We have developed a quantitative method for examining the removal ofN-methylpurines from specific genes to investigate their possible differential repair throughout the genome. Chinese hamster ovary cells were exposed to dimethyl sulfate, and the isolated DNA was treated with an appropriate restriction endonuclease. The DNA was heated to convert remaining N-methylpurines to apurinic sites to render them alkaline-labile. Duplicate samples heated in the presence of methoxyamine to protect the apurinic sites from alkaline hydrolysis provided controls to assess total DNA. After alkaline hydrolysis, agarose gel electrophoresis, Southern transfer, and probing for the fragment of interest, the ratios of band intensities of the test DNA sample to its methoxyamine-treated control counterpart were calculated to yield the percentage of fragments containing no alkaline-labile sites. The frequency of N-methylpurines was measured at different times after dimethyl sulfate treatment to study repair. We found no differences between the rates of repair of N-methylpurines in the active dihydrofolate reductase gene and a nontranscribed region located downstream from it in treated cells. Also, similar rates of repair were observed in the transcribed and nontranscribed strands of the gene, in contrast to previous results for the removal of cyclobutane pyrimidine dimers. Thus, there does not appear to be a coupling of N-methylpurine repair to transcription in Chinese hamster ovary cells. However, the repair in the dihydrofolate reductase domain appears to be somewhat more efficient than that in the genome overall. Our method permits the quantifying at the defined gene level of abasic sites or of any DNA adduct that can be converted to them.The excision-repair of structure-distorting DNA lesions in mammalian chromatin is generally heterogeneous (reviewed in refs. 1-3). Cyclobutane pyrimidine dimers are excised rapidly from the active dihydrofolate reductase (DHFR) gene but persist in the bulk DNA in UV-irradiated Chinese hamster ovary (CHO) cells (4). Furthermore, efficient DNA repair is selective for the transcribed strand of the DHFR gene (5). Some bulky chemical DNA adducts also exhibit differential repair in mammalian cells, including aflatoxin B1 and psoralen photoadducts, which are removed less efficiently from the unexpressed a-satellite DNA sequences than from the bulk DNA in African green monkey cells (6-8).Within the DHFR gene in human cells, the interstrand cross-linking of DNA by psoralen diadducts is repaired at a faster rate than are psoralen monoadducts in the same region (9). Thus, efficiency of DNA repair may be affected by a variety of factors, including the nature of the damage, its location within the genome, the state of expression of the affected DNA sequences, and the chromatin configuration of the region under study.Simple methylating agents, such as dimethyl sulfate (DMS), produce a variety ofdamaged bases in DNA of which 7-methylguanine and 3-methyladenine constitute approximately 90%o of the alteratio...