Glucose Transporter Isoforms GLUT1 and GLUT3 Transport Dehydroascorbic Acid*

Dehydroascorbic acid (DHA) is rapidly taken up by cells and reduced to ascorbic acid (AA). Using the Xenopus laevis oocyte expression system we examined transport of DHA and AA via glucose transporter isoforms GLUT1–5 and SGLT1. The apparentK m of DHA transport via GLUT1 and GLUT3 was 1.1 ± 0.2 and 1.7 ± 0.3 mm, respectively. High performance liquid chromatography analysis confirmed 100% reduction of DHA to AA within oocytes. GLUT4 transport of DHA was only 2–4-fold above control and transport kinetics could not be calculated. GLUT2, GLUT5, and SGLT1 did not transport DHA and none of the isoforms transported AA. Radiolabeled sugar transport confirmed transporter function and identity of all cDNA clones was confirmed by restriction fragment mapping. GLUT1 and GLUT3 cDNA were further verified by polymerase chain reaction. DHA transport activity in both GLUT1 and GLUT3 was inhibited by 2-deoxyglucose, d-glucose, and 3-O-methylglucose among other hexoses while fructose and l-glucose showed no inhibition. Inhibition by the endofacial inhibitor, cytochalasin B, was non-competitive and inhibition by the exofacial inhibitor, 4,6-O-ethylidene-α-glucose, was competitive. Expressed mutant constructs of GLUT1 and GLUT3 did not transport DHA. DHA and 2-deoxyglucose uptake by Chinese hamster ovary cells overexpressing either GLUT1 or GLUT3 was increased 2–8-fold over control cells. These studies suggest GLUT1 and GLUT3 isoforms are the specific glucose transporter isoforms which mediate DHA transport and subsequent accumulation of AA.

Dehydroascorbic acid is structurally similar to glucose.
Therefore, DHA entry has been proposed to be mediated by glucose transporters (12,13,19,20). Despite investigations in several cell types, this hypothesis has not been proven. The ideal means to verify it is to express glucose transporters using an expression system, and to study DHA transport activity. If any transporters were active, transport kinetics could be characterized only under conditions of 100% internal reduction to ascorbate, consistent with DHA transport into cells being ratelimiting (7). If internal DHA reduction were incomplete, kinetics could not be calculated. Although one study characterized DHA transport by expressed GLUT1 (21), there were a number of flaws in this report. Experiments were performed using mixtures of ascorbic acid and ascorbic acid oxidase instead of pure DHA as substrate. There was insufficient data about internal DHA reduction at each external DHA concentration, and calculations of high affinity transport were based on incorrect mathematical assumptions. In addition, although DHA transport was attributed to GLUT2 and GLUT4 as well, no data were presented to support these conclusions.
To characterize dehydroascorbic acid and ascorbate transport we utilized a Xenopus laevis oocyte expression system to express glucose transport isoforms GLUT1-5 (22)(23)(24)(25)(26) and SGLT1 (27). The data here indicate that DHA is transported by GLUT1 and GLUT3 but not other isoforms, while ascorbate is not transported by any of the proteins studied.
Preparation of [ 14 C]Dehydroascorbic Acid-[ 14 C]DHA was prepared from crystalline [ 14 C]ascorbic acid (NEN Life Sciences Products Inc., 6.6 mCi/mmol) as described (15). Briefly, 5 l of bromine solution (Fluka, Ronkonkoma, NY) was added to 600 l of [ 14 C]ascorbic acid solubilized in ultrapure water at a concentration of 20 mM, vortexed briefly and immediately purged with nitrogen on ice in the dark for 10 min. HPLC with radiomatic detection confirmed 100% conversion of AA to DHA, which could be completely recovered upon reduction with 2,3-dimercapto-1-propanol (37). Inhibitors or competitors were added to the incubation as described in the text. Individual oocytes were either dissolved in 500 l of 10% SDS and internalized radioactivity was measured using scintillation spectrometry, or oocytes were frozen to Ϫ70°C in 50 l of 60% MeOH, 1 mM EDTA for later HPLC analysis.
HPLC of Ascorbic Acid and [ 14 C]Dehydroascorbic Acid-Oocyte total AA mass and internalized AA and DHA radioactivity were measured using HPLC. AA mass was determined using electrochemical detection (4). Internalized radioactivity was determined using the same HPLC system followed by on-line scintillation spectrometry (Packard Series A-120 Flo-one radiomatic detector (Downers Grove, IL)). Prior to HPLC analysis, oocytes previously frozen in 60% MeOH, 1 mM EDTA were thawed on ice, lysed by agitation with a pipette tip, and centrifuged at 14,000 rpm for 10 min. The supernatant was removed and analyzed by HPLC. MeOH was confirmed to extract 100% of AA and DHA from oocytes (data not shown).
Statistics and Kinetic Calculations-Data are expressed as the arithmetic mean Ϯ S.D. of 10 -20 oocytes at each data point, unless otherwise indicated. S.D. is not displayed when smaller than the symbol size. Transport kinetics were analyzed by best-fit analysis of data points utilizing curve-fitting (Jandel Scientific, San Rafael, CA) or Eadie-Hofstee transformation. IC 50 values for DHA inhibition were determined by fitting data to a logit-log plot.

RESULTS
We investigated DHA and ascorbate transport by GLUT1-5 and SGLT1. mRNA coding for the individual isoforms was injected into Xenopus oocytes and concentration-dependent DHA transport activity was assessed (Fig. 1). Radiolabeled 2-DG, fructose, or glucose uptake performed within the same experiment was a positive control for transporter activity. [ 14 C]DHA transport in GLUT1 and GLUT3 expressing oocytes was over 100-fold greater than control sham-injected oocytes and similar to 2-[ 3 H]DG transport on a mole for mole basis. DHA transport by oocytes expressing GLUT2, GLUT5, and SGLT1 did not differ from sham-injected controls. Oocytes expressing GLUT4 transported 2-4-fold more DHA than control, but 2-DG transport was an order of magnitude greater. Uptake of radiolabeled sugars for the different transporters was in the range expected (38,39). Ascorbate transport by GLUT1-5 and SGLT1 was not different from sham-injected controls (Ͻ1 pmol/ oocyte/10-min incubation) (data not shown).
We verified the identity of the GLUT1 and GLUT3 cDNA constructs utilizing restriction digestion and PCR (Fig. 2). Restriction digestion of GLUT1 and GLUT3 insert DNA using HindIII and EcoRI gave predicted fragments based on known sequences of GLUT1 and GLUT3 cDNA (22,24). In addition, PCR performed using cDNA primers specific for either GLUT1 or GLUT3 cDNA, produced DNA products only when the appropriate primers were used. The identities of the other glucose transporter constructs were also confirmed by restriction digest mapping (data not shown).
Injection conditions for GLUT1 and GLUT3 mRNA were determined based on post-injection time and amount of mRNA injected. Variation in injection amount of either mRNA from 2 to 10 ng/oocyte resulted in a linear increase in transport activity for both DHA and 2-DG, and activity achieved plateau at 20 -40 ng mRNA/oocyte (data not shown). Transport activity increased over time post-injection, with maximal activity occurring at 3-5 days (data not shown).
To determine DHA transport kinetics internal reduction of DHA to AA must be complete and efflux of DHA should not occur. To establish these conditions, we first examined concentration-dependent [ 14 C]DHA uptake (0.1-8 mM) over 10 min into oocytes injected with 30 ng of GLUT1 mRNA (21). Total radiolabeled uptake was measured, representing the sum of AA, DHA, and metabolites. The percentage of label present intracellularly as DHA, AA, or metabolites was also analyzed by HPLC and is displayed as % reduction to ascorbate. The results of total radiolabel data suggest that uptake saturated at 4 mM external DHA (Fig. 3). However, these observations can be explained by incomplete internal reduction of DHA to ascorbate (Fig. 3). Complete internal reduction occurred at [ 14 C]DHA external concentrations Յ1 mM but reduction was incomplete at higher concentrations. [ 14 C]DHA metabolites were only present when DHA reduction was incomplete (not shown). Under conditions of incomplete reduction transport kinetics cannot be calculated because reduction rather than transport becomes limiting. Similar results were obtained for GLUT3 (data not shown). Consistent with these observations, DHA efflux occurred from oocytes expressing GLUT1 or GLUT3 only when reduction was incomplete (data not shown).
To achieve complete internal reduction, incubation time with substrates was decreased to Յ1 min for oocytes expressing either GLUT1 or GLUT3. For GLUT1 expressing oocytes, injected mRNA was also decreased to 2 ng/oocyte. Using these conditions, concentration-dependent DHA transport occurred in GLUT1 and GLUT3 expressing oocytes (Fig. 4, A and B). At all concentrations of DHA, HPLC analysis confirmed 100% reduction of internalized label to AA (data not shown). For each concentration selected, uptake was linear with respect to time. Kinetic parameters of GLUT1-and GLUT3-mediated DHA transport were calculated using best-fit analysis and Eadie-Hofstee transformation (Fig. 4, A and B, inset). Using best-fit analysis, apparent K m was 1.1 Ϯ 0.2 mM and V max was 108 pmol/min/oocyte for GLUT1, and apparent K m was 1.7 Ϯ 0.3 mM with V max of 241 pmol/min/oocyte for GLUT3. Eadie-Hofstee transformation yielded similar results. For GLUT1, apparent K m was 1.2 mM and V max was 124 pmol/min/oocyte, and for GLUT3 apparent K m was 1.1 mM and V max was 201 pmol/min/oocyte. We examined the ability of different sugars, ascorbic acid, and cytochalasins B and E to inhibit [ 14 C]DHA transport in both GLUT1 and GLUT3 expressing oocytes (Table I). The relative ability of different sugars to inhibit DHA uptake was similar for both transport proteins (2-DG Ն glucose Ն 3-Omethylglucose Ͼ maltose Ͼ mannose Ͼ xylose). IC 50 values were lower for GLUT3 for all of the sugars with an inhibitory effect. Cytochalasin B strongly inhibited DHA transport via both isoforms with an IC 50 similar to that seen with transport of 2-[ 3 H]DG under similar conditions (data not shown), while cytochalasin E had no effect. As expected, AA did not inhibit DHA uptake through either transporter.
Glucose transporters possess both endofacial and exofacial substrate-binding sites (40). When DHA is present externally, internal DHA is absent under conditions of complete internal reduction. Therefore, we anticipated that exofacial and endofacial inhibitors of glucose transport would behave differently with DHA as the substrate. We predicted that the endofacial glucose transport inhibitor cytochalasin B (41, 42) would behave as a non-competitive inhibitor of DHA transport, and that the exofacial inhibitor 4,6-O-ethylidene-␣-glucose (43, 44) would behave as a competitive inhibitor. Oocytes expressing either GLUT1 or GLUT3 were incubated with increasing con-  (Fig. 5). The results show ethylidene glucose was a competitive inhibitor of DHA uptake by GLUT1 and GLUT3. These data suggest that the external binding sites for ethylidene glucose and DHA are identical. Cytochalasin B inhibited DHA transport by GLUT1 and GLUT3 non-competitively.
Glucose transporter mutants, previously constructed with single amino acid substitutions in domains believed to be important for glucose transport, were demonstrated to be defective in their ability to transport 2-DG (28,30). If similar sites are involved in both DHA and 2-DG transport, transport of either substrate by these mutant constructs should be similar. We measured the ability of two mutants, GLUT1 161 (Gln 161 3 Leu) and GLUT3 410 (Trp 410 3 Leu), to transport 2-[ 3 H]DG or [ 14 C]DHA (Table II). Transport of both substrates via GLUT1 and GLUT3 was Ͼ100-fold higher than control and was virtually eliminated by the mutations. Western blotting demonstrated that both mutant proteins were present in the plasma membrane (data not shown).
To confirm that GLUT1 and GLUT3 also transport DHA in mammalian cells, we examined DHA transport in cells overexpressing these proteins. DHA transport was measured in Chinese hamster ovary cells stably transfected with rat GLUT1, human GLUT3, or rat GLUT5. GLUT1 and GLUT3 overexpressing cells demonstrated a 2-8-fold increase in [ 14 C]DHA uptake and a 2-18-fold increase in 2-DG uptake compared with control cells (Fig. 6). Increased transport activity of both substrates was inhibited by up to 95% by cytochalasin B (data not shown). GLUT5 overexpressing cells showed higher [ 14 C]fructose transport but [ 14 C]DHA uptake was no different from control (data not shown). DISCUSSION In the present report we demonstrate that glucose transporter isoforms GLUT1 and GLUT3 mediate the transport of dehydroascorbic acid. Transport activity was demonstrated both in the Xenopus oocyte expression system and in CHO cells overexpressing these transport proteins. Mutant constructs of both GLUT1 and GLUT3 failed to transport DHA. Determination of DHA transport kinetics was performed under conditions of complete internal reduction of DHA to AA. This was confirmed at all external concentrations of DHA by HPLC. Without complete reduction, kinetics cannot be calculated because efflux of substrate occurs and the reduction process becomes  rate-limiting. Transport of DHA by other glucose transporter isoforms was either very low or nondetectable. DHA transport by GLUT4 expressing oocytes was 2-4-fold above control, but 10-fold less than 2-DG transport. This suggests that GLUT4 may have a low affinity for DHA, but because DHA transport was so low, it was not possible to characterize transport kinetics in oocytes. GLUT2, GLUT5, and SGLT1 expressing oocytes did not transport DHA differently from control sham-injected oocytes. Within experiment sugar transport and restriction enzyme mapping confirmed the identity and functional integrity of each construct.
DHA transport by at least 2 glucose transporters is consistent with the structural similarity of DHA and glucose. The lack of DHA transport by both GLUT2 and GLUT5 may be associated with sequence-specific differences responsible for their ability to transport fructose (30,31). These mechanisms and those responsible for the lower affinity of DHA for GLUT4 remain to be elucidated. Nevertheless, because of specificity for GLUT1 and GLUT3, DHA may prove to be a useful tool to discriminate transport activity between different glucose transporter isoforms.
The apparent transport affinities (K m ) of DHA for GLUT1 and GLUT3 of 1.1 and 1.7 mM, respectively, are similar to, or less than, those reported previously for glucose (45,46). Despite data showing that the apparent affinities of both glucose transporters for DHA are comparable, inhibition of DHA transport with various sugars demonstrated lower IC 50 for GLUT3. This finding is consistent with previously reported sugar inhibition of 2-DG uptake in oocytes (30). In addition, although the zero-trans K m of both transporters for glucose entry is similar (45,46), the equilibrium exchange K m for GLUT3 is lower than for GLUT1 (45,47). The higher apparent affinity of GLUT3 for glucose and other sugars compared with GLUT1 in competition assays of DHA transport could therefore reflect differences in sugar binding affinity to intracellular sites on the glucose transporter between the two isoforms.
Because glucose transport proteins possess both internal and external binding sites for substrate (41,43), we examined if inhibition of DHA uptake utilizing either the exofacial inhibitor 4,6-O-ethylidene ␣-glucose (43,44), or the endofacial inhibitor cytochalasin B (41,42) would demonstrate different kinetics. Competitive inhibition by ethylidene glucose suggests that the external binding site of DHA is the same as that of 2-DG for both GLUT1 and GLUT3. Other data supporting this conclusion are the lack of transport of DHA by single amino acid substitution mutants of GLUT1 and GLUT3, and the similarity in rank order of inhibition of various sugars on DHA transport through GLUT1 and GLUT3. In experiments with cytochalasin B, competition at the internal site could not occur because DHA was completely reduced intracellularly to AA. Consequently, noncompetitive competition was expected and was observed. These findings, however, do not rule out the possibility that intracellular binding sites are the same. A non-reducible DHA analog would be a helpful tool to examine this issue once it becomes available.
