Gemcitabine [2',2'-difluoro-2'-deoxycytidine (dFdCyd)] is a potent ionizing radiation sensitizer in solid tumor cells in vitro and in vivo. Previously, we have demonstrated (Shewach et al., Cancer Res., 54: 3218-3223, 1994) a strong correlation between depletion of dATP (caused by dFdCyd diphosphate-mediated inhibition of ribonucleotide reductase) and radiosensitization. In addition, we and others (Latz et al., Int. J. Radiat. Oncol. Biol. Phys., 41: 875-882, 1998; Ostruszka and Shewach, Cancer Res., 60: 6080-6088, 2000) have shown that the accumulation of cells in S phase prior to irradiation is also important for radiosensitization with dFdCyd. This led us to hypothesize that the incorporation of incorrect nucleotides because of the dATP pool imbalance was important for radiosensitization with dFdCyd, and, therefore, cells deficient in mismatch repair (MMR) would exhibit greater radiosensitization. We tested this hypothesis by evaluating the ability of HCT116 colon carcinoma cell lines, which differ in MMR proficiency, to be radiosensitized by dFdCyd. The MMR-proficient cell line (HCT116 + ch3) was more sensitive to dFdCyd alone than were the MMR-deficient cell lines (HCT116, HCT116 + ch2, and HCT116 p53(-/-)). Interestingly, the MMR-proficient cells could not be radiosensitized at concentrations of dFdCyd <or=IC(90), although extremely high concentrations of dFdCyd (>IC(96)) enhanced cell killing with radiation. In contrast, the MMR-deficient cells were radiosensitized at concentrations of dFdCyd <or=IC(50), with radiation enhancement ratios of approximately 1.5. Cell cycle analysis, using dual parameter flow cytometry, demonstrated that all of the cell lines accumulated in S phase after dFdCyd treatment, and, shortly after irradiation, a prominent but transient G(2)-M block was observed. In the MMR-deficient cells, the IC(50) for dFdCyd produced a >or=80% decrease in dATP within 4 h after drug addition, and this low dATP level was maintained for another 12-20 h. Although the IC(50) of dFdCyd was unable to sustain a >80% decrease in the dATP level in the MMR-proficient cells, the IC(90) did achieve this level of dATP depletion; however, it was unable to radiosensitize the MMR-proficient cells. Similar results were obtained with HCT116 cells, in which the MMR deficiency was corrected by transfection with a vector containing the hMLH1 cDNA. In addition, the deletion of p53 did not increase radiation enhancement ratios. These results demonstrate that MMR deficiency promotes radiosensitization with dFdCyd. We suggest that dATP depletion produces errors of replication in MMR-deficient cells, which, if left unrepaired, enhances cell death by ionizing radiation.