Evidence that strong positive selection drives neofunctionalization in the tandemly duplicated polyhomeotic genes in Drosophila (original) (raw)

• Journal List
• Proc Natl Acad Sci U S A
• v.105(14); 2008 Apr 8
• PMC2291077

Proc Natl Acad Sci U S A. 2008 Apr 8; 105(14): 5447–5452.

Steffen Beisswanger

Section of Evolutionary Biology, BioCenter, University of Munich, Grosshaderner Strasse 2, 82152 Planegg-Martinsried, Germany

Wolfgang Stephan

Section of Evolutionary Biology, BioCenter, University of Munich, Grosshaderner Strasse 2, 82152 Planegg-Martinsried, Germany

Section of Evolutionary Biology, BioCenter, University of Munich, Grosshaderner Strasse 2, 82152 Planegg-Martinsried, Germany

Edited by Tomoko Ohta, National Institute of Genetics, Mishima, Japan, and approved February 6, 2008

Author contributions: S.B. and W.S. designed research; S.B. performed research; S.B. and W.S. analyzed data; and S.B. and W.S. wrote the paper.

Copyright © 2008 by The National Academy of Sciences of the USA

Abstract

The polyhomeotic (ph) locus in Drosophila melanogaster consists of the two tandemly duplicated genes ph-d (distal) and ph-p (proximal). They code for transcriptional repressors belonging to the Polycomb group proteins, which regulate homeotic genes and hundreds of other loci. Although the duplication of ph occurred at least 25 million to 30 million years ago, both copies are very similar to each other at both the DNA and the protein levels, probably because of the action of frequent gene conversion. Despite this homogenizing force, differential regulation of both transcriptional units suggests that the functions of the duplicates have begun to diverge. Here, we provide evidence that this functional divergence is driven by positive selection. Based on resequencing of an ≈30-kb region around the ph locus in an African sample of D. melanogaster X chromosomes, we identified a selective sweep, estimated its age and the strength of selection, and mapped the target of selection to a narrow interval of the ph-p gene. This noncoding region contains a large intron with several regulatory elements that are absent in the ph-d duplicate. Our results suggest that neofunctionalization has been achieved in the Drosophila ph genes through the action of strong positive selection and the inactivation of gene conversion in part of the gene.

Keywords: gene duplication, selective sweep

Gene duplication is a major evolutionary mechanism for generating new genes and new functions, thus increasing organismal complexity. Since Ohno's (1) pioneering ideas, this topic has become a main stay of evolutionary biology research. Ohta (2) laid the groundwork by exploring the population genetic mechanisms leading to the maintenance of gene duplications and the evolution of multigene families. The early evolution and functional divergence of duplicated genes, on the other hand, remained somewhat obscure. Theoretical work suggests that both genetic drift and positive selection may play a role in the fixation and early evolution of duplicate genes (3–5). In particular, positive selection is thought to drive the fixation of a duplicate gene that has gained a new function through acquisition of a beneficial mutation, a process referred to as neofunctionalization (6). However, there is currently limited evidence for this suggestion. The reasons for this may be twofold. First, it is difficult to detect newly diverging gene copies and, at the same time, identify selection at one and/or the other copy at the population level. Second, recent mathematical modeling predicts that neofunctionalization is unlikely in the presence of gene conversion, unless selection is very strong (7).

One possible method for detecting strong positive selection and localizing the target of selection is the search for selective sweeps. A selective sweep denotes a signature of variation in the genome that results from the recent fixation of a new, strongly selected beneficial mutation (8) or standing low-frequency variants (9). Such footprints last not much longer than 0.1 _N_e generations, where _N_e is the effective population size (10, 11). In the past 5 years, several approaches have been proposed to detect sweeps in SNP data (10, 12, 13). The hallmark of a selective sweep is a severe reduction of genetic variation at the site of the beneficial mutation in the genome and an increase of nucleotide diversity at both sides of the selected site. The rate of increase depends on the ratio of the frequency of crossing-over to the strength of selection (8). Thus, if the rate of crossing-over is sufficiently high and the sweep was recent, this diversity-reducing feature of a sweep can be used to map the target of selection in the genome relatively accurately (potentially down to the gene level; refs. 14–16).

