Perspectives
The MutS and MutL mismatch repair proteins are highly conserved through evolution. However, there are notable differences between the systems. Thus, only gram-negative bacteria possess MutH proteins; the remainder of the prokaryotic kingdom, as well as all eukaryotes have found another solution to complete the all-important task of strand discrimination. Even within eukaryotes there are differences; for example, the N-terminal region of the yeast and human MSH6 protein differ in structure and function. To date, these differences have been largely ignored, because the main questions that the field was attempting to answer concerned the elucidation of the general principles of the MMR pathway. This has resulted in several studies of mismatch repair complex formation in E.coli, yeast or humans. To further expand and comprehensively integrate these studies into quantitative models for the pro- and eukaryotic pathways, we need to gather detailed, quantitative information about the constituent proteins and about their interaction with heteroduplex and homoduplex substrates, as well as with one another. This is only feasible if the individual steps of the process are analyzed with kinetic and thermodynamic principles in mind, using a combination of state-of-the-art technology from different disciplines, and through the development of novel technologies, assays and reporter systems. Some of the key aspects of this program are described below.
Construction of a quantitative model of DNA mismatch repair with relevance for human health
Mutations in human mismatch repair genes, primarily MSH2 and MLH1, lead to cancer of the colon, endometrium and other organs with high penetrance. Mutations in the other key mismatch repair genes, such as MSH6 and PMS2 have also been detected, but these lead to somewhat attenuated phenotypes, due to functional redundancy with MSH3 and MLH3, respectively. In addition, changes in MMR protein concentration (e.g. during the E. coli life cycle, or due to the hypermethylation of MLH1 promoter in human cancer) will affect the function and capacity of the mismatch repair machinery.
In order to understand the key events of DNA mismatch repair, we will study the kinetics of
in vitro and in vivo mismatch repair in the model system E.coli and in the human system. We will systematically vary parameters such as protein concentrations, and type and number of lesions (mismatches). We will also make use of protein variants containing cancer-related mutations. Individual rate constants for mismatch binding, repairosome assembly, daughter strand incision and degradation will be determined in biochemical and biophysical experiments, as well as in experiments focussing on the in vivo assembly of the MMR repairosome. These rate constants will be used as input for the construction of mathematical models describing the individual steps and sub-steps of DNA mismatch repair. The starting models will be simple, depicting a sequential assembly of the MMR components on a DNA mismatch. Iterative cycling between experimental rate constant determinations and mathematical modelling will result in the refinement of the models and suggestions for experiments addressing additional steps or constraints in the pathway. Because we address the system from many sides (E.coli versus human; in vitro versus in vivo, bulk versus single molecules, empirical versus experimental, correlation of kinetic data obtained from different techniques), we expect to make significant progress towards a mathematical description of DNA mismatch repair.
Multidisciplinary approach
DNA mismatch repair is capable of detecting very rare mistakes in DNA pairing upon replication (one in a million) and coupling that to identification of a distant strand discrimination signal. This fascinating combination of exquisite sensitivity and long-distance signalling makes for a system that is difficult to study with a single technique. We will therefore address this coupling using multidisciplinary technologies. Starting at the single molecule level, we will use real-time single-molecule DNA nanomanipulation (Paris) to address kinetics of DNA structural changes induced by repair complex assembly and correlate these changes to location of DNA mismatch and strand discrimination signals engineered into the nanomanipulated DNA. We will furthermore directly visualize the repair complex on a DNA mismatch using scanning force microscopy (SFM, Rotterdam). This technique will resolve the complex at nanometer resolution, allowing us to determine if a single SLH complex is assembled or, for example if one or more of the components form filaments along the DNA. In addition it will permit us to visualize conformational changes in the DNA such as kinking, wrapping or looping which, on DNA containing a hemimethylated GATC site in addition to the mismatch, will allow correlation to the data obtained from DNA nanomanipulation and identification of the relevant coupling mechanism. These single-molecule data will be correlated to kinetics of complex formation using Biacore real-time protein interaction analysis (Amsterdam), and MutH activation assays in which kinetics of activation can be correlated to distance between mismatch and strand discrimination signal (Giessen). Furthermore enzymatic analysis of the MutS and MutL ATPase activities will reveal if this high-energy cofactor is used to switch between functional states or drives active protein translocation or DNA spooling (Rotterdam). In addition we envisage that these analyses
provide insight into the unknown factors or mechanism that would explain the discrepancy between specificity for a DNA mismatch at different steps in the repair reaction.
Combining these studies will provide a full picture of when, where and how the repair initiation complex couples mismatch recognition to determination of which
DNA strand contains the actual mispaired base.