Different functional domains of GLUT2 glucose transporter are required for glucose affinity and substrate specificity.

GLUT2 is the major glucose transporter in pancreatic beta-cells and hepatocytes. It plays an important role in insulin secretion from beta-cells and glucose metabolism in hepatocytes. To better understand the molecular determinants for GLUT2's distinctive glucose affinity and its ability to transport fructose, we constructed a series of chimeric GLUT2/GLUT3 proteins and analyzed them in both Xenopus oocytes and mammalian cells. The results showed the following. 1) GLUT3/GLUT2 chimera containing a region from transmembrane segment 9 to part of the COOH-terminus of GLUT2 had Km values for 3-O-methylglucose similar to those of wild-type GLUT2. Further narrowing of the GLUT2 component in the chimeric GLUTs lowered the Km values to those of wild-type GLUT3. 2) GLUT3/GLUT2 chimera containing a region from transmembrane segment 7 to part of the COOH-terminus of GLUT2 retained the ability to transport fructose. Further narrowing of this region in the chimeric GLUTs resulted in a complete loss of the fructose transport ability. 3) Chimeric GLUTs with the NH2-terminal portion of GLUT2 were unable to express glucose transporter proteins in either Xenopus oocytes or mammalian RIN 1046-38 cells. These results indicate that amino acid sequences in transmembrane segments 9-12 are primarily responsible for GLUT2's distinctive glucose affinity, whereas amino acid sequences in transmembrane segments 7-8 enable GLUT2 to transport fructose. In addition, certain region(s) of the amino-terminus of GLUT2 impose strict structural requirements on the carboxy-terminus of the glucose transporter protein. Interactions between these regions and the carboxy-terminus of GLUT2 are essential for GLUT2 expression.

G LUT2 is the major glucose transporter in ␤-cells of pancreatic islets and hepatocytes. In both cell types, GLUT2 mediates the facilitated diffusion of glucose across the cell membranes, and then intracellular glucose metabolism is initiated by the glucose-phosphorylating enzyme, hexokinase IV or glucokinase. In the ␤-cell, the rate of glucose metabolism controls insulin secretion, whereas in the liver, glucose metabolism and transport are essential to subsequent glycogen synthesis and gluconeogenesis (1)(2)(3)(4)(5)(6). GLUT2 is a member of a family of facilitative glucose transporters (GLUT) that share a high degree of protein homology and a putative protein structure (7,8). All GLUT isoforms are predicted to traverse the plasma membrane 12 times, with intracellular amino-and carboxy-termini, an extracellular loop between the first and second transmembrane segments, and a large intracellular loop between transmembrane segments 6 and 7 (9 -11). The primary amino acid sequences of the putative transmembrane segments and the short loops connecting the transmembrane regions are highly conserved among the different glucose transporter isoforms (7,8). Despite these overall similarities, several lines of evidence suggest that GLUT isoforms are not interchangeable and that features of the GLUT2 protein may allow it to perform specialized functions. First, each GLUT isoform can be distinguished by its glucose affinity and pattern of tissue-specific expression (7,8,12). The kinetic properties of different glucose transporters are determined by their primary amino acid sequences (13)(14)(15). Because of GLUT2's lower glucose affinity (ϳ30 mm compared with 1-10 mm for the other GLUT isoforms), elevations in the blood glucose level during the postprandial period linearly increase the velocity of glucose transport and the intracellular glucose concentration. Both transport of glucose into the liver during the postprandial state and transport of glucose out of the hepatocyte after gluconeogenesis benefit from GLUT2's high K m , because increases in the velocity transport would parallel increases in either the postprandial glucose concentration or the intracellular hepatocyte glucose concentration. Furthermore, there is a correlation between the glucose affinity of GLUT isoforms and the glucose affinity of the hexokinases with which the GLUT isoforms are expressed. For example, GLUT2 is usually expressed in tissues that express glucokinase, which has the highest affinity of the hexokinases (16,17). Second, introduction of GLUT2, as opposed to GLUT1, into insulin-secreting cells results in a more normal insulin secretory profile (18,19). Third, animal studies of transplanted pancreatic islets suggest that GLUT2 expression is required for normal glucose-stimulated insulin secretion (20,21). Fourth, although all members, except GLUT5, can efficiently transport glucose, only GLUT2 and GLUT5, are capable of transporting fructose (14,22,23). This property allows for the uptake of fructose by liver and for transport of fructose across the basolateral membrane of epithelial cells of the small intestine. GLUT2's ability to transport fructose may have physiological relevance, because fructose utilization by the liver influences glucose metabolism, and fructose metabolites modulate the activity of glucokinase and its regulatory protein (24,25).
The molecular determinants that endow GLUT2 with these specialized functions are incompletely understood. One hypothesis is that regions of the GLUT2 protein may be important in the specialized functions of GLUT2 in glucose utilization and insulin secretion. To address this question, we constructed a series of chimeric proteins comprised of GLUT2 and GLUT3 components and analyzed the function of the chimeric GLUT proteins in both Xenopus oocytes and mammalian cells. GLUT3, the primary glucose transporter of neurons, has a high affinity for glucose (14,26). We used these chimeric proteins to identify the region(s) that is responsible for GLUT2's distinctive high K m and its ability to transport fructose. Oligolabeling kits for synthesizing double stranded DNA probes were purchased from Pharmacia Biotech (Piscataway, NJ). Affinity-purified rabbit antihuman GLUT3 was obtained from Charles River Pharmservices (Southbridge MA). Peroxidase-conjugated affinity-purified goat antirabbit IgG was purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). ECL Western blotting detection kits were purchased from Amersham (Arlington Heights, IL). Plasmids pCA14 and pJM17 as well as 293 cells were purchased from Microbix Biosystems (Toronto, Canada). Oligonucleotides were synthesized in the DNA Core Laboratory of the Vanderbilt Diabetes Research and Training Center.

