Chromosoma (1998) 107:166±172
Springer-Verlag 1998
Centromeric genotyping and direct analysis of nondisjunction
in humans: Down syndrome
J.J. Shen1, S.L. Sherman1, 2, Terry J. Hassold1
1
Department of Genetics and the Center for Human Genetics, Case Western Reserve University School of Medicine and University Hospitals
of Cleveland, Cleveland, OH 44106, USA
2
Department of Genetics, Emory University School of Medicine, Atlanta, GA 30322, USA
Received: 12 January 1997; in revised form: 16 February 1998 / Accepted: 19 February 1998
Abstract. In species with chiasmate meioses, alterations
in genetic recombination are an important correlate of
nondisjunction. In general, these alterations fall into one
of two categories: either homologous chromosomes fail
to pair and/or recombine at meiosis I, or they are united
by chiasmata that are suboptimally positioned. Recent
studies of human nondisjunction suggest that these relationships apply to our species as well. However, methodological limitations in human genetic mapping have made
it difficult to determine whether the important determinant(s) in human nondisjunction is absent recombination,
altered recombination, or both. In the present report, we
describe somatic cell hybrid studies of chromosome 21
nondisjunction aimed at overcoming this limitation. By
using hybrids to ªcaptureº individual chromosomes 21
of the proband and parent of origin of trisomy, it is possible to identify complementary recombinant meiotic
products, and thereby to uncover crossovers that cannot
be detected by conventional mapping methods. In the
present report, we summarize studies of 23 cases. Our results indicate that recombination in proximal 21q is infrequent in trisomy-generating meioses and that, in a proportion of the meioses, recombination does not occur anywhere on 21q. Thus, our observations indicate that failure
to recombine is responsible for a proportion of trisomy 21
cases.
Introduction
In species with chiasmate meioses, alterations in genetic
recombination are an important correlate of nondisjunction. For example, in Saccharomyces cerevisiae and in
Drosophila, meiotic mutants that reduce or abolish recombination typically increase the likelihood of meiotic
malsegregation (e.g., Hawley 1988; Rockmill and Roeder
1994). Similarly, spontaneous nondisjunction also has
Edited by: H.F. Willard
Correspondence to: T.J. Hassold
been linked to abnormalities in recombination. In Drosophila, nondisjunction of the X chromosome is associated with failure to recombine or, if recombination has occurred, with chiasmata positioned either extremely proximally or distally (Koehler et al. 1996a). In S. cerevisiae,
disjunction of artificial chromosomes is dependent on the
presence of genetic exchange; however, like Drosophila,
the presence of an exchange does not necessarily ensure
proper segregation, with distally located exchanges being
less able to ensure segregation than more proximal events
(e.g., Bascom-Slack et al. 1997).
In humans, meiotic nondisjunction is extraordinarily
common, with at least 3%±4% of all clinically recognized
pregnancies being trisomic (Hassold and Jacobs 1984).
Recently, considerable information has been obtained
on the meiotic origin of the additional chromosome in human trisomy. For example, data from DNA polymorphism studies of over 1000 trisomy families are now
available on: (1) the parent of origin of nondisjunction,
(2) the meiotic stage of the error, and (3) the frequency
and placement of recombinational events in the nondisjunctional meioses (reviewed in Hassold et al. 1996).
The results of these studies suggest that most human trisomy originates in oogenesis, typically at the first meiotic
division, and indicate that genetic recombination is an important correlate of nondisjunction.
However, the correctness of these conclusions depends
on the reliability of the test assays used to determine the
parent and meiotic stage of origin and the occurrence of
recombination. The information on parental origin is likely to be accurate, since it can be verified by genotyping
multiple loci. However, the other two types of analysis
are less straightforward. For example, specifying the meiotic stage of nondisjunction requires analysis of the inheritance of centromeric polymorphisms. In practice, this has
been accomplished by analyzing centromeric alpha-satellite polymorphisms, using conventional or pulsed-field
gel electrophoresis (PFGE) and Southern blotting techniques (e.g., MacDonald et al. 1994; Hassold et al.
1995), or by using the polymerase chain reaction (PCR)
to study tightly linked microsatellite polymorphisms that
167
flank the centromere (e.g., Fisher et al. 1995; Hassold
et al. 1995).
