Birds have female heterogamety with Z and W sex chromosomes. These evolved from different autosomal precursor chromosomes than the mammalian X and Y. However, previous work has suggested that the pattern and process of sex chromosome evolution show many similarities across distantly related organisms. Here we show that stepwise restriction of recombination between the protosex chromosomes of birds has resulted in regions of the chicken Z chromosome showing discrete levels of divergence from W homologs (gametologs). The 12 genes analyzed fall into three levels of estimated divergence values, with the most recent divergence (dS = 0.18–0.21) displayed by 6 genes in a region on the Z chromosome corresponding to the interval 1–11 Mb of the assembled genome sequence. Another 4 genes show intermediate divergence (dS = 0.27–0.38) and are located in the interval 16–53 Mb. Two genes (at positions 42 and 50 Mb) with higher dS values are located proximal to the most distal of the 4 genes with intermediate divergence, suggesting an inversion event. The distribution of genes and their divergence indicate at least three evolutionary strata, with estimated times for cessation of recombination between Z and W of 132–150 (stratum 1), 71–99 (stratum 2), and 47–57 (stratum 3) million years ago. An inversion event, or some other form of intrachromosomal rearrangement, subsequent to the formation of strata 1 and 2 has scrambled the gene order to give rise to the nonlinear arrangement of evolutionary strata currently seen on the chicken Z chromosome. These observations suggest that the progressive restriction of recombination is an integral feature of sex chromosome evolution and occurs also in systems of female heterogamety.
DIFFERENTIATED sex chromosomes bear testimony of an ancestral autosomal state in the form of homologous sequences shared between the X and Y chromosomes. In the absence of recombination, such homologous sequences will gradually diverge. Sequence comparison of sex chromosome homologs, or gametologs (García-Moreno and Mindell 2000), can thus shed light on both sex chromosome evolution and the molecular evolutionary forces that differentially affect X and Y chromosomes (Hurst and Ellegren 1998; Charlesworth and Charlesworth 2000; Li et al. 2002; Marais and Galtier 2003; Charlesworth et al. 2005). For example, for neutral sequences and assuming a molecular clock, the observed divergence between gametologs can be used to estimate the time when recombination was effectively suppressed between protosex chromosomes. While this type of analysis is increasingly difficult for noncoding sequences as divergence time increases, the rate of synonymous substitution in coding sequences (dS) shared between sex chromosomes provides a useful means for studies of how and when sex chromosome differentiation was initiated.
Lahn and Page (1999) found a distinct divergence pattern of human gametologs with respect to the location of genes on the X (but not Y) chromosome. Specifically, genes located close to the pseudoautosomal region (PAR) on the terminal part of Xp had the lowest X–Y divergence while genes on Xq had the highest. Between these regions, they found two segments on Xp with intermediate divergence, giving a total of four “evolutionary strata.” Each stratum was characterized by rather uniform dS estimates, with stratum 1 on Xq corresponding to 240–320 million years of divergence and stratum 4 at Xp of 20–30 million years of divergence. From an analysis of the finished sequence of the human X chromosome, a fifth stratum has subsequently also been suggested (Ross et al. 2005). These observations suggest that sex chromosome differentiation occurred in a stepwise fashion, with more or less simultaneous cessation of recombination in large blocks of the protosex chromosomes (Charlesworth et al. 2005).
In birds, females are the heterogametic sex with Z and W sex chromosomes; males are ZZ. Studies of sex chromosome evolution in birds and other systems with female heterogamety are important because they offer independent replication of observations from X–Y species (Ellegren 2000). One such example is the demonstration of evolutionary strata on the chicken (Gallus gallus) Z chromosome (Ellegren and Carmichael 2001; Lawson Handley et al. 2004), suggesting that discontinuous recombination suppression over large chromosomal segments is a general feature of sex chromosome evolution, irrespective of male or female heterogamety. However, information on the evolutionary history of avian sex chromosomes is sparse, with only five incomplete sequenced chicken genes analyzed by Lawson Handley et al. (2004). Here, we present an extended analysis of how and when the avian Z and W chromosomes diverged, on the basis of data from an additional seven gametologous gene pairs shared between these chromosomes. The new analysis reveals at least three evolutionary strata on the chicken Z chromosome that, in contrast to the situation on the human X chromosome, are not linearly ordered with increasing divergence by increasing distance from the PAR.
