REVIEWS NA TURE REVIEWS GENETICS VO UME ANUARY used the X chromosome particularly for - PDF document

MICROSATELLITEA class ofrepetitive DNA that ismade up ofrepeats that are 2 to8 nucleotides in length.They canbe highly polymorphic and arefrequently used as molecularmarkers in population-geneticLINKAGE DISEQUILIBRIUMA nonrandom correlationbetween alleles ofphysicallylinked loci.llinearity in the order ofgenes(or ofother DNA sequences) ina chromosomal region oftwoANUARY 2004 www.nature.com/reviews/genetics Low diversity.The distinctive characteristics ofthe Xmosome largely derive from how it is inherited.ales have only one copy ofthe X chromosome,soery existing X chromosome has spent two-thirds ofits history in females.Consequently,mutations occurless frequently on the X chromosome than on auto-somes because the nucleotide mutation rate in femalesis several-fold lower than in malesthe result islower genetic diversity,as well as smaller interspecieshas not lost its autosomal character:the X chromosomeis physically the most stable nuclear chromosome,atleast among placental mammals,and is the only one totain complete large-scale between mouse andhe size ofthe X chromosome Ñ 150 millionbase pairs (Mb) in humans Ñ is consistently ~5% ofthe genome among mammals.Its gene density is low,anking about seventeenth among the 24 humanclear chromosomes Marker typeReferencesmtDNAY chromosomeX chromosomeAutosomesSize (Mb)0.017601503,00015,16,59Number of usable loci11HundredsThousandsÐMutation rate (mutations Very highHighLowModerate19,60,61per Mb per generation)(1Ð300)(0.033)(0.015)(0.020)Recombination rate (cM/Mb)000.81.121Diversity (fraction of Very highLowModerateHigh42,62discordant base pairs)(0.4%)(0.02%)(0.04%)(0.08%)Accessible haplotypes*YesYesYesNoÐHighHighModerateLowÐ100,000 years100,000 years750,000 years1,000,000 yearsÐEffective population size 1/41/43/41Ð (relative to autosomes)*A haplotype is a set of genetic markers that is present on one chromosome; àgenetic drift describes the random changes in allefrequency that occur because genes that appear in offspring are not a perfectly representative sample of the parental genes (fothese entries are approximate population genetics inferences, based on the consensus estimate for theeffective population size in humans. cM, centiMorgan; Mb, megabase; mt, mitochondrial. WhatÕs in a locus?The usefulness ofa locus for population-genetic studies depends on three main characteristics:its age,its mutation rateand its recombination rate.These charactersitics are described below,and the different markers used in population-genetic studies are compared in ith respect to these characteristics and their consequences.AgeThe age ofa locus is the time to the most recent common ancestor (MRCA) ofall extant copies ofit.The age ofa locusdefines the period from which genetic variation has been preserved,and therefore delimits the historical period that canbe investigated using that locus.r human population-genetic studies,interesting timescales include:a few thousand years for recent historicalents;~50,000 years for the expansion ofomo sapiensout ofAfrica;and 100,000Ð200,000 years for the emergenceofmodern omo sapienstation rateThe mutation rate can make a locus uninformative by being either too high or too low.Ifthe rate is too low (comparedith the age ofthe locus),there will be too little genetic variation to study,but ifthe rate is too high,recurrent mutationsll occur at every site and will obscure the process under study.In practice,single-base substitution rates (thoseproducing single nucleotide polymorphisms,or SNPs) are low enough that recurrent mutation can be neglected,at leasten studying a single species.The exception is mtDNA,which mutates at a much higher rate than the rest ofthegenome.Mutation rates at MICROSATELLITEmarkers,on the other hand,are high enough that recurrent mutations becomeproblematic after a few tens ofthousands ofyears.mbination rateThe recombination rate controls the size ofa chromosomal region that shares a single genealogical history.In the absenceofrecombination,every chromosome falls on a unique branch ofthe phylogenetic tree,whereas recombination combinesdifferent branches onto a single physical chromosome.Too much recombination,therefore,makes reconstruction ofphylogenetic trees impossible,as individual markers represent different trees.tation and recombination rates are biologically determined,whereas the age ofa locus is primarily a function ofthesize ofthe population Ñ larger populations tend to have older loci.Together,the recombination rate and the agedetermine the amount ofLINKAGE DISEQUILIBRIUM(LD) that is observed because recombination over time breaks down LD. TURE REVIEWS GENETICSANUARY 2004 chromosome and the size ofregions with a singlegenetic history to be larger.This effect is reinforced bythe younger age ofthe X chromosome,as younger locihave had less time for recombination to break down LD.stematic comparisons ofLD on the X chromosomeand the autosomes have not been published for mam-mals;relevant data sets do exist for humans,how-er,and it is expected that their analysis will support theprediction that LD is greater on the X chromosome.The biological questions that can be studied withthe X chromosome flow from the characteristicsdescribed above.The female-dominated history oftheX chromosome makes it an ideal system for studyingpulation-genetic differences between males andfemales,particularly the differences in mutation rateand in patterns ofrecombination.The presence ofasingle copy ofthe X chromosome in males means thatX-linked alleles are more exposed to natural selection,making the X chromosome an attractive place