Rtt107 association with Mms22 prevents genotoxin sensitivity and prolonged checkpoint
Both Rtt107 and Mms22 are important for genome stability; however, the biological functions of their interaction have been unclear. To address this question, we examined a separation-of-function allele of Mms22 that specifically disrupts the Rtt107 binding without affecting other known interactions29. Given that Rtt107 and Mms22 each have multiple binding partners, this allele provides a valuable reagent to define the biological functions of their interaction25, 28, 30. High-resolution structure has revealed that Rtt107 binds to the N-terminal Rtt107-interaction-motif (RIM) of Mms2229. Further, alanine substitution of two residues (D13 and Y33) within Mms22’s RIM (mms22RIM) disrupts the Mms22 and Rtt107 interaction (Fig. 1a)29. mms22RIM supports the wild-type level of protein expression and interactions with other binding partners, such as the Mms1 protein that associates with the C-terminal region of Mms22 (Supplementary Fig. 1a)25, 29. We confirmed here that mms22RIM led to sensitivity to the DNA methylation agent MMS (methyl methanesulfonate) and further showed that it also caused sensitivity to the Top1 trapping compound CPT (camptothecin) and the replication fork blocking agent HU (hydroxyurea) (Fig. 1b). Compared with mms22RIM, mms22∆ led to greater sensitivity to MMS, CPT, and HU, as well as slower growth (Fig. 1b). This agrees with the notion that mms22RIM is a separation-of-function allele affecting the Mms22-Rtt107 binding without interfering with other roles of Mms22.
Sensitivity to several genotoxins raised the possibility that mms22RIM may affect common processes during different genotoxin responses such as the DNA damage checkpoint. To test this notion and determine the underlying causes of the genotoxin sensitivity exhibited by mms22RIM cells, we focused on MMS treatment since DDC has been well examined in this condition. Specifically, we examined how synchronized cells moved through the cell cycle in a regimen that allows monitoring of both the initial response to MMS and the recovery. Briefly, cells synchronized in G1 were released to the cell cycle in the presence of MMS for 45 minutes, and then allowed to recover in MMS-free media (Fig. 1c, right). As expected, wild-type cells progressed slowly through S phase and were able to enter the second cell cycle after MMS washout (Fig. 1c). We also examined Rad53 activation, a marker for DDC signaling. The F9 antibody is widely used to detect Rad53 activation as it specifically recognizes the phosphorylated but not unmodified form of Rad53, although the antibody also binds to a nonspecific band only in G1 cells31. As previously reported, we observed that in wild-type cells, Rad53 activation was induced in S phase when cells were treated with MMS (30–45 min) and gradually diminished during the recovery phase (Fig. 1d and Supplementary Fig. 1b).
In contrast to wild-type cells, mms22RIM cells delayed the re-entry into the second G1 phase after MMS washout, though showed no obvious defects during S phase progression in the presence of MMS (Fig. 1c). Concomitantly, Rad53 phosphorylation induced by MMS failed to decline during the recovery phase (Fig. 1d and Supplementary Fig. 1b). The results from these two assays are consistent with each other and suggest that mms22RIM cells exhibit persistent DDC during the recovery phase, which can delay entry to the next cell cycle.
Genotoxin sensitivity associated with the loss of the Mms22-Rtt107 interaction is rescued by reducing the DNA damage checkpoint
Persistent checkpoint can be caused by delayed genome replication and repair or by reduced ability to dampen the checkpoint itself. The two scenarios differ in how cells would respond to reduced levels of DDC. In the first scenario, cell viability would suffer when checkpoint function is weakened, because cells rely on optimal checkpoint to complete DNA replication and repair. In contrast, cells in the second scenario could show better survival upon checkpoint weakening. Based on this rationale, we tested how mms22RIM cells responded to reduced DDC levels conferred by mild alleles of the checkpoint mediator protein Rad9 or Ddc1. The two mutants used, namely rad9-K1088M and ddc1-T602A, contain single point mutations that reduce Rad9 binding to γH2A and Ddc1 binding to another checkpoint factor Dpb1120, 32. These studies have shown that rad9-K1088M and ddc1-T602A mildly reduce DDC. Significantly, we found that either allele conferred strong rescue of the MMS sensitivity of mms22RIM cells (Fig. 2a).
