Proc.Nat!.Acad. Sci. USA Vol. 85,pp.4794-4798,July 1988 Genetics
Lack of evidence for association of meiotic nondisjunction with particular DNA haplotypes on chromosome 21
(Down syndrome)
NICOLETTA SACCHI*t, JAMES F. GUSELLAt, LUCIA PERRONI*, FRANCA DAGNA BRICARELLI§, AND TAKIS S. PAPAS*
*Laboratory of MolecularOncology,National CancerInstitute,Frederick,MD 21701-1013;tNeurogenetics Laboratory,Massachusetts GeneralHospital, Departmentof Genetics, Harvard University, Boston,MA02114; and§Centrodi Genetica Umana, E.0. Ospedali Galliera, Genoa, Italy
Communicated by George J. Todaro, January 28, 1988
ABSTRACT Thehypothesisofapredisposition to meiotic nondisjunctionfor chromosome 21 carrying aspecificmolec- ularhaplotype has been tested. Thehaplotypeinquestion is definedby the restrictionfragmentlength polymorphismsfor theD21S1/D21S11 loci.Ourresultsobtainedon asampleof Northern Italian families with the occurrence oftrisomy 21 (Down syndrome)failedtosupportthishypothesis,contradict- ingaprevious study [Antonarakis, S. E., Kittur,S.D., Me- taxotou, C., Watkins,P. C. &Patel, A. S.(1985)-Proc. Natl.
Acad. Sci. USA 82, 3360-3364]. These findings rule out an associationbetween anyspecificD21SI/D21SII haplotype (as wellasotherhaplotypes for theD21S13, ETS2, and D21S23 loci) and a putative cis-acting genetic element favoring the meioticmissegregation ofchromosome21.For this reason, no preventive screeningfor couples at riskfortrisomy 21 may be basedonanyof thehaplotypestested.
Errors inthe transmission ofgenetic material-a process that must occur with a high degree of precision-lead to aneu- ploidy in eukaryotic. organisms. Little is known about the causes for incorrect pairing and aberrant segregation of chromosomes, but these mechanisms are likely tobe influ- encedby bothgeneticandenvironmentalfactors (1). It has beenproposedthatcorrectchromosomepairingorsynapsis plays a critical role in ensuring appropriate segregation, although the evidence todateis not conclusive. Therefore, genes encoding proteins involved incontrollingorcarrying outchromosomepairing andsegregationmightbe involved in abnormalmeioses. Studiesof mutants inDrosophila pointto alarge number of loci thatcanaffettsegregationof chromo- some pairs and suggest an inverse relationship between recombination and nondisjunction (NDJ) (2, 3). Genetic effectsonmeiosismightalsobeexpectedtooperatethrough particularDNA sequences thateukaryoteshave evolved to fulfillspecificmeiotic functions.IntheyeastSaccharomyces cerevisiae, centromeric DNA sequences essential for both reductional and equational meiotic divisions have been de- fined(4). InLilium, DNA sequences have been found that mayplay a role in the chromosome pairing or synapsis (5). No specificgene loci orstructural DNAsegments in man have yet been implicated in chromosome NDJ, although aneu- ploidy is the basis of many severe human disorders.
The mostcommon humandisorder resulting from aberrant chromosomesegregationis Downsyndrome(DS) or trisomy 21,affecting " 1 birth in 1000 (6). Recently, an association has been reported between NDJ ofchromosome 21 and ahap- lotypedefinedby restriction fragment length polymorphism (RFLP)atananonymous DNA locus onthis autosome (7). In view of the profound impact of such a finding for genetic
counseling andpreventionofDS,weinvestigatedthegeneral validity of this observation. In contrast to theprevious study, wedid not observe any particular haplotype on chromosome 21 associated with NDJ chromosomes-i.e., chromosomes thatdid not undergodisjunction (DJ).
MATERIALS AND METHODS
FamilySamples. Weexamined 23 Italianfamilies(groupA) asacontrol group and 37 families withaDS child(groupB).
All of thecontrol families consisted ofmother,father, and one normal child (NC). Of 37 families withaDS child, 23 consisted of mother, father, and the DS child, and the remaining 14consisted ofthe parentsand twochildren,one NC and one withDS;of the 37families,35wereinformative for the origin ofnondisjunction (8), as described in detail elsewhere. Our DS familieswere notselected for advanced maternal age,having 5 mothers s 25 years, 16 between 25 and 35 yr, and 6 . 35 yr.
Cytogenetic Analysis. Theparentaloriginand stageofNDJ were determined byevaluating the chromosome 21 satellite polymorphisms (8). Double-blind microscope scoring was performedinall cases, and theorigin ofNDJattributedonly incaseof concordance(8).
Southern Blot Analysis. DNA was obtained from 20-ml samples of blood with EDTAasanticoagulant. Erythrocytes were lysedin 20 mM Trischloride, pH 8.0/10mM EDTA.
The pelletwas washed threetimes with the same solution, resuspended in 5 mlof150 mMNaCl/2mMEDTA/10mM Trischloride, pH7.5/0.05%NaDodSO4 and incubatedover- nightat55°Cin the presence ofproteinase Kat 100
,ug/ml.
DNA was extracted three times with phenol/chloroform,
24:1 (vol/vol) and once with chloroform. The last upper phasewasprecipitatedwith 2.5 volof absolute ethanol in the presence of 0.3 M NaC2H302. DNA was recovered by spoolingandwasdissolvedin 1 ml of 10mMTrischloride/i mM EDTA, pH 8.0. DNAs were digested with several restriction enzymes under conditions recommended by the suppliers (Boehringer Mannheim), electrophoresed in0.8%
agarose gel, and blotted onto a nylon membrane (Nytran;
Schleicher & Schuell). Prehybridization was performed at 42°Cforatleast2hrin1 MNaCl/50mMTrischloride, pH 7.5/1%NaDodSO4/50 ,ugofsalmonsperm DNA perml/10x Denhardt's solution (0.02% polyvinylpyrrolidone/0.02%
Ficoll/0.02%bovine serumalbumin).Thehybridizationwas overnight at 42°C in 1 M NaCl/50 mM Tris chloride, pH 7.5/1%NaDodSO4/50%formamide/50,ugof salmon sperm DNAper ml. The32P-labeledprobes (specificactivity,2 x Abbreviations: DS, Down syndrome; RFLP, restriction fragment lengthpolymorphism; NC,normalchild;NDJ,nondisjunction; DJ, disjunction.
tTowhomreprintrequestsshould beaddressed.
4794 Thepublicationcostsof this article were defrayed in part by page charge payment.This articlemust therefore behereby marked"advertisement"
in accordancewith 18U.S.C. §1734solely to indicate this fact.
Proc.Natl. Acad. Sci. USA8S (1988) 4795 109cpm/,ug) were added to 5-10 ng/ml of thehybridization
solution.
Filters were washedtwicefor15min at room temperature in 6x SSC (1x SSC = 0.15 M NaCl/0.015 M sodium citrate)/0.1% NaDodSO4, twice for 15 min at 370C in 1 x SSC/1% NaDodSO4, and finally for 1 hr at 650C in 0.1 x SSC/1% NaDodSO4. The membranes then were exposed to Kodak XAR-2 films at -700C for 24 hr.
Chromosome 21 Sequences. The probes used for the hap- lotype analysis were single-copy chromosome 21 sequences chosen for mapping in differentregions of the chromosome (Fig. 1).Withtheexclusion of H33 ETS2 (9), which is part of a transcriptionally active gene, ETS2, all the others are anonymous sequences (ref. 10; R. E. Tanzi, personal com- munication). D21S13, a 7-kilobase (kb) EcoRI fragment located closest to the centromere, identifies a Taq I poly- morphic site. D21J1, a 1.5-kb EcoRI fragment, recognizes two polymorphicsites (BamHI and MspI).D21J1S,a1.85-kb EcoRI fragment, identifies an EcoRI and a Taq Ipolymorphic site. D21JS andD2JSJJ map in close proximity in the q21 region of chromosome 21. H33 ETS2, a 1-kb EcoRIfragment mappingtothe q22.3region of chromosome 21,identifiestwo polymorphic sites (Msp I and TaqI). D21S23, an EcoRI fragment of 0.95-kb located in the q22.3 region of the chromosome,recognizesapolymorphism with either of two enzymes (EcoRI orMsp I).
