Introduction

Physical interactions (pairing and synapsis) between homologous chromosomes are prerequisite for ensuring accurate chromosome segregation during meiosis (Garcia-Muse and Boulton 2007). In C. elegans, a family of four related C2H2-type zinc-finger proteins has been demonstrated to play an essential role in these events (Phillips and Dernburg 2006; Phillips et al. 2005). The genes coding for these four proteins are organized as an operon, sharing a common set of transcriptional regulatory sequences, with zim-1 (zinc finger in meiosis 1) and him-8 (high incidence of males 8) being the most up- and downstream genes, respectively (Blumenthal et al. 2002). It appears that HIM-8 may interact with the “pairing center” (PC) of the X chromosome to promote the sex chromosome-specific homologous synapsis during meiosis (Phillips et al. 2005), whereas all three ZIM proteins (ZIM-1, ZIM-2, and ZIM-3) are required for meiotic pairing and synapsis of specific pairs of autosomes (Phillips and Dernburg 2006). Indeed, the activity of a member of the ZIM/HIM-8 protein family seems to be specifically required to mediate efficient meiotic pairing and synapsis of each chromosome pair (Phillips and Dernburg 2006). Nevertheless, neither the role of the PC nor the identity of the corresponding ZIM/HIM-8 family member is sufficient to explain the specificity of homologue recognition (Garcia-Muse and Boulton 2007; McKim 2007).

Genes involved in meiosis, such as zinc-finger proteins like HIM-8, diverge very rapidly during evolution (Englbrecht et al. 2004; Phillips et al. 2005). Thus, sequence comparisons with known meiotic factors have not revealed any obvious orthologues of HIM-8 (Phillips et al. 2005). In C. elegans, HIM-8 also functions outside of meiosis to antagonize EGL-13 Sox protein function (Nelms and Hanna-Rose 2006), a negative regulatory activity that is also shared by the HIM-8-related ZIM proteins (Sun et al. 2007). The fact that ZIM/HIM-8 proteins form an operon (Blumenthal and Gleason 2003) may partially explain their similar functions in the negative regulation of EGL-13 activity (Sun et al. 2007). However, compared with the ZIM proteins, HIM-8 has broader regulatory activity through its effects on transcription factors other than EGL-13 (Sun et al. 2007), suggesting the possibility of their rapid functional divergence of ZIM/HIM-8 proteins.

Phillips and Dernburg (2006) identified four ZIM/HIM-8 proteins in C. remanei, and five in C. briggsae, which had an additional ZIM protein (CbZIM-4) located between ZIM-1 and ZIM-2. Like their counterparts in C. elegans, these genes have the same orientation and are also arranged in tandem (Phillips and Dernburg 2006). The conserved organization of ZIM/HIM-8 proteins in different Caenorhabditis species suggests that orthologous proteins might perform similar functions. Coincidently, inactivation of the C. briggsae him-8 gene by RNAi results in F1 animals with a strong Him phenotype (Phillips and Dernburg 2006). However, no meiotic defects were observed following RNAi inactivation of Cbzim-4 (Phillips and Dernburg 2006). Further experiments are needed to determine whether the Cbzim-4 function is truly associated with meiosis.

A comparison of the C. elegans genes with those from the two other species indicates that zim/him-8 genes have likely arisen from a common ancestor through gene duplication, divergence, and selection (McKim 2007; Phillips and Dernburg 2006). The zim/him-8 family in C. elegans may represent an intermediate state in evolution, and has undergone recent changes in copy number. C. briggsae has gained an extra zim gene (Cbzim-4), which may have been derived from a recent gene duplication event (Phillips and Dernburg 2006). It is worth noting that detailed mechanisms driving the evolution of the zim/him-8 gene family still remain unclear, although Phillips and Dernburg (2006) observed that the C-terminal region shows high conservation among orthologues in different species, whereas the N termini are more closely related to the other ZIM/HIM-8 proteins within the same species. Here I show that gene conversion and positive selection are two primary mechanisms driving the evolution of the ZIM/HIM-8 protein family in Caenorhabditis species. Gene conversion in the N-terminal domain limits the divergence of paralogous proteins, whereas positive selection acting on the sequence regions believed to not have experienced gene conversions may be responsible for the functional diversification of ZIM/HIM-8 proteins.