MATERIALS AND METHODS
W-linked chicken genes were obtained from three sources. First, we extracted all genes assigned to the W chromosome in the chicken genome assembly version 2.1 (May 2006) (http://www.ensembl.org/Gallus_gallus/index.html). Second, we used published information on two W-linked genes from Wahlberg et al. (2007). Third, we surveyed chicken microarray expression data (Ellegren et al. 2007) from male and female somatic and reproductive tissue for genes of unknown chromosomal location consistently showing log2 fivefold higher expression level in females than in males, representing putative W chromosomal genes. W linkage of these genes was subsequently confirmed by PCR amplification of male and female DNA (see supplemental information).
The sequences of all W-linked genes were BLASTed to the chicken genome sequence, which in all cases revealed a closely related paralog (gametolog) on the Z chromosome. Map location (physical position) of these genes on the Z was taken from the genome assembly.
Estimates of divergence:
The full coding sequences for chicken genes were downloaded through BIOMART (http://www.biomart.org). After translation to protein sequences, gametologous gene pairs were aligned by ClustalW and then checked by eye, and the corresponding DNA sequence alignments were then used for further analysis. The synonymous (dS) and nonsynonymous (dN) divergence values were estimated by the maximum likelihood method, using CODEML in the PAML package version 3.15 (Yang 1997), after removing poorly aligned sequences. Alignments are available upon request. Our divergence estimates differ from those of Lawson Handley et al. (2004), because (although we used the same sequences) we used a different alignment method (aligning protein rather DNA sequences) and calculated divergence estimates differently (using maximum likelihood rather than the Nei and Gojobori method with Jukes–Cantor correction).
Estimating divergence times:
The fossil record of birds is poor and hence there are few data points available for calibrating an avian molecular clock. Birds and mammals diverged 320 million years ago (MYA) (Hedges 2002). Substitution rates clearly vary among avian genes (Webster et al. 2006); in a comparison of >7500 chicken–human orthologs, a median dS of 1.66 was obtained (International Chicken Genome Sequencing Consortium 2004). This translates into a mean lineage-specific rate of 2.6 × 10−9/site/year since the split of the lineages leading to chicken and humans (divergence is twice the mean lineage-specific rates). It has been suggested that chicken and turkey (Meleagris gallopavo) diverged ∼30 MYA (Dimcheff et al. 2002). Axelsson et al. (2004) observed 10% divergence between chicken and turkey autosomal introns, which corresponds to lineage-specific rates of 1.8 × 10−9. A preliminary analysis of orthologous autosomal single-copy coding sequences of chicken and turkey indicates that dS is 0.08, giving a rate of 2 × 10−9 (data not shown). Due to male-biased mutation, the higher mutation rate in the male than in the female germline (Ellegren 2007), neutral Z and W chromosome sequences will evolve with different rates. The Z chromosome is in the male germline two-thirds of the time, and thus, assuming a 1:1 sex ratio, its mutation rate is two-thirds of the male mutation rate plus one-third of the female rate. The W chromosome is only transmitted through the female germline and has therefore the female mutation rate. Previous work in birds has indicated that the male mutation rate is two to four times higher than the female rate (Bartosch-Härlid et al. 2003; Axelsson et al. 2004; Berlin et al. 2006). In galliforms, the Z chromosome rate is slightly higher than the autosomal rate, whereas the W chromosome rate is about half the autosomal rate (Axelsson et al. 2004). If we assume, from the estimates mentioned above, an autosomal lineage-specific rate of 2.5 × 10−9, the combined rate of Z–W divergence may be ∼3.8 × 10−9 (2.6 + 1.2 × 10−9). This rate was used to estimate divergence times between gametologs on the basis of estimates of dS.
