pH-dependent pyridoxine transport by SLC19A2 and SLC19A3: Implications for absorption in acidic microclimates

SLC19A2 and SLC19A3, also known as thiamine transporters (THTR) 1 and 2, respectively, transport the positively charged thiamine (vitamin B1) into cells to enable its efficient utiliza-tion. SLC19A2 and SLC19A3 are also known to transport structurally unrelated cationic drugs, such as metformin, but whether this charge selectivity extends to other molecules, such as pyridoxine (vitamin B6), is unknown. We tested this possibility using Madin-Darby canine kidney II (MDCKII) cells and human embryonic kidney 293 (HEK293) cells for transfection experiments, and also using Caco-2 cells as human intestinal epithelial model cells. The stable expression of SLC19A2 and SLC19A3 in MDCKII cells (as well as their transient expression in HEK293 cells) led to a significant induction in pyridoxine uptake at pH 5.5 compared with control cells. The induced uptake was pH-dependent, favoring acidic conditions over neutral to basic conditions, and protonophore-sensitive. It was saturable as a function of pyridoxine concentration, with an apparent K m of 37.8 and 18.5 m M , for SLC19A2 and SLC19A3, respectively, and inhibited by the pyridoxine analogs pyridoxal and pyridoxamine as well as thiamine. We also found that silencing the endogenous SLC19A3, but not SLC19A2, of Caco-2 cells with gene-specific siRNAs lead to a significant reduction in carrier-mediated pyridoxine uptake. These results show that SLC19A2 and SLC19A3 are capable of recognizing/transporting pyridoxine, favoring acidic conditions for operation, and suggest a possible role for these transporters in pyridoxine transport mainly in tissues with an acidic environment like the small intestine, which has an acidic surface microclimate.

SLC19A2 and SLC19A3, also known as thiamine transporters (THTR) 1 and 2, respectively, transport the positively charged thiamine (vitamin B1) into cells to enable its efficient utilization. SLC19A2 and SLC19A3 are also known to transport structurally unrelated cationic drugs, such as metformin, but whether this charge selectivity extends to other molecules, such as pyridoxine (vitamin B6), is unknown. We tested this possibility using Madin-Darby canine kidney II (MDCKII) cells and human embryonic kidney 293 (HEK293) cells for transfection experiments, and also using Caco-2 cells as human intestinal epithelial model cells. The stable expression of SLC19A2 and SLC19A3 in MDCKII cells (as well as their transient expression in HEK293 cells) led to a significant induction in pyridoxine uptake at pH 5.5 compared with control cells. The induced uptake was pH-dependent, favoring acidic conditions over neutral to basic conditions, and protonophore-sensitive. It was saturable as a function of pyridoxine concentration, with an apparent K m of 37.8 and 18.5 mM, for SLC19A2 and SLC19A3, respectively, and inhibited by the pyridoxine analogs pyridoxal and pyridoxamine as well as thiamine. We also found that silencing the endogenous SLC19A3, but not SLC19A2, of Caco-2 cells with gene-specific siRNAs lead to a significant reduction in carrier-mediated pyridoxine uptake. These results show that SLC19A2 and SLC19A3 are capable of recognizing/transporting pyridoxine, favoring acidic conditions for operation, and suggest a possible role for these transporters in pyridoxine transport mainly in tissues with an acidic environment like the small intestine, which has an acidic surface microclimate.
Pyridoxine is one of the vitamin B6 compounds, which include the structurally analogous pyridoxal and pyridoxamine. Because of its hydrophilic nature as a water-soluble vitamin, it cannot easily permeate across the cellular membrane by simple diffusion and, hence, is required for intestinal absorption a specialized carrier-mediated transport system (1). Indeed, carriermediated transport of pyridoxine has been observed in the Caco-2 cell line, a human intestinal epithelial cell model (2). Although transporters have been identified for a variety of water-soluble vitamins (1), none have been identified as yet for pyridoxine, although transport properties consistent with a carrier-mediated mechanism have been noted in many different cell types (3)(4)(5). In the current study, we explored whether SLC19A2 and SLC19A3, known thiamine transporter 1 (THTR1) and THTR2 (6,7), respectively, can also transport pyridoxine, based upon the fact that both pyridoxine and thiamine (vitamin B1) have cationic charge and that other structurally unrelated cationic drugs, such as metformin, fedratinib, and trimethoprim, are substrates for these transporters (8,9). The data indicate that pyridoxine is indeed also their substrate.