Materials and Methods

Sequence Data

The zim/him-8 gene sequences from three Caenorhabditis species (C. elegans, C. remanei, and C. briggsae) were downloaded from the GenBank database (release 090707) in National Center for Biotechnology Information (NCBI). The accession numbers are Z82282 for C. elegans, BK005903 for C. remanei, and DQ498827 for C. briggsae. For accuracy, zim/him-8 genes in other nematodes that were only predicted using computer programs were not included in this dataset.

Multiple Sequence Alignment and Phylogenetic Tree Reconstruction

ZIM/HIM-8 protein sequences were aligned using the program E-INS-i implemented in MAFFT v6.5 (Katoh et al. 2005). The resulting protein alignment was subsequently employed to generate the alignment of corresponding coding DNA sequences. Neighbor-joining phylogenies, based on the parameters p-distance matrix and pairwise deletion of gaps/missing data, were reconstructed with MEGA v3.1 (Kumar et al. 2004). Bootstrap confidence values were obtained by 1000 replications.

Tests for Gene Conversion

Genetic Algorithms for Recombination Detection (GARD) was used to screen for putative recombination breakpoints. This method splits the alignment at every variable site and searches for discordant phylogenetic signals on either side using c-AIC (Kosakovsky Pond et al. 2006). The parameters HKY85 model of nucleotide substitution and beta-gamma rate variation with three rate classes were used for GARD analysis. As GeneConv is the program reported to have the least propensity for false-positives in detecting gene conversion (Posada 2002), GeneConv v1.81 (Sawyer 1989) was used with 100,000 permutations to assign probability to matching fragments in aligned DNA sequences for the zim/him-8 loci. In these two analyses, the genomic sequences of paralogous zim/him-8 genes within each species were separately aligned using MAFFT v6.5 (Katoh et al. 2005), and subsequently, the alignments were used to detect gene conversion events.

Analysis of Selection

Two statistics, Tajima (1989) D and Fu (1997) F S were used to examine whether the frequency distribution of polymorphic sites deviated from the neutral equilibrium expectation. Tests of neutrality for the zim/him-8 genes were conducted in DnaSP v4.10 (Rozas et al. 2003). Further, the PAML package v4.0 (Yang 2007) was utilized to test the hypothesis of positive selection in the zim/him-8 gene family. In this test, two pairs of models were contrasted. First, models M0 (one ratio) and M3 (discrete k = 3) were compared, using a test for heterogeneity between codon sites in the d N/d S ratio value, ω. The second comparison was M7 (beta) vs. M8 (beta + ω > 1); this is the most stringent test of positive selection (Anisimova et al. 2001). In such comparisons, positive selection is indicated if the models that allow for selection are significantly better than the null model (no selection) in the likelihood ratio test (LRT).

Results and Discussion

Analysis of Gene Conversion

Upon alignment based on nucleotide difference, the N-terminal regions of paralogous genes showed relatively higher sequence similarity than those of orthologues (Fig. 1). To determine whether the similarity was the consequence of recombination, GARD was employed to identify single recombination breakpoint (SRB) in the aligned coding DNA sequences. As expected, an SRB located at the nucleotide position 1146 was identified in the zim/him-8 gene family (Fig. 1). The alignment was then cut at this point and subsequently used to reconstruct separate phylogenies. Phylogenetic trees of the two sequence regions show significant phylogenetic incongruence (p < 0.01; Fig. 2). In the C-terminal region phylogeny (1146-end), orthologous sequences, e.g., Cezim-1, Cbzim-1, and Crzim-1, cluster together (Fig. 2b). In contrast, the N-terminal region phylogeny (1–1145) reveals that paralogous zim genes within each species cluster together, e.g., Cezim-1, Cezim-2, and Cezim-3, as well as Crzim-1 and Crzim-3 (Fig. 2a). Thus, the significant discrepancy between phylogenetic trees reconstructed using the separate sequences up- and downstream of the breakpoint constitutes powerful evidence for recombination (Holmes et al. 1999).

