Mismatch Repair (MMR) Pathways and Deficiencies

Published Date: 29 Mar 2018

1- Mismatch Repair

The DNA mismatch repair (MMR) pathway is a bidirectional excision-resynthesis system that corrects mismatches generated during DNA replication or homologous recompination (HR). MMR adds up to 1000 fold increase in the fidelity of DNA replication (Kolodner, 1996; Preston et al., 2010). Mismatches fall into two groups, base-base mispairs resulting from incorrect nucleotide insertion by DNA polymerses and insertion-deletion loops (IDLs) resulting from slippage of DNA polymerase at simple sequence repeats (microsatelite) (Kunkel and Erie, 2005; Li, 2008).

1.1- Prokaryotic

MMR is well conserved from bacteria to mammals. The primary bacteria proteins involved in MMR were designated MutS, MutL, and MutH. MutS recognizes mismatch as homodimers. MutL (also called molecular mismatch maker) facilitates the interaction between DNA-MutS-MutL and MutH (Matson and Robertson, 2006; Robertson et al., 2006). MutH cleaves GATC sequences selectively in the nascent strand, which remains transiently unmethylated because deoxyadenine methylase lags behind the replication fork (Lee et al. 2005). The nicked strand is cleaved by one of four single-strand exonucleases (the 5' → 3' exonucleases ExoI and ExoX, or the 3' → 5' exonucleases RecJ and Exo VII) (Burdett et al., 2001). The single-strand gap is bound by single strand DNA binding protein (SSB). DNA polymerase III then completes the gap, and DNA ligase seals the nick (Hsieh and Yamane, 2008).

1.2 Eukaryotic

MMR in eukaryotes retains many of the key features of the E . coli MMR pathway. The MutS equivalents in humans (MSH) exist in two heterodimeric forms: MutSα (MSH2 and MSH6) that identifies base-base mismatches and small loops, and MutSβ (MSH2 and MSH3) that identifies larger loops, with some overlap in substrate specificity among the two MutS complexes (Li et al., 2008). Recently, MutSα has been shown to have strong bias for insertion loops repair, while MutSβ has an even stronger bias for deletion loops repair (Romanova and Crouse, 2013). Human cells express more MSH6 than MSH3, leading to a MutSα:MutSβ ratio of 10:1 (Drummond et al., 1997; Marra et al., 1998). Despite their redundant activities, both complexes are required for MMR and defective or abnormal expression of MSH6 or MSH3 lead to spontonous mutation (mutator phenotype) (Drummond et al., 1995; Marsischky et al., 1996; Drummond et al., 1997; Harrington and Kolodner, 2007).

Heterodimeric MutS complex (MSH2/6 or MSH2/3) recognizes the substrate, likely by recognizing increased flexibility at the site of the mismatch (Isaacs and Spielmann, 2004). MutS complex then recruits MutL. MutL equivalents in humans exist in three heterodimeric forms: MutLα (MLH1–PMS2), MutLβ (MLH1–MLH3), and MutLγ (MLH1–PMS1). MutLα is the major MutL homolog that participates in MMR and it has endonuclease activity (Cannavo et al., 2005). MutLα binds several MMR proteins and modulates their activity in a mismatch-dependent manner (Kunkel and Erie, 2005; Li, 2008). EXO1, the only exonuclease implicated in MMR to date, has an obligate 5′→3′ polarity, which seems inconsistent with the bidirectional MMR, but close analysis revealed that MutLα harbors cryptic endonuclease activity. PMS2 introduces a nick in the daughter strand, 5' or 3' of the mismatch, and this nick serves as entry point for the EXO1 that carries out the excision step (Kadyrov et al., 2006; 2007). In support, MLH1–PMS2 is required for 3'-excision but not 5'-excision (Constantin et al., 2005; Zhang et al., 2005; Pluciennik et al., 2010). RPA (replication protein A) protects the MMR excision intermediate from nuclease degradation and the excised DNA strand is resynthesized by Polδ (Kunkel and Erie, 2005; Li, 2008). The basic human MMR system includes MutSα or MutSβ, MutLα, EXOI, PCNA, RFC (loads PCNA onto DNA), RPA, polymerase δ, and DNA ligase I (Constantin et al., 2005; Zhang et al., 2005). The 5' to 3' mismatch-directed strand excision requires only MutSα, EXOI, and RPA, whereas substrates with a 3' nick require also MutLα, PCNA, and RFC (Dzantiev et al., 2004).

