Biosynthetic errors that escape the proofreading activity of DNA polymerases are processed by the mismatch repair (MMR) system. Their correction entails the degradation of the mismatch-containing section of the newly synthesized strand and its error-free resynthesis. Although the main MMR players have been identified, the key mechanistic details of the process remain obscure. One of these concerns strand discrimination. How is MMR targeted to the newly synthesized strand? In PNAS, Pluciennik et al. (1) help elucidate this phenomenon. Moreover, their findings suggest that MMR proteins might be involved in the metabolism of noncanonical structures that arise in nonreplicating DNA.

In eukaryotes, the MMR system deploys specific functions such as the base/base mispair and insertion/deletion loop-binding factors MutSα (MSH2/MSH6) and MutSβ (MSH2/MSH3), respectively; MutLα (MLH1/PMS2); and components of the replication machinery proliferating cell nuclear antigen (PCNA), replication factor C (RFC), replication protein A, exonuclease 1 (EXO1), polymerase-δ (pol-δ), and DNA ligase (2). These purified constituents can catalyze the repair of mismatches in circular substrates if there is a preexisting strand discontinuity (a nick or a gap) within ∼1 kb of the mismatch (3). On these substrates, mismatch-activated MutSα was proposed to translocate to the discontinuity and recruit MutLα and EXO1. Both nicks positioned 5′ or 3′ from the mismatch can be used as initiation sites for exonucleolytic degradation, which travels toward the mismatch and generally terminates ∼150 nucleotides beyond it. The resulting gap is then resynthesized by pol-δ (3).

The caveat of this mechanism lies in the fact that EXO1, the only exonuclease implicated in MMR to date, has an obligate 5′→3′ polarity, which made it difficult to imagine how this enzyme could also degrade DNA from a nick positioned 3′ from the mismatch. This puzzle was partially solved by Kadyrov et al. (4), who showed that MutLα was actually a cryptic endonuclease, which could, upon stimulation by mismatch-activated MutSα or MutSβ, introduce additional nicks into the discontinuous strand. However, it was unclear how MutLα could distinguish between the intact strand and that which already contained a break several hundred nucleotides downstream. The second caveat concerned the question of the directionality of these cleavage events. Why did activated MutLα travel from the break toward the mismatch rather than away from it? The answer to this puzzle appears to lie in an interaction between MutLα and PCNA.

PCNA is a homotrimeric, ring-shaped molecule that diffuses along the DNA contour in the form of a sliding clamp (5). It is loaded onto DNA by RFC, which opens the PCNA ring and closes it again once the DNA helix is encircled (6). Once loaded, PCNA remains on the DNA until actively unloaded. Thus, through interacting with PCNA, proteins increase their residence time on DNA.

The sliding clamp has two distinct faces. Although the distal face has no known function, the proximal face of PCNA contains the residues that interact with its plethora of partners (7), mostly via a PCNA-interacting peptide (PIP) motif.

RFC loads PCNA preferably at the interface of single-stranded DNA (ssDNA) and double-stranded DNA (dsDNA), with the proximal side facing the primer 3′ terminus. In this way, interaction with a DNA polymerase will ensure that the enzyme is correctly oriented for DNA synthesis. Pluciennik et al. (1) show that a similar mechanism may also operate in MMR, but in reverse. Mismatch-activated MutSα or MutSβ can interact with PCNA (8) and may displace the polymerase, thus freeing the 3′ terminus of the newly synthesized DNA strand. Loading of MutLα may generate a complex that can slide along the dsDNA toward the mismatch, whereby MutLα may introduce nicks into one of the strands. These nicks can be used as EXO1 loading sites. In a complex with PCNA, it is easy to imagine how MutLα cleaves only the correct strand. Assuming that it hydrolyzes only 5′→3′-oriented phosphodiester linkages, only one strand emerging from the PCNA/MutLα complex has this polarity and will therefore be cleaved, even though the terminus where the complex was loaded might be many hundred nucleotides distant (Fig. 1A). This mechanism provides an explanation for the puzzling finding made many years ago that PCNA was required for a step of MMR preceding DNA resynthesis (9).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Hypothetical mechanism of MMR in the leading strand. (A) A G/T mismatch made by pol-δ is detected by MutSα, which translocates along the DNA contour and displaces the polymerase from its complex with PCNA. Loading of MutLα and reversal of the direction of travel of the protein complex allow MutLα to introduce nicks in the leading strand, which are used as initiation sites for EXO1-catalyzed degradation. Once the mismatch is removed by EXO1, MutSα is no longer stimulated, and pol-δ can resume leading-strand synthesis. (B) Substrates where RFC might load PCNA and thus give rise to MutLα-dependent nicking. Only one orientation of the PCNA/MutLa complex is shown for simplicity, but the complex could also load in an inverse orientation such that either strand could be incised.

The interaction of PCNA and MutLα requires further study. Pluciennik et al. (1) show that it can be abolished by a peptide originating from p21^CIP1/WAF1 that has a high-affinity PIP site (10) and is known to interfere with many PCNA-dependent processes. However, neither MLH1 nor PMS2 carries a well-conserved PIP motif, and although human PCNA and MutLα interact in vitro (2), it is of interest to find out how they interact and whether their interaction tolerates other known PCNA partners such as MutSα, given that PCNA has three equivalent PIP-binding sites.

By clarifying the mechanism of strand discrimination, Pluciennik et al. (1) show how MMR contributes to replication fidelity. However, their work has a further dimension, whereby MMR may contribute to genomic instability. Some years ago, genetic experiments in mice implicated the MMR system in triplet repeat expansions (TRE), a phenomenon linked to several neurodegenerative diseases (11). In these syndromes, runs of CAG/CTG trinucleotides can increase in size, which can bring about gene silencing or loss of function. The mechanism of TRE has been puzzling and its link to MMR even more so. Pluciennik et al. (1) show that RFC can load PCNA—albeit with low efficiency—also at ssDNA/dsDNA junctions that arise in supercoiled DNA or in a substrate containing a stretch of noncomplementary nucleotides (Fig. 1B). Although PCNA is, in this scenario, loaded onto DNA that has no discontinuity, it still appears to recruit MutLα, which then introduces nicks into the DNA, albeit without strand bias. Triplet repeats form relatively stable stem-loop extrusions that are bound by MutSβ in vitro (12). A MutSβ-stimulated, MutLα-catalyzed break in either strand could cause these structures to collapse, and their subsequent metabolism might lead to changes in repeat length. In addition to triplet repeats, PCNA/MutLα might also nick other structures in nonreplicating DNA. This could trigger metabolic processes that implicate MMR proteins but cannot be explained by our current understanding of this process.

The work of Pluciennik et al. (1) represents another major achievement of Modrich’s laboratory and demonstrates how relatively simple biochemistry can help solve complex biological phenomena.