Fig. 1
figure 1

Alignment of all zim/him-8 coding DNA sequences from C. elegans, C. briggsae, and C. remanei using the program MAFFT v6.5. Dashes represent alignment gaps. Identity with the master sequence is indicated by dots, except at gap positions. Dark shading indicates 100% conserved nucleotides. The black arrow designates the putative single recombination breakpoint detected by GARD. Open triangles and arrows indicate the four (47, 184, 185, and 186) and six (444, 448, 449, 517, 568, and 620) amino acid sites that are under the influence of positive selection for the N- and C-terminal sequence regions, respectively. Black triangles indicate the three mutation positions (me4, mn253, and e1489) in the him-8 gene of C. elegans. Asterisks denote the two zinc fingers

Full size image
Fig. 2
figure 2

Phylogenetic trees reconstructed using a N-terminal and b C-terminal sequences of ZIM/HIM-8 proteins from C. elegans, C. briggsae, and C. remanei. The scale is amino acid substitutions per site. Bootstrap values >60% are shown beside nodes. The C. elegans gene C02F5.12, which possesses two C2H2 zinc-finger domains, was used as outgroup sequence to root the trees. Cb, Caenorhabditis briggsae; Ce, C. elegans; Cr, C. remanei

Full size image

To further establish whether gene conversion has occurred, an analysis was conducted using the program GeneConv (Sawyer 1989). As gene conversion between zim/him-8 genes could only occur during meiotic division in the ancestor of a species, and not between species, only pairwise comparisons of paralogous sequences in each species were considered. In this analysis, GeneConv detected six highly significant global inner fragments in C. elegans, as well as three each in C. remanei and C. briggsae (p < 0.05; Table 1). Moreover, one significant global outer fragment, which is evidence of a past gene conversion event, was detected in Cehim-8 (the corresponding fragments are marked in Supplementary Figs. S1–S3). It is observed that the lengths of inferred conversion tracts vary extensively, ranging from 29 bp for Cbzim-1/Cbzim-2 to 524 bp for Cezim-1/Cezim-2 (Table 1). As gene conversion occurs at the genomic level, it may be expected to occur in both exons and introns. Indeed, this is the case for the Crzim-1/Crzim-3 gene pair, wherein the whole first intron also appears to have experienced gene conversion (Supplementary Fig. S2). The GARD analysis identified seven, five, and two recombination breakpoints in C. elegans, C. remanei, and C. briggsae, respectively, which is generally consistent with the GeneConv results (Supplementary Figs. S1–S3). Together, these results strongly indicate the occurrence of gene conversion. However, in C. remanei and C. briggsae, there was no evidence of gene conversion events either for him-8 with zim genes or for Crzim-2 and Cbzim-4 with other zim paralogues within species.

Table 1 Gene conversion analysis for zim sequences from three Caenorhabditis species
Full size table

