Molecular Analysis of Nondisjunction in Down Syndrome Patients With and Without Atrioventricular Septal Defects
Background Congenital heart disease is common in Down syndrome patients, with atrioventricular septal defects accounting for a majority of the abnormalities. The molecular mechanisms of meiotic nondisjunction resulting in Down syndrome were studied for associations with the presence of atrioventricular septal defects.
Methods and Results Twenty highly polymorphic chromosome 21 microsatellite markers were used to genotype two groups of patients (group 1: Down syndrome with atrioventricular septal defects, n=43; and group 2: Down syndrome without cardiac defects, n=51) to determine (1) the parental origin of the extra chromosome, (2) the stage of meiotic nondisjunction resulting in the trisomy, (3) the presence or absence of disomic homozygosity or heterozygosity, and (4) the degree of recombination in the nondisjoined chromosomes. The parental origin of the nondisjoined chromosome was maternal in 86.2% of the families, with no significant differences between groups. The most centromeric marker was nonreduced, indicating a meiosis I nondisjunction in 76.5% of maternally derived trisomies, and reduced, indicating a meiosis II nondisjunction in 76.9% of paternally derived trisomies, with no significant differences between groups. There were no significant differences in the proportion of reduced markers at any locus between groups. The distribution of the number of crossovers was significantly different between groups (χ2=14.12, P<.001), with less recombination observed in group 1.
Conclusions In Down syndrome patients, no association was found between the presence of an atrioventricular septal defect and the parent of origin, stage of meiotic nondisjunction, or disomic homozygosity or heterozygosity. A significant association was found between the presence of an atrioventricular septal defect and reduced frequency of recombination.
The association between Down syndrome and atrioventricular septal defects is well known. Of the 40% to 50% of all Down syndrome patients with congenital heart disease, approximately 30% to 60% have atrioventricular septal defects.1 2 3 4 These figures can be compared with the general population, in which the estimated incidence of congenital heart disease is 8 per 1000 births,5 6 with atrioventricular septal defects accounting for only 2% to 3% of the total number.7 The striking increase in incidence of atrioventricular septal defects in Down syndrome suggests that one or more genes responsible for heart development may be found on chromosome 21. Previous studies have identified a region on chromosome 21 from q22 to qter as the Down syndrome critical region.8 9 Several candidate genes within this region, such as COL6A110 and SOD,11 are being investigated for their possible roles in cardiac embryogenesis.
In addition to the high incidence of atrioventricular septal defects in Down syndrome, these defects are also known to occur in an autosomal dominant fashion with variable penetrance.12 13 14 Nonsyndromic parents with atrioventricular septal defects have a high risk (9.6%) of producing children with congenital heart disease, and the risk is known to be greater (14.3%) if the mother has the atrioventricular septal defect.3
The normal mechanisms of meiosis result in the production of a daughter germ cell that contains only one copy of each of the parental chromosomes (Fig 1a⇓). The majority of numerical chromosomal aberrations, such as Down syndrome, result from a nondisjunction error in meiosis. A daughter cell that has received a pair of a specific parental chromosome will result in a proband trisomic for that chromosome once conception has occurred. As illustrated in Figs 1b⇓ and 1c⇓, a daughter cell may receive one of two possible sets of chromosomes based on the stage of meiosis in which the nondisjunction occurred. Assuming no recombination between parental chromosomes, a meiosis I nondisjunction (Fig 1b⇓) will result in the proband receiving one copy of each of his nondisjoining parent’s chromosomes (disomic heterozygosity or nonreduced). A meiosis II nondisjunction (Fig 1c⇓) results in the proband receiving two identical copies of one of his parent’s chromosomes (disomic homozygosity or reduced).
Another fundamental feature of meiosis is the process of crossing over between parental chromosomes that results in the recombination of genetic material. Recombination allows for specific segments of the nondisjoined chromosomes to be disomic homozygous and other areas to be disomic heterozygous (Figs 1d⇑ and 1e⇑).
In addition to searching for the gene(s) responsible for the development of atrioventricular septal defects in Down syndrome, a second line of inquiry thus needs to address whether the mechanisms by which the trisomic state arises are themselves associated with the trisomic phenotype, including atrioventricular septal defects. The possibility of gene dosage effects resulting from disomic homozygosity or heterozygosity is one such intriguing possibility.
In this article we describe the degree of association between several meiotic mechanisms and the presence of atrioventricular septal defects in Down syndrome patients. The specific questions addressed were as follows. (1) Is there an association between the parent of origin of the nondisjoined chromosomes and the presence of atrioventricular septal defects in the proband? (2) Is there an association between the stage of parental meiotic nondisjunction and the presence of atrioventricular septal defects in the proband? (3) Is there an association between the presence of disomic homozygosity or heterozygosity at any studied locus and the presence of atrioventricular septal defects? (4) Is there any difference in the amount of recombination in the nondisjoined chromosomes in Down syndrome patients with atrioventricular septal defects compared to those without congenital heart disease?
Down syndrome patients with and without atrioventricular septal defects were identified from University of Iowa Hospitals and Clinics’ computer files, echocardiographic lists, catheterization files, surgical files, and the Iowa Birth Defects Registry and the Child Health Specialty Clinics Registry as well as similar registries from the Division of Pediatric Cardiology at the University of Nebraska College of Medicine. The ascertainment process, recruitment, enrollment, and subsequent study of probands and their parents were approved by the University of Iowa’s Internal Review Board, and informed consent was obtained from all families who participated. All probands were nonadopted and under the age of 40 years. There were 43 families in which the proband had an atrioventricular septal defect (group 1) and 51 families in which the proband had no congenital heart disease (group 2).
All probands were examined to diagnose or exclude atrioventricular septal defects. The diagnosis of atrioventricular septal defect was made in all group 1 probands by one or more of the following techniques: echocardiography, catheterization, surgical inspection, or postmortem examination. The diagnosis of atrioventricular septal defect required one of the following: (1) an ostium primum atrial septal defect, with or without atrioventricular valve abnormality; (2) a defect in the ventricular septum related to the posterior inflow portion of the right ventricle and both the inflow and outflow portions of the left ventricle, with or without atrioventricular valve abnormality; or (3) the complete form of atrioventricular canal defect, which consists of both an ostium primum atrial septal defect and an inflow ventricular septal defect as well as an atrioventricular valve common to both ventricles. All probands in group 2 underwent echocardiography to exclude congenital heart disease. Complete two-dimensional echocardiograms were done using a Hewlett-Packard sector scanner with standard views. All echocardiograms were obtained by a trained echocardiographer. All abnormalities were reviewed by two independent observers from the Division of Pediatric Cardiology at the University of Iowa.
Blood for DNA preparation was obtained by venipuncture from all probands and as many parents as possible. In group 1, complete sets of blood (proband and both parents) were available for 41 of the 43 families (95.3%) and incomplete sets (proband with one parent) in the remaining 2 families. In group 2, complete sets of blood were available for 50 of the 51 families (98.0%), and an incomplete set was available for the remaining family. DNA was prepared from whole blood specimens according to the technique of Grimberg et al.15
Markers were selected based on availability, degree of polymorphism, and quality. Care was taken to assure locations along 21q that spanned as much of the chromosome as possible to detect recombination. The primer pairs for D21S11, D21S126, D21S267, D21S266, and PFKL were dinucleotide CA-repeat markers supplied by Research Genetics; the remaining 15 markers included 2 trinucleotide repeat markers (ATA class) and 13 tetranucleotide repeat markers (11 GATA and 2 GGAA), which have been developed by the Cooperative Human Linkage Centers.16 The map order and location for 15 of the markers (D21S1432, D21S11, D21S1436, D21S1437, D21S1435, GATA26A04, GATA24H09, D21S1270, ATA27F01, ATA22G04, D21S167, D21S267, D21S266, PFKL, GATA70B08) are known and published16 17 ; these markers span 21q from q11 to q22.3. The remaining five markers (D21S1433, D21S1434, GATA45C03, GATA29D01, GATA71H10) were placed into order following genotyping by reducing the number of recombinations among all families. Each of these additional five markers could be placed successfully between two other markers with the use of this technique. The final map order for these markers can be found along the top of Fig 3⇓. The known recombination fractions between the mapped markers are D21S1432-0.07–D21S11-0.04–D21S1436-0.04–D21S1437-0.12–D21S1435-0.03–GATA24H09-0.02–D21S1270-0.10–ATA27F01-0.01–ATA22G04-0.01–D21S267-0.10–D21S266-0.10–GATA70B08 (unpublished data, Cooperative Human Linkage Centers, 1995).
Amplification of each marker was performed using 40 ng of genomic DNA in an 8.4 μL polymerase chain reaction (PCR) containing 1.25 μL PCR buffer (100 mmol/L Tris-HCl, pH 8.8, 500 mmol/L KCl, 15 mmol/L MgCl2, 0.01% wt/vol gelatin), 200 μmol/L each of dCTP, dGTP, dTTP, 7.5 μmol/L dATP, 2.5 pmol of each primer, 0.20 μL 35S-dATP (Amersham, >1000 Ci/mmol), and 0.25 U Taq DNA polymerase. Some reactions were performed with the use of 200 μmol/L of unlabeled dATP instead of the 35S-dATP. PCR conditions were identical for all markers: an initial 94° for 5 minutes, then 35 cycles of denaturing at 94° for 30 seconds, annealing at 55° for 30 seconds, and extension at 72° for 30 seconds followed by one final extension period of 72° for 5 minutes. The PCR products then were mixed with 5 μL of loading dye (95% formamide, 10 mmol/L NaOH, 0.5% bromophenol blue, and 0.05% xylene cyanol), denatured at 95° for 5 minutes and electrophoresed on a denaturing 6% polyacrylamide sequencing gel (7.7 mol/L urea). The gels containing radiolabeled PCR products were transferred to Whatman 3-mm paper and dried. Autoradiograms were made by exposing Kodak X-Omat AR film to the dried gels for 24 to 48 hours. Gels containing nonlabeled PCR products were silver stained using the method of Bassem et al,18 and genotypes were read directly from the stained gels.
Analysis of Genotyping
The method used to analyze marker data for each family has been described by Chakravarti and Slaugenhaupt.19 Each family unit was scored as either reduced, nonreduced, or uninformative (Fig 2⇓). The parent of origin of the nondisjoined chromosomes was assigned. This assignment was possible from a single marker when the proband was nonreduced, with two alleles found only in one of his parents who was heterozygous at that locus. For an individual family, a crossover event leading to recombination between the nondisjoined chromosomes was determined when a nonreduced and a reduced marker were both present in the family. Marker data were placed in graphic format (Figs 3⇓ and 4⇓) to assist with this determination.
Data were tabulated for both groups. Contingency table data were analyzed with the use of χ2 analysis with significance placed at P<.05. For tables with small expected frequencies, Fisher’s exact test was used.
Sex Distribution of the Probands
There was no significant proband sex distribution difference between the two study groups (χ2=0.39, NS). Twenty (46.5%) of the probands in group 1 and 27 (52.9%) of the probands in group 2 were male.
Sex Distribution of the Parent of Origin of Nondisjunction
Assignment of the parent of origin of the nondisjoined chromosomes resulting in trisomy 21 was unequivocally made in all 94 families. There were no discrepancies noted in this assignment, that is, no two markers from the same family indicated a different parent as the parent of origin. The nondisjoined chromosomes originated from the mother in 37 (86.0%) group 1 families and 44 (86.3%) group 2 families (χ2=0.001, NS). The percentage of maternally derived trisomies for all the families combined was 86.2%.
Analysis of the Most Centromeric Marker
Analysis of the most centromeric marker can be used as an indicator of the meiotic stage of nondisjunction when it is assumed that no recombination occurs between the centromere and the first informative marker for each family. The most centromeric marker used in this study, D21S1432, has been mapped to the D21S120 locus (unpublished data, CHLC, 1994), which lies in the region known to be 5 to 10 cM from alphoid repeat units.20 Therefore, the likelihood of a significant amount of recombination between the centromere and the first marker, while not zero, may be assumed to be small.
A meiosis I nondisjunction was inferred if the most centromeric marker for a family was nonreduced; likewise, a meiosis II nondisjunction was inferred if the most centromeric marker was reduced. The proportion of the centromeric markers that were nonreduced and reduced for each group was analyzed based on the parental origin of the nondisjunction as well as for maternal and paternal nondisjunctions combined (Table 1⇓). There were no significant differences in the proportion of nonreduced centromeric markers between groups 1 and 2 when the meiotic stage of nondisjunction was analyzed based on a maternal origin (χ2=2.04, NS), a paternal origin (χ2=0.66, NS), or combined origin (χ2=2.14, NS).
The status of the most centromeric marker based solely on parent of origin was analyzed for all the probands combined. A maternal origin for the extra chromosome was due to a meiosis I nondisjunction in 62 probands (76.5%) and due to a meiosis II nondisjunction in 19 probands (23.5%); a paternal origin for the extra chromosome was due to a meiosis I nondisjunction in 3 probands (23.1%) and a meiosis II nondisjunction in 10 probands (76.9%) (χ2=15.01, P<.001).
Analysis of the Proportion of Reduced Markers at Each Locus
The proportion of reduced markers at each locus was compared between the two study groups (Table 2⇓). There were no statistically significant differences between the two groups at any of the loci.
Analysis of Detectable Crossovers
A crossover event between two markers is evident if one marker is reduced and the other nonreduced. However, not all crossovers will be detectable. As illustrated in Fig 1d⇑ and discussed by Howard et al,21 a meiosis I nondisjunction with one prior crossover event will produce four possible gametes: two of these (Figs 1d[i] and 1d[ii]) will have crossovers detectable by the methods used in this study, one (Fig 1d⇑[iv]) will have a crossover event that will not be detectable (all markers appear nonreduced), and one (Fig 1d⇑[iii]) will have no crossover events and all markers will appear nonreduced. Therefore, the absolute number of crossovers for these groups is indeterminable from this study. However, because the proportion of meiosis I and meiosis II nondisjunctions does not significantly differ between the two groups, one may assume that the detection factor is the same for both groups and that the detectable crossover differences between groups may still be compared. The number of detected crossovers for each family was thus determined by analysis of Figs 3⇑ and 4⇑, and group data based on meiotic stage of nondisjunction are presented in Table 3⇓.
In group 1, of the 33 probands with a nonreduced centromeric marker (meiosis I nondisjunction), 25 probands had no crossovers detected and 8 probands had one crossover detected. This proportion was not significantly different (χ2=0.13, P=1.0) from those group 1 probands with a reduced centromeric marker (meiosis II nondisjunction) in which 7 of the 10 probands had no crossovers detected and the remaining 3 had one crossover detected. In group 2, of the 32 probands with a nonreduced centromeric marker, 18 probands had no crossovers, 11 probands had one crossover, 2 probands had two crossovers, and 1 proband had 3 crossovers detected. This distribution of crossovers was significantly different (χ2=15.5, P<.0005) from those with a reduced centromeric marker in which of the 19 probands, 1 had no crossovers, 17 had one crossover, and 1 had two crossovers detected. There was no significant difference in the distribution of detectable crossovers between groups for those probands with a nonreduced centromeric marker (χ2=4.6, P>.16). A significant difference between groups was present for those probands with a reduced centromeric marker (χ2=13.8, P<.001). In addition, a significant difference in the distribution of crossovers was present between the total number of group 1 and group 2 probands (χ2=14.15, P<.001).
Examination of Table 3⇑ reveals that there were 32 probands (74.4%) in group 1 with no detectable crossovers, whereas only 19 (37.3%) of the group 2 probands had no detectable crossovers. All the probands with evidence of recombination in group 1 had single crossovers detected. The 32 probands in group 2 with evidence of recombination consisted of 28 probands (87.5%) with one crossover, 3 probands (9.4%) with two crossovers, and 1 proband (3.1%) with three crossovers (P<.001 for group 1 versus group 2 distributions).
The proportion of informative meioses at each locus between the two groups was compared to rule out a possible difference in number, which might account for decreased detection of crossovers in group 1. These data are presented in Table 4⇓. Only two of the markers (D21S1437, GATA70B08) had a significantly different proportion of informative meioses between the groups. D21S1437 had a higher proportion of informative markers in group 1 (χ2=9.86, P<.01), and GATA70B08 had a higher proportion of informative markers in group 2 (χ2=7.07, P<.01). The remaining 18 markers had no significant difference in the proportion of informative meioses between the two groups.
The mechanisms of meiotic nondisjunction provide several interesting possibilities for investigation into the development of specific trisomic phenotypes. These investigations have been made possible in recent years by the development and utilization of microsatellite DNA markers. Such markers provide a relatively simple, reproducible, and accurate assay for investigating allelic transmission from parents to offspring. In the trisomic proband it is possible to determine from even one informative marker the parent of origin of the nondisjunction and whether the alleles are reduced or nonreduced. Recombination events can be detected based on differences in reduced or nonreduced status (Fig 2⇑) at different loci along the chromosome.
The atrioventricular septal defect occurs with such an increased frequency in Down syndrome that it is likely that a gene or genes necessary for normal cardiac embryogenesis must lie on chromosome 21. However, recent linkage analysis studies13 14 have excluded chromosome 21 as the locus for the gene responsible for atrioventricular septal defects in families where the defect was inherited in an autosomal dominant fashion with incomplete penetrance. Thus, it appears that there may be several atrioventricular septal defect susceptibility genes.
Our study was designed to investigate possible differences in the origin and mechanism of meiotic nondisjunction between two large groups of Down syndrome patients that might be associated with the presence of atrioventricular septal defects in one of the groups. Several pertinent findings were evident from this study. No association was found between the parent of origin of the nondisjoined chromosomes and the development of atrioventricular septal defects. In our study, the parent of origin for both groups was most likely to be maternal (86.2% of total cases). This finding was not significantly different by χ2 analysis (χ2=8.85, NS) from several previous studies using DNA markers to study parental origin of Down syndrome.22 23 24 25
The second pertinent finding is that no specific locus had a significant difference in the proportion of reduced markers between the two groups. Specific attention was given to a number of markers between q22.2 and qter, the Down syndrome “critical region.”8 9 We found no evidence to support the hypothesis that disomic homozygosity or heterozygosity in this region, or at any other locus along 21q, was associated with the development of atrioventricular septal defects in Down syndrome patients. It is unlikely that our study missed a locus where such a difference might be found. Based on data presented in Figs 3⇑ and 4⇑ and in Table 4⇑, it is evident that informative markers were spread evenly across the chromosome from D21S1432 (q11) to GATA70B08 (q22.3). In addition, seven markers from the Down syndrome critical region were used to ensure good coverage of that region. Therefore, unless the locus in question falls outside this range, it is unlikely that an association exists between disomic homozygosity or heterozygosity and the development of atrioventricular septal defects in Down syndrome patients.
A third pertinent finding is the lack of association between the stage of meiotic nondisjunction and the development of atrioventricular septal defects. This was true for both maternally and paternally derived trisomies and leads one to conclude that the stage of meiotic nondisjunction is not associated with the development of atrioventricular septal defects in Down syndrome patients.
Our data indicate that a meiosis I nondisjunction was responsible for 76.5% of the total maternal cases of nondisjunction. These data are consistent with findings made by Juberg and Mowrey,25 Sherman et al,24 and Howard et al,21 who found meiosis I errors responsible for 77.1%, 76.6%, and 83.3% of maternal cases, respectively. We found that a majority of paternal cases resulted from a meiosis II nondisjunction. The difference between the majority of maternal cases being due to a meiosis I nondisjunction and the paternal cases being due to a meiosis II nondisjunction was significant (P<.001). Our ability to detect this previously unreported difference may be due to the relatively large number of paternal cases available for analysis.
The frequency of recombination between nondisjoined chromosomes has been a topic of much interest and investigation. Although several studies have noted a decreased recombination frequency in nondisjoined chromosomes,22 24 there are also data to suggest that recombination is not less frequent.21 26 The study by Howard et al21 evaluated the amount of recombination in a cohort of 27 Down syndrome children with several forms of congenital heart disease and found no decrease from that expected.
In our study, there was a marked decrease in the number of detected crossovers for the group with atrioventricular septal defects compared with the group without heart defects. This significant difference in the distribution of crossovers was observed in probands with reduced centromeric markers (meiosis II nondisjunction) and for the group totals. This decreased recombination was not significantly apparent between the groups for probands with nonreduced centromeric markers. As mentioned previously, this lack of significance may be secondary to the inability to detect all crossovers in nonreduced, nondisjoined chromosomes. This inability to detect crossovers in the group 2 patients with nonreduced centromeric markers may also explain why a significant difference was observed between the crossover distributions of the reduced centromeric marker patients and the nonreduced centromeric marker patients of group 2. Finally, only single crossovers were noted in group 1 families. Multiple crossovers were noted in four of the group 2 families.
There are several possible reasons why our results differ from those reported by Howard et al.21 The latter study analyzed Down syndrome patients with a variety of congenital heart defects; only 56% of those probands had atrioventricular septal defects. Since it is known that different congenital heart defects result from different embryological mechanisms, we chose to analyze a specific subgroup of Down syndrome patients who all had atrioventricular septal defects. In addition, our use of a control group of Down syndrome patients without congenital heart disease allowed us to look for differences in the crossover frequency distributions between the groups.
Several studies have investigated these meiotic mechanisms in Down syndrome individuals who have developed myeloproliferative disorders, such as leukemia. Abe et al27 used chromosome banding to identify disomic homozygosity in 9 of 9 patients with Down syndrome and transient myeloproliferative disorder and hypothesized gene duplication as a potential mechanism in the development of this myeloproliferative disorder. A study by Lorber et al28 also analyzed meiotic mechanisms in 18 Down syndrome patients with myeloproliferative disease. This latter study, which was done with the use of DNA markers, failed to confirm disomic homozygosity as a sole mechanism in the development of myeloproliferative disease. It is interesting to note in the Lorber et al study28 that although the number of informative meioses for each family was small, no recombination in the nondisjoined chromosomes was detected.
Our data support the findings from these studies of myeloproliferative disorders in Down syndrome that there may be phenotypic subsets of patients with Down syndrome based on the meiotic mechanism of nondisjunction. Warren et al22 proposed that nondisjunction from asynapsis (the homologous chromosomes never pair in meiosis I) results from lack of recombination, whereas desynapsis (the homologous chromosomes unpair before disjunction) would not be expected to exhibit reduced recombination. Our data indicate that the presence of atrioventricular septal defects is associated with asynapsis in a majority of cases.
The biological mechanism of this association is unknown. We have no data to suggest that the decreased recombination is directly responsible for the development of atrioventricular septal defects. In addition, there does not appear to be any biological mechanism that would support a cause-and-effect relation. Rather, it is our hypothesis that there is a more basic inherent chromosomal abnormality or pleiotropic gene that produces both a meiotic error in the parent and a development error in the conceptus. Thus, in certain Down syndrome individuals with an atrioventricular septal defect, we speculate that the specific abnormality responsible for the meiotic events of decreased crossing over, decreased recombination, and ultimately nondisjunction in the individual’s parent may be the same abnormality responsible for the defective fetal gene expression producing the atrioventricular defect. This relation will need to be reproduced and investigated further.
This work was supported by National Institutes of Health (NIH) SCOR in Congenital Heart Disease P50-HL-42266 (R.M.L., T.L.B., V.C.S.); NIH grant P50-HG00835 (J.C.M., V.C.S.); and the NIH Research Training Program in Pediatric Cardiology HL-07413 (M.M.Z.). We wish to thank the families who made this research possible by their participation. We would like to acknowledge the excellent technical assistance of Sara Sunden, Anne Kwitek-Black, Ann McClain, Tanya Rokhlina, Jinx Tracy, and Kail Denison. We also wish to thank Dr John Kugler for his assistance in enlisting the families under care of the staff at the University of Nebraska.
- Received April 4, 1995.
- Revision received June 19, 1995.
- Accepted July 3, 1995.
- Copyright © 1995 by American Heart Association
Emanuel R, Somerville J, Inns A, Withers R. Evidence of congenital heart disease in the offspring of parents with atrioventricular defects. Br Heart J. 1983;49:144-147.
Mitchell SC, Korones MD, Berendes HW. Congenital heart disease in 56,109 births: incidence and natural history. Circulation. 1971;63:323-331.
Campbell M. Incidence of cardiac malformations at birth and later, and neonatal mortality. Br Heart J. 1973;35:189-200.
Santoro G, Ambrosio G, Formigari R, Marcelletti C, Chiariello M, Marino B. Low levels of myocardial superoxide dismutase in patients with atrioventricular canal. J Am Coll Cardiol. 1994;23:306A. Abstract.
O’Nuallain S, Hall JG, Stamm SJ. Autosomal dominant inheritance of endocardial cushion defects. Birth Defects. 1977;13(3A):143-147.
Grimberg J, Nawoschik S, Belluscio L, McKee R, Turck A, Eisenberg A. A simple and efficient non-organic procedure for the isolation of genomic DNA from blood. Nucleic Acids Res. 1989;17:83-90.
Murray JC, Buetow KH, Weber JL, Ludwigsen S, Scherpbier-Heddema T, Manion F, Quillen J, Sheffield VC, Sunden S, Duyk GM, Weissenbach J, Gyapay G, Dib C, Morrissette J, Lathrop GM, Vignal A, White R, Matsunami N, Gerken S, Melis R, Albertsen H, Plaetke R, Odelberg S, Ward D, Dausset J, Cohen D, Cann H. A comprehensive human linkage map with centimorgan density. Science. 1994;265:2049-2054.
Warren AC, Chakravarti A, Wong C, Slaugenhaupt SA, Halloran SL, Watkins PC, Metaxotou C, Antonarakis SE. Evidence for reduced recombination on the nondisjoined chromosomes 21 in Down syndrome. Science. 1987;237:652-654.
Sherman SL, Takaesu N, Freeman SB, Grantham M, Phillips C, Blackston RD, Jacobs PA, Cockwell AE, Freeman V, Uchida I, Mikkelsen M, Kurnit DM, Buraczynska M, Keats BJB, Hassold TJ. Trisomy 21: association between reduced recombination and nondisjunction. Am J Hum Genet. 1991;49:608-620.
Hultén M, Lawrie NM, Laurie DA. Chiasma-based genetic maps of chromosome 21. Am J Med Genet. 1990;(suppl 7):148-154.