Mismatch Repair and Somatic Hypermutation--A Tale of a Double-Edged Sword

Marilia Cascalho

Mutation is at the heart of evolution. But because most mutations are deleterious, genomic stability requires a low mutation rate, which is achieved by powerful repair systems. Paradoxically, in parallel to the repair systems, a mutator system has evolved in higher organisms. This mutator system increases the mutation rate at the immunoglobulin loci a million times over the spontaneous rate at other loci. It introduces somatic point mutations into the exons encoding the variable region of the immunoglobulin and into its flanking regions, thereby adding to the diversity of the antibody repertoire. Antigen can select somatically mutated B cells to produce antibodies of higher affinity. Thus, hypermutation is intrinsically linked to the generation of a memory response and is a reliable marker of memory B cells (1).

The possibility of potent DNA repair mechanisms coexisting with hypermutation in B cells represented a major conundrum that fascinated me and motivated my thesis work. Hypermutation can be thought of as the result of two steps: (i) introduction of mutations in one strand, resulting in mismatches; and (ii) resolution of the mismatches, leading to the fixation of mutations. Clearly, DNA repair has to be considered in this process, at least in the events that follow the introduction of mutations. This is why I chose to study the role of mismatch repair in somatic hypermutation.

The mismatch repair system corrects DNA replication errors and blocks recombination between divergent DNA sequences. In Escherichia coli, the prototypic mismatch repair requires the mutS, mutL, and mutH gene products. It corrects the newly synthesized strand, which is marked by its transient lack of methyl groups. In eukaryotic cells single-strand nicks have been proposed as a marker for the new strand. Inactivation of the mismatch repair has been shown to result in a mutator phenotype in bacteria (2) and in mice (3). In both mice and men, mismatch repair deficiency has been related to tumor development; for example, nonfunctional human homologs of mutS and mutL have been implicated in hereditary nonpolyposis colon cancer (4).

Considering this, what is the role of mismatch repair in somatic hypermutation? A priori, there are three possibilities: the immunoglobulin mutator may overwhelm the DNA repair system with so many mutations that they become fixed before repair can occur; the immunoglobulin mutator also may deactivate some or all of the repair systems in hypermutating B cells, or they simply may not be present; and, finally, the mismatch repair may be subverted to fix the incipient mutations rather than to correct them.

To test these hypotheses I first generated a mouse with a very limited antibody repertoire by gene targeting. In this "Quasi-Monoclonal" (QM) mouse only a few germline elements contribute to the primary immunoglobulin repertoire, which is developed by standard V(D)J rearrangement: at birth, all B cells produce the same heavy chain and just a few light chains, making identifications of mutations easy. The very limited repertoire also creates an enormous selection pressure for diversification, and consequently increases the frequency of mutants, which can be easily sorted from the peripheral blood for sequence analysis (5).

The QM mouse was made deficient in mismatch repair by crossing it with the targeted "knockout" allele of Pms2, a mutL homolog (3). When I first compared the numbers of B cells between QM mice proficient and deficient in mismatch repair, I was surprised to find no difference between equivalent populations; this included the populations I knew to be hypermutated in the QM mice. But when I sequenced both the immunoglobulin heavy chain and light chain variable regions in these cells, it became clear that absence of mismatch repair leads to a decrease rather than an increase in hypermutation. Apparently mismatch repair contributed to, rather than stifled, somatic hypermutation. So, how is mismatch repair co-opted by the immunoglobulin mutator to increase the number of mutations? An attractive suggestion would be that after mismatches have been introduced at the immunoglobulin locus, the mismatch repair identifies the "wrong," mutated strand as a template and thus makes the mutations permanent(6).

The findings described above identified Pms2 as the first trans-acting factor for somatic hypermutation; they also provide a tool for identifying other components of the immunoglobulin mutator system. The suggestion of an active role for a DNA repair mechanism in the creation of mutations ought to have implications in tumor biology. Because many tumors exhibit high mutation rates (7), a mutator phenotype may not just arise because of a loss of function, as generally held, but may also result from a gain of function of a DNA repair component. It is possible that the co-option of the mismatch repair, as described here for the immunoglobulin genes, could play a role. Consistent with this view is the fact that mismatch repair null mutants do not reproduce the clinical features of missense mutations.

In more general terms, the study of hypermutation may lead the way for mutation research over the next years. The high frequency of mutations makes it easier to do experiments, and the idea that mutations are introduced not just by error, but by design, might bear out in cells other than lymphocytes.


    • M. Weigert et al., Nature228, 1045 (1970); D. McKean et al., Proc. Natl. Acad. Sci. U.S.A.81, 3180 (1984); C. A. Reynaud et al., Semin. Immunol.8, 125 (1996); M. Wabl et al., Curr. Opin. Immunol.11, 186 (1999).
    • Reviewed in R. Lahue and P. Modrich, Annu. Rev. Biochem.65, 101 (1996).
    • S. M. Baker et al., Cell82, 309 (1995); L. Narayanan et al., Proc. Natl. Acad. Sci. U.S.A.94, 3122 (1997).
    • R. A. Fishel et al., Cell75, 1027 (1993); F. S. Leach et al., Cell75, 1215 (1993); C. E. Bronner et al., Nature368, 258 (1994); N. C. Nicolaides et al., Nature371, 75 (1994); N. Papadopoulos et al., Science263, 1625 (1994).
    • M. Cascalho et al., Science272, 1649 (1996); M. Cascalho et al., J. Immunol.159, 5795 (1997).
    • M. Cascalho et al., Science279, 1207 (1998).
    • L. A. Loeb, Science277, 1449 (1998).