Construction of GLUT2/GLUT3 chimeric complementary DNAs (cDNAs)
The cDNAs encoding the human glucose transporters GLUT2 and GLUT3 were gifts from Dr. Graeme Bell (University of Chicago, Chicago, IL) and were used as previously reported to construct chimeric GLUT cDNAs (27). The chimeric glucose transporter cDNAs, comprised of fragments from human GLUT2 and GLUT3, were constructed by gene overlap extension PCR or the introduction of unique restriction enzyme sites into GLUT2 cDNA by PCR and exchange of the cDNA fragments between GLUT2 and GLUT3. The design of these chimeric cDNAs was based on the predicted two-dimensional structure of GLUT proteins proposed by Mueckler et al. (8,9). All chimeric GLUTs were created so that the predicted transmembrane segments were not disrupted but were replaced by the corresponding regions of the other glucose transporter protein. For overlap extension PCR, the GLUT2 and GLUT3 cDNA fragments were amplified separately in a first round of PCR from the coding sequence of human GLUT2 and human GLUT3 cDNAs. The generated GLUT2 and GLUT3 cDNA fragments were subsequently annealed in a second round of PCR and extended in a third round of PCR as described previously (27). DNA polymerase pfu (Stratagene, La Jolla, CA) was chosen to increase the fidelity of DNA synthesis during the PCR. The chimeric GLUT cDNAs were subcloned into a modified oocyte expression vector (pGOV) and checked by sequencing as previously described (27).
The chimeric GLUTs created in the present study are listed in Table  1. The nomenclature for these chimeric proteins is based on the primary structure of the glucose transporter proteins. Each name starts with the NH 2 -terminal portion of the chimeric GLUT protein, with the number in parentheses indicating the amino acid boundary of this portion. GLUT2 is referred to as G2, and GLUT3 is referred to as G3. As GLUT2 is 524 amino acids in length and GLUT3 is 479 amino acids in length (due to a larger extracellular loop between transmembrane segments 1 and 2 of GLUT2), the locations of predicted transmembrane segments do not occur at identical amino acid positions. The predicted two-dimensional structures of these chimeric GLUT proteins are shown in Fig. 2.

In vitro transcription of capped RNA
The pGOV oocyte expression vector, containing different GLUT cDNAs, was linearized downstream of the 3Ј-globin untranslated region and in vitro transcribed into capped RNA by T7 RNA polymerase with a messenger RNA (mRNA) capping kit. The yield of in vitro transcription reactions was assessed by spectrophotometry at 260 nm. Verification of full-length transcription of the cDNA was confirmed by electrophoresis on formaldehyde-agarose gels. RNA for microinjection was dissolved in diethylpyrocarbonate-treated water and stored at Ϫ70 C until use.

