`
`Alec J. Jeffreys
`Maxine J, Allen
`John A. L. Armour
`Andrew Collick
`Yurl Dubrova
`Neale Fretwell
`Tara Guram
`Mark Jobling
`Cella A. May
`David L. Neil
`Rita Neumann
`Department or GenetJcs·,
`University of Leicester
`
`1 Introduction
`
`Muta.Hon processes lU human miniuteHites
`
`1577
`
`Mutation processes at human minisatellites
`
`Minisatellites provide one of the most experimentally tractable systems for stu(cid:173)
`dying tandem repeat instability in man. Analysis of mutation processes has
`been greatly aided by the development of single molecule methods for recove(cid:173)
`ring de novo mutants, and of techniques for exploring allele structure in detail.
`Application of these approaches to man has shown that minisatellites do not
`primarily mutate by processes such as replication slippage and unequal cross(cid:173)
`over intrinsic to the tandem repeat array. Instead, germline repeat instability is
`largely regulated by cis-acting elements near the array and involves unexpec(cid:173)
`tedly complex processes of gene conversion, of potential relevance to the biol(cid:173)
`ogy of meiosis. These processes can be explored both in humans and, in prin(cid:173)
`ciple, in lransgenlc mouse models of human repeal instability.
`
`Minisatellites are tandem repeat loci typically 0.5-30
`kbp long with repeat units in the range 6-100 bp. Thou·
`sands of minisatellites exist in the human genome, pref·
`erentially located near the ends of chromosomes, and
`frequently show variability in repeat copy number and
`therefore allcto length . The extreme informativeness of
`the most variable minisatellites has led to their wide(cid:173)
`spread use for identification and parentage analysis in
`legal medicine, using both multi-locus
`forensic and
`DNA fingerprinting nnd single-locus DNA profiling ap(cid:173)
`proa(;hes. While much ls known about allele length varia·
`tion at hul1)an minisateltites, and how this variation can
`be exploited in DNA typing (seo [1)), we are still largely
`igno ra m of the mutation processes lhat generate Lh.is
`variability and maintain it ill human populations. To
`explore these mechanisms, we and others are using mini(cid:173)
`satellites ns a test-bed for developing new approaches for
`the detection and characterization of de novo mutation
`events in the human germline. One particular interest is
`to determine which of the "classic" models for 'tandem
`repeat instability, such as unequal exchange [2] and repli(cid:173)
`cation slippage [3], are involved in minisatellite muta·
`tlon, and whether direct evidence can be adduced to sup(cid:173)
`port earlier speculations that rninisatcllilcs m11y play a
`role in meiotic recombination [4]. Taudern repeat insta(cid:173)
`bility is also of profound interest in relation to the phe(cid:173)
`nomenon of triplet repeat expansion now known to be
`involved
`in seven different neurological disorders
`including Fragile-X mental retardation, Huntington's
`disease and myotonic dystro phy (see [5, 6]); again, our
`understanding of triplet expansion processes is still far
`from complete.
`
`Knowledge of mutation rates nod processes is also of
`relevance to forensic DNA typing, not only in paternity
`testing where paternal mutation could lead to a false
`exclusion, but also in fully understanding the population
`genetics of loci where genetic drill Is llUbstantially coun-
`
`teracted by recurrent mutation, and in assessing the
`potential for somatic mutation to lead to divergent geno(cid:173)
`types in diJJerent tissues of the same individual.
`
`2 Minlsatellite mutation mtes In pedigrees
`
`The first indication that some minlsatellites can show
`remarkably high germline mutation rates to new length
`alleles, arising from spontaneous changes in repeat copy
`number, came from multi-locus ON/\ fingerprint anal·
`ysis of human families [4, 71 and recombinant inbred
`mouse strains [8} which revealed the frequent appear(cid:173)
`ance of new mutant DNI\ fragment.~ in offspring. Anal(cid:173)
`ysis of single human mlnisatellite loci showed mutation
`rates as high as 5 uh per gamete for the most variable
`minisatellite MSI , with mutation rates at different loci
`Increasing with allele length heterozygosity in accor(cid:173)
`dance wilb the predictions of lhe neutral mutation/
`random drm model [9]. At some loci such as MSl,
`paternal and maternal mutatioM arise with similar fre(cid:173)
`quency, despite diiTercnces in numbers of mitotic cell
`divisions in the germ line leading to sperm or oocytes. AI
`other foci, evidence is accumulating that mutations pre(cid:173)
`ferentially :trise In the male germline; the most extreme
`example to date is locus CEBl with a mutation rate of
`15% per sperm but only 0.3% per oocyte [10]. There is
`thus no clear correlation of mutation rate with germinal
`cell turnover, as might be expected for a mitotic muta(cid:173)
`tion process. lt is also not clear why mutation rates can
`vary by orders of magnitude between different loci; sug·
`gestions that instability may be promoted by factors such
`as short repeat units, high copy number arrays and
`repeat unit sequence homogeneity along U1e array [111
`arc contradicted by unstable loci sucb as CElli, which
`have relatively long repeat units, low copy number arrays
`and extensive sequence diversity between repeat units in
`a single array [10, 12].