In a previous study, Beisswanger et al. (17) identified a genomic region in an African population (from the ancestral species range) that most likely has been affected by positive directional selection in the recent past. However, because of partial sequencing in this region, they were unable to identify the target of selection with high confidence. Interestingly, this region encompasses the locus polyhomeotic (ph) with the two tandemly duplicated copies ph-p (proximal) and ph-d (distal) that were in the approximate center of the swept segment. The ph genes code for transcriptional repressors belonging to the Polycomb group of proteins, which is known to regulate homeotic genes and also hundreds of other genes in mammals and insects (18). The distal protein is very similar to the proximal product, except for the absence of an amino acid terminal region and a small region near the carboxyl terminus (19). Over a stretch of 1.3 kb both proteins are completely identical. On the other hand, the first intron of ph-p is much larger than that of ph-d. With regard to functional divergence, there is no clear evidence that the distal and proximal products bind to and modify chromatin in different ways. However, other results (such as the differential regulation of both transcriptional units at the mRNA level) suggest that the functions of these proteins have begun to diverge (20). Thus, although the ph duplication is relatively old (see Discussion), the fact that the duplicates show little or no amino acid differences but significantly different expression patterns suggests that the two genes are still in an early phase of their sequence and functional divergence.

Here, we present results from a resequencing study of an ≈30-kb region around the ph genes, based on a sample of 12 Drosophila melanogaster X chromosomes. We report strong evidence for a selective sweep in this region and map the target of selection to ph-p, which indicates that positive selection drives the early sequence and functional divergence of the ph genes.

Results

Polymorphism in the _ph-d_–CG3835 Region.

In a previous study we reported evidence for a selective sweep that affected the genomic region around the wapl gene of an African population of D. melanogaster (17). Several neutrality tests and the recently proposed composite likelihood ratio (CLR) test (10) pointed toward a target of selection located in the center of the region (within or near the gene ph-p). Therefore, we decided to sequence the _ph-d_–CG3835 region, i.e., the segment between the 3′ flanking region of ph-d and the 3′ flanking region of CG3835 encompassing the gene ph-p (Fig. 1). This segment comprises 31,700 annotated base pairs (D. melanogaster genome release 4.3; ref. 21). However, small parts of the region could not be sequenced because of primer malfunction (see below). Fragment positions and individual summaries are provided in Table 1. In total, we analyzed 18 fragments with an overall length of ≈27 kb (per line) in 12 African D. melanogaster lines.

Illustration of the _ph-d_–CG3835 region. (Upper) Modified from ref. 17, the entire wapl region is shown. The wapl fragment is indicated by a star. It harbors no variation in the European sample and was detected in a previous genome scan (34). The numbers on the x axis are the absolute genomic position (in Mb), and those on the y axis are nucleotide variability in the African sample (the solid line corresponds to θ and the dashed line corresponds to π). The arrow indicates the position of the target of selection predicted by Beisswanger et al. (17). (Lower) The region investigated in detail in this study. Exons are illustrated by thick black lines and introns by thin black lines. The direction of transcription is indicated by dotted arrows below each gene. Sequenced fragments are denoted by gray lines and numbers. Gray arrow heads indicate the position of the three SNPs shared between both ph copies. These SNPs are located within a region of 135 bp.

Table 1.

Summaries of SNP data

Because ph-d and ph-p are paralogous genes (19, 22) sequencing was sometimes difficult. Hence, we blasted each sequenced fragment against the published D. melanogaster genome sequence and used only those fragments that resulted in a blast identity of >99.9%. We identified two segments where primers erroneously annealed to the paralogous gene. These fragments were excluded from subsequent analyses. Excluding sequence overlaps and indels, a total of 27,372 bp were used for further investigation.

As shown in Table 1, levels of nucleotide variability across the newly sequenced region are heterogeneous, but low, particularly in the vicinity of ph-p. For noncoding and synonymous sites average θ across the region is 0.0037. This value is lower than the chromosome-wide average (θ = 0.0126; 23). In addition, average θ is ≈40% lower than the average level of nucleotide diversity reported for the 12 loci around the _ph-d_–CG3835 region (see ref. 17 and Fig. 1). In total, we observed 181 SNPs in our dataset. Here, we focus on noncoding sequences.

Haplotype Structure.

The number of haplotypes across the region ranges from 2 to 12 with an average of 6 haplotypes per locus. Haplotype diversity (Hd) ranges from 0.17 to 1 (Table 1). Linkage disequilibrium (LD), as measured by Kelly's _Z_nS statistic (24), is low to intermediate (average: 0.48), and only three values were significant (Table 1). Two of these loci (fragments 4 and 17) represent coding sequences. For eight loci no _Z_nS estimates could be obtained, as all polymorphic sites were noninformative. Five of these fragments, i.e., loci 8 and 10–13, are located in ph-p and its 5′ flanking region. We also estimated LD in terms of _r_2 (25) for all pairwise SNPs across the whole region. Only 14 significant associations were observed, of which 8 are located within ph-d, 5 in the _ph-d_–ph-p intergenic region, and 1 in the _ph-p_–CG3835 intergenic region. Assuming the selected site was estimated correctly by Beisswanger et al. (17), this result is in qualitative agreement with Stephan et al. (26), who predicted that after a hitchhiking event LD is reduced between polymorphisms across the selected site (relative to LD between pairs of polymorphisms located on the same side of the target of selection). Because Hd estimates were generally high (on average 0.67), we were also able to determine the minimum number of recombination events. A minimum of 23 crossovers was identified across the region investigated.

