Accessory gland proteins are a major component of Drosophila seminal fluid. These proteins have a variety of functions and may be subject to sexual selection and/or antagonistic evolution between the sexes. Most population genetic data from these proteins are from D. melanogaster and D. simulans. Here, we extend the population genetic analysis of Acp genes to the other simulans complex species, D. mauritiana and D. sechellia. We sequenced population samples of seven Acp's from D. mauritiana, D. sechellia, and D. simulans. We investigated the population genetics of these genes on individual simulans complex lineages and compared Acp polymorphism and divergence to polymorphism and divergence from a set of non-Acp loci in the same species. Polymorphism and divergence data from the simulans complex revealed little evidence for adaptive protein evolution at individual loci. However, we observed a dramatically inflated index of dispersion for amino acid substitutions in the simulans complex at Acp genes, but not at non-Acp genes. This pattern of episodic bursts of protein evolution in Acp's provides the strongest evidence to date that the population genetic mechanisms driving Acp divergence are different from the mechanisms driving evolution at most Drosophila genes.

---

THE evolution of proteins involved in reproduction has attracted considerable interest. One generality emerging from this research is that reproduction-related genes tend to evolve quickly (reviewed in SWANSON and VACQUIER 2002). However, the importance of directional selection in driving this rapid protein evolution in reproduction-related genes is still unclear. In a few proteins, the ratio of nonsynonymous to synonymous substitution (d_N/d_S) is significantly greater than one, suggesting that the rapid evolution is a result of selection (LEE et al. 1995; METZ and PALUMBI 1996; TSAUR and WU 1997; AGUADé 1998, 1999; TING et al. 1998, 2000; WYCKOFF et al. 2000). In most cases, however, there is little support from d_N/d_S ratios that adaptive protein divergence is a general property of genes functioning in reproduction. Similarly, molecular population genetic analysis suggests the action of directional selection in some reproduction-related genes (AGUADé et al. 1992; AGUADé 1998, 1999; NURMINSKY et al. 1998; TSAUR et al. 1998, 2001; BEGUN et al. 2000), but for most such genes there is no compelling evidence for adaptive protein divergence.

One particular group of reproduction-related proteins, the male accessory gland proteins (Acp's) of Drosophila, has been the object of much research. Acp's, a major component of seminal fluid, have several demonstrated effects on female physiology and sperm use (CLARK et al. 1995; HERNDON and WOLFNER 1995; NEUBAUM and WOLFNER 1999; TRAM and WOLFNER 1999; reviewed in CHEN 1996; WOLFNER 1997). Natural variation of Acp's was first investigated by protein electrophoresis, which revealed that accessory gland proteins are more polymorphic and evolve faster than proteins from many other nonreproductive male tissues (COULTHART and SINGH 1988; THOMAS and SINGH 1992; CIVETTA and SINGH 1995). These inferences from protein gels were subsequently supported by DNA sequence analyses, which revealed that Acp's are much more polymorphic and diverge more quickly than "typical" proteins in the Drosophila melanogaster vs. D. simulans comparison (BEGUN et al. 2000; SWANSON et al. 2001). Data from Acp26Aa, Acp29AB, and Acp36DE provided some evidence for adaptive protein divergence (AGUADé et al. 1992; TSAUR and WU 1997; AGUADé 1998, 1999; TSAUR et al. 1998, 2001; BEGUN et al. 2000). Data from most Acp's, however, were not suggestive of adaptive protein evolution in D. melanogaster and D. simulans (BEGUN et al. 2000).

Despite the relatively detailed study of Acp's in D. melanogaster and D. simulans, little attention has been paid to Acp variation in the close relatives of these species, D. mauritiana and D. sechellia. Studies of Acp's in these species allow us to address several key population genetic issues. Are patterns of polymorphism and divergence in Acp's different from patterns in "non-Acp" genes in all melanogaster subgroup species? Are patterns of substitution in Acp's similar across individual simulans complex lineages? Does analysis of Acp's along individual lineages provide evidence for adaptive protein evolution? Do some lineages experience more directional selection than others? Here, we address these questions through an analysis of variation in seven Acp's in D. mauritiana, D. sechellia, and D. simulans.

Stocks/loci:

New D. simulans sequences reported here were collected from inbred lines derived from flies collected in the Wolfskill Orchard in Winters, California (BEGUN and WHITLEY 2000). D. sechellia sequences were from lines obtained from the species stock center and J. Coyne. Line sech01, also known as "Robertson," was collected in 1980 on Cousin Island by L. Tsacas. Line sech77 (a.k.a. s77 25x) was collected by K. Kimura on Praslin Island in 1987. Lines sech33, sech34, and sech35 were collected by J. W. O. Ballard on Mahe in 1998. All remaining lines were collected by J. R. David in 1985 on Cousin island. D. mauritiana sequences were from lines kindly provided by J. Coyne and D. Barbash. The D. mauritiana lines were originally collected in Mauritius in 1981 by O. Kitagawa. All lines were originally established from single, inseminated females and show very low levels of residual heterozygosity. In all cases, direct sequencing of PCR products in both directions was determined on an ABI 3700 sequencer. Raw data were analyzed using phred (EWING et al. 1998) and phrap (EWING and GREEN 1998). Sequences were then examined by eye using Consed (GORDON et al. 1998). Polymorphic sites in multiple alignments were verified using the MACE package of scripts (B. GILLILAND, unpublished results; http://ludwig.ucdavis.edu/MACE/). New sequences reported in this article can be found in GenBank under accession nos. AY505178, AY505293.

