Identification and Cleavage Site Analysis of DNA Sequences Bound Strongly by Bleomycin

A hairpin DNA library containing an 8-base pair sequence-randomized region was employed in a SELEX-type procedure to identify DNAs that bound strongly to bleomycin A5, the latter of which was immobilized on a solid support. Ten hairpin DNAs that bound BLM A(5) strongly were identified and sequenced, and used to characterize DNA binding by the antitumor antibiotic. While all 10 selected hairpin DNAs bound to BLM strongly, they did exhibit a range of binding specificities. Further, while the binding specificity was generally the greatest for hairpin DNAs that contained at least one of the sequences (5'-GC-3' and 5'-GT-3') cleaved most frequently by Fe(II)-bleomycin, the hairpin DNA exhibiting the poorest binding specificity also contained a 5'-GT-3' site. Four of the hairpin DNA substrates were 5 132 P end-labeled and used to assess the preferred sites of cleavage by Fe(II).BLM. The substrate DNAs included two lacking any 5'-GT-3' or 5'-GC-3' site; these were cleaved at 5'-AA-3' and (more strongly) at 5'-AT-3' and 5'-GA-3' sequences. For two hairpin DNAs containing 5'-GT-3' or 5'-GC-3' sequences, cleavage was observed at these sequences as well, but the three aforementioned sequences were also cleaved efficiently. For hairpin DNA 3, which was bound the least well of the 10 DNAs studied, a 5'-TA-3' site was also cleaved efficiently. Thus, the pattern and intensities of cleavage of the four DNAs studied were not entirely consistent with the preferred pattern of DNA cleavage reported for Fe(II).BLM in numerous published studies that have employed arbitrarily chosen DNA substrates. Also studied were the chemistry of DNA cleavage for one of the hairpin DNAs, and competition experiments in which the diminution of cleavage was measured following admixture of a molar excess of a smaller hairpin DNA shown to be an exceptionally good substrate for cleavage by Fe(II)-BLM. In the aggregate, the data indicate that the relationship between DNA binding and degradation by Fe-BLIVI, as well as the chemistry leading to DNA degradation, are more complex than suggested by earlier studies that employed only DNA degradation product analysis as an end point.