Introduction

The DNA damage response promotes organismal survival of genotoxic stress by eliciting many cellular changes. This includes a critical signaling process termed the DNA damage checkpoint (DDC)(Enoch, Carr, & Nurse, 1992; Hartwell & Weinert, 1989; Lanz, Dibitetto, & Smolka, 2019). Upon encountering stress, the apical DDC kinases are recruited to DNA lesion sites by the damage sensing proteins and are then activated by activator proteins(Usui, Ogawa, & Petrini, 2001; L. Zou & S. J. Elledge, 2003). A universal DDC sensor is the RPA complex that binds with high affinity and specificity to single-strand DNA (ssDNA), whose abundance rises under almost all types of genotoxic stress(Maréchal & Zou, 2015). Using the budding yeast as an example, RPA bound to the ssDNA filament can interact with Ddc2, the co-factor of the Mec1 checkpoint kinase, and recruit the Mec1-Ddc2 complex to DNA lesion sites(Elledge, 1996; Foiani et al., 2000). Subsequently, Mec1 is activated and can phosphorylate multiple substrates(Harrison & Haber, 2006). Among these is a key downstream effector kinase Rad53, the phosphorylation of which turns on its own kinase activity and yields further phosphorylation. Collectively, the DDC kinases can modify hundreds of proteins, inducing multitude of pathway changes (Lanz, Dibitetto, & Smolka, 2019). For example, DDC can arrest the cell cycle progression to allow more time for genome repair, which is required for cellular survival (Sanchez et al., 1999; Wang et al., 2001).

While DDC activation is crucial for the cell to cope with genotoxic stress, its timely downregulation is equally vital (Waterman, Haber, & Smolka, 2020). Many changes induced by the DDC kinases need to be reversed to allow continued cell growth and recovery. For example, the DDC-mediated cell cycle arrest has to be lifted to allow proliferation and alternative DNA repair permitted at other cell cycle phases (Pellicioli et al., 2001; Waterman, Haber, & Smolka, 2020). Thus far, DDC downregulation has been mainly elucidated using the model organism yeast, with several strategies being identified (Waterman, Haber, & Smolka, 2020). One of these is mediated by the DNA helicase Srs2 that targets the most upstream step of the DDC pathway through removing RPA and the associated Mec1-Ddc2 from chromatin (Dhingra et al., 2021). The DDC dampening (anti-checkpoint) function of Srs2 was shown to be separable from its roles in removing the recombinase Rad51 from ssDNA (anti-recombination) during homologous recombination (Dhingra et al., 2021). Considering that other organisms also possess DNA helicases capable of removing RPA from ssDNA (Hormeno et al., 2022; Jenkins et al., 2021; Shorrocks et al., 2021), it is plausible that an anti-checkpoint strategy involving RPA antagonism may be present beyond yeast.

Currently, little is known about how checkpoint dampening can be temporally controlled to minimize interference with the initial DDC activation that must occur in the early phase of the DNA damage response. In the case of the Srs2-mediated checkpoint regulation, it is also unclear how the process can minimize superfluous removal of RPA from sites of other RPA-mediated processes, such as at ongoing replication forks and transcription bubbles. In addressing these questions, we hypothesize that Srs2-based RPA antagonism should be selective in location and timing, and that the regulatory elements within the Srs2 protein may contribute to the selectivity.

To test the above hypothesis and to define Srs2 features relevant to its RPA antagonism during DDC regulation, we examined six Srs2 protein sites involved in its post-translational modifications and interactions with other proteins. Our genetic analyses of mutations affecting each of these sites suggest that while most of the examined features do not affect RPA antagonism, Srs2 binding to PCNA, which marks the 3’ junction of ssDNA gaps, and Srs2 sumoylation contribute to RPA regulation in the context of DDC. Srs2 binding to PCNA was known to recruit Srs2 to replication perturbation sites, while the roles of Srs2 sumoylation were less studied but can affect rDNA recombination (Kolesar et al., 2012; Moldovan, Pfander, & Jentsch, 2007; Papouli et al., 2005; Pfander et al., 2005; Saponaro et al., 2010). Unexpectedly, we found that Srs2 PCNA binding and its sumoylation are functionally related, as Srs2 binding to PCNA promotes its sumoylation. Moreover, Srs2 sumoylation level peaks late in the response to genomic stress after the DDC kinases are hyper-activated, and this modification depends on Mec1. Collectively, our data support a model wherein Srs2 recruitment to DNA lesion sites via binding to PCNA and subsequent sumoylation in a Mec1-dependent manner help to counter RPA functions and the Mec1 checkpoint.