Several analyses used previously published sequences. For Acp26Aa in D. mauritiana, 22 published sequences were used (TSAUR et al. 2001). Previously published D. melanogaster and D. simulans Acp sequences used in this analysis, with the exception of D. simulans Acp26Aa, Acp26Ab, and Acp62F (which were collected for this paper), were from BEGUN et al. (2000) and references therein. Comparisons between Acp's and non-Acp's in D. sechellia and D. mauritiana were made using sequences from HEY and KLIMAN (1993), KLIMAN et al. (2000), PARSCH et al. (2001), and references therein, as well as new sequence data from Relish (AY505188–AY505192; see supplementary table at http://www.genetics.org/supplemental/ for polymorphism data). This non-Acp data set is composed of 11 loci, Adh, ase, ci, est-6, janA, janB, per, Rel, yp2, z, and Zw, surveyed for variation in simulans complex species. For polymorphism comparisons across species, ase and ci were excluded as they are from regions of low recombination in D. melanogaster. Similarly, janA was omitted for comparisons of replacement and silent variation across species due to a lack of polymorphism data in D. sechellia. All analyses were performed on gene regions for which homologous sequence was available from the simulans complex species as well as for D. melanogaster. Note that the polymorphism statistics reported in BEGUN et al. (2000) for D. simulans Acp62F disagree with those reported here, which are from a separate population sample. We use the new data in this article. Additionally our analysis of silent and replacement mutations included sites for which a base could not be called in every sequence sampled, leading to some small differences in mutation counts between the present report and BEGUN et al. (2000).

Expression analysis:

We used reverse transcription-PCR (RT-PCR) on cDNA from whole adult males and whole adult females to validate that putative Acp's were male limited in expression in the simulans complex. Gene-specific primers were used in all cases (sequences available from the authors). Poly(A)+ RNA was prepared from whole flies using a MicroPoly(A) kit (Ambion, Austin, TX). cDNA for reverse transcriptase-PCR and rapid amplification of cDNA ends (RACE) was prepared from this RNA using the SMART RACE cDNA amplification kit (CLONTECH, Palo Alto, CA). SuperScript II reverse transcriptase (GIBCO BRL, Rockville, MD) was used for all RT reactions. In all cases we found that simulans complex homologs of D. melanogaster Acp's showed male-limited expression. This demonstrates that genes that are Acp's in D. melanogaster show similar patterns of expression in simulans clade species.

Sequence analysis:

Most population genetic analyses were performed using software developed by the authors. Source code in Smalltalk is available upon request. Throughout the article, "silent sites" refers to synonymous coding positions in exons. Polarized mutations are those changes that could be unambiguously assigned to individual lineages of the simulans complex under parsimony (using D. melanogaster as an outgroup). Divergence estimates were calculated using the maximum-likelihood method of GOLDMAN and YANG (1994) as implemented in PAML (YANG 1997). For maximum-likelihood (ML) divergence estimation we allowed equilibrium codon frequencies to be free parameters of the model. In addition we estimated the transition/transversion ratio for each locus. Estimated substitution rates along individual lineages of the simulans complex and indices of dispersion [R(t)] for individual loci were calculated in a manner adapted from GILLESPIE (1989). GILLESPIE's (1989) analysis uses sequence comparisons of one allele from each of three species for each of several loci and then corrects for lineage effects by weighting each lineage by the mean divergence along that lineage. Explicitly, let N_i, i = 1, 2, 3, be the number of substitutions at a particular locus on the ith lineage. These N_i can be estimated using estimates of pairwise divergence between all possible pairs of species such that

where D_ij is the estimate of divergence between species i and j. The N_i represent random variables drawn from at most three different distributions. As such, the moments of N_i can be written as

Here, w_i is the weighting factor that will be shared by all loci on a particular lineage. µ and ² are the locus-specific contributions to the mean and variance in the number of substitutions. The w_i are defined such that

This correction adjusts the mean number of substitutions along a lineage without changing the average rate of evolution at a particular locus. In practice, lineage-specific weights are calculated by choosing values such that the mean number of substitutions per 100 sites is equal among lineages. We are now in a position to write down the index of dispersion in a manner that is free of lineage effects:

Lineage effects may be removed by dividing the number of substitutions along the ith lineage by the weight w_i. Thus the mean number of substitutions for a particular locus will be estimated as

The expectation of this estimator is then

which is an unbiased estimator of µ. The variance may be estimated in a straightforward manner by

This estimator of variance is also unbiased. From the estimators of the mean and variance in the numbers of substitutions one can write an estimator for R(t),

As we had polymorphism data for multiple loci, we developed a computational procedure to use this information in our analyses of R(t). Briefly, we iterated Gillespie's routine 1000 times, where each iteration consisted of choosing one allele at random from each species at every locus examined, estimating divergences, calculating lineage-specific weights, and then estimating R(t) for each locus. From each iteration we recorded the values of R(t), thereby generating an empirical distribution of the test statistic R(t) given different configurations of the data. These distributions allow us to make statistical inferences about the index of dispersion at individual loci and across classes of loci. Zeste was omitted from this analysis because there were no interspecific differences. Tests for deviations of R(t) from neutral, equilibrium expectations were determined through simulations of the substitution process. We assumed a Poisson molecular clock and conditioned on the mean number of estimated substitutions at a locus across our iterations. For each locus, 10⁶ independent replicates of this procedure were performed. Negative branch lengths were set to zero for subsequent analyses.

Contingency table analyses for McDonald-Kreitman (MK; MCDONALD and KREITMAN 1991) tests were done using 10⁵ replicates of a Monte Carlo procedure (ENGELS 1988) because numerous cells contained small values or values of zero. A resampling procedure to detect small-scale hitchhiking effects associated with polarized fixations was implemented according to KERN et al. [2002; Kern-Jones-Begun (KJB) test].

