RPA is a eukaryotic heterotrimeric single-stranded DNA binding protein and is essential for DNA replication, recombination, and repair. In this study, crystal structures of the soluble RPA heterodimer, composed of the RPA14 and RPA32 subunits, have been determined for the full-length protein in multiple crystal forms. In all crystals, the electron density for the N-terminal and C terminal regions of RPA32 is weak and of poor quality indicating that these regions are disordered and assume multiple positions in the crystals. Thus, the RPA32 N-terminus appears to be inherently disordered in the unphosphorylated state. The C-terminal, protein-protein interaction domain, adopts several conformations perhaps to facilitate its interaction with various proteins. Although the ordered regions of RPA14/32 resemble the previously solved protease-resistant core crystal structure, the quaternary structures between the heterodimers are quite different. Thus, the 4-helix bundle quaternary assembly noted in the original core structure is unlikely to be related to the quaternary structure of the intact heterotrimer. An organic ligand binding site between subunits RPA14 and RPA32 was identified to bind dioxane.

Rad52 and RPA, along with other members of the Rad52 epistasis group, repair double-stranded DNA breaks (DSBs). In this thesis, it provided critical structural information and indicated that wild-type Rad52 formed an undecameric ring in solution. Surprisingly, these large complexes of wild-type Rad52 rings were disrupted by the binding of ssDNA alone, RPA alone, and most effectively by preformed ssDNA-RPA. The ssDNA-RPA formed complexes with several forms of Rad52, including rings, horseshoes, monomers and other intermediates. Interestingly, in vitro phosphorylated RPA promoted complex formation with monomeric Rad52 and caused the transfer of ssDNA from RPA to Rad52. This indicated that RPA phosphorylation may regulate the first steps of DSB repair and in particular single-strand annealing reactions. Also, since RPA is in excess over Rad52 in the cell, these data suggest that the active form of Rad52 in vivo is probably monomeric and not the multimeric ring form.