Results

Srs2 regulatory features and their disabling mutants

To understand which Srs2 attributes are relevant to RPA regulation during the DNA damage response, we examined mutant alleles that perturb known features of the Srs2 protein. Srs2 was reported to contain a helicase domain at its N-terminal region and three protein-protein interaction sites at its C-terminal region (Fig. 1A) (Niu & Klein, 2017). We have previously shown that the Srs2 ATPase-dead allele (srs2-K41A) fails to downregulate RPA and Mec1-mediated DDC, indicating a requirement of ATPase-driven stripping of RPA from ssDNA (Dhingra et al., 2021) . The reported Srs2 protein-bindings sites include a Rad51 binding domain (Rad51-BD), a PCNA interaction motif (PIM), and a SUMO interaction motif (SIM) (Fig. 1A) (Colavito et al., 2009; Kolesar et al., 2016; Kolesar et al., 2012). We used CRISPR-based editing to generate the srs2-ΔPIM (Δ1159-1163 amino acids) and the srs2-SIMmut (₁₁₇₀ IIVID ₁₁₇₃ to ₁₁₇₀ AAAAD ₁₁₇₃) alleles based on previous reports and examined an established srs2-ΔRad51BD allele (Δ875-902 amino acids) (Colavito et al., 2009; Kolesar et al., 2016; Kolesar et al., 2012). We note that despite Srs2 removal of RPA from ssDNA, an interaction between the two proteins has yet to be reported.

Figure 1.

We also queried post-translational modifications reported for Srs2 by mutating the corresponding modification sites (Fig. 1A). First, seven Cdk1 phosphorylation consensus sites of Srs2 were mutated to produce the srs2-7AV (T604V, S698A, S879A, S893A, S938A, S950A, and S965A) allele (Chiolo et al., 2005), which was previously shown to affect recombinational repair. Second, three sumoylation sites of Srs2 were mutated to generate srs2-3KR (K1081R, K1089R, and K1142R) (Saponaro et al., 2010). Finally, we mutated two serine residues (S890 and S933) to alanine to generate srs2-2SA, since phosphorylation at these sites was shown in proteomic studies as dependent on Mec1 and genotoxic treatments (Faca et al., 2020). Because the Srs2 C-terminal end contributes to protein functions (Antony et al., 2009; Colavito et al., 2009), the six mutant alleles described above were not tagged with epitopes. Using an anti-Srs2 specific antibody, we showed that all six alleles supported wild-type levels of Srs2 protein based on immunoblotting (Fig. 1B).

Mutating PCNA binding and sumoylation sites of Srs2 rescues rfa1-zm2 CPT sensitivity

Srs2 loss leads to persistent DDC and perturbed recombination, both of which contribute to srs2Δ sensitivity to genotoxins, including camptothecin (CPT) that traps Top1 on DNA and methyl methanesulfonate (MMS) that methylates DNA (Fig. 2A) (Niu & Klein, 2017). Significantly, our recent study has shown that RPA mutants that reduce its ssDNA binding affinity can rescue the persistent DDC levels in srs2Δ cells but not their recombination defects (Dhingra et al., 2021). rfa1-zm2 (N492D, K493R, K494R) is one of the RPA alleles examined that contains mutations at three DNA contacting residues in the Rfa1 subunit of RPA (Fan & Pavletich, 2012). An RPA variant complex with these mutations exhibits a two-fold increase in dissociation constant compared to wild type RPA (Dhingra et al., 2021). In cells, rfa1-zm2 supports normal Rfa1 protein level, cell growth, DNA replication, and recombinational repair (Dhingra et al., 2021). However, rfa1-zm2 and srs2Δ are mutually suppressive for genotoxic sensitivities and DDC abnormalities (Fig. 2B, S1A) (Dhingra et al., 2021). These data suggest that the srs2 and rfa1-zm2 genetic relationship can be used as a readout for the antagonism between Srs2 and RPA during the DNA damage response. Based on this rationale, we employed the genetic suppression as an initial readout in querying whether any of the six srs2 mutant alleles described above are related to RPA regulation. Given that Srs2’s role in DDC dampening was best examined in CPT (Dhingra et al., 2021), experiments were mostly carried out in this condition.

Figure 2.

We found that on their own, srs2-SIM^mut, -ΔPIM, and -7AV showed different levels of sensitivity toward CPT at a concentration ranging from 2 to 8 µg/ml, while srs2-ΔRad51BD and -3KR exhibited close to wild-type level of CPT resistance, which are consistent with previous reports (Colavito et al., 2009; Saponaro et al., 2010) (Fig. S1A). In addition, srs2-2SA, which has not been examined previously, did not show CPT sensitivity (Fig. S1A). Since rfa1-zm2 cells exhibited stronger CPT sensitivity than the tested srs2 alleles (Fig. S1A), we asked if the latter could suppress rfa1-zm2 CPT sensitivity. We found that both srs2-ΔPIM and -3KR improved the CPT resistance of rfa1-zm2 cells on media containing 2 µg/ml CPT (Fig. 2C). Rescue was also seen at higher CPT concentrations, with srs2-ΔPIM exhibiting a stronger effect than srs2-3KR (Fig. S1B). In contrast to srs2-ΔPIM and srs2-3KR, srs2-ΔRad51BD and -2SA showed no effect on rfa1-zm2 growth on media containing CPT, whereas srs2-7AV and -SIM^mut worsened it (Fig. 2D). The differential effects seen for the srs2 alleles are consistent with the notion that Srs2 has multiple roles in cellular survival of genotoxic stress (Bronstein et al., 2018; Niu & Klein, 2017). The unique suppression seen for mutants affecting Srs2 binding to PCNA and its sumoylation suggests that these features are relevant to RPA regulation during the DNA damage response.

