Search

More than 50 years after the introduction of the dihydrofolate reductase inhibitor, methotrexate, new antifolates have emerged and have been incorporated into the chemotherapeutic armamentarium. These include pralatrexate, with the same target as methotrexate, but with enhanced properties, and pemetrexed, with different enzyme targets and properties. Current synthetic efforts are focused on developing antifolates that are selectively delivered to cancer cells, but not to normal proliferating cells, exploiting the different properties of folate transporters. In another novel approach, drugs structurally and mechanistically unrelated to folates are linked to and use folic acid as a carrier to be endocytosed by folate receptors and then released to inhibit their cellular targets. This review describes the evolution and current status of antifolate pharmacology and prospects for the development of the next generation of folate analogs.

Purpose: Lometrexol [(6R)-5,10-dideaza-5,6,7,8-tetrahydrofolate] is the prototype folate antimetabolite that targets the de novo purine synthesis pathway. Early phase I trials were confounded by cumulative myelosuppression that prevented repetitive administration. Subsequent preclinical and clinical studies suggested that coadministration of folic acid might favorably modulate lometrexol toxicity without eliminating potential antitumor activity. We set out to determine if concurrent folic acid would allow administration of lometrexol on a weekly schedule, and, if so, to identify an appropriate dose combination for phase II trials. Pharmacokinetic and metabolism studies were undertaken in an attempt to improve our understanding of lometrexol pharmacodynamics., Methods: Patients with advanced cancer received daily oral folic acid beginning 7 days before lometrexol and continuing for 7 days beyond the last lometrexol dose. Lometrexol was administered by short i.v. infusion weekly for 8 weeks. Scheduled lometrexol doses were omitted for toxicity of more than grade 2 present on the day of treatment, and dose-limiting toxicity was prospectively defined in terms of frequency of dose omission as well as the occurrence of severe toxic events. Plasma and whole blood total lometrexol contents (lometrexol plus lometrexol polyglutamates) were measured in samples taken just prior to each lometrexol dose., Results: A total of 18 patients were treated at five lometrexol dose levels. The maximum tolerated dose was identified by frequent dose omission due to thrombocytopenia and mucositis. The recommended phase II dose combination is lometrexol 10.4 mg/m(2) per week i.v. with folic acid 3 mg/m(2) per day orally. One patient with melanoma experienced a partial response, and three patients, two with melanoma and one with renal cell carcinoma, experienced stable disease. Lometrexol was not detectable in any predose plasma sample tested. The total red blood cell content of lometrexol increased over several weeks and then appeared to plateau., Conclusions: Weekly administration of lometrexol is feasible and well-tolerated when coadministered with daily oral folic acid. The nature of the interaction between natural folates and lometrexol that renders this schedule feasible remains unclear. A definition of dose-limiting toxicity that incorporated attention to dose omissions allowed efficient identification of a recommended phase II dose that reflects the maximum feasible dose intensity for a weekly schedule. Lometrexol is a promising, anticancer agent.

Folylpoly-gamma-glutamate synthetase (FPGS) catalyzes the addition of several equivalents of glutamic acid to the gamma-carboxyl group in the side chain of folate cofactors and analogs. Folylpoly-gamma-glutamate synthetase has three functions in folate homeostasis in mammals: polyglutamation prevents efflux of folate cofactors from the cell, it increases the binding of folate cofactors to some of the enzymes of folate interconversion and biosynthesis, and it appears to allow the accumulation of folates in the mitochondria that are required for glycine synthesis. The efficient substrate activity of the newer generations of tetrahydrofolate analogs results in levels of intracellular accumulation of cytotoxic drug in any cell expressing FPGS in which the enzyme activity is not suppressed by feedback, and the binding of folate inhibitors of thymidylate synthase and glycinamide ribonucleotide formyltransferase is substantially increased by polyglutamation. Resistance to these drugs appears to be most frequently due to mutations that change the level of polyglutamation of parent compound, a clear indication of the centrality of the process to the cytotoxicity of these drugs. Folylpoly-gamma-glutamate synthetase is widely expressed in human tumors and is tightly linked either to proliferation or to a lack of differentiation. The cytotoxicity of both thymidylate synthase and purine inhibitors requires continued inhibition of target for greater than one generation time, so that the integrative function of FPGS adds considerably to the efficiency of folate antimetabolites.

The binding of the prototypical folate inhibitor of de novo purine synthesis, 5,10-dideazatetrahydrofolate (DDATHF), and its hexaglutamate to recombinant trifunctional mouse glycinamide ribonucleotide formyltransferase (rmGARFT) was studied by equilibrium dialysis and by steady-state kinetics using sensitive assays that allowed initial rate calculations. rmGARFT was expressed in insect cells infected with a recombinant baculovirus and purified by a two-step procedure that allowed production of about 25 mg of pure protein/L of culture. The binding of DDATHF to GARFT was approximately 50-fold tighter than previously reported, with Kd and Ki values of 2-9 nM, making the parent form of this antifolate a tight-binding inhibitor. The binding of the hexaglutamate of DDATHF to rmGARFT had Kd and Ki values of 0.1-0.3 nM, consistent with the view that polyglutamation enhances binding of antifolates to GARFT. Kinetic analyses using either mono- or hexaglutamate substrate did not yield different values for the Ki for the hexaglutamate form of DDATHF, in contradiction with previous reports. Both the folate substrate commonly used to study GARFT, 10-formyl-5,8-dideazafolate, and its hexaglutamate were found to have very low Km values, namely, 75 and 7.4 nM, respectively, and the folate reaction products for these substrates were equally potent inhibitors, results which modify the interpretation of previous kinetic experiments. The product analog DDATHF and beta-glycinamide ribonucleotide bound to enzyme equally well in the presence and absence of the other, an observation at variance with the concept that GARFT obeys an ordered sequential binding of the substrates. We conclude that the kinetics of mouse GARFT are most consistent with a random order of substrate binding, that both the inhibitor DDATHF and the folate substrate are tight-binding ligands, and that polyglutamate forms enhance the affinity of both substrate and inhibitor by an order of magnitude.

ZD1694 (Tomudex; TDX) is a quinazoline antifolate that, when polyglutamated, is a potent inhibitor of thymidylate synthase (TS), the enzyme that converts dUMP to dTMP. Continuous exposure of MCF-7 breast and NCI H630 colon cells to TDX, with stepwise increases in TDX up to 2.0 microM, resulted in stably resistant cell lines (MCFTDX and H630TDX) that were highly resistant to TDX. Initial studies revealed 34-fold increase in TS protein levels in MCFTDX and a 52-fold increase in TS levels in H630TDX cell lines. Despite continued exposure of these cells to 2.0 microM TDX, TS protein and TS mRNA expression decreased to parental levels in H630TDX cells, whereas in MCFTDX cells TS mRNA expression and TS protein levels remained elevated. Southern blot analysis revealed a 20-fold TS gene amplification in the MCFTDX cell line. TDX uptake was 2-fold higher in resistant MCFTDX cells than in parental MCF-7 cells, whereas in H630TDX cells TDX uptake was 50-fold less than that observed in parental H630 cells. In contrast, no change in the transport of either leucovorin or methotrexate into H630TDX cells was noted when compared with the H630 parental cells. In H630TDX cells, folylpolyglutamate synthetase (FPGS) activity was 48-fold less compared to parent H630 cells; however, FPGS mRNA expression was similar in both lines. H630TDX cells were also highly resistant to ZD9331, a novel quinazoline TS inhibitor that does not require polyglutamation, suggesting that defective transport by the reduced folate carrier was also an important mechanism of resistance in these cells. In MCFTDX and H630TDX resistant cells, several mechanisms of resistance are apparent: one increased TS expression; the others evolved over time from increased TS expression to decreased FPGS levels and decreased TDX transport.

Recent clinical trials with lometrexol [(6R)-5,10-dideazatetrahydrofolate] have revealed a level of toxicity in humans that was not predicted on the basis of previous in vivo preclinical studies. Because standard laboratory animal diets contain high levels of folic acid relative to human folate intake, the toxicity and therapeutic activity of lometrexol was studied in mice under conditions of restricted dietary folate intake. Remarkably, the lethality of this drug increased by three orders of magnitude in mildly folate-deficient mice, mimicking the unexpected toxicity seen in humans. Lometrexol had limited therapeutic activity in folate-deficient mice bearing the C3H mammary adenocarcinoma, compared with the substantial therapeutic index for treatment of this tumor in animals on standard diet. When folic acid was administered p.o. to mice that were mildly folate deficient, antitumor activity was again observed at nontoxic doses of lometrexol, and the range of lometrexol doses that allowed safe therapeutic use of this drug increased at higher dietary folate intake. At a fixed dose of lometrexol, the antitumor effects in animals were dependent on the level of dietary folate and went through a distinct optimum. Excessively high folate intake reversed the antitumor effects of lometrexol. Optimization of the folic acid content in the diet and of the lometrexol dosage are predicted to have substantial impact on the clinical activity of this class of drugs.

