Regulation of Elg1 activity by phosphorylation

Abstract

ELG1 is a conserved gene with important roles in the maintenance of genome stability. Elg1's activity prevents gross chromosomal rearrangements, maintains proper telomere length regulation, helps repairing DNA damage created by a number of genotoxins and participates in sister chromatid cohesion. Elg1 is evolutionarily conserved, and its Fanconi Anemia-related mammalian ortholog (also known as ATAD5) is embryonic lethal when lost in mice and acts as a tumor suppressor in mice and humans. Elg1 encodes a protein that forms an RFC-like complex that unloads the replicative clamp, PCNA, from DNA, mainly in its SUMOylated form. We have identified 2 different regions in yeast Elg1 that undergo phosphorylation. Phosphorylation of one of them, S112, is dependent on the ATR yeast ortholog, Mec1, and probably is a direct target of this kinase. We show that phosphorylation of Elg1 is important for its role at telomeres. Mutants unable to undergo phosphorylation suppress the DNA damage sensitivity of Δrad5 mutants, defective for an error-free post-replicational bypass pathway. This indicates a role of phosphorylation in the regulation of DNA repair. Our results open the way to investigate the mechanisms by which the activity of Elg1 is regulated during DNA replication and in response to DNA damage.

Introduction

The genomes of all organisms are under continuous attack from endogenous metabolic processes and exogenous environmental factors that can modify their chemical structure. DNA damage can lead to mutations, deletions, insertions, translocations, and loss of essential genetic information.¹ It is therefore of vital importance that cells should repair these lesions accurately. Several surveillance and repair mechanisms operate in eukaryotic cells to ensure the stability of the genome, and dire consequences to the cell ensue when they fail to act properly. Indeed, genomic instability is a hallmark of cancer cells. Most human cancer cells show signs of genome instability, ranging from elevated mutation rates, to gross chromosomal rearrangements, including deletions and translocations.

Most spontaneous chromosomal rearrangements and mutations in the genome arise during DNA replication. The activity of the DNA polymerases may be impaired by the presence of secondary structures, bound proteins or DNA lesions; they may also be halted by collisions with other DNA-interacting proteins (such as RNA polymerases, or topoisomerases.^2,3 These encounters may lead to stalling or even collapse of replication forks, creating single-stranded gaps or double-strand breaks (DSBs). In response, cellular mechanisms are activated that arrest cell cycle progression, induce DNA repair or lesion bypass, and restore replication, in what is commonly called the “DNA damage checkpoint” or the “DNA Damage Response” (DDR). The checkpoint pathways are activated by delays in DNA replication (replication stress pathway) or by DNA lesions that obstruct replication (DNA damage pathway).⁴ This protective cellular response is evolutionarily conserved, a fact that underscores its centrality and importance. Because of this conservation, simple organisms, such as budding yeast, are extremely useful for studying the basic principles of genome stability and maintenance. Not surprisingly, genes that affect genome stability in yeast were later found to be tumor suppressors in human cells.⁵ Unless otherwise specified, we will hereafter refer to the vast knowledge on the mechanisms that maintain genome stability accumulated in yeast.

PCNA is a homotrimeric ring that encircles the double stranded DNA and plays a central role in DNA replication. It acts as a processivity factor for the replicative DNA polymerases. Many proteins are known to interact with PCNA, including factors involved in DNA replication, repair and other DNA-related activities.⁶ When DNA is damaged, PCNA may be mono-ubiquitinated at lysine 164 by the E2/E3 pair Rad6/Rad18.⁷ This modification of PCNA activates the post-replication repair (PRR) pathway, which is composed of 2 branches. When lysine164 of PCNA is mono-ubiquitinated, it allows the binding of special DNA polymerases able to replicate damaged DNA molecules, at the expense of accuracy. These translesion synthesis polymerases participate in an error-prone bypass mechanism⁸ that may generate mutations. Alternatively, PCNA can be further poly-ubiquitinated on the same lysine residue by an alternative mechanism that also requires an E2 heterodimer composed of the Ubc13 and Mms2 proteins, and the E3 protein Rad5.⁹ This poly-ubiquitination coordinates an error-free repair mechanism.¹⁰ The same residue of PCNA (K164) can be modified by the ubiquitin-like molecule SUMO. This modification takes place during S-phase or after high doses of DNA damage and requires the SUMO-specific E2 Ubc9 and the SUMO ligase Siz1.⁷ An additional residue, lysine 127, can also be SUMOylated, but not ubiquitinated. SUMOylation seems to prevent homologous recombination, favoring ubiquitin-dependent lesion bypass.^11,12

