Nonrandom Homolog Segregation at Meiosis I in Schizosaccharomyces pombe Mutants Lacking Recombination

Corresponding author: Gerald R. Smith, 1100 Fairview Ave. North, A1-162, P.O. Box 19024, Seattle, WA 98109-1024., gsmith{at}fhcrc.org (E-mail)

Communicating editor: M. LICHTEN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Physical connection between homologous chromosomes is normally required for their proper segregation to opposite poles at the first meiotic division (MI). This connection is generally provided by the combination of reciprocal recombination and sister-chromatid cohesion. In the absence of meiotic recombination, homologs are predicted to segregate randomly at MI. Here we demonstrate that in rec12 mutants of the fission yeast Schizosaccharomyces pombe, which are devoid of meiosis-induced recombination, homologs segregate to opposite poles at MI 63% of the time. Residual, Rec12-independent recombination appears insufficient to account for the observed nonrandom homolog segregation. Dyad asci are frequently produced by rec12 mutants. More than half of these dyad asci contain two viable homozygous-diploid spores, the products of a single reductional division. This set of phenotypes is shared by other S. pombe mutants that lack meiotic recombination, suggesting that nonrandom MI segregation and dyad formation are a general feature of meiosis in the absence of recombination and are not peculiar to rec12 mutants. Rec8, a meiosis-specific sister-chromatid cohesin, is required for the segregation phenotypes displayed by rec12 mutants. We propose that S. pombe possesses a system independent of recombination that promotes homolog segregation and discuss possible mechanisms.

IN sexually reproducing organisms two gametes fuse to form a zygote, which then develops into an adult. To maintain a constant number of chromosomes from generation to generation, the gametes must contain precisely one-half the diploid number of chromosomes. Meiosis is the specialized form of cell division that accomplishes this task (Fig 1). A diploid cell undergoes meiotic DNA replication, followed by an extended prophase during which homologous chromosomes (homologs), each consisting of two sister chromatids, must find each other, align, synapse, and recombine. At the first meiotic division (MI) homologs segregate to opposite poles, accomplishing the most critical function of meiosis, reducing by half the number of chromosomes. For this reason MI is called a reductional division. The second meiotic division (MII), unlike mitosis and MI, is not immediately preceded by a round of DNA replication. However, like mitosis, MII segregates sister chromatids to opposite poles and is called an equational division. The net result of these processes is the production of four haploid progeny from one diploid cell.

View larger version (35K): In this window In a new window Download PPT slide   Figure 1. Meiotic chromosome segregation. Meiosis consists of one round of DNA replication followed by an extended prophase. Prophase is followed by the first (reductional) meiotic division, which segregates homologous chromosomes to opposite poles, and by the second (equational) meiotic division, which segregates sister chromatids to opposite poles.

Clustering of the ends of the chromosomes or telomeres, one of the earliest events in meiosis, facilitates the initial interaction between homologs (reviewed in SCHERTHAN 2001 ; YAMAMOTO and HIRAOKA 2001 ) and, in the fission yeast Schizosaccharomyces pombe, is required for wild-type levels of pairing and meiotic recombination (SHIMANUKI et al. 1997 ; COOPER et al. 1998 ; NIMMO et al. 1998 ; YAMAMOTO et al. 1999 ). In S. pombe, meiotic replication is thought to be required for wild-type levels of meiotic recombination and for the reductional nature of MI (WATANABE et al. 2001 ). Recombination is enhanced indirectly by these early processes. However, Rec12, like its homologs in other organisms, is thought to directly promote meiotic recombination by catalyzing the formation of meiotic DNA double-strand breaks (DSBs) via a topoisomerase-like mechanism (BERGERAT et al. 1997 ; KEENEY et al. 1997 ; CERVANTES et al. 2000 ). The mutant protein Rec12(Y98F), in which a tyrosine at the presumptive catalytic site is replaced by a phenylalanine, differs from wild type only by the absence of the presumptive reactive oxygen atom. The rec12-164 allele, which encodes Rec12(Y98F), is strongly recombination deficient (CERVANTES et al. 2000 ).

Recombination is thought to be required for proper MI segregation in the following way (Fig 1): Proper segregation of homologs at MI requires that each homolog in a pair attach to spindle microtubules from opposite poles. Bipolar attachment produces tension when the poleward forces exerted by the spindle apparatus are opposed by a physical connection between homologous chromosomes. This tension stabilizes the spindle's attachment to the chromosomes, reinforcing the bipolar attachment (reviewed in NICKLAS 1997 ). In most organisms, reciprocal recombination, or crossing over, in conjunction with sister-chromatid cohesion (SCC) provides the connection between homologous chromosomes. In many organisms these connections are cytologically visible and are called chiasmata.

Meiotic chromosome segregation is initiated when tension signals the bipolar attachment of microtubules to each homolog pair. MI segregation is triggered by the release of SCC distal to the crossovers, disentangling the homologs and permitting them to segregate to opposite poles. However, centromere-proximal SCC is maintained until anaphase of MII to promote bipolar attachment of microtubules to sister chromatids. Finally, release of the remaining SCC triggers the segregation of sisters to opposite poles at MII.

In the budding yeast Saccharomyces cerevisiae, separin-dependent cleavage of Rec8, a meiosis-specific homolog of the sister-chromatid cohesin Scc1 (KLEIN et al. 1999 ; PARISI et al. 1999 ; WATANABE and NURSE 1999 ), is required for the segregation of recombined homologs at MI (BUONOMO et al. 2000 ). S. pombe Rec8 possesses a sequence that is homologous to separin cleavage sites (UHLMANN et al. 1999 ; NASMYTH 2001 ), suggesting that Rec8 cleavage may trigger the release of meiotic SCC in S. pombe also.

Meiotic chromosome missegregation results in aneuploid gametes and, eventually, aneuploid zygotes. Aneuploidy generated during meiosis is of considerable medical interest. Aneuploidy is associated with 35% of lost pregnancies in humans (reviewed in HASSOLD and HUNT 2001 ). Aneuploidy, including trisomy 21, which results in Down syndrome, is the leading known cause of mental retardation. In humans, decreased meiotic recombination is associated with trisomy 21 and other MI missegregation events (reviewed in HASSOLD and HUNT 2001 ). Understanding the role of recombination in meiotic chromosome missegregation is therefore of great importance.

Mutants deficient in meiotic recombination may display several types of segregation defects (Fig 2). In the absence of the physical connection between homologs promoted by recombination, homologs are expected to segregate randomly at MI. In this case homologous chromosomes will move to the same pole in one-half of the MI divisions (Fig 2B). The absence of recombination could also perturb sister-chromatid interactions. Precocious separation of sister chromatids (PSSC) prior to MI results in random segregation of sister chromatids at MI. The chromatids may move either to opposite poles at MI (Fig 2C) or to the same pole, where they are free to segregate randomly again at MII (Fig 2D). Additionally, PSSC after MI but prior to anaphase of MII would allow sisters to segregate randomly at MII (Fig 2D). Although not shown, homologs and/or sister chromatids may also be lost (i.e., not packaged into a spore).

View larger version (37K): In this window In a new window Download PPT slide   Figure 2. Meiotic chromosome missegregation. (A) Proper segregation of one set of homologs, as in Fig 1. (B) Meiosis I homolog missegregation. (C and D) Precocious separation of sister chromatids can result in premature segregation of sister chromatids at MI (C) or missegregation at MII (D).

While reciprocal recombination is usually required for proper MI homolog segregation, examples of recombination-independent, or achiasmate, segregation exist. Achiasmate chromosomes segregate properly during meiosis in the fruit fly Drosophila melanogaster (reviewed in HAWLEY 1989 ; HAWLEY and THEURKAUF 1993 ; MCKEE 1998 ). S. cerevisiae also has an achiasmate system capable of promoting the segregation of achiasmate artificial chromosomes (DAWSON et al. 1986 ) or two monosomic nonhomologous chromosomes (GUACCI and KABACK 1991 ). Additionally, data from the Kohli and Hiraoka labs (MOLNAR et al. 2001A , MOLNAR et al. 2001B ) and in this article indicate that a mechanism exists in S. pombe that promotes proper MI segregation in the absence of meiotic recombination.

We chose S. pombe as an experimental organism to study the role of recombination in meiotic chromosome segregation for the following reasons: as in S. cerevisiae, and most likely Arabidopsis, Caenorhabditis elegans, Coprinus, Drosophila females, and mouse (reviewed in LICHTEN 2001 ), meiotic recombination in S. pombe is initiated by the formation of DNA double-strand breaks (CERVANTES et al. 2000 ). In the absence of Rec12, which is essential for the formation of meiotic DNA double-strand breaks (CERVANTES et al. 2000 ), meiotic recombination in S. pombe is reduced by as much as 1000-fold (DEVEAUX et al. 1992 ; this article). Because S. pombe has only three chromosomes, mutants that randomly segregate homologs are expected to produce a significant number of viable progeny. This simple feature allows the genetic analysis of meiotic products characteristic of missegregation as well as the rapid identification of mutations that specifically affect segregation. Here we use both genetic and cytological assays to show that in the absence of meiotic recombination segregation of homologs to opposite poles at MI is significantly more frequent than random segregation would predict. Possible bases for this nonrandom segregation are discussed.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Yeast strains, media, and culture conditions:
The yeast strains used in this work are described in Table 1. Complete genealogies are available upon request. The cyh1-101 allele used in this work was derived in this lab by selection on plates containing cycloheximide at a final concentration of 100 µg/ml. Tight linkage (1.6 cM; data not shown) to lys1 confirmed the classification of the mutation as an allele of cyh1. Other mutations are described below or have been previously described.

Replacement of the rec12 coding sequence with 3HA-6His-kanMX was performed as follows: plasmid pFA6a-3HA-kanMX6 (BAHLER et al. 1998 ) was modified to contain six histidine codons, using the Morph kit (5'–3', Inc) and 5'-CGTTCCAGATTACGCTGCTCAGTGCCACCATCACCACCATCATTGAGGCGCGCCACTTCTAAATAAGCG-3' as a replication primer. Candidate plasmids were screened by restriction digestion and confirmed by sequencing. Plasmid pFA6a-3HA-6His-kanMX6 was then used as template for the polymerase chain reaction (PCR) that created the rec12 deletion construct. The deletion construct was used to transform S. pombe strain GP2232 (h⁹⁰ ade6-M216 leu1-32 ura4-D18 mes1::LEU2) as described (BAHLER et al. 1998 ). The resulting rec12-169::3HA6His-kanMX allele was confirmed by a PCR.

To integrate the rec12-164 allele (Y98F; CERVANTES et al. 2000 ) onto the chromosome, the rec12-171::ura4⁺ deletion of the rec12 coding sequence was created as follows: The ura4⁺-containing HindIII fragment (GRIMM et al. 1988 ) was amplified from plasmid pKS-ura4 (BAHLER et al. 1998 ), and PmeI and BglII sites were added to the ends by a PCR. The PCR product was restricted with PmeI and BglII and ligated into the PmeI and BglII sites of pFA6a-kanMX6 (BAHLER et al. 1998 ), creating pLD206. The same primers that are used with the pFA6a series of vectors to target integration of kanMX6 (BAHLER et al. 1998 ) can be used with pLD206 to target integration of ura4⁺. This plasmid was used as template in a PCR to generate the rec12-171::ura4⁺ allele using the method of BAHLER et al. 1998 . The resulting PCR product was used to transform GP2682 (h⁺ ade6-M216 lys1-37 leu1-32 ura4-D18 rec12-152::LEU2) to Ura⁺, and the transformants were screened for the loss of the LEU2 marker. Among 50 Ura⁺ transformants, 2 (GP3398 and GP3399) were Leu^-. Deletion of rec12 was confirmed by PCR. The rec12-164-containing plasmid, pJF03 (CERVANTES et al. 2000 ), was restricted with HindIII to release the rec12 gene and used to transform GP3398 to Ura^-. The Ura^- transformants were selected on supplemented EMM2 + 1 mg/ml 5-fluoroorotic acid (GRIMM et al. 1988 ). The rec12-164 (Y98F) replacement was confirmed by sequencing the entire coding sequence.

To enhance the signal from the lacO-tagged chromosome I (ChrI), a previously described variant of the green fluorescent protein (GFP)-LacI-nuclear localization signal (NLS) fusion with increased fluorescence, thermostability, and lacO binding (STRAIGHT et al. 1998 ) was adapted for S. pombe. Plasmid pAFS144 (STRAIGHT et al. 1998 ) was restricted with NotI and the ends were made blunt with the Klenow fragment of Escherichia coli DNA polymerase I. The 1.8-kb NotI-XhoI fragment containing GFP13-LacI12-NLS was ligated into the SmaI and XhoI sites of pREP41x (FORSBURG 1993 ), placing GFP13-LacI12-NLS between the nmt1 promoter and terminator and creating pLD203. Plasmid pLD203 was restricted with SphI and the ends were made blunt with T4 DNA polymerase. The nmt1p-GFP13-LacI12-NLS-nmt1t cassette was released by restriction with SacI and ligated into the NotI (made blunt with the Klenow fragment of E. coli DNA polymerase I) and SacI sites of pBluescript KS+ (Stratagene, La Jolla, CA), creating pLD207. The his7⁺ open reading frame (APOLINARIO et al. 1993 ) plus 500 bp of upstream and 212 bp of downstream sequence was amplified, and NarI and KpnI sites were added by a PCR using genomic DNA as template. The PCR product was restricted with NarI and KpnI and ligated into the ClaI and KpnI sites of pLD207, creating pLD213. The dis1⁺ promoter sequence (from -1 to -1012 relative to the ATG; NABESHIMA et al. 1995 ) was amplified and XhoI and SpeI sites were added by PCR. The sites were used to replace the nmt1 promoter with the dis1 promoter, upstream of the GFP13-LacI12-NLS coding sequence, in pLD213. This created pLD214, which was linearized with ClaI and used to transform GP3420 (h^- ade6-M210 leu1-32 lys1-37 his7-366 rec12-169::3HA6His-kanMX). His⁺ transformants were selected and correct integration confirmed by PCR.

Strains were grown on YEA + 4S [YEA (GUTZ et al. 1974 ) plus 50 µg/ml histidine, 100 µg/ml leucine, 100 µg/ml lysine, and 100 µg/ml uracil], YEA + 5S (YEA + 4S plus 100 µg/ml adenine), or supplemented EMM2 (as modified in NURSE 1975 ) solid media at 32°. When needed, cycloheximide was added to YEA + 5S to a final concentration of 100 µg/ml. Liquid cultures were grown in YEL + 5S [YEL (GUTZ et al. 1974 ) plus five supplements as in YEA + 5S] liquid media at 30°. Sporulation was at 25° on supplemented SPA (GUTZ et al. 1974 ) for 2–4 days.

Calculation of the frequency of aneuploid meiotic products:
We calculated the frequency of meiotic products that would occur at various levels of homolog missegregation at MI. For these calculations we assumed that (1) the only type of segregation error is MI homolog missegregation, (2) missegregation of homologs does not preclude microscopically visible spore formation, (3) all three homologs missegregate with equal frequency, and (4) ChrIII disomes are viable, while ChrI and ChrII aneuploids are inviable. The validity of these assumptions and the effect of deviations from them are considered in the DISCUSSION.

Let x be the frequency of homolog segregation to opposite poles, and 1 - x the frequency of missegregation. The frequencies of zero, one, two, and three homolog segregation errors can be calculated as shown in the second column of Table 2. The frequencies of one and two errors occurring must include an additional factor of three since one error can occur with any one of the three chromosomes, and two errors can occur with any of the three possible combinations of two chromosomes. Therefore, one and two errors can occur three times as often as zero and three errors.

As well as the number of errors, the exact nature of the meiotic products that result from missegregation will depend on which of the two MI poles the homologs move to. That is, either pole A or pole B can receive a pair of homologs (designated I and I', II and II', and III and III'). Therefore, a factor of eight is introduced (2³ is the no. of poles^{no. of chromosomes}) for all the classes, zero, one, two, and three errors. For instance, in the class with three errors, pole A can receive any of the following eight groupings of three chromosomes: (1) none; (2) I, I'; (3) II, II'; (4) III, III'; (5) I, I' and II, II'; (6) I, I' and III, III'; (7) II, II' and III, III'; and (8) I, I' and II, II' and III, III'. The chromosomes that occupy pole B are determined by those at A; thus, there are eight ways to accomplish three errors. The same is true of the class with zero errors and all the subclasses of one and two errors (one error involving ChrI, one error involving ChrII, etc.). The importance of the random assortment of errors to either of two poles is underscored in the example above with three errors. One of the above groupings of chromosomes [(8) I, I' and II, II' and III, III'] results in two diploid spores after MII, while all the other groupings are nullosomic for at least one chromosome and will therefore give rise to two inviable spores after MII. This final step in the calculation of the frequency of zero, one, two, and three errors occurring is shown in the third column of Table 2. The fourth column of Table 2 indicates the nature of the meiotic products that result. Using column four of Table 2, we have graphed the frequency of viable spores, heterozygous diploids, and ChrIII disomes expected at reductional segregation rates of 50% (random; x = 0.5) to 100% (x = 1.0; Fig 3).

View larger version (34K): In this window In a new window Download PPT slide   Figure 3. Expected frequency of rec12 meiotic products characteristic of MI homolog missegregation at segregation frequencies of 50–100%. The derivation of the formulas used to construct the graphs is described in MATERIALS AND METHODS. The observed frequencies of rec12 meiotic products characteristic of MI homolog missegregation (Table 5) have been plotted on the appropriate expectation curve and are indicated by open stars for rec12 and solid stars for rec12(Y98F). Observed spore viabilities have been normalized to wild type. The vertical arrow indicates the frequency of proper segregation by ChrI as observed cytologically (Table 7).

Detecting aneuploid meiotic products:
The frequency of aneuploid meiotic products was determined by random spore analysis. While random spore analysis cannot differentiate between homolog missegregation and PSSC, we chose this method because the abnormal spore morphology of rec12 mutants (see Fig 4) could bias the selection of asci for dissection and analysis of tetrads. Additionally, differential plating of random spores allowed us to maintain selection for the unstable ChrIII disomes.

View larger version (81K): In this window In a new window Download PPT slide   Figure 4. Ascus morphology and DNA distribution in rec mutants indicates the occurrence of chromosome missegregation and frequent single-division meiosis. Photomicrographs are shown combining both visible light and fluorescence images of Hoechst 33342-stained asci (A–F) or zygotes (G–I). Hoechst staining is distributed evenly between the spores in rec+ asci (A). (B and C) rec12 dyad asci containing one or two nuclei per spore, respectively. (D) rec12 ascus with unevenly distributed DNA. (E and F) Additional rec12 asci, including one with four spores containing apparently equivalent amounts of DNA. (G) rec+ mes1 zygotes. Hoechst 33342 staining is distributed evenly between the MI poles. (H and I) rec12 mes1 zygotes. Black arrows indicate lagging chromosomes. White arrowheads indicate zygotes with one pole apparently devoid of DNA.

Spores were liberated from asci, and vegetative cells were killed, by treatment with glusulase and ethanol (PONTICELLI and SMITH 1989 ). Spore suspensions were plated on supplemented EMM2 to detect I₂-staining spore colonies (mat1-P/mat1-M heterozygous diploids; BRESCH et al. 1968 ) and on YEA + 4S + 100 µg/ml guanine (YEAG), which inhibits uptake of adenine (CUMMINS and MITCHISON 1967 ), to detect adenine-prototrophic spore colonies. Both ade6-M210/ade6-M216 heterozygous diploids and heterozygous ChrIII disomes grow on YEAG. Heterozygous diploids form large colonies after 3 days, while heterozygous ChrIII disomes form small colonies only after 4 or more days. Because heterozygous diploids were detected in the Spo⁺ assay above, only the small colonies (ChrIII disomes) were counted on YEAG. Spore suspensions were plated on YEA + 4S to determine the total number of viable spores and on YEA + 5S to determine the frequency of diploid spore colonies. The cells from random spore colonies were examined microscopically. Ploidy was initially determined by cell size and confirmed by flow cytometry (data not shown). The phenotypes of diploid spore colonies were checked by streaking to YEA + 4S and, after 3 or 4 days growth, replica plating to supplemented EMM2 ± 100 µg/ml lysine and YEA + 5S with or without 100 µg/ml cycloheximide. The phenotypes of the spores dissected from rec12 dyads were determined in a similar manner. Spore viability was determined as follows: asci were suspended in sterile water and spores were liberated from asci, and vegetative cells were killed by treatment with glusulase and ethanol (PONTICELLI and SMITH 1989 ). The spore suspension was examined by light microscopy to determine the number of total spores and dilutions were plated to YEA + 4S to determine the number of viable spores.

Microscopy:
Approximately 10⁷ cells of each parent strain were mated on supplemented SPA and, after 2–3 days, the matings were collected and suspended in 500 µl of sterile water. A 100-µl aliquot of this suspension was added to 1 ml of prechilled -20° methanol. After 20–30 min at -20°, the cells were pelleted, washed once in water, resuspended in 200 µl PEMS (1.2 M sorbitol, 100 mM Pipes pH 6.9, 1 mM EGTA, 1 mM MgSO₄), and stored at 4°. To stain chromatin, 20 µl of the fixed ascus suspension was added to 1 ml of PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 2 mM KH₂PO₄ pH 7.4) and pelleted. The pellet was resuspended in 50 µl of PBS and 10 µl was placed on a microscope slide. The slide was heated for 1–2 min at 70° to provide a monolayer of cells adhered to the slide. Mounting medium (50% glycerol, 1 mg/ml p-phenylenediamine, 1 µg/ml of the fluorescent DNA-dye Hoechst 33342) was used. The same procedure was used for mes1 zygotes except in this case the cells were not fixed prior to viewing.

For detection of GFP13-LacI12-NLS-marked chromosomes, live cells were observed. Cells were mated on supplemented SPA and, after 2–3 days, zygotes were examined. By this time 95% of zygotes had undergone MI. Fluorescence microscopy was performed on a Nikon Eclipse 600 microscope using a Nikon 60x 1.4 NA Plan Apo objective (Nikon, Melville, NY). Images were captured on a CoolSNAP CCD camera (Roper Scientific, San Diego).

Recombination frequencies:
The appropriate strains were crossed on supplemented SPA and, after 2–4 days, spores free of viable vegetative cells were prepared by treatment with glusulase and ethanol (PONTICELLI and SMITH 1989 ). Spore suspensions were plated on YEA + 5S and, after 3–5 days, colonies were toothpicked to grids on YEA + 5S. After growth overnight, the segregants were replicated to the appropriate test media: EMM2 ± adenine for ade4-31, ± leucine for leu2-120, and ± uracil for ura1-61 and spc1::ura4⁺. All of the rec12 mutant recombinants were purified and retested to verify the scoring.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Expected frequency of aneuploid meiotic products in segregation-defective mutants:
Missegregation of homologous chromosomes at MI should result in spore inviability and heterozygous diploid and aneuploid spores (Fig 2). If homologs segregate randomly at MI in the absence of recombination, the spores of a rec mutant meiosis should yield a characteristic frequency of viable haploids, heterozygous diploids, and heterozygous ChrIII disomes, the only aneuploids that have been propagated in S. pombe (NIWA and YANAGIDA 1985 ; MOLNAR and SIPICZKI 1993 ). As a framework for interpretation of our segregation data, we calculated the frequency of viable spores, heterozygous diploids, and ChrIII disomes expected at reductional segregation rates of 50% (random) to 100% (Fig 3; see MATERIALS AND METHODS for derivation).

Several assumptions were made to simplify the calculations: (1) The only type of segregation error is MI homolog missegregation, (2) missegregation of homologs does not preclude microscopically visible spore formation, (3) all three homologs segregate with equal efficiency, and (4) ChrIII disomes are viable, while ChrI and ChrII aneuploids are inviable. The validity of these assumptions and the effect of deviations from them are considered in the DISCUSSION. Importantly, significant deviation from any of these assumptions should be obvious when the results of genetic and physical analysis of segregation are compared.

Spore morphology and DNA staining in recombinationless meiosis:
To begin to assess meiotic segregation in the absence of recombination, we characterized the morphology and DNA content of Hoechst 33342-stained rec⁺ and rec12 asci using differential interference contrast and fluorescence microscopy. The four spores in rec⁺ asci (tetrads) all appeared to be the same size and contained apparently equivalent amounts of DNA (Fig 4A). Many asci with morphologies not seen in wild type were frequently observed in rec12 crosses. Dyads were frequently formed in rec12 meiosis (Fig 4B and Fig C; Table 3). The spores in the dyad asci frequently contained two Hoechst 33342-staining bodies (Fig 4C). Three-spored asci containing one large spore, with two Hoechst 33342-staining bodies, and two smaller spores were also observed (Fig 4F). Other less common classes of aberrant asci were also observed in rec12 crosses (data not shown). The size of spores and the distribution of DNA were often uneven in both dyad and tetrad rec12 asci (Fig 4, B–F). This suggests that chromosomes missegregated at MI and/or MII in rec12 mutants.

To eliminate the contribution of any MII errors in DNA distribution, the mes1::LEU2 (mes1) mutation, which blocks MII (SHIMODA et al. 1985 ; KISHIDA et al. 1994 ; Table 7), was used. Both rec⁺ and rec12 strains containing the mes1 mutation were induced to mate and to undergo MI. Cells were examined as above. DNA was evenly distributed between the two MI poles in rec⁺ mes1 (Fig 4G) zygotes. Uneven distribution of DNA between the two MI poles in rec12 mes1 zygotes (Fig 4H and Fig I) indicates that chromosomes at least occasionally missegregate at MI. Additionally, lagging chromosomes were occasionally observed in rec12 mes1 zygotes (Fig 4I). The abnormalities we observed, including dyad spore formation, uneven distribution of DNA, and lagging chromosomes in rec12 meioses, are similar to those of rec7, rec14, and rec15 mutants (MOLNAR et al. 2001A , MOLNAR et al. 2001B ).

The dyads observed in rec12 meiosis frequently contain homozygous diploid spores, the products of proper reductional division:
A striking feature of rec12 meioses was the high frequency of dyad asci (Fig 4B and Fig C; Table 3). Analysis of random spore colonies indicated that rec12 meioses yielded a high frequency of diploid spores. Strains marked on all three chromosomes were used to determine the types of meiotic errors that gave rise to these diploids. Chromosomes I, II, and III were marked with lys1⁺ cyh1-101/lys1-37 cyh1⁺, mat1-P/mat1-M, and ade6-M210/ade6-M216 alleles, respectively. In wild-type strains the lys1 and ade6 alleles are centromere linked (lys1, 4 cM; ade6, 13 cM; KOHLI et al. 1977 ). In rec12 mutants the mat1 alleles were also centromere linked (2 cM; data not shown).

Both MI and MII segregation errors can give rise to diploid spore colonies. Centromere-linked markers will be heterozygous if the diploids are due to MI errors or homozygous if the diploids are due to MII errors (Fig 2). To determine which type of error occurs in rec12 meioses, the marked strains were sporulated and random spores plated on rich medium. Cells from the resulting colonies were examined. Diploid cells were easily distinguished by their size: haploid and diploid cells are 14 and 22 µm long, respectively, when cell division occurs (SVEICZER et al. 1996 ). Ploidy was confirmed by flow cytometry (data not shown). While only 3% of the rec⁺ spore colonies were diploid, >25% of rec12 spore colonies were diploid (Table 3). The marker phenotypes were determined and indicated that the rec12 diploid spores were predominantly (88%) homozygous for the centromere markers (Table 3). This indicates that these rec12 diploid spores were the result of a single reductional (MI) division. A substantial portion of the observed heterozygosity was in mixed segregation diploids. These may have resulted from a single reductional division, during which homologs missegregated, followed by mitotic chromosome loss.

Dissection of the rec12 dyad asci was performed to determine whether they were the source of the rec12 diploid spores. Of the 69 dyads dissected, 37 contained two viable spores (Table 4). The vast majority (34/37) of the two-spore viable dyads contained two homozygous diploid spores. Furthermore, 10 of 17 one-spore viable dyads contained a homozygous diploid spore, suggesting that chromosome loss occurred either during meiosis or after germination of an aneuploid product of MI missegregation. This indicates that the rec12 dyad asci were at least a major source of the rec12 diploid spores.

The dyad dissections also suggest that MI homolog segregation was not random in the rec12 meioses. Random segregation is expected to result in proper segregation of all three chromosomes in only 12.5% of meioses (0.5³ = 0.125). However, 49% (34/69) of rec12 dyads contained two homozygous diploid spores and were therefore the result of a single MI division in which all homologs segregate properly (Table 4). Together with the high spore viability of rec12 dyads (66%; Table 4), this indicates that MI homolog segregation is not random in rec12 meioses, at least not for those that form dyads.

The frequency of meiotic products characteristic of MI missegregation indicates rec12 mutants segregate homologs nonrandomly:
Although analysis of rec12 diploid spore colonies indicated that segregation was nonrandom, several one-spore viable dyads contained aneuploid spores indicative of MI homolog missegregation. As discussed above (see Fig 3), missegregation of homologs at MI should result in a predictable level of spore inviability and in the formation of heterozygous diploids and ChrIII disomic spores. Values for these parameters were determined for both rec12 and the presumptive catalytically inactive mutant rec12(Y98F) to begin to assess the degree of MI homolog missegregation in rec12 meiosis. The validity of this approach depends on the unbiased encapsulation of meiotic products into spores. Although a few zygotes did not form visible spores (data not shown), in the DISCUSSION we argue, on the basis of the entirety of our data, that encapsulation is essentially unbiased.

Random segregation of the three homologs at MI is predicted to decrease spore viability to 20.3% (Fig 3). As the abnormal spore morphology of rec12 mutants could bias the selection of asci for dissection, we chose to assay viability by determining the fraction of random spores, enumerated by microscopy, that formed visible colonies. Although there was significant variability from experiment to experiment, rec⁺ meioses resulted in 70% spore viability, most likely due to difficulty in scoring spores microscopically and/or to excess treatment with glusulase. However, as expected, rec⁺ spore viability [69.4 ± 8.6% (mean ± SEM)] was significantly higher than that of the rec12 mutants. Spore viability in rec12 and rec12(Y98F) meioses was only 14.5 ± 4.2% and 21.8 ± 1.7%, respectively (Table 5).

Random segregation at MI also predicts that 7.7 and 30.7% of total viable spores would be heterozygous diploids and ChrIII disomes, respectively (Fig 3). To reduce bias, as with spore viability, we chose to assay heterozygous diploids and ChrIII disomes among random spores rather than by tetrad dissection. Heterozygous diploid spores, which have one copy each of the mat1 alleles P and M, are sporulation proficient (Spo⁺). Only 1.3 ± 0.2% and 1.4 ± 0.2% of the viable spores produced in rec12 and rec12(Y98F) meioses, respectively, were Spo⁺ (Table 5). This is not greatly different from the 0.6 ± 0.1% Spo⁺ spores observed in rec⁺ crosses (Table 5). The intragenic complementation displayed by the ade6 alleles M210 and M216 allows missegregation of ChrIII to be assayed by formation of Ade⁺ (white) colonies. While 0.1 ± 0.01% of rec⁺ viable meiotic products were Ade⁺, 10.5 ± 1.2% of rec12 and 19.3 ± 2.0% of rec12(Y98F) viable meiotic products were heterozygous ChrIII disomes (Table 5). Both the rec12 and rec12(Y98F) heterozygous ChrIII disome frequencies were significantly <31%, the frequency predicted for random segregation at MI (² = 137, P << 0.001 and ² = 80, P << 0.001, respectively). These results are all consistent with the hypothesis that, in the absence of meiotic recombination, chromosomes segregate reductionally more frequently than random segregation predicts.

Visualization of ChrI indicates that rec12 mutants segregate homologs nonrandomly:
Genetic analysis of segregation is limited to viable meiotic products. The results above, which are consistent with nonrandom segregation of homologs at MI, could also be explained if the products of properly executed meiotic divisions are preferentially encapsulated into spores. To overcome this limitation, we used strains whose ChrI centromeres were marked with a tandem array of lacO DNA (NABESHIMA et al. 1998 ). This array can be visualized by fluorescence microscopic observation of GFP-LacI-NLS fusion proteins that bind it. The mes1 mutation was again used to simplify the analysis by blocking MII and thereby eliminating the contribution of any MII errors. None of the results described below were dependent on the mes1 mutation, as in all cases tested similar results were obtained in a mes1⁺ background (data not shown).

First, we determined whether sister chromatids separate prematurely prior to MI in rec12 mutant meioses (Fig 5A and Table 6). In crosses heterozygous for the lacO array, if the first division is reductional, both sister chromatids containing the array will segregate to one pole. One lacO-containing sister chromatid at each pole indicates either an equational first division or a PSSC. In rec⁺ mes1 crosses 98% of cells had a GFP signal at only one pole (Table 6). Similarly, in 96% of rec12 mes1 meioses the lacO arrays segregated to only one pole (Table 6). These data establish that sister chromatids do not frequently separate prior to MI in rec12 mutants.

View larger version (104K): In this window In a new window Download PPT slide   Figure 5. Chromosome I homolog segregation is nonrandom in rec12 mutant meiosis. A lacO array integrated near the centromere of chromosome I is detected by a GFP-LacI fusion protein in mes1 meioses. Photographs are of rec12 mes1 strains with (A) heterozygous lacO arrays and (B) homozygous lacO arrays. In A, black arrows indicate proper segregation of sister chromatids to one pole and white arrowheads indicate a single sister-chromatid missegregation event. In B, black arrows indicate proper segregation of homologs to opposite poles in three zygotes and white arrowheads indicate poles without a marked ChrI homolog indicative of zygotes that have undergone homolog missegregation.

To segregate properly at MII, sister chromatids must maintain cohesion at the centromere until the metaphase-to-anaphase transition of MII. Even if MI segregation is reductional, loss of cohesion prior to MII would allow the lacO arrays to diffuse away from each other after MI (BERNARD et al. 2001 ), resulting in two lacO signals at the same pole. Two distinct lacO signals were seen at the same pole 17% of the time in rec⁺ mes1 (Table 6) and 11% of the time in rec12 mes1 (Table 6). The relatively high number of zygotes with two distinct signals at one pole was likely due to the duration of the mes1 arrest. However, these data do indicate that sister-chromatid cohesion is maintained after MI in rec12 meioses as well as it is in rec⁺ meioses.

To determine the frequency of proper homolog segregation at MI, mes1 crosses homozygous for the lacO array were performed and zygotes were examined by fluorescence microscopy after 2–3 days (Fig 5B). In these crosses proper segregation will result in one lacO-containing homolog at each pole. In rec⁺ mes1 crosses the lacO arrays segregated to opposite poles in 99% of the meioses (Table 7). In rec12 mes1 crosses the lacO arrays segregated to opposite poles in 63% of the meioses (Table 7). The frequency of proper segregation in rec12 (63%) is significantly higher than the expected frequency of 50% (² = 21: P < 0.001).

Spontaneous, Rec12-independent recombination is not sufficient to account for the observed level of proper homolog segregation:
The degree of nonrandom segregation observed using the lacO array on ChrI in rec12 mutants could be explained if Rec12-independent recombination resulted in one or more crossovers on ChrI in 26% of meioses. This is equivalent to ChrI having a genetic length of 13 cM. To address this possibility, we performed random spore analysis of crosses in which we could measure recombination between spc1 (SHIOZAKI and RUSSELL 1995 ) and ade4 near the ends of ChrI. These markers are separated by 5.1 Mb (WOOD et al. 2002 ), thus covering 89% of ChrI (5.7 Mb). Although spc1 has not been placed on the genetic map, its physical proximity to cdc12 (WOOD et al. 2002 ) indicates that the genetic distance separating spc1 and ade4 in a rec⁺ meiosis would be 900 cM (MUNZ et al. 1989 ). As expected, spc1 and ade4 were unlinked in rec⁺ crosses (Table 8). In rec12 crosses the recombination frequency between spc1 and ade4 was 2.1% (= 2.2 cM, Table 8). Similar results were obtained with ura1 and leu2 markers, which cover 63% of ChrI (Table 8). Extrapolating from the spc1 to ade4 data gives a genetic length of 2.5 cM for ChrI (2.2 cM/0.89). This is significantly lower (² = 61: P << 0.001) than the 13 cM required to explain the observed degree of nonrandom segregation as being directed by Rec12-independent recombination.

The absence of recombination, rather than of some other function of Rec12, results in the observed segregation phenotypes:
It has been reported that the Rec12 homolog, Spo11, has roles in meiotic progression other than DSB formation in S. cerevisiae and Coprinus (CHA et al. 2000 ; MERINO et al. 2000 ). Additionally, localization of Spo11 along synapsed chromosomes in mouse (ROMANIENKO and CAMERINI-OTERO 2000 ) suggests a role in stabilizing homolog interactions. This raises an important question: Are the segregation phenotypes that we observe in rec12 strains due solely to the absence of recombination? The segregation phenotype is not specific to rec12 as all the strong rec mutants whose only known meiotic defect is in recombination, rec6, rec7, rec12, rec14, and rec15 (LIN et al. 1992 ; LIN and SMITH 1994 , 1995; EVANS et al. 1997 ), displayed an increased frequency of dyad asci and diploid spores and a low level of Spo⁺ diploid spores (Table 9; DEVEAUX and SMITH 1994 ; MOLNAR et al. 2001B ). This suggests that these gene products act together, perhaps in a complex, or that the absence of recombination per se results in the common phenotype.

Tyrosine-135 of Spo11 is thought to be directly involved in creating meiotic DNA double-strand breaks and presumably forms a phosphodiester link between Spo11 and DNA by a topoisomerase-like mechanism (BERGERAT et al. 1997 ; KEENEY et al. 1997 ). Spo11 (Y135F) mutant protein is not capable of promoting meiotic recombination yet appears to perform Spo11's additional roles in meiotic progression (CHA et al. 2000 ). The homologous tyrosine is replaced by a phenylalanine in Rec12(Y98F), which is encoded by the rec12-164 allele, differing from wild type only by the absence of the presumptive reactive oxygen atom. The rec12-164 allele is strongly recombination deficient (CERVANTES et al. 2000 ) and, like the other strong rec mutants (Table 9), displayed a low level of Spo⁺ diploid spores and an increased frequency of dyad asci and diploid spores (Table 5; data not shown). The rec12-164 allele (Rec12-Y98F) may result in production of an unstable protein or of a protein that does not properly form a complex. We think, however, that these data are most consistent with the notion that the absence of recombination, rather than of some other function of Rec12, results in the observed segregation phenotypes.

Rec8 is required for the segregation phenotypes displayed by rec12 mutants:
To determine whether the Rec8 cohesin protein is required for nonrandom segregation, rec8::ura4⁺ (rec8) single-mutant and rec12 rec8 double-mutant strains were investigated. These strains were marked on all three chromosomes to determine the frequency of meiotic products characteristic of MI missegregation. The frequency of heterozygous diploid (Spo⁺) spores from both rec8 and rec12 rec8 mutant crosses (Table 5; 10.3 ± 0.6% and 9.3 ± 0.4%, respectively) was consistent with random segregation, which predicts 7.7% heterozygous diploid spores (Fig 3). Interestingly, rec12 rec8 mutant crosses produced many fewer dyad asci than rec12 mutant crosses and about the same as rec⁺ (Table 3). While these data indicate that rec8 is epistatic to rec12 with regard to meiotic segregation, the frequency of meiotic products disomic for ChrIII (Ade⁺) suggests a more complicated relationship. Random segregation predicts that 30.7% of the viable meiotic products would be disomic for ChrIII (Fig 3); however, we observed 21.7 ± 2.0% in rec8 meioses and 30.0 ± 1.3% in rec12 rec8 meioses (Table 5). This difference between rec8 and rec12 rec8 is statistically significant (P < 0.01; Table 5) and indicates that Rec12 and Rec8 independently promote homolog segregation.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In most organisms, reciprocal recombination in conjunction with SCC provides the connection between homologous chromosomes required for proper MI segregation. While this appears to be generally true, mechanisms exist in many organisms that promote the proper segregation of achiasmate chromosomes. In Drosophila, which possesses chromosomes that never recombine, these mechanism are nearly 100% effective (reviewed in HAWLEY and THEURKAUF 1993 ). S. cerevisiae, in which even the shortest chromosomes only rarely fail to undergo at least one reciprocal recombination event (KABACK et al. 1989 ), can promote the proper segregation of achiasmate artificial chromosomes (DAWSON et al. 1986 ) as well as two monosomic nonhomologous chromosomes (GUACCI and KABACK 1991 ). Here, we first discuss the gross cytology of the meiotic divisions and spore formation in S. pombe recombination-deficient mutants. We then provide evidence that in the absence of meiotic recombination homologous chromosomes segregate to opposite poles more frequently than predicted by random segregation.

The dyads produced in rec12 meioses frequently contain two homozygous diploid spores:
We frequently observed that rec12 meioses produced two-spored (dyad) asci (Fig 4B and Fig C). The spores in these dyads frequently contained two nuclei (Fig 4C). After submission of this article, the presence of dyad asci containing spores with two nuclei in rec12 mutants was also reported by SHARIF et al. 2002 . Additionally, rec12 dyads frequently contain two homozygous diploid spores (Table 4); these are clearly the products of a single reductional division. Meioses in rec7, rec14, and rec15 mutants also produce dyad asci that contain spores with two nuclei (MOLNAR et al. 2001A ). Like rec12 dyads, rec7 dyads frequently contain homozygous diploid spores (MOLNAR et al. 2001B ).

The rec12 dyads are reminiscent of those formed by several cell division cycle (cdc) mutants. Cells with mutations in the mitotic initiation genes cdc2, cdc25, or cdc13 are defective in the initiation or completion of MII, frequently undergoing a single reductional meiotic division and forming dyad asci when sporulated at semipermissive temperature (NAKASEKO et al. 1984 ; NIWA and YANAGIDA 1988 ; GRALLERT and SIPICZKI 1991 ). Like rec12 dyads (Fig 4C), cdc25 dyads often contain spores that enclose two nuclei (GRALLERT and SIPICZKI 1989 ). One difference between the dyads formed in rec12 mutants and those formed in cdc2 mutants is that rec12 dyad formation requires the meiosis-specific sister-chromatid cohesin Rec8 (Table 3) while cdc2 dyad formation does not (WATANABE and NURSE 1999 ). There are at least two possible interpretations of the dependence of dyad formation on Rec8.

First, the absence of tension in rec12 meiosis may activate the spindle checkpoint, leading to the formation of dyads. Rec8 is required for dyad formation because in its absence MI segregation becomes equational (WATANABE and NURSE 1999 ). In this equational division sister chromatids segregate from each other with high fidelity, suggesting that in the absence of Rec8 SCC is maintained until MI, perhaps by Rad21. Therefore, the absence of Rec8 restores tension to the MI spindle in a rec12 mutant background, the spindle checkpoint is not activated, and dyads are not formed. This idea is supported by results in S. cerevisiae showing that the spindle checkpoint is activated during spo11 meioses (SHONN et al. 2000 ). Additionally, expression of a nondegradable form of the anaphase inhibitor Pds1 during meiosis in S. cerevisiae leads to the formation of dyad asci. The authors suggested that a fixed time exists between formation of the MI spindle and spore formation and that delaying MI results in spore formation before the completion of MII (SHONN et al. 2000 ). This fits well with the observation that in rec7 mutants the MI division lasts, on average, more than twice as long as it lasts in wild type (MOLNAR et al. 2001A ). Finally, the frequent occurrence in rec12 dyads of spores that enclose two nuclei (Fig 4C) suggests that at least in some rec12 meioses MII and spore formation may occur concurrently.

Second, in the absence of recombination Rec8 may not be properly removed from chromosomes, thus delaying or preventing the separation of sister chromatids at MII. In this scenario the incomplete penetrance of dyad formation in rec12 mutants indicates that the Rec8 remaining on the chromosomes is not always able to resist the pulling forces exerted by the spindle. These possibilities are not mutually exclusive, especially as cleavage of Rec8 may be one target of the spindle checkpoint in meiosis.

Chromosome segregation errors at MI in rec12 mutants—uneven distribution of DNA and lagging chromosomes:
The uneven distribution of DNA between the two MI poles that we observed in rec12 mes1 zygotes (Fig 4H and Fig I) is consistent with the occurrence of chromosome missegregation at MI. Uneven distribution of DNA at MI has also been observed in rec7, rec14, and rec15 mutants (MOLNAR et al. 2001A ). While this analysis does not distinguish between homolog missegregation and precocious separation of sister chromatids, using a lacO-tagged ChrI we showed that in rec12 meioses, as in rec⁺, sister chromatids segregate to the same pole at MI (Table 6). The uneven distribution of DNA is therefore likely to reflect homolog missegregation.

Lagging chromosomes were occasionally observed in rec12 mes1 zygotes (Fig 4I). Lagging chromosomes have also been observed in rec7, rec14, and rec15 mutants where chromosomes were observed to move back and forth between the poles during MI (MOLNAR et al. 2001A ). Importantly, lagging chromosomes have been observed in many S. pombe mutants that perturb mitotic chromosome segregation (PIDOUX et al. 2000 and references therein; GARCIA et al. 2002 ; RAJAGOPALAN and BALASUBRAMANIAN 2002 ). At least in some of these mutants the lagging chromosomes also move back and forth between the poles (PIDOUX et al. 2000 ). This is thought to be the result of either an unstable connection between chromosomes and spindles or the attachment of chromosomes to spindles from both poles. While this could also be the cause of the lagging chromosomes seen in the absence of recombination, it is possible, as proposed by MOLNAR et al. 2001A , that these chromosome movements represent an active part of a mechanism for achiasmate segregation.

The expected frequency of aneuploid meiotic products as a framework for the interpretation of both genetic and lacO segregation data:
We have calculated the frequency of viable spores, heterozygous diploids, and ChrIII disomes expected from reductional segregation rates of 50% (random) to 100% (Fig 3). The calculations used to derive Fig 3 were based on four assumptions:

1. The only type of segregation error is MI homolog missegregation. Sister-chromatid cohesion appears normal in rec12 mutants (NABESHIMA et al. 2001 ). Our data support this conclusion; the frequency of precocious separation of ChrI sister chromatids in rec12 mutants was not significantly different from that in rec⁺ (Table 6). As we mentioned above, two-spored (dyad) asci were common (Fig 4B and Fig C; Table 3). These frequently were the products of a single reductional division (Table 4). The frequent occurrence in these dyads of spores containing two nuclei (Fig 4C) suggests that the coordination of MII and spore formation may be perturbed in rec12 meiosis. Although only half as many spores are produced in a dyad, the types and proportions of aneuploid products produced by MI missegregation will be preserved, in this case with twice as much DNA. As long as dyad formation is random and not biased by the presence or absence of any MI missegregation events, and to the extent that a 2n + 2 aneuploid behaves like a 1n + 1 and a 4n tetraploid behaves like a 2n diploid, our calculations remain valid. Additionally, lagging chromosomes were occasionally observed in rec12 mes1 meioses (Fig 4I). This could result in meiotic chromosome loss and decrease spore viability accordingly. However, as long as chromosome loss is random and not biased by the presence or absence of any other MI missegregation event, the calculated frequencies of heterozygous diploids and ChrIII disomes among viable spores remain valid.

2. Missegregation of homologs does not preclude microscopically visible spore formation. We infer from the Hoechst 33342 staining of asci (Fig 4) that visible spores are formed that contain only one or two chromosomes. If MI homolog segregation were completely random, only 1.6% (= 2^-6, the reciprocal of the number of ways that three pairs of objects can be arranged in two groups) of MII spindle poles would be expected to be nullosomic for all three chromosomes. Therefore, even if completely nullosomic MII spindle poles do not form spores, it is likely an insignificant source of error. Additionally, although a few zygotes did not form visible spores (data not shown), we think that encapsulation of meiotic products into spores is essentially unbiased by the presence or absence of MI missegregation events. Absence of bias is suggested by the internal consistency of the lacO segregation data (Table 7), which are not dependent on encapsulation into spores, the calculations of aneuploid frequencies (Fig 3), and the corresponding genetic data (Table 5).

3. All three homologs missegregate with equal frequency. As our chromosome-specific cytological examination of segregation included only ChrI, we have no direct support for this assumption. However, the genetic data (Table 3 Table 4 Table 5) are consistent with the reductional segregation of all three chromosomes occurring more frequently than random segregation predicts.

4. ChrIII disomes (1n + 1) are viable, while ChrI and ChrII disomes are inviable. ChrIII disomes can be propagated but are unstable (NIWA and YANAGIDA 1985 ). Consequently, we might underestimate their frequency. Additionally, although ChrI and ChrII disomes cannot be propagated, they occasionally germinate and can infrequently lose the extra chromosome and resolve to euploidy (NIWA and YANAGIDA 1985 ; MOLNAR et al. 2001B ). The occasional germination and loss of the extra chromosome in ChrII and ChrIII disomes would result in a euploid spore colony from a meiotic product that we assumed would be inviable.

These possibilities seem not to contribute significantly to the frequency of heterozygous ChrIII disomes or to spore viability for the following reasons: rec8 mutant meioses, which consist of an equational MI followed by a random MII (WATANABE and NURSE 1999 ), should yield the same frequency and classes of aneuploids as we calculated for a random MI followed by an equational MII. Our data for rec8 and rec8 rec12 (Table 5) are consistent with this notion. These data also suggest that we were able to efficiently detect unstable heterozygous ChrIII disomes: random MI segregation predicts 30.7% ChrIII disomes (Fig 3), whereas 21.7 ± 2.0% were observed in rec8 meioses and 30.0 ± 1.3% in rec8 rec12 meioses (Table 5). The data also suggest that any viability of ChrI and ChrII disomes does not contribute significantly to spore viability: random MI segregation predicts 20.3% viability (Fig 3), whereas spore viability in rec8 and rec8 rec12 meioses, normalized to wild type, was 21 and 24%, respectively (Table 5). These considerations strongly suggest that, despite deviations from these assumptions, our calculations remain valid. Therefore, the differences between the observed values for rec12 mutants and the values predicted for random segregation indicate MI homolog segregation is nonrandom in rec12 mutants.

Genetic analysis of meiotic segregation in the absence of recombination:
To determine whether, in the absence of recombination, homologs segregate randomly at MI, the observed values for rec12 and rec12(Y98F) mutant meioses from Table 5 were plotted onto the corresponding expectation curves (Fig 3). The observed values were most consistent with nonrandom segregation of homologs at MI in the absence of recombination. While random segregation predicts that 7.7% of viable spores will be heterozygous diploids, only 1.3 ± 0.2% and 1.4 ± 0.2% of the viable spores produced in rec12 and rec12(Y98F) meioses, respectively, were heterozygous diploids (Spo⁺, Table 5). Additionally, while random segregation predicts that 30.7% of viable spores will be heterozygous ChrIII disomes, 10.5 ± 1.2% of rec12 and 19.3 ± 2.0% of rec12(Y98F) viable meiotic products were heterozygous ChrIII disomes (Table 5). Finally, random segregation predicts that 20.3% of spores will be viable. Spore viability in rec12 and rec12(Y98F) meioses, normalized to wild type, was 21 and 31%, respectively (Table 5). All of these values, except rec12 spore viability, are consistent with nonrandom segregation at MI in the absence of recombination.

Our data also suggest that Rec12 and Rec8 independently promote homolog segregation. Rec8, a meiosis-specific homolog of the sister-chromatid cohesin Scc1 (PARISI et al. 1999 ), promotes sister-chromatid cohesion (MOLNAR et al. 1995 ). In the absence of Rec8, the first meiotic division is equational rather than reductional (WATANABE and NURSE 1999 ). This, together with protein localization data, led to the proposal that Rec8 orients kinetechores in such a way that sister chromatids move to the same pole at MI (i.e., homologs segregate; WATANABE and NURSE 1999 ). rec12 rec8 meioses produced significantly more heterozygous ChrIII disomic spores than rec12 or rec8 meioses (P < 0.01; Table 5), indicating that Rec