Profiling DNA damage-induced phosphorylation in budding yeast reveals diverse signaling networks

Significance

The DNA damage response (DDR) promotes survival and genome maintenance. It involves a network of kinases that phosphorylate a multitude of effector proteins. Although the protein kinases involved have been studied extensively, many targets remain to be discovered. We have used an unbiased approach to profile DDR phosphorylation in budding yeast. We reveal a link between DDR signaling and the metabolic pathways of inositol phosphate and phosphatidyl inositol synthesis, which are required for resistance to DNA damage. Taken together, these data shed new light on the organization of DDR signaling in budding yeast.

Abstract

The DNA damage response (DDR) is regulated by a protein kinase signaling cascade that orchestrates DNA repair and other processes. Identifying the substrate effectors of these kinases is critical for understanding the underlying physiology and mechanism of the response. We have used quantitative mass spectrometry to profile DDR-dependent phosphorylation in budding yeast and genetically explored the dependency of these phosphorylation events on the DDR kinases MEC1, RAD53, CHK1, and DUN1. Based on these screens, a database containing many novel DDR-regulated phosphorylation events has been established. Phosphorylation of many of these proteins has been validated by quantitative peptide phospho-immunoprecipitation and examined for functional relevance to the DDR through large-scale analysis of sensitivity to DNA damage in yeast deletion strains. We reveal a link between DDR signaling and the metabolic pathways of inositol phosphate and phosphatidyl inositol synthesis, which are required for resistance to DNA damage. We also uncover links between the DDR and TOR signaling as well as translation regulation. Taken together, these data shed new light on the organization of DDR signaling in budding yeast.

Genomes of all organisms constantly experience life-threatening chemical and structural alterations because of the highly reactive chemical environments in which they reside and endogenous errors that occur during genome replication (1, 2). The ability to repair these lesions and maintain genomic stability is critical to organismal survival. A failure to maintain this stability leads to deleterious events such as mutagenesis, chromosomal rearrangements, gene amplifications or deletions, and the gain or loss of entire chromosomes. These events reduce the fitness and may even endanger the life of unicellular organisms while leading to developmental abnormalities and tumorigenesis in metazoans. Selective pressure from these DNA insults has resulted in the evolution of a higher-order regulatory pathway referred to as the DNA-damage response (DDR), which emerged to coordinate repair processes both by directing the types of repair to be used for particular lesions and by coordinating this repair with other cellular events such as cell-cycle progression (1).

The eukaryotic DDR consists of two central protein kinases of the PIKK family, ataxia-telangiectasia mutated (ATM) and ataxia telangiectasia and Rad3-related (ATR) (3, 4). These proteins participate in sensing common intermediates in DNA repair, such as double-strand breaks or stalled replication forks, and thereupon phosphorylate downstream effectors to promote appropriate repair and cell-cycle coordination. The outlines of these pathways were originally established largely through studies in budding and fission yeast (1) and subsequently were expanded in mammals. In budding yeast, Tel1 (ATM homolog) and MEC1 (ATR homolog) carry out the bulk of the DNA-damage signaling. Unlike ATM in mammals, Tel1 has a relatively minor role in the budding yeast DDR (5). Downstream of these kinases is a second layer of protein kinases, the CHK1 and CHK2 (RAD53 and DUN1 in budding yeast) kinases (6, 7). These proteins, called checkpoint kinases, were identified based either on their ability to activate the transcription of DNA damage-inducible genes or their roles in arresting the cell cycle in response to DNA damage. In budding yeast, RAD53 has been shown to activate DUN1 (7, 8). Thus, one hallmark of DNA-damage signaling is a sequential activation of protein kinases.

Understanding a kinase-regulated signal-transduction pathway at the systems level requires a thorough knowledge of the phosphorylation substrates of these kinases. In budding yeast, multiple substrates of MEC1 and RAD53 have been identified, including RAD9 (9), MRC1 (10), RAD53 (8), RTT107/ESC4 (11), RFA1 (12), HTA1 (13), DUN1 (14), and CRT1 (15), among others. Previous proteomic studies in our laboratory and others have identified a multitude of proteins that are phosphorylated in mammals on consensus ATM/ATR sites and other motifs in response to DNA damage (16, 17). These results suggest that numerous substrates may exist in other organisms as well. While proteomic studies in Saccharomyces cerevisiae have indeed supported this (18–20), we predicted that many additional substrates remained to be discovered. We profiled DDR phosphorylation in budding yeast with the hope that discovery of new substrates would uncover pathways that reveal critical connections between species. We took a global approach using stable amino acid labeling in culture (SILAC) coupled with strong cation exchange (SCX) chromatography and immobilized metal affinity chromatography (IMAC) to profile phosphorylation quantitatively in yeast cells exposed to methyl methanesulfonate (MMS), hydroxyurea (HU), and ionizing irradiation (IR). We also determined the dependency of these events on the activity of MEC1, RAD53, DUN1, and CHK1. We identified over 15,000 unique phosphopeptides, more than 1,500 of which are increased more than fourfold in response to DNA damage or replication stress. Nearly 30% of DNA damage-induced or MEC1-regulated pS/TQ phosphorylation events were independently verified using phospho-antibody–mediated peptide immunoprecipitation (IP). A group of core DDR phosphorylated factors was identified by determining overlap among substrates phosphorylated in response to MMS, HU, and IR. Moreover, examination of sensitivity to DNA damage revealed a role for inositol phosphate and phosphatidyl inositol synthesis in facilitating DNA repair in budding yeast.

Results

Profiling DDR Phosphorylation in Budding Yeast.

To profile DDR-regulated phosphorylation in S. cerevisiae, we created a yeast SILAC strain (lys1∆arg4∆car1∆sml1∆, hereafter designated “WT”) to allow arginine and lysine labeling with stable isotopes, to suppress the conversion of arginine to proline with car1∆, and to suppress the lethality of mec1∆ and rad53∆ mutations with sml1∆. Isogenic mec1Δ-, rad53Δ-, dun1Δ-, and chk1Δ-null SILAC strains were generated in this background to determine dependency of phosphorylation events on DDR kinases (see Fig. S7 for strain information). Yeast cells were synchronized to control cell-cycle– dependent effects on protein phosphorylation and stability. We used α-factor arrest/release and nocodazole arrest to collect S-phase cells or mitotic cells, respectively. We then triggered DDRs with three different agents: MMS, HU, and IR. Cells cultured in light medium were treated with DNA damage, while cells grown in heavy medium were left untreated (Fig. 1A). MMS and HU were applied to S-phase cells to cause DNA alkylation or replication stress (Fig. 1B), whereas IR was applied to mitotically arrested cells to generate DNA double-strand breaks (Fig. 1C). We used flow cytometry to monitor cell-cycle progression and phospho-RAD53 immunoblotting to ensure activation of the DDR (Fig. 1 B and C and Fig. S1). Phosphopeptides were purified by SCX followed by IMAC and were identified by mass spectrometry. To determine the dependency of phosphorylation events on DDR kinases, we performed seven additional SILAC screens in which WT cells were labeled with heavy medium and kinase-mutant cells (mec1Δ, rad53Δ, dun1Δ, or chk1Δ) were labeled with light medium. Both light- and heavy-labeled cells then were subjected to DNA damage (Fig. 1D). Overall, we performed 10 screens, as detailed in Fig. S2. For simplicity, we named these screens by pairwise comparison as WT–MMS, WT–IR, WT–HU, MEC1–MMS, MEC1–IR, MEC1–HU, RAD53–MMS, DUN1–MMS, CHK1–MMS, and RAD53–MMS asynchronous (e.g., “WT–MMS” refers to the dataset from the WT strain with and without MMS treatment; “MEC1–MMS” refers to the comparison of WT and MEC1 mutant strains, both with MMS treatment). The total numbers of unique phosphopeptides and corresponding proteins from S-phase cells treated with MMS are summarized in Fig. 2A. Complete lists of phosphopeptides identified for all damage agents and for kinase-null screens are provided in Datasets S1–S3.

Fig. S7.

Description of yeast strains used in this study.

Fig. 1.

Phosphorylation profiling by SCX-IMAC and pS/TQ IP. (A) Schematic for the SILAC approach to profiling MMS-, HU-, or IR-induced phosphorylation events by SCX-IMAC and pS/TQ peptide IP (see Fig. S2 for details). (B) Strategy for S-phase synchronization before MMS or HU treatment (see Materials and Methods for details). Flow cytometry plots depict progression into S-phase before MMS or HU treatment. (C) Strategy for cell-cycle synchronization before IR treatment (see Materials and Methods for details). Flow cytometry plots depict progression into G2/M before irradiation. (D) Schematic for the SILAC approach to profiling phosphorylation events dependent upon the kinases MEC1, RAD53, DUN1, or CHK1. WT and kinase-null cells (mec1Δ, rad53Δ, dun1Δ, or chk1Δ) were subjected to the same type of DNA damage (see Fig. S2 for details). RAD53 phosphorylation detected by immunoblot was used to monitor DDR activation.

Fig. S1.

Analysis of cell-cycle synchronization and DDR activation conditions used in phosphorylation screens. (A) Morphology of yeast cells 3 h after the addition of α-factor (Fig. 1B). (B) Yeast cells were treated with 0.05% MMS or 200 mM HU and were harvested at the indicated times; lysates were immunoblotted for RAD53. RAD53 is fully phosphorylated at 1 h. (C) Cell-cycle profile for yeast cells released from α-factor arrest into regular medium or medium containing 0.05% MMS or 200 mM HU for the indicated times. Nondamaged cells progress through the cell cycle normally, whereas treatment with MMS or HU causes cell-cycle arrest in S-phase. (D) Morphology of yeast cells 2 h after the addition of nocodazole (Fig. 1C). (E) Yeast cells were treated with 120 Gy IR and were harvested at the indicated times; lysates were immunoblotted for RAD53. RAD53 is fully phosphorylated at 30 min.

Fig. S2.

Overview of phosphorylation screens performed using SCX IMAC. Ten phosphorylation screens were performed using SCX-IMAC. Datasets are named according to the type of DNA damage stimulus and yeast strain genotype. Three screens examined induction of phosphorylation by DNA damage (WT–MMS, WT–HU, and WT–IR), and seven screens examined the dependency of phosphorylation events on DDR kinases using kinase-null yeast strains (MEC1–MMS, MEC1–HU, MEC1–IR, RAD53–MMS, DUN1–MMS, CHK1–MMS, and RAD53–MMS asynchronous). For these seven screens, WT and kinase-null strains were subjected to the same type of DNA damage. For screens involving IR treatment, cells were arrested with nocodazole before irradiation. For screens involving MMS or HU treatment, cells were synchronized with α-factor before DNA damage, except for the RAD53–MMS asynchronous screen.

Fig. 2.

Phosphopeptide yields and quality assessment for MMS datasets. (A) Numbers of total and DDR-regulated (more than fourfold) unique phosphopeptides and phosphoproteins in datasets acquired from MMS-treated cells synchronized in S-phase. (B and C) Number (B) and ratio (C) of fourfold-regulated pS/TQ-containing peptides in the WT–MMS, MEC1–MMS, RAD53–MMS, DUN1–MMS, and CHK1–MMS datasets. The ratio corresponds to the number of fourfold-regulated phosphopeptides over the total number of phosphopeptides in each dataset. pS/TQ-containing peptides are elevated in the MEC1 dataset, consistent with the known pS/TQ substrate specificity of the MEC1 kinase. (D) Known direct MEC1 substrates RFA1, RFA2, and RTT107 are regulated in the WT–MMS and MEC1–MMS datasets but not in the RAD53–MMS, DUN1–MMS, or CHK1–MMS datasets, thus supporting the quality of our datasets (L/H ratio = light/heavy ratio).

Quality Assessment of Our Yeast DDR Phosphoproteome.

To assess the quality of our DDR phosphorylation database, we determined the frequency of regulated pS/TQ phosphorylation in the MMS datasets upon DDR kinase deletion. While the number of unique fourfold upregulated pS/TQ-containing peptides from WT–MMS and MEC1–MMS was 89 and 42, respectively, fewer unique pS/TQ DDR peptides were identified from the RAD53–MMS, DUN1–MMS, and CHK1–MMS datasets (13, 3, and 1, respectively) (Fig. 2B). This decrease in pS/TQ phosphorylation is consistent with the known pS/TQ substrate specificity of the MEC1 kinase. Because the total number of unique pS/TQ DDR peptides is influenced by the size of each phospho-dataset, we also calculated the percentage of unique fourfold upregulated pS/TQ peptides per total unique pS/TQ peptides in each dataset. In agreement with the total numbers of unique peptides, the percentage of DDR pS/TQ events also dropped from 23.18% (89 of 384) in the WT–MMS dataset and 12.07% (42 of 348) in the MEC1–MMS dataset to 1.65% (13 of 785) for RAD53–MMS, 0.92% (3 of 327) for DUN1–MMS, and 0.24% (1 of 419) for CHK1–MMS (Fig. 2C). Consistent with the order of kinase activation in yeast, we found that the total number of DDR-induced phosphopeptides decreased from 641 in the WT dataset to 396 in the MEC1-, 225 in the RAD53-, 65 in the DUN1-, and 33 in the CHK1-dependent datasets (Fig. 2A). Moreover, the percentage of fourfold-induced unique phosphopeptides also dropped from 6.4% for WT–MMS and 4.5% for MEC1–MMS to 1.7% for RAD53–MMS, 0.8% for DUN1–MMS, and 0.3% for CHK1–MMS (Fig. 2A).

The quality of our data is further supported by our finding that pS/TQ phosphorylation occurred on previously described sites of well-known MEC1 substrates: S178 of RFA1 (12), S122 of RFA2 (21), and S806 of RTT107/ESC4 (11). We found these sites to be highly upregulated (>16-fold) in response to MMS (Fig. 2D and Dataset S1). We also found significant reduction in the phosphorylation of all three sites upon deletion of MEC1 (Fig. 2D and Dataset S2). However, we observed no change in the phosphorylation of any of the sites (log₂ of light/heavy ratio ∼0) upon deletion of RAD53, DUN1, or CHK1 (Fig. 2D and Dataset S3). The lack of regulation in the latter datasets is consistent with direct MEC1 phosphorylation of these sites and confirms the specificity of the kinase-null screens.

Validation of DDR-Regulated Phosphorylation Substrates.

To determine whether the DNA damage-induced phosphopeptides in our screens represent bona fide DDR phosphorylation events, we took several approaches. First, we randomly selected and tagged candidate proteins with a C-terminal TAP tag or a 13Myc tag. We then examined the mobility of these proteins using SDS/PAGE gels embedded with 20-μM Phos-Tag (22), a phosphate-binding ligand that is incorporated into the gel and slows the migration of phosphorylated proteins. We found that GIS1 and TOF1 clearly showed slower migration in a DDR-dependent manner (Fig. 3A). Next, we overexpressed a group of N-terminal GST fusion proteins containing DDR-induced pSQ motifs, purified them using glutathione beads, and probed them with phospho-antibodies against human or mouse pS/TQ motifs. Two proteins, ISW2 and MLH1, could be detected by these antibodies, likely through cross-reaction with pSQ motifs on the examined proteins (Fig. 3B).

Fig. 3.

Validation of candidate DDR phosphorylation substrates. (A) Yeast cells expressing TAP-tagged GIS1 and Myc-tagged TOF1 were treated with 0.05% MMS for 1 h. Lysates were run on an SDS/PAGE gel embedded with Phos-Tag and blotted with the indicated antibodies. Gel mobility shift was used to monitor phosphorylation. (B) Yeast cells expressing GST-tagged ISW2 and GST-tagged MLH1 were treated with 0.05% MMS for 1 h. GST-tagged proteins were purified by glutathione pulldown and blotted with the indicated cross-reactive pS/TQ antibodies. Anti-pSMC refers to the antibody cocktail for SMC in Fig. S3. (C) Overlap of fourfold-regulated pS/TQ-containing proteins identified by IMAC and by a high-throughput validation approach using peptide IP with pS/TQ antibodies. See Figs. S3 and S4 for additional details. (D) Among 21 fourfold-regulated phosphoproteins in the IMAC datasets validated by pS/TQ IP with identical peptides, quantification ratios were strongly correlated (R = 0.93) between the IMAC and IP approaches.

To perform a higher-volume evaluation of our data, we attempted to validate our IMAC-identified phosphorylation events using an independent phosphopeptide purification approach involving pS/TQ-peptide IP (17). To simplify this analysis, we focused on IMAC-identified peptides observed in our six WT and MEC1 datasets (WT–MMS, WT–IR, WT–HU, MEC1–MMS, MEC1–IR, and MEC1–HU). These datasets contained a total of 84 proteins with fourfold-regulated phosphopeptides containing pS/TQ motifs. Our pS/TQ-peptide IP enrichment approach is outlined in Fig. S3. To increase the yield of phosphopeptides recovered from each peptide pool, we used a sequential pull-down strategy as described in Fig. S3. As summarized in Fig. 3C and Fig. S4, a total of 63 pS/TQ-bearing proteins from our IP approach were found to undergo a fourfold increase in phosphorylation upon DNA damage or a fourfold decrease in phosphorylation upon MEC1 deletion. Twenty-four of these proteins overlapped with the 84 pS/TQ-containing proteins identified by IMAC. Twenty-one of the 24 proteins were validated with identical pS/TQ sites on these DDR candidates. Among the 21 proteins with identical sites, quantification ratios were strongly correlated (R = 0.93) between the IMAC and IP approaches (Fig. 3D). Overall, nearly 30% (24 of 84) of our IMAC-identified pS/TQ DDR candidate proteins were verified by phosphopeptide IP (Fig. 3C and Fig. S4). These findings suggest a high degree of concordance between our IMAC discovery and IP validation datasets. Notably, our IP approach uncovered numerous substrates not identified by the IMAC strategy, including POL31/HYS2, UBP6, and SMC2.

Fig. S3.

Sequential phospho-IP approach to validate pS/TQ DDR phosphorylation. To validate IMAC-identified phosphorylation events, an independent phosphopeptide purification strategy involving pS/TQ-peptide IP was used. To increase the yield of phosphopeptides recovered from each peptide pool, a sequential pull-down strategy was performed with the indicated commercial pS/TQ antibodies. This figure is related to Fig. 3C.

Fig. S4.

Overlap of fourfold-induced pS/TQ proteins identified by IMAC and by phospho-IP. This figure identifies the fourfold-induced pS/TQ-containing phosphoproteins depicted in the Venn diagrams in Fig. 3C.

General DDR Signaling Triggered by Different Types of DNA Damage.

We next sought to identify substrates whose phosphorylation is induced by all three of the agents, MMS, HU, and IR. We found that 30–40% of proteins undergoing twofold-increased phosphorylation upon MMS, HU, or IR treatment were unique to that specific agent (Fig. S5). These agent-specific events may reflect differences not only in the types of induced DNA lesions but also in the cell-cycle phase in which damage occurred and stochastic fluctuations in the ability to detect low-abundance peptides between experiments. Notably, we found 133 proteins (Dataset S4) whose phosphorylation was induced twofold in response to all three agents (i.e., upregulated by twofold in each of the WT–MMS, WT–HU, and WT–IR datasets). These 133 phosphoproteins likely represent the most abundant and commonly used factors in DDR signaling. They include numerous core DDR components such as RAD53, DUN1, RAD9, MRC1, MRE11, RFA2, SLX4, MLH1, and MSH6. Phosphorylation of MLH1 and MSH6 in response to exogenous agents may indicate a role for mismatch repair (MMR) machinery in functions other than simple postreplicative mismatch correction, such as DNA recombination repair (23, 24), checkpoint activation in response to methylguanine lesions (25), or polymerase-η recruitment to sites of oxidative or methylguanine damage (26, 27). Beyond these known core factors, other proteins in this list may represent a group of general DDR effectors. These proteins include factors that regulate gene expression at multiple steps, including chromatin silencing (SIR4, ESC1), chromatin remodeling (ISW2 and ITC1), mRNA nuclear export (MLP1, NPL3, and HPR1), and gene transcription (CRT1/RFX1, MBP1, NDD1, and SWI6). Previously, we identified the repressor CRT1/RFX1 as a DUN1 substrate in response to MMS (15), and our screens confirmed dependency on this kinase (Datasets S3 and S4). Also included in our list of general DDR effectors are structural proteins, such as mitotic spindle factors (FIN1, BIM1, STU2, and SLK19) and nuclear pore components (NUP60, NUP1, NUP159, and NUP2), the latter of which is discussed in more detail below.

Fig. S5.

Overlap of proteins whose phosphorylation is induced by MMS, HU, or IR. The Venn diagram shows the overlap of proteins whose phosphorylation is induced more than twofold by MMS, HU, or IR. There are 133 phosphoproteins common to all three stimuli.

Pathways and Networks in Yeast DDR Signaling.

To determine pathways regulated by DDR phosphorylation, we performed gene set enrichment analysis on proteins whose phosphorylation was regulated more than twofold in response to DNA damage or DDR kinase deletion. We used data from all 10 IMAC screens (Fig. S2). To simplify our analysis, these 10 screens were reduced to seven datasets: three for the different DNA-damage agents (MMS, HU, and IR) and four for the kinase-null strains (MEC1, RAD53, DUN1, and CHK1). Pathway analyses on these seven datasets (Dataset S5) enriched for multiple repair processes, including homologous recombination (Fig. 4A and Fig. S6B), nonhomologous end-joining (Fig. S6D), and MMR (Fig. 4A and Fig. S6A) as well as cell-cycle checkpoint pathways (Fig. S6 A, B, and D). These analyses also enriched for DNA replication (Fig. 4A and Fig. S6 B–D), which we previously identified as a regulated pathway in the mammalian DDR (16). As shown in Fig. 4B, numerous phosphoproteins involved in replication were regulated in our screens. These proteins include replication initiation factors (CDC45, SLD3, and DPB11), components of the origin recognition complex (ORC1 and ORC2), MCM helicase proteins (MCM3, MCM4, and MCM6), DNA polymerase components (DPB2 and POL1), and the CDC7 kinase regulatory subunit DBF4. Phosphorylation of both SLD3 and DBF4 was dependent on RAD53 (Fig. 4B), as is consistent with a prior report (28), demonstrating the quality of our data.

Fig. 4.

Pathways and functional modules enriched in DDR-regulated phosphoproteins. (A) GO processes and functions significantly enriched (P < 0.02) (Dataset S5) among proteins phosphorylated (more than twofold) in the WT–MMS dataset. Relative enrichment is calculated as the percentage of regulated phosphoproteins with the particular GO annotation divided by the percentage of all S. cerevisiae proteins with the particular GO annotation. (B–F) Functional modules enriched among regulated (more than twofold) phosphoproteins. In each module, orange lines represent physical interactions, and green lines represent genetic interactions. Colored bars indicate whether phosphorylation of the protein was induced (more than twofold) by the indicated damage agent or inhibited (more than twofold) by deletion of the indicated kinase. Identified modules are DNA replication (B), mRNA nuclear export (C), translation (D), target of rapamycin (TOR) signaling (E), and mitotic exit (F).

Fig. S6.

Pathways enriched in DDR-regulated phosphoproteins. GO processes and functions significantly enriched (P < 0.02) (Dataset S5) among proteins phosphorylated (more than twofold) in the WT–HU (A), WT–IR (B), MEC1–MMS (C), and RAD53–MMS (D) datasets. Relative enrichment is calculated as the percentage of regulated phosphoproteins with the GO annotation divided by the percentage of all S. cerevisiae proteins with the GO annotation.

Similar to the general DDR effectors described above, we enriched for multiple pathways that regulate gene expression, including transcriptional activation (Fig. 4A and Fig. S6 A, B, and D), chromatin assembly (Fig. 4A and Fig. S6 A–D), and mRNA nuclear export (Fig. 4C and Fig. S6B). We also enriched for proteins involved in translation initiation and regulation (Fig. 4A). As shown in Fig. 4D, these proteins include numerous core translation-initiation factors: SUI3 (eIF2), GCD2/6 (eIF2B), TIF35 (eIF3G), TIF3 (eIF4B), TIF4631/2 (eIF4G), and CDC33 (eIF4E). Notably, CDC33 (eIF4E) binding to the 5′ methylguanosine cap of mRNA is necessary for the initiation of translation. The protein EAP1 (4EBP1) inhibits the start of translation by sequestering CDC33 (eiF4E) from other initiation factors. We found that EAP1 was phosphorylated by MMS and HU in a MEC1- and RAD53-dependent fashion (Fig. 4D). Interestingly, a previous study demonstrated phosphorylation of human 4EBP1 by ATM in response to insulin stimulation and IR (29). This study suggested that ATM-mediated 4EBP1 phosphorylation may dissociate it from eIF4E to initiate translation. MEC1- and RAD53-mediated phosphorylation of EAP1 in yeast may fulfill a similar role.

Related to translation, multiple factors involved in TOR signaling were also found to be phosphorylated in response to DNA damage. Components of both the TORC1 and TORC2 complexes were identified as DDR substrates (Fig. 4E). TORC1 regulates cell growth and protein synthesis in response to nutrient availability. The cellular functions of TORC2 are less well understood, but recent reports have implicated it in the maintenance of genome stability (30, 31). We found that more components of the TORC2 than of the TORC1 complex were regulated by DDR phosphorylation (Fig. 4E). Phosphorylated TORC2 components included TOR2, AVO1, AVO2, TSC11, and BIT61. The TORC2 substrates and effector proteins SLM1 and SLM2 also were regulated by DDR phosphorylation (Fig. 4E). These results support a role for TORC2 in the DDR and provide putative mechanisms by which the DDR may regulate TORC2 signaling. Additional pathways enriched in our DDR phosphoproteome include mitotic exit (Fig. 4F and Fig. S6 A and C) and inositol phosphate synthesis (Figs. 4A and 6B and Fig. S6C), as discussed below.

Fig. 6.

A role for inositol phosphate and phosphatidyl inositol synthesis pathways in the DDR. (A) Sensitivity of diploid strains bearing homozygous null mutations of the indicated inositol pathway genes to 0.025% MMS. (B) Diagram of inositol phosphate (IP) and phosphatidyl inositol (PI) synthesis pathways. Depicted are inositol metabolites and the enzymes responsible for their synthesis. Colored bars indicate whether phosphorylation of the enzyme was induced (more than twofold) by the indicated DNA damage agent or inhibited (more than twofold) by deletion of the indicated kinase. Sphingolipids include IPC, MIPC, and M(IP)₂C. DAG, diacylglycerol; IPC, inositolphosphoceramide; MIPC, mannose inositol phosphoceramide; M(IP)₂C, mannose-(inositol-P)₂-ceramide; PP, pyrophosphate; Ptd Ins, phosphatidyl inositol.

DDR-Regulated Phosphoproteins Are Enriched for Proteins Required for DNA Damage Resistance.

Our phosphorylation screens implicated hundreds of potentially new proteins in DDR signaling. To determine which proteins are functionally critical to the DDR, we examined sensitivity to multiple DNA-damaging agents upon deletion of these DDR candidates. A list of 258 nonessential genes was compiled from hits identified in all 10 IMAC screens (Fig. S2) with more than twofold regulation. Diploid deletion mutants were cherry-picked from the Saccharomyces Genome Deletion Project collection and assayed for sensitivity to four agents: 150 mM HU, 250 Gy IR, 0.025% MMS, and 100 J/m² UV. Diploid- rather than haploid-deletion strains were assayed, because the lithium acetate transformation process for deleting ORFs can introduce recessive mutations. These genetic analyses demonstrated that, among the 258 candidate genes, 108 (42%) were sensitive to at least one of these agents (Dataset S6A).

To determine whether our IMAC-based hits were enriched for damage sensitivity, 100 randomly picked diploid deletions were assayed also, and 18 (18%) of these deletion strains demonstrated sensitivity to at least one DNA-damaging agent. These findings indicate a greater than twofold enrichment for damage sensitivity among our IMAC-based hits relative to random picks. This twofold enrichment persisted when examining sensitivity to multiple (two, three, or four) agents or different levels (mild, moderate, or severe) of sensitivity (Fig. 5). Interestingly, a predominance of mutants exhibiting HU sensitivity was observed (33% of all mutants) (Dataset S6A); this result may be related to the fact that 70% of our IMAC screens involved administration of DNA damage during S phase (Fig. S2). Together, these findings validate the utility of our phosphorylation-based screening in identifying new DDR players.

Fig. 5.

DDR-regulated phosphoproteins are enriched for proteins necessary for resistance to DNA damage. Diploid deletion strains of DDR-regulated phosphoprotein candidates from our IMAC screen or from 100 randomly selected genes were examined for sensitivity to MMS, HU, IR, and UV. The degree of sensitivity was categorized as mild, moderate, or severe based on the most severe sensitivity to any one agent. Also indicated are the number of agents to which mutant strains are sensitive. Asterisks denote significant differences between IMAC candidates and randomly selected genes (P < 0.05, Student’s t test).

DDR Regulation of the Inositol Phosphate and Phosphatidyl Inositol Synthesis Pathways.

Among the 108 damage-sensitive diploid deletions, we noted three kinases of the inositol phosphate pathway: IPK2/ARG82, KCS1, and VIP1 (Dataset S6A). Furthermore, our bioinformatic analyses demonstrated that DDR phosphoproteins identified through IMAC were enriched for inositol phosphate and phosphatidyl inositol synthesis pathways (Fig. 4A and Fig. S6C). Members of these pathways, which include kinases, phosphatases, and lipases, have been implicated in a variety of cellular processes, including chromatin remodeling, mRNA export, and DNA repair (32). These gene products fall into three biological processes: sphingolipid synthesis, phosphatidyl inositol synthesis, and inositol phosphate synthesis. The biosynthetic pathways in which these proteins are involved, which members of these pathways are phosphorylated in response to DNA damage, and which kinases control these phosphorylation events are summarized in Fig. 6B. To understand further the role of inositol phosphates in the DDR, 25 nonessential genes that control the inositide “metabolome” (32) were cherry-picked from the diploid deletion collection and assayed for sensitivity to 150 mM HU, 250 Gy IR, 0.025% MMS, and 100 J/m² UV. Forty percent (10/25) of diploid deletions were sensitive to at least one type of damage (Dataset S6B), suggesting potential DDR regulation of the inositol phosphate and phosphatidylinositol pathways. Sensitive mutants included deletions of csg2, fab1, ipk2/ar82, ipt1, kcs1, plc1, sac1, csg1/sur1, vac7, and vip1 (Fig. 6A and Dataset S6B), six of which encoded DDR-regulated phosphoproteins in our screens.

Discussion

The DDR is a vast signal-transduction network that responds to life-threatening genomic insults and promotes processes that increase the probability of organismal survival. Phosphorylation is central to DDR signaling and is used to regulate the activity, stability, and localization of DDR proteins. Understanding how this network reorganizes cellular physiology to promote genomic stability and survival requires a deep knowledge of DDR phosphorylation. Here we combined SCX-IMAC and mass spectrometry to profile MMS-, HU-, and IR-induced phosphorylation in budding yeast. Cell synchronization was used to capture phosphorylation events that may be cell cycle phase-specific, such as those occurring when a DNA polymerase encounters DNA damage in S phase. Our screens additionally examined the dependency of DDR phosphorylation events on the kinases MEC1, RAD53, DUN1, and CHK1 through genetic deletion of these kinases. Moreover, we validated IMAC-identified phosphorylation events using an independent enrichment strategy involving pS/TQ-peptide IP.

Through these efforts, we have generated a large dataset of DDR phosphorylation substrates. These substrates are enriched for factors involved in DNA repair, cell-cycle control, DNA replication, transcription, chromatin assembly, mRNA export and processing, and translation. We found that 133 substrates were phosphorylated in response to the three stimuli, MMS, HU, and IR. These proteins may represent a set of general DDR effectors and include known DDR factors such as RAD53, DUN1, RAD9, MRC1, MRE11, RFA2, SLX4, CRT1/RFX1, TOP1, TOP2, MLH1, and MSH6. Clastogen-induced phosphorylation of MLH1 and MSH6 may indicate MMR involvement in functions other than canonical MMR such as recombination (23, 24) or DNA-damage sensing (25–27). Other general DDR effectors include transcription factors, mRNA regulatory proteins, chromatin remodeling factors, and structural proteins.