Standard Neutrality Tests and Frequency Spectra.

As shown in Table 1, all fragments under investigation show a trend toward a negative Tajima's D (27), except fragment 15 for which D = 0.05. We obtained significantly negative estimates for four loci in the region analyzed (D = −1.97, −1.83, −1.62, and −1.66 for fragments 12, 13, 14, and 16, respectively). Three of these fragments are located in or immediately around ph-p (Fig. 1). Similarly, Fu and Li's D (28) was significantly different from zero (P < 0.05) for these loci (D = −2.52, −2.67, −1.67, and −2.26). This finding indicates an excess of low-frequency variants. We also applied a multilocus HKA test (29) to our data and obtained a significant result (_X_2 = 17.87, P = 0.037). In addition, we analyzed our data in combination with the data of Beisswanger et al. (17). The result obtained by this approach was also significant (_X_2 = 32.79; P = 0.012), which indicates that the _ph-d_–CG3835 region is not evolving according to the neutral equilibrium model. We note that in the presence of gene conversion (which may have played a role in the evolution of the ph genes; see below), the tests we have used here are conservative (30).

With regard to the site frequency spectrum, we observed a strong deviation from standard neutral expectations at ph-p. In a >4-kb stretch (encompassing the large intron of ph-p, its 5′ UTR and 417 bp of the _ph-p_–CG3835 intergenic region), we detected only singletons, with the exception of a doubleton at position 15,785 (Fig. 2Inset, a). A similar, although less extreme, pattern is evident in the _ph-p_–CG3835 intergenic region. Here, the frequency spectrum of the first 40 segregating sites (corresponding to the first half of that region) also reveals an excess of low-frequency variants (Fig. 2 Inset, b). Then, going further away from ph-p toward CG3835, the frequency spectrum of the second half of 40 SNPs (Fig. 2 Inset, c) approaches that of the entire X chromosome (see Fig. 2).

Site frequency spectra. The folded frequency spectrum of 27 loci randomly chosen from ref. 23. (Inset) Shown are the spectra of the ph-p intron and the 5′ UTR (a), the first half of the _ph-p_–CG3835 intergenic region (b), and the second half of that region (c). The gray lines indicate the equilibrium frequencies.

Application of the CLR Test and the Strength of Selection.

We applied the CLR test of Kim and Stephan (10) to our data to examine two different scenarios. First, we tested our new data (i.e., the _ph-d_–CG3835 region), assuming an expected nucleotide diversity of θ = 0.0037, which is the average across the region. Second, we tested our dataset assuming θ = 0.0067, which is the mean across the entire wapl region (17). As shown in Table 2, we obtained highly significant results for both scenarios (P < 0.01). Following Jensen _et al._ (31), our results were also robust against simple demographic scenarios (_P_ > 0.9). In other words, positive selection rather than demography is the likely cause of the pattern observed. Estimates of the strength of selection (α) are 1,141 and 2,608 for both scenarios tested, which suggests that selection at ph was strong. Because the method for the estimation of α assumes that the sweep ended at present time (10), which is not the case here (see below), the strength of selection is somewhat underestimated.

Table 2.

Likelihood analysis of selective sweep

Target of Selection.

The CLR test (10) also provides an estimate of the target of selection. For both scenarios the likelihood ratio was maximized at 15.2 kb. This position corresponds to the first intron of the gene ph-p, which is 4,291 bp in length according to release 4.3 of the D. melanogaster genome (21). Because the 1,445-bp transposable element (TE) FB{}23 is annotated in this region (positions 15,516 to 16,961 in our dataset), we tested all D. melanogaster lines and Drosophila simulans for the presence of the TE by diagnostic PCR. We could not detect any insertion in this genomic region in the Drosophila lines investigated (data not shown), which indicates that the first ph-p intron is actually 1,445 bp smaller than annotated.

