The purpose of this study was to determine the conditions for optimum synergistic efficacy of the two-drug combination of trimetrexate and 5-fluorouracil. Synergistic cell killing of Chinese hamster ovary cells in these clonogenic survival assays was observed only when the cells had been exposed to trimetrexate (25 microM) for 2 to 4 h prior to 5-fluorouracil exposure (either 125 or 250 microM). The schedule dependence of the observed synergy in vitro was closely linked to trimetrexate-induced changes in cellular 5-phosphoribosyl 1-pyrophosphate (PRPP) pools. Exposure to 25 microM trimetrexate induced increases in PRPP pools to 398% and 761% of control values at 2 and 4 h, respectively. Methotrexate (20 microM) also increased Chinese hamster ovary cell PRPP content in a time-dependent fashion to values of 280 and 511% of control after 2 and 4 h of drug exposure. Previous in vivo studies demonstrated a modest degree of therapeutic synergy between trimetrexate and 5-fluorouracil against P388 leukemia. Our in vitro results suggested that the degree of synergy seen in vivo could be increased with appropriate schedule changes. Mice were implanted i.p. with 10(6) P388 leukemia cells on Day 0 and were treated with trimetrexate (every 3 h for eight injections; Days 1, 5, and 9) and 5-fluorouracil (Days 1, 5, and 9) as single agents or in combination on one of two schedules; 5-fluorouracil was administered with either the first or the last of the eight trimetrexate doses on Days 1, 5, and 9. Both treatment regimens demonstrated therapeutic synergy but, as predicted from the in vitro data, the "5-fluorouracil last" was superior to the "5-fluorouracil first" sequence. Treatment with the optimal doses on the "5-fluorouracil last" sequence (trimetrexate, 31; 5-fluorouracil, 33 mg/kg/injection) produced an increased life span of 183% and a net reduction in tumor cell burden of 6.7 logs compared with a 111% increased life span (net reduction in tumor burden of 2.6 logs) produced by the most active of the single agents, 5-fluorouracil. Thus the efficacy of the combination of trimetrexate with 5-fluorouracil was sequence and time dependent both in vitro and in vivo. The synergy, observed in vitro and probably in vivo, was linked to a trimetrexate-induced elevation of intracellular PRPP, thus facilitating the production of 5-fluoropyrimidine nucleotides. These data are similar to the sequence and schedule dependency of the methotrexate/5-fluorouracil combination with important differences.(ABSTRACT TRUNCATED AT 400 WORDS)