetics 164:977–988.
Celniker, S. E., D. A. Wheeler, B. Kronmiller et al. (29 Co-authors). 2002. Finishing a whole-genome shotgun: release 3 of the Drosophila melanogaster euchromatic genome sequence. Genome Biol. 3:research0079. 1–0079.
Civetta, A., and R. S. Singh. 1995. High divergence of reproductive tract proteins and their association with postzygotic reproductive isolation in Drosophila melanogaster and Drosophila virilis group species. J. Mol. Evol. 41:1085–1095.
———. 1999. Broad-sense sexual selection, sex gene pool evolution, and speciation. Genome 42:1033–1041.
Coulthart, M. B., and R. S Singh. 1988. High level of divergence of male-reproductive-tract proteins, between Drosophila melanogaster and its sibling species, D. simulans. Mol. Biol. Evol. 5:182–191.
Darwin, C. 1871. The descent of man, and selection in relation to sex. John Murray, London.
Domazet-Loso, T., and D. Tautz. 2003. An evolutionary analysis of orphan genes in Drosophila. Genome Res. 13:2213–2219.
Eberhard, W. G. 1985. Sexual selection and animal genitalia. Harvard University Press, Cambridge, Mass.
Haldane, J. B. S. 1922. Sex-ratio and unisexual sterility in hybrid animals. J. Genetics 12:101–109.
Hosken, D. J., and P. Stockley. 2004. Sexual selection and genital evolution. Trends Ecol. Evol. 19:87–93.
Ikemura, T. 1981. Correlation between the abundance of Escherichia coli transfer RNAs and the occurrence of the respective codons in its protein genes: a proposal for a synonymous codon choice that is optimal for the E. coli translational system. J. Mol. Biol. 151:389–409.
Jamieson, B. G. M. 1987. The ultrastructure and phylogeny of insect spermatozoa. Cambridge University Press, Cambridge, UK.
Kim, Y. 2004. Effect of strong directional selection on weakly selected mutations at linked sites: implications for synonymous codon usage. Mol. Biol. Evol. 21:286–294.
McDonald, J. H., and M. Kreitman. 1991. Adaptive protein evolution at the Adh locus in Drosophila. Nature 351:652–654.
Meiklejohn, C. M., J. Parsch, J. M. Ranz, and D. L. Hartl. 2003. Rapid evolution of male-biased gene expression in Drosophila. Proc. Natl. Acad. Sci. USA 100:9894–9899.
Nurminsky, D., D. De Aguiar, C. D. Bustamante, and D. L. Hartl. 2001. Chromosomal effects of rapid gene evolution in Drosophila melanogaster. Science 291:128–130.
Nurminsky, D. I., M. V. Nurminskaya, D. De Aguiar, and D. L. Hartl. 1998. Selective sweep of a newly evolved sperm-specific gene in Drosophila. Nature 396:572–575.
Orr, H. A., and A. J. Betancourt. 2001. Haldane's sieve and adaptation from the standing genetic variation. Genetics 157:875–884.
Parisi, M., R. Nuttal, D. Naiman, G. Bouffard, J. Malley, J. Andrews, S. Eastman, and B. Oliver. 2003. Paucity of genes on the Drosophila X chromosome showing male-biased expression. Science 299:697–700.
Parsch, J., C. D. Meiklejohn, and D. L. Hartl. 2001. Patterns of DNA sequence variation suggest the recent action of positive selection in the janus-ocnus region of Drosophila simulans. Genetics 159:647–657.
Parsch, J., C. D. Meiklejohn, E. Hauschteck-Jungen, P. Hunziker, and D. L. Hartl. 2001. Molecular evolution of the ocnus and janus genes in the Drosophila melanogaster species subgroup. Mol. Biol. Evol. 18:801–811.
Presgraves, D. C., and H. A. Orr. 1998. Haldane's rule in taxa lacking a hemizygous X. Science 282:952–954.
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.
Rozas, J., J. C. Sanchez-DelBarrio, X. Messeguer, and R. Rozas. 2003. DnaSP, DNA polymorphism analyses by the coalescent and other methods. Bioinformatics 19:2496–2497.
Rubin, G. M., L. Hong, P. Brokstein, M. Evans-Holm, E. Frise, M. Stapleton, and D. A. Harvey. 2000. A Drosophila complementary DNA resource. Science 287:2222–2224.
Singh, R. S., and R. J. Kulathinal. 2000. Sex gene pool evolution and speciation: a new paradigm. Genes Genet. Syst. 75:119–130.
Swanson, W. J., A. G. Clark, H. M. Waldrip-Dail, M. F. Wolfner, and C. F. Aquadro. 2001. Evolutionary EST analysis identifies rapidly evolving male reproductive proteins in Drosophila. Proc. Natl. Acad. Sci. USA 98:7375–7379.
Swanson, W. J., and V. D. Vacquier. 2002. The rapid evolution of reproductive proteins. Nat. Rev. Genet. 3:137–144.
Tao, Y., S. Chen, D. L. Hartl, and C. C. Laurie. 2003. Genetic dissection of hybrid incompatibilities between Drosophila simulans and D. mauritiana. I. Differential accumulation of hybrid male sterility effects on the X and autosomes. Genetics 164:1383–1397.
Ting, C. T., S. C. Tsaur, M. L. Wu, and C. I. Wu. 1998. A rapidly evolving homeobox at the site of a hybrid sterility gene. Science 282:1501–1504.
Townsend, J. P., and D. L. Hartl. 2002. Bayesian analysis of gene expression levels: statistical quantification of relative mRNA level across multiple strains or treatments. Genome Biol. 3:research0071.1–0071.16.
True, J. R., B. S. Weir, and C. C. Laurie. 1996. A genome-wide survey of hybrid incompatibility factors by the introgression of marked segments of Drosophila mauritiana chromosomes into Drosophila simulans. Genetics 142:819–837.
Tsaur, S. C., C. T. Ting, and C. I. Wu. 1998. Positive selection driving the evolution of a gene of male reproduction, Acp26Aa, of Drosophila: II. Divergence versus polymorphism. Mol. Biol. Evol. 15:1040–1046.
Tsaur, S. C., and C. I. Wu. 1997. Positive selection and the molecular evolution of a gene of male reproduction, Acp26Aa of Drosophila. Mol. Biol. Evol. 14:544–549.
Turelli, M., and H. A. Orr. 1995. The dominance theory of Haldane's rule. Genetics 140:389–402.
Wolfner, M. F. 1997. Tokens of love: functions and regulation of Drosophila male accessory gland products. Insect Biochem. Mol. Biol. 27:179–192.
Wright, F. 1990. The ‘effective number of codons’ used in a gene. Gene 87:23–29.
Wu, C. I., and A. W. Davis. 1993. Evolution of postmating reproductive isolation: the composite nature of Haldane's rule and its genetic bases. Am. Nat. 142:187–212.
Yang, Z. 1997. PAML: a program package for phylogenetic analysis by maximum likelihood. CABIOS 13:555–556.
———. 1998. Likelihood ratio tests for detecting positive selection and application to primate lysozyme evolution. Mol. Biol. Evol. 15:568–573.(Zhi Zhang, Tina M. Hambuc)