We have studied the cytotoxicity of 5,10-dideazatetrahydrofolate (DDATHF) and of D-1694 to human WiDr colonic carcinoma cells as a model system for the effects of pure inhibitors of either the de novo purine synthesis pathway or thymidylate synthesis. The growth of this cell line was inhibited by very low concentrations of either agent and the lethality of DDATHF and D-1694 was completely prevented by continuous exposure to either hypoxanthine or thymidine, respectively, indicating that these compounds were very potent metabolic inhibitors, each specific for one of these pathways. D-1694 was highly cytotoxic (> 3 logs of kill) after a 4-h exposure to 1 microM drug, or a 24-h exposure to very low concentrations (0.04 microM). On the other hand, the cytotoxicity of DDATHF was substantially lower, with 2 logs of cell kill requiring >> 100 microM with 4 h of exposure or approximately 40 microM for 72 h of exposure. Maximal cell kill induced by D-1694 was 5-6 logs, consistent with elimination of all viable cells except preexisting mutants. A maximum of 2-3 logs of cell kill was observed with DDATHF. Exposure of WiDr cells to either D-1694 or DDATHF caused striking cellular changes, but the morphologies of cells treated with the two drugs were remarkably different. D-1694-treated cells detached from the dish within 1-2 days after a megaloblastosis, whereas DDATHF-treated cells remained adherent to the dishes for at least 10 days after treatment. The addition of thymidine to D-1694-treated cultures or hypoxanthine to DDATHF-treated cells after up to 20 h of drug exposure completely prevented cytotoxicity of either drug. With longer exposures, cytotoxicity of both drugs progressively increased in spite of such rescue. Our results indicate that substantial (99-99.9%) tumor cell kill can be induced by a pure inhibitor of purine synthesis, but that the rate of commitment to cell death and the extent of cell kill is greater with a pure inhibitor of thymidylate synthesis.

Background: Cancer chemotherapy with folate antimetabolites has been traditionally targeted at the enzyme dihydrofolate reductase and is based on the requirement of dividing tumor cells for a supply of thymidylate and purines. However, a new compound, 5,10-dideazatetrahydrofolate (DDATHF, whose 6R diastereomer is also known as Lometrexol), has become available that prevents tumor cell growth by inhibiting the first of the folate-dependent enzymes involved in de novo purine synthesis, glycinamide ribonucleotide formyltransferase., Purpose: We investigated the toxicity and therapeutic activity of DDATHF in a phase I clinical trial., Methods: DDATHF was given at one of the following dose levels to 33 patients (16 females and 17 males) with malignant solid tumors: 3.0 mg/m2 per week (level A) to 10 patients, 4.5 mg/m2 per week (level B) to 13 patients, or 6.0 mg/m2 per week (level C) to 10 patients. Each drug cycle consisted of three weekly injections of DDATHF followed by a 2-week rest prior to redosing in the next cycle., Results: Of 33 patients, 27 received at least one full cycle of DDATHF. Thrombocytopenia was the major dose-limiting toxicity, and it was severe in one of 10 patients during the first cycle and in two of four patients during the second cycle. Because of cumulative toxicity at 6.0 mg/m2, second or later cycles were abbreviated to two weekly doses. Stomatitis was generally mild, but it was dose-limiting in one patient. Neutropenia was infrequent and mild, and normocytic anemia requiring blood transfusion was common with repeat dosing. Leucovorin was given for grade 2 or greater thrombocytopenia and resulted in hematologic recovery within 1 week in all eight patients so treated. Without leucovorin, the thrombocytopenia lasted from 7 to 49 days in three patients. A partial response was noted in one patient with non-small-cell lung cancer and a minor response in one patient with breast cancer. Three patients with colorectal cancer achieved stable disease for greater than 3 months with improvement in carcinoembryonic antigen levels in one patient., Conclusions: DDATHF has an unusual pattern of toxicity with repetitive dosing, and humans with advanced cancer are considerably more sensitive than would be predicted from previous animal studies. Although doses of 6.0 mg/m2 per week on our schedule have been determined to be safe, repeated cycles require careful monitoring because of cumulative toxic effects., Implications: Additional phase I studies of DDATHF that relate toxicity to folate intake and tissue folate pools appear warranted.

We have investigated the structural features of 5,10-dideaza-5,6,7,8-tetrahydrofolate (DDATHF) that determine the activity of this compound as an inhibitor of glycinamide ribonucleotide formyltransferase (GARFT) purified from mouse L1210 cells. 5-Deazatetrahydrofolate was as good an inhibitor of GARFT as DDATHF, indicating that isosteric replacement of nitrogen by carbon at the 5-position of tetrahydrofolate is sufficient for inhibition of GARFT. 5,10-Dideazafolic acid, 5,8,10-trideazatetrahydrofolate, and 2-desamino-5,10-dideazatetrahydrofolate were poor inhibitors of GARFT, indicating that a reduced pyridopyrimidine ring, N-8, and the 2-amino group of DDATHF, respectively, play an important role in the binding of tetrahydrofolate analogues to this enzyme. DDATHF analogues in which the phenyl ring was replaced either by a cyclohexyl ring or by methylene groups retained activity as inhibitors. 5,10-Dideazatetrahydrohomofolate was about 6 times more potent as an inhibitor of GARFT than DDATHF, but 5,10-dideazatetrahydronorfolate had about one-fifth of the activity of DDATHF. An analogue of DDATHF in which the glutamic acid side chain was replaced by aspartic acid (which was not a substrate for polyglutamation and was only weakly cytotoxic) was equiactive with DDATHF as an inhibitor of purified GARFT. Surprisingly, 5,10-dideazatetrahydropteroic acid was about as active as DDATHF as an inhibitor of GARFT, an indication that the glutamic acid in the side chain of DDATHF does not play a role in this ligand-enzyme interaction. The polyglutamate derivatives of DDATHF bound up to 100 times tighter to GARFT than DDATHF itself; longer chain polyglutamates conformed to Goldstein's zone B behavior under experimental conditions and were projected to be in zone C, i.e., stoichiometric inhibition, in vivo. We conclude that the presence of carbon at the 5-position of tetrahydrofolate analogues is sufficient for inhibition of GARFT, that N-8 and the 2-amino group are involved in binding of DDATHF to GARFT, probably through hydrogen bonds, and that the structures of the phenyl ring and amino acid side chain of DDATHF analogues are not primary determinants of GARFT inhibition by monoglutamate forms of these compounds. We also conclude that polyglutamation plays a major role in the potent cytotoxicity of DDATHF.

The previously undescribed 2-desamino and 2-desamino-2-methyl analogues of aminopterin (AMT) and methotrexate (MTX) were synthesized from 2-amino-5-(chloromethyl)pyrazine-3-carbonitrile. The AMT analogues were obtained via a three-step sequence consisting of condensation with di-tert-butyl N-(4-aminobenzoyl)-L-glutamate, heating with formamidine or acetamidine acetate, and mild acidolysis with trifluoroacetic acid. The MTX analogues were prepared similarly, except that 2-amino-5-(chloromethyl)pyrazine-3-carbonitrile was condensed with 4-(N-methylamino)benzoic acid and the resulting product was annulated with formamidine or acetamidine acetate to obtain the 2-desamino and 2-desamino-2-methyl analogues, respectively, of 4-amino-4-deoxy-N10-methylpteroic acid. Condensation with di-tert-butyl L-glutamate in the presence of diethyl phosphorocyanidate followed by ester cleavage with trifluoroacetic acid was then carried out. Retention of the L configuration in the glutamate moiety during this synthesis was demonstrated by rapid and essentially complete hydrolysis with carboxypeptidase G1 under conditions that likewise cleaved the L enantiomer of MTX but left the D enantiomer unaffected. The 2-desamino and 2-desamino-2-methyl analogues of AMT and MTX inhibited the growth of tumor cells, but were very poor inhibitors of dihydrofolate reductase (DHFR). These unexpected results suggested that activity in intact cells was due to metabolism of the 2-desamino compounds to polyglutamates.

Heterocyclic para-aminobenzoate modifications of 2-desamino-2-methyl-5,8-dideazafolic acid and a series of its N10-substituted analogs have produced a number of interesting compounds that have enabled a deeper understanding of the biochemical events required for activity in this class of antimetabolite. There is a relationship that has become apparent between compound potency and both uptake via the reduced-folate carrier and FPGS substrate activity. Rapid cellular uptake and metabolism of polyglutamate forms that are approximately 100-fold more potent as inhibitors of TS can translate a modest TS inhibitor such as ICI D1694 into a very potent inhibitor of cell growth (approximately 500- and approximately 10-fold more potent than CB3717 or ICI 198583, respectively). Polyglutamation may therefore act as an almost essential activation step and ICI D1694 may be highly specific for tumors expressing both the reduced-folate carrier and FPGS. Polyglutamation of folate analogs also leads to drug retention which may play a major role in the pharmacodynamics of TS inhibition by ICI D1694 in vivo. Current studies with 3H-ICI D1694 are aimed at demonstrating metabolism to polyglutamates in tumor cells. The serious toxic limitations of CB3717, i.e., liver and kidney toxicities, are not seen with ICI D1694 reflecting the good water solubility of the drug compared with CB3717. The toxicities observed in mice are however to hematological tissues and are due to its TS inhibitory effects. Thus ICI D1694 may elicit toxicities in man more typical of an antimetabolite than of CB3717. The clinical evaluation of ICI D1694 may further our understanding of the role that metabolism to polyglutamates may have in therapeutic activity.

We have investigated some unusual aspects of the inhibition of mammalian thymidylate synthase (TS) by the folate antimetabolite, 10-propargyl-5,8-dideaza-folic acid (CB 3717). From our results, we conclude that binding of CB 3717 metabolites to one subunit of L1210 TS modified the conformation of the second active site of this enzyme so that it retained the ability to bind 5-fluro-2'-deoxyuridine-5'-monophosphate (FdUMP) but not its catalytic activity. Exposure of intact mouse L1210 cells to CB 3717 resulted in inactivation of cellular TS activity, yet desalted cytosol preparations from these cells retained the ability to bind FdUMP. The same effect was found with several analogs of CB 3717. Complexes of FdUMP formed in vitro with TS from cells exposed to CB 3717 were covalent and co-migrated on sodium dodecyl sulfate-polyacrylamide gel electrophoresis with complexes of FdUMP, folate cofactor, and TS from cells not exposed to CB 3717. In the presence of dUMP, a tightly bound complex rapidly formed between isolated pure TS and the pentaglutamate of CB 3717 but not the monoglutamate form of this compound. Binding experiments using CB 3717 pentaglutamate-inhibited TS suggested a stoichiometry of 1 mol of FdUMP bound per mol of dimeric TS.

A series of "stretched" methotrexate (MTX) analogues containing up to five 4-aminobutyryl (Gab) spacers between the 4-amino-4-deoxy-N10-methylpteroyl (MeAPA) moiety and the glutamate (Glu) side chain was prepared. Interest in these compounds stemmed from their relationship to MTX gamma-polyglutamates, from which they differ only in lacking "internal" alpha-carboxyl groups. The ability of the MeAPA-Gabn-Glu derivatives to inhibit dihydrofolate reductase (DHFR) and thymidylate synthase (TS) in vitro and to inhibit the growth of tumor cells in culture was evaluated. The IC50 for DHFR inhibition increased progressively from 0.082 to 0.84 microM as the number of Gab spacers was varied from one to five. At the same time the introduction of Gab spacers was found to produce substantial TS inhibition (Ki 0.1-0.4 microM) similar to that reported for MTX polyglutamates. Despite the activity of the MeAPA-Gabn-Glu derivatives as combined inhibitors of TS and DHFR, there was a steep loss of cell growth inhibitory potency as the number of Gab spacers was increased. This most likely reflects low cell uptake and the fact that when n greater than 1 there is almost total abolition of substrate activity for folylpolyglutamate synthetase, which had previously been observed with n = 1.

Replacement of the glutamic acid (Glu) moiety in methotrexate (MTX) and aminopterin (AMT) by 2-amino-4-phosphonobutyric acid (APBA) and ornithine (Orn) has been found to give analogs that retain the ability to inhibit dihydrofolate reductase (DHFR) while also displaying high activity against folylpolyglutamate synthetase (FPGS). One of these compounds, the Orn analog of AMT, is the most potent FPGS inhibitor we have found to date. A model to account for the fact that side-chain analogs containing a basic and those containing an acidic terminal group can both competitively inhibit FPGS is proposed. According to this model, binding may involve interaction of an acidic terminal group on the inhibitor with a positively charged active-site residue to which the gamma-carboxyl of the folate-antifolate substrate normally binds. It may also involve the interaction of a basic terminal group on the inhibitor with a different active-site residue which is negatively rather than positively charged and to which the alpha-amino group of the incoming Glu cosubstrate must bind before an amide bond to the gamma-carboxyl of the folate-antifolate can form. The 2 oppositely charged active-site residues assumed to take part in this binding are probably situated near each other and at approximately the same distance from the pteridine-binding site. The ability of compounds to inhibit both DHFR and FPGS makes it possible in principle for such compounds to kill cells via a "self-potentiation" mechanism in which inhibition of tetrahydrofolate synthesis is complemented by interference with the subsequent conversion of tetrahydrofolates to their polyglutamate conjugates. Possible exploitation of this mechanism to overcome MTX resistance is considered.

gamma-Phosphonate analogues of methotrexate (MTX) and aminopterin (AMT) were synthesized from 4-amino-4-deoxy-N10-methylpteroic acid and 4-amino-4-deoxy-N10-formylpteroic acid, respectively, by reaction with methyl D,L-2-amino-4-phosphonobutyrate followed by gentle alkaline hydrolysis. The products were compared with the corresponding D,L-homocysteic acid derivatives as inhibitors of dihydrofolate reductase and folylpolyglutamate synthetase, and as inhibitors of cell growth in culture. The gamma-phosphonates were somewhat less active than either the gamma-sulfonates or the parent drugs as inhibitors of murine dihydrofolate reductase. The MTX gamma-sulfonate and gamma-phosphonate analogues were equally inhibitory toward mouse liver folylpolyglutamate synthetase (Ki = 190 microM), but in the AMT series the gamma-phosphonate (Ki = 8.4 microM) was more potent than the gamma-sulfonate (Ki = 45 microM). The AMT analogues were consistently more inhibitory than the MTX analogues against cultured L1210 murine leukemia cells, but neither the gamma-phosphonates nor the gamma-sulfonates were as potent as their respective parent drugs. The gamma-phosphonate analogue of MTX was three times more potent than MTX against the MTX-resistant mutant line L1210/R81, but the AMT gamma-phosphonate was less potent than AMT; however, these differences were small in comparison with the level of resistance to all these compounds in the L1210/R81 line. The results suggest that N10-methyl and N10-unsubstituted compounds altered at the gamma-position do not necessarily follow identical structure-activity patterns in every test system.

The diasteromers of 5,10-dideaza-5,6,7,8-tetrahydrofolate (DDATHF) differing in chirality about carbon 6 were resolved and studied as inhibitors of folate-dependent processes in mouse leukemia cells. Both diastereomers of DDATHF were found to be potent inhibitors of leukemia cell growth due to effects on de novo purine synthesis. Cell growth inhibition by these compounds was prevented by 5-formyltetrahydrofolate in a dose-dependent manner. This indicated that the effects of the DDATHF diastereomers were due to inhibition of folate-dependent processes. Metabolite reversal experiments indicated that 5'-phosphoribosylglycinamide formyltransferase was the major site of action of these compounds in mouse cells. Another site in de novo purine synthesis was affected at higher concentrations of diastereomer B in L1210 cells. Low concentrations of both diastereomers were found to inhibit pure L1210 5'-phosphoribosylglycinamide formyltransferase competitively with the folate substrate. The two diastereomers were also efficient substrates for mouse liver folylpolyglutamate synthetase. We conclude that the 6R- and 6S-diastereomers of DDATHF are remarkably similar and equiactive antimetabolites inhibitory to de novo purine synthesis and that the biochemical processes involved in their cytotoxicity display little stereochemical specificity.

Analogues of the antitumor antifolate methotrexate (MTX) were synthesized in which the glutamate (Glu) moiety was replaced by ornithine (Orn), 2,4-diaminobutyric acid (Dab), or 2,3-diaminopropionic acid (Dap). An aminopterin (AMT) analogue with Orn in place of Glu was also synthesized. The MTX analogues were obtained by reaction of 4-amino-4-deoxy-N10-methylpteroic acid (mAPA) and N omega-Boc-alpha,omega-diaminoalkanoic acids in the presence of diethyl phosphorocyanidate, followed by deprotection with trifluoroacetic acid (TFA) or by reaction of p-nitrophenyl-mAPA and N omega-Boc-alpha,omega-diaminoalkanoic acids and subsequent treatment with TFA. The AMT analogue (APA-Orn) was synthesized by reaction of p-nitrophenyl 4-amino-4-deoxy-N10-formylpteroate with silylated N delta-Boc-L-ornithine in DMF at 55 degrees C for 3 days (45% yield), saponification (83%), and TFA cleavage (89%). APA-Orn was a potent inhibitor of both dihydrofolate reductase (DHFR) from L1210 mouse leukemia (IC50 = 0.072 microM) and partly purified folylpolyglutamate synthetase (FPGS) from mouse liver (Ki = 0.15 +/- 0.06 microM). The MTX analogue (mAPA-Orn) was likewise active against both enzymes, with an IC50 of 0.160 microM for DHFR and a Ki of 20.4 +/- 7.7 microM for FPGS inhibition. The other MTX analogues and the previously reported lysine derivative (mAPA-Lys) showed DHFR affinity similar to that of mAPA-Orn but lacked activity as FPGS inhibitors. The positively charged amino group appears to be detrimental to cellular uptake, as evidenced by the low cytotoxicity of these compounds (IC50 = 0.40-2.4 microM) in comparison with MTX and AMT (IC50 = 0.002 microM) against wild-type L1210 cells. On the other hand, mAPA-Orn and APA-Orn were both more potent than the corresponding Glu derivatives MTX and AMT against L1210/R81 cells, suggesting that in these MTX-resistant cells there may occur a "self-potentiation" process involving enhanced antifolate activity via interference with the polyglutamylation of reduced folates. APA-Orn is the most potent dual inhibitor of DHFR and FPGS discovered to date, but its effectiveness as a therapeutic agent may require some form of prodrug modification to neutralize the terminal amino group of the side chain.

The activity of a series of folic acid analogues as substrates for partially purified mouse liver folylpolyglutamate synthetase was determined and the effects of substituents on the binding to, and catalytic processes of, this enzyme were inferred. A 4-amino group improved substrate activity primarily by decreasing the apparent Km while N10-methyl substitution substantially diminished utilization as a substrate, again, by effects on Km. Isosteric replacement of N-10 altered substrate activity. A free alpha-carboxyl group in the amino acid side chain was required for catalysis as was the presence of the side chain amide carbonyl group. Modification of the amino acid side chain length profoundly affected activity. Several observations were made that may be relevant to chemotherapy with folate antimetabolites: 1) 7-hydroxymethotrexate was a substrate for this enzyme; 2) substrate activity and substrate inhibition were observed with CB 3717, a potent inhibitor of thymidylate synthase; 3) potent classical dihydrofolate reductase inhibitors were identified that were either not substrates for mouse liver folylpolyglutamate synthetase (e.g., 4-amino-4-deoxy-N10-methylpteroyl-L-alpha-aminoadipate) or were much better substrates than methotrexate for this enzyme (e.g., aminopterin); and 4) leucovorin and methotrexate appeared to be substrates for the same synthetase, but leucovorin saturated the reaction at much lower concentrations. These results have implications for the design of folylpolyglutamate synthetase inhibitors and for the selection of dihydrofolate reductase inhibitors that are either not polyglutamated or are efficiently polyglutamated in vivo.

The activity of mouse-liver folylpolyglutamate synthetase (FPGS) was compared using a number of folates and folate analogs in order to determine which structure modifications were compatible with enzyme catalysis and with efficient binding to enzyme. Most structural alterations in the amino acid side chain eliminated activity as a substrate for this enzyme, whereas modifications of any of several positions in the pteridine ring were tolerated with retention of FPGS substrate activity. Folate analogs with the lowest apparent Michaelis constants (Km,app) had a) a 4-amino group, b) a 5,6,7,8 reduced ring system, c) a quinazolate ring, and/or d) an unsubstituted 10-position. There was some correlation between FPGS substrate activity and the potency of folate antimetabolites as cytotoxic compounds but not necessarily as compounds selectively cytotoxic to tumor cells.

A new analogue of methotrexate was synthesized from 4-amino-4-deoxy-N10-methylpteroic acid and D,L-homocysteic acid. The product (mAPA-HCysA) was bound tightly to L1210 mouse leukemia dihydrofolate reductase (IC50 = 1 nM), inhibited L1210 cell proliferation in culture (IC50 = 0.3 microM), and prolonged the survival of L1210 leukemic mice (98% increase in lifespan at 120 mg/kg, qdx9). Studies on the interaction of mAPA-HCysA with partially purified mouse liver folyl polyglutamate synthetase revealed that mAPA-HCysA was not a substrate. Hence, the increased dose of mAPA-HCysA required to inhibit tumor growth in vitro and in vivo relative to methotrexate may reflect, in part, the inability of this compound to form non-effluxing polyglutamates. Folyl polyglutamate synthetase was competitively inhibited by mAPA-HCysA (K1 = 190 +/- 70 microM) when folate was the variable substrate. Thus, mAPA-HCysA is the first known compound to inhibit both mammalian dihydrofolate reductase and mammalian folyl polyglutamate synthetase.

Methotrexate (MTX) and aminopterin (AMT) analogues containing L-homocysteic acid or L-cysteic acid in place of L-glutamic acid were synthesized and tested as inhibitors of dihydrofolate reductase from L1210 cells and folyl polyglutamate synthetase from mouse liver. The ID50 against dihydrofolate reductase was comparable for the MTX and AMT analogues (0.04-0.07 microM), whereas the ID50 against folyl polyglutamate synthetase was 3- to 4-fold lower for the AMT analogues (40-60 microM) than for the MTX analogues (100-200 microM). Thus, N10-substitution has a greater effect on binding to folyl polyglutamate synthetase than dihydrofolate reductase. The cytotoxicity of these compounds was assayed in vitro against L1210 cells, and the AMT analogues again proved more potent (ID50 = 0.03-0.05 microM) than the MTX analogues (ID50 = 0.1-0.4 microM). A similarly increased potency was observed for the AMT analogues against L1210 leukemia in vivo. Though differential cell uptake cannot be ruled out as the basis of increased potency, it is possible that part of the activity of the AMT analogues involves interference with the intracellular polyglutamation of reduced folate cofactors, i.e., that they are "self-potentiating antifolates". Of the four compounds reported, the most active was N-(4-amino-4- deoxypteroyl )-L-homocysteic acid, which produced a 138% increase in life span (ILS) in L1210 leukemic mice when given on a modified bid X 10 schedule at a dose of 2 mg/kg. A comparable ILS was obtained with AMT itself at 0.24 mg/kg. Thus, replacement of gamma-CO2H by gamma-SO3H in the side chain does not decrease therapeutic effect. However, a higher dose is required, presumably to offset pharmacological differences reflecting the inability of the sulfonate group to be polyglutamated .

(6R,6S)-5,8,10-Trideaza-5,6,7,8-tetrahydropteroic acid was synthesized in several steps from 4,4-(ethylenedioxy)-cyclohexanone and [4-(tert-butyloxycarbonyl)benzyl]triphenylphosphonium bromide and was elaborated to (6R,6S)-5,8,10-trideaza-5,6,7,8-tetrahydropteroyl-L-glutamic acid and (6R,6S)-5,8,10-trideaza-5,6,7,8-tetrahydropteroyl-L-ornithin e. Compound 1 was found to be a good substrate for partially purified mouse liver folypolyglutamate synthetase (FPGS), with a Michaelis constant (Km = 15 microM) comparable to that reported for the reduced folate substrate (6S)-5,6,7,8-tetrahydropteroyl-L-glutamic acid and for (6R,6S)-5,10-dideaza-5,6,7,8-tetrahydropteroyl-L-glutamic acid (DDATHF). However, in striking contrast to DDATHF, which is potently cytotoxic, 1 failed to inhibit tumor cell growth in culture at concentrations of up to 100 microM. These results suggested that the NH at position 8 of DDATHF is important for cytotoxic activity but not for polyglutamylation. Just as 1 was a good substrate for FPGS, the ornithine analogue 2 proved to be among the more potent competitive inhibitors of this enzyme discovered to date, with a Ki,s of 10 microM. While the binding affinity of 2 was lower than that reported for 5,6,7,8-tetrahydropteroyl-L-ornithine (H4PteOrn), very substantial FPGS inhibition was observed even though N5,N8, and N10 in H4PteOrn were replaced by carbon. Binding to FPGS thus appears to be tolerant of bioisosteric replacements made simultaneously in ring B and the bridge region. Neither 1 nor 2 was active in preventing cell growth in culture at concentrations of up 100 microM. The N delta-hemiphthaloyl derivative of 2, synthesized as a potential prodrug, was also inactive.

Analogues of methotrexate (MTX) and aminopterin (AMT) with aminophosphonoalkanoic, aminoalkanesulfonic, and aminoalkanephosphonic acid side chains in place of glutamate were synthesized and tested as inhibitors of folylpolyglutamate synthetase (FPGS) from mouse liver. The aminophosphonoalkanoic acid analogues were also tested as inhibitors of dihydrofolate reductase (DHFR) from L1210 murine leukemia cells and as inhibitors of the growth of MTX-sensitive (L1210) and MTX-resistant (L1210/R81) cells in culture. The optimal number of CH2 groups in aminophosphonoalkanoic acid analogues of AMT was found to be two for both enzyme inhibition and cell growth inhibition but was especially critical for activity against FPGS. Deletion of the alpha-carboxyl also led to diminished anti-FPGS activity in comparison with previously studied homocysteic acid and 2-amino-4-phosphonobutyric acid analogues. In the aminoalkanesulfonic acid analogues of MTX without an alpha-carboxyl, anti-FPGS activity was low and showed minimal variation as the number of CH2 groups between the carboxamide and sulfonate moieties was changed from one to four. In similar aminoalkanephosphonic acid analogues of MTX, anti-FPGS activity was also low, was comparable for two and three CH2 groups between the carboxamide and phosphonate moieties, and was diminished by monoesterification of the phosphonate group. These effects demonstrate that the alpha-carboxyl group of folate analogues is involved in binding to the active site of FPGS, and that an alpha-carboxyl group should be retained as part of the structure of FPGS inhibitors.