A symmetrical model for the domain structure of type I DNA methyltransferases

Type I DNA methyltransferases are complex multisubunit enzymes that methylate a specific base in each half of an asymmetric bipartite DNA recognition sequence. The specificity (S) subunit contains two corresponding DNA sequence recognition domains, plus a number of conserved regions which interact with two modification (M) subunits to form a trimeric enzyme of the form M2S. The way in which the subunits interact with DNA in a pseudo-symmetric fashion has long been unclear. Analysis of internal sequence repeats in the S-subunit shows the occurrence of significant homologies between the central conserved domain and sequences near the N and C termini. On the basis of this "split repeat", a "circular" organisation of the domains of this subunit is proposed that provides the required symmetry for interacting with the M-subunits and with the target DNA sequence. In the proposed model, one M-subunit interacts with the N and C-terminal conserved regions of the S-subunit, which are thereby brought into close proximity. The second M-subunit makes equivalent contacts with repeated sequences in the central conserved domain. The model suggests a more general scheme for the imposition of pseudo-dyad symmetry on protein subunits that have internal repeats by making equivalent contacts with additional subunits.