The nucleotide divergence between zim genes within species was calculated for every exon and intron (Supplementary Table S1) using K-estimator v6.0 (Comeron 1995). The numbers of synonymous (Ks) and nonsynonymous (Ka) substitutions per site in exon 1 were 0.071 and 0.024 for Cezim-1/Cezim-2, and 0.094 and 0.015 for Crzim-1/Crzim-3, and the K value of intron 1 for Crzim-1/Crzim-3 was 0.052. Similar results were also found in exon 2, especially in sequence regions predicted to be involved in gene conversion. The Ks and Ka values for such regions are significantly lower than for other regions not implicated in gene conversion (Supplementary Table S1). From these data, it is clear that some sequences in the N-terminal region, such as exon 2 and/or exon 1, have been almost completely homogenized. The patterns of homogenization observed might be the results of repeated gene conversion events, rather than functional constraints on protein sequence content, an inference supported by the observation of strong sequence homogenization at neutral sites (Noonan et al. 2004). The homogenization is not likely to have been due to unequal crossing-over, because the adjacent sequence regions have not been homogenized (Zhao et al. 1998).

Kruithof et al. (2007) showed that amino acid sequences that became identical after gene conversion were most likely to be encoded by the same triplets compared to sequences that became identical through convergent evolution. Accordingly, the codon usage of conserved amino acids in the identified sequence regions was analyzed to determine if the above results could be attributed to gene conversion rather than convergent evolution. On average, identical codons were significantly (Mann-Whitney test, p = 0.001) more frequently used to encode conserved amino acids in the identified regions, compared to other sequence regions (93.6% vs. 41.7%; Table 2). As a consequence, gene conversion appears to be the most reasonable explanation for the observed results.

Table 2 Comparison of codon usage pattern for conserved amino acids between sequence regions involved in and those not involved in gene conversion
Full size table

The N-terminal portion of ZIM/HIM-8 proteins contains numerous potential modification sites but lacks obvious structural motifs (Phillips and Dernburg 2006). Moreover, several but not all paralogous zim/him-8 genes appear to have experienced gene conversion events. Thus, I infer that the high degree of sequence similarity in the identified regions between special paralogues could not be due to functional constraint on particular regions of each protein, because functional constraint at the protein level does not act on synonymous sites and, thereby, they show a high Ks value relative to nonsynonymous sites (Miyata et al. 1980). This inference is supported by the observations of very low Ks and Ka values in the identified sequence regions (Supplementary Table S1). Moreover, functional constraints are likely to act on the same or similar regions in orthologous proteins, resulting in high sequence conservation among closely related species (Noonan et al. 2004). For the zim/him-8 genes, the sequence identity in the N-terminal portion between paralogous genes is substantially higher than that between orthologous genes (Supplementary Table S2), implicating gene conversion rather than functional constraints. Gene conversion homogenizes the sequence of multigene family members, and selection subsequently eliminates individuals in which gene conversions occurred in regions where it is important that different genes have different sequences. Therefore, the observation that the sequence regions, particularly exon 2, have been converted between the zim-1/zim-3 paralogues suggests that selection has not eliminated the individuals in which these conversions occurred. It can therefore be inferred that there is no selective pressure to keep differences in these regions of the two genes. Another reasonable explanation is that the zim genes, especially zim-1 and zim-3, may contain short, conserved DNA segments that are particularly prone to gene conversion (Kruithof et al. 2007). However, him-8 genes are not as conserved as their zim paralogues in the N-terminal region, where only one possible past gene conversion in Cehim-8 was detected, perhaps indicating the functional specificity of him-8 genes (Phillips and Dernburg 2006).

Gene conversion, a type of intrachromosomal nonreciprocal recombination that may lead to genetic homogenization, is an important evolutionary mechanism in many families of viruses and eukaryotes (Chen et al. 2007; Kruithof et al. 2007). Gene conversion is more likely between similar paralogues, as increased sequence similarity facilitates ectopic strand invasion (Elliott et al. 1998). Thus, tandem gene arrays are particularly prone to gene conversion (Drouin et al. 1999). When paralogues cluster on the same chromosome in direct orientation, and at a distance of <40 kb, gene conversion events are particularly likely (Mondragon-Palomino and Gaut 2005), as is the case for zim genes in the three Caenorhabditis species studied here. The zim/him-8 family members are close together (224–1855 bp) on the same chromosome, and have the same orientation (Phillips and Dernburg 2006). Thus, gene conversion is very likely to have been an important evolutionary mechanism for the zim/him-8 gene family, although it may not be uniformly strong across species (Table 1).