We scrutinized this intron and the adjacent 5′ flanking region (noncoding) for fixed differences between D. melanogaster and its sibling species D. simulans, and Drosophila yakuba. In addition, we analyzed divergence between species. In parts of the intron and at the beginning of the 5′ flanking region we detected relatively little divergence among species. However, two segments of the intron, located at 16.3–16.6 kb and 17.3–17.7 kb, show high levels of divergence (≥ 10% to D. simulans and >25% to D. yakuba). A similar pattern can be observed in the 5′ flanking region. In the exons we could not detect any significant changes, except for a few single amino acid substitutions. However, we found indel differences between species in both the intron and 5′ region. There are two deletions of 28 and 37 bp in comparison with D. simulans (D. yakuba) at intron positions 15,747 and 17,226, respectively. In addition, a fixed 6-bp deletion can be found at position 18,925.

We also searched the first ph-p intron and the 5′ flanking region in the three Drosophilids for putative transcription factor binding sites (TFBs). We detected several putative TFBs with high confidence (P > 0.95), which are located in the first intron and the 5′ flanking region. Several of these are shared between D. simulans and D. yakuba but not between these species and D. melanogaster. These TFBs are located at the 5′ end of the intron or the proximate part of the 5′ flanking region. Interestingly, all of the detected TFBs are associated with the regulation of homeotic genes. For instance, at position 18,801 a GAGA factor binding site is predicted for D. melanogaster, but not for D. simulans and D. yakuba. At position 18,519 a fushi tarazu binding site is predicted for D. melanogaster, whereas a caudal binding site is found for the other two species. This difference is caused by an A to T substitution in the D. melanogaster lineage. Similarly, at intron position 17,383 a retained TFB is predicted for D. melanogaster, but not for D. yakuba and D. simulans, because of a nucleotide substitution.

Age of Selective Sweep.

We estimated the age of the selective sweep by using two different approaches. First, we applied the method of Przeworski (11) to ≈4 kb of the ph-p 5′ flanking region. We assumed a distance of either 1 bp, 1.7 kb, or 3.2 kb to the selected site. These values correspond to a putatively selected site located in either the proximate 5′ flanking region (at the proximate end of fragment 14), the 5′ end of the ph-p intron, or the middle of that intron. For all distances tested this method returned age estimates between 45,000 and 55,000 years (assuming 10 generations per year; Table 3). Very similar values are obtained if ≈ 4 kb of the downstream region of the ph-p intron are used. We note that these estimates are surprisingly consistent. For both regions analyzed, the lowest values are obtained if the selected site is assumed to be in the 5′ flanking region.

Table 3.

Age of selective sweep (according to ref. 11)

Second, we estimated the time since the fixation of the beneficial allele following Slatkin and Hudson (32) and Ayala et al. (33). This method assumes a star-like genealogy, i.e., a complete selective sweep and subsequent accumulation of new mutations. We applied the method to two regions. First, we considered the first ph-p intron, which harbors 13 segregating sites in 2,613 bp, of which 12 are singletons. Assuming a local mutation rate of 0.9 × 10 −9 per year (estimated from K = 0.041), this method yields an age estimate of 46,086 years. In addition, we applied this method to polymorphism data from 4,219 consecutive sites of the 5′ flanking region where 41 SNPs were observed. We obtained an estimate of 44,990 years since fixation assuming μ = 1.8 × 10−9 (K = 0.085). Similar values were found for various stretches of the center of the sweep region.

In the presence of gene conversion, the first method should overestimate the age of the sweep because this process is expected to shift the frequency spectrum of polymorphic sites to intermediate values (30), whereas the second method (which is based on the number of segregating sites) should be less sensitive. The fact that both approaches returned comparable values (with regard to the method and the region analyzed) suggests that gene conversion does not introduce a major bias on our age estimates.

Discussion

Evidence for a Selective Sweep in the ph-p Gene Region of African D. melanogaster.

By resequencing ≈80% of an ≈30-kb region around the ph genes in a sample of 12 African D. melanogaster X chromosomes, we found strong evidence for a selective sweep in this region. Furthermore, our mapping showed that the target of selection is most likely located at the 5′ end of the large intron of ph-p or the proximate 5′ flanking region (within a region of ≈3 kb). We arrived at this conclusion after two rounds of mapping. First, we selected from an initial genome scan (34) a region with low variation (the wapl fragment) and found evidence for a sweep by resequencing several short fragments around wapl in a European and an African sample (17). For the European population a very large valley of very low variation (of ≈60 kb) was detected, whereas in the African population the valley was only ≈15 kb wide. Statistical tests showed evidence for a sweep in both populations. However, it was also plausible that the European pattern of variation was caused by a sweep that originated in Africa. For these reasons we decided to use the African population for a second round of mapping. Our estimates of the age of the sweep now support this rationale: the sweep occurred ≈50,000 years ago, most likely before the European D. melanogaster lineage split off from the African one (≈16,000 years ago; ref. 35). The estimated age of the sweep is also consistent with the observation that the sequences of the ph-p alleles from our European sample are entirely identical with that of the African haplotype in the swept region (data not shown).

