ABSTRACT: We describe in detail a direct assay for the substrate-inactivated DNA-repair enzyme, O6-methylguanine-DNA methyltransferase (O6-MT), which measures the transfer of radiolabelled methyl groups from a prepared O6-methylguanine-DNA substrate to the protein fraction of an enzyme-containing cell/tissue extract. This assay, a modification of a previously suggested method for monitoring O6-ethylguanine-DNA repair [Renard, Verly, Mehta & Ludlum (1983) Eur. J. Biochem. 136, 461-467], is sensitive, highly reproducible, accurate and is, as described here and relative to previously published methods, well suited for use with a large number of test samples. We identified two problems with the O6-[Me-3H]methylguanine-DNA substrate used in the present work and in other reported assay: firstly, that of progressively higher assay backgrounds with increasing age of substrate, which was nullified by once-only purification of the double-stranded substrate by hydroxyapatite chromatography; secondly, a substrate of high specific radioactivity (30 Ci/mmol), made with freshly prepared tritiated methylnitrosourea, behaved as a substrate of 5 Ci/mmol when referenced against radiolabelled O6-methylguanine-DNA made with either [3H]- or [14C]-methylnitrosourea at the lower specific radioactivities of 1 Ci/mmol and 61 mCi/mmol respectively. This apparently stemmed from the known instability of high-specific-radioactivity [3H]methylnitrosourea and indicated that an expected increase in sensitivity of the assay does not necessarily result from increasing the specific radioactivity of substrates above approx. 1 Ci/mmol. Although O6-MT was stable to preincubation at 25 degrees C, marked losses of activity were observed at 37 degrees C, and more so at 45 degrees C. Enzyme lability at the higher temperatures was not, however, seen during preincubation in the presence of its substrate. O6-[Me-3H]methylguanine-DNA, which apparently protected O6-MT against thermal inactivation. As previously seen with other human cells and tissues, extracts of human spleen in the present study showed wide interindividual differences in O6-MT specific activity (18-fold), which spanned the range 50-900 fmol/mg of protein. Cultured human lymphoblastoid Jurkat cells contained approx. 57,000 enzyme molecules/cell. Substrate-inactivated [Me-3H]methylated O6-MT was analysed by SDS/PAGE and electroblotting. The different but similarly sized forms of this enzyme that we previously detected in human spleen [Major, Gardner, Carne & Lawley (1990) Nucleic Acids Res. 18, 1351-1359] were clearly resolved by fluorography of electroblots, but only at considerable expense of time. As expected, scintillation counting of the protein extracted from gel slices and linear-wire scanning of enzyme-associated radioactivity on electroblots were quicker methods for detecting the [Me-3H]methylated inactivated O6-MT.