The most recent chicken genome assembly contains eight genes on the W chromosome that have homologous sequences on chromosome Z (ATP5A1, CHD1, HINT, KCMF1, NIPBL, SMADS, SPIN, and UBAP2). Two additional genes, ZFR and ZNF532, have recently been reported to map to the chicken W chromosome (Wahlberg et al. 2007). Moreover, from chicken microarray data (Ellegren et al. 2007) we identified two genes, MIER3 and hnRNPK, which consistently show strong female-biased expression across tissues, indicative of W-linkage, confirmed by PCR with W-specific primers (supplemental Figure 1). A closely related gene copy on the Z chromosome was also identified for the four latter genes by BLAST searches (percentage amino acid identity of 95.6% for ZFR, 89.6% for ZNF532, 92.0% for MIER3, and 99.3 for hnRNPK). In analogy with the nomenclature for gametologous gene pairs on the avian sex chromosomes (Ceplitis and Ellegren 2004), we refer to these gene pairs as ZFRW/ZFRZ, ZNF532W/ZNF532Z, MIER3W/MIER3Z, and hnRNPKW/hnRNPKZ.
The conservation between gametologs is generally high, with dN (the non-synonymous substitution rate) in the range of 0.003–0.068 (Table 1, and dN/dS of 0.01–0.32). The only exception is HINT for which the W-linked gametolog HINTW is known to have evolved rapidly under positive selection (Ceplitis and Ellegren 2004). Silent divergence (dS) ranges from 0.12 to 0.57 (Table 1). Divergence estimates cluster in three different groups, corresponding to dS = 0.18–0.22 (six genes), 0.28–0.36 (four genes), and 0.50–0.52 (two genes) (Figure 1). These genes' dS values differ significantly. In post hoc tests, the group located in the Z region from 1–11 Mb (stratum 3) have significantly lower divergence than the group located more distally (stratum 2) (t = 5.09, P = 0.0009, two-tailed). The third group (stratum 1), despite being located among the latter genes, have a significantly higher dS and thus being the earliest region to have lost recombination (t = 5.42, P = 0.0056). We therefore conclude that the different regions stopped recombining at three different times.
Subsequent to the formation of the two oldest clusters or strata (strata 1 and 2), a chromosomal inversion in the lineage leading to the chicken seems to have erased the orderly pattern to give a nonperfect correlation between gametologous divergence and position on the Z chromosome. Critical to this interpretation is the distal location and relatively moderate divergence of KCMF1Z/KCMF1W. One potential alternative possibility would be that this gene pair has undergone gene conversion that homogenized the sequences of otherwise independently evolving Z and W chromosomes copies. Gene conversion between differentiated sex chromosome gametologs seems rare but has been reported for several mammals (Pecon Slattery et al. 2000; Ross et al. 2005).
GC content and recombination rate tend to be positively correlated in vertebrate genomes (Galtier 2003; Meunier and Duret 2004), including in birds (Webster et al. 2006), due to biased gene conversion. For sex chromosomes, this leads to the predictions of the highest GC for genes that recombine in both sexes (the pseudoautosomal region), intermediate GC for genes that recombine in just one sex (X linked/Z linked), and lowest GC for nonrecombining genes (Y linked/W linked), predictions that are supported by empirical data (Iwase et al. 2003; Backström et al. 2005). A comparison of GC content in gametologous genes can be used to reveal possible cases of gene conversion (Marias and Galtier 2003). Figure 2 shows the GC content of all gametologs on the chicken Z and W chromosomes. All Z-linked gene sequences, except HINTZ (see discussion), have higher GC3 than their W-linked gametologs. This suggests that interchromosomal gene conversion has not played an important role in the evolution of avian gametologs since they ceased to recombine. The high GC of HINTW is in accordance with the fact that while HINTZ is a single-copy gene, HINTW has been amplified on the avian W chromosome to become a multicopy gene (in the chicken there are ≈40 copies, Hori et al. 2000) and undergoes frequent intrachromosomal gene conversion on the W (Backström et al. 2005). HINTW became amplified after the original HINT gene ceased to recombine, hence all HINTW copies are orthologs to HINTZ.