We found four nuclear pore components (NUP60, NUP1, NUP2, and NUP159) that were phosphorylated in response to all three agents, MMS, HU, and IR (Fig. 4C). Phosphorylation of all four components occurred in a MEC1- and RAD53-dependent manner (Fig. 4C), consistent with a prior report that NUP60, NUP1, and NUP3 are phosphorylated by RAD53 in vitro (20). A fifth nuclear pore component, NUP145, was phosphorylated upon treatment with MMS and IR in a MEC1-dependent but RAD53-independent fashion (Fig. 4C). In pathway enrichment analyses, these five proteins clustered with the mRNA nuclear export factors NPL3, HPR1, and MLP1, all of which were phosphorylated in response to DNA damage in a RAD53-dependent fashion (Fig. 4C). Deletion of either NPL3 or HPR1 is known to cause sensitivity to DNA damage (33). It has been proposed that these proteins promote genome stability through the proper formation of nuclear export-competent messenger ribonucleoprotein particles that promote the dissociation of mRNA from DNA and thereby prevent the collision of replication forks with R loops (33). Nuclear pore components may be targeted by DDR signaling to regulate this process or perhaps other functions. For example, the nuclear localization of many proteins is affected by DNA damage (34), and DDR kinase targeting of nuclear pore components may affect the localization of these proteins. Lastly, the nuclear pore is known to associate with damaged DNA, and this association promotes certain types of DNA repair (35). Phosphorylation of nuclear pore components may thus regulate repair processes that occur at the nuclear periphery.

Numerous players in the mitotic exit pathway were phosphorylated in our screens (Fig. 4F). Among these substrates are the separase inhibitor securin (PDS1) and multiple factors in the mitotic exit network (MEN) signaling pathway, including TEM1, CDC15, and MOB1. Additional substrates include components of the CDC14 early anaphase release (FEAR) network such as SLK19 and CDC5. Both the MEN and FEAR networks activate mitotic exit by stimulating the release of the phosphatase CDC14 from the nucleolar protein NET1 during anaphase. Importantly, CDC14 and NET1 also were detected as phospho-substrates in our screens (Fig. 4F). Previous evidence has shown that DNA damage impairs mitotic exit through both RAD53- and CHK1-dependent processes that promote the nucleolar localization of CDC14 (36). We found that, consistent with this report, phosphorylation of multiple components in the MEN and FEAR pathways depends on RAD53 and CHK1 (Fig. 4F).

We found an unexpected concentration of DDR phosphorylation events on phosphatidyl inositol and inositol phosphate kinases such as FAB1, PIK1, MSS44, ARG82/IPK2, KCS1, and VIP1 (Fig. 6B). Furthermore, among 25 nonessential genes that regulate the inositol “metabolome,” we observed that 40% were sensitive to DNA damage (Dataset S6B). These findings suggest that inositol metabolites promote DNA repair. Prior reports have implicated inositol signaling in the DDR. In a study nearly 20 years ago, we found that IPK2/ARG82 overexpression suppressed the lethality caused by RAD53 deletion (37), suggesting that it may act downstream of RAD53. Subsequently, the inositol metabolite IP6 was purified as a KU70/KU80-binding factor that stimulates nonhomologous end-joining in mammalian cells in vitro (38, 39). It also has been shown that the metabolite IP7 (5-PP-IP5) activates CK2 in mammalian cells to phosphorylate the TTT complex, which stabilizes ATM and thereby promotes p53-mediated apoptosis (40). Moreover, IP7 is necessary for efficient homologous recombination repair in mouse embryonic fibroblasts (41) and activates CRL4 in response to UV radiation by causing its dissociation from the COP9 signalosome (42). Our data provide potential regulatory mechanisms through which DDR kinases may influence inositol synthesis pathways. Because mammalian PI3K and the mammalian PI3 phosphatase PTEN (phosphatase and tensin homolog) have been shown to have critical roles in tumorigenesis, it will be important to understand further how inositol phosphate synthesis is linked to DDR signaling in yeast.

In summary, we have established a large DDR phosphorylation database in yeast, genetically examined the dependency of DNA damage-induced phosphorylation events on DDR kinases, and genetically interrogated phosphorylation substrates for roles in resistance to DNA damage. This database will serve as a useful resource for future studies of DDR pathways and mechanisms.

Materials and Methods

SILAC Sample Preparation.

Yeast SILAC strains (see Fig. S7 for detailed information) were cultured in either light synthetic complete (SC) medium supplemented with l-lysine and l-arginine or heavy SC medium supplemented with [¹³C₆,¹⁵N₂] l-lysine and [¹³C₆¹⁵N₄] l-arginine (Cambridge Isotope Laboratories). See Fig. S2 for the assignment of light versus heavy medium to DNA-damage stimuli and kinase-null strains. For MMS and HU treatment (Fig. 1B), cells were cultured in SC medium until OD_596nm = 1.5. They were then synchronized in G1 by adding α-factor. After 1.5 h, α-factor was added again. Cells were incubated for another 1.5 h and then were washed and released into medium containing 0.05% MMS or 200 mM HU for 1 h before harvest. For IR treatment (Fig. 1C), cells were cultured in SC medium until OD_596nm = 1.2. They were then arrested in mitosis by adding nocodazole to a final concentration of 15 µg/mL. After 1 h, an equal amount of nocodazole was added again. Cells were incubated for 1 h and then were irradiated with 120 Gy IR and harvested 30 min later. Harvested cells were resuspended in 8 M urea lysis buffer and were lysed by bead-beater disruption. Lysates were subsequently combined in a 1:1 ratio, reduced with 4.5 mM DTT, and alkylated with iodoacetamide. Proteins were digested with trypsin, acidified with trifluoroacetic acid (TFA), and clarified by centrifugation. The supernatant was then desalted using C18 solid-phase extraction cartridges, and peptides were lyophilized.

SCX/IMAC and pS/TQ IP Enrichment of Phosphopeptides.

SCX and IMAC were performed as previously described (16). Briefly, heavy- and light-labeled peptides (20 mg) were separated on a 9.4 × 200-mm column packed with polySULFOETHYL Aspartamide (The Nest Group). Twelve fractions were collected by salt gradient elution, desalted using 1-mL tC18 SepPak cartridges (Waters), and lyophilized. One-half of each peptide solution was resuspended in 200 µL IMAC buffer [250 mM acetic acid, 30% (vol/vol) acetonitrile] and enriched by 90-min incubation with 50 µL of IMAC resin (PHOS-Select Iron Affinity Gel; Sigma-Aldrich). The other half was resuspended in 200 µL of TiO₂ buffer [2 M dihydroxybenzoic acid, 50% (vol/vol) acetonitrile, 0.1% TFA] and enriched by 90-min incubation with 50 µL of Titansphere TiO₂ beads (GL Sciences). Beads were washed with IMAC or TiO₂ buffer, respectively, and were eluted with 50 mM Tris, 300 mM NH₄OH (pH 10.0). For pS/TQ IP, peptides were dissolved in IP buffer [100 mM 3-(N-morpholino)propanesulfonic acid (pH 7.2), 10 mM sodium phosphate, 50 mM NaCl], incubated with 50 µg anti-pSQ motif antibody (from Cell Signaling Technology or Bethyl) (Fig. S3), immobilized on Protein A Sepharose beads, and incubated overnight at 4 °C. Beads were washed three times with IP buffer and twice with water. Bound phosphopeptides were eluted with 0.015% TFA, lyophilized, and desalted by stage tip chromatography before LC-MS/MS.

Mass Spectrometry and Quantification.

Lyophilized peptides enriched by IMAC were dissolved in 5% acetonitrile/5% formic acid and were loaded using a Famos autosampler (LC Packings) onto a reversed-phase microcapillary column (100 µm i.d.) packed first with 5 mm of Magic C4 resin (5 µm, 100 Å; Michrom Bioresources) followed by 20 cm of Maccel C18AQ resin (3 µm, 200 Å; The Nest Group). The peptides were then separated using a gradient of 5–25% (vol/vol) acetonitrile in 0.125% formic acid over 95 min and were detected in a hybrid Orbitrap XL mass spectrometer (Thermo Fisher). MS/MS spectra were searched using Sequest with the following parameters: three missed tryptic cleavages; static modification of 57.02146 Da (carboxyamidomethylation) on cysteine; and dynamic modifications of 79.96633 Da (phosphorylation) on serine, threonine, and tyrosine, 15.99491 Da (oxidation) on methionine, 10.00827 Da on arginine, and 8.01420 Da on lysine. Matches were filtered to a false-discovery rate of <1% by simultaneous searching of a reverse-sequence database and linear discriminant analysis. Automated peptide quantification was performed using the Vista program (43). Phosphopeptides were required to have Vista confidence score greater than 80, and the sum of signal-to-noise values for the heavy and light peptides was required to be greater than 8.0.

Bioinformatic Analysis.

Gene set enrichment analysis was performed using custom Perl scripts and Gene Ontology (GO) terms for S. cerevisiae. Independent analyses were performed on eight different datasets: three for the different DNA damage agents (MMS, HU, IR), four for the kinase-null strains (MEC1, RAD53, DUN1, and CHK1), and one for the pS/TQ IP screen (Dataset S5). Enrichment terms from all eight analyses were then pooled together and regrouped into a single table of functional clusters representing data for our entire phosphoproteome. Using these composite clusters, we constructed a matrix indicating the number of proteins within each cluster regulated by each damage agent or each DDR kinase deletion (Dataset S5). Relative enrichment is calculated as the percentage of regulated phosphoproteins with the GO annotation divided by the percentage of all S. cerevisiae proteins with the GO annotation. Interaction modules in Fig. 4 were created with esyN using S. cerevisiae BioGRID physical and genetic interactions (44) and were annotated manually with colored bars denoting which damage agents or kinase deletions influenced the regulation of phosphorylation.

Validation of Individual Candidate DDR Phospho-Substrates.

N-terminal GST fusions of ISW2 and MLH1 were induced by 2% (wt/vol) galactose in synthetic defined (SD)-Ura medium for 4 h. MMS was added to a final concentration of 0.05%, and cells were harvested 1 h later. Ten mg of cellular lysates were incubated with 50 μL anti-GST antibody beads. Bound proteins were separated by SDS/PAGE and probed with phospho-antibodies as indicated in Fig. 3B. For the mobility shift experiments in Fig. 3A, TAP-tagged GIS1 or Myc-tagged TOF1 were run on 7.5% SDS/PAGE gels containing 25 uM AAL-107 (Phos-TAG) and were then immunoblotted with anti-Protein A or anti-Myc antibodies, respectively.

Analysis of Sensitivity to DNA Damage.

Diploid deletion strains for DDR-regulated phosphoproteins, inositol “metabolome” proteins, or randomly selected genes were grown to log-phase, normalized by cell number, and serially diluted by factors of 10. Cells were spotted onto rich medium alone or rich medium containing 150 mM HU or 0.025% MMS or were exposed to 250 Gy IR or 100 J/m² UV and allowed to grow at 30 °C for 3 d. The degree of sensitivity to damaging agents was categorized as mild, moderate, or severe based on cell growth.