Unfortunately, neither of these approaches is suitable
for trisomy 21, clinically the most important of all human
chromosomal abnormalities. Centromeric alpha-satellite
polymorphisms exist for chromosome 21 but, as the alphoid array shares extensive sequence homology with the
centromere of chromosome 13 (Jorgensen et al. 1987), interpreting the pattern of inheritance of centromeric alleles
of chromosome 21 is difficult (Charlieu et al. 1993). Similarly, the approach of utilizing microsatellite markers of
chromosome 21 in making meiotic stage determinations
is questionable. PCR-based 21p polymorphisms are not
available, making it impossible to bracket the centromere
of chromosome 21, and there is little information on the
genetic distance between the centromere and the proximal 21q microsatellite markers used to infer the meiotic
stage of origin (Jabs et al. 1991; Van Hul et al. 1993).
Thus, a proportion of meiotic stage assignments likely
are incorrect, making previous estimates of the relative
frequency of meiosis I and II (MI and MII) nondisjunctional errors in the genesis of trisomy 21 (Sherman et al.
1994) uncertain.
Similarly, the third type of analysis, determining the
location and frequency of crossover events in nondisjunctional meioses, is also subject to question. While there is
little doubt that reduced recombination is an important
contributor to human nondisjunction (e.g., Sherman et al.
1994; Koehler et al. 1996b), there is less certainty regarding the basis for the reduction; i.e., whether it is attributable to failure of recombination in a proportion of cases,
or to altered ± but not necessarily absent ± recombination
in most cases. This uncertainty derives from the fact that
only two of the four meiotic products are recovered in the
trisomic conceptus; thus it is impossible to detect all recombinational events between the nondisjoining chromosomes. For example, assuming a single exchange between
the chromosomes 21 in a nondisjunctional meiosis, we
expect 1/2 of the exchanges to be ªundetectableº in conventional mapping studies: 1/4 because the two recovered
products will not have been involved in the exchange and
1/4 because complementary recombinant products will
have been recovered.
To address these uncertainties, we have recently initiated somatic cell hybrid studies of human nondisjunction,
using trisomy 21 as a model. By using hybrids to ªcaptureº individual chromosomes 21 of the proband and parent of origin of trisomy, it is possible to follow the inheritance of chromosome 21 alpha-satellite alleles, and
thereby to determine unequivocally the meiotic stage of
origin of trisomy. Further, this approach makes it possible
to identify complementary recombinant meiotic products,
and thereby to uncover crossovers that cannot be detected
by conventional mapping methods. In the present report
we describe the basis for this experimental approach,
and summarize results on an initial series of 23 cases of
trisomy 21.
Materials and methods
Down syndrome families. Trisomy 21 families were ascertained as
part of ongoing cytogenetic, molecular and epidemiological studies
of Down syndrome (Sherman et al. 1994), or as part of studies of
Down syndrome and leukemia (Shen et al. 1995); in these studies,
conventional DNA polymorphisms were used to determine the parent and meiotic stage of origin of the extra chromosome 21, and to
assess the frequency of recombination between the nondisjoining
chromosomes. All cases selected for inclusion in the present study
were nonmosaic, maternally derived trisomies that fell into one of
two categories: (1) those in which the extra chromosome had been
scored as arising at maternal MI and in which there was no evidence
for recombination in the trisomy-generating meiosis (n=16) and (2)
those in which the extra chromosome had been scored as originating
at maternal MII (n=7).
Somatic cell hybrids. Lymphoblastoid cell lines were established
from trisomic probands and their parents and maintained in RPMI
1640 with 15% fetal bovine serum (Gibco-BRL). To create somatic
cell hybrids, lymphoblastoid cell lines from trisomic probands and
their mothers (the parent of origin of trisomy in all cases in the present study) were first expanded, then fused using polyethylene glycol
(Boehringer Mannheim), to the CHO purine auxotrophic cell line
Ade-C. Ade-C is deficient for glycinamide ribonucleotide formyltransferase, encoded by a gene located on human chromosome 21.
Thus, to select for the retention of human chromosome 21, somatic
cell hybrids were maintained in hypoxanthine-free medium with
15% serum. Following initial fusion experiments, individual colonies were transferred to 24-well plates for subsequent passaging
and analysis.
