Mismatch repair rectifies base-pairing errors within the DNA helix. Although a mismatched base pair is a rare occurrence in DNA, inactivation of this repair system has profound consequences for a living cell: a 100- to 1,000-fold increase in mutation production. In humans, mismatch repair defects are the cause of the most common form of hereditary colon cancer (HNPCC) and are believed to contribute to the development of a subset of sporadic tumors that occur in a variety of tissues.
Correction of DNA Replication Errors in Escherichia coli
Mismatch repair stabilizes the cellular genome in several ways. The best understood of these is its function as a molecular editor, which corrects errors that occur during the course of chromosome replication. Because replication errors are composed of incorrectly paired but otherwise normal Watson-Crick bases, mismatch repair systems rely on secondary signals to direct repair to the newly synthesized DNA strand that contains the mistake. In E. coli, these strand signals are based on methylation at GATC sequences by the DNA adenine methylase. Newly synthesized DNA is transiently unmethylated at GATC sequences, and it is the absence of this modification that directs repair to the new strand.
E. coli methyl-directed repair depends on 11 activities, including the products of the four mutator genes: mutH, mutL, mutS, and mutU. Our biochemical studies have shown that MutS is responsible for mismatch recognition, and MutL couples mismatch recognition by MutS to activation of downstream activities. One downstream activity is MutH, an endonuclease that incises the unmethylated strand at a GATC sequence in newly synthesized DNA. MutS and MutL also activate the excision system, which loads at the MutH strand break and is composed of the mutU gene product DNA helicase II and one of several single-strand-specific exonucleases. The strand break introduced by MutH can be located either 3’ or 5’ to the mismatch, and MutS and MutL coordinate recognition of these two DNA sites in a manner that permits orientation-dependent loading of the appropriate 3’ to 5’ or 5’ to 3’ hydrolytic system. Excision terminates upon mismatch removal, and DNA polymerase III holoenzyme repairs the ensuing gap. Ligase restores covalent continuity to the helix.
Correction of DNA Replication Errors in Human Cells
In contrast to the E. coli reaction, DNA methylation does not provide strand direction to mismatch repair in higher cells. Although the strand signals that direct eukaryotic repair have not been genetically identified, we found that a strand break located either 5’ or 3’ to the mismatch is sufficient to direct the reaction in human cell extracts. This permitted us to identify two key activities involved in the initiation of human mismatch repair: MutSα, a heterodimer of the MutS homologs MSH2 and MSH6 that is responsible for most mismatch recognition events in human cells, and MutLα, a heterodimer of the MutL homologs MLH1 and PMS2 that is recruited to the mismatch in a MutSα-dependent manner. These two heterodimers are of interest because genetic inactivation of either causes cancer.
To study how mismatch repair occurs in the mammalian cell, we have examined purified forms of the proteins that have been implicated in the process, including MutSα, MutLα, Exo1 (an exonuclease that hydrolyzes duplex DNA with 5’ to 3’ polarity), RPA (a protein that binds to and stabilizes single-stranded DNA), PARP-1 (a protein that binds to DNA strand breaks and signals their presence), PCNA (a trimeric protein that forms a ring-like clamp around DNA and functions at the replication fork), RFC (the enzyme that loads the PCNA clamp onto the helix), and DNA polymerase δ (an enzyme that plays a major role in chromosome replication). This has revealed several reactions that likely contribute to correcting replication errors in higher cells.
The simplest of these reactions is mismatch repair directed by a 5’ strand break (Figure 1). MutSα activates mismatch-dependent excision by Exo1, which hydrolyzes the incised strand from the 5’ terminus. RPA, which binds to the single-stranded gap produced in this manner, controls the excision process and down-regulates hydrolysis after mismatch removal. MutLα and PARP-1 are not required for excision in this system; however, by suppressing Exo1 action on mismatch-free DNA they substantially enhance the mismatch dependence of the reaction. The RPA-filled single-stranded DNA gap produced in this manner is repaired by DNA polymerase δ with the assistance of RFC and PCNA. We think it likely that this simple pathway contributes to replication error removal from the lagging strand of the replication fork where 5’ termini are produced during the course of DNA synthesis.
We also identified a more complex mode of mismatch rectification in which repair can be directed by either a 5’ or 3’ strand break. Initiation of this mode of repair involves activation of a latent endonuclease within MutLα in a reaction that requires a mismatch, MutSα, RFC, PCNA, ATP, and a DNA strand break. Strand direction is manifested at this step of repair: action of the MutLα endonuclease is directed to the heteroduplex strand that contains a preexisting break and is biased to the distal side of the mismatch (Figure 2, left pathway). Incision in this manner brackets the mispair with 5’ and 3’ strand breaks. These multiply incised molecules are substrates for MutSα-activated Exo1, which removes the mismatch.
The PCNA replication clamp plays critical roles in the activation and strand direction of the MutLα endonuclease. The two faces of the PCNA clamp are not equivalent, and several other laboratories, including those of Michael O’Donnell (HHMI, Rockefeller University) and John Kuriyan (HHMI, University of California, Berkeley), have shown that PCNA is loaded with a unique orientation at a DNA strand break. We have found that the heteroduplex strand break serves as a site for PCNA loading and that the orientation of clamp loading, relative to the absolute orientation of the helix, determines the strand direction of MutLα incision. These findings also provide a hint concerning the nature of the strand signals that direct mismatch repair in the eukaryotic cell. Because the orientation of PCNA loading at the replication fork is determined by 3’ termini on the leading and lagging strands, such termini are obvious candidates for the strand signals that direct MutLα endonuclease action within the eukaryotic cell.
While Exo1 is believed to play an important role in mammalian mismatch repair, mouse genetic studies indicate that a substantial fraction of repair events can occur in an Exo1-independent manner. One possibility is that other exonucleases may substitute for Exo1, but attempts to identify such enzymes have not been successful. However, we have identified an exonuclease-independent mode of mismatch repair that may partially resolve this paradox. Exo1-independent repair, which depends on MutLα endonuclease action, occurs by the mechanism shown in Figure 2 (right pathway). After incision by MutLα, the multiply incised heteroduplex serves as substrate for DNA synthesis–driven strand displacement by polymerase δ. This leads to removal of a DNA segment spanning the mismatch and concomitant mismatch repair. This reaction occurs more slowly than Exo1-dependent mismatch correction but is nevertheless sufficient to account for the rate of Exo1-independent repair that has been observed in mouse cell extracts.
Mutagenic Action of Mismatch Repair
Although we usually think of mismatch repair as a mutation avoidance system, there are two known instances where action of the pathway is required for mutation production. One of these is the expansion of (CAG)n/(CTG)n triplet repeat sequences, the cause of a number of neurodegenerative disorders such as Huntington's disease and myotonic dystrophy. Work of others with mouse models has shown that (CAG)n/(CTG)n expansion in somatic cells depends on the mismatch repair activities MutSβ (MSH2-MSH3 heterodimer that recognizes extrahelical extrusions of 2-10 nucleotides) and MutLα. A popular model invokes MutSβ recognition of small (CAG)x or (CTG)x extrusions produced by "strand slippage" and processing by the repair system as key events that may lead to repeat expansion. However, because expansion can occur in postmitotic cells, it has been unclear how mismatch repair initiates on nonreplicating DNA. We have identified a simple mechanism by which this may occur.
We have found that (CAG)n/(CTG)n extrusions of 2- or 3-repeat units serve as moderately effective PCNA loading sites when present in covalently continuous DNA. Because these lesions are also well-recognized by MutSβ, this leads to activation of MutLα endonuclease, which can incise either DNA strand, thus providing an intermediate for downstream repair events. Indeed, covalently continuous heteroduplex DNA containing a 2- or 3-repeat unit extrusion also triggers MutSβ- and MutLα-endonuclease-dependent mismatch repair in nuclear extracts of human cells, and the reaction occurs without obvious strand bias. These findings thus provide a simple mechanism for initiation of triplet repeat processing in nonreplicating DNA.
Structural Analysis of Human Mismatch Repair Proteins
Our long-term collaboration with Lorena Beese’s laboratory (Duke University) has yielded structures for MutSα bound to a variety of DNA lesions, as well as several structures of the Exo1 amino-terminal catalytic domain bound to a 5’-recessed DNA duplex. In addition to clarifying the mode of DNA lesion recognition, the MutSα structures have permitted visualization in a structural context of those amino acid residues that are altered by missense mutations in HNPCC patients. Exo1 structures have revealed that this nuclease introduces a sharp bend into the helix at a nick or gap while fraying the 5’ terminus that is destined for hydrolysis. Structural features of the Exo1-DNA complex and comparative biochemical analysis of the catalytic domain and full-length forms of the enzyme have also suggested a simple mechanism for allosteric activation of Exo1 by partner proteins like MutSα.
Work on the bacterial and human mismatch repair systems has been supported in part by grants from the National Institutes of Health.
As of September 9, 2013