Simulation of sperm competition:

CLARK (2002) proposed a scramble competition model of sperm displacement. We used computer simulations of this model to shed light on patterns of molecular evolution that might be expected for genes functioning in sperm competition. In this haploid model, the fitness of an allele is defined by its ability to displace other genotypes after competitively mating. Consider two alleles, x_i and x_j, which mate in succession such that x_j mates first, followed by x_i. Let s_ij represent the proportion of offspring sired by x_i (also known as the P₂ score of x_i), where s_ij = ¹/₂ for (i = j). Assuming random mating, the frequency of x_i after a generation of selection is

The distribution of s_ij is an important parameter because distributions with greater variances will produce weaker frequency dependence. Fortunately several experiments provide some information about the variation in P₂ scores (s_ij values) in D. melanogaster (CLARK et al. 1995). P₂ has a modal value near 1.0 and a tail of smaller values that disappears at 0.5 in laboratory experiments. We simulated Clark's model using three different distributions of s_ij values: a Uniform (0, 1), which is probably inappropriate, a Uniform (0.5, 1), which is slightly closer to empirical data, and a Beta (5, 1), which is likely to be the most biologically relevant distribution examined here. These simulations used an infinite-site, no-recombination representation of a single gene (WATTERSON 1975). Allele frequencies change each generation according to the effects of selection, drift, and mutation, in that order. The procedure for specifying allelic fitness means that many, but certainly not all, new mutations will have different fitness values from that of their parent allele. Further, allelic fitness can change purely as a function of the other alleles segregating in the population at any point during the simulation. Thus, the mutation parameter reflects both mutation rates to neutral and selected alleles. We employed a simple allelic genealogy (TAKAHATA 1990) to maintain a tree of alleles in the population at any time. This structure keeps track of the relationships among alleles, the generation in which individual alleles arose by mutation, and, for alleles that ultimately fix, their fixation times. Thus, the allelic genealogy provides statistics of polymorphism and divergence.

Polymorphism:

Table 1 gives summary statistics of DNA polymorphism from seven Acp's from D. sechellia, D. simulans, and D. mauritiana. Acp's harbor more replacement polymorphism than non-Acp's in all species. In D. mauritiana, average silent heterozygosity is not significantly greater for Acp's than for non-Acp's. However, D. mauritiana Acp's are significantly more polymorphic at replacement sites than are non-Acp's (Mann-Whitney U-test: P = 0.0017 for _w; P = 0.0013 for ). Surprisingly, one allele in our population sample from D. mauritiana (mau205) contained a premature stop codon at Acp29. This premature termination codon was a change from a lysine at residue 110 (base 325 in our alignment) of our sampled region, which is about two-thirds of the length of the full Acp29 protein and was supported by a phred quality score of 90 (i.e., the probability this call is in error is 10^–9). No other polymorphisms occurred on this allele downstream of this premature termination codon. D. sechellia Acp replacement polymorphism is almost eightfold higher than non-Acp's replacement polymorphism, although this difference is not statistically significant (Mann-Whitney U-test: P = 0.0851 for _w; P = 0.0851 for ). Finally, as previously reported (BEGUN et al. 2000), D. simulans Acp replacement polymorphism is much greater than replacement polymorphism at non-Acp's (Mann-Whitney U-test: P = 0.0013 for _w; P = 0.0013 for ).

View this table: In this window In a new window   TABLE 1 Summary statistics of DNA polymorphism for seven Acp's from D. mauritiana, D. sechellia, and D. simulans

 

Divergence:

Consistent with previous observations (BEGUN et al. 2000; SWANSON et al. 2001), Acp's show high replacement substitution rates. Table 1 presents maximum-likelihood estimates (GOLDMAN and YANG 1994) of silent and replacement divergence of D. simulans, D. mauritiana, and D. sechellia vs. D. melanogaster. Silent divergence between D. melanogaster and each of the three simulans complex species is not significantly different for Acp's than for non-Acp's. However, rates of amino acid evolution are significantly greater from D. melanogaster to D. mauritiana, D. simulans, and D. sechellia (D. mauritiana, Mann-Whitney U-test, P = 0.0049; D. simulans, Mann-Whitney U-test, P = 0.0083; D. sechellia, Mann-Whitney U-test, P = 0.011). This effect is approximately threefold for each species.

Divergences of D. mauritiana, D. sechellia, and D. simulans vs. D. melanogaster are nonindependent as a result of the shared evolutionary history of these species. We used unrooted three-taxon trees to investigate substitution rates along recently separated, individual simulans-complex lineages (Table 2) . Replacement divergence since the speciation event(s) is significantly greater at Acp's than at non-Acp's for all lineages (D. mauritiana, Mann-Whitney U-test, P = 0.0063; D. simulans, Mann-Whitney U-test, P = 0.0027; D. sechellia, Mann-Whitney U-test, P = 0.0192). Silent divergence, however, is comparable between Acp's and non-Acp's. Thus, rapid rates of Acp protein evolution are characteristic of both the more recent and more ancient histories of the melanogaster subgroup.

View this table: In this window In a new window   TABLE 2 Lineage-specific divergence for Acp genes and non-Acp genes from D. mauritiana, D. sechellia, and D. simulans

 
In addition to characterizing mean replacement divergence among Acp's and non-Acp's in the simulans complex, we estimated the index of dispersion [R(t)] for replacement sites (Table 3) . The mean R(t) across loci, across iterations of our computational procedure (see MATERIALS AND METHODS) for all Acp's is significantly overdispersed [R(t) = 6.50; P < 0.001] as are values of R(t) for several loci individually. Non-Acp's do not reject a Poisson clock [mean R(t) = 2.56]. Additionally estimates of R(t) are significantly greater for Acp's than for non-Acp's using a nonparametric test (Mann-Whitney U-test: P = 0.0096). It is worth noting that the only two non-Acp loci that showed significant overdispersion, janA and janB, are both reproductive proteins.