toxamine the role ofselection in human history.utation rates in males and females.The higher muta-tion rate in male mammals,including humans,is gener-ally attributed to the larger number ofmitoses thatgermline cells go through in males.Direct measurementofthe difference in rates is difficult,however,as theabsolute rates are low (~2 x 10tations/bp/genera-tion).The easiest way to obtain an accurate measure-ment has been to examine homologous regions on theX and Y chromosomes and to compare their divergencefrom the inferred ancestral sequence;a higher maletation rate will be reflected in a higher rate ofsubsti-tutions on the Y chromosome copy.Such studies havebeen done for humansand for several otherorganisms (other primatessheep and goatsrode).The consistent finding has been that malesdo have higher mutation rates than females;moststudies have also concluded that in humans and apes,the ratio is particularly high,presumably because ofng generation time and unusually large numbermale mitoses in those species.By comparing allsystems (the X and Y chromosomes,and thetosomes),it has been possible to test an alternativepothesis for the lower mutation rate on the X chro-namely that,independent ofwhether it is inmales or females,the X chromosome has evolved anunusually low mutation rate to protect itselfagainstdeleterious mutations.The available data,however,donot support this hypothesisDespite the apparent simplicity ofthese measure-ments,controversy continues about the magnitude ofthe difference between the mutation rate in males andost studies have estimated about a fivefoldhigher rate in males,but two studies have yielded ratiosclose to twohis lower value might be caused by afailure to take into account polymorphism present inthe ancestral population,a possibility that is supportedthe observation that divergences over longer evolu-tionary times yield higher ratiosn the other hand,adivergence,on the X chromosome.According to onestudy,for example,the divergence between human andhimpanzee X chromosomes is 83% ofthat observedbetween the autosomes ofthese two species.Diversityis further reduced by the EFFECTIVE POPULATION SIZEthe X chromosome,which,because males lack a sec-ond copy,is three-quarters that ofthe autosomes.Thenet effect is that diversity on the X chromosome (mostoften measured as HETEROZYGOSITY,or about halfofthat on the autosomes.This expectationhas been borne out by several experiments (see BOX 2Diversity on the X chromosome might be low com-pared with the autosomes,but it is still twice that onthe Y chromosome,and it provides enough variantsites (~1Ð10 per 1,000 base pairs (bp)) for reasonablysized regions to be informative.In terms ofage,auto-somes record slightly older time periods than the X chromosome,but both record substantially olderhistories than either the Y chromosome or mtDNABOX 3Pronounced population structure.nother consequenceofthe smaller population size ofthe X chromosome isthat genetic driftis faster for the X chromosome thanfor the autosomes.As a result,POPULATION STRUCTUREbe more pronounced on the X chromosome;that is,pop-ulations should differ more in their X chromosomesthan in their autosomes.Indeed,as shown in GENETIC DISTANCESbetween populations are significantlylarger on the X chromosome.ong linkage disequilibrium intervals.X chromosomesmbine only in females as males have a single copy;therefore,only two-thirds ofX chromosomes recom-bine in each generation.The measured recombinationfor the X chromosome is,in fact,almost exactlytwo-thirds ofthe genome averages a result,we canpect linkage disequilibrium (LD) to be greater on the EFFECTIVE POPULATION SIZEThe size ofthe idealpopulation in which the effectsofrandom drift would be thesame as those seen in the actualHETEROZYGOSITYA measure ofthe geneticvariation in a population:themean number ofdifferencesfound when comparing twopies ofa sequence.Usuallypressed as the number ofdifferences per base pair.THE SNP CONSORTIUMSC).A publicÐprivate effortthat mapped approximately 1million SNPs across the humanPOPULATION STRUCTUREA departure from randommating as a consequence offactors such as inbreeding,erlapping generations,finitepopulation size andgeographical subdivision.GENETIC DISTANCEThe degree ofgeneticdifferentiation between twopopulations.It is measured bymparing allele frequencies(and in the case ofmicrosatellitemarkers,by comparing allelesizes) between populations. Genetic diversity of the X chromosome and autosomesThe lower mutation rate and the smaller population size ofthe X chromosome,comparedith autosomes,lead to an unambiguous prediction that genetic diversity should also belower there.As measurements ofthe diversity at individual loci on the X chromosome and on autosomes vary widely (as discussed in BOX 3),a direct comparison ofthe overalldiversity between the X chromosome and autosomes is difficult.Despite experimentaldifficulty,however,several human studies support this prediction.An analysisof47 kb ofsequence on the X chromosome found diversity to be 80% ofthat on the autosomes Ñich is surprisingly high.Larger studies ofthe X chromosome have found values more line with expectation:a study ofvariation in genesin a total of~70 kb found theio ofthe diversity between the X chromosome and autosomes to be 48%,and thelarge SNP-discovery project carried out by THE SNP CONSORTIUMSC),which analysed base pairs (Mb),found the ratio to be 59%,which corresponds to diversity onthe chromosome ofone difference in every 2,000 bpnon-African populations,which are genetically less diverse than African ones,the diversity on the X chromosome can be markedly lower than on the autosomes.one study (S.F.S.unpublished observations,listed