Expression of GLUT isoforms in Xenopus oocytes
Oocytes for microinjection were isolated from female Xenopus laevis (Nasco, Fort Atkinson, WI) as previously described (27). Capped RNA (10 -20 ng) in a volume of 10 -20 nl was microinjected into each oocyte within 24 h of isolation, using a Picospritzer II (General Valve Corp., Fairfield, NJ) and a micromanipulator (Singer Instruments Co., Somerset, UK). Microinjected oocytes were incubated at 18 C for 48 h before assay, during which time the medium was changed every 12 h, and unhealthy oocytes were removed.

3OMG and fructose transport assays
Two days after RNA injection, healthy oocytes were selected for 3OMG transport assay. To determine maximal GLUT expression, oocytes injected with increasing amounts of GLUTs RNA were incubated in 1 mm 3OMG and 1.2 Ci [ 3 H]3OMG at room temperature for 15 min. The transport of methylglucose was stopped by washing oocytes four times with ice-cold buffer containing 100 m phloretin. The oocytes were transferred to scintillation vials and lysed; radioactivity was quantitated by liquid scintillation counting. To determine the equilibrium exchange K m for methylglucose, the oocytes were incubated in a range of 3OMG concentrations (1,5,10,15,20,30, and 40 mm) at 18 C for 15-18 h to reach equilibrium exchange conditions. Ten oocytes for each glucose concentration were used as a group, and each assay was performed in triplicate. Before each assay, a preliminary uptake was performed by incubating oocytes (under equilibrium exchange conditions with 1 mm 3OMG) with 1.2 Ci [ 3 H]3OMG at room temperature for 60 min. This incubation time is sufficient for the oocytes to reach maximal uptake of [ 3 H]3OMG. The uptake of [ 3 H]3OMG reached with the 60-min incubation and with incubations of 1-10 min was used to select a transport assay time (1)(2)(3)(4)(5), so that the uptake of [ 3 H]3OMG was less than 20% of that reached during the 60-min incubation. For the final assay, transport was initiated by adding 1.2 Ci [ 3 H]3OMG into each group of oocytes equilibrated within the different 3OMG concentrations. The transport of methylglucose was stopped by washing oocytes four times with ice-cold buffer containing 100 m phloretin. The oocytes were transferred to scintillation vials and lysed; radioactivity was quantitated by liquid scintillation counting. Each assay was repeated at least four times for each chimeric glucose transporter.
For measurement of fructose transport, 2 days after RNA injection healthy oocytes were selected and incubated in 1 mm of fructose at 18 C for 15-18 h to reach equilibrium exchange conditions. Ten healthy oocytes were used as a group, and each assay was performed in triplicate. To initiate the transport assay, 0.13 Ci [ 14 C]fructose was added to each group of oocytes. The oocytes were then incubated at room temperature for 60 min. The uptake of fructose was stopped by washing the oocytes four times with ice-cold buffer containing 100 m phloretin. The oocytes were transferred to scintillation vials and lysed. The radioactivity was quantitated by liquid scintillation counting. A transport assay for 1 mm 3OMG was performed in parallel for each glucose transporter.

Kinetic analysis of 3OMG and fructose transport
The kinetic analysis of sugar uptake was performed by calculating the clearance of the oocyte medium at various sugar concentrations. Linear regression of a Hanes plot [(substrate concentration/velocity) vs. substrate concentration] was used to calculate the K m for methylglucose transport. A time course of methylglucose transport at 1 mm 3OMG was performed for each experiment to ensure that initial rates of transport were obtained. The K m values from individual experiments were averaged and presented as the mean and sem. Statistical comparisons of the K m for chimeric GLUTs used Student's unpaired t test.