`
`3 Somatic mutation
`
`Correspondence: Professor Sir A. J. Jeffreys, Department of Genetics,
`University of Leicester, Leicester, LEI 7RH, UK
`
`Nonstandard abbreviations: MVR, rninisatellite variant repeat; SP·PCR,
`small pool·polymorase chain reaction
`
`Keywords: Minisatellite I Mutation I Recombination I Conversion I
`Sperm
`
`Minisatellite mutation events are not restricted to the
`germllne hut also occur in somatic tisRues. However,
`standard hybridization 'analysis will only reveal such
`mut8Jlts if the tissue is partially or completely clonal fo r
`the new mutant allele, as can occur in cell lines and
`tumours [13}. Mutational clonality in lymphoblastoid
`lines derived fro m human pedigrees can lead lo
`cell
`
`~ VCH Vcr\agsgc•cllscllofl mbH, 69451 Weinheim, 1995
`
`~
`
`0173·0835/95/0909-1577 $5.00+.25/0
`
`The Johns Hopkins University Exhibit JHU2010 - Page 1 of 9
`
`
`
`1578
`
`A. I. IoiTroys •t a/,
`
`Elwrophoresls 1995, 16, 1577-1585
`
`somalic mutation being mis-scored as germinal in origin,
`as bas also been noted for microsatellite mutation [14].
`Mutational mosaicism can also arise in normal tissue as
`a resu.lt of early embryonic mutation, to generate
`somatic DNA with three, not two, alleles detectable by
`hybridization . Such events at minisatellitcs are extreme(cid:173)
`ly rare in humans (only one has been seen to date) but
`common at two mouse minlsatelliles [15-18}. Detailed
`analysis of mulanL alle'le dosage and the segregation of
`mutant cells in embryonic and extra-embryonic tissues
`has S\Jgge~>ted that many of these somatic mutation
`events in mice arc confined to a very early develop(cid:173)
`mental window of instability, during the first two cell
`[19]. However, these
`divisions following fertilization
`unstable mouse loci both consist of long arrays of very
`short repeat units (GGCA or GGGCA), unlike the
`human loci studied, and it is possible that these mouse
`loci mutate via pathways distinct from those operating at
`human mlnisateWtes but more analogous to postzygotic
`trlplet repeat instability, seen for example in Pragile-X
`mental retardation [20] and myotonic dystrophy [21].
`
`4 No unequal exchange
`
`If minisatellites mutate to new length alleles by unequal
`exchange between alleles, then new mutant alleles
`should be recombinant for flanking DNA markers. Anal(cid:173)
`ysis of limited numbers of germline mutants at three dif(cid:173)
`ferent .loci has failed to detect an exchange of flanking
`markers [10, 22, 23], suggesting that inter-allelic unequal
`exchange is not the dominant mode of germline muta(cid:173)
`tion at human minisatellites. Similarly, analysis of
`tumour DNAs and cell lines carrying somatic mutations
`has again excluded unequal mitotic exchange between
`alleles as a m!ijor process [13]. These studies unfortun(cid:173)
`ately do not rule out unequal exchange as a contributory
`mechanism, nor do they clarify which of a host of alter·
`native processes might be involved in minisatcllite insta·
`bility. Further investigation of minisatelllte mutation
`required the development of methods lor investigating
`allelic structure in detail before and after mutation, and
`of techniques capable of recovering unlimited numbers
`of new mutant alleles from any source of human DNA,
`rather than the limited numbers identifiable by pedigree,
`cell line and tumour DNA analysis.
`
`5 Internal allelic structure
`
`All human hypervurlable minisat.e!llLes characterized to
`dale vary not only in repeaL copy number (allele length)
`but also in the l.nterspersion pattern of variant repeat
`units within the array (Fig. l A) [24]. This internal varin·
`tion pmvides a powerful approach to the study of aJJelic
`variation and IJI'ocesses of muta.tion. Interspersion pal(cid:173)
`terns can be determined by minisatellite variant repeat
`(MVR) mapping by PCR 125]. MVR-PCR uses a PCR
`primer ill a fi'l';ed site in lhe DNA flanking the. repeat
`array, together with primers specific for different repeat
`variants, to produce ladders of PCR products extending
`from the flanking D A lo each repeat unit of a given
`type. MVR distribution patterns can be determined by
`subsequent electrophoretic analysis of these MVR-PCR
`
`products. If only one site of repeal unit sequence varia(cid:173)
`tion is targeted, then MVR·PCR CIUl be used Lo distin(cid:173)
`guish just two types of repent, to generate binary codes
`of the two repeat types interspersed along an allele
`("two-state" MVR-PCR). More Internal structural infor(cid:173)
`mation can be recovered by analysing additional sites of
`variability, if tbey exist, with appropriate MVR-PCR
`primers. For example, the 29 bp repeat unit of. minisatel(cid:173)
`llte MS32 contains two base substitullonal polymorphic
`sites eparated by 1 bp ; simultaneous analysis of both
`sites by "four-state» MYR·PCR generates a quaternary
`code from an allele and doubles the infomlalion recover(cid:173)
`able by analysis of either site alone [26].
`
`MVR-PCR analysis of total genomic DNA generates
`PCR products from both alleles simultaneously, to pro(cid:173)
`duce extremely variable diploid digital codes of consider(cid:173)
`able potential usc in forensic Identification (24, 251. For
`allelic diversity slud.!es, however, it is necessary to deter(cid:173)
`mine codes from individual alleles. This can be achieved
`in three ways: fll'St, by physical separalion or differential
`PCR amplifiCation of alleles of different length prior to
`MVR-PCR (25]; second, by deducing parental single(cid:173)
`allele codes from !he diploid codes of parents and their
`offspring [25]; and third and mosl simply, by using allele(cid:173)
`specific nanldng PCR primers targeted to known sites of
`base substitu tional polymorphism in lhe DNA flanking
`the minisatellite so as to produce MVR-PCR products
`from only one allele in an appropriate flanking heterozy(cid:173)
`gote [27].
`
`Single-allele coding has been most extensively carried
`out on minisatellite MS32 [251. Two- tate MVR codes
`have been established for more than 1100 alleles to date
`and have allowed a detailed analysis of allelic variability
`at this locus. Allelic diversity so revealed is far greater
`than can be dJstinguished by standard allele length anal(cid:173)
`ysis ; approximately 50 different length alleles can be
`resolved nt MS32, compared wit11
`an estimated
`100 000 000 or more different alleles distinguishable by
`MVR-PCR in
`tbe global human population. While
`almost all alleles llu far typed at MS32 are different, il is
`possible to compare allele codes to identify groups of
`alleles that nrc closely related and which have therefore
`diverged from 11 recent common ancestral allele (Fig.
`ill). Such allele groups show a curious phenomenon,
`namely lhnt most variability between alleles in repeat
`copy number and MVR code is restricted ro the begin·
`ning of lhe repeat array [25, 28]. TI1is indicates a polarity
`of vari.abilily along the array and implies the existence of
`a terminal mutntion hotspot. Similar polarity has also
`been seen at three other minisatellites and may be a gen(cid:173)
`eral feature of VNTR loci [29-31]. Such polarity would
`not be predicted by clo.ssic processes of slippage or une(cid:173)
`qual exchange. Analogous polarity has also been ob(cid:173)
`serve{! at an unstable letraoucleotide repeat on tbe X
`c:hr<>mosome [32!, and also
`in
`the Fragi le-X (CGG).
`repeal array such that most variation is confined to one
`end where aU repeals are homogeneous and not dis(cid:173)
`rupted by AGG varia nts which can occur in the less vari(cid:173)
`able segment of the array [33, 34}. For Fragile·X, it is sug·
`gested that repeat homogeneity promotes instability, for
`example by slippage, and ihus creates polar variation .
`TI1is does not apply to mlnisatellite polarity, where the
`
`The Johns Hopkins University Exhibit JHU2010 - Page 2 of 9
`
`
`
`E/ewvphomls 1995, 16, lSTI-1 585
`
`Mutation processes al huma n minis1delliles
`
`1579
`
`A
`
`allele
`binary code ataattaaaata ......