in ),40% ofalllymorphisms showed no variation in a sample ofEast Asian chromosomes,whereasonly 10% showed no variation in a West African sample.In a similar vein,extensive ÔSNPdesertsÕÑ regions (some ofwhich are more than a million base pairs long) with veryw SNPs Ñ were found on European X chromosomes ANUARY 2004 www.nature.com/reviews/genetics that they leave behind.The obvious platform for such astudy will be the X chromosome,because it is only therethat female recombination occurs in the absence ofmale recombination.tural selection.tural selection has two importantfunctions in evolution:it allows adaptation (positiveselection) and it suppresses deleterious mutations(background selection).The direct effect ofthe lattercan be easily measured by observing the reduced poly-morphism and interspecies divergence at coding sites.The extent ofpositive selection is less clear.It is notknown how often episodes ofpositive selection occur orhow important they have been in shaping our genome;nor is it clear what effect background selection has onerall neutral variation.The reason for considering theX chromosome in this context is that selection might bemore visible there than on the autosomes.Two explana-tions could account for this;first,as recessive alleles onthe X chromosome are exposed to selection in malesgardless oftheir frequency,a larger fraction ofmuta-tions undergo selection;and second,higher LD on the X chromosome means that when selection does act on alocus,it is likely to affect a larger region than it would onrecent studythat accounted for this effect still reporteda fairly low value for the ratio (~threefold).Studies intothis unresolved question will probably continue.combination in males and females.A related issuentres on the differences between males and females inthe pattern ofrecombination.The overall rate ofrecom-bination is higher in women,and the rate varies alongmosomes differently for the two sexes;for example,mbination in males tends to be higher near thelomere,whereas in females it is higher near the cen-tromeren males,the distribution ofrecombinationcan be studied directly at a very fine scale by typingDNA from a large number ofspermo far,thesestudies have shown that recombination is highly clus-d in Ôhot spotsÕ,with little recombination occurringelsewhere.It is not known how universal or variable thisphenomenon is across the genome,and little is knownabout what determines where recombination occurs oneither fine or coarse scales.The hope is that furthersperm studies will identify these determinants.No com-parable studies offemale recombination will be possi-ble,however,so the only access to fine-scale patterns offemale recombination will be through the LD patterns OUT OF AFRICA MODELSdels for the origin ofmodernuman populations in whichanatomically modern sapiens evolved~100,000Ð150,000 years ago inica and expanded from therethe rest ofthe world.A marked reduction inpopulation size followed by thesurvival and expansion ofasmall random sample oftheoriginal population. nlike the Y chromosome and mtDNA,theX chromosome contains many independentloci,each with its own phylogenetic tree.Itis a characteristic ofgenealogies,whatevermosome they occur on,that they varyandomly;that is,under identicalcircumstances,the phylogenetic trees fortwo loci can be very different,both in shapeThe two trees shown in panel are thesult ofsimulations ofa constant-sizedpulation for two loci,and are typical oftheamount ofvariation observed.Although thetwo simulated loci share an identicalpulation history,the age (and thereforethe diversity) oflocus 1 is many times that oflocus 2;inferring the characteristics ofthepulation from either tree alone willtherefore give a badly skewed result.shows the full range ofages expectedr the three types ofchromosome,on thebasis ofan OUT OF AFRICA MODELofhumanorigins.As the X chromosome has threemes the effective population size ofthe Ymosome or mtDNA,loci on the Xmosome can be expected to be mucholder;the same is true for autosomes,whichhave four times the effective population ofthe Y chromosome.Note the broad age rangeexpected for different loci from the sameofchromosome.The histogram showspublished estimates ofthe age ofvarious loci;all have large uncertainties (not shown)imilar variation from locus tolocus occurs in other inferences,such as those concerning in population size,or about the source ofmigrations into a region. 0.511.522.53ime to most recent common ancestor (millions of years) 0.511.522.53ime to most recent common ancestor (millions of years) X chromosome loci Number of loci expected(arbitrary scale)Number of loci observedba Locus 2Locus 1 ime X chromosome loci 1 2 TURE REVIEWS GENETICSANUARY 2004 attached to them,so a discrepancy Ñ albeit probablyonly a small one Ñ might still emerge.At present,there-there is no compelling evidenfor a larger role fornatural selection on the X chromosome.Other avenuesmain to be explored,however,such as comparing theX chromosome and the autosomes in terms ofgeneticdistance between populations,or comparing the fre-quencies ofcoding polymorphisms for the two types ofthe autosomes.So,a ll typically covermore sequence on the X chromosome.A mildly delete-ious mutation,which reduces the number ofviablemosomes in the population,has a similar effect ofducing variation in a larger region than on autosomes.aces ofselective sweeps (or at least good candidatesfor them) have indeed been found on the X chromo-ore general interest is the relative impor-tance ofsweeps,and ofselection generally,in the twopes ofchromosome.Several tests have been applied inthe search for traces ofselection.Perhaps the mostbust ofthese involves detecting whether diversityvaries in step with recombination.Because selectivesweeps lower diversity,and because they extend furtherwhere there is little recombination,we expect less diver-sity where the recombination rate is low.