Yeast strains, media, and culture conditions:

Strains were grown at 32° on yeast extract agar (YEA) plus histidine (50 μg/ml), leucine (100 μg/ml), lysine (100 μg/ml), and uracil (100 μg/ml) (YEA + 4S; Gutz et al. 1974); YEA + 4S plus adenine (100 μg/ml) (YEA + 5S); or supplemented Edinburgh minimal medium 2 (EMM2 as modified in Nurse 1975) solid media. Liquid cultures were grown at 30° in yeast extract liquid + 5S [YEL (Gutz et al. 1974) plus five supplements as in YEA + 5S]. Sporulation was at 25° on supplemented sporulation agar (SPA; Gutz et al. 1974) for 2–4 days. The yeast strains and mutant alleles used are described below or in references in Table 1.

A complete replacement of the dhc1 coding sequence with 3HA-6His-kanMX was performed as follows. Plasmid pFA6a-3HA-6His-kanMX6 (Davis and Smith 2003) was used as template in a polymerase chain reaction (PCR) to generate the dhc1-D126::kanMX6 allele using the method of Bahler et al. (1998). The forward and reverse primers in this reaction contained homology to the 5′ and 3′ ends of dhc1⁺ corresponding, respectively, to nucleotides 7022–7101 of cosmid c30C2 (GenBank accession no. AL355652) and 12826–12905 of cosmid c1093 (GenBank accession no. AL132839). The resulting PCR product was used to transform S. pombe strain GP3845 (h⁻ ade6-M216 ura4-D18 dhc1-D2::ura4⁺), which contained a ura4⁺ insertion in the dhc1 gene. The resulting G418-resistant transformants were screened for the loss of the ura4⁺ marker. Deletion of dhc1 was confirmed by a PCR.

Detecting aneuploid meiotic products:

The frequency of aneuploid meiotic products was determined by random spore analysis. Spores were liberated from asci, and vegetative cells killed, by treatment with glusulase and ethanol (Ponticelli and Smith 1989). Spore suspensions were plated on YEA + 5S and replicated to supplemented SPA after 3–4 days to detect I₂-staining spore colonies (mat1-P/mat1-M heterozygous diploids; Bresch et al. 1968). Spore suspensions were also plated on YEA + 4S + guanine (100 μg/ml) (YEAG), which inhibits uptake of adenine (Cummins and Mitchison 1967), to detect adenine-prototrophic spores and on YEA + 4S to determine the total number of viable spores. Both ade6-M210/ade6-M216 heterozygous diploids and heterozygous chromosome III (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. Only the small colonies (ChrIII disomes) were counted on YEAG.

Microscopy:

For detection of lacO-tagged chromosomes in live cells we used a previously described variant of the green fluorescent protein–LacI–nuclear localization signal fusion (GFP13–LacI12–NLS; Straight et al. 1998), adapted for S. pombe (Davis and Smith 2003). Cells were mated on supplemented SPA and, after 18–24 hr, zygotes were examined. By this time, many zygotes had undergone both MI and MII. Only those zygotes with four nuclei, as detected by background GFP13–LacI12–NLS fluorescence, were counted. In crosses homozygous for the lacO array, zygotes with segregation patterns consistent with premature segregation of sister chromatids at MI were eliminated from the analysis. Stacks of images from 24 focal planes separated by 0.2 μm were captured using SoftWoRx software and a DeltaVision microscope system (Applied Precision), which included an Olympus IX70 microscope and an Olympus UPlan Apo 100× 1.35 NA objective.

Visualization of ChrI segregation indicates a role for both Dhc1 and Dlc1 in achiasmate segregation of homologs at MI:

To determine the frequency of homolog segregation at MI, we used strains in which chromosome I (ChrI) was marked near the centromere with a tandem array of lacO DNA (Nabeshima et al. 1998). This array can be visualized by fluorescence microscopic observation of the GFP–LacI–NLS fusion protein that binds it (Straight et al. 1996). We performed crosses homozygous for the lacO array in both a rec⁺ and rec12Δ background and examined zygotes that had undergone MII. Faithful MI homolog segregation would result in one lacO-containing chromatid in each of the four nuclei. As the nuclei in S. pombe asci are ordered (Kitajima et al. 2003), missegregation of a homolog at MI would result in four lacO-containing chromatids in one pair of sister nuclei (the two nuclei at one end of a zygote) and none in the other pair (see Figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.—

Characteristic patterns of GFP–LacI–NLS staining in strains heterozygous (a) and homozygous (b) for a lacO-tagged ChrI. Solid circles represent the lacO array. Open circles represent the centromere of ChrI. See results for description. Not all missegregation patterns are illustrated.

In a rec⁺ background, the dlc1Δ mutant showed high levels of proper homolog segregation (99%, Table 2) that were not significantly different from wild type (χ² = 0.20; P > 0.5). In contrast, the frequency of proper homolog segregation in the dhc1Δ mutant (91%, Table 2) was reduced relative to that in wild type (χ² = 12.9; P < 0.0005). The frequency of proper segregation in a rec12Δ background, 66%, was reduced significantly by both the dlc1Δ and dhc1Δ mutations (52 and 53%, Table 2; χ² = 8.32, P < 0.005; χ² = 6.07, P < 0.02, respectively). Segregation was not significantly different from random, 50%, in either double mutant (χ² = 0.10, P > 0.7 and χ² = 0.22, P > 0.5, respectively). These data indicate that, while Dlc1 played a role preferentially in achiasmate segregation, Dhc1 was required for chromosome segregation in both the presence and the absence of recombination.

We occasionally observed segregation patterns consistent with premature segregation of sister chromatids (PSSC) in MI and also missegregation of sister chromatids at MII (data not shown). In these crosses, which were homozygous for the lacO array, it was not always possible to distinguish these two segregation patterns. To more precisely determine the frequency of other types of segregation errors, we examined the behavior of sister chromatids in crosses heterozygous for the lacO array. The results of the heterozygous crosses (below) were consistent with the observed frequency of PSSC and MII missegregation in homozygous crosses.

Visualization of ChrI segregation indicates that the absence of Rec12, Dlc1, or Dhc1 modestly affects the cosegregation of sister chromatids at MI:

MI of wild-type meiosis is reductional, that is, sister chromatids move together to the same pole and homologs move to opposite poles. To determine whether sister chromatids move together to the same pole at MI (cosegregate) in dlc1Δ, dhc1Δ, and rec12Δ mutant meioses, we performed crosses heterozygous for the lacO array. We examined zygotes that had undergone MII and determined whether or not lacO-containing sister chromatids were in sister nuclei. If the first division is reductional-like, both sister chromatids containing the array will be in one pair of sister nuclei and none in the other pair. In contrast, one lacO-containing sister chromatid in each pair of sister nuclei indicates an equational first division (see Figure 1). In wild-type crosses, lacO cosegregated in 87% of MI divisions (Table 3). The lacO array is integrated at the lys1 gene, ∼4 cM from the centromere in wild-type meiosis (Watanabe and Nurse 1999). The frequency of apparent sister chromatid segregation at MI in wild type can be explained by recombination between the centromere and the lacO array and is consistent with previously published results (Watanabe and Nurse 1999). In rec12Δ, dlc1Δ, dhc1Δ, rec12Δ dlc1Δ, and rec12Δ dhc1Δ mutants, sister chromatids cosegregated at MI at least 92% of the time (Table 3). Because meiotic recombination is reduced or abolished in these mutants (DeVeaux et al. 1992; Yamamoto et al. 1999; Miki et al. 2002; Davis and Smith 2003), recombination between the centromere and the lacO array is unlikely to significantly influence these results. These data indicate that Rec12, Dlc1, and Dhc1 play a small but significant role in ensuring that sister chromatids cosegregate at MI.

We were also able to detect MII errors. If sister chromatids cosegregate at MI but missegregate at MII, both lacO-containing sister chromatids would be contained in a single nucleus (see Figure 1). While sister chromatid segregation at MII in dlc1Δ or dhc1Δ single mutants was not different from that in wild type, the frequency of MII missegregation in rec12Δ and rec12Δ dhc1Δ mutants was 8 and 9%, respectively (Table 3). These data indicate that, although none of the mutants tested were severely impaired, sister chromatid missegregation at MII was elevated in rec12Δ relative to wild type (χ² = 5.85, P < 0.02). A similar conclusion, based on genetic assays of segregation, was reached by Sharif et al. (2002).

Genetic assays of segregation suggest a role for Dhc1 in the presence of recombination:

In the first meiotic division, homologous chromosomes segregate to opposite poles. In the absence of recombination between a heterozygous marker and the centromere, this reductional segregation eliminates heterozygosity in the products of meiosis. However, if homologous chromosomes missegregate at MI (i.e., if both homologs remain together), heterozygosity is retained. Because S. pombe has only three chromosomes, even mutants that segregate homologs at random are expected to produce a significant number of viable progeny. Nonetheless, if homologs missegregate at MI, an elevated frequency of heterozygous meiotic products is expected (Figure 2 and Davis and Smith 2003). These meiotic products include heterozygous diploids and heterozygous ChrIII disomes, the only aneuploids that have been propagated in S. pombe (Niwa and Yanagida 1985; Molnar and Sipiczki 1993). These features of S. pombe biology allowed us to characterize meiotic segregation by genetic analysis of the products of meiosis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.—

Expected frequency of heterozygous meiotic products at homolog segregation frequencies of 50–100%. The formulas used to construct the graph have been described (Davis and Smith 2003). The observed frequencies of heterozygous spore colonies characteristic of MI homolog missegregation (Table 4) are plotted on the appropriate expectation curve and are indicated by solid circles (rec12Δ), open circles (rec12Δ dlc1Δ), and shaded circles (rec12Δ dhc1Δ). The solid (rec12Δ), open (rec12Δ dlc1Δ), and shaded (rec12Δ dhc1Δ) arrows indicate the frequency of proper segregation by ChrI as observed cytologically (Table 2).

Mutations that specifically abolish achiasmate segregation by definition would not alter segregation nor, therefore, the frequency of heterozygous meiotic products, in the presence of meiotic recombination. To determine if this was true of dhc1Δ and dlc1Δ mutants, we analyzed their segregation phenotype in a rec12⁺ (chiasmate) background. Crosses heterozygous for the complementing ade6 alleles, M210 and M216, allowed missegregation of ChrIII to be assayed by formation of Ade⁺ colonies (Moreno et al. 1991). Because formation of a heterozygous ChrIII disome requires only one missegregation event (that of ChrIII), the frequency of heterozygous ChrIII disomes among viable spores increases linearly as the frequency of MI missegregation increases (Figure 2 and Davis and Smith 2003). The frequency of heterozygous ChrIII disomes is predicted to increase from 0.06 ± 0.01% of viable spores, as observed in wild-type meioses (Table 4), to 30.7% of viable spores if homolog segregation at MI is random (Figure 2 and Davis and Smith 2003), an increase of 500-fold. The observed ChrIII disome frequencies of dlc1Δ mutant crosses were 1.4 ± 0.1%, ∼25 times higher than that of wild type (Table 4). The dhc1Δ mutant was more severely affected, with a ChrIII disome frequency (9.1 ± 0.7%; Table 4) that, although >150 times higher than that of wild type, was lower than that of the rec12Δ mutant (23 ± 1.7%; Table 4). These data suggest that Dhc1 plays a significant role in MI segregation even in the presence of recombination.

Formation of a heterozygous diploid requires three missegregation events (those of ChrI, II, and III). Therefore, the frequency of heterozygous diploids among viable spores does not increase markedly until the frequency of MI missegregation is >∼30% (Figure 2 and Davis and Smith 2003). The frequencies of heterozygous diploids (assayed as Spo⁺, resulting from mat1-P/mat1-M heterozygosity on ChrII) among viable spores in the dhc1Δ and dlc1Δ mutant crosses was low (Table 4), consistent with ChrIII disome results.

Genetic assays of segregation suggest a role for Dlc1 specific to achiasmate segregation:

Mutations that specifically abolish achiasmate segregation should result in an increased frequency of heterozygous meiotic products only when combined with a mutation, such as rec12Δ, that eliminates meiotic recombination. To determine if dhc1Δ and dlc1Δ had a more dramatic effect on MI segregation in the absence of recombination, we analyzed their segregation phenotype in a rec12Δ (achiasmate) background. As above, crosses were performed heterozygous for the complementing ade6 alleles, M210 and M216. In a rec12Δ background the formation of heterozygous ChrIII disomes among viable spores is not predicted to be very sensitive to increased missegregation because the missegregation frequency is already high (Figure 2 and Davis and Smith 2003). Accordingly, neither the rec12Δ dhc1Δ nor the rec12Δ dlc1Δ crosses had a frequency of heterozygous ChrIII disomes significantly higher than that in the rec12Δ single mutant (Table 4).

In contrast, the formation of heterozygous diploid spores in a rec12Δ background is predicted to be very sensitive to increased missegregation. The frequency of heterozygous diploids is predicted to increase from 1.1 ± 0.2% of viable spores, as observed in rec12Δ meioses (Table 4), to 7.7% of viable spores if homolog segregation at MI is random (Figure 2 and Davis and Smith 2003). In a rec12Δ mutant background dhc1Δ mutant crosses produced no more heterozygous diploid meiotic products than did the rec12Δ single-mutant cross (Table 4). The dlc1Δ rec12Δ double-mutant crosses resulted in 8.1 ± 0.7% heterozygous diploid meiotic products (Table 4), consistent with random segregation of homologs at MI (7.7% heterozygous diploids predicted; Figure 2). Together with the data from rec12⁺ meioses, these data suggest that Dlc1 is required preferentially for achiasmate segregation. Interestingly, while dlc1Δ rec12Δ double mutants yielded elevated frequencies of heterozygous diploid spores, the dhc1Δ rec12Δ crosses did not result in more heterozygous diploid spores than the rec12Δ crosses (1.5 ± 0.3% and 1.1 ± 0.2%, respectively; Table 4). The discrepancy between the genetic results, which indicate no significant role for Dhc1 in achiasmate segregation, and the cytological results is discussed below.

The dynein light chain, Dlc1, is required for achiasmate segregation:

Analysis of dlc1Δ meioses indicated that chromosome missegregation was only modestly elevated by the absence of Dlc1 in rec⁺ (chiasmate) meioses. The frequency of ChrIII disomes (Table 4) observed in dlc1Δ crosses was consistent with a homolog missegregation frequency of ∼3% (Figure 2). The frequency of missegregation in dlc1Δ measured by direct observation of lacO-tagged ChrI, 1% (Table 2), was not significantly different from that expectation (χ² = 0.81; P > 0.3). Miki et al. (2002) measured crossovers in dlc1Δ mutants in five nonoverlapping intervals that constitute ∼4% of the genome; recombination frequencies were reduced <10-fold relative to wild type. If recombination were decreased by a factor of five uniformly throughout the genome, the mean number of crossovers per homolog would be 3.8, 3, and 2.2 for chromosomes I, II, and III, respectively (Munz 1994). Since S. pombe lacks crossover interference (Munz 1994), the frequency of achiasmate ChrI, -II, and -III would be expected to be the corresponding Poisson null terms of 0.02, 0.05, and 0.11, respectively. Missegregation of the infrequent achiasmate chromosomes thus appears to be sufficient to explain the segregation defect we observed in dlc1Δ (rec⁺) meioses. Our data indicate that when crossovers are present, Dlc1 plays at most a modest role in faithful chromosome segregation at MI. In contrast, dlc1Δ had a more severe effect in a rec12Δ background. Direct observation of lacO-tagged ChrI showed that segregation of homologs was reduced from 66% in rec12Δ to 52% in dlc1Δ rec12Δ, i.e., nearly random (Table 2). The frequency of proper segregation in dlc1Δ rec12Δ crosses was significantly different from that of rec12Δ (χ² = 8.32; P < 0.005) but was not significantly different from random (χ² = 0.10; P > 0.7). Our genetic results also are consistent with completely random segregation of homologs at MI; the level of heterozygous diploid meiotic products seen in dlc1Δ rec12Δ meioses (8.1%, Table 4) was not significantly different from that predicted for random segregation (7.7%; χ² = 0.08, P > 0.7). Our data indicate that Dlc1 is required preferentially for achiasmate segregation.

There are two ways to think about the role of Dlc1 in achiasmate segregation. Dlc1 may play a novel role in achiasmate segregation that is distinct from its minor role in wild-type meiosis. For instance, in rec12Δ, but not rec12⁺, meioses, Dlc1 may be required for homologous association of centromeres (discussed below). Alternatively, Dlc1 may play the same role as in wild-type meiotic chromosome segregation but this role becomes crucial only in the absence of recombination. For instance, the aberrant horsetail movement observed in the absence of Dlc1 (Miki et al. 2002) may not perturb homologous interactions that are stabilized by crossovers (rec12⁺), but it may severely perturb homologous interactions that are not stabilized by crossovers (rec12Δ).

Cytological data indicate that the dynein heavy chain, Dhc1, is required for achiasmate segregation:

Analysis of dhc1Δ meioses indicates that horsetail movement or some other function of Dhc1 is required for faithful MI homolog segregation in rec⁺ meiosis. The observed ChrIII disome frequency (Table 4), our most sensitive genetic assay, is consistent with a homolog missegregation frequency of ∼15% (Figure 2). The frequency of missegregation in dhc1Δ measured by direct observation of lacO-tagged ChrI, 9% (Table 2), is not significantly different from that expectation (χ² = 2.21; P > 0.10). The level of missegregation we observed in dhc1Δ meioses cannot be explained by the modest decrease in recombination observed by others in dhc1 mutants (Yamamoto et al. 1999; see above). In fact, meiotic recombination is reduced less in the dhc1Δ mutant than in the dlc1Δ mutant in all three genetic intervals tested in both mutants (Yamamoto et al. 1999; Miki et al. 2002). The level of missegregation that we observe in the dhc1Δ mutant suggests that, in addition to its essential role in horsetail movement (Yamamoto et al. 1999), Dhc1 may be directly involved in the segregation of homologs on the MI spindle. Alternatively, the recombination events in dhc1Δ meioses may not efficiently hold homologs together. Strikingly, while direct observation of lacO-tagged ChrI indicated that Dhc1 was required for achiasmate segregation, the genetic assays did not suggest a role for Dhc1 in achiasmate segregation (Table 2). In the genetic assay, heterozygous diploid formation in the dhc1Δ rec12Δ mutant was not significantly different from that in the rec12Δ mutant (Table 4). This is in contrast to direct observation using lacO-tagged ChrI in which proper segregation of homologs was reduced from 66% in rec12Δ to 53% in dhc1Δ rec12Δ, a significant difference (Table 2; χ² = 6.07, P < 0.02). Segregation in dhc1Δ rec12Δ was not significantly different from random (χ² = 0.22; P > 0.5). Because, unlike the genetic assays, observation of lacO-tagged ChrI is a direct measure of homolog segregation, we consider the cytological data to indicate an important role for Dhc1 in achiasmate segregation.

A likely cause of the disparity between our genetic and cytological results for the dhc1Δ rec12Δ mutant (as well as the failure to measure an increase, relative to rec12Δ, in ChrIII disome formation in dhc1Δ rec12Δ and dlc1Δ rec12Δ mutants) is the indirect nature of the genetic assays. While genetic analysis can easily generate large data sets, it is limited to viable meiotic products. Additionally, the genetic assays (heterozygous diploid and ChrIII disomic spore formation; Figure 2, Table 4) are interpretable only if several assumptions are made (see Davis and Smith 2003). Two of these assumptions are particularly relevant:

1. The only type of segregation error is MI homolog missegregation. In the case of the mutants tested in this article by direct observation using lacO-tagged ChrI, we know that this is not strictly true. Sister chromatids occasionally segregated from each other at MI in the mutants tested, and rec12Δ mutants occasionally missegregated sister chromatids at MII (Table 3).

2. Heterozygous diploids and ChrIII disomes are as likely to form a colony as a haploid cell. This is most likely to be a problem for mutants with a mitotic phenotype. Several mutants of S. pombe are viable as haploids but are not stable as diploids (Bernard et al. 1998; Catlett and Forsburg 2003). Both Dhc1 and Dlc1 play a role in attachment of chromosomes to the mitotic spindle (referenced in Yamamoto and Hiraoka 2003). Perhaps dhc1Δ diploids (as well as dhc1Δ and dlc1Δ ChrIII disomes) are less stable than wild type. While it is difficult to predict the degree to which a deviation from these assumptions might influence the results of the genetic assays, these caveats complicate interpretation of the genetic assays and may explain the difference between the dhc1Δ rec12Δ genetic and cytological results.

Because of the ease of assaying many meioses with the genetic assays, we tested other likely candidates for missegregation in the presence and absence of Rec12. These candidates include Taz1 (Cooper et al. 1998; Nimmo et al. 1998; Ding et al. 2004), Kms1 (Shimanuki et al. 1997; Niwa et al. 2000), Meu13 (Nabeshima et al. 2001), Hopl (Lorenz et al. 2004), and Mad2 (He et al. 1997). The results are provided in the supplementary materials (Tables S1–S3; http://www.genetics.org/supplemental/) and suggest that some of these proteins may play a role in achiasmate segregation.

The role of dynein in achiasmate segregation:

Taken together, our results indicate that Dhc1 and Dlc1 promote the proper segregation of achiasmate chromosomes. Because both Dhc1 and Dlc1 are required for proper horsetail movement (Yamamoto et al. 1999; Miki et al. 2002), the recent examination of the role of recombination and horsetail movement in homologous association during prophase by Hiraoka and colleagues (Ding et al. 2004) is directly relevant. In brief, telomere clustering and horsetail movement are required to bring chromosomes into close alignment; recombination then stabilizes homologous associations. Significant levels of homologous association are retained in rec12 mutants, especially at the centromeres and telomeres. Importantly, homologous association of centromeres is eliminated in the dhc1 rec12 mutant (Ding et al. 2004), and proper segregation in the dhc1Δ rec12Δ mutant was reduced to random (Table 2). It is interesting to note that pairing of centric heterochromatin is required in one of the two distinct achiasmate segregation mechanisms possessed by Drosophila females (Dernburg et al. 1996; Karpen et al. 1996). Additionally, centromere pairing of achiasmate chromosomes has recently been observed in S. cerevisiae (Kemp et al. 2004). These data lead us to suggest that pairing of centric heterochromatin is important for achiasmate segregation in S. pombe. A conserved mechanism of centromere pairing may be widely important in the segregation of achiasmate chromosomes.

Other, though not necessarily mutually exclusive, mechanisms may promote achiasmate segregation in S. pombe. One such mechanism is suggested by the second achiasmate system in Drosophila females. As mentioned above, one mechanism of achiasmate segregation in Drosophila females requires pairing of homologous centromere-proximal heterochromatin (Dernburg et al. 1996; Karpen et al. 1996). However, in the second mechanism two achiasmate chromosomes, heterologous in this case, segregate from each other with high fidelity. Interestingly, pairing is not observed (Dernburg et al. 1996), and the heterologous chromosomes are thought to orient toward the less crowded pole, thus assuring that one will go to each pole. It is interesting to speculate that this type of mechanism might be especially effective in an organism, like S. pombe, that makes only a small number of microtubule connections per kinetechore (Ding et al. 1993). A mechanism, such as this, that relies on spatial or numerical constraints might be more effective if fewer chromosomes were achiasmate. This possibility has not been sufficiently tested in S. pombe.

This type of mechanism might be especially sensitive to perturbation of meiotic spindle function. Given the localization of Dlc1 to the spindle pole body at MI (Miki et al. 2002), and the function of dynein in spindle assembly and mitotic chromosome movement in metazoans (reviewed in Banks and Heald 2001), mutations in Dlc1 or Dhc1 may result in perturbation of meiotic spindle function. In fact, regardless of the underlying mechanism promoting achiasmate segregation, altered spindle function in the absence of Dlc1 or Dhc1 could lead to the observed failure of achiasmate segregation.