SIB1 -1: A Mutation Affecting Meiotic Sister-Chromatid ... - Europe PMC

0 downloads 0 Views 3MB Size Report
arms of the sister chromatids appear closely associated ...... The role of sisterthromatid association in ensuring disjunction of recombinant chromosomes.
Copyright 0 1993 by the Genetics Society of America

SIB1 -1: A Mutation Affecting Meiotic Sister-Chromatid Association in Yeast Michele Flatters and Dean Dawson' Department of Molecular Biology, Tufts University, Boston, Massachusetts021 11 Manuscript received October23, 1992 Accepted for publication February24, 1993

ABSTRACT Meiotic chromosome segregationmust occur with high fidelity in order to prevent the generation of deleterious aneuploidies. In meiosis I, homologous chromosomes pair, then migrate to opposite poles of the spindle. This process uses a collection of unique structures and mechanisms that haveyet to be thoroughly characterized.To acquire a collection of informative meiotic mutants, we carried out a novel genetic screen in Saccharomyces cerevisiae. This screen was designed to identify dominant mutants in which meiosis I chromosomesegregationoccurs with decreasedfidelity.One mutant recovered using this screen, SIDl-1 (sister disjunction), showed an incidence of spores disomic for a marked chromosome IZI that was 25-fold greater than the wild-type level. Crossing-over is slightly, butnotdramatically,reduced in S I D l - 1 . Both recombinantandnonrecombinant chromosomes segregate with reduced fidelity in the presence of SZDl-1. We present evidence that the mutant is defective in sister-chromatid association. "

M

EIOTIC chromosome segregation is an essential process for sexual reproduction and must occur with high fidelity in order to ensure that viable meiotic products are produced. In meiosis, there are two rounds of chromosome segregation and cell division following only one round of DNAreplication (Figure 1A). T h e first meiotic division results in the segregation of homologous chromosomes to opposite spindle poles in a reductional division. T h e second meiotic division is equational, in that sister chromatids separate from one another to generate four haploid nuclei. Through a still mysteriousprocess,homologous chromosomes pair early in meiosis I and establish a durable association. This association is maintained until anaphase I , at which time the homologs separate and segregate to opposite spindle poles. In cytologically tractable organisms (such as grasshoppers), the arms of the sister chromatids appearclosely associated untilanaphaseI. Following anaphaseI,the sister chromatids remain joined, apparently by virtue of an association atthecentromere region(reviewed in MOENS 1987). With few exceptions, in order forhomologous chromosomes to move away from one another with high fidelity at anaphase I , they must undergo reciprocal genetic exchange (crossing-over) [reviewed in BAKER et al. (1976)l. This conclusion is supported by the observation that in many organisms mutations which eliminate crossing-over lead to aneuploidy and/or sterility. Recombination eventsare thought to participate in holding homologous chromosomes together prior to anaphase I [reviewed in HAWLEY (1988)l.

' T o whom correspondence should be addressed. Generics 134 423-433 (June, 1993)

Although crossing-over is clearly necessary to ensure correct segregation at anaphaseI, it is not sufficient. DARLINGTONand later MAGUIREpointed o u t that while a crossover would establish a physical linkage between homologs, other factors would likely be necessary tosecurethe association (DARLINGTON 1932; MAGUIRE1974). Genetic studies inyeast and other organismshave yielded a small collection of mutants whose phenotypes are consistent with this idea. For example, mutations at theRED1 and MER1 loci of yeast result in elevated levels of nondisjunction (ROCKMILL and ROEDER1988; ENCEBRECHT and ROEDER 1989). In the aneuploidprogeny, the disomic chromosomes are sometimes recombinant, indicating that in thesemutants, crossovers are notensuring segregation. Similarly, cytological experiments with the desynaptic (dy) mutants of maize demonstrate that these mutants can form apparently normal chiasmata (the cytological manifestationsof crossovers), but these fail to securely join homologs until anaphase I (MAGUIRE1978b). These studies suggest that something else must be acting, in addition to crossovers, to ensure disjunction of recombinant homologs. The natureof additional factorsthat, in conjunction with chiasmata, ensure disjunction has remained elusive. Numerous proposals have been proffered (MAGUIRE 1974). These models fall into two general categories. In the first,all factors or conditions necessary for chiasma maintenance are at the chiasmata. In the second groupare models that suggest that sisterchromatid cohesiveness distal to the crossover is necessary for chiasma maintenance. Models formulated to explain this cohesiveness attribute it to synaptonemal complex remnants (MAGUIRE1978a), chromatid catenation (MURRAYand SZOSTAK 1985) or accessory

424

M. Flatters and D. Dawson

FIGURE1 .-Segregation of marked copies of chromosome ZZZ in meiosis. Chromosome ZZZ homologs in DM32 were differentially marked (see MATERIALS AND METHODS): one carries the URA3 and HIS3 genes (Ura+, His’), and the other carries the LYSP and TRPI genes (Lys+, Trp+). (A) A normal meiosis. Homologous chromosomes pair, recombine and synapse, then segregate to opposite poles at meiosis I (recombination not shown for simplicity). (B) A meiosis I homolog nondisjunction. Homologs erroneously move to the same pole at meiosis I, producing two Ura+His+Lys+Trp+disomic sister spores following meiosis 11. (C) A precocious separation of sister chromatids. One chromatid precociously separates from its sister and segregates with the other homolog at meiosis I. A three spore viable tetrad results. Two spores have one parental chromosome; the third spore is a Ura+His+Lys+Trp+chromosome I21 disome; the fourth is inviable. D. A meiosis I1 nondisjunction. Both sister chromatids move to the same pole at meiosis 11. A three spore viable tetrad results. Two spores have one copy of one of the parental chromosomes, while the third spore is disomic for the other parental chromosome. The fourth spore is inviable.

cohesive proteins (RATTNER,KINCWELLand FRITZLER 1988). The model that sister-chromatid association may be required to help maintain chiasmata is supported by both genetic and cytological studies. T h e ord mutant of Drosophila exhibits high levels of both precocious separation of sister chromatids (PSSC) and homolog nondisjunction (MASON 1976; MIYAZAKIand ORRWEAVER1992). Cytological analysis has shown that in ord mutants sister chromatids fail to remain associated along the lengths of their arms during early meiosis I (GOLDSTEIN1980; MIYAZAKI and ORR-WEAVER 1992). This phenotype correlates with premature sister separation and homolog nondisjunction. This suggests that sister-chromatid cohesion may be required to maintain homologous associations. The DISI-I mutation in Saccharomyces cerevisiae (ROCKMILL and FOGEL 1988) also generates disomes as a result of precocious separation of sister chromatids at meiosis I. It has been proposed thatthe DISl defect results in equational divisions at meiosis I, due to inappropriate expression of factors required for kinetochore differentiation or sister separation (ROCKMILLand FOGEL 1988). Interestingly, this mutant also exhibits mitotic nondisjunction, perhaps reflecting the need for regulated sister chromatid association in mitosis. We have undertaken a mutant hunt to detect diploids that generate disomes at high frequencies as a result of errors in meiosis I segregation. We describe the genetic characterization of adominant allele, SIDI-I, recovered in this screen. MATERIALSANDMETHODS Media: Synthetic complete (CM), rich media (YPD) and sporulation media (SPO) were prepared as described (SHERMAN, FINKand LAWRENCE 1979). Synthetic complete me-

