|Home | About | Journals | Submit | Contact Us | Français|
The cellular response to DNA damage employs multiple dynamic protein modifications to exert rapid and adaptable effects. Substantial work has detailed the roles of canonical checkpoint-mediated phosphorylation in this program. Recent studies have also implicated sumoylation in the DNA damage response; however, a systematic view of the contribution of sumoylation to replication and repair and its interplay with checkpoints is lacking. Here, using a biochemical screen in yeast, we establish that DNA damage-induced sumoylation occurs on a large scale. We identify MRX (Mre11-Rad50-Xrs2) as a positive regulator of this induction for a subset of repair targets. In addition, we find that defective sumoylation results in failure to complete replication of a damaged genome and impaired DNA end processing, highlighting the importance of the SUMO-mediated response in genome integrity. We also show that DNA damage-induced sumoylation does not require Mec1 checkpoint signaling, and the presence of both enables optimal DNA damage resistance.
Maintenance of genome integrity is crucial for cell survival. A key program in genome protection is the DNA damage response (DDR), which detects DNA lesions resulting from intrinsic or extrinsic sources and coordinates multiple cellular processes necessary for recovery. Canonical Mec1 (ATR)-mediated checkpoint signaling is the best understood aspect of this response, using a cascade of phosphorylation events to induce protective mechanisms, such as inhibition of late origin firing, stabilization of stalled replication forks, and cell cycle arrest (Putnam et al., 2009; Zegerman and Diffley, 2009). The effectiveness of these mechanisms relies on the fast and reversible nature of protein phosphorylation. Recently, other types of protein modification that are readily reversible and capable of dynamic signaling and regulation were also found to be important for genome maintenance (Bergink and Jentsch, 2009; Ulrich and Walden, 2010; Nagai et al., 2011). One such modification is sumoylation, which involves the covalent addition of the small ubiquitin-like modifier SUMO to one or more lysines of the target protein.
Sumoylation is catalyzed in a three-step process analogous to ubiquitination by E1, E2, and E3 enzymes, and desumoylation is catalyzed by proteases specifically recognizing SUMO conjugates (Kerscher et al., 2006; Geiss-Friedlander and Melchior, 2007). Most organisms employ a single E1 and E2 that are essential, and multiple E3s that have partly redundant functions. In budding yeast, three mitotic SUMO E3s have been identified, including two homologous Siz proteins (PIAS in humans) and the evolutionarily conserved Mms21 subunit of the essential Smc5/6 complex (Johnson and Gupta, 2001; Zhao and Blobel, 2005). Mutation or depletion of sumoylation enzymes in yeasts and humans results in defects in DNA repair, including recombination abnormalities and impaired double-strand break repair (Maeda et al., 2004; Branzei et al., 2006; Galanty et al., 2009; Morris et al., 2009). In addition, SUMO modifies the recombination protein Rad52, the DNA polymerase processivity factor PCNA, and the ssDNA binding complex RPA, thereby regulating recombination efficiency and pathway choices (Hoege et al., 2002; Papouli et al., 2005; Pfander et al., 2005; Sacher et al., 2006; Altmannova et al., 2010; Dou et al., 2010). In these examples as well as others, sumoylation influences substrate interactions, localization, or activity to achieve specific biological effects.
These individual cases show that sumoylation is important for DNA repair. However, a systematic view of the contribution of sumoylation to the DNA damage response is lacking. It is unclear whether there is a coordinated induction of protein sumoylation in response to increased DNA lesions, analogous to checkpoint phosphorylation, and if so, how sumoylation induction affects recovery and integrates with checkpoints as a whole. The above evidence and the observation that sumoylation of a few known substrates, such as Rad52, PCNA and RPA, is strongly induced by genotoxins led us to hypothesize that there is a sumoylation-based response to DNA damage that can induce the modification of many proteins needed for replication and repair to improve cellular survival. Increased sumoylation of relevant substrates could occur either as a result of checkpoint signaling or independently. We reasoned that if such a response operates in parallel with checkpoint, the combination of two independent signaling systems would strengthen the overall DDR program.
Here, we test these hypotheses in the model system Saccharomyces cerevisiae. Our results show that DNA damage-induced sumoylation is widespread among proteins that function in six DNA repair pathways and multiple steps of replication. This large-scale induction of sumoylation can operate in the absence of the Mec1 checkpoint. In addition, reduced overall sumoylation leads to multiple defects in DNA metabolism. Therefore, our results suggest that the cell employs a bifurcated system using different protein modification mechanisms to maximize the protection of the genome.
To test our hypothesis that DNA damage induces a sumoylation response affecting many replication and repair proteins, we set out to identify putative new substrates using a biochemical screen that detects sumoylation of individual proteins. We examined 179 proteins directly involved in replication and six major DNA repair pathways in yeast (Experimental Procedures and Table S1). Proteins were mostly tagged with TAP, with a few fused to other epitopes, and were expressed from the endogenous promoter to avoid possible artifacts caused by overexpression. Protein extracts were made under denaturing conditions to preserve sumoylated forms and prevent the copurification of interacting proteins (Ulrich and Davies, 2009). To maximize the detection of low-abundance proteins, we immunoprecipitated proteins individually and examined each one by western blotting with an antibody specific to the tag and another recognizing SUMO (see Experimental Procedures). Because a small percentage of substrate is typically sumoylated (Johnson, 2004), the anti-tag antibody often detects only the unmodified form at the exposures shown. Bands preferentially recognized by the anti-SUMO antibody are specifically decreased in SUMO E2 mutant ubc9 cells, indicating that they represent sumoylated forms (Figure 1A). Sumoylation patterns (mono- and poly- or multi-SUMO forms) vary among substrates, and mono-SUMO forms appear in expected positions approximately 20 kDa above the unmodified band (Figures 1A and 1B). To assess whether sumoylation, like Mec1 checkpoint-mediated phosphorylation, increases after DNA damage, the panel of strains was cultured with and without the DNA methylation agent methyl methanesulfonate (MMS), which is known to induce replication stress.
From this screen, we confirmed the sumoylation of all previously identified substrates, such as the recombination proteins Rad52 and Rad59, two subunits of RPA, and the nonhomologous end joining (NHEJ) protein Yku70, after MMS treatment (Table 1 and Figures S1A and S1B available online; Zhao and Blobel, 2005; Sacher et al., 2006; Burgess et al., 2007). In addition, we expanded the total number of SUMO substrates that function in different steps of replication and repair pathways to 67 (Table 1). Figures 1A and 1B show examples of proteins screened, including replisome subunits, DNA repair enzymes, and regulators of chromatin structure. Four subunits of the replicative helicase MCM show sumoylated forms (Mcm2, Mcm4–6), whereas no detectable bands other than the unmodified forms were found for the other two subunits, Mcm3 and Mcm7 (Figure 1A). Figure 1B shows the sumoylation of three other proteins involved in replication (the rDNA replication barrier component Tof2, replication origin-binding protein Abf1, and the Pob3 subunit of the chromosome remodeling complex FACT) as well as four representative repair proteins (Rad5 [postreplicative repair], Rad25 [nucleotide excision repair], Ogg1 [base excision repair], and Mlh1 [mismatch repair]). In summary, our results show that SUMO targets many proteins functioning in replication and DNA repair pathways, suggesting that sumoylation has a broader role in replicating and repairing the genome than was previously realized.
This expanded set of substrates allowed us to assess the extent of sumoylation after DNA damage compared to unchallenged conditions. We found that about half (37/67) of the identified targets were sumoylated only after MMS treatment (Table 1; e.g., Figures 1A, 1B, S1A, and S1B). In addition, the level of sumoylation for several constitutively modified proteins increased substantially in MMS compared to the untreated cultures (e.g., Figures 1A and 1B). However, MMS treatment does not induce sumoylation indiscriminately. For example, although sumoylation of several MCM subunits increased upon MMS treatment, that of other replisome components, such as Pob3, was not enhanced by MMS (Figures 1A and 1B). Overall, the number of proteins whose sumoylation increases in response to DNA damage rivals the number of Mec1/Tel1 phosphorylation substrates induced by checkpoint activation (Chen et al., 2010). The evidence that DNA damage-induced sumoylation is both widespread and directed to specific proteins points toward this response as an integral part of the DDR.
While a full understanding of the biological effects of DNA damage-induced sumoylation awaits detailed examination of each substrate, the importance of sumoylation in coping with genome insults can be addressed using a global approach that examines how sumoylation enzyme mutations affect cell survival, replication, and DNA repair. In accordance with previous reports, we found that the hypomorphic SUMO E2 mutant ubc9-10 is sensitive to MMS and the replication blocking agent hydroxyurea (HU) (Figure 1C; Jacquiau et al., 2005). In addition, we found that the removal of two nonessential SUMO E3s—Siz1 and Siz2—also leads to MMS and HU sensitivity (Figure 1C). Moreover, another E3 double mutant, siz1Δ mms21-CH, showed greater sensitivity to DNA damage than siz1Δ siz2Δ or the siz1Δ and mms21-CH single mutants (Figures 1C and S1C). The third E3 double mutant, siz2Δ mms21-CH, grew poorly and was not used (Figure S1D). These observations confirm and extend previous data and show that perturbing sumoylation using different mutants can produce a range of sensitivities to replication stress.
Consistent with the observed sensitivity to replication stress in sumoylation mutants, the set of replication proteins contains the largest number of sumoylation targets (Table 1). To assess how sumoylation affects replication at a global level, we compared the duplication of damaged genomes in wild-type and sumoylation-defective cells using pulsed-field gel electrophoresis (PFGE). G1 cells were treated with a pulse of MMS and then allowed to replicate in drug-free medium. While wild-type cells completed their chromosomal replication within 90 min, chromosomes from siz1Δ mms21-CH cells, the most MMS-sensitive sumoylation mutant, barely entered the gel even after 180 min of recovery, indicating that replication was not completed (Figure 1D). The accompanying FACS data show that the majority of siz1Δ mms21-CH mutant cells reached a 2N DNA content in a similar time to that of wild-type cells (Figure 1D). We infer that these cells can duplicate the bulk of the genome but fail to complete replication and/or perform replication-associated repair. In either case, these results show that Siz1/Mms21-mediated sumoylation is critical for complete duplication of a damaged genome.
Enzymes involved in recombination constitute another major category of sumoylated proteins (Table 1). Recombination at DNA double-strand breaks (DSBs) is initiated by degradation of the 5′ strand to form 3′ ssDNA overhangs. This resection step requires 12 proteins (Mimitou and Symington, 2011); significantly, 7 of them are sumoylated, including the subunits of the multifunctional MRX complex (Mre11, Rad50, and Xrs2), the nuclease Sae2, the helicase Sgs1, and 2 subunits of RPA. The sumoylation of all these proteins appears or increases after DNA damage, suggesting a potential role of induced sumoylation in DNA end resection (Figures S1A and S2A, Branzei et al., 2006; Burgess et al., 2007).
To test this directly, resection was examined using a physical assay that detects ssDNA formation distal to a single DSB introduced by the HO endonuclease (Figure 2A; see Experimental Procedures). Since ssDNA is resistant to digestion by restriction enzymes, resection can be monitored by the progressive disappearance of restriction fragments detected by labeled probes. In ubc9-10 cells, these fragments persist longer after HO cleavage compared with wild-type cells in both asynchronous and G2/M-arrested cultures, indicating that DNA end processing is impaired (Figures 2B, 2C, S2B, and S2C). Consistent with this, the phosphorylation of the checkpoint protein Rad53 that is elicited by increased amounts of ssDNA was also decreased in ubc9-10 cells upon HO induction compared to wild-type (Figures 2B and S2B). End processing occurs in two stages: initial end clipping by MRX/Sae2, followed by extensive resection involving either Sgs1 or Exo1 (Mimitou and Symington, 2011). Whereas defective end clipping due to the removal of Sae2 or MRX results in a delay in the disappearance of unresected fragments, deleting both SGS1 and EXO1 results in the accumulation of partially resected fragments (Figure S2D; Zhu et al., 2008; Mimitou and Symington, 2008). Because unresected fragments persisted, but no partially resected fragments accumulated in ubc9-10 cells, end clipping appears to be defective in these cells (Figure S2D). Taken together, these results show that defective sumoylation impairs DNA end resection and subsequent Rad53 phosphorylation.
The above results support the hypothesis that DNA damage-induced sumoylation of a large number of proteins improves cellular survival by facilitating both DNA replication and repair. An important question that follows is how cells induce sumoylation following DNA damage. In particular, are substrate modifications controlled individually or as a group? To begin to address this issue, we asked whether the key DNA damage sensor and repair complex MRX, which has been shown to affect sumoylation of Rad52, contributes to sumoylation of other recombination proteins. In accordance with a previous study, we found that the absence of Mre11, which disrupts the complex, decreased the sumoylation of Rad52 (Usui et al., 1998; Sacher et al., 2006; Figure 2D). In addition, mre11Δ reduced the sumoylation of Rad59 and the RPA subunits Rfa1 and Rfa2 (Figures 2D and 2E). However, sumoylation of the NHEJ proteins Yku70, Yku80, and Lif1 was not affected by mre11Δ, even though MRX is involved in NHEJ, nor was Sae2 sumoylation reduced (Figures 2F and S2E). Representative proteins from other repair pathways, such as Rad5, Rad25, Ogg1 or Mlh1, were also unaffected (Figure S2F). These results demonstrate that Mre11 or, by extension, the MRX complex is necessary for the MMS-induced sumoylation of a subset of proteins involved in recombinational repair, suggesting that DNA damage could induce sumoylation of multiple substrate “modules,” at least one of which requires MRX.
Since MRX is multifunctional, we next sought to understand which of its roles contributes to sumoylation induction. We found that cells lacking Sae2, which collaborates with MRX in end clipping, exhibited a sumoylation defect similar to mre11Δ cells, suggesting that initial end processing contributes to sumoylation induction (Figure 2G). In line with this, promoting Exo1-mediated extensive resection by deleting YKU70 did not rescue the sumoylation defect in mre11Δ cells (Figure S2G; Mimitou and Symington, 2010; Shim et al., 2010). Finally, cells lacking the Tel1 (ATM) kinase that mediates the checkpoint function of MRX did not alter the sumoylation of Rad52 or Rad59, and affected Rfa1 and Rfa2 sumoylation to a much lesser extent than mre11Δ (Figures 2D and 2E), indicating that MRX's role in DNA damage-induced sumoylation is mainly independent of Tel1. Like mre11Δ, tel1Δ also did not affect the sumoylation of NHEJ proteins (Figure 2F). These results collectively suggest that the initial DNA end resection role of MRX is important for sumoylation induction.
Since the main checkpoint kinase Mec1 transduces signals from DNA damage sensors, one possibility is that cells induce sumoylation via a Mec1-mediated mechanism. To test this, we examined whether the damage-induced sumoylation of repair proteins tested above requires Mec1. When a high dose of MMS was used, mec1Δ had little detectable effect on the sumoylation of any of the 16 proteins tested, indicating that DNA damage-induced sumoylation, at least for the tested substrates, does not rely on the Mec1 kinase (examples in Figures 3A and 3B). When a lower dose of MMS was used, mec1Δ dramatically increased the sumoylation of several repair proteins, notably those involved in DSB repair (i.e., recombination and NHEJ) but not other proteins (Figure 3A and 3B). As shown in Figure 4A, global levels of sumoylation also increased in mec1Δ cells, particularly following a low dose of MMS. Although the reasons for this increase are unclear (see Discussion), the fact that protein sumoylation is not reduced in mec1Δ cells indicates that canonical checkpoint activation is not required for the DNA damage-induced sumoylation of a range of proteins representing different repair pathways.
Although we found that checkpoint defects do not diminish sumoylation, results from Figure 2B showed that sumoylation defects do reduce Rad53 phosphorylation, a key readout of checkpoint activation, by impairing end processing at DSBs. To understand if such an effect applies to other damaging conditions, we examined checkpoint activation in MMS-treated cells. We examined the three sumoylation enzyme mutant strains used in Figure 1C that greatly reduce global sumoylation (Figure 4A). Like wild-type cells, none of these mutants had detectable Rad53 phosphorylation before MMS treatment, consistent with the healthy growth of these strains (Figure 4A). Upon MMS treatment, Rad53 phosphorylation was detected in all three mutants, although the ratio of phosphorylated to unmodified Rad53 was somewhat reduced (Figure 4A). Consistently, a slower onset of Rad53 activation was observed in these mutants when G1 cultures were released into MMS-containing medium (Figure 4B, see 30 min). However, this did not lead to obvious faster S phase progression as measured by FACS, a phenotype characteristic of checkpoint inactivation (Figure 4B; Paulovich and Hartwell, 1995). Thus, while normal sumoylation levels allow optimal Rad53 phosphorylation, the checkpoint response to MMS is largely intact in sumoylation mutants. These results suggest that sumoylation can affect checkpoint differently depending on the type of lesion. The differential effects on checkpoint by sumoylation mutants in MMS and HO-induced DSB conditions are consistent with the notion that ssDNA is generated differently in these situations and only the latter requires resection.
Of our panel of 67 sumoylated proteins, 18 are known to be phosphorylated by checkpoint kinases (~27%), based on individual tests and proteomic results (Table 1 and Figure S3; Paciotti et al., 1998; D'Amours and Jackson, 2001; Baroni et al., 2004; Smolka et al., 2007; Puddu et al., 2008; Makovets and Blackburn, 2009; Chen et al., 2010; Randell et al., 2010). In total, among 39 checkpoint kinase substrates tested, 18 were found to be sumoylated (~46%, Table S1 and Figure S3). We note that among 11 checkpoint signaling proteins, most of which are checkpoint kinase substrates, only 2 were sumoylated (Figure 4C, Table 1, and Table S1), suggesting that checkpoint signaling proteins as a group are largely unaffected by sumoylation.
The set of dual-modified proteins provides a possible intersection between the two DDR branches. On the other hand, the large number of unique substrates suggests that phosphorylation and sumoylation also have nonoverlapping roles in coping with DNA stress. To test this idea genetically, we performed epistasis analysis using MMS and HU sensitivity as the measured phenotype. When three different sumoylation mutants, siz1Δ siz2Δ, siz1Δ mms21-CH, and ubc9–10, were combined with MEC1 deletion, the resulting strains were more sensitive to MMS and HU than either mutant alone, strongly supporting the idea that sumoylation and checkpoint have separate functions in damage resistance (Figure 4D). In addition, although mec1Δ cells were more sensitive to HU than SUMO E3 mutants, siz1Δ mms21-CH cells grew more slowly on MMS-containing medium than mec1Δ mutants, suggesting a differential requirement for the two pathways in different damage conditions (Figure 4D).
Recent studies have implicated sumoylation in coping with DNA damage. To provide a systematic view of the contribution of sumoylation to the DNA damage response, we show that, in parallel to checkpoint phosphorylation, cells also induce sumoylation of many proteins needed for replication and repair in the presence of DNA damage. Consistent with a broader role for sumoylation, we found that effective sumoylation is important for replication under damage conditions and DSB end processing, as well as survival in the presence of genotoxins. Moreover, sumoylation induction does not require the key checkpoint kinases, and sumoylation and the Mec1 checkpoint function largely autonomously in MMS conditions. These findings support a model in which DNA damage-induced sumoylation (or DDIS) is an integral part of the DNA damage response, and the effects of checkpoint phosphorylation and sumoylation combine to generate a robust response to DNA damage (Figure 4E).
Our identification of many SUMO substrates at endogenous expression levels was made possible by biochemical examination of individual proteins. Because the sumoylation/desumoylation cycle is highly dynamic, and because sumoylated forms are prone to deconjugation during purification and are refractory to mass spectrometry-based proteomic approaches, identifying SUMO substrates has thus far been challenging (Wilson and Heaton, 2008; Jeram et al., 2009). Immunoprecipitating individual tagged proteins provides an alternative, perhaps a more sensitive and direct way to detect endogenous protein sumoylation, although we do not exclude the possibility that some substrates still elude detection. Quantification of a quarter of the substrates that exhibited moderate to high levels of sumoylation showed that the percentage of sumoylated forms ranged from 2% to almost 20% of the protein (see Experimental Procedures). In addition, specific sumoylation patterns were associated with each substrate, varying from monosumoylation (e.g., Mcm4, Mcm5, and Mlh1 in Figures 1A and 1B) to poly- or multisumoylation (e.g., Mcm2, Mcm6, and Rad25 in Figures 1A and 1B). While it is not entirely clear how these SUMO forms can differentially affect protein functions, poly- or multisumoylation may confer stronger affinity to proteins containing multiple SUMO interaction domains, thus favoring these interactions (Bruderer et al., 2011). Regardless of the forms of SUMO modification, sumoylation of some of the new targets likely modulates replication or repair processes, in light of the defects shown by mutants with global reductions in sumoylation. This extensive list of SUMO substrates enables future work to elucidate the specific roles of sumoylation in individual genome protection programs.
Our results suggest that cells can sense the insults from DNA damaging agents and respond by increasing protein sumoylation. We have identified MRX as a positive regulator of a subset of DNA damage-induced sumoylation events, specifically affecting proteins involved in recombinational repair (Figures 2D, 2E, 2F, and S2F). Because the end clipping function but not Tel1 activation is required for sumoylation induction (Figures 2D, 2E, 2F, and 2G), MRX may exert a local effect at DSBs or DNA lesions undergoing recombination. Since the role of MRX in supporting sumoylation appears to be targeted to a small group of substrates, cells likely rely on additional proteins besides MRX to induce the modification of other functional groups of SUMO substrates.
Additional positive regulators of DDIS are unlikely to include the canonical checkpoint kinase Mec1 because removal of Mec1 had a negligible effect on the sumoylation of all 16 proteins tested following treatment with a high concentration of MMS (e.g., Figures 3A and 3B). At lower concentrations of MMS, sumoylation of several repair proteins, particularly those involved in DSB repair, even increased in mec1Δ cells (Figures 3A and 3B). This could be due to a higher level of DNA lesions, such as DSBs, present in mec1Δ cells (Cha and Kleckner, 2002) or could be related to Mec1's known suppression of recombination foci (Lisby et al., 2004; Alabert et al., 2009); it may also be a compensatory effect, with increased sumoylation buffering the repair defects of checkpoint mutants. This effect might become less noticeable at high doses of MMS, when sumoylation would already be strongly induced in wild-type cells. The separation between DDIS and the Mec1 checkpoint is demonstrated by the finding that simultaneous defects in both SUMO and checkpoint pathways exacerbate DNA damage sensitivity (Figure 4D). However, there is a certain amount of interplay within the framework of SUMO and checkpoint responses working alongside each other. For example, the minor checkpoint defects in sumoylation mutants upon MMS treatment may indicate that the sumoylation of some DNA lesion detector proteins could contribute to optimal checkpoint activation (Figures 4A, 4B, and Table 1). In addition, a stronger impact on checkpoint activation by sumoylation was detected in the presence of DSBs, due to the effect on resection (Figures 2B and S2B).
In summary, we present evidence that DDIS occurs on a large scale, that effective sumoylation is important for coping with DNA damage, and that induced sumoylation requires at least one known DNA damage sensor but not key checkpoint kinases. Together, these data support the idea that a SUMO-based response is an integral part of the DDR and acts alongside checkpoint signaling to boost cells' ability to replicate and repair DNA under damage conditions. This work is likely to have important implications in human cells because sumoylation and pathways governing genomic stability are highly conserved between yeast and humans. Indeed, human homologs of some of the SUMO substrates identified here were recently found to be sumoylated (Golebiowski et al., 2009; Table 1). Future work examining the interplay between the SUMO and checkpoint responses in yeast and humans will lead to a more comprehensive picture of the DNA damage response. Given the increased complexity associated with both sumoylation and the checkpoint response in human cells, their relationship may be more intricate than in yeast. However, if these two systems make largely separable contributions, as seen here, an approach simultaneously targeting both these branches of the DDR, or ablating DDIS in checkpoint-deficient cells, may lead to more effective cancer treatment strategies.
Based on Gene Ontology (GO) annotation and literature, 185 proteins have primary roles in replication and repair, and 179 of them were examined here (97% coverage; Table S1). Strains containing TAP-tagged proteins used in the screen were obtained from Open Biosystems (Ghaemmaghami et al., 2003). Where TAP-tagged strains were missing or the TAP-protein could not be detected, the gene was alternatively tagged with Myc or Flag at its endogenous locus. Other yeast strains are listed in Table S2. Standard yeast protocols were used for strain construction, growth, medium preparation, spot assays, and synchrony experiments. Because siz1Δ siz2Δ leads to amplification of 2-micron plasmids and nibbled colonies (Chen et al., 2005), we examined cells lacking 2-micron. Removal of 2-micron was carried out as described (Tsalik and Gartenberg, 1998). MMS was added to log phase cultures for 2 hr before harvesting, at a final concentration of 0.05% or 0.3% as indicated.
Protein extracts and immunoprecipitates were prepared essentially as described (Hang et al., 2011). In brief, cells were disrupted by bead beating under denaturing conditions and diluted protein extracts were immunoprecipitated, using IgG-sepharose to pull down TAP-tagged proteins, or Protein G-agarose plus anti-Myc (9E10) or anti-Flag (Sigma) antibodies to pull down Myc- or Flag-tagged proteins, respectively. Immunoprecipitated proteins were washed and eluted with loading dye before separating by standard SDS-PAGE and immunoblotting with anti-Smt3 (SUMO) (Zhao and Blobel, 2005) and the appropriate tag-specific antibody. Due to their low abundance, sumoylated species are not detected by tag-specific antibodies at normal exposures but they are readily detected by anti-Smt3. Because the Fc region of the antibody interacts with the Protein A portion of the TAP tag, anti-Smt3 binds unmodified TAP-protein forms, but preferentially recognizes sumoylated forms. As such, the SUMO blot does not reflect the true proportion of sumoylated protein in the sample. The bands detected by anti-Smt3 are specific to the tagged proteins because untagged strains did not yield these bands on either TAP or SUMO blots (e.g., Figure 2D, UN), and because these bands are not present in samples from differently sized proteins, meaning that the proteins serve as controls for one another. Interacting proteins are not pulled down because cells are lysed in denaturing conditions. We quantified the percentage of sumoylated forms for 17 representative substrates. Protein blots probed with the anti-tag antibodies were scanned and the percentage of modified forms was quantified using the ImageGauge program. The level of sumoylated forms ranges from 2%–5% (Rad52, Rad59, Rfa2, Yku70, Ogg1) or 5%–10% (Rfa1, Lif1, Mcm4, Rad1) to 10%–20% (Mcm2, Mcm5, Mcm6, Rad25, Rad5, Pob3, Srs2, Tof2).
Cultures synchronized with alpha factor were released into 0.03% MMS for 1 hr, followed by neutralization with 5% sodium thiosulfate. Washed cells were allowed to recover in normal media and samples taken at appropriate intervals. Plugs were prepared from these samples; 1% Megabase agarose gels were run on a BioRad CHEF apparatus and stained with ethidium bromide according to published methods (Maringele and Lydall, 2006).
The DSB resection assay was performed as described (Mimitou and Symington, 2010). In brief, cultures were grown to early-mid log phase in YP-Lactate medium. HO was induced by the addition of 2% galactose and cell samples were harvested before and after induction at indicated time points. Genomic DNA was isolated, digested with XbaI and StyI, and separated on 1% agarose gels. DNA fragments were transferred to nylon membranes (Hybond-XL, GE Healthcare) and hybridized with radiolabelled DNA probes. The probes for 0.7 kb and 3 kb fragments were generated by PCR amplification of MAT sequences distal to the HO-cut site with coordinates 201176–201580 and 204184–204893 on chromosome III, respectively. Quantities of DNA loaded at each time point were normalized using a DNL4 probe (coordinates 334672–335378 on chromosome XV). Intensities of detected bands on Southern blots were analyzed using ImageGauge. DSB end resection was calculated as the ratio of the signal intensity at each time point relative to the signal 30 min after HO induction, and represents the mean of two independent experiments.
To assay checkpoint activation, cultures synchronized in G1 were released into 0.03% MMS and samples taken at intervals for 2 hr for FACS analysis and protein extraction. FACS was performed as in Zhao and Rothstein (2002). To detect Rad53 phosphorylation, TCA extracts of total cell lysate prepared as described (Foiani et al., 1994) were separated on standard SDS-PAGE gels and western blotted, followed by probing with anti-Flag (Sigma) or anti-Rad53 antibodies (a kind gift from Marco Muzi-Falconi). For G2/M arrest, cells were treated with 15 μg/ml nocodazole for 3 hr. For spot assays, log phase cultures were serially diluted 10-fold and spotted onto agar plates containing YPD media either alone or with the addition of the indicated concentration of MMS or HU. Plates were incubated at 30°C (or 35°C for the plate containing ubc9-10 strains in Figure 4D) and photographed after 24–48 hr.
We thank Lorraine Symington, Rodney Rothstein, John Diffley, Maria Pia Longhese, and John Petrini for providing yeast strains, Eleni Mimitou and Huan Chen for kind help in setting up DSB processing assays, Marco Muzi-Falconi for kindly donating the Rad53 antibody, and John Diffley for sharing unpublished results on Rfa1 sumoylation. We also thank Scott Keeney, Stewart Shuman, Isabel Lam, and all Zhao lab members for comments on the manuscript. This work was supported by the NIH grant R01GM080670 to X.Z.
SUPPLEMENTAL INFORMATION Supplemental Information includes three figures and two tables and can be found with this article online at doi:10.1016/j.molcel.2011.11.028.
The authors declare no conflict of interest.