Human Small Intestine Facilitative Fructose/Glucose Transporter (GLUT5) Is Also Present in Insulin-Responsive and Brain Investigation of Biochemical Characteristics and Translocation PETER R. SHEPHERD, E. MICHAEL GIBBS, CHRISTIAN WESSLAU, GWYN W. GOULD, AND BARBARA B. KAHN A recent study by C.F. Burant et al. (13) demonstrates that GLUT5 is a high-affinity fructose transporter with a much lower capacity to transport glucose. To arbohydrate metabolism is vital to all mamma- characterize the potential role of GLUT5 in fructose lian cells. Recently, molecular cloning studies and glucose transport in insulin-sensitive tissues, we investigated the distribution and insulin-stimulated have identified a family of at least five facilitated translocation of the GLUTS protein in human tissues diffusion glucose transporters in mammalian Ccells (GLUT1-5) (1,2). One of these isoforms (GLUT4) is by immunoblotting with an antibody to the COOH-terminus of the human GLUTS sequence. expressed primarily in muscle and fat, the tissues in GLUTS was detected in postnuclear membranes from which insulin markedly stimulates glucose transport by the small intestine, kidney, heart, four different skeletal recruiting GLUT4 from an intracellular pool to the plasma muscle groups, and the brain, and in plasma membrane (3). GLUT1 is present in much lower amounts membranes from adipocytes. Cytochalasin-B than GLUT4 in adipose cells (4) and muscle (5) and photolabeled a 53,000-/tfr protein in small intestine undergoes a much smaller translocation to the plasma membranes that was immunoprecipitated by the membrane. In addition, growing evidence suggests that GLUT5 antibody; labeling was inhibited by D- but not under some circumstances, changes in glucose trans- L-glucose. M-glycanase treatment resulted in a band of porter intrinsic activity (i.e., number of glucose molecules 45,000 Afr in all tissues. Plasma membranes were prepared from isolated adipocytes from 5 nonobese transported/transporter/unit time) may be an additional and 4 obese subjects. Incubation of adipocytes from mechanism for alterations in the rate of glucose transport either group with 7 nM insulin did not recruit GLUT5 to (6-9). Alternatively, other glucose transporter isoforms the plasma membrane, in spite of a 54% may be present in these tissues and may contribute insulin-stimulated increase in GLUT4 in nonobese significantly to basal and/or insulin-stimulated glucose subjects. Thus, GLUT5 appears to be a constitutive transport. GLUT2 is not present in human muscle or sugar transporter that is expressed in many tissues. adipose tissue (10), and a previous study demonstrates Further studies are needed to define its overall that GLUT3 is not present in human adipose tissue (11). contribution to fructose and glucose transport in The tissue distribution of GLUT5 protein has not been insulin-responsive tissues and brain. Diabetes investigated previously. Although northern blotting stud- 41:1360-65,1992 ies indicate that GLUT5 mRNA is abundant in small intestine and kidney, it is present at much lower levels in human muscle and adipose tissue (1,12). Thus, this From the Charles A. Dana Research Institute and Harvard-Thorndike Labo- study was designed to determine the tissue distribution ratory of Beth Israel Hospital, Department of Medicine, Beth Israel Hospital of the GLUT5 protein to investigate the possibility that and Harvard Medical School, Boston, Massachusetts; the Department of Biochemistry, University of Glasgow, Glasgow, Scotland; the Department of GLUT5 could contribute to sugar transport in highly Medicine, University of Goteborg, Goteborg, Sweden; and the Pfizer Central insulin-responsive tissues. During the preparation of this Research, Groton, Connecticut. manuscript, we became aware that GLUT5 is a high- Address correspondence and reprint requests to Barbara B. Kahn, MD, Diabetes Unit/Beth Israel Hospital, 330 Brookline Ave., Boston, MA 02215. affinity fructose transporter with much less capacity to Received for publication 8 June 1992 and accepted in revised form 16 July transport glucose than reported previously (1,12,13). 1992. Thus, our demonstration in this study that GLUT5 is BMI, body mass index; SDS, sodium dodecyl sulfate; SDS-PAGE, sodium dodecylsulfate-polyacrylamide gel electrophoresis; TBS, Tris- buffered saline; present in human muscle, adipose cell plasma mem- PBS, phosphate-buffered saline; type II diabetes, non-insulin-dependent branes, and brain should now stimulate further investiga- diabetes mellitus; ANOVA, analysis of variance. 1360 DIABETES, VOL. 41, OCTOBER 1992 PR. SHEPHERD AND ASSOCIATES tion of both the potential contribution of GLUT5 to was synthesized by Biomac (Dept of Biochemistry, Uni- glucose transport and the importance of fructose as a versity of Glasgow, Scotland), and the GLUT1 peptide metabolic substrate for these tissues. (NH2-KTEPEELFHPLGADSQV-COOH) was synthesized by the Peptide Synthesis Laboratory (Pfizer Central Re- RESEARCH DESIGN AND METHODS search, Groton, CT). a-G5 antibodies were affinity-puri- Preparation of membranes from tissues. Normal hu- fied over a peptide column as described (18) and a-G1 man duodenum and ileum were obtained from the clearly antibodies were affinity-purified using human erythro- demarcated disease-free margin of small intestine re- cyte membranes depleted of peripheral proteins (19). moved from two patients with inflammatory bowel dis- RaGLUT5, an affinity-purified antibody to the 12 COOH- ease. Rat small intestine was obtained from 200-g male terminal amino acids of human GLUT5 was purchased Sprague-Dawley rats (Charles River Breeding Laborato- from East Acres Biologicals (Southborough, MA). An ries, Wilmington, MA). Epithelial cells were isolated and a antiserum to the COOH-terminus of rat GLUT4 (from Dr. postnuclear membrane fraction was prepared (10). Post- Mike Mueckler) was purified using a protein-A sepharose nuclear membranes were prepared using the method of column (Pierce, Rockford, III). Thorens et al. (10) from placenta obtained from a full-term Western blotting. Samples were solubilized in Laemelli uncomplicated pregnancy and kidney and brain ob- loading buffer containing 2% SDS, and proteins were tained at autopsy. Muscle postnuclear membranes were separated by SDS-PAGE (10% acrylamide). Mr was prepared as described previously (14) from human vas- assessed with prestained (Biorad, Richmond, CA) and tus lateralis muscle obtained from lean, nondiabetic unstained (Sigma, St. Louis, MO) Mr markers. Proteins volunteers by percutaneous needle biopsy; rectus ab- were transferred to nitrocellulose at 250 mAmps for 16 h dominus and soleus obtained during elective surgery; in tris-glycine buffer containing 20% methanol and 0.1% and heart and psoas major obtained at autopsy. SDS. Uniformity of gel loading and protein transfer to the Human subcutaneous adipose tissue was obtained filters was assessed by ponceau-S staining. from the upper abdomen of 9 subjects undergoing Dot blots. Before dot blotting, western blots were per- elective surgery. The age range was 29-64 yr. Five formed to confirm that only one immunoreactive band 2 subjects were nonobese (BMI 25-27.7 kg/m ), three was present in human fat cell membrane samples. Trip- 2 were obese (BMI 28-34 kg/m ), and one was massively licate samples (2 |xg) of human adipocyte plasma mem- 2 obese (BMI 53.1 kg/m ) with impaired glucose tolerance. branes were applied to nitrocellulose (BA85, Schleicher One moderately obese subject had type II diabetes and and Schuell, Keene, NH), using a dot blot apparatus was being treated with glibenclamide. Samples from two (BRL, Gaithersberg, MD), and allowed to air dry. A of the subjects were pooled because of low yields. standard curve of small intestine membranes was run on Biopsies were taken immediately after the induction of the same filter to assess the linear range for quantitation. anesthesia and placed in medium 199 containing 4% Immunoblotting. Filters were blocked for 60 min at 20°C bovine serum albumin and 5.5 mM glucose at pH 7.4 and in PBS (pH 7.4)2, 5% nonfat dry milk/0.1% Triton-X-100 37°C. Adipocytes were prepared by collagenase diges- and then incubated for 2 h at room temperature with 10 tion (15). Isolated adipocytes were incubated for 20 min ng/ml of GLUT1, GLUT4, or GLUT5 antibody in PBS/1% in the presence of 1 U/ml adenosine deaminase and 0 or nonfat dry milk/0.1% Triton-X-100. Filters then were 7.0 nM insulin. Plasma membranes then were prepared washed at 20°C, once with PBS/5% nonfat dry milk/ by differential centrifugation (16). 0.1% Triton-X-100 and twice in PBS/0.1% Triton-X-100. Postnuclear membrane fractions also were prepared Immunoreactive bands shown in Figures 1 and 4 were 125 from frozen human brain frontal lobe, fresh whole mouse visualized with either l-goat anti-rabbit IgG or 125 brain, and the following frozen monkey tissues: brain, I-protein-A (both Du Pont-NEN, Boston, MA) using soleus muscle, liver, pancreas, omental fat, and omental autoradiography with XAR-5 film (Eastman-Kodak, Roch- adipocytes (prepared as described above for human ester, NY) with an intensifying screen at -70°C or phos- adipocytes) by homogenizing the samples with a Brink- phoimaging on a Phosphorlmager (Molecular Dynamics, man polytron at setting of 4 (human and mouse tissues) Sunnyvale, CA). Immunoreactive bands shown in Figure or 7 (monkey tissues) for 30 s (or 60 s for monkey soleus) 2 were visualized using enhanced chemiluminescence in a volume of 10 ml of HEPES/sucrose/EDTA buffer (pH (ECL, Amersham, Arlington Heights, IL) according to 7.4, 0.25 M sucrose, 10 mM HEPES, 5 mM EDTA with 2.5 manufacturer's instructions. Autoradiographic bands ixg/ml leupeptin, pepstatin, and aprotinin) per gram of were scanned with a Hewlett-Packard ScanJet. Bands tissue. Nuclei, mitochondria, and connective tissue were were quantitated with Imagequant software (Molecular removed by centrifugation at 2000 g for 10 min, and then Dynamics). postnuclear membranes were obtained by centrifugation Cytochalasin-B photolabelling of GLUT5. Duodenal at 300,000 g for 1 h.