It is clear that the evidence we provided is subject to uncertainty. First, the test of Kim and Stephan (10) for a selective sweep does not take the proper demographic model into account. However, we corrected for this by using the approach of Jensen et al. (31). Furthermore, we applied the test in a conservative fashion using a low (local) estimate of θ (instead of the chromosome-wide value reported in ref. 23). The 95% C.I. around the selected site are much smaller than in the first mapping study, but still considerably large (although 80% of the ph region was sequenced). However, the frequency spectra (Fig. 2) support our conclusion that the most likely target of selection is located in a region of only a few kilobases between the 5′ end of the large ph-p intron and the proximal half of the _ph-p_–CG3835 intergenic region. Finally, although the 95% C.I. of the age of the sweep are also relatively large, the lower bound corresponds to an age of >20,000 years, suggesting that the sweep is older than D. melanogaster's migration out of Africa. A more serious problem may be that the estimation method is based on an equilibrium population (11), which is probably not the case for the African population (35). Thus, taking into account the evidence that the African population increased (13) may change our estimates. However, two observations argue against this: first, according to the estimation of Li and Stephan (13) the last major population size expansion in Africa occurred ≈60,000 years ago, and second the method by Slatkin and Hudson (32), which is independent of _N_e, yields results that are consistent with the estimates produced by Przeworski's (11) method.

Can polymorphism patterns at ph-p be explained by other factors? Functional elements within genes tend to be located within the first intron (36). In addition, it has been observed that long introns are more likely to harbor functional elements than short ones (37). It has therefore been argued that intron sequence evolution is more constrained than synonymous site evolution (38). This is partially reflected in the first ph-p intron, which shows only low to intermediate divergence between D. melanogaster and D. simulans. However, the amount of polymorphism at ph-p is severely reduced over >3 kb and results of common summary statistics (Table 1) observed at ph-p are more extreme than generally observed for conserved noncoding sequences (39). This suggests that purifying selection is not a likely cause of the observed reduction of variation.

Significance of the Selective Sweep and Neofunctionalization in the ph Genes.

Which nucleotide site was the target of selection? Despite the relatively accurate mapping of the selective target, this question is difficult to answer. It may even be possible that several molecular variants within the small target region (of ≈3 kb) we identified were under positive selection, as this region contains several putative TFBs that are new in D. melanogaster or show differences between D. melanogaster and both D. simulans and D. yakuba.

Although the exact selective target(s) cannot be identified, the fact that the target region of selection contains new (or altered) TFBs relative to D. simulans and D. yakuba suggests that it is involved in the functional divergence of the two ph genes, which are differentially regulated (20). Furthermore, the evidence of a selective sweep that has been mapped to this region indicates that the functional divergence of these genes is driven by strong positive selection.

The observation that the distal and proximal Ph proteins are very similar and bind to and modify chromatin in similar ways, but that the transcriptional units of ph-d and ph-p are differentially regulated suggests that divergence of these two genes is still in its early stage. This is surprising, as the ph duplication likely occurred before the split of the obscura and melanogaster groups: a comparison of the 12 sequenced Drosophila genomes (40) shows that duplicated ph genes are already present in all species of the melanogaster group and Drosophila pseudoobscura. Thus, the duplication appears to be at least 25 million to 30 million years old (41). Given this relatively old age, it is interesting that the ph genes show little sequence divergence (5% for coding regions that could be aligned), and in fact, are completely identical at the protein level over a distance of ≈1.3 kb (19). In D. pseudoobscura sequence divergence between the ph duplicates is even smaller (< 3% for coding regions). There are two possible explanations for this. First, Ph is a member of the Polycomb group of proteins, which form very conserved protein complexes, suggesting that purifying selection at the ph locus is strong. This finding is in agreement with the generally low divergence at these genes (Table 1). Second, the identity of both genes may be maintained by frequent gene conversion. The number of segregating sites is too small to test the latter prediction (for instance, using the method described in ref. 42). However, there is some evidence for gene conversion at ph as a few polymorphic sites are shared between the duplicates (3 of 28 SNPs), which may suggest that the forces counteracting the functional divergence of the ph genes are relatively severe and can only be overcome by very strong positive selection (7). However, why was neofunctionalization eventually achieved? The likely reason appears to be that gene conversion was at some point inactivated in the first introns of ph-p and ph-d, because of drastic sequence differences caused by insertion/deletion events (43). As a result, the first intron of ph-p (which is much larger than that of ph-d) contains several regulatory elements that may be potential targets of selection (in that they can escape the counteracting force of gene conversion). We have shown here that some of these potential target sequences are located in precisely the narrow region of the genome that experienced recent strong positive selection. Taken together, these results provide consistent evidence for the process of neofunctionalization in duplicated genes under concerted evolution.