Divergence estimates for gametologous genes in the three different evolutionary strata on the chicken Z chromosome can be used to estimate when recombination between homologs ceased. As diverging sex chromosomes are differentially affected by male and female mutation rates, we assumed a molecular clock that took into account sex-specific mutation rates for the Z and the W chromosome lineage (3.8 × 10−9/site/year; see materials and methods). This gives estimates of divergence times for stratum 1 of 132–150 MYA; for stratum 2, 71–99 MYA; and for stratum 3, 47–57 MYA, respectively (Table 1).
The evolutionary history of genes shared between the chicken Z and W chromosomes is consistent with a process of sex chromosome evolution including arrest of recombination at different times. Moreover, the consistent observation of a Z-linked gametolog for genes on the chicken W chromosome suggests that the protein-coding content of avian W largely derives from a degenerating protosex chromosome, without the addition of autosomal sequences, through processes that include translocation or transposition. The consecutive expansions of the nonrecombining region on the avian Z chromosome mirrors the situation in both mammals (Lahn and Page 1999; Sandstedt and Tucker 2004; Ross et al. 2005) and dioecious plants of the genus Silene (Filatov 2005; Nicolas et al. 2005; Zluvova et al. 2005; Bergero et al. 2007), two systems with male heterogamety. The observation of this pattern in highly divergent phylogenetic lineages of male as well as female heterogamety suggests that progressive restriction of recombination generating evolutionary strata is an inherent theme of sex chromosome evolution (Charlesworth et al. 2005). Two possible, but not mutually exclusive, mechanisms for such segmental steps of sex chromosome divergence are inversions on Y (W) and recombination restriction without inversions. Neither of these alternatives can currently be tested, as the poor assembly of the repeat-rich chicken W chromosome precludes gene-order analysis, and no antagonistic effects of W genes are known. When a W chromosome physical map becomes available, comparisons with the Z gene order should reveal any inversions that have occurred.
We cannot exclude the possibility that additional strata exist in those regions where we lack data from gametologous genes (17–39 and 53–75 Mb, respectively). Moreover, it is theoretically possible that the heterogeneity in divergence estimates seen within each inferred stratum reflects the existence of additional strata within these regions. However, the stochastic variation in divergence estimates from finite sequences coupled with mutation rate variation on a local scale, as previously demonstrated for the avian sex chromosomes (Berlin et al. 2006), seem to be more likely explanations to such heterogeneity. In any case, the interpretation of three evolutionary strata on the chicken Z chromosome should be seen as a minimum number. Moreover, although divergence estimates cluster in three different intervals, we cannot formally distinguish between a process of discrete events of cessation of recombination and a process of more gradual recombination restriction.
In mammals, the evolutionary strata identified on the human X chromosome are also seen on the mouse X chromosome (Sandstedt and Tucker 2004). However, these strata are not ordered linearly along the mouse X, similar to what we observe for the chicken Z. This has been explained by X chromosome rearrangements in the rodent lineage subsequent to the split from primates, breaking up the strict correlation between divergence of X–Y gametologs and physical position on the X (Sandstedt and Tucker 2004). A similar explanation is also reasonable for birds. While the body of evidence suggests that the gene content of the avian Z chromosome has remained conserved during the evolution of extant birds (Shetty et al. 1999; Backström et al. 2006; Itoh et al. 2006), comparative mapping indicates that intrachromosomal rearrangements have occurred several times during avian Z chromosome evolution (Shibusawa et al. 2004a,b; Backström et al. 2006; Griffin et al. 2007), including within Galliformes (chicken and its allies). The scrambled location of gametologous genes on the chicken Z chromosome, with respect to Z–W divergence, is consistent with this and suggests at least one inversion event that distinguishes the contemporary gene order in the chicken from the ancestral organization of the chromosome. One possibility is that a fragment including HINTZ, CHD1Z, and KCMF1Z was inverted after the evolution of strata 1 and 2. However, there are several alternative scenarios that cannot be excluded.