Gene Conversion Limits the Divergence of Paralogous zim Genes

Within a gene family, gene conversion can homogenize genetic variation among family members and thereby retard divergence among paralogous genes (Kruithof et al. 2007), although it can also generate diversity by reassorting variants among paralogues (Ohta 1997). Moreover, the rate of gene conversion is thought to be strongly associated with both the degree of homogenization between paralogous sequences and the extent of divergence between orthologous sequences (Hurles et al. 2004). To validate this speculation, the K value (the average number of nucleotide divergence between sequences) was calculated using DnaSP v4.10 (Rozas et al. 2003) to assess the divergence between zim/him-8 sequences. The results show that the K values between genes involved in gene conversion are significantly (p < 0.001) lower than those between genes not involved in gene conversion for each species (Table 3), a result supporting the proposal that gene conversion limits the divergence of paralogous sequences. The C. elegans zim genes, in particular, are very similar to each other, indicating that gene conversions in this species have been more frequent than in the other two species. Moreover, greater sequence divergence was observed between orthologous genes compared to paralogous genes (Table 4). The average extent of sequence divergence, as measured by K values, was as follows: him-8 (0.542) > zim-2 (0.519) > zim-3 (0.469) > zim-1 (0.457). These results suggest that the zim/him-8 genes have greatly diverged from each other, since descent from a common ancestor of the three Caenorhabditis species.

Table 3 Variance analysis of nucleotide divergence between gene sequences involved in and those not involved in gene conversion
Full size table
Table 4 Analysis of sequence divergence between orthologous genes
Full size table

Intron number variation was also observed among zim/him-8 genes (Table 5; see also Phillips and Dernburg 2006), with six and five introns for the zim and him-8 genes in C. elegans, respectively. In contrast, with two exceptions, C. briggsae and C. remanei zim/him-8 genes have five introns each. However, the C. briggsae zim-2 gene has only four introns. Because nematode introns are lost at a very high rate during evolution, almost 400-fold higher than that in mammals (Cho et al. 2004), it is more parsimonious to propose that intron losses have occurred for the zim genes after the speciation of C. elegans and C. briggsae. Indeed, alignment of zim/him-8 genomic sequences revealed that Cehim-8 and Cbzim-2 have probably lost the third and first intron, respectively (Supplementary Figs. S1, S3). On the contrary, the additional intron found in Crhim-8 (Supplementary Fig. S2) may be due to intron gain rather than independent instances of intron losses in Cehim-8 and Cbhim-8, although intron gains are far less common than losses in nematode genes (Cho et al. 2004). Nonetheless, it is observed that with one exception (Cbzim-2), the intron number is conserved for zim genes in each species, with six introns in C. elegans and five in the other two species, although the intron length varies extensively (Supplementary Table S3).

Table 5 Intron number variation in zim and him-8 genes
Full size table

Exon lengths in the paralogous zim genes, especially for Cezims, have diverged significantly, which may partially explain their sequence divergence at the protein level. However, the length of exon 1 is highly conserved (about 99 bp) in all three species. In C. elegans, the two zinc-finger domains are mainly encoded by the fifth exon, whose length is 195 bp in all Cezim genes. Gene conversions between Cezim paralogues are primarily found in the second exon. Coincidently, lengths of the corresponding exon vary slightly (Supplementary Table S3). Similar results are also found in C. remanei, for which the fourth exon codes for the two zinc fingers, and gene conversions have occurred in exons 1 and 2, and in intron 1, between Crzim-1 and Crzim-3. Again, the lengths of the above regions are nearly identical. Accordingly it seems that gene conversion and the zinc-finger domains are responsible for maintaining the lengths of the corresponding sequence regions. However, significant divergence of exon lengths and numbers is observed in C. briggsae (Supplementary Table S3, Table 5). Therefore, further experiments are needed to determine whether such divergences are relevant to the functional diversification of Cbzim/Cbhim-8 genes.

Positive Selection Analysis

Some studies have revealed that the most complete record of gene conversion might be most evident in gene families commonly subjected to positive selection (Goldstone and Stegeman 2006; Zhang 2008). Hence, it is of interest to test whether positive selection has also acted on the nematode zim/him-8 gene family. Given that phylogenetic trees reconstructed using the N- (1–1145) and C-terminal (1146-end) sequences are significantly discordant because of the existence of gene conversions, it is reasonable to investigate the corresponding two regions separately. As expected, the results of neutrality test show that both these regions may be under positive selection. The Tajima (1989) D values are 2.28 (p < 0.05) and 3.49 (p < 0.001) for the N- and C-terminal region, respectively. However, the Fu (1997) Fs values (0.321 and 0.662) are not significant (p > 0.05). Importantly, application of the methods of Yang et al. (2000) to each region revealed strong heterogeneity in ω among codon sites (Table 6), which was concordant with a model of adaptive evolution where some sites might undergo amino acid changing substitutions at a high rate.

Table 6 Results of positive selection analysis using a variety of codon substitution models
Full size table

The estimation of positive selection in the N- and C-terminal regions is based on the separate phylogenies presented in Fig. 2a and b. Table 6 shows that for both these regions, models that allow for positively selected sites fit the data significantly better than those that do not permit positive selection. In the maximum likelihood (ML) analysis, the comparison of M3 (discrete, k = 3) vs. M0 (one ratio) reveals that ω is not uniformly distributed along the zim/him-8 coding DNA sequences, and about 4.5% and 6.1% of codon sites in the N- and C-region may be under the influence of positive selection (ω = 1.16 and 18.0, respectively). Compared with the M7 (beta) model, the M8 model suggests that ~2.5% and ~3.0% of codons fall in a category with estimated ω values of 1.73 and 2.76 in the N- and C-regions, respectively, a result indicative of strong positive selection, while other codons fall into ω categories with values between 0.04 and 0.34 and between 0.03 and 0.30, respectively.

Based on the Bayesian posterior probabilities, four and three codon site candidates for positive selection were identified in the N-terminal region for the M3 and M8 models, respectively, of which the posterior probability of site 185 is >0.95 in both analyses. Codon site 47, which is found by comparison of M3 and M0, is located in exon 1, while the other three sites (184–186) are in exon 2. Further analysis shows that the localizations of codon sites 184, 185, and 186 are outside of the sequence regions that were supposed to have experienced gene conversions (Supplementary Figs. S1–S3). In the C-terminal region, six and five codon sites were identified to be under positive selection, with site 517 within the zinc-finger domain (Table 6; Fig. 1), although the posterior probabilities for some of sites are lower than the specified 95% threshold in one of the models. Under positive selection, these codon sites may, however, strongly contribute to the functional divergence of the zim/him-8 family members.

Genetic diversity in genes under strong selection is generally expected to be very low. On the contrary, neutral (unselected) genes are expected to exhibit more diversity than selected genes, although diversity of neutral genes can be reduced by bottlenecks or random genetic drift (Mita et al. 2007; Sengupta et al. 2007). Table 4 shows the average K value of orthologues, revealing considerable nucleotide diversity in the coding sequence region for all zim/him-8 genes. If the full-length gene sequences are taken into account, these differences are more apparent (Table 5 and Supplementary Figs. S1–S3). Considering that frequent gene conversions have greatly limited the divergence between paralogues within each species, the extent of sequence divergence between orthologues at each locus would probably increase. Furthermore, positive selection has clearly acted on the zim/him-8 family members (Table 6), which may introduce more nucleotide diversity in evolution. Therefore, it appears that the bottleneck effects are unlikely to be the cause of the observed results. In addition, previous research (Moeller and Tiffin 2005; Sengupta et al. 2007) showed that the Fu (1997) Fs values are particularly sensitive to demographic history, in the sense that if only Fs is significantly negative while the other statistics are not significant, then it is more likely to be due to population expansion rather than natural selection. However, in this study, the Fs values of the four loci are all positive, and the Tajima (1989) D values are significantly positive. Accordingly, the results can be better explained as positive selection, rather than bottleneck effect and/or population expansion.

Shannon et al. (2003) demonstrated that in zinc-finger genes, mutations in the zinc-finger could alter the specificity of their DNA binding sites but would not perturb their regulatory function or interaction with binding partners. In C. elegans, the four ZIM/HIM-8 proteins bind specifically to the PC region of six chromosomes (HIM-8 for the X chromosome, ZIM-2 for V, ZIM-1 for II and III, and ZIM-3 for I and IV [Phillips and Dernburg 2006; Phillips et al. 2005]). In addition, it was found that the C-terminal sequence following the zinc-finger might also plays important roles in binding to the PC region of the sex chromosome for HIM-8 in C. elegans (Phillips et al. 2005). Therefore, the identification of one site (517) in the first zinc-finger domain and two sites (568 and 620) in the sequence region following the zinc finger that are under positive selection may be, at least partially, responsible for their binding specificity of ZIM/HIM-8 proteins.

The wild-type him-8 gene product may act antagonistically to transcription factors like EGL-13 (Nelms and Hanna-Rose 2006; Sun et al. 2007). The C-terminal C2H2 zinc fingers, rather than the N-terminal sequence, play an important role in the mechanism of him-8 function in suppression of egl-13 defects (Nelms and Hanna-Rose 2006). For functioning in meiosis, it was found that one mutation in the N-terminal portion (me4, S85F) and two mutations in the zinc-fingers (mn253, G259R; e1489, C281Y) of the C. elegans CeHIM-8 can cause severe consequences for X chromosome crossing-over and segregation. Thus, the mutation of these sites is likely to eliminate the normal meiotic function of the him-8 gene (Phillips et al. 2005). On the alignment (Fig. 1), of the three mutational amino acid (AA) sites, the first and third sites are the invariant serine (S) and cysteine (C), respectively. The second site (523) is the nearly identical glycine (G), except in the C. briggsae HIM-8. Thus, the invariant AA sites may likely play crucial roles in maintaining the normal meiotic function of ZIM/HIM-8 proteins.

Conclusion

The present study shows that the homogenizing force of gene conversion and the diversifying force of positive Darwinian selection have been operating on a tandem array of zim duplicates, a phenomenon of co-occurrence of concerted evolution and adaptive diversifying evolution now documented in an increasing number of genes and gene families (Beisswanger and Stephan 2008; Goldstone and Stegeman 2006; Zhang 2008). In the zim/him-8 gene family, the N-terminal portions of zim duplicates, particularly between zim-1 and zim-3, are very similar to each other, probably because of the action of frequent gene conversion. Despite this homogenizing force, the functions of the family members have begun to diverge. Experiments show that each of the zim/him-8 genes mediates the pairing and synapsis of a specific chromosome(s) during meiosis (Phillips and Dernburg 2006; Phillips et al. 2005), implying that these genes have evolved specific functions after duplication. Here, I have provided evidence that this functional divergence is driven by positive selection. Innan (2003) predicted that neofunctionaliztion is unlikely in the presence of gene conversion, unless selection is very strong. Hence, it is no surprise that many genes and gene families (e.g., the zim/him-8 gene family) are found to be simultaneously under the influence of homogenizing as well as positive selection forces. Another common feature of these genes is that they are arranged in tandem. Thus, it will be interesting to perform detailed analyses to investigate whether tandemly arranged genes generally have a greater propensity to exchange genetic information and undergo rapid, adaptive evolution.