Additional suppressive effects conferred by srs2-ΔPIM and -3KR toward rfa1 mutants

We next tested whether the observed srs2-ΔPIM and -3KR suppression was specific towards rfa1-zm2 or applicable to other rfa1 mutants that are also defective in ssDNA binding. While rfa1-zm2 affects the ssDNA binding domain C (DBD-C) of Rfa1, a widely used rfa1 allele, rfa1-t33, impairs the DBD-B domain (Fig. 2E) (Umezu et al., 1998). Unlike rfa1-zm2, rfa1-t33 additionally decreases Rfa1 protein level and interferes with DNA replication and repair (Deng et al., 2014; Umezu et al., 1998). Despite having these additional defects, CPT survival of rfa1-t33 cells was also improved by srs2-ΔPIM or srs2-3KR (Fig. 2E). As seen in the case of rfa1-zm2, srs2-ΔPIM conferred better rescue than srs2-3KR toward rfa1-t33 (Fig. 2E). When comparing the two rfa1 alleles, srs2 rescue was stronger toward rfa1-zm2, likely because rfa1-t33 has other defects besides reduced ssDNA binding (Fig. 2C, 2E). In either case, srs2 mutants only conferred partial suppression, consistent with the notion that Srs2 has roles beyond RPA regulation (Bronstein et al., 2018; Niu & Klein, 2017). Nevertheless, the observed rescue indicates that Srs2 PCNA binding and sumoylation are likely involved the regulation of RPA during the DNA damage response.

Srs2 regulation of RPA during DDC downregulation is not only observed under CPT treatment, but also in MMS treated conditions (Dhingra et al., 2021). Consistent with this observation, srs2-ΔPIM moderately improved MMS sensitivity of rfa1-zm2 and rfa1-t33, while srs2-3KR had a small effect on the MMS sensitivity of rfa1-zm2 (Fig. S1B, S1C). Collectively, the genetic results provide evidence that the roles of Srs2 binding to PCNA and sumoylation in RPA antagonism can be applicable to more than one genomic stress situations.

srs2-ΔPIM and -3KR rescue checkpoint abnormality of rfa1-zm2 cells

We next investigated whether srs2-ΔPIM and -3KR suppression of rfa1-zm2 pertained to DNA damage checkpoint regulation. We employed two readouts for DDC, including the levels of the phosphorylated and active form of Rad53 detectable by the F9 antibody (Fiorani et al., 2008) and cellular ability to exit the G2/M phase after CPT treatment. We have previously shown that srs2Δ cells exhibit increased levels of Rad53 phosphorylation and diminished ability to exit G2/M phase in CPT, and that both are suppressed by rfa1-zm2 (Dhingra et al., 2021). Interestingly, rfa1-zm2 leads to similar defects and these are rescued by srs2Δ (Dhingra et al., 2021). The mutual suppression seen in these assays is consistent with that seen for CPT sensitivity (Dhingra et al., 2021). We note that several other RPA ssDNA binding mutants, such as rfa1-t33, also exhibit phenotypes indicative of increased DDC levels (Pellicioli et al., 2001; Seeber et al., 2016; Smith & Rothstein, 1995; Lee Zou & Stephen J. Elledge, 2003). This is likely because compromised ssDNA protection can lead to increased levels of DNA lesions thus enhancing the activation of Mec1 and/or its homolog Tel1.

We reasoned that if srs2-ΔPIM and srs2-3KR compromise Srs2’s ability to remove RPA from DNA in cells, they should alleviate rfa1-zm2 cells’ defects in ssDNA protection and the associated DDC abnormalities. Indeed, while rfa1-zm2 exhibited about 5-fold increase in Rad53 phosphorylation levels in CPT compared with wild-type cells as seen before (Dhingra et al., 2021), srs2-ΔPIM and -3KR reduced this level by about 50% and 80%, respectively (Fig. 3A). Similar to earlier data, we observed that wild-type cells arrested in G₂/M after 1 hour of CPT treatment and about 25% of these cells entered G₁ after another hour (Fig. 3B) (Dhingra et al., 2021). The ability to exit from G₂/M phase was reduced in rfa1-zm2 cells so that only 13% cells moved to G1 phase after 2 h treatment with CPT, as seen before (Fig. 3B) (Dhingra et al., 2021). Significantly, srs2-ΔPIM and -3KR increased the percentage of rfa1-zm2 cells transitioning into the G1 phase (Fig. 3B). The suppressive effects seen in both assays were stronger for srs2-ΔPIM than srs2-3KR as seen in genotoxicity assays (Fig. S1B; 3A, 3B). Collectively, the genetic and DDC assay data are consistent with each other and suggest that Srs2 binding to PCNA and its sumoylation contribute to RPA antagonism in DDC regulation.

Figure 3.

srs2-ΔPIM is additive with a slx4 mutant defective in DDC recovery

We found that srs2-ΔPIM and srs2-3KR mutants on their own behaved normally in the two DDC assays described above, which differs from srs2Δ cells that exhibit increased levels of Rad53 phosphorylation and reduced G1 entry (Fig. 3A, 3B) (Dhingra et al., 2021). One plausible interpretation of this difference is that unlike srs2Δ, these partial srs2 mutants only mildly impair Srs2-mediated RPA regulation during DDC downregulation, such that their defects can be compensated for by other DDC dampening systems. To test this idea, we examined the consequences of removing the Slx4-mediated anti-checkpoint function in srs2-ΔPIM cells. Slx4 is known to bind the scaffold protein Rtt107, and their complex can compete with the DDC protein Rad9 for binding to damaged chromatin, consequently disfavoring Rad9’s role in promoting Rad53 phosphorylation (Ohouo et al., 2010; Ohouo et al., 2013). Slx4 uses an Rtt107 interaction motif (RIM) to bind Rtt107 and a previously established RIM mutant, slx4^RIM (T423A, T424A, and S567A), abolishes Rtt107 binding without affecting other Slx4 protein attributes (Wan et al., 2019). slx4^RIMwas tagged with a TAP tag at its C-terminal end and examined together with a control wherein the wild-type Slx4 was tagged in the same manner. Compared with the control, slx4^RIMexhibited mild sensitivity toward CPT (Fig. S2A). Significantly, when srs2-ΔPIM was combined with slx4^RIM, the double mutant exhibited stronger CPT sensitivity compared with corresponding single mutants (Fig. S2A).

We further examined DDC levels when cells were exposed to CPT using both Rad53 phosphorylation and ability to exit G2/M as readouts. Compared with Slx4-TAP cells, slx4^RIM-TAP cells exhibited 2-fold increase in Rad53 phosphorylation level, consistent with a role of RIM of Slx4 in downregulating DDC (Fig. S2B). In Slx4-TAP background, srs2-ΔPIM led to 3-fold increase in Rad53 phosphorylation level (Fig. S2B). As cells containing untagged Slx4 and srs2-ΔPIM did not show Rad53 phosphorylation changes (Fig. 3A), tagging Slx4 likely sensitizes srs2-ΔPIM. Significantly, the double mutant of srs2-ΔPIM and slx4^RIM-TAP exhibited further increase in Rad53 phosphorylation (∼5.5 fold) compared with srs2-ΔPIM and slx4^RIM-TAP single mutant (Fig. S2B). Congruous with this finding, the srs2-ΔPIM slx4^RIM-TAP double mutant also reduced the percentage of cells entering G₁ from G2/M arrest compared with their single mutants (Fig. S2C). Together, these three lines of evidence support the notion that defects in the Srs2-mediated DDC downregulation can be compensated for by Slx4-mediated mechanisms.

Srs2 binding to PCNA facilitates Srs2 sumoylation

We next addressed whether the two Srs2 features involved in RPA antagonism are functionally related. To this end, we tested if Srs2 binding to PCNA affected Srs2 sumoylation in cells. We confirmed that endogenous sumoylation of Srs2 could be detected in cell extracts after enriching SUMOylated proteins using 8His-tagged SUMO (8His-Smt3) (Fig. 4A). Briefly, proteins were extracted from cells under denaturing conditions to disrupt protein-protein association and prevent desumoylation, and sumoylated proteins were enriched on Ni-NTA (nickel-nitrilotriacetic acid) resin (Ulrich & Davies, 2009). Consistent with previous findings, sumoylated forms of Srs2 in the Ni-NTA eluate were detected on immunoblots using an anti-Srs2 antibody only when cells contained 8His-Smt3, where unmodified Srs2 was seen regardless of the presence or absence of 8His-Smt3 due to nonspecific binding to the resin (Fig. 4A) (Saponaro et al., 2010). These bands were only seen in wild-type cells but not srs2Δ cells (Fig. S3A). We found that Srs2 sumoylation level was reduced in srs2-3KR strains in CPT conditions, as expected from mutating the three major sumoylation sites (Fig. 4A) (Saponaro et al., 2010). Interestingly, srs2-ΔPIM cells exhibited a similar reduction in Srs2 sumoylation as seen in srs2-3KR cells (Fig. 4A). This result suggests that Srs2 sumoylation largely depends on Srs2 binding to PCNA, supporting the notion that the two events occur sequentially.

Figure 4.

Srs2 sumoylation peaks after maximal Mec1 activation and depends on Mec1

While RPA can engage with ssDNA generated in different processes, such as transcription and replication, only a subset of ssDNA regions is adjacent to 3’-junctions loaded PCNA. This is because PCNA loading occurs mainly during S phase and requires an ssDNA and dsDNA junction site permissive for loading (Moldovan, Pfander, & Jentsch, 2007). We thus reasoned that Srs2 binding to PCNA may prevent Srs2 from removing RPA bound to ssDNA that forms transiently or is not flanked by PCNA, such as within transcription bubbles, R-loop regions, or at ongoing replication forks. This could provide a means for spatial control of Srs2’s anti-RPA role. We explored whether this role could also be regulated temporally. As the genetic data described above suggest that Srs2 sumoylation acts downstream of Srs2 binding to PCNA in antagonizing RPA, we examined the timing of Srs2 sumoylation after CPT treatment.

We released G1 synchronized cells into media containing CPT and examined Srs2 sumoylation at several time points (Fig. 4B). A similar experimental scheme was used to monitor the level of Mec1 activation using Rad9 phosphorylation as a readout (Fig. 4C). Quantification of Srs2 sumoylated forms relative to its unmodified forms showed that its sumoylation level peaked between 1.5 and 2 h after cells were released into CPT (Fig. 4D). Quantification of phosphorylated forms of Rad9 relative to its unmodified form showed that checkpoint activation peaked 1 hour after cell release (Fig. 4D). That checkpoint level peaked before Srs2 sumoylation level peaked suggests a possible dependence of the latter on the checkpoint. In testing this idea, we found that mec1Δ cells, which contain sml1Δ to support viability, abolished Srs2 sumoylation (Fig. 4E). Neither sml1Δ alone nor removal of Rad53 affected Srs2 sumoylation (Fig. 4E), suggesting that Mec1’s effect on this modification does not require the downstream Rad53 effector kinase. As Mec1 is not generally required for protein sumoylation during the DNA damage response (Cremona et al., 2012), its involvement in Srs2 sumoylation presents a rather unique effect.

Examination of Mec1-S1964 phosphorylation in Srs2-mediated DDC regulation

The combined results that Srs2 sumoylation levels peak after those of Rad9 phosphorylation and that Srs2 sumoylation depends on Mec1 raised the possibility that Mec1 hyper-activation during the late part of the DNA damage response may favor Srs2’s role in DDC dampening. Recently, Mec1 protein bound to Ddc2 was shown to be auto-phosphorylated at Ser1964 in a late part of the DNA damage response after a single DNA break is induced by the HO endonuclease (Memisoglu et al., 2019). Further, the phosphorylation-defective mutant mec1-S1964A impaired checkpoint downregulation in this condition, leading to the model that Mec1-S1964 phosphorylation contributes to DDC dampening at least after HO-induced break (Memisoglu et al., 2019). We tested the idea that Mec1-S1964 phosphorylation is also involved in Srs2-mediated DDC regulation under CPT conditions.

First, we asked whether Mec1-S1964 phosphorylation occurs under CPT conditions using a time course scheme similar to the one used in testing Srs2 sumoylation. Ddc2-bound Mec1 was recovered after immunoprecipitating Ddc2 at each timepoint and the proteins were examined by immunoblotting. To detect Mec1-S1964 phosphorylation, we used an antibody raised against the peptide containing this modification (Memisoglu et al., 2019). The antibody detected the Mec1 S1964 phosphorylation band in immunoprecipitated fractions only when Ddc2 was tagged with the Myc tag but not when Ddc2 was untagged (Fig. S3B). Further, the Mec1 S1964 phosphorylation band was largely abolished in the mec1-S1964A cells. (Fig. S3B; 5A). Given that the antibody also detects a closely spaced background band (Fig. S3B; 5A), we calculated the relative level of Mec1 S1964 phosphorylation in each sample by normalizing to this band. Quantification results showed that Mec1 S1964 phosphorylation level peaked around the same time as seen for Rad9 phosphorylation, which is 1 hour after CPT treatment (Fig. 5A).

Figure 5.

After confirming that Mec1 S1964 phosphorylation also took place under CPT conditions, we tested whether the mec1-S1964A or the phosphorylation-mimetic mutant, mec1-S1964E, influenced rfa1-zm2 or srs2 growth on media containing CPT. If Mec1 S1964 phosphorylation aids Srs2-mediated RPA antagonism in CPT, we would expect mec1-S1964A to behave similarly to srs2 mutants in rescuing rfa1-zm2 CPT sensitivity, while mec1-S1964E should have the opposite effects. We found that mec1-S1964A, which showed normal CPT resistance on its own, did not affect the growth of rfa1-zm2 cells on CPT-containing meidua (Fig. 5B). mec1-S1964A also did not influence the growth of srs2-ΔPIM and srs2-3KR cells regardless of rfa1-zm2 (Fig. 5B, S5A).

Though mec1-S1964E on its own did not cause CPT sensitivity, it led to slight but reproducible slow growth in cells containing srs2-3KR and srs2-ΔPIM with or without rfa1-zm2 (Fig. S4B). We found that neither mec1-S1964A nor mec1-S1964E affected Srs2 sumoylation levels (Fig. S4C). Thus, Mec1, but not its S1964 phosphorylation, is required for Srs2 sumoylation. These data provide evidence that even though an HO-induced DNA break and CPT treatment both induce Mec1 S1964 phosphorylation, this modification appears to not affect Srs2-mediated DDC dampening in the former situation.

Genetic data suggest that known RPA phosphorylation sites do not affect DDC recovery

While we mainly focused on the identification of Srs2 features that contribute to RPA antagonism during DDC regulation, we also considered the possibility that certain regulatory RPA features may contribute to this regulation as well. In particular, RPA is known to be phosphorylated by Mec1 (Chen et al., 2010; Faca et al., 2020; Lanz et al., 2021). We thus tested whether this phosphorylation may render RPA more susceptible to be removed from ssDNA by Srs2. To this end, we mutated previously identified Mec1 phosphorylation sites on the Rfa1 and Rfa2 subunits. These include Rfa1 Ser178 and three sites on Rfa2 (S122, S187, S189) (Albuquerque et al., 2015; Chen et al., 2010). We generated corresponding non-phosphorylatable mutants, rfa1-S178A and rfa2-3SA (S122A, S187A, and S189A), and phosphomimetic mutants, rfa1-S178D and rfa2-3SD.

