Interstrand cross-links (ICLs) are formed by many chemotherapeutic agents and may also arise endogenously. The mechanisms used to repair these lesions remain unclear in mammalian cells. Repair in Escherichia coli and Saccharomyces cerevisiae requires an initial unhooking step to release the tethered DNA strands. We used a panel of linear substrates containing different site-specific ICLs in order to characterize how structure affects ICL processing in mammalian cell extracts. We demonstrate that ICL-induced distortions affect NER-dependent and independent processing events. The NER-dependent pathway produces dual incisions 5' to the site of the ICL as described previously [Bessho et. al. (1997) Mol Cell Bio 17(12): 6822-30] but does not release the cross-link. Surprisingly, we also found that the interstrand cross-linked duplexes were unhooked in mammalian cell extracts independent of the NER pathway. Unhooking occurred identically in extracts prepared from human and rodent cells and is dependent on ATP hydrolysis and metal ions. The structure of the unhooked product was characterized and was found to contain the remnant of the cross-link. Both the NER mediated dual 5' incisions and unhooking reactions were greatly stimulated by ICL-induced distortions including increased local flexibility and disruption of base-pairs surrounding the site of the ICL. These results suggest that in DNA not undergoing transcription or replication, distortions induced by the presence of an ICL could contribute significantly to initial cross-link recognition and processing. KeywordsInterstrand cross-links; DNA repair; unhooking; mammalian cell extracts Interstrand cross-links (ICLs) covalently tether both strands of the DNA duplex preventing strand separation and therefore transcription and replication (1). This modification poses a unique challenge to the cellular DNA repair machinery because a non-damaged template is not readily available during repair synthesis. An understanding of ICL repair pathways has important clinical implications in chemotherapy as well as aging (1-3).Much of what is known about the mechanisms of ICL repair has come from studies in E. coli and S. cerevisiae, where repair is initiated by the nucleotide excision repair (NER) pathway (1,4,5). NER-dependent incisions are made on either side of the cross-link, thereby 'unhooking' the two DNA strands. Unhooking is then followed by error-free or error-prone repair of the †

The study of DNA repair has been facilitated by the development of extract-based in vitro assay systems and the use of synthetic DNA duplexes that contain site-specific lesions as repair substrates. Unfortunately, exposed DNA termini can be a liability when working in crude cell extracts because they are targets for DNA end-modifying enzymes and binding sites for proteins that recognize DNA termini. In particular, the double-strand break repair protein Ku is an abundant DNA end-binding protein that has been shown to interfere with nucleotide excision repair (NER) in vitro. To facilitate the investigation of NER in whole-cell extracts, we explored ways of modifying the exposed ends of synthetic repair substrates to prevent Ku binding and improve in vitro NER efficiency. Replacement of six contiguous phosphodiester linkages at the 3'-ends of the duplex repair substrate with nuclease-resistant nonionic methylphosphonate linkages resulted in a 280-fold decrease in binding affinity between Ku and the modified duplex. These results are consistent with the published crystal structure of a Ku/DNA complex [Walker et al. (2001) Nature 412, 607-614] and show that the 3'-terminal phosphodiester linkages of linear DNA duplexes are important determinants in DNA end-binding by Ku. Using HeLa whole-cell extracts and a 149-base pair DNA duplex repair substrate, we tested the effects of modification of exposed DNA termini on NER-mediated in vitro excision of a 1,3-GTG-Pt(II) intrastrand cross-link. Methylphosphonate modification at the 3'-ends of the repair substrate resulted in a 1.6-fold increase in excision. Derivatization of the 5'-ends of the duplex with biotin and subsequent conjugation with streptavidin to block Ku binding resulted in a 2.3-fold increase excision. By combining these modifications, we were able to effectively reduce Ku-derived interference of NER excision in vitro and observed a 4.4-fold increase in platinum lesion excision. These modifications are easy to incorporate into synthetic oligonucleotides and may find general utility whenever synthetic linear duplex DNAs are used as substrates to investigate DNA repair in whole-cell extracts.