While the DNA damage checkpoint can operate throughout the cell cycle, cells also employ the DNA replication checkpoint (DRC) during S phase3. We next examined whether mild hypomorphic DRC mutants could affect the MMS sensitivity of mms22RIM cells. We tested two well-characterized alleles affecting either the Mrc1 mediator protein of the DRC pathway, namely mrc1-AQ, or mec1-100 that is specifically defective in DRC33, 34. As shown in Fig. 2b, neither allele affected the MMS sensitivity of mms22RIM cells. Together, these data suggest that defects of mms22RIM cells can be rescued by reducing DDC, but not DRC, functions. Collectively, the genetic findings raise the possibility that Mms22 binding to Rtt107 contributes to the downregulation of DDC but not DRC, and that this role can be partly responsible for the genotoxin sensitivity caused by the loss of the Mms22-Rtt107 interaction.
Persistent checkpoint in mms22-RIM cells is rescued by rad9-K1088M and ddc1-T602A
To further test the above hypothesis, we examined whether the suppression of MMS sensitivity of mms22RIM cells by the DDC mutants is associated with a correction of the persistent checkpoint seen in mms22RIM cells. We used a similar experimental scheme as depicted in Fig. 1c to monitor cell cycle progression and Rad53 activation, except a longer period of recovery was examined. Similar to observations described above (Fig. 1c), while wild-type cells were able to enter the next cell cycle after recovery from MMS treatment, mms22RIM delayed the entry (Fig. 2c and Supplementary Fig. 2a). This delay was quantified for two late time points (260 and 300 min; Fig. 2d). Significantly, this delay was improved by either rad9-K1088M or ddc1-T602A (Fig. 2c and Supplementary Fig. 2a). Compared with mms22RIM, more cells in the mms22RIM rad9-K1088M or mms22RIM ddc1-T602A double mutants exited G2/M and entered the next G1 phase between 200 to 300 minutes, with the most prominent differences seen at 260 and 300 minutes (Fig. 2d). rad9-K1088M and ddc1-T602A behaved similarly to wild-type cells, reflecting redundancy among DDC mediator functions (Figs. 2c and 2d, Supplementary Fig. 2a). Importantly, rad9-K1088M or ddc1-T602A also reduced the persistent Rad53 activation seen in mms22RIM cells (Fig. 2e and Supplementary Fig. 2b). Collectively, the suppression of prolonged Rad53 activation and cell cycle arrest seen in mms22RIM by rad9-K1088M or ddc1-T602A provides further evidence that Mms22 binding to Rtt107 plays a role in dampening the DNA damage checkpoint.
Mms22 and Slx4 act in parallel to dampen the DNA damage checkpoint
Rtt107 has an established role in DDC dampening through pairing with Slx414, 23. Rtt107 employs the same surface to engage with short motifs within Slx4 or Mms22, engendering mutually exclusive pairwise interactions29. A separation-of-function allele, slx4TTS (T423, T424, and S567A), has been constructed that specifically abolishes the Slx4-Rtt107 interaction without affecting protein level or binding to other known partners, such as the Slx1 nuclease29. We thus asked how the Mms22-Rtt107 mediated effect on DDC recovery is functionally related to that mediated by Slx4-Rtt107.
Applying the experimental scheme described in Fig. 2A, we found that slx4TTS led to a more pronounced delay in late S phase compared with mms22RIM (80 and 120 min; Fig. 3a). In contrast, slx4TTS showed a milder delay in re-entry into the next cell cycle compared with mms22RIM (240 and 300 min; Fig. 3a). These observations raised the possibility that the two Rtt107 interactors may differentially affect DDC at different stages of the cell cycle.
Significantly, the mms22RIM slx4TTS double mutant exhibited more severe defects in exiting the G2/M arrest in the first cell cycle than either single mutant (Fig. 3a). At the end of the time course (300 min), the double mutant showed significantly fewer cells exited the G2/M phase than either single mutant (Fig. 3a). A similar additive effect was seen when assaying for the active form of Rad53. The mms22RIMslx4TTS double mutant showed higher levels of active Rad53 than either single mutant at the two last time points of the time course (240 and 300 min; Fig. 3b). Results from cell cycle analysis and active Rad53 forms can be best explained by that Mms22-Rtt107 and Slx4-Rtt107 complexes acting in different pathways to downregulate the DNA damage checkpoint.
Additional evidence supports the independence of Mms22- and Slx4-mediated functions
Further supporting the functional independence of Mms22 and Slx4, we found that the double mutant of mms22RIMslx4TTS led to stronger MMS sensitivity than either single mutant (Fig. 3c). The additive effect of mms22RIM and slx4TTS was also seen when assaying the stability of the repetitive ribosomal DNA (rDNA) locus and of a non-rDNA locus during growth, suggesting that their independence is a general feature (Supplementary Fig. 3). We showed that each single mutant caused a 4 to 5-fold increase in marker loss at rDNA, while the mms22RIMslx4TTS double mutant led to a 10-fold increase (Supplementary Fig. 3a). Similarly, in the gross chromosome rearrangement (GCR) assay that assesses the stability of a non-rDNA locus, we detected a 59-fold increase of marker loss in the mms22RIMslx4TTS double mutant and a 16 to 20-fold increase in the corresponding single mutants (Supplementary Fig. 3b). These genetic data are in line with the results from two cell cycle checkpoint assays described above as well as previous biochemical findings that Mms22 and Slx4 partner with Rtt107 in a mutually exclusive manner29. Together, they strongly suggest that Mms22 and Slx4 represent parallel pathways in regulating both the DNA damage checkpoint and genomic stability.
While our data support the functional independence of Mms22 and Slx4, the two proteins have the potential to indirectly affect each other due to sharing a common partner, Rtt107, that facilitates the chromatin recruitment of both29. To examine this possibility, we asked if Mms22 and Slx4 affect each other’s chromatin association. We queried how the loss of the Slx4-Rtt107 interaction could affect Mms22 chromatin association and vice versa. We confirmed the reported findings that mms22RIM and slx4TTS each reduced its own chromatin association (Fig. 3d). In contrast, mms22RIM led to an increase in Slx4 level on chromatin while slx4TTS showed no effect on the chromatin association of Mms22 (Fig. 3d and Supplementary Fig. 3c). These results suggest that the checkpoint dampening defects seen for mms22RIM cells are not due to an indirect effect of reducing Slx4 chromatin association, and vice versa. This notion is consistent with data describing the additive effect of mms22RIM and slx4TTS shown above (Figs. 3a-3c); together they suggest that Mms22 and Slx4 can work in parallel pathways with each collaborating with Rtt107 (Fig. 3e).
The Mms22-Rtt107 interaction and the Rtt101 E3 promote Rad9 degradation
We next investigated the mechanisms by which the Mms22-Rtt107 interaction can promote DDC dampening. As Mms22 is a subunit of the Rtt101Mms1 ubiquitin E327, a likely means for it to downregulate the checkpoint is through protein degradation. Given the strong genetic suppression of mms22RIM by rad9 and ddc1 mutant alleles (Figs. 2a and 2c-2e, Supplementary Fig. 2), we examined the stability of these two proteins. We performed a standard procedure that utilizes cycloheximide (CHX) to block new protein synthesis, thus allowing the monitoring of protein stability during a time course. We first examined Ddc1 and Rad9 during normal growth. While the Ddc1 protein level showed minimal changes during an eight-hour time course, the Rad9 protein level exhibited a strong reduction over time (Figs. 4a and 4b, Supplementary Fig. 4a). Importantly, Rad9 instability is improved by the removal of Rtt101 or Rtt107, and by mms22RIM (Fig. 4a). Similar improvement was also seen in mms22∆ and mms1∆ cells (Supplementary Fig. 4b). Rad9 protein level quantification based on at least two biological isolates per genotype showed that all examined mutants stabilized Rad9 levels to a similar degree (Fig. 4b and Supplementary Fig. 4c). In contrast to this group of mutants, Rad9 degradation was not affected by the lack of Slx4 (Supplementary Fig. 4d). We thus conclude that the Rtt107-Mms22-Rtt101Mms1 axis but not the Rtt107-Slx4 axis affects Rad9 protein stability. As mutants of Rtt107, Mms22, and Rtt101 did not fully stabilize Rad9, additional means also exist to promote Rad9 degradation.
We next monitored Rad9 stability when cells were treated by MMS. Again, we observed Rad9 degradation in wild-type cells and this was reduced upon the removal of Rtt107, Rtt101, and in mms22RIM cells (Figs. 4c and 4d). This group of mutants also showed persistent Rad9 phosphorylation, which is catalyzed by the Mec1 kinase and serves as a marker for DDC activation (Fig. 4c)21, 35. These results are consistent with data presented above and further support the conclusion that the Mms22-Rtt107 interaction is required for degrading Rad9. Collectively, our findings suggest that the Mms22-Rtt107 interaction as well as the Rtt101Mms1 E3 are partly responsible for Rad9 degradation.
