Biology and therapeutic applications of the proton-coupled folate transporter

Abstract

Introduction:

The proton-coupled folate transporter (PCFT; SLC46A1) was discovered in 2006 as the principal mechanism by which folates are absorbed in the intestine and the causal basis for hereditary folate malabsorption (HFM). In 2011, it was found that PCFT is highly expressed in many tumors. This stimulated interest in using PCFT for cytotoxic drug targeting, taking advantage of the substantial levels of PCFT transport and acidic pH conditions commonly associated with tumors.

Areas covered:

We summarize the literature from 2006–2022 that explores the role of PCFT in the intestinal absorption of dietary folates and its role in HFM, and as a transporter of folates and antifolates such as pemetrexed (Alimta) in relation to cancer. We provide the rationale for the discovery of a new generation of targeted pyrrolo[2,3-d]pyrimidine antifolates with selective PCFT transport and inhibitory activity toward de novo purine biosynthesis in solid tumors. We summarize the benefits of this approach to cancer therapy and exciting new developments in the structural biology of PCFT and its potential to foster refinement of active structures of PCFT-targeted anti-cancer drugs.

Expert opinion:

We summarize the promising future and potential challenges of implementing PCFT-targeted therapeutics for HFM and a variety of cancers.

1. Introduction and general background

Folates are B9 vitamins that are essential for the biosynthesis of purine nucleotides, thymidylate, serine and methionine [1]. Mammals cannot synthesize folates de novo so these cofactors must be obtained from the diet. Folates are hydrophilic anions and diffuse poorly across plasma membranes. Mammalian cells principally use facilitative transporters to accumulate folates and related compounds including the reduced folate carrier (RFC or SLC19A1) and the proton-coupled folate transporter (PCFT or SLC46A1) (Figure 1) [2–4].

Figure 1.

Folate transport and C1 metabolism. The schematic depicts cellular uptake of folates by facilitated transport via RFC and PCFT. Following internalization, folates are transported into the mitochondria by the mitochondrial folate transporter or SLC25A32. In the mitochondria, serine is catabolized by sequential SHMT2, MTHFD2/L and MTHFD1L through which the C1 moiety from serine C3 is incorporated into formate, thus providing C1 units for cellular biosynthesis in the cytosol. Intracellular targets of the clinically approved antifolates and PCFT-targeted antifolates, as described in the text, are noted. Abbreviations are as follows: 10-CHO-THF, 10-formyl tetrahydrofolate; 5,10-me-THF, 5,10-methylene tetrahydrofolate; AICAR, 5-aminoimidazole-4-carboxamide ribonucleotide; ATIC, 5-aminoimidazole-4-carboxamide ribonucleotide formyltransferase/inosine monophosphate cyclohydrolase; DHF, dihydrofolate; DHFR, dihydrofolate reductase; FAICAR, formyl 5-aminoimidazole-4-carboxamide ribonucleotide; fGAR, formyl glycinamide ribonucleotide; GAR, glycinamide ribonucleotide; GARFTase, glycinamide ribonucleotide formyltransferase; MTHFD1, 5,10-methylene tetrahydrofolate dehydrogenase 1; MTHFD2(L), 5,10-methylene tetrahydrofolate dehydrogenase 2(-like); MTHFR, 5,10-methylene tetrahydrofolate reductase; MTR, methionine synthase; MTX, methotrexate; PCFT, proton-coupled folate transporter; PDX, pralatrexate; PMX, pemetrexed; PRPP, phosphoribosyl pyrophosphate; RFC, reduced folate carrier; RTX, raltitrexed; SAM, S-adenosylmethionine; SHMT1/2, serine hydroxymethyltransferase 1/2; THF, tetrahydrofolate; TS, thymidylate synthase.

Although high affinity folate receptors (FRs) (α,β, and δ) can facilitate accumulation of folates and related compounds by endocytosis, cellular uptake by folate receptors is inefficient relative to RFC and PCFT [5,6].

RFC is the major tissue folate transporter and is expressed ubiquitously [3]. Loss of RFC in mice is embryonic lethal [7]. The preferred substrate for RFC is (6S)-5-methyltetrahydrofolate, the primary circulating folate form [2]. Antifolate drugs approved for cancer include methotrexate, pemetrexed, pralatrexate and raltitrexed (Figure 2A) and all are transport substrates for RFC [2,8,9].

Figure 2.

