CopyTight 0 1995 by the Genetics Society of America

Nondisjunction Rates and Abnormal Embryonic Development in a Mouse Cross Between Heterozygotes Carrying a (7, 18) Robertsonian Translocation Chromosome

Rebecca J. Oakey," Paul G. Matteson,+ SamuelLitwin," Shirley M. Tilghmant and Robert L. Nussbaumf

*Institute for Cancer Research, Fox Chase Cancer Center, Philadelphia, Pennsylvania191 11, tDepartment of Molecular Biology, Lewis Thomas Laboratory, Princeton University, Princeton, New Jersey 08544, and :Laboratory of Genetic Disease Research, National Center for Human Gennome Research, National Institutes of Health, Bethesda, Maryland 20892 Manuscript received April 6, 1995 Accepted for publication July 18, 1995

ABSTRACT Mice bearing Robertsonian translocation chromosomes frequently produce aneuploid gametes. They are therefore excellent tools for studying nondisjunction in mammals. Genotypic analysis of embryos from a mouse cross between two different strainsof mice carrying a ( 7, 18) Robertsonian chromosome enabled us to measure the rate of nondisjunction for chromosomes 7 and 18. Embryos (429) were harvested from 76 litters of mice and the parental origin of each chromosome 7 and 18 determined. Genotyping these embryos has allowedus to conclude the following:(1) there were 96 embryos in which at least one nondisjunction event had taken place; (2) the rate of maternal nondisjunction was greater than paternal nondiqunction for the chromosomes sampled mice;in these (3) a bias against chromosome 7 and 18 nullisomic gametes was observed, reflectedin a smaller than expected number of uniparental disomic embryos; (4) nondisjunction events did not seem to occur at random throughout the 76 mouse litters, but were clusteredinto fewer than would be expectedby chance; and(5) a deficiency of paternal chromosome 18 uniparental disomic embryos was observed along with a higher than normal rate of developmental retardation at 8.5 days post coitum, raising the possibility that this chromosome has at least one imprinted gene.

UTOSOMAL. aneuploidy in mammals adversely af- some without apparent loss of genes, and displaying no A fects developmental processes. It has been esti- phenotypic abnormalities. Robertsonian chromosomes mated that 0.3% of human newborns are aneuploid segregate normally when present in ahomozygous state (HASSOLD1985). In humans, >70% of thechromo- in mice, butexperience a higher than normal fre- somal errors associated with early spontaneous abortion quency of nondisjunction when heterozygous. Comple- and almost half of those detected among newborns in- mentation of unbalanced gametes produced by inter- volve aneuploidy (BOND and CHANDLEY1983). The crossing mice carrying Robertsonian chromosomes has most common cause of genetic mental retardation in been used extensively to study the frequencyand effects humans is due to trisomy of chromosome 21 (GEAR- of nondisjunction events (SNELL 1946; SEARLE et al. HART et al. 1987) with 2-3% of trisomy 21 casesre- 1971; LYONet al. 1976; SEARLEand BEECHEY1978,1982) sulting from Robertsonian translocation chromosomes where embryo lethality or phenotypic markers were (GARDNERand SUTHERLAND1989). Themost likely ex- used to analyze the products of nondisjunction. planation for the profound phenotypiceffects resulting In this study, highly polymorphic molecular markers from aneuploidy of whole or parts of a chromosome is were used to distinguish five strains of inbred mice from an imbalance in gene dosage. one another (Figure 1).Mice carrying a ( 7, 18) Robert- The mouse offers several advantages for studying sonian chromosome were used to determine the out- chromosomalaberrations such as aneuploidy. Many comes of meiosis under conditions of elevated rates of mouse strains exist that carry Robertsonian transloca- nondisjunction. Two genetically distinguishable strains tion chromosomes thathave been maintained in inbred of mice, each with two copies of a (7, 18) Robertsonian strains allowing studies to be performed in a defined chromosome were crossed to a second strain with no geneticbackground. Robertsonian chromosomes in Robertsonian chromosome (Figure 2, top). Mice in the mice are the result of a centric fusion between two F1 generation were phenotypically normal, but carried one copy each of the ( 7,18) Robertsonian chromosome chromosomes, producing asingle metacentric chromo- and one normal chromosome 7 and 18. These F1 mice were intercrossed (for example, Figure 2, middle) and Cmesponding author; Robert L. Nussbaum, LGDR/NCHGR/NIH, Building 49 Room 4A72, 49 Convent Dr. MSC-1470, Bethesda, MD chromosomes 7 and 18 typed in the resulting F2 prog- 208924470. E-mail: [email protected] eny, harvested before birth. Several different combina-

Genetics 141: 667-674 (October, 1995) 668 R. J. Oakey et al.

A scheme of the (7,1@ Robertsonian mouse intercross Locus Size in bp BRCD3 D7MIT25 110 88 144108 96 D7MIT222 147155 178 123 123 D7MIT52 164170 146 160 162 D18MIT14 108112 130 103 107 D18MIT110 149 110125 153 117 D18MIT87 144130 176 152 140 RbRb(2.r8)2Lw7*18))9Lub D8 MIT4 175 I8O DbDW D2 MIT32102 107 98110 102 BbC578LIBJE!-Rtf7,18)9Lub 3 b C3HItie.J PATERNAL GAMETES

I I I I I I Balanced I Trisomy I Mono I Trisomy I Mono I I18I1817 I7 7 18 2 8 I I I

FIGURE1.-Locus, fragment size and chromosomal loca- 18 tion of the polymorphic markers usedto genotype F2 embryos. I I I I I I I I Null 18 I Trisomy 7 I Mono 7 I I I 1 Mono 18 I and 18 tions of parental strains were crossed to produce the I I I I I Balanced I I I Tei 7 I uniparental I I I 17 embryos (Table 1, first column). Because all of the pa- 18 rental strains could be distinguished by molecular methods, the parentalorigin and numbers of copies of J R Is Rb(z.18)2LutV7, 18)W chromosome 7 and 18 were determined for each F2 Gametes and resultingprogeny from D is OBAQJ Nondisjunctionand correct segregation B )a C57BLlglE&RM7,18)9Llh offspring. The phenotype (normal or retarded develop- Chrommes 7 and 78 C 1s Mus ulblBnBv(l of 3 Ls WWHeJ ment at the day of harvest) of each embryo was then correlated withits chromosome 7 and 18 genotype FIGURE2.-Top shows two of the six parental genotype (normal, trisomic or uniparental disomic). combinations used to generate theF, mice used in this study. The middleshows one of the 10 combinationsof F1genotypes used to generate the F4 embryos analyzed in this study. The MATERZALSAND METHODS bottom shows the theoretical outcomes that result from non- Mice: All mouse strains were obtained from The Jackson disjunction events in the F, gametes. Laboratory. Rb(2, 8)2Lub(7, 18)9Lub and C57BL/6JEi-Rb(7, 18)9Lub strains were obtainedfrom the Robertsonian Re- when the embryo appeared to be the correct size and at the source at The Jackson Laboratory. Rb(2, 8)2Lub(7, 18)9Lub appropriate stage of development for its time of harvest. An has been maintained onits own inbred background, a combi- abnormal phenotype was scored when the embryo either ap- nation of -50% wildderived Mus m. domesticusand 50%labo- peared smaller that expected or had a visible developmental ratory mousestrain background (DAVISSONand &SON problem, as determined by comparison with normal embryos 1993). The second Robertsonian mouse was derived by re- detailed in KAUFMAN'S Atlas of Mouse Development (KAUF- peated backcrossing to a C57BL/6J-Ei background. Rb(2, MAN 1992). After staging, a tissue sample was removed from 8)2Lub(7, 18)9Lub was crossed to DBA/2J and C57BL/6JEi- the tip of the embryo tail and frozen for DNA preparations. Rb(7, 18)9Lubwas crossed to either M. castaneus or C3H/HeJ DNA preparation: The embryo tail samples were vortexed to generate mice carrying one copy of the ( 7, 18) Robertson- in 100 p1 of a solution containing 100 mM KCl, 20 mM Tris ian chromosome (Figure 2, middle). Male F1 mice from the HCI (pH 8.3), 5 mM MgC12, 1.4 p~ sodium dodecyl sulfate, C57BL/6JEi-Rb(7, 18)9Lub X M. castaneus cross were not fer- 40 mM dithiothreitol, 2 mg/ml gelatin and 0.1 mg/ml protein- tile but male F1 mice from the reciprocal cross were fertile ase K. The samples were incubated at 37" for 1 hr, placed at [therefore C3H/HeJ mice were introduced into the cross to 95" for 5 min and stored at -20". generate F, males with a maternally inherited C57BL/6JEi- Primers and PCR Simple sequence repeat (SSR) polymor- Rb(7, 18) chromosome]. Allof the combinations of inter- phism primers (Map Pairs') were purchased from Research crosses used are detailed in Table 1 (first column). Females Genetics, Huntsville, AL. The names of the Map Pairs' used were used as generated and notespecially selected, but some and the polymorphic bands for each mouse strain along with females were tested more often, because, for example, DBA/ chromosomal locations are detailed in Figure 1. Genotyping 25 females had more litters than Castaneus females. was performed on 1pl samples of DNA using PCR with radio- F, mice with a single copy of the (7, 18) Robertsonian actively labeled primers (DIETRICHet al. 1992). The PCR prod- chromosome were intercrossed to produce 429 F2 progeny ucts were visualized on polyacrylamide gels. DNA was ampli- that were harvested from 8.5 (349 embryos), 9.5 (23embryos) fied ina 10 p1 reaction using 1 unit of AmpliTaq DNA and 10.5 (57 embryos) days post coitum (dpc) in timed mat- polymerase (Perkin Elmer Cetus) with manufacture's buffer ings. Immediately after removal from the uterus, theembryos in the presence of 0.4 mM dNTPs and both primers at a were dissected free from extra embryonic tissue and yolk sac concentration of 100 mM unlabeled, with one labeled 20 mM and placed in a drop of PBS. Each embryo was examined primer. Primers were labeled with gamma .?"P ATP using T4 under a dissection microscope, and the phenotype scored as polynucleotide kinase (NEB) (TABOR1994). Amplification either normal or retarded. A normal phenotype was scored conditions were: 3 min at 94" followed by 30 cycles of 94" for 7, 18 Translocation Nondisjunction 669

TABLE 1 Parental genotypes of all embryo categories investigated

Total No Double P arents litters nondisjunctionParentslitters 7 Tri 18 Tri 7 UPD 18 UPD Tri

BXC X DXR 10 2 3 8 1 1 4 BXC X RXD 7 2 3 3 1 1 2 CXB X DXR 0 0 0 0 0 0 0 CXB X RXD 3 1 0 1 0 0 0 BX3 X DXR 3 1 0 2 0 0 0 BX3 X RXD 8 2 3 1 0 2 1 3XB X DXR 0 0 0 0 0 0 0 3XB X RXD 1 1 0 0 0 0 0 DXR X CXB 15 5 10 4 2 6 0 RXD X CXB 5 1 1 3 3 0 0 DXR X BX3 15 6 5 6 2 4 0 9 4 RXD X BX3 9 6 6 2 0 0 Total 76 22 31 34 11 14 7 The first column lists all of the combinations of parental strains used in the crosses described in the paper. The second column lists the number of litters from each of these crosses. The third column lists the parental origin of litters with no nondisjunction events. The subsequentcolumns list the parental origins of all of the aneuploid categories.

15 sec, 55" for 30 sec, then 72" for 30 sec, followed by a final vidual probabilities of abnormality, p( i). Then Generalized 72" for 7 min. PCR products were diluted 1:l with 100% Likelihood Ratio (chi-squared) = -2 log(I,,/L,). Where I!,, formamide loading buffer containing bromophenolblue and was datalikelihood under the null hypothesis and L, was xylene cyanol, heated at 80" for 2 min, 6 pl was loaded onto data likelihood under thealternative. Each likelihood was the an 8% denaturing polyacrylamide gel until the bromophenol product of binomial distributions: b[a( i),n(i), p(i)], where blue dye had reached the bottom. Gels were dried, exposed a(i) was the number of abnormal embryos in litter i, n(i) to Kodak X-ray film and the genotypes scored. Each embryo the number of embryos and p(i) the maximum likelihood was typed for chromosomes 7 and 18 with one marker poly- estimate of the true probability of abnormality under the ap- morphic among thefive strains used in this study. To confirm propriate hypothesis. the genotyping for embryos with trisomies or UPD for either Maximum Likelihood Estimates (MU)of the rate of nmmal versus chromosome, theembryos were retyped with additional mark- disomic and nullisomic gametes: A generalized likelihood ratio ers. A subset of embryos were also typed for chromosomes 2 statistic was used to test two hypotheses. In the first, a null and 8. hypothesis, that the probability of a normal gamete,p(n), the Statistical analyses: A number of different statistical tests probability of a disomic gamete, p( d), and the probability of were employed for analysis of the data. a nullisomic gamete, p(o), are all equalto was tested Chi-squared analyses to determine the distribution of nondisjunc- against an alternative hypothesis that the probability of nor- tionevents between mouse litterx The distribution of embryos mal, disomic, and nullisomic gametes can assume any value with nondisjunction events was examined among the76 litters and sum to 1. In the second hypothesis, p(d) and p(o) were of mice. The expected number of abnormal embryos per litter assumed equal to each other but not equal to p(n),us. the was too small for the chi-squared distribution to be a good same alternative hypothesis described above. The assumptions approximation to that of our chi-squared statistic. The litter made in generating the likelihood ratio statistics were that size varied from 3 to 10 pups, and direct use of the Poisson there would be no monosomic embryos at or after 8.5 dpc distribution of numbers of abnormalities does not take this and trisomies, UPDs and normal embryos were equally likely into account. Therefore two statistical methods were applied to survive to 8.5-10.5 dpc. Themaximum likelihoodestimates to investigate randomness of nondisjunction among litters. were made for the data (Table 2) by setting derivatives to 0 In one,we rigorously simulated the null hypothesis of random and solving and the search method. The data were entered allocation using a Monte Carlomethod. Abnormality, defined into a computer program designed with the above criteria to as aneuploidy or uniparentaldisomy, was randomly allocated calculate the best estimates forthe generalizedlikelihood to each of the 429 embryos, and the resulting number of ratio statistic and Pvalues for each hypothesis. abnormals were counted in each litter and the chi-squared recomputed. This random reallocation process was repeated 10,000 times and a tally kept of how often the random alloca- RESULTS tion chi-squared statistic equaled or exceeded thatof the data chi-squared statistic. The Pvalue was taken and this tally di- Incidence of nondisjunction: Embryos harvested at vided by 10,000. 8.5 (349), 9.5 (23) and 10.5 dpc (57) were scored for Maximum Likelihood test for randomness of the distribution of the number and parentalorigin of chromosomes 7 and nondisjunction events between mouse litters: The second method The total number of embryos genotyped was 429 applied a likelihood ratioas an additional test of significance. 18. In this test, the null hypothesis was that p was equal to the and the incidence of chromosomal aneuploidies for probability of abnormality being constant over all of the em- 7 and 18 are summarized in Table 2 along with the bryos. The alternative hypothesis was that all litters have indi- chromosome parental origins. As expected, there were 670 R. J. Oakey et al.

TABLE 2 Frequency of various chromosome complementsin recovered embryos

No. of Maternal Paternal C hromosom e embryos nondisjunctions nondisjunctionsnondisjunctions embryos Chromosome Chromosome 7 Trisomy 31 15 16 Chromosome 18 Trisomy 34 20 14 Chromosomes 7 + 18 Trisomies 7 8 6 Paternal 7 uniparental disomy 4 0 8 Maternal 7 uniparental disomy 7 14 0 Paternal 18 uniparental disomy 2 0 4 Maternal 18 uniparental disomy 12 24 0 Normal complement 330 - - Total number of embryos 429 81 48 Number, classification and parental origin of aneuploid embryos identified in 429 individuals. no monosomic chromosome 7 or 18 embryos detected genotyped (not including 15 that were not assigned a at 8.5-10.5 dpc, probably because either nullisomic ga- phenotype). metes were unlikely to participate in conception or mo- Normal embryos: Among normal genotypes, 18% of nosomy was lethal in the embryo before 8.5 dpc, or embryos were retarded in phenotype at8.5 dpc, similar both. to the rate expected in any normal mouse litter. Clustering of nondisjunction events: Table 3 shows Trisomic embryos: There were 31 phenotyped embryos the distribution of embryos with nondisjunction events with trisomy of chromosome 7 (Table 4). Twenty-seven across the 76 litters in the study. Embryos with nondis- embryos were found at 8.5 dpc, 10 (37%) had devel- junction events did notseem to be randomly distributed oped normally and 17 (63%)were retarded. Therewere among the litters of mice but were clustered into fewer three at 10.5 dpc, one at 9.5 dpc, each retarded. Em- litters than expected. To investigate this observation, bryos with trisomy 7 experience a higher than normal statistical analyses wereperformed to test the hypothesis rate of retardation at 8.5 dpc; thirty-four embryos had of random allocation of abnormalities (including triso- trisomy 18, of these, 21 (61%) had developed normally mic and UPD embryos for chromosomes 7 and 18) and 13 (38%)were retarded. Of the 34, seven embryos among litters. The value of the chi-squared statistic was were harvested at 10.5 dpc, two at 9.5 dpc and 25 at calculated. For a sample of 429 embryos, there were 65 8.5 dpc. Chromosome 18 trisomy was not as severe as trisomies, 25 UPDs and 7 double trisomies distributed chromosome 7 trisomy, because almost twice as many amongst the 76 litters (Table 3). Thechi-squared statis- embryos developed normally to 8.5 dpc. Of the seven tic calculated for these data is 79.81389 with a Pvalue embryos with trisomy of both chromosomes 7 and 18, of 0.009. For the generalized likelihood ratio, the P only one outof seven was normally developed (five were value is 0.013, similar to that of the chi-squared statistic. found at 10.5 dpc and two at 8.5 dpc). Both ofthese statistical analysessupport theobservation that nondisjunction events occurring in those gametes TABLE 4 that produced aneuploid or UPD embryos were clus- Phenotype of embryos with various chromosome tered among particular matings and not randomly dis- complements tributed between the 76 litters (P= 0.009-0.013). Phenotype Comparison of chromosome aneuploidy and embryo development: The phenotype (normal us. retarded de- Normal Retarded velopment) was compared with the genotype of chro- Genotype n No. Percent No. Percent mosomes 7 and 18 for each embryo in the study. The results are summarized in Table 4. The total number Normal 333 275 (82) 56 (18) of embryos were those that have been phenotyped and Trisomy 7 31 10 (32) 21 (68) Trisomy I8 34 21 (61) 13 (38) 6 TABLE 3 Trisomy 7and 18 7 1 (14) (86) UPD 7 11 3 (27) 8 (72) Distribution of nondisjunction events among litters UPD 18 14" 7 (54) 6 (46) Total embryos 429' 298 (69) 1I5 (27) N o. nondisjunctionNo. events 0 1 2 3 >3 N o. of No. litters 25 13 21 9 8 Comparison of genotype with the phenotype of both nor- mal and aneuploid embryos. Distribution of nondisjunction events among 76 litters of One embryo of unknown phenotype. mice. 'Fifteen embryos of unknown phenotype. T ranslocation Nondisjunction7. 18 Translocation 671

Because a second Robertsonian chromosome involv- mosome 14,a trisomic embryo results from a normal ing chromosomes 2 and 8 segregated in the Rb(2,8)- and disomic gamete, and most UPD embryos are the 2Lub( 7, 18)9Lub strain of mouse, we questioned product of two abnormal gametes, a disomic and anulli- whether the retarded phenotype of the trisomy 18 or somic gamete. Monosomic 7 or 18 embryos would have trisomy 7 embryos was related to their genotype for been the products of one nullisomic and one monos@ chromosomes 2 or 8. In other words, were the more mic gamete but, as expected, none were detected in this severe trisomy 18 or trisomy 7 phenotypes due to non- protocol. Thus, the number of nullisomic gametes may disjunction of the (2, 8) Robertsonian chromosome? be inferred only from the numbers of UPD embryos. When the embryos with trisomy 18 were genotyped for Although the number of disomic and nullisomic gametes chromosomes 2 and 8 (data not shown), only one tri- might be theoretically expected to be the same (Figure somy 18 embryo was also trisomic for chromosome 8 2), there appeared to be a smaller than expected num- and surprisingly showed a normal phenotype at8.5 dpc. ber of UPD embryos reflecting a deficit of nullisomic No trisomy I8 embryos were found tohave a trisomy of gametes, just as other empirically determined studies chromosome 2. Among a subset of normally developed have shown.A maximum likelihood estimateof the num- trisomy 7 embryos, no trisomies of chromosome 2 or 8 bers of disomic compared with nullisomic gametes was were found (data not shown). A single example of a performed to investigate this observation. trisomy 7 embryo was found to alsohave trisomy of There were two statistical tests performed: (1) a null chromosome 8 but this embryo had retarded develop- hypothesis that monosomic, disomic and nullisomic ment at 10.5 dpc. Therefore we concluded that aneu- gametes are equally probable, or p(o) = p( n) = p(d) ploidy for chromosomes 2or 8due to the (2, 8) Robert- = us. the most general testhypothesis that each can sonian chromosome segregating in one of the parental assume any valuesumming to one, orp( o) + p( n) + p( d) strains of mice could not explain why some trisomy 18 = 1. For thesetwo hypotheses, the generalized likelihood or trisomy 7mice appeared to be phenotypically normal ratio of the chi-squared statistic with 2 degrees of free- while others were not. Strain differences also could not dom was 600, where the probability of x2 > 600 < 10”. be correlated with the degree of normal or retarded (2) The second test had as its null hypothesis that the developmental phenotype in embryos with trisomy of fraction of gametes with an extra chromosome (disomic) 7or 18. was the same asthe fraction of gametes lackingan entire WDembryos: Three out of 11 embryos with UPD of chromosome (nullisomic), or p(d) = p(o), but that nei- chromosome 7 developed normally to 8.5 dpc. The re- ther equaled I/:+, us. the same general alternative. For maining eight were 0.5-1 day retarded. As there were these hypotheses, the generalized likelihood ratio of the seven maternal UPDs and four paternal UPDs; no obvi- chisquared statistic with 1 degree of freedom was 23 ous difference in the frequency of each parental origin where the probability of the x‘ > 32 = 1.4 X Both for UPD could be detected. Seven out of 14 embryos of thesehypotheses were therefore strongly rejected. with UPD of chromosome 18 developed normally, six Based on the data for 429 embryos, the maximum likeli- were retarded and one was of unknown phenotype; this hood estimates of the three probabilities p( n), p( d) and is a higher rate of retardation than among embryos with p(o) were 0.85, 0.10 and 0.04, respectively. Thus, it is a normal genotype. In contrast to the situation with chro- overwhelmingly likelythat there is a bias against finding mosome 7, only two paternal UPD 18 embryos were de- embryos resulting from aneuploid gametes,especially tected compared with 12 maternal UPD embryos. gametes that lack a chromosome. Frequency of maternal and paternal nondisjunction: The frequency of nondisjunction in this mouse cross DISCUSSION was 129 events out of a possible 858 gametes, This was Incidence of nondisjunction: The germ cells of mice calculated based on a trisomy being the product of 1 carrying a single Robertsonian chromosome undergo a and a UPD the product of two nondisjunction events. higher than normal frequency of nondisjunction. The There were 81 out of 858 maternal nondisjunctions resulting gametes would theoretically be predicted to and 48 out of 858 paternal nondisjunctions (Table 2, be normal about a third of the time, disomic, a third columns 3 and 4). Therewas a larger number of mater- of the time and nullisomic, a third of the time (Figure nal nondisjunction events than paternal ones. A two- 2). Short of genotyping or karyotyping individual ga- tailed test of binomial proportions showed this to be metes, the only way to determine empirically the fre- significant P = 0.002325. quency of aneuploid gametes is to allow fertilization Maximum likelihood estimates for nullisomic anddi- and then infer thegametic genotypes from the resulting somic gametes: Figure 2 shows the possible outcomes in embryonic genotypes. Under the assumption that no the embryo from the combinations of gametesgenerated monosomic embryos would be detected at or after 8.5 in the mouse cross in this study. Each embryo (Table 2) dpc, and assuming normal, trisomic, and UPD embryos is the product of the parental gametes after meiosis. A all develop equally well to at least to 8.5 dpc, we used normal embryo is the product of two normal gametes the incidenceof aneuploidy in the embryos to infer the (containing one copy of chromosome 7and oneof chro- relative frequency of aneuploidy in gametes. 672 R. J. Oakey et al.

The rate of nondisjunction in this mouse cross is in a trisomic conceptus, then the estimate of 4% for 15%. This is consistent with rates of nondisjunction nullisomic gametes would be an overestimate, which seen in other Robertsonian crosses. For a (2, 6)Robert- only further strengthens the argument forsevere selec- sonian heterozygote intercross the rate of nondisjunc- tion against nullisomic gametes. tion was reported to be 17% when measured in oocytes Development of uniparental disomic embryos: Gy- (CHEWBOTARand BARILYAK1994), whereas a double nogenetic and androgenetic embryos generated from Robertsonian heterozygote intercross experienced non- nucleartransplantation experiments (MCGRATHand disjunction in 28-36% of embryos (BEECHEYand SOLATER1983,1984) can progress to 10.5 dpc(SURAN1 et SEARI.E 1988). Crosses involving female heterozygotes al. 1986). However, these were exceptional occurrences and male hemizygotes for an Xautosomeintercross un- and most embryos of this type do not survive this long. derwent nondisjunction at-10% in spermatocytes and In these studies, developmental syndromes affecting 5% in offspring (TEASEand FISHER1991; ADLERet al. largely thetrophoblast andthe embryo respectively, 1989) although Xautosome Robertsonians are not en- generally arrest normal development at amuch earlier tirely comparable with rearrangements involving au- stage. Because these embryos have a correct chromo- tosomes only (TEASEand FISHER1991). Overall, nondis- somal complement, the lethality is due to imprinting, junction events during female meiosis I caused more namely unequal expression of certain genes on the pa- (63%) of the trisomies and UPDs than male meiosis I rental alleles. Studies of intercrosses of mice carrying (37%),implying a genderdifference in the incidenceof one Robertsonian chromosome or reciprocal transloca- nondisjunction. Of course, nondisjunction events that tion chromosome led to the definition of particular occurred during meiosis I1 would not have been de- chromosomal regions that show imprinting effects tected in thiscross. However, because Robertsonian (SEARLEand BEECHEY1985). Chromosomal regions chromosomes are not likely to cause an increase in that showed noncomplementation, i.e., two copies from frequency of nondisjunction of chromatids in meiosis a single parent could not substitute for a missing copy 11,we would not expect such events to have distorted from the other parent included portions of chromo- these data appreciably. Only limited data have been somes 2, 6, 7, 8, 11, 12 and 17. Lack of complementa- available forcomparing rates of nondisjunction in tion is occasionally nonreciprocal. For example, mice males us. females experiencing embryonic lethality be- with uniparental disomy (UPD) of mouse chromosome cause most aneuploid embryos have been characterized 6 are normal and survive when both chromosomes are by karyotypic analysis, however, first cleavage embryos of paternal origin but maternal UPD embryos are not may be sex distinguished because the female chromo- viable (SEARLEand BEECHEY1985; CATTANACH1986). somes are slightly ahead of the male ones (BEECHEY For chromosome 7, complementation studies with a and SEARLE1988). In one case, equal rates of nondis- translocation chromosome called 19H has established junction were detectedin males and females of an that embryos with paternal duplication, maternal defi- intercross involving chromosome 15 Robertsonians ciency of distal mouse chromosome 7 die before day (BEECHEYand SEARLE1988). Others have reported a 11 post coitum (SEM,E and BEECHEY1990). Embryos slightly higherincidence of nondisjunction in male with maternal duplication and paternal deficiency of compared with female mice when karyotyping sperma- distal chromosome 7 are growth retarded and die tocytes and oocytes (TEASEand FISHER1991); in con- around day 16 post coitum (FERGUSON-SMITHet al. trast however, others have found the reverse (GROPP 1991). Embryos with maternal duplication of the cen- and WINKINC,1981). tral portion of chromosome 7 are subject to postnatal The results of the maximum likelihood estimate lethality whereas paternal duplication has not been as- strongly supported the suggestion that there was a bias sociated with an affect (CATTANACHet al. 1992). Mater- against aneuploid gametes in general, especially those nal duplication of the proximal part of chromosome 7 that lack an entire chromosome. The best estimate for causes neonatal lethality and the corresponding pater- the frequency of disomic gametes was 10% and for nulli- nal duplication has affects on postnatal growth and via- somic gametes 4%. As this estimate was based on the bility (SEARLEand BEECHEY1990; CATTANACHut al. assumption that we would not see any monosomic em- 1992). This chromosome must therefore be a very rich bryos, as monosomy is usually a more severe phenotype source of imprinted genes. It is not clear at what devel- than other aneuploidies (BEECHEYand SEARLE1988), opmental stage the whole chromosome 7 maternal or the frequency of nullisomic gametes was estimated from paternal UPD is lethal. In these molecular studies, the the incidence of UPD. Previously,this categoly of non- latest stage at which normally developed maternal or disjunction would not have been detected because an- paternal UPDs of chromosome 7 were observed was at euploid embryos have been analyzed by chromosome 8 dpc. Our results aretherefore consistent with the karyotyping and thus parentalorigin could not be ascer- wealth of previous data demonstrating that chromo- tained. In this study, if UPD were not the result of a some 7 is imprinted and that particular genes on that conception between a nullisomic and disomic gamete, chromosome areexpressed exclusively from one paren- but instead were the result of mitotic nondisjunction tal allele or the other. 7, 18 Translocation Nondisjunction 673

For chromosome 18 UPD embryos, the rate of retar- TABLE 5 dation was 46%, higher than in the normal genotype Distribution of litters among 12 FI males category (18%).This was an unexpected result because chromosome 18 is not known to have imprinted genes. Litter with Litter without In one cross involving the DOH and the TI 8H translo- Male nondisjunction nondisjunction cation chromosomes, most of the maternal and the CXB #9* 8 1 most distal portions of the paternal chromosome 18 CXB #1 2 3 were shown not to exhibit imprintingeffects (BEECHEY RXD #3 1 1 et al. 1991). This left a small part of the maternal chrc- CXB #8 3 2 mosome and most of the paternal chromosome un- RXD #1 1 0 tested for imprinting. Thus, one explanation for these BX3 #1 2 4 BX3 #3* 6 2 observations is that the untested portions of chromo- DXR #2 2 0 some 18 do indeed have one ormore imprinted genes. DXR #7 1 1 CATTANACHand BEECHEYhave recently argued that BX3 #4 4 1 in crosses involving translocation chromosomes, differ- RXD #2 3 0 ences in the frequency of paternal us. maternal UPD RXD #4 1 0 offspring may not be theresult of imprinting at all but Total 34 15 instead may reflect biases against aneuploid gametes The genotypes of males and the numberof litters with one duringmaternal or paternal gametogenesis (CATTA- or more and without nondisjunction events are detailed. Of NACH and BEECHEY1994). Based on evidence from the 76 litters, 49 litters have fathers of an assigned number, other translocation chromosomes, the discordance be- for the remaining 27, the strain of the fatheronly is recorded and these data do not appear here. BXC males were sterile tween maternal us. paternal frequencies was attributed and 3XB males were not used. to either nonbiological causes or to nonrandom segre- gation of chromosomes to the polar bodies in the oo- cytes of the female (EICHENLAUB-RITTERand WINKING and seven double trisomies gave a chi-squared statistic 1990; TEASEand FISHER1991; CATTENACHand BEECHEY of 79.81389 and a P value of 0.009. The generalized 1994). Inour cross, indeed therewere onlytwo paternal likelihood ratio Pvalue for this data set was 0.013, simi- chromosome 18 UPD embryos out of429 compared lar to that of the chi-squared statistic. This suggests that with 12 maternal UPDs, suggesting an increased recov- the nondisjunction events in these mice were not dis- ery of UPDembryos of maternal us. paternal origin.An tributed randomly among the litters, but that some lit- excess of maternal over paternal UPD among offspring ters were more likely to contain aneuploid or UPD em- has also been seen in studies of chromosome 1, while bryos than others. The factors that could have caused a reciprocal excess of paternal over maternal has been this clustering are not clear. Because females in this seen with chromosomes 5, 9 and 14 (GTTANACHand cross were sacrificed atthe time embryos were har- BEECHEY1994). However, the major argument for im- vested, we could not determine if females in which we printing of chromosome 18 rather than nonrandom observed clustering of nondisjunction events were more segregation during gametogenesis is the observation likely to have more clustering of nondisjunction events that nearly half of all chromosome 18 UPD embryos in future litters. The distribution of nondisjunction showed growth retardation, regardless of whether ma- events was not equal between all of the males. Forty- ternal or paternal UPD for chromosome18was present. nine out of the 76 litters had a father with an assigned Thus, although there appears to be an imbalance in number (Table 5).Among the 12 differentmales, most the frequency of paternal versus maternal UPD for produced approximately equal numbers of litters with chromosome 18 in our cross, which may reflect in part entirely normal embryos and litters with embryos with some nonrandom segregation of chromosomes to polar one or more nondisjunction event. However, two out- bodies and oocytes, the occurrence of growth retarda- standing males produced six and eight litters con- tion in UPD embryos of either parental origin serves as taining oneor more offspring with nondisjunction evidence for an imprintingeffect for this chromosome. events (Table 5). It is therefore possible that events in A number of other translocation chromosomes exist male meiosis and gametogenesis as well as in female that could be used to test particular regions of chromo- meiosis, gametogenesis, or pregnancy, may contribute some I8 more thoroughly for imprinted genes. to thenonrandom distribution of nondisjunction Clustering of nondisjunction events: The distribu- events seen in these litters. tion of embryos with nondisjunction events (Table 3) There are a number of possible explanations for the suggests that they are clustered among certain matings clustering. These explanationsinclude mitotic nondisjunc- and therefore not distributed randomly among the 76 tion leading to germline mosaicism for aneuploid gamete litters. Statistical analyses wereperformed on these data precursors in some individuals, increased tendency to mei- to asses the validity of this observation. In the sample otic nondisjunction in some individuals, or factors that of 429 genotyped embryos, the 65 trisomies, 25 UPDs allow some females to carry chromosomally abnormal fe- 674 R. J. Oakey et al. tuses longer than other females. As we did not observe Willi syndrome which shows an absence of Snrpn expression. Nat. Genet. 2: 270-274. clustering of aneuploid or UPD embryos in crossesinvolv- CHEWXITAR,N. A., and I. R. BARILYAK,1994 Comparison of the ing one or the other F1 hybrid strain of mice in our cross features of chromosomal nondisjunction during thefirst division (Table l),we could draw no conclusions as to the role, if of oocyte maturation in mice which are heterozygous for the balanced structural chromosomal rearrangements. Tsitol. Genet. my, played by genetic background in the frequency of 28: 11- 19. nondisjunction or in generating any differencesin a puta- DAVISSON,M. T., and E. C. &SON, 1993 Recombination suppres- tive tolerance toaneuploidy. Consequently, the mecha- sion by heterozygous Robertsonian chromosomes in the mouse. Genetics 133: 649-667. nism of the observed clustering remains obscure. DIETRICH,W., H. KAT%, S. E. LINCOLN,H.3. SHIN,J. FRIEDMANet al., Inhumans, there also appears to be clustering of 1992 A genetic map of the mouse suitable for typing intraspe- cific crosses. Genetics 131: 423-447. nondisjunction events in sibships in which trisomy 21 EICHENIAUB-RITTER,U., and H. WINKING,1990 Nondisjunction, dis- has occurred. The recurrence risk for nontranslocation turbances in spindle structure, and characteristics of chromo- trisomy 21 in the offspring of young women is elevated some alignment in maturing oocytes of mice heterozygous for Robertsonian translocations. Cytogenet. Cell Genet. 54: 47-54. three- to fivefold over the general population risk for FERGUSON-SMITH,A. C., B. M. CATTANACH, S. C. BARTON,C. BEECHEY age-matched control women (GARDNERand SUTHER- and M. A. SURANI,1991 Embryological and molecular investiga- tions of parental imprinting on mouse chromosome 7. Nature LAND 1989). Although one has to be cautious in com- 351: 667-670. paring the clustering of aneuploid offspring in humans GARIINER,R. J. M., and G. R. SUTHERIAND,1989 Down syndrome, with that in the mouse, the notion that there could pp. 137-143 in Chromosome Abnormalities and Genetic Counsrling. Oxford monographs onmedical genetics no. 17. Oxford Univer- be some common mechanism that explains either a sity Press, Oxford. predisposition to nondisjunction and/or aneuploid em- GWHART, J.D., M. L. OSTER-GRANITE,R. H. REEVES and J. Y. COYIX, bryo survival is suggested by these observations. 1987 Developmental consequences of autosomal aneuploidy in mammals. Dev. Genet. 8: 248-265. In summary, the use of Robertsonian translocation GROW,A,, and H. WINKING,1981 Robertsonian translocations: cy- chromosome harboring mice can be used to study ele- tology, meiosis, segregation pattern and biological consequences of heterozygosity, pp. 141-181 in Biology ofthe House Moust, ed- vated rates of nondisjunction in meiosis. The frequency itcd by R. J. BERRY.Academic Press, New York. of nondisjunction in the gametes was inferred from an HASXLD,T. J., 1985 The origin of aneuploidy in humans, pp. 103- examination of the genotypes of the embryos. These 115 in Aneuploidy, edited by V. I.. DEI.IARCO,P. E. VOYTEK, and A. HOLLAENDER.Plenum Press, New York. kinds of studies are useful for examining thefrequency KAUFMAN, M. H., 1992 Assessment of developmental stage of pre- of, and potentially the mechanisms involved in, gamete and postimplantation mouse embryos ba5ed on the staging sys- tem of Theiler (1989), pp. 35- 11 1in The Atlas of Muusr Da1r1qI.1- selection and success in mice. In addition, the maternal ment. Academic Press, London. and paternal UPD embryos generated in this cross will LYON,M. F., H. C. WARDand G. M. SIMPSON,1976 A genetic method be used to search for additional imprinted genes on for measuring nondisjunction in mice with Robertsonian translo- cations. Genet. Res. 26: 283-295. chromosome 7 and to explore chromosome 18 for im- MCGRATII,J.. and D. SOI.TF.R,1983 Nuclear transplantation in the printed loci as well. mouse embryo by microsurgery and cell fusion. Science 220: 1300-1302. Thanks go to STEPHENA. LIERHABERand MURRAY H.BRII.I.IANT MCGRATH,,J., and D. SOLTER,1984 Completion of mouse em- for generously allowing this work to continue in their respective labo- bryogenesis requires both the maternal and paternal genomes. ratories. Thanks to MICHAEI.H. MAI.IM for his patient help and for Cell 37: 179-183. SEARI.E,A. G., and C. V. BEECHEY,1978 Complementation studies reading this manuscript. Thanks to PA51 A. JANNE for his encourage- with mouse translocations. Cytogenet. Cell Genet. 20 282-303. ment and to DONNA DURHAM-PIERREand MURRAY H. BRIILLANTfor SEARIX,A. G., and C. V. BEECHEY,1982 The use of Robertsonian their critical reading of the paper. This work was supported by the translocations in the mouse for studies on nondisjunction. Cyte Howard Hughes Medical Institute at the University of Pennsylvania genet. Cell Genet. 33: 81-87. and Princeton University. SEARI.E,A. G., and C. V. BEECHEY,1985 Noncomplementation phe- nomena and their bearing on nondi$unction events, pp. 363- 376 in Aneuploidy, edited by V. L. DEIARCO,P. E. VOYTEK,and LITERATURE CITED A. HOIMENDER.Plenum Press, New York. SEARI.E,A. G., and C. V. BEECHEY,1990 Genome imprinting phe- ADI.ER,I.-D., R. JOHANNISSON and H. WINKING,1989 The influence nomena on mouse chromosome 7. Genet. Res. 56: 237-244. of the Robertsonian translocation Rb(X.2)2Ad on anaphase I SEARI.E,A. G., C. E. FORU and C. V. BEECHEY,1971 Meiotic disjunc- nondisjunction in male laboratory mice. Genet. Res. 53: 77-86. tion in mouse translocations and the determination of centro- BEECHEY,C. V., and A. G. SEARI.E,1988 Effects of zero to four copies mere position. Genet. Res. 18: 215-235. of chromosome15 on mouseembryonic development. Cyto- SNEI-I.,G. D., 1946 An analysis of translocations in the mouse. Genet- genet. Cell Genet. 47: 66-71. ics 31: 157-180. BEECHEY,C. V., B.M. CATTANACHand A. G. SEARI-E,1991 Genetic SURANI,M. A,, S. C. BARTONand M. L. NORRIS,1986 Nuclear trans imprinting map. Mouse Genome 89: 60-61. plantation in the mouse: heritable differences between parental ge- BOND,D. J., and A.C. CHANDI.EY,1983 The origins and causes of nomes after activation of the embryonic genome. Cell 45 127-136. TABOR,S., 1994 Enzymatic manipulation of DNA and RNA, pp. aneuploidy in experimental organisms. pp. 27-54 in Aneuploidy, 3.10.2-3.10.4in CurrentProtocolsinMoleculurBiology, Ed. 8, edited edited byD. J. BONDand A. C. CHANDLEY.Oxford University by F. M. ALJSUBEI.,R. BRENT,R. E. KINGSTON, D. D. Moow and Press, Oxford. J. D. SEIDMANtt al. Greene Publishing Associates Inc and Wiley CATTANACH,B. M., 1986 Parental origin effects in mice. J. Embryol. and Sons Inc., New York. Exp. Morph. 97 (Suppl.): 137-150. TEASE,C., and G, FISHER,1991 Two new X-autosome Robertsonian CATTANACH,B. M., and C. V. BEECHEY,1994 Furthur evidence that translocations in the mouse. Meiotic chromosome segregation differential recoveries of maternal and paternaldisomies are not in male hemizygotes and female heterozygotes. Genet. Res. 58: attributable to imprinting. Mouse Genome 92: 504. 115-121. CATTANACH, B.M., J. A. BARR, E. P. EVANS,M. BURTENSHAW,C. V. BFECHEY,et al., 1992 A candidate mousemodel for Prader- Communicating editor: R. E. GANSCHOW