DNA Mismatch Repair: Dr. Jekyll and Mr. Hyde?

Abstract

In this issue, Peña-Diaz et al. (2012) describe a pathway for somatic mutation in nonlymphoid cells termed noncanonical DNA mismatch repair, whereby the error-prone translesion polymerase Pol-η substitutes for high-fidelity replicative polymerases to resynthesize excised regions opposite DNA damage.

Genome instability can occur via gross chromosome rearrangement, loss or duplication, or via more discrete alterations, such as point mutations, frame-shifts, or DNA repeat instabilities. Replication-linked DNA mismatch repair (MMR) targets the more discrete alterations by correcting replication errors; thus, MMR contributes 100- to 1000-fold to overall replication fidelity. Loss of MMR confers a strong mutator phenotype and is often characterized by microsatellite instability at short repeats. Mutations in human MMR genes segregate with Lynch syndrome resulting in a predisposition to colorectal and other cancers. Although MMR is normally a high-fidelity process that preserves genomic integrity, it can also be a source of somatic hypermutation during antibody diversification in activated B cells. In this issue Peña-Diaz et al. (2012) present data demonstrating that the MMR pathway can be hijacked to promote genome instability in other nonlymphoid cells, providing insight into DNA-damage-induced mutagenesis and genome rearrangements.

Canonical MMR begins with the recognition of base-base mispairs or insertion/deletion loops containing one or a few unpaired bases by the MutSα heterodimer (MSH2-MSH6) and, less frequently, the MutSβ heterodimer (MSH2-MSH3) (Figure 1A) (Hsieh and Yamane, 2008); (Kunkel and Erie, 2005; Li, 2008). Subsequent recruitment of the MutLα heterodimer (hMLH1-hPMS2) in a process modulated by ATP and the replication processivity factor proliferating cell nuclear antigen (PCNA) (which normally tethers the replication machinery to DNA) results in activation of an endonuclease function in the PMS2 subunit of MutLα and EXO1, a 5′ to 3′ exonuclease, to carry out strand excision (Pluciennik et al., 2010). Importantly, these nucleases direct excision exclusively to the newly synthesized strand containing the error. In the final step, resynthesis of the RPA-coated single-strand gap to restore an intact duplex is catalyzed by high-fidelity replicative polymerases, Polδ or Polε, and ligase.

Figure 1. Mutagenic and Nonmutagenic Outcomes of DNA Mismatch Repair.

(A) General scheme for replication-linked MMR in which recognition of a mismatch by a MutSα-MutLα complex modulated by interaction with PCNA and ATP/ADP binding by MMR proteins licenses excision of the newly synthesized strand by EXO1 and an endonuclease function in the PMS2 subunit of MutLα. The resulting RPA-coated single-strand gap is a substrate for resynthesis by high-fidelity Pol-δ.

(B) In activated B cells, AID creates G/U mispairs that are substrates for MutSα-MutLα. An ncMMR pathway is utilized in which mUbPCNA recruits Pol-η to carry out error-prone gap filling to achieve somatic hypermutation at variable regions of Ig genes.

(C) A scenario for a ncMMR pathway operating in nonlymphoid cells that responds to a variety of DNA lesions that are substrates for MutSα including uracil, alkylated bases, and distorted DNA structures that impede replication. Recruitment of Pol-η by monoubiquitinated PCNA leads to error-prone gap filling.

In stark contrast to replication-linked MMR, a highly mutagenic noncanonical MMR (ncMMR) pathway is required for somatic hypermutation and chromosome rearrangements during antibody diversification in activated B cells (see review by Chahwan et al., 2012). Two distinct but related pathways, somatic hypermutation (SHM) of immunoglobulin (Ig) variable regions and class-switch recombination (CSR) at Ig switch regions are dependent on both MutSα and MutLα. SHM begins at Ig variable regions with deamination of deoxycytosines to deoxyuracil by activation-induced deaminase (AID) (Figure 1B). The conversion of G/C to G/U leads to A/T transitions upon replication. However, both transversions and transitions are observed at G/C as well as frequent mutations at A/T base pairs. In mice missing MSH2, MSH6, or EXO1, mutations in Ig variable regions at A/T drop by 80%–90%. Residual A/T mutations are attributable to another repair pathway, base excision repair (BER), which removes uracil from DNA. Since MMR is usually a high-fidelity process, how is it co-opted by the SHM program? The key link is the recruitment of monoubiquitinated PCNA (mUb-PCNA) following the recognition of the G/U mispair by MutSα and MutLα. In replication-associated MMR nonubiquitinated PCNA modulates MutSα and MutLα interactions at mismatches and mediates faithful resynthesis by replicative polymerases. In SHM, mUb-PCNA instead recruits DNA polymerase η (Pol-η), a translesion polymerase that is most error-prone when copying opposite A and T. Thus, SHM in activated B cells depends in part on ncMMR using the mismatch recognition and incision components of MMR and substituting high-fidelity DNA synthesis with error-prone gap filling by Pol-η. The formation of DNA double-strand breaks (DSBs), an early event in CSR, also depends on MutSα and MutLα. These DSBs are prerequisites for the chromosomal rearrangements that juxtapose different Ig switch regions. Thus, in CSR and SHM, MMR has been co-opted for programmed genomic alterations.