View this table: In this window In a new window   TABLE 3 Index of dispersion for replacement divergence among the simulans complex species

 
A potential caveat associated with these R(t) analyses is that the simulans complex species have only recently diverged and thus may segregate ancestral polymorphism (e.g., KLIMAN et al. 2000). For example, although sequence comparisons of a single randomly selected Acp allele from each of the three simulans complex species may reveal a number of amino acid differences, few of these differences will be fixed differences between species. This stands in contrast to typical analyses of R(t), which are applied to situations in which polymorphism would be expected to have negligible effects on estimates of substitution rates (e.g., GILLESPIE 1989). Further, the method used to correct for lineage effects in this analysis weights each gene along a lineage equally, thus ignoring any lineage-by-population size contributions to the variance in standing levels of polymorphism among loci. There are three reasons to think that polymorphism will not bias either our rejection of the neutral model for Acp's or our Acp's vs. non-Acp's comparison. First, under the strictly neutral model, the expectation that R(t) = 1 should not depend on the particular time during the substitution process that the populations are sampled. That is, the number of mutations occurring on each lineage is Poisson distributed regardless of the "length" of that lineage. Second, our analysis shows that the inflated R(t) for Acp's is not dependent on the particular alleles sampled. Finally, and perhaps most importantly, both Acp's and non-Acp's are segregating ancestral polymorphism, yet only Acp's show an inflated R(t).

Model of sperm competition:

The highly significant replacement R(t) rejects the simple neutral model for Acp protein evolution. However, it remains unclear which models of evolution might fit the data. Several models with different flavors of selection produce an overdispersed molecular clock (e.g., GILLESPIE 1993, 1994a,b; CUTLER 2000). CLARK (2002) proposed a scramble competition model for sperm competition, but did not characterize the properties of this model in a stochastic framework. Here we present a preliminary analysis of whether scramble competition can produce values of R(t) as high as those observed for Acp's. Figure 1 shows R(t) as a function of 4Nu (where N is held constant at 10⁶ and u is allowed to vary) for three different distributions of s_ij (see MATERIALS AND METHODS). It is clear that R(t) 1 for only a small portion of the parameter space for each distribution of s_ij. To inform our expectations about actual data, we drew samples of size n = 8 (the typical sample size for our D. simulans data) from simulated populations at independent intervals and then recorded polymorphism statistics. This procedure allows us to determine the approximate values of 4Nu, which, under scramble competition, give the average number of segregating sites observed among D. simulans Acp's (assuming N = 10⁶). For the Beta(5, 1) case, this value corresponded to 4Nu 1. For the Uniform(0.5, 1) case, this value was 4Nu 3. Notably, the Uniform(0, 1) did not produce the high levels of polymorphism observed within the parameter space examined here. Thus, irrespective of the distribution of selection coefficients, observed values of R(t) appear to be incompatible with Clark's scramble competition model.

View larger version (10K): In this window In a new window Download PPT slide   FIGURE 1.— Simulation results from Clark's scramble model. Plotted is the index of dispersion R(t) vs. 4Nu, where N = 106 and u is allowed to vary. For each simulation N = 106 haploid individuals. For each parameter set, the simulation was burnt in for the first 2000 fixations, and then results were recorded for the next 100,000 substitutions. R(t) was estimated from C0 in the simulation (see GILLESPIE 1993).

 

Frequency distribution:

We used TAJIMA's (1989) D to test for deviations of the site frequency distribution from the neutral equilibrium expectation (Table 1) at silent and replacement sites independently. Statistical significance was assessed using 10⁵ standard coalescent simulations conditioning on the observed number of segregating sites. D values were nonsignificant in all cases.

Polymorphic vs. fixed mutations:

Tables 4 and 5 show results of several MK tests. Unpolarized MK tests from each of the simulans complex species to D. melanogaster (Table 4) are nonsignificant, with the exception of Acp26Aa in the D. mauritiana and D. simulans lineages (the MK test of Acp26Aa variation in D. sechellia is not significant after a conservative Bonferroni correction for multiple tests although the data show the same pattern observed in the other species). These results, as well as previously reported significant deviations for Acp26Aa, which were obtained without polymorphism data from D. simulans or D. sechellia (AGUADé 1998; TSAUR et al. 1998), are consistent with an excess of amino acid fixations at this locus. Polarized MK tests (Table 5) provide no support for adaptive evolution at individual loci after speciation of the three sister taxa, with the notable exception of D. simulans Acp76A, which is significant even after corrections for multiple tests. Interestingly, this pattern is not nearly as strong in the unpolarized comparison to D. melanogaster. Thus, it may represent a lineage-specific phenomenon in D. simulans.

View this table: In this window In a new window   TABLE 4 Unpolarized McDonald-Kreitman tests between D. melanogaster and each of the simulans clade species

 

View this table: In this window In a new window   TABLE 5 Polarized McDonald-Kreitman tests for each of the simulans complex species

 
Table 6 shows polarized Acp mutations summed across loci for each simulans complex lineage. Two patterns are of note. First, if we restrict our attention to the two highly variable species D. simulans and D. mauritiana, D. mauritiana is much more polymorphic than D. simulans at replacement sites. Second, the contingency table for D. sechellia is individually significant (P = 0.037), most plausibly as the result of an excess of silent fixations (see below).

View this table: In this window In a new window   TABLE 6 Polarized mutations summed across loci for individual lineages of the simulans complex species

 
Table 7 shows a contingency table of polarized silent mutations categorized as preferred (a mutation from an unpreferred to a preferred codon), unpreferred (a mutation from a preferred codon to unpreferred codon), and no change (mutations from preferred to preferred or from unpreferred to unpreferred) mutations. No single-locus contingency table is significantly heterogeneous, nor is the contingency table of pooled mutations significantly heterogeneous for any species. Nevertheless, there is a strong lineage effect for D. sechellia silent sites. In contrast to D. simulans and D. mauritiana, which have fixed approximately equal numbers of preferred and unpreferred mutations, D. sechellia has fixed significantly more unpreferred codons than preferred codons (10:4; binomial test, P = 0.028). This supports the conclusion from KLIMAN et al. (2000) that D. sechellia shows a genomic trend toward reduced codon bias.