(A similar effectalso occurs for background selection.) A positive cor-lation between heterozygosity and recombination rateis,therefore,a signature ofextensive selection.Such aelation has been observed on human autosomes,but it is weak and can probably better be explained byfactors other than selection(FIG.1a).Severhave indicated that the correlation is larger on the X chromosomeere,also,the evidence is weak,butis consistent with a greater role for selection on thismosome.The large data set from the SNP consor-tium (TSC),however,shows no correlation betweenheterozygosity and recombination rate at all on the X chromosome (FIG.1b)second indicator ofselectionould be a difference in diversity between the X chromo-and the autosomes,even after adjusting for theirdifferent and mutation rates.Positive selection (butnot background selection) would lower the diversity ofthe X chromosome compared with autosomes.Here,supporting evidence ofselection seems to be miss-ing.The relative X-to-autosome diversity cited in BOX 2(59% for the TSC data) agrees almost perfectly with theed value of58% that is obtained by assuming atio ofmale-to-female mutation rate offivefold.Boththe observed and estimated values have uncertainties The process by which new,favourable mutations becomefixed so quickly that physicallylinked alleles also become fixedhitchhikingÕ. Human X-linkedNumberNumber ofSize of DNA sequenceMean F(African Referencemarker or geneof locimarkersanalysed (kb)versus non-African)0.6280.087dystrophin13680.045Xq13.3133100.0990.0244Multilocus survey162501100.26Unpublished (S.F.S. otal213981430.24Ð*Values for and Xq13.3 were calculated from the published data; other values are quoted from the original publications; measure of genetic distance, based on allele frequencies, that indicates the proportion of genetic diversity found between popuelative to the amount within populations. Values range from 0.0 to 1.0. Because studies sample different populations, it is often difficult tocompare directly genetic distance measurements from different studies; for this reason, only the large-scale measurement betweeand non-African populations is given. On average, the divergence values for X-chromosome-linked loci, shown here, are significa(REF.47); this indicates that the X chromosome has amore pronounced population structure than that of autosomes. Note also the large variation in genetic distance among the X-chro, pyruvate dehydrogenase (lipoamide) , zinc finger protein, X-linked. 0.140.120.10.080.020.0 0 1 2 3 4 5 6 0 1 2 3 4 5 Heterozygosity (%)Heterozygosity (%) X chromosome Figure 1 |Natural selection produces acorrelation between diversity (shown here as heterozygosity)and recombination rate| Heterozygosity measurementsfrom The SNP Consortium SNP-discovery surveycorrelation on the autosomes, but the correlation is weak and probably results from factors other than selection| A much stronger correlation on the X chromosome wouldindicate a larger effect of natural selection there, but the samedata show no correlation between diversity and recombinationrate on this chromosome. ANUARY 2004 www.nature.com/reviews/genetics oups.As discussed below,these three categories Ñobal population characteristics,large-scale history andlocal history Ñ have varied considerably in the extent towhich they have used the X chromosome.Global population characteristics.Global features ofathe occurrence ofexpansionsor bottlenecks and the extent ofsubdivision within thepopulation.All three features can be inferred from sta-tistical information about the population.For exam-ple,because genetic diversity and population size arecorrelated,can be inferred from the measuredamount ofdiversity.Studies ofthis kind that have usedX-chromosome loci have yielded estimates ofange from 4,900 to 37,000 for humansthis wideange is expected because ofthe large stochastic varia-tion between loci (BOX 3)obal estimate based onesequencing a third ofthe X chromosome yields anof12,000 (REF.42)imilar studies can be done usingother genetic systems,and the autosomes,Y chromo-some and mtDNA have all been used for them.Theonly reason the X chromosome is over-representedhere is that X-chromosome loci are often chosen forhaplotype analysis,and because summary statistics caneniently be studied at the same time,using theThe X chromosome could provide a unique resourcefor one topic to which it has so far not contributed:theway in which population-genetic parameters differbetween males and females.Numerous mechanismsmight contribute to this phenomenon.For example,different breeding and mortality patterns can producedifferences in the average length ofa generation,andpolygamy increases female lative to that ofmales.TRILOCALITYproduces larger genetic distances betweenoups for males than for females,whereas military con-quest has the opposite effect.The question is notwhether these processes have operated in human historyÑ they all have Ñ but how important the differenteffects have been.Comparisons between mtDNA andthe Y chromosome have shown that genetic distancesnd to be larger for males than for females;the cause ofthe difference is very controversials the X chro-mosome spends twice as much time in females as inmales,differences between males and females will alsobe reflected in differences between the X chromosomeand the autosomes.Given the complexity ofthe ques-tions,it would be useful to bring the large body ofgenetic information available on the X chromosome