Materials and Methods

Fly Samples and Data.

We collected data from 12 highly inbred D. melanogaster lines derived from an African population (44). For interspecific comparisons we used the publicly available sequences of D. simulans, D. yakuba, and D. pseudoobcura (40). For generating the frequency spectrum (Fig. 2), we randomly chose 27 putatively neutrally evolving fragments from ref. 23: loci 57, 60, 78, 84, 93, 106, 166, 178, 205, 206, 231, 237, 241, 259, 272, 276, 277, 286, 326, 359, 392, 422, 431, 439, 465, 721, and 727.

Molecular Techniques.

Sequences of the region between the annotated genes ph-d and CG3835 (the _ph-d_–CG3835 region) were generated as described (17), except that sequences were run on an ABI 3730 DNA Analyzer (Applied Biosystems). For alignments we used SeqMan and MegAlign (DNAstar). The ≈30-kb region under investigation was sequenced in overlapping fragments.

Statistical Analysis.

Basic analyses were done by using DnaSP 4.10 (45). Nucleotide diversity was estimated in terms of θ (46) and π (47) using the number of haplotypes and Hd (48). Indels and singletons were not included in the analysis. We estimated the minimum number of recombination events in the whole region applying the four-gamete rule as implemented in LDhat (49). We tested the neutral equilibrium model by using Tajima's D (27), Fu and Li's D (28), and the HKA test (29). Significance was inferred by 10,000 neutral coalescent simulations under the assumption of equilibrium and zero recombination. Tests were performed as one-sided.

Likelihood Analysis of Selective Sweep.

We analyzed our SNP data by using the CLR test of Kim and Stephan (10). We tested the significance of the resulting likelihood ratio by 10,000 Monte Carlo simulations under neutrality. The CLR test result was evaluated by the goodness-of-fit test (31). This test was also used to obtain C.I. for X̂ by parametric bootstrapping. For the likelihood analyses we assumed θ = 0.0037 and 0.0067 (instead of the chromosomewide estimate of θ = 0.0126; ref. 23), which correspond to the average level of nucleotide diversity observed in the _ph-d_–CG3835 region and the entire wapl region (17), respectively. These lower θ values make the test more conservative.

Identifying the Target of Selection.

After the identification of a putative selective sweep and localizing the approximate position of the selected site in the African sample, we screened the proximity of that site for candidate fixations that occurred along the D. melanogaster lineage. D. simulans and D. yakuba were used as outgroups. We also tested whether candidate fixations in intronic and 5′ flanking regions result in regulatory changes (i.e., differences in TFBs) by applying the MatInspector tool (50).

Age of Selective Sweep.

We used two methods to estimate the age of the selective sweep. First, we ran the algorithm of Przeworski (11) to obtain 2,000 successful matches. The time since the fixation of the selected site was determined by finding the mode of a histogram. Parametrization of the prior distributions is as follows: _N_e = 2.9 × 106, the mean mutation rate μ = 1.45 × 10 −9 per site per generation (assuming 10 generations per year), and crossing-over rate ρ = 0.48 × 10 −8. The first two estimates are inferred from the entire X chromosome dataset (13), whereas the last one is for the (local) wapl region under consideration (17).

Second, we assumed a star-like genealogy subsequent to the sweep (for the region surrounding the presumed selected site) and computed the time since the sweep as T = number of SNPs / (number of sites × sample size × mutation rate per year) (32, 33). Local mutation rates μ were estimated as μ = K / (2_t_), where K is the local divergence between D. melanogaster and D. simulans, and t is the divergence time between the two species. We assumed t = 2.3 million years (13, 51).

Acknowledgments.

We thank Anne Wilken and Simone Lange for excellent technical assistance; David De Lorenzo, Hideki Innan, Pleuni Pennings, and Molly Przeworski for valuable suggestions; Tim Sackton for help with the Drosophila assemblies; and Neel Randsholt for introducing us to polyhomeotic. This work was supported by the Deutsche Forschungsgemeinschaft Grant STE 325/7.

Footnotes

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The sequences reported in this paper have been deposited in the GenBank database (accession nos. AM943662–AM943805).