Rad9 degradation is mediated by proteasomes
To further test the above notion, we asked whether Rad9 protein instability is mediated by the proteasomes. If Mms22 and Rtt107 collaborate with the Rtt101Mms1 E3 in Rad9 degradation, we would expect that blocking proteasomal functions can hinder Rad9 degradation. We treated cells with MG132 that inhibits proteasomal activity, and with CHX that blocks new protein synthesis. In the control DMSO treatment, Rad9 protein level reduced during a four-hour time course (Fig. 5a). However, the Rad9 protein was stabilized in the presence of MG132 (Fig. 5a). This result suggests that Rad9 degradation is mediated by proteasomes. This conclusion is in line with the involvement of the Rtt101Mms1 E3 in Rad9 degradation in cells.
rad9-K1088M rescues the MMS sensitivity caused by the loss of Rtt101 and Mms1
We next examined the functional significance of the Rtt101Mms1 E3’s involvement in Rad9 degradation. If this role is important for cell survival in genotoxins, we would expect that as seen for mms22RIM mutant, genotoxin sensitivity caused by the loss of the Rtt101Mms1 E3 could be rescued by a mildly defective Rad9 allele. Indeed, we found that rad9-K1088M greatly increased the viability of either rtt101∆ or mms1∆ mutant cells on media containing MMS, CPT, or HU (Fig. 5b). This data provides evidence that like Mms22, Rtt101Mms1 E3’s involvement in Rad9 degradation can also be important for cellular survival in the face of genotoxins.
Mms22-Rtt107 interaction helps the Rtt101Mms1 E3 associate with chromatin and regulating Rad9 stability on chromatin
Our data thus far suggest that Rtt107 dampens DNA damage checkpoint signaling through binding to Mms22, in addition to binding to Slx4. Previous studies have shown that Rtt107 helps to recruit both Mms22 and Slx4 to chromatin via its ability to recognize γH2A14, 29, 36, 37. These findings raise the possibility that Rtt107 binding to Mms22 may help the chromatin association of the Rtt101Mms1 E3 for degrading Rad9 on chromatin.
We tested the above notion first by examining the chromatin association of Rtt101 and Mms1 when the Mms22-Rtt107 interaction is disrupted by mms22RIM. We used a well-established chromatin fraction method to separate chromatin and soluble fractions38. We found that mms22RIM not only reduced its own chromatin association but also lessened those for Rtt101 and Mms1 (Fig. 6a). We moved on to monitor Rad9 levels on chromatin and found that mms22RIM cells exhibited an increased amount of Rad9 on chromatin (Fig. 6b), suggesting that the Mms22-Rtt107 interaction is required for Rad9 loss from chromatin. Importantly, in the presence of CHX that allows the examination of protein degradation, we found that degradation of Rad9 in the chromatin fraction was largely blocked in mms22RIM cells (Fig. 6c). The Rad9 stabilization effect conferred by mms22RIM is much stronger for the chromatin pools of Rad9 compared with the whole cell extract (WCE) pool of Rad9 (Fig. 6c), suggesting that the main effect of Mms22-Rtt107 stems from the regulation of the stability of Rad9 in the chromatin fraction. Collectively, these data provide evidence that the Mms22-Rtt107 interaction is important for the chromatin association of the Rtt101Mms1 E3 and degrading Rad9 in the chromatin fraction.
Mms22-Rtt107 binding does not affect the Rad53-Asf1 or Rtt107-Dpb11 association
The Rtt101Mms1 E3 was previously shown to collaborate with Asf1 in down-regulating DDC when cells suffer from two double-strand DNA breaks (DSBs)39. The proposed model suggests that Rtt101Mms1 E3 may help Asf1 binding to Rad53, an interaction that could disfavor Rad53 phosphorylation thus leading to reduced DDC levels39. To discern if the Mms22 collaboration with Rtt101Mms1 during DDC dampening might be related to Asf1-Rad53 association, we examined this interaction by co-immunoprecipitation. We found that the amount of Asf1 recovered from Rad53 co-immunoprecipitation was similar between mms22RIM and wild-type cells, in both normal growth and MMS treated conditions (Supplementary Fig. 5a). This result suggests that the effect of the Mms22-Rtt107 interaction on DDC dampening is not via regulating the Asf1-Rad53 interaction. Finally, considering that Rtt107 also associates with another checkpoint factor, Dpb1123, we asked whether this interaction is perturbed in mms22RIM cells. We found that while mms22-RIM disrupted the Rtt107-Mms22 interaction, it did not affect the Rtt107-Dpb11 association in the presence or absence of MMS treatment (Supplementary Fig. 5b). These results suggest the DDC recovery role of the Rtt107-Mms22 interaction is not mediated by altering the Rtt107-Dpb11 interaction.