Characterization of hybrids. In initial studies of two families, we
used both fluorescence in situ hybridization (FISH) and molecular
techniques to examine the number and structural integrity of the
chromosomes 21 in the hybrids. However, as we observed no obvious discrepancies between the two approaches, we eliminated the
FISH studies and relied on the molecular approach in studies of
all other families. For all hybrid characterizations, DNA was extracted with the Puregene kit (Gentra Systems). Initially, the hybrids
were screened using two or more microsatellite markers of chromosome 21: one pericentromeric (D21S215, D21S258, D21S120, or
D21S192) to infer the presence of a centromere, and at least one
in medial/distal 21q (D21S11, D21S214, D21S232, D21S210,
D21S226, D21S213, IFNAR, D21S167, D21S156, HMG14, PFKL,
D21S212, D21S171, or D21S1575) to assess whether or not the
chromosome was intact; all information regarding primer sequences, chromosomal location, and amplification conditions are provided in the Genome Database (Johns Hopkins University, Baltimore
Md.). Hybrids that displayed only one allele at each locus were considered to be monochromosomal 21 ªcandidatesº and were passaged
for further characterization. Subsequently, these were genotyped at
all 18 loci and those with evidence for a single allele at all informative loci, including markers in proximal, medial and distal 21, were
considered to be monochromosomal 21 hybrids. To screen for the
presence or absence of pericentromeric sequences of chromosome
13, all such hybrids were also studied using proximal 13q markers
(D13S175, D13S232, D13S1316, or D13S1236); for these studies,
multiplex PCR was performed with chromosome 21 markers acting
as a positive internal control.
For each family in the study, we attempted to obtain five monochromosomal 21 hybrids: two from the mother (containing each of
her two chromosomes 21), and three hybrids from the Down syndrome individual (containing each of the maternally derived chromosomes 21 and the single paternally derived chromosome 21).
Pulsed-field gel electrophoresis. For each family, agarose plugs of
high molecular weight DNA were obtained from lymphoblastoid
cell lines or hybrids, using standard methods. DNA samples were
then digested with either BamHI (Gibco-BRL) or BanI (New
England Biolabs), as previous PFGE studies of monochromosomal
168
21 hybrids (Trowell et al. 1993) and CEPH families (Marcais et al.
1991) indicated that these enzymes might be useful in revealing
chromosome 21-specific polymorphisms. Pulsed-field gels were
run in a CHEF-DR III electrophoresis system (Bio-Rad), the products blotted on Zeta-Probe membranes (Bio-Rad) and the membranes hybridized with the probe aRI680 (D21Z1), detecting centromeric alpha-satellite sequences of chromosomes 13 and 21.
Results
Somatic cell hybrids were generated from 23 trisomy
21 individuals and their mothers. On the basis of conventional DNA polymorphism studies, 16 of the trisomies had been scored as originating at maternal MI,
with no evidence for recombination between the nondisjoining chromosomes, and 7 cases had been scored
as arising at maternal MII (data not shown). Information on the PFGE analyses and the data on the studies
of recombination are summarized in Tables 1 and 2, respectively.
Analysis of the meiotic stage of origin of nondisjunction
and the frequency of pericentromeric recombination
Approach. Centromeric polymorphisms are necessary to
specify unequivocally the meiotic stage of origin of trisomy, but currently available probes to the centromeric
alpha-satellite sequences of chromosome 21 cross-hybridize to those of chromosome 13 (Jorgensen et al. 1987).
Thus, studies of nondisjunction of chromosome 21 have
had to rely on proximal 21q DNA markers to infer the
meiotic stage of origin.
As an alternative approach to this problem, we decided
to eliminate the ªcontaminatingº chromosomes 13 by generating somatic cell hybrids that retained chromosome 21
but not chromosome 13. This strategy requires establishing at least four monochromosomal 21 hybrids from each
Down syndrome family: two from the mother (the parent
of origin in each of our cases), representing her two different centromeres, and two from the trisomic individual,
representing the two maternally derived chromosomes.
Hybrid DNAs are then hybridized to a chromosome 21 alpha-satellite-detecting probe (aRI680) to reveal polymorphic fragments (alleles) associated with the individual
chromosome 21 centromeres. The presence of both maternal centromeric alleles in different hybrids of the trisomic
individual indicates MI nondisjunction; the presence of
only one of the two maternal centromeric alleles indicates
MII nondisjunction (see Fig. 1 for example).