MMR in eukaryotic cells has to deal with the nucleosome to reach the mismatch. Previous reports have demonstrated that DNA mismatches within tightly associated nucleosomes, in contrast to naked DNA, are poor MMR substrates (Li et al., 2009; Schöpf et al., 2012), so there must be a signal that allow a timely recruiting of MMR to the nucleosome. Li et al (2013) have made a breakthrough by demonstrating that, an epigenetic histone mark, H3K36me3, during G1 and early S-phase recruits the MutSα onto the chromatin before replication independent of the presence of mismatch. Cells that lack STD2 (H3K36 trimethyltransferase) display microsatellite instability (MSI) and spontonus mutations frequencies characterized of MMR deficient cell.

1.3- Strand Discrimination

Eukaryotic cells do not use methylation for strand discrimination, alternatively a nick in DNA can signal for strand-specific eukaryotic MMR in vitro. The first biochemical studies, carried out with extracts of human or Drosophila melanogaster cells showed that covalently closed circular DNA substrates with a single mismatch were refractory to MMR, but a nick in either strand situated up to 1 kb away from the mismatch was necessary and sufficient to activate the MMR process (Holmes et al., 1990; Thomas et al., 1991). Discontinuous lagging strand synthesis of Okazaki fragments (aproximately 200 bp long in eukaryotes) introduces high number of 5' DNA ends that discriminate the nascent lagging strand (Burgers, 2009). On the other hand, the leading strand is replicated in a continuous manner. This raises question about how the MMR directs the nascent leading strand.

The answer to this question appears to lie in an interaction between MutLα and PCNA. RFC loads PCNA at 3' primer termini (boundaries between double- and single-stranded DNA) with the same side facing the DNA terminus (McNally et al., 2010). Mismatch made by pol-δ is detected by MutSα or MutSβ, which slides along the DNA, interacts with PCNA and displaces the polymerase. Loading of MutLα generate a protein complex that travel toward the mismatch where MutLα can introduce nicks in the leading strand, which are used as loading sites for EXO1. Since only one strand of DNA has the correct orientation (5' → 3' or 3' → 5') for hydrolysis, the enzyme will cleave only a single strand (Peña-Diaz and Jiricny, 2010; Peña-Diaz and Jiricny, 2012). According to this model, on the leading strand, the MutLα/PCNA complex needs to travel from the 3' terminus to the mismatch whcih could be hundreds of nucleotides distant (Schöpf et al., 2012), so it was suggested that MMR is less efficient on the leading strand compare to the lagging strand, where strand discontinuities are available (Nick McElhinny et al., 2010a).

Recently, an additional mechanism was proposed for the nascent leading strand discrimination. During replication, more than one million of ribonucleotides are introduced into mouse genome (Hiller et al., 2012; Reijns et al., 2012), and a similar situation occurs in Saccharomyces cerevisiae, where Polε (the leading strand polymerase) incorporates into the nascent DNA about four times more ribonucleotides than polδ (the lagging strand polymerase) (Nick McElhinny et al., 2010b; 2010c; Lujan et al., 2012). Recent reports have shown that RNase H2-dependent processing of the ribonucleotides incorporated by Polε, acts as a signal that can direct MMR to the nascent leading strand. Inhibition of RNase H2 has no effect on the lagging strand, because of the high number of nicks introduced by Okazaki fragments. This mechanism has small contripution to MMR fidelity, because it requires that the mispair and the ribonucleotide are within less than 1 kb of each other (Ghodgaonkar et al., 2013; Lujan et al., 2013).

2- Mismatch Repair Deficiency, Microsatelite Instability, and Lynch Syndrom

Microsatellites are short repetitive DNA sequences 1-6 (Ellegren et al., 2004). Because of their repetitive sequence structure, microsatellites exhibit a particularly high mutation rate. During replication, DNA polymerases often fail to correctly duplicate the microsatellite repeats, due to slippage, which results in insertion/deletion loops (Peña-Diaz and Jiricny; 2012). This phenomenon is known as MSI, and is recognized as length changes in the microsatellites. The MSI status is determined using a panel of five microsatellites (BAT25, BAT26, D2S123, D5S346, and D17S250) (Medina-Arana et al., 2012).

In 1993, MSI was detected in about 10–15% of sporadic colorectal carcinomas as well as in >90% of Lynch syndrome (LS) patients, also referred to as Hereditary Non-Polyposis Colon Cancer (HNPCC) (Peña-Diaz and Jiricny, 2012). The finding that MMR deficiency in Saccharomyces cerevisiae induced MSI led to the suggestion that cancers with MSI might also have defects in MMR (Jiricny, 1994). LS is a prevalent autosomal dominant hereditary cancer syndrome caused by heterozygous mutations in one of MMR genes MSH2, MSH6, MLH1, or PMS2 (Rasmussen et al., 2012). The mutations in MMR genes that lead to truncation or deletion can be securely classified as pathogenic, but in significant fraction of individuals suspected to develop LS, subtle alterations in MMR genes are identified such as misense mutations or mutations in splice sites. These types of mutation are called variation of uncertain significant (VUS) (Rasmussen et al., 2012). The pathogenesis of many of these VUS is not clear due to the absence of data on the consequences of these mutations on gene function. Many functional analysis assays have been developed in vitro and in vivo to identify the pathogencity of VUS (Rasmussen et al., 2012; Drost et al., 2013). Recently, using yeast as a model system have demonstrated that more than half of the deleterious missense mutation in MSH2 results in lower level of the protein due to ubiquitin-mediated proteasomal degradation (the primary ubiquitin ligase is san1). Increase the expression of the unstable variants, deletion of san1 or the use proteasomal inhibitor restores MMR function (Arlow et al., 2013).

MSI is known to occur due to defects in MMR genes such as germline mutation in MSH2 or MLH1 in most LS cases and epigenetic silencing of MLH1 in most sporadic cases (Leach et al., 1993; Bronner et al., 1994; Herman et al., 1998; Veigl et al., 1998). Nevertheless, many colorectal and several other MSI-positive cancers do not have genetic or epigenetic defects in MMR genes. Recently, Li et al. (2013) have shown that depletion of SETD2 impaires MutSα chromatin binding, leads to MSI, and increases the mutation rates. Intriguingly, they have found a renal cell carcinoma and a Burkitt’s lymphoma cell line, both without defects in MMR genes but MSI positive, to be mutated in SETD2. This report provides explanation for the discrepancy between the genotypes and phenotypes of such cancers.

In addition, recent studies support the idea that defects in MMR pathway could be independent from defects in MMR-genes. POLE and POLD1 are related B family polymerases, and they represent the main catalytic and proofreading subunits of the Polε and Polδ enzyme complexes (Pursell et al., 2007; McElhinny et al., 2008). POLE and POLD1 contain a 3′–5′ exonuclease (proofreading) domain which recognize and excise the mispair and inturn increases replication fidelity by aproximately 100-fold. Recent reports have shown that, POLE and POLD1 exonuclease domain mutations (EDMs) increase the susptability to colorectal cancer (CRC) and, in the latter case, to endometrial cancer (EC). In addition, somatic POLE EDMs have been reported in sporadic CRCs and ECs (Church et al., 2013; Palles et al., 2013).

Microsatellites have been identified within the coding sequences of a number of genes (Duval and Hamelin, 2002a; Duval and Hamelin et al., 2002b). The DNA polymerase slippage within these coding sequences can induce frameshifting mutations. In case of CRC genomes, cancer-associated genes frequently affected by MSI (e.g., TGFBR2, ACVR2A, and BAX) have been investigated (Markowitz et al., 1995; Rampino et al., 1997; Jung et al., 2004). Recently, Kim et al. (2013a) have provided comprehensive analysis of the prevalence and functional consequence of MSI in CRC and EC. Using exome and whole genome sequencing, they show that recurrent MSI events in coding sequences have 1) elevated frameshift-to-inframe ratios, so they hypothesized that the genes inactivated by recurrent MSI may have tumor suppressor roles, and the high fame shift (nonnutral) could provide selection advantage on coding sequence 2) lower transcript levels than wild-type alleles, which may be due to RNA surveillance pathway that eliminates mRNA containing a premature stop codon and 3) tumor type specificity.

3- Mismatch Repair, Monofunctional Alkylating Agent, and Therapy-Related Myeloid Neoplasms

Conventional chemotherapeutic agents used in clinics operate by inducing DNA damage in cancer cells. Unfortunately, normal cells are also targeted by these chemotherapeutic agents which induce mutations and in turn the development of secondary cancers in normal cells. The most prevalence forms are therapy-related myeloid neoplasms (t-MN) which account for about 10-20% of myeloid neoplasms and can be subdivided into therapy-related myelodysplastic syndrome (t-MDS), therapy-related acute myeloid leukemia (t-AML), and therapy-related myelodysplastic/myeloproliferative neoplasms(t-MDS/MPN) (Vardiman et al., 2008). Based on the type of chemotherapeutic agents, two main subtypes of t-MN with different characteristic have been identified. The first subtype of t-MN is related to exposure to alkylating agents, and it is characterized by long latency period of 3-10 years, a proceeding myelodysplasia and loss of all or parts of chromosomes 5 or 7 or both. The second subtype of t-MN is related to exposure to Topoisomerases2-poisons (TOP2-poison) and it is characterized by a short latency period of 1-3 years, often lacks a preceding myelodysplasia, and balanced chromosomal rearrangements involving MLL at 11q23 and t(15,17)(PML-RARA) (Pedersen-Bjergaard and Rowley, 1994; Godley and Larson, 2008; Pascual et al., 2009).

Alkylating agents are divided into monofunctional (e.g., temozolomide, dacarbazine, and Methylnitronitrosoguanidine (MNNG)) or bifunctional alkylating agents such as nitrogen mustards (chlorambucil and cyclophosphamide), and chloroethylating agents (e.g., nimustine (ACNU), lomustine (CCNU), and carmustine (BCNU)) (Kondo et al., 2010). BER can repair the majority of the alkylated DNA adducts induced by monofunctinal alkylators except for O6meG, which is largely responsible for the cytotoxicity of this class of chemotherapeutic agents. Methylguanine methyltransferase (MGMT) can directly repair the O6meG by covalent attachment of the methyl group from the O6meG to a cysteine residue on MGMT, leading to irreversible inactivation of MGMT (Margison et al., 2003; Gerson, 2004; Mueser and Williams, 2011). During replication, DNA polymerases frequently mispair O6meG with thymine, which in turn, activate the MMR (Spratt and Levy, 1997). Interestingly, MMR rather than repairing O6MeG, it induces DNA damage signaling, cell cycle arrest and apoptosis (Hickman et al., 2004; Jiricny, 2006; Mojas et al., 2007). This means that, the cytotoxicity of monofunctional alkylating agents requires a proficient MMR. Indeed, cells proficient for MMR and deficient for MGMT show high sensitivity to monofunctional alkylating agent, while cells deficient for MMR and MGMT are resistant to cell death and have increased mutation rates (Branch et al., 1993; de Wind et al., 1995; Sharma et al., 2009). The mutator phenotype that characterizes MMR deficient cells may accelerate t-MN development. In support t-AML occurring after exposure to alkylating agents displays MSI (Casorelli et al., 2012).

The mechanism by which MMR mediates the cytotoxicity of monofunctional alkylating agents is not fully understood yet. Two models have been proposed, the "futile cycle" and "direct signaling" models. The “futile cycle” model suggests that, since the MMR machinery can only target the newly synthesized DNA strand containing the mismatched thymine, the O6MeG will never be removed and another thymine opposite to O6meG will be inserted in the following replication. The repeated excision and regeneration of O6meG:T mispairs will induce cytotoxic DNA DSBs. In this model the ATR is indirectly activated after DNA Damage (York and Modrich, 2006; Mojas et al., 2007).The “direct signaling” model suggests that, MMR proteins binding to O6meG:T mispairs act as scaffold for direct recruitment and activation of ATR DNA damage signaling pathway. This model has been supported by separation of function mutations in mice containing mutations in Msh2 and Msh6 ATPase domains which are essential for MMR activation but not for MMR-dependent DNA damage-induced apoptosis. These mice showed that MMR activity can be inhibited without affecting MMR-induced DNA damage response (Lin et al., 2004; Yang et al., 2004; Yoshioka et al., 2006). Interestingly, many studies have shown that RPA is not essential for MMR-dependent ATR activation (Liu et al., 2010; Pabla et al., 2011). In a remarkable contrast to the "direct signaling" model, a recent study have showed that O6MeG induce ATM and ATR activation, and inhibition of ATM and ATR sensitizes the cell to monofunctional alkylating agent with more rule to ATR (Eich et al., 2013).