View this table: In this window In a new window   TABLE 7 Polarized silent fixations and polymorphisms

 

An interesting new result from our analyses is that the index of dispersion [R(t)] for Acp's, in contrast to that for non-Acp's, is overdispersed. While R(t) has been shown to have limitations (GOLDMAN 1994; NIELSEN 1997), our results strengthen the idea that Acp protein evolution is generally incompatible with simple neutrality (Table 3). Overdispersion does not appear to be a general property of Drosophila proteins (this report; ZENG et al. 1998). Thus the evolution of the amino acid sequences of Acp proteins requires a special explanation. The overdispersed R(t) for Acp's rejects not only most neutral models of molecular evolution (but see TAKAHATA 1987), but also many models of selection. For example, recurrent genetic hitchhiking models (such as the normal shift model of GILLESPIE 1994b) produce R(t) < 1. Overdominance can also produce R(t) < 1, as can random environment models, such as the SAS-CFF model (GILLESPIE 1978) with a low environmental autocorrelation. However, random environment models with fluctuations on the timescale of molecular evolution can produce R(t) > 1 (GILLESPIE 1993), as can the house of cards model (OHTA and TACHIDA 1990; GILLESPIE 1994b) and other deleterious mutation models (CUTLER 2000).

CLARK (2002) proposed a simple model of scramble competition to describe the molecular evolution of Acp's. Given that this model has a balancing component in the absence of fluctuating parameters, we might expect R(t) < 1 (see CUTLER 2000). Indeed, simulation results (Figure 1) of the scramble competition model clearly demonstrate that R(t) 1 for most of the parameter space examined. Interestingly, in these simulations R(t) decreases with increasing mutation rate. An intuitive explanation for this result is that the strength of the balancing component (i.e., frequency-dependent selection) in the scramble competition model increases with the number of unique alleles in the population. Thus, scramble competition cannot explain the observed patterns of divergence for the levels of heterozygosity observed in our samples. Nor can the Acp data be explained by shift models of hitchhiking (e.g., GILLESPIE 1997). Instead, deleterious allele models and random environment models with high environmental autocorrelations seem to provide a better fit to the observations. Overdispersed molecular clocks are a signature of episodic bursts of substitution. If Acp evolution is driven by sexual conflict (e.g., RICE 1996; HOLLAND and RICE 1999), perhaps these data suggest that there are periods of escalated conflict interspersed with periods of relative quiescence.

Despite the evidence for nonneutral evolution from estimates of R(t), neither contrasts of polymorphism and divergence nor analyses of the frequency spectrum show convincing evidence of selection on Acp's. Certainly there is little evidence for adaptive evolution at individual loci. For example, no gene/species examined here deviates from the neutral site frequency spectrum (Table 1). Furthermore, most genes/species are consistent with the neutral model in McDonald-Kreitman tests. There are, however, exceptions. The unpolarized test of Acp26Aa between D. melanogaster and each simulans complex species rejects neutrality in the direction of an excess of amino acid fixations (this report; AGUADé et al. 1992; AGUADé 1998; TSAUR and WU 1997; TSAUR et al. 1998, 2001). AGUADé (1999) reported evidence for directional selection in Acp29AB in an unpolarized comparison between D. melanogaster and D. simulans. Data from the slightly smaller region of the gene surveyed here do not deviate from neutrality. Furthermore, polarized analyses of polymorphism and divergence provide little evidence for adaptive protein divergence between simulans complex species. Even Acp26Aa data are compatible with neutrality. This contrast between the results for polarized and unpolarized MK tests at Acp26Aa is consistent with a burst of adaptive protein evolution on the lineage connecting the most recent common ancestor of the simulans complex to D. melanogaster. However, further investigation of this hypothesis is problematic because of uncertain alignments with the outgroup, D. yakuba.

A notable result from analyses of polymorphism and divergence at the single-locus level comes from Acp76A. A polarized MK test rejects the neutral model in a manner consistent with an excess of amino acid fixations in D. simulans. If amino acid mutations fix under directional selection, one might predict that heterozygosity would be reduced near such fixations (MAYNARD SMITH and HAIGH 1974; KAPLAN et al. 1989; STEPHAN et al. 1992; GILLESPIE 1997, 2000). We used a KJB test (KERN et al. 2002) to test this prediction. This test uses a resampling method to determine whether levels of variation near amino acid fixations (or any defined fixation type) are inconsistent with the average level of variation in a gene. Surprisingly, Acp76A amino acid fixations are associated with a significant excess of polymorphism (two-tailed test and a 200-bp window; P = 0.025 for _w; P = 0.013 for ) rather than with reduced polymorphism. In fact, all six fixations occurred in a region of Acp76A that is highly polymorphic at silent and replacement sites (Figure 2) . Thus, two patterns require explanation. First, there is an apparent excess of amino acid fixations. Second, there is a nonrandom physical organization of polymorphism—the region of the protein that presumably shows an excess of amino acid fixations has more polymorphism than other regions of the same gene. This is certainly not the expectation under hitchhiking models. Note that the fact that the region showing rapid protein divergence also is highly polymorphic does not negate the conclusion that there is an excess of amino acid fixation in the gene as a whole. A possible explanation for this pattern is that one region of Acp76A is a "hotspot" for both adaptive fixations and balanced polymorphisms (cf. GILLESPIE 1994a). Perhaps mutations under balancing selection rapidly fix when the environment changes, and the biology of the Acp76A protein is such that these selected sites are physically clustered. If this hypothesis is correct then one might be able to detect functional variation at this locus in D. simulans populations. Alternatively, it remains possible given the relatively few observed mutations that the significant MK test is a case of falsely rejecting the null hypothesis of neutral evolution just by chance (type I error).