Remarkably, Peña-Diaz et al. (2012) report that an ncMMR mechanism analogous to that utilized in SHM exists in nonlymphoid cells and is activated in response to a variety of genotoxic agents with the potential for widespread somatic mutagenesis (Figure 1C). Using an in vitro repair system to monitor correction of G/U, the authors observe MutSα- and MutLα-dependent DNA synthesis that, unlike replication-linked MMR, exhibits no strand selectivity for excision and recruits mUb-PCNA associated with the persistence of single-strand gaps. Most notably this ncMMR is not restricted to B cells as it occurs in 293T cells, a human embryonic kidney-derived cell line, and occurs in cells held in G1 by serum starvation. The existence of a MMR pathway uncoupled to replication receives support from live-cell imaging and mutational studies in S. cerevisiae (Hombauer et al., 2011). In cells treated with the S_N1 alkylator N-methyl-N’-nitro-N-nitrosoguanidine (MNNG), which gives rise to cytotoxic O⁶-methyl guanine (O⁶meG), MMR-dependent mUb-PCNA recruits Pol-η to chromatin. As ubiquitination of PCNA is catalyzed by the E3-ubiquitin ligase RAD18, Peña-Diaz et al. (2012) analyzed HPRT mutation frequencies in MNNG-treated MMR-deficient and -proficient cells in the absence of RAD18, and demonstrate that the increase in MNNG-induced mutation is dependent on both MMR proteins and RAD18. An analogous association of MutSα, mUb-PCNA, and Pol-η is also observed in response to oxidative DNA damage (Zlatanou et al., 2011). Taken together, these findings lend support to the conclusion of Peña-Diaz et al. (2012) that ncMMR involving mUb-PCNA and error-prone Pol-η generates mutations in nonlymphoid cells.

These findings raise many questions regarding the regulation of excision-repair pathways. MMR and BER trigger cell-cycle arrest and apoptosis in response to DNA damage, but they generate strand breaks and gaps of varying sizes that are targets for alternate pathways if repair is somehow compromised. Clearly the fidelity of MMR in somatic cells is dictated in large part by the identity of the polymerase that is recruited behind an advancing replication fork or in nonreplicating chromatin. What are the molecular details underlying polymerase recruitment? Fluctuations in the level and composition of nucleotide pools perhaps tied to metabolic status and the presence of noncanonical bases such as uracil and O⁶meG divert MMR to mutagenic pathways. Might other chemotherapeutic base analogs such as fluorouracil act similarly? Likewise, thiopurines, e.g., 6-thioguanine (6-TG), are commonly prescribed for chronic inflammatory and autoimmune diseases, and their long-term use brings elevated cancer risk (Karran and Attard, 2008). 6-TG, like MNNG, ultimately yields cytotoxic O⁶me-G/T mispairs recognized by MutSα. In these patients, there is tremendous selective pressure to inactivate MMR, undoubtedly a major pathway for the development of cancer. However, there is reason to suspect other pathways, as well. Could ncMMR function in this context? Can error-prone polymerases complete MMR in S phase under conditions of replication stress? In this regard, Nick McElhinny et al., (2010) report that ribonucleotides, which are frequently incorporated into DNA in yeast, are mutagenic and present obstacles for DNA Pol-ε as it directs leading-strand synthesis. What is the nature of interactions between MMR and BER that share substrate specificities? Finally, as Peña-Diaz et al. (2012) point out, the breadth of MutSα substrate specificity encompassing distorted DNA structures that might block resynthesis, e.g., cisplatin crosslinks, hairpins, and G-quartets, begs the question of whether ncMMR promotes gross chromosomal rearrangements in nonlymphoid cells as it does in B cells. In summary, Peña-Diaz et al. (2012) reveal a general pathway for mutagenesis triggered by a broad array of DNA damage and challenge us to understand the cellular contexts under which normally high-fidelity repair systems like MMR can be redirected to create genome instability.