Mechanism-Based Inhibitors of Deoxythymidine Monophosphate Synthesis as Antineoplastic Agents

Transcription

1 Mechanism-Based Inhibitors of Deoxythymidine Monophosphate Synthesis as Antineoplastic Agents Victoria F. Roche 1 School of Pharmacy and Allied Health Professions, Creighton University, Omaha NE INTRODUCTION 5-Fluorouracil and methotrexate are antimetabolites widely used in the treatment of neoplastic disease. They act by inactivating the two enzymes involved in the biosynthesis of the pyrimidine nucleotide 2 -deoxythymidine 5 -monophosphate, namely thymidylate synthase and dihydrofolate reductase. The three dimensional structure of each enzyme is known and the mechanisms of the reactions they catalyze are well understood. This allows the medicinal chemistry professor the opportunity to engage students in a mechanism-based discussion of how the structure of each antimetabolite permits selective enzyme targeting and inhibition. This article will provide an overview of the structure and function of the thymidylate synthase and dihydrofolate reductase enzymes, take a mechanistic look at the structureactivity relationships of 5-fluorouracil, methotrexate and their clinically important analogs, and highlight some of the newer pyrimidine antagonist molecules recently reported in 1 Associate Professor of Pharmaceutical Sciences. the literature. A brief discussion of the role of antifolates in inhibiting purine nucleotide biosynthesis is provided. Among the more widely used antineoplastic agents in the anticancer armamentarium are the pyrimidine antimetabolites 5-fluorouracil (5-FU) and methotrexate. Distinctly different in structure, these two pyrimidine antagonists have different enzymatic sites of action, but both ultimately inhibit the rate-limiting enzyme in the dtmp synthesis pathway, thymidylate synthase. 5-FU is a close structural analog of the endogenous substrate deoxyuridylic acid, and inhibits this sulfhydryl enzyme directly. The antifolate methotrexate, through a direct inhibition of dihydrofolate reductase, induces a feedback inhibition of the synthase. The structures of both thymidylate synthase and dihydrofolate reductase enzymes have been well characterized, and the mechanisms of the reactions they catalyze are well understood. Thus a discussion of the action of these two drug structures provides a great opportunity to demonstrate to students the importance of chemistry to the rational (and therefore efficient) design of therapeutic agents, and to the true understanding of the mechanism of drug action. 196 American Journal of Pharmaceutical Education Vol. 58, Summer 1994

2 Fig. 1. Endogenous substrates and cofactors in pyrimidine and purine nucleotide biosynthesis. SYNTHESIS OF DEOXYTHYMIDINEMONOPHOSPHATE (dtmp) The targeting of the dtmp synthesis pathway in the treatment of selected neoplasms is a logical one, since cells deficient in this nucleotide cannot provide the 2'- deoxythymidine-5'-triphosphate (dttp) which, along with dctp, serves as a primary pyrimidine substrate for DNA polymerase(l). The substrate for dtmp synthesis is the 5- desmethyl analog 2'-deoxyuridine 5'-monophosphate (also known as deoxyuridylic acid or dump). This pyrimidine nucleotide is generated from orotidylic acid (also a pyrimidine nucleotide) through the action of several enzymes, including orotidylate decarboxylase (which provides UMP), UMP kinase (which provides UDP), ribonucleotide reductase (which provides dudp), and a phosphatase (which provides dump)(l,2). A second route to this key intermediate involves the hydrolysis and ribonucleotide reductase-' mediated reduction of CTP to form dcdp, which is then converted by the action of phosphatase and deaminase enzymes to dump. In yet another biosynthetic pathway, dutp, which forms through deamination of dctp, can hydrolyze via a pyrophosphatase to form dump(l). dump undergoes a methylation reaction catalyzed by thymidylate synthase to form dtmp, a process which is significantly accelerated in cancer and other rapidly proliferating cells. 5,10-Methylenetetrahydrofolate serves as the cofactor which will donate the one carbon unit to the substrate. The structures of the nucleotide substrate and folate cofactor are found in Figure 1. The complete reaction sequence ateo requires the enzyme dihydrofolate reductase. The reaction mechanism is depicted in Scheme 1. In brief, the synthase enzyme binds both dump substrate and folate cofactor, resulting in a reversible ternary complex. Properly positioned, the nucleophilic sulfhydryl group of thymidylate synthase is attracted to the electrophilic C 6 of the dump substrate. The sulfhydryl group Scheme 1. Synthesis of 2'deoxythymidine-5'-phosphate (dtmp). attacks in a Michael fashion to displace the electrons of the double bond, which subsequently attack the electrophilic methylene group of the cofactor. Students are expected to recognize the inductive and resonance effects which promote the electron deficient character of both of these electrophilic targets. This second attack, which breaks the bond between the methylene group and N 10 of the cofactor(3), results in the joining of enzyme, substrate and cofactor into a covalent ternary complex. This covalent complex undergoes oxidative decomposition to product (dtmp) only through the transfer of the dump C 5 hydrogen atom to a basic center of