dium lacking one or more nutrient(s)is denoted by CM-the nutrients lacking (e.g., CM-Arg isCM withoutarginine). Cyclohexamide (Cyh) and canavanine (Can) were added as indicated to 10 pg/ml and 60 pg/ml, respectively. Media containing5-fluoroorotic acid (5-FOA) was prepared by adding 5-FOA at 1 mg/ml and uracil at 50 rg/ml to CMUra. Media containing a-amino-adipate (a-AA) was prepared as described (TRECO 1989). Strain construction: S. cerevisiae strains are described in Table 1. T h e mutant hunt strain (DM32) carries a pair of specially marked chromosome I l l s . On one chromosome two marker genes (URA3 and HIS3) have been inserted at the LEU2 and PGKl loci, and on the other a different pair of marker genes (LYS2 and T R P l ) was similarly inserted. T h e chromosome ZZZ pair in DM32 (see Figure 1) was constructed using the single-step replacementtechnique (ROTHSTEIN 1983). Each chromosome ZZZ homolog was constructed in a haploid, and the two haploids were mated to generate DM32. These insertions were added to chromosomes using restriction fragments from a set of plasmids bearing LYS2 and URA3 inserts in LEU2, and HIS3 and T R P l inserts adjacent to P G K I . A HpaI/SalI fragment of pMH4 in which a 1.l-kbp fragment carrying the URA3 gene was cloned into the Clal site of LEU2 was used to put URA3 into LEU2. This strain was constructed by MAYA HANNA, who generously allowed us to use it for our experiments. A BamHI fragment carrying the HIS3 gene was inserted into the BglII site upstream of the P G K l geneon acloned fragment to create plasmid pD40. An EcoRI fragment from this plasmid was then used to replace the PGK locus on chromosome IZZ. A SalI/SacI fragment from pMH7 (a gift from MAYAHANNA) bearing theLYS2 gene on an EcoRI/ ClaI fragment inserted into the ClaI site of LEU2 was used to targetLYS2 to theLEU2 locus. An EcoRI TRPl fragment was inserted into the BglII site of a cloned portion of the PGK region to create plasmid pD133. An EcoRI fragment from this plasmid was then used to target the insertion to chromosome ZIZ. All constructions were confirmed both by tetrad dissection and by Southern analysis (data not shown). T h e chromosome VIZZ arg4::URA3 marker in DM162 and

425

Meiotic Sister-Chromatid Association TABLE I Yeast strains Strain

DM32 DM101 DM 155D

Genotype

Source

a / a , adel/ADEl, lysZ-BglZA, ura3-52, trpl-289, leu2::URA3/leu2::LYS2, his3-11,15, PGKl::HIS3/PGKI::TRpl, sidl+, CYHZ/cyh2, CANllcanl-I a/a, ADEl, lysZ-BglZA, ura3-52, trpl-289, leu2::URA3/leu2::LYS2, hid-I 1,15, PGKI::HIS3/PGKl::TRPI, SIDl-l/sidl+, CYH2/cyh2, CANllcanl-1 a / a , adel/ADEl, lys2-Bg12A, ura3-52, trpl-289, leu2::URA3/leu2::LYS2, his3-11,15,

This study This study This study

PGKl::HIS3/PGKI::TRpl,SIDI-l/sidl+ DM165 154 DM DMl6lA 162 DM DM163

a/a, ADEl, IysZ-BglZA,ura3-52, trpl-289, leu2::URA3/leu2::LYS2, his3-11,15, PGKl::HIS3/PGKl::TRPl, SIDI-l/SIDI-I, CYH2lcyh2, canl-1 a/a, adel/ADEI, ura3-52, trpl-289, leu2::URA3/leu2, his3-l1,15,PGKI::HIS3/ PGKl, sidl+, CANl, arg4AlARG4, [YLp: leu2, TRP1, ura3, ARG4, his31 ala, adel/ADEl, ura3-52, trpl-289, leu2::URA3/leu2, his3-11,15, PGKI::HIS?/ PGKl, SIDl-l/sidl+, CANllcanl-I, ARG4, [YLp: leu2, TRPl, ura3, ARG4,his31 a / a , ADEl, lysZ-BglZA, ura3-52, trpl-289, leu2::LYS2/LEU2, his3-11,15, pgkI::TRPl/PGKl, SIDl-l/sidl+, argl::URA3/ARG4 a/a,ADEl, lysZ-BglZA, ura3-52, trpl-289, leu2::LYS2/LEV2, his3-I 1,15, pgkl::TRPl/PGKI, SIDl-l/SIDl-I, arg4::URA3/ARG4

DM163 was obtained by dissecting MGD254 (SCHULTES and SZOSTAK 1990), andmating to aSZDI-1 or wild-type haploid. Confirmation of chromosome ZZZ disomes: T o estimate the background level of chromosome ZZZ disomic spores generated by our wild-type mutanthunt strain, DM32, singlecolonieswere sporulated onSPO plates and then replica plated to CM-Arg,Ura,His,Lys,Trp+Cyh+Can plates. These plates select for chromosome ZZZ disomes by the Ura, His, Lys and T r p selection, and for haploid cells with the Cyh and Can resistances. T o test whether these colonies were truly disomic, we took advantage of 5-FOA to select against cells containing a URA3 gene. Papillae arising withincolony imprints were streaked onto FOA plates, to select for haploids that had lost the leu2::URA3, PGKl::HZS3 chromosome IZI. FOA' colonies were found to have lost the ability to grow on CM-Ura and CM-His, while retaining the ability to grow on CM-Trp and CM-Lys, and regaining strong mating competence. This indicates that papillae arising on CM-Arg,Ura,His,Lys,Trp+Cyh+Can following meiosisare truechromosome ZZZ disomes. Isolation of mutants: About 50,000 DM32cells were plated on 200 YPD plates. These were then immediately UV irradiated to 80% survival (40sec from 1 m with a 30W germicidal lamp). Plates were incubated for 3days at 30" to allow cells to grow into well-separated colonies. These were then replica-plated to SPO plates to induce meiosis. Once sporulation was complete (4 days at 30°), the plates were replica plated to CM-Arg,Ura,His,Lys,Trp+Cyh+Can to select for chromosome ZZZ disomichaploids. Approximately 40,000 mutagenized diploids were screened for increased production of chromosome ZZZ disomes in this manner. Quantitation of meiotic and mitotic levels of nondisjunction: For the purposes of initial and rapid quantitation of the levels of meiotic segregation errors, a quantitative tetrad plating technique was used. Sporulated patches of cells were resuspended in sterile water, and dilutions were plated on CM-Arg,Ura,His,Lys,Trp+Cyh+Can to determine the numbers of chromosome ZZZ disomes. Dilutions were alsoplated on CM-Arg+Can+Cyh medium to measure the number of spores that have acquired both the cyh2 and c a n l - 1 alleles. By comparing the numbers of colonies ob-

This study This study This study This study This study

served on these media, we were able to estimate the relative frequencies of chromosome ZZI disome production. For a more accurate determination of the levels of chromosome 111 missegregation, tetrads were dissected, and tetrad analysis was used to determine the number of meioses in which chromosome ZZZ experienced a segregation error. This method was also used to examine the segregation of chromosome VZZZ in the SZDl-I mutant. To determine the levelsofmitotic loss, single diploid colonieswere resuspended in sterile water, and dilutions were plated on YPD, todeterminethe concentration of viable cells, and on FOA plates, to detect the number of diploids which had become monosomic for chromosome ZZZ through the loss of the leuZ::URA3,PGKI::HIS3 version of chromosome ZZZ. FOA' colonies could be generated by loss or by mitotic crossing-over. Colonies generated by loss events would also become His-, while those generated by mitotic crossovers remain His+. Quantitation of recombination in tetrads and disomic spores: T o determine the levels of crossing-over in SZDI-I and sidl+ strains, tetrads were dissected, and the fraction of four-spore viable tetrads in which there had beena crossover in the leu2-PGKl and the PGKl-MAT intervals was determined. Disomic spores are prototrophic for uracil, histidine, lysine and tryptophan. However, it is not possible to determine the linkage of the markers in disomic spores simply from the prototrophies. T o determine whether the chromosome ZZZs in disomic spores were recombinant or not, spores that were identified as disomic by tetrad dissection were streaked onto a-AA (a-amino-adipate) plates. This medium allows the growth of cells that have lost the LYSZ bearing chromosome ZIZ homolog. By examining the linkage of the markers on the remaining homolog, it is possible to determine whether the chromosomes in the disomic spore were recombinant or nonrecombinant. Colonies arising on these plates were then tested by replica-plating to determine which of the chromosome 111 markers were lost simultaneously with the LYSZ gene. RESULTS

Identification of mutants that exhibit high levels of meiotic chromosome segregation errors: A screen

426

M. Flatters and D. Dawson

was developed to allow the detection of mutants that generate viable disomic spores at increased levels relative to wild-type cells. Briefly, diploid cellsbearing a specially marked chromosome ZZZ pair were plated and then UV irradiated, and allowed to grow into colonies on YPD. These were sporulated by replicaplating to SPO medium. Sporulated colonies were then replica-plated to amedium that selects for spores bearing chromosome ZZZ disomes. This allowed a visual screen for colonies that yielded increased levels of these disomic spores, which appeared as papillae within the colony imprint. The parent diploid, DM32 (Table l), has two notable features. First, it has distinctly marked copies of chromosome ZZZ, each homolog bearing a different pair of markers. The pairs of markers are inserted at identical positionson each homolog, each flanking the centromere by approximately 10 CM (Figure 1). The opposing position of the pairs of genes (URA3 and LYSP, and HIS3 and T R P l ) makes it impossible for all four markers to end up on one chromosome as the result of recombination. A second feature of DM32 is that it is heterozygous for both canavanine and cyclohexamide resistances. In both cases, the sensitive allele is dominant. Thus, unsporulated diploids can be selectedagainst by demanding simultaneous cyclohexamide and canavanine resistance. Sincethese segregate independently and are not centromere linked, approximately one in four spores are cyh'andcan'. This assures that we assay only haploid meiotic products. Mutagenesis was done in a diploid strain, to isolate dominant mutants, and the screen required that the meiotic products be viable. Thus, we did not expect to recover any meiotic lethals,or any other mutations that dramatically affected chromosome segregation such that inviable aneuploids were generated at high frequencies. Instead, we expected thatthe screen would detect mutations that were more subtle and merely tweaked the meiotic segregation process. Approximately 40,000 mutagenized colonies were screened. In the initial screen, 181 colonies showed elevated levels of papillation. These were retested to eliminate false positives.A collection of9 1 potentially interesting mutants was obtained and subjected to successive backcrossing to eliminate extraneous mutations and to establish whether 2:2 segregation of a mutant phenotype could be demonstrated. Backcrosses were performed by sporulating mutant d i p loids, scoring the spore colonies for the relevant auxotrophies and drug resistances, and thenmating these with appropriate haploids, derived from DM32, to create diploids that could be tested in the papillation assay. In these backcrosses, most of the colonies that had tested as positive in the initial screen either behaved as polyploid strains or had new linkages of the markers we were following, suggesting they carried

FIGURE2.-Segregation of S I D l - I in heterozygous diploids. A SIDI-l/sidl' diploid was dissected. Relevant genotypes in each spore were scored, and each spore was mated to a suitable wildtype haploid to generate diploids suitable for our papillation assay. The resultingdiploids(A-D) were patched onto YPD, replica-plated to SPO, and following sporulation, replica-plated to media to select haploid cells disomic for chromosome III. The high papillation phenotype of S I D I - l segregates 2:2, indicating the phenotype is attributable to a single dominant mutation. The top patch at the right represents the original S I D I - l / s i d l + diploid isolate, while the two remaining patches are wild-type diploids.

chromosome ZZZ translocations.Both wouldbeexin our screen. pected to show high levels of papillation Followingsix rounds ofbackcrossing we wereleft with a pair of putative mutants. @Cl (Qu'est-ce que c'est)has a weak nondisjunctionphenotype (not shown) and will be described elsewhere. The classification of the other,SZDl-l, is described below. SZDl-l; adominantmutation affecting meiotic chromosome segregation:SZD1-l behaves as a single dominant allele in our genetic tests (Figure 2). The four patches labeled A-D represent the four spores of one tetrad from a S D I - l / s i d l + diploid. Each of the spores was mated to a wild-type sidl+ haploid with complementary chromosome ZZZ and CYH and CAN alleles to create diploids suitable for testing with our papillation assay. These diploids were sporulated and replica-plated to disomic selection media. The high papillation phenotype of SZDl-l segregates 2:2 (patches A and C). This 2:2 pattern was observed in 24 of 24 tetrads examined. The three patches on the right represent two wild-type diploid strains (middle and bottom) and a patch of the original SZDl-l/sidl+ diploid isolate (top). Mitotic chromosome behavior in SZDl-I strains: Two types of assays were used to evaluate the effect of SZDl-l mutants on mitotic chromosome behavior. First, quantitative plating assays (see MATERIALS AND

Association Sister-Chromatid Meiotic were performed to crudely assess the mitotic stability of the URA3 bearing version of chromosome ZZZ in wild-type (DM32) and SZDI-l/sidl+ (DM 101) diploids. Colonies grown on CM-Arg-UraHis-Lys-Trp were plated on 5-FOA plates to determine the fraction of cells that were Ura-. Such cells could arise by chromosome loss or mitotic crossingover. I n both wild-type and SZDI-l/sidl+ mutants about one in lo6 cells were Ura-. Tetrad analysis also allowed us to evaluate mitotic instability. Among the 243 SZDI-l/sidl+ tetrads we dissected, four carried three copies of chromosome IZZ, presumably as the result of a mitotic nondisjunction. In the homozygous SZDI-I/SZDI-I mutant, three such tetrads of 31 1 were noted. Similar numbers of wild-type tetrads showed no evidence of mitotic nondisjunction. Taken together, our results suggest that the SZDI-I mutation slightly decreases the fidelity of mitotic chromosome segregation. Meiotic segregationof chromosome ZZZ in a SZD11 mutant: From patch assays like that shown in Figure 2, it was apparent that SZDI-1 strains generate chromosome ZZZ disomes at frequencies approximately 25fold greater than wild-type. DM32 generateschromosome ZZZ disomes at a frequencyof 4 X 10-4 errors/ meiosis by a quantitative platingassay (see MATERIALS AND METHODS), so SZDI-I mutantstrains were expected to make an error in about one of every 100 meioses. We thus reasoned that it would be possible to examine the nature of the segregational defect in SZDI-1 mutant strains directly by dissecting tetrads, and following the segregation of the chromosome ZZZ markers. We dissected tetradsfrom SZDI-I/SZDI-l (DM165), SZDI-l/sidl+ (DM155D) and sidl+/sidl+ (DM32) diploids. The spore viability patterns of the SZDl-1 strains are strikingly different from thewild-type (Figure 3). When compared with wild-type strains, the S I D I - I / sid1+ strain produces a large excess of three, two and one spore viable tetrads at the expense of four spore viable tetrads. These spore viability patterns are consistent with a defect in meiosis I chromosome segregation. Meiosis I nondisjunction gives tetrads with two viable spores (Figure 1B) and PSSC of a single chromosome gives tetrads with three viable spores, where the disomic spore has a live sister (Figure IC). Note that multiple PSSCs and/or meiosis I nondisjunctions in a single meiosis could also give rise to tetrads with zero, one or two viable spores. T h e phenotype of the homozygous SIDI-1 mutant is even more pronounced than that of the heterozygote, showing greater proportions of two, one and zero spore viable tetrads (Figure 3). Tetrad analysis revealed that three classes of segregational errors in the SZDI-I diploids are significantly elevated over the wild-type (Table 2). Meiosis I homolog nondisjunctions and precocious separation METHODS)

427

of sister chromatids are particularly prominent. This was expected, as the screen was designed to detect meiosis I errors. Of the 243 SZDI-l/sidl+ tetrads we examined, 23 (or 9.5%) showed evidence of meiosis I homolog nondisjunction, PSSC or meiosis I1 nondisjunction. These chromosome ZZZ errors partially explain thereduction inlevels of fourspore viable tetrads in the SZDI-1 strains (wild-type, 96%four spore viable tetrads; SZDI-l/sidl+, 24% fourspore viable tetrads). The remainder of this lossof spore viability might be due to segregational errors of the remaining chromosomes. It is possible to roughly estimate the reduction in four sporeviability that would occur if all 16 chromosomes missegregate independently at the frequency measured for chromosome III (Table 2). This calculation predictsthata SZDI-1 mutant would reducethefrequency of four spore viable tetradsto 20%, a value similar to that we observed (24%). The homozygous SZDI-I mutant showed two- to threefold higher levels of PSSCthan the heterozygote and similar levels of meiosis I homolog nondisjunction. As spore death (presumably due to missegregation of other chromosomes) prevents us from detecting allof the chromosome ZZZ errors,the values presented in Table 2 are probably underestimates. This is particularly true of DM 165, the homozygous SZDI-1 mutant, because of its elevated levels of spore death. The heterozygous SZDI-I mutant also showed eleI1 nondisjunctions of the vated levelsofmeiosis marked chromosome ZZZ. A meiosis I1 nondisjunction that follows reciprocal recombination in o u r marked interval on chromosome ZZZ would be predicted to give unique tetrad classes that are distinguishable from precocious separation of sister chromatids (see Table 2 footnote). About 1% of the tetrads were of this type. This reasonably high number of recombinant meiosis I1 nondisjunctions led us to believe that there might be a large number of meiosis I1 nondisjunctions that were nonrecombinant in our marked interval on chromosome ZZZ “hidden” among the three spore viable tetrads. We tested this by screening a number of three spore viable tetrads for evidence of meiosis I1 nondisjunctions. A meiosis I1 nondisjunction yields a three spore viable tetrad containing a pair of aneuploid sister spores (Figure 1D). One sister is nullosomic and dead. The living sister is disomic. For our test of hidden meiosis I1 nondisjunctions, sisterless spores of threespore viable tetrads were mated towild-type haploids, and theresulting diploids were dissected. The presence of a pairof chromosome IZZs in the spore being tested would be revealed by the characteristic segregation pattern they would exhibit in this tetrad analysis. Of 10 three spore viable tetrads tested in this way, one showed evidence of an extra chromosome ZZZ in the sisterless spore. This

428

M. Flatters and D.Dawson 100

90 80

70 v)

U

60

T "

"

"

4 SV Tetrads

"

03 SV Tetrads

"

2 SV Tetrads

E

% 50

I-

.c

0

40

30 20 10

0

1 SV Tetrads

"

0 SV Tetrads

"

"

"

"

=Dl-l/sidl+ sidl+/sidl+

SIDl-l/SID1-1

FIGURE3.-Spore viability in wild-typeand heterozygous and homozygous SIDI-I strains. Tetrads from wild-type, SIDI-l/sidl+ and SIDIIISID1-I strains were dissected, and thenumber of viable spores in each tetrad was determined.

TABLE 2 Segregation of chromosome III in wild-type and heterozygous and homozygous SIDI-I strains Strain

Total tetrads Meiosis I homolog NDJ Total Recombinant Nonrecombinant

DM32

DM 155D

(sidl+fsidl*)

(SIDI-llsidl*)

235

243

293

0 0 0

5 4 1

ND ND

1 0 1 0

16

45

7

ND ND

DM 165 (SIDI-IISIDI-I)

6

PSSC Total Recombinant Nonrecombinant ND

Recombinant meiosis I1 NDI'

0

6 3 2

45

0

In DM32 and DM155D we were able to score only those meiosis I1 nondisjunctions in which there was a crossover in our marked interval. These events are distinguishable from a precocious separation of sister chromatids (PSSC)of a recombinant chromosome by virtue of the heterozygosity of the centromere-linked ADEl gene in these diploids, which allowed us to identify sister spores in the tetrad. It is formally possible that the events we scored as meiosis I1 nondisjunctions are in fact PSSCs in which there was a crossover between ADEl and its centromere. We think this is unlikely. We wouldexpect a maximum of approximately 5% crossing-over between ADEl and its centromere. If this is true, then the apparent meiosis I1 nondisjunctions should represent only 5% of the total PSSC events. Instead, apparent meiosis I1 nondisjunctions are more common (approximately 25% of the PSSC levels (219)). although we are limited by a relatively small sample size. Because DM165 was homozygous at ADEI, the six tetrads that could be either recombinant PSSCs or meiosis I1 nondisjunctions could not be unambiguously assigned to eitherclass and areincluded in the 45 PSSC tetrads. Of the DM155D tetrads 23, or 9.5%, showed segregation errors forchromosome 111. If all 16 chromosomes missegregate independently at this frequency, 20% of the tetrads would have experienced no error(1 - 0.095)16 = 0.2. This value (20%) is similar to the number of four spore viable tetrads observed in this cross (24%. Figure 3).

suggests thatnonrecombinant meiosis I1 nondisjuncregation are intimately related. Crossing-over and auxiliary factors are normally necessary toensure tions are occurring,although our sample size is too small to allow an accurate estimation of the frequency segregation of homologs to opposite spindle poles at at which these events occur. anaphase I. The tetrad analysis we performed on our The relationship between crossing-over and segheterozygous SIDI-I strain allowed us to ask whether regation in SZDZ-Z mutants: Recombination and segcrossovers maintained their ability to ensure disjunc-

Association Sister-Chromatid Meiotic

429 TABLE 3

A. InterVal X1: leuZ-CEN3

Crossing-over inSIDI-I heterozygous and wild-type strains Dead : UH : LyH. UT : LyT

Dead: UH : LyH. LyT : UT

Dead : LyH : UH, UT : LyT

Dead : LyH : UH. LyT : UT

Dead : UT : UH. LyT : LyH

Dead : UT : LyH, LyT : UH

Dead : LyT : UH. UT : LyH

Dead : LyT ' UH. LyH : UT

8. Interval 12: CEN3-fGK1

Dead : UH : UT, LyH : LyT

Dead : UH : UT, LyT : LyH

Dead : UT : UH. LyH : LyT

Dead : UT : UH. LyT : LyH

Dead : LyH : UH. LyT : UT

Dead : LyH : UT, LyT : UH

Dead : LyT : UH, LyH : UT

Dead : LyT : UH. UT : LyH

FIGURE4.-Possible tetrad types resulting from reciprocal recombination followed by precocious separation of sister chromatids. The version of chromosome ZZZ in each spore is indicated byits markers. Spores within each tetrad are separated by colons, and sister spores in each tetrad are underlined. (A) Tetrads resulting from crossing-over between leu2 (URA3, LYS2) and CENZZZ and subsequent PSSC. Note that the first tetrad is indistinguishable from a tetrad in which there was a PSSC inthe absenceof exchange. (B) Tetrads resulting from crossing-over between CENZZZ and PGKZ (HZS3, TRPI) and subsequent PSSC. Note that the first tetrad is indistinguishable from a tetrad in which there was a PSSC in the absence of exchange.

tion in the mutant background. Our approach was to look for recombinant chromosome ZZZs in the exceptional disomic spores. Among the exceptional tetrads were those apparently resulting from PSSC and those resulting from homolog nondisjunctions. We examined the disomic spores from each of these exceptional classes to look forevidencethat missegregated chromosomeshad crossed-over. We will describe our analysis of the PSSC tetrads here, and the homolog nondisjunction tetrads below. PSSC tetrads were of two classes: those that by tetrad analysis alone seemed to contain missegregated crossover chromosomes (the phenotypes of these are shown in Figure 4), and those that showed no evidence of containing a crossover chromosome. ThesetetradscontainoneUra+,His+,Lys+,Trp+ disomic spore, and two sporeseach with one of the parental chromosomes. We found two tetrads of the first class, and 14 tetrads of the second class. Disomic spores in this latter class could contain either a pair of parental or a pair of reciprocal recombinant chromosome ZZZs. T o distinguish between these possibilities, disomic spore colonies were grown on a-AA plates. This results in the growth of cells that have lost the LYS2 bearing chromosome ZZZ and allows a determination of the linkage of the remaining markers (see MATERIALS AND METHODS). In this way we were able to detectfive additional PSSC events involving recom-

Strain

Total tetrads Crossover tetrads leu2-PGK 1 Crossover tetrads PGK 1 -MAT

DM101

DM32 (sidZ+/sidl+)

(SIDI-l/sidl+)

137 28 (20%)

110 '13 (12%)

48 (35%)

+ l l (10%)

These values are not differentin a chi-squared contingency test (0.1 > P > 0.05). t These values are significantly different in a chi-squared contingency test ( P < 0.005).

binant chromosomes, to give a total of seven recombinant tetrads (Table 2). These results, coupled with our ability to identify sister spores by thesegregation of thecentromere linked heterozygous ADEl marker in DM155D allowed us to determine whetherit was the recombinant or the nonrecombinant chromatid that segregated to the same pole as the other homolog in these recombinant PSSC events. In six of the seven recombinant PSSC events, the recombinant chromatid segregated with the other homolog. A similar approach was used to analyze whether the chromosomes in two spore viable tetradsresulting from ameiosis I homolog nondisjunction were recombinant. Of the fivemeiosis I nondisjunction tetrads we observed,four involved recombinant homologs (Table 2). Thus, recombinant chromosomes can nondisjoin in the SZDI-1 mutant. Reciprocal recombination is slightly reduced in SZDl-1 mutants: We measured crossover frequencies in two intervals on chromosome ZZI the leu2-PGKI interval, markedby the insertions of the four markers; and thePGKI-MAT interval. Four sporeviable tetrads from aSZDI-l/sidl+ strain (DM101)and thewild-type DM32 were used todeterminethesefrequencies (Table 3). In the leu2 to P G K l interval, the wild-type strain had crossovers in 20% of all four spore viable tetrads, while in the SZDI-1 heterozygote 12% of the four spore viable tetrads showed a crossover in this interval. In the interval from P G K l to MAT the wildtype had crossovers in 35% of the four spore viable tetrads, while the SZDI-1 heterozygote showed 10% crossing-over. Together theseexperiments suggest that the SZDl-I mutation causes a two- to threefold decrease in crossover frequency. Another observation from these results is that in SZDI-1 strains crossingover in our marked interval seems to be elevated in the exceptionaltetrads (13 crossovers in 23 exceptional tetrads vs. 13 crossovers among 110 unexceptional tetrads). SZDl-I segregation defects are not chromosome ZZZ specific: Chromosome ZZZ was used as a marker in the mutant screen and in the characterization of the

and M. Dawson Flatters

430

D.

TABLE 4

TABLE 5

Segregation of chromosome VIII in heterozygous SIDI-1 and wild-typediploids

Segregation of a single artificial chromosomein heterozygous SIDI-I and wild-typediploids

Strain

Strain

DM163 DM162 (sidl+/sidl+) (SIDl-l/sidl+)

Total tetrads Meiosis I homolog NDJ PSSC' 7

332 0 1

DMl61A

472 0

Three spore viable tetrads with one Arg+ Ura+ spore were scored as PSSCs. DM163 produced three tetrads with one Arg+ Ura+ spore, allofwhich were likely the result of ectopic gene conversion. One of these was a three spore viable tetrad with three Arg' spores, another was a four spore viable tetrad with three Ura+ spores the other was a two spore viable tetrad. DM 162 produced a three spore viable tetrad with one Arg+ Ura+ spore thatwas likely the result of an ectopic conversion, as there were three Ura+ spores in the tetrad. In addition, DM 162 also had aone spore viable tetrad and a two spore viable tetrad, each of which contained an Arg+ Ura+ spore. Note that the number of PSSCs is an underestimate as the SID-l/sidl+ heterozygote exhibits low spore viability (Figure 3) and any chromosome VIII PSSC events that occur in two, one, or zero spore viable tetrads will go undetected.T o statistically compare the chromosome VIII PSSC frequency in DM163 and DM162, we did not include the two, one andzero spore viable tetrads. DM 163 showed one PSSC in 176 + 112 = 288 four and threespore viable tetrads while DM162 showed seven PSSCs in 91 146 = 237 four and three spore viable tetrads. These frequencies are significantly different in a G test of independence (Sokal and Rohlf 1969) G = 6.40, 0.025 > P > 0.01.

+

SZDl-1 phenotypes. T o test whether the SIDI-1 defect affects the behavior of other chromosomes, we examined the meiotic segregation of chromosome VZIZ homologs that weredifferentially marked at theARC4 locus. One homolog carries the ARC4 gene, while the other carries a URA3 disruption of the ARC4 gene. Thus, spores that are disomic for chromosome VIII will score as Arg+Ura+.Thesechromosomes were crossed into aSZDl-l/sidl+ strain (DM162) and a wildtype strain (DM 163). T h e results of our tetradanalysis are shown in Table 4. The SZDl-1 mutant showed an elevated level of chromosome VIII PSSC (1.4%)when compared with the level exhibited by the wild-type (0.3%). In contrast to our experiments with chromosome 111, we saw no examples of meiosis I nondisjunction of chromosome VZIZ in the SIDl-1 mutant. T h e way in which this experiment was done did notallow us to measure or detect meiosis I1 nondisjunctions. The effect of the SZDZ-2 mutation on the segregation of a single artificial chromosome: T h e phenotypes exhibited by the SID1-1 mutant areconsistent with a defect in sister-chromatid association and segregation. Alternatively, bivalent formation and segregation could be defective in the SIDI-I mutant. It is conceivable that either failure to form bivalents in prophase I, orresolve them in anaphase I, could result in elevated levels of PSSC. Univalents resulting from failed bivalents have been shown in some organisms to be extremely unstableand prone to undergo equational divisions at anaphaseI (HAWLEY1988).To

Total tetrads' PSCD tetradst

DM154 (sidl+/sidl+)

107 0

(SZDI-l/sidl+)

35 7 (20%)

In all, 168 DM154 and 310 DM16lA tetrads were dissected. Only four spore viable tetrads are reported here. The artificial chromosome shows some fluctuation in copy number. Only tetrads bearing asingle copy of the artificial chromosome (two chromatids) were included in the data shown above. To maximize our chances of correctly designating spores as sisters, we only used tetrads in which sister spore designation by both the heterozygous centromere linked ADEl gene, and by the URA3 and HIS3 genes on chromosome III were in agreement. For DM154, we eliminated a total of 50 four spore viable tetrads, and for DMl61A a total of 16 four spore viable tetrads in whicheither the copy number of the artificial chromosome was not one, or inwhich the designation of sister spores by the centromere linked markers was not in agreement. These frequencies (0 of 107 and 7 of 35) are significantly different in a G test of independence with the Yates correction (Sokal and Rohlf 1969). P < 0.025.

determine whether the SIDl-1 defectdependson bivalent formation, we tested the behavior of a partnerless artificial chromosome in a SZDI-1 mutant. Pairs of these artificial chromosomes show relatively low levels of PSSC in meiosis I, suggesting that they are reasonable substrates with which to examine the fidelity of meiotic sister-chromatid association (DAWSON, MURRAYand SZOSTAK 1986; Ross et al. 1992). T h e artificial chromosome used for our experiments is composed of a backbone of bacteriophage lambda DNA, is terminated with yeast telomeres and carries an originof replication, acentromere andyeast genes, whichallow it to be followed using tetrad analysis (DAWSON, MURRAYand SZOSTAK 1986). The artificial chromosome carries no essential genes, so missegregations do not lead to cell deathand analyses of meiotic behavior are thereforesimplified. We crossed haploids carrying asingle artificial chromosome to SIDl-1 and wild-type haploids to create the diploids DM161 and DM154. The artificial chromosome was marked with a TRPl gene. If the artificial chromosome undergoes PSSC then the Trp+ phenotype will be exhibited by nonsister spores. The results of tetrad analysis are shown in Table 5 . The artificial chromosome showed 20% PSSC in the SIDl-1 strain (seven of 35 tetrads) but noPSSC (zero of 107 tetrads) in the wild-type. Thus, the high level of PSSC seen in SIDI-1 appears to be the resultof altered sister-chromatid association, and is unlikely to require chromosome pairing and association. DISCUSSION

We havedescribedthe isolation of adominant mutation (SZDI-1) that results in high levels of both meiosis I and meiosis I1 segregational errors. Both

Association Sister-Chromatid Meiotic recombinant and nonrecombinant chromosomes are affected in the mutant background.Reciprocal recombination is slightly reduced in the SIDI-1 mutant. T h e segregation defect is not restricted to one chromosome, indicating that a global pathway has been altered. SIDl-1 affects sister-chromatid association in meiosis I: We believe that the simplest interpretation of our observations is that SIDI-I causes a defect in what might most broadly be described as sister-chromatid association. There are several results that support this. First, the SZDI-1 mutation causes high levels of PSSC. Cytological studies in tractable organisms (e.g., grasshoppers (MOENS 1987)) suggest that there are two forms of sister-chromatid association. First, there areassociations along sister-chromatidarms that persist until anaphase of meiosis I. Second, there are associations of the sister chromatids at the kinetochores that are maintained until meiosis 11, at which time the kinetochore differentiates and duplicates to allow the homologs to make a bipolar spindle attachment at the equational division. The high levelsof PSSC of chromosomes 111 and VIII seen in SIDI-1 mutants are consistent with the idea that one or both of these associations are weakened or prematurely dissolved in the SIDI-I background. Secondly, the SIDI-1 mutant showed high levels of meiosis I1 nondisjunctions of chromosome III. We favor the notion that the observed meiosis I1 nondisjunctions are the manifestations of a defect in meiosis I. Infact,anelevated level of meiosisI1 nondisjunction would be one of the predicted phenotypes of a mutant whose primary defect was in the association of sister chromatids in meiosis I. In such a mutant, sister chromatids that have lost contact with each other in meiosis I would be expected to face a variety of possible segregation fates. In some meioses thechromatids will migrate to opposite poles at meiosis I to give PSSC tetrads. In other meioses, the dissociated sister chromatids would beexpectedto move independently to the same pole at meiosis I. At meiosisI1 they will be expected to segregate independently once again. If they move to opposite poles the tetrad will appear normal. However, if they move to the same pole,they will generateatetradthat would be scored as a meiosis I1 nondisjunction. Another argument that the meiosis I1 defects we see have a meiosis I origin is that a meiosis I1 specific defect would not give rise to PSSC events.Additionally, given the similarity of meiosis I1 and mitosis, a mutant with a meiosis I1 defect might be expected to have a mitotic phenotype. We have found only a weak mitotic chromosome stability defect in S I D I - I . These data areconsistent with the model that SZDI-1 results in a loss of sister-chromatid association prior to anaphase 1. LOSSof sister-chromatid association early in pro-

43 1

phase might be expected to result in dramatic alterations in recombination behavior. However, reciprocal recombination is reduced only two- to threefold in the S I D I - I mutant. This result suggests that the SIDI-1 defect does not alter chromosome structure in a way that substantially affects the recombination pathway or, alternatively, if the defect does not manifest itself until recombination has been completed. We are intrigued by the observation that crossingover in our marked interval seems to be elevated in SIDI-1 exceptional tetrads relative to unexceptional tetrads.Ordinarily, crossing-over would notbe expected to disrupt segregation. We offer two explanations of this result. First, we speculate that in a S I D I 1 mutant, chromosome structure in the centromere region may be compromised, and that a crossover in this region may exacerbate the segregational defect imposed by S I D I - I . Alternatively, there may be some variability in the duration of prophase in the SIDI-I strain. Cells might experience an extended prophase either as a direct consequence of the SIDI-1 defect, or indirectly because the S I D I - I defect affects a process that must be completed before a meiotic checkpoint can be passed. In either case, cells experiencing a segregational problem would spend extended periods in prophase.Longerprophase would provide the opportunity to accumulate additional crossovers. Thus, exceptional tetrads would be expected to look somewhat hyper-rec. In cells without aprolonged prophase, there would be less opportunity for crossing-over and fourspore viable tetrads would thus look slightly hypo-rec. Structural defects that result in a meiosis I loss of sister-chromatid association might be imagined to do so by affecting any of a number of processes. We alluded to the associations observed between sisterchromatid arms. Defects that reduce the durabilityof these associations or cause premature kinetochore differentiation could result in the phenotype we have observed. Alternatively, improper bivalent formation might also lead to PSSCin at least two ways. First, failure to formbivalents will yield unpaired univalents that have been shown in some organisms to disjoin precociously (HAWLEY1988). Second, if the factors acting to hold homologous chromosomes togetherare extraordinarilystrong in a SIDI-1 mutant, then at anaphase I a single chromatid might be ripped away from its sister, which remains paired with the other homolog. Segregation of the separated chromatids to opposite poles at meiosis I would give PSSC tetrads. Our experiment with the single artificial chromosome allows us to address aspects of these models. The fact that univalent artificial chromosomes show elevated levels of PSSC in SIDI-I mutants disproves the model that bivalent formation is required for manifestation of the SIDI-1 defect. Additionally, our results show that univalents in SIDI-1 are inherently less stable

432

M. Flatters and D. Dawson

Sister cohesiveness may promote Meiosis

i

C.

B.

A.

I Homolog

Nondisjunction Meiosis

Separation Precocious stability bivalent

1 Homolog NondisjunctionOR Chromatids of Sister

OR

Meiosis II Nondisjunction

FIGURE5.-The role of sisterthromatid association in ensuring disjunction of recombinant chromosomes. Sisterthromatid cohesion is indicated along the lengths of chromatid arms, but may also be acting at the centromere to maintain an undifferentiated kinetochore. (A) Sister cohesiveness may promote bivalent stability, by providing a means for homologs to remain associated prior to the anaphase I division. (B) Weakened sister-chromatid association may allow bivalents to fall apart prior to anaphase I. Homologs then segregate independently of one another, resulting in meiosis I homolog nondisjunction. (C) Weakened sisterchromatid association may result in the completeseparation of a chromatid from its sister prior to anaphase I. Various assortments of the remaining homolog and lonesister can result in both meiosis I homolog nondisjunctions and precociousseparation of sister chromatids.

than in wild-type strains. Thus, the SIDI-1 segregational errors arenot solely the result of elevated levels of wild-type univalents, but must somehow be related to defective sister-chromatid association. The role of sidl+ in homolog segregation: Additional factors are required to enable crossovers to ensure disjunction. These factors somehow act to stabilize the chiasmata prior to anaphase I . T h e elevated levels of segregation errors of exchange chromosomes in the SIDI-1 mutant, and the proposed role of s i d P in sister-chromatid association, are consistent with the idea that sister-chromatid association is required to ensure the correct segregation of recombinant chromosomes. Thus, sister-chromatid association may be one of these chiasma stabilizing factors. Models illustrating this are shown in Figure 5. Sisterchromatid association along chromatid arms may act to maintain associations between recombinant homologs prior to anaphase I (Figure 5A). By this model, crossovers act to join homologs while sister-chromatid associations act to keep the chiasmata in place. If chromosomes are joined in this manner, then dissolution of sister-chromatid association would berequired for progression through anaphase. T h e ways in which defects in sister-chromatid association mightresult inPSSC or meiosis I homolog nondisjunctions are illustrated in Figure 5A and Figure 5B. Even a partial reduction of sister-chromatid association might result in the premature separation of recombinant homologs (Figure 5B). Once sepa-

rated, the homologs can segregate independently of one another atmeiosis I, leading to meiosis I homolog nondisjunctions as well as apparently normal disjunctions. Alternatively, a weakened sister chromatid association or premature kinetochoredifferentiation may allow an entire chromatid to prematurely separate from its sister (Figure 5C). This scenario might result in PSSC, meiosis I and meiosis I1 nondisjunctions, dependingonthesegregation of the precociously separated sisters relative to each other and their homologous partners. Fortuitous assortmentsof these independent entities can also result in apparently normal disjunctions. Relationship of SZDl-1 defects to those seen in other mutants: T h e SIDI-I mutant shares a number of important phenotypes with other meiotic mutants inyeast and Drosophila. These shared phenotypes may illuminate the rolethat s i d P playsin meiotic chromosome segregation. In yeast, genetic analysis has been used to identify a variety of mutant genes that confer defective meiosis I phenotypes. At this time it is unclear whether SIDII represents a novel meiotic gene, or a new allele of a previously identified gene. We are in the process of cloning the sidl+ gene. Once cloned, we will be able to test the allelic relationship ofSID1-I toother meiotic genes. However, three characterized genes exhibit mutant phenotypes especially similar to SIDI1. Mutations in REDI, MERI, and DISl show high frequencies of meiosis I segregation errors (ROCKMILL

Meiotic Sister-Chromatid Association

and FOCEL1988; ROCKMILL and ROEDER1988; ENGEBRECHT and ROEDER1989). These errors can involve recombinant chromosomes, as we have seen in the SIDI-I mutant. For M E R I and R E D l mutants, it is not known whetherthe meiosis I errors include PSSCs, meiosis I homolog nondisjunctions, or both. M E R I and R E D l mutants show pronounced reductions in meiotic crossing-over, not seen in SIDI-I mutants. T h e dominant DISl mutants, like the SIDZ-I mutant, result in high levels of meiosis I PSSC events (ROCKMILL and FOCEL1988). Both D I S l and S I D l mutants have close to wild-type levels of meiotic recombination.Inaddition,bothmutants cause the missegregation of recombinant chromosomes. D I S I - I causes a mild, and DZSI-2 a more pronounced, elevation in the level of mitotic nondisjunction. SIDI-1 displays a wider range of meiotic segregational errors than DISI-I. It has been proposed that DISI-1 may prematurely express thedisl+ product, which may be required for sister separation. This would result in the premature separation of sister chromatids. This idea would nicely explain the phenotypes exhibited by SIDI-I. If SIDI-I is a new allele of D I S I , it becomes particularly interesting both for its apparent separation of the mitotic and meiotic phenotypes, as well a s the broader rangeof meiotic errors it causes. Recessive ord mutations in Drosophila have been examined both cytologically and genetically. The genetic characterization has shown that ord is required for meiotic segregation in both males and females (MIYAZAKIand ORR-WEAVER 1992).females, In both exchange and nonexchangechromosomes are affected in ord mutants. In addition, there is a decrease in the levels of meiotic recombination in ord mutant females. Like SIDI-I mutants, ord mutants show PSSC, meiosis I1 and meiosis I homolog nondisjunctions. Cytological evidence shows that ord mutants lose sister chromatid cohesion along the lengths of chromatid arms inmeiosis I (GOLDSTEIN 1980; MIYAZAKI andORR-WEAVER1992). It has been suggested that the reductionin recombination levels seen in ord mutant females is an indirect affect of the altered structure of the sister chromatids (MIYAZAKI and ORR-WEAVER 1992), a rationale that could also be applied to SIDI-1 and DISI-I mutants. Given the similarity of the ord and SIDI-I mutants, it is tempting to speculate that their phenotypes are attributable to defects in similar structures. We thank JACK SZOSTAK, MAYAHANNA and NEILSHULTES for ORRproviding strains and plasmids, LINDAHYMAN andTERRY WEAVER for critical review of the manuscript, and members of the Dawson Lab for helpful discussions and advice. This work was supported by a National Institutes of Health grant GM40452, and a Medical Research Council of Canada Studentship grant to M.F.

433

LITERATURECITED BAKER,B., A. CARPENTER, M. ESPOSITO and L. SANDLER, 1976 The genetic control of meiosis. Annu. Rev. Genet. 1 0 53-134. DARLINGTON, C. D., 1932 Recent Advances in Cytology. The Blakiston Co., Philadelphia. DAWSON, D. S., A. W. MURRAYand J. W. SZOSTAK, 1986 An alternate pathway for meiotic chromosome segregation in yeast. Science 2 3 4 7 13-7 17. ENGEBRECHT, J., and G . S. ROEDER,1989 Yeast mer1 mutants display reduced levels of meiotic recombination. Genetics 121: 237-247. GOLDSTEIN, L. S. B., 1980 Mechanismsof chromosome orientation revealed by two meiotic mutants in Drosophilamelanogaster. Chromosoma 78: 79-1 1 1 . HAWLEY, S. R. , 1988 Exchange and chromosomal segregation in eukaryotes, pp. 497-527 in Genetic Recombination,edited by R. KUCHERLAPATIand G. R. SMITH.American Society of Microbiology, Washington, D.C. MAGUIRE,M. P., 1974 The need for a chiasma binder. J. Theor. Biol. 4 8 485-487. MAGUIRE,M., 1978a A possible role for the synaptonemal complex in chiasma maintenance. Exp. Cell Res. 112: 297-308. MAGUIRE,M. P., 1978b Evidence for separate genetic control of crossing over and chiasma maintenance in maize. Chromosoma (Berlin) 65: 173-1 83. MASON,J. M., 1976 Orientation disruptor (ord): a recombinationdefective and disjunction-defective meiotic mutant in Drosophila melanogaster. Genetics 84: 545-572. MIYAZAKI,W. Y., and T . L. ORR-WEAVER, 1992 Sister-chromatid misbehavior in Drosophila ord mutants. Genetics 132: 10471061. MOENS, P. , 1987 Introduction to meiosis, pp. 1-17 in Meiosis, edited by P. MOENS.Academic Press Inc., Orlando, Fla. MURRAY,A. W., and J. W. SZOSTAK, 1985 Chromosome segregation in mitosis and meiosis. Ann. Rev. Cell Biol. 1: 289-315. RATTNER,J. B., B. G. KINGWELL and M. J. FRITZLER, 1988 Detection of distinct structural domains within the primary constriction using autoantibodies. Chromosoma 96: 360367. ROCKMILL, B., and S. FOGEL,1988 DIS1: a yeast gene required forproper meiotic chromosome disjunction. Genetics 1 1 9 261-272. ROCKMILL, B., and G . S. ROEDER,1988 R E D l : a yeast gene required for the segregation of chromosomes during the reductional division of meiosis. Proc. Natl. Acad. Sci. USA 85: 6057-6061. Ross, L. O., D. TRECO,A. NICOLAS, J. W. SZOSTAK and D. DAWSON, 1992 Meiotic recombination on artificial chromosomes in yeast. Genetics 131: 541-550. ROTHSTEIN, R. J., 1983 One-step gene disruption in yeast. Methods Enzymol. 101: 202-21 1 . SCHULTES, N. P., and J. W. SZOSTAK, 1990 Decreasing gradients of geneconversion on both sides of the initiation site for meiotic recombination at the ARG4 locus in yeast. Genetics 126: 8 13822. SHERMAN, F., G . R. FINKand C. LAWRENCE, 1979 Methods in Yeast Genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. SOKAL,R. R., and F. J. ROHLF,1969 pp. 549-620 in Biometry, Freeman, San Francisco. TRECO, D.A. , 1989 Basic techniques ofyeast genetics, pp. 13.1.1-1 3.1.7in Current Protocols in Molecular Biology, edited by F. M. AUSUBEL, R. BRENT,R. E. KINGSTON et al. Wiley Interscience, Brooklyn. Communicating editor: F. WINSTON