Structures of clinically approved antifolates including methotrexate, pemetrexed (Alimta), pralatrexate, and raltitrexed (A) and PCFT-selective 6-substituted pyrrolo[2,3-d]pyrimidine antifolates (B). Panel C summarizes Ki values for various pyrrolo[2,3-d]pyrimidine antifolates measured at pH 5.5 with [³H]MTX in R2/PCFT4 Chinese hamster ovary cells expressing human PCFT [37]. Data are presented as mean values ± standard errors (in parentheses) (n=4). For fluorinated antifolates (AGF264, AGF270, AGF278, AGF283), values significantly different from those of parent antifolates (AGF23, AGF17, AGF107, AGF117, and AGF150, respectively) (p < 0.05) are marked with *. These data are from Ravindra et al. [91].

PCFT was identified in 2006 as the major transporter of dietary folates at the acidic pH of the upper gastrointestinal (GI) tract [10]. PCFT was originally described as a low affinity intestinal heme transporter important for iron homeostasis [11]; however, later studies established its principal role is as a proton-coupled folate transporter [10]. PCFT is expressed in other tissues [10,12–14], although given its modest transport activity at neutral pH, outside of the choroid plexus, its broader physiologic role is not well established [14]. With the discovery that PCFT is widely expressed in human tumors [15–18], attention soon turned to its potential for therapeutic targeting of cancer. This led to the discovery of novel cytotoxic 6-substituted pyrrolo[2,3-d]pyrimidine antifolates (Figure 2B) with limited transport by RFC that are transported into tumors by PCFT at the acidic pH commonly associated with the tumor microenvironment [2,12,19]. As a result of these studies, a unique pharmacophore for PCFT-drug targeting has begun to emerge [8,12,19]. Indeed, results of structure-activity studies of PCFT for a number of compounds with varied capacities for transport by PCFT versus RFC were recently validated by cryo-electron microscopy (cryo-EM), including structures of chicken PCFT (cPCFT) in its apo form and cPCFT complexed with pemetrexed [2,20]. These are poised to provide an increasingly rational approach for developing improved PCFT-targeted therapies for cancer.

2. Discovery and expression of PCFT

The proximal small intestine is the major site of absorption of dietary folates and PCFT is highly expressed in the duodenum and jejunum [21]. PCFT levels are lower in more distal regions of the intestine [13]. While RFC is expressed throughout the intestine [22], the acidic pH of jejunum (pH 5.8–6.0) does not support transport by RFC. Dietary folates are transported by PCFT across the apical brush border membrane in the proximal small intestine into the enterocytes [21]. Folates are exported across the basolateral membrane into the vascular system by ABCC proteins (principally ABCC3) [21].

It was studies of the genetic basis for a rare autosomal recessive condition termed hereditary folate malabsorption (HFM) syndrome associated with impaired intestinal absorption of folates that eventually led to the identification of PCFT as the primary mechanism of folate transport into enterocytes in the proximal intestine [10,23]. The human SLC46A1 gene which encodes PCFT is located on chromosome 17q11.2. Inherited variants in the SLC46A1 gene in patients result in loss of PCFT expression and/or transport with severely depressed levels of systemic and cerebral folates [23]. HFM syndrome manifests shortly after birth and is associated with anemia, GI symptoms, immune deficiency and neurologic effects [23,24], and can be reversed by parenteral administration of exogenous folate.

In addition to its presence in the upper GI, PCFT can be detected in the choroid plexus, kidney, sinusoidal membrane of the liver, retinal pigment epithelium, spleen and placenta [10,12–14]. PCFT expression extends to a growing number of human tumors including breast, lung, ovary and pancreas cancers, as well as malignant pleural mesothelioma [15,17,18]. In contrast, PCFT levels in human leukemias are very low [17].

A number of studies have begun to explore the transcriptional regulation of the human PCFT (hPCFT) gene [25–30]. A large (~1 kb) CpG island is located in the hPCFT promoter; hypermethylation of this region results in very low levels of hPCFT [15,28,31]. Further, 5-aza-2’-deoxycytidine treatment substantially restored hPCFT levels to low-hPCFT expressing cells [15,28,31]. Assorted transcription factors were implicated in regulating hPCFT including SP1, AP1, AP2, NRF1, KLF4, KLF15, HNF4α, YY1 and vitamin D receptor [25–30]. The vastly different levels of hPCFT between HepG2 human hepatocellular carcinoma (high PCFT) and HT1080 fibrosarcoma (undetectable PCFT) cells were reflected in divergent levels of NRF1, KLF15 and SP1 proteins [25]. Interestingly, dietary folate levels appear to regulate murine PCFT levels, as mice fed a folate-deficient diet experienced a ~13-fold increase in PCFT transcripts in the proximal small intestine compared to mice fed a folate-replete diet [13].

3. PCFT mechanism and structure

PCFT is a member of the major facilitator superfamily (MFS) of transporters. Other MFS transporters include the inorganic phosphate/glycerol-3-phosphate antiporter (GlpT), the bacterial lactate/proton symporter (LacY) and RFC [32,33]. The hPCFT protein is composed of 459 amino acids and includes 12 transmembrane domain (TMD) segments with cytosolic oriented N- and C-termini (Figure 3). hPCFT is glycosylated at Asn58 and Asn68 in the first extracellular TMD. Both RFC and PCFT exist as oligomeric forms (i.e., dimers) [34,35]. For hPCFT, oligomerization is associated with optimal transport activity [34].

Figure 3.

Schematic structure of hPCFT membrane topology. A topology model for hPCFT is shown (HMMTOP prediction algorithm), which depicts 12 TMDs and internal N- and C-termini. N-Glycosylation sites at Asn58 and Asn68 are shown as green circles. Structurally and functionally important residues in hPCFT based on the cPCFT structural studies and biochemical studies, as described in the text, are shown as red circles, including Gln39, Asn90, Asp109, Gly112, Arg113, Gln127, Arg148, Asp156, Glu185, Hi247, His281, Phe282, Tyr315, Asn342, Arg376, and Met403.

RFC substrates are also transported by PCFT, including 5-methyl tetrahydrofolate and 5-formyl tetrahydrofolate, and the clinically used antifolates methotrexate, pemetrexed, raltitrexed and pralatrexate [4,8,10,12] (Figure 2A). PCFT transport is specific for (6S)-5-formyl tetrahydrofolate and for L- over D-aminopterin [36]. In contrast to RFC, folic acid is a good substrate for PCFT [4]. Conversely, the novel antifolates PT523 and GW1843U89 are excellent substrates for RFC but are poorly transported by PCFT [37].

PCFT is a proton symporter for which the proton gradient across the plasma membrane drives uphill transport of folates. In contrast to RFC which functions as an organic anion exchanger with optimal transport at pH 7.2–7.4 [3], transport by PCFT is optimal at pH 5–5.5 with appreciable activity up to pH 6.5–6.8 [4,38,39]. Tissues that express PCFT invariably express RFC as well, and a dynamic functional interplay can occur between RFC and PCFT that depends on transporter levels and extracellular pH [38].

Interestingly, the anti-inflammatory drugs sulfasalazine and indomethacin inhibit PCFT transport [39]; sulfasalazine has also been reported to inhibit intestinal folate absorption [40]. Removal of Na⁺, K⁺, Ca²⁺, Mg²⁺ or Cl⁻ has no impact on PCFT transport; however treatment with the proton ionophore carbonylcyanide p-trifluoro-methoxyphenylhydrazone or nigericin (a K⁺/H⁺ exchanging ionophore) decreases PCFT transport [41]. Treatment of cells with bicarbonate, bisulfite or nitrate abolished the pH gradient and inhibited PCFT transport at slightly acid or neutral pH [42]. In Xenopus oocytes, PCFT transport is electrogenic and results in intracellular acidification [43]. At acid pH, proton and folate fluxes are uncoupled.

Structurally or functionally important amino acids in the hPCFT have been identified from characterizing loss-of-function mutations in hPCFT from patients with HFM syndrome [10,23,24,44–61]. Additional studies used systematic site-directed mutagenesis based on sequence homologies between species and predictions of membrane topology to identify critical residues [47,61–73]. Several reports described results with substituted-cysteine accessibility methods with thiol-reactive reagents (i.e., 2-aminoethyl methane thiosulfonate) to probe aqueous accessible residues/domains and identify potential ligand binding sites in hPCFT [60,63–67,73,74]. Mechanistically important amino acids in hPCFT are highlighted in Figure 3 and are summarized in Figure 4B.

Figure 4.

cPCFT structure and identified substrate-binding sites. A, rearrangements of key side-chains are observed after pemetrexed and proton binding with cPCFT. The insets show the cryo-EM densities for the two salt-bridge interactions of cPCFT, which are broken in the presence of substrate. Arrows indicate the direction of side-chain movement. TMDs 4, 5 and 10 in cPCFT are shown. This panel is from the publication by Parker et al. [20] with copyright permission. B, table lists key residues in the cPCFT structure and the corresponding residues in hPCFT. C, schematic of the deduced binding interactions for pemetrexed bound to cPCFT based on the cPCFT/pemetrexed structure [20]. This panel is adapted from a schematic by Parker et al. [20] with modifications. D shows predicted binding interactions of AGF94 with cPCFT. This panel is adapted from a schematic in the publication by Parker et al. [20] with modifications.

Cryo-EM structures of cPCFT were reported without bound substrate (3.2 Å) and in complex with pemetrexed (3.3 Å) [20]. The cPCFT protein assumed the canonical “MFS fold” with the 12 TMD segments in two discrete N- and C-terminal domains in an “outward-open” conformation [20]. Salt bridges between Arg156 and Asp164 in TMD4, and between Glu193 in TMD5 and Arg384 in TMD10 stabilized this conformation which was further stabilized by a hydrogen-bond between His289 in TMD7 and Asn350 in TM9 (for select functionally important residues in cPCFT, the homologous residues in hPCFT are noted in Figure 4B). In the cPCFT structure, TMDs 1, 2, 4, 5, 7, 8, 10, and 11 formed a hydrophilic cavity for binding anionic (anti)folates, and TMDs 1, 2, 7 and 8 formed an open extracellular gate and TMDs 4, 5, 10 and 11 formed a closed intracellular gate (Figure 4A).

When complexed with cPCFT, pemetrexed assumes a “kinked” conformation [20]. The pyrimidine moiety fits in a polar cavity constructed from the side chains in the C-terminal domain with the carbonyl and 3-nitrogen of pemetrexed interacting with Tyr323 (TMD8), and the 2-amino group of pemetrexed interacting with Glu193 (TMD5) (Figure 4C). The pemetrexed pyrrole nitrogen interacts with Glu407 in TMD11; the benzoyl side chain of pemetrexed makes a π–π stacking interaction with Phe290 in TMD7. The L-glutamate of pemetrexed interacts with Arg156 in TMD4 and Asn98 in TMD2 in a positively charged pocket (Figure 4C). The L-glutamate of pemetrexed does not interact with specific amino acids in the binding site of cPCFT; however, both the amide and carbonyl moieties flanking the benzoyl side chain of pemetrexed interact with Ser411 in TMD11.

The cPCFT cryo-EM structure permits consideration of the structural changes in the protein that accompany and facilitate transport [2,20]. As with other MFS transporters, PCFT transport is presumed to mediate folate transport by an “alternating access” mechanism. This involves distinct conformational changes in the transporter which enable the ligand binding site to access either side of the plasma membrane in an alternating fashion [75,76]. During each transport cycle, the extracellular and intracellular gates alternate between open and closed states via formation of intra- and interhelical salt bridges, thus providing a facile mechanism to couple (anti)folate transport to pH gradients.

Using cPCFT and pemetrexed transport as an example, this is suggested to occur in 5 steps (Figure 5). (i) In the outward open state, Asp164 and Glu193 are protonated, resulting in weakened salt bridges between Arg156 and Asp164, and between Glu193 and Arg384. (ii) Increased pKa values for Asp164 and Glu193 upon pemetrexed binding further weaken these salt bridges. This enables the formation of an additional salt bridge between Arg156 and the γ-carboxylate of pemetrexed while the 2-amine group on the pyrimidine associates with Glu193. Binding of pemetrexed also results in disruption of a hydrogen bond between His289 (TMD7) and Asn350 (TMD9). (iii) His289 protonation effects an interaction between the pemetrexed benzoyl group with Phe290 and the repositioning of TMD7 toward TMD1. This causes the extracellular gate to close. (iv) Pemetrexed binding to cPCFT facilitates opening of the cytoplasmic half of TMD4 and TMD5, resulting in the intracellular release of substrate. Finally, (v) Asp164, Glu193, and His289 are deprotonated, the salt bridge and hydrogen bond networks reform, and cPCFT reorients to the outward-facing state (Figure 5).

Figure 5.

“Alternating access” transport model for cPCFT. The schematic is based on the alternating access model for secondary transporters [75,76] and structural changes in the protein that facilitate transport by cPCFT [2,20]. PCFT transport likely facilitates folate transport by an alternating access mechanism such that the ligand binding site is alternately accessible to either side of the membrane through distinct conformational changes. During each transport cycle, the extracellular and intracellular gates alternate between open and closed states via the formation of intra- and interhelical salt bridges, thus providing a facile mechanism to couple (anti)folate transport to pH gradients. This image is adapted from a schematic in a publication by Hou et al. [2] with modifications.

This mechanism accounts for how protons facilitate co-transport of folates, indeed the characteristic acid pH dependence for transport of folates by PCFT. While transport in Figure 5 is depicted for monomeric cPCFT, should cPCFT exist as dimers (analogous to hPCFT [34]), both cPCFT monomers would be expected to undergo the transport cycle concurrently such that functional cooperativity occurs between the individual monomers [34].

4. Discovery of PCFT-targeted therapeutics for cancer

Antifolates continue to be important agents for treating cancer along with other conditions (Figure 2) [8,9]. Methotrexate is widely used to treat lymphoblastic leukemia, osteogenic sarcoma, lymphomas and breast cancer [9]. Pemetrexed (Alimta) is approved by the US FDA for treating malignant pleural mesothelioma and lung adenocarcinoma [77–80], whereas pralatrexate is approved for treating relapsed and refractory peripheral T-cell lymphoma [81]. Raltitrexed is approved for the treatment of advanced colorectal cancer in Europe, Canada and Australia [82]; raltitrexed is also approved for treating malignant pleural mesothelioma in several European countries [83].

Classic antifolates inhibit essential metabolic targets in one-carbon (C1) metabolism (Figure 1). Methotrexate and pralatrexate are both dihydrofolate reductase inhibitors [9]; pemetrexed and raltirexed inhibit thymidylate synthase, and pemetrexed also inhibits folate-dependent enzymes in de novo purine biosynthesis [5-aminoimidazole-4-carboxamide ribonucleotide formyltransferase (ATIC) and glycinamide ribonucleotide (GAR) formyltransferase (GARFTase)] [9,77]. All these antifolates are excellent transport substrates for RFC and substrates for polyglutamylation [9]. For methotrexate, transport by RFC is essential for generating sufficient intracellular drug to maximize inhibition of dihydrofolate reductase and to provide substrate for metabolism to methotrexate polyglutamates required for cellular retention and sustained dihydrofolate reductase inhibition [84]. For pemetrexed, polyglutamylation results in enhanced inhibition of its intracellular targets [77]. Loss of RFC transport results in resistance to methotrexate in preclinical models [3] and in patients with acute lymphoblastic leukemia and osteosarcoma [85–87]. For antifolates that are principally transported by RFC, selectivity toward cancer cells over normal proliferative tissues (e.g., bone marrow) is limited, as both cancer and normal tissues express RFC [3].

Following the discovery of PCFT, studies began to explore the association between hPCFT expression and the cancer phenotype [8,12,19]. Initial studies examined hPCFT transcript levels by real-time RT-PCR in 80 human tumor and leukemia cell lines [17]. hPCFT was detected in 52 of 53 tumor cell lines from a wide range of tumor subtypes, with especially high levels of hPCFT in epithelial ovarian cancer, lung adenocarcinoma, malignant pleural mesothelioma and pancreatic adenocarcinoma [17]. Low-to-undetectable levels of hPCFT were measured in 27 human leukemias [17]. In tumor cell lines, hPCFT transcripts paralleled levels of hPCFT protein on western blots and transport activity at pH 5.5 (pH optimum for PCFT) [15,17,72]. High levels of hPCFT were also measured by real-time RT-PCR and immunohistochemistry in primary human tumors including non-small cell lung cancer [72], epithelial ovarian cancer [16], malignant pleural mesothelioma [15], and pancreatic adenocarcinoma (J. Frühauf, Z. Hou, and L.H. Matherly, unpublished).

The antifolate pemetrexed is among the best substrates for PCFT [8]. Transport of pemetrexed by PCFT is an important determinant of its anti-tumor efficacy for lung cancer and malignant pleural mesothelioma [8,9]. For malignant mesothelioma patients with low levels of PCFT expression, treatment with pemetrexed was accompanied by significantly lower rates of disease control and shorter overall survival [15]. Multivariant analysis further confirmed a prognostic role for PCFT in malignant pleural mesothelioma [15].

While all these clinically approved antifolates are transported by PCFT, their transport by RFC results in loss of tumor selectivity [8]. Comprehensive studies by Matherly and Gangjee have explored the structure-activity determinants of antifolate substrates for PCFT versus RFC [8,12,19]. Pemetrexed is a 5-substituted 2-amino-4-oxo-pyrrolo[2,3-d]pyrimidine antifolate with a 2-carbon bridge linked to a p-aminobenzoyl L-glutamate [77] (Figure 2A). Matherly and Gangjee systematically modified the pemetrexed structure with the goal of identifying structural determinants for optimal PCFT transport, and to identify novel targeted antifolates with tumor specificities based on their selective membrane transport by PCFT over RFC [12,88–95]. The modifications involved substitutions on the bicyclic pyrrolo[2,3-d]pyrimidine ring, the length and nature of the side-chain bridge, the character of the side-chain aromatic ring and the nature of the terminal amino acid (Figure 2B). To assess the impact of these modifications on PCFT or RFC transport activity, initial studies used isogenic Chinese hamster ovary or HeLa cell lines engineered to express a single transporter (PCFT or RFC) such that differences in transport are reflected as inhibition of cell proliferation [37,89,96]. Other studies directly measured PCFT transport using radiolabeled antifolates [15,18,89,96,97] and clinically relevant human tumor cell lines [12,18,88–97].

All the PCFT selective compounds of this series were 6-substituted pyrrolo[2,3-d]pyrimidine antifolates. While the 6-subsituted pyrrolo[2,3-d]pyrimidine regioisomer of pemetrexed with a two carbon bridge was pharmacologically inert [98,99], analogs with 3- (AGF17) or 4- (AGF23) bridge carbons (Figure 2B) inhibited cell proliferation at nanomolar concentrations, with near absolute PCFT selectivity over RFC [89]. PCFT transport and cell inhibition decreased with 5- or 6-carbons in the bridge region [89]. When isosteric heteroatom replacements (i.e., N, O and S) were introduced in the bridge region of AGF17 (i.e., AGF183, AGF182 and AGF140, respectively) (Figure 2B), PCFT transport was preserved [88]. Further, N-formyl (AGF219), N-acetyl (AGF174) or N-trifluoroacetyl (AGF209) substituted compounds (Figure 2B) showed no differences in PCFT transport from the N-unsubstituted compound (AGF183) [88]. Importantly, none of these compounds exhibited any RFC transport activity.

Replacement of the side-chain phenyl ring (AGF17 and AGF23) with a straight chain alkyl moiety abolished PCFT transport [100], whereas replacement with a thienoyl (AGF71, AGF94, AGF117, AGF118, AGF150, AGF154) [92–95] or pyridyl (AGF107) (Figure 2B) ring preserved PCFT transport [90]. AGF94 and AGF154 with a 3-carbon bridge and side-chain thienoyl ring were substantially more active for PCFT transport than the side-chain phenyl compound AGF17 [94,95] and the corresponding thienoyl compounds with a 4-carbon bridge (AGF71, AGF117, AGF118) [92–95] (Figure 2B). K_is for the most active PCFT substrates compared to pemetrexed were determined [91] (Figure 2C). Fluorine substitutions on the side-chain phenyl or thiophene rings resulted in decreased K_is for binding to PCFT (AGF264 and AGF23; AGF279 and AGF17; AGF278 and AGF117; AGF283 and AGF150) [91]. This was accompanied by increased inhibition of cell proliferation [91], although factors in addition to PCFT transport were likely involved. Finally, replacement of the L-glutamate in PCFT selective substrates (e.g., AGF94) with unnatural amino acids (α-amino adipate, 4-amino butanoate, α-amino pentanoate) abolished PCFT transport although for certain compounds, binding to PCFT was preserved [101].

The reported cryo-EM cPCFT structure provided further structural insights into the impact of the side-chain bridge length and the aromatic ring on PCFT binding and transport of pyrrolo[2,3-d]pyrimidine antifolates [20]. Thus, the cPCFT structural model predicts that the pemetrexed γ-carboxyl coordinates with Arg156 and that the 2-amino group of the pemetrexed pyrimidine moiety coordinates with Glu193 so as to define a minimal distance between these residues in the ligand binding site for related substrates (Figure 4C). Lengthening the bridge to 3- or 4-carbons as in AGF94 and AGF71 would cause the γ-carboxylate to extend further into a polar pocket including Gln43 (TMD1) and Gln135 (TMD3). The net effect is that the α-carboxylate of the bound antifolate now interacts with Arg156 in lieu of the γ-carboxylate (Figure 4D) [20]. 6-Subsitituted pyrrolo[2,3-d]pyrimidine compounds with fewer than 3-bridge carbons are inactive [17], likely due to loss of PCFT binding.

It is of interest that all the pharmacologically active selective PCFT-targeted antifolates to date (Figure 2B) are inhibitors of de novo purine nucleotide biosynthesis at the first folate-dependent step in the 10-step sequence from phosphoribosyl pyrophosphate (PRPP) to IMP (Figure 1) in which GAR is formylated by 10-formyl tetrahydrofolate to generate formyl GAR, mediated by GARFTase in the GART trifunctional enzyme [12,88–95]. By in vitro assays with purified human GARFTase (formyltransferase domain), inhibition by 6-substituted pyrrolo[2,3-d]pyrimidine compounds including AGF71, AGF94, AGF117, AGF118, AGF150, AGF154 and AGF183 was confirmed with K_i values at low-to-mid nanomolar concentrations [88,92–95,102]. GARFTase inhibition in cells was further confirmed by in situ assays [12,88–95] . From crystal structures of ternary complexes with GARFTase, β-GAR, and monoglutamyl pyrrolo[2,3-d]pyrimidine compounds, in vitro inhibitory potencies correlated with drug binding and positioning of the terminal L-glutamate carboxylates [88,95,102].

In vivo efficacy toward human tumor xenografts in mouse models was recorded for the PCFT-targeted antifolates (e.g., AGF71, AGF94, AGF154 and AGF278), including hepatocellular carcinoma [17], epithelial ovarian cancer [16,103], malignant pleural mesothelioma [97] and lung adenocarcinoma [72]. Recent studies extended studies with AGF94 in immune-compromised mouse models of ovarian cancer to a novel syngeneic mouse model of high grade serous ovarian cancer [103]. Interestingly, all of these compounds are also substrates for endocytosis by FRs, along with PCFT [91,92,94,95]. Thus, tumors that express both FRs and PCFT such as epithelial ovarian cancer [16,104] and lung adenocarcinoma [105] would be uniquely sensitive to pyrrolo[2,3-d]pyrimidine drugs such as AGF94. Indeed, in epithelial ovarian cancer cells, in vitro and in vivo efficacies to the pyrrolo[2,3-d]pyrimidine antifolates AGF94 and AGF154 were significantly preserved in spite of variations in the level of FRα [16].

AGF94, AGF154 and AGF278 (Figure 2B) are prototypes of PCFT-selective antitumor agents with substantially reduced transport by RFC and potent inhibition of de novo purine biosynthesis at GARFTase [16,18,91,94–96]. The basis for tumor selectivity by these drugs is as follows. (i) The normal tissue milieu is generally at a neutral pH which favors RFC transport. Even if some PCFT is expressed, the reduced electrochemical proton gradient in normal tissues would result in limited accumulation of PCFT-targeted compounds such as AGF94. Transport by PCFT at neutral pH would be further decreased via inhibition by bicarbonate anions [42]. (ii) Further, RFC transport of extracellular folates into normal tissues at neutral pH would likely protect cells from antifolate cytotoxicity by competing for polyglutamylation and/or for binding to intracellular targets; additional effects of high levels of intracellular folates include decreased activity of folypolyglutamate synthetase [106] and/or increased or altered cellular distributions of ABC transporters [107]. (iii) In the acidic microenvironment of tumors, the elevated electrochemical proton gradient favors high levels of uptake of cytotoxic PCFT substrates such as AGF94 which are metabolized to polyglutamates [96]. (iv) The 6-substituted pyrrolo[2,3-d]pyrimidine antifolate polyglutamates inhibit GARFTase and de novo purine biosynthesis independent of tumor p53 mutation status [108], resulting in decreased pools of purine nucleotides and related metabolites. Selectivity of GARFTase inhibition toward tumors versus normal tissues is further enhanced by differences in rates of de novo purine nucleotide biosynthesis and purine salvage between normal tissues and tumors [109,110]. (v) The net effect is potent and targeted killing of tumor cells that express PCFT with limited toxicity toward normal tissues. For tumors that express PCFT with FRα, transport redundancy preserves uptake independent of variations in one or the other transporter. Thus, cytotoxic PCFT selective agents without RFC transport should exhibit far greater tumor selectivity and much less toxicity toward normal tissues than previous generations of antifolates such as methotrexate or pemetrexed.

5. Conclusion

PCFT was discovered in 2006 as the major folate transport system involved in the intestinal absorption of dietary folates [10]. Studies of mutations in PCFT as the causal basis for HFM [10] led to identification of mechanistically important residues/domains involved in (anti)folate substrate binding and proton coupling [10,23,24,44–61]. Like other MFS proteins, PCFT was identified as a dimer with functional and regulatory significance [34]; however, the broader functional significance of PCFT dimerization is still evolving [2]. Additional studies identified key gene promoter elements and critical transcription factors, and demonstrated promoter hypermethylation, as important regulators of hPCFT levels between tissues and tumors [25–30]. The substantial levels of hPCFT expressed in human tumors [15–18], combined with high levels of PCFT transport activity at acidic pHs [4,38], have emerged as defining features that offer substantial promise for future clinical translation. With this goal in mind, an important step involves the discovery of novel 6-substituted pyrrolo[2,3-d]pyrimidine antifolates with PCFT transport selectivity over RFC and potent in vitro and in vivo anti-tumor efficacies toward a range of human tumors [12,88–95]. As these agents are transported by FRs as well as PCFT, for certain tumors (e.g., epithelial ovarian cancer), this preserves an important redundancy in the face of variations in the levels of either system [16].

The recent results of cryo-EM structures of cPCFT in its apo form and in complex with pemetrexed [20] have validated several features of optimal PCFT substrates. Moving forward, these PCFT structures are especially promising for continued development of PCFT-targeted therapies for cancer, drawing from structural biology to complement medicinal chemistry and multiparameter optimization of lead compounds.

6. Expert opinion

Over the past 16 years since PCFT was discovered as the principal transporter of dietary folates in the upper GI [10], research progress on the biology and clinical and potential therapeutic applications of this transport system has been staggering. Particularly noteworthy are the dramatic advances in characterizing causal roles of hPCFT mutations in loss-of-function phenotypes of HFM patients [23], identification of hPCFT as an important folate transporter in human tumors and discovery of novel PCFT-targeted antifolates as tumor-selective antitumor agents [12]. While cPCFT cryo-EM structures [20] have validated substrate determinants for binding and transport by PCFT, the hPCFT structure will be essential to enable an increasingly rational approach for design of a new generation of targeted therapeutics for cancer. This will be further advanced now that a human RFC structure has been reported [111].

Additional studies should focus on advancing research discoveries toward clinical applications. Of interest is the transcriptional and posttranscriptional regulation of PCFT. While PCFT CpG methylation was described [28,31], it is unclear whether coadministration of demethylating agents (i.e., azacytidine, decitabine) can be used therapeutically to increase PCFT levels in tissues and tumors. For HFM and potentially cancer, it is appealing to consider approaches to “rescue” misfolded or inactive PCFT proteins potentially through use of pharmacologic chaperones such as ligands, peptidomimetics or even interfacial lipids [2]. From studies of hPCFT dimerization, positive cooperativity between PCFT monomers and functional rescue of mutant hPCFT by wild-type hPCFT was described [34]. Heterologous protein interactions could also play an important role in PCFT regulation and function in disease, suggested by recent proximity-dependent biotin identification (BioID) studies including hPCFT [112], as protein interaction networks (“interactome”) are important for cellular homeostasis and could serve vital biological roles and provide potential drug targets.

In tumors, ischemia and acidosis promote progression, invasion and metastases, reduce drug efficacy through ion-trapping and cell cycle arrest, and result in evasion of immune recognition [113–119]. While variations in oxygenation in tumors occur chronically as well as acutely and could impact the success of PCFT therapeutic targeting due to potential effects on expression of PCFT and target enzyme(s) [120], this should be rigorously tested although in vivo efficacy of PCFT-targeted antifolates with human tumor xenografts has been unequivocally demonstrated [16–18,72,89,91,94,97,103]. Of course, given their selectivity for acidic tumors, it is also intriguing to consider other applications of PCFT-targeted antifolates such as tumor imaging with positron emission tomography. If realized, this could offer an exciting non-invasive strategy for stratifying patients for therapy.

An untested consideration in assessing the therapeutic potential of PCFT-targeted drugs involves the role of the tumor microenvironment, including infiltrating immune cells (e.g., tumor associated macrophages (TAMs) and T-lymphocytes) which impact disease progression and responses to therapy [121]. While a recent report described targeting TAMs in a syngeneic model of high grade serous ovarian cancer with the pyrrolo[2,3-d]pyrimidine antifolate AGF94 [103], this was suggested to involve FRβ, although TAMs also express PCFT [122]. Of course, as PCFT-targeted compounds such as AGF94 inhibit purine biosynthesis, another potential impact of this series includes their effects on purine nucleotide levels in tumors and adenosine pools in the tumor microenvironment which act as immune modulators via effects on macrophage polarization and inhibition of cytotoxic effector functions of both NK and CD8⁺ T cells [123–125]. Understanding of the role of C1 metabolism in inflammatory responses to therapy in relation to cancer progression is an essential area for future studies.

Other areas for future study of PCFT-targeted antifolates for cancer include mechanisms of drug resistance, whether discovery of PCFT selective antifolates without FR activity is possible or necessary, whether selective PCFT-targeting can be rationally expanded to include a wider array of intracellular targets including those mitochondrial C1 metabolism [126], and the extent to which PCFT-selective purine biosynthesis inhibitors can be combined with other drugs used for cancer such as kinase (e.g., EGFR) inhibitors. The vast number of clinical trials of kinase inhibitors with cytotoxic agents attests to the potential significance of this latter approach.

Article highlights:

• The proton-coupled folate transporter (SLC46A1) is the principal mechanism by which folates are absorbed in the intestine and mutations in the SLC46A1 gene are the causal basis for hereditary folate malabsorption.

• The proton-coupled folate transporter is expressed in many tumors and is active under acidic conditions associated with tumors.

• New pyrrolo[2,3-d]pyrimidine antifolates are being discovered for cancer therapy with selective membrane transport via the proton-coupled folate transporter and as inhibitors of one-carbon metabolism.

• The recent structure of the proton-coupled folate transporter by cryo-electron microscopy provides new insights into transporter structure and function and should foster refinement of drugs for tumor-targeting based on their selective membrane transport.

• Future studies should emphasize further refining structures of drugs with selective transport by the proton-coupled folate transporter and inhibition of one-carbon metabolism, developing approaches for selectively increasing transporter levels or activity in tumors and for treating patients with hereditary folate malabsorption, and adapting discoveries in targeted drug delivery to include tumor imaging and key immune populations in the tumor microenvironment.