Cell Biology of the Human Thiamine Transporter-1 (hTHTR1)

The human thiamine transporter hTHTR1 is involved in the cellular accumulation of thiamine (vitamin B1) in many tissues. Thiamine deficiency disorders, such as thiamine-responsive megaloblastic anemia (TRMA), which is associated with specific mutations within hTHTR1, likely impairs the functionality and/or intracellular targeting of hTHTR1. Unfortunately, nothing is known about the mechanisms that control the intracellular trafficking or membrane targeting of hTHTR1. To identify molecular determinants involved in hTHTR1 targeting, we generated a series of hTHTR1 truncations fused with the green fluorescent protein and imaged the targeting and trafficking dynamics of each construct in living duodenal epithelial cells. Whereas the full-length fusion protein was functionally expressed at the plasma membrane, analysis of the truncated mutants demonstrated an essential role for both NH2-terminal sequence and the integrity of the backbone polypeptide for cell surface expression. Most notably, truncation of hTHTR1 within a region where several TRMA truncations are clustered resulted in intracellular retention of the mutant protein. Finally, confocal imaging of the dynamics of intracellular hTHTR1 vesicles revealed a critical role for microtubules, but not microfilaments, in hTHTR1 trafficking. Taken together, these results correlate hTHTR1 structure with cellular expression profile and reveal a critical dependence on hTHTR1 backbone integrity and microtubule-based trafficking processes for functional expression of hTHTR1.

The human thiamine transporter hTHTR1 is involved in the cellular accumulation of thiamine (vitamin B1) in many tissues. Thiamine deficiency disorders, such as thiamine-responsive megaloblastic anemia (TRMA), which is associated with specific mutations within hTHTR1, likely impairs the functionality and/or intracellular targeting of hTHTR1. Unfortunately, nothing is known about the mechanisms that control the intracellular trafficking or membrane targeting of hTHTR1. To identify molecular determinants involved in hTHTR1 targeting, we generated a series of hTHTR1 truncations fused with the green fluorescent protein and imaged the targeting and trafficking dynamics of each construct in living duodenal epithelial cells. Whereas the full-length fusion protein was functionally expressed at the plasma membrane, analysis of the truncated mutants demonstrated an essential role for both NH 2 -terminal sequence and the integrity of the backbone polypeptide for cell surface expression. Most notably, truncation of hTHTR1 within a region where several TRMA truncations are clustered resulted in intracellular retention of the mutant protein. Finally, confocal imaging of the dynamics of intracellular hTHTR1 vesicles revealed a critical role for microtubules, but not microfilaments, in hTHTR1 trafficking. Taken together, these results correlate hTHTR1 structure with cellular expression profile and reveal a critical dependence on hTHTR1 backbone integrity and microtubule-based trafficking processes for functional expression of hTHTR1.
Thiamine (vitamin B1) is a water-soluble micronutrient that is essential for many cellular functions relating to growth and development. For example, thiamine pyrophosphate (the coenzyme form) is important for normal carbohydrate metabolism and energy production (1,2). Thiamine is not synthesized by humans and other mammals and is obtained from dietary sources via absorption into intestinal epithelia. Deficiencies in cellular thiamine accumulation lead to cardiovascular and neurological disorders and are associated with the inherited con-dition of thiamine-responsive megaloblastic anemia (TRMA) 1 (1,(3)(4)(5)(6)(7). Recent molecular analyses have identified the protein product of the SLC19A2 gene in humans (chromosome 1q23. 3) as the site of mutations that link with TRMA inheritance (2,(7)(8)(9)(10). SLC19A2 encodes a saturable, high affinity human thiamine transporter known as hTHTR1 (2,(7)(8)(9)(10), which is expressed in many tissues, including duodenal epithelium (11). The functional properties of hTHTR1 mimic the biochemical characteristics of thiamine uptake in intestinal preparations (reviewed in Refs. 12 and 13). The full-length hTHTR1 protein encodes a 497-amino acid polypeptide with 12 predicted transmembrane-spanning segments and cytoplasmic NH 2 -and COOH-terminal regions (2,8). Several clinically identified mutations have been characterized in hTHTR1 (2,(7)(8)(9)(10)14), including ten prematurely truncated mutants, resulting from either point mutation, deletion, or insertion of nucleotides (reviewed in Refs. 7 and 15). Unfortunately, because nothing is known about the molecular determinants that control the intracellular trafficking and cell surface targeting of hTHTR1, the precise cellular defect of such clinically isolated mutants remains undefined.
Through analogy with other nutrient transporters, the intracellular trafficking and cell surface targeting processes likely involve specific molecular "motifs" within the amino acid sequence of hTHTR1. Such determinants can either be localized within specific regions of the polypeptide or result from protein conformation generated from sequences throughout the entire polypeptide sequence (16,17). To identify regions of hTHTR1 important for cell surface targeting, we applied confocal imaging techniques to monitor both the steady-state distribution and real-time trafficking of seven hTHTR1 fusion proteins tagged with the enhanced green fluorescent protein (EGFP). We have chosen a human duodenal cell line (HuTu-80) for these studies, because hTHTR1 is expressed at higher levels in human duodenum than other intestinal epithelia (11). Our results show that (i) integrity of both the NH 2 -terminal sequence and the transmembrane backbone play an important role in the cell surface expression of hTHTR1; (ii) truncation of the backbone of hTHTR1 between the sixth and seventh transmembrane domains, at a locus where several clinical TRMA mutations cluster (7,15), results in intracellular retention of hTHTR1; (iii) the cytoplasmic COOH-terminal domain of hTHTR1 is not required for plasma membrane targeting; (iv) the intracellular trafficking dynamics of hTHTR1 are critically dependent on an intact microtubules but not microfilaments. These results highlight how hTHTR1 is delivered to the cell surface of mammalian epithelia within a physiologically relevant context and define the regions of hTHTR1 in which mutation would likely impair thiamine absorption through disruption of normal cellular targeting mechanisms.
Generation of hTHTR1-EGFP and Truncated Constructs-cDNA of the full-length hTHTR1 and truncated constructs were generated by PCR using the primer combinations shown in Table I and conditions specified previously (18,19). Both the PCR products and the EGFP-N3 vector were digested with the restriction enzymes BamHI and XhoI, and the products were gel-isolated and then ligated together to generate in-frame fusion proteins with enhanced green fluorescent protein (EGFP) fused to the COOH terminus of each construct. The nucleotide sequence of each construct was confirmed by sequencing. A schematic representation of each construct is shown below in Fig. 1.
Expression of hTHTR1-EGFP in Xenopus Oocytes-Capped cRNA for oocyte microinjection was synthesized from linearized plasmid DNA using the mMessage mMachine in vitro transcription kit (Ambion, Austin, TX). Pigmented oocytes (stages V-VI) were isolated and defolliculated using standard methods (19 -21) and then microinjected with either cRNA into the cytoplasm or cDNA into the nucleus. Injected oocytes were separated individually into 96-well plates and maintained in Barth's solution (88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO 3 , 0.83 mM MgSO 4 , 0.33 mM Ca(NO 3 ) 2 , 0.41 mM CaCl 2 , 10 mM HEPES, 550 mg/liter sodium pyruvate; 0.05 mg/ml gentamicin, pH 7.4, at 18°C) with repeated solution changes at least every 18 h. Additional oocytes from the same donor were injected with water as parallel controls for both viability and [ 3 H]thiamine uptake assays. For antisense studies, an equal amount of hTHTR1 cRNA was injected into two batches of oocytes, one batch of which was co-injected with a 5Ј-cholesterol-modified DNA oligonucleotide (5Ј-CATCCATCCGGGGCGCGAGGGGAGG) complementary to a sequence upstream of the hTHTR1 start codon (-18 to  H]thiamine. Uptake was terminated by several washes in ice-cold Barth's solution (5 ml), after which individual oocytes were transferred to scintillation vials and dissolved in 10% SDS (250 l) before scintillation counting.
Generation of Stable Cell Lines-Three stable cell lines were generated by selection of hTHTR1-EGFP, hTHTR1-pcDNA3.1, and EGFP-N3 expressing HuTu-80 cells with G418 (0.8 mg/ml) in minimal essential medium supplemented with 10% fetal bovine serum. Molecular analyses of stable cell lines were performed using semi-quantitative RT-PCR. Total RNA (5 g) was isolated from stably transfected and untransfected HuTu-80 cells and used to generate first-stranded cDNA with hTHRT1 gene-specific primers (Table I) and a superscript First-Strand synthesis kit (22). After subsequent PCR reactions, agarose gels (0.7%) were stained with ethidium bromide, and the intensity of the product band was quantified by densitometry. For [ 3 H]thiamine uptake exper-iments, cell monolayers were incubated for 3 min at 37°C in Krebs-Ringer buffer (pH 7.4) supplemented with [ 3 H]thiamine (30 nM). Uptake was terminated by addition of ice-cold Krebs-Ringer buffer, and accumulated radioactivity was measured by scintillation counting.
Confocal Microscopy-Cells were imaged using either a customized laser scan confocal microscope (23) or a video-rate scanning confocal microscope (24). Both microscopes were based on an Olympus IX70 inverted microscope fitted with a 40ϫ oil-immersion objective. Fluorophores were excited using the 488-nm line from an argon ion-laser, and emitted fluorescence was monitored with a 530 Ϯ 20-nm band pass (EGFP) or a 650-nm long-pass filter (FM4 -64). Using the linescan microscope (Figs. 2A, 3, 4, and 5A), frame scans were obtained by scanning the laser line either laterally (x-y scan) or axially (x-z scan) within the cell. For real-time images of vesicular trafficking, hTHTR1-EGFP fluorescence was monitored using the video-rate confocal microscope (Figs. 6 and 7). Petri dishes were maintained at 22°C or 37°C using a thermostatted stage mount. Images were captured at 30 Hz over a 40-m square region using a video-acquisition board/frame grabber and archived onto digital tape. For image processing, video data were digitized into image stacks using the stream acquisition function of the Metamorph imaging package (Universal Imaging, Downingtown, PA). Raw data stacks were averaged (as merges of two sequential 33-ms frames) and low pass-filtered before analysis. Frame-to-frame particle tracking was performed using customized macro routines in Metamorph to generate two-dimension (x-y) tracks of discrete particles over time. From such plots, quantitative data concerning the linear velocities and associated vectorality of individual structures could be analyzed throughout the entire recording period. Quick-time movies of video sequences are appended as supplementary material (videos 1-5).

RESULTS
Design of hTHTR1 Truncations-The schematic representation of the full-length hTHTR1-EGFP fusion protein in Fig. 1A illustrates the structural organization of the protein, comprising an NH 2 -terminal cytoplasmic region (amino acid residues 1-29), a transmembrane domain with 12 predicted transmembrane-spanning regions (amino acid residues 30 -479), and a COOH-terminal cytoplasmic domain (amino acid residues 480 -497) that was fused to EGFP. To identify regions important for cell surface expression, we generated a series of seven truncated fusion proteins (Fig. 1, B and C). First, we analyzed the role of NH 2 -terminal, transmembrane backbone, and COOH-terminal regions in hTHTR1 targeting (Fig. 1B). Second, we examined the effect of premature truncation between the sixth and seventh transmembrane domains of hTHTR1 (Fig. 1C), where three premature truncations (amino acid residues 239, 250, and 259) occur in clinical presentations of TRMA (7,15).
Functionality of the hTHTR1-EGFP-We first confirmed the functionality and appropriate cellular targeting of the fulllength hTHTR1-EGFP fusion construct by expressing the protein in Xenopus oocytes following nuclear injection of cDNA encoding hTHTR1-EGFP (Fig. 1A). Two days after microinjection, oocytes were imaged by confocal microscopy for hTHTR1-EGFP expression. In all oocytes examined (n ϭ 8 cells), hTHTR1-EGFP fluorescence was evident with an average 11.0 Ϯ 2.5-fold greater peak fluorescence in the animal compared with the vegetal pole ( Fig. 2A). Fluorescence was confined to a radial band 5.2 Ϯ 1.4 m wide, consistent with the thickness of the invaginated oocyte plasma membrane (19). To confirm localization of hTHTR1-EGFP to the cell surface, and to compare the functionality of hTHTR1-EGFP relative to hTHTR1, we measured [ 3 H]thiamine uptake in Xenopus oocytes heterologously expressing either construct. Oocytes expressing hTHTR1 or hTHTR1-EGFP exhibited [ 3 H]thiamine accumulation (108.0 Ϯ 9.6 and 131.5 Ϯ 35.5 fmol/oocyte/h, respectively) that was about double that in water-injected controls (53.0 Ϯ 1.0 fmol/oocyte/h) or oocytes co-injected with hTHTR1 cRNA and an antisense oligonucleotide to prevent translation of hTHTR1 (58.6 Ϯ 2.4 fmol/oocyte/h). These values for [ 3 H]thiamine uptake in cells expressing hTHTR1-EGFP (ϳ1.5-fold uptake) were similar to those reported elsewhere on expression of hTHTR1 cDNA (2,14). Thus, the functionality of the hTHTR1-EGFP fusion protein was not impaired by COOHterminal fusion to GFP, and the fusion protein correctly targeted to the cell surface.
Plasma Membrane Localization of hTHTR1-EGFP in Epithelial Cell Lines-To study the targeting of hTHTR1-EGFP in mammalian epithelia, cDNA encoding hTHTR1-EGFP was transiently transfected into a variety of epithelial cell lines, and the resulting fluorescence distribution was analyzed by confocal microscopy. In HuTu-80 cells, hTHTR1-EGFP expres- sion was evident in the cell membrane as well as in processes extending from the cell surface and a variety of intracellular structures (Fig. 3A). This distribution differed markedly from HuTu-80 cells transfected with EGFP alone, in which the entire cytoplasmic volume was fluorescence (Fig. 3B). These contrasting distributions were also apparent in axial scans (side views) of HuTu-80 cells expressing either hTHTR1-EGFP or EGFP alone (Fig. 3, C and D, respectively). A similar plasma membrane targeting of hTHTR1-EGFP was also apparent in several other cell lines, including MDCK, OK, HEK-293, and HepG2 (Fig. 3E).
Determination of Regions of hTHTR1 Important for Cell Surface Targeting-To investigate the role of specific regions of hTHTR1 structure in cell surface targeting, we constructed a series of hTHTR1 truncations (Fig. 1, B and C) and compared their expression profile with the full-length hTHTR1-EGFP fusion protein. Fig. 4 (A-C) shows representative confocal images (x-y scans) of HuTu-80 cells imaged 48 h after transient transfection with each construct. Several expression patterns were apparent, which could be grouped into three classes: 1) expression predominantly at the cell surface (hTHTR1-(1-479)-EGFP and hTHTR1- (19 -486), Fig. 4A Targeting of different hTHTR1-EGFP constructs to the cell surface was quantified utilizing a distribution overlap assay (18,25), in which cells transfected with individual hTHTR1-EGFP fusion constructs were co-stained by extracellular application of the red lipophilic marker FM4 -64 to selectively stain the plasma membrane. Fig. 4 illustrates this method for cells expressing hTHTR1-EGFP (Fig. 4D) and hTHTR1-(1-240)-EGFP (Fig. 4E). The green fluorescence of hTHTR1-EGFP overlapped considerably with the red fluorescence of FM4 -64 to yield a yellow color indicating a high degree of localization of hTHTR1-EGFP at the cell surface (Fig. 4D). In contrast, little fluorescence co-localization was observed with hTHTR1-(1-240)-EGFP (Fig. 4E), which remained localized within intracellular membranes. Fig. 4F shows collated measurements of fluorescence co-localization obtained by this method for hTHTR1-EGFP and all seven truncated constructs. A high degree (ϳ60%) of fluorescence overlap was observed in cells expressing hTHTR1-EGFP, hTHTR1-(1-479)-EGFP, and hTHTR1-(19 -486)-EGFP, confirming that each of these constructs targeted to the cell surface. Little co-localization (ϳ10%) was seen with the remaining constructs confirming their inability to reach the plasma membrane.
Establishment of a Stable HuTu-80 Cell Line for Analysis of hTHTR1-EGFP Trafficking-To facilitate investigation of the role of the cytoskeleton in controlling the delivery to hTHTR1-EGFP to the cell surface, we generated a stably expressing HuTu-80 cell line. Fig. 5A shows a confocal image of a HuTu-80 monolayer after 8 weeks of antibiotic selection, when the majority of cells (ϳ60%) exhibited the characteristic cell surface and intracellular distribution of hTHTR1-EGFP. Two further stable cell lines, expressing either hTHTR1 or EGFP alone, were established in parallel as controls. RT-PCR analysis of all three stable cell lines showed that hTHTR1 mRNA expression was ϳ5-fold greater in both the hTHTR1 and hTHTR1-EGFP stable HuTu-80 clones compared with the EGFP-expressing stable cell line or mock transfected HuTu-80 cells (Fig. 5B). Furthermore, uptake of [ 3 H]thiamine in both the hTHTR1-EGFP (113 Ϯ 6 fmol/mg of protein/3 min) and hTHTR1 (128 Ϯ 3 fmol/mg of protein/3 min) stable cell lines was significantly greater than in the EGFP stable cell line (73 Ϯ 5 fmol/mg of protein/3 min) or in untransfected HuTu-80 cells (59 Ϯ 4 fmol/mg of protein/3 min). Taken together, these data further support the functionality of the hTHTR1-EGFP fusion construct and establish the validity of this stable cell line for analyses of hTHTR1-EGFP properties.
Intracellular Trafficking Dynamics of hTHTR1-Confocal imaging of the stable hTHTR1 HuTu-80 cell line revealed numerous vesicular-like structures within the cytoplasm (Figs. 5A and 6A). The dimensions of these structures ranged between 0.5 and 1.5 m (Fig. 6B), suggesting that hTHTR1-EGFP localized within a heterogeneous population of intracellular vesicles. Vesicles were evident throughout the cytoplasm, but fluorescence was most concentrated close to the nucleus, which presumably represents the initial sites of hTHTR1-EGFP biosynthesis (see "Discussion"). To investigate the dynamics of individual vesicles trafficking toward the cell surface, we employed video-rate confocal imaging to capture image sequences with sufficient temporal resolution (30 frames per second) to track the movement of single vesicles in real time. An analysis of the video data (supplementary video 1), illustrated by a single image frame in Fig. 6A, is shown schematically in Fig. 6C. The coordinates of fluorescent vesicles were tracked at 66-ms intervals (two-frame averages) to generate color-coded tracks depicting their motion. Several tracks are shown expanded in Fig. 6C and illustrate three generalized features of hTHTR1-EGFP dynamics.
First, in any given cell, the population of hTHTR1-EGFPcontaining vesicles exhibited a wide variation in motility, some vesicles remained predominantly static (e.g. vesicle iv, Fig. 6C), whereas others were highly dynamic (e.g. vesicles i, ii, and iii, Fig. 6C). The distribution between the "dynamic" and "static" phenotype was temperature-dependent, with progressively more vesicles becoming motionless at lower temperature (data not shown). Furthermore, the velocity of vesicular movements increased with temperature. At 22°C, the average vesicular velocity (including stationary periods) was 0.43 Ϯ 0.13 m s Ϫ1 (n ϭ 25 vesicles), compared with 0.92 Ϯ 0.21 m s Ϫ1 at 37°C (n ϭ 93 vesicles).
Second, the high temporal resolution of the video-rate confocal records revealed two discrete components to the dynamics of individual vesicles, namely periods of rapid, approximately linear motion (e.g. black lines, vesicle iii, Fig. 6C) interspersed with periods of relatively immobility during which vesicles displayed small Brownian-like movements (e.g. arrow, vesicle iii, Fig. 6C). Because of this "stop-start" behavior, "average velocity" measurements presented above underestimate the peak velocity of hTHTR1-EGFP-containing vesicles. Therefore, we quantified velocities during the periods of rapid, directed linear movements (Fig. 6D). At 37°C the average linear velocity was 1.71 Ϯ 0.13 m/s, decreasing to 0.73 Ϯ 0.36 m/s at 22°C (n ϭ 38 vesicles, 5 cells). However, at both temperatures the "run length" (the length of linear segments) was similar, 2.75 Ϯ 0.22 m at 22°C and 3.22 Ϯ 0.22 m at 37°C.
Third, the movement of vesicles was strikingly multidirectional. Rather than displaying a consistent progression toward  . vesicle i, Fig. 6C). Similar behavior was also apparent when vesicles were in close proximity to the cell boundary, where they appeared to track cirumferentially beneath the plasma membrane (e.g. vesicle ii, Fig. 6C). Despite the multidirectional nature of vesicular motion, vesicles moving through the same area of the cytoplasm showed significant overlap in their tracks (Fig. 6C).
Role of Microtubules and Actin Microfilaments in hTHTR1-EGFP Trafficking-Having quantified the characteristics of vesicular motion at 37°C (Fig. 6D), we proceeded to examine the effect of cytoskeletal disruption on the motion of individual hTHTR1-EGFP-containing vesicles. Image sequences were recorded before and after addition of either the microtubule disrupting drugs nocodazole or colchicine (with ␥-lumicolchicine as a negative control) or the microfilament disrupting drug cytochalasin D. Fig. 7 shows the effect of incubation of HuTu-80 cells with nocodazole (Fig. 7A, 10 M for 5 min), colchicine (Fig. 7B, 10 M  for 15 min), ␥-lumicolchicine (Fig. 7C, 50 M for 30 min), or cytochalasin D (Fig. 7D, 10 M for 30 min), together with the measurements of the effects of each compound on the linear velocities of hTHTR1-EGFP motion at various times after drug addition (Fig. 7E). Each diagram of vesicular motion (Fig. 7,  A-D) is associated with a supplementary video (videos 2-5 in supplemental material) from which the tracking data were derived. Incubation of cells with nocodazole rapidly inhibited the linear motion of hTHTR1-EGFP-containing vesicles within 5 min (Fig. 7, A and E). Colchicine exhibited a similar effect albeit over a slower time course, with maximal inhibition taking up to 15 min (Fig. 7, B and E). In contrast, addition of ␥-lumicolchicine (50 M) failed to impede vesicular motion, even with incubation periods as long as 30 min (Fig. 7, C and  E). Similarly, cytochalasin D had little effect on the linear motions of hTHTR1-containing vesicles (Fig. 7, D and E). tions, we employed confocal microscopy to image the dynamics of hTHTR1-EGFP fusion proteins. This strategy has proved a powerful approach to monitor the targeting and expression of nutrient transporters and other proteins, because it allows localization with high spatial and temporal resolution in live cells (18,25,26). Heterologous expression of hTHTR1-EGFP in either mammalian epithelial cell lines (Figs. 3 and 5A) or the Xenopus oocyte expression system ( Fig. 2A), resulted predominantly in expression of hTHTR1-EGFP fluorescence at the cell surface, as expected for a transporter protein, as well as in cellular processes extending from the plasma membrane and a variety of intracellular structures (see below). Ligation of EGFP to the COOH terminus of hTHTR1 did not impair functionality, as assessed by [ 3 H]thiamine accumulation in both Xenopus oocytes (Fig. 2B), and a stably expressing human duodenal cell line (Fig. 5C).
Regions of hTHTR1 Important for Cellular Targeting-The delivery of proteins to specific cellular compartments depends on the presence of targeting commands embedded as primary sequence, structural conformations, or post-translational processing signals within the protein structure (27,28). Many "motifs" responsible for directing nutrient transporters to the plasma membrane of epithelial cells have been localized within the NH 2 terminus (29), COOH terminus (30), and/or the transmembrane backbone regions of individual proteins (18). Here we show that both NH 2 -terminal sequence (within amino acids 19 -29) as well as determinants within the transmembrane backbone are important in directing the expression and consequent export of hTHTR1 from the endoplasmic reticulum to the plasma membrane. This conclusion is based on the following evidence. First, the COOH-terminal cytoplasmic region (amino acids 480 -497) was unimportant for plasma membrane targeting, because removal of this region failed to prevent cell surface localization (hTHTR1-(1-479)-EGFP). In contrast, removal of the NH 2 -terminal sequence (amino acids 1-29) prevented cell surface expression of the truncated protein (hTHTR1-(30 -497)-EGFP), which was retained within intracellular membranes (Fig. 4C). Second, the NH 2 -terminal sequence alone was not sufficient for cell surface expression, because hTHTR1-(1-240)-EGFP, a construct in which the entire NH 2 -terminal sequence was preserved, but the polypeptide backbone was truncated after six transmembrane helices, was retained within intracellular membranes. The role of the polypeptide backbone was underscored by the observation that hTHTR1- (19 -486) in which both the NH 2 -and COOH-terminal regions were partially truncated, but the integrity of the transmembrane domains preserved, was targeted to the cell surface (Fig. 4, A and  F). Further experiments are needed to delimit whether polypeptide backbone integrity per se, or specific determinants downstream of the truncation site, are needed for plasma membrane expression. The important point is that, if COOH determinants exist, they do not function in isolation, because hTHTR1-(241-479) was also defective at export from intracellular membranes. Most likely, plasma membrane targeting is dependent on both NH 2 -terminal sequence and determinants distributed throughout the hTHTR1 polypeptide backbone.
A final important point relates to our knowledge of targeting determinants in other members of the SLC19A gene family. As the protein product of the SLC19A2 gene, hTHTR1 shares significant sequence identity (ϳ40% (2,8)) with the human reduced folate carrier (hRFC), the protein product of the SLC19A1 gene (31). Topological analysis of both proteins suggests a similar 12 transmembrane-spanning topology (Refs. 2, 32, and 33, but see Ref. 10). We recently analyzed the effect of specific domain ablations on hRFC targeting (18,19) and demonstrated that polypeptide backbone integrity plays a crucial role in the plasma membrane expression of hRFC. This result mimics that observed here with hTHTR1, in that the polypeptide backbone of either transporter (hTHTR1-(19 -486)-EGFP or hRFC-(19 -466)-EGFP) with minimal flanking sequence is sufficient to direct cell surface expression. One subtle difference with hTHTR-1 is that the NH 2 -terminal sequence (amino acids 19 -29) is more critical for cell surface expression than the corresponding region in hRFC (18). This may relate to the difference in net positive charge of the NH 2 -terminal domain of hTHTR1 ("ϩ4") compared with hRFC ("0"), which may be crucial in directing appropriate protein translocation and folding within the endoplasmic reticulum membranes (34). Overall these results suggest the possible conservation of targeting mechanisms between these closely related cousins within the major facilitator superfamily of transporters (35). It remains to be seen whether this homology extends to hTHTR2, the protein product of the recently characterized SLC19A3 gene (36,37), which encodes a second human thiamine transporter with even greater amino acid residue identity to hTHTR1 (ϳ48% (36, 37)).
Clinical Relevance of Truncations within the hTHTR1 Polypeptide Backbone-TRMA is a rare autosomal recessive disorder, attributed to mutational impairment of hTHTR1 function caused by specific mutations in the SLC19A2 gene (2, 7-10, 14, 15). At least fourteen distinct mutations have been identified from TRMA patients (7,15) of which 10 result in premature truncation of hTHTR1 (Fig. 1A). Three of these truncations (239, 250, and 259 amino acids (8)) result in premature termination within a region normally between the sixth and seventh transmembrane helices (10). Confocal imaging of hTHTR1-(1-240)-EGFP showed that the truncated protein is retained within intracellular membranes (Fig. 4C), with no visible cell surface expression (Fig. 4E). Therefore, the three clinically relevant truncations occurring within this same region likely result in abrogation of cell surface expression. Because the four upstream truncations (after amino acids 66, 97, 127, and 162) lack further amino acids (Refs. 7, 8, 10, and 15), each of these truncations likely results in intracellular retention of the protein product. Further experiments are needed to identify the cellular localization of the three remaining clinically identified truncations (313, 358, and 385 amino acids in length) (Refs. 9, 15, and 38), although we note that similar sized truncations of hRFC backbone resulted in intracellular retention (18).
Intracellular Trafficking of hTHTR1-EGFP-Generation of a stable hTHTR1-EGFP-expressing duodenal cell line facilitated analysis of hTHTR1-EGFP trafficking as synthesis of hTHTR1-EGFP was ongoing and the population of intracellular fluorescent structures more numerous than with transient transfection. The steady-state distribution of hTHTR1-EGFP was defined by a strong cell surface expression, together with considerable fluorescence within a juxtanuclear pool, most likely the Golgi apparatus. Vesicular motility was critically dependent on an intact microtubular cytoskeleton, as shown by the effects of pharmacological disruption with nocodazole or colchicine, but not ␥-lumicolchicine as a negative control (Fig.  7). Furthermore, video-rate confocal measurements of vesicular dynamics (Fig. 6, average moving velocities of ϳ1.7 m/s and run lengths of ϳ3 m) are consistent with microtubule-based transport of different cargos in other cell types (39,40), as well as speeds of microtubule-based motors measured in vitro (41). A striking result was the observation of multidirectional vesicular motion, both toward and away from the cell surface, as well as circumferential movements beneath the plasma membrane (Fig. 6). These images suggest that the trafficking and insertion of hTHTR1 into the plasma membrane may be physiologically regulated events, and we are currently investigating mechanisms that may influence the net progression of hTHTR1 transporters toward, and removal from, the cell surface.
Summary-These studies show that, unlike many nutrient transporters, the molecular determinants that dictate hTHTR1 targeting to the cell surface are not located within the COOHterminal cytoplasmic tail of the polypeptide but are dependent on residues within both the NH 2 -terminal region and the polypeptide backbone. Premature truncation of the polypeptide backbone, as seen in several independent clinical mutations in hTHTR1, causes intracellular retention of the protein. Finally, trafficking of hTHTR1 to the cell surface is critically dependent on intact microtubules but not microfilaments.