Answers are grounded in this paper and cite their sources.
The dimorphic sex chromosomes likely evolved from a pair of autosomes, followed by Y-chromosome degeneracy and X-chromosome dosage compensation (Ohno 1967 ). The Y-linked genes generally degenerate by the processes of Muller's ratchet, genetic hitchhiking, or background selection after it has ceased to recombine with the X chromosome (Bachtrog and Charlesworth 2002 ). There remain, however, several functional genes on the Y chromosome outside of the pseudoautosomal region (Lahn and Page 1997 ). We wish to know whether the processes that contribute to Y-degeneracy also leave a mark on these functional genes. Furthermore, there is often a degree of functional divergence between the X- and Y-linked genes. When genes are evolving toward new functions, the rate of amino acid substitutions is expected to be high during the initial phase of divergence and eventually returns to a lower rate of evolution (Ohno 1970 ; Li 1982 ; Lynch and Conery 2000 ). We ask whether Y-linked functional genes in mammals have completed the process of functional divergence since more than 100 MYA. Genes in this study were collected from GenBank. The human X and Y chromosome genes were described by Lahn and Page (1999) , and orthologous outgroup sequences were found using BLAST (Altschul 1990 ). Orthology was assigned using the assumption that the best scoring outgroup sequence under BLAST was indeed the ortholog; in most cases the genes are also syntenic, and orthology was easily assigned. After speciation, X-Y gene conversion may occur, and we excluded two gene pairs (AMELX/Y and RBMX/Y) from this analysis because they show evidence of gene conversion between X and Y homologs. This results in X-Y sequence pairs within a species having large homology tracts. Sequences for six genes (DBX/Y [AF000983, NM_004660, Z38117, AJ007376], SMCX/Y [Z29630, NM_004653, AF127245, AF127244], Sox3/SRY [X71135, X53772, X94125, X67204], UTX/Y [AF000993, NM_007125, AJ002730, NM_009484], ZFX/Y [NM_003410, NM_003411, M32309, X14382], and RPS4X/Y [NM_001007, NM_001008, AB024285, AB024286]) were aligned in the Megalign program from DNASTAR to minimize protein differences, and KA/KS comparisons were performed with NewDiverge in The Wisconsin Package (GCG 1999 ) using a modification of the method of Li (1993) . NewDiverge calculates the numbers of transitions and transversions at twofold, fourfold, and nondegenerate sites corrected for multiple hits, and these numbers are used for the A:S calculations in table 1 . The Kimura two-parameter correction for multiple hits ensures that we are not grossly underestimating the silent divergence rate for either Y or X lineage. In addition, we calculated KA and KS using other methods (such as that of Yang, Nielson, and Hasegawa [1998] , who denote them as dN and dS, respectively). Although there was some variance in the actual change rates across methods, these did not affect the significance of the differences between X and Y genes. When compared with the previous work (such as that of Smith and Hurst [1999] , specifically table 1 ), it is not surprising that the conclusion of this report is the same regardless of the methods used in calculating substitutions. The branch length between the Ancestral Y (AY) and Ancestral X (AX) nodes (fig. 1 and table 2 ) was calculated using the method of Tsaur and Wu (1997) for A and S separately; specifically, this is an arithmetic method of calculating branch length. The recent branch calculation as shown in table 2 is the average of the Xh-Xm and Yh-Ym branches. We would like to compare the O (the top node when X and Y started to diverge)-AY branch with the Yh-Ym branch, but substitutions along the O-AY branch could not be calculated. We, therefore, compare the KA/KS ratio of the ancient branch (AX-AY) with that of the recent branch. Assuming that the selective constraint along the X lineage maintained a constant KA/KS ratio, the difference between the ancient and recent branches, if present, is more likely caused by the changes in the Y lineage. This assumption should hold true if the genes on the X chromosome have maintained their ancestral function (see Discussion). As illustrated in figure 1 , the X-linked genes of human and mouse grouped together separated distinctly from the Y-linked genes. Figure 1 shows the average evolutionary distances given next to the X, Y, and ancestral branches. Gene conversion between the nonrecombining portion of the X and Y