The role of multidrug resistance efflux transporters in antifolate resistance and folate homeostasis

Abstract

Members of the ATP-binding cassette (ABC) transporters including P-glycoprotein (Pgp/ABCB1), multidrug resistance proteins (MRPs/ABCC) as well as breast cancer resistance protein (BCRP/ABCG2) function as ATP-dependent drug efflux transporters, which form a unique defense network against multiple chemotherapeutic drugs and cellular toxins. Among antitumor agents is the important group of folic acid antimetabolites known as antifolates. Antifolates such as methotrexate (MTX), pemetrexed and raltitrexed exert their cytotoxic activity via potent inhibition of folate-dependent enzymes essential for purine and pyrimidine nucleotide biosynthesis and thereby block DNA replication. Overexpression of MRPs and BCRP confers resistance upon malignant cells to various hydrophilic and lipophilic antifolates. Apart from their central role in mediating resistance to antifolates and other anticancer drugs, MRPs and BCRP have been recently shown to transport naturally occurring reduced folates. This was inferred from various complementary systems as follows: (a) Cell-free systems including ATP-dependent uptake of radiolabeled folate/MTX into purified inside-out membrane vesicles from stable transfectants and/or cells overexpressing these transporters, (b) Decreased accumulation of radiolabeled folate/MTX in cultured tumor cells overexpressing these transporters, as well as (c) In vivo rodent models such as Eisi hyperbillirubinemic rats (EHBR) that hereditarily lack MRP2 in their canalicular membrane and thereby display a bile that is highly deficient in various reduced folate cofactors and MTX, when compared with wild type Sprague–Dawley (SD) rats. In all cases, these folate/antifolate transporters functioned as high capacity, low affinity ATP-driven exporters. While the mechanism of cellular retention of (anti)folates is mediated via (anti)folylpolyglutamylation, certain efflux transporters including MRP5 (ABCC5) and BCRP were shown to transport both mono-, di- as well as triglutamate derivatives of MTX and folic acid. Furthermore, overexpression of MRPs and BCRP has been shown to result in decreased cellular folate pools, whereas loss of ABC transporter expression brought about a significant expansion in the intracellular reduced folate pool. The latter finding has important implications to antifolate-based chemotherapy as an augmented cellular folate pool results in a significant level of resistance to certain antifolates. Hence, the aims of the present review are: (a) To summarize and discuss the cumulative evidence supporting a functional role for various multidrug resistance efflux transporters of the ABC superfamily which mediate resistance to hydrophilic and lipophilic antifolates, (b) To describe and evaluate the recent data suggesting a role for these efflux transporters in regulation of cellular folate homeostasis under folate replete and deplete conditions. Furthermore, novel developments and future perspectives regarding the identification of novel antifolate target proteins and mechanisms of action, as well as rationally designed emerging drug combinations containing antifolates along with receptor tyrosine kinase inhibitors are being discussed.

Introduction

The efflux of folate cofactors (Fig. 1) and folic acid antagonists known as antifolates (Fig. 2) in cultured malignant cells has been long recognized (Hakala, 1965, Goldman, 1969, Dembo et al., 1984, Assaraf and Schimke, 1987). In detailed kinetic studies initiated more than 2 decades ago by Henderson and Zevely (1984) it was shown that mammalian cells including murine L1210 and human CCRF-CEM leukemia cells as well as erythrocytes contain several ATP-driven multispecific organic anion transporters (MOAT) and/or glutathione (GSH)-conjugate (i.e. GS-X) pumps (Henderson et al., 1986, Saxena and Henderson, 1996a, Saxena and Henderson, 1996b). It was found that these efflux transporters are able to transport (i.e. efflux) various organic compounds including the antifolate methotrexate (MTX). This initial dissection, which discerned between the multiple MTX efflux routes has been achieved through the utilization of an assortment of transport inhibitors (Henderson et al., 1986, Sirotnak and O’Leary, 1991). Shortly thereafter, by comparing wild-type Sprague–Dawley (SD) rats and mutant Eisai hyperbillirubinemic rats (EHBRs), whose canalicular multispecific organic anion transporter (cMOAT/MRP2/ABCC2) is absent from the canalicular membrane as a consequence of heredity, it was shown that MTX is transported into the bile (Masuda et al., 1997). Using canalicular membrane vesicles (CMV) from wild type SD rats an ATP-dependent transport of L-MTX with a Km of ∼0.3 mM was identified. Moreover, L-MTX competitively inhibited the ATP-driven transport of a classical substrate of cMOAT, [³H]2,4-dinitrophenyl-S-glutathione (DNP-SG), with an L-MTX inhibition constant that was comparable with its own Km. Consistently, increased uptake of MTX was directly demonstrated in membrane vesicles isolated from GLC4/ADR cells, which overexpress MRP1 (ABCC1) (Heijn et al., 1997). Subsequent key studies by Hooijberg et al. (1999) and Kool et al. (1999) have conclusively demonstrated that MRP1, MRP2 (ABCC2) and MRP3 (ABCC3) are able to transport and thereby mediate a high level of resistance to various hydrophilic antifolates including MTX.

The molecular identification of the multiple efflux pathways initially studied by Henderson and colleagues has not been addressed thereby resulting in a growing confusion as to the specific identity of these efflux transporters (Henderson et al., 1986, Saxena and Henderson, 1996a, Saxena and Henderson, 1996b). However, this field of ATP-driven drug efflux transporters (Danø, 1973) recently received a renewed interest after several studies reported on the molecular cloning of various genes of the MRP family (Cole et al., 1992, Dean and Allikmets, 2001; reviewed in Borst and Oude Elferink, 2002, Kruh and Belinsky, 2003, Deeley et al., 2006).

The demonstration that MRP1, MRP2 and MRP3 are involved in MTX efflux raised the intriguing possibility that these efflux transporters may also have the facility to transport naturally occurring tetrahydrofolate (THF) cofactors (Fig. 1) and thereby may potentially play a physiological role in regulation of cellular folate homeostasis (Assaraf and Goldman, 1997). Indeed, early studies have shown the loss of function of an ATP-driven folic acid/MTX efflux pump in Chinese hamster ovary (CHO) cells that are highly resistant to pyrimethamine, a lipophilic antifolate inhibitor of dihydrofolate reductase (DHFR) (Fig. 3; Assaraf and Goldman, 1997, Jansen et al., 1999). Importantly, the loss of this ATP-dependent folate efflux function resulted in a 3-fold expansion in the cellular THF pool, thereby allowing these cells to bypass and overcome the folate-depleting effects exerted by DHFR inhibitors; indeed, these cells displayed a 1000-fold resistance to the lipophilic antifolate pyrimethamine (Assaraf and Slotky, 1993). Consistent with these results was the key finding that naturally occurring THF coenzymes including tetrahydrofolate (THF), 5-methyl-tetrahydrofolate (5-CH₃-THF), 10-formyl-tetrahydrofolate (10-CHO-THF) and 5,10-methylenetetrahydrofolate (5-CH₂-THF) are endogenous transport substrates for MRP2 in vitro and are secreted into the bile in wild type SD rats in vivo (Kusuhara et al., 1998). In this study, the ATP-dependent transport of the primary circulatory reduced folate 5-CH₃-THF showed a Km of 121 μM in CMV isolated from wild type SD rats. Remarkably, in the model of EHBR, rats that are hereditarily deficient in MRP2 function, the transport of these reduced folate derivatives into the bile was severely impaired or completely absent.

Hence, the primary goals of this review are to summarize and discuss the existing evidence for the mediation of antifolate resistance by MRPs and BCRP. Moreover, the cumulative data supporting a possible role for the regulation of intracellular folate pool and folate homeostasis by MRPs and BCRP are also presented and evaluated.

Unlike bacteria and plants, mammalian cells lack the ability to synthesize their own reduced folate derivatives (Fig. 1) and therefore must obtain them from exogenous sources (i.e. from the diet). Reduced folate coenzymes are essential B-complex vitamins. The most common and well-known vitamin in this group is folic acid, an oxidized stable folate form. Folic acid and reduced folates are comprised of a pteridine ring structure, a p-aminobenzoic acid (PABA) and a glutamate residue. Reduced folate cofactors act as donors and acceptors of one-carbon units in a series of interconnected metabolic pathways involving de novo biosynthesis of purines and thymidylate, amino acids including methionine, serine and glycine, catabolism of histidine and formic acid, as well as methyl group metabolism including CpG island DNA methylation (Fig. 3; Stokstad, 1990). Reduced folate derivatives including 10-CHO-THF contribute one-carbon units in two key de novo biosynthetic transformylase reactions. The first one involves the generation of the imidazole ring of purines through the enzyme glycinamide ribonucleotide transformylase (GARTF), whereas the second reaction mediated by 5-aminoimidazole-4-carboxamide ribonucleotide transformylase (AICARTF) results in the formation of the purine intermediate inosinic acid (Fig. 3). The latter also known as inosine monophosphate (IMP) serves as a regulated branch point intermediate for the formation of the purine nucleotides AMP and GMP. Moreover, another important THF coenzyme is 5,10-CH₂-THF which serves as a cofactor for the key enzyme thymidylate synthase (TS) which extracts a methyl group from 5,10-CH₂-THF and attaches it to dUMP thereby resulting in the formation of dTMP. Hence, these folate-dependent enzymes provide the cells with both purine and thymine nucleotides essential for DNA replication. Dihydrofolic acid (DHF), the resultant byproduct of this biosynthetic reaction is efficiently recycled to THF via an NADPH-dependent reduction catalyzed by the important enzyme dihydrofolate reductase (DHFR). THF cofactors are then interconverted enzymatically to the abovementioned various one-carbon donors (Fig. 1) thereby forming a cyclic system.

Recognition of the central role that folic acid metabolism plays in the de novo biosynthesis of purine and pyrimidine nucleotides and DNA replication paved the way for the introduction of folic acid antagonists, the first class of antimetabolite anticancer drugs (Farber et al., 1948, Gorlick et al., 1996, Purcell and Ettinger, 2003). Nearly 60 years ago, this class of hydrophilic antifolates entered the clinics for the treatment of childhood acute lymphoblastic leukemia (ALL). Specifically, aminopterin (4-amino-folic acid), a homologue of the widely used anticancer drug MTX (4-amino-10-methylfolic acid), achieved remarkable remissions in childhood ALL (Farber et al., 1948). MTX and other antifolates (Fig. 2) are integral components of different chemotherapeutic regimens currently used for the treatment of various human malignancies. These include ALL, osteosarcoma, breast cancer, colorectal cancer, malignant pleural mesothelioma, non-small cell lung cancer, primary central nervous system (CNS) lymphoma, choriocarcinoma and gestational trophoblastic neoplasia (Walling, 2006).

Antifolates inhibit key enzymes in folate metabolism (Fig. 2, Fig. 3). Aminopterin and MTX are potent inhibitors of DHFR, a key enzyme in folate metabolism. The pharmacologic activity of MTX as an efficacious cytotoxic agent is highly attributable to its extremely potent and almost irreversible inhibition of DHFR (Ki = 5 pM; Johnson et al., 1997). Inhibition of the catalytic activity of DHFR or one of other key enzymes in folic acid metabolism leads to the disruption of purine and pyrimidine nucleotide biosynthesis, thereby resulting in cessation of DNA replication and cell death. Over the recent decades, second and third generation novel antifolates were synthesized; these include more potent inhibitors of DHFR such as PT523 and some of its derivatives which inhibit DHFR with a 15-fold greater affinity (Ki = 0.35 pM; Rosowsky et al., 2000). Furthermore, DHFR inhibitors like edatrexate (10-ethyl-10-deazaaminopterin) have been rationally synthesized which display unique properties such as much better transport via the reduced folate carrier (RFC) and increased polyglutamylation leading to augmented cellular accumulation and retention (Sirotnak et al., 1984, Schornagel et al., 1995). Rationally-designed, novel generation antifolates also include potent TS inhibitors such as raltitrexed (Tomudex; ZD1694) (Jackman et al., 1991) that has been approved for the treatment of advanced colorectal cancer (Cocconi et al., 1998) as well as pemetrexed (Alimta; LY231514) (Shih et al., 1997) that was registered in both the USA and Europe for the treatment of malignant pleural mesothelioma (Vogelzang et al., 2003) and non-small cell lung cancer (Hanna et al., 2004). Additionally, novel antifolates have been synthesized that are currently undergoing clinical evaluation as antitumor agents including the lipophilic antifolates nolatrexed (Thymitaq; AG337) (Pivot et al., 2001), neutrexin (trimetrexate) (Matin et al., 2005) and piritrexim (BW301U (Huie et al., 2005), the polyglutamatable hydrophilic TS inhibitor OSI-7904(L) (GW1843) (Duch et al., 1993) as well as the non-polyglutamatable TS inhibitor plevitrexed (BGC9331; ZD9331) (Jackman et al., 1997). Antifolates including raltitrexed, pemetrexed, OSI-7904(L) and plevitrexed were rationally designed to be efficiently taken up by the RFC, the primary transport system of reduced folates and antifolates (Fig. 4; see discussion in the next section). Whereas others, including the lipid-soluble antifolates nolatrexed and neutrexin were designed to enter cells by diffusion, thereby bypassing drug resistance phenomena associated with impaired drug uptake due to qualitative (i.e. inactivating mutations) and quantitative (i.e. down-regulation) alterations in the RFC (Jansen et al., 1998, Drori et al., 2000, Rothem et al., 2002). Other non-polyglutamtable hydrophilic antifolates including plevitrexed and PT523 were designed to overcome antifolate-resistance phenomena associated with impaired antifolate polyglutamylation due to decreased activity of folylpolyglutamate synthetase (FPGS), the enzyme responsible for polyglutamyl conjugation of both THF cofactors and antifolates (Liani et al., 2003). Recently, a novel hydrophilic antifolate BGC945 has been synthesized that is on the one hand a potent inhibitor of TS (Ki = 1.2 nM) and on the other hand is an excellent substrate of folate receptor α (FR α; Fig. 4), a high affinity folate uptake route, but a very poor substrate for the RFC (Gibbs et al., 2005). As such, this antifolate can selectively target human malignancies such as ovarian cancers that highly overexpress FRα thereby avoiding cytotoxic side effects to most normal tissues that ubiquitously express the RFC but not FRα. Validation of GARTF as an anticancer drug target came in the 1980s with the discovery of the first selective and potent inhibitor (Ki = 65 nM) lometrexol (5,10-dideaza-5,6,7,8-tetrahydrofolic acid (DDATHF) (Taylor et al., 1985). Recently, a more potent analogue of lometrexol, 10-(trifluoroacetyl)-5,10-dideazacyclic-5,6,7,8-tetrahydrofolic acid has been synthesized that proved to be a tight binder of GARTF with a remarkable inhibition constant (Ki = 15 nM) and cytotoxic activity against human leukemia cells (Zhang et al., 2003). Recently, a computerized hybrid functional Petri nets (HFPN) modeling of folate metabolism under physiologoical conditions and antifolate inhibitory conditions has been achieved (Assaraf et al., 2006). This HFPN-based simulation offers an inexpensive, user-friendly and reliable means of pre-clinical evaluation of the inhibitory profiles of various antifolates. Hence, various key enzymes in folate metabolism have been successfully targeted thereby proving excellent determinants for the cytotoxic activity of different antifolates several of which are currently used in the combination chemotherapeutic treatment of various human cancers (Walling, 2006).