EFFECT OF O6-ALKYLGUANINE DNA ALKYLTRANSFERASE ON THE FREQUENCY AND SPECTRUM OF MUTATIONS INDUCED BY N-METHYL-N'-NITRO-N-NITROSOGUANIDINE IN THE HPRT GENE OF DIPLOID HUMAN FIBROBLASTS

N-Methyl-N'-nitro-N-nitrosoguanidine (MNNG) reacts with 12 nucleophilic sites in DNA to induce a variety of lesions, but O6-methylguanine (O6-MeG) and O4-methylthymine are the most effective premutagenic lesions produced, mispairing with thymine and guanine, respectively. O6-MeG is repaired by O6-alkylguanine-DNA alkyltransferase (AGT), which removes the methyl group from the O6 position and transfers it to itself, rendering the transferase inactive. When diploid human fibroblasts were exposed to 25-mu-M, O6-benzylguanine (O6-BzG) in the medium for 3 h, their level of AGT activity was dramatically reduced, to a level of at most 1.6% of the control. Populations of cells pretreated with this level of O6-BzG for 2 h or not pretreated, were exposed to MNNG at a concentration of 2, 4 or 6-mu-M in the presence or absence of O6-BzG and assayed for survival of colony-forming ability and the frequency of 6-thioguanine-resistant cells (mutations induced in the HPRT gene). O6-BzG (25-mu-M) was also present in the appropriate half of the cells during the 24 h immediately following exposure to MNNG. This 27-h exposure to O6-BzG alone had no cytotoxic or mutagenic effect on the cells but significantly increased the cytotoxicity and mutagenicity of MNNG, increasing the mutant frequency to that found previously in human cells constitutively devoid of AGT activity. At doses of 2-mu-M and 4-mu-M MNNG, the mutant frequency observed with the AGT-depleted cells was 120 X 10(-6) and 240 X 10(-6), respectively; in the cells with abundant AGT activity, these values were 10 x 10(-6) and 20 X 10(-6), respectively. DNA-sequence analysis of the coding region of the HPRT gene in 36 independent mutants obtained from MNNG-treated AGT-depleted populations and 36 from the control populations showed that even though AGT repair lowered the frequency of mutants by more than 90%, it did not affect the kinds of mutations induced by MNNG nor the strand distribution of the premutagenic guanine lesions. In mutants from the AGT-depleted cells, there were 26 base substitutions and 13 putative splice site mutations; in the control, there were 25 base substitutions and 11 splice site mutations. All but two substitutions involved G . C with 92% being G . C --> A . T. In both sets, approximately 73% of the premutagenic lesions were located in the nontranscribed strand. Many 'hot spots' were seen, and there was evidence that AGT repaired more lesions from the 5' half of the gene than from the 3' half.