chromosomes (Slattery, Sanner-Wachter, and O'Brien 2000 ) would have distorted the phylogenetic history of X- and Y-linked genes. Six genes clearly support the XX/YY grouping, with the vast majority of changes in these genes supporting the XX/YY branch (133/141 changes). Apparently, the X-Y divergence predates the mammalian radiation. As shown in table 1 , KS is uniformly higher for the Y chromosome genes than for the X-linked ones. This fits in well with previous findings regarding male-driven evolution (Miyata et al. 1987 ; Shimmin, Chang, and Li 1993 ; Agulnik et al. 1997 ). The salient feature of table 1 is that KA/KS is also higher for Y-linked genes than for their X chromosome counterparts. Because of the poor statistical properties of the KA/KS ratio, we used the observed numbers of nonsynonymous (A) and synonymous (S) substitutions (corrected for multiple hits) between human and mouse for statistical tests. If the A:S ratios for the X- and Y-linked genes are the same, G-test or Fisher's Exact test should yield insignificant results. For example, the Y copies of DB have an A:S ratio of 88:188, whereas the ratio between the X copies is 13:177. In every case involving human and mouse, the difference is highly significant (P < 0.0001). In the case of RPS4, the Fisher's Exact test is not significant because the divergence between the X-linked copies of human and macaque is small (7:0 for A:S). In addition, we wanted to test whether the average A:S ratios for both the X and Y lineages were significantly different from one another. To do this we performed a t-test for each lineage, using each gene as a separate observation of the lineage A:S ratio. The t-test was highly significant (two-tailed paired t-test, P = 0.0038). A general theory on the evolution of new function in duplicated genes is that one or both copies would initially have an elevated KA/KS ratio until the new function has been obtained (Lynch and Conery 2000 ). The elevation in KA can be caused either by the relaxation of negative selection or occasionally by the action of positive selection. Thus, immediately after the X-Y divergence, the Y-linked genes would have experienced accelerated evolution along the top branch between node O and node AY. Because the X-Y divergence predates the mammalian radiation by 10s of millions of years (Lahn and Page 1999 ), one might have expected the KA/KS ratio to have returned to normalcy at the time of mammalian radiation. By this reasoning the KA/KS ratio should be consistently larger along the ancient branch than along the recent branch. But as shown in table 2 , such an expectation is not fulfilled. In other words, the transient phase of elevation in the KA/KS ratio may persist for a long period of time (in this example, because of the divergence of rodents and primates). Why have Y-linked genes been accumulating more amino acid substitutions for every synonymous change than their X-linked homologs? In addition to the smaller population size (1/3 of the X chromosome) and the high variance in male reproductive success, the absence of recombination has led to a reduction in the effectiveness of purifying selection, allowing many slightly deleterious mutations to become fixed. Both genetic hitchhiking and background selection contribute to reduce Y's effective population size (Steinemann, Steinemann, and Lottspeich 1993 ; Charlesworth 1996 ). Many genes on the Y chromosome have become nonfunctional, as shown by the presence of a large number of Y-chromosome pseudogenes (Lahn and Page 1999 ). Recent work in Drosophila has also observed increased KA/KS ratios in genes on the neo-Y chromosome compared with their homologs on the neo-X chromosome (Yi and Charlesworth 2000 ). Still, the elevated nonsynonymous substitutions could be the result of the positive selection for advantageous mutations. Because Y-linked genes mostly pertain to male functions, sexual selection could conceivably play a role (Wyckoff, Wang, and Wu 2000 ). Moreover, Y-linked mutations do not encounter the problem of recessivity (Haldane 1964 ; Charlesworth, Coyne, and Barton 1987 ) or sexual antagonism (Rice 1992 , 1996 ), and thus any advantageous mutation has a much better chance of becoming fixed in the population than the autosomal or X-linked ones have. Unfortunately, the observations of this study would not allow us to distinguish positive selection from the relaxation of negative selection. To do so, a comparison between high-frequency polymorphism and divergence would be