`
`8
`Gl:roup A
`
`alATATAAATAAAAAAATATAAAATATAATAAAAATTAAAAAAA--AAAAAAAAT~ .. . , ... . .. ... .
`TATATAAATAAAAAAATATAAAATATAATAAAAAT~AAAAAAA- - AAAAAAAATA~ .. .. . ......... , ,
`7TATAAATAAAAAAATATAAAATATAATAAAAATTAAAAAAA--AAAAAAAAT~~AAAAAAAATAATA? .... . . .
`ATATAAATAAAAAAA?- ·AAAATATAATAAAAAT~AAAAAAA- -AAAAAAAA? AAAliAAAAAMJIATAATAAA ..... .
`tatttatataATAATAA-AATTAAAAAAA--AAAAAAAAT~AATAAAATAAA.
`a tAATATAATAA- MTTAAAAAl\11·- AAAAMAA'rf..AAAAAAAJIAA.TAATAAAATAAA.
`taATAATAA-AATTAAAAAAAaaAAAAAAAATAAAAAAAAJJV.1\JTAATAAAATAAA.
`
`Group B
`
`?ttaaaaaaaaaataattataATAAATATTAAAAAATTAAATTAAAAATTA . . . . .. ... . . .. .. . . ... . .... ... . .
`TATATATAAAJ.),:U : r:TATAAATATTNIMMT'f AAATTAMAATTM-ATTAA . , •.. . • • , • •• . • • , , .•• , , •
`TATATATAhAAAAATTATAAATATTAAAAAATTAAATTAAAAATTAAaATT~TTMA ...•...
`tt "NITJ\W\ANIJITTIITMATATTAAAA.ANl'TMATTMAAAT'rAA-ATTA'AAAAAJ;.A.AAA~"l'A .. ..... . .
`aaATAA.AAAAATTATAAATATTAAAAAATTAAATT'AAAAATTAA·ATTAAAA ... .. . . .. ... ........ .
`ataaaaaatAATTATAA.ATATTAAAA.MTTAAATTM AAATTAA-ATTAAAAAAAAAA . . , . .. ... . .. . . .
`TATMAAAAATTATAA.ATATTMAAAATTAAATTAAAAATTAA-ATTAAAAAAA.AAAMATTAAATAT'l'AA.
`
`Group c
`
`?? aTTTATAA-AAATAATTAATT<
`aaaTTATAA-AMTAATTAATT<
`at aATAAaAMTAATTAATT<
`TTTATAA·AAATAATTAATT<
`TTATAA·AAATA.ATTAATT<
`
`allele
`1. Minisatellite
`Figure
`analysi~ by MVR·PCR. (A) The
`principles of MVR·PCR. DNA
`assays minisatellite
`profiling
`allele length variability by diges(cid:173)
`tion with a restriction enzyme X
`which cleaves in Lhe genomic
`DNA flanking the tandem repeat
`array, followed by gel electro·
`phoresis
`and Southern blot
`bybridisatlon . In contra~t. MVR(cid:173)
`PCR assays allelic varl•bllity in
`the
`interspersion pattern of
`variant repeats (white, shaded
`boxes) within an allele. 1\vo(cid:173)
`state MVR·PCR uses a primer at
`a fixed sfte fn the DNA flanking
`the mlnfsatellite, plus primers
`specific for one or other repeat
`type, to generate PCR products
`from
`the
`flanking
`extending
`DNA lo each repeat of a given
`from which
`the allelic
`type,
`binary code of repeat types (a-,
`t-type) can be read. (B) Exam(cid:173)
`ples of groups of closely related
`MS32 alleles discovered
`in a
`sample of 1100 alleles mapped
`by two·state MVR-PCR. Related
`alleles are aligned with gap•
`introduced
`to
`improve align·
`ments. Repeats shared amongst
`related alleles are shown
`ln
`uppercase. Alignable groups a.re
`usually population-specific; thus
`Group A alleles are aU Cauca·
`stan
`ln origin, Group B all
`African and Group C all Japa(cid:173)
`nese. Note that most variability
`is concentrated at the beginning
`of the alleles, even wlthln the
`very short (18-22 repeat) alleles
`of Group C.
`
`most variable part of the array contains normal levels of
`repeat unit sequence heterogeneity (Fig. lB ).