Uptake of pyridoxine in transient transfectant HEK293 cells
We first examined the uptake of pyridoxine at its trace concentration of 5 nM in transient transfectant human embryonic kidney 293 (HEK293) cells expressing each of SLC19A2 and SLC19A3. The trace concentration was used so that the potential transport activities could be detected most effectively, as transporters generally operate more efficiently at lower concentrations. This low concentration is also physiologically relevant, as pyridoxin concentration can be in the nanomolar range at the lowest end in body fluids, typically in plasma (10). As shown in Fig. 1, the uptake of pyridoxine was found to be much greater in cells expressing those SLC19As than in mock cells at an acidic pH of 5.5, suggesting that those SLC19As can transport pyridoxine efficiently. On the other hand, pyridoxine uptake was not altered by transient introduction of SLC19A1, suggesting that this transporter, which is another member of SLC19A family and known as reduced folate carrier 1, does not have such a capability for pyridoxine transport.

Time course of pyridoxine uptake in stable transfectant MDCKII cells
For detailed characterization of pyridoxine transport by SLC19A2 and SLC19A3, we prepared stable transfectant Madin-Darby canine kidney II (MDCKII) cells expressing each of them. As shown in Fig. 2, the uptake of pyridoxine (5 nM) was much greater in cells expressing those SLC19As than in mock cells at pH 5.5, further demonstrating the pyridoxine transport capability of both transporters, which was initially found in the transient transfectant HEK293 cells. As pyridoxine uptake increased in proportion to time up to 3 min in all the transfectant and mock cells, the uptake period was set to be 3 min in subsequent experiments for the evaluation of pyridoxine transport by those SLC19As in the initial uptake phase.
The uptake of pyridoxine was almost equilibrated at 30 min in both SLC19A2-transfected and SLC19A3-transfected cells. Using the cellular volumes of 6.5 6 1.1 and 6.4 6 1.1 ml/mg of protein (mean 6 S.E., n = 3), respectively, estimated from the equilibrium uptake of 3-O-methyl-D-glucose (3-O-MG) in those cells, the apparent intracellular concentrations of pyridoxine were estimated to be 43 and 52 nM, respectively. They were about 10 times higher than the pyridoxine concentration in the extracellular solution, suggesting that those SLC19As could possibly operate in an active transport mode that enables pyridoxine to be concentrated in the cells. In this analysis, the cellular volume was estimated as the uptake/concentration ratio of 3-O-MG (11), as the uptake of this unmetabolizable analog of D-glucose could be quickly equilibrated in MDCKII cells at the intracellular concentration equal to the concentration in the extracellular solution by nonconcentrative transport, in which facilitative glucose transporters have been suggested to be involved (12,13). Consistent with that suggestion, 3-O-MG uptake was confirmed to be Na 1 -independent and quite rapid, being equilibrated within 10 min, in those transfectant MDCKII cells.
Effect of pH and ionic conditions on pyridoxine transport by SLC19A2 and SLC19A3 Both SLC19A2 and SLC19A3 have been reported to operate in a pH-dependent manner, favoring neutral to basic conditions over acidic conditions, for thiamine transport (6,7). However, for pyridoxine transport, they were both found to operate in an opposite manner of pH dependence, favoring acidic conditions over neutral to basic conditions, as indicated by an increase in the specific uptake of pyridoxine (5 nM) by each transporter with a decrease in pH in stable transfectant MDCKII cells (Fig. 3). Moreover, the specific uptake by SLC19A2 and SLC19A3 were both extensively reduced by protonophores of carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) and carbonyl cyanide m-chlorophenylhydrazone (CCCP), suggesting that they may operate for pyridoxine transport in a manner coupled with proton or, at least, facilitated by an inward proton gradient (Fig. 4). Supporting these additionally, the specific uptakes by both SLC19As were, as assessed using uptake solution in which NaCl was replaced with KCl, reduced by intracellular acidification for dissipation of H 1 gradient by nigericin (10 mM), a K 1 /H 1 -exchanging protonophore, and also by an NH 4 Cl prepulse method, in which    the cells were pretreated with substrate-free uptake solution containing NH 4 Cl (20 mM) and subsequently with the one free of NH 4 Cl (14). The values of specific uptake rate were 42.9 6 2.0% of control (39.5 6 2.3 fmol/min/mg of protein) and 38.6 6 3.3% of control (71.6 6 3.8 fmol/min/mg of protein), respectively, for SLC19A2 and SLC19A3 in the experiments using nigericin, and 62.9 6 7.7% (47.6 6 5.2 fmol/min/mg of protein) and 43.7 6 6.6% of control (65.9 6 11.2 fmol/min/ mg of protein), respectively, in those by NH 4 Cl prepulse, where the values are presented as the mean 6 S.E. (n = 4) and all the differences are significant at p , 0.05 compared with respective control values. It should also be noted in addition that pyridoxine, which is a basic compound with the pK a of 5.1 (15), becomes progressively cationized with a decrease in pH in the near neutral to acidic pH range, coinciding to the pH-dependent increase in the specific uptake of pyridoxine. Hence, the increase in the cationized fraction of pyridoxine may also be involved in that, implicating the preferential transport of its cationized form by those SLC19As as a possibility. Thus, those SLC19As were suggested to operate in a different mode for pyridoxine than for thiamine, discriminating them. Although the mechanism of the substrate-dependent difference in pH dependence is unknown at this time, we conducted all the other experiments in this study at pH 5.5 as an acidic pH where both SLC19As can operate efficiently.
The replacement of NaCl in the medium with various salts (KCl, Na-gluconate, K-gluconate, and mannitol) had insignificant or only modest impacts on the specific uptake of pyridoxine (5 nM) by SLC19A2 (Fig. 5A) and SLC19A3 (Fig.  5B). The specific uptake by SLC19A2 was reduced slightly by replacement of NaCl with KCl and, to a comparable extent, by replacement with mannitol, suggesting that Na 1 may be slightly involved in the SLC19A2 operation but Cl 2 may not. The specific uptake by SLC19A3 was reduced modestly by replacement of NaCl with mannitol, but not by replacement with KCl, suggesting that Cl 2 may be slightly involved in the SLC19A3 operation but Na 1 may not. However, overall, such effects of those ions seemed to be only modest, suggesting at least that both SLC19A2 and SLC19A3 are unlikely to operate in a manner coupled with Na 1 or Cl 2 for pyridoxine transport.
Effect of pyridoxine-related compounds and water-soluble vitamins on pyridoxine transport by SLC19A2 and SLC19A3 To examine a possibility that some other pyridoxine-related compounds may also be recognized by SLC19A2 and SLC19A3, we here assessed the effect of such compounds (200 mM) on the specific uptake of pyridoxine (5 nM) by those SLC19As in stable transfectant MDCKII cells. As shown in Fig.  6A, the specific uptake by SLC19A2 was extensively inhibited by pyridoxamine and, to a lesser extent, by 4-deoxypyridoxine. Pyridoxal, pyridoxal 5-phosphate (PLP), and 4-pyridoxic acid were, however, found not to inhibit the specific uptake. SLC19A3 exhibited almost the same characteristics, although  pyridoxal exhibited a slight inhibitory effect as a minor difference (Fig. 6B). These results suggested the potential interaction of pyridoxamine and 4-deoxypyridoxine with both SLC19As and of pyridoxal with SLC19A3 to elicit the inhibitory effects.
Among water-soluble vitamins, the SLC19A2/3 substrate of thiamine was an only and potent inhibitor for both SLC19A2 and SLC19A3 (Fig. 7). All the other vitamins were confirmed not to inhibit the specific uptake by either of those SLC19As. As a note, specific transporters have already been identified for almost all of them, with the only exception being nicotinamide.
Focusing on thiamine and the vitamin B6 compounds of pyridoxal and pyridoxamine, analyses were conducted for detailed characterization of their concentration-dependent inhibitory effects on the specific uptake of pyridoxine (5 nM) by SLC19A2 and SLC19A3, as shown in Fig. 8 and Table 1. The IC 50 values of thiamine, the most potent inhibitor for the both, were comparable, being 0.71 and 0.99 mM, respectively, for SLC19A2 and SLC19A3. Those of pyridoxamine, the second most potent of the two, were quite different, being much higher for SLC19A2 (55.16 mM) than for SLC19A3 (2.97 mM). Those of pyridoxal, the least potent of the two, were relatively close, although being somewhat higher for SLC19A2 (1.49 mM) than for SLC19A3 (0.52 mM). The IC 50 of pyridoxal for SLC19A2 was high enough not to elicit detectable inhibition at its concentration of 200 mM, whereas that for SLC19A3 was at a level sufficient to elicit a modest inhibition, consistent with the results in Fig. 6. All   Table 1. Data are presented as the mean 6 S.E. (n = 3).

Table 1
Parameters for the inhibition of SLC19A2-and SLC19A3-mediated pyridoxine transport by selected compounds in stable transfectant MDCKII cells The values of the Hill coefficient (n) and IC 50 were determined for the inhibition profiles shown in Fig. 8 and those of the inhibition constant (K i ) were determined for those shown in Fig. 9. The parameter values are presented as the mean 6 S.E. (n = 3).

Compound
Transporter

Kinetic analyses of pyridoxine transport by SLC19A2 and SLC19A3
Both SLC19A2 and SLC19A3 were found to mediate pyridoxine transport in a highly saturable manner, as indicated by the concentration-dependent profiles of specific pyridoxine uptake (Fig. 9). According to kinetic analyses using the Michaelis-Menten model, the V max and K m were 332 pmol/min/mg of protein and 37.8 mM, respectively, for SLC19A2, and 264 pmol/ min/mg of protein and 18.5 mM, respectively, for SLC19A3 (Table 2). Thus, the K m values for those SLC19As were comparable, suggesting their comparable affinities for pyridoxine. It should also be noted that the pyridoxine concentration of 5 nM, which was used in regular experiments, was confirmed to be in the linear phase of transport kinetics at concentrations much below the K m values for both SLC19As, where their transport activities were highest and could be evaluated most effectively.
The profiles of the inhibition of the specific uptake of pyridoxine by thiamine, pyridoxamine, and pyridoxal were all found to conform to the competitive inhibition model (Fig. 9), with the values of inhibition constant (K i ) being comparable with the respective IC 50 values determined in Fig. 8 (Table 1). These results suggest that the vitamin B6 compounds of pyridoxamine and pyridoxal may be competing substrates recognized at the pyridoxine-binding site in both SLC19As. Cationic characteristic in addition to structural similarity to pyridoxine may be important for pyridoxine-related compounds to be recognized by SLC19A2 and SLC19A3, as PLP and 4-pyridoxic acid, which are less cationic pyridoxine derivatives because of having anionic moieties, did not inhibit the specific uptake of pyridoxine by either of the SLC19As (Fig. 6), suggesting no affinity for them. Thus, those SLC19As may have some multispecific characteristics within that limitation. 4-Deoxypyridoxine, a pyridoxine derivative that does not have an acidic moiety, may also be recognized as a competing substrate by those SLC19As, as this pyridoxine derivative inhibited specific pyridoxine uptake by them (Fig. 6).

Kinetic analyses of the effect of pyridoxine on thiamine transport by SLC19A2 and SLC19A3
To further clarify the interrelation between pyridoxine transport and thiamine transport in the operation of each of SLC19A2 and SLC19A3, we examined the effect of pyridoxine on thiamine transport by those SLC19As. The uptake of thiamine was first examined at a trace concentration of 5 nM, the same low concentration as that in regular assessments of pyridoxine transport, and at pH 7.4, as those SLC19As have been reported to operate more efficiently under neutral to basic conditions than acidic conditions (6, 7). As shown in Fig. 10, the uptake was much greater in cells expressing those SLC19As than in mock cells, demonstrating their known ability to transport thiamine. As thiamine uptake increased in proportion to time up to 1 min in all the transfectant and mock cells, the uptake period was set to be 1 min in subsequent experiments for the evaluation of thiamine transport by those SLC19As in the initial uptake phase. Figure 9. Effect of selected compounds on concentration-dependent pyridoxine uptake by SLC19A2 and SLC19A3 stably expressed in MDCKII cells. The specific uptake of [ 3 H]pyridoxine by SLC19A2 (A) and SLC19A3 (B) was evaluated at varied concentrations for 3 min at pH 5.5 and 37°C in the presence of a test compound or in its absence for control. The profiles are transformed into the Eadie-Hofstee format in the right panels.
The estimated values of V max , K m , and the inhibition constant (K i ) are presented in Table 2. The K i values of thiamine are also presented in Table 1 for comparison with the IC 50 values. Data are presented as the mean 6 S.E. (n = 3).

Table 2
Parameters for the SLC19A2-and SLC19A3-mediated transport of pyridoxine and thiamine in stable transfectant MDCKII cells The values of V max and K m were determined for the profiles shown in Figs. 9 (pyridoxine), 11 (thiamine, pH 7.4), and 12 (thiamine, pH 5.5). The values of the inhibition constant (K i ) of pyridoxine and thiamine were determined for their inhibitory effects on thiamine transport (Fig. 12) and pyridoxine transport (Fig. 9), respectively. The parameter values are presented as the mean 6 S.E. (n = 3).

Compound
Transporter The specific uptakes of thiamine (5 nM) by SLC19A2 and SLC19A3 were smaller at pH 5.5 than at pH 7.4 by 34 and 64%, respectively (Table 3). These moderately pH-dependent characteristics of thiamine transport were in agreement with those reported earlier (6,7). The inhibitory effects of pyridoxine (200 mM) on the specific uptakes of thiamine by SLC19A2 and SLC19A3 were quite extensive, reducing them to 24.0 and 8.9% of control, respectively, at pH 5.5, whereas its inhibitory effects were insignificant or only minimal at pH 7.4. These pH-dependent characteristics of the effect of pyridoxine on thiamine transport were consistent with those of pyridoxine transport by those SLC19As, which were indicative of the higher interaction of pyridoxine with them for its greater transport at pH 5.5 than at pH 7.4 (Fig. 3).
The saturable profiles of the specific uptake of thiamine by SLC19A2 and SLC19A3 were kinetically assessed at pH 7.4 (Fig. 11) and also at pH 5.5 (Fig. 12), and the estimated parameters are summarized in Table 2. At pH 7.4, the V max values were 106 and 65.9 pmol/min/mg of protein, respectively, for SLC19A2 and SLC19A3, and the K m values were 2.83 and 2.36 mM, respectively. These K m values were comparable with the previously reported values of 2.5 and 3.16 mM, respectively (6,8). At pH 5.5, the V max values were 100 and 26.9 pmol/min/mg of protein, respectively, for SLC19A2 and SLC19A3, and K m values were 3.66 and 2.33 mM, respectively. These K m values were comparable with the K i values of 1.11 and 0.86 mM, respectively, for the inhibition of pyridoxine transport by thiamine ( Table 2), suggesting that the interaction of thiamine with those SLC19As for its transport is possibly linked to its inhibitory action for pyridoxine transport. It is notable that, for SLC19A2, the K m was indicated to be increased with a decrease in pH, whereas V max was unchanged. Although the extent of decrease in K m was not large enough to be statistically significant, the trend was indicated at p , 0.1. To the contrary, for SLC19A3, the V max was indicated to be decreased, whereas K m was unchanged. Therefore, different mechanisms may be involved in the pH-dependent alterations in thiamine transport by those SLC19As. It should also be noted that the thiamine concentration of 5 nM, which was used in preceding experiments, was confirmed to be in the linear phase of transport kinetics at concentrations much below the K m values for both SLC19As, where their transport activities were highest and could be evaluated most effectively.
The analyses of the inhibitory effect of pyridoxine on the specific uptake of thiamine at pH 5.5 indicated the noncompetitive type of inhibition for both SLC19As (Fig. 12). This is an unexpected result, as pyridoxine, of which the transport was indicated to be competitively inhibited by thiamine (Fig. 9), should inhibit thiamine transport competitively, according to the Michaelis-Menten model. It is notable, however, that the K i values of 49.6 and 13.8 mM, respectively, for SLC19A2 and  Table 3 Effect of pyridoxine on thiamine uptake by SLC19A2 and SLC19A3 stably expressed in MDCKII cells  SLC19A3 were comparable with the K m values of 37.8 and 18.5 mM, respectively, for pyridoxine transport, suggesting that the interaction of pyridoxine with those SLC19As for its transport is possibly linked to its apparently noncompetitive type of inhibitory action for thiamine transport. It seems that some kind of unknown mechanism is involved in the interrelationship between pyridoxine transport and thiamine transport in the operation of each of those SLC19As, although each of the profiles of uptake and inhibition apparently conformed to the Michaelis-Menten model. There is a slightly curvilinear trend in the low concentration region in the Eadie-Hofstee plot of pyridoxine transport by SLC19A3, which may suggest the presence of a second recognition site with a high affinity for pyridoxine (Fig. 9B). However, it is not evident and, besides, the major part of the concentration-dependent profile of pyridoxine uptake is well characterized by the estimated K m , which is comparable with the K i of pyridoxine for thiamine transport.

Effect of silencing endogenous SLC19A2 and SLC19A3 in Caco-2 cells
The effect of silencing the endogenous SLC19A2 and SLC19A3 of Caco-2 cells (with the use of gene-specific RNAi) on pyridoxine uptake was examined to confirm the roles of these transporters in pyridoxine uptake. The expression of SLC19A2 was specifically suppressed by a siRNA specific to SLC19A2, as indicated by a significant reduction in the level of SLC19A2 mRNA but not in that of SLC19A3 mRNA (Fig.  13A). Similarly, the suppressing effect of an SLC19A3-specific siRNA was specific to the expression of SLC19A3 mRNA with no effect on the expression of SLC19A2 mRNA (Fig. 13B). The uptake of pyridoxine (5 nM) at pH 5.5 was reduced by silencing SLC19A3 (Fig. 13C), in accordance with the suppression of its mRNA expression and indicating its role in pyridoxine uptake. The SLC19A2 silencing-induced reduction of pyridoxine uptake was, however, not evident, although the trend was indicated at p , 0.1.
The ratio of the uptake rate to concentration (transport coefficient) at a pyridoxine concentration as high as 1 mM, which can represent the transport coefficient for simple diffusion (2), was estimated to be 1.7 ml/min/mg of protein. Because the transport coefficient was 5.8 ml/min/mg of protein for control at low concentrations (Fig. 13C), about 30% of the uptake can be accounted for by basal uptake by simple diffusion and, hence, the remaining 70% can be by the carrier-mediated mechanism. The reduction in the uptake by about 50% by the silencing of SLC19A3 suggests that this transporter could be  the main carrier-mediated pyridoxine uptake system in this cell model, accounting for about 70% of the carrier-mediated uptake. The remaining 30% of the carrier-mediated uptake could be accounted for by the modest contribution of SLC19A2, where its silencing led to ;20% reduction in uptake.

Discussion
The present study demonstrates that both SLC19A2 and SLC19A3 transport pyridoxine, and that this process is optimal under acidic conditions. The K m values for SLC19A2 (37.8 mM) and SLC19A3 (18.5 mM) were comparable in stable transfectant MDCKII cells (Fig. 9 and Table 2) and similar to the K m of 12 mM reported for pyridoxine uptake in Caco-2 cells (2). Unlike the modest difference in K m for the two transporters in stable transfectant MDCKII cells, the K m of two transporters in Caco-2 cells was deemed similar. The low-pH optimum of pyridoxine uptake mediated by SLC19A2 and SLC19A3 (Figs. 3 and 4) and its inhibition by protonophores is in agreement with the properties of pyridoxine transport in Caco-2 cells (2) and is consistent with a proton-coupled process. Interestingly, the bsu1 pyridoxine transporter in yeast is also proton-coupled; however, it is genetically distinct from the eukaryotic thiamine transporters (1, 16). 4-Pyridoxic acid was not effective in inhibiting SLC19A2-or SLC19A3-mediated pyridoxine transport in stable transfectant MDCKII cells (Fig. 6), although this compound was reported to be a potent inhibitor of pyridoxine transport in Caco-2 cells (2). The effects of all other pyridoxine-related compounds on pyridoxine transport by these SLC19As in stable transfectant MDCKII cells (Figs. 6 and 8) were similar to the pattern of inhibition in Caco-2 cells.
The similarity of SLC19A2-and SLC19A3-mediated pyridoxine transport in stable transfectant MDCKII cells and pyridoxine transport reported for Caco-2 cells, the latter a model for intestinal transport, along with the impact of siRNA suppression of SLC19A3 in Caco-2 cells, suggests the possibility that this transporter plays a role in the intestinal absorption of pyridoxine. SLC19A2 and SLC19A3 have been known to be localized to the basolateral and brush border membranes (17)(18)(19)(20)(21), respectively, in intestinal epithelial cells and could constitute a pathway of pyridoxine absorption. According to this localization of the transporters, the latter would mediate uptake across the brush border membrane driven by the H 1 gradient within the acidic microenvironment at the intestinal surface. A possible role for SLC19A2 in transport across the basolateral membrane in the absence of H 1 gradient is uncertain but this transporter might facilitate export driven by a pyridoxine concentration gradient. More likely, there is some expression of SLC19A2 at the brush border membrane (17)(18)(19)(20), where it would contribute to pyridoxine uptake. Although SLC19A2 and SLC19A3 are expressed in a variety of systemic tissues (6,8), the extent to which they deliver pyridoxine to these tissues within their near neutral microenvironments is not clear.
The pH dependence of pyridoxine transport systems in other cell types (opossum kidney cells, mouse colonocytes, and pancreatic acinar 266-6 cells) differs from that of SLC19A2 and SLC19A3 in that these systems have neutral to basic pH optima (3)(4)(5). In addition, the K m for pyridoxine transport in these cells is 1-2 orders of magnitude lower than those observed for SLC19A2 and SLC19A3 (3)(4)(5). A similar low K m has also been reported for pyridoxine transport in human primary pancreatic acinar cells (5). There are also differences in sensitivity to pyridoxine-related compounds among pyridoxine transport systems in these cell types and as compared with SLC19A2 and SLC19A3. However, in the absence of additional information regarding the mechanism of transport of pyridoxine and related compounds, it is not possible to determine whether these observations are due to different properties of SLC19A2 and SLC19A3 expressed in these cells and/or the presence of other pyridoxine transporter(s).
The current study indicates that SLC19A2 and SLC19A3 transport both thiamine and pyridoxine. However, there are marked differences in the pH at which transport of these substrates is optimal; this occurs, for thiamine, at near neutral pH and, for pyridoxine, at acidic pH ( Fig. 3 and Tables 2 and 3). There was also a qualitative difference in the kinetics of inhibition of pyridoxine transport by thiamine and inhibition of thiamine transport by pyridoxine; the former was competitive and the latter noncompetitive (Figs. 9 and 12). This suggests that the two substrates may bind to different sites of the transporter proteins, a likely possibility in view of the structural diversity of cationic compounds transported by these transporters (6)(7)(8)(9). There is precedent for this; for example, novel organic cation transporter 2 (SLC22A5) functions as a Na 1 -dependent transporter for L-carnitine, whereas its mediated transport of organic cations such as tetraethylammonium is Na 1 -independent (22).
In summary, we report for the first time the molecular identity of transporters for pyridoxine. We demonstrate that SLC19A2 and SLC19A3, originally identified as thiamine transporters, are multispecific transporters shared by pyridoxine and other structurally unrelated cationic compounds. The proton-coupled transport manifested for pyridoxine raises the possibility that these transporters may be the carrier-mediated mechanism by which pyridoxine and related compounds are absorbed in the small intestine. These findings can be a basis for further studies on the role these transporters play in the physiological processes involving pyridoxine and the pathophysiology of deficiency disorders involving this vitamin. Center (Tsukuba, Japan). The cells were maintained at 37°C and 5% CO 2 in DMEM supplemented with 10% FBS, 100 units/ ml of penicillin, and 100 mg/ml of streptomycin, as described previously (23).

Preparation of plasmids
The cDNA of human SLC19A1 (GenBank accession number NM_194255.4) was cloned by an RT-PCR method, as described previously (23). In brief, an RT reaction was carried out to obtain a cDNA mixture from human placenta total RNA (Bio-Chain Institute, Newark, CA), using 1 mg of the total RNA, an oligo(dT) primer, and ReverTra Ace (Toyobo, Osaka, Japan) as a reverse transcriptase. The cDNA of SLC19A1 was amplified by PCR, using PrimeSTAR Max DNA Polymerase (Takara Bio, Kusatsu, Japan). Then, the second PCR was performed using the amplified product as a template to incorporate restriction sites. Similarly, the cDNAs of SLC19A2 and SLC19A3 were cloned from human placenta total RNA. Their GenBank accession numbers are NM_006996.3 and NM_025243.4, respectively. The primers for PCR are shown in Table 4.
All final cDNA products were incorporated into pCI-neo vector (Promega, Madison, WI) to prepare plasmids for transfection and their sequences were determined with an automated sequencer (ABI PRISM 3130, Applied Biosystems, Foster City, CA), as described previously (24).

Preparation of stable transfectant MDCKII cells
MDCKII cells (4.0 3 10 5 cells/ml, 0.5 ml/well) were seeded on regular 24-well-plates, transfected with 1 mg/well of the plasmid for the designated transporter (SLC19A2 or SLC19A3), using 2 ml/well of Lipofectamine 2000 (Invitrogen), and cultured in DMEM supplemented with 10% FBS and 800 mg/ml of G418 (Sigma-Aldrich) for 2 to 3 weeks for stable transfection, as described previously (25). G418-resistant clones were selected to obtain stable transfectant MDCKII cells. Mock cells were similarly prepared using empty pCI-neo vector.

Silencing of SLC19A2 and SLC19A3 in Caco-2 cells
Caco-2 cells (5.0 3 10 4 cells/ml, 1 ml/well) were grown on 24-well-coated plates for 6 h, transfected with 15 pmol/well of the siRNA specific to the mRNA of SLC19A2 or SLC19A3 (Sigma-Aldrich), using 1.5 ml/well of Lipofectamine RNAi MAX (Invitrogen), and cultured for 5 days for silencing of the designated transporter by RNAi. The sequences of the siRNAs are shown in Table 5. For control, Negative Control DsiRNA (Integrated DNA Technologies, Coralville, IA) was used.

Real-time PCR analysis
Total RNA isolated from Caco-2 cells by a guanidine isothiocyanate extraction method (26) was used to obtain a cDNA mixture by an RT reaction, using ReverTra Ace as a reverse transcriptase. Real-time quantitative PCR analysis was performed with a CFX Connect Real-Time PCR Detection System (Bio-Rad Laboratories), using SsoAdvanced Universal SYBR   Green Supermix (Bio-Rad Laboratories) with specific detection primers for SLC19A2 and SLC19A3 (Table 6). GAPDH was used as the internal control, of which the mRNA expression level was used to normalize those of the SLC19As.

Transport study
Uptake assays were conducted as described previously (23), using transient transfectant HEK293 cells cultured on 24-wellcoated plates, stable transfectant MDCKII cells cultured on regular 24-well-plates, and Caco-2 cells cultured on 24-wellcoated plates. In brief, uptake solutions were prepared using Hanks' solution modified by supplementation with 10 mM MES (pH 6.5 and below) or 10 mM HEPES (pH 7.0 and above) and added with [ 3 H]pyridoxine as the substrate. Cells in each well were preincubated for 5 min in 1 ml of substrate-free uptake solution. Uptake assays were started by replacing the substratefree uptake solution with the one containing [ 3 H]pyridoxine (0.25 ml). All the procedures were conducted at 37°C. In experiments for intracellular acidification using the protonophores of FCCP and CCCP, they were added in the solution for preincubation as well as that for the uptake period. In those using nigericin and by an NH 4 Cl prepulse method, NaCl in the regular uptake solution was replaced with KCl. Nigericin was added only in the solution for preincubation. In the NH 4 Cl prepulse experiments, cells were preincubated in the solution added with 20 mM NH 4 Cl for 10 min and then in the solution without NH 4 Cl for 5 min before starting uptake using the solution without NH 4 Cl for the uptake period, as described previously (14). To examine the effect of ionic conditions, NaCl was replaced as indicated. In inhibition experiments, test compounds were added to the uptake solution to be present only during the uptake period. After termination of the uptake of [ 3 H]pyridoxine into the cells, the cells were solubilized and the associated radioactivity was determined by liquid scintillation counting for the evaluation of the uptake. The uptake was normalized to cellular protein content, which was determined by the bicinchoninic acid method (BCA Protein Assay Reagent Kit, Thermo Fisher Scientific, Waltham, MA), using BSA as the standard. In experiments using stable transfectant MDCKII cells, the specific uptake of pyridoxine was estimated by subtracting the uptake in mock cells from that in transfectant cells. The uptake of [ 3 H]thiamine was similarly assessed in stable transfectant MDCKII cells.

Estimation of cellular volume
The uptake of [ 3 H]3-O-MG (1 nM) was evaluated at equilibrium (10 min after the initiation of uptake) for the estimation of the cellular volume, as reported previously (11), in SLC19A2transfected and SLC19A3-transfected MDCKII cells.

Data analysis
The saturable uptake of a substrate by each transporter was analyzed using the Michaelis-Menten model equation as follows: v = V max 3 s/(K m 1 s). The V max and K m were estimated by fitting this equation to the experimental profile of the uptake rate (v) versus the concentration (s) of the substrate, using a nonlinear least-squares regression analysis program, WinNonlin (Certara, Princeton, NJ).
Analysis of the competitive type of inhibition of saturable substrate uptake by each transporter was conducted using the Michaelis-Menten model equation for competitive inhibition as follows: v = V max 3 s/(K m (1 1 i/K i ) 1 s). The K i was estimated by fitting this equation to the experimental profile of v versus s in the presence of an inhibitor at the concentration of i, with the V max and K m fixed at those estimated in its absence. Analysis of the noncompetitive type of inhibition was conducted similarly by using the equation for noncompetitive inhibition as follows: v = (V max /(1 1 i/K i )) 3 s/(K m 1 s).
The concentration-dependent inhibition of pyridoxine uptake by selected compounds was analyzed by using the following equation, which describes v as a function of i at s much lower than K m (s K m ): v = v 0 /(1 1 (i/IC 50 ) n ). The v 0 represents v in the absence of inhibitors. The IC 50 was estimated together with n by fitting this equation to the experimental profile of v versus i, with v 0 fixed at the observed value.
Data are presented as the mean 6 S.E. or S.D. with the number of experiments conducted using different preparations of cells, where each experiment was conducted in duplicate as biological repeats. Statistical analysis was performed by using Student's t test or, when multiple comparisons were needed, ANOVA followed by Dunnett's test, with p , 0.05 considered significant.

Data availability
All data are contained within the article.