The ELG1 gene was identified as a yeast mutant that causes enhanced levels of genomic instability.^13-15 Deletion of ELG1 leads to increased recombination levels,^14,16 as well as elevated levels of chromosome loss^14,17 and gross chromosomal rearrangements.¹⁷ elg1 mutants also exhibit elongated telomeres¹⁸ and increased levels of Ty transposition.¹⁹ Elg1 function is thus clearly required for maintaining genome stability.

Replication Factor C (RFC) is a 5-subunit complex in charge of loading PCNA onto DNA to allow its replication. Elg1 shares sequence homology with Rfc1 and interacts with the 4 small subunits of RFC to create an RFC-like complex (RLC).^13-15 The Elg1 RLC was shown to interact with PCNA, particularly when it is SUMOylated, and plays a role in unloading it from chromatin.^20,21

Human ELG1/ATAD5 has been shown to play an important role in maintaining genome stability in S phase. hELG1 knockdown leads to increased levels of recombination and chromosomal aberrations such as chromosomal fusions and inversions.^22,23 Homozygous Elg1^−/− mice die as embryos, underscoring the important role played by this gene.²⁴ Heterozygous mice can be created, but display a haploinsufficient phenotype: Elg1^+/− mouse embryonic fibroblasts exhibited molecular defects in PCNA deubiquitination in response to DNA damage, as well as DNA damage hypersensitivity and high levels of genomic instability, apoptosis, and aneuploidy. Moreover, ˜90% of heterozygous Elg1^+/− mice developed tumors, including sarcomas, carcinomas, and adenocarcinomas, and these exhibited high levels of genomic alterations.²⁴ In addition, somatic mutations of hELG1 are seen in 4.6% of sporadic human endometrial tumors. These results suggest that the genome stability seen in the absence of proper ELG1 function can promote the development of cancer.²⁴

Strains lacking ELG1 are sensitive to DNA damage agents such as methyl methane sulfonate (MMS). Intriguingly, however, deletion of ELG1 partially suppresses the (greater) sensitivity of Δrad5, Δmms2 and Δubc13 mutants, defective for the error-free form of post-replicational repair.²¹ Genetic and biochemical experiments have shown that Elg1 interacts with PCNA, and particularly with SUMOylated PCNA.^20,21 Current models of Elg1 RLC activity²⁵ propose that it may be in charge of PCNA unloading during S-phase, or during repair. Elg1 may be important for unloading PCNA upon DNA Polymerase stalling at sites difficult to replicate or upon encounter with DNA lesions. Since ELG1 is not essential for viability, in its absence PCNA unloading can apparently be carried out by alternative RLCs or by RFC.

Here we report that Elg1 undergoes phosphorylation following DNA damage. Through mass spectrometry analysis, we have found that Serine 112 is phosphorylated in response to MMS, and that this phosphorylation is dependent on Mec1, the yeast ortholog of ATR. A second phosphorylation site was detected at serines 6/8. We have investigated the biological significance of Elg1 phosphorylation by creating mutants either unable to undergo phosphorylation, or mimicking a phosphorylated status. Although no combination of mutations affected the ability of cells to grow in the presence of DNA damaging agents such as MMS, we found that Elg1 phosphorylation at least at one of the 2 positions is important for correct telomere length maintenance. Moreover, we also show that preventing phosphorylation of Elg1 partially suppress the sensitivity of Δrad5 mutants, defective in the error-free post-replicational repair.