Construction of recombinant adenoviruses carrying wild-type or chimeric GLUT cDNAs
Recombinant adenoviruses carrying wild-type GLUT2, wild-type GLUT3, or chimeric GLUT2-(1-303)/GLUT3-(270 -497) (chimera 9 in Table 1) cDNAs were generated by homologous recombination as previously described (28). The cDNA encoding each GLUT was subcloned into plasmid pCA14. The resulting plasmid was cotransfected with pJM17 into 293 cells by calcium phosphate transfection. Two to 3 weeks later, individual clones were chosen and confirmed by Southern blot analysis. The recombinant adenovirus was amplified in 293 cells and purified by cesium chloride centrifugation. The multiplicity of infection (MOI) of the viral stock was determined by a plaque assay with 293 cells and expressed as infectious units per microliter. The viral stocks were stored at Ϫ70 C until use. The terms for the recombinant adenoviruses used in this study are AdGLUT2 (containing wild-type GLUT2 cDNA), AdGLUT3 (containing wild-type GLUT3 cDNA), and AdGLUT2/ GLUT3 [containing chimeric GLUT2-(1-303)/GLUT3-(270 -497) cDNA].

In vitro transduction of RIN 1046-38 cells
RIN 1046-38 cells were cultured in DMEM supplemented with 10% FBS, 100 U/ml penicillin, and 100 g/ml streptomycin. The cells were infected with a recombinant adenovirus by adding the desired amount of viral stock directly to the culture medium. The MOI was selected by examining cell survival and transgene expression as a function of increasing MOI. AdGLUT2 was used at a MOI of 80. AdGLUT3 was used at a MOI from 10 -50. AdGLUT2/GLUT3 was used at a MOI from 5-50. The viruses were removed 24 h later by washing the cells twice with PBS. The cells were harvested for Northern blot analysis after washing or were cultured for another 24 h in culture medium for immunoblot analysis.

Northern blot analysis of GLUT RNA
Total RNA was prepared from RIN 1046-38 cells using the RNeasy minikit. Northern blot analysis was performed by separating 15 g total RNA on 1.5% formaldehyde-agarose gels. The RNA was transferred onto a nylon membrane by capillary blotting. 32 P-Labeled double stranded DNA probes were synthesized from human GLUT2-(1-303)/ GLUT3-(270 -497) cDNA using an oligolabeling kit. This probe hybridizes to both GLUT2 and GLUT3 mRNA. The hybridizations were performed at 42 C overnight with formamide hybridization solution. The membranes were then washed with 2 ϫ SSC-0.1% SDS twice at room temperature and with 0.1 ϫ SSC-0.1% SDS twice at 65 C, then subjected to autoradiography.

Western blot analysis of GLUT protein
RIN 1046-38 cells were lysed in ice-cold lysis buffer [0.5% Nonidet P-40, 5 mm MgCl 2 , 50 mm Tris (pH 7.4), and 0.1 mm phenymethylsulfonylfluoride]. The total cellular proteins were separated on 10% SDS-PAGE gels and electrotransferred to an Immobilon-P membrane (Millipore, Bedford, MA). The affinity-purified rabbit antihuman GLUT3 was used as primary antibody at a 1:500 dilution. This antibody reacts only with the carboxy-terminus of human GLUT3. The affinity-purified peroxidase-conjugated goat antirabbit IgG was used as a secondary antibody at a dilution of 1:10,000. Antibody binding was detected by chemiluminescence using the ECL Western blotting detection kit according to the manufacturer's protocol.

Expression of wild-type and chimeric GLUTs in Xenopus oocytes
After injection of increasing amounts of GLUT2 or GLUT3 RNA, methylglucose uptake by oocytes reached a maximum at 10 -15 ng of both GLUT2 and GLUT3 RNA (Fig. 1). A time course of 3OMG accumulation, performed under the condition of maximal GLUT expression, was linear for approximately 15 min and reached a maximum at 60 min for GLUT2and GLUT3-expressing oocytes (data not shown). The oocytes expressing chimeric GLUTs demonstrated properties similar to those of GLUT2-or GLUT3-expressing oocytes for methylglucose transport (data not shown). Under the maximal GLUT expression and equilibrium exchange conditions, the average velocity of 1 mm 3OMG transport was 24.83 Ϯ 0.41 pmol/min⅐oocyte (mean Ϯ sem) for wild-type GLUT2 and 25.21 Ϯ 0.14 pmol/min⅐oocyte (mean Ϯ sem) for wildtype GLUT3 (Table 1). All chimeric proteins containing the NH 2 -terminal portion of GLUT3 (chimeras 2-7 in Table 1) had uptake rates for 3OMG similar to those of wild-type  GLUT2 and wild-type GLUT3 (P Ͼ 0.05), with the exception of chimera 1, which had a rate significantly lower than those of the wild-type GLUTs (P Ͻ 0.01). The two chimeric proteins with the NH 2 -terminal portion of GLUT2 (chimeras 8 and 9 in Table 1) did not transport more 3OMG than sham-injected oocytes (Table 1).