References

1. Ohno S. Evolution by Gene Duplication. Berlin: Springer; 1970. [Google Scholar]

2. Ohta T. Evolution and Variation of Multigene Families: Lecture Notes in Biomathematics 37. Berlin: Springer; 1980. [Google Scholar]

3. Clark AG. Invasion and maintenance of a gene duplication. Proc Natl Acad Sci USA. 1994;91:2950–2954. [PMC free article] [PubMed] [Google Scholar]

4. Lynch M, Force A. The probability of duplicate gene preservation by subfunctionalization. Genetics. 2000;154:459–473. [PMC free article] [PubMed] [Google Scholar]

5. Walsh JB. Population-genetic models of the fates of duplicate genes. Genetica. 2003;118:279–294. [PubMed] [Google Scholar]

7. Innan H. A two-locus gene conversion model with selection and its application to the human RHCE and RHD genes. Proc Natl Acad Sci USA. 2003;100:8793–8798. [PMC free article] [PubMed] [Google Scholar]

8. Maynard Smith J, Haigh J. The hitch-hiking effect of a favourable gene. Genet Res. 1974;23:23–35. [PubMed] [Google Scholar]

9. Innan H, Kim Y. Pattern of polymorphism after strong artificial selection in a domestication event. Proc Natl Acad Sci USA. 2004;101:10667–10672. [PMC free article] [PubMed] [Google Scholar]

10. Kim Y, Stephan W. Detecting a local signature of genetic hitchhiking along a recombining chromosome. Genetics. 2002;160:765–777. [PMC free article] [PubMed] [Google Scholar]

11. Przeworski M. Estimating the time since the fixation of a beneficial allele. Genetics. 2003;164:1667–1676. [PMC free article] [PubMed] [Google Scholar]

12. Nielsen R, et al. Genomic scans for selective sweeps using SNP data. Genome Res. 2005;15:1566–1575. [PMC free article] [PubMed] [Google Scholar]

13. Li H, Stephan W. Inferring the demographic history and rate of adaptive substitution in Drosophila. PLoS Genet. 2006;13:e166. [PMC free article] [PubMed] [Google Scholar]

14. Nurminsky DI, De Aguiar D, Bustamante CD, Hartl DL. Chromosomal effects of rapid gene evolution in Drosophila melanogaster. Science. 2001;291:128–130. [PubMed] [Google Scholar]

15. Schlenke TA, Begun DJ. Strong selective sweep associated with a transposon insertion in Drosophila simulans. Proc Natl Acad Sci USA. 2004;101:1626–1631. [PMC free article] [PubMed] [Google Scholar]

16. Orengo DJ, Aguadé M. Genome scans of variation and adaptive change: Extended analysis of a candidate locus close to the phantom gene region in Drosophila melanogaster. Mol Biol Evol. 2007;24:1122–1129. [PubMed] [Google Scholar]

17. Beisswanger S, Stephan W, De Lorenzo D. Evidence for a selective sweep in the wapl region of Drosophila melanogaster. Genetics. 2006;172:265–274. [PMC free article] [PubMed] [Google Scholar]

18. Schwartz YB, Pirrotta V. Polycomb silencing mechanisms and the management of genomic programs. Nat Rev Gen. 2007;8:9–22. [PubMed] [Google Scholar]

19. Deatrick J, Daly M, Randsholt NB, Brock HW. The complex genetic locus polyhomeotic in Drosophila melanogaster potentially encodes two homologous zinc-finger proteins. Gene. 1991;105:185–195. [PubMed] [Google Scholar]

20. Hodgson JW, et al. The polyhomeotic locus of Drosophila melanogaster is transcriptionally and posttranscriptionally regulated during embryogenesis. Mech Dev. 1997;66:69–81. [PubMed] [Google Scholar]

21. Grumbling G, Strelets V The FlyBase Consortium. FlyBase: Anatomical data, images and queries. Nucleic Acids Res. 2006;34:D484–D488. [PMC free article] [PubMed] [Google Scholar]

22. Dura J-M, et al. A complex genetic locus, polyhomeotic, is required for segmental specification and epidermal development in D. melanogaster. Cell. 1987;51:829–839. [PubMed] [Google Scholar]

23. Ometto L, Glinka S, De Lorenzo D, Stephan W. Inferring the impact of demography and selection on Drosophila melanogaster from a chromosome-wide DNA polymorphism study. Mol Biol Evol. 2005;22:2119–2130. [PubMed] [Google Scholar]

25. Hill WG, Robertson A. Linkage disequilibrium in finite populations. Theor Appl Genet. 1968;38:226–231. [PubMed] [Google Scholar]

26. Stephan W, Song YS, Langley CH. The hitchhiking effect on linkage disequilibrium between linked neutral loci. Genetics. 2006;172:2647–2663. [PMC free article] [PubMed] [Google Scholar]

27. Tajima F. Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics. 1989;123:585–595. [PMC free article] [PubMed] [Google Scholar]