tobear on them.Until now,however,the power ofanyn X chromosome-linked locus to address thesequestions has been too low,simply because ofthe mix-ture ofmale and female histories on the X chromosome.r example,a parameter (such as diversity) that is(hypothetically) 50% higher in a purely female lin-eage than in a male one will only be 7% higher on theX chromosome than on the autosomes.In the longhowever,the volume ofdata from the X chromo-some promises to provide an independent perspectiveon questions that neither the Y chromosome nor mtDNAis able to resolve.Studies of human historyis a different set offeatures that makes the X chromo-some particularly valuable for historical research.First,there is a matter ofpracticality.Because there is a singleofthe X chromosome in males,it is easy to deter-mine haplotypes (BOX 4)feature that is not present inthe autosomes.As haplotypes are needed to infer thephylogeny ofa region,the X and Y chromosomes,andmtDNA are the natural choices for studies that use phy-logenies,an advantage that explains their dominant rolein historical studies.Until technology to read single mol-ecules ofDNA becomes practical,their pre-eminencell probably continue,or even increase:development ofhigh-throughput genotyping and sequencing technol-ogy has markedly expanded the possibilities for large-scale haplotype studies,whereas extracting haplotypesfrom autosomal loci remains labour intensive.Second,there is the presence ofrecombination.Ifease ofhaplotyping distinguishes the X chromosomefrom the autosomes,it is the occurrence ofrecombina-tion that distinguishes it from the Y chromosome andmtDNA.In the latter two systems,the entire chromo-some acts as a single locus and shares a single genealogi-cal history.The X chromosome and the autosomes,onthe other hand,are broken up by recombination atery generation,so that different regions have differenthistories.This difference has important implications forhistorical investigations.In practical terms,recombina-tion makes it harder to study phylogenetic trees and soX chromosome-based phylogenies must be restricted toons with very strong LD.In broader terms,however,mbination creates a tremendous resource:it meansthat the X chromosome records hundreds or thousandsofdifferent snapshots ofthe populationÕs history,whereas the Y chromosome and mtDNA each recordonly a single one.Because the history ofany single locusonly crudely records the history ofthe population inwhich it lived (see BOX 3),information from a recombin-ing system is crucial for providing as complete a view aspossible ofthe history ofhuman populations.It is thembination ofaccessible haplotypes and multiplegenetic histories,therefore,that makes the X chromo-some a uniquely powerful tool for historical studies.istorical studies can be divided into two types:those that reconstruct gene phylogenies on the basis ofhaplotypes and those that use SUMMARY STATISTICSto properties ofancestral populations.Owing to theirdependence on haplotypes,it is the phylogenetic studiesthat can be expected to have a particular focus on the Xand Y chromosomes,and mtDNA.Studies can also be classified by their focus.Thebroadest studies look at global characteristics ofa popu-lation,typically using summary statistics as their basis.Asecond class looks at the geographical history ofsub-populations and the phylogenetic relationships betweenthem.In studies ofhuman history,an approximate butuseful distinction can be made,within the second class,between studies that focus on large-scale patterns,espe-cially those that aim to distinguish between Out ofMULTIREGIONAL MODELSand those that analysethe history ofsmaller geographical areas or ethnic SUMMARY STATISTICA single number thatsummarizes complex data;xamples include mean andvariance.MULTIREGIONAL MODELSdels for the origin ofmodernuman populations in whichanatomically modern sapiensolved simultaneouslythroughout Africa,Europe andTRILOCALITYA residential pattern in which amarried couple settles in theusbandÕs home or community. TURE REVIEWS GENETICSANUARY 2004 nclusions about the basic structure ofhuman popula-tions.It is for this reason that many loci have been (andwill continue to be) needed.ine-scale geographical studies.Studies on smaller geo-aphical and ethnic units abound,with the focus rang-ing from the colonization ofthe Americas and thecific to the history ofsmall,isolated populations suchas the Andaman Islanders.The studies that use the Xmosome have so far been virtually absent fromthese efforts,not because they would not be valuable butbecause the bulk ofthe work,and all ofthe phylogeneticeffort,has been done with the Y chromosome andmtDNA.The reason for this dominance is partly theamount ofwork that has already been done using thesesystems;the existing phylogenetic trees and geographi-cal information about Y chromosome and mtDNAhaplotypes provide a context for all new studies.Inaddition,the Y chromosome and mtDNA are peculiarlysuited to tracking recent local history.They have highertation rates and much higher rates ofgenetic driftthan the X chromosome,which mean that populationsll tend to differ more at these loci than they do at anX-chromosome locus,allowing finer resolution for localstudies.On the other hand,this last characteristic mightnot always be an advantage:faster drift also means thatlationships between populations are more easilyerased,sometimes making it harder to identify sourcepopulations.