View larger version (20K): In this window In a new window Download PPT slide   FIGURE 2.— All six replacement fixations in D. simulans Acp76A occurred in a region that is highly polymorphic at silent and replacement sites. The heterozygosity statistic is . Position refers to the location within the sampled region of Acp76A. This sampled region includes almost all of this gene. The window size shown is 200 bp with a displacement of 10 bp. There are 26 segregating sites within this coding region.

 
Acp's show greater amounts of amino acid polymorphism and divergence than do a set of non-Acp genes in all species of the simulans complex (as well as D. melanogaster; BEGUN et al. 2000). Not surprisingly (e.g., KLIMAN et al. 2000), D. sechellia Acp's were less polymorphic than D. mauritiana and D. simulans Acp's. Interestingly, D. mauritiana Acp's harbor more amino acid polymorphism than do D. simulans Acp's (Mann-Whitney U-test: P = 0.032 for _w; P = 0.044 for ; see Tables 5 and 6), in spite of the fact that D. simulans is more polymorphic than D. mauritiana at silent sites (HEY and KLIMAN 1993; KLIMAN et al. 2000; this report) overall. More generally, the rank order of replacement to silent heterozygosity at Acp's (Table 8) is negatively correlated with the presumed rank order effective population size (as inferred from silent heterozygosity). Given that D. mauritiana and D. sechellia are island endemics, a possible explanation for the relative excess of amino acid polymorphism is that reduced population sizes of D. mauritiana and D. sechellia result in a greater contribution of mildly deleterious amino acid mutations to polymorphism in these species relative to D. simulans. Nevertheless, the pattern does not appear to hold at non-Acp loci in D. mauritiana. Thus, one would have to explain why the biology of Acp's might result in a distribution of fitness effects for segregating variation in the ancestral population that would be more nearly neutral than the distribution at other loci.

View this table: In this window In a new window   TABLE 8 Ratio of replacement to silent heterozygosity for Acp's

 

We thank J. Gillespie, P. Awadalla, A. Holloway, and the Nuzhdin lab for their input on this work. J. Coyne and D. Barbash provided stocks. S. Shih, M. Kerber, and R. Repking gave technical assistance. A.D.K. is a Howard Hughes Medical Institute predoctoral fellow. C.D.J. and D.J.B. were funded by the National Science Foundation.

AGUADé, M., 1998 Different forces drive the evolution of Acp26Aa and Acp26Ab accessory gland genes in the Drosophila melanogaster species complex. Genetics 150: 1079–1089.[Abstract/Free Full Text]

AGUADé, M., 1999 Positive selection drives the evolution of the Acp29AB accessory gland protein in Drosophila. Genetics 152: 543–551.[Abstract/Free Full Text]

AGUADé, M., N. MIYASHITA and C. H. LANGLEY, 1992 Polymorphism and divergence in the mst 355 male accessory gland gene region. Genetics 132: 755–770.[Abstract]

BEGUN, D. J., and P. WHITLEY, 2000 Reduced X-linked nucleotide polymorphism in Drosophila simulans. Proc. Natl. Acad. Sci. USA 97(11): 5960–5965.[Abstract/Free Full Text]

BEGUN, D. J., P. WHITLEY, B. L. TODD, H. M. WALDRIP-DAIL and A. G. CLARK, 2000 Molecular population genetics of male accessory gland proteins in Drosophila. Genetics 156: 1879–1888.[Abstract/Free Full Text]

CHEN, P. S., 1996 The accessory gland proteins in male Drosophila: structural, reproductive, and evolutionary aspects. Experientia 52: 503–510.[CrossRef][Medline]

CIVETTA, A., and R. S. SINGH, 1995 High divergence of reproductive tract proteins and their association with postzygostic reproductive isolation in Drosophila melanogaster and Drosophila virilis group species. J. Mol. Evol. 41: 1085–1095.[Medline]

CLARK, A. G., 2002 Sperm competition and the maintenance of polymorphism. Heredity 88: 148–153.[CrossRef][Medline]

CLARK, A. G., M. AGAUDé, T. PROUT, L. G. HARSHMAN and C. H. LANGLEY, 1995 Variation in sperm displacement and its association with Accessory gland protein loci in Drosophila melanogaster. Genetics 139: 189–201.[Abstract]

COULTHART, M. B., and R. S. SINGH, 1988 Differing amounts of genetic polymorphism in testes and male accessory glands of Drosophila melanogaster and D. simulans. Biochem. Genet. 26: 153–164.[CrossRef][Medline]

CUTLER, D. J., 2000 Understanding the overdispersed molecular clock. Genetics 154: 1403–1417.[Abstract/Free Full Text]

ENGELS, B., 1988 Monte Carlo Contingency Table Test, Version 2.1. University of Wisconsin, Madison.

EWING, B., and P. GREEN, 1998 Base-calling of automated sequencer traces using phred. II. Error probabilities. Genome Res. 8: 186–194.[Abstract/Free Full Text]

EWING, B., L. HILLIER, M. C. WENDL and P. GREEN, 1998 Base-calling of automated sequencer traces using phred. I. Accuracy assessment. Genome Res. 8: 175–185.[Abstract/Free Full Text]

GILLESPIE, J. H., 1978 A general model to account for enzyme variation in natural population. V. The SAS-CFF model. Theor. Popul. Biol. 14: 1–45.[CrossRef][Medline]

GILLESPIE, J. H., 1989 Lineage effects and the index of dispersion of molecular evolution. Mol. Biol. Evol. 6: 636–647.[Abstract]

GILLESPIE, J. H., 1993 Substitution processes in molecular evolution. I. Uniform and clustered substitutions in a haploid model. Genetics 134: 971–981.[Abstract]

