S. cerevisiae Mre11 recruits conjugated SUMO moieties to facilitate the assembly and function of the Mre11-Rad50-Xrs2 complex.

Double-strand breaks (DSBs) in chromosomes are the most challenging type of DNA damage. The yeast and mammalian Mre11-Rad50-Xrs2/Nbs1 (MRX/N)-Sae2/Ctp1 complex catalyzes the resection of DSBs induced by secondary structures, chemical adducts or covalently-attached proteins. MRX/N also initiates two parallel DNA damage responses-checkpoint phosphorylation and global SUMOylation-to boost a cell's ability to repair DSBs. However, the molecular mechanism of this SUMO-mediated response is not completely known. In this study, we report that Saccharomyces cerevisiae Mre11 can non-covalently recruit the conjugated SUMO moieties, particularly the poly-SUMO chain. Mre11 has two evolutionarily-conserved SUMO-interacting motifs, Mre11(SIM1) and Mre11(SIM2), which reside on the outermost surface of Mre11. Mre11(SIM1) is indispensable for MRX assembly. Mre11(SIM2) non-covalently links MRX with the SUMO enzymes (E2/Ubc9 and E3/Siz2) to promote global SUMOylation of DNA repair proteins. Mre11(SIM2) acts independently of checkpoint phosphorylation. During meiosis, the mre11(SIM2) mutant, as for mre11S, rad50S and sae2Δ, allows initiation but not processing of Spo11-induced DSBs. Using MRX and DSB repair as a model, our work reveals a general principle in which the conjugated SUMO moieties non-covalently facilitate the assembly and functions of multi-subunit protein complexes.

INTRODUCTION

Small ubiquitin-like modifier (SUMO) is a small regulatory protein found in almost all eukaryotic organisms (1–3). In the yeast Saccharomyces cerevisiae, the essential gene SMT3 encodes the SUMO protein. SUMOylation is a post-translational modification, in which SUMO is covalently attached to a substrate protein including SUMO itself, to form a poly-SUMO chain. SUMOylation is mediated by an enzymatic cascade that is analogous to the one that is involved in ubiquitination. The removal of the SUMO adduct from targets is catalyzed by specific SUMO proteases, e.g. Ulp1 and Ulp2. The conjugated SUMO moieties (CSMs) are recognized by two types of SUMO-binding motifs, short hydrophobic sequences known as SUMO-interacting motifs (SIMs) (4) and the ZZ zinc fingers (5,6). The addition, removal and recognition of SUMO are influenced by and affect a plethora of cellular pathways. Because SUMOylation frequently targets entire groups of physically interacting proteins rather than individual proteins, it has been proposed that protein-group SUMOylation functions to establish new physical interactions between proteins that have SUMO-binding motifs (7). Alternatively, CSMs can covalently or non-covalently prevent premature aggregation by increasing the water solubility of individual protein subunits (8,9) prior to their assembly into a functional protein complex (10–12). A proof-of-concept has been provided by simultaneously expressing three capsid proteins of the foot-and-mouth disease virus (FMDV); these three SUMO fusion proteins formed a stable heterotrimeric complex. The proteolytic removal of SUMO moieties from the ternary complexes resulted in virus-like particles with a size and shape resembling the authentic FMDV, which contains 20 heterotrimers of the capsid proteins (10).

SUMOylation is strongly connected to the repair of DNA double-strand breaks (DSBs) (13). In S. cerevisiae vegetative cells, mutations or deletion of SUMOylation genes cause a pronounced sensitivity to DNA damage and genomic instability, including the poly-SUMO chain mutant (smt3-allR) (14), E2 conjugating enzyme (ubc9), E3 ligase enzymes (siz1Δ, siz2Δ, mms21) (7,15), deSUMOylation proteases (ulp1Δ) (16,17) as well as SUMO-targeted ubiquitin ligases (STUbLs; slx5Δ, slx8Δ) (18,19). In parallel, with checkpoint phosphorylation, vegetative yeast cells also induce SUMOylation of many proteins that are needed for replication and repair in response to DNA damage (7,15). The MRX (Mre11-Rad50-Xrs2)-Sae2/Com1 dsDNA endonuclease complex has been implicated as a positive regulator for DSB-induced global SUMOylation (15). Mre11 exhibits notable two-hybrid interactions with Ubc9 and Siz2, and it has been proposed that the binding of Siz2 might be achieved through Ubc9-catalyzed SUMOylation of Mre11 (7). Sae2 (also called Com1)(20) is a SUMOylated protein during vegetative growth, and SUMOylation of Sae2 increases both soluble Sae2 and the MRX function in DNA end resection (21). Sae2 apparently mediates removal of the MRX complex from the DNA damaged sites during vegetative growth. The MRX complex is retained at DSB ends in sae2Δ, thus turning on the DNA damage checkpoint to stall cell cycle progression (22,23). Recently, it has been reported that, following the resection of DSBs, Siz2 also collaborates with the ssDNA binding complex RPA (replication protein A) to enhance global SUMOylation (7,24). The relationship between MRX and RPA in promoting DSB-induced SUMOylation and the molecular mechanism of the interaction between MRX and Sae2 interactions remains unclear.

MRX has multiple functions during S. cerevisiae meiosis (25,26). First, it has a unique role in Spo11-induced DSBs independently of its catalytic activity, and the C-terminal portion of Mre11 is specifically required for this function (27,28). Second, the MRX-Sae2 endonuclease complex acts at each 5′-end of DSBs to generate 3′-end ssDNA tails through the removal of a covalently linked Spo11-oligonucleotide complex (29,30). The 3′-end ssDNA tails subsequently assemble into nucleoprotein filaments comprised of two RecA-family recombinases (Rad51 and Dmc1) and their accessory factors to catalyze DSB repair via homologous recombination (31–33). Third, MRX senses DSBs and activates the Tel1ATM checkpoint kinase for target phosphorylation. This checkpoint phosphorylation has dual roles in preventing superfluous Spo11-induced DSBs (34,35) and in promoting interhomolog recombination (12,35,36). Interhomolog recombination is a hallmark of meiotic recombination. A few Spo11-induced DSBs must be repaired using a homologous non-sister chromosome (but not a sister chromatin) as template to generate new combinations of DNA sequences (26).

Accumulating evidence has also revealed that SUMOylation functionally links two groups of S. cerevisiae proteins that are essential for interhomolog recombination. The first group includes three meiosis-specific chromosomal proteins Hop1, Red1 and Mek1. These proteins are the axial components of the synaptonemal complex (SC)—a zipper-like proteinaceous structure that mediates chromosome synapsis between homologous chromosomes during meiotic prophase. The SC consists of two dense lateral/axial elements and a central element. To assemble the SC, both the SC central protein, Zip1, and the SC axial protein, Red1, non-covalently interact with conjugated SUMO moieties (CSMs), such as poly-SUMO chains or conjugates. During SC assembly, the SC initiation protein, Zip3, acts as a SUMO E3 ligase that promotes the formation of additional CSMs (11,37). Consistent with these findings, it has been shown that the SUMOylation of Ubc9 promotes the formation of a poly-SUMO chain, which is a key event for SC formation (38). Furthermore, SUMOylation and the ubiquitin-mediated removal of CSMs (e.g. SUMOylated topoisomerase II or Red1) have been implicated in SC-mediated crossover interference (39). Crossover interference is a genetic phenomenon in which crossovers tend to be evenly spaced along any given meiotic chromosomes. SUMOylation is also critical in the regulation of meiotic recombination or chromosomal morphogenesis in other sexually-reproductive organisms, including the fission yeast S. pombe (40), the fungus Sordaria (41), Arabidopsis thaliana (42) and mammals (43).

The DNA damage checkpoint proteins Mec1ATR and Tel1ATM are the second group required for establishing interhomolog bias during meiotic recombination. Tel1ATM is activated by non-resected DSBs via an Xrs2-dependent mechanism (44–47), and Mec1ATR is recruited to RPA-coated ssDNA tails via its binding partner, Ddc2 (48,49). Mec1ATR activation also requires three additional DNA damage sensors: the yeast 9-1-1 complex (Ddc1-Mec3-Rad17), its clamp loader, the Rad24-RFC complex and Dpb11 (48,50–52). These two protein kinases phosphorylate the SC axial protein Hop1 to ensure interhomolog recombination (11,12,36,53,54). Notably, both Tel1ATM- and Mec1ATR-dependent Hop1 phosphorylation requires Red1 and the Red1-CSM interaction (11,36). Red1 first non-covalently associates with CSMs (11) and then with the 9-1-1 complex (55) to activate Mec1ATR resulting in Hop1 phosphorylation via its binding to the 9-1-1 complex (12). It is still unclear how the Red1-CSM ensemble couples with MRX at the non-resected DSB ends during meiosis.

Here, we investigate the molecular mechanism and physiological impacts of the Mre11-SUMO interaction in response to DSBs during vegetative growth and meiosis. Our results reveal that the yeast S. cerevisiae Mre11 can non-covalently recruit CSMs to facilitate both global SUMOylation and DSB repair.

MATERIALS AND METHODS

Yeast strains, two-hybrid assay and physical analysis

All vegetative experiments were performed using haploid cells from isogenic W303 strains as described previously (15,56). All meiotic experiments were performed using diploid cells from isogenic SK1 strains. Quantitative yeast two-hybrid assays, tetrad dissection, fluorescence-activated cell sorting (FACS), pulsed-field gel electrophoresis (PFGE) and Southern hybridization were carried out as previously described (11,36,37).

Antisera, immunoblot, dephosphorylation assay and cytology

The antisera used against Hop1, phosphorylated Hop1-T318, Zip1, phosphorylated Zip1-S75, H2A and phosphorylated H2A-S129 have been described previously (36). Peroxidase-anti-peroxidase (PAP) antibody (Sigma, CA, USA), IgG Sepharose beads (GE Healthcare, Bucks, UK), anti-HA antibody (Roche, Basel, SWZ) and anti-Rad53 antibody (Santa Cruz Biotechnology, TX, USA) were purchased commercially. Western blotting analyses were performed as recently described and repeated 2-4 times (36). The dephosphorylation assay was performed as described previously (54). Cytology analyses were carried out as previously described (11,37).

Immunoprecipitation

Yeast vegetative cultures (20 ml; OD600 ≈0.5) were harvested and washed once with ice-cold water. Cells were resuspended in 250 μl of lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM EDTA pH 8.0, 0.1% Triton X-100) containing an EDTA-free protease inhibitor complete cocktail (Roche) and 20 mM N-ethylmaleimide (Sigma). To prepare the total cell lysates, the cell suspension was mixed with 1/2 volume of 0.5-mm acid-washed glass beads (Sigma), vigorously vortexed (30 s of vortexing and 30 s on ice) five times, and then microcentrifuged (16 000 x g) at 4°C for 10 min. The supernatants were collected and incubated with IgG Sepharose beads (Sigma) or immobilized anti-HA affinity resin (Sigma) at 4°C for 2 h. The precipitants were washed three times with 1 ml of lysis buffer and then resuspended in 100 μl Laemmli loading buffer containing 100 mM dithiothreitol. The proteins were analyzed by SDS-PAGE and immunoblotting as described previously (11).

RESULTS

Identification of two SUMO-interacting motifs (SIMs) in Mre11 that preferentially interacts with the poly-SUMO chain

We identified, in silico, two putative SIMs in S. cerevisiae Mre11: Mre11SIM1 (IRIL, residues 9-12) and Mre11SIM2 (ESDKIKVV, residues 154-161) (57). Both SIMs are evolutionarily conserved in S. pombe Mre11: Mre11SIM1 (IRIL, residues 18-21) and Mre11SIM2 (ENDNIVV, residues 163-169) (Figure 1A). Furthermore, they reside at the outermost surface of the S. pombe Mre11-Nbs1 complex crystal structure (58) (Figure 1A). Nbs1 is the S. pombe homolog of Xrs2. Thus, Mre11 might non-covalently associate with the SUMO monomer or CSMs.

Figure 1.

S. cerevisiae Mre11 protein has two SUMO-interacting motifs (SIMs) that are differentially required for MMS resistance and telomere length maintenance. (A) Schematic representation of full-length S. cerevisiae Mre11, Mre11SIM1 and Mre11SIM2, and their amino acid sequences. Mre11SIM1 and Mre11SIM2 are evolutionarily conserved in S. pombe Mre11. Shown in the lower panel is the crystal structure of Nbs1 (474–531 amino acid residues; in blue) and the two full-length Mre11 subunits (in gray and dark green)(58). SIM1 and SIM2 are indicated in purple and orange, respectively. The catalytic site of S. pombe Mre11 is indicated by a red arrow. (B, C) Spot assay showing the five-fold serial dilutions of W303 (B) and (C) SK1 haploid strains grown on YPD plates and YPD plates containing MMS at the indicated concentrations. The SK1 mre11Δ haploid strains were transformed with an pYC6-P vector for expressing various Mre11 proteins as indicated. P is the promoter of the wild-type MRE11 gene. (D) Telomere length. XhoI-digested genomic DNA was separated on a 1.2% agarose gel and Southern hybridized with a Y′ probe. The terminal telomere repeats form heterogeneous fragments of ≈1.3 kb, whereas the Y′ long and short subtelomeric repeat sequences appear at the top of the gel (56).

These two distinct possibilities were further examined by two-hybrid assays with either vegetative or meiotic two-hybrid reporter cells (11,59). We found that Mre11 preferentially interacts with CSMs rather than the Smt3 monomer; the hierarchy for two-hybrid interactions with Mre11 was Smt3 > Smt3-allR > Smt3-ΔGG ≅ mock control in both vegetative and meiotic reporter cells (Table 1). Smt3-allR cannot form a polymeric chain because the nine lysine residues in the wild type Smt3 are replaced by arginine, but it remains competent in the SUMO conjugation with all target proteins (including wild-type Smt3) (60). Smt3-ΔGG is a conjugation-incompetent Smt3 mutant that lacks the C-terminal pair of glycines required for E1-mediated Smt3 activation (61). Next, we constructed two SIM mutant proteins (Mre11I9R and Mre11I158R), each with a mutation from isoleucine (I) to arginine (R). We found that neither of these two mutants exhibited notable two-hybrid interactions with Smt3, Smt3-allR or Smt3-ΔGG (Table 1).

Table 1.

Two-hybrid analyses1

	X-Y interaction1
Gal4AD-X	Host cell	LexA-Y
		Mre11	Mre11I9R	Mre11I158R	Mre11D16A	Mre11P84S	Mre11T188I
mock	Vegetative growth	0.4 ± 0.1	0.3 ± 0.0	0.5 ± 0.3	1.2 ± 0.6	0.2 ± 0.0	0.2 ± 0.1
Smt3		71.9 ± 0.6	0.8 ± 0.0	7.6 ± 1.3	8.1 ± 5.5	51.7 ± 3.1	60.1 ± 3.6
Smt3allR		21.8 ± 0.1	0.6 ± 0.0	1.9 ± 0.3	1.7 ± 2.0	11.2 ± 1.3	16.6 ± 2.8
Smt3ΔGG		0.7 ± 0.0	0.3 ± 0.0	0.2 ± 0.0	1.8 ± 0.6	0.2 ± 0.0	0.3 ± 0.0
Ubc9		63.6 ± 2.0	0.8 ± 0.4	2.4 ± 1.0	n.d.2	54.6 ± 1.7	48.3 ± 2.5
Siz2348–726		79.8 ± 2.1	0.3 ± 0.2	7.8 ± 1.2	n.d.2	84.2 ± 6.2	41.4 ± 2.1
Rad50		135.2 ± 5.7	2.6 ± 0.1	106.0 ± 4.8	5.8 ± 2.9	121.1 ± 4.2	55.2 ± 4.9
Xrs2		188.1 ± 8.5	0.3 ± 0.0	37.9 ± 4.0	76.2 ± 7.0	161.9 ± 6.4	153.4 ± 4.7
mock	Meiosis3 (ndt80Δ)	0.7 ± 0.3	0.6 ± 0.3	0.3 ± 0.1	n.d.2	n.d.2	n.d.2
Smt3		38.8 ± 5.8	0.6 ± 0.3	1.4 ± 0.3
Smt3allR		5.4 ± 1.9	0.9 ± 0.5	0.8 ± 0.2
Smt3ΔGG		0.7 ± 0.3	0.7 ± 0.3	0.4 ± 0.1
Rad50		54.7 ± 0.9	1.0 ±0.5	26.0 ± 1.0
Xrs2		125.3 ± 9.5	0.6 ± 0.3	35.1 ± 1.0

1The two-hybrid interaction was determined by measuring the β-galactosidase activity. One unit of β-galactosidase hydrolyzes 1 μmol of o-nitrophenyl β-galactopyranoside per min per OD600 (optical density at 600 nm) unit.

2n.d. (not determined).

3Meiotic two-hybrid analysis using an ndt80Δ diploid strain (11,59)

S. cerevisiae Mre11 exhibits strong two-hybrid interactions with the SUMO E2 enzyme Ubc9 and the SUMO E3 ligase Siz2 (7), and Mre11 is also a SUMOylated protein (7,11). We examined whether the Mre11-Ubc9 and Mre11-Siz2 interactions are mediated via SUMOylated Mre11 or non-covalently via the Mre11-CSM ensembles. We found that the hierarchy for the two-hybrid interaction with Ubc9 or Siz2348–726 was Mre11 > Mre11I158R ≅ Mre11I9R ≅ mock control (Table I). Siz2348–726 is the C-terminal domain of Siz2 and harbors two SIMs (11). These results suggest that both interactions are likely mediated via the Mre11-CSM ensembles rather than SUMOylated Mre11 (also see below).

To minimize the potential effects of arginine replacement on protein structure or folding, we also constructed additional Mre11SIM1 and Mre11SIM2 mutants with one, two or three alanine mutations, respectively. All these mutants exhibited reduced or no two-hybrid interactions with Smt3, Smt3-allR or Smt3-ΔGG (Supplementary file, Supplementary Table S1). The hierarchy for the two-hybrid interaction of the Mre11SIM1 mutants with Smt3 was Mre11 >> Mre11I11A > Mre11I9A > Mre11I9A,I11A ≅ Mre11I9R, and that of the Mre11SIM2 mutants with Smt3 was Mre11 >> Mre11I158A > Mre11I158A,V161A ≅ Mre11V160A, V161A ≅ Mre11I158R ≅ Mre11V160A > Mre11V161A ≅ Mre11I158A,V160A.

Mre11I9R is defective in MRX assembly

Mre11 displays strong two-hybrid interactions with Xrs2 and Rad50 (62). Using both vegetative and meiotic two-hybrid reporter cells (11,59), we found that Mre11I158R (as for Mre11) exhibited notable two-hybrid interactions with Xrs2 and Rad50. In contrast, Mre11I9R failed to interact with Rad50 and Xrs2 in the same assay (Table 1).

Next, we constructed three W303 strains (MRE11-6HA, mre11) that express an HA-epitope-tagged wild-type and mutant Mre11, respectively. Compared to MRE11-6HA and mre11, the mre11 mutant displayed a slow vegetative growth phenotype on the YPD plate (Figure 1B). The hierarchy for resistance to the DNA damage agent methyl methanesulfonate (MMS) is MRE11 ≅ MRE11-6HA > mre11 >> mre11 ≅ mre11Δ (Figure 1B). Again, to minimize the potential effects of arginine replacements on protein structure or folding, we also expressed and compared different Mre11SIM1-6HA and Mre11SIM2-6HA mutants for MMS resistance in a SK1 mre11Δ mutant (Figure 1C). For all the mutants examined here, their two-hybrid interactions with Smt3 correlated well with the hierarchy for MMS resistance (Figure 1C): MRE11-6HA > mre11 > mre11 > mre11 ≅ mre11 ≅ mre11Δ and MRE11-6HA >> mre11 ≅ mre11 > mre11 ≅ mre11 ≅ mre11 > mre11 ≅ mre11 > mre11Δ.

The S. cerevisiae MRX is also required for telomere length maintenance. Like mre11Δ, mre11 exhibits severe phenotypes in DNA repair and telomere maintenance. Mre11D16A is defective in MRX assembly and lacks endonuclease activity (56). We first confirmed that Mre11D16A exhibits defects in two-hybrid interactions with Smt3 and Rad50 but not Xrs2 (Table 1). Next, we showed that the hierarchy for telomere length maintenance was MRE11 ≥ MRE11-6HA ≅ sae2Δ > mre11 > mre11 > mre11 ≅ mre11Δ (Figure 1D). Genomic DNA was isolated from all strains, digested with XhoI and Southern hybridized for telomeric sequences. Consistent with previous reports, the telomeres in mre11Δ are shorter than those in mre11 (56) and the sae2Δ deletion does not affect the maintenance of normal telomere length (63). These results indicate that mre11 is phenotypically similar to mre11Δ, confirming that Mre11I9R is defective in MRX assembly. In contrast, mre11 showed much milder defects in MMS-resistance (Figure 1B,C) and telomere maintenance (Figure 1D). Thus, Mre11I158R apparently can form a functional (at least partly) MRX complex.

D16 is one of seven essential and evolutionarily-conserved amino acid residues that coordinate two Mn2+ ions to form the phosphodiesterase active site (64). Because Mre11D16A is defective in association with the Mn2+, A16 and its neighboring amino acid residues (such as SIM1: IRIL, residues 9–12) might be improperly folded. As a result, Mre11D16A is unable to interact with SUMO. Alternatively, D16 should be classified as being part of SIM1, or the SIM1-CSM interaction might mask D16 from chelating with Mn2+. In this scenario, Mre11 first recruits CSMs to facilitate Mre11 folding and/or MRX assembly. After replacement of CSMs with Mn2+, the folded Mre11 protein or the assembled MRX complex is no longer able to associate with CSMs via SIM1. This possibility is more logical, because it is consistent with the results of our two-hybrid (Table 1) and pull-down experiments (see below) showing neither Mre11I9R nor Mre11I158R is unable to interact with CSMs.

Mre11I9R and Mre11I158R are differentially defective in their ability to repair spontaneous and MMS-induced DNA damage

When DNA replication forks are stalled by spontaneous or MMS-induced DNA lesions, checkpoint proteins are activated to stabilize those stalled forks. Stalled forks eventually collapse and produce broken DSBs. The recovery of DNA replication is typically controlled by MRX-dependent repair of DSBs (65–67). Here, we compared the recovery of DNA replication in MRE11-6HA, mre11 and mre11 (Figure 2A). Cells were first arrested in G1 by mating pheromone α-factor and then released into S phase; this time point was referred to as T0. After 10 min, different amounts of MMS (0, 0.01%, 0.033%) were added to the cells for 45 min. This time point was referred to as T55, i.e. 55 min after release from α-factor arrest. Next, MMS was removed from the cells for 3 h (referred to as T235) to determine the recovery of DNA replication (Figure 2A). DNA replication was monitored by fluorescence-activated cell sorting (FACS; Figure 2B) and pulsed-field gel electrophoresis (PFGE; Figure 2C). Checkpoint activation was determined by Western blot analysis with anti-Rad53 antisera (Figure 2D). Rad53 is an essential DNA damage checkpoint protein kinase that is required for cell-cycle arrest in response to DNA damage. Rad53 activation occurs through direct phosphorylation by Mec1ATR and Tel1ATM followed by Rad53 autophosphorylation (68,69). The hyperphosphorylated Rad53 protein is distinguishable from the unphosphorylated Rad53 protein, as the latter migrates faster in a SDS-PAGE gel (68).

Figure 2.

DNA replication during recovery from MMS-induced stalling. (A) Yeast cells were grown to mid-log phase, arrested in G1 by the presence of α-factor (i.e. T0) and then released from G1 arrest into YPD. After 10 min, MMS (0, 0.01% or 0.033%) was added for 45 min (i.e. T55). Cells were then washed extensively and incubated in YPD at 30°C for an additional 180 min (i.e. T235). Samples were taken at three different time points: T0 (lanes 1, 4, 7, 10, 13, 16, 19, 22 and 25), T55 (lanes 2, 5, 8, 11, 14, 17, 20, 23 and 26) and T235 (lanes 3, 6, 9, 12, 15, 18, 21, 24 and 27). (B) Progression of DNA replication. Yeast cells were harvested at the indicated time points after being transferred into sporulation medium. The cells were then stained with SYTOX Green, and the DNA content was measured by FACS. (C) PFGE. Yeast chromosomes were separated by PFGE and then visualized by staining with ethidium bromide. (D) DNA damage checkpoint. Total protein extracts were prepared and analyzed by immunoblot using a HA antibody and a Rad53 antibody. Hsp104 was used as a loading control. Rad53 phosphorylation—an indicator of DNA damage checkpoint activation—was assayed as the phosphorylation-dependent shift of the protein.

The steady-state levels of these three HA-tagged Mre11 proteins were not significantly different at T0, T55 or T235 (Figure 2D). To examine their protein stability, we performed cycloheximide shut-off experiments (Supplementary Figure S1A). Protein synthesis was inhibited by 200 μg/ml of cycloheximide added into the vegetative cultures. Samples were taken at 0, 30, 60, 90, 120 and 180 min after the addition of cycloheximide (Supplementary Figure S1A, lower panel). The immunoblotting results revealed that Mre11-6HA, Mre11I9R-6HA and Mre11I158R-6HA exhibited similar half-lives (t1/2 ≈45-60 min) in the presence of cycloheximide up to 180 min (Supplementary Figure S1C), suggesting that these two SIM mutations have little or no effect on Mre11-6HA protein stability during vegetative growth.

Our results also indicate that, in the absence of MMS, the DNA damage checkpoint was activated at T55 and T235 in mre11 but not in MRE11-6HA or mre11, as a proportion of the Rad53 proteins had become hyperphosphorylated and migrated more slowly in an SDS-PAGE gel (Figure 2D, middle panel). In the presence of 0.01% MMS, the hierarchy for the DNA damage checkpoint responses was mre11 > mre11 > MRE11-6HA. In contrast, in the presence of 0.033% MMS, the DNA damage checkpoint responses in all three strains were distinctly activated at T55. The activated checkpoint in MRE11-6HA was then attenuated at T235, as the majority of the Rad53 proteins became unphosphorylated. Hyperphosphorylated Rad53 was still detected in mre11 at T235 (Figure 2D, middle panel). These results suggest that mre11 cannot repair both spontaneous and MMS-induced DSBs due to its defect in MRX assembly. The mre11 mutant, compared to MRE11-6HA, is partly defective in repairing MMS-induced DSBs.

The results of PFGE (Figure 2C) are also consistent with those of MMS sensitivity (Figure 1B) and checkpoint activation (Figure 2D); the tendency for recovery of MMS-induced replication fork stalls is MRE11-6HA > mre11 > mre11. Due to the presence of forks and bubbles that impede chromosome migration, the incompletely replicated chromosomes did not enter the gel at T55 (Figure 2C, lanes 8, 14, 17 and 26). Chromosome replication was recovered and almost completed in MRE11-6HA at T235 (Figure 2C, lane 9), enabling the chromosomes to enter the gel. In contrast, a significant proportion of the chromosomes was never completely replicated in mre11 and mre11 at T235 (Figure 2C, lanes 18 and 27).

The MRX complex associates with the conjugated Smt3 moieties (CSMs) via SIM1 and SIM2 in vivo

Next, we performed co-immunoprecipitation experiments to compare the capability of Mre11-6HA, Mre11I9R-6HA and Mre11I158R-6HA in recruiting Smt3 and/or CSMs in vivo. The MRE11-6HA, mre11 and mre11 vegetative cells were treated with 0.3% MMS for 90 min. These cells also expressed V5-tagged Smt3 proteins (V5-Smt3). Total cell lysates were then prepared and used to carry out immunoprecipitation. The anti-HA affinity resin was used to pull down Mre11-6HA, Mre11I9R-6HA or Mre11I158R-6HA. Both total cell lysates and the bound protein complexes were subjected to immunoblotting analysis with anti-HA and anti-V5 antibody, respectively. The apparent molecular weight (Mrapp.) of V5-Smt3 determined by SDS-PAGE was ≈15 000 (Figure 3A, lanes 1-3). Our results revealed that MMS induced much more V5-tagged CSMs (Mrapp. ≥ 30 000) in MRE11-6HA and mre11 than that in mre11 (Figure 3A, lanes 4–6). Since MRX is a positive regulator for DSB-induced global SUMOylation (15) and Mre11I9R-6HA cannot form a functional MRX complex, the mre11 mutant is defective in DSB-induced global SUMOylation. In contrast, mre11, compared to mre11Δ or mre11, confers higher MMS resistance (Figure 1) but still can partly repair DSBs (Figure 2), so we inferred that RPA (24) or other novel factors in mre11 might promote MMS-induced SUMOylation during DSB repair.

Figure 3.

Mre11 promotes MMS-induced protein SUMOylation. (A) Mre11I9R-6HA and Mre11I158R-6HA mutants are defective in recruiting conjugated SUMO moieties (CSMs). Total cell lysates from vegetative cells (MRE11-6HA V5-SMT3, mre11) untreated or treated with 0.3% MMS were prepared and used to carry out chromatin precipitation. An immobilized anti-HA affinity resin (Sigma) was used to pull down HA-tagged Mre11 proteins. The bound protein complexes and total cell lysates were subjected to immunoblotting analysis with anti-V5 antibody (top and middle panels) and anti-HA antibody (bottom panel), respectively. (B,C) The Mre11I158R-6HA mutant is defective in promoting SUMOylation of DNA repair proteins. TAP (Tandem affinity purification)-tagged Rad59 (B) and TAP-tagged Rfa1 (C) in yeast cells untreated or treated with 0.3% MMS were analyzed by immunoblotting using the peroxidase-anti-peroxidase (PAP) antibody (left panel). TAP-tagged proteins were immunoprecipitated using IgG-Sepharose (Sigma), washed and eluted with loading buffer, before separating by SDS-PAGE and immunoblotting with anti-V5 antibodies (right panel). Arp7 was used as a loading control.

Next, we showed that Mre11-6HA co-immunoprecipitated much more V5-tagged CSMs (but not the V5-tagged Smt3 monomers) than Mre11I9R-6HA or Mre11I158R-6HA in the absence (Figure 3A upper panel, lanes 7-9) or presence (Figure 3 upper panel, lanes 10-12) of 0.3% MMS. These results are consistent with those of our two-hybrid assays (Table 1), indicating that Mre11 and/or the MRX complex preferentially interact with CSMs but not the Smt3 monomer via SIM1 and SIM2.

The immunoblotting results with anti-HA antibodies (Figure 3 lower and right panel, lanes 7-12) also revealed that Mre11-6HA and Mre11I158R-6HA, but not Mre11I9R-6HA, are phosphorylated in response to MMS (Figure 3 lowest panels, lane 4-6, 10-12). Notably, all these three Mre11-6HA proteins exhibited similar higher mobility patterns in SDS-PAGE (Figure 3 lowest right panels, lanes 7-12), indicating that neither SIM1 nor SIM2 significantly affects Mre11 SUMOylation (11). From these results, we can infer SUMOylation of Mre11 occurs prior to the assembly of the MRX complex. Consistent with the results of our two-hybrid assays (Table 1), the results here also indicate that Mre11-dependent global SUMOylation is mediated via the Mre11-CSM ensembles rather than the SUMOylated Mre11.

Finally, the SUMOylation of two TAP (tandem affinity purification)-tagged DNA repair proteins (Rad59 and Rfa1) was profoundly affected by mre11Δ or sae2Δ (15). Using the same approach, we found that the hierarchy for the levels of SUMOylated Rad59-TAP (Figure 3B) or Rfa1-TAP (Figure 3C) in response to 0.3% MMS were MRE11-6HA > mre11 > mre11Δ, though mre11 could induce global protein SUMOylation in response to MMS (Figure 3A).

In conclusion, the wild-type vegetative cells can repair both spontaneous and MMS-induced DSBs, initiate DSB-induced global SUMOylation and maintain normal telomere length during vegetative growth. The mre11 mutant fails to assemble the MRX complex, and thus it is defective in all physiological functions of the MRX complex. In contrast, mre11 can repair spontaneous DSBs and maintain normal telomere length, while it is partly defective in repairing MMS-induced DSBs and thus confers higher MMS resistance than mre11 and mre11Δ. Finally, the mre11 mutant can induce global SUMOylation in response to MMS, but is less efficient than the wild-type cells in promoting SUMOylation of DNA repair proteins.

Mre11I9R and Mre11I158R differentially affect the formation of Spo11-induced DSBs

S. cerevisiae MRX is required for both initiation and processing of Spo11-induced DSBs (25,26). To reveal the meiotic functions of Mre11SIM1 and Mre11SIM2, we constructed three SK1 diploid strains: MRE11-6HA, mre11 and mre11. Tetrad dissection analyses revealed that MRE11-6HA generated many viable spores (>97%), whereas mre11 and mre11 yielded no viable spores (<1%) (Table 2).

Table 2.

Sporulation efficiency and spore viability

Strain	Sporulation				Spore viability
	Ascus with 4, 3, 2, 1 and 0 spores
	4	3	2	1 or 0
MRE11–6HA	69%	4%	20%	8%	98% (n = 208)
mre11I9R-6HA	4%	6%	39%	53%	0% (n = 112)
mre11I158R-6HA	2%	2%	50%	48%	0% (n = 52)
mre11P84S-6HA	49%	5%	21%	25%	89% (n = 212)
mre11T188I-6HA	4%	4%	33%	59%	0% (n = 128)
mre11P84S, T188I(mre11S)-6HA	1%	3%	13%	83%	0% (n = 52)
MRE11–6HA, sae2Δ	4%	3%	33%	60%	1% (n = 104)
mre11I9R-6HA, sae2Δ	0%	2%	29%	69%	n.d.*
mre11I158R-6HA, sae2Δ	1%	3%	30%	68%	0% (n = 58)
SAE2–3HA	68%	5%	17%	10%	99%(n = 216)
sae2K97R-3HA	66%	3%	22%	9%	99% (n = 208)
sae2K319R-3HA	67%	2%	18%	13%	96% (n = 216)
sae2K97R, K319R-3HA	64%	3%	19%	14%	99% (n = 216)

Sporulation efficiencies were counted after 5 days on sporulation media at 30°C. To score for spore viability, only tetrads (but not dyads or triads) were dissected on YPD. *n.d. (not determined).

All yeast strains were then induced to undergo relatively synchronous meiosis in the sporulation medium (SPM). At the indicated time points, cells were harvested to monitor key meiotic events as described previously (11,37). First, the FACS results indicate that these three strains exhibited no apparent differences in their completion of pre-meiotic DNA replication, as the majority of cells (>80%) became 4N at the 6h time point (Figure 4A). Second, we performed 4, 6-diamidino-2-phenylindole (DAPI) staining to monitor if meiotic progress in terms of MI nuclear division was delayed in mre11 and mre11. The completion of MI nuclear division took about ≈4.5 h for 50% MRE11-6HA and ≈6.0 h for 50% mre11 or 50% mre11 cells, respectively. Moreover, more mre11 (45%) and mre11 (40%) than MRE11-6HA (≈15%) never underwent MI nuclear division (Figure 4B).

Figure 4.

Pre-meiotic DNA replication, initiation and resection of Spo11-induced DSBs. (A) Progression of pre-meiotic DNA replication. Yeast cells were harvested at the indicated time points after being transferred into sporulation medium. The cells were then stained with SYTOX Green, and the DNA content was measured by FACS. (B) Timing of nuclear division (MI). The number of DAPI–stained foci per cell (n = 200) was counted for each time point after transfer of cells into sporulation medium (SPM). (C, D) Detection of DNA double strand breaks (DSBs) at the YCR047C –a DSB hot spot on chromosome III. Yeast genomic DNAs were digested by BglII, separated by electrophoresis on a 1% agarose gel, probed with YOR052W and visualized using a Fujifilm phosphorimager. Unresected DSBs at YOR047C DSB are indicated by arrows and those resected are indicated by an asterisk on the right, respectively. DSB signal/total lane signal ratio from the Southern blot as in (C) are shown (D).

Next, we examined DSB formation at YCR047C, a DSB hot spot on chromosome III. To more precisely quantify to the level of Spo11-induced DSBs, we introduced the sae2Δ (com1Δ) mutation into these three strains. The sae2Δ mutant (20,70,71), like rad50S (72) and mre11S (73), blocks the resection of Spo11-induced DSBs into 3′-end ssDNA tails. Genomic DNA isolated from sporulating cultures at various time points was digested with BglII, separated by gel electrophoresis, blotted and hybridized with the YCR052W DNA probe as described before (11) (Figure 4C). The order for the maximum overall quantity of DSBs at this specific site was MRE11-6HA (≈3%), mre11 (<1%), mre11 (≈7%), MRE11-6HA sae2Δ (≈12%), mre11Δ (<1%) and mre11Δ (≈7%) (Figure 4D).

The overall levels of Spo11-induced DSBs along chromosome IV were also revealed by PFGE (Figure 5A) and Southern hybridization using a FDC1 DNA probe (Figure 5B). The results further confirm that mre11 is defective in the formation of Spo11-induced DSBs because mre11Δ, compared to MRE11–6HA sae2Δ or mre11Δ, accumulated much less unresected FDC1-containing chromosome fragments (Figure 5B, left panel). This conclusion is also supported by the results of immunoblot time course analyses using antibodies specifically against phosphorylated Hop1-T318 and phosphorylated Zip1-S75 (Figure 6). We found that mre11 (Figure 6A, middle panels) and mre11Δ (Figure 6B, middle panels) generated very low levels of phosphorylated Hop1-T318 and phosphorylated Zip1-S75. In contrast, more phosphorylated Hop1-T318 and Zip1-S75 proteins appeared in MRE11-6HA, mre11 (Figure 6A), MRE11-6HA sae2Δ and mre11Δ at the 3 h time point and thereafter (Figure 6B).

Figure 5.

The overall distribution of Spo11-induced DSBs along chromosome IV. Yeast chromosomes were separated by PFGE (A), analyzed by Southern hybridization with a FDC1 (YDR539W) DNA probe and visualized using a Fujifilm phosphorimager (B). All sporulation time-course experiments were repeated twice, and the representative results are shown.

Figure 6.

Meiotic checkpoint activation. (A, B) Immunoblotting time-course analysis of yeast cells in wild-type (A) and sae2Δ (B) backgrounds. All of the antibodies used have been described previously (12,36,54). Hsp104 was used as a loading control. (C) Quantitation of immunoblotting results in (A and B) was determined as previously described. The relative levels of different phosphoprotein versus Hsp104 (loading control) at each time point are shown. (D) Gel mobility shift analysis. Total cell lysates of meiotic cells at 6 h after being transferred into sporulation medium were treated with calf intestinal alkaline phosphatase (CIAP; 120U) in the absence or presence of 2-glycerophosphate (2PG; 16 mM). (E) Western blot time-course analysis of MRE11-6HA and mre11 in sae2Δ and dmc1Δ backgrounds.

Mre11I158R is phenotypically similar to sae2Δ, but not dmc1Δ, in processing of Spo11-induced DSBs

Next, we introduced the dmc1Δ mutation into these three strains to show that mre11 is phenotypically equal to sae2Δ during meiosis. Unlike sae2Δ, the dmc1Δ mutant removes the Spo11-oligonucleotide complex from meiotic DNA and accumulates unrepaired 3′-end ssDNA tails (74). We found that MRE11-6HA dmc1Δ cells accumulate shorter, YCR052W-containing BglII digested DNA fragments due to excessive DSB resection (Figure 4C, right panel). In contrast, the meiotic DSBs at YCR047C were not resected in mre11Δ (right panel) as in mre11 (middle panel) or in mre11Δ (left panel). The order for the overall quantity of DSBs at YCR047C was MRE11-6HA dmc1Δ (≈12%), mre11Δ (<1%) and mre11Δ (≈7%) (Figure 4D, right panel).

The results of PFGE (Figure 5A, right panel) and Southern hybridization (Figure 5B, right panel) also indicate that MRE11-6HA dmc1Δ cells accumulate high levels of shorter, FDC1-containing chromosome fragments due to both multiple breaks and excessive DSB resection. In contrast, mre11Δ (right panel), as for MRE11-6HA sae2Δ (left panel) and mre11Δ (right and left panels), accumulated longer and unresected FDC1-containing chromosome fragments (Figure 5B). We conclude that mre11, like sae2Δ, blocks the resection of Spo11-induced DSB ends.

Our conclusion is further supported by the results of immunoblot time course analyses (Figure 6) because only Tel1ATM (but not Mec1ATR) was activated in mre11 as in sae2Δ and rad50S (12,36,53). Tel1ATM is recruited by the MRX complex to unresected DSBs, whereas Mec1ATR is recruited to replication protein A (RPA)-coated ssDNA tails. The hierarchy for the steady-state levels of phosphorylated Hop1-T318 and phosphorylated Zip1-S75 was MRE11-6HA SAE2 > MRE11-6HA sae2Δ ≅ mre11 ≅ mre11Δ (Figure 6A-C) and MRE11-6HA dmc1Δ >> mre11Δ ≅ mre11Δ ≅ mre11 (Figure 6E). Due to the lack of 3′-end ssDNA tails in mre11 (Figure 6A,E), mre11Δ (Figure 6B,E) and mre11Δ (Figure 6E), these three strains accumulated low but detectable levels of phosphorylated Hop1-T318 and Zip1-S75 proteins. In contrast, both Hop1-T318 and Zip1-S75 were hyperphosphorylated in the sporulating MRE11-6HA dmc1Δ cells (Figure 6E, right panels). Therefore, mre11, as for sae2Δ or rad50S, does not affect Xrs2-dependent Tel1ATM activation, further confirming that the MRX complex is properly assembled in mre11. Notably, all four strains (Figure 6E) produced similar levels of phosphorylated H2A-S129 (γH2A) during meiosis. H2A-S129 phosphorylation occurs during the onset of pre-meiotic DNA replication and is independent of Spo11-induced DSBs (12). In contrast, the levels of γH2A were much reduced in mre11Δ (Figure 6B,C) due to the lack of properly assembled MRX complex.

Mre11 and Mre11I158R, but not Mre11I9R, are phosphorylated during meiosis

Notably, before they were transferred to the sporulation medium (i.e. 0-h), all Mre11-6HA proteins in the three strains exhibited the same mobility in an SDS-PAGE gel (Figure 3A,B, upper panel). Mre11I9R-6HA exhibited the same mobility throughout sporulation, whereas Mre11-6HA and Mre11I158R-6HA were post-translationally modified at the 3-h time point and thereafter, migrated more slowly in the SDS-PAGE gel. A dephosphorylation assay was performed as described previously (54) using calf intestinal alkaline phosphatase (CIAP) and 2-glycerophosphate (2-GP), a general phosphatase inhibitor. The results reveal that the faster and slower migrating species represent nonphosphorylated and phosphorylated proteins, respectively (Figure 6D). These results suggest that Mre11 phosphorylation occurs in the context of the MRX complex but is independent of Mre11SIM2. Although Mre11 is a SUMOylated protein during vegetative growth (Figure 3A, lowest right panel), we hardly detected any SUMOylated Mre11-6HA or Mre11I158R-6HA in meiotic cells by immunoblotting (Figure 6A,B).

The steady-state levels of these three HA-tagged Mre11 proteins in meiosis were also compared by cycloheximide shut-off experiments (Supplementary Figure S1B). Protein synthesis was inhibited by 200 μg/ml of cycloheximide added into the meiotic cultures at the 3-h time point. Samples were taken at 30, 60, 90, 120 and 180 min after the addition of cycloheximide (Supplementary Figure S1B, lower panel). We found that Mre11I9R-6HA (t1/2 ≈40 min) is less stable than Mre11-6HA (t1/2 ≈60 min) and Mre11I158R-6HA (t1/2 ≈70 min). These results might account for the lower steady-state levels of Mre11I9R-6HA protein in mre11 (Figure 6A, upper panel) and mre11Δ (Figure 6B, upper panel) after 6-h in SPM.

A comparison between mre11 and mre11S

Our results indicate that mre11 allows initiation but not processing and repair of Spo11-induced DSBs similar to sae2Δ, rad50S and mre11S. The mre11S allele contains two point mutations, P84S and T188I (73). P84 and T188 are evolutionarily conserved. In the S. pombe Mre11-Nbs1 complex crystal structure (58), P93 and S207 (the amino acid equivalents of P84 and T188) reside nearby SIM1 and SIM2, respectively (Supplementary Figure S2). However, neither Mre11P84S nor Mre11T188I exhibit apparent defects in two-hybrid interactions with Smt3, Ubc9, Siz2, Xrs2 or Rad50 (Table 1). We found that the S. cerevisiae mre11S-6HA homozygous diploid, like the S. pombe mre11 homozygous diploid (75), displayed no apparent (or modest) defects in sporulation or spore viability (≈89%). In contrast, mre11 and mre11S-6HA (mre11S-6HA) are unable to produce any viable spore (Table 2). Therefore, the T188I mutation is the main cause for mre11S (64). As S. cerevisiae T188 and S. pombe S207 are located in a surface loop that contains conserved positively-charged residues critical for DNA binding (Supplementary Figure S2), the T188I mutation might distort Mre11 DNA binding during processing of meiotic DSBs (58,64,76). Since T188, as for S. pombe S207 (58), is close to SIM2, we speculated that the Mre11SIM2-CSM interaction might affect the Mre11 DNA binding during meiosis. Next, we carried out meiotic nuclear spread immunostaining experiments to determine whether Mre11-6HA, Mre11I9R-6HA and Mre11I158R-6HA were properly targeted to meiotic chromosomes using the anti-HA antisera (Figure 7A). It is known that sae2Δ or rad50S accumulate Mre11 foci, which is not seen in the wild type (27). Our cytology data (Figure 7B) revealed that the Mre11-6HA foci appeared at the 3.5-h sporulation time point in wild-type and sae2Δ, but only persisted at the 5-h sporulation time point in sae2Δ. Neither of these two strains examined here formed Mre11I9R-6HA foci because Mre11I9R-6HA is unable to assemble the MRX complex. In contrast, Mre11I158R-6HA foci accumulated in both wild-type and sae2Δ, indicating that the Mre11SIM2-CSM interaction is dispensable for the Mre11 DNA binding during meiosis. It is of interest to further determine how the Mre11SIM2-CSM interaction affects the catalytic function of Mre11 in processing Spo11-induced DSBs.

Figure 7.

Accumulation of Mre11I158R-6HA foci along meiotic chromosomes in the wild-type and sae2Δ meiotic cells. (A) Nucleoids prepared at indicated meiotic time points were stained by anti-HA serum (red1) and DAPI (blue). Representative images are shown. The white bar represents 5 μm. (B) Quantitation of Mre11–6HA, Mre11I9R-6HA and Mre11I158R-6HA foci per nucleoids meiotic time points (A and B). The percentage of nucleoids with more than five anti-HA positive foci was obtained from 120–150 randomly selected nucleoids.

Mre11 is unlikely to recruit SUMOylated Sae2 for end processing of Spo11-induced DSBs

Sae2 can directly interact with Mre11 (77) and the MRX complex (30) in vitro. There are two putative SUMOylation sites (K97 and K319) in Sae2. The SUMOylation at K97 of Sae2 increases both soluble Sae2 and the MRX function in dsDNA end resection during vegetative growth (21). We postulated that Mre11SIM2 might recruit SUMOylated Sae2 to reconstitute a functional MRX-Sae2 dsDNA endonuclease (at least) during vegetative growth. However, this hypothesis is not applicable to the repair of Spo11-induced DSBs: (i) Sae2 SUMOylation might not occur in vivo during meiosis because SUMOylated Sae2-3HA was not detected in meiotic cells by immunoblot time course experiments (Supplementary Figure S3). (ii) Both sae2 (21) and sae2 mutants, as compared to wild-type SAE2-3HA diploid cells, underwent normal sporulation and produced (>90%) viable spore (Table 2). (iii) It has been reported that the phosphorylation of Sae2 is required to initiate resection and to improve the efficiency of resection through cooperation with the MRX complex during meiosis (78,79). Our results also confirm that Sae2-3HA, Sae2K97R-3HA and Sae2K97R,K319R-3HA are phosphorylated during meiosis (Supplementary Figure S3).

DISCUSSION

The results in this report indicate that S. cerevisiae Mre11 can non-covalently recruit CSMs (particularly the poly-SUMO chains or their conjugates) to facilitate its assembly and functions during both vegetative growth and meiosis. Mre11 first recruits CSMs via SIM1 to facilitate MRX assembly and/or Mre11 folding, after which, Mre11 recruits CSMs via SIM2 to promote its interaction with the SUMO enzymes (E2/Ubc9 and E3/Siz2). These enzymes then induce global SUMOylation at DSB sites, particularly SUMOylation of DNA repair enzymes. However, the Mre11SIM2-CSM interaction does not affect Xrs2-dependent Tel1 activation. We also show that the primary defect of the mre11 mutation, as for rad50S, sae2Δ and mre11S, is to block the resection of the Spo11-induced DSB ends during meiosis. Mre11SIM1 and Mre11SIM2 are evolutionarily conserved in S. cerevisiae and S. pombe. It would be interesting to investigate further whether the S. pombe Mre11SIM1 and Mre11SIM2 are also important for the assembly of MRN and/or its function in mitosis and meiosis.

Our results support the hypothesis that CSMs can non-covalently prevent premature aggregation by increasing the water solubility of individual protein subunits (8,9) prior to their assembly into a functional MRX complex (10,11,37). Similar to this scenario, SUMOylation of Sae2 increases both soluble Sae2 and the Sae2's function in DNA end resection (21). To coordinate this task in a timely manner, these CSMs or SUMOylation are negatively regulated by deSUMOylation proteases and SUMO-targeted ubiquitin ligases. Consistent with this, mutations in either deSUMOylation protease genes (ulp1Δ) (16,17) or SUMO-targeted ubiquitin ligase genes (slx5Δ, slx8Δ) (18,19) cause pronounced effects on DSB repair and genomic instability. Accordingly, we suggest that the SIM2-dependent MRX-CSM interaction can provide a regulatory function in regulating MRX function during DSB repair.

A key finding of this study is that SUMOylation of S. cerevisiae Mre11 is unlikely to be required for the formation of the MRX complex and the DSB-induced SUMOylation of DNA repair proteins (Figure 3A). Mre11-6HA, Mre11I9R-6HA and Mre11I158R-6HA all can be SUMOylated during vegetative growth. However, we failed to detect SUMOylated Mre11–6HA by immunoblotting during meiosis (Figure 6). Our results also indicate that Mre11-6HA and Mre11I158R-6HA, but not Mre11I9R-6HA, are phosphorylated and can form MRX complexes during both vegetative growth and meiosis. Because Mre11I9R-6HA is unable to form an MRX complex, we further inferred that Mre11 SUMOylation and phosphorylation occur before and after Mre11 assembly, respectively. Further investigation will reveal whether phosphorylation of Mre11 is functionally relevant to MRX assembly, DSB-dependent checkpoint activation or global SUMOylation.

SUMOylation of Sae2 increases both soluble Sae2 and the MRX function in DNA end resection during vegetative growth (21). Our results have revealed that SUMOylated Sae2 protein not only was not detected by immunoblotting but also that it is functionally unimportant for meiosis, sporulation and spore viability (Supplementary Figure S3 and Table 2). Like Mre11, Sae2 is also a phosphorylated protein during meiosis. Phosphorylation of Sae2 Ser-267 by cyclin-dependent kinase 1 (Cdk1) is required to initiate meiotic DSB resection by allowing Spo11 removal from DSB ends (78,79). It is still unclear how phosphorylated Sae2 regulates the Sae2-Mre11 interaction. In S. pombe, Ctp1 is the functional counterpart of Sae2 in S. cerevisiae and CtIP in mammals (80). Sae2, Ctp1 and CtIP share limited homology (approximately 30 amino acids) at their C-terminal RHR motifs (80), which are responsible for dsDNA binding (81). Ctp1 also is referred to as an Nbs1 interacting protein (Nip1) because Nbs1 recruits phosphorylated Ctp1 to DSBs via its binding to the Nbs1 FHA domain of Ctp1 phosphorylated SXT motifs (Ctp1 residues 71-79) (82,83). In contrast, Sae2 interacts with Mre11 but not Xrs2 (77). Furthermore, the SXT motif is not conserved in Sae2. The coupling mechanism between MRX and Sae2 in S. cerevisiae may be different from that of MRX and Ctp1 in S. pombe.

Finally, we previously proposed that the meiosis-specific axial protein Red1 might recruit CSMs to activate Tel1 for Spo11-dependent Hop1 (or Hop1-T318) phosphorylation (11,12). The results in this report indicate that the Red1-CSM complex does not exert this function via an interaction with Mre11 (or the SIM2 in Mre11), because Xrs2-dependent Tel1 activation still occurs in the mre11 diploid during meiosis (Figure 6A). We intend to conduct further studies to decipher the coupling mechanism between Red1 and Tel1.