DELETION MUTANTS OF POLY(ADP-RIBOSE) POLYMERASE SUPPORT A MODEL OF CYCLIC ASSOCIATION AND DISSOCIATION OF ENZYME FROM DNA ENDS DURING DNA-REPAIR

With an in vitro DNA repair system, Satoh and Lindah [(1992) Nature 356, 356-358] demonstrated that unmodified poly(ADP-ribose) polymerase (PADPRP) binds to radiation-damaged DNA and inhibits repair in the absence of NAD. However, in the presence of NAD, PADPRP undergoes automodification and the DNA is repaired. It was hypothesized that PADPRP cycles between an unmodified form, which protects DNA breaks, and an automodified form, which is released from DNA, thereby allowing access to repair enzymes. We have now tested this model with bacterially expressed mutants of PADPRP with deletions in the three major functional domains of the enzyme [Cherney, B. W., Chaudry, B., Bhatia, K., Butt, T. R., and Smulson, M. E. (1991) Biochemistry 30, 10420-10427]. Deletion mutants with an intact amino-terminal DNA-binding domain, and therefore capable of binding to DNA strand breaks in the in vitro assay, inhibited repair; however, whether the deletion was in the NAD-binding, active site domain or the automodification domain, the inhibition of repair exerted by these mutant proteins was not alleviated by NAD. A PADPRP mutant with a deletion in the DNA-binding domain did not inhibit DNA repair. Thus, the behavior of these PADPRP deletion mutants is consistent with the model proposed earlier. The model was also supported by experiments with Manley extracts of HeLa cells stably transfected with a PADPRP antisense RNA construct. Extracts of cells induced to express antisense RNA did not markedly inhibit in vitro DNA repair, nor did the addition of NAD influence the assay. In contrast, noninduced cell extracts inhibited repair and inhibition was alleviated by NAD. Finally, exposure of nuclei from mid-S phase HeLa cells to NAD increased single deoxynucleotide incorporation as catalyzed by exogenous Escherichia coli DNA polymerase. This effect, probably attributable to chromatin restructuring, also may be explained, at least in part, by the proposed model for PADPRP cycling and DNA strand break rejoining.