`
`Groups of closely rel oted or identical mlnisatellite alleles
`usually share a common haplotype of flanking DNA var(cid:173)
`Iants, suggesting that whatever mutation processes oper·
`11te, they do nol normally
`involve recombinational
`exchanges of flanking DNA. This confirms thaL inter(cid:173)
`allelic um:qual excb.ange is not the major mutational
`process. l-lowever, related grou ps do frequently show cvi·
`dcnce of flanking haplotype swi tching [30} although it is
`difficult 10 deduce rel iable rates of exchange from these
`population data, and impossible lo deteratioe whether
`these exchanges occur ns a riOlsu\t of tandem repeat muta(cid:173)
`tion or, lnsLead, represent recombillation or conversion
`events operating strictly in the flanking DNA.
`
`6 New approaches to mutation detection
`
`Mutation processes cannot be reliably deduced from
`popuJution data, and can only be investigated by dl reel
`st ructural analysis of new mulnnt alleles. The re covery of
`mutants from pedigrees is too inei'Jlciem for detailed
`
`mutation analysis, and further can give no information
`about individual mutation rates. Pedigree analysis, of
`course, uses offspring
`to determine whether either
`parental gamete carried a mutation. Given the potential
`fo r PCR to analyse DNA at the single molecule level, an
`alternative is to detect de novo mutalions dl rectly in
`single gametes. Since 10' or more sperm can be recov(cid:173)
`ered from a man, equivalent to 108 offspring, unlimited
`numbers of mutants could be identified and character(cid:173)
`ized. This approach is not appropriate for studying muta(cid:173)
`tion in the fe male germline, in view of the limited availa(cid:173)
`bility of human oocytes, but can be easily adapted to the
`detection of mutants in bulk somatic DNA.
`
`We have developed two single-molecule PCR approaches
`for det.ectlng new mutations Ln sperm and somatic DNA.
`1l1e first approach uses gel electrophoretic enrichment of
`minisatollitc mutants of abnormal length., compared with
`progenitor alleles, from bulk sperm or somatic DNA
`prior to recovery of indi vidual mutant molecules by PCR
`[28]. For technical reasons, thls approach is most effec(cid:173)
`tive for recoved ng large deletion mutants and can be
`used to detect mutants at a frequency as low as lo-1/ cell.
`However, the sperm deletion mutants recovered by this
`method from minlsateiUte MS32, for wh ich the tech-
`
`The Johns Hopkins University Exhibit JHU2010 - Page 3 of 9
`
`
`
`1580
`
`A J. letT!ey, <1 a/.
`
`Eleclrophoresis 1995, 115, JS77-1S85
`
`A ~
`
`~
`~
`=-mmm:mm:m:rril
`.....
`-=-m:mmm:m:n:rr
`~
`~
`~
`
`~
`
`...
`
`B
`
`80
`
`input mole~ules
`
`1.60
`
`·tlO.
`·repeats Figure 2. Minisatellite mutation dotec·
`40
`!ion by small pool PCR. (A) The princi·
`pies of SP-PCR. Genomic DNA
`is
`diluted until an aliquot contains a
`limiting number of minisatellite mole·
`cules. If a mutant molecule (shaded) is
`present, it will now make up a significant
`proportion of all molecules. Each mole·
`cule Is PCR amplified using primers in
`the flanking DNA (arrows). Products of
`the mutant molecule
`can now be
`resolved
`from
`the progenitor
`•llele
`(white) by agarose gel electrophoresis
`and detected by Southern blot hybridisa·
`tion (right). (B) Example of sperm muta·
`tion detection at MS32 by SP·PCR on
`multiple aliquots of dilute sperm DNA.
`The Individual tested was heterozygous
`for a 29-repeat allele plus a 200·repeat
`allele
`too
`large
`to be amplified
`in
`SP· PCR. Mutant allele• are detected as
`PCR products different in size from the
`29-repeat progenitor, which shows an
`intense signal.
`
`29
`
`nique was developed, are now known to be rare and
`highly atypical of the bulk: of de novo mutationa arising
`in the male germline.
`
`technically
`is
`The second single-molecule approach
`simpler and involves PCR amplification of the entire
`allele from multiple dilute aliquots of sperm DNA
`(small pool PCR, SP-PCR; Pig. 2; [351). Since the
`haploid genome of man con.tnlns 3 pg DNA, o small
`pool of 300 pg sperm DNA will contain 100 molecules
`of a given locus (SO per allele) . For a locus such as MS32
`with a mutation rate of 1 %/gamete, each small pool will
`contain on average one mutant molecule. The PCR prod(cid:173)
`ucts from the mutant will therefore make up 1% of the
`total products following SP-PCR, an amount whi.ch ena(cid:173)
`bles the mutant PCR products to be readily detected by
`gel electrophoresis 1.111d Southern blot hybridlzntion.
`Analysis of multiple pools of DNA enables 10000s of
`sperm to be surveyed for mutallon In 11 single experi(cid:173)
`ment. SP-PCR can only be applied to alleles short
`enough to be ampllfied efficiently in tbelr eniirety
`(<5 kbp); recent developments in
`long-range PCR
`[36, 37] may make it possible to analyse longer alleles for
`
`mutation, though electrophoretic resolution of mutants
`from their progenitors wiJl remain problematical.
`
`SP-PCR has been extensively validated at MS32, in par(cid:173)
`ticular to determine whether abnormal length PCR prod(cid:173)
`ucts are authentic mutants rather than artefacts arising
`durin~:~ PCR {35]. Various lines of evidence indicate
`authenlicily, including the quanta! nature and appro·
`priate intensity of mutant sperm PCR products, the cor(cid:173)
`rect '))roportionalily between input DNA and yield of
`mutants, the much lower frequency and different spec(cid:173)
`trum of mutants detected by corresponding analyses of
`somatic (blood) DNA, and the curious structures of new
`mutant alleles (see below) which are incompatible with
`PCR artefacts. Mutation rates as low a J0--4/sperm can
`be reliably measured by SP-PCR; below this level, .PCR
`artefact noise will progressively impede mutation detec(cid:173)
`tion. SP·.PCR is also being adapted to otber mini ·atellile
`loci, and has been applied to the (CAG). triplet repeal
`array in myotonic dystrophy to enable lhe heteroge(cid:173)
`neous smear of new roulant alleles detected in the
`sperm and somatic DNA of affected individuals to be
`resolved into individual mutant molecules [38].
`
`The Johns Hopkins University Exhibit JHU2010 - Page 4 of 9
`
`
`
`6/mrvphiJm/s 1995, 16, JS77-1S8S
`
`Mutotion processes ot human mlnisatellites
`
`1581
`
`7 Sperm mutation rates at MS32
`
`SP-PCR has now been used to measure sperm mutation
`rares directly in numerous men (35). The mean mutation
`rale is 0.8 °Al per sperm, as predicted from pedigree anal(cid:173)
`ysis and suggesting that SP-PCR is not being heavily
`biased by sampling mutation in abnormal sperm that
`cannot contribute to the nex.l generation, Curiou~ly, the
`•perm mutation rate per allele does not increase wi[b
`alLele length (over the sampled range of 22-164 repeats),
`a result which appenrs to be incompatible with mutation
`processes intrinsic to the repeat array, where the mula(cid:173)
`lion rate should increase with target size (array lenglh).
`Most mutation events involve the gain or loss of one or
`a few repeat units, with the size distribution of mutant
`alleles n:lative to the progenitor being appan~nlly con(cid:173)
`stant, and again independent of aHele length (Fig. 3).
`Most of these mutants are specific to sperm in lhnt they
`are not seen in blood DNA (Fig. 3). Sperm mutation at
`MS32 shows a remarkable 3:1 bias in favour of gains
`rather than losses of repeat units; both alleles in a man
`show the same bias, indicating an asymmetric mutation
`process rather than one allele gaining repeats at Lhe
`expense of the other. Pedigree analysis at other minisa(cid:173)
`tel)ltes has also provided evidence for expansion biOB in
`the mule gero,llne, suggesting that this may be a
`common feature of VNTR loci (10, 35]. The scale of U1e
`bias at MS32 has mf\ior population genetic implications;
`in particular, array lengths will not drift in a stochastic
`fashion but will increase determinjstically at a mean rate
`or l repeat added per 1000 years . A major question there(cid:173)
`fore is what forces act to prevent such loci from ex(cid:173)
`panding
`indefinitely. There are various possibililie ,
`Including counteracting delellons in the female germline
`(although such deletions have not yet been seen at
`MS32), increased rate of large deletions as alleles grow
`In length, truncating selection acting on chromosome
`c.Jysfunction induced by long arrays, failure of very long
`alleles to engage In the mutation process, and the popu(cid:173)
`lation spread of mutations that reduce or eliminate the
`mutational competence of minisatellites (see below).
`Which of these processes operate to prevent minisatcl(cid:173)
`lltcs from swamping the genome is ru~ yet unknown.