In practice, there are at least two complications to
this approach. First, the alpha-satellite fragments detected in conventional Southern blotting studies are
complex, and insufficiently polymorphic to be useful
in meiotic stage studies (data not shown). Thus, we used
a PFGE-based approach, which resulted in a more easily interpretable pattern of fragments (see Fig. 1). Second, chromosome 13 is not always eliminated by the selection process. We addressed this problem by maintaining and analyzing several extra ªpartially informa-
Fig. 1. Pulsed-field gel electrophoresis (PFGE) analyses of monochromosomal 21 hybrids to determine the meiotic stage of origin
of trisomy 21. In this family (D399), analysis of pericentromeric microsatellite polymorphisms had indicated a maternal meiosis I (MI)
error. BanI-digested DNA samples from lymphoblastoid cell lines
of the proband (DS), father and mother, and from different monochromosomal 21 hybrids of the proband (DS 1±3) and mother
(Mo 1, 2) were size-fractionated using PFGE and probed with
aRI680 to identify alpha-satellite fragments of approximately
150±600 kb (Saccharomyces cerevisiae marker not shown). On
the basis of microsatellite marker analysis, DS1 was known to contain the paternal chromosome 21; on PFGE, it retained the 460 kg
fragment, which was also observed in paternal genomic DNA. From
microsatellite studies, DS2 and DS3 were known to contain different maternally derived chromosomes 21; on PFGE, the chromosome
21-derived fragment arrays of DS2 were identical to those of Mo1
and those of DS3 identical to those of Mo2 (the additional fragments observed in the proband and maternal genomic DNA and
in hybrid DS2 were derived from chromosome 13). Thus, each of
the two maternal centromeric alleles was transmitted to the Down
syndrome offspring, consistent with an MI nondisjunctional event
tiveº hybrids per family. For example, in some instances two hybrids were available that, on the basis
of microsatellite analysis, contained the same chromosome 13 but different maternally derived chromosomes
21. In such cases, alpha-satellite fragments common to
the two hybrids were assumed to come from chromosome 13, while fragments unique to each hybrid were
assumed to derive from the chromosome 21 of that cell
line.
Results. PFGE was used to study the inheritance of centromeric alpha-satellite fragments of chromosome 21 in
ten Down syndrome families, six scored as arising at
maternal MI and four at maternal MII on the basis of
conventional centromere-mapping studies using proximal 21q markers. Examples of the PFGE analysis are
provided in Fig. 1 and a summary of the data is given
in Table 1.
We were able to specify the meiotic stage of origin in
all ten families: six on the basis of both BamHI studies
and BanI studies, two on the basis of BamHI studies
alone and two on the basis of BanI studies alone. In
169
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
Yes
No
No
No
No
No
No
No
Yes
No
No
D88
D342
D399
D492
D727
D728
D733
D4000
D4001
D4013
D4019
Informative loci are indicated by dots, and the locations of previously undetected crossovers detected in two of the cases are indicated by Xs
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
X
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
IF
NAR
D21
S213
D21
S226
D21
S210
D21
S232
D21
S214
D21
S11
D21
S192
D21
S120
D21
S258
D21
S215
Locus
Crossover
detected?
Down
syndrome
family
ID no.
all ten Down syndrome families, the PFGE results were
in agreement with previous meiotic stage assignments
based on analysis of proximal 21q microsatellite markers. Thus, there was no evidence for recombination between the centromere and the proximal 21q markers in
these cases.
Table 2. Summary of somatic cell hybrid studies of recombination in 11 cases of trisomy 21
Fig. 2. Examples of meiotic configurations leading to trisomy, assuming at most two exchanges. Three types of nondisjunctional
meiotic products are possible: parental (in which both of the transmitted chromosomes are nonrecombinant), recombinant (in which
one chromosome is recombinant and the other nonrecombinant,
making it possible to detect the crossover(s) on conventional microsatellite mapping studies) and complementary recombinant (in
which complementary recombinant chromosomes are recovered,
leading to a crossover detectable by the somatic cell hybrid approach but not conventional microsatellite analysis)
l
l
l
l
l
l
l
l
l
l
l
l
l
X
MI
MI
MI
Uninformative
MI
Uninformative
MII
MII
MII
MII
l
l
l
l
MI
MI
MI
MI
MI
MI
Uninformative
MII
Uninformative
MII
l
MI
MI
MI
MI
MI
MI
MII
MII
MII
MII
D21
S171
D399
D492
D727
D4000
D4001
D4019
D693
D704
D708
D3112
PFKL
BanI
D21
S212
BamHI
HMG
14
Results of PFGE analysis using:
D21
S156
Results
of studies
of proximal
21q markers
D21
S167
Down
syndrome
family
ID no.
D21
S1575
Table 1. Comparison of results of studies of the meiotic stage of origin of nondisjunction using proximal 21q markers versus PFGE
analysis of alpha-satellite polymorphisms
170
Fig. 3. Example of the detection of a previously undetected exchange in case D4001, based on results at seven informative 21q microsatellite markers. At each locus, the alleles of genomic DNA
samples of the proband (DS), father and mother, and different
monochromosomal 21 hybrids of the proband (DS1±3) and mother
(Mo 1, 2) were compared. Hybrid DS1 contained the paternally de-
rived chromosome 21, while DS2 and DS3 were maternally derived.
In proximal 21q, the haplotype of DS2 was identical to that of Mo1,
and DS3 identical to that of Mo2; distally, this relationship changed,
with DS2 identical to Mo2 and D32 identical to Mo1. This ªswitchº
occurred between IFNAR and HMG14, localizing the site of the
crossover between the nondisjoining chromosomes
Analysis of recombination and the demonstration
of previously undetectable crossovers
if the Down syndrome individual inherits two maternally
derived chromosomes 21, but with haplotypes differing
from those observed in the mother, we conclude that crossing over occurred in the trisomy-generating meiosis.
The number of previously undetectable crossovers detected by this approach depends on the number and arrangement of chiasmata in the trisomy-generating meiosis. For example, if the nondisjoining chromosomes are
held together by a single chiasma, or by two chiasmata
involving all four chromatids, our approach allows us to
detect a single exchange. In the event of a two-strand
double event, the somatic cell hybrid approach should detect both exchanges (Fig. 2).
Approach. Each of the 16 MI trisomies in the present study
was selected as being maternally derived, and as having no
evidence of recombination between the nondisjoined chromosomes in standard mapping analyses (Sherman et al.
1994). Conceptually, trisomies scored as being ªnonrecombinantº can arise in one of three ways: (1) as a result of a
nondisjunctional event in which there is no crossing over between the chromosomes 21, (2) as a result of a nondisjunctional event in which crossing over occurs, but nonrecombinant chromatids are recovered or (3) as a result of a nondisjunctional event in which crossing over occurs, but complementary recombinant chromatids are recovered (Fig. 2).
It is not possible to distinguish between the first two alternatives, but the somatic cell hybrid approach allows us to
uncover the previously undetectable crossover(s) associated
with the third category. DNA samples from individual hybrids of the trisomic individual and the mother are genotyped to define the different chromosome 21 haplotypes
and the trisomic and maternal haplotypes are then compared with one another. If both maternal chromosome 21
haplotypes are transmitted intact to the Down syndrome individual, we conclude that either recombination did not occur between the nondisjoining chromosomes, or that the
products of recombination were not transmitted. In contrast,
Results. Somatic cell hybrids were established from the trisomic individual and/or mother in all 16 MI cases. However, in five instances we were unable to generate the different types of hybrids necessary to define the chromosome 21
haplotypes of both the trisomic individual and the mother.
In the 11 remaining MI cases, all requisite hybrids
were generated, allowing us to define unequivocally the
trisomic and maternal chromosome 21 haplotypes. The
results are summarized in Table 2, and an example is provided in Fig. 3. In 9 of the 11 cases, the two maternal
chromosome 21 haplotypes were identical to the two maternally derived haplotypes of the trisomic individual,
consistent with no recombination in the trisomy-generat-
171
ing meiosis. In the other two cases (D88 and D4001) a
single distally located exchange was identified.
To evaluate the statistical significance of this result,
we calculated the expected frequency of previously undetectable crossovers; i.e., the conditional probability that
the somatic cell hybrid approach would detect an exchange among ªnonrecombinantº MI cases, assuming
that at least one exchange had occurred in the trisomygenerating meiosis. For this calculation, we assumed (1)
that one or two exchanges had occurred between the nondisjoining chromosomes 21, (2) that the proportion of single:double exchange events was the same as that for normally disjoining chromosomes 21 and (3) that the normal
female genetic map of chromosome 21 is approximately
80 cM (A. Lynn, personal communication).
Using these assumptions, we first estimated that about
2/5 of the nondisjoining chromosomes 21 had a single exchange, and 3/5 two exchanges, as follows. If x=the number of single exchanges and (1Ÿx)=the number of double
exchanges, then x+2(1Ÿx)=80/50 or x=2/5. We then estimated the proportion of single and double exchange
among apparently nonrecombinant trisomies. For a single
exchange, 1/2 of the 4 possible meiotic outcomes will appear recombinant and 1/2 nonrecombinant by conventional
mapping approaches, while for a double exchange 1/4 of
all possible outcomes will appear nonrecombinant (Robinson et al. 1993). Thus, only (2/5)(1/2)+(3/5)(1/4)=3/10 of
the total population will be included in this nonrecombinant group, of which 4/7 should involve single exchanges
and 3/7 double exchanges. Finally, we calculated the likelihood of detecting a previously undetectable exchange for
each of these two situations. For nondisjunction involving
a single exchange, 1/2 of the apparently nonrecombinant
events involve transmission of both parental chromatids.
The other 1/2 involve transmission of the complementary
recombinant chromatids (Fig. 2) ± those which the somatic
cell hybrid approach should detect. Applying the same logic to nondisjunctional meiosis involving two exchanges,
we estimate being able to detect at least one exchange in
3/4 of all apparently nonrecombinant trisomies. Thus,we
expect to be able to detect a crossover in (1/2)(4/7)+
(3/4)(3/7), or 61% of apparently nonrecombinant trisomies.
This expectation is significantly higher than the 18% (2/11)
we observed in our series (X2=8.48; P<0.005).
Discussion
Technical limitations have hampered previous studies of
chromosome 21 nondisjunction in three important respects. First, the absence of a useful chromosome 21-specific centromeric polymorphism has made it impossible
to specify unequivocally the meiotic stage of origin. Second, the absence of a centromeric marker precludes genetic mapping in proximal 21q, therefore it has not been
possible to examine the role of pericentromeric recombination on nondisjunction. Third, standard centromeric
mapping techniques cannot detect all exchanges in trisomy-generating meioses in humans, thus complicating
efforts to distinguish between reduced versus absent recombination in the genesis of trisomy 21.
The results of the present study indicate that proximal
21q markers are, in fact, reliable indicators of the meiotic
stage of origin of nondisjunction. That is, in each of the six
MI and four MII cases we analyzed, the direct PFGE results were in agreement with indirect interferences based
on proximal 21q microsatellite markers. This suggests that
previous estimates of the proportion of MI, MII and mitotic errors in trisomy 21 approximate the real occurrence of
these events; i.e., about 65%, 23%, 3%, 5% and 3% for
maternal MI, maternal MII, paternal MI, paternal MII
and mitotic errors, respectively (Lamb et al. 1996). Further, our results indicate that, in future studies of chromosome 21 nondisjunction, analyses of proximal 21q markers
will be sufficient for making meiotic stage determinations.
Our results also suggest that pericentromeric recombination is infrequent in nondisjunctional meioses, since we
found no differences between the meiotic stage assignments
based on PFGE and those based on proximal 21q microsatellite markers. Previously, we (Lorber et al. 1992) and others (Petersen et al. 1992) reported a number of discrepancies between meiotic stage determinations based on proximal 21q DNA markers and those based on 21q chromosomal heteromorphisms. To reconcile these discrepancies,
Lorber et al. (1992) and Petersen et al. (1992) proposed that
recombination in extremely proximal regions of 21p or 21q
might be a factor in nondisjunctional meioses. The present
results suggest that ± if such exchanges occur ± they may be
restricted to 21p, since we found no evidence for recombination between centromeric alphoid sequences and proximal 21q microsatellite markers.
Finally, our results provide strong evidence for zeroexchange or ªachiasmateº chromosomes 21 in the genesis
of trisomy 21 of maternal MI origin. In previous studies
of maternal MI trisomy 21, we reported a highly significant reduction in recombination in the trisomy-generating
meioses (Sherman et al. 1994; Lamb et al. 1996). However, the basis for this effect has not been clear. In early
studies, our data were consistent with reduced and abnormally placed recombinational events in the vast majority
of maternal MI trisomies (Sherman et al. 1994). However, with the addition of more material and using an approach to estimate 0, 1 and multiple exchanges, we recently suggested that a large proportion (45%) of cases
were achiasmate (Lamb et al. 1997). The results of the
present study support this interpretation.
Further, the results of the present study allow us to derive a crude estimate of the proportion of achiasmate
cases, one that is remarkably similar to that reported by
Lamb et al. (1997). That is, of those trisomies that appear
to be nonrecombinant, our somatic cell hybrid assay
should detect approximately 1/2 of those in which an exchange actually occurred (i.e., those in which the complementary recombinant chromosomes were recovered) and
miss approximately 1/2 (i.e., those in which the parental
chromosomes were recovered). In the present study population of 11 apparently nonrecombinant cases, we identified a previously undetectable exchange in two cases; thus
we estimate that recombination actually occurred in 4, and
that the remaining 7 (64%) were achiasmate. Assuming
that about 65% of maternal MI trisomy 21 appear to be
nonrecombinant on conventional analysis (Sherman et al.
1994), we conclude that approximately 42% of all instances of maternal MI trisomy 21 involve bivalents that failed
to pair and/or recombine. This is virtually identical to the
172
estimate provided by Lamb et al. (1997), and indicates
that there are multiple, recombination-associated routes
to nondisjunction of chromosome 21: meioses in which
the chromosomes fail to recombine, leading to an MI nondisjunction (Lamb et al. 1997 and the present study); meioses in which exchanges occur, but are distally placed,
leading to MI nondisjunction (Sherman et al. 1994; Lamb
et al. 1997); and meioses in which too many, and proximally located, exchanges occur, leading to MII nondisjunction (Lamb et al. 1996, 1997).
These obervations are remarkably similar to those
from recent studies of other organisms. For example,
Koehler et al. (1996a) analyzed spontaneous meiotic nondisjunction of the X chromosome in female Drosophila,
identifying 103 exceptional progeny. Approximately
70% of these were attributable to failure of recombination
at MI; the majority of the remaining MI cases were associated with a single, distally placed exchange, while all
six MII cases were associated with extremely proximal
exchanges. Similarly, in systematic analyses of chromosomal segregation in yeast model chromosomes, Ross
et al. (1992, 1996a, b) found that exchange chromosomes
were more likely to segregate properly than were nonexchange chromosomes, but noted that exchange did not ensure segregation. Specifically, they found that distal exchanges were less able to ensure normal segregation than
were more proximally located events. Thus, in species
from yeast to humans, both achiasmate and ªunusualº
chiasmate configurations are associated with nondisjunction. The challenge now facing us is to determine the reasons why these, but not other, chiasmate configurations
confer an increased susceptibility to malsegregation.
Acknowledgements. We gratefully acknowledge Dr. David Patterson for donating the Ade-C cell line, and Elise Millie and Dorothy
Pettay for their technical expertise. This work was supported by
NIH grants HD21341 and HD32111.
References
Bascom-Slack C, Ross L, Dawson D (1997) Chiasmata, cross-overs,
and meiotic chromosome segregation. Adv Genet 35:253±283
Charlieu J-P, Marcais B, Laurent A-M, Roizes G (1993) New tools
for the study of chromosome segregation and aneuploidy at the
molecular level. In: Vig B (ed) Chromosome segregation and
aneuploidy. Springer, Berlin Heidelberg New York, pp 75±86
Fisher JM, Harvey JF, Morton NE, Jacobs PA (1995) Trisomy 18:
studies of the parent and cell division of origin and the effect
of aberrant recombination on nondisjunction. Am J Hum Genet
56:669±675
Hassold T, Jacobs PA (1984) Trisomy in man. Annu Rev Genet
18:69±97
Hassold T, Merrill M, Adkins K, Freeman S, Sherman S (1995) Recombination and maternal age-dependent nondisjunction: molecular studies of trisomy 16. Am J Hum Genet 57:867±874
Hassold T, Abruzzo M, Adkins K, Griffin D, Merrill M, Millie E,
Saker D, et al. (1996) Human aneuploidy: incidence, origin,
and etiology. Environ Mol Mutagen 28:167±175
Hawley RS (1988) Exchange and chromosome segregation in eucaryotes. In: Kucherlapati R (ed) Genetic recombination. American Society of Microbiology, Washington, DC, pp 497±527
Jabs EW, Warren AC, Taylor EW, Colyer CR, Meyers DA, Antonarakis SE (1991) Alphoid DNA polymorphisms for chromosome 21 can be distinguished from those of chromosome 13 using probes homologous to both. Genomics 9:141±146
Jorgensen AL, Bostock CJ, Bak AL (1987) Homologous subfamilies of human alphoid repetitive DNA on different nucleolus organizing chromosomes. Proc Natl Acad Sci USA 84:1075±
1079
Koehler KE, Boulton CL, Collins HE, French RL, Herman KC,
Lacefield SM, Madden LD, Schuetz CD, Hawley RS (1996a)
Spontaneous X chromosome MI and MI nondisjunction events
in Drosophila melanogaster oocytes have different recombinational histories. Nat Genet 14:406±413
Koehler KE, Hawley RS, Sherman S, Hassold T (1996b) Recombination and nondisjunction in humans and flies. Hum Mol Genet
5:1495±1504
Lamb NE, Freeman SB, Savage-Austin A, Pettay D, Taft L, Hersey
J, Gu Y, et al. (1996) Susceptible chiasmate configurations of
chromosome 21 predispose to non-disjunction in both maternal
meiosis I and meiosis II. Nat Genet 14:400±405
Lamb NE, Feingold E, Savage A, Avramopoulos D, Freeman S, Gu
Y, Hallberg A, et al. (1997) Characterization of susceptible chiasma configurations that increase the risk for maternal nondisjunction of chromosome 21. Hum Mol Genet 6:1391±1399
Lorber BJ, Grantham M, Peters J, Willard HF, Hassold TJ (1992)
Nondisjunction of chromosome 21: comparisons of cytogenetic
and molecular studies of the meiotic stage and parent of origin.
Am J Hum Genet 51:1265±1276
MacDonald M, Hassold T, Harvey J, Wang LH, Morton NE, Jacobs
P (1994) The origin of 47, XXY, and 47, XXX aneuploidy: heterogeneous mechanisms and role of aberrant recombination.
Hum Mol Genet 3:1365±1371
Marcais B, Bellis M, Gerard A, Pages M, Boublik Y, Roizes G
(1991) Structural organization and polymorphism of the alpha
satellite DNA sequences of chromosomes 13 and 21 as revealed
by pulse field gel electrophoresis. Hum Genet 86:311±316
Petersen MB, Frantzen M, Antonarakis SE, Warren AC, Van Broeckhoven C, Chakravarti A, Cox TK, et al. (1992) Comparative
study of microsatellite and cytogenetic markers for detecting
the origin of the nondisjoined chromosome 21 in Down syndrome. Am J Hum Genet 51:516±525
Robinson WP, Bernasconi F, Mutirangura A, Ledbetter DH, Langlois S, Malcolm S, Morris MA, et al. (1993) Nondisjunction of
chromosome 15: origin and recombination. Am J Hum Genet
53:740±751
Rockmill B, Roeder S (1994) The yeast med1 mutant undergoes
both meiotic homolog nondisjunction and precocious separation
of sister chromatids. Genetics 136:65±74
Ross L, Treco D, Nicolas A, Szostak J, Dawson D (1992) Meiotic
recombination on artificial chromosomes in yeast. Genetics
131:541±550
Ross L, Rankin S, Flatters M, Dawson D (1996a) The effects of homology, size and exchange on the meiotic segregation of model
chromosomes in Saccharomyces cerevisiae. Genetics 142:79±
89
Ross L, Maxfield R, Dawson D (1996b) Exchanges are not equally
able to enhance meiotic chromosome segregation in yeast. Proc
Natl Acad Sci USA 93:4979±4983
Shen JJ, Williams BJ, Zipursky A, Doyle J, Sherman SL, Jacobs
PA, Shugar AL, et al. (1995) Cytogenetic and molecular studies
of Down syndrome individuals with leukemia. Am J Hum Genet
56:915±925
Sherman SL, Petersen MB, Freeman SB, Hersey J, Pettay D, Taft L,
Frantzen M, et al. (1994) Non-disjunction of chromosome 21 in
maternal meiosis I: evidence for a maternal age-dependent
mechanism involving reduced recombination. Hum Mol Genet
3:1529±1535
Trowell HE, Nagy A, Vissel B, Choo KHA (1993) Long-range analyses of the centromeric regions of human chromosomes 13, 14
and 21: identification of a narrow domain containing two key
centromeric DNA elements. Hum Mol Genet 2:1639±1649
Van Hul W, Van Camp G, Stuyver L, Delabar JM, McInnis MG,
Warren AC, Antonarakis SE, et al. (1993) A contiguous physical map of the pericentromeric region of chromosome 21q between D21Z1 and D21S13E. Genomics 15:626±630