needed (Fay, Wyckoff, and Wu 2001 , 2002 ). A second question concerns the change in the KA/KS ratio during and after the emergence of new functions. Evidence suggests that many of the Y-linked genes analyzed here have been functionally altered, vis-à-vis their X-linked homologs. (It is indeed difficult to see how Y-linked genes can escape nonfunctionalization after more than 100 Myr unless they have gained some new functions not performed by the X-linked genes.) SRY, the testis-determining factor that is homologous to Sox3, not only has a critical role in determining maleness (Graves 1998a , 1998b ) but also has expression in the brain of adult males (Mayer et al. 1998 ). ZFY also has brain expression and is differentiated from ZFX in several ways, including differences in the zinc finger region between X and Y homologs and functional differences by in vivo assay in the mouse model (Mayer et al. 1998 ). RBMY has a testis-specific transcript of a different length compared with the RBMX transcript, and DBY maps to a region known to be crucial for sperm formation (Sun et al. 1999 ). UTY, SMCY, and DFFRY all code for various male-specific H-Y antigens that elicit antigenic responses in females (Warren et al. 2000 ). The fact that these genes are antigenic in females might mean that they have some role in male-female interplay during embryonic development or that rapid evolution is a consequence of an antigenic interaction causing an “evolutionary arms race” (Dawkins 1976, pp. 1–176). There is other evidence, which suggests subtly different roles for X- and Y-specific genes; for example, UT, SMC, and DFFR might be expressed differently on X- or Y-bearing sperm (Hendricksen 1999 ). It should be noted that X- and Y-linked genes would retain some basic overlapping functions, although the Y-linked genes evolved new ones. The ability of either human RPS4X or RPS4Y to rescue hamster cell lines from temperature sensitivity is one such example (Watanabe et al. 1993 ). A brief period of heightened KA/KS ratio followed by the return to a slower evolutionary rate has been reported for duplicated functional genes (Lynch and Conery 2000 ). Either each of the duplicated genes becomes specialized in a subset of the original functions (Force et al. 1999 ), or one of them retains the original function, allowing the other to evolve a new function (Ohno 1970 ; Li 1982 ). Because the X-linked gene has to retain the original function for females, the homologous Y-linked gene is less constrained by the old function than are the true duplicated genes. Our observation, thus, suggests that the process of evolving new functions on the Y-chromosome may take a long time (more than 100 Myr in mammals) before the new function becomes subjected to the normal selective constraint of the genome. Alternatively, genes on the Y chromosome have achieved their equilibrium KA/KS ratio, but this new equilibrium is very different because of the specific biology of the Y chromosome. If this is the case, it opens the possibility that Y-linked genes will continue to be subjected to more rapid evolution, possibly aiding in the acquisition of new functions within the rather changeable world of male reproduction. David Rand, Reviewing Editor Keywords: sex chromosomes molecular evolution X chromosome Y chromosome divergence rate duplicate genes Address for correspondence and reprints: Gerald J. Wyckoff, Department of Molecular Biology and Biochemistry, the University of Missouri-Kansas City, 5100 Rock Hill Road, Kansas City, MO 64110. E-mail: wyckoffg@umkc.edu Table 1 Number of Nonsynonymous (A) and Synonymous Substitutions (S) for X-linked and Y-linked Genes. KA and KS are the Number of Substitutions Per Site (Li 1993). The Significance Level in the Difference Between the A:S Ratios of X- and Y-linked Genes is Given Table 1 Number of Nonsynonymous (A) and Synonymous Substitutions (S) for X-linked and Y-linked Genes. KA and KS are the Number of Substitutions Per Site (Li 1993). The Significance Level in the Difference Between the A:S Ratios of X- and Y-linked Genes is Given Table 2 KA/KS Ratios Along Ancestral and Recent Branches Table 2 KA/KS Ratios Along Ancestral and Recent Branches Fig. 1.—Average substitution rates along the three branches, between X's of human and mouse, between Y's of human and mouse, and between the ancestral X and the ancestral Y chromosome genes
No takes yet. Share an insight, caveat, or question.
Wyckoff et al. (2002) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: