Microsatellites, Transposable Elements and the X Chromosome

Transcription

1 Microsatellites, Transposable Elements and the X Chromosome Philippe Jarne,* Patrice David,* and Frédérique Viard* *Génétique & Environnement CC065, Institut des Sciences de l Evolution, Université Montpellier II, France; Department of Biology, University College London, United Kingdom; and Department of Forest Sciences, University of British Columbia, Canada Variability at microsatellite (MS) loci is generally perceived as resulting from an interaction between mutation and genetic drift and, to a lesser extent, selection and recombination. Less investigated has been the reason for MS accumulation in genomes. We present here a simple model that could account for the variation in density of MS loci, assuming that they are created either through replication slippage or in association with transposable elements. Microsatellites then evolve under the forces cited above. We use this framework to revisit two results obtained from high-density genomic maps of the human and mouse genomes built with thousands of CA repeats: MS loci are (1) less variable and (2) less dense on the X chromosome than on autosomes. The first result is most likely explained by differential mutation on the X chromosome and the autosomes. The second result may be explained by differential mutation, provided the distributions of MS loci are still not at equilibrium. Selection, acting either directly on large allele size or indirectly on the transposable elements associated with MS, may explain the same result. The framework developed here is a first step toward more rigorous models, calling for additional data. Introduction Microsatellite (MS) sequences are now extensively used in at least three fields of biology, namely, the study of genetic diseases, genetic mapping, and population biology (Charlesworth, Sniegowski, and Stephan 1994). Knowledge gained in each one of these fields has been used to inseminate the others through an interchange between the molecular and populational levels (Freimer and Slatkin 1996; Jarne and Lagoda 1996). However, many aspects of the evolution of MS remain to be clarified. Two recently published studies reporting on highdensity genetic maps in humans (Dib et al. 1996) and mice (Dietrich et al. 1996) exemplify this. These studies are of exceptional value, since they both mapped the distribution of thousands of MSs. Both are exclusively concerned with dinucleotide (CA) repeats. This particular array represents only one category of MS, although it is the second most frequent category in the human genome, amounting to about 8% of all MSs (Nadir et al. 1996), with a number of CA loci which can be inferred to be about 100,000. Moreover, the loci studied for building the human and mouse maps follow an approximately random distribution along chromosomes and are distributed on all chromosomes, including the X chromosome. This suggests that their distribution and polymorphism may be explained by general forces acting at the level of the whole genome. These studies also reported, as a by-product, two interesting results: (1) Microsatellites are less polymorphic on the X chromosome than on the autosomes. Dietrich et al. (1996) found a 30% reduction based on the comparison of 10 inbred lines of mice representing widely distributed sampling sites. Based on 28 unrelated caucasoid individuals from a CEPH panel, Dib et al. Key words: microsatellites, X chromosome, autosomes, replication slippage, transposable elements, humans. Address for correspondence and reprints: Philippe Jarne, Génétique & Environnement CC065, Institut des Sciences de l Evolution, Université Montpellier II, Place E. Bataillon, Montpellier cedex 5, France. jarne@isem.univ-montp2.fr. Mol. Biol. Evol. 15(1): by the Society for Molecular Biology and Evolution. ISSN: (1996) found a gene diversity (sensu Nei 1987) of 0.70 on the autosomes versus 0.65 on the X chromosomes. (2) The X chromosome exhibits 25% and 38% deficits in density of CA repeats compared to the average values for autosomes in humans and mice respectively. This is in line with previous work, based on 10 times fewer loci, reporting a 25% deficit in the rat genome (Jacob et al. 1995). A possible caveat with respect to such mapping studies is that polymorphic markers are screened for. However, Dietrich et al. (1996) showed that the X chromosome is actually depauperate in monomorphic CA repeats as well. Similar data are not available for the human genome. However, an indirect argument can be used based on the data from Genethon (Dib et al. 1996). The primary reason for differentially rejecting microsatellites on the X and autosomes at the screening stage would be their size (see table 1 of Weissenbach et al. 1992). However, there is no (or a positive) correlation between the sizes of loci and variability (e.g., r , P and r 0.124, P 10 6 for the comparison between the number of alleles and the minimum and maximum sizes, respectively, per locus). The sizes of loci are rather larger on the X chromosome than on the autosomes (one-tailed Student test, P 0.03 and 0.05 for the minimum and maximum sizes, respectively). This suggests a conclusion similar to that for the mouse genome. Thus, while lower microsatellite variation appears to be explained by recent selection for specific X chromosomes, sex differences in mutation rate (since the X chromosome spends only two thirds of the time spent by autosomes in the male germline, and most mutations are thought to occur in males) and hemizygosity of the X chromosome in males affecting selection and effective population size (see Dietrich et al. 1996), the lower density of microsatellites on the X chromosome than on the autosomes remains unexplained. We develop here a simple compartment model in order to analyze the forces acting on the creation and subsequent buildup of MS loci. We also review the forces acting on their polymorphism. On this basis, we reanalyze the results on 28

2 Microsatellites and the X Chromosome 29 variability and density of microsatellites, considering the roles of genetic drift, differential mutation, selection, recombination, and transposable elements (TEs) in determining MS polymorphism and density. Note that, although we will focus on CA repeats, the theory developed here could be generalized to any category of MS. Data and Analysis The Model Microsatellite loci may be generated in two ways. First, they may originate from random sequences, from which short repeated sequences (proto-mss) may be generated through mutation. These proto-mss can then evolve through replication slippage, which will increase or decrease the current size by one or more repeat units (reviewed in Amos et al. 1996). They will be classified as true MSs once they have reached a threshold size (fig. 1). Stephan and Cho (1994) showed by a modeling approach that this sequence of events is quite plausible. The random sequences and proto-ms compartments are separated by a threshold size which corresponds to the lower number of repeats at which replication slippage may add or substract one repeat (of the order of three to five repeats). Similarly the threshold between the proto-ms and MS compartments corresponds to the lower size of MS searched for in genetic mapping projects (of the order of 10; see Dib et al. 1996; Dietrich et al. 1996). u 0 and u 2 are the rates at which proto-mss and MSs are generated from random sequences and proto-mss, respectively. The reverse evolution occurs at rates u 1 and u 3, respectively. The densities of proto-mss and MSs are N 1 and N 2, respectively. In our model, MSs can also evolve into other sequences (e.g., CAGA from CA). This includes any mutation that generates impure MSs from pure ones. On the other hand, the transition from an impure to a pure MS seems far less likely, as a point mutation at a specific site is required. Therefore, if we define u 4 as the net exchange between pure CA repeats and other sequences, this parameter is largely biased toward positive values. The second pathway through which MS may be created is TEs (Jurka and Pethiyagoda 1995). Nadir et al. (1996), analyzing about three million base pairs of the human genome drawn from a database (GenBank), showed that the majority of MSs are poly-a arrays, and most other MSs have A-rich core sequences of the type A 2 5 C. They also showed that most MSs are associated with retroposons, especially with Alu sequences. However, the percentage of CA repeats associated with such sequences reaches only 35%. Nadir et al. (1996) suggest that A-rich MSs are generated from the poly-a tail of retroposons when these sequences insert into the host genome. Poly-A tails are not always pure A stretches, and they may also evolve after retroposition. This may explain why not all MSs associated with retroposons are poly-a. Whether the association between MS and TE represents a general trend across genomes still has to be shown. Primmer et al. (1997) indeed failed to find any association between the most widely spread element (CR1) and MSs in the chicken genome. However, we consider the association possible, occurring at rate u 5. Based on these processes creating and destroying MSs (see fig. 1), the variation in N i across generations (t) is: N 1 N (t 1) N (t) 1 1 dt u un(t) (u u )N (t) N 2 N (t 1) N (t) 2 2 dt u un(t) (u u )N (t). (1) Equilibrium solutions for the N i s can be found provided u 1, u 2, and u 4 0: uu 3 5 u 0(u3 u 4) N 1 (u u )(u u ) uu and uu 0 2 u 5(u1 u 2) N2. (2) (u u )(u u ) uu General solutions for the N i s as a function of both time t and the u i s were also found using (1): d t(c)/2 t(c)/2 1 i 1 2 N (t, u ) ce ce b a t(c)/2 t(c)/2 2 i 3 4 N (t, u ) ce b ce, (3) where a uu u (u u ), b uu u (u u ), c u u u u, d uu 3 5 u 0(u3 u 4), c2 4b, d 1 cd c1 u 0, 2b 2b d 1 cd c2 u 0, 2b 2b a 1 ac c3 u 5, 2b 2b a 1 ac c4 u 5. 2b 2b We considered that MSs are evolving under mutational pressure only, and we did not take into account transitory polymorphic states in which alleles belonging to different categories (e.g., proto-ms and MS) coexist at the same locus. Indeed, we assume that such transitory states are not relevant to the study of the density

3 30 Jarne et al. FIG. 1. A schematic representation of the life of microsatellite (MS) loci. They are generated/destroyed by replication slippage (u 0 to u 3 ), and may also be created by transposable elements (u 5 ). Microsatellites can also evolve into other MSs or into higher order structures at rate u 4. The differences between random sequences and proto-mss and between proto-ms and MSs relies on a threshold size. of MSs, which is true, provided the u i s are interpreted as transition rather than mutation rates. However, a number of factors other than mutation rates, including genetic drift and selective pressures, can alter actual transition rates. We therefore proposed a very general framework for analyzing the evolution of microsatellites. In the next sections, we evaluate whether genetic drift, differential mutation and selective pressures can explain the observed differences between the X chromosome and the autosomes with regard to the amount of polymorphism (result 1 in Introduction) and, based on our model, MS density (result 2). Genetic Drift From a population genetics point of view, MSs have mainly been considered under two models of mutation, the infinite alleles and the stepwise mutation models (for a discussion and references, see Freimer and Slatkin 1996; Jarne and Lagoda 1996). Under both models, the gene diversity of a neutral locus in a population at equilibrium between genetic drift and mutation is a simple function of Nu (where N is the effective population size, and u is the mutation rate per locus; see Ohta and Kimura 1973; Nei 1987). Based on the gene diversity observed at human microsatellite loci (see above; Dib et al. 1996), we can estimate N for the X chromosome to be on average 0.71 and 0.80 that of autosomes under the stepwise and infinite alleles mutation models, respectively, assuming equal mutation rates on the X chromosome and the autosomes. This result fits nicely with the assumed lower effective population size of the X chromosome (Dietrich et al. 1996). However, the effective size of the X chromosome is of the order of those of autosomes (or even larger) when the effective size of the male population is much lower than that of the female one (see Caballero 1995). This is likely to be the case in mammals, in which sexual selection processes may maintain a high variance in reproductive success in males compared to females (Andersson 1994; Thornhill and Gangestad 1996). In this situation, genetic drift is unlikely to explain a lower variability on the X chromosome than on the autosomes. Genetic drift is also unlikely to explain the lower density on the X chromosome. This may be envisaged as follows. Let us assume an MS locus with a new mutant allele belonging to another compartment (say proto- MS). The probability of transition from one compartment to another will be the mutation rate to a new allele (Kimura 1983), and does not depend on the population size. Without other differences than in effective size between the X chromosome and the autosomes, the u i s will be identical, and there is therefore no reason to believe that genetic drift will lead to a buildup of MSs on autosomes. Differential Mutation If we assume that the main source of mutation at MS loci is replication slippage and that mutation generally occurs in the male germline (Weber and Wong 1993), the mutation rate on the X chromosome is reduced compared to that on the autosomes, since the X chromosome spends only two thirds of the time