Single Nucleotide Polymorphisms in the Human Reduced Folate Carrier: Characterization of a High-Frequency G/A Variant at Position 80 and Transport Properties of the His²⁷ and Arg²⁷ Carriers¹

The natural folates are members of the B-class of vitamins and are essential to cell proliferation (1). Because mammalian cells cannot synthesize these essential cofactors, folates must be actively transported into cells, whereupon they participate in one-carbon transfer biosynthetic reactions leading to purines, pyrimidines, serine, and methionine. The primary route for membrane transport of reduced folates into mammalian cells involves the RFC³ (2, 3). Membrane transport via RFC is also an important determinant of antifolate therapeutics used in cancer chemotherapy (e.g., MTX and Tomudex), and impaired uptake of antifolates by RFC is a frequent mode of drug resistance (4). Likewise, the loss of RFC function conceivably could contribute to disease states associated with decreased accumulations of folate cofactors (e.g., fetal abnormalities, cardiovascular disease, and cancer; Refs. 5, 6, 7).

The RFC gene has been cloned and characterized from a number of sources (8, 9, 10, 11, 12, 13, 14, 15). For the hRFC, an intriguing feature involves transcript heterogeneity. For instance, multiple cDNAs were described for hRFC, differing in their 5′-untranslated regions (13). Furthermore, variable deletions of hRFC open reading frame sequence have been reported, including a functional hRFC form encoded by a transcript with a 626-bp deletion (13) and a truncated hRFC resulting from a 987-bp deletion (15, 16). Point substitutions in the hRFC coding sequence have been described for hRFC cDNA clones isolated from different laboratories (see above), and point- and frame-shift mutations have been reported in transport-impaired cells selected in vitro for antifolate resistance (17, 18, 19, 20). However, the relevance of these assorted sequence alterations to patient specimens is uncertain.

With the completion of the sequencing of the human genome (21, 22), there has been a surge of interest in the possible roles of gene variants in human disease, and potential opportunities for therapeutic intervention on this basis (23, 24). From the National Center for Biotechnology Information database, seven potential SNPs have been tentatively identified in hRFC, including three [positions 80 (G/A), 246 (C/G), and 696 (C/T)] in the hRFC coding sequence. Only the alteration at position 80 results in an amino acid substitution (i.e., Arg/His²⁷). A recent study implied an effect of G/A80 in combination with C/T677 in 5,10-methylene tetrahydrofolate reductase on plasma folate levels and homocysteine pools (25); however, the effects of G/A80 on hRFC transport function were not addressed.

On these bases, we began a series of experiments to confirm the presence of putative hRFC sequence variants in leukemia blasts obtained from children and in normal (non-leukemia) specimens. For childhood ALL, we reasoned that the existence of such sequence variants might result in differences in the cellular uptake of MTX in ALL blasts, thus contributing to interpatient variations in response to chemotherapy with MTX. In normal tissues, differences in membrane transport of reduced folate cofactors could result in downstream effects on folate-dependent anabolic pathways so as to contribute to interpatient differences in susceptibilities to cardiovascular disease, fetal abnormalities, and/or cancer. In this report, we document the occurrence of high-frequency SNPs in leukemia and non-leukemia patient specimens and compare the functional properties of Arg²⁷-hRFC to His²⁷-hRFC resulting from a G to A SNP at position 80.

[3′, 5′, 7-³H]MTX (20 Ci/mmol) and [3′, 5′, 7, 9-³H] (6S) 5-CHO-H₄PteGlu (17 Ci/mmol) were purchased from Moravek Biochemicals. Unlabeled MTX and (6R,S) 5-CHO-H₄PteGlu were provided by the Drug Development Branch, National Cancer Institute, Bethesda, MD. Both labeled and unlabeled MTX were purified by HPLC before use (26). Several antifolate drugs were provided as follows: 10-ethyl-10-deazaaminopterin (CIBA-GEIGY Corp.; Ref. 27), GW184389 (GlaxoWellcome Pharmaceuticals; Ref. 28); Tomudex (Ann Jackman, Institute of Cancer Research, Surrey, England; Ref. 29), and (6R,S) dideaza-5,6,7,8-tetrahydrofolate (Lometrexol or DDATHF; Lily Research Laboratories; Ref. 30).

Restriction and modifying enzymes were obtained from Promega and New England Biolabs. Synthetic oligonucleotides were obtained from Genosys Biotechnologies, Inc., and Life Technologies, Inc. Tissue culture reagents and supplies were purchased from assorted vendors with the exception of fetal bovine serum and iron-supplemented calf serum, which were from Life Technologies, Inc., and Hyclone Laboratories, Inc., respectively.

The wild-type K562 erythroleukemia cell line was obtained from American Type Culture Collection (Rockville, MD). Cells were maintained in RPMI 1640 containing 10% heat-inactivated iron-supplemented calf serum, 2 mm l-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin in a humidified atmosphere at 37°C in the presence of 5% CO₂/95% air. The transport-deficient K500E subline was selected from wild-type K562 cells by cloning in soft agar in the presence of 500 nm MTX (31). K500E cells were transfected with the wild-type Arg²⁷-hRFC cDNA (KS43) to generate the K43-6 subline, as described previously (31).

For cytotoxicity determinations with various antifolates (MTX, Tomudex, and GW1843U89), cells were cultured in 24-well dishes at 20,000 cells/ml in 2 ml of complete RPMI 1640 containing 10% dialyzed fetal bovine serum. Cells were counted with a hemacytometer after 120 h. The IC₅₀s were calculated as the concentrations of antifolate necessary to inhibit growth by 50% compared with control cells grown under identical conditions, except that the inhibitor was omitted. For assays of (6R,S) 5-CHO-H₄PteGlu growth requirements, cells were depleted of folates by culturing in folate-free RPMI 1640 (Life Technologies, Inc.) and 10% dialyzed serum containing adenosine (60 μm) and thymidine (10 μm), for 2 weeks. Growth was assayed after 120 h after culturing in complete folate-free medium without nucleosides but containing various concentrations of (6R,S) 5-CHO-H₄PteGlu (0–500 nm).

The childhood acute leukemia specimens used for this study were obtained from the Children’s Hospital of Michigan (Detroit, MI) and the Pediatric Oncology Group (Chicago, IL) and are those used in our previous studies of MTX response and resistance in pediatric ALL (32, 33). Patient samples were obtained at diagnosis after informed consent, and in accordance with protocols approved by the Committee on Investigation Involving Human Subjects at Wayne State University. Blasts were purified previously by standard Ficoll Hypaque density centrifugation. Total RNAs were prepared from 1–5 × 10⁶ leukemia cells using the RNeasy mini-kit (Qiagen). cDNAs including the complete hRFC open reading frame (∼1.8 kb) were prepared by reverse transcription with Superscript II (Life Technologies, Inc.) and oligo(dT) primer and PCR amplified with Taq polymerase (Promega) using PCR primers based on 5′- and 3′-untranslated region sequences (P8, 5′-CAGTGTCACCTTCGTCCCCT CCG-3′, and P10, 5′-CACCACCTC TTCCAGCAACAAA-3′). PCR conditions were 94°C for 1 min (1 cycle) and then 94°C for 30 s, 63°C for 45 s, and 72°C for 2 min (35 cycles) and ending with 72°C for 7 min (1 cycle). PCR products were subcloned into pGEM-T-Easy plasmid and five to seven separate clones were completely sequenced (see below) with a combination of universal and gene-specific primers. For amplifying the exon 3 sequence only, reverse transcription was performed with random hexamers and a RT-PCR kit from Perkin-Elmer, and the cDNAs were amplified with P8 and JW-RFC1 (5′-GGGTGATGAAGC TCTCCCCTGG-3′) primers [94°C for 5 min (1 cycle), 94°C for 15 s, 63°C for 45 s, and 72°C for 45 s (35 cycles), ending with 72°C for 7 min (1 cycle)]. PCR reactions were electrophoresed on a 2% LE agarose gel, and the amplicons were isolated from the gel with the Concert Rapid Gel Extraction kit (Life Technologies, Inc.). DNA sequences were obtained by automated DNA sequencing using the P8 primer and BigDye Terminator cycle sequencing reactions (Perkin-Elmer) at the DNA sequencing core of the Center of Molecular Medicine and Genetics (Wayne State University, Detroit, MI) and the Western Kentucky University Biotechnology Center (Bowling Green, KY).

gDNAs were isolated from 3–5 × 10⁶ ALL blasts using the DNA Isolation Kit from Puregene (Gentra Systems, Inc.). gDNAs from non-leukemic adult individuals were prepared from peripheral blood samples and provided by the Children’s Cancer Institute Australia for Medical Research. Exons 3–7 were amplified from gDNAs (200 ng) with Taq polymerase (Promega) over 35 cycles [94°C for 5 min (1 cycle), 94°C for 15 s, 61–63°C for 45 s, and 72°C for 45 s (35 cycles), ending with 72°C for 7 min (1 cycle)]. The primers are summarized in Table 1. GC-rich reagent (Roche) was included in the amplifications from gDNAs. The amplicons were isolated and directly sequenced with the upstream primers by automated DNA sequencing.

For PCR-RFLP screening for the presence of a CATG insertion at position 191 in hRFC transcripts, RNAs were reverse transcribed and PCR amplified with P8 and P7 (5′-GCCAGCGAGATGTAGTTGAGCGT-3′) primers. The PCR products were subcloned into pGem-T-Easy plasmid (Promega). For each patient, 10–11 bacterial colonies were PCR amplified and digested with Nsp1, and products were electrophoresed on a 2% LE agarose gel with ethidium bromide staining. By this approach, Nsp1 digestion of wild-type sequence (i.e., no CATG) would generate 215- and 403-bp fragments, whereas the presence of CATG-191 would result in 240-, 167-, and 215-bp restriction fragments.

“SOE” PCR (34) was used generate the His²⁷- (A80) hRFC cDNA, using Arg²⁷- (G80) hRFC cDNA (KS43) in pCDNA3 as template (13). The primary PCR step was designed to amplify two separate products for the secondary PCR step and used two primer sets: (a) R27H, 5′-CTCCGGTCCTGGCGGCACCTC- GTGTGCTACCTT-3′ and reHA-7, 5′-CTCGTTCCACAGGA- TGTGCAC-3′; and (b) T7 and reR27H, 5′-AAGGTAGCACACGAGGTGCCGCCAGGACCGGAG-3′. The R27H/reR27H primers were completely complementary and included the desired A80 mutation. PCR conditions for the primary amplification were 95°C for 30 s, 56°C for 30 s, and 72°C for 1 min (30 cycles). Secondary PCR was carried out under identical conditions with the T7/reHA-7 primer set and the primary PCR products as template. The amplicon was isolated, digested with BbrPI and NotI, and ligated into BbrPI and NotI-digested KS43 hRFC cDNA in pCDNA3. The presence of A80 in the His²⁷-hRFC cDNA construct was confirmed by automated DNA sequencing.

His²⁷-hRFC cDNA in pCDNA3 was transfected into transport-impaired K500E cells with lipofectin, as described previously (31). G418-resistant clones were isolated, expanded and screened for immunoreactive hRFC protein on Western blots and for [³H]MTX uptake.

Plasma membranes were prepared by differential centrifugation as described earlier (35). Membrane proteins were electrophoresed on 7.5% gels in the presence of SDS (36) and electroblotted onto polyvinylidene difluoride membranes (37) for detection with protein A-purified hRFC antibody (20) and enhanced chemiluminescence (Roche). Densitometry was performed on a Kodak Digital Science Image Station 440CF.

Initial [³H]MTX or [³H](6S) 5-CHO-H₄PteGlu uptake was measured over 180 s at 1 μm, as described previously (13, 20, 31). Levels of intracellular radioactivity were expressed as pmol/mg protein, calculated from direct measurements of radioactivity and protein contents of the cell homogenates. Protein assays were by the method of Lowry et al. (38). Kinetic constants (K_t, V_max, K_i) were calculated from Lineweaver-Burk and Dixon plots as in our previous reports (13, 20, 31).

By sequencing multiple (five to seven) full-length hRFC cDNA clones amplified from each of four B-precursor ALL specimens, we identified three distinct and reproducible sequence alterations. (a) A CATG insertion was detected at position 191 (relative to the ATG translation start at position 1) in one or two clones from all four ALL specimens. This 4-bp insertion is identical to that previously described in MTX-resistant CCRF-CEM cells (CEM/MTX-1; Ref. 20) and occurs at the donor splice junction for exon 3, suggesting its origin by alternative splicing. The result of the frame shift is the synthesis of a M_r ∼48,000 protein with mostly nonsense sequence (20). (b) For all four lymphoblast specimens, cDNAs were identified with both a G (CGC) and an A (CAC) at position 80 (i.e., codon 27). (c) A C to T transition at position 696 (codon 232) was also identified in these cDNAs. The former would result in a substitution of His for Arg²⁷, whereas the latter would be phenotypically silent (i.e., Pro→Pro).

An RT-PCR/RFLP strategy was developed for screening total RNAs for the CATG insertion at position 191, using RNAs obtained from 16 ALL specimens (10 B-precursor and 6 T-ALL). After the initial RT-PCR, the amplicons were subcloned and multiple (10 or 11) clones for each sample were screened for the presence of a novel Nsp1 restriction site generated by the CATG-191 insertion, resulting in a unique pattern of restriction fragments (215 and 403 bp in wild type or 240, 167, and 215 bp with CATG insertion; Fig. 1,A). Of the 16 ALL specimens tested, 10 ALL specimens expressed transcripts with CATG-191 at frequencies ranging from 10 to 60% (Fig. 1 B). Thus, the CATG-191 phenotype, previously described for CEM/MTX-1 cells in vitro (20), is indeed clinically relevant. At high frequencies, the presence of CATG-191 could result in significantly decreased levels of functional hRFC protein and levels of MTX transport despite high levels of hRFC transcripts. Although we were unable to establish any direct correlation between levels of MTX uptake and frequency of CATG-191 among our patient specimens (not shown), for the B-precursor ALL patient with the highest frequency of CATG-191 (60%), there was a strikingly disproportionate increase in the level of hRFC transcripts relative to MTX uptake (33).

To assess the frequency of a G/A80 transition in the hRFC coding sequence, we screened 54 gDNAs prepared from ALL specimens (34 BP-ALL and 20 T-ALL). gDNAs were PCR amplified with specific primers for exon 3 (includes the ATG translation start site), and the 194 bp amplicons were sequenced by automated sequencing. For comparison, 51 gDNAs from normal (non-leukemia) individuals were screened for G/A80, as well. Among these 105 leukemia and non-leukemia amplifications of exon 3, the only alteration detected was G/A80. The allelic frequencies for G/A80, based on the Hardy-Weinberg equation, are summarized in Table 2 and are virtually identical for the non-leukemia (42.2% CGC and 57.8% CAC) and leukemia gDNAs (40.7% CGC and 59.3% CAC). These results confirm that G/A80 represents a high-frequency SNP in the hRFC coding sequence.

For 10 of the ALL patients, including the 4 originally screened for hRFC mutations, total RNAs were reverse transcribed and PCR amplified for sequencing. In this analysis, the allelic distributions for codon 27 were in near-perfect agreement with those obtained by genomic sequencing and included 1 of 10 CGC/CGC (Arg²⁷), 6 of 10 CGC/CAC (Arg²⁷/His²⁷), and 3 of 10 CAC/CAC (His²⁷). The allelic frequencies for the cDNAs are 40% CGC and 60% CAC. Thus, both the CGC and CAC alleles are transcribed equally well.

The allelic frequencies obtained by direct sequencing of 105 gDNA specimens are distinctly different from those previously reported by PCR-RFLP analysis with CfoI (25) and in our own experience using DraIII or HaeII digestions to distinguish the CGC allele (DraIII-insensitive and HaeII-sensitive) from the CAC allele (DraIII sensitive and HaeII insensitive) in gDNAs from normal and leukemia patients (39), including many of those characterized by sequencing. The allelic frequencies by PCR amplification-DraIII digestions were 60% CGC and 40% CAC (not shown), whereas those by HaeII digestion were 56.6% CGC and 43.4% CAC, a striking contrast from those obtained by direct sequencing. These data suggest that the use of RFLP analysis to identify and quantitate this hRFC polymorphism can lead to an appreciable level of inaccuracy.

The demonstration of a frequent G/A variant at codon 27 of the hRFC coding sequence among 105 individuals led us to expand the scope of our search for high frequency polymorphisms that may occur in exons 4 through 7. Coding exons 4–7 were PCR amplified from 53 gDNAs (47 ALL and 6 non-ALL) and directly sequenced. The sequences for each exon were then compared with wild-type hRFC coding sequence (GenBank accession no. U19720). For an additional three specimens (two ALL and one non-ALL) exons 3, 4, 5, and 7 (i.e., not exon 6) were also sequenced; for an additional 6 non-ALL specimens, sequence data were obtained for exons 3–6 (i.e., not exon 7). Similar to our cDNA data (see above), C/T696 was identified as a frequent polymorphism that had an allelic frequency nearly equal among the 62 gDNAs (30 CCT and 32 CCC) sequenced. G/A80 and C/T696 did not cosegregate in our patients, indicating there is no linkage between these two SNPs. The only other alteration detected in 62 gDNAs was a heterozygous C/A transversion at position 1242 in two ALL specimens, resulting in a silent mutation (i.e., both ATC and ATA at codon 414 encode isoleucine).

From topology models for hRFC, Arg²⁷ is predicted to lie in the first transmembrane domain, a region implicated as important to carrier function (17, 19, 40, 41, 42). To directly compare the transport properties of His²⁷-hRFC with Arg²⁷-hRFC, we transfected the His²⁷-hRFC cDNA into hRFC-null K500E cells, exactly as in our earlier study with the Arg²⁷-hRFC-expressing K43-6 cells (31). Two stable clones (designated R27H2-4 and R27H2-8) that expressed moderate levels of His²⁷-hRFC protein on Western blots (only slightly lower than for the K43-6 cells) were isolated and characterized (Fig. 2 A).

In transport assays of [³H]MTX (1 μm) and [³H](6S) 5-CHO-H₄PteGlu (1 μm), uptake rates over 180 s in K43-6 (Arg²⁷-hRFC) exceeded that for the His²⁷-hRFC-expressing R27H2-4/R27H2-8 sublines by ∼39 and 31%, respectively (Fig. 2 B).

Relative uptake rates for the KS43-6 cells and the two His²⁷-hRFC-expressing lines were identical when uptake data were normalized for hRFC protein expression (Fig. 2 C). However, these normalized values may be slightly biased, because hRFC expression and transport are not always proportional in transfected cells (31).

Transport kinetic parameters (i.e., V_max and K_t) for MTX were, likewise, nearly identical between K43-6 cells, and the R27H2-4 and -8 sublines (Table 3). Furthermore, there were, at most, only minor (∼2-fold) differences in the K_is for folate/antifolate transport substrates between the Arg²⁷-hRFC- and His²⁷-hRFC-expressing cells (Table 3). Finally, there were no significant differences between K43-6 and R27H2-4 cells in their sensitivities to growth inhibition by MTX (IC₅₀s of 54 and 56 nm, respectively), Tomudex (IC₅₀s of 4.1 nm for both lines), or GW1843U89 (IC₅₀s of 1.1 and 1.2 nm, respectively), nor were there differences in their (6R,S) 5-CHO-H₄PteGlu growth requirements (50% maximal growth at 1.9 and 1.6 nm, respectively). Collectively, these results demonstrate that the His²⁷-hRFC protein exhibits, at most, only minor functional differences from Arg²⁷-hRFC in terms of substrate affinities and/or transport efficiencies.

A major goal of human genetics is to identify the roles of common genetic variants that result in altered susceptibilities to disease. This requires identifying gene variations in human populations and assembling an extensive catalogue of SNPs and characterizing their functional significance and associations with particular diseases. The identification of SNPs that result in altered proteins can also provide molecular insights into the bases for interpatient variability in therapeutic responses and toxicities to drugs (23, 24). Given the narrow therapeutic window for most cancer therapies, this is particularly important for the clinical management of cancer (43). By identifying specific gene polymorphisms involved in altered drug metabolism or transport, pharmacogenetic screens can be developed to determine whether particular groups of patients may be predisposed to unacceptable drug toxicities and/or poor antitumor responses for specific classes of anticancer drugs. This potentially could lead to more individualized therapies for cancer.

From the critical role of hRFC in membrane transport of folates for cellular proliferation and tissue regeneration and of antifolates for cancer therapy, interest in hRFC gene variants is particularly acute. In this report, we have identified four sequence alterations in hRFC in patient specimens. These include a four bp (CATG) frameshift mutation at position 191, apparently resulting from aberrant mRNA splicing, a silent C to T alteration at position 696 in exon 4, a silent C to A transversion at position 1242 in exon 6, and a high-frequency (G to A) SNP at position 80 in exon 3 that results in a substitution of Arg²⁷ by His²⁷.

The high frequency CATG insertion would result in a nonfunctional carrier and could possibly contribute to low levels of hRFC-protein and -MTX transport in ALL lymphoblasts from patients. The allelic frequencies of CAC and CGC at codon 27 were ∼60 and 40%, respectively, among the 105 patients in our analysis. The change of a strongly basic amino acid (Arg) to a weak base (His) in a region of the carrier documented to influence folate substrate binding and rates of uptake (19, 40, 41, 42) might be expected to alter hRFC transport properties. However, by directly comparing the transport properties of Arg²⁷-hRFC to His²⁷-hRFC in stable K562 transfectants, no significant differences in the uptake rates of either MTX or 5-CHO-H₄PteGlu were observed, and only minor differences were calculated in the relative affinities for an assortment of transport substrates. Collectively, these data strongly argue for a lack of major functional differences between the Arg²⁷- and His²⁷-hRFCs for reduced folate cofactors and for various antifolates used in cancer chemotherapy.

We gratefully acknowledge the contributions of Kenneth Bark and Dawn Clifford in certain of the formative experiments that led to this study. We thank the Pediatric Oncology Group for providing childhood ALL specimens for our ongoing studies of MTX response and resistance in pediatric ALL.