the cofactor. Once this abstraction occurs, the methylene unit is released to the substrate, the cofactor oxidizes elimination, and a resultant hydride shift from cofactor to the substrate's exocyclic methylene unit provides the 5-CH 3 group of the product dtmp. The original electron flow pattern reverses to regenerate the pyrimidine olefinic linkage and release free thymidylate synthase(l). Left behind to fend for itself is the oxidized cofactor, dihydrofolate. In order to participate in another dtmp synthesis cycle, this molecule must convert to tetrahydrofolate via the enzyme dihydrofolate reductase, and submit to methylene insertion by serine with the assistance of pyridoxal phosphate and serine hydroxymethylase(4). THE THYMIDYLATE SYNTHASE TARGET The rate limiting reaction in the dtmp synthesis sequence is the formation of the covalent enzyme-substrate-cofactor ternary complex catalyzed by thymidylate synthase. The American Journal of Pharmaceutical Education Vol. 58, Summer

3 structure of the enzyme isolated from Lactobacillus casei was elucidated in 1987 by Hardy, et al.(5) and has been subsequently well characterized by others. Thymidylate synthase is the most highly conserved enzyme known(6), and this fact supports the involvement of multiple residues in the catalysis reaction(7). Through crystallographic analysis investigators have shown that the active site is a large, open cavity which accommodates both substrate and cofac-tor. When the cof actor binds, conf ormational changes in the C- terminus cause this cavity to close down. This facilitates the orientation of reaction participants and the subsequent production of dtmp(8,9). Complementarity of the faces of the pteridine portion of the cof actor and the pyrimidine ring of the substrate ensures proper placement for ternary complex stability(10). The synthase enzyme is dimeric, and binds substrate and cofactor equally, but sequentially, to both active sites. Key active site residues which interact with substrate through hydrogen bonding include Arg23 (which binds with the 5 - phosphate), Tyr261 and His259 (which bind with the 3 - OH), Asn229 (which binds with the pyrimidine N 3 -H) and, of course, Cysl98 (the nucleophile which attacks C 6 of dump). A critical active site conformation-maintaining role has been proposed for the highly conserved Tyr33, as mutation to His33 in the mammalian enzyme results in a 3-4 fold increase in resistance to 5-FU(11). Ordered molecules of water in the active site cavity are important in promoting enzyme specificity. One water molecule, activated by His l99, forms a hydrogen bond with the C 4 oxygen atom of dump(12). As indicated above, the tetrahydrofolate cofactor also binds to thymidylate synthase, and does so most effectively in the natural 6R, polyglutamated form. Monoglutamated folate is selectively transported into cells where a glutamyl transferase enzyme (also known as folyl polyglutamyl synthase) adds several (most commonly 3-6) additional γ- glutamate units to the original residue. This not only dramatically enhances affinity for the synthase enzyme, but also augments the catalytic capability of the synthase(13). The folate binding loop of thymidylate synthase includes residues and contains several basic amino acids, among them Lys50 and His53. With the assistance of highly ordered water molecules, these residues anchor the first Glu residue of the polyglutamate side chain of the cofactor(7,13). The oxygen atoms of Leu224 and Ile310, as well as Asp221, are also involved in cofactor anchoring. Diffuse electrostatic interactions hold subsequent Glu residues to the synthase enzyme. In studies withsynthase isolated from E. coli, Kamb et al. found that the first glutamate residue of the 5,10- methylenetetrahydrofolate cofactor is anchored rigidly to the synthase, with each successive glutamate being more and more disordered. The reduction in K i contributed by each glutamate residue in the chain is highly species specific. Citing the fact that the region of the synthase enzyme which binds the polyglutamate chain of the cofactor is less highly conserved than the active dump-binding site, Kamb and coworkers hypothesize that more species-selective dtmp synthesis inhibitors might be generated by targeting the polyglutamate binding area as a drug receptor(13). While not a part of the active site or folate binding loop, the C-terminal Val316 is critical in the catalytic action of thymidylate synthase. Studies with L. casei synthase have shown that, when substrate and cofactor are bound, a significant conformational change moves the flexible C-terminus approximately 4 angstroms. This forms a tight cover or lid over the active site. In this conformation, the C-terminal Val316 interacts hydrophobically with Thr24, and through hydrogen bonds with Arg23 and Trp85 residues on the opposite side of the binding pocket(14). As previously mentioned, this closing down of the active site aligns all dtmp synthesis players for effective covalent ternary complex formation. Synthase mutants lacking Val316 retain the ability to bind substrate, albeit in a different conformation from the wild type enzyme(14), but lack the ability to catalyze dtmp synthesis. This lack of catalytic capability is presumed due to an approximate 40 fold decrease in folate affinity, and difficulty in forming the covalent ternary complex(9,15). 