Sister chromatid cohesion and interhomologue recombination are coordinated to promote the segregation of homologous chromosomes instead of sister chromatids at the first meiotic division. During meiotic prophase in Saccharomyces cerevisiae, the meiosis-specific cohesin Rec8p localizes along chromosome axes and mediates most of the cohesion. The mitotic cohesin Mcd1p/Scc1p localizes to discrete spots along chromosome arms, and its function is not clear. In cells lacking Tid1p, which is a member of the SWI2/SNF2 family of helicase-like proteins that are involved in chromatin remodeling, Mcd1p and Rec8p persist abnormally through both meiotic divisions, and chromosome segregation fails in the majority of cells. Genetic results indicate that the primary defect in these cells is a failure to resolve Mcd1p-mediated connections. Tid1p interacts with recombination enzymes Dmc1p and Rad51p and has an established role in recombination repair. We propose that Tid1p remodels Mcd1p-mediated cohesion early in meiotic prophase to facilitate interhomologue recombination and the subsequent segregation of homologous chromosomes.
Chromosome structure and nuclear architecture are modified during meiosis to ensure that homologous chromosomes become connected and oriented to disjoin properly in the first meiotic division. These modifications likely begin with the establishment of meiosis-specific cohesin and condensin organization in premeiotic S-phase (Klein et al., 1999; Yu and Koshland, 2003). Additional meiosis-specific proteins create programmed DNA double strand breaks (DSBs) and regulate their repair during meiotic prophase so that a homologous chromatid is used as the template for repair rather than the sister chromatid. This process generates crossovers between the homologues. Sister chromatid cohesion that is distal to each crossover could be sufficient to provide stable interhomologue connection until release at the first meiotic anaphase (Maguire, 1995; Roeder, 1997; Nasmyth, 2001; Petronczki et al., 2003). How recombination repair of meiotic DSBs is prevented from using the sister chromatid, is directed to use the homologous chromosome, and is regulated to produce the appropriate connections remain important unanswered questions.
In meiosis, the Mcd1p subunit of cohesin largely is replaced by the meiosis-specific subunit Rec8p, which localizes to the meiotic chromosome axes (Klein et al., 1999). Meiotic recombination is also associated with chromosome axes. In meiosis, several cohesin subunits are required both for axis formation and successful completion of DSB repair (Klein et al., 1999; Revenkova et al., 2004). However, close associations between sister chromatids near DSBs could be an impediment for interhomologue recombination. Several lines of evidence indicate requirements for cohesion and cohesins in the repair of DSBs in mitotic cells, where sister-biased recombination is the preferred pathway for the repair of DSBs (Kadyk and Hartwell, 1992). First, in several organisms, genes encoding proteins that are required for cohesion were described to be required for DNA repair (Huynh et al., 1986; Birkenbihl and Subramani, 1992; Denison et al., 1993). Second, mutations in genes encoding subunits of cohesin complexes lead to precocious sister chromatid separation and defective postreplicative repair in yeast (Sjogren and Nasmyth, 2001). Similarly, the depletion of chicken Rad21p/Mcd1p/Scc1p leads to premature sister separation, reduced efficiency of DSB repair, and reduced frequency of sister chromatid exchange (Sonoda et al., 2001). Third, the establishment of cohesion in the preceding S-phase is required for successful DSB repair in G2 (Sjogren and Nasmyth, 2001). Fourth, cohesins are recruited to the sites of induced DSBs both in yeast and in mammals (Kim et al., 2002; Strom et al., 2004; Unal et al., 2004). Newly loaded cohesins are able to hold sister chromatids together (Strom et al., 2004), suggesting that cohesins could promote mitotic DSB repair by ensuring close proximity of sister chromatids. Bias toward interhomologue recombination in meiosis would require disfavoring the sister chromatid as a donor for DSB repair (for example, by locally destroying the close association between sister chromatids). Thus, components of the recombination machinery could remodel cohesion around DSBs to ensure use of the homologue for DSB repair in meiosis.
Efficient interhomologue recombination during vegetative growth and during meiosis requires Tid1p, which is a member of the SWI2/SNF2 family of helicase-like chromatin-remodeling proteins and is a paralogue of the recombination repair protein Rad54p (Eisen et al., 1995; Shiratori et al., 1999). However, in mitotic cells, the major repair pathway for DSBs uses the sister chromatid as a template even when a homologue is present and requires Rad54p and an interacting strand exchange enzyme, Rad51p (Kadyk and Hartwell, 1992; Paques and Haber, 1999; Symington, 2002; Tan et al., 2003). Thus, although Tid1p may interact with Rad51p (Dresser et al., 1997) and can promote the strand exchange activity of Rad51p in vitro (Petukhova et al., 2000), the effect of TID1 deletion on the repair of DSBs and on survival in mitotic cells is subtle unless sister chromatid–based repair is eliminated (Arbel et al., 1999; Lee et al., 2001; Signon et al., 2001; Ira and Haber, 2002).
During meiosis, most DSBs are repaired by a pathway that normally requires Dmc1p (a meiosis-specific paralogue of Rad51p; Bishop et al., 1992) in addition to Rad51p (Shinohara et al., 1992). Tid1p interacts with Dmc1p in the two-hybrid system (Dresser et al., 1997) and promotes Dmc1p colocalization with Rad51p on meiotic chromosomes (Shinohara et al., 2000), suggesting that Tid1p may serve together with Dmc1p during recombination repair to bias meiotic recombination specifically toward interhomologue repair (Dresser et al., 1997; Shinohara et al., 1997, 2003). The deletion of TID1 causes a delay in the processing of meiotic DSBs and increased resection at the broken DNA ends (Shinohara et al., 1997). However, even though mature recombination products do eventually form, at least at an artificial DSB hot spot (Shinohara et al., 1997), the majority of tid1Δ cells fail to sporulate (Klein, 1997; Shinohara et al., 1997).
We describe a novel phenotype of the deletion of TID1 that implicates Tid1p in the remodeling of chromosome structure (specifically sister chromatid cohesion). Tid1p is required for normal chromosome segregation in both meiotic divisions. The segregation defect in tid1Δ cells results from a failure of sister chromatid separation. In a spo11Δ spo13Δ background, there is no meiotic recombination, and the sole meiotic division requires that sister chromatids (not homologues) are separated. When tid1Δ is placed in the spo11Δ spo13Δ background, two thirds of the cells block in anaphase. This block is not suppressed by REC8 deletion alone. However, in the spo11Δ spo13Δ background, the tid1Δ block is suppressed by mcd1-1 (heat sensitive) if shifted to the nonpermissive temperature before or during meiotic prophase. It is also suppressed by mcd1-1 rec8Δ if shifted to the nonpermissive temperature at any time before cells begin to lose viability at the block. Moreover, both Mcd1p and Rec8p persist abnormally long in tid1Δ in an otherwise wild-type background. Our observations suggest that Tid1p plays a role in cohesion remodeling in meiotic prophase that is required for the successful separation of sister chromatids in anaphase.
Deletion of TID1 prevents sporulation by arresting the majority of cells in pachytene and by blocking cells that progress past pachytene in anaphase I and II
To determine the kinetics of progression through meiotic divisions, we used a strain with a single TUB1-GFP allele to visualize the spindles. Cells were stained with DAPI to visualize DNA. The majority of tid1Δ cells are blocked as mononucleate cells with no spindle (Fig. 1 A) and remain blocked even 26 h after the shift into sporulation. At the same time, tid1Δ does not affect the kinetics of formation and the appearance of the synaptonemal complex (SC), suggesting that Tid1p is not required for synapsis. The accumulation of cells with fully formed SC in tid1Δ indicates that the deletion of TID1 blocks cells in pachytene (Fig. S1, A and B, available). tid1Δ cells that were blocked in pachytene could be arrested by the pachytene checkpoint, which is triggered by unrepaired DSBs. Turnover of DSBs and formation of mature recombination products are delayed in tid1Δ (Fig. S1, C–F), confirming that pachytene arrest in tid1Δ is associated with the defective processing of DSBs.
The small percentage of tid1Δ cells that progress past pachytene (∼16% of cells progress into divisions after an ∼3-h delay; Fig. 1 A) reveals a new defect. Two thirds of these cells are blocked in anaphase I (single long spindle), whereas the remaining third enter the second division and are blocked in anaphase II (two long spindles). Chromatin in both types of blocked cells remains as a single stretched mass (“dumbbell”), indicating that the chromosomes stay largely unsegregated even though one or both divisions had occurred according to spindle behavior (Fig. 1 B). A very small fraction of cells (≤1%) successfully completes the meiotic divisions and segregates chromatin into four distinct masses. Consistent with the defect in chromatin segregation, the deletion of TID1 increases the missegregation of chromosome IV in the spo13Δ background (Fig. S2, available).
Deletion of SPO11 suppresses the tid1Δ pachytene arrest and allows anaphase I segregation
To prove that pachytene arrest of the majority of tid1Δ cells depends on DSB formation, the SPO11 gene, which encodes the meiosis-specific endonuclease that is responsible for producing DSBs and chiasmata (Cao et al., 1990; Keeney et al., 1997), was deleted, and the progression of cells through meiotic divisions was monitored by staining chromatin with DAPI. spo11Δ suppresses the pachytene arrest of tid1Δ, allowing 68% of spo11Δ tid1Δ cells to enter the divisions without delay (Fig. 2 B), which supports the idea that the tid1Δ pachytene arrest is caused by unrepaired DSBs.
The deletion of SPO11 also permits testing of whether or not the tid1Δ defect in chromosome segregation in the first meiotic division is caused by a failure to dissolve chiasmate connections between homologues. Chiasmata are produced as a result of interhomologue recombination and are stabilized by cohesion distal to crossover sites (see Introduction). spo11Δ eliminates interhomologue recombination and, thus, chiasmata and leads to random segregation of homologous chromosomes in the first meiotic division (Fig. 2 A; Klein et al., 1999). spo11Δ suppresses the anaphase I block of tid1Δ. In spo11Δ tid1Δ, most of the dumbbells are resolved after 10 h of sporulation, whereas in tid1Δ, they persist until their chromatin becomes fragmented after 16 h of sporulation (Fig. 2 C and not depicted). This suggests that the tid1Δ failure of segregation in the first division is caused by unresolved connections between homologues. Homologues may remain connected in tid1Δ because of a failure to resolve sister chromatid cohesion that is distal to crossover sites. Even though spo11Δ suppresses the defects of tid1Δ in the first meiotic division, in spo11Δ tid1Δ, anaphase II dumbbells accumulate, and fewer cells finish the second meiotic division than in spo11Δ (this difference was seen in four independent trials; unpublished data). This raises the possibility that tid1Δ affects sister chromatid separation.
tid1Δ blocks sister chromatid segregation in the spo11Δ spo13Δ background
The effect of TID1 deletion on the ability of cells to separate sister chromatids was tested in a background where sister chromatids segregate instead of homologues. The introduction of spo13Δ in the spo11Δ background, where meiotic recombination is eliminated, leads to biorientation of sister kinetochores and subsequent segregation of sister chromatids in the single meiotic division (Fig. 3 A; Klapholz et al., 1985; Shonn et al., 2002). In spo11Δ spo13Δ tid1Δ, 23% of cells complete the separation of chromatin into two masses (binucleates) compared with 66% of cells in spo11Δ spo13Δ (Fig. 3 B). Thus, even in the absence of interhomologue recombination, the deletion of TID1 prevents some spo11Δ spo13Δ cells from successfully finishing the division, presumably because of the failure to segregate sister chromatids in the absence of Tid1p. Consistently, one reason for the reduction in the percentage of binucleates is the accumulation of dumbbells (Fig. 3 C), which begin degrading at ∼10 h into sporulation (unpublished data). Another reason is an accumulation of cells with a single unstretched mass of chromatin (mononucleates) and a short spindle. In spo11Δ spo13Δ tid1Δ, only 40% of cells progress past mononucleate stage (Fig. 3 D) compared with 70% in spo11Δ tid1Δ (Fig. 2 B), which is higher than the 15–25% in tid1Δ but lower than the 66% in spo11Δ spo13Δ (Fig. 3 B). Tagging of spindle tubulin with GFP reveals that the mononucleate cells consist of two classes: cells with a single spindle pole body (SPB) and cells with a short spindle. To determine what percentage of the population is represented by each class, we scored 50 spo11Δ spo13Δ mononucleate cells (29% of the total population is mononucleate at 8 h) and 68 spo11Δ spo13Δ tid1Δ mononucleate cells (53% of the total population is mononucleate at 8 h). The fraction of cells with a single SPB is similar in spo11Δ spo13Δ and spo11Δ spo13Δ tid1Δ (14% of each total population). However, the fraction of cells with a short spindle is lower in spo11Δ spo13Δ than in spo11Δ spo13Δ tid1Δ: 15 versus 39% of the total population, respectively. In addition, the fraction of cells with a short spindle subsequently decreases in spo11Δ spo13Δ, whereas it persists essentially unchanged in spo11Δ spo13Δ tid1Δ (unpublished data). Thus, the failure of spo11Δ spo13Δ tid1Δ cells to divide their chromatin into two masses can be attributed to a block with two phenotypes: dumbbells and mononucleate cells with a short spindle. The tid1 K351R allele, which is designed to eliminate ATP hydrolysis (Petukhova et al., 2000), causes a similar block in the spo11Δ spo13Δ background (unpublished data). A similar phenotype has been reported in other mutants where sister chromatids fail to segregate, leading to a block with a metaphase-length spindle (Toth et al., 2000; Clyne et al., 2003; Lee and Amon, 2003; Rabitsch et al., 2003).
Connections between sister chromatids persist in cells blocked in anaphase by the deletion of TID1
To examine the segregation of sister chromatids in greater detail, one of the homologues of chromosome IV was labeled with an array of tetO less than 2 kb away from the centromere (small GFP spot; He et al., 2000) and an array of lacO near the right telomere (∼1,071 kb away from centromere IV; large GFP spot). These arrays bind GFP-linked tetR and lacI, respectively. Spots were visualized by using anti-GFP antibodies in fixed whole cells. Tubulin that was visualized with antitubulin antibodies and chromatin that was stained with DAPI were used to distinguish between different cell classes. Cells with two or four spots indicate unseparated or separated IV sister chromatids, respectively, whereas cells with three spots indicate the separation of chromatids at the centromeres but not at the telomeres (Fig. 3 E). To avoid scoring cells that have begun to degrade, samples were scored at the 8-h time point. The introduction of tid1Δ decreases the percentage of cells with four spots, whereas it increases the percentage of cells with two or three spots in the spo11Δ spo13Δ background (Fig. 3 F). Most of the increase in the percentage of cells with three GFP spots is accounted for by dumbbells, and most of the increase in the percentage of cells with two spots is accounted for by cells with a short spindle (cells with a single SPB are responsible for the high background of cells with two GFP spots in spo11Δ spo13Δ and spo11Δ spo13Δ tid1Δ; 14% in each; Fig. 3 G). In ∼25% of the spo11Δ spo13Δ tid1Δ binucleates, the spots either segregate 4:0 (indicating missegregation) or 3:1 (suggesting breakage), whereas these classes are relatively rare in spo11Δ spo13Δ (∼2%). The persistence of sister association in cells with short spindles suggests that the missegregation classes result from persistent connections between the chromosome IV sisters. Importantly, attachment of centromeres IV to the spindle appears unaffected by tid1Δ, as the centromeres are closer to the spindle pole than are the telomeres in all dumbbells, even when telomeres IVR are still unseparated. This observation makes it unlikely that a defect in kinetochore attachment is responsible for the tid1Δ failure of chromosome segregation. Altogether, these observations are consistent with persistent sister chromatid connections, which are variable in position and/or number, causing the tid1Δ block to chromosome segregation. Presumably, connections relatively far from centromeres allow spindle elongation and dumbbell formation, whereas connections closer to centromeres limit spindle length and prevent stretching of the chromatin mass.
Pds1p turnover is not affected by tid1Δ in the spo11Δ spo13Δ background
Failure to separate sister chromatids in tid1Δ could result from an inhibition of the mechanism that is responsible for dissolving sister chromatid cohesion. Pds1p is down-regulated at anaphase I, and its loss is thought to enable cohesion dissolution by Esp1p/separase action (Petronczki et al., 2003). Therefore, we monitored the turnover of HA-tagged Pds1p in whole permeabilized cells by using anti-HA and antitubulin antibodies after 8 h of sporulation, when the spo11Δ spo13Δ tid1Δ and spo11Δ spo13Δ behaviors have just begun to diverge (Fig. 3 B) but before cell degradation. No Pds1p-positive cells with stretched dumbbell nuclei were seen either in spo11Δ spo13Δ (eight dumbbells were scored) or spo11Δ spo13Δ tid1Δ (52 dumbbells were scored). The increase in the percentage of cells with a short spindle that was observed in spo11Δ spo13Δ tid1Δ is almost exclusively accounted for by Pds1p-negative cells (Table I). This indicates that the short spindle arrest observed in spo11Δ spo13Δ tid1Δ occurs downstream of Pds1p destruction. Therefore, we conclude that the deletion of TID1 in spo11Δ spo13Δ results in a block of two thirds of cells in anaphase (short spindle cells plus dumbbells). The kinetics of Pds1p turnover that was measured in whole cell protein extracts on Western blots is also nearly identical in spo11Δ spo13Δ and spo11Δ spo13Δ tid1Δ (unpublished data). Thus, the persistent sister chromatid association that is caused by tid1Δ does not arise from any mechanism that leads to the persistence of Pds1p. Note that in dumbbells, where there is no detectable Pds1p, a significant fraction nevertheless has tightly associated sisters at telomere IVR (Fig. 3 G, 3 spot class). This result suggests that the persistent connections occur in the presence of active Esp1p.
The sister chromatid segregation defect of tid1Δ is not rescued by the deletion of REC8 in a spo11Δ spo13Δ background
The timely turnover of Pds1p in spo11Δ spo13Δ tid1Δ indicates that cells are not blocked in metaphase but, rather, progress into a defective anaphase in which sister chromatids fail to segregate. In meiosis, the maintenance of sister chromatid cohesion requires the meiosis-specific cohesin Rec8p, and its cleavage is required for chromosome segregation in anaphase (Klein et al., 1999; Buonomo et al., 2000). One possible reason for the failure of sister segregation in tid1Δ cells is the persistence of Rec8p-based cohesin complexes on chromosomes. To test this possibility, REC8 was deleted in the spo11Δ spo13Δ background, where the effect of tid1Δ on sister chromatid separation is most obvious. The deletion of REC8 does not rescue the defect of sister chromatid separation in spo11Δ spo13Δ tid1Δ. The percentage of cells that progress past anaphase (binucleates) is similar in spo11Δ spo13Δ tid1Δ and in spo11Δ spo13Δ rec8Δ tid1Δ (Fig. 4 A). The percentage of dumbbells is reduced in spo11Δ spo13Δ rec8Δ tid1Δ (Fig. 4 B) because more cells are blocked earlier with a single unstretched mass of chromatin (not depicted). To determine the fraction of cells that were blocked with short spindles, spread preparations of nuclei were prepared after 12 h of sporulation, and the spindle was visualized by immunostaining (Fig. 4 C). A large fraction of spo11Δ spo13Δ rec8Δ tid1Δ cells is arrested with short spindles (43% in spo11Δ spo13Δ rec8Δ tid1Δ vs. 2% in spo11Δ spo13Δ rec8Δ at 12 h of sporulation). Thus, the deletion of REC8 fails to suppress the tid1Δ-mediated block to sister segregation.
Inactivation of the mitotic cohesin complex subunit Mcd1p rescues the tid1Δ anaphase block in the spo11Δ spo13Δ background
In mitotic cells, the maintenance of cohesion between sister chromatids requires Mcd1p, which is a homologue of Rec8p (Guacci et al., 1997; Michaelis et al., 1997). Although Mcd1p has been reported not to have a major role in the maintenance of cohesion during meiosis in budding yeast (Klein et al., 1999), it does play a role in meiotic sister chromatid cohesion in fission yeast (Yokobayashi et al., 2003). Therefore, we tested whether Mcd1p is required for persistent sister chromatid connections in tid1Δ by using a heat-sensitive allele of MCD1 (mcd1-1), which was previously shown to be defective in cohesion maintenance at nonpermissive temperature in mitotic cells (Guacci et al., 1997). Progression of cells through the division was monitored by using TUB1-GFP and DAPI-stained chromatin. Sporulating cells were shifted from permissive for mcd1-1 temperature (21°C) to nonpermissive temperature (33°C) every 2 h, and binucleates were scored after at least 24 h of sporulation. Early shift to 33°C completely rescued the tid1Δ anaphase block (compare spo11Δ spo13Δ mcd1-1 with tid1Δ vs. without tid1Δ; Fig. 5 A). However, this suppression disappears by metaphase (compare spo11Δ spo13Δ mcd1-1 tid1Δ in Fig. 5 A with spo11Δ spo13Δ mcd1-1 in Fig. S3 C, in which cells with short spindles represent metaphase cells; available). To more precisely determine the effect of Mcd1p inactivation on the tid1Δ anaphase block, the percentage of binucleates, dumbbells, and cells with a short spindle was compared in cells sporulating at 21°C and in cells shifted to 33°C at 4 h (Fig. 5, B–D, and Fig. S3, A–C), which is a time when mcd1-1 completely suppresses tid1Δ. Consistently, identical levels of binucleates are reached by spo11Δ spo13Δ mcd1-1 and spo11Δ spo13Δ mcd1-1 tid1Δ after a shift to 33°C at 4 h of sporulation. No accumulation of cells with a short spindle or of dumbbells was observed in spo11Δ spo13Δ mcd1-1 tid1Δ cells compared with spo11Δ spo13Δ mcd1-1. At 21°C, the phenotype of tid1Δ is unchanged by mcd1-1. These data suggest a direct involvement of Mcd1p in generating persistent connections in tid1Δ.
Deletion of REC8 in spo11Δ spo13Δ mcd1-1 allows late suppression of the tid1Δ sister chromosome segregation defect
Our data demonstrate that in the spo11Δ spo13Δ background, Mcd1p inactivation efficiently suppresses the tid1Δ anaphase block but only at early times in sporulation. This suggests that some other protein mediates persistent sister chromatid connections later in meiosis in the absence of Mcd1p. To determine whether the Mcd1p-independent sister connections are mediated by Rec8p, the effect of rec8Δ on the anaphase block of tid1Δ was studied in spo11Δ spo13Δ mcd1-1. Progression of cells through the divisions was monitored by using TUB1-GFP and DAPI-stained chromatin. Sporulating cells were shifted from permissive to nonpermissive temperature every 2 h, and binucleates were scored after 24 h of sporulation. The anaphase block of tid1Δ is suppressed by mcd1-1 rec8Δ at early times as well as at late times at 33°C (Fig. 6 A) despite the fact that cells with tid1Δ reach full anaphase block by 8–10 h of sporulation at 21°C (Fig. S3, D–F). A slight decrease in the level of suppression after 10 h coincides with the onset of chromatin fragmentation in a fraction of cells (not depicted). To confirm that persistent sister chromatid connections is the main reason for the anaphase block, spo11Δ spo13Δ mcd1-1 rec8Δ tid1Δ cells were allowed to block at anaphase (10 h of sporulation at 21°C) and were transferred to 33°C (Fig. 6, B–D). After a shift to 33°C, the turnover of anaphase-blocked cells (dumbbells and cells with a short spindle) in spo11Δ spo13Δ mcd1-1 rec8Δ tid1Δ occurs within 1 h, coinciding with an increase in binucleates. This quick and complete rescue indicates that both Mcd1p and Rec8p are required to maintain the sister chromatid connections that persist in spo11Δ spo13Δ tid1Δ.
Dissociation of cohesins Mcd1p and Rec8p from chromatin is delayed in tid1Δ in an otherwise wild-type background
Failure to segregate chromosomes in both divisions in tid1Δ is likely caused by Mcd1p- and Rec8p-dependent connections that persist in anaphase I and II. To confirm this possibility, the association of Mcd1p and Rec8p with chromatin was tested in both anaphase I and II in spread preparations of meiotic nuclei. Spindle behavior was used as an internal marker for progression through the meiotic divisions in order to compare cohesin dissociation from chromatin in tid1Δ with wild-type. Rec8p and Mcd1p were visualized by immunolabeling in spread nuclei of sporulating tid1Δ and wild-type cells. Although none of the wild-type cells have Rec8p that is associated with chromatin in anaphase II, Rec8p signal remains in 24% of tid1Δ cells that have two fully elongated spindles. Thus, the dissociation of Rec8p from chromatin is delayed in tid1Δ (Fig. 7, A and B). Aside from this delay, there were no obvious differences in the appearance of Rec8p in wild-type and tid1Δ cells (not depicted), which is consistent with the normal appearance of SCs in tid1Δ (Fig. S1 A).
Analogously, none of the wild-type cells have Mcd1p signal after metaphase I, whereas Mcd1p persists as spots on chromatin in tid1Δ through anaphase I (82% of cells) and II (90% of cells). Mcd1p foci in wild-type and tid1Δ are not apparently different in prepachytene nuclei (unpublished data). These data are consistent with a direct role for Mcd1p in the maintenance as well as in the establishment of persistent sister chromatid connections in tid1Δ.
Early suppression of tid1Δ anaphase block by mcd1-1 in a spo11Δ spo13Δ background suggests that the misregulation of Mcd1p in the absence of Tid1p occurs before or during prophase. To confirm that tid1Δ has an effect on Mcd1p during prophase, Mcd1p was visualized by immunolabeling in spread nuclei of pachytene tid1Δ and wild-type cells, which were identified by the appearance of well-condensed chromosome bivalents. Mcd1p appears as spots on chromatin in spread preparations of pachytene cells (Klein et al., 1999; unpublished data). The number of Mcd1p spots on pachytene chromosomes in tid1Δ is significantly increased compared with wild type (Fig. 8), suggesting that the functional interaction between Mcd1p and Tid1p begins before or during pachytene. Thus, the misregulation of Mcd1p and, perhaps as a consequence, of Rec8p in or before prophase may lead to sister chromatid connections that persist through both anaphases and prevent chromosome segregation in both divisions of tid1Δ.
Numerous observations indicate that Tid1p functions in conjunction with strand exchange enzymes in meiotic and mitotic cells, presumably to facilitate single strand invasion or subsequent processing during DNA DSB repair (see Introduction). However, a significant component of the tid1Δ phenotype is the failure of chromosome segregation during the meiotic divisions as a result of unresolved connections between sister chromatids. This tid1Δ phenotype provides a new perspective on the role of Tid1p in recombination repair.
Tid1p is required to remodel cohesion before the onset of divisions to allow severing of sister chromatid connections in anaphase
Our data indicate that persistent sister chromatid connections mediated by Mcd1p and Rec8p are the primary reason for the anaphase block in tid1Δ. We propose that Tid1p is required for Mcd1p (and possibly Rec8p) to be removed by an Esp1p-independent mechanism in prophase in preparation for or during DNA repair.
The timely progression into anaphase suggests that pachytene (DNA damage) and spindle checkpoints are not triggered by tid1Δ in the spo11Δ spo13Δ background. The involvement of a checkpoint that was triggered directly by DNA damage arising before the division (for example, in premeiotic S-phase) is possible but seems unlikely because no anaphase arrest was observed in the first division in spo11Δ tid1Δ.
A nucleosome remodeling function has been proposed for the Tid1p paralogue Rad54p (Alexeev et al., 2003; Jaskelioff et al., 2003; Wolner and Peterson, 2005). Another member of the SWI2/SNF2 family, SNF2hp, regulates cohesin loading on chromosomes, and ATPase activity is required for this process (Hakimi et al., 2002). We propose that a Tid1p chromatin remodeling function regulates Mcd1p and Rec8p association with chromosomes.
Mcd1p is required for local sister chromatid cohesion in budding yeast meiosis
In budding yeast, Mcd1p appears not to be essential for cohesion between sister chromatids in meiotic prophase, yet it forms discrete foci on pachytene chromosomes rather than localizing along their full lengths as with Rec8p (Klein et al., 1999). However, our results prove that Mcd1p can provide connections between sister chromatids in meiosis of Saccharomyces cerevisiae. Variability of the connections from cell to cell suggests that they could be independent of regular cohesin-binding sites. What leads Mcd1p to be loaded onto chromosomes in discrete, irregularly located spots (unpublished data) rather than along the chromosome axes (as it is in mitotic cells) and loaded as Rec8p is in meiotic cells is an important and unanswered question.
The Tid1p effect on recombination could be exerted through regulation of sister chromatid connections
We propose that Tid1p remodels Mcd1p-mediated cohesion to promote and regulate interhomologue recombination (Fig. 9). In this model, domains of Mcd1p would colocalize with DSBs that initiate meiotic recombination (and may be hot spots for DSBs; Petes, 2001). These domains would initially be prohibitive for impending interhomologue recombination for the following reason. In mitosis, cohesion between sister chromatids that was established during replication and de novo loading of cohesins at a DSB site are required for postreplicative DSB repair (Sjogren and Nasmyth, 2001; Strom et al., 2004; Unal et al., 2004). One of the roles proposed for cohesins in postreplicative repair in mitotic cells is to keep sister chromatids in proximity (Strom et al., 2004; Unal et al., 2004). If so, then in meiosis, additional factors would have to be incorporated at the Mcd1p domains to guide DSB repair to lead to the separation of, rather than the alignment of, sister chromatids. We suggest that Tid1p specifically facilitates this separation, which allows interaction with one of the chromatids of the homologue, perhaps by promoting displacement from the loop to the axis (Blat et al., 2002). In the absence of this loop displacement, the DSB ends would be free to separate and could give rise to the separation of Dmc1p and Rad51p foci that were reported for tid1Δ (Shinohara et al., 2000). This two-step mechanism for control over the fate of multiple programmed DSBs in meiosis would provide an escape route through sister-based repair in case of a failure to initiate repair involving the homologue as a template. Remodeling of cohesins by Tid1p could be promoted by Tid1p interaction with Dmc1p, which is required for the interhomologue bias of recombination in meiosis (Bishop et al., 1992; Dresser et al., 1997). Thus, Tid1p would serve to remodel chromosome structure, which is promoted by but is not necessarily dependent on DSB metabolism. The model predicts that Mcd1p and proteins involved in DSB repair should colocalize to some degree depending on their relative times of activity and on whether Mcd1p is involved at all or at a subset of DSB sites.
According to this model, the defect in DSB processing and the failure to segregate chromosomes in tid1Δ arise as a result of the failure of the same Tid1p-dependent process. Proteins that were loaded in preparation for DSB processing could obscure the cleavage site of Mcd1p, creating an area of separase-resistant cohesion. The absence of Mcd1p remodeling in tid1Δ would cause Mcd1p-bound domains to persist and impede DSB repair. Recombination in tid1Δ would slowly overcome the obstacle if Mcd1p (and subsequently Rec8p) could be dislodged without Tid1p activity (for example, during the extended resection of DSBs that was observed in tid1Δ; Shinohara et al., 1997), thus resulting in a number of crossovers being flanked by a region of separase-resistant cohesion. It is interesting to note that flanking regions of separase-resistant sister chromatid cohesion could serve to stabilize chiasmata in wild-type cells and, thus, could provide a system for orientation that is distinct from the separase-regulated pathway.
Very little is known about how sister chromatid cohesion influences interhomologue recombination repair. However, it is clear that chromosome structure is adapted early to promote the programmed essential recombination repair that occurs in meiosis. Our results indicate that Tid1p is involved in this adaptation. Regulation of sister chromatid associations by Tid1p could similarly play a role during vegetative growth in the Tid1p-dependent subset of DSB repair that takes place between homologues in diploid cells (Klein, 1997; Shinohara et al., 1997; Arbel et al., 1999) and during break-induced replication (Signon et al., 2001; Ira and Haber, 2002). Structure–function genetic analysis by mutations in TID1 and a biochemical description of the proposed remodeling activity of Tid1p will be required to establish the molecular functions of and the biological requirements for Tid1p in mitotic and meiotic cells.
Materials And Methods
Strains and plasmids
Diploid strains that were used in the assays were obtained by matings of MDY431 and MDY433 or their isogenic derivatives (Table S1, available). spo13Δ was made by single step replacement using p(spo13)16 (Wang et al., 1987). spo11Δ was introduced into strains by transformation with pGB518 (Giroux et al., 1989). The plasmid to transfer tid1 K351R into our strains was made by cloning the PstI–SalI fragment of TID1 containing the K351R mutation from prdh54K/R.1 (Petukhova et al., 2000) into YIplac204. Integration of this construct into wild-type strains creates a partial duplication containing tid1 K351R and a truncation of TID1 preceded by a plasmid sequence instead of the TID1 promoter. rec8Δ was made by replacing REC8 with a PCR product containing TRP1 from pFA6a-TRP1 and small fragments of homology to flanking regions of the REC8 gene (Longtine et al., 1998). The construct to replace REC8 with REC8-GFP (pMDE820) was made by adding GFP coding sequence (EcoRI–XhoI) from pAFS135 (Straight et al., 1998) to the 3′ terminus of REC8, which was cloned by PCR from genomic template into pRS306. MCD1 was replaced with mcd1-1 by transformation of a fragment containing mcd1-1 from pVG327 (provided by V. Guacci, Fox Chase Cancer Center, Philadelphia, PA). Spindles were marked by integrating TUB1-GFP at the URA3 locus by transformation with pAFS125 (provided by A. Straight and A.W. Murray, Harvard University, Cambridge, MA) cut with StuI. One additional copy of PDS1-HA was introduced into strains by transformation of one of the haploids with pVG319 (provided by V. Guacci) cut with KpnI. The presence of Pds1p-HA in these strains was confirmed on Western blots.
Strains with a chromosome IV centromere and right telomere marked by GFP spots were constructed in several steps. To integrate a tandem array of 256 copies of the lac operator near the right telomere of chromosome IV, plasmid pMDE780 was constructed by cloning a HindIII–SphI fragment of chromosome IV (nt 1,521,038–1,522,195) into pAFS59 (Straight et al., 1996). PCYC1-lacI/GFP was introduced by the integration of pAFS152 (provided by A. Straight and A.W. Murray) at URA3 in derivatives of MDY431 and at CYC1 in derivatives of MDY433. PDMC1-lacI/GFP was introduced into strains at LYS2 by the integration of pMDE798, which contains lacI/GFP-NLS from pAFS152 (provided by A. Straight and A.W. Murray) under control of the DMC1 promoter in order to ensure expression during meiosis. To mark chromosome IV with a centromere GFP spot, pXH115 containing an array of 224 tet operator repeats was integrated into the chromosome. PURA3-tetR/GFP was introduced by the integration of pXH123 at LEU2 (He et al., 2000). Construction of strains with a paracentric inversion on chromosome VIIR have been described previously (Dresser et al., 1994). The presence of correct alleles in all strains was confirmed by genetic complementation and/or by PCR.
For all cytological experiments, cultures were prepared from freshly mated diploids that were grown in acetate-containing rich medium and were transferred to sporulation medium to sporulate at 8 × 107 cells + buds/ml at 30°C. For an analysis of progression through meiosis, cells were stained with 0.5 μg/ml DAPI in 50% ethanol and were examined by fluorescence microscopy. For experiments requiring the preservation of GFP fluorescence, cells were fixed in 1% PFA for 15 min at RT, washed with PBS and 30% ethanol, and stained with 5 μg/ml DAPI in PBS. Stained cells were embedded in 1% low melting point agarose (GIBCO BRL) in PBS to prevent flattening.
Pds1p-HA and tubulin were visualized in whole cells using antibodies against HA (12CA5; Babco) and tubulin (YL1/2; Abcam). Cells were fixed with 4% PFA in 50 mM phosphate buffer, pH 6.5, containing 0.5 mM magnesium chloride for 10 min and were washed twice with 0.1 M phosphate buffer containing 1.2 M sorbitol (immunofluorescence assay buffer B [IFB]). The cell wall was digested with 20 μg/ml Zymolyase-100T and was washed with IFB. Spheroplasts were permeabilized with 1% Triton X-100 in IFB, washed twice with IFB, and placed on coverslips that were treated with 1% polyethyleneimine. Nonspecific binding of antibodies was blocked by incubating the preparations in 4% nonfat dried milk (Bio-Rad Laboratories) in PBS for 20 min. Coverslips were incubated for 1 h with primary mouse anti-HA antibodies (1:500 final dilution) and rat antitubulin YL1/2 (1:1,000), washed three times with PBS, and incubated with secondary FITC-conjugated donkey anti–mouse and Rhodamine Red-X–conjugated donkey anti–rat antibodies (both 1:400 dilution; Jackson ImmunoResearch Laboratories). The preparations were then washed three times with PBS, stained with 10 μg/ml DAPI in PBS for 20 min, rinsed in PBS, and mounted on slides with 1.5% low melting point agarose (GIBCO BRL) in 70% glycerol in PBS. To slow fluorescence fading, Citifluor in glycerol (Ted Pella) was added under the coverslip before examination. This same method was used to visualize telomere and centromere GFP spots with primary rabbit anti-GFP antibodies (1:500; Invitrogen) and Cy2-conjugated donkey anti–rabbit secondary antibodies (1:400; Jackson ImmunoResearch Laboratories).
Spreads were prepared as described previously (Dresser and Giroux, 1988) on slides that were treated with 1% polyethyleneimine. Labeling with antibodies was performed on slides as described above with the exception of using TBS. To compare the percentage of cells that were arrested with a short spindle in spo11Δ spo13Δ rec8 tid1Δ with in spo11Δ spo13Δ rec8Δ, tubulin was visualized in spread preparations of meiotic nuclei using rat antitubulin YOL1/34 serum (Serotec) diluted 1:50 and Rhodamine Red-X–conjugated donkey anti–rat antibodies (Jackson ImmunoResearch Laboratories) diluted 1:400. To compare the dissociation of Rec8p from chromatin in wild-type with tid1Δ, Rec8p-GFP was visualized with chicken anti-GFP antibodies (Chemicon International) diluted 1:5,000 and FITC-conjugated donkey anti–chicken secondary antibodies (Jackson ImmunoResearch Laboratories) diluted 1:400. Tubulin was visualized as described above. Cross-reaction of the primary chicken anti-GFP antibodies with the rat antitubulin antibodies was noted but did not interfere with the assay. To compare the dissociation of Mcd1p from chromatin in wild-type with tid1Δ, Mcd1p was visualized with affinity-purified rabbit anti-Mcd1p (559–1) antibodies (provided by V. Guacci) diluted 1:4,000 and with Cy3-conjugated donkey anti–rabbit secondary antibodies (Jackson ImmunoResearch Laboratories) diluted 1:400. Tubulin was visualized using rat antitubulin YL1/2 antibodies (Abcam) diluted 1:1,000 and Rhodamine Red-X–conjugated donkey anti–rat antibodies (Jackson ImmunoResearch Laboratories) diluted 1:400.
All images were acquired at 12-bit depth at RT on a microscope (Axioplan 2ie; Carl Zeiss MicroImaging, Inc.) using a plan-Apochromat 100× NA 1.4 oil differential interference contrast objective (Carl Zeiss MicroImaging, Inc.) and a camera (Quantix 57; Photometrics) driven by MetaMorph software (Universal Imaging Corp.). Stacks of images were produced by using either a nanopositioner (PolytecPI) or a stage (Carl Zeiss MicroImaging, Inc.) for focusing movements. Image processing was limited to scaling by using either MetaMorph software or AdobePhotoshop. The isointensity surface extraction that was used for three-dimensional reconstruction and maximum intensity projection images were generated by using OMRFQANT (written by M.E. Dresser).
Silver-stained spread preparations of meiotic nuclei were prepared as described previously (Dresser and Giroux, 1988) and were examined in an electron microscope (model 1200EX; JEOL). Electron micrographs were acquired by using a high resolution digital camera (model ES4; Advanced Microscopy Techniques Corp.).
DNA DSB and recombination products analysis
The formation of DNA DSBs was assayed on CHEF gels (Gerring et al., 1991). Samples were run on an electrophoresis system (CHEF II; Bio-Rad Laboratories) in 1% FastLane agarose and 0.25× Tris-borate/EDTA electrophoresis buffer at 12°C and 6 V/cm by using a 20–40-s ramp over 24 h. Southern transfer was performed by alkaline capillary transfer (Chomczynski, 1992) onto ZetaProbe membrane (Bio-Rad Laboratories). DSB fragments were visualized by using a radioactively labeled probe that was complementary to a near-telomeric region of chromosome III (Dresser et al., 1997) and were quantified using a phosphoimager (PhosphorImager 425; Molecular Dynamics). The formation of mature recombination products was assayed in a strain that was hererozygous for a paracentric inversion on chromosome VII (Dresser et al., 1994).
Online supplemental material
Table S1 lists yeast strains that were used in this study. Fig. S1 shows the kinetics of SC formation, turnover of DSBs, and formation of mature recombination products in wild-type and tid1Δ. Fig. S2 shows the segregation of chromosome IV with a telomere GFP spot on both homologues in spo13Δ and spo13Δ tid1Δ. Fig. S3 shows control data for Figs. 5 and 6. More detailed information about these results is located in the supplemental text.
We thank Debra Ewing, Danese Coody, Valerie Besseyrias, and Ben Fowler (Oklahoma Medical Research Foundation Imaging Facility) for technical help; Aaron Straight, Marion Shonn, and Andrew Murray for GFP constructs; and Douglas Koshland (Carnegie Institution of Washington, Baltimore, MD) for anti-Pds1p antibody.
This work was supported by grants from the National Institutes of Health (GM-45250-04) and the Oklahoma Center for the Advancement of Science and Technology (HR01-032) to M.E. Dresser.
Abbreviations used in this paper: DSB, double strand break; SC, synaptonemal complex; SPB, spindle pole body.