DNA mismatch repair and the DNA damage response

Abstract

This review discusses the role of DNA mismatch repair (MMR) in the DNA damage response (DDR) that triggers cell cycle arrest and, in some cases, apoptosis. Although the focus is on findings from mammalian cells, much has been learned from studies in other organisms including bacteria and yeast [1,2]. MMR promotes a DDR mediated by a key signaling kinase, ATM and Rad3-related (ATR), in response to various types of DNA damage including some encountered in widely used chemotherapy regimes. An introduction to the DDR mediated by ATR reveals its immense complexity and highlights the many biological and mechanistic questions that remain. Recent findings and future directions are highlighted.

1. Introduction

In addition to its roles in editing replication errors and other functions (see other reviews in this issue and [3]), the MMR system is also implicated in the repair and cytotoxicity of a subset of DNA lesions caused by S_N1 DNA alkylators, 6-thioguanine, fluoropyrim-idines, cisplatin, UV light and certain environmental carcinogens that form DNA adducts (reviewed in [4–7]). Defining the exact role of MMR in cell killing resulting from exposure to these DNA damaging agents is complicated by the sometimes broad spectrum of DNA damage and the convergence of multiple repair pathways such as base excision repair (BER), nucleotide excision repair (NER) and double-strand break (DSB) repair pathways and attendant DNA damage signaling pathways (see, e.g., [8–11]). The S_N1 DNA alkylators, e.g., N-methyl-N′-nitro-N-nitrosoguanidine (MNNG), methylnitrosourea (MNU) and the chemotherapy drug temozolomide, methylate all four DNA bases producing a variety of potentially cytotoxic lesions that are substrates for BER. O⁶-methylguanine-DNA methyltransferase (MGMT) directly reverses O⁶meG and plays an important role in protecting against cytotoxic effects of S_N1 alkylators and preventing tumor formation in vivo [7]. Not unexpectedly, there are numerous clinical implications, and these are discussed in this issue (minireviews by Begum, Heinen, Sijmons in this issue).

In the case of S_N1 DNA alkylators, the DDR requires components of the MMR system; the loss of functional MMR proteins, e.g., hMutSα (MSH2-MSH6) or hMutLα (MLH1-PMS2) gives rise to tolerance in which the persistence of potentially cytotoxic lesions is no longer linked to cell death. Tolerance to the S_N1 class of DNA alkylating agents was first observed in Escherichia coli strains defective in MMR that exhibited greatly increased resistance to cell killing and was subsequently demonstrated in MMR-deficient mammalian cell lines some of which are almost two orders of magnitude more resistant to cell killing than comparable MMR-proficient cells (reviewed in [12]). In a similar vein, rare cells that survive exposure to alkylating agents oftentimes have accrued mutations that inactivate MMR [13]. Despite constituting only a small fraction of total alkylated DNA lesions, O⁶me-G is the key contributor to the mutagenic and cytotoxic effects of S_N1 alkylators [14]. Low doses of MNNG induce a G₂/M cell cycle arrest in the second cell cycle after exposure that is dependent on MMR proteins (reviewed in [15,16]). A DDR signaling kinase, ATM and Rad3-related (ATR) is activated and licenses a G₂/M cell cycle arrest mediated by downstream targets including the checkpoint kinases CHK1, CHK2, and SMC1 and cell division control 25 (CDC25) phosphatases. Apoptosis ensues directed in most cases by phosphorylation of p53 that also requires functional MutSα and MutLα [17].

2. The DNA damage response

The cellular responses to DNA damage are collectively termed the DNA damage response. The DDR engages signaling pathways that regulate the recognition of DNA damage, the recruitment of DNA repair factors, the initiation and coordination of DNA repair pathways, transit through the cell cycle and apoptosis [18]. The large number of human diseases and syndromes that arise from defects in components of the DNA damage response reflect the importance of the DDR for health and viability [19].

Three protein kinases, DNA-dependent protein kinase (DNA-PK); ataxia-telangiectasia-mutated (ATM); and ATM and Rad3-related (ATR), have prominent roles in the DDR pathways that respond to genotoxic stress. These master regulator kinases are members of the phosphoinositide three-kinase-related kinase (PIKK) family, a class which also includes suppressor of morphological effect on genitalia family member (SMG1), mammalian target of rapamycin (mTOR), and transformation/transcription domain-associated protein (TRRAP) [20,21]. DNA-PK and ATM are best known for their role in the double-strand DNA breaks (DSB) response though it is increasingly apparent that they function in multiple contexts [22,23]. In contrast to ATM, ATR is essential for the survival of proliferating cells most likely due to its roles in the response to replication stress, i.e., the rescue of stalled or collapsed replication forks and the regulation of replication origin firing. In addition, it is activated by DNA damage that poses a threat to replication including certain base adducts, interstrand cross-links, and DSBs. Despite differences in substrate specificity and activation, the kinases share similar structures and regulatory themes involving localization to sites of damage and reliance on interacting protein partners [24,25]. ATR, like ATM, phosphorylates hundreds of protein targets at Ser/Thr-Gln motifs and other sites. The phosphorylated substrates in turn execute functions affecting DNA repair, replication, transcription, cell cycle checkpoint signaling, and cell fate pathways such as apoptosis or senescence.

3. Upstream events and activation of ATR

Recruitment of ATR and its constitutively interacting partner, ATR interacting protein (ATRIP), to damaged DNA was observed to be dependent on an interaction between ATRIP and replication protein A (RPA) bound to single-stranded DNA (ssDNA) [26]. Subsequent work has supported a model in which processing of DNA damage by various repair systems yields a common intermediate consisting of RPA-ssDNA, that, together with a ssDNA–dsDNA junction, serves to activate ATR [24,27]. Such structures are generated at stalled replication forks in S phase where fork reversal or uncoupling of replication factor leads to exposure of ssDNA. They can also be generated by resection at ends of DSBs during break repair. Excision repair pathways such as nucleotide excision repair (NER) acting in G₁ and G₂/M similarly can generate RPA-ssDNA structures that elicit ATR activation.

Activation of ATR requires not only localization to sites of DNA damage but also a combination of interacting partners. Recognition and recruitment to RPA-ssDNA requires ATR interacting protein (ATRIP), an obligate partner of ATR that interacts directly with RPA [26,28,29]. In addition, RPA stimulates the Rad17-replication factor C (RFC) clamp loader complex, directing it to load the Rad9-Rad1-Hus1 (9-1-1) clamp complex at the 5′ end of the ssDNA-dsDNA junction [30–32]. The 9–1–1 complex bound to DNA recruits topoisomerase binding protein 1 (TopBP1) that activates ATR through interactions with ATRIP [33–36]. The recruitment to damage sites of ATR and its key activator TopBP1 utilize distinct interactions within multiple protein complexes helping to ensure that ATR is activated only when appropriate [27].

Detailed mechanisms of each step in the activation pathway as well as the ways in which activation may differ under different damage contexts or physiological conditions remain poorly understood. A growing list of posttranslational modifications reveals their important roles in coordinating the assembly and activity of signaling complex components at sites of damage [27,24,37]. After binding of ATR–ATRIP to RPA-ssDNA, ATR undergoes trans-autophosphorylation. This phosphorylation is essential for further ATR activation as it generates a docking site for TopBP1 [38,39]. Phosphorylation of TopBP1 by ATM further enhances ATR-TopBP1 interaction and thereby ATR activity [40,41]. RPA phosphorylation by DNA-PK has also been implicated in the ATR checkpoint response [42–44]. RPA is a direct substrate of ATR in executing the DDR [45–47]. Other proteins necessary for ATR activation, including Rad17 and TopBP1, have also been identified as substrates of ATR. Other post-translational modifications of ATR and activation complex proteins are important for signaling regulation. SUMOylation of ATRIP, for instance, has been shown to enhance ATR activation by promoting interaction of ATRIP with other proteins in the pathway, including ATR, RPA, and TopBP1 [48]. The ubiquitin ligase PRP19 is recruited to RPA-ssDNA, where it enhances signaling by ubiquitinating RPA and potentially other substrates at the signaling complex [49].

Recent work also reveals new players that promote ATR activation, many of which still warrant further characterization. The Triple T complex (TTT) acts with Hsp90 to chaperone PIKKs, including ATR, promoting PIKK maturation and checkpoint signaling [50–52]. The 9–1–1 interacting nuclear orphan (RHINO) protein is an enhancer of ATR activation through interactions with the 9-1-1 clamp and TopBP1 [53,54]. Other recently identified examples include CDK2-interacting protein (CINP), which interacts with ATRIP, and the Mre11-Rad50-Nbs1 (MRN) complex that functions in DSB repair, that is required for recruitment of TopBP1 to ss-dsDNA junctions in Xenopus nuclear extracts [55,56]. Regulation may be fine-tuned to respond to specific types of damage or specific contexts such as cell cycle control [57,58].

4. Downstream signaling from ATR

Following activation, ATR acts locally at sites of damage and more globally to phosphorylate a range of substrates that execute downstream functions regulating DNA replication origin firing, stabilizing or restarting replication forks, carrying out DNA repair, activating cell cycle checkpoints and signaling apoptosis [25,27]. Although detailed understanding is not yet in hand, there is evidence of ATR's involvement in the regulation of replication origin firing to minimize the collision of forks with DNA damage, stabilization of protein machinery at the fork, modulation of the movement of forks from replication origins and protection or reinitiation of synthesis at stalled forks. ATR directly phosphorylates proteins that carry out DNA repair, e.g., DNA helicases like the Werner syndrome ATP-dependent helicase (WRN) or Bloom syndrome protein (BLM) that target stalled or damaged forks and proteins involved in inter-strand cross-link repair, e.g., the Fanconi anemia (FA) proteins and xeroderma pigmentosum complementation group A (XPA). The DDR also arrests transit through the cell cycle giving time for repair before replication or mitosis. A key example is the activation of the kinase CHK1 by ATR. This requires the mediator protein claspin [59–64]. Phosphorylation of CHK1 is further mediated by the MRN complex, the Timeless-Tipin complex, and by the protein Treslin [65–67]. pCHK1 subsequently phosphorylates CDC25 phosphatases preventing their activity on cyclin-dependent kinases (CDKs). The result is a reduction in CDK activation in S phase and a block to entering mitosis.

5. DNA methylation and the DDR

A longstanding explanation for how O⁶meG triggers an MMR-dependent DDR is grounded in the targeting of MMR excision and resynthesis exclusively to the newly synthesized DNA strand (see the minireview by Kadyrova in this issue). DNA synthesis opposite O⁶meG by replicative polymerases or possibly utilizing a translesion synthesis (TLS) pathway and error-prone polymerases causes misincorporation and produces non-Watson-Crick base pairs including O⁶meG:T and O⁶meG:C. These mispairs are recognized by MutSα (see 5]). However, the MMR-directed excision and resynthesis directed exclusively to the newly synthesized strand dictate that O⁶meG remains in the DNA leading to the recreation of O⁶meG:T mismatches and iterative rounds of MMR known as futile cycling. Eventually, a persistent single-stranded gap coated with RPA activates the ATR–ATRIP complex leading to a full-blown DDR [7]. Alternatively, the gap leads to replication fork collapse in the second round of synthesis and a DSB that activates a checkpoint response and p53-mediated apoptosis. A related pathway invokes aberrant or abortive MMR processing at sites of damage without a requirement for iterative rounds of MMR leading to the production of toxic repair intermediates including single-strand gaps [68].

There is experimental support for persistent single-strand gaps generated during processing of lesions by the MMR system. In vitro MMR assays demonstrate that the MMR system can engage in multiple rounds of excision and resynthesis supporting a futile cycling mechanism [69]. At higher doses, the presence of multiple gaps, however transient, can lead to DSBs and an ATM/ATR-mediated DNA damage response in the first cell cycle perhaps because gaps overlap on opposing strands leading to frank DSBs or because a component of MMR becomes limiting due to the high density of O⁶meG. At low doses, single-strand gaps persist into the next cell cycle leading to double-strand breaks in the subsequent round of replication [68,70]. Collapsed replication forks and DSBs can be rescued by homologous recombination yielding increased incidence of sister chromatid exchange. This has been observed in cells treated with alkylating drugs in which MGMT has been inhibited [71]. In another study, Rad51d Trp53-negative MEFs that are deficient for homologous recombination are hypersensitive to MNNG, but the additional deletion of Mlh1 resulting in loss of MMR partially rescues the sensitivity to MNNG [72]. This provides support for a role for homologous recombination in rescuing cells from otherwise lethal MMR-dependent processing of O⁶meG mismatches.

A central role of single-strand gaps in mediating the DDR to S_N1 alkylators is supported by the involvement in O⁶meG-induced cell death of the only known exonuclease that functions in MMR, EXO1 [73,74]. Recently, in vitro experiments utilizing Xenopus laevis egg extracts that couple replication with MMR implicate MMR processing in a methylation-induced DDR in which persistent single-strand gaps opposite O⁶meG and subsequent DSBs give rise to ATR activation in the second cell cycle [75]. The endonuclease activity of murine PMS2 is also required for the DDR to 6-TG, an O⁶meG mimetic, as Pms2^−/− null MEFs expressing a catalytic mutant in human PMS2, hPMS2-E705K, were highly resistant to 6TG similar to Pms2 -null cells [76]. Thus, MMR-induced excision/incision in response to O⁶meG yields gaps that are not efficiently repaired. Understanding why this is so might reveal much about the molecular mechanism of the gap-filling step of MMR.

An alternative model for the MMR-induced DDR is one involving ‘direct signaling’ in which the MutSα and MutLα MMR proteins recognize sites of damage and recruit components of the DDR pathway either directly or via interactions with other entities [77] Thus, MSH2 is reported to associate with ATR in large multi-protein complexes [78] and MSH2 with CHK2 and MLH1 with ATM [79,80]. MutSα and MutLα proteins are required to activate the ATR kinase in vitro leading to CHK1 phosphorylation in the presence of O⁶meG:T mispairs [81]. Liu et al. using purified proteins observe a direct interaction of MutSα with ATR, TopBP1 and Chk1, and MutLα with TopBP1 confirming findings from nuclear co-immunoprecipitation in cells treated with MNNG [82]. In addition, ATR, TopBP1 and Chk1, but not RPA, Claspin, Rad17-RFC or Rad9-Rad1-Hus1 are recruited to chromatin in a MutSα- and MutLα-dependent manner. Pabla et al. describe recruitment and activation of ATR at sites of damage in cisplatin-treated cells in a pathway that is not dependent on RPA, Rad17 or the 9-1-1 complex proteins consistent with a direct signaling mechanism [83].

Determining the role of MMR-mediated excision in the DDR is complicated by the fact that the excision step is not fully understood. Recent studies in Saccharomyces cerevisiae point to both EXO1-dependent and -independent excision pathways, the latter being dependent on the PMS1 (mammalian PMS2) endonuclease activity (discussed in [84]). Paradoxically, although EXO1 is the only exonuclease known to participate in the excision step of MMR and is required for in vitro MMR reactions, inactivation of EXO1 in S. cerevisiae confers only a mild mutator phenotype and, in mice, fails to fully recapitulate either the mutational or tumor spectrum observed in animals missing essential MMR proteins (see [2,84]). The notion that EXO1 plays distinct structural and catalytic roles in MMR derives from experiments in S. cerevisiae in which it is hypothesized that EXO1 interacts with MutS and/or MutL homologues to promote the stability of protein complexes [85,86]. Although an EXO1-MSH2/MLH1 interaction at physiological protein levels is not detected, ectopic expression of a nuclease-dead EXO1 [JL6] construct in MEF's in which endogenous EXO1 is absent restores a MSH2-CHK1 interaction and MNNG sensitivity [87]. A homozygous knock-in mouse harboring the Exo1-E109IK mutation is MMR-proficient but corresponding embryonic fibroblasts resemble those from Exo1 null strains with regard to loss of apoptotic signaling in response to various DNA damaging agents [74,84] The EXO1 E1091K mutant protein has subsequently been shown to be catalytically competent for excision suggesting that the loss of the DDR could be attributable to an as yet undefined structural role for Exo1 in this pathway [88,89]. Thus, a role for the scaffolding function of EXO1 in direct signaling is not ruled out.

Mice harboring one of two missense mutations in murine MutSα, Msh2^G674A or Msh6^T1217D, retain the apoptotic response to MNNG but have lost MMR activity though the mutant MutSα proteins still bind to G:T mismatches in vitro [90,91]. A mutant human MutSα protein harboring the equivalent MSH6T1219D substitution fails to interact with MutLα and does not support any detectable excision in an in vitro MMR assay suggesting that a DDR can occur in cells in the absence of excision [92]. The uncoupling of MMR and apoptotic signaling is consistent with a direct signaling model though a subsequent study suggests an alternative explanation. In double null mice missing MGMT and EXO1 there is greatly enhanced survival in response to MNU methylation in some tissues akin to MGMT/MSH6-null mice, but only partial rescue from cell killing in other tissues [73]. While this residual killing might reflect a scaffolding role for EXO1 in the DDR, the authors suggest that the residual cytotoxicity is due to PMS2 endonuclease activity that is modulated by EXO1 or MutSα in some way. Conclusive evidence in support of direct signaling remains to be established and would be aided by a robust in vitro reconstitution of the ATR-directed DDR in response to specific types of DNA damage and relevant DNA repair proteins [82,93].

6. Fluorouracil

Exposure of mammalian cells to fluoropyrimidines such as 5-fluorouracil (FU) or 5-fluoro-2′-deoxyuridine (FdU) also evinces a DDR that is dependent on the MutSα and MutLα MMR proteins [94–96]. FU is widely used in the treatment of a variety of solid tumors including colorectal tumors, and resistance is a challenge in the clinical setting (reviewed in [97]). Determining the mechanism of cell killing by FU is complicated as its metabolite, 5-fluoro-2′-deoxyuridine monophosphate, inhibits thymidylate synthase, a key enzyme in de novo pyrimidine biosynthesis, resulting in imbalances in nucleotide precursor pools and the incorporation of uracil and fluorouracil in DNA and RNA. Cytoxicities of MMR-proficient and -deficient cells after treatment with Tomudex™, a specific inhibitor of thymidylate synthase, are very similar suggesting that incorporation of FdU into DNA is the major pathway for MMR-dependent differential cell killing observed in these cells [95]. Demonstration of a direct correlation between cell survival and excision from DNA of FdU but not uracil by SMUG1 glycosylase also points to the incorporation of FdU into DNA as being the primary pathway for cell killing [98]. Four uracil glycosylases, UNG2, SMUG1, TDG and MBD4 target FU for removal though with varying efficiencies and in different contexts (see discussion in [99–101]). These glycosylases excise a damaged base generating an abasic site that is converted to a single nucleotide gap by AP endonuclease. A 5′-deoxyribose phosphate at the gap is removed by DNA Pol β that, together with a DNA ligase, restores a repaired DNA [10,7].

During replication, FdU is incorporated by DNA polymerases opposite adenine or guanine where it can be removed by MMR following detection of altered base stacking and hydrogen bonding at mismatches and excision on the newly synthesized strand. Whereas UNG2 will remove A in FU:A mispairs, MutSα targets rarer FU:G mispairs resulting in the activation of MutSα ATPase activity [95]. FU stimulates an MMR-dependent cell cycle checkpoint in G₂ and subsequent cell death (reviewed in [96]). This DDR in MMR-proficient cells occurs within the first cell cycle and upon activation of the ATR kinase leads to phosphorylation of CHK1 and SMC1 [102]. Prolonged exposure leads to an irreversible arrest in G₂ that is independent of MMR and coincides with extensive involvement of BER pathways. MMR-mediated DDR in response to FU may involve similar signaling pathways to methylation damage discussed above, but the DDR differs in some notable ways–whereas O⁶meG more likely resides in the template strand where it is not accessed by MMR, FdU by virtue of being directly incorporated into DNA is a substrate for MMR removal; cell cycle arrest occurs in the first cell cycle in FU-treated cells whereas it is occurs in the second cell cycle in the case of alkylation damage; FU does not appear to involve p53-mediated apoptosis as alkylation does [103,104]. No single pathway may predominate as these studies identify an ATR/ATM-independent signaling pathway in the case of both alkylation and FU that utilizes an hMLH1/c-Abl/GADD45α signaling pathway. In addition, BER processing of FU imposes its own cytoxicity as evidenced by the accumulation of toxic processing intermediates by thymine DNA glycosylase [101].

MLH1 interacts with a DNA glycosylase, methyl-binding domain protein 4 (MBD4), a BER enzyme that functions prominently in the repair of deaminated C and 5-methylcytosine within CpG sites (reviewed in [11]). MBD4 also removes T opposite O⁶meG and halogenated pyrimidines including FU. MEFs deficient for MBD4 have a diminished apoptotic response to several DNA damaging agents including MNNG, and FU, and epistasis analysis of MBD4^−/− MLH1^−/− mice exposed to FU suggests that MBD4 and MLH1 function in the same pathway. Additional studies are required to understand the relationship between MMR and BER pathways in the apoptotic response.

Recent studies in Caenorhabditis elegans and human cells suggest that a hybrid MMR-BER pathway exists for the activation of autophagy in response to FU in which MSH6 acts as a sensor of DNA damage, and in which autophagy is a significant contributor to cell death. Epistasis analyses in C. elegans are consistent with the BER AP endonucleases APN-1 and EXO-3 acting downstream of MSH6 with EXO-3 postulated to create nicks in the vicinity of FU residues that are used by the MMR pathway to initiate excision repair while APN-1 modulates the MMR-mediated resection step [105]. Global chromatin decompaction rather than DNA breaks is observed, and the authors speculate that a relaxed chromatin conformation induced by MSH-6 processing of DNA containing FU alters rDNA organization in nucleoli leading to autophagy. These findings raise interesting questions with respect to the interaction between BER and MMR pathways. What are the substrates in vivo that trigger MMR activation? What aspects of DNA damage determine whether BER and MMR proteins compete or cooperate at sites of DNA damage and how might this affect the DDR? What is the spectrum of DNA lesions that serve to activate MMR proteins as DNA damage sensors? When in the cell cycle does this pathway operate, and is it an example of ‘noncanonical’ MMR broadly defined as MMR occurring outside S phase (see, for example, [106])? Finally, within the context of clinical relevance, it is clear that MMR status alone is probably insufficient to assess treatment outcomes using fluoropyrimidines and similar drugs given the complicated interplay between MMR and BER.

7. Oxidative damage and DNA adducts

The primary repair pathway for oxidized DNA damage is BER utilizing specific DNA glycosylases that excise the damaged bases followed by cleavage at the abasic site by AP-endonuclease 1 and gap repair involving polβ (see, for example [107]). However, studies in E. coli, S. cerevisiae, murine embryonic fibroblasts and stem cells and human cancer cell lines support a role for MMR in the repair of oxidative DNA damage, specifically 7,8-dihydro-8-oxo-guanine (8-oxoG) mispairs [15,108]. MMR can excise 8-oxoG incorporated into DNA from dNTP pools and targets replication errors leading to 8-oxo-G:A mispairs (see minireview by Crouse in this issue). In vitro studies support recognition of mismatched 8-oxoG by hMutSα [109]. Studies in S. cerevisiae and human cells support redundant but distinct roles for MSH2 and a BER glycosylase, MutY homolog (MYH), that removes A opposite 8-oxoG [110,111]). In addition, a physical and functional interaction between hMutSα and MYH has been reported [112].

Ziatounou et al. describe a noncanonical MMR pathway requiring MSH2-MSH6, but not MLH1, in the direct repair of oxidative damage during replication [113]. In this BER-independent pathway, recognition of clustered oxidative lesions by MutSα and ensuing excision results in the MutSα-dependent monoubiquitination of PCNA and the recruitment of Poltη instead of or in addition to high fidelity replicative Polδ/ε for gap filling. This pathway is consistent with findings that human cells deficient in MSH2, but not MLH1 are sensitized to oxidizing agents [114], and the co-association of MutSα and Polη both in vitro and in mammalian cells [115,116].

MMR proteins have a role in the imposition of cell cycle checkpoints and apoptotic signaling in response to UV. Among the DNA lesions caused by UV exposure are nontemplating DNA lesions such as cyclobutane pyrimidine dimers (CPD) and genotoxic (6-4) pyrimidine pyrimidone dimers (6-4)PP. Although NER is the primary repair pathway for UV damage, Msh2-deficient mice exhibit an increased incidence of UVB-induced skin tumors, an effect that is additive with the loss of XPA, a key NER protein [117,118]. Both in vitro and in vivo studies of UVB-irradiated NER-deficient mice also implicate MMR in UV-induced DNA damage signaling and cell cycle arrest [119,120]. Similarly, hMSH2 is implicated in cell cycle arrest and apoptosis in melanocytes [121]. MMR-deficient mammalian cell lines or murine embryonic stem cells exhibit increased mutagenesis upon UV exposure compared to MMR-proficient cells [122,123]. MutSα can bind to mismatched but not matched CPD and (6-4)PP in vitro [124,125].

A recent study of ES cells derived from mice harboring defects in NER and/or MMR and exposed to short wavelength UVC irradiation suggests a novel “post-TLS” repair pathway in proliferating cells for MMR-dependent, UV-induced mutagenesis and DNA damage signaling. In this pathway, MutSα targets mismatches formed by TLS opposite (6-4)PP lesions thereby acting downstream of NER and TLS [126]. The presence of long-lived excision intermediates consisting of gapped DNA coated with RPA serves as a signal for ATR kinase and will give rise to broken chromosomes upon replication in the next cell cycle. Additional study will reveal whether this intersection of TLS and MMR occurs in other contexts. Deducing a molecular underpinning for organ tropism of tumors is oftentimes challenging. In the case of Lynch syndrome colorectal cancer, the rate of cellular proliferation is surely an important contributor, but it is unlikely to be the only factor. Tsaalbi-Shtylik et al. suggest that the loss of post-TLS repair and attendant elevated mutagenesis leads to disruption of multiple tumor suppressing functions, and in combination with constant exposure to intestinal genotoxins, may explain the colorectal tropism of Lynch syndrome [126].

MMR proteins have also been implicated in the repair and cytotoxicity of agents such as psoralen and cisplatin that form bulky adducts on DNA (reviewed in [127]). The most common cisplatin adducts are intrastrand cross-links involving adjacent purines, and these are handled by the nucleotide excision repair pathway. Cisplatin-induced interstrand cross-links (ICLs) are much more toxic as they cause a complete block to replication and transcription. A clear understanding of how MMR participates in ICL repair or modulates cytotoxicity is not currently available due to the complexity of ICL repair and other types of damage caused by these agents including intrastrand cross-links and DSBs. hMutSβ binds to cisplatin and psoralen ICLs in vitro [128,129]. However, the effects of MMR on cell survival tend to be fairly small, several-fold, or even undetectable [130], and MMR is only one of several pathways that affect drug cytotoxicity [131]. Nonetheless, MMR has been shown to mediate cisplatin cytotoxicity via a caspase signaling pathway in human cancer cell lines [132]. ICL repair involves large numbers of repair factors, many of which have additional roles in other repair pathways [133]; the Fanconi anemia (FA) pathway proteins that are mutated in the FA cancer predisposition syndrome figure prominently though FA-independent pathways exist [134]. In one model, a subset of FA proteins is recruited to the stalled fork leading to the activation of ATR, phosphorylation of CHK1 and activation of other repair factors. Additional core FA proteins are recruited, and unhooking of the ICL is achieved by structure-specific nucleases that sequentially nick on either side of the ICL [135,136]. TLS polymerases bypass the lesion and extend synthesis providing a substrate for homologous recombination to repair a DSB. NER removes the unhooked ICL giving rise to two replicated DNA duplexes. To complicate matters further, the repair of ICLs can also occur in a DNA replication-independent pathway in the G0/1 phase of the cell cycle [137].

There is an extensive literature on the effects of cisplatin and related drugs on the relative survival of MMR-proficient and -deficient cells and on the interaction between MMR proteins and members of various other pathways implicated in ICL repair including FA, homologous recombination, DSB repair and NER (reviewed in [138] and in the minireviews by Heinen, this issue and Begum and Martin, this issue). Given our knowledge of MutSα, MutSβ and MutLα in the editing function of MMR and the current models for ICL repair, it seems unlikely that MMR proteins participate directly in the removal of ICLs associated with replication [128,129]. Given the participation of homologous recombination and error-prone TLS in ICL repair, it is possible that MMR proteins might modulate these steps [139]. In addition, it is proposed that crosstalk between FA and MMR pathways is important to regulate repair, and that a FANCJ–MLH1 interaction suppresses MSH2 activity to promote restart at stalled replication forks [140,141]. A p53-independent apoptotic pathway involving the p53-related transcription factor p73 has been described for the MLH1-mediated damage response to cisplatin [142]. In cells treated with cisplatin, levels of p73 rise in a manner requiring both MLH1 and PMS2; a physical interaction between PMS2 and p73 is proposed to stabilize p73. Pabla et al. report that hMSH2 facilitates the recruitment of ATR to foci in cisplatin-treated cells independently of RPA, Rad17 or Rad9-Hus1-Rad1 activating CHK2, p53 and PUMA-α [83]. The role of MMR in ICL-mediated DDR and chemoresistance in the patient population remains an area of intense investigation.

MMR has also been implicated in the cytotoxicity of other bulky adducts that act as carcinogens. Benzo[c]phenanthrene dihydrodiol epoxide (B[c]PhDE), an environmental chemical carcinogen, modifies adenine residues in DNA and induces apoptosis in a MMR-dependent manner both in cancer cell lines and in the intestinal crypt of mice [143]. Specific binding of MutSα and MutSβ to B[c]PhDE-DNA adducts in vitro is also observed. In mammalian cells, low or moderate doses of the carcinogen hexavalent chromium Cr(VI) induces rapid DSB formation and γH2AX foci, a subset of which are dependent on MutSα, MutSβ and MutLα MMR proteins [144,145]. Although the exact molecular mechanism for MMR-induced cytotoxicity is not known, the rapid response to Cr(VI) is qualitatively different from that caused by S_N1 alkylators discussed above. Msh2-deficient mice exposed to a polycyclic aromatic hydrocarbon, benzo[a]pyrene (B[a]P) exhibit increased susceptibility to lymphomagenesis though levels of B[a]P adducts are not affected by Msh2 status [146]. Two studies describe a several-fold increase in the induction of mutation by the heterocyclic amine 2-amino-1-methyl-6-phenolimidazo[4,5-b] pyridine (PhIP) in the colon of Mlh^−/− or Msh2^−/− mice compared to MMR-proficient mice [147,148]. NER is the predominant repair pathway, and Mlh1^−/− mice are not more sensitive to killing by PhIP in contrast to NER-deficient mice suggesting that MMR operates downstream of NER possibly editing mutagenic bypass of PhIP adducts. In summary, MMR proteins are unlikely to be involved in the direct removal of these DNA adducts, but function instead in concert with other repair pathways such as NER, BER, FANC or the DSB repair pathway to elicit a DNA damage response.

8. Context and environment

Both the immediate and more global cellular environment can influence the DDR. Recent work reveals important links between MMR function/regulation and chromatin dynamics and the proteins that regulate chromatin structure (see the minireview by Li, this issue) [149–151]. Lin and colleagues have examined the MMR-dependent DDR in human pluripotent stem cells (PSCs) exposed to MNNG and find a surprising result. In contrast to somatic cells exposed to MNNG alkylation damage that undergo a G₂/M cell cycle arrest in the second cell cycle, PSCs undergo apoptosis in the first cell cycle without activating CHK1 or CHK2 and in the absence of a prior G₂/M arrest [152]. The authors speculate that the immediate imposition of apoptosis upon DNA damage may reflect a lower tolerance for DNA damage in PSCs that eventually give rise to all tissues in the developing embryo.

9. Conclusions and future directions

The MMR system plays important roles in a number of cellular processes including the DDR. The challenges as have been noted above center on elucidating molecular mechanisms, exploring the interplay with other repair pathways and defining the effects of cellular and spatiotemporal context, all important for understanding fundamental processes relevant to cancer and therapeutic approaches. In vitro reconstitutions of a DDR that is activated by defined DNA damage and includes relevant repair proteins is progressing (see [153]). ‘Omics’ technology approaches are being implemented to define RNA-mediated transcriptional and regulatory features of the DDR, identify the DNA damage interaction network in cells, catalog relevant post-translational modifications that regulate the process and define the three-dimensional space in which the DDR takes place [154–156]. High-throughput assays measuring DNA repair capacity surveying multiple DNA repair pathways simultaneously are being developed [157]. Although much has been deduced from the use of established tumor cell lines deficient for MMR, tunable promoters and gene targeting approaches such as the CRISPR/Cas9 system are gaining wide use as are rapidly advancing imaging techniques [158,159]. Finally, in vivo studies provide a critical testing ground [160,161]. Collectively, these approaches are poised to reveal novel insights into the functions of the MMR system in the DNA damage response.