The estimated times of origin of evolutionary strata on the chicken Z chromosome are of relevance for temporal aspects of avian evolution. Highly differentiated sex chromosomes are characteristic of birds from the dominant clade of contemporary bird lineages, the Neognathae (everything but ratites and tinamous, which belong to Paleognathae). In Paleognathae, the Z and W chromosomes have largely remained recombining, including HINT, CHD1, and other genes that do not recombine in the chicken (Nishida-Umehara et al. 2007). As we estimate the age of stratum 1 to 132–150 MYA, the Neognathae and Paleognathae in this view should have diverged earlier than this. The estimated divergence time between Neognathae and Paleognathae is, however, slightly later (120 MYA) (Van Tuinen and Hedges 2001), but the discrepancy is small and is probably due to the uncertainty of rate of the molecular clock.
Our previous work showed that genes from stratum 1 (HINTZ/HINTW and CHD1Z/CHD1W) and stratum 2 (SPINZ/SPINW) had ceased to recombine prior to the split of the majority of living Neognathae bird orders (Ellegren and Carmichael 2001; Lawson Handley et al. 2004), thought to have occurred prior to the Cretaceous/Tertiary (K/T) boundary 65 MYA (Cooper and Penny 1997; Kumar and Hedges 1998; Cracraft 2001; Clarke et al. 2005). With an estimated age of stratum 2 of 71–99 MYA, this would suggest that the major avian radiation occurred in late Cretaceous, not long before the K/T boundary. Moreover, this is consistent with the dating of stratum 3 (47–57 MYA), as previous work showed that two genes from this region (ATP5A1 and UBAP2) still recombined when the major avian lineages split (Ellegren and Carmichael 2001; Lawson Handley et al. 2004).
One notable observation in the distribution of genes on the chicken Z chromosome is that the incidence of genes with a still active (not degenerated) gene copy on the W is highest in the youngest stratum close to the PAR at Zp. There are six gametologs out of a total of 155 genes in the first 11 Mb of the Z genome assembly (approximately stratum 3) compared to a total of six out of a total of 606 genes in the rest of the chromosome (P = 0.022, Fisher's exact test) (the chicken W chromosome is not completely sequenced yet, these 12 genes represent the gametologs known so far). It seems reasonable to assume that the number of “surviving” genes in the nonrecombining region of W will decrease with the time since recombination stopped. The density of remaining gametologs shared between Z and W should then decrease with the age of the evolutionary strata, as we observe. A similar pattern is seen on the human X chromosome (Lahn and Page 1999).
The observed pattern of avian sex chromosome organization is similar to that seen in male heterogametic animals and plants. Interestingly, the three most well-studied systems with respect to the formation of sex chromosomal strata represent three different temporal scales of sex chromosome evolution. Mammalian sex chromosome evolution was initiated 320–240 MYA (stratum 1; but see Wallis et al. 2007, who place the start at 210–180 MYA, and Potrzebowski et al. 2007) and was followed by the formation of a second stratum 170–130 MYA. In contrast, in Silene sex chromosomes evolved <10 MYA. The avian sex chromosomes are intermediate, with Z and W starting to diverge ∼150 MYA. The three systems thus offer possibilities for comparative studies of sex chromosome evolution on different time scales.
Niclas Backström, Judith Mank, Deborah Charlesworth, and two anonymous reviewers are acknowledged for helpful comments. Financial support was obtained from the Swedish Research Council.
Communicating editor: D. Charlesworth
- Received April 14, 2008.
- Accepted August 19, 2008.
- Copyright © 2008 by the Genetics Society of America