Temperature-Dependent Modulation of Chromosome Segregation in msh4 Mutants of Budding Yeast

Background In many organisms, homologous chromosomes rely upon recombination-mediated linkages, termed crossovers, to promote their accurate segregation at meiosis I. In budding yeast, the evolutionarily conserved mismatch-repair paralogues, Msh4 and Msh5, promote crossover formation in conjunction with several other proteins, collectively termed the Synapsis Initiation Complex (SIC) proteins or ‘ZMM’s (Zip1-Zip2-Zip3-Zip4-Spo16, Msh4-Msh5, Mer3). zmm mutants show decreased levels of crossovers and increased chromosome missegregation, which is thought to cause decreased spore viability. Principal Findings In contrast to other ZMM mutants, msh4 and msh5 mutants show improved spore viability and chromosome segregation in response to elevated temperature (23°C versus 33°C). Crossover frequencies in the population of viable spores in msh4 and msh5 mutants are similar at both temperatures, suggesting that temperature-mediated chromosome segregation does not occur by increasing crossover frequencies. Furthermore, meiotic progression defects at elevated temperature do not select for a subpopulation of cells with improved segregation. Instead, another ZMM protein, Zip1, is important for the temperature-dependent improvement in spore viability. Conclusions Our data demonstrate interactions between genetic (zmm status) and environmental factors in determining chromosome segregation.


Introduction
During gamete production, germline cells undergo a specialized cell division (meiosis) where two consecutive nuclear divisions follow a single DNA replication event thereby reducing the chromosome number by half. To ensure that each gamete inherits an entire complement of chromosomes, organisms employ an array of different mechanisms, all of which rely on partner recognition followed by their separation ('segregation') at the first meiotic division.
Many, but not all, organisms depend upon the formation of chiasmata between the segregating partners, termed homologous chromosomes (homologs) [1]. Chiasmata have been proposed to counteract the pulling forces that are generated when kinetochores attach to microtubules emanating from opposite spindle poles, thereby facilitating the inactivation of the spindle assembly checkpoint once homologs are bioriented [2,3].
Holliday Junctions are substrates of the mismatch-repair paralogues, Msh4 and Msh5 [8]. Msh4 and Msh5 function as a heterodimer during meiosis [9] and do not contain the mismatch-repair binding domain found in other MutS paralogues [10,11]. Consistent with this, the absence of MSH4 or MSH5 influences neither mitotic postreplicative mismatch repair nor meiotic heteroduplex DNA repair, since no significant appearance of post-meiotic segregation events are observed [12,13]. Furthermore, the human MSH4-MSH5 heterodimer does not show a preference for mismatch-containing heteroduplex DNA over homoduplex DNA [8].
Msh4-Msh5 may promote crossing over by protecting crossover-specific intermediates from being resolved as noncrossovers, for example by Sgs1/BLM [14,15]. In turn, formation of Msh4-Msh5 foci on meiotic chromosomes depends upon synaptonemal complex proteins, including Zip3 and Zip1.
In the absence of any one of the zmm genes, crossover frequencies are reduced, but not abolished [16]. The decreased levels of crossovers are generally thought to cause increased levels of homolog missegregation, resulting in decreased viability of gametes, known as spores in budding yeast. Several observations support the notion that the decreased spore viability is due to increased missegregation of homologous chromosomes. First, when assessed genetically, many of the mutants display increased non-disjunction of homologous chromosomes (reviewed in [17]).
Second, partial restoration of crossover levels in zmm mutants by mutation of SGS1/BLM helicase improves segregation and spore viability [14,15]. Third, spore viability is decreased when precursors to crossovers, double-strand breaks, are decreased by mutation of Spo11, a topoisomerase II-type protein that catalyzes the DSBs [18]. Fourth, the zmm mutants analysed show an increase in the formation of non-exchange chromosomes, particular of smaller chromosomes [19,20]. This, together with the observation that artificial or sequence-diverged (homeologous) non-exchange chromosome 'pairs' in budding yeast missegregate ,10-20% of meioses (compared to ,1% for exchange pairs in wild-type cells) indicate that the generation and subsequent missegregation of non-exchange chromosome pairs contribute to spore death [21,22,23,24].
Here, we have identified temperature as a modulator of chromosome segregation in msh4 and msh5 mutant strains. We show that this occurs in different budding yeast strains, without any apparent concomitant increase in crossover levels. We identify Spo11 and Zip1 as important factors for this phenotype.

Results
Temperature modulates spore viability in msh4 and msh5 mutants zmm mutants show decreased levels of crossovers at both 23 and 33uC, with a more severe defect at 33uC [16]. Nevertheless, msh4 and msh5 mutants showed improved gamete viability at 33uC ( Figure 1A and B). This was not the case for the remaining zmm mutants (Tables 1 and 2). At 23uC, all of the zmm mutant strains showed the characteristic increase in the two-and zero-viable spore classes, consistent with meiosis I non-disjunction ( Table 1). The Y55 msh5 and the msh4 msh5 mutants displayed a more severe decrease in the proportion of four-viable spored tetrads compared to the msh4 strain at this lower temperature. Since Msh5 has previously been suggested to be important for chromosome segregation in crossoverdefective mms4 mlh1 strains [25], it is possible that Msh5 has a role independently of Msh4. If so, this is strain-specific as no differences in gamete viability between the msh4 and msh5 strains was observed in SK1 ( Table 2). Regardless of this, the temperature-dependent  Supplementary Table S1 and viable-spore class distributions in Table 1 (Y55) and Table 2 (SK1). Spore viability was measured by counting viable spores in dissected tetrads. (C) Assessment of meiosis I non-disjunction in spores containing LacO or TetO labelled chromosome IIIs and expressing LacI-GFP or TetR-GFP, respectively. More than 150 tetrads containing four distinct GFP foci were examined using standard fluorescence microscopy. For isogenic wild-type strains, the missegregation frequencies were less than 1/150 (data not shown). The asterisks indicate statistically significantly improved viability frequencies (A and B) and improved segregation (C) compared to 23u (P,0.01, G-test for homogeneity). Hatched bars represent assays at 23u and black bars at 33u. doi:10.1371/journal.pone.0007284.g001 improvement in gamete viability was observed in all mutants defective for MSH4, MSH5 or both, irrespective of strain background. Thus, our data suggest two temperature-dependent processes: Msh5 may have a strain-specific chromosome segregation function at 23uC, whereas elevated temperature (33uC) positively impacts chromosome segregation in the absence of Msh4 or Msh5. We focus on the latter phenotype.
To verify that the improved spore viability at 33uC was due to improved chromosome segregation, we assessed the segregation of chromosome III in tetrads. To this end, we generated diploid strains where LacO (Y55, BR) or TetO (SK1) repeats had been inserted near the centromere of chromosome III. These strains also express the recombinant LacI-GFP or TetR-GFP that recognize the LacO and TetO repeats, respectively. Upon the completion of meiosis, each of the four spores in the ascus should contain a single GFP focus, provided meiosis I segregation occurred normally. In contrast, a meiosis I non-disjunction event will generate an ascus where two of the spores contain two GFP signals and two spores lack GFP altogether. In all of the three strains, meiosis I non-disjunction frequencies were decreased ,3fold at 33uC compared to 23uC ( Figure 1C). Thus, improved chromosome segregation is the most likely explanation for the enhanced spore viability in the msh4 and msh5 strains at 33uC.
Crossover frequencies are not increased on chromosome III msh4 mutants are defective in crossing over [12,16,26]. Therefore, we considered whether crossover formation was different at 33uC in the msh4 mutant. Classical genetic analysis of crossing over on chromosome III ( Figure 1D) in the four-viable spored tetrads revealed similar crossover levels in the msh4 mutant at 23uC and 33uC (Table 3). These observations suggest that total levels of crossovers within the population of viable spores are unaffected by temperature in the msh4 mutant.
A non-exchange homeologous chromosome III pair displays improved chromosome segregation at 33uC compared with 23uC in the absence of MSH4 To determine unambiguously that msh4 mutants have improved chromosome segregation independently of enhanced crossover frequencies, we assessed the segregation of a homeologous chromosome pair. In this diploid strain of Y55, one of the chromosome IIIs had been replaced by the co-linear chromosome from the sibling species, S. paradoxus. The resulting sequence divergence (10-20%) [27,28] suppresses crossing over in ,90% of wild-type meioses [24]. We failed to observe a single crossover in the LEU2-MAT interval in the inspected tetrads at 23uC or 33uC in the msh4 strain (data not shown), suggesting that crossing over is not increased. The homeologous chromosome pair missegregates in 7.3% and 1.7% of meioses at 23uC and at 33uC respectively (Table 4). These observations support the hypothesis that temperature regulates segregation independently of modulating crossover levels on chromosome III.
It is unclear whether temperature improves the segregation of the homeologous chromosome pairs in the presence of MSH4, since the homeolog pair did not display any detectable temperature-dependent decrease in its non-disjunction frequency at 33uC. In the wild-type strain, the homeolog pair missegregated in 8.2% of meioses at 23uC, similar to previous observations [24], and 6.9% at 33uC (Table 4). These observations raise the possibility that the temperature-dependent modulation of chromosome segregation might be specific to msh4 mutants.

A defect in meiotic progression does not select for improved viability amongst successfully sporulated cells
Progression to meiosis I in zmm mutants in the SK1 background is more severely abrogated at 33uC than at 23uC. This is accompanied by a more severe defect in crossing over at the HIS4LEU2 hotspot at 33uC (15% of wild type) compared to  23uC (40-50% of wild type) [16]. Nevertheless, both the msh4 and msh5 mutants in SK1 displayed relatively good viability, particularly at 33uC (Table 2). It is therefore possible that preferential progression of the meiotic cells with relatively high accuracy of chromosome segregation may be selected. Although this could explain the phenotype in SK1, in Y55 and BR, sporulation frequencies were similar at the two temperatures in the msh4 strains (.45%, Table 1 and data not shown). Importantly, further decreasing the sporulation frequency of msh4 in Y55 by increasing temperature to 37 or 39uC, does not influence viability or crossover frequencies (Tables 1 and 3).
Deletion of ZIP1 reduces spore viability of the msh4 mutant at 33uC to levels observed at 23uC Homolog pairing was also not affected by temperature. Using fluorescent in situ hybridization (FISH) with probes against chromosome III as well as V in the msh4 mutant (EY137), we observed homolog pairing frequencies that were similarly high at both 23 and 33uC (.95%, data not shown), consistent with previous observations that deletion of MSH4 or MSH5 does not influence the juxtapositioning required for recombination in the Spo11-independent Cre/Lox system [29]. Nevertheless, homolog recognition mediated by meiotic recombination is presumably still required for the improved segregation in msh4 since spo11 msh4 double mutants displayed ,2% spore viability (irrespective of temperature), similar to a spo11 mutant (Table 5).
Finally, since other zmm mutants did not display the improvement in chromosome segregation at 33uC, we carried out epistasis analysis. ZIP1, which encodes the transverse element of the synaptonemal complex, is required for the temperature-mediated improvement in chromosome segregation ( Table 5). The msh4 zip1 mutant displayed similar spore viabilities at 23 and 33uC.

Discussion
The zmm mutants have similar defects in crossover frequencies, however, msh4 and msh5 display improved chromosome segregation at 33uC. At least two observations suggest that temperature modulates chromosome segregation independently of increasing crossover frequencies. First, crossover frequencies on chromosome III were similar at 23 and 33uC (Table 3). Second, missegregation of the non-exchange homeolog chromosome pair was also improved at 33uC in msh4 cells.
Although crossover frequencies are important, another consideration is the crossover position relative to the centromere, since centromere distal crossovers are less likely to facilitate biorientation of homologs, at least in mad2 mutants of budding yeast [30]. In our dataset, crossovers falling within the LEU2-MAT interval that spans CEN3 (220 kb to +100 kb) were similar at 33uC and 23uC ( Table 3), suggesting that crossovers were not being redistributed to the centromeric region at 33uC thereby promoting segregation.
The enhanced segregation at 33uC depends upon Spo11, since deleting SPO11 in the msh4 strains caused spore viability to mimic those observed in a spo11 strain (Table 5). Therefore, some aspect of recombination is presumably required for chromosome segregation in the msh4 strain.
Moreover, another ZMM protein, Zip1, is important for the temperature-mediated chromosome segregation phenotype. Although a previous study failed to observe an effect of deleting ZIP1 affecting the segregation of a non-exchange homeologous chromosome pair [31], Zip1 might be important for chromosome segregation of homologous chromosomes that fail to crossover. Mechanisms that aid the segregation of non-exchange chromosome pairs (also originally referred to as distributive segregation) in S. cerevisiae are well-documented [21,30,32]. One hypothesis would be that homolog recognition, which depends upon Spo11dependent recombination (noncrossover or even crossover), is followed by a Zip1-mediated segregation mechanism that occurs independent of its function in crossing over. Another possibility is that different recombination intermediates-that might support chromosome segregation-could be formed in the msh4 mutant compared to the zip1 mutant. Differences in the recombination phenotypes of zmm mutants have been reported [14,16].
Finally, we note that at 23uC, the zmm mutants, including msh4 and msh5, display similar reductions in spore viability. Thus, whichever mechanism(s) promotes chromosome segregation in the msh4 strain, it requires and responds to elevated temperature (33uC).

Yeast genetics and sporulation conditions
All strains are listed in Supplementary Table S1 and were verified by PCR and/or Southern blot as well as meiotic segregation. Initial screening of gamete viability in zmm mutants was carried out in SK1 on 2% KAC-COM-agar 2% KAC, 0.22% yeast extract, 0.05% D-glucose, 0.087% complete drop-out mixture (Abdullah and Borts 2001), 2% agar, pH 7.0. This medium was also used for Y55. For BR, sporulation on solid medium was carried out on 2% KAC, 0.2% yeast extract, 0.1% D-glucose, 0.1% complete amino acid mix (Bio101), and 2% agar, pH 7.0. Spore viability was determined by standard tetrad dissection.

Microscopy
Meiotic spreads and FISH analysis was carried out, as described [33]. GFP was visualised using a Deltavision IX70 system (Applied Precision) using the softWoRx software, and an Olympus Plan Apo 10061.4 numerical aperture objective lens.

Statistics
Map distances were calculated according to Perkins [34]. We used the Fisher exact test, t-test for proportions, or G-test for homogeneity, adjusting the P-values according to Dunn-Sidak or Tukey-Kramer methods when multiple comparisions where made.

Supporting Information
Table S1 a iNAT and iHPH indicate the insertion of NATMX4 and HPHMX4 cassettes (GOLDSTEIN and MCCUSKER 1999) as illustrated in Figure 1D and