(It is interesting to speculate,for instance,that X-chromosome loci will identify the nearest Asianlatives ofnative Americans more successfully than theY chromosome and mtDNA have done.)Many more loci are needed,however,to unravel themplexities ofthe history ofpopulations.This is why,despite its minimal contribution so far,the X chromo-some has tremendous potential for this kind ofwork.This potential will only be realized ifthere actuallymany suitable loci on the X chromosome Ñ thatons with markers in strong LD and contain-ing enough markers to construct phylogenetic trees.unately,this seems likely to be the case.It has becomeclear in the past several years that haplotypes in much ofthe genome form block-like structures,which are a fewkilobases to many tens ofkilobases in length,with littlembination visible within themblock on the Xmosome that is a few tens ofkilobases long can bepected to contain ~50 to 100 SNPs.On the basis ofobservations on the autosomes,we can expect to findundreds ofthem on this chromosome.Furthermore,many ofthese blocks will also contain one or moremicrosatellite markers.This is useful because micro-satellites,with their much higher mutation rate,make itpossible to distinguish between closely related haplo-pes,therefore improving the resolution for local stud-ies.Shorter regions can also be used Ñ no region studiedso far has been much longer than 10 kb Ñ but to achievethe resolution needed for localized studies,greaterlength is preferable.The amount ofinformation thatis available on the X chromosome is large enough todaunting:duplicating the efforts that have alreadybeen made for the Y chromosome and mtDNA for eachLarge-scale geographical studies: Africa and beyond.The second class ofstudy provides the best illustrationofthe potential ofthe X chromosome among existingphylogenetic studies.The central issue in these studieshas been the debate about the origin ofnon-Africanpopulations,between multiregional and Out ofAfricamodels,a debate that has largely been settled in favourofthe latter.Much ofthe evidence invoked in theourse ofthe debate has been in the form ofphyloge-netic studies that give clues to the source ofall mod-ern populations.Naturally,the Y chromosome andmtDNA have played a part,each contributing onephylogeny;several autosomal studies have also beenhe largest source ofevidence,however,hasbeen the X chromosomen together,these studies indicate an Out ofica origin for modern humans.Individually,how-er,the studies point to diverse conclusions.For exam-ple,two studies reconstructed haplotypes in intronsX chromosome-linked genes,using chromosomesdrawn from many populations around the world.st studyfound a pattern that was strongly indica-ofa recent expansion (and subsequent isolationdistance) ofhumans from a single geographicalsource.The chromosomes showed two common,old(200,000Ð800,000 years) haplotypes with worldwidedistribution,as well as many younger,derived haplo-pes (~100,000 years old) with limited geographicaldistribution.Africa was identified as the probable sourcefor the expansion on the basis ofthe higher geneticdiversity there,and by the continued presence oftheancestral human haplotype (as determined by compari-son with other primates).The second study,on other hand,found a pattern that was more consistentith multiregionalism.At this locus,most haplotypesnot shared between Africans and non-Africans,indicating independent histories.Moreover,within thephylogenetic tree,haplotypes from different geographi-cal areas were often intermingled,indicating no singlesource.The exception was one group ofrelated haplo-pes that were exclusively non-African,and that wereestimated to be old enough (66,000Ð264,000 years) tomake it likely that they predated the putative Out ofica migration (thought to have occurred ~50Ð60,000ears ago).These two studies,therefore,point to opposite nstruct phylogenetic trees,it is first necessary to determine which haplotypes arepresent in the population.This is straightforward for mtDNA,the Y chromosome and(in males) the X chromosome,because in all three cases,an individual possesses a singlehaplotype.For the autosomes,however,the task is harder:as both copies ofthe locus areexamined together,it is difficult to distinguish between alleles that belong to onehaplotype and those that belong to the other.It is technically possible to examine only asingle haplotype (for example,by using cloning or allele-specific PCR),but suchmethods add significantly to the cost and effort ofa study.An alternative solution is tonstruct the haplotypes statistically Ñ that is,to estimate which alleles typically gogether in the population as a whole.This technique works well for commonhaplotypes,but becomes increasingly unreliable for rarer ones,to the point that it isunworkable for haplotypes that combine SNPs and microsatellites. ANUARY 2004 www.nature.com/reviews/genetics son ofthe two genomes will give a detailed map ofthevariation in mutation rate across the chromosome,allowing better calibration ofages,at least for SNPs.The second resource will be one generated by theapMapoject (see online links box),aneffort to describe the haplotype and LD structure forery region ofthe human genome;initial data fromthis project are due in 2003.For the studies described inthis article,the map will provide an easy way to identifyons ofstrong LD Ñ regions with a single genealogi-cal history Ñ and therefore candidates for phylogeneticinvestigation.Although the map will also provide fre-quency information (in three populations) about morethan 30,000 SNPs on the X chromosome,this informa-tion will be ofuncertain value,partly because there hasbeen a strong bias in the selection ofmarkers towardsthose present at high frequency.Discovery ofmarkers(SNPs and microsatellites) within the regions will stillquire extensive screening,but the identification oftheons themselves will be a considerable benefit.the long term,for investigations that rely onintrinsic features ofthe X chromosome (such as studiesofdifferences between males and females in mutation,mbination and history,and studies into the behav-iour ofexposed recessives under natural selection),theole ofthis chromosome will remain unaltered.By con-ast,for investigations that rely on X chromosome locibecause they are easy to haplotype,including most his-ical studies,the situation might change.Iftechnologybecomes available that can cheaply and reliably extracthaplotype information directly from diploid chromo-somes (for example,by sequencing individual mole-cules ofDNA),there will no longer be a reason to preferthe X chromosome to the autosomes.This does notmean that the role I have described for the X chromo-some will be eliminated;it simply means that it will bebroadened to include the 20-fold larger store ofdatathat is present on the autosomes.X-chromosome locus in turn will be a major task,butone that will become easier as high-throughput screen-ing and genotyping methods become more efficient.The use ofthe X chromosome in population genetics isstill in its infancy.It has already proved its worth in studiesofthe early history ofmodern omo sapiensut in mostesearch areas its potential remains largely untapped.That potential is needed Ñ the Y chromosome andmtDNA,despite their enormously fruitful contributions,are not very informative about some questions (such asthe size ofancestral populations),and the informationthat they can provide about others (such as populationhistory before the Out ofAfricamigration) has largelyalready been mined.The X chromosome is therefore thelogical place to turn for more information.Many ofthesame questions can be addressed by either the X chromo-some or the autosomes,but the X chromosome has aclear advantage in allowing easy access to haplotypes;thest ofextracting haplotypes from autosomes remainshigh,even as sequencing and genotyping become muchfaster and cheaper.is needed to realize the promise ofthe Xmosome.Most important,ofcourse,will be moredata:more markers,from more loci,studied in morepopulations.We will also need a better understanding ofpatterns ofrecombination and their causes and oftation rates,as well as improvements in modellingthe history ofpopulations and in extracting informa-tion from summary statistics.Two continuing projectspromise to provide some assistance.The first is thesequencing ofthe chimpanzee genome,which is due tobe completed before the end of2003.This will provideuseful information ofat least two kinds.The chim-panzee sequence usually records the ancestral state ofsites that are variable in humans,and so provides a rootphylogenetic trees.More importantly,the compari-1.Cann, R. L., Stoneking, M. & Wilson, A. C. Mitochondrial DNA and human evolution. 2.Lucotte, G., Guerin, P., Halle, L., Loirat, F. & Hazout, S. Y chromosome DNA polymorphisms in two AfricanAm. J. Hum. Genet. 3.Hey, J. Mitochondrial and nuclear genes present conflicting4.Zietkiewicz, E. 5.Zietkiewicz, E. . Genetic structure of the ancestralpopulation of modern humans. An early summary-statistic study of a human X-chromosome locus (8 kb of intronic sequence in the dystrophin gene). It is distinguished by the verylarge data samples that were used (250 chromosomeswere used for SNP ascertainment and 860chromosomes were genotyped).Nachman, M. W., Bauer, V. L., Crowell, S. L. & Aquadro, C. F.DNA variability and recombination rates at X-linked loci inA study of diversity at seven X-chromosome loci in7.Jaruzelska, J. neutral polymorphisms in the last ZFX intron: analysis of thehaplotype structure and genealogy. X-chromosome locus (1.1 kb) that supported a classic Out of Africa model of human migration. 8.Harris, E. E. & Hey, J. X chromosome evidence for ancientProc. Natl Acad. Sci. USAA study of haplotypes in a 4.2-kb X-chromosomelocus, showing surprisingly large differences in9.Kaessmann, H., Heissig, F., von Haeseler, A. & PŠŠbo, S.DNA sequence variation in a non-coding region of lowecombination on the human X chromosome. non-coding region on the human X chromosome.10.Lahn, B. T. & Page, D. C. Four evolutionary strata on the human Y chromosome. 11.May, C. A., Shone, A. C., Kalaydjieva, L., Sajantila, A. &Jeffreys A. J. Crossover clustering and rapid decay oflinkage disequilibrium in the Xp/Yp pseudoautosomal gene12.Skaletsky, H. The male-specific region of the human Ychromosome: a mosaic of discrete sequence classes.13.Jobling, M. A. & Tyler-Smith, C. The human Y chromosome:Nature Rev. Genet. A recent comprehensive review of the Y chromosome.14.Waterston, R. H. analysis of the mouse genome. 15.Venter, J. C. . The Sequence of the Human Genome.16.International Human Genome Sequencing Consortium.Initial sequencing and analysis of the human genome.17.Bohossian, H. B., Skaletsky, H. & Page, D. C. Unexpectedly18.Li, W. H., Yi, S. & Makova, K. Male-driven evolution. Curr. Opin. Genet. Dev. A report that summarizes many of the issues involvedin estimating male and female mutation rates.19.Makova, K. D. & Li, W. H. Strong male-driven evolution of20.Ebersberger, I., Metzler, D., Schwarz, C. & PŠŠbo, S.Am. J. Hum. Genet.,A comprehensive study of genetic divergencebetween humans and chimpanzees, using autosomesand sex chromosomes.21.Kong, A. . A high-resolution recombination map of the22.Lin, S., Cutler, D. J., Zwick, M. E. & Chakravarti, A.Haplotype inference in random population samples. Am. J. Hum. Genet. TURE REVIEWS GENETICSANUARY 2004 23.Stephens, J. C. 24.Anagnostopoulos, T., Green, P. M., Rowley, G., Lewis, C. M.& Giannelli, F. DNA variation in a 5-Mb region of the Xchromosome and estimates of sex-specific/type-specificAm. J. Hum. Genet. A summary-statistic study that focuses on differing25.Huang, W. . Sex differences in mutation rate in higherprimates estimated from AMG intron sequences. J. Mol.26.Lawson, L. J. & Hewitt G. M. Comparison of substitutionrates in ZFX and ZFY introns of sheep and goat related27.Chang, B. H. Weak male-driven molecular evolutionin rodents. 28.Chang, B. H. & Li, W. H. Estimating the intensity of male-driven evolution in rodents by using X-linked and Y-linked29.McVean, G. T. & Hurst L. D. Evidence for a selectivelyfavourable reduction in the mutation rate of the Xchromosome. 30.Hubert, R., MacDonald, M., Gusella, J. & Arnheim, N. Highesolution localization of recombination hot spots using31.Jeffreys, A. J., Ritchie, A. & Neumann, R. High resolutionanalysis of haplotype diversity and meiotic crossover in thehuman TAP2 recombination hotspot. Hum. Mol. Genet. 32.Jeffreys, A. J., Kauppi, L. & Neumann, R. Intenselypunctuate meiotic recombination in the class II region of the33.Schneider, J. A., Peto, T. E., Boone, R. A., Boyce, A. J. &Clegg, J. B. Direct measurement of the male recombinationfraction in the human Hum. Mol. Genet.34.Ruwende, C. & Hill, A. Glucose-6-phosphatedehydrogenase deficiency and malaria. J. Mol. Med. 35.Sabeti, P. C. . Detecting recent positive selection in thehuman genome from haplotype structure. 36.Charlesworth, B., Morgan, M. T. & Charlesworth, D. Theeffect of deleterious mutations on neutral molecular37.Nordborg, M., Charlesworth, B. & Charlesworth, D. Theeffect of recombination on background selection. 38.Lercher, M. J. & Hurst, L. D. Human SNP variability andmutation rate are higher in regions of high recombination.39.Hellmann, I., Ebersberger, I., Ptak, S. E., PŠŠbo, S. &Przeworski, M. A neutral explanation for the correlation ofdiversity with recombination rates in humans. 40.Payseur, B. A., Cutter, A. D. & Nachman, M. W. Searchingfor evidence of positive selection in the human genomeusing patterns of microsatellite variability. A broad search for evidence of selection in the humangenome that includes differences in selectionbetween the X chromosome and the autosomes.41.Nachman, M. W. Single nucleotide polymorphisms andecombination rate in humans.42.Sachidanandam, R. . A map of human genome43.Begun, D. J. & Whitley, P. Reduced X-linked nucleotideDrosophila simulans.Proc. Natl Acad. Sci.44.Y. DNA Polymorphism in a worldwide sample ofhuman X chromosomes. A large haplotype-based study of a 10-kb region onthe X chromosome that indicates a multiregional45.Seielstad, M. T., Minch, E. & Cavalli-Sforza, L. L. Geneticevidence for a higher female migration rate in humans.46.Oota, H., Settheetham-Ishida, W., Tiwawech, D., Ishida, T. &Stoneking, M. Human mtDNA and Y-chromosome variationis correlated with matrilocal versus patrilocal residence.47.Laporte, V. & Charlesworth, B. Effective population size andpopulation subdivision in demographically structured48.Oota, H. . Extreme mtDNA homogeneity in continental49.Pennisi, E. Tracking the sexes by their genes. 50.Harding, R. M. . Archaic African and Asian lineages inthe genetic ancestry of modern humans. Am. J. Hum.51.Rana, B. K. melanocortin 1 receptor locus. 52.Jin, L. . Distribution of haplotypes from a chromosome 21egion distinguishes multiple prehistoric human migrations.53.Rogers, E. J. . Integrated analysis of sequence54.Yu, N. & Li, W. H. No fixed nucleotide difference betweenAfricans and Non Africans at the pyruvate dehydrogenase55.Gabriel, S. B. . The structure of haplotype blocks in the56.Yu, N. . Larger genetic differences within Africans thanvariation within and between human populations.57.Miller, R. D., Taillon-Miller, P. & Kwok, P. Y. Regions of loworangutan Xq: deserts and recent coalescences.58.Takahata, N., Lee, S. H. & Satta, Y. Testing multiregionalityof modern human origins. 59.Anderson, S. Sequence and organization of the60.Heyer, E. mitochondrial substitution rates: study of control regionmutations in deep-rooting pedigrees. Am. J. Hum. Genet.61.Nachman, M. W. & Crowell, S., L. Estimate of the mutation62.Ingman, M., Kaessmann, H., PŠŠbo, S. & Gyllensten, U.Mitochondrial genome variation and the origin of modernI wish to thank D. Reich, M. Daly and two anonymous reviewers forhelpful comments on the manuscript, and G. Thorisson for provid-ing remapped TSC data.Competing interests statementThe authors declare that they have no competing financial interests. FURTHER INFORMATIONInternational HapMap Project: http://www.hapmap.orgMIT Center for Genome Research:http://www.genome.wi.mit.eduMedical and Population Genetics Group:http://www.genome.wi.mit.edu/mpgAccess to this interactive links box is free online. TURE REVIEWS GENETICSANUARY 2004 used the X chromosome,particularly for investigatingthe origin ofnon-African populations,and the muchlarger potential ofthe X chromosome for addressing thehistory ofpopulations and anthropological questionsabout historical differences between males and females. THE X CHROMOSOME INPOPULATION GENETICS phen F.SchaffnerGenetic variation records a large amount of information about the history of a species and aboutthe processes that create and shape that variation. Owing to the way in which it is inherited, theX chromosome is a rich resource of easily accessible genetic data, and therefore provides aunique tool for population-genetic studies. The potential of the human X chromosome, whichrivals that of the more traditional mtDNA and Y chromosome, has only just begun to be tapped. HYLOGENETIC TREEA graph that depicts theancestorÐdescendantlationships between organismsor gene sequences.Thesequences are the tips ofthe tree.anches ofthe tree connect thetips to their (unobservable)ancestral sequences.