5-FLUOROURACIL AND RELATED ANALOGS: A CASE OF CHEMICAL ENTRAPMENT The prototypical substrate-based inhibitor of thymidylate synthase is 5-fluorouracil (Figure 1) Once students have a clear understanding of the mechanisms involved in dump binding and subsequent conversion to dtmp via the synthase enzyme, the elegant rationale of the simple replacement of fluorine for hydrogen in dump to produce this antimetabolite becomes clear. Originally conceived in 1957 by Heidelberger(16), the mechanism of dtmp synthesis inhibition by 5-FU represents chemically-based enzyme entrapment at its finest. No self-respecting enzyme that naturally binds substrate as nucleotide would be fooled by a drug structure missing the large phosphorylated deoxyribose unit. Thus, 5- FU must be activated by conversion to 5-fluoro-2 - deoxyuridylic acid (5-FdUMP). This occurs in vivo through a number of pathways(17,18). In one, a phosphorylated ribose unit is transferred to N 1 of 5-FU through the action of orotate phosphoribosyl transferase and 5-phosphoribosyl- 1-pyrophosphate to form 5-FUMP. UMP kinase converts this nucleotide to the diphosphorylated derivative, 5-FUDP, which is then vulnerable to reduction by ribonucleotide reductase, forming 5-FdUDP. Hydrolysis of 5-FdUDP by a phosphatase yields the desired 5-FdUMP. Alternatively, ribose-1-phosphate can react with 5-FU with enzymatic assistance from uridine phosphorylase to form the fluorinated nucleoside. Uridine kinase phosphorylation provides 5-FUMP, which can follow the pathway described above to yield 5-FdUMP. In yet another pathway, 5-FU is converted to 5-FdUMP through the action of thymidine phosphorylase and thymidine kinase. Once activated, the drug molecule is ready for the chemical sting, and the properties of the 5-fluoro group (the only functional group distinguishing natural from false substrate) will be solely responsible for setting the trap. As a small atom (about the size of the endogenous C 5 -hydrogen), fluorine provides no steric hindrance to binding in the active site cavity of the synthase. As the most electronegative atom known, its negative inductive effects dramatically increase the electrophilic character of the adjacent C 6, which is the target for attack by the nucleophilic synthase Cysl98. The result is a very rapid nucleophilic attack by the enzyme, with the resultant formation of a highly stable covalent ternary enzyme-substrate-cofactor complex. At this point the drug has directly and permanently inhibited the synthase enzyme, since lack of an abstractable C 5 -hydrogen means 198 American Journal of Pharmaceutical Education Vol. 58, Summer 1994

4 Fig. 2. Fluoropyrimidine inhibitors of thymidylate synthase. the covalent ternary complex will be unable to undergo oxidative breakdown. No breakdown, no product and, more importantly, no regeneration of thymidylate synthase. Consequently, there will be no conversion of endogenous molecules of dump to dtmp by the inhibited enzyme, and the cell must generate new thymidylate synthase de novo for DNA production. Until new enzyme is synthesized, DNA production ceases which, of course, is the therapeutic goal. Supporting the unique chemical role of the fluorine atom in halting dtmp synthesis is the almost total lack of synthase inhibiting action of other 5-halogenated dump analogs. Chlorine, bromine and iodine are all bulkier and less electronegative than fluorine. Therefore, they interfere significantly with substrate anchoring, and promote a less electrophilic C 6 than the fluorinated analog(1). Other fluoropyrimidine structures have demonstrated effective inhibition of the synthase enzyme (Figure 2). The deoxyribonucleoside version of 5-FU, (5-fluorodeoxyuridine or floxuridine), is commercially available. Since this molecule must be phosphorylated by uridine kinase to the same active false nucleotide substrate as is formed with 5-FU, it has a similar activity profile. The 5-trifluoromethyl version of floxuridine (trifluridine) was another brainchild of Heidelberger(16). After thymidine kinase-catalyzed phosphorylation to give the nucleotide, the electron withdrawing trifluoromethyl group promotes rapid nucleophilic attack at adjacent C 6 by the synthase Cys198. This results in a displacement of HF and generates a highly reactive difluoromethylene unit. This reactive intermediate then presumably interacts covalently with the cofactor to irreversibly inhibit thymidylate synthase(4). Tegafer, the 1- tetrahydrofuranyl analog of 5-FU is slowly cleaved, possibly through thymidine phosphorylase, to 5-FU(1). Another unique 5-FUprodrug is 5-fluoropyrimidin-4(1H)-one, which is oxidized in vivo to 5-FU by xanthine oxidase(1). Other nonpyrimidine based direct thymidylate synthase inhibitors are being reported in the literature (Figure 3). 4',5'-Dichloro, dibromo, and diiodo derivatives of pyridoxine (vitamin B 6 ) have demonstrated an irreversible inhibition of synthase. Pretreatment with 10µmol dump completely protects against enzyme inactivation, which indicates inhibitor interaction at the active catalytic site(19). Stoichet and coworkers, through the use of a molecular Fig. 3. Nonpyrimidine-based inhibitors of thymidylate synthase. docking computer program, identified unique phenolphthalein inhibitors which bind to an area of the synthase active site that is distinct from the area which binds substratebased inhibitor molecules. The tetraiodo analog provided a synthase IC 50 of 3µM(20).6,7-Imidazotetrahydroquinolines conceptualized by iterative ligand design to occupy the folate ligand binding pocket of thymidylate synthase have recently been synthesized(21). Webber et al. reported the synthesis of 5-(arylthio)quinazolin-4-ones which were designed by repetitive crystallographic analysis to bind to the folate binding area of the synthase(22). Other quinoline and quinazoline antifolates, ICID1694 for example, have demonstrated effectiveness as inhibitors of thymidylate synthase(23-25). The 2-CH 3 group of the quinazoline synthase inhibitors provides increased enzymatic specificity compared to 2-NH 2 precursor antagonist molecules. In addition, better solubility properties result in lower nephrotoxicity. The better inhibition of cell growth observed from the 2-CH 3 quinazolines is presumably due to more effective polyglutamation(26). THE DIHYDROFOLATE REDUCTASE TARGET The family of isoenzymes known as the dihydrofolate reductases (DHFRs) have been extensively characterized. DHFRs are relatively small proteins (approximately 20,000 Da), and key active site residues of enzymes isolated from mammalian and bacterial sources are highly conserved. The tetrahydrofolates they produce accept single carbon units in various forms, and are of considerable biological importance as one carbon donors to numerous endogenous substrates(27). DHFR requires NADPH as cofactor to reduce natural folates and antifolates. Catalytic reduction involves the formation of a ternary complex of enzyme, folate and American Journal of Pharmaceutical Education Vol.58, Summer

5 levels of folyl polyglutamyl synthase should respond better to methotrexate or other antifolates which do not require polyglutamation for high affinity binding to DHFR. The key residues of the active site of Echerichia coli derived DHFR which bind the natural substrate dihydrofolate (DHF) have been identified. Primary among them is Asp27 which donates a proton to N 5 of DHF and initiates the reduction reaction. Significantly, N 5 is the most basic nitrogen atom in the substrate. The ion-ion bond formed between Asp27 and N 5 is complemented by hydrogen bonds between Arg57 and the α-carboxyl group, and between Thr43 and the 2-NH 2 of DHF. As might be anticipated, stereochemistry of the glutamate side chain is important to activity, with the natural L isomer showing a 10 fold increase in potency over the unnatural D form(33). The carbons of the pteridine and phenyl rings bind hydrophobically to Ile5, Ala7, Phe31, and to Leu28 and Ile50 respectively. Both rings bind through hydrophobic bonds to Ile94(27,34). Fig. 4. Classical and nonclassical inhibitors of dihydrofolate reductase. cofactor. The order of binding of substrate and cofactor to DHFR is random, but the anchoring of one entity greatly facilitates the rapid binding of the other. Once bound, the C 4 of NADPH stereospecifically transfers hydride to the substrate, a process which is promoted by the protonation of the N 5 of the folate by Asp27(27,28). Anionic polyglutamate chains are attached enzymatically to natural folates and synthetic antifolates, which ensures their intracellular localization(13). While some folate polyglutamates demonstrate enhanced affinity for DHFR compared to monoglutamated analogs, the effect is generally much lower (less than 10 fold) than observed with other folate-binding enzymes. In classical antifolates like methotrexate, this diminished effect is believed due to the very strong, essentially stoichiometric, binding to the reductase enzyme provided by the amino groups at positions 2 and 4. Interestingly, in studies with recombinant human DHFR, Rosowsky et al. found that affinity for DHFR of 2-desamino- 2-methylaminopterin modified with three or four γ-glutamyl residues was two orders of magnitude higher than the parent antifolate. This observation caused these investigators to propose a domain in the distal region of the active site which might contain cationic Lys or Arg residues that could interact electrostatically with the polyglutamate anions of antifolates(29). Many investigators have noted that folyl polyglutamyl synthase activity is higher in certain tumor cells than in normal, but rapidly proliferating, cells(30-32). Thus the development of antifolates which require polyglutamation prior to anchoring to their target enzyme could permit more selective, and therefore safer, antifolate chemotherapy than is currently available(29). On the flip side, tumors with low METHOTREXATE AND OTHER CLASSICAL DHFR INHIBITORS Methotrexate (Figure 4) is termed a classical antifolate because it retains the p-aminobenzoylglutamic acid side chain of natural DHF. The structural distinctions between this antifolate and the natural folate include the replacement of the 4-oxo group of DHF with an amino group, the presence of a 7,8-double bond, and the addition of a methyl group at N 10. However these slight structural changes result in highly significant changes in reductase binding and activity. While the electron withdrawing 4-oxo group of DHF decreases lone pair availability at N 1 (recall N 5 is the most basic nitrogen of DHF), the 4-amino group of methotrexate can donate electrons to N 1 through the pi electron system, which increases its basic strength over 1000 fold. N 1 is the most basic nitrogen of methotrexate, and is sought out by Asp27 at the expense of N 5. This ion-ion anchoring between Asp27 and N 1 alters the pteridine ring orientation 180 compared to the dihydropteridine ring of DHF. The binding that occurs between the rest of the methotrexate structure and the DHFR active site is believed to parallel that observed with DHF(27). The importance of unhindered hydrogen bonds between the NH 2 groups at C 2 and C 4 is emphasized by the significant loss of DHFR inhibiting potency when these groups are alkylated(35). Methotrexate can undergo polyglutamation in vivo, but this structural modification produces only a modest enhancement of DHFR affinity at best. It is interesting to note that, while the polyglutamate analogs of methotrexate are capable of directly inhibiting thymidylate synthase at commonly achieved intracellular concentrations, the affinity for the synthase is two to four orders of magnitude below that observed for the primary target, DHFR(36). Cody et al. have recently determined the crystal structure of a recombinant human DHFR-NADPHmethotrexate-γ-tetrazole complex to 2.3 angstrom resolution(37). In their complex, N 1 is protonated by Glu30, and the methotrexate 2-amino group is hydrogen bonded to a water molecule which also interacts with Glu30 and Thr136. The 4-amino group binds with Ile7 and Tyr121, and the α- carboxylate of the glutamate side chain forms a bridge with Arg American Journal of Pharmaceutical Education Vol. 58, Summer 1994

6 While technically a competitive antagonist of DHF, methotrexate is so tightly bound as a result of the N 1 -enzyme interaction that it appears noncompetitive or pseudoirreversible (4). While K i s will differ with the DHFR source, values in the 4-7 picomolar range are common(27,38). These K i values reflect the affinity of methotrexate for DHFR in the presence of bound NADPH. The route from initially reversible to pseudoirreversible binding has been proposed to involve a slow isomerization of the antifolate, which could induce subtle changes in the conformation of active site amino acids and associated water molecules(39). Interestingly, whether single or multiple conformations exist when the DHFR enzyme is initially bound to substrate or cofactor (binary complexes) appears to be species dependent(40). The ternary complexes studied in E. coli maintain a single conformation. Armed with an understanding of the differences in substrate chemistry and DHFR binding induced by the structural differences in DHF and methotrexate, the mechanism of DHFR inhibition by this antifolate becomes clear. Since N 5 of methotrexate will not be protonated by DHFR residues, reduction of the 5,6-double bond to form tetrahydrofolate does not occur. The presence of the methotrexate 7,8-double bond, which provides for a totally aromatic pteridine ring, should also retard reduction attempts. The net result is that the antifolate structure anchors very tightly to DHFR through protonated N 1, and the C 2 and C 4 amino groups, but will not react with it, thus there is no release of reduced product nor regeneration of free enzyme. As a permanent occupant of the DHFR active site, methotrexate keeps the natural byproduct of dtmp synthesis, dihydrofolate, from binding to the reductase. Therefore, no reduced tetrahydrofolate can form which, in turn, means that no 5,10-methylenetetrahydrofolate cofactor can be synthesized. More importantly, the buildup of DHF in the cell signals the rate-limiting thymidylate synthase enzyme via a feedback mechanism to shut down, and dtmp synthesis (and therefore DNA synthesis) grinds to a halt. In terms of cancer chemotherapy, mission accomplished. The structures of some classical and nonclassical DHFR inhibitors are provided in Figure 4. The classical antifolate most closely related to methotrexate is the N 10 desmethyl analog aminopterin. This molecule, while effective as an antileukemic agent, is decidedly more toxic than methotrexate(27), and has not been developed for clinical use. However 10-ethyl-lO-deazaaminopterin (10-EDAM) has shown promise as a therapeutic agent in a non-small cell lung tumor trial(41,42). This molecule appears to have affinity for DHFR comparable to methotrexate (K i = 2-3 pm), but is less toxic due to favorable uptake and polyglutamation in tumor cells(26) coupled with rapid clearance from susceptible healthy tissues(27). Rosowsky(26) provides a review of the favorable results obtained with 10- EDAM in patients with advanced cancer in Phase I and II clinical trials. Some deaza-and dideazatetrahydrofolate analogs, among them the nonclassical antifolate trimetrexate (Figure 4), have been investigated for reductase inhibiting activity. Trimetrexate is a 5-deazafolate analog with a trimethoxy substituted phenyl ring. Because it lacks the p- aminobenzoylglutamic acid side chain it is not polyglutamated in vivo. Thus it may show enhanced efficiency in tumors resistant to methotrexate due to folyl polyglugamyl synthase deficiency, as long as drug transport mechanisms, DHFR binding affinity and thymidylate synthase levels are not impaired(43). While not currently marketed as an antineoplastic agent, this nonclassical antifolate is available for the treatment of Pneumocystis carinii infections. Mono-and diester derivatives of the methotrexate glutamate carboxylate groups have been evaluated for DHFR inhibiting activity. The rationale behind glutamate esterification is enhancement of cell membrane penetration. While generally not as potent as methotrexate, the esters do retain DHFR inhibiting action, with potency increasing directly with ester lipophilicity up to a maximum chain length of 4-5 carbons(44). The α, γ- dibutyl ester was less active than either monobutyl ester or methotrexate when assayed with rabbit liver DHFR. In the rabbit liver DHFR system, the two monobutyl esters were of equal potency, but when the DHFR was obtained from murine leukemia (L1210) cells, a 10 fold potency increase of the γ-butyl ester over the a-butyl derivative was noted. The potency of the γ-butyl ester was comparable with methotrexate in this system(45,46). Capitalizing on the pseudoirreversible methotrexate- DHFR interaction in a unique way, Hawkins et al. have designed a novel system for delivering radionuclides to tumor cells(47). In this in vitro study, a nontoxic, bivalent alpha-transferrin monoclonal antibody targeted to human K562 erythroleukemia cells was conjugated to recombinant human DHFR and administered. Then a methotrexate analog labelled with m In was administered, and highly specific delivery of nuclide to the tumor target was demonstrated. PURINE SYNTHESIS INHIBITION BY ANTIFOLATES: THE GAR FORMYLTRANSFERASE TARGET While peripheral to a discussion of mechanism-based pyrimidine antagonists, it is important to note that antifolates such as methotrexate have the ability to inhibit purine, as well as pyrimidine, nucleotide biosynthesis. The two folaterequiring enzymes in the purine synthesis pathway are glycinamide ribonucleotide formyltransferase (also known as GAR transformylase, GART and GARFT) and aminoimidazolecarboxamide ribonucleotide formyltransferase (AICARFT). The single carbon-donating folate cofactor required for both formylation reactions is 10- formyltetrahydrofolate. This folate is produced from the thymidylate synthase cofactor 5,10-methylenetetrahydrofolate through oxidation. GARFT catalyzes the transfer of the 10-formyl group of 10-formyltetrahydrofolate to the substrate glycinamide ribonucleotide to generate the purine nucleotide intermediate American Journal of Pharmaceutical Education Vol. 58, Summer

7 formylglycinamide ribonucleotide and tetrahydrofolate. AICARFT catalyzes a later formylation of an aminoimidazolecarboxamide ribonucleotide to produce 5- formylaminoimidazole-4-carboxamide ribonucleotide, the immediate precursor to inosine. The structures of these important substrates and the formylated folate cofactor are provided in Figure 1. In studies with mammalian GARFT, Caperelli demonstrated a substrate: cofactor stoichiometry of 1:1(48). While methotrexate itself has a relatively low affinity for these purine synthesis enzymes, polyglutamate analogs of methotrexate anchor much more strongly (29,49,50). Antineoplastic agents which more selectively antagonize purine biosynthesis through GARFT inhibition can be found in 5,10-dideaza analogs of tetrahydrofolate(51-53). Racemic 5,10-dideazatetrahydrofolic acid, also known as DDATHF or lometrexol, inhibits GARFT with a K i of 39nM. A 100 fold decrease in K i is noted in the derivative containing 5 γ-glutamate residues(26). While the two enantiomers have similar enzymatic activity profiles and IC 50 values, the 6R isomer was selected for clinical evaluation(26). Alkylation at C 5 or C 10 in the dideazatetrahydrofolate analogs often decreases affinity for GARFT, but may enhance transport into tumor cells(53,54). Preliminary data from lometrexol Phase I clinical trials have been promising. However the side effects of myelosuppression and anemia proved problematic(26,55-57). Investigators found that the cumulative toxicity induced by lometrexol could be significantly attenuated or reversed with the concomitant administration of folic acid or leucovorin (5-formyltetrahydrofolic acid). A discussion of other new antifolate structures currently in preclinical development is provided by Rosowsky(26). In summary, the topic of nucleotide antagonism in the treatment of neoplastic disease provides the medicinal chemistry instructor with an excellent opportunity to discuss rational, mechanistic approaches to drug design and action. The number of antifolate structures which have been designed, synthesized and evaluated for selective inhibition of key enzymes in the pyrimidine and purine biosynthetic pathways continues to burgeon(58), and many new therapeutically useful compounds are on the clinical horizon. And, while fewer in number, the fluoropyrimidine-based inhibitors of dtmp synthesis dramatically reinforce that an understanding of the chemical properties and behavior of endogenous macromolecules and drug structures is the key to interpreting and predicting drug action. Am. J. Pharm. Educ., 58, (1994); received 2/1/94, accepted 4/13/94. References (1) Hobbs, J.B., Purine and pyrimidine targets in Comprehensive Medicinal Chemistry, The Rational Design, Mechanistic Study and Therapeutic Application of Chemical Compounds, Volume 2. (edit. Sammes, P.G.) Pergamon Pres s, New York NY (1990) pp (2) Stryer, L., Biosynthesis of nucleotides in Biochemistry, Third Edition, Freeman and Company, New York NY (1988) pp (3) Pellino, A.M. and Danenberg, P.V., Evidence from chemical degradation studies for a covalent bond from 5-fluoro-2 -deoxyuridylate to N-5 of tetrahydrofolate in the ternary complex of thymidylate synthase-5- fluoro-2 -deoxyuridylate-5,10-methylenetetrahydrofolate, J. Biol. Chem., 260, (1985). (4) Remers, W.A., Antineoplastic agents in Wilson and Gisvolds Textbook of Organic Medicinal and Pharmaceutical Chemistry, Ninth Edition, (edits. Delgado, J.N and Remers, W.A.) J.B. Lippincott Company, Philadelphia (1991) pp (5)Hardy, L.W., Finer-Moore, J.S., Montfort, W.R., Jones, M.O., Santi, D.V. and Stroud, R.M., Atomic structure of thymidylate synthase: Target for rational drug design, Science, 235, (1987). (6) Perry, K.M., Fauman, E.B., Finer-Moore, J.S., Montfort, W.R., Maley, G.F., Maley, F. and Stroud, R.M. Plastic adaptation toward mutations in proteins: Structural comparison of thymidylate synthase. Proteins, 8, (1990). (7) Finer-Moore, J.S., Fauman, E.B., Foster, P.G., Perry, K.M., Santi, D.V. and Stroud, R.M., Refined structures of substrate-bound and phosphate-bound thymidylate synthase from Lactobadllus casei, J. Mol. Biol., 232, (1993). (8) Stroud, R.M. and Finer-Moore, J.S., Stereochemistry of a multistep/ bipartite methyltransfer reaction: Thymidylate synthase, FASEB. 7, (1993). (9) Carreras, C.W., Climie, S.C. and Santi, D.V., Thymidylate synthase with a C-terminal deletion catalyzes partial reactions, but is unable to catalyze thymidylate formation, Biochemistry, 31, (1992). (10) Finer-Moore, J.S., Montfort, W.R. and Stroud, R., Pairwise specificity and sequential binding in enzyme catalysis: Thymidylate synthase, ibid., 29, (1990). (11) Hughey, C.T., Barbour, K.W., Berger, F.G. and Berger, S.H., Functional effects of a naturally-occuring amino acid substitution in human thymidylate synthase, Mol. Pharmacol., 44, (1993). (12) Liu, L. and Santi, D.V., Exclusion of 2 -deoxycytidine 5 -monophosphate by asparagine 229 of thymidylate synthase, Biochemistry, 32, (1993). (13) Kamb, A., Finer-Moore, J.S., Calvert, A.H. and Stroud, R.M., Structural basis for recognition of polyglutamyl folates by thymidylate synthase, ibid., 31, (1992). (14) Perry, K.M., Carreras, C.W., Chang, L.C., Santi, D.V. and Stroud, R.M., Structures of thymidylate synthase with a C-terminal deletion: Role of the C-terminus in alignment of 2 -deoxyuridine 5 -monophosphateand5,10-methylenetetrahydrofolate, ibid., 32, (1993). (15) Cisneros, R.J., Zapf, J.W. and Dunlap, R.B., Studies of 5- fluorodeoxyuridine 5 -monophosphate binding to carboxypeptidase A-inactivated thymidylate synthase from Lactobadllus casei, J. Biochem., 268, (1993). (16) Heidelberger, C, Fluorinated pyrimidines and their nucleosides, Handbook Exper. Pharmacol., 38,193(1975). (17) Chabner, B.A., Pyrimidine antagonists, in Pharmacologic Principles of Cancer Treatment. W.B. Saunders Company, Philadelphia (1982) p (18) Calabresi, P. and Chabner, B.A., Chemotherapy of neoplastic diseases in Goodman and Gilman s The Pharmacological Basis of Therapeutics, Eighth Edition (edits., Gilman, A.G., Rail, T.W., Nies, A.S, Taylor, P.) McGraw-Hill, Inc., New York NY (1993) pp (19) Huang, S., Parish, E.J. and Aull, J.L., Irreversible inhibition of thymidylate synthase by pyridoxine (B 6 ) analogs. J. Enzym. Inhib., 6, (1992). (20) Schoichet, B.K., Stroud, R.M., Santi, D.V., Kuntz, I.D. and Perry, K.M., Structure-based discovery of inhibitors of thymidylate synthase, Science, 259, (1993). (21) Reich, S.H. et al., Design and synthesis of novel 6,7- imidazoltetrahydroquinoline inhibitors tf thymidylate synthase using iterative protein crystal structure analysis, J. Med. Chem., 35, (1992). (22) Webber, S.E. et al., Design of thymidylate synthase inhibitors using protein crystal structures: The synthesis and biological evaluation of a novel class of 5-substituted quinazolinones, ibid., 36, (1993). (23) Warner, P., et al., Quinoline antifolate thymidylate synthase inhibitors: Variation of the C2-and C4-substitutents, ibid., 35, (1992). (24) Thornton, TJ. et al., Quinazoline antifolate thymidylate synthase inhibitors, ibid., 35, (1992). (25) Bisset, G.M.F., Pawelczak, K., Jackman, A.L. and Calvert, A.H., Synthesis and thymidylate synthase inhibitory activity of the poly-γglutamyl conjugates of N-[5-[N-(3,4,-dihydro-2-methyl-4- oxoquinazolin-6-ylmethyl)-n-methylamino]2-thenoyi]-l-glutamic acid (ICI D1694) and Other Quinazoline antifolates, ibid., 35, (1992). (26) Rosowsky, A., Development of antifolate analogs as anticancer agents, Am. J. Pharm. Educ., 56, (1992). (27) McCormack,J.J., Reductases, in Comprehensive Medicinal Chemistry, The Rational Design, Mechanistic Study and Therapeutic Application of Chemical Compounds, Volume 2, (edit. Sammes, P.G.) Pergamon Press, New York NY (1990) pp (28) Coward, J.K., Parameswaran, K.N., Cashmore, A.R. and Bertino, J.R., 7,8-Dihydropteroyl oligo-gamma-l-glutamates: Synthesis and 202 American Journal of Pharmaceutical Education Vol. 58, Summer 1994

8 kinetic studies with purified dihydrofolate reductase from mammalian sources, Biochemistry, 13, (1974). (29) Rosowsky, A., et al., Biochemical and biological studies on 2- desamino-2-methylaminopterin, an antifolate the polyglutamates of which are more potent than the monoglutatmate against three key enzymes of folate metabolism, Cancer Res., 52, (1992). (30) Poser, R.G., Sirotnak, F.M. and Chello, P.L., Differential synthesis of methotrexate polyglutamates in normal proliferative and neoplastic tissues in vivo, ibid., 41, (1981). (31) Fabre, I., Fabre, G. and Goldman, I.D., Polyglutamation, an important element in methotrexate cytotoxicity and selectivity in tumor versus murine granulocyte progenitor cells in vitro, ibid., 44, (1984). (32) Runberger, B.G.,Barrueco,J.R. and Sirotnak, F.M., Differing specificities for 4-aminofolate analogues of folyl polyglutamyl synthetase from tumors and proliferative intestinal epithelium of the mouse with significance for selective antitumor action, ibid., 50, (1990). (33) Cramer, S.M., Schornagel, J.H., Kalghatgi, K.K., Bertino, J.R. and Horvath, C., Occurnace and significance of D-methotrexate as a contaminant ofcommerical methotrexate, ibid., 44, (1984). (34) Roth, B., Selective inhibitors of bacterial dihydrofolate reductase: Structure-activitity relationships, Handbook of Exp. Pharmacol.,64, (1983). (35) Baker, B.R., Design of Active Site Directed Irreversivle Enzyme Inhibitors, John Wiley & Sons, New York NY (1967) p (36) Rhee, M.S., Coward, J.K. and Galivan, J., Depletion of 5,10- methylenetetrahydrofolate and 10-formyltetrahydrofolate by methotrexate in cultured hepatoma cells, Mol. Pharmacol., 42, (1992). (37) Cody, V., Luft, J.R., Ciszak, E., Kalman, T.I. and Freisheim, J.H., Crystal structure determination at 2.3 angstrom of recombinant human dihydrofolate reductase ternary complex with NADPH and methotrexate-γ-tetrazole, Anticancer Drug Des., 7, (1992). (38) Jackson, R.C., Hart, L.I. and Harrap. K.R., Intrinsic resistance to methotrexate of cultured mammalian cells in relation to the inhibition kinetics of their dihydrofolate reductases, Cancer Res., 36, (1976). (39) Blakley, R.L. and Cocco, L., Role of isomerization of initial complexes in the binding of inhibitors to dihydrofolate reductase, Biochemistry, 24, (1985). (40) Cheung, H.T.A., Birdsall, B. and Feeney, J., 13 C-NMR studies of complexes of Escherichia coli dihydrofolate reductase formed with methotrexate and with folic acid, FEBS, 312, (1992). (41) DeGraw, J.I., Brown, V.H., Tagawa, FL, Kisliuk, R.L., Gaumont, Y. and Sirotnak, F.M., Synthesis and antitumor activity of 10-alkyl-10- deazaminopterins. A convenient synthesis of 10-deazaminopterin, J. Med. Chem., 25, (1982). (42) Shum, K.Y., Kris, M.G., Gralla, R.J., Burke, M.T., Marks, L.D. and Heelan, R.T., Phase II study of 10-ethyl-10-deazaaminopterin in patients with Stage III and IV non-small-cell lung cancer, J. Clin. Oncol., 6, (1988). (43) Koizumi, S. and Allegra, C.J., Enzyme studies of methotrexateresistant human leukemic cell (K562) subclones. Leuk. Res., 16, (1992). (44) Johns, D.G., Farquhar, D.. Chabner, B.A., Wolpert, M.K. and Adamson, R.H., Antineoplastic activity of lipid-soluble dialkyl esters of methotrexate, Experientia, 29, (1973). (45) Rosowsky, A., Beardsley, G.P., Ensminger, W.D., Lazarus, H. and Yu, C-S., Methotrexate analogues. 11. Unambiguous chemical synthesis and in vitro biological evaluation of α-and γ-monoesters as potential prodrugs, J. Med. Chem., 21, (1978). (46) Rosowsky, A. et al., Methotrexate analogues. 25. Chemical and biological studies on the γ-t-butyl esters of methotrexate and aminopterin, ibid., 28, (1985). (47) Hawkins, G.A. et al., Delivery of radionuclides to pretargeted monoclonal antibodies using dihydrofolate reductase and methotrexate, Cancer Res., 53(10 Suppl), (1993). (48) Caperelli, C.A., Mammalian glycinamide ribonucleotide transformylase. Kinetic mechanism, and associated de novo purine biosynthetic activities, J. Biol. Chem., 264, (1989). (49) Baram, J., Chabner, B.A., Drake, J.C., Fitzhugh, A.L., Sholar and P.W., Allegra, C.J., Identification and biochemical properties of 10- formyldihydrofolate, a novel folate found in methotrexate-treated cells, ibid., 263, (1988). (50) Allegra, C.J., Drake, J.C., Jolivet, J. and Chabner, B.A., Inhibition of phosphoribosyl-aminoimidazolecarboxamide transformylase by methotrexate and dihydrofolic acid polyglutamates, Proc. Natl. Acad., Sci, USA, 82, (1985). (51) Chen, P., et al., Crystal structure of glycinamide ribonucleotide transformylase from Escherichia coli at 3.0 angstrom resolution, J. Mol. Biol. 227, (1992). (52) Beardsley, G.P., Moroson, B.A., Taylor, E.C. and Moran, R.G., A new folate antimetabolite, 5,10-dideaza-5,6,7,8-tetrahydrofolate is a potent inhibitor of de novo purine synthesis, J. Biol. Chem., 264, (1989). (53) Piper, J.R. et al., 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, J. Med. Chem., 31, (1988). (54) DeGraw, J.I., Christie, P.H., Kisliuk, R.L., Gaumont, Y. and Sirotnak, F.M., Synthesis and antifolate properties of 10-alkyl-5,10-dideaza analogues of methotrexate and tetrahydrofolic acid, ibid., 33, (1990). (55) Muggia, F. et al., Phase I clinical trial of weekly 5,10- dideazatetrahydrofolate (LY , DDATHF-B), Proc. Am. Assoc. Clin. Oncol., 9, 74(1990). (56) Young, C.W. et al., Improved clinical tolerance of lometrexol with oral folic acid, Proc. Am. Assoc. Cancer Res., 33, 406(1992). (57) Cole, J.T., Gralla, R.J., Kardinal, C.G. and Rivera, N.P., Lometrexol (DD AHF): Phase I trial of a weekly schedule of this new antifolate, ibid., 33, 413(1992). (58) Rosowsky, A., Chemical and biological activity of antifolates, Prog. Med. Chem., 26, 1-252(1989). American Journal of Pharmaceutical Education Vol. 58, Summer