|Home | About | Journals | Submit | Contact Us | Français|
Structure-specific endonucleases resolve DNA secondary structures generated during DNA repair and recombination. The yeast 5′-flap endonuclease Slx1-Slx4 has received particular attention with the finding that Slx4 has Slx1-independent key functions in genome maintenance. Although Slx1 is a highly conserved protein in eukaryotes, no orthologs of Slx4 were reported other than in fungi. Here we report the identification of Slx4 orthologs in metazoa, including fly MUS312, essential for meiotic recombination, and human BTBD12, an ATM/ATR checkpoint kinase substrate. Human SLX1-SLX4 displays robust Holliday junction resolvase activity in addition to 5′-flap endonuclease activity. Depletion of SLX1 and SLX4 results in 53BP1 foci accumulation and H2AX phosphorylation as well as cellular hypersensitivity to MMS. Furthermore, we show that SLX4 binds the XPF and MUS81 subunits of the XPF-ERCC1 and MUS81-EME1 endonucleases and is required for DNA interstrand crosslink repair. We propose that SLX4 acts as a docking platform for multiple structure-specific endonucleases.
Maintenance of genome integrity depends on a rich variety of DNA repair and recombination mechanisms, many of which require cleavage of DNA secondary structures by structure-specific endonucleases.
A well-known example is the highly conserved family of XPF/MUS81 endonucleases (Ciccia et al., 2008). XPF-ERCC1 enzymes function as 3′-flap endonucleases in nucleotide excision repair (NER) and some homologous recombination (HR) mechanisms (Brookman et al., 1996; Fishman-Lobell and Haber, 1992; Niedernhofer et al., 2001; Schiestl and Prakash, 1990; Sijbers et al., 1996). MUS81-EME1Mms4 endonucleases are required for resolution of Holliday junctions (HJ) and DNA structures that precede HJ maturation generated during meiotic recombination and a subset of mitotic double-strand break (DSB) repair pathways (Boddy et al., 2001; Chen et al., 2001; Gaillard et al., 2003; Gaskell et al., 2007).
Paradoxically, despite their pivotal contribution to genome maintenance, structure-specific endonucleases constitute a potential source of genome instability. Indeed, cutting DNA opens a window of opportunity for chromosome rearrangements and it is likely that their action needs to be kept under tight control. In particular, coordination with upstream and downstream events of the repair mechanisms in which they are involved should ensure rapid processing of the appropriate substrate and restoration of the original chromosome structure. Despite their fundamental nature, the mechanisms that control structure-specific endonucleases remain poorly understood. Recent studies carried out in S. cerevisiae on the Slx1-Slx4 complex shed a new light on how this may be achieved in some cases (Flott et al., 2007; Flott and Rouse, 2005).
Slx1 and Slx4 were identified along with Mus81-Mms4Eme1 in a synthetic lethal screen with mutations that inactivate Sgs1, a member of the RecQ family of DNA helicases (Mullen et al., 2001). These helicases are thought to have an anti-recombinogenic function that unwinds and dissolves DNA secondary structures (Wu and Hickson, 2006). Slx1 and Slx4, were subsequently shown to be catalytic and regulatory subunits, respectively, of a 5′-flap endonuclease (Coulon et al., 2004; Fricke and Brill, 2003). Studies in S. cerevisiae and S. pombe revealed that Slx1-Slx4 plays a critical role in maintaining the integrity of the ribosomal DNA (rDNA) loci, which are made of tandem rDNA repeats (Coulon et al., 2004; Coulon et al., 2006; Kaliraman and Brill, 2002). Slx1-Slx4 was proposed to initiate DNA recombination by collapsing replication forks stalled at natural replication pause sites. Further investigations in budding yeast uncovered Slx1-independent activities of Slx4. The SLX4 gene interacts with genes involved in error-free DNA damage bypass and Slx4 is required for resistance to DNA alkylation during S-phase (Flott et al., 2007). This is thought to involve the interaction between Slx4 and the multi-BRCT domain protein Rtt107 (Flott et al., 2007; Roberts et al., 2006; Roberts et al., 2008). Rtt107 itself associates with Rtt101CUL4 that is part of a ubiquitin (Ub) E3 ligase proposed to control protein turnover at stalled replication forks and promote replication through natural pause sites and DNA damage (Luke et al., 2006; Zaidi et al., 2008). SLX4 was also found in genome wide screens for genes involved in the response to interstrand cross-linking (ICL) agents (Lee et al., 2005; Wu et al., 2004). Another Slx4-binding protein is Rad1, which together with Rad10 forms the budding yeast XPF-ERCC1 ortholog. Slx4 is essential for the removal of 3′-flaps during HR repair by single-strand annealing (SSA) and is proposed to recruit, along with the recently identified Saw1 protein, Rad1-Rad10 to ongoing SSA events (Li et al., 2008; Lyndaker et al., 2008). Taken together, these studies suggest that Slx4 coordinates multiple DNA repair and recombination mechanisms in budding yeast.
Surprisingly, although the Slx1 and Rad1-Rad10 partners of Slx4 are highly conserved from yeast to man, orthologs of Slx4 outside of the fungal kingdom have not been reported (Coulon et al., 2004; Mullen et al., 2001). Sequence comparisons of the fungal Slx4 proteins show that it has undergone rapid evolutionary divergence. Here we describe the identification of Slx4 orthologs in metazoa and the characterization of the SLX4 family of protein in eukaryotes. We present a biochemical characterization of the enzymatic function of the human SLX1-SLX4 complex. In addition to 5′-flap endonuclease activity, SLX1-SLX4 has a robust Holliday junction resolvase activity. Our data show that SLX1 and SLX4 repair endogenous DNA lesions and damage caused by the DNA alkylating agent methyl methanesulfonate (MMS). Furthermore, SLX4 but not SLX1 is required for ICL repair. Finally, we show that SLX4 associates with XPF and MUS81, suggesting that it coordinates the activities of multiple structure-specific endonucleases.
Whereas SLX1 is a highly conserved protein from yeast to man, with an N-terminal Uri nuclease domain (Aravind et al., 1999) and a C-terminal PHD finger (Figure 1A), no SLX4 orthologs are known other than in fungi (Coulon et al., 2004; Mullen et al., 2001). We therefore undertook a PSI-BLAST search using the conserved C-terminal domain (CCD) of fungal Slx4 proteins. Two potentially relevant and related hits were Drosophila melanogaster MUS312 and vertebrate BTBD12 proteins. MUS312 is a MEI9XPF binding partner essential for ICL repair and meiotic recombination (Yildiz et al., 2002). The first two properties are reminiscent of S. cerevisiae Slx4, suggesting that Slx4 and MUS312 may be orthologs. In support of this idea, we detected a strong yeast two hybrid (Y2H) interaction between MUS312 and D. melanogaster SLX1 (Figure 1B).
This finding prompted a PSI-BLAST search using MUS312, whereby we once again detected similarities to vertebrate BTBD12 proteins. Human BTBD12 was identified in two large-scale screens for ATM/ATR targets (Matsuoka et al., 2007; Mu et al., 2007), but is otherwise uncharacterized. Vertebrate BTBD12 proteins have large N-terminal extensions that are not found in fungal Slx4 proteins (Figure 1D). In addition to the BTB/POZ protein-protein interaction domain, these extensions have one or several C2HC zinc finger related to the UBZ4 family of Ub-binding domains (Hofmann, 2009) (Figure 1E). We also detected a conserved motif contained within a region required for the MUS312SLX4-MEI9XPF interaction (Yildiz et al., 2002), which we named MLR (MUS312/MEI9-interaction Like Region) (Figure 1F). Interestingly, this conserved motif is absent in yeasts, although there is some sequence homology within the rest of the MLR between yeasts and metazoa (Figure S1).
We found that human BTBD12 has a Y2H interaction with human SLX1 (Figure 2A), suggesting that BTBD12 is indeed the human SLX4 ortholog. Hereafter we refer to BTBD12 as SLX4. Truncation analyses showed that only full length SLX1 binds specifically to SLX4 (Figure 2A). The CCD of SLX4 is sufficient for its interaction with SLX1, indicating that the function of this domain is conserved throughout evolution (Figure 2B).
Having established that human SLX4 and SLX1 interact in Y2H assays, we next investigated whether they exist in a complex in human cells. We found that transiently expressed FLAG-SLX1 co-precipitated with full-length SLX4 or the SLX4-4 N-terminal truncation fused to GFP (Figure 2C) and that SLX1 and SLX4 colocalized in the nucleus (Figure S2).
To assess whether the MUS312SLX4-MEI9XPF and Slx4-Rad1XPF interaction is conserved in human cells, we probed for endogenous XPF in the FLAG-SLX1 immunoprecipitates. XPF readily co-precipitated with SLX1-SLX4 but not SLX1-SLX4-4 (Figure 2C), suggesting that XPF associates with the N-terminal domain of SLX4, which contains the MLR region (Figure 1D,F), but not with SLX1. ERCC1 also co-precipitated with SLX4 and not SLX4-4, indicating that SLX4 interacts with XPF-ERCC1 (Figure S3). Y2H assays showed that SLX4 interacts with XPF but not ERCC1, whereas neither subunit interacts with SLX1 (Figure 2D). Consistent with the immunoprecipitation results (Figure 2C), XPF did not associate with the SLX4-4 [684-1834] C-terminal fragment. However, it did associate with the SLX4-3 [1-1043] N-terminal fragment, which contains the MLR. Taken together, these data indicate that the SLX4-XPF interaction is direct, conserved from yeast to man, and that XPF binding involves the MLR domain.
Yeast Slx1-Slx4 complexes are 5′-flap endonucleases (Coulon et al., 2004; Fricke and Brill, 2003). To assess whether the human complex has this activity, the SLX1-SLX4 eluates described in Figures 2C and S4 were incubated with a model DNA stem loop substrate (SL) (Figure 3A). This substrate allows detection of single-strand (ss) incisions in duplex DNA on the 5′ or 3′ side of a junction with ss DNA, as shown for the 3′-flap endonuclease activity of human XPF-ERCC1 (Sijbers et al., 1996) and the 5′-flap endonuclease activity of S. pombe Slx1-Slx4 (Coulon et al., 2004). As shown in Figure 3A, human SLX1-SLX4 displayed a 5′-flap endonuclease activity, introducing a nick in the duplex DNA on the strand that runs from a 3′ double-strand (ds) to a 5′ ss configuration. Remarkably, the cutting patterns of human and fission yeast SLX1-SLX4 were essentially identical, occurring within 1 or 2 nucleotides from the ds/ss junction. In contrast, recombinant XPF-ERCC1 introduced cuts on the opposite strand of the junction. Although endogenous XPF and ERCC1 associated with over-produced SLX1-SLX4 (Figure 2C), no cleavage products indicative of XPF-ERCC1 were detected in reactions with immunoprecipitated human SLX1-SLX4 (Figure 3A). The XPF-ERCC1 activity may be overwhelmed by much more SLX1-SLX4 present in these assays (Figure S4).
Cation titration experiments with human SLX1-SLX4 showed that Mn++ is strongly favored over Mg++, with optimum cutting at 0.5-1 mM Mn++ (Figure 3B). This contrasted with the S. pombe complex that efficiently processed the SL in 0.5 mM Mn++, as well as in 0.5-5 mM Mg++ (Coulon et al. 2004).
To confirm that the endonuclease activity we were monitoring was SLX1-dependent, we engineered two SLX1 mutants where one of two highly conserved charged residues in the nuclease domain, arginine-41 or glutamate-82, were mutated to alanine. The equivalent residues are essential for the nuclease activity of S. pombe Slx1 (Coulon et al., 2004). Both mutations abolished the 5′-flap endonuclease activity of human SLX1-SLX4, confirming that the activity we detected in our FLAG-peptide eluates was SLX1-dependent.
D. melanogaster MUS312SLX4 is essential for chromosome segregation during meiosis and was proposed to collaborate with MEI9 to resolve HJs (Yildiz et al., 2002). We therefore investigated whether human SLX1-SLX4 processes HJs in vitro. We used the model substrate X12 capable of spontaneous branch migration within its 12 bp core of homology (Rass and West, 2006). The reactions shown in Figure 4A were carried out in parallel with the SL substrate reactions in Figure 3C, with identical conditions and amounts of substrate. In contrast to the partial digestion of the SL substrate, X12 was completely processed and converted into linear duplex products. This activity was SLX1-dependent, being abolished by the R41A and E82A mutations in the SLX1 nuclease domain (Figure 4A). Significant X12 processing was observed only when both SLX1 and SLX4 were co-produced in the cell but not with separately produced SLX1 and SLX4 proteins alone or combined (Figure S5). Human SLX1-SLX4 produced symmetric cleavage patterns within the region of homology on opposite strands (Figure 4B). SLX1-SLX4 also converted into linear duplex products a fixed intact or nicked X0 HJ lacking a central core of homology, as well as a model replication fork (RF), albeit less efficiently than the X12 structure (Figure S6). Dual incisions on opposite strands of the non-physiological fixed X0 showed partial symmetry (Figure S7).
We assessed the divalent cation requirements for processing of X12 and RF structures (Figure S8). Whereas SL cleavage was observed only in Mn++, some X12 and RF processing occurred in Mg++, but it was significantly less efficient than in Mn++. Noteworthy, X12 remained a favored substrate in both Mg++ and Mn++. This prompted us to test whether Mn++ may also help S. pombe Slx1-Slx4 to process a HJ into linear duplex products. No significant processing of X12 was seen in Mg++, in agreement with our previous report (Coulon et al. 2004). In contrast, we observed robust conversion of the HJ into linear duplex products in Mn++ (Figure S9), suggesting that this is a common feature of SLX1-SLX4 in eukaryotes in general.
To further characterize the HJ processing activity of human SLX1-SLX4, we compared its ability to resolve a χ structure with that of the canonical RusA bacterial HJ resolvase (Bolt and Lloyd, 2002). This much larger HJ substrate more closely resembles HJs found in vivo with a large 300 bp core of homology and long duplex arms ranging from 0.45 to 2.15 kb. It was previously used to characterize the HJ resolving activities of bacterial and phages resolvases (Sharples et al., 2004; Zerbib et al., 1997). Resolution of the χ structure yields two pairs of nicked duplex products following dual incisions on strands A and C or B and D (Figure 4C). Like RusA, SLX1-SLX4 efficiently resolved the χ structure into all four duplex products (Figure 4D). A bias towards dual incisions on strands B and D was observed for RusA in agreement with a previous report (Figure 4D) (Sharples et al., 2004). In contrast, like RuvC (Zerbib et al., 1997), SLX1-SLX4 did not display any bias (Figure 4D).
To gain insight into which part of SLX4 contributes to HJ resolution by SLX1, we co-produced C-terminal SLX4 fragments (Figure 2B) with full length FLAG-SLX1 in HEK cells (Figure S10). Robust HJ resolution was observed even with the shortest C-terminal SLX4-7 fragment consisting of just the CCD domain of SLX4 (Figure 4E). These results indicated that the SAP domain, which was proposed to be important in the substrate recognition and tethering of SLX1 to its substrate (Aravind and Koonin, 2000; Coulon et al., 2004; Fricke and Brill, 2003), is in fact not required for the DNA processing activity of the complex. HJ resolution remained more efficient than processing of a 5′-flap structure (Figure S11). The overall nuclease activity of the SLX1-SLX4-7 complex was superior to that of the SLX1-SLX4 complex on both structures, reflecting the fact that higher amounts of the SLX4-7 complex are produced and purified from HEK cells (Figure S10).
To confirm that the HJ resolving activity we were monitoring could be carried out by SLX1-SLX4 alone and did not require any co-purifying factor, SLX1-SLX4-7 was produced in E. coli and purified to near homogeneity (Figure 4F). This complex resolved X12 with an impressive efficiency (Figure 4G), and displayed the same incision pattern (Figure 4H) as immunoprecipitated FLAG-SLX1-GFP-SLX4 (Figure 4B). Using an asymmetric X12 structure, which contained an extended strand 1/strand 4 heteroduplex arm (Figure 4H), we tested whether HJ resolution by SLX1-SLX4 involved symmetric incisions across the junction and production of religatable nicks on linear duplex products. Close to 25% of resolution products contained readily religatable nicks indicating that SLX1-SLX4 is capable of HJ resolution by symmetric cleavage (Figure 4H, S12). Taken together, our data establish SLX1-SLX4 as a novel human HJ resolvase.
As a first step to assess the functions of SLX1 and SLX4 in human cells, we undertook an siRNA based strategy. Knockdown efficiencies were verified for each siRNA (Figures S13, S14). Spontaneous genome instability following SLX1 or SLX4 depletion was suggested by a significant increase in the number of 53BP1 foci compared to scrambled siRNA transfected cells (Figure 5A). This correlated with an induction of H2AX phosphorylation (Figure 5B). Elevated 53BP1 foci and H2AX phosphorylation indicate increased DNA damage or defects in DNA repair.
We next assessed the roles of SLX1 and SLX4 in response to exogenous DNA damaging agents. Consistent with slx1 and slx4 mutants in yeasts and mus312 mutants in D. melanogaster, depletion of SLX1 or SLX4 in human cells did not increase UV sensitivity (Figure 5C). In contrast, following MMS treatment, increased cell death and H2AX phosphorylation was observed in SLX1- or SLX4-depleted cells (Figure 5B,C).
Taken together, these data indicate that human SLX1 and SLX4 have important functions in the maintenance of genome integrity and in the tolerance to DNA damage inflicted by an alkylating agent.
Yeast Slx4 and fly MUS312 are required for ICL repair (Lee et al., 2005; Wu et al., 2004; Yildiz et al., 2002). This activity may involve interactions with their respective XPF orthologs, since XPF-ERCC1 complexes are key players in ICL repair. We assessed the impact of depleting SLX4 in human cells exposed to various ICL agents. An acute sensitivity to both cisplatin (cisPt) and nitrogen mustard (HN2) was observed in SLX4- but not SLX1- or mock-depleted cells (Figure 5C). Interestingly, SLX4-depleted cells also displayed a more sustained level of H2AX phosphorylation (Figure 5B, lane 9 compared to lanes 7 and 8).
SLX4 associates with SLX1 and XPF, suggesting that it might be a flexible targeting platform for structure-specific endonucleases. To explore this possibility further, we reanalyzed our earlier proteomic studies aimed at identifying proteins that interact with the structure-specific endonuclease MUS81-EME1. We found that both SLX1 and SLX4 were among the proteins present in partially purified human EME1 preparations (Figure 6A). To assess whether endogenous MUS81 co-purified with SLX1-SLX4, FLAG-eluates described in Figures S10 and used in nuclease assays shown in Figure 4E were probed with an anti-MUS81 antibody. As shown in Figure 6B, MUS81 co-immunoprecipitated with all SLX1-SLX4 to SLX1-SLX4-6 complexes, but not with the SLX1-SLX4-7 complex lacking the SAP domain. These data suggested that the association of MUS81 with the SLX1-SLX4 complex is mediated through SLX4 rather than SLX1 and involves the SAP domain of SLX4. MUS81 binding to SLX4 was further supported by the efficient co-purification of HA-MUS81 with FLAG-SLX4 (Figure 6C). Truncation analysis indicated that the first 86 residues of MUS81 contain the SLX4-binding region (Figure 6B). We further found that whilst GFP-SLX4 co-precipitated with full length FLAG-EME1, it did not co-precipitate with the FLAG-EME1 [1-467] fragment that cannot bind to MUS81 (Taylor and McGowan, 2008) (Figure 6D), suggesting that the association of SLX4 with MUS81 is not mediated through EME1.
Thus our data reveal that in human cells SLX4 associates with three different structure-specific endonucleases: SLX1, XPF-ERCC1 and MUS81-EME1.
The identification of metazoan orthologs of yeast Slx4 has allowed us to define the eukaryotic SLX4 family of proteins (Figure 1D). All members of the SLX4 family share at their carboxyl-terminus a CCD motif preceded by a SAP domain. While we show that the CCD is required for the conserved SLX1-SLX4 interaction (Figure 2B), our data reveal an as yet undescribed role of the SAP domain in mediating binding to the MUS81 subunit of the MUS81-EME1 endonuclease (Figure 6). Whether, the SLX4-MUS81 interaction is specific to human SLX4 or a general feature of SLX4 family members will need further investigation. Metazoan SLX4 proteins have a large N-terminal extension where we have mapped a region critical for the association of human SLX4 with the XPF subunit of the XPF-ERCC1 endonuclease (Figure 2E). This interaction is conserved in yeast and flies (Flott et al., 2007; Ito et al., 2001) (Yildiz et al., 2002). Interestingly, this large N-terminal extension harbors at least two additional conserved functional domains not found in fungal orthologs. The first domain is composed of one or two UBZ4 Ub binding motifs primarily found in proteins involved in the DNA damage response (Hofmann, 2009). The second one is the multifunctional BTB/POZ protein-protein interaction domain (Perez-Torrado et al., 2006), which has recently emerged as a central architectural element of adaptor subunits in cullin 3 Ub-ligases (Geyer et al., 2003; Xu et al., 2003). Interestingly, SLX4BTBD12 was recently found as an ATM/ATR substrate with proteins of the Ub-proteasome system involved in mammalian DNA damage checkpoint control (Matsuoka et al., 2007; Mu et al., 2007).
Taken together, our observations suggest that human SLX4 is at the crossroads of several DNA repair mechanisms and DNA damage signalling processes that involve protein ubiquitination and phosphorylation.
The mechanisms of HJ resolution in eukaryotic cells have been the object of intense effort and debate over recent years (Haber and Heyer, 2001; Symington and Holloman, 2008). Three pathways have been proposed in human cells: dissolution by the BLM/TOPIII/RIM1 helicase/topoisomerase complex (Wu and Hickson, 2003, 2006); resolution by MUS81-EME1 (Chen et al., 2001; Constantinou et al., 2002; Taylor and McGowan, 2008); or resolution by GEN1 (Ip et al., 2008). The relative contributions of these pathways are unknown. Our identification and characterization of human SLX1-SLX4 defines a fourth pathway.
HJ processing by MUS312SLX4 in flies (Yildiz et al., 2002) has been proposed to explain the severe chromosome segregation defect during meiotic recombination in mus312 mutants. It appears that meiotic crossing-over (CO) relies on different endonucleases in differents organisms. While flies rely essentially on MEI9XPF (Yildiz et al., 2002) and fission yeast on Mus81 (Boddy et al., 2001; Cromie et al., 2006), crossing-over in budding yeast appears to rely on multiple pathways (Hollingsworth and Brill, 2004). Interestingly, large scale transcriptome studies show that the highest SLX4 mRNA levels in mouse tissues are in testes, oocytes and fertilized eggs (Su et al., 2002), consistent with the idea that SLX4 is involved in meiotic recombination in mammals. Furthermore, the fact that SLX4 has remarkable HJ resolving activity when associated with SLX1 and associates with XPF-ERCC1 and MUS81-EME1, each involved in meiotic recombination in different organisms, constitutes further support for a possible role of SLX4 in meiotic recombination in mammals.
We provide evidence suggesting that both SLX1 and SLX4 are required to prevent spontaneous genome instability (Figure 5A,B). An important source of spontaneous genomic instability comes from the replication of genomic loci rich in repeated sequences such as the rDNA. In yeast, Slx1-Slx4 maintains the stability of the rDNA locus where it is thought to initiate HR by collapsing stalled replication forks (Coulon et al., 2004; Coulon et al., 2006; Kaliraman and Brill, 2002). It is conceivable that SLX1-SLX4 not only initiates HR by collapsing the stalled fork but also resolves HJs generated after fork recapture. Human rDNA loci contain a broad range of palindromic structures (Caburet et al., 2005), which lead to frequent temporary and permanent fork arrest (Lebofsky and Bensimon, 2005). Palindromic structures can form HJ-like structures efficiently resolved in vitro (Giraud-Panis and Lilley, 1997; Taylor and McGowan, 2008) and in vivo (Coté and Lewis, 2008) by HJ resolvases. The robust HJ resolving activity of human SLX1-SLX4 (Figure 4) may thus require tight control to prevent unscheduled conversion of palindromes into DSBs (Lewis and Coté, 2006). Interestingly, over-produced SLX1-SLX4 accumulates outside of the nucleoli (Figure S15).
Telomeres are other regions of the genome prone to replication-induced genomic instability (Gilson and Géli, 2007). Interestingly, XPF plays a role in telomere maintenance through its interaction with the telomeric capping protein TRF2 (Muñoz et al., 2005; Zhu et al., 2003) but this may be independent of its nuclease activity (Wu et al., 2007). Considering that SLX1 and SLX4 seem to prevent spontaneous genomic instability and that SLX4 binds XPF, it is conceivable that SLX4 and SLX1 may also participate in telomere maintenance. This possibility is strongly supported by our finding that TRF2 readily co-immunoprecipitates, independently of its XPF interaction, with FLAG-SLX4 (Figure S16).
By analogy to what has been described for yeast Slx4, it is probable that human SLX4 prevents spontaneous genome instability through the rescue of stalled or collapsed replication forks. In agreement, we find that SLX4 is also required for cell survival following MMS treatment, which potently blocks replication fork progression (Figure 5B,C) (Vázquez et al., 2008). In that respect, the presence of SLX4 and MUS81-EME1 in a same complex may also be relevant (Figure 6).
A striking consequence of SLX4 depletion in human cells is the acute sensitivity to ICL agents (Figure 5B,C). This is in agreement with the ICL-hypersensitivity described for fly mus312 and yeast slx4 mutants (Yildiz et al., 2002; Lee et al., 2005; Wu et al., 2004). In contrast, SLX1 depletion did not lead to any significant ICL-sensitivity (Figure 5C). Although consistent with the phenotype of S. cerevisiae slx1 mutants, this was somewhat surprising considering the clear MMS hypersensitivity we observed in SLX1-depleted human cells. Indeed, HN2 and cisPt induce a broad range of mono and divalent intra-strand adducts that can also hinder replication fork progression. Noteworthy, SLX1 and SLX4 depleted cells are not sensitive to UV (Figure 5C). UV, cisPt and HN2 have in common that the majority of DNA lesions they induce are efficiently repaired by NER. In contrast, repair of MMS induced damage is repaired by base excision repair (BER). Furthermore, the requirements for fork progression through MMS or UV damage are different. While HR and replicative DNA damage tolerance pathways are required, along with BER, to allow fork movement through alkylated DNA (Vázquez et al., 2008) they are not required for fork progression through irreparable UV damage (Lopes et al., 2006). The lack of UV-sensitivity following depletion of SLX1 or SLX4 combined with the cisPt and HN2 sensitivity of SLX4- but not SLX1-depleted cells strongly supports a role for SLX4 specifically in the processing of ICL. We postulate that SLX4 may have a key role in coordinating MUS81-EME1 and XPF-ERCC1 and coupling their action with the rest of the ICL repair machinery. In this regard, the presence of the UBZ4 Ub-binding motifs at the N-terminus of SLX4 may be of pivotal importance in coordinating the action of both endonucleases and the Fanconi anemia (FA) machinery. Following DSB induction at ICL sites, the FANCD2 and I are mono-Ub by the FA core complex (Patel and Joenje, 2007). This is an essential step in ICL repair and is required for subsequent recruitment of the FANCD2/I complex to chromatin by an as yet unknown mechanism. An appealing possibility is that SLX4 may play a role in this process by coordinating the action of the two endonucleases and the recruitment via its UBZ4 domain of mono-Ub FANCD2/I to the ongoing ICL repair process.
Human SLX4 shares with its yeast counterparts many functional features, i.e. it forms an active endonuclease with SLX1, associates with XPF-ERCC1 and is required for protecting the genome against both spontaneous and drug induced instability. However, human SLX4 also contains N-terminal UBZ4 and BTB/POZ domains, which suggests that it has additional functions at the interface of DNA repair/recombination pathways and regulatory mechanisms involving protein ubiquitination.
We propose that SLX4 acts as a platform involved in channeling to the appropriate substrates and/or coordinating with ongoing repair and recombination mechanisms, the action of at least three structure-specific endonucleases (Figure 7). Considering the marked spontaneous and drug induced phenotypes associated with depletion of SLX4 in human cells and the variety of DNA repair/recombination pathways it appears to be involved in, we believe that the identification of human SLX4 opens new avenues for understanding the processes involved in the maintenance of genome stability and the prevention of the onset of cancer and other human diseases.
See Supplemental Data.
Y2H assays were set up using the ProQuest Two-Hybrid System (Invitrogen) according to the manufacturers instructions. SLX4 and MUS312 could only be fused to the GAL4-AD because of autoactivation when they are fused to the GAL4-DBD.
See Supplemental data for information on cell growth conditions and siRNA transfection and validation.
For colony survival assays, Hela cells were plated in 60 mm dishes (500 cells) 48 h post siRNA transfection. After 15 h incubation, MMS, cisPt or HN2 were added for 1h in fresh media. Cells were then washed twice with PBS and incubated with fresh medium for 6-8 days. Media was removed for UVC irradiation. Cells were then incubated as above in fresh media. Cells were finally fixed and stained for colony counting. All experiments were performed in triplicate.
HEK 293F cells from 10 cm plates, transiently transfected with mammalian expression vectors (Supplemental data) were lysed and sonicated 24 hrs post transfection in 1 ml cold lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1μM EDTA, 15% Glycerol, 1% Triton and EDTA-free protease inhibitor cocktail (Roche)). Total lysates were collected by centrifugation at 12000 rpm and incubated with α-FLAG conjugated M2-agarose (Sigma) 16 hrs at 4°C. Immunoprecipitated protein complexes were eluted in 100 μl elution buffer (50mM Tris-HCl pH 7.4, 150 mM NaCl, 15% glycerol, 15pg/μl Flag peptide (Sigma)). Protein content was analyzed by coomassie-blue staining or western blotting following SDS-PAGE in 3-8% Tris-acetate or 4-12% Bis-Tris NuPAGE gels (Invitrogen).
Rosetta (DE3) (Novagen) cells transformed with E. coli expression vectors (Supplemental data) were grown to OD600=0.5. Following induction with 0.1 mM IPTG for 20 h at 16°C, cells were pelleted, frozen at -20°C and resuspended in lysis buffer (500 mM NaCl, 50 mM Tris-HCl pH 8.0, 10% glycerol, 1% Triton X-100, EDTA-free protease inhibitor cocktail (Roche), 1 mg/ml lysozyme). 6His-tagged proteins were eluted from Ni-NTA agarose (Qiagen) with lysis buffer containing 160 mM imidazole. For RusA-6his, protein production was induced with 1 mM IPTG for 1 h at 35 °C.
DNA susbtrates (≈ 1 nM) were incubated in 15 μl reactions with FLAG-eluates (30, 60, 120 or 240 nl), XPF-ERCC1 (150 ng), RusA or SLX1-SLX4-7 (10 to 30 nM) or 2-5 μl Tev-Slx1 (Coulon et al., 2004), in presence of 0.5 mM Mn++ for SLX1-SLX4 and XPF-ERCC1, 2.5 mM Mg++ for TEV-Slx1 or 10 mM Mg++ for RusA. DNA substrates were prepared and reactions carried out and analyzed as previously described (Coulon et al., 2004; Gaillard, 2003). The χ structure was kindly provided by Mauro Modesti (IGC, IMM, CNRS). Resolution assays using this structure were carried out as described above except that reaction products were analyzed in 1.2% agarose gels followed by Southern blotting with a 32P-labeled χ structure-specific probe. Dried gels and hybridized nitrocellulose membranes were analyzed with a FUJI FLA5100 scanner.
We thank all members of the IGC department as well as Angelos Constantinou for encouragement and stimulating discussions. We thank Mauro Modesti, Geneviève Almouzni and Michael N. Boddy for critical reading of the manuscript, Benoît Di Martino for technical help, Denise Aragnol and Laurence Röder for MUS312 cDNA and Rick Wood for providing rXPF-ERCC1. This work was primarily funded by an ATIP-CNRS grant, Marie Curie IRG MIRG-CT-2006-046581 and Association pour la Recherche sur le Cancer grant ARC-SF1052 that were awarded to P.-H.L.G. as well as by NIH grants GM59447 and CA77325 awarded to P.R., NIH P41 RR011823 awarded to J.R.Y.III.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.