Statistical Methods. The difference between comparable haplotypedistributions was tested by
theX2
testfor indepen- dence. When 2 x 2 tables were being compared, the continuity correction wasapplied (11); when expected fre- quencies in these tables were small, the Fisher exact test was usedin lieu of the x2 test.RESULTS
The total number of chromosomes 21 taken into account in the23-control families (group A) was 92. In the second group of families with aDS child (group B), the total number of chromosomes 21 analyzed was 140. The origin, whether
Chromosome21 Sequences
D21S13
qll
RFLPs
Identifiedby Chr21 Sequence
Taql 7.0(-) 6.0(+)
paternal or maternal, of the extra chromosome 21 was attributed in 35 DS childrenout of 37on the basis of both cytogeneticandmolecular data thatwereeitherinformative additivelyorconcordant(8).MaternalNDJwasthecauseof trisomy in 25 (72%) of the cases, occurring in the great majorityatthe first meiotic division (22 cases). The origin of NDJ waspaternalin 10 (28%) of the cases.Therefore, as far astheorigin of NDJ is concerned, the families of group Bare sufficiently representative of the values in the literature (12, 13), having a 3:1 ratio of maternal versus paternal. The chromosomes 21 within group B could be allocated to several subgroups: 62 NDJ chromosomes [NDJ at the first meiotic division (NDJI) in 54 and at the second meiotic division (NDJII) in 8] and 78 DJ chromosomes (chromosomes that underwentDJ).
The DNA sequences used toidentify specific chromosome 21haplotypes are reported inFig. 1. Notably, D21S13 isan anonymous sequence on theproximal long arm. D2JSI and D21SJJ are the two closely associated sequences about 15 centimorgans from the centromere (R. E. Tanzi, personal communication) thatwereused in the previouslymentioned study (7). RFLPs at D21SJ and D21SJJ display linkage disequilibrium and can be considered as asingle haplotype, since we haveobserved norecombination between the two loci in 145 informative meiotic events. D21S23 and the genomic sequence for the H33 ETS2 gene both map to the q22.3 region. Thehaplotypes identified by the RFLPs of the different sequences, together with their distribution in the controlgroupA ofchromosomes 21 wereas follows. Two haplotypes + and -, with afrequency of 0.58 and 0.42, were identified by the Taq I RFLP for theD21S13 sequence; the BamHI andMsp I RFLPs fortheD2JSJ sequence and the EcoRI and Taq I RFLPs for theD2JS11 sequence identify mainly three haplotypes, + ++ + (0.40), + - - - (0.26), and - + + + (0.25), out of the 16 expected haplotypes.
Three,- - (0.63), + + (0.21), and +-(0.15), werethe most recognizable haplotypes of the four expected for the H33 ETS2 Msp I and Taq I RFLPs. Finally, the EcoRI, Msp I RFLPs fortheD21S23 sequencerecognizetwohaplotypes,
Haplotype Precentage
DS inGroup A
+ 0.58
- 0.42
13 (
P 12
11.2
11.1 11.2 21 q
22.1 22.2 22.3
D21S1 q21
D21S11 q21
H33ets-2 q22.3 HumanChromosome21
D21S23
q22.3
BamHl Mspl
_ 7.0(-) 7.6(-) 6.3(+) 5.0(+)
+.+ ++ 0.40 + - - - 0.26
-+ + + 0.25
Others 0.09
EcoRI Taql
2.9(-) 5.5(-) 1.9(+) *, 4.0(+)
Mspl Taql
146' 4.0(-)
1.4(-)
1.1(+) 2.1*
1.2(+)
EcoRI Mspi
1.4(-) 2.1(-) 1.1 I+) 1.8(+)
FIG.1. RFLPsof thearbitrary
_
o0.63
chromosome 21sequencesused in+ + 0.21 thisstudy. Their location is indi-
+ - 0.15 cated. The + and - indicate the Others 0.01 presence or the absence of a poly- morphic fragment. (The asterisk refers to an invariant fragment.) Notallpotentialhaplotypeswere observed. For theD21JS/D2IS11, mainly 3 haplotypes occurred of the 16expected,and for the H33- + + 0.61 ETS2,onlythreeoccurred of four - - 0.39 expected.
Genetics: SacchietA
Proc. Natl. Acad. Sci. USA 85 (1988) c
* .
M V o
? ?
DS
* v
I
D21Sl/D21S11°l
d
F M
QQ53J
OL
? OENC
?0 ?@ +
D21SIID21S11
FIG. 2. The assignment of the haplotypefor each individual chro- mosome 21was possiblein group B on the basis of knowledge of the haplotype of theNC(a) andknowl- edgeof the parental origin of NDJ (morphologic and molecular poly- morphism) coupled to a suitable pat- ternof RFLPs (b). The assignment was notpossible when theinforma- tion was represented only by the NDJ data (c) or when bothparents and NC in group A were heterozy- gousforoneormore sites(d).
+ + and --, withfrequencies of0.61 and 0.39, respectively (Fig. 1).
The assignment of the haplotypes was possible in the majority of cases for groups A and B, combining both the cytogenetic and -molecular data of the father (F), mother (M), and children(DS or NC) (8). The criteria used are exemplified inFig.2. IngroupB,either the knowledge of thehaplotype of the NC(Fig.2a) orthe knowledge of the cytogenetic origin of NDJ (morphology of satellites) coupled to a suitable pattern of RFLPs (Fig. 2b) was helpful in determining the linkage phase. A few recombinational events preceding the NDJs at meiosis I and II were recognized by combining both the cytogenetic and the molecular analysis (8). In all cases,
DJ
i, 78chr
Group A =0.86 lo
0
NDJ
62 chr
D21S13
theserecombinations had an effect on the repositioning of the more distal haplotypes for the ETS2 and D21S23 loci. In theory, therepositioning of a given haplotype occurring in prophase Imightaffectthe subsequent segregationalphases.
Forthis reason, we took into accountthe effect of recombi- nations. In these cases, the haplotypes were assigned to the recombinant chromosomes. The recombinant chromosomes weredefined by their centromere and accordingly distributed in theproper grouping. The assignment of thehaplotype was impossible either when the information representedby the NDJ data wasinsufficient(Fig- 2c) or when (in group A) the heterozygosity for one or more polymorphic sites was pre- sent in both the parents and the child (Fig. 2d). Forthese
DJ
11% 70chr
9 X*
GroupA P=0.18 .1 Group B
71chr 126 chr
NDJ
56chr
H33 ets-2
DJ
s 77 chr
Group A = 0.65 - Group B
84chr 138chr
NDJ
61chr
D21S23
FIG. 3. Thediagrams summarize how the distribution of each hap- lotypewascomparedamongthevar-
iousgroupsofchromosomes 21. Each
arrowpointstothetwogroupsunder comparison. The size of eachgroup
for the haplotype in question is also reported. The P valueswereobtained bythe %2test.
a F M
NC
i++0+i+
DS D21SIID21S11b
F
* 0
:Qf O
M
V o
DS
* v o
H33ots-2
SA
9DJchrGroup
B Group A P= 0.55 ' A Group B140 chr 75chr 122chr
NDJ
53chr
D2IS1-D21SII 47% Genetics: Sacchi etal.
Proc.Nati. Acad. Sci. USA 85 (1988) 4797
Table 1. Distribution analysis ofthe haplotypes identified byH33ETS2 probe
Haplotypes, no. Total Statisticaldata
Group + + +- -- no. X2 P df
A 15 11 45 71
B 33 30 63 126 A--B 3.47 0.18 2
DJ 17 15 38 70 A--DJ 1.32 0.52 2
NDJ 16 15 25 56 A-.NDJ 4.65 0.098 2
DJ--NDJ 1.17 0.56 2
NDJI 14 13 22 49 A- NDJI 4.2 0.12 2
DJ--NDJI 1.02 0.60 2
NDJII 2 2 3 7 A-.NDJII 1.25 0.53 2
DJ-*NDJII 0.35 0.84 2
Example of statistical elaboration to compare the distribution of the haplotypes in the different groups ofchromosomes. The example refers to the haplotypes identified by theETS2 probe. Groups: A, chromosomes 21 from control parents; B, chromosomes 21 from parents with a DS child; DJ, chromosomes 21 of groupBthat underwent disjunction; NDJ, chromosomes 21 that did not undergo disjunctionatthefirst meiotic division (NDJI)or at the second (NDJII).
reasons and for the exclusion of the most infrequent hap- lotypes, the total numberof chromosomes considered for the statistical analysis in each group varies slightly for the different haplotypes.
Fig. 3 represents diagrammatically the summary of the overall statistical analysis. The comparison of thedistribu- tions of the haplotypes between each group and the other groups wasperformed bythe x2 testforindependence. The probability values(P) were inall cases calculated as exem- plified in Table 1 that contains the data for the ETS2 haplotypes. In the diagrams of Figs. 3, the final P values deducedfrom the overall calculationsarereported. From the various comparisons it isapparentthat no significant differ- ence(P>0.01) exists betweengroups AandB,even when the NDJ and DJgroups areconsidered separately.Moreover, eventhedistinction of NDJgroups onthebasis offirst(NDJI) andsecond (NDJII) meioticerrordoesnotsignificantly affect thePvalues(Fig. 4).Inthe lattercase,because ofthe small numberof chromosomes of theNDJII group, thestatistical test usedwas theFisherexact test(FET).
A study considering only the D21S1/D21S11 haplotypes wasperformedonGreek subjectsby otherauthors (1). That study seemedtopointto acorrelation betweenthehaplotype -+ + + andapropensitytomissegregation ofthe chromo- some21 marked with that haplotype. For this reason, the Greek and Northern Italian chromosomegroups were com- paredfor thehaplotype in question. Table2shows that the
NDJ I
10. 54 chr *
Group A DJ
92 chr 78 chr
% NDJ II
8 chr
D21S13
9
Group A
75chr
.Z*
Greek andItalian controlgroups(Group A) areindistinguish- able (P = 0.95) as are the B (P = 0.63) and DJ groups (P = 0.52). The only significant difference (P = 0.013) was found for theNDJ groups.
DISCUSSION
In experimental organisms (Drosophila), there is evidence thatgenetic factors may increase the frequencyof chromo- somemissegregation during the meioticprocess. Thepossi- bility that this mayalso be thecasein humans isintriguing.
A recentstudy has suggestedagenetic predispositiontoNDJ for human chromosome 21-that is,apredispositiontoDown syndrome (7). Specifically, inaninvestigation on agroupof Greek families, chromosomes 21 marked with a particular DNAhaplotype seemedto bepreferentially involved in the NDJ events. The haplotype in question (- + ++) was definedby four closely linked polymorphicsites: BamHI and MspI inD21JS and EcoRI andTaqIin D21S11. Thisfinding implied some association between this haplotype and a putativecis-acting genetic element influencing NDJ. If sub- stantiated, it wouldrepresent a way to screenforcouplesat increased risk for trisomy 21 offspring.
To test the general applicability of that observation, we undertook an extensive investigation ofa larger cohort of Northern Italian individuals. In addition to the haplotype used in theprevious work, weanalyzed controlhaplotypes for chromosome 21sequenceslocalized indistinctregionsof NDJ I
.0
46chr 0
DJ
69chr
NDJ II
7chr
D21SI/D21S11
NS NDJ I O
1*10 48chr * 00
Group A DJ
71chr 70 chr
0k NDJII 9
8chr
H33ets-2
41 NDJ I
If,10 53 chr
Group A DJ
84 chr 77 chr
.? NDJ II °
8chr
D21S23
FIG. 4. Haplotype distributions of groupsAand DJarecomparedwith thoseof the NDJ group, whichwasclassified intwo
subgroups,NDJ Iand NDJ II,accordingto the stage ofoccurrenceof the meioticerror.
The Pvalues indicatedwereobtainedbythe
X2test orFisherexacttest.
Genetics: Sacchietal.
Proc. NatL. Acad. Sci. USA 85 (1988)
Table 2. Comparison of thedistribution of haplotypes identified by D2121 andD21SJ1 withinthe groupsinGreek and Italian subjects
Haplotypes, no. Statisticalanalysis
Group Subjects + ++ + - ++ + + - - - Total X2 P df
A Greek 43 23 24 90 0.098 0.95 2
Italian 34 20 21 75 008 09
B Greek 15 17 16 48 0.91 0.63 2
Italian 46 35 41 122 0.1 .6 2
DJ Greek 14 5 10 29
Italian 27 19 23 69 1.3 0.52 2
NDJ Greek 1 12 6 19 8.64 0.013 2
Italian 19 16 18 53
Distribution of haplotypes defined by polymorphic sites identified by D21S1(BamHIand MspI)and D21S1I(EcoRIand TaqI)withinthe various chromosome groups in Greek and Italiansubjects. The values forx2,P, and the degree of freedom (df) are reported.
thechromosome.Fromtheoverallstatistical comparisons of the distribution of the various haplotypes, we could not conclude thata particular haplotype is characteristic in the NDJ chromosomes that we considered, even when we distinguished those that were involved in the reductional segregation of meiosisI (NDJ I)from those involved in the equational segregation of meiosis II (NDJ II).
Therefore, ourresults donotshowanygenetic predispo- sitiontoNDJ linkedtothearbitrary chromosome 21-specific DNAsequencesanalyzed,including the D21SJ and D21S11 sequencesused in the other study (7).
The difference between ourfindings and the conclusion based ontheGreek population is puzzling. Thetwo ethnic groupsdo not seemtodiffer because thedistribution of the haplotypes in question is identical in control subjects ofgroup A(P=0.95) andsimilar in theB(P=0.63)andDJ(P= 0.52) groups (Table2). The onlysignificant difference is in theP values obtained from thecomparison ofgroupsAandNDJ, givinga Pvalue of 0.56 for the Italian sample and of0.0005 for the Greeksample (Fig. 5). The probability of 0.013 (Table 2 and Fig. 5) associated with the NDJ groups is astrong indication of a difference between the Greek and Italian subjects. While the relatively small size of the Greek sample doesnotinvalidate the aboveconclusion,alarger sample of Greeksubjects might have producedalessstriking difference in the ++ + + haplotype (Table 2), which is the largest componentof the overallx2.
Fromourdata,weexcludeanassociation between specific haplotypesdetected bythesechromosome 21 sequences in NorthernItalian individuals.As a consequence,it wouldnot beappropriateto usethesespecificmolecularhaplotypes for identifying couples at risk for having a DS child in this population.Whetherageneticcomponent onchromosome21 itselfmayinfluence the tendencytochromosome21misseg- regation remainstobe determined. Success in identifying this componentmightbeachieved by focusingonprobes closeto the centromere. Thequite consistent distance (15centimor- gans) (R. E.Tanzi, personal communication) oftheD21S1/
GroupA 108 chr.
Greek
p=0.0005
p=0.54 GroupA -
75chr. Italian
NDJ 20 chr.
NDJ 53chr.
p=0.013
FIG. 5. Thedistribution of the haplotype for theD2151/S21SII polymorphicsites iscomparedbetween the A andNDJgroupsin the Italian andGreeksubjects.
D21JSI loci fromthe centromere issupportiveofthefindings we described. It is difficult to perceive how an aberrant cis-actingelement onchromosome21would have remained linkedoverthe course of the evolution to both the centromere and the distantD2JSI/D21SII haplotype (- +++). This latter argument makes itunlikelytointerpretthediscrepancy between theprevious study and ourstudy onthe basis ofa difference in such linkage between Greeks and Northern Italians. Mostlikely the discrepancy might be reconciled by applying a more rigorous definition of the origin of the chromosomes in the sample ofGreeksubjects.
Inthecontextof the abovediscussion,anyapplicationof thehaplotypetest(7)toclinicalpractice for Down syndrome prevention should inouropinion be viewed with caution.
Wethank P. C. Watkins(IntegratedGenetics, Framingham, MA) and D. K. Watson(NationalCancerInstitute, Frederick, MD),for the probes; A. Arslanian, M. Pierluigi, and M. Grasso (E. 0.
Ospedali Galliera, Genoa, Italy)forcytogenetics studies;C.Riggs (Information Management Services, Inc., Frederick, MD) for the statistical analysis; and K. Cannon for typingthe manuscript. A special acknowledgment goestotheAssociation of DownSyndrome Ce.PiM, Genoa, Italy.
1. Hoffmann,G. R.(1985) inAneuploidy: EtiologyandMechanism, eds.Dellarco,V. L., Voytek,P. E. &Hollaender,A.(Plenum,New York), pp.539-549.
2. Baker,B. S., Carpenter, A.T.C., Esposito, M. S., Esposito,R. E.
&Sandler,L. (1976)Annu. Rev. Genet. 10,53-134.
3. Sandler,L.(1981)inTrisomy21:Research Perspectives,eds. De La Cruz,F. F.&Gerald,P. S.(UniversityParkPress,Baltimore) pp.
189-1%.
4. Yeh,E. &Bloom,K.(1985)inAneuploidy:Etiology andMecha- nism, eds. Dellarco, V. L., Voytek, P.E. & Hollaender, A.
(Plenum,NewYork), pp.231-241.
5. Hotta, Y., Tabata, S., Stubbs, L. & Stem, H. (1985) Cell 40, 785-793.
6. Hook, E. B. (1985) Aneuploidy: Etiology and Mechanism, eds.
Dellarco, V.L., Voytek,P. E. &Hollaender, A. (Plenum, New York), pp.7-33.
7. Antonarakis,S. E., Kittur,S.D., Metaxotou, C., Watkins,P.C. &
Patel,A.S. (1985)Proc.Natl. Acad. Sci. USA82,3360-3364.
8. DagnaBricarelli, F., Pierluigi, M., Perroni, L., Grasso, M.,Arsla- nian,A. &Sacchi,N.(1988)HumanGenet.,inpress.
9. Watson,D.K., McWilliams-Smith,M.J., Nunn,M.F., Duesberg, P.H.,O'Brien,S. J. &Papas,T.S.(1985)Proc.Natl. Acad. Sci.
USA82,7294-7298.
10. Watkins, P.C., Tanzi, R.E., Gibbons, K.T., Tricoli, J.V., Landes, G., Eddy, R., Shows,T. B.&Gusella,J. F.(1985)Nucleic Acids Res. 13,6075-6088.
11. Armitage, P. (1971) Statistical Methods in Medical Research (Blackwell, Oxford), pp.121-211.
12. Frateschi, M., Arslanian, A., Pierluigi, M., Ferro,M. A.,Gessaga, M.,Coviello,D. A., Strigini, P. &DagnaBricarelli, F. (1984)in Aspetti epidemiologici, genetici, clinici, riabilitativi socialidella Sindrome di Down(Ce.Pi. M.,Genova), pp.79-93.
13. Schmidt, R., Dar, H.&Nitowsky,H.M. (1975) Pediatr.Res.9, 318-325.
4798 Genetics: Sacchietal.