Using genetic suppression as a readout for possible involvement of RPA mutants in Srs2-based DDC regulation, we examined the interactions between these mutants and srs2Δ. If RPA phosphorylation promote its removal by Srs2, the SD mutants may show suppressive interactions with srs2Δ, while SA mutants would behave like srs2 mutants. We found that neither rfa1-S178D nor S178A, both of which showed normal CPT resistance on their own, affected srs2Δ cell growth on CPT-containing media (Fig. 5C). In contrast, rfa2-3SD or rfa2-3SA, which also behaved like wild-type regarding CPT sensitivity, sensitized srs2Δ growth on CPT-containing media (Fig. 5D). This negative genetic interaction is in contrast to the positive interaction seen between srs2Δ and rfa1-zm2 or rfa1-t33 (Fig. S1A) (Dhingra et al., 2021). The differential Srs2 genetic interactions among distinct RPA alleles likely reflect the multifunctional natures of Srs2 and RPA. Collectively, these analyses suggest that the known phosphorylation sites on RPA are unlikely to be involved in Srs2-based DDC regulation.

Discussion

Timely termination of DDC is crucial for cell survival, yet the underlying mechanisms are not well understood. Using the budding yeast model system, we investigated how the Srs2-mediated DDC dampening through antagonizing RPA can be regulated. Our previous work reports a requirement of Srs2’s ATPase activity in this role; here we examined six additional Srs2 features involved in either binding to other proteins or post-translational modifications. Our results suggest that among the examined features, only Srs2 binding to PCNA and its sumoylation are involved in RPA antagonism during DDC downregulation. Interestingly, these two features are functionally linked, as optimal Srs2 sumoylation requires Srs2 binding to PCNA. Further, the Srs2 sumoylation level peaks after Mec1 activity reaches its maximum and that Mec1 is required for this modification. We posit a model that can interpret both previous and current data (Fig. 6). We envision that Srs2 binding to PCNA that is loaded at the 3’ junction site flanking ssDNA gaps help to bring Srs2 in proximity to the RPA-ssDNA filament and promote Srs2 sumoylation, both favoring downregulation of the checkpoint (Fig. 6). These regulatory mechanisms could in principle help to minimize undesirable RPA loss from PCNA-free regions and premature RPA removal from DNA during checkpoint induction phase.

Figure 6.

Our initial assessment of six Srs2 mutants used genetic suppression of the rfa1-zm2 CPT sensitivity as a readout for Srs2 and RPA antagonism. The results suggest that RPA regulation involves Srs2 binding to PCNA and Srs2 sumoylation, but not Srs2 binding to either Rad51 or SUMO nor its phosphorylation at Cdk1 sites and two potential Mec1 kinase sites. Genetic analysis showed that the suppression conferred by mutating PCNA binding sites on Srs2 (srs2Δ-PIM) in rfa1-zm2 cells also extended to MMS conditions. Additionally, a similar suppression for another rfa1 mutant defective in ssDNA binding was seen. These results ruled out suppression being specific to one allele or one genotoxin. Data from two checkpoint assays further showed that srs2Δ-PIM and srs2-3KR suppressed DDC abnormalities in rfa1-zm2 cells, strengthening the conclusion that Srs2 binding to PCNA and its sumoylation help to counter RPA during the DNA damage response.

RPA coats ssDNA generated from many DNA metabolic processes; thus, selective RPA stripping can be beneficial in cells. Our data suggest that this selectivity could be partially achieved by Srs2 binding to PCNA. In principle, ssDNA regions not adjacent to stable 3’ ss-dsDNA junction sites, such as those formed during transcription and in the R-loop regions, could be spared from Srs2-mediated RPA stripping. In contrast, ssDNA regions adjacent to PCNA-loaded junction sites, such as at ssDNA gap regions behind the replication forks, can be better targeted for Srs2-mediated RPA removal. An involvement of PCNA binding in Srs2-based RPA regulation could also compensate for a lack of Srs2 interaction with RPA. This scenario differs from Srs2-based Rad51 stripping that requires their interaction (Antony et al., 2009; Colavito et al., 2009; Krejci et al., 2003). It is thus likely that though both the anti-recombinase and anti-checkpoint roles of Srs2 can benefit from selectivity, they are achieved via different means. Previous studies have suggested that both PIM and SIM aid Srs2 recruitment to sumoylated form of PCNA (Papouli et al., 2005; Pfander et al., 2005). Our data reveal that at least in CPT conditions, Srs2-PIM, but not Srs2-SIM, is involved in RPA regulation. This observation is congruent with phenotypic differences between Srs2’s PIM and SIM seen in other studies and here, such as much stronger genotoxic sensitivity of the latter (Fig. S1A) (Fan et al., 2023; Kolesar et al., 2016; Kolesar et al., 2012). Considering that only a small percentage of PCNA is sumoylated and mainly during S phase (Papouli et al., 2005; Pfander et al., 2005), it is probably advantageous for Srs2 not being restricted to sumoylated PCNA for efficient RPA regulation.

How could Srs2 binding to PCNA helps Srs2 sumoylation? While a full answer awaits future studies, we speculate that this interaction may render Srs2 more permissive for sumoylation and/or position Srs2 in proximity with its sumoylation E3 ligase Siz2 that binds RPA (Chung & Zhao, 2015; Kolesar et al., 2012). Considering that Mec1, Ddc2, and RPA contain SUMO interaction motifs (Psakhye & Jentsch, 2012), it is possible that Srs2 sumoylation could help enrich the helicase at ssDNA-RPA sites via interacting with these motifs, though other effects are also possible. Compared with srs2-3KR, srs2-ΔPIM showed better suppression of rfa1-zm2 defects. This indicates that during RPA regulation, PCNA plays additional roles besides promoting Srs2 sumoylation.

We found that Srs2 sumoylation in cells also requires Mec1, which does not promote sumoylation in general (Cremona et al., 2012). On the contrary, Mec1 loss causes increased sumoylation levels (Cremona et al., 2012). It is currently unclear how Mec1 can specifically promote Srs2 sumoylation, but this may involve Mec1-mediated phosphorylation of SUMO substrates and/or enzymes. A requirement of Mec1 for Srs2 sumoylation and the dynamic nature of sumoylation suggest that Srs2 sumoylation may be a potential timing regulator. It is possible that hyperactivation of Mec1 favors Srs2 sumoylation that may aid its own downregulation. While fully testing this idea requires time-course phosphoproteomic studies under genotoxin treatment, we examined a Mec1 auto-phosphorylation site implicated in its downregulation after induction of a single DNA break (Memisoglu et al., 2019). We did not find evidence that this is involved in Mec1 dampening in CPT condition, raising the possibility that distinct mechanisms may be used in different genomic stress conditions. We also found that two phosphorylation sites on Srs2 that showed Mec1 dependency in proteomic studies had no effect in the CPT situation (Albuquerque et al., 2015; Faca et al., 2020); however we note that these sites do not fit with the typical S/TQ Mec1 consensus sites, thus may not be the bona fide Mec1 sites. Other phosphorylation sites examined here include five Mec1 sites on Rfa1 and Rfa2 (Albuquerque et al., 2015; Chen et al., 2010). Genetic analyses did not support their roles in the Srs2 and RPA antagonism. Thus, future studies to comprehensively map DDC kinase phosphorylation sites on Srs2 and RPA, as well as their regulators, such as the RPA chaperone Rtt105 (Li et al., 2018), will be needed to test how the kinases can communicate with Srs2-based RPA regulation during the DNA damage response.

While we provide multiple lines of evidence in support of the involvement of Srs2 binding to PCNA and its sumoylation in RPA regulation during checkpoint downregulation, srs2-ΔPIM or -3KR alone behaved normally in checkpoint assays. The lack of a defect here can be due to buffering effects, both from not-yet characterized Srs2 features and from other checkpoint dampening pathways. We provided evidence for the latter since srs2-ΔPIM or -3KR showed additive phenotype when combined with a slx4 mutant defective in DDC dampening function. The presence of multiple checkpoint dampening factors that target different checkpoint modules highlights the importance of the process. Recent demonstration of other DNA helicases, such as the human HELB, HELQ, and BLM proteins, in stripping of RPA from ssDNA in vitro suggests that Srs2-like checkpoint downregulation may present in human cells as well (Hormeno et al., 2022; Jenkins et al., 2021; Shorrocks et al., 2021). Since RPA-ssDNA binding not only plays a key role in checkpoint, but also in many other processes, the Srs2 features uncovered here may help to better understand how RPA dynamics is controlled by helicases and other factors during additional processes beyond DDC.

Methods

Yeast strains and genetic techniques

Standard procedures were used for cell growth and media preparation. The strains used in this work are listed in Table S1and are isogenic to W1588-4C, a RAD5 derivative of W303 (Zhao & Blobel, 2005). Mutant alleles of srs2, rfa1, rfa2 and mec1 used in this work were generated by CRISPR-Cas9-based gene replacement, except for srs2-3KR that was introduced using a PCR-based method (DiCarlo et al., 2013). All mutations were verified by sequencing. Standard procedures were used for tetrad analyses and genotoxin testing. For each assay, at least two biological duplicates were used per each genotype. Cells were grown at 30 °C unless otherwise stated.

Cell synchronization and cell cycle analyses

Cells from log-phase cultures were treated with 5 μg/mL α factor (GenScript RP01002) for 1 h, followed by an additional 2.5 μg/mL α factor for 30 min. When at least 95% cells were arrested in G1 phase based on the percentage of unbudded cells, they were released into yeast extract–peptone–dextrose (YPD) media containing 100 μg/mL Protease (Millipore 53702) and 16 μg/mL CPT (Sigma C9911) for 2 h. Cell cycle progression was monitored by standard flow cytometry analyses as described previously (Zhao & Rothstein, 2002).

Protein extraction using trichloroacetic acid (TCA)

To examine the protein levels of Srs2, the phosphorylation form of Rad53, and Rad9, cell extracts were prepared as reported (Regan-Mochrie et al., 2022) . Briefly, 2 × 10⁸ cells were collected at indicated time points. Cell pellets were resuspended in 20% TCA and lysed by glass bead beating. The lysate was then centrifuged to remove supernatant. Precipitated proteins were suspended in Laemmli buffer (65 mM Tris-HCl at pH 6.8, 2% SDS, 5% 2-mercaptoethanol, 10% glycerol, 0.025% bromophenol blue) with 2 M Tris added to neutralize the solution. Prior to loading, samples were boiled for 5 min and spun down at 14,000 rpm for 10 min to remove insoluble materials. Samples were separated on 4-20% Tris-glycine gels (Bio-Rad 456-1096) to examine Srs2 level and active Rad53 form or 7% Tris-acetate gels (Thermo Fisher EA03555) to detect Rad9 phosphorylation.

Detection of Srs2 sumoylation

Sumoylated proteins were pulled down from cells containing 8His-tagged SUMO expressed from its endogenous locus using the standard Ni-NTA pull-down method as previously described (Ulrich & Davies, 2009). Briefly, protein extracts prepared in 55% TCA were incubated in buffer A (6 M guanidine HCl, 100 mM sodium phosphate at pH 8.0, 10 mM Tris-HCl at pH 8.0) with rotation for 1 h at room temperature. The cleared supernatant was obtained after centrifuging for 20 min and was then incubated overnight at room temperature with Ni-NTA resin (Qiagen 30210) in the presence of 0.05% Tween-20 and 4.4 mM imidazole with rotation. Beads were washed twice with buffer A supplemented with 0.05% Tween 20 and then four times with buffer C (8 M urea, 100 nM sodium phosphate at pH 6.3, 10 mM Tris-HCl at pH 6.3) supplemented with 0.05% Tween 20. Proteins were eluted from the beads using HU buffer (8 M urea, 200 mM Tris-HCl at pH 6.8, 1 mM EDTA, 5% SDS, 0.1% bromophenol blue, 1.5% DTT, 200 mM imidazole). Samples were loaded onto NuPAGE^TM 3– 8% Tris-acetate gels (Thermo Fisher EA03752) for immunoblotting to detect both sumoylated and unmodified Srs2. Equal loading was verified using Ponceau S staining.

Detection of Mec1 S1964 phosphorylation during CPT treatment

G1-arrested cells were filtered and resuspended in prewarmed YPD media containing 16 μg/mL CPT to allow entry into S phase. Samples were collected at indicated time points. Cells were disrupted by glass bead beating in lysis buffer, followed by centrifugation at 20,000 g for 15 min to obtain whole-cell extract. The lysis buffer consisted of 25 mM K-HEPES (pH 7.6), 100 mM NaCl, 100 mM K-glutamate, 5 mM Mg(OAc)₂, 0.02% NP40, and 0.5% Triton X-100 supplemented by protease inhibitor cocktail (Sigma P8215) and Complete Ultra EDTA-free protease inhibitor (Roche 11836145001). The activated form of Mec1 was enriched by immunoprecipitating myc-tagged Ddc2 using Protein A beads and an anti-Myc antibody (Thermo Fisher 9E10) for 2–4 h at 4°C. Beads were washed five to six times with lysis buffer, and proteins were eluted using Laemmli buffer (65 mM Tris-HCl at pH 6.8, 2% SDS, 5% 2-mercaptoethanol, 10% glycerol, 0.025% bromophenol blue). After boiling for 5 min, eluted proteins were loaded onto NuPAGE^TM 3–8% Tris-acetate gels for SDS-PAGE and subsequent immunoblotting analyses.

Immunoblotting analysis and antibodies

After SDS-PAGE, proteins were transferred to a 0.2-μm nitrocellulose membrane (GE, #G5678144) for immunoblotting. Pgk1 was used as a loading control. Antibodies used were anti-myc (Thermo, Fisher 9E10). anti-Srs2 (Santa Cruz, yC-18), anti-Pgk1 (Invitrogen, 22C5D8), F9 (a kind gift from Marco Foiani and Daniele Piccini, The FIRC Institute of Molecular Oncology, Milan, Italy) (Pellicioli et al., 2001), anti-Rad9 (a kind gift from John Petrini, MSKCC, NY, USA) (Usui, Foster, & Petrini, 2009), anti-Mec1-S1964-p (a kind gift from James E. Haber, Brandeis University, MA, USA) (Memisoglu et al., 2019).

Validation of antibodies is provided either by the manufacturers or in the cited references. Membranes were scanned with a Fujifilm LAS-3000 luminescent image analyzer, which has a linear dynamic range of 10⁴. The signal intensities of nonsaturated bands were measured using ImageJ software. For graphs, data are shown as mean and SD. Statistical differences were determined using two-sided Student’s t tests.

Acknowledgements

We thank Drs. James Haber, Marco Foiani, Daniele Piccini, John Petrini for sharing antibodies, and Tzippora Chwat-Edelstein for critical reading of the manuscript. X.Z. is supported by National Institute of General Medical Sciences (NIGMS) grants R35GM145260.

Author contributions

All authors were involved in research design and data analyses, J. F., N. D., T. Y., and V. Y. performed research. J. F. and X.Z. wrote the paper with all authors’ input.

Competing interest statement

The authors declare no competing interests.

Supplemental Materials

Figure S1.

Figure S2.

Figure S3.

Figure S4.

Table S1: