Eur. J. Biochem. 270, 4469–4477 (2003) FEBS2003 doi:10.1046/j.1432-1033.2003.03839.x

Disruption of transport activity in a D93H mutant thiamine transporter 1, from a Rogers Syndrome family

Dana Baron1, Yehuda G. Assaraf2, Stavit Drori2 and Ami Aronheim1 1Department of Molecular Genetics, The Rappaport Institute for Research in the Medical Sciences and the B. Rappaport Faculty of Medicine, and 2Department of Biology, The Technion-Israel Institute of Technology, Haifa, Israel

Rogers syndrome is an autosomal recessive disorder result- transport null cells of a homologous D88H in ing in megaloblastic , mellitus, and sensori- the hRFC did not result in restoration of MTX trans- neural deafness. The associated with this disease port activity, thereby suggesting that D88 is an essential encodes for thiamine transporter 1 (THTR1), a member of residue for MTX transport activity. These results suggest the SLC19 including THTR2 and the that the D93H mutation does not interfere with trans- reduced carrier (RFC). Using transient transfections porter expression, glycosylation and plasma membrane into NIH3T3 cells of a D93H mutant THTR1 derived targeting. However, the substitution of this negatively from a Rogers syndrome family, we determined the charged (Asp93) by a positively charged expression, post-translational modification, plasma mem- residue (His) in an extremely conserved region (the bor- brane targeting and thiamine transport activity. We also der of transmembrane domain 2/intracellular loop 2) in explored the impact on methotrexate (MTX) transport the SLC19 family, presumably inflicts deleterious struc- activity of a homologous missense D88H mutation in the tural alterations that abolish thiamine binding and/or human RFC, a close homologue of THTR1. Western blot translocation. Hence, this functional characterization analysis revealed that the D93H mutant THTR1 was nor- of the D93H mutation provides a molecular basis for mally expressed and underwent a complete N-glycosylation. Rogers syndrome. However, while this mutant THTR1 was targeted to the Keywords: Rogers syndrome; thiamine; transporter; muta- plasma membrane, it was completely devoid of thiamine tions. transport activity. Consistently, introduction into MTX

Thiamine responsive megaloblastic anemia (TRMA) also progression, insulin secretion gradually declines to levels known as Rogers syndrome [1], is an early onset autosomal that require insulin therapy. On the other hand, in all recessive disorder that was described in only 20 families TRMA patients, anemia is fully reversed following the all over the world. Patients with Rogers syndrome are administration of high doses of thiamine [5]. diagnosed by the occurrence of multiple clinical manifesta- Fibroblasts isolated from Rogers syndrome patients, tions including: megaloblastic anemia, diabetes mellitus and display only  5–10% of thiamine uptake as compared with sensorineural deafness [Online Mendelian Inheritance in fibroblasts derived from healthy individuals. This uptake Man (OMIM) 249270, http://www.ncbi.nlm.nih.gov/ apparently occurs via a low-affinity, unsaturable thiamine Omim/]. The administration of high doses of thiamine or route [6]. These results are consistent with an entry of alternatively the use of medications such as sulfonylureas thiamine via passive diffusion and may explain the fact that recovers insulin secretion [2–4]. However, with disease high doses of thiamine are able to induce minimally sufficient thiamine levels in Rogers syndrome patients. The gene responsible for Rogers syndrome, SLC19A2, Correspondence to A. Aronheim, Department of Molecular Genetics, was identified by positional cloning [7–9]. SLC19A2 encodes The B. Rappaport Faculty of Medicine, 7th Efron St. Bat-Galim, for a thiamine transporter (THTR1), a member of the solute The Technion-Israel Institute of Technology, Haifa 31096, Israel. carrier family. This family includes SLC19A1 that encodes Fax: + 972 4 8295225, Tel.: + 972 4 8295226, for the reduced folate carrier (RFC) and SLC19A3 that E-mail: [email protected] encodes for , THTR2 [10]. THTR1 Y. G. Assaraf, Department of Biology, The Technion-Israel contains 497-amino acids and has 12 putative transmem- Institute of Technology, Haifa 32000, Israel. brane domains (TMD) [7–9]. THTR1 is responsible for + Fax: + 972 4 8225153, Tel.: + 972 4 8293744, thiamine transport in a Na independent manner and is + E-mail: [email protected] stimulated by H efflux gradient [11]. Abbreviations: DMEM, Dulbecco’s modified Eagle’s medium; MTX, Although all Rogers syndrome patients exhibit similar Methotrexate; THTR1, thiamine transporter 1; TRMA, thiamine clinical manifestations, multiple were identified responsive megaloblastic anemia; hRFC, human reduced folate resulting in premature translation termination due to base carrier; TMD, transmembrane domain. pair insertions or deletions which result in frame-shift. In (Received 23 July 2003, revised 11 September 2003, addition, three missense mutations were described resulting accepted 18 September 2003) in single amino-acid substitutions. Recently we have initiated 4470 D. Baron et al.(Eur. J. Biochem. 270) FEBS2003 studies that are aimed at characterization of the impact BRL Inc.), whereas XtremeGENE reagent (Roche Inc.) was that these missense THTR1 mutations have on the expres- used for Swiss 3T3 cells. Typically, for NIH3T3 cell sion, post-translational modification, plasma membrane transfections, 6 lg of plasmid DNA were used with 9 lL targeting and thiamine transport activity. In a recent paper, of the LipofectAMINE reagent whereas for Swiss 3T3 cells, we described the molecular consequences of a single amino- 2 lg of plasmid DNA were used along with 10 lLof acid substitution identified in an Italian Rogers syndrome XtremeGENE reagent according to the instructions of the family harboring a glycine to aspartate substitution at manufacturer. position 172 of the THTR1 [12]. Although this mutant THTR1 gene is properly transcribed and translated in vitro, Cell extract no could be detected in NIH3T3 cells transfected with the mutant G172D THTR1 cultured at 37 C. Shifting Whole cell extracts were isolated as previously described transfected cells to 28 C resulted in a substantial expression [17]. of the mutant G172D THTR1 protein thereby allowing for biochemical and functional characterization [12]. The Isolation of THTR1 cDNA G172D mutant THTR1 failed to undergo a complete N-glycosylation, on the glutamine 63 acceptor which is THTR1 cDNA was isolated as previously described [12]. found to be conserved in both THTR1 and THTR2 [12] resulting in its cytoplasmic retention in the endoplasmic Mammalian expression plasmids reticulum. This is in contrast to the other members of the solute carrier family, including the hRFC, in which pCan-THTR1 expression plasmids; pCan is a pcDNA- N-glycosylation plays neither a role in its plasma membrane based (Invitrogen Inc.) mammalian expression vector localization nor in MTX transport activity [13]. Here we expressing downstream of a Myc-epitope tag. describe the characterization of another mutation identified Protein expression is under the control of cytomegalovirus in a French Rogers family harboring an aspartate to (CMV) immediate early promoter. An oligonucleotide histidine substitution at position 93 (D93H) in THTR1 [14]. based PCR approach was used to amplify the human Unlike in a previous analysis of the D93H mutation [15], THTR1 cDNA that was fused, in frame, with the Myc- here we show that the mutant protein was efficiently epitope tag using 5¢-EcoRI and 3¢-XhoI restriction sites. The expressed both in vitro and in vivo and upon immunoflu- Myc tag and polylinker regions are composed of the orescence studies this protein was properly targeted to the following 17 amino acids: MVQKLISEEDLRIHRCR. For plasma membrane. Uptake studies with the D93H mutant fusion of Myc tag to the C-terminus of the THTR1, an protein revealed no [3H]thiamine transport activity. Import- oligonucleotide-based PCR approach was used to amplify antly, this D93 region in THTR1 and the homologous D88 the human THTR1 cDNA that was fused in frame, using vicinity in the hRFC are extremely conserved among all 5¢-HindIII and 3¢-HindIII restriction sites with the Myc tag members of the solute carrier family (SLC19A1, SLC19A2 in the 3¢-end. The correct orientation was verified by XbaI and SLC19A3). Indeed, a hRFC harboring the same D88H digestion generating a 1100-bp fragment due to a unique mutation displayed no methotrexate transport activity. This XbaI site at position 377 within THTR1 cDNA and at is consistent with a previous report in which a D88V mutant the 3¢-end of the pCan expression vector’s multiple cloning RFC showed no transport activity [16]. We suggest that this site. site (i.e. D93 in THTR1 and D88 in hRFC) represents a highly conserved region, which is absolutely essential for the GFP expression plasmid structure and function of all members of the solute carrier family. Hence, the present study constitutes the first This plasmid encodes for the green fluorescent protein under demonstration of a THTR1 mutation that does not interfere the control of the CMV promoter (Life Technologies Inc.). with transporter expression, N-linked glycosylation and plasma membrane targeting but completely disrupts thi- Site-directed mutagenesis amine transport activity. The corresponding oligonucleotides were designed to yield Experimental procedures an aspartate 93 to histidine substitution and to generate a diagnostic AatII restriction site. Site directed mutagenesis was performed using the QuikChange XL site directed Cell cultures mutagenesis kit according to the instructions of the NIH3T3 and Swiss 3T3 cells were cultured under mono- manufacturer (Stratagene Inc.) Primers used for muta- layer conditions in a humidified atmosphere of 5% CO2 at genesis of THTR1 (D93H) were: Upstream oligo: 37 C. Cells were grown in Dulbecco’s modified Eagle’s 5¢-CCTGTGTTCCTTGCCACACACTACCTCCGTTA medium (DMEM) containing 10% fetal bovine serum TAAACC-3¢ and downstream oligo: 5¢-GGTTTATAACG )1 (FBS), 2 mML-glutamine and 100 unitsÆmL penicillin and GAGGTAGTGTGTGGCAAGGAACACAGG-3¢. DNA 100 lgÆmL)1 streptomycin. sequencing was used to verify that THTR1 was free of any other mutations. The primers for the introduction of the D88H mutation in the hRFC were: Upstream oligo: Transient and stable transfections 5¢-GTGTTCCTGCTCACCCACTACCTGCGCTACAC Transient NIH3T3 cell transfections were carried out in G-3¢ and downstream oligo: 5¢-CGTGTAGCGCAGGT 60 mm plates using the LipofectAMINE reagent (Gibco- AGTGGGTGAGCAGGAACAC-3¢. FEBS2003 Lack of thiamine transport in D93H THTR1 ( Eur. J. Biochem. 270) 4471

sham)for20minat37C as previously described [11]. Cells In vitro translation studies were then washed twice with 10 mL of ice-cold NaCl/Pi and Coupled in vitro transcription and translation were lysed with 0.2 mL of 0.2 M NaOH, 0.1% SDS at 65 Cfor performed using the T7 TNT quick transcription and 30 min [18]. The lysate was used to determine intracellular translation kit according to the instructions of the manu- radioactivity in a liquid scintillation spectrometer. facturer (Promega Inc.). Stable transfections with wild-type and D88H mutant Tunicamycin treatment hRFC cDNA Tunicamycin (Sigma Cat. # T7765) was dissolved in A wild-type hRFC cDNA clone containing the entire dimethyl sulfoxide to yield a stock solution of 10 mgÆmL)1. coding region as well as a D88 mutant RFC cDNA This N-glycosylation inhibitor was applied to monolayer prepared by site-directed mutagenesis (QuickChange XL cells at a final concentration of 10 lgÆmL)1 for 24 h prior kit, Stratagene) were directionally cloned into pcDNA3 at to cell harvesting and immunofluorescent staining. the BamHI/XhoI site (Ivitrogen). Mouse leukemia cells, L1210 and their MTX transport null cells, MTXrA, were used for transfections. Exponentially growing cells (2 · 107) Peptide N-glycosidase F (PNGaseF) treatment were harvested by centrifugation and stably transfected by Aliquots of cell extracts (50 lg) were incubated with a electroporation (1000 lF, 234 V) with 10 lg of an expres- denaturing buffer containing sodium phosphate pH 7.5, sion vector (pcDNA3-hRFC1) harboring the normal hRFC 0.5% SDS and 1% 2-mercaptoethanol in the presence or or the D88H mutant hRFC cDNA. Following 24 h of ) absence of 2000 U of PNGaseF (NEB Inc.) for 45 min at growth at 37 C, cells were exposed to 600–750 lgÆmL 1 37 C. active G-418 (Calbiochem-Novabiochem, San Diego, CA). Cells were then distributed into 96-well plates at  2 · 104)4 · 104 cells per well and after 10–20 days of SDS/PAGE and Western blotting incubation at 37 C, individual G418-resistant clones were Aliquots of protein extracts (50 lg) or 5 lL of a pro- picked and expanded for further studies. Only clones grammed rabbit reticulocyte lysate were resolved on a 10% expressing comparable RFC mRNA levels were used. SDS/PAGE followed by immobilization onto Hybond nitrocellulose membrane (Amersham Inc.). Western blots Assay of MTX transport were then blocked with a buffer containing: 10% low fat milk in NaCl/Pi for 1 h and incubated with mouse anti-Myc Antifolate transport measurements were performed as 9E11 monoclonal IgG (Babco Ltd) followed by a secondary previously described [19]. Briefly, exponentially growing goat anti-(mouse Ig) Ig conjugated to horse radish peroxi- cells were harvested, washed twice with Hepes/NaCl/Pi at ) dase (HRP). Detection was performed with a chemilumi- pH 7.4 and suspended at a density of 2 · 107 cellsÆmL 1 at nescent substrate (SuperSignal, Pierce Inc.) and membranes 37 C. The uptake of [3H]MTX (specific activity 0.5 Ci were exposed to X-ray films and autoradiogram with linear mmol)1) was measured at an extracellular concentration of exposures were obtained. 2 lM in 1 mL of cell suspension at 37 C. Influx measure- ments were performed at 0.5, 1 and 2 min during which MTX transport rates were linear. Uptake was terminated by Immunofluorescence adding 10 mL of ice-cold Hepes/NaCl/Pi, following which Swiss 3T3 cells were grown on 22-mm coverslips and centrifugation and cell wash with another 10 mL ice-cold transfected as described above. Forty-eight hours after Hepes/NaCl/Pi was performed. The final cell pellet was transfection, cells were fixed and immunostained as des- processed for liquid scintillation counting. cribed [12]. Each coverslip contained at least 5 · 104 cells and the efficiency of transfection varied between 10 and Results 20%. About 90% of transfected cells exhibited red and green staining and all of these cells displayed the staining Expression of THTR1 and D93H THTR1 in NIH3T3 cells presented. At least 100 transfected cells per slide from each transfection were examined. The immunofluorescence ana- A Rogers syndrome patient from a French family was lysis, which was reproduced in five independent transfection found to harbor an aspartate 93 to histidine substitution experiments, revealed an identical staining pattern. (D93H) in THTR1. In order to study the effect of this mutation on THTR1 expression and function, we used site- directed mutagenesis to express D93H mutant THTR1 Assay of thiamine uptake protein fused to an N-terminal Myc tag epitope. NIH3T3 NIH3T3 cells transiently transfected with wild-type and cells were transiently transfected with expression plasmids mutant D93H THTR1 and vector control were detached encoding for either the wild-type or the D93H mutant with NaCl/Pi containing 1 mM EDTA 2 h prior to thiamine THTR1. Cell lysates isolated from transfected cells were transport measurements. Cells were incubated in a 15-mL subjected to SDS/PAGE followed by Western blot analysis ) tube (Greiner Ltd) at a density of 3 · 105 cellsÆmL 1 in using anti-Myc 9E11 monoclonal Igs (Fig. 1A). A strong serum-free DMEM lacking thiamine. NIH3T3 cells were signal represented by a broadly migrating THTR1 protein centrifuged and resuspended in 100 lL of transport buffer with an average molecular mass of 68 kDa was detected in 3 )1 containing 0.6 lM of [ H]thiamine (30 CiÆmmol ;Amer- cell lysates derived from both the wild-type and D93H 4472 D. Baron et al.(Eur. J. Biochem. 270) FEBS2003

Fig. 1. Expression of D93H mutant THTR1 and wild-type THTR1 proteins in NIH3T3 cell transfectants. (A) NIH3T3 cells were transiently transfected either with wild-type or mutant THTR1 in which aspartate 93 was substituted by histidine (D93H). Cells transfected with an empty vector were used as a negative control (Vector). Total cell protein extracts were isolated and proteins (50 lg) were resolved on a 10% SDS/PAGE followed by Western blotting using an anti-Myc 9E11 monoclonal Ig and anti-(mouse IgG) Ig conjugated to HRP as a second antibody (Santa Cruz Inc.) following which chemiluminescence reaction was performed. Molecular mass (MW) markers (ProSieve BioWhittaker Molecular Applications Inc.) are given in kDa. (B) NIH3T3 cells were transiently transfected with wild-type or mutant D93H THTR1 constructs containing the Myc epitope tag in the C-terminus (THTR1-Myc, D93H-Myc). Wild-type THTR1 with the Myc tag in the N-terminus was used as a positive control. Proteins (50 lg) were resolved on a 10% SDS/PAGE and THTR1 expression was followed as described in A. (C and D) The indicated expression plasmids (with the Myc epitope tag at the C-terminus) were used in a coupled in vitro transcription-translation system using the T7 RNA polymerase in the presence of [35S]methionine. Proteins were then resolved on 10% SDS/PAGE followed by either autoradiography (C) or Western blotting probed with anti-cMyc Ig as described above (D). mutant THTR1 transfected cells. As expected, no THTR1 antibody (Fig. 1D). Both wild-type THTR1 and D93H was observed in cell lysates derived from vector transfected mutant THTR1 proteins were detected as sharp protein cells (Fig. 1A). To test the effect of the Myc epitope tag bands of 52 kDa. These results suggest that the Myc tag fused to the N-terminal of the protein we also designed fusion in the C-terminus of THTR1 protein may interfere expression plasmids encoding for THTR1 fused to a Myc with the cellular expression and/or stability of the protein. tag epitope at the C-terminus. Western blot analysis In view of these results, all subsequent experiments were performed with cell lysates derived from NIH3T3 cells performed using expression plasmids encoding for the transfected with expression plasmids encoding for either the N-terminal Myc tag fusions of THTR1. wild-type THTR1 or the D93H mutant THTR1 fused to Myc epitope at the C-terminus, did not reveal any The mutant D93H THTR1 is N-glycosylated expression when probed with the anti-Myc Ig (Fig. 1B). To confirm whether THTR1-Myc and D93H mutant The broad migration of the THTR1 proteins (Fig. 1A) THTR1-Myc can be properly transcribed and translated, suggested that the well expressed D93H mutant protein we used an in vitro system of coupled transcription and apparently undergoes a normal N-glycosylation. To test translation (rabbit reticulocyte lysate) in the presence of whether the slower migration of the D93H mutant THTR1 [35S]methionine. In vitro translated, radiolabeled proteins is indeed due to N-glycosylation, we treated transfected were then resolved on SDS/PAGE. Both wild-type THTR1- NIH3T3 cells with tunicamycin (10 lgÆmL)1), an estab- Myc and D93H mutant THTR1-Myc were efficiently lished N-glycosylation inhibitor [12]. Whole cell extracts transcribed and translated resulting in a sharp protein band derived from mutant D93H THTR1 transfected cells of 52 kDa (Fig. 1C), consistent with the core molecular treated with tunicamycin exhibited the same migration mass of the unglycosylated transporter protein, as previ- pattern on SDS/PAGE when compared to cells transfected ously described [12]. The small difference in the migration with the wild-type THTR1 after treatment with tunicamycin between the N-terminal and C-terminal Myc-fusions is due (Fig. 2, compare lane 2 and 5). In order to assess whether to the longer epitope tag of the latter. In order to confirm N-glycosylation is the post-translational modification that these constructs contained a Myc epitope that can be responsible for the markedly decreased electrophoretic recognized by the anti-Myc antibody, we resolved the mobility of the transfected wild-type and D93H mutant in vitro translated protein products on SDS/PAGE followed THTR1 proteins, we used peptide N-glycosidase F by Western blotting and probed them with the anti-Myc (PNGaseF). PNGaseF is an endoglycosidase, which FEBS2003 Lack of thiamine transport in D93H THTR1 ( Eur. J. Biochem. 270) 4473

THTR1 proteins (Fig. 3A). Wild-type and D93H mutant THTR1 proteins were stained with anti-Myc 9E11 mono- clonal Ig followed by CY3-conjugated goat anti-mouse IgG as a second antibody (Fig. 3B). Co-staining with both green (GFP) and red fluorescence (THTR1) is shown (Fig. 3C). Thus, colocalization is represented by a yellow stain. Cells transfected with THTR1 expression plasmids exhi- bit perinuclear-endoplasmic reticulum (ER) staining, as well as cytoplasmatic and plasma membrane staining as previ- ously reported (Fig. 3, THTR1) [12]. Similarly, the immu- nofluorescent localization pattern obtained with the D93H mutant THTR1 was indistinguishable from that observed Fig. 2. The N-glycosylation pattern of the wild-type and D93H mutant with the wild-type THTR1 (Fig. 3 compare D93H and THTR1proteinsinNIH3T3cells.NIH3T3 cells were transfected as wild-type THTR1). described in Fig. 1 legend. Cell lysates were prepared and proteins To test whether the glycosylation of the mutant D93H (50 lg) were resolved on a 10% SDS/PAGE and THTR1 expression THTR1 plays a role in the plasma membrane localization, was followed as described in Fig. 1 legend. Where indicated, cells were as compared with the wild-type THTR1, we treated )1 treated with tunicamycin (Tun.; 10 lgÆmL)1) for 24 h prior to cell transfected cells with tunicamycin (10 lgÆmL ), 24 h prior harvesting. Cell lysates (50 lg protein) were treated with PNGaseF to immunofluorescence analysis. As previously reported, where indicated and treated as described in experimental procedures treatment with tunicamycin resulted in a complete loss of and subjected to 10% SDS/PAGE followed by Western blotting using plasma membrane localization of the THTR1 protein an anti-cMyc monoclonal antibody. (Fig. 3, THTR1 ± tunicamycin) [12]. Cells transfected with the D93H mutant THTR1 and treated with tunicamycin displayed a similar lack of plasma membrane localization of cleaves the covalent bond between the innermost N-acetyl- the mutant transporter as well as a typical subcellular D-glucosamine and the acceptor asparagine, thereby resulting localization that was confined to the perinuclear ER in the complete removal of the N-glycan from glycosylated membrane (Fig. 3, D93H ± tunicamycin). proteins. Treatment of total cell extracts from transfected cells with PNGaseF resulted in an almost complete removal Transport in cells transfected with THTR1 or D93H THTR1 of the carbohydrate, thereby revealing a sharp 52 kDa protein band with a similar electrophoretic mobility for The fact that the D93H mutant THTR1 protein is both the core wild-type THTR1 protein and the in vitro expressed, fully glycosylated and sorted out to the plasma translated THTR1 or the mutant D93H THTR1 proteins membrane as does the wild-type THTR1 protein raised the (Fig. 2). Thus, these results strongly suggest that the slower question of the functionality (i.e. thiamine transport migration of both the wild-type and mutant D93H THTR1 activity) of this mutant THTR1. To test the ability of the proteins is largely due to an N-linked glycosylation. These mutant THTR1 to transport thiamine, we transiently results suggest that the D93H mutant THTR1 protein is transfected NIH3T3 cells with expression plasmids repre- efficiently transcribed, translated and apparently undergoes senting the empty vector as well as expression plasmids a complete N-glycosylation as compared to the wild-type encoding for either the wild-type THTR1 or the D93H THTR1 protein. mutant THTR1 and measured [3H]thiamine uptake. NIH3T3 cells transfected with an empty expression vector exhibited an endogenous background thiamine transport Mutant D93H THTR1 is targeted to the plasma membrane (Fig. 4). Expectantly, NIH3T3 cells transfected with an We have previously shown that only the properly folded expression vector harboring the wild-type THTR1 exhibited THTR1 protein is targeted to the plasma membrane and a 2.5-fold increase in thiamine uptake as compared with that only the fully glycosylated mature THTR1 protein cells transfected with the empty expression vector. In reaches its plasma membrane destination [12]. We deter- contrast, cells transfected with the D93H mutant THTR1 mined whether or not the D93H mutant THTR1 is indeed showed thiamine uptake reflecting the background levels properly glycosylated such that it will be properly trafficked that were obtained with cells transfected with the empty to the plasma membrane. We used immunofluorescence vector (Fig. 4) or untransfected cells (data not shown). analysis with Swiss 3T3 cells in order to explore THTR1 These results establish that the mutant D93H THTR1 protein localization. We employed Swiss 3T3 cells for the protein lacks thiamine transport activity. Furthermore, this immunofluorescence analysis as these cells are larger in size finding is in accord with Rogers syndrome patients carrying and exhibit a distinct cell morphology as compared with this D93H mutant protein which apparently lost thiamine NIH3T3 cells. Swiss 3T3 cells were cotransfected with transport activity. expression plasmids encoding for wild-type or D93H mutant THTR1 along with an expression plasmid encoding Aspartate 93 is conserved in all members for green fluorescence protein (GFP) (Fig. 3). The GFP of the solute carrier SLC19 family expression plasmid is expected to cosegregate with THTR1 or mutant D93H THTR1 expression plasmids. Therefore, Aspartate 93 is located in the border of TMD2 and the cells that exhibit green fluorescence in all cell compartments beginning of intracellular loop 2 that is composed of only assist in the localization of the wild-type or D93H mutant eight amino acids [11]. This region is extremely conserved in 4474 D. Baron et al.(Eur. J. Biochem. 270) FEBS2003

Fig. 3. Sub-cellular localization of D93H mutant THTR1 proteins byimmunofluorescence. Swiss 3T3 cells were cotransfected with a GFP expression plasmid, with the Myc-THTR1 or mutant D93H THTR1 expression plasmids as indicated. Cells were treated (where indicated) with 10 lgÆmL)1 tunicamycin for 24 h prior to cell fixation and staining with anti-Myc 9E11 monoclonal Ig and secondary goat-anti-(mouse Cy3 Ig) Ig. Cells were then analyzed using confocal microscopy. The localization of GFP alone (green, A) or CY3-conjugated goat-anti-(mouse Ig) Ig staining alone (red, B) and costaining (orange/yellow, Panel C) is shown. Orange/yellow stain indicates colocalization. Cell transfections and treatments were as indicated: Wild-type THTR1 (THTR) and mutant D93H THTR1 (D93H) either untreated or treated with tunicamycin (10 lgÆmL)1)(+tun.). FEBS2003 Lack of thiamine transport in D93H THTR1 ( Eur. J. Biochem. 270) 4475

3 Fig. 4. [ H]Thiamine uptake in NIH3T3 cell transfectants expressing Fig. 5. [3H]Methotrexate transport in cell transfectants expressing the 3 the wild-type and mutant D93H THTR1 proteins. [ H]Thiamine uptake wild-type and D88H mutant hRFC proteins. MTX transport null, was determined in NIH3T3 cells expressing the indicated proteins. Net MTXrA cells were stably transfected with pcDNA-based expression thiamine uptake was calculated by subtraction of the radioactivity vectors harboring the wild-type and D88H mutant RFC . Clonal obtained at 37 C with that observed at 4 C. Endogenous thiamine cell lines stably expressing comparable levels of the wild-type and transport in NIH3T3 was represented by cells transfected with an mutant RFC were used to determine [3H]MTX transport. Exponen- 3 3 empty vector. The cells were incubated with 0.6 lM [ H]thiamine for tially growing cells were incubated with 2 lM [ H]MTX for 30, 60 and 20 min in a transport buffer at pH 7.4 as described in Experimental 120 s at 37 C following which the initial rates of labeled radio MTX procedures. were determined as detailed in Experimental procedures. all members of the solute carrier SLC19 family including members of the SLC19 family, this D93H THTR1 mutation THTR1, THTR2 as well as RFC and across mammalian and D88H in the hRFC abolished transport for both species (i.e. human, mouse, and hamster) (Table 1). The thiamine and MTX, respectively, thereby, suggesting that substitution of aspartate by histidine reverses the net charge this domain plays an important functional role in substrate at this position and this may strongly affect transporter binding and/or translocation. function (Table 1). In order to test whether or not the conserved aspartic acid at position 93 is also essential for the transport function of the hRFC, a THTR1 and THTR2 Discussion homologue, we used site-directed mutagenesis to substitute Rogers syndrome, is a rare uni-gene autosomal recessive the corresponding aspartic acid at position 88 by histidine disease, currently identified in 20 families all over the in the hRFC. Methotrexate (i.e. an established transport world. Rogers syndrome patients suffer from constant r substrate of RFC) transport null MTX Amurineleukemia thiamine deficiency and display the three main disease cells were stably transfected with the vehicle vector plasmid manifestations including megaloblastic anemia, diabetes and expression plasmids encoding for either the wild-type mellitus and sensorineural deafness. Some of the non- or the D88H mutant RFC proteins. Northern blot analysis developmental manifestations in Rogers syndrome can be confirmed the similar levels of RFC mRNA in the overcome by treatment with high doses of thiamine thus transfected cells with both the wild-type and D88H mutant suggesting that alterations in thiamine transporter activity hRFC (data not shown). Cells were then tested for their may play a role in the pathogenesis. Consistently, Rogers 3 ability to take up [ H]methotrexate. RFC transfected cells syndrome patients were found to harbor mutations in the exhibited  16-fold higher MTX transport as compared to SLC19A2, encoding for THTR1 [7–9]. Different muta- nontransfected cells or vector transfected cells (Fig. 5). In tions in the THTR1 (SLC19A2) gene were described in all contrast, cells expressing the D88H mutant hRFC failed to Rogers syndrome patients [14]. There is no correlation exhibit MTX uptake that was above the poor background between the type of mutation identified and the severity of r levels present in the transport null MTX A cells and empty disease symptoms. We therefore initiated studies aimed at vector-transfected cells (Fig. 5). Based on the high conser- the characterization of missense mutations resulting in vation of TMD2 and intracellular loop 2 between the amino-acid substitutions. Recently we have described one of the three-missense mutations that occur in Rogers Table 1. Sequence comparison of various members of the solute carrier syndrome families [12]. The mutation was identified in an family. Amino-acid alignment of the aspartate 93 in human THTR1 Italian family, in which, glycine 172 is substituted by with various members of the SLC19 family. The conserved aspartate 93 aspartate thereby resulting in an apparently missfolded is indicated in bold. Light shading indicates the amino acids showing THTR1 protein, that fails to undergo complete glycosy- conservation in all members of the solute carrier family. lation, and is retained in the Golgi-ER compartment this resulted in the targeting of this mutant THTR1 protein to degradation [12]. We undertook the present study to determine the impact that the D93H THTR1 mutation from a French family with Rogers syndrome has on thiamine transport activity. During the early stages of our research, we were interested in selecting the appropriate Myc tag constructs (N- or C-terminal fusion) of the D93H mutant THTR1. We find here that insertion of a Myc-tag epitope at the N-terminus 4476 D. Baron et al.(Eur. J. Biochem. 270) FEBS2003 of the wild-type THTR1 retains normal expression, plasma kinase resulted in a spontaneous a-N-6 gluconoylation membrane trafficking, and thiamine transport activity, which led to a severe interference with its crystallization [26]. whereas no detectable expression could be detected in Using transient transfections and immunofluorescent NIH3T3 cells transfected with a THTR1 harboring a subcellular localization we find that the D93H mutant C-terminal Myc-tag. Apparently, insertion of a six-histidine THTR1 is well expressed, undergoes an apparently com- tag to the N-terminal end of the wild-type THTR1 did not plete N-linked glycosylation and is targeted to the plasma interfere with transporter expression and thiamine transport membrane. However, despite this apparently normal activity [15]. This latter result is consistent with recent expression, post-translational modification (i.e. glycosyla- studies that reported on the introduction of an N-terminal tion) and trafficking to the plasma membrane, the D93H Myc-tag to the human Na+-glucose [20] and mutant transporter did not exhibit any thiamine transport the [21] without interfering with their activity. These results establish that the D93H THTR1 expression and transport function. The human THTR1 is mutation disrupts thiamine uptake activity and may there- shorter by 94 amino acids than its close counterpart, the fore explain the thiamine deficiency in this French family hRFC, as it lacks the long C-terminus present in the latter. It with Rogers syndrome that clinically responds only to high is possible that the insertion of a highly charged Myc-tag doses of thiamine. (i.e. 9/17 amino acids are either negatively or positively We find here that the D93H mutant THTR1 as well as charged) to the C-terminus in the hTHTR1 may result in its D88H mutant hRFC counterpart lost thiamine and protein missfolding that could be identified by the quality MTX transport activity, respectively. The proposed topo- control mechanism in the ER, thereby leading to rapid logical structure of THTR1 predicts that D93 is the first transporter degradation. This putative degradation may amino acid in the short intracellular loop 2 (IL2), whereas gain support from the fact that the D93H-Myc mutant its D88 counterpart in the hRFC is embedded in the end THTR1 (i.e. the Myc tag placed in the C-terminus) was of TMD2 [13,23]. An alignment of the deduced amino- readily translated in vitro (Fig. 1C,D). However, one should acid sequence of the human THTR1 with that of the keep in mind that not any epitope tagging at the C-terminal various members of the SLC19A family including rodent end of the THTR1 may necessarily result in lack of protein (i.e. mouse and hamster) THTR1, human and rodent expression and/or rapid degradation; introduction of an THTR2, as well as RFC reveals a striking conservation at enhanced green fluorescent protein (EGFP) tag at the the vicinity of D93 in THTR1, THTR2 and of D88 in its C-terminus of THTR1 neither resulted in interference with counterpart, RFC. Specifically, not only is D93/D88 transporter expression nor with thiamine transport activity absolutely conserved across species in both THTR1, [22]. Consistently, either hemagglutinin (HA) or green THTR2 and RFC, TMD2 and IL2 are extremely well fluorescent protein (GFP) tagging at the C-terminus of the conserved (Table 1). Consistently, we recently identified human RFC did not interfere with transporter expression, the very same D88H mutation (along with additional plasma membrane targeting and MTX transport activity mutations in TMD2) in GW1843-resistant leukemia [23]. Taken together, the tag of choice has to be carefully GW70/LF cells that displayed a markedly altered trans- taken into considerations. We show here that the D93H port of antifolates and [27]. Moreover, recent mutant THTR1, while fused to Myc epitope tag at the studies demonstrate that elimination of the negative N-terminal, is properly translated and targeted to the charge at the D88 of the hRFC (e.g. D88V) results in a plasma membrane. Yet a previous study has shown that, complete loss of its MTX transport activity [16]. It was upon fusion of His and Xpress tags derived from pcDNA3.1 further found that this negatively charged residue (i.e. expression plasmid (Invitrogen Inc.) with D93H mutant D88) is absolutely essential for folate (i.e. leucovorin) and THTR1, this mutant protein is neither detected in the antifolate (i.e. MTX) transport as it presumably forms an cytosolic nor in the plasma membrane fractions of trans- ion-pair with a positively charged residue (i.e. R133) in fected cells [15]. According to the topological model TMD4 of the hRFC. It was therefore concluded that this proposed by these authors [15], the N-terminal domain charge-pair presumably allows for a proper tertiary including the six His-Xpress tags and the D93H residue structure to occur that is absolutely essential for folate located in the intracellular loop between TMD2 and TMD3 and antifolate transport. Hence, it is possible that the are expected to come to a close proximity in the cytosolic conserved D93 residue in THTR1 also plays an important milieu. This could result in the formation of nonphysiolog- role in the formation of a certain tertiary structure ical ion-pairs (i.e. His-Asp) as the Xpress tag introduces as that is crucial for thiamine binding and/or substrate much as five Asp residues; alternatively, His 93 could translocation. augment and/or alter the metal-cation chelation capabilities Multiple amino-acid substitutions are known to disrupt of the proximal six His domain. Clearly, in both cases RFC transport function [28]. As the number of patients transporter misfolding and rapid protein degradation could carrying THTR1 missense mutations is rather limited due be envisaged. In fact, among all mutations introduced in the to the fact that Rogers syndrome is a relatively rare His/Xpress tagged THTR1, only the D93H was undetect- disorder [14], it would be interesting to test whether other able [15]. Support to the notion of His tag mediated protein substitutions in conserved residues of the orthologue misfolding derives from several recent papers. First, intro- hRFC can inactivate the thiamine transport function of duction of a six His tag to lactate dehydrogenase [24] and THTR1 and THTR2 as well. Therefore, further compar- reverse transcriptase [25] resulted in misfolding and/or loss ative mutagenesis studies between the three members of of catalytic activity. Second, introduction of a His tag can the solute carrier family should provide useful information also result in detrimental post-tanslational modifications; regarding the specificity and functionality of these trans- for example, His tagging of the SH3 domains of Src tyrosine porters. FEBS2003 Lack of thiamine transport in D93H THTR1 ( Eur. J. Biochem. 270) 4477

Acknowledgements 14. Raz, T., Labay, V., Baron, D., Szargel, R., Anbinder, Y., Barrett, T., Rabl, W., Viana, M.B., Mandel, H., Baruchel, A., Cayuela, r We thank Prof. I.D. Goldman for the MTX A cells, Dr O. Shenkar for J.M. & Cohen, N. (2000) The spectrum of mutations, including assistance with confocal microscopy and Ms A. Cohen for technical four novel ones, in the thiamine-responsive megaloblastic anemia assistance. gene SLC19A2 of eight families. Hum. Mutat. 16, 37–42. 15. Balamurugan, K. & Said, H.M. (2002) Functional role of specific References amino acid residues in human thiamine transporter SLC19A2: mutational analysis. Am. J. Physiol. Gastrointest. Physiol. 1. Rogers, L., Porter, R. & Sidbury, J.J. (1967) Thiamine-responsive 283, G37–G43. megaloblastic anemia. J. Pediat. 74, 494–504. 16. Liu, X.Y. & Matherly, L.H. (2001) Functional interactions 2. Mandel, H., Berant, M., Hazani, A. & Naveh, Y. (1984) Thiam- between arginine-133 and aspartate-88 in the human reduced ine-dependent beriberi in the thiamine–responsive anemia syn- folate carrier: evidence for a charge–pair association. Biochem. J. drome. N.Engl.J.Med.311, 836–838. 358, 511–516. 3. Luzi, L. & DeFronzo, R.A. (1989) Effect of loss of first-phase 17. Hibi, M., Lin, A., Smeal, T., Minden, A. & Karin, M. (1993) insulin secretion on hepatic glucose production and tissue glucose Identification of an oncoprotein- and UV-responsive protein disposal in humans. Am. J Physiol. 257, E241–E246. kinase that binds and potentiates the c-Jun activation domain. 4. Groop, L.C., Barzilai, N., Ratheiser, K., Luzi, L., Wahlin-Boll, E., Genes Dev. 7, 2135–2148. Melander, A. & DeFronzo, R.A. (1991) Dose-dependent effects of 18. Sharif, K.A. & Goldman, I.D. (2000) Rapid determination of glyburide on insulin secretion and glucose uptake in humans. membrane transport parameters in adherent cells. Biotechniques Diabetes Care 14, 724–727. 28, 926–928,930,932. 5. Rindi, G., Patrini, C., Laforenza, U., Mandel, H., Berant, M., 19. Rothem, L., Aronheim, A. & Assaraf, Y.G. (2003) Alterations in Viana, M.B., Poggi, V. & Zarra, A.N. (1994) Further studies on the expression of transcription factors and the reduced folate erythrocyte thiamin transport and phosphorylation in seven carrier as a novel mechanism of antifolate resistance in human patients with thiamin-responsive megaloblastic anaemia. leukemia cells. J. Biol. Chem. 278, 8935–8941. J. Inherit. Metab. Dis. 17, 667–677. 20. Bissonnette, P., Noel, J., Coady, M.J. & Lapointe, J.Y. (1999) 6. Stagg, A.R., Fleming, J.C., Baker, M.A., Sakamoto, M., Cohen, Functional expression of tagged human Na+-glucose cotrans- N. & Neufeld, E.J. (1999) Defective high-affinity thiamine trans- porter in Xenopus laevis oocytes. J. Physiol. 520, 359–371. porter leads to cell death in thiamine–responsive megaloblastic 21. Kilic, F. & Rudnick, G. (2000) Oligomerization of serotonin anemia syndrome fibroblasts. J. Clin. Invest. 103, 723–729. transporter and its functional consequences. Proc. Natl Acad. Sci. 7. Labay, V., Raz, T., Baron, D., Mandel, H., Williams, H., Barrett, USA 97, 3106–3111. T.,Szargel,R.,McDonald,L.,Shalata,A.,Nosaka,K.,Gregory, 22. Subramanian, V.S., Marchant, J.S., Parker, I. & Said, H.M. S. & Cohen, N. (1999) Mutations in SLC19A2 cause thiamine- (2003) Cell biology of the human thiamine transporter-1 responsive megaloblastic anemia with diabetes mellitus and deaf- (hTHTR1). Intracellular trafficking and membrane targeting ness. Nat. Genet 22, 300–304. mechanisms. J. Biol. Chem. 278, 3976–3984. 8. Banikazemi, M., Diaz, G.A., Vossough, P., Jalali, M., Desnick, 23. Ferguson, P.L. & Flintoff, W.F. (1999) Topological and func- R.J. & Gelb, B.D. (1999) Localization of the thiamine–responsive tional analysis of the human reduced folate carrier by hemagglu- megaloblastic anemia syndrome locus to a 1.4-cM region of 1q23. tinin epitope insertion. J. Biol. Chem. 274, 16269–16278. Mol. Genet. Metab. 66, 193–198. 24. Halliwell, C., Morgan, G., Ou, C.-P. & Cass, A. (2001) Intro- 9. Fleming, J., Tartaglini, E., Steinkamp, M., Schorderet, D., Cohen, duction of a (poly) histidine tag in 1-lactate dehydrogenase pro- N. & Neufeld, E.J. (1999) The gene mutated in thiamine-respon- duces a mixture of active and inactive molecules. Anal Biochem. sive anemia with diabetes and deafness (TRMA) encodes a func- 295, 257–261. tional thiamine transporter. Nat. Genet. 22, 305–308. 25. Pekrun, K., Petry, H., Jentsch, K., Hunsmann, G. & Luke, W. 10. Eudy, J.D., Spiegelstein, O., Barber, R.C., Wlodarczyk, B.J., (1995) Reverse transcriptase enzyme from type 1 human immu- Talbot, J. & Finnell, R.H. (2000) Identification and characteriza- nodeficiency virus using different baculoviral vector systems. Eur. tion of the human and mouse SLC19A3 gene: a novel member of J. Biochem. 234, 811–818. the reduced folate family of micronutrient transporter genes. Mol. 26. Kim, K., Yi, E., Baker, D. & Zhang, Y. (2001) Post-translational Genet. Metab. 71, 581–590. modification of the N-terminal His tag interfers with the crystal- 11.Dutta,B.,Huang,W.,Molero,M.,Kekuda,R.,Leibach,F.H., lization of the wild-type and mutant SH3 domains from chicken Devoe, L.D., Ganapathy, V. & Prasad, P.D. (1999) Cloning of the src tyrosine kinase. Acta Crystallogr. D57, 759–762. human thiamine transporter, a member of the folate transporter 27. Drori, S., Jansen, G., Mauritz, R., Peters, G.J. & Assaraf, Y.G. family. J. Biol. Chem. 274, 31925–31929. (2000) Clustering of mutations in the first transmembrane domain 12. Baron,D.,Assaraf,Y.G.,Cohen,N.&Aronheim,A.(2002)Lack of the human reduced folate carrier in GW1843U89-resistant of plasma membrane targeting of a G172D mutant thiamine leukemia cells with impaired antifolate transport and augmented transporter derived from Rogers syndrome family. Mol. Med. 8, folate uptake. J. Biol. Chem. 275, 30855–30863. 462–474. 28. Rothem, L., Ifergan, I., Kaufman, Y., Priest, D., Jansen, G. & 13. Wong, S.C., Zhang, L., Proefke, S.A. & Matherly, L.H. (1998) Assaraf, Y.G. (2002) Resistance to multiple novel antifolates is Effects of the loss of capacity for N-glycosylation on the transport mediated via defective drug transport resulting from clustered activity and cellular localization of the human reduced folate mutations in the reduced folate carrier gene in human leukaemia carrier. Biochim. Biophys. Acta 1375, 6–12. cell lines. Biochem. J. 367, 741–750.