GILLESPIE, J. H., 1994a Substitution processes in molecular evolution. II. Exchangeable models from population genetics. Evolution 48: 1101–1113.[CrossRef]

GILLESPIE, J. H., 1994b Substitution processes in molecular evolution. III. Deleterious alleles. Genetics 138: 943–952.[Abstract]

GILLESPIE, J. H., 1997 Junk ain't what junk does: neutral alleles in a selected context. Gene 205: 291–299.[CrossRef][Medline]

GILLESPIE, J. H., 2000 Genetic drift in an infinite population: the pseudohitchhiking model. Genetics 155: 909–919.[Abstract/Free Full Text]

GOLDMAN, N., 1994 Variance to mean ratio, R(t), for Poisson processes on phylogenetic trees. Mol. Phylogenet. Evol. 3: 230–239.[CrossRef][Medline]

GOLDMAN, N., and Z. YANG, 1994 A codon-based model of nucleotide substitution for protein-coding DNA sequences. Mol. Biol. Evol. 11: 725–736.[Abstract]

GORDON, D., C. ABAJIAN and P. GREEN, 1998 Consed: a graphical tool for sequence finishing. Genome Res. 8: 195–202.[Abstract/Free Full Text]

HERNDON, L. A., and M. F. WOLFNER, 1995 A Drosophila seminal fluid protein, Acp26Aa, stimulates egg laying in females for 1 day after mating. Proc. Natl. Acad. Sci. USA 92: 10114–10118.[Abstract/Free Full Text]

HEY, J., and R. N. KLIMAN, 1993 Population genetics and phylogenetics of DNA sequence variation at multiple loci within the Drosophila melanogaster species complex. Mol. Biol. Evol. 10: 804–822.[Abstract]

HOLLAND, B., and W. R. RICE, 1999 Experimental removal of sexual selection reverses intersexual antagonistic coevolution and removes a reproductive load. Proc. Natl. Acad. Sci. USA 96: 5083–5088.[Abstract/Free Full Text]

KAPLAN, N. L., R. R. HUDSON and C. H. LANGLEY, 1989 The "hitchhiking effect" revisited. Genetics 123: 887–899.[Abstract/Free Full Text]

KERN, A. D., C. D. JONES and D. J. BEGUN, 2002 Genomic effects of nucleotide substitutions in Drosophila simulans. Genetics 162: 1753–1761.[Abstract/Free Full Text]

KLIMAN, R. M., P. ANDOLFATTO, J. A. COYNE, F. DEPAULIS, M. KREITMAN et al., 2000 The population genetics of the origin and divergence of the Drosophila simulans complex of species. Genetics 156: 1913–1931.[Abstract/Free Full Text]

LEE, Y.-H., T. OTA and V. D. VACQUIER, 1995 Positive selection is a general phenomenon in the evolution of the abalone sperm lysine. Mol. Biol. Evol. 12: 231–238.[Abstract]

MAYNARD SMITH, J., and J. HAIGH, 1974 The hitch-hiking effect of a favourable gene. Genet. Res. 23: 23–35.[Medline]

MCDONALD, J. H., and M. KREITMAN, 1991 Adaptive protein evolution at the Adh locus in Drosophila. Nature 351: 652–654.[CrossRef][Medline]

METZ, E. C., and S. R. PALUMBI, 1996 Positive selection and sequence rearrangements generate extensive polymorphism in the gamete recognition protein bindin. Mol. Biol. Evol. 13: 397–406.[Abstract]

NEUBAUM, D. M., and M. WOLFNER, 1999 Mated Drosophila melanogaster females require a seminal fluid protein, Acp36DE, to store sperm efficiently. Genetics 153: 845–857.[Abstract/Free Full Text]

NIELSEN, R., 1997 Robustness of the estimator of the index of dispersion for DNA sequences. Mol. Phylogenet. Evol. 7: 346–351.[CrossRef][Medline]

NURMINSKY, D. I., M. V. NURMINSKAYA, D. D. AGUIAR and D. L. HARTL, 1998 Selective sweep of a newly evolved sperm-specific gene in Drosophila. Nature 396: 572–575.[CrossRef][Medline]

OHTA, T., and H. TACHIDA, 1990 Theoretical study of near neutrality. I. Heterozygosity and rate of mutant substitution. Genetics 126: 219–229.[Abstract]

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.[Abstract/Free Full Text]

RICE, W. R., 1996 Sexually antagonistic male adaptation triggered by experimental arrest of female evolution. Nature 381: 232–234.[CrossRef][Medline]

STEPHAN, W., T. H. E. WIEHE and M. W. LENZ, 1992 The effect of strongly selected substitutions on neutral polymorphism: analytical results based on diffusion theory. Theor. Popul. Biol. 41: 237–254.

SWANSON, W. J., and V. D. VACQUIER, 2002 The rapid evolution of reproductive proteins. Nat. Rev. Genet. 3: 137–144.[Medline]

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.[Abstract/Free Full Text]

TAJIMA, F., 1989 Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123: 585–595.[Abstract/Free Full Text]

TAKAHATA, N., 1987 On the overdispersed molecular clock. Genetics 116: 169–179.[Abstract/Free Full Text]

TAKAHATA, N., 1990 A simple genealogical structure of strongly balanced allelic lines and trans-species evolution of polymorphism. Proc. Natl. Acad. Sci. USA 87: 2419–2423.[Abstract/Free Full Text]

THOMAS, S., and R. S. SINGH, 1992 A comprehensive study of genic variation in natural populations of Drosophila melanogaster. VII. Varying rates of genic divergence as revealed by two-dimensional electrophoresis. Mol. Biol. Evol. 9: 507–525.[Abstract]

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.[Abstract/Free Full Text]

TING, C.-T., S. C. TSAUR and C.-I WU, 2000 The phylogeny of closely related species as revealed by the genealogy of a speciation gene, Odysseus. Proc. Natl. Acad. Sci. USA 97: 5313–5316.[Abstract/Free Full Text]