29. Hudson RR, Kreitman M, Aguadé M. A test of neutral molecular evolution based on nucleotide data. Genetics. 1987;116:153–159. [PMC free article] [PubMed] [Google Scholar]

30. Innan H. The coalescent and infinite-site model of a small multigene family. Genetics. 2003;163:803–810. [PMC free article] [PubMed] [Google Scholar]

31. Jensen JD, et al. Distinguishing between selective sweeps and demography using DNA polymorphism data. Genetics. 2005;170:1401–1410. [PMC free article] [PubMed] [Google Scholar]

32. Slatkin M, Hudson RR. Pairwise comparisons of mitochondrial DNA sequences in stable and exponentially growing populations. Genetics. 1991;129:555–562. [PMC free article] [PubMed] [Google Scholar]

33. Ayala FJ, Balakirev ES, Saez AG. Genetic polymorphism at two linked loci, Sod and Est-6, in Drosophila melanogaster. Gene. 2002;300:19–29. [PubMed] [Google Scholar]

34. Glinka S, Ometto L, Mousset S, Stephan W, De Lorenzo D. Demography and natural selection have shaped genetic variation in Drosophila melanogaster: A multilocus approach. Genetics. 2003;165:1269–1278. [PMC free article] [PubMed] [Google Scholar]

35. Stephan W, Li H. The recent demographic and adaptive history of Drosophila melanogaster. Heredity. 2007;98:65–68. [PubMed] [Google Scholar]

36. Majewski J, Ott J. Distribution and characterization of regulatory elements in the human genome. Genome Res. 2002;12:1827–1836. [PMC free article] [PubMed] [Google Scholar]

37. Haddrill PR, Charlesworth B, Halligan DL, Andolfatto P. Patterns of intron evolution in Drosophila are dependent upon length and GC content. Genome Biol. 2005;6:R67. [PMC free article] [PubMed] [Google Scholar]

38. Halligan DL, Keightley PD. Ubiquitous selective constraints in the Drosophila genome revealed by a genome-wide interspecies comparison. Genome Res. 2006;16:875–884. [PMC free article] [PubMed] [Google Scholar]

39. Casillas S, Barbadilla A, Bergman CM. Purifying selection maintains highly conserved noncoding sequences in Drosophila. Mol Biol Evol. 2007;24:2222–2234. [PubMed] [Google Scholar]

40. Drosophila 12 Genomes Consortium. Evolution of genes and genomes on the Drosophila phylogeny. Nature. 2007;450:203–218. [PubMed] [Google Scholar]

41. Russo CAM, Takezsaki N, Nei M. Molecular phylogeny and divergence times of Drosophilid species. Mol Biol Evol. 1995;12:391–404. [PubMed] [Google Scholar]

42. Innan H. A method for estimating the mutation, gene conversion, and recombination parameters in small multigene families. Genetics. 2002;161:865–872. [PMC free article] [PubMed] [Google Scholar]

43. Teshima KM, Innan H. Neofunctionalization of duplicated genes under the pressure of gene conversion. Genetics. 2008 doi: 10.1534/genetics.107.082933. [PMC free article] [PubMed] [CrossRef] [Google Scholar]

44. Begun DJ, Aquadro CF. African and North American populations of Drosophila melanogaster are very different at the DNA level. Nature. 1993;365:548–550. [PubMed] [Google Scholar]

45. Rozas J, Sánchez-Del Barrio JC, Messeguer X, Rozas R. DnaSP, DNA polymorphism analyses by the coalescent and other methods. Bioinformatics. 2003;19:2496–2497. [PubMed] [Google Scholar]

46. Watterson GA. On the number of segregating sites in genetical models without recombination. Theor Pop Biol. 1975;7:256–276. [PubMed] [Google Scholar]

47. Tajima F. Evolutionary relationship of DNA sequences in finite populations. Genetics. 1983;105:437–460. [PMC free article] [PubMed] [Google Scholar]

48. Nei M. Molecular Evolutionary Genetics. New York: Columbia Univ Press; 1987. pp. 177–178. [Google Scholar]

49. McVean G, Awadalla P, Fearnhead P. A coalescent-based method for detecting and estimating recombination rates from gene sequences. Genetics. 2002;160:1231–1241. [PMC free article] [PubMed] [Google Scholar]

50. Cartharius K, et al. MatInspector and beyond: Promoter analysis based on transcription factor binding sites. Bioinformatics. 2005;21:2933–2942. [PubMed] [Google Scholar]

51. Li Y-J, Satta Y, Takahata N. Paleo-demography of the Drosophila melanogaster subgroup: Application of the maximum-likelihood method. Genes Genet Syst. 1999;74:117–127. [PubMed] [Google Scholar]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences