Phosphatase Complex Pph3/Psy2 Is Involved in Regulation of Efficient Non-Homologous End-Joining Pathway in the Yeast Saccharomyces cerevisiae

One of the main mechanisms for double stranded DNA break (DSB) repair is through the non-homologous end-joining (NHEJ) pathway. Using plasmid and chromosomal repair assays, we showed that deletion mutant strains for interacting proteins Pph3p and Psy2p had reduced efficiencies in NHEJ. We further observed that this activity of Pph3p and Psy2p appeared linked to cell cycle Rad53p and Chk1p checkpoint proteins. Pph3/Psy2 is a phosphatase complex, which regulates recovery from the Rad53p DNA damage checkpoint. Overexpression of Chk1p checkpoint protein in a parallel pathway to Rad53p compensated for the deletion of PPH3 or PSY2 in a chromosomal repair assay. Double mutant strains Δpph3/Δchk1 and Δpsy2/Δchk1 showed additional reductions in the efficiency of plasmid repair, compared to both single deletions which is in agreement with the activity of Pph3p and Psy2p in a parallel pathway to Chk1p. Genetic interaction analyses also supported a role for Pph3p and Psy2p in DNA damage repair, the NHEJ pathway, as well as cell cycle progression. Collectively, we report that the activity of Pph3p and Psy2p further connects NHEJ repair to cell cycle progression.


Introduction
Among DNA lesions, double-stranded DNA breaks (DSBs) are regarded as the most severe form of DNA damage. The mechanisms for DSB repair are divided in two independent pathways, Homologous Recombination (HR), and Non-Homologous End Joining (NHEJ). HR utilizes an undamaged homologous template, preferably the sister chromatid or homologous chromosomes, to repair the broken sites of DSBs [1], [2], and is considered to be an error free repair pathway [3]. A more flexible alternative to the HR repair system is NHEJ [4], [5]. In NHEJ, the two broken strands of DNA can be ligated directly. Because NHEJ does not use a homologous template, there is a higher risk of errors in repair, which can result in mutations [6]. NHEJ is the main pathway to repair DSBs in mammals [7].
The NHEJ pathway is highly conserved from yeast to human. Yku70p and Yku80p are S. cerevisiae homologs of Ku70p and Ku80p, respectively, which bind to DSB ends; they form a ring which is required as a factor for protecting and stabilizing the broken ends of DNA from degradation. The MRX (Mre11p, Rad50p, Xrs2p) complex in yeast is homologous to MRN (Mre11p, Rad50p, Nbs1p) in mammalian cells. It forms a bridge between the two broken ends of DNA and brings the broken ends closer to each other preparing them for ligation. The MRX complex is recruited by Yku70/Yku80 to the site of a DNA break.
It is thought that Xrs2p is one the key protein for targeting of the MRX complex to the damage site, although both the complex and all individual members of the complex can bind to DNA directly [8], [9]. Recent evidence suggests that MRN may function in multiple steps of NHEJ in mammalian cells [10]. The Dnl4/Lif1 complex is the homolog of mammalian DNA ligase XRCC4 which has ligase activities. Lif1p interacts with Xrs2p and Dnl4p, and Dnl4p performs the ligation of DNA [11], [12], [13]. Nej1p binds to the Dnl4/Lif1 complex through an interaction with Lif1p. Although its exact role is still unclear, recent investigations suggest it is recruited to the site of break, interacts with DNA and participates in the final steps of ligation [14]. Plasmid repair analyses have demonstrated that NEJ1p is required for NHEJ to function at high efficiency [14].
The efficiency of NHEJ depends on a growing number of factors. For example, different histone acetyltransferases are shown to be required for NHEJ efficiency [15], [16]. Another study reported that NHEJ is dependent on different stages of the cell cycle; NHEJ activity NHEJ is higher in G1 compared to G2/M [17]. In a recent study, methylation of histone H3 lysine 36 was shown to enhance the efficiency of NHEJ [18].
Before committing to mitosis, cells pass through different cell cycle checkpoints. Checkpoints can be activated in response to DNA damage, incomplete DNA replication and damaged replication complexes. By recognizing DNA damage and regulat-ing cell cycle arrest, they delay cell cycle progression to provide additional opportunity for DNA repair. Defects in checkpoint function can cause genomic instability [19]. Temporal association between the cell cycle and DNA damage is thought to begin with Mec1p, a DNA damage dependent checkpoint gene [20]. Mec1p phosphorylates Rad9p [21], [22]. Phosphorylation of Rad9p further stimulates the activity of Mec1p to trigger several kinases including Rad53p and Chk1p [23], [24], [25]. The checkpoint Rad53p is a key protein in response to DNA damage. Activation of Rad53p up-regulates repair genes, down-regulates cyclins and delays cell cycle progression. It is shown that the phosphatase complex Pph3/Psy2 negatively regulates Rad53p activity by dephosphorylating it and allowing cell cycle progression to continue [26]. Recently, it was shown that the deletion of PPH3 reduced the ability of cells to complete DSB repair via HR [27].
Here, we report that the deletion of Pph3 and Psy2 reduces the efficiency of NHEJ in S. cerevisiae. We further illustrate that this activity appears connected to cell cycle regulation.

Plasmid Repair Assay
A unique XbaI restriction site was used for plasmid (p416) linearization and the repair assay was performed as in [15]. Each experiment was repeated at least five times.

Chromosomal Repair Assay
Chromosomal double stranded breaks were induced by exposing the cells to galactose. Serial dilutions of cells (10 23 -10 25 ) were exposed to galactose for 90 minutes to induce HO endonuclease and compared to those before exposure. Number of colonies formed before and after induction of HO endonuclease was used as a measure of survival and the efficiency of the cell to repair induced DSBs. Each experiment was repeated at least five times. For compensation experiments, gene overexpression in a single mutant background was generated by transforming the JKM139-based gene deletion strains using a corresponding plasmid carrying the target gene [33], or an empty vector as a control. To study NHEJ efficiency in different phases of the cell cycle, cells were synchronized in G1, S, and G2/M phases using drug treatment with 10 mg/ml alpha-factor, 0.2 M hydroxyurea, and 15 mg/ml nocodazole, for 2.5 hours, before exposure to galactose.

Drug Sensitivity Spot Test
A series of single and double deletion mutants grown to mid-log were diluted (10 22 -10 25 ) and spots of 15 ml of each dilution were placed on YPD plates containing 60 mM hydroxyurea (HU), 4 mg/ml bleomycin, or no drug as a control. Reduced colony size and numbers represented increased sensitivity.

Genetic Interaction Analysis
Genetic interaction between target genes and DNA damage array (DDA) was examined using a miniaturized version of Synthetic Genetic Array (SGA) analysis [34]. In miniaturized SGA (mSGA) a target gene is deleted or overexpressed (plasmid-based), in an alpha mating type strain and crossed to two arrays of 384 gene deletion strains, one for target genes (DD array) and the other random (as a control) [32]. Double mutant strains were scored for fitness as in [35], [36] with some modifications. In brief, average colony size (S ave ) was calculated by summing the size of all colonies on a plate and dividing by the total number (384). S ave was subtracted from each colony to derive a relative size for individual colonies. Each experiment was repeated three times and those colonies that had a reduction of 30% or more in two of the three repeats were deemed ''positive''. Synthetic sick interactions (positives) were categorized as follows: moderate (30-49% reduction), strong (50-69%), and very strong (70-99%), as in [37]. For conditional interactions, the above analysis was repeated in the presence of low (sub-inhibitory) concentrations of DNA damage-inducing drugs. Hits were confirmed by random sporulation. Synthetic dosage lethality (SDL) analysis was performed as above with the exception that overexpression plasmids were transformed into the above deletion arrays as in [37]. Gene classification on the basis of cellular process and function was performed by Yeast Features (http://software.dumontierlab.com/ yeastfeatures/), Yeast Genome Database (http://www. yeastgenome.org/) and GeneMANIA (http://www.genemania. org/).

Protein-Protein Interaction Prediction
Protein-Protein Interactions (PPIs) were predicted on the basis of co-occurring polypeptide regions as in [38]. An updated high confidence PPI database (approximately 55,000 interactions) was generated from published data (BioGRID: www.thebiogrid.org and DIP: www.dip.doe-mbi.ucla.edu). The analysis was performed at 99.95% specificity (a measure for false positive prediction) generating a sensitivity (percentage of interactions that can be identified from the total interactions that a protein makes) of 28% (in comparison to the sensitivity of 14.6% in [38] estimated by leave-one-out analysis. The local regions that mediate PPIs were predicted using PIPE-site algorithm [39].

Deletions of PPH3 and PSY2 Reduced the Efficiency of NHEJ in a Plasmid Based Repair Assay
To evaluate the activity of Pph3/Psy2 complex on the efficiency of NHEJ, a plasmid repair assay was utilized [40], [41]. Equal amounts of circular and linearized plasmids were transformed separately to both wild-type and deletion mutants for PPH3 and PSY2. Transformed cells were plated on a selective media in a way that only cells receiving an intact plasmid or cells capable of repairing a received digested plasmid would form a colony. In this case, DNA repair is limited to NHEJ because the break site has no homologous region within the genome of S. cerevisiae. The number of colonies formed from linearized plasmids is related to colonies formed from intact plasmids, and this ratio reflects the proportion of successful NHEJ events that have occurred. Previously, using this assay, deletion effects for the RSC2, a member of the RSC, chromatin remodeling complex [40], SIR2, SIR3, SIR4 proteins involved in telomere maintenance [41], and yeast histone acetyltransferase RTT109 have been evaluated [15]. Deletion of YKU70 or YKU80 reduced NHEJ efficiency to approximately 6% and has been used as a positive control [40], [15].
It was observed that efficiency of NHEJ for individual deletions of PPH3 and PSY2 was approximately 24% and 28%, respectively ( Figure 1). Deletion of both PPH3 and PSY2 had a NHEJ efficiency of approximately 25%. This data is in agreement with the involvement of Pph3/Psy2 phosphatase complex in efficient NHEJ of a plasmid DNA.

The Effect of Pph3p and Psy2p on Efficient NHEJ is Relevant in a Chromosomal Context
We subsequently sought to confirm the involvements of Pph3p and Psy2p in efficient NHEJ in a chromosomal context, using a JKM139 strain-based chromosomal break assay [29]. In this assay, the target genes are knocked-out in a JKM139 strain background and the viability of target gene deletion mutants are evaluated after exposure to galactose. JKM139 strain carries a GAL promoter in  Wild-type, Dpph3, and Dpsy2 cells (JKM139 background) were exposed to DSB inducing conditions for 90 minutes and allowed to form colonies (Figure 2A). Fractions of colonies formed before and after exposure to galactose were used as a measure of survival and were related to the ability of the cell to repair induced DSBs ( Figure 2A). As expected, Dpph3 and Dpsy2 strains had a reduced ability to survive when DSBs were induced compared to wild-type, further supporting the involvement of Pph3p and Psy2p in the efficiency of NHEJ.
Cell cycle dependency for Dpph3 and Dpsy2 strains was investigated by synchronizing the cells in G1, S, and G2/M phases by treating the cells with alpha-factor, hydroxyurea, and nocodazole, respectively, before HO endonuclease induction. It Pph3/Psy2 Complex Affects NHEJ PLOS ONE | www.plosone.org was observed ( Figure 2B) that Dpph3 and Dpsy2 strains had their lowest NHEJ efficiencies in S phase (1% and 14%, respectively). The significant reduction in the efficiency of NHEJ for Dpph3 strain appears to separate the activity of PPH3 from SPY2 during S phase. A possible explanation is that in S phase, in addition to its SPY2-dependent activity, PPH3 might also affect NHEJ efficiency through an additional pathway, which is independent of PSY2.

The Pph3/Psy2 Complex Functions in Association with Components of Cell Cycle
The PPH3/PSY2 complex is associated with a cell cycle checkpoint through dephosphorylation of the checkpoint protein Rad53p [26], [27]. Above, we showed that deletion of individual and both members of this complex reduced efficiency in NHEJ as measured by plasmid repair analysis. To determine if these results are in fact associated with checkpoint activity, the activity of other related checkpoint proteins was investigated for their effect on NHEJ using plasmid repair assay. We observed that NHEJ efficiency for deletion of CHK1 was 25%. Deletion of RAD53 [42], [43], which works in parallel with CHK1, reduced NHEJ efficiency to 57% (Figure 1).
Double mutant strains Dpph3/Dchk1 and Dpsy2/Dchk1 showed NHEJ efficiency of 19% and 20%, respectively. NHEJ efficiency for Dpph3/Drad53 and Dpsy2/Drad53 double mutant strains were 57% and 64%, respectively, which were similar to that for Drad53 (57%) single mutant suggesting that the effect of these two genes on NHEJ is likely within the same pathway as Rad53p. In this context, Rad53p appears to be upstream of Pph3p and Psy2p that the activity of these two proteins is dependent on the presence of Rad53p. A possible explanation is that deletion of RAD53 triggers a second parallel pathway, for example Chk1p-dependent pathway, which works independent of Pph3p and Psy2p. This second parallel pathway is not triggered when RAD53 is intact.

Overexpression of CHK1 can Recover DNA Damage Sensitivity Phenotypes in Dpph3 and Dpsy2 Mutants in JKM139
We also used the JKM139 strain to detect phenotypic compensation in a chromosomal assay. Overexpression of genes in the DNA damage repair pathways was evaluated for its ability to compensate for a phenotype caused by deletion of PPH3 and PSY2. In this way, genes that have compensating functions can be identified.
It was observed that overexpression of CHK1 compensated for the absence of either PPH3 or PSY2 in a JKM139 assay (Figure 2A). Such a recovery provides a strong support for a functional association for CHK1 with PPH3 and PSY2. This is explained by the activity of Chk1p being in a parallel pathway which is compensatory to that of Pph3p and Psy2p, in response to activation of Mec1 ( Figure 2C). Of interest, overexpression of RAD53 had a compounding effect on phenotypes of PPH3 and PSY2 deletions (Figure 2A and 2D). Deletion strains for PPH3 and PSY2 grew very poorly (sick phenotype) if RAD53 was overexpressed when DSB was induced. This observation is in accordance with the assumption that a certain equilibrium between ''enzyme and substrate'' can be important for cell viability [33] ( Figure 2D). Rad53p (substrate) is known to be dephosphorylated by Pph3/ Psy2 complex (enzyme). In this context, overexpression of the substrate in the absence of the enzyme caused a conditional sick phenotype. Overexpression of CHK1 or RAD53 alone did not affect the phenotype of a wild-type JKM139 strain.

Drug Sensitivity Analysis
It is expected that deletion of genes involved in DNA repair pathway might change (usually elevate) the sensitivity of their deletion strains to DNA damage-inducing drugs. We used drug sensitivity to bleomycin and hydroxyurea (HU), to further study   [44], [45]. Drad53 strain showed sensitivity to HU ( Figure S1) and Dpph3, Dpsy2, Dchk1 and Drad53 strains all showed increased sensitivity to bleomycin (Figure 3), confirming previously reported observations [46], [26], [27]. Double mutant strains Dpph3Dchk1 and Dpsy2Dchk1 had elevated sensitivity in comparison with single mutants. This is in agreement with the activity of Chk1p in a parallel pathway to Pph3p and Psy2p. Double deletion mutants Dpph3Drad53 and Dpsy2Drad53 showed similar sensitivity to bleomycin as Drad53. This can be explained by the activity of Pph3p and Psy2p which is dependent on the presence of Rad53p, as above.

Genetic Interactions Analysis for PPH3 and PSY2
A genetic interaction refers to phenotypes of overexpression/ deletion of two genes together that are not easily explained by the investigation of two single genes alone [47]. It reveals a higher order pathway association between genes and their functions. Since functionally related genes often genetically interact with one another [48], one way that the function(s) of a gene is studied is through the genetic interactions that it makes with other genes with known functions. In this context, genetic interactions are divided into two groups of negative and positive interactions. A more extreme phenotype for a double mutant than expected infers a negative or aggravating interaction, whereas in positive or alleviating interaction the phenotype of the double mutant is less severe. A negative genetic interaction is often observed when two genes interact through parallel pathways [49]. To further study the activity of PPH3 and PSY2 we examined their negative genetic interactions under standard laboratory growth condition and in the presence of sub-inhibitory concentrations of DNA damaging agents bleomycin and HU. In this way, conditional genetic interactions were investigated when DNA damage was induced. We used the method of synthetic genetic array (SGA) analysis [34] to examine sick phenotypes (negative interactions) for two miniarrays, one for DNA damage (DD) which is a collection of 384 deletion strains for genes associated with DNA damage response, DNA replication, cell cycle progression and other interesting genes whose products are localized to nucleus, and a second array that contains 384 random deletion strains, used as a control. Using a DD array, 25 and 12 synthetic sick interactions were observed for PPH3 and PSY2 respectively (Figure 4) in comparison to 4 and 3 in a random array. Illustrated in Figure 4 on the basis of their cellular process, the interacting genes can be grouped into two categories of cell cycle progression or DNA repair (or both) connecting the activity of PPH3 and PSY2 to both cell cycle progression and DNA repair with P values of 2.65610 211 and 2.95610 227 for PPH3 and 8610 213 and 5.98610 210 for PSY2, respectively, with the assumption that random array represents the global distribution of negative interactions. This is in agreement with the enrichment of negative interactions previously reported for PPH3, 6.78610 29 and 5.73610 26 , and PSY2, 5.73610 26 and 2.3610 24 , for response to DNA damage and cell cycle progression, respectively [50], [51]. Differences in the genetic interaction profiles for PPH3 and PSY2 may underscore their additional functions within the cell that are independent of each other. For example, unlike PSY2, PPH3 does not form negative genetic interactions with HR genes, suggesting that a previously reported role for PPH3 in HR [27] appears independent of PSY2.
Presence of sub-inhibitory concentrations of bleomycin (3 mg/ml; MIC = 7.5 mg/ml) or HU (45 mM; MIC = 150 mM) generated a number of previously unreported conditional negative interactions ( Figure 4). As expected, majority of these new interactions are linked to the DNA damage response. For example, MAG1 encodes for a 3methyl-adenine DNA glycosylase that initiates base excision repair, PMS1 encodes for a mismatch repair protein, and XRS2 is a DSB repair protein, among others. Of interest, CHK1 formed a conditional negative genetic interaction with both PPH3 and PSY2 in the presence of bleomycin. In the presence of HU, CHK1 also interacted with both PPH3 and PSY2. These conditional interactions are in agreement with a DNA damage dependent functional association for Pph3p and Psy2p with Chk1p.
Overexpression of certain genes may have no phenotypic consequence for a wild-type strain, however when a second gene is deleted, the same overexpression may result in an unexpected phenotype such as sickness or lethality. This type of interaction is termed Synthetic Dosage Lethality (SDL) [52], [53] and is often used to study the relationship between regulator and substrate where the overexpression of the substrate in the absence of the regulator often causes a severe phenotype [54], [55]. To study potential regulators for the Pph3/Psy2 complex, we examined the overexpression phenotypes for PPH3 and PSY2 on DD array in the presence and absence of sub-inhibitory concentration of DNA damage drugs bleomycin and HU as above. In the absence of DNA damage, overexpression of PPH3 or PSY2 did not form any SDL interactions. However, when DNA damage was induced, overexpression of either PPH3 or PSY2 formed SDL interactions with gene deletion strains for each of the three members of MRX complex MRE11, RAD50 and XRS2 (P value of 5.43610 216 ) ( Figure 5). This