Results

Elg1 is phosphorylated in response to exposure to MMS

To follow the endogenic Elg1 protein, we tagged the ELG1 gene at its genomic locus with a C-terminal HA-tag. Figure 1A shows that Elg1 is immuno-precipitated with an anti-HA antibody (hereafter referred to as Elg1-IP) and, as expected, co-IPs with the Rfc4 protein, which, together with Rfc2, 3 and 5 binds Elg1 and forms a functional RFC-like complex.¹⁴ No band is detected in the control strain lacking the HA tag (Fig. 1A, B), and Rfc4 is not immunoprecipitated (Fig. 1A) in this strain. A ratio of 1:1:1:1 for Rfc2, Rfc3, Rfc4 and Rfc5 was observed in our mass spectrometry (MS) experiments with Elg1-IP under all conditions [untreated, or exposed to methylmethane sulfonate (MMS) or hydroxyurea (HU); data not shown].

Figure 1.

(A, B) Elg1 is immuno-precipitated together with the Rfc small subunits. Anti-HA and anti Rfc4 immunoblot of Immunoprecipitation (IP) and whole cell extract (WCE) shown. (C, D) Elg1 undergoes phosphorylation upon exposure to MMS treatment, but not to HU. Immunoprecipitation with an anti-HA antibody precipitates a phosphorylated version of Elg1. An untagged strain (Elg1) was used as a negative control. Cells were treated with 0.15% MMS or 200 mM HU for 2 h, or left untreated before the Immunoprecipitation (C). Lack of response in HU is not due to lack of activity of the antibody or failure to elicit the DNA Damage Response, as can be seen by detection of phosphorylated proteins by the anti-P-SQ antibody and phosphorylated Rad53 in whole cell extracts (D). As expected, Rad53 is more hyperphosphorylated in MMS than in HU. Histone H2B is not phosphorylated but other others proteins are phosphorylated at SQ sites upon exposure to HU.

Previous results have suggested that Elg1 may be under control of the DNA damage response, which consists of a cascade of post-translational modifications, especially phosphorylation events. In order to search for Elg1 phosphorylation events upon DNA damage, we used LC-MS/MS. We subjected the Elg1-HA strain to treatment with either MMS or HU, and, after IP with an anti-HA specific antibody, samples were used for LC-MS/MS phosphorylation analysis. We identified, in MMS-treated cells, a phosphorylated serine, at position 112 of Elg1. The phosphorylated S112 is followed by a glutamine (Q). ATM/ATR kinases and their homologs, Tel1/Mec1, preferentially phosphorylate their substrates on serine or threonine residues followed by glutamine residues, the so-called SQ/TQ motifs.²⁶ Serine 112 also conforms to the motif LSQE, which was found in a peptide library analysis as the preferentially phosphorylated central core motif of ATM.²⁷ In order to validate Elg1s SQ-phosphorylation upon DNA damage, we used an antibody that specifically detects phosphorylated SQ motifs (hereafter referred to as anti-P-SQ antibody). We subjected the Elg1-HA strain to treatment with MMS or HU. After IP with an anti-HA antibody, the anti-P-SQ antibody was able to detect a phosphorylated version of Elg1 in cells treated with MMS, but not in cells exposed to HU (Fig. 1C). The specificity of the response is not due to lack of activity of the antibody or lack of checkpoint activation in the cells treated by HU (Fig. 1D): many phospho-proteins are observed both in MMS and in HU with the anti-P-SQ antibody when total cell extract was used. Fig. 1D also shows that Rad53, the yeast ortholog of Chk2, is phosphorylated under the DNA damaging conditions (although, as expected, Rad53 is less phosphorylated in HU, and histone H2A, which marks the presence of double-strand breaks,²⁸ is not much changed). We thus conclude that although the cells respond to MMS and HU by activating the DDR, Elg1 is phosphorylated at serine 112 only in cells treated with MMS.

Next, we explored the kinetics of Elg1 phosphorylation: cells were grown in 0.1% MMS for different time intervals and HA-tagged Elg1 was immunoprecipitated from cell extracts using anti-HA antibody. A strong phosphorylated Elg1 band was detected by the anti-P-SQ antibody after 15 min of incubation; its levels continued to increase in further time points, reaching its peak already after 60 minutes (Fig. 2A). Note that in this experiment, as in others, a low but detectable level of Elg1 phosphorylation can be detected in cells not exposed to DNA damage, indicating the presence of spontaneous DNA damage (Fig. 2A). Again, controls in Figure 2B demonstrate the normal kinetics of induction of the DDR in whole cell extracts of these cells.

Figure 2.

Elg1 is phosphorylated rapidly following MMS treatment. (A) Kinetics of SQ phosphorylation of Elg1-IP. Total HA-Elg1-IP levels and Rfc4 co-IPed with Elg1 are also shown. MMS (0.1% or 0.04%) were added at time=0 and samples were taken at various time points. (B) Kinetics of Rad53 phosphorylation and total SQ/TQ phosphorylation in whole cell extracts of the same samples as in (A).

Our MS experiment detected S112 as the target of phosphorylation of the DDR. By bioinformatic analysis, however, we identified 2 additional SQ motifs in the yeast Elg1 protein (S540 and S579). To identify the residues undergoing phosphorylation we individually mutated each Serine in the 3 SQ motifs to alanine. Figure 3A shows that mutation of S112 to A was sufficient to prevent phosphorylation of Elg1. The mutation did not alter the cellular DDR, as seen in Figure 3B. No Effect was observed for the 2 other SQ sites (data not shown).

Figure 3.

Damage-induced phosphorylation of Elg1 at S112 site. Cells were incubated in the presence, or not, of 0.2% MMS for 2 hours before Elg1-HA was immunoprecipitated and analyzed. (A) Mutation of the Serine residue at position 112 to Alanine abolishes phosphorylation of Elg1 after MMS treatment. (B) The cellular response to MMS is not affected by the mutation.

Phosphorylation of Elg1 following DNA damage is Mec1 dependent

In the budding yeast S.cerevisiae, Mec1 plays a critical role in checkpoint control,²⁹ whereas Tel1 plays a lesser role and shares an overlapping role with Mec1 in maintaining survival after DNA damage and telomere length homeostasis.³⁰ We next asked whether Elg1 phosphorylation is Mec1 or Tel1 dependent. We created Δmec1, Δtel1, or Δmec1 Δtel1 double mutants expressing an HA-tagged Elg1 protein; since Mec1 is essential to maintain a minimal dNTP level in the cells, all strains were deleted for the SML1 gene.³¹ Fig. 4 shows that whereas deletion of TEL1 has no effect on Elg1 phosphorylation, deletion of MEC1 completely abolishes the phosphorylation. We thus conclude that the phosphorylation of Elg1 at its S112 residue upon MMS treatment is a Mec1-dependent process.

Figure 4.

Elg1 phosphorylation is Mec1-dependent. Cells were incubated in the presence, or not, of 0.1% MMS for 2 hours before Elg1-HA was immunoprecipitated and analyzed, as before A. Elg1 phosphorylation following MMS treatment is not seen in strains deleted for MEC1. B. Whole cell extract controls.

Elg1 phosphorylation is not required for MMS resistance

Elg1 plays important roles during DNA replication and in maintaining the genome stable.^20,32 Δelg1 mutants are sensitive to various DNA damaging agents, among them MMS. Surprisingly, the S112A mutation, which abolishes the MMS-induced phosphorylation, had no effect on the sensitivity to MMS (Fig. 5A).

Figure 5.

Phenotype of elg1phosphorylation mutants. (A) MMS sensitivity assay of different elg1 mutants. None of the mutants defective for phosphorylation sites shows phenotype upon MMS exposure. (B) Southern blot analysis to check the telomere phenotype in various elg1 mutants. Phosphorylation of at least one of the 2 sites, serines 6/8 or serine 112, is necessary in order to maintain normal telomere length. A continuous white line marks the length of the wt telomere XIR (2214 bp). A dotted white line marks the telomere length of Δelg1 and elg1-SSS6,8,112AAA mutants. A 2044 bp genomic band serves as size reference.

In addition to S112 phosphorylation, our Mass Spectrometry experiments also detected peptides in which Serine 6 or Serine 8 of Elg1 was phosphorylated. Indeed, a phosphorylation event at Ser6 has been observed in a proteomic study.³³ We reasoned that phosphorylation at S6 may also play a regulatory role and affect the sensitivity to MMS. Since mutations in one serine residue sometimes lead to phosphorylation to another serine residue close by, we generated different yeast strains mutated at Serine 6 and 8 (SS6,8AA, SS6,8EE) alone, or in combination with mutations in serine 112 (S112A, S112E). Serines were mutated either to alanine, a small, neutral amino acid, or to glutamic acid, a positively charged residue that often mimics a phosphorylated state. However, none of these mutations had any effect on the sensitivity to yeast strains to MMS (Fig. 5A).

Elg1 phosphorylation affects its role in telomere length maintenance

One of the most striking phenotypes of Δelg1 strains is that they exhibit long, but stable, telomeres.^18,34 To learn whether Elg1 phosphorylation plays a role in telomere length maintenance, we carried out a Southern blot analysis of elg1 mutants carrying mutations at either serines 6/8 or serine 112, in all possible combinations. Fig. 5B shows that mutating S112 or the serines at positions 6 and 8 to alanine (S112A; SS6,8AA) did not result in changes in telomere length. No effect was also observed in the S112E or the SSS6,8,112AAE mutants. In contrast, mutating the 3 serine residues to alanine (SSS6,8,112AAA) led to long telomeres, similarly to what is seen in Δelg1 mutants (Fig. 5B). In all the strains in which serines 6 and 8 were changed to glutamic acid (SS6,8EE; SSS6,8,112EEA; SSS6,8,112EEE), telomeres of intermediate length are observed.

We conclude that telomere length maintenance by Elg1 requires phosphorylation at both the serines 6/8 and the serine 112. However, phosphorylation of at least one of the 2 sites is sufficient to maintain normal telomere length.

Elg1 phosphorylation mutations suppress the sensitivity to MMS of Δrad5 cells

Rad5 is a multifunctional protein important for genome stability: it plays a role in the error-free post-replication repair pathway⁹ and in the repair of double-stranded breaks.³⁵ Rad5 contains 2 domains, an E3 ubiquitin ligase motif and a conserved helicase-like domain of the SWI/SNF family of ATPases.⁷ Previous work has shown that deletion of ELG1 partially suppresses the sensitivity of Δrad5 and rad5-E3 mutants to MMS,²¹ presumably by preventing the unloading of PCNA. We therefore tested whether Elg1 phosphorylation may play a role in this process. Fig. 6A shows that the SS6,8AA and SS6,8EE mutants show a significant suppression of the Δrad5 sensitivity to MMS, which is not further affected by additional mutation of S112. The nature of the changed amino acid had only a small effect: suppression of MMS sensitivity of Δrad5 by Elg1-S6,8,112 to alanine (A) is only slightly lower than Elg1-S6, 8,112 to glutamic acid (E) (Fig. 6A). One possible explanation for this suppression could be that the level of Elg1 protein is reduced in Δrad5 mutants. Fig. 6B, however, shows that there is no change in the level of Elg1 in the mutants. Taken together, these results suggest that any changes in the Serines 6, 8 or 112 are important for an activity of Elg1 that becomes toxic in the absence of PCNA polyubiquitination.

Figure 6.

Sensitivity to MMS of elg1 mutants in Δrad5 mutant background. (A) All the elg1 phosphorylation mutants show a phenotype similar to that of Δelg1 in the Δrad5 mutant background and suppress the MMS sensitivity. (B) Elg1 protein levels in the Δrad5 mutant background. Protein levels of Elg1 remain unchanged in the Δrad5 background.

Discussion

Elg1 is a central protein with many roles in the maintenance of genome stability. Lack of Elg1 causes sensitivity to DNA damaging agents, gross chromosomal rearrangements, chromosome loss, high levels of recombination and transposition, and elongated telomeres.^13-15 Here we have found that the Elg1 protein undergoes phosphorylation at 2 different sites: Serines 6/8 and Serine 112. Whereas phosphorylation at Serine 112 is clearly induced by DNA damage and is dependent on an active DDR, the nature of the phosphorylation at Serines 6/8 and the identity of the kinase responsible for the modification is still unknown. The sequence context of these serines, however, is incompatible with the CDK1 cell cycle kinase, which is known to regulate genome stability proteins in coordination with DDR kinases.^36-38

The DNA damage response (DDR) has been genetically dissected in the yeast S. cerevisie: The main control is carried out by the phosphatidyl-inositol 3-kinase-like kinase (PI3KK) Mec1, an ortholog of the human ATM and ATR kinases. Mec1 controls the downstream kinases Chk1 and Rad53 through the Rad9 and Mrc1 adaptor proteins, which amplify signals created by chromosomal damage (e.g., DSBs) and replication slowdown or stalling, respectively. The localized Mec1 activation is thus amplified into a nuclear response that regulates many different processes, including cell cycle control, chromatin remodeling and movement, and several types of DNA damage repair and tolerance pathways.⁴ The DDR also induces transcription of a specific group of genes, many of which (although not all), are involved in DNA replication or repair. The DNA damage response (DDR) thus coordinates the cellular response to genotoxic stress and controls effector pathways that maintain genomic stability and promote survival.⁴ A relatively small number of proteins have been shown to become phosphorylated by the DDR; these include Ddc2, Mec1s binding partner,³⁹ the ssDNA binding protein RPA,⁴⁰ the Mcm4 and Mcm6 components of the replicative helicase,⁴¹ the DNA primase, which synthesizes the RNA primer during DNA replication,⁴² and the Rad55 protein, which participates in homologous recombination.⁴³ Here we add Elg1 to this list. Since the S112 serine phosphorylated is an SQ site, usually recognized and phosphorylated by PI3KKs, and deletion of MEC1 abolishes the modification (Fig. 4A), we consider very likely that Elg1 is a direct target of Mec1 phosphorylation. Interestingly, S112 phosphorylation is elicited by treatment with MMS, a drug that acts during DNA replication, but not by hydroxyurea, a drug that arrests cells at the border between G1 and S, due to a deficiency in dNTPs.

Stalled replication forks are a major source of genomic instability,¹ and several pathways operate at stalled forks, presumably in a highly regulated hierarchy. Mec1 plays a central role in preventing irreversible breakdown of stalled replication forks in budding yeast.⁴⁴ PCNA modifications play a central role in signaling the presence of stalled replication forks. PCNA poly-ubiquitination coordinates an error-free repair mechanism¹⁰ that bypasses the damaged nucleotides, possibly by copying information from the recently duplicated sister chromatid, allowing completion of DNA replication. PCNA SUMOylation at lysines 127 and 164, on the other hand, has been suggested to prevent homologous recombination, favoring repair via ubiquitin-dependent lesion bypass.^11,12

By interacting with the 4 small subunits of RFC, Elg1 forms a conserved alternative RFC-like protein complex (RLC).^13-15 This RLC unloads PCNA, possibly during DNA replication or as a response to fork stalling or collapse.^21,45 However, it is important to point out that the Elg1-RLC activity cannot be the only mechanism able to unload PCNA, as PCNA unloading must be an essential function and mutants deleted for ELG1 show only slight growth defects.^13,15 Elg1 may be responsible for the unloading of PCNA at specific genomic regions, such as cohesion sites,^46,47 or upon replication fork arrest.⁴⁵ It may also participate in polymerase switches during DNA replication and in the coordination between replication and telomere elongation.⁴⁸

Despite the fact that a mutation of Serine 112 to Alanine completely abolishes phosphorylation by MMS, it has no discernible effect on the sensitivity to this DNA damaging agent (Fig. 5A). Since sometimes targets of the DDR are phosphorylated by more than one kinase [e.g.,^41,49], we created yeast strains carrying multiple mutations at serines 6, 8 and 112. Surprisingly, these mutations did not have an effect on the ability of cells to repair MMS DNA damage. We therefore have to assume that additional layers of regulation on Elg1 activity remain to be uncovered.

Elg1 controls telomere length in yeast.^18,34 Our working hypothesis is that Elg1 plays a role in the coordination between leading strand elongation by telomerase and lagging strand DNA synthesis, which should occur during genome replication. Mutation of the 3 serines (residues 6, 8 and 112) to alanine led to long telomeres, of similar size of those seen in Δelg1 strains. In contrast, strains having alanine replacements only at the 6/8 or at the 112 positions showed telomeres of wt length (Fig. 5B). This implies that these 2 sites carry overlapping functions, and at least one of them must undergo phosphorylation to allow Elg1 to function normally. Mutations that change serine residues to alanine prevent phosphorylation; however, mutations that change serine to glutamic acid sometimes do, and sometimes do not, mimic phosphorylation. It is clear that for telomere length control, lack of phosphorylation at both sites (elg1 6,8, 112 AAA) is similar to ELG1 deletion. However, it appears as if the changes to glutamic acid do not truly mimic phosphorylation. Thus, changing serines 6 and 8 by glutamate caused a slight elongation of telomeres that was not further affected by mutations at position 112. It is likely that the S-to-E mutants, without fully mimicking a phosphorylated state, are also slightly defective for phosphorylation. Phosphorylation of Elg1 may play a role in its recruitment or its unloading after it has carried out its role. Future work will center on identifying the kinase involved in phosphorylating serines 6/8, and the proteins able to interact with the phosphorylated protein.

Despite the fact that by themselves the phosphorylation mutations did not impair Elg1s capacity of DNA repair/bypass, they did show an effect if the error-free branch of the post-replication repair was abolished. Strains in which the RAD5 gene is deleted are extremely sensitive to MMS, presumably because of an inability to bypass lesions left at replicative gaps.⁵⁰ Deletion of ELG1 in this background reduces the sensitivity to that of the single Δelg1 mutant. This can be interpreted as follows: Unloading of SUMOylated PCNA by Elg1 prevents an alternative repair pathway that is active when PCNA cannot be poly-ubiquitinated. If ELG1 is deleted, this alternative pathway is more efficient and sensitivity is reduced. We have shown in the past that this toxicity requires PCNA SUMOylation.²¹ The phosphorylation-site mutations show suppression of Δrad5 to almost the same extent of Δelg1 (Fig. 6A). These results thus suggest that lack of phosphorylation of Elg1 impairs a SUMOylated-PCNA-dependent pathway (possibly involving PCNA unloading from chromatin); this inactivation is enough to show a phenotype in the absence of Rad5, but it is masked if the Rad5-dependent pathway is active.

Materials and Methods

Yeast strains and plasmid

All yeast strains used are derivatives of MK166.⁵¹ Mutants were created by standard yeast manipulation techniques.

Yeast cells used in this study were grown at 30°C in either YPD medium or in Synthetic Dextrose (SD) medium supplemented with essential nutrients as required.

Immunoprecipitation

Yeast cells were grown at 30°C to mid-log phase. Cells were treated with 0.15% MMS or 200 mM HU for 2 h, or left untreated. The cells were then harvested and washed 3 times with distilled cold water. The cells were lysed with glass beads in pre-chilled lysis buffer for 40 min at 4°C with rotation. The cell lysate was pelleted down by centrifugation for 20 min at 14,000 RPM at 4°C and the supernatant (extract) used for immunoprecipitation (IP). Protein concentrations were measured and equal amount of proteins were taken for IP: Extracts were incubated with anti-HA antibody overnight at 4°C. Protein A-/Protein G-sepharose were added and the tubes were incubated with rotation, at 4°C, for 3 more hours. The beads were washed 5 times with lysis buffer. The precipitated proteins next to samples of extracts were separated on 10% acrylamide SDS-PAGE and were transferred to a nitrocellulose membrane for Immunoblotting. For identification of post translational modifications of proteins, gels were washed with ultra pure water overnight and subjected to protein analysis by LC- MS/MS.

Immunoblotting

Immunoblotting (IB) for Elg1 was carried out using an anti-HA antibody (Santa Cruz sc-7392). Phosphorylated S112 was detected with a rabbit polyclonal anti-pSQ/pTQ antibody (code number: #2851, Cell Signaling Technologies). Other antibodies used for IB: anti-RFC4 (Abcam, ab2627), anti-actin (ImmunO #69001), anti-pH2AX (Abcam, ab81299), and anti-Rad53 (Santa Cruz sc-6749).

DNA damage sensitivity assay

Yeast cells were grown overnight in SD-complete medium at 30°C to mid-log phase. Serial fold10- dilutions were spotted on SD-complete plates with or without MMS (Sigma) and incubated at 30°C for 3 d

Site specific mutagenesis

A DNA fragment carrying full length ELG1::HA marked with a KANMX cassette was cloned into pGEM-T Easy Vector (Promega) and subjected to PCR using different forward and reverse primers containing the desired mutations (Table 1). Amplified PCR product was transformed in DH5α cells after digestion with Dpn1 (NEB). Plasmids were isolated and sequenced.

Table 1.

List of primers used in creating different mutations through site directed mutagenesis

Mutation	Forward primer (5′- - - 3′)	Reverse primer (5′- - - 3′)
Elg1-S112A	TTGCACTTGCGCAGGAGCATG	CATGCTCCTGCGCAAGTGCAA
Elg1-S112E	TTGCACTTGAGCAGGAGCATG	CATGCTCCTGCTCAAGTGCAA
Elg1-S6,8A	TGGCTTTAGCTGATATATTGACAG	ATATCAGCTAAAGCCACGTGCC
Elg1-S6,8E	TGGAGTTAGAGGATATATTGACAG	ATATCCTCTAACTCCACGTGCC

To transfer the mutation to the yeast genome, the mutagenized plasmids were digested with NotI (NEB) and transformed to elg1::HYG cells. Transformed cells were selected on G418 (Sigma) plates. G418 resistant, hygromycin sensitive colonies were subjected to DNA sequencing to confirm the presence of the particular mutation, and the lack of spurious ones.

Southern teloblots

After creating new mutants, cells underwent 10 consecutive re-streaks (∼25 generations each) on YEPD to make sure that the cells reach a stable telomere length phenotype. Southern blotting was carried out as previously described,⁵² except that a probe specific for telomere XIR was used. The size-control probe is a specific region of chromosome II (positions 558490 to 559790) that detects 2 bands in the XhoI digested genomic DNA (2044 and 779bp long).⁵²

Funding Statement

This work was supported by grants from the Israel Science Foundation, the Israel cancer association and the Israel Cancer Research Foundation to M.K.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank all members of the Kupiec group for help and encouragement. We also thank Tevi Mehlman and the MS facility of the Weizmann Institute of Science for their assistance.