TRAM, U., and M. WOLFNER, 1999 Male seminal fluid proteins are essential for sperm storage in Drosophila melanogaster. Genetics 153: 837–844.[Abstract/Free Full Text]

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.[Abstract]

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 vs. polymorphism. Mol. Biol. Evol. 15: 1040–1046.[Abstract]

TSAUR, S. C., C.-T. TING and C.-I WU, 2001 Sex in Drosophila mauritiana: a very high level of amino acid polymorphism in a male reproductive gene, Acp26Aa. Mol. Biol. Evol. 18: 22–26.[Abstract/Free Full Text]

WATTERSON, G. A., 1975 On the number of segregating sites in genetical models without recombination. Theor. Popul. Biol. 7: 256–276.[CrossRef][Medline]

WOLFNER, M. F., 1997 Tokens of love: functions and regulation of Drosophila male accessory gland products. Insect Biochem. Mol. Biol. 27: 179–192.[CrossRef][Medline]

WYCKOFF, G. J., W. WANG and C.-I WU, 2000 Rapid evolution of male reproductive proteins in the descent of man. Nature 403: 304–309.[CrossRef][Medline]

YANG, Z., 1997 PAML: a program package for phylogenetic analysis by maximum likelihood. Comput. Appl. Biosci. 13: 555–556.[Free Full Text]

ZENG, L.-W., J. M. COMERON, B. CHEN and M. KREITMAN, 1998 The molecular clock revisited: the rate of synonymous vs. replacement change in Drosophila. Genetica 102/103: 369–382.[CrossRef]

This article has been cited by other articles:

				T. Bedford and D. L. Hartl Overdispersion of the Molecular Clock: Temporal Variation of Gene-Specific Substitution Rates in Drosophila Mol. Biol. Evol., August 1, 2008; 25(8): 1631 - 1638. [Abstract] [Full Text] [PDF]

				L. A. McGraw, A. G. Clark, and M. F. Wolfner Post-mating Gene Expression Profiles of Female Drosophila melanogaster in Response to Time and to Four Male Accessory Gland Proteins Genetics, July 1, 2008; 179(3): 1395 - 1408. [Abstract] [Full Text] [PDF]

				T. Bedford, I. Wapinski, and D. L. Hartl Overdispersion of the Molecular Clock Varies Between Yeast, Drosophila and Mammals Genetics, June 1, 2008; 179(2): 977 - 984. [Abstract] [Full Text] [PDF]

				A. Wong, M. C. Turchin, M. F. Wolfner, and C. F. Aquadro Evidence for Positive Selection on Drosophila melanogaster Seminal Fluid Protease Homologs Mol. Biol. Evol., March 1, 2008; 25(3): 497 - 506. [Abstract] [Full Text] [PDF]

				W. Haerty, S. Jagadeeshan, R. J. Kulathinal, A. Wong, K. Ravi Ram, L. K. Sirot, L. Levesque, C. G. Artieri, M. F. Wolfner, A. Civetta, et al. Evolution in the Fast Lane: Rapidly Evolving Sex-Related Genes in Drosophila Genetics, November 1, 2007; 177(3): 1321 - 1335. [Abstract] [Full Text] [PDF]

				C. S. McBride, J. R. Arguello, and B. C. O'Meara Five Drosophila Genomes Reveal Nonneutral Evolution and the Signature of Host Specialization in the Chemoreceptor Superfamily Genetics, November 1, 2007; 177(3): 1395 - 1416. [Abstract] [Full Text] [PDF]

				B. J. Wagstaff and D. J. Begun Adaptive Evolution of Recently Duplicated Accessory Gland Protein Genes in Desert Drosophila Genetics, October 1, 2007; 177(2): 1023 - 1030. [Abstract] [Full Text] [PDF]

				K. Ravi Ram and M. F. Wolfner Seminal influences: Drosophila Acps and the molecular interplay between males and females during reproduction Integr. Comp. Biol., September 1, 2007; 47(3): 427 - 445. [Abstract] [Full Text] [PDF]

				A. C. Fiumera, B. L. Dumont, and A. G. Clark Associations Between Sperm Competition and Natural Variation in Male Reproductive Genes on the Third Chromosome of Drosophila melanogaster Genetics, June 1, 2007; 176(2): 1245 - 1260. [Abstract] [Full Text] [PDF]

				A. Sanchez-Gracia and J. Rozas Unusual Pattern of Nucleotide Sequence Variation at the OS-E and OS-F Genomic Regions of Drosophila simulans Genetics, April 1, 2007; 175(4): 1923 - 1935. [Abstract] [Full Text] [PDF]

				J. A. Andres, L. S. Maroja, S. M. Bogdanowicz, W. J. Swanson, and R. G. Harrison Molecular Evolution of Seminal Proteins in Field Crickets Mol. Biol. Evol., August 1, 2006; 23(8): 1574 - 1584. [Abstract] [Full Text] [PDF]

				J. J. Welch Estimating the Genomewide Rate of Adaptive Protein Evolution in Drosophila Genetics, June 1, 2006; 173(2): 821 - 837. [Abstract] [Full Text] [PDF]

				D. J. Begun, H. A. Lindfors, M. E. Thompson, and A. K. Holloway Recently Evolved Genes Identified From Drosophila yakuba and D. erecta Accessory Gland Expressed Sequence Tags Genetics, March 1, 2006; 172(3): 1675 - 1681. [Abstract] [Full Text] [PDF]

				D. A. Moeller and P. Tiffin Genetic Diversity and the Evolutionary History of Plant Immunity Genes in Two Species of Zea Mol. Biol. Evol., December 1, 2005; 22(12): 2480 - 2490. [Abstract] [Full Text] [PDF]