Equilibrium exchange K m for methylglucose of wild-type and chimeric GLUTs
Under equilibrium exchange conditions, the apparent K m for 3OMG was 34.63 Ϯ 2.97 mm (mean Ϯ sem) for wild-type GLUT2 and 15.57 Ϯ 1.97 mm for wild-type GLUT3 (Fig. 2). GLUT3/GLUT2 chimeric proteins containing a region from transmembrane segment 9 to the initial 20 amino acid residues of the COOH-terminus of GLUT2 (chimeras 1, 2, and 4 in Table 1 and Fig. 2) had K m values similar to that of wildtype GLUT2 and different from that of wild-type GLUT3 (P Ͼ 0.05 compared with the K m of wild-type GLUT2 and P Ͻ 0.01 compared with the K m of wild-type GLUT3). Based on these results, we constructed two additional chimeric proteins to determine the specific region(s) responsible for GLUT2's high K m . Replacing the region from transmembrane segment 9 to the loop connecting transmembrane segments 10 and 11 of GLUT2 with the corresponding region of GLUT3 (chimera 5) resulted in a K m intermediate between those of wild-type GLUT2 and wild-type GLUT3 (P Ͻ 0.05 compared with the K m of wild-type GLUT3 and P ϭ 0.055 compared with the K m of wild-type GLUT2). When only transmembrane segments 9 and 10 and the connecting loop between these two segments of GLUT2 were present in the chimeric GLUT3/GLUT2 (chimera 6), the K m was similar to that of wild-type GLUT3 and different from that of wild-type GLUT2 (P Ͼ 0.05 compared with the K m of wild-type GLUT3 and P Ͻ 0.01 compared with the K m of wild-type GLUT2; Fig.  2). These results indicate that the amino acid residues responsible for GLUT2's distinctive high K m reside mainly in the region from transmembrane segments 9 -12. Chimera 3 contains a region from GLUT2 that is similar to chimera 2 and also has nine amino acid residues from GLUT2's intracellular loop between transmembrane segments 6 and 7. This chimera had a K m of 48 Ϯ 4.62 mm (mean Ϯ sem), which is significantly higher than the K m of wild-type GLUT2 (Fig. 2).

Fructose transport of wild-type and chimeric GLUTs
GLUT2 can also transport fructose across the cell membrane, whereas GLUT3 does not have this ability (Fig. 3). Under the condition of maximal GLUT expression, the equilibrium exchange uptake rate for 1 mm fructose was 2.46 Ϯ 0.21 pmol/min⅐oocyte (mean Ϯ sem) for wild-type GLUT2. The wild-type GLUT3 had an uptake rate of 0.58 Ϯ 0.02 pmol/min⅐oocyte (mean Ϯ sem), which was not significantly different from that in sham-injected oocytes (data not shown). To localize the region(s) of GLUT2 responsible for this property, fructose transport by our series of GLUT3/ GLUT2 chimeric proteins was examined. The GLUT3/ GLUT2 chimera containing a region from transmembrane segment 7 to part of the COOH-terminus of GLUT2 (chimera 3) retained the ability to transport fructose (P Ͼ 0.05 compared with wild-type GLUT2 and P Ͻ 0.01 compared with wild-type GLUT3). Further narrowing of this region in the GLUT3/GLUT2 chimeric proteins (chimeras 4 -6) resulted in a complete loss of the fructose transport ability (Fig. 3). In FIG. 2. Equilibrium exchange K m for 3OMG of wild-type and chimeric glucose transporters. Twenty nanograms of each GLUT RNA were injected into oocytes, which were subsequently incubated at 18 C as previously described. Two days after RNA injection, the oocytes were incubated in a range of 3OMG concentrations (1,5,10,15,20,30, and 40 mM) at 18 C for 15-18 h to reach equilibrium. The transport assay was performed as described in Materials and Methods. The kinetic analysis of 3OMG uptake was performed by calculating the clearance of the oocyte medium at various 3OMG concentrations. Linear regression of a Hanes plot [(substrate concentration/velocity) vs. substrate concentration] was used to calculate the K m of 3OMG transport. The experiment was performed at least four times for each glucose transporter. The K m values from individual experiments were averaged and presented as the mean and SEM. Refer to Table 1 for amino acid numbering of each chimera. *, Not different from each other (P Ͼ 0.05), but significantly different from wild-type GLUT3 (P Ͻ 0.01); **, significantly different from wild-type GLUT3 (P Ͻ 0.05); P ϭ 0.055 compared with wild-type GLUT2 and P Ͻ 0.05 compared with chimera 2; ***, not different from wild-type GLUT3, but significantly different from wild-type GLUT2; ****, significantly higher than that of wild-type GLUT2 (P Ͻ 0.05).
parallel experiments, all chimeric proteins had similar uptake rates for 1 mm 3OMG (Fig. 3). This suggests, but does not prove, that GLUT expression is similar among the chimeric proteins (the differences in K m may somewhat influence the transport rate).

Expression of wild-type GLUTs and chimeric GLUT2/GLUT3 in RIN 1046-38 cells
Two chimeric proteins that contain the NH 2 -terminal portion of GLUT2 did not demonstrate transport activity for methylglucose when injected into Xenopus oocytes (Table 1). To investigate the molecular basis for the inability to transport glucose, we also expressed chimeric GLUT2-(1-303)/ GLUT3-(270 -497), wild-type GLUT2, and wild-type GLUT3 in RIN 1046-38 cells using adenovirus-mediated gene transfer. Transduction of RIN 1046-38 cells with AdGLUT2, Ad-GLUT3, and AdGLUT2/GLUT3 resulted in expression of mRNAs with the expected sizes, as demonstrated by Northern blot analysis (Fig. 4A). However, no GLUT2/GLUT3 protein was detected by immunoblotting (Fig. 4B).

Discussion
The GLUT2 glucose transporter plays a key role in glucose homeostatic processes in both the liver and the ␤-cell. By creating and expressing a series of GLUT2/GLUT3 chimeric proteins in Xenopus oocytes and mammalian cells, the current study has localized molecular determinants that enable GLUT2 to perform some of these physiological functions. Amino acid sequences in transmembrane segments 9 -12 of GLUT2 are primarily responsible for its glucose affinity, whereas amino acid sequences in transmembrane segments 7-8 enable it to transport fructose. In addition, interactions between the amino-terminus and the carboxy-terminus of GLUT2 are essential for GLUT2 expression.

Glucose affinity of GLUT2
Based on prior studies, the distinctive glucose affinity of GLUT2 is a function of the carboxy half of the protein (27,29). Using a series of chimeric proteins that exchanged transmembrane segments 7-12 of GLUT2 for corresponding regions of GLUT3, we found that transmembrane segments 9, 10, 11, and 12 are required for this property. GLUT2's distinctive glucose affinity requires the presence of all four transmembrane segments, as a chimeric protein that excluded transmembrane segments 9 and 10 (chimera 5) had a K m intermediate between those of wild-type GLUT2 and wild-type GLUT3, whereas a chimeric protein that contained only transmembrane segments 9 and 10 (chimera 6) had a K m similar to that of GLUT3. These results suggest that amino acids in transmembrane segments 9 -12, which are GLUT2 specific, interact with the glucose molecule to facilitate its transport. Models of glucose transporter topography predict FIG. 3. Uptake rates for fructose and 3OMG of wild-type and chimeric glucose transporters. Twenty nanograms of each GLUT RNA were injected into oocytes, which were subsequently incubated at 18 C. Two days after RNA injection, the oocytes were incubated in 1 mM fructose or 3OMG at 18 C for 15-18 h to reach equilibrium. The transport assays were performed as described in Materials and Methods. The uptake rates were calculated based on the clearance of the oocyte medium. The results from individual experiments were averaged and are presented as the mean and SEM. Refer to Table 1 and Fig.  2 for amino acid numbering of each chimera. *, Not different from each other (P Ͼ 0.05), but significantly different from water-injected oocytes (data not shown in the figure); **, not different from waterinjected oocytes, but significantly different from wild-type GLUT2 (P Ͻ 0.01). Total cellular proteins were separated on 10% SDS-PAGE gels and transferred onto Immobilon-P membranes. Immunoblotting was performed with an ECL Western blotting detection kit. Affinity-purified rabbit antihuman GLUT3 was used as primary antibody, and affinitypurified, peroxidase-conjugated goat antirabbit IgG was used as secondary antibody. Lane 1, Cellular proteins from untransduced cells. that these transmembrane segments are part of the pore through which glucose traverses the membrane (9). One envisions that amino acids specific for GLUT2 in these transmembrane segments determine its glucose affinity, but that the overall membrane pore and the ability to transport glucose are common to all glucose transporter isoforms. These results also point out a limitation of using chimeric proteins to study protein function, in that regions that are not contiguous by primary amino acid sequence may cooperate functionally, and this must be taken into consideration when designing chimeric proteins. In addition, further narrowing of this regions using additional chimeric proteins is unlikely. As continuous amino acids are not required, future efforts will require assessment of glucose affinity after mutation of combinations of amino acids widely separated in the primary amino acid sequence of GLUT2.
The present results contrast with those of Arbuckle et al., who concluded that transmembrane segment 7 alone was responsible for GLUT2's K m (30). The reason for this disagreement is not clear, as both laboratories examined the function of chimeric GLUT2/GLUT3 proteins in Xenopus oocytes. One possible explanation is that the current study used 3OMG, which measures only glucose transport, whereas Arbuckle et al. used 2-deoxyglucose, which requires both glucose transport and phosphorylation (31). In addition, the current study used a series of chimeric proteins that span transmembrane segments 7-10, whereas Arbuckle et al. used only a single chimeric protein to reach their conclusion. The K m of this chimeric protein in the work of Arbuckle et al. was lower than the K m of GLUT2, but higher than the K m of GLUT3. This suggests that additional regions of GLUT2 participate in fully determining GLUT2's glucose affinity. As none of the chimeric proteins examined by Arbuckle et al. involved exchanges of transmembrane segments 8 -10, the contribution of this area to GLUT2's K m was not detected. The reason why chimera 3 (containing transmembrane segments 7-12, part of the carboxy-terminal tail, and nine amino acids of the intracellular loop of GLUT2) had a K m significantly higher than that of wild-type GLUT2 is not clear. One possible explanation is that this altered the glucose-binding site and therefore affected the glucose affinity (32,33). Prior work by our laboratory and by Noel and Newgard has demonstrated a minor contribution of GLUT2's intracellular carboxy-tail to GLUT2's glucose affinity (27,29,34). Thus, the current findings, integrated with our prior work and the findings of Arbuckle et al. and Noel and Newgard, support the concept that several regions in the carboxy half of a glucose transporter protein cooperate to determine its glucose affinity.

Fructose transport
Fructose transport is a property of only GLUT2 and GLUT5 (14,22), and the current study indicates that transmembrane segments 7 and 8 of GLUT2 are essential for this property (chimera 3 in Fig. 3). These results indicate that the requirements for binding and transporting fructose are slightly different from the protein regions responsible for GLUT2's glucose affinity. Our findings clarify conflicting results from other laboratories regarding fructose transport.
Based on a series of GLUT5/GLUT1 chimeric proteins expressed in selected clones of Chinese hamster ovary cells, Inukai et al. concluded that the entire GLUT5 protein was required for fructose transport (35). However, none of the chimeric proteins used in that study exhibited increased glucose transport, so a conclusion about their ability to transport fructose is not possible. Noel and Newgard found that a GLUT2/GLUT1 chimera consisting of the amino-terminus, the first transmembrane segment, and the extracellular loop of GLUT1, with the remainder of the protein being GLUT2, was able to transport fructose but at a reduced velocity compared with wild-type GLUT2 (34). These researchers concluded that the amino-terminal region of GLUT2 was important for fructose transport. The current study and the work of Arbuckle et al. (30) indicate that the amino-terminus of GLUT2 is not required for fructose transport and that transmembrane segments 7 and 8 are the essential regions. Both studies examined a series of GLUT2/GLUT3 chimeric proteins, whereas Noel and Newgard studied only one GLUT2/GLUT1 chimeric protein. It is not known whether using GLUT1 rather than GLUT3 is responsible for the different results. Arbuckle et al. further suggested that the affinity for fructose was modulated by transmembrane segments 8 -12. Interestingly, this is the same region identified by the current study to determine GLUT2's glucose affinity. The current study cannot assess whether transmembrane segments 9 -12 modulate the affinity for fructose. Thus, the ability of GLUT2 to transport fructose requires amino acids in transmembrane segments 7 and 8, but transmembrane segments 9 -12 may modulate the affinity for fructose.