spent by the autosomes in this germline. The relationship between the ratio R of the mutation rate on the X chromosome over that on the autosomes and the ratio of the mutation rate in the male over that in the female germline is R {2( 2)}/{3( 1)} (Miyata et al. 1987). R can be derived from the observed gene diversities in humans given in the Introduction using formulas from Ohta and Kimura (1973), Nei (1987) and Valdès, Slatkin, and Freimer (1993), and assuming that the effective sizes of the X chromosome and the autosomes are about the same. This gives values of 3 (stepwise mutation model) and 4 (infinite alleles model) for, which may be a bit high when compared to values for neutral nucleotide substitutions (see McVean and Hurst 1997). The few available data suggest that is much higher than 1 at microsatellite loci, even if precise estimates are not available (see Weber and Wong 1993; Amos et al. 1996). Alternatively, the difference in gene diversity between the X chromosome and the autosomes

4 Microsatellites and the X Chromosome 31 may be due to a lower mutation rate on the X chromosome (McVean and Hurst 1997). These authors estimated that the ratio of neutral nucleotide substitutions on the X chromosome over that on the autosomes is about 0.6. This is compatible with the 0.8 value for microsatellites. A difference in mutation rate between the X chromosome and the autosomes may therefore fully explain a lower variability on the former. A difference in mutation rate between the X chromosome and the autosomes, whatever its source, affects not only the variability at MS loci, but also the rate at which they are created and destroyed, that is, the ratio of u 0 to u 4 (fig. 1). If TEs make a negligible contribution to the creation of MSs (u 5 small compared to u 0 ), it can be shown from equation (2) that the equilibrium value of N 2 will not vary with a difference in mutation rates. Thus, this difference does not explain the lower density on the X chromosome than on the autosomes at equilibrium. However, the dynamics toward equilibrium depend significantly on the mutation rate. We analyzed the variation of N 2 over time using equation (3), assuming an initial situation with no MS loci and a reduction of the u i s on the X chromosome by a factor of two thirds (when most mutations are in the male germline). N 2 (X)/ N 2 (autosomes) is a continuously increasing function of time with limits 0.46 (t 0) and 1 (t tending toward infinity). Situations may therefore be found in which this ratio takes the observed value of ca (data not shown). Under these conditions, reaching the equilibrium requires millions of generations. Whether enough time has elapsed since the creation of the X chromosome in mammals to reach equilibrium densities of MSs is hard to judge. Of course, initiating the simulations with no MSs provides the most favorable conditions to observe a deficit of MSs on the X chromosome. A more plausible scenario is that MSs were already present on all chromosomes when a proto-x arose. In this situation, the initial value of N 2, N 2 (0), has no influence on the expected density of MS at equilibrium (see eq. 2), and the general form of the curves drawn from equation (3) is not greatly affected. However, when N 2 (0) is positive, the ratio N 2 (X)/N 2 (autosomes) is higher than when N 2 (0) 0, and cannot be below N 2 (0)/N 2 (equilibrium), although it remains below 1 as long as equilibrium is not reached. The time window to reach equilibrium will also be shorter, and the conditions for a deficit of MSs on the X chromosome will be harsher. We can conclude that differential mutation may have a role on the relative density of MS, provided the distributions on the X chromosome and the autosomes are out of equilibrium and the initial density was not too large. On the other hand, if u 5 is large compared to other u i s, the equilibrium value of N 2 reduces to about u 5(u1 u 2) N2. (4) (u u )(u u ) uu As u 5 is not expected to differ on the X chromosome and the autosomes, and given that the other u i s on the X chromosome are expected to be two thirds that on autosomes, we expect 50% more MSs on the X. Therefore, if MSs mainly originate from TEs, we expect them to build up on the X chromosome relative to the autosomes. This is in contradiction to result 2. Selection Microsatellite alleles are generally considered neutral, even if loci involved in genetic diseases are clearly under selective pressure (Sutherlands and Richards 1995). The question here is whether selection may act differentially on the X chromosome