Differential introduction and repair of psoralen photoadducts to DNA in specific human genes.

We have developed a novel procedure to measure interstrand DNA cross-linking in specific DNA sequences. After alkaline denaturation, CsCl gradient equilibrium sedimentation at pH 10.8 is used to resolve cross-linked double-stranded DNA from un-cross-linked single-stranded DNA. The DNA in gradient fractions is slot-blotted and hybridized with 32P-labeled DNA probes for the sequences of interest. After densitometric quantitation of the autoradiograms, the fraction of DNA cross-linked is determined by the ratio of cross-linked DNA to total DNA (the sum of cross-linked and un-cross-linked DNA). We have used this approach to measure the initial levels of production and extent of repair of the photoadducts of 4'-hydroxymethyl-4,5',8-trimethylpsoralen, i.e., both interstrand cross-links and cross-linkable monoadducts, in specific DNA sequences in cultured human cells. Under conditions in which DNA fragments carrying the expressed dihydrofolate reductase gene were extensively modified, with approximately 92% of the fragments cross-linked, only 37% of the fragments containing the unexpressed fms protooncogene were cross-linked. The overall level of cross-linking for bulk DNA was 74%. Within 24 h, 90% of the cross-linking had been removed from the dihydrofolate reductase gene, whereas little removal was detected in fms, and the bulk DNA showed 31% removal. From this study, we conclude that both the introduction and removal of 4'-hydroxymethyl-4,5',8-trimethylpsoralen adducts are dependent upon the target DNA sequence and its transcriptional activity. The implications for DNA repair of chromatin structure and active transcription are discussed in relation to our results.