Role of amino-terminal region(s) of GLUT2
Chimeric proteins that contained the amino-terminus, the first transmembrane segment, and the extracellular loop of GLUT2 were not expressed either in the oocyte system or in mammalian cells using adenovirus-mediated gene transfer. Even when the entire amino half of the chimeric protein was GLUT2, there was no glucose transport. This finding with GLUT2/GLUT3 chimeric proteins confirms our earlier observation with similar GLUT2/GLUT4 chimeric proteins (27), the findings of Noel and Newgard with GLUT2/GLUT1 chimeric proteins (34), and the findings of Arbuckle et al. with GLUT2/GLUT3 chimeric proteins (30). Both our laboratory and that of Noel and Newgard were able to in vitro translate chimeric proteins with the amino-terminal regions of GLUT2 into a protein of the expected size (34). Neither the current study nor that by Noel and Newgard could detect the chimeric GLUT protein from cells that express the mRNA for the chimeric protein. Taken together, these findings from three different laboratories with chimeric proteins of GLUT1, GLUT3, and GLUT4 indicate that the amino-terminal region(s) of GLUT2 imposes strict structural requirements on the carboxy-terminus of the glucose transporter protein. As the amino-terminus of GLUT1, GLUT3, and GLUT4 can participate with the carboxy-terminus of GLUT2, this property is peculiar to GLUT2. As GLUT2's extracellular loop is distinctive (64 amino acids compared with 32 amino acids for GLUT1, GLUT3, and GLUT4, with no sequence homology) (7), it is likely that this region is responsible for the lack of expression of these chimeric proteins. These findings suggest that a failure of the amino-terminal regions of GLUT2 to interact with other GLUT isoforms interferes with protein stability within the cell. Cope et al. proposed that the aminoterminus of GLUT1 provides a scaffold for the folding of the carboxy-terminus of the protein (33), and these findings with GLUT2 support this hypothesis. The amino-terminal region of GLUT2 does not influence glucose or fructose transport substantially, so the functional purpose of GLUT2's distinctive extracellular loop remains obscure.

Model of functional domains of GLUT2
Based on the results of the current study and the work of other laboratories, a model of functional domains of glucose transporter proteins is emerging. Transmembrane segments 7-8 appear to form a functional domain that is responsible for binding the substrate (glucose/fructose), whereas another functional domain in transmembrane segments 9 -12 may modulate the affinity of the transporter for the substrate. The intracellular carboxy-tail of the transporter also appears to independently module affinity for the substrate and is involved in the targeting of GLUT4 to its distinctive intracellular location (36,37). These functional domains interact to produce the final substrate specificity and substrate affinity that are distinctive for each GLUT isoform. As yet, no functional role has been identified for the extracellular loop between transmembrane segments 1 and 2 or the intracellular loop between transmembrane segments 6 and 7, but specific interactions of the amino-terminus of GLUT2 are required for protein stability.
Whether regions of any GLUT isoforms interact with other membrane or cytoplasmic proteins is not known, but such interactions could provide yet another mechanism for modifying transporter function. It is also unresolved whether GLUT isoforms are interchangeable. For example, could any GLUT isoform substitute for GLUT2 in the liver or pancreatic ␤-cell, or are certain features of GLUT2 required? Have the distinctive transporter kinetics and the tissue-specific expression patterns evolved in response to certain requirements of cellular physiology? Studies in vivo to ablate a GLUT isoform in a certain tissue, then introduce a different GLUT isoform and assess the physiological consequences will be required to answer such questions.