and the autosomes. This may derive from the fact that the X chromosome occurs as a single copy and does not recombine in males (we neglected the possible recombination with the pseudo-autosomal region of the Y chromosome, since this accounts for less than 2% of the X chromosome; see Foote et al. 1992). The action of selection on MSs may therefore take different forms, depending on whether it bears directly on alleles, indirectly on ectopic exchanges, or on the TEs from which they may derive. Direct Selection Selection with a different fitness attributed to different alleles and genotypes is unlikely for microsatellites, and hemizygosity of the X chromosome in males has no consequence on either the relative variability or the relative density on the X chromosome and the autosomes. Selection may, however, act on the absolute lengths of alleles, truncating allelic distributions above a given threshold, as suggested by studies on some genetic diseases (Sutherlands and Richards 1995). Distributions of MS loci in humans seem to have a lower maximum size on autosomes than on the X chromosome (see Introduction). Distributions of alleles at given loci are therefore expected to be wider on the X chromosome (suggesting a milder selection on the X chromosome), and less variability is expected on autosomes, running counter to result 1. This type of selection may also act on the density of MSs, as larger alleles are more likely to be subject to point mutation, and contribute more to u 4. The size distribution bias toward small sizes observed at autosomal loci may result from intense selection against large allele size, which would decrease u 4. Assuming that u 5 is negligible and u 3 K u 4, equation (2) at equilibrium gives: N 2(X) u 4(autosomes) (5) N (autosomes) u (X) 2 4 If truncation selection is milder on the X chromosome, we may expect more loci on autosomes, which is compatible with the observed results. However, this effect is at best mild, and no difference is expected under other conditions on the u i s. Moreover, the minimum size of human MSs is also lower on autosomes than on the X chromosome (see Introduction), and autosomes and the X chromosome do not differ in mean allelic range per locus (size of larger allele minus size of smaller allele; data from Genethon, P 0.19). Finally the difference observed in humans (see above) was not observed in the mouse (data from Whitehead/MIT center for genome research; Dietrich et al. 1996; two-tailed Student test on mean size per locus, P 0.9). This casts some doubt

5 32 Jarne et al. Table 1 A Summary of the Influence of Various Forces on the Relative Polymorphism and Density (Results 1 and 2 in the Introduction Section, Respectively) of Microsatellite Loci on the X Chromosome and the Autosomes Evolutionary Forces Polymorphism Density Genetic drift... Nodiff. (if intense sexual selection in males) No diff. Mutation a Replication slippage... Yes() No diff. (at equilibrium) Transposable elements... Yes() Selection Direct Size... Size difference... Ectopic exchanges Intrachromosomal... Interchromosomal... Transposable elements Intrachromosomal... Interchromosomal... Direct selection... Yes () b Yes () Yes () b No diff. No diff. Yes () No diff. Yes () Yes () (in part only) NOTE. No diff. no difference is expected between the X-chromosome and the autosomes; Yes a difference is expected; ()/() means that the difference goes in the same/opposite direction of the observations in the human and mouse genomes; irrelevant. a Microsatellites are generated either through replication slippage or via transposable elements. b Under restrictive conditions. on the previous interpretations based on milder truncation selection on the X chromosome than on the autosomes. Ectopic Exchanges Ectopic exchanges are defined as crossing over between homologous elements located at different chromosome sites, and they occur either within a chromosome or between homologous chromosomes, provoking a decrease of fitness because they result in gametes with a chromosomal loss or addition (Charlesworth, Sniegowski, and Stephan 1994). Ectopic exchanges are meiotic processes in yeast and Drosophila (Charlesworth and Charlesworth 1995), and they have been considered the main force containing the buildup of TEs in the Drosophila genome (Charlesworth, Sniegowski, and Stephan 1994; but see Hoogland and Biémont 1996). By analogy, they may occur between MS loci at nonhomologous sites, although we have no data on such a process. If interchromosomal exchanges are frequent, they will be more frequent on autosomes, as the X chromosome does not recombine in males. This would prevent the buildup of MSs on autosomes, in contrast with what has been observed. This reasoning assumes that the recombination rate does not differ on the X chromosome and the autosomes, as suggested by comparisons in both the human genome (data from Dib et al. [1996] on the female genetic map; Student test, P 0.20) and the mouse genome (data from Nachman and Churchill [1996] on the average map over males and females; P 0.30). There is no reason to believe that intrachromosomal exchanges differ between the X chromosome and the autosomes. More generally, the role of recombination on MS density seems limited, since the latter does not correlate with the recombination rate across autosomes in either humans (data from Dib et al. [1996] on the female genetic map; r 0.203, P 0.364) or the mouse (data from Nachman and Churchill [1996] on the average map; r 0.265, P 0.613). Transposable Elements If MSs derive from, and are strongly linked to, transposable elements, then some selective effects are expected, either directly or through ectopic exchanges as envisaged in the two previous paragraphs. If ectopic exchange is the main force driving the evolution of TEs (Charlesworth, Sniegowski, and Stephan 1994), we then expect relative densities of MSs on the X chromosome and the autosomes to be as described in the previous paragraph. On the other hand, direct selection against TE insertion might be a more significant force than ectopic exchanges (Hoogland and Biémont 1996). If this holds, noting again that about 35% of CA repeats in humans are related to TEs, we expect a shortage of MSs on the X chromosome. This probably would explain only part of the observed result. The forces considered in the last two sections (ectopic exchanges and TEs) have no bearing on the variability at MS loci and do not explain the lower variability on the X chromosome than on the autosomes. Conclusion From the simple model and considerations above, it can certainly be concluded that there is no simple (and probably no single) explanation for the two results obtained from the mapping of the human and mouse genomes using CA repeats. A summary of the possibilities envisaged here is presented in table 1. The lower microsatellite variability on the X chromosome than on the autosomes cannot be explained by genetic drift if, as is likely, the effective size of the male population is lower than that of the female population. We can also reject possible direct selective effects. The most likely expla-

6 Microsatellites and the X Chromosome 33 nation is differential mutation between the male and female germlines. On the other hand, mutation explains the lower density on the X chromosome than on the autosomes only if the distribution of MSs is out of equilibrium, a point difficult to evaluate. However, very slow dynamics toward equilibrium are obtained with reasonable sets of parameters, consistent with nonequilibrium distributions. Other possible explanations include direct selection on allele size (under restrictive conditions) and selection mediated through TEs (which explains only part of the result). A further mechanism can generate some differences in the rate of evolution between the X chromosome and the autosomes. We assumed that replication slippage, the addition or subtraction of a single repeat unit, is the main mechanism generating variability at MS loci. However, we cannot reject the possibility of events that add or subtract a much larger number of repeats (DiRienzo et al. 1994; Amos et al. 1996). If such events are related to recombination, they would be less frequent on the X chromosome, as recombination is also less frequent. Less polymorphism would therefore be expected on the X chromosome. The model and considerations presented above can be generalized to other situations. For example, variation in recombination rates across genomes (e.g., sexual vs. parthenogenetic species) or parts of genomes (e.g., Y vs. other chromosomes) may generate patterns similar to those observed here. Similarly, species with conspicuous haploid phases, such as some red algae, may be submitted to different selective regimes than diploid species. However, other forces acting on the maintenance of polymorphism would have to be considered, such as selective sweeps (Begun and Aquadro 1991) and background selection (Charlesworth, Morgan, and Charlesworth 1993; Slatkin 1995). Even if the framework provided here for analyzing the evolution of microsatellites is general and of an exploratory nature, we hope that it will provoke the development of more precise models as additional data become available. Acknowledgments The authors thank C. Aquadro, B. Charlesworth, A. Estoup, F. Rousset, S. Samadi, and one anonymous referee for discussion and/or comments on the manuscript. This is contribution from Institut des Sciences de l Evolution. LITERATURE CITED AMOS, W., S. J. SAWCER, R.W.FEAKES, and D. C. RUBINZ- STEIN Microsatellites show directional bias and heterozygote instability. Nat. Genet. 13: ANDERSSON, M Sexual selection. Princeton University Press. BEGUN, D. J., and C. F. AQUADRO Molecular population genetics of the distal portion of the X chromosome in Drosophila: evidence for genetic hitchhiking of the yellowachaete region. Genetics 129: CABALLERO, A On the effective size of populations with separate sexes, with particular reference to sex-linked loci. Genetics 139: CHARLESWORTH, D., and B. CHARLESWORTH Transposable elements in inbreeding and outbreeding populations. Genetics 140: CHARLESWORTH, B., M. T. MORGAN, and D. CHARLESWORTH The effect of deleterious mutations on neutral molecular variation. Genetics 134: CHARLESWORTH, B., P. SNIEGOWSKI, and W. STEPHAN The evolutionary dynamics of repetitive DNA in eukaryotes. Nature 371: DIB, C., S. FAURÉ, C. FIZAMES et al. (14 co-authors) A comprehensive genetic map of the human genome based on 5264 microsatellites. Nature 380: DIETRICH, W. F., J. MILLER, R. STEEN et al. (22 co-authors) A comprehensive genetic map of the mouse genome. Nature 380: DIRIENZO, A., A. C. PETERSON, J. C. GARZA, A. M. VALDES, M. SLATKIN, and N. B. FREIMER Mutational processes of simple-sequence repeat loci in human populations. Proc. Natl. Acad. Sci. USA 91: FOOTE, S., D. VOLLRATH, A. HILTON, and D. C. PAGE The human Y chromosome: overlapping DNA clones spanning the euchromatic region. Science 258: FREIMER, N. B., and M. SLATKIN Microsatellites: evolution and mutational processes. Pp in Variation in the human genome. Wiley, Chichester. HOOGLAND, C., and C. BIÉMONT Chromosomal distribution of transposable elements in Drosophila melanogaster: test of the ectopic recombination model for maintenance of insertion site number. Genetics 144: JACOB, H. J., D. M. BROWN, R.K.BUNKER et al. (20 coauthors) A genetic linkage map of the laboratory rat, Rattus norvegicus. Nat. Genet. 9: JARNE, P., and P. LAGODA Microsatellites, from molecules to populations and back. Trends Ecol. Evol. 11: JURKA, J., and C. PETHIYAGODA Simple repetitive DNA sequences from primates: compilation and analysis. J. Mol. Evol. 40: KIMURA, M The neutral theory of molecular evolution. Cambridge University Press, Cambridge. MCVEAN, G. T., and L. D. HURST Evidence for a selectively favourable reduction in the mutation rate of the X chromosome. Nature 386: MIYATA, T., H. HAYASHIDA, K. KUMA, K. MITSUYASU, and T. YASUNAGA Male-driven molecular evolution: a model and nucleotide sequence analysis. Cold Spring Harb. Symp. Quant. Biol. 52: NACHMAN, M. W., and G. A. CHURCHILL Heterogeneity in rates of recombination across the mouse genome. Genetics 142: NADIR, E., H. MARGALIT, T. GALLILY, and S. A. BEN-SASSON Microsatellite spreading in the human genome: evolutionary mechanisms and structural implications. Proc. Natl. Acad. Sci. USA 93: NEI, M Molecular evolutionary genetics. Columbia University Press, New York. OHTA, T., and M. KIMURA A model of mutation appropriate to estimate the number of electrophoretically detectable alleles in a finite population. Genet. Res. 22: PRIMMER, C. R., T. RAUDSEPP, B. P. CHOWDHARY, A. P. MOLL- ER, and H. ELLEGREN Low frequency of microsatellites in the avian genome. Gen. Res. 7: SLATKIN, M Hitchiking and associative overdominance at a microsatellite locus. Mol. Biol. Evol. 12:

7 34 Jarne et al. STEPHAN, W., and S. CHO Possible role of natural selection in the formation of tandem-repetitive noncoding DNA. Genetics 136: SUTHERLANDS, G. R., and R. I. RICHARDS Simple tandem DNA repeats and human genetic disease. Proc. Natl. Acad. Sci. USA 92: THORNHILL, R., and S. W. GANGESTAD The evolution of human sexuality. Trends Ecol. Evol. 11: VALDÈS, A. M., M. SLATKIN, and N. B. FREIMER Allele frequencies at microsatellite loci: the stepwise mutation model revisited. Genetics 133: WEBER, J. L., and C. WONG Mutation of human short tandem repeats. Hum. Mol. Genet. 2: WEISSENBACH, J., G. GYAPAY, C. DIB, A. VIGNAL, J. MORIS- SETTE, P. MILLASSEAU, G. VAYSSEX, and M. LATHROP A second-generation linkage map of the human genome. Nature 359: CHARLES F. AQUADRO, reviewing editor Accepted September 19, 1997