One report examined DHA transport by glucose transporter isoforms GLUT1, -2, and -4 in oocytes and concluded that DHA transport via GLUT1 demonstrated both high (60 M) and low affinity (3.5 mM) transport processes (21). These data as presented, however, had several flaws which were previously explained in detail (7). Briefly, prior experiments were performed utilizing mixtures of ascorbic acid and ascorbic acid oxidase instead of pure DHA as substrate. Uptake kinetics may therefore have been confounded by the rate of ascorbate oxidation in the extracellular medium. Although the investigators stated that intracellular reduction was complete, these data were not well supported. We demonstrated here that intracellular reduction can alter apparent DHA uptake kinetics, and verification of DHA reduction at all external DHA concentrations must be performed to calculate transport kinetics. Previously, DHA transport activity was also attributed to both GLUT2 and GLUT4, although no data was presented (21). We found no transport activity in GLUT2 expressing oocytes and very low activity in GLUT4 expressing oocytes under circumstances where hexose transport was observed.
AA was not transported by any of the glucose isoforms tested. This is consistent with the results of previous studies demonstrating that AA and DHA are transported by separate mechanisms (7). The ascorbic acid transport protein has not yet been cloned, although transport activity has been reported in oocytes injected with fractionated rabbit renal mRNA (48).
We did not examine DHA or AA transport by either GLUT7 or SGLT2. GLUT7 is localized to the microsomal membrane  1 N), 1% CHAPS, and cell associated radioactivity was quantified. Data shown represents the difference between total radiolabel uptake and nonspecific uptake in the presence of 10 M cytochalasin B (5-10% of total uptake). Mean Ϯ S.D. (n ϭ 4). (49) and thus is not pertinent to the transport of DHA or AA from extracellular domain. SGLT2 demonstrates very low glucose transport activity in oocyte expression systems (50), and long incubation times of up to 2 h are required. This is problematic for DHA transport studies because DHA oxidation becomes significant after 20 min at 23°C.
The extent to which glucose isoforms mediate DHA transport in vivo remains to be determined. GLUT1 has wide tissue distribution (28,38) while GLUT3 is primarily expressed in brain, placenta, testis, and platelets (46,(51)(52)(53)(54). DHA uptake has been demonstrated in many human tissues including neutrophils (7,21), fibroblasts (20), erythrocytes (55), platelets (56), and placenta (57), all of which express relatively high levels of either GLUT1 or GLUT3. Little is known about DHA uptake in brain. Despite the ubiquitous nature of glucose transporters in vivo and evidence presented here and elsewhere (21) that glucose transporter isoforms mediate the transport of DHA, it remains possible that there are other mechanisms of DHA transport in mammalian cells (55).
DHA uptake could theoretically be modulated by either DHA reduction or transport itself. In neutrophils internal DHA reduction is complete at external DHA concentrations Ͻ800 M (7). Based on these observations, DHA transport appears to be rate-limiting in neutrophils, although reduction rates may differ in other tissues. Physiologic concentrations of plasma DHA are certainly Ͻ100 M, and are most likely less than 1-5 M (58). Because DHA concentrations are well below the apparent K m for DHA transport of both GLUT1 and GLUT3 (approximately 1.5 mM), regulation of DHA transport is potentially sensitive to a number of alterations including substrate availability, transporter affinity, and transporter number. GLUT1mediated glucose transport is increased by a number of growth factors and mitogens (59). The resulting increased transport has both an early and a late phase response, likely reflecting both recruitment of transporter to the cell surface and new protein synthesis (59). Less is known about the regulation of GLUT3 transport, although recruitment of GLUT3 has been recently demonstrated in platelets in response to thrombin stimulation. 2,3 Both GLUT1 and GLUT3 expression have been shown to be elevated in various human carcinomas (60).
Variations in plasma glucose could effect DHA uptake through these transport proteins. In the present study the IC 50 of D-glucose on transport of 150 M DHA was 10 and 4 mM, for GLUT1 and GLUT3, respectively, which are within the physiologic range of plasma glucose. Consistent with such inhibition, higher levels of DHA have been reported in diabetic plasma (61)(62)(63)(64)(65). The significance of these findings (66,67) and whether DHA transport and cellular AA accumulation is aberrant in diabetic individuals remains to be elucidated.