Protein-Protein Interaction Prediction
Proteins often realize their function through interactions with one another. The overall profiles of such interactions can reveal information about the function as well as the cellular process in which proteins participate. Some protein-protein interactions (PPIs) are mediated by a finite number of short interaction motifs [60], [61]. Such interactions can be studied by examining the cooccurrence of small polypeptide regions which are significantly enriched in the dataset of high confidence interacting proteins [38], [62]. One advantage of this method is that the polypeptide regions which are responsible for a physical interaction between proteins can be identified [39]. Here, we examined the possible proteome-wide interactions that Pph3p and Psy2p make on the basis of short interacting motifs. Their predicted interaction partners, along with their proposed site of interactions, are represented in Table 1. A potential interaction between amino acids 16-231 for Pph3p and 317-337 for Psy2p was identified. Previously, Pph3p was reported to interact with Psy2p [26] however the region responsible for this physical association remained unclear. Similarly, interactions between Pph3p (amino acids 298-336) and Rad53p (amino acids 266-268), as well as Psy2p (amino acids 587-608) and Rad53p (amino acids 298-336) were proposed. A number of these interactions appear to be competing for the same binding site on Pph3p and Psy2p. Such competing interactions may function as regulators of activity. For example, Rrd1p, a cell cycle regulator that activates PP2A phosphatase, competes for the same region of Pph3p (amino acids 213-286) as Rad53p. Rrd1p abundance is reported to increase in response to DNA damage [63] and can potentially outcompete the interaction of Pph3p with Rad53p, preventing the dephosphorylation of Rad53p by Pph3p in response to DNA damage. Further investigations are needed to examine the validity of this model of regulation for Pph3/Psy2 activity. The interacting partners of Pph3p and Psy2p can be grouped into two general categories of DNA damage response and cell cycle progression. These PPI profiles are in agreement with the activity of Pph3/Psy2 in regulating the cell cycle in response to DNA damage.

Concluding Remarks
In this study, we show that interacting proteins Pph3p and Psy2p affect the efficiency of NHEJ in the unicellular budding yeast S. cerevisiae. Deletion of either PPH3 or PSY2 genes reduced NHEJ efficiency both in the context of chromosomal and plasmid DSB repair. Pph3p and Psy2p form a phosphatase complex, which dephosphorylates Rad53 checkpoint kinase [26]. Our analyses using a plasmid repair assay suggested a functional connection between the activity of Pph3p and Psy2p on NHEJ through checkpoint protein Rad53. Similarly, phenotypic suppression analysis revealed that overexpression of Chk1p, another checkpoint kinase that works in parallel to Rad53p, compensated for the absence of either PPH3 or/and PSY2 genes in a chromosomal based repair assay. Double deletion mutant strains for either PPH3 or PSY2 with CHK1 showed additional reduction in the efficiency of plasmid repair through NHEJ than single mutant. Our genetic interaction analyses revealed synthetic sick phenotypes for both PPH3 and PSY2 with DNA damage response genes that function in regulation and upstream to DNA damage repair pathway, in addition to genes involved in cell cycle progression. This observation is in clear agreement with the activity of Pph3/Psy2 in cell cycle progression and further indicates that their effect on NHEJ is not at the mechanistic but rather at the regulatory level. This is consistent with previously reported activity of Pph3p in HR pathway [27]. In support of a role for Pph3/Psy2 in regulation of DNA damage response via cell cycle, the PPI analysis reported here suggested that both Pph3p and Psy2p interact with both DNA damage response and cell cycle progression proteins.
Dephosphorylation of Rad53p by Pph3/Psy2 releases cell cycle arrest. Pph3p also dephosphorylates cH2AX which regulates DNA damage checkpoint proteins activity. This regulation is through chromatin modification [46]. A recent study by Kim et al. reported a role for Pphp3 in DSB repair through HR [27]. Here, we show that Pph3p and Psy2p also affect the efficiency of NHEJ. We also present genetic evidence for conditional cross-talk and functional associations between Pph3p and Psy2p with checkpoint kinases Rad53p and Chk1p. These associations can be triggered by bleomycin, HU and HO endonuclease. Figure S1 Strain sensitivity analysis to 60 mM hydroxyurea (HU). (PDF)