`
`8 Complex conversion events
`
`MVR-PCR sLructural analysis of new mutant alleles
`identified in pedigrees or recovered by SP-PCR anaJysis
`of sperm DNA has revealed an unexpectedly complex
`and bizarre mutation process operating at MS32 In the
`male germJine (Pig. 4) [35). As predicted from allele
`diversity studies, most mutation events are extremely
`polar, !nvolvi.ng repeat unit changes at the extreme end
`of the array over the region previously identified as a
`variability llotspol. In al least half oi' the sperm gain
`mutants, mutation involves the transfer of repeats or
`repeat unit blocks from one allele (tbe "donor") into Lhe
`other allele (the "recipient"). These inter-al.lelic transfers
`are most obvious when the donor allele contains MVR
`types not present in tbc recipient (Fig. 4). Transfer is fre(cid:173)
`quemly accompanied by complex rearrangements io the
`recipient allele, including "target site" duplications in
`repeat blocks flanking, or adjacent to, the site of inser-
`
`10 .------------------------------------r
`sperm
`01<3-llnked allel
`
`0
`
`8
`6
`
`X
`<l>
`i§
`<=
`0
`
`"" l1l
`5
`E
`
`8
`
`6
`
`4
`
`2
`
`0
`
`4
`
`2
`
`0 .
`
`4
`
`2
`
`blood : OtG-Iinked alleles
`
`sperm
`
`01 C-linked alleles
`
`· 20
`
`-10
`
`0
`change in no. repeats
`
`, 0
`
`20
`
`Flgur. 3, 0Qrmlino nnrl somatic mutation roles to new length alleles
`Bt MSJ2 !Ill meuured by SP-PCR. Rate! are given RS the frequenny
`per progenitor molecule of mutant molecules nf a given size class rela·
`tlvo to the progenitor. Top , mean distribution for 25 different OIG (cid:173)
`Ilnkotl ttllcte..q In sperm , determined from l80 000 progenitor mole·
`cules. Middle, m~nu dlslrlbulion fm 4 different 010-llnked nltcles In
`blood (56 000 progenitor molecule8 tested), Bottom, me en distribution
`for? diffcrenl OlC-linkcd ~llcl~ in sperm (90000 molecules tos!ed) .
`
`tlon, and occasional multiple rounds of imperfect ampli(cid:173)
`fication of repent blocks at or near the sJte of insertion
`to create mutant alleles much longer than the recipient
`and .in which the beginning of lhe mutating allele is pro(cid:173)
`found ly remodeUed. In some cases, scrambling of MVR
`types caa be so extensive as to create repeat segments in
`the recipient allele which have no obvious origin in
`either parental allele. lotcr-allelic transfer usually, but
`not always, involves the acquislti.on of repent unit blocks
`from
`the corresponding region of the donor allele,
`implying that alleles of different lengths are paired at
`their 5' end pdor to transfer. Even closely-flanking DNA
`markers are seldom colransferred during Inter-allelic
`transfer, though one example has recently been discov(cid:173)
`ered by pedigree analysis of minisatellile MS3l
`(D.L.Neil and A. J. Jelireys, unpublished). This suggests
`that the mutation process is largely restricted to the
`repeat array itself. Some sperm mutants do not involve
`inter-allelic tTan.sfer but can nevertheless show polarity
`and complex rearrangements compati ble with an anal(cid:173)
`ogous mutation process involving transfer between sister
`chromatids rather l.han ttlleles. TheTe is evidence thal the
`b!llance between inter- and intra-allelic events, and the
`degree of polarity of intra-ollellc mutation, can vary
`between loci [12].
`
`The Johns Hopkins University Exhibit JHU2010 - Page 5 of 9
`
`
`
`1582
`
`A. J. Jeffrey• el a/.
`
`Eltmophoml> 1995, 16, !5TI-l585
`
`allele l
`
`allele :a
`
`allele 2 mutantsz
`
`A
`
`B
`
`c
`
`D
`
`l!
`
`F
`
`eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeoooooeeeeeeeeoeeooeeo . ... . .
`Figure 4. Structures of mutant MS32
`alleles
`in sperm. These examples are
`taken from a man heterozygous for a
`long ( -800 repeal) and short (32 repeat)
`allele (alleles 1, 2, respectively). These
`alleles were mapped using four-state
`MVR-PCR to classify repeats as E-, e-, Y·,
`or y-type. Additional "null" repeats which
`to amplify with these MVR-PCR
`fail
`primers are shown as "o". The beginning
`of allele 1 is l8rgely homogeni~ed for
`e-lype repeat~ which are absent from
`allele 2 except for the first repeat. Sperm
`mutants derived from allele 2 were rccov-
`ered by SP-PCR and MVR mapped.
`Duplications Are shown hy double under(cid:173)
`lining of the 5' segment and single under-
`llnlng of the 3' segment. Repeats ac(cid:173)
`quired from all!!lc 1 ore 8.'1teri9ted, 11nd
`repeal~ of Wlch:ar origin tnarked ''I". The
`six mutant~ were welccted to show typical
`mutant structures. (A) A maJor expan(cid:173)
`sion Involving Imperfect amplificutlon of a donor/recipient junction. (B) insertion from allele l interrupted by anomnlous repeats, without loss
`of repeats in allele 2. (C) Extreme terminal repeat tran~fer. (D) Tran•f~r with target site duplication. (E) Deletion with lntor-allelic transfer.
`(F) Simple intra-allelic deletion.
`
`eYYoEEoEEEEoEEYEyyyYYYEYEYYYYYYY
`
`..
`
`*!••
`
`•• !
`
`eYYoEEoeeeeEEYeeeEEEEoEEYEyyyYYYEYEYYYYYYY
`
`eeeeYYoEEoEEEEoEEYEyyyYYYEYEi'YYYYYY
`
`eieeXYoEEoEEEEoEEYEyyyYYYEYEYYYYYYY
`
`eYYoEE-eEEEoEEYEyyyYYYilYEYYYYYYY
`
`eYYo---EEEEoEEYEyyyYYYEYEYYYYYYY
`
`Most of these complex mutanLS recovered from sperm
`are different, compatible with, though not proving, a
`meiotic origin . Somatic mutations by contrast involve
`much simpler intra-allelic duplications and deletions
`with less obvious polarity. Little is known about muta(cid:173)
`tions in the female g:ermllne in view of the limited
`number of mutants that nave been identified in pedi(cid:173)
`grees; current data arc, however, beginning to suggest
`that, wblle polarity does occur in oocytes, the details of
`the mutation process may well clitlcr substantially from
`that seen in sperm, in particular with respect to the fre(cid:173)
`quency of inter-allelic transfer.
`
`The major mode of m.injsatelllte mutation !n the male
`gormlinc therefore involv~ complex inter- and lnlra(cid:173)
`allelic gene conversion events. While many models can
`be postulated for this process, the simplest borrows from
`current models for the initiation of meiotic recombina(cid:173)
`in yeast [39-41], and
`tion
`involves double-s trand
`breakage followed by gap repair (Fig. 5) (35]. To account
`for polarity, we invoke a regulator of mlttation in the
`flanking DNA. This regulator could, for example, serve
`to direct an endonuclease to the beginning of !he array
`which cleaves the allele, thereby activating a mutation
`repair pathway; the presence of target site dunlications
`following mutation suggeslS that cleavage may involve
`staggered breaks
`in
`the recipient allele. Following
`cleavage, tile double-strand break can expand to form a
`gap (thereby explaining lhe bias towards galnln~ repeat
`units) which is then bridged by strand invasion from the
`donor allele or sister chromatid synapsed with lhe begin·
`ning of the recipient allele. Tho donor strnnd tllen acts
`as a template for gap repair syn thesis before being
`extruded from the repaired recipient allele, whicll now
`contains a segment of donor allele. This mutation ini(cid:173)
`tiator model accounts for polarity, pius constancy of
`mutation rate and spectrum irrespective of allele length,
`but does not explain some of the more complex prod·
`ucts of mutation; these may for example result from
`rounds of doublc-stTand breakage/repair
`multiple
`synthesis in tho conversion complex itself. This mutation
`
`process actively creat~ MVR interspersion ln alleles,
`and it is lb.erefore paradoxical lhot some MS32 alleles
`are largely if not completely homogenlsed for just one
`clnss of MVR [28]. One possibility ls that these alleles
`are defective as recipients in inter-allelic transfer, and
`are confined to intra-allelic repair pathways that could
`eventually lead to repeat unit fixation analogous to cross(cid:173)
`over fixation (2].
`
`A
`
`a
`
`B
`
`B
`
`IT'
`
`~~
`
`~2
`
`c
`
`donor
`
`recipient
`
`D
`
`0
`
`Figure J. A model for sperm mutallon Ill. human mini¥atellltes. ( l)
`Protein binds to a cis-act ing mutation Initiator element In the DNA
`l'lltnklng the tandem repeat moy (