Synthesis and antifolate activity of 5-methyl-5,10-dideaza analogues of aminopterin and folic acid and an alternative synthesis of 5,10-dideazatetrahydrofolic acid, a potent inhibitor of glycinamide ribonucleotide formyltransferase.

The title compounds were prepared in extensions of a general synthetic approach used earlier to prepare 5-alkyl-5-deaza analogues of classical antifolates. Wittig condensation of 2,4-diaminopyrido[2,3-d]pyrimidine-6-carboxaldehyde (2a) and its 5-methyl analogue 2b with [4-(methoxycarbonyl)benzylidene] triphenylphosphorane gave 9,10-ethenyl precursors 3a and 3b. Hydrogenation (DMF, ambient, 5% Pd/C) of the 9,10-ethenyl group of 3b followed by ester hydrolysis led to 4-[2-(2,4-diamino-5-methylpyrido[2,3-d]pyrimidin-6-yl)ethyl]ben zoi c acid (5), which was converted to 5-methyl-5,10-dideazaaminopterin (6) via coupling with dimethyl L-glutamate (mixed-anhydride method using i-BuOCOCl) followed by ester hydrolysis. Standard hydrolytic deamination of 6 gave 5-methyl-5,10-dideazafolic acid (7). Intermediates 3a and 3b were converted through concomitant deamination and ester hydrolysis to 8a and 8b. Peptide coupling of 8a,b (using (EtO)2POCN) with diesters of L-glutamic acid gave intermediate esters 9a and 9b. Hydrogenation of both the 9,10 double bond and the pyrido ring of 9a and 9b (MeOH-0.1 N HCl, 3.5 atm, Pt) was followed by ester hydrolysis to give 5,10-dideaza-5,6,7,8-tetrahydrofolic acid (11a) and the 5-methyl analogue 11b. Biological evaluation of 6, 7, 11a, and 11b for inhibition of dihydrofolate reductase (DHFR) isolated from L1210 cells and for growth inhibition and transport characteristics toward L1210 cells revealed 6 to be less potent than methotrexate in the inhibition of DHFR and cell growth. Compounds 6, 11a, and 11b were transported into cells more efficiently than methotrexate. Growth inhibition IC50 values for 11a and 11b were 57 and 490 nM, respectively; the value for 11a is in good agreement with that previously reported (20-50 nM). In tests against other folate-utilizing enzymes, 11a and 11b were found to be inhibitors of glycinamide ribonucleotide formyltransferase (GAR formyltransferase) from one bacterial (Lactobacillus casei) and two mammalian (Manca and L1210) sources with 11a being decidedly more inhibitory than 11b. Neither 11a nor 11b inhibited aminoimidazolecarboxamide ribonucleotide formyltransferase. These results support reported evidence that 11a owes its observed antitumor activity to interference with the purine de novo pathway with the site of action being GAR formyltransferase.