Orientation of the glucose transporter in the human erythrocyte membrane. Investigation by in situ proteolytic dissection

Glucose transporter proteins (zone 4.5) which had been photoaffinity labeled with [3H]cytochalasin B in human erythrocyte ghosts were subjected to enzymatic dissection in order to study the transmembrane disposition of the protein in situ. Proteolytic enzymes as well as glycosidases were used to treat unsealed and resealed ghosts in order to explore the various membrane domains of the transporter in a topographically defined manner. Limited digestion of sealed ghosts with trypsin had no effect on the apparent Mr of the transporter (55,000). Similar treatment in unsealed ghosts, however, resulted in the generation of a major fragment of 21.5 kDa, along with several minor fragments. Thermolysin also had no effect on sealed ghosts but caused a complete loss of radiolabel from the zone 4.5 region with no lower-molecular-weight fragments being retained on the gel. Chymotrypsin treatment resulted in the generation of a single peak, Mr = 18,400, in both sealed and unsealed ghosts indicating its action occurs at the outer surface. Digestion with carboxypeptidase and aminopeptidase indicate the C-terminal end of the transporter is located exterior to the membrane with the N terminus located at the cytoplasmic surface. Treatment with endoglycosidase resulted in a shift of mobility of the transporter to a lower Mr of 49,000. The results obtained indicate that the carbohydrate is located near the C-terminal end and that the cytochalasin B-binding site is located near the cytoplasmic N-terminal end.     

Proteolytic and chemical dissection of the human erythrocyte glucose transporter.

Treatment of the purified, reconstituted, human erythrocyte glucose transporter with trypsin lowered its affinity for cytochalasin B more than 2-fold, and produced two large, membrane-bound fragments. The smaller fragment (apparent Mr 18000) ran as a sharp band on sodium dodecyl sulphate (SDS)/polyacrylamide-gel electrophoresis. When the transporter was photoaffinity labelled with [4-3H]cytochalasin B before tryptic digestion, this fragment became radiolabelled and so probably comprises a part of the cytochalasin B binding site, which is known to lie on the cytoplasmic face of the erythrocyte membrane. In contrast, the larger fragment was not radiolabelled, and ran as a diffuse band on electrophoresis (apparent Mr 23000-42000). It could be converted to a sharper band (apparent Mr 23000) by treatment with endo-beta-galactosidase from Bacteroides fragilis and so probably contains one or more sites at which an oligosaccharide of the poly(N-acetyl-lactosamine) type is attached. Since the transporter bears oligosaccharides only on its extracellular domain, whereas trypsin is known to cleave the protein only at the cytoplasmic surface, this fragment must span the membrane. Cleavage of the intact, endo-beta-galactosidase-treated, photoaffinity-labelled protein at its cysteine residues with 2-nitro-5-thiocyanobenzoic acid yielded a prominent, unlabelled fragment of apparent Mr 38000 and several smaller fragments which stained less intensely on SDS/polyacrylamide gels. Radioactivity was found predominantly in a fragment of apparent Mr 15500. Therefore it appears that the site(s) labelled by [4-3H]cytochalasin B lies within the N-terminal or C-terminal third of the intact polypeptide chain.     

Structural basis of human erythrocyte glucose transporter function in reconstituted vesicles

The transmembrane orientation of the human erythrocyte glucose transporter was assessed based on polarized Fourier transform infrared and ultraviolet circular dichroism spectroscopic data obtained from oriented multilamellar films of the reconstituted transporter vesicles. Infrared spectra revealed that there are distinct vibrations for alpha-helical structure while the vibrational frequencies specific to beta-structure are characteristically absent. Analysis of linear dichroism of the infrared spectra further indicated that these alpha-helices in the transporter are preferentially oriented perpendicular to the lipid bilayer plane forming an effective tilt of less than 38 degrees from the membrane normal. Such a preferential orientation was further supported by ultraviolet circular dichroism spectra which reveal that the 208 nm Moffit band found in the detergent-solubilized preparation is absent in the film preparation. Linear dichroism data further indicated that D-glucose, a typical substrate, further reduces this effective tilt angle slightly.     

Facilitated glucose transporter of human erythrocyte: proteolytic mapping of the [3H]cytochalasin B photoaffinity-labeled transporter polypeptide

The human erythrocyte D-glucose transporter is an integral membrane glycoprotein with an heterogeneous molecular mass spanning a range 45-70 kDa. The protein structure of the transporter was investigated by photoaffinity labeling with [3H]cytochalasin B and fractionating the labeled transporter according to molecular mass by preparative SDS-polyacrylamide gel electrophoresis. Each fraction was digested with either papain or S. aureus V8 proteinase, and the labeled proteolytically derived peptide fragments were compared by SDS polyacrylamide gel electrophoresis. Papain digestion yielded two major peptide fragments, of approx. molecular mass 39 +/- 2 and 22 +/- 2 kDa; treatment with V8 proteinase resulted in two fragments, with mass of 24 +/- 2 and 15 +/- 2. Proteolysis of each transporter fraction produced the same pattern of labeled peptide fragments, irrespective of the molecular mass of the original fractions. The binding characteristics of [3H]cytochalasin-B-labeled transporter to Ricinis communis agglutinin lectin was examined for each transporter molecular mass fraction. It was found that higher-molecular-mass fractions of intact transporter had a 2-fold greater affinity for the lectin than lower-molecular-mass fractions (i.e., 67 kDa greater than 45 kDa fraction). However, proteolytically derived labeled peptide fragments from each fraction had minimal affinity for the lectin. These results suggest that the labeled peptide fragments have been separated from the glycosylated regions of the parent transporter protein. The present findings indicate that, although transporter proteins have an apparently heterogeneous molecular mass, some regions of the protein share a common peptide. Furthermore, the glycosylated regions appear to be located some distance from the [3H]cytochalasin-B-labeled site(s).     

Site-specific antibodies as probes of the topology and function of the human erythrocyte glucose transporter.

Antibodies were raised against synthetic peptides corresponding to most of the regions of the human erythrocyte glucose transporter predicted to be extramembranous in the model of Mueckler, Caruso, Baldwin, Panico, Blench, Morris, Lienhard, Allard & Lodish [(1985) Science 229, 941-945]. Most of the antibodies (17 out of a total of 19) recognized the intact denatured protein on Western blots. However, only seven of the antibodies recognized the native membrane-bound protein, even after its deglycosylation. These antibodies, against peptides encompassing residues 217-272 and 450-492 in the hydrophilic central and C-terminal regions of the transporter, bound to the cytoplasmic surface of the erythrocyte membrane. This finding is in agreement with the prediction of the model that these regions of the sequence are cytoplasmic. Antibodies against peptides from the central cytoplasmic loop of the transporter were found to inhibit the binding of cytochalasin B to the membrane-bound protein, whereas antibodies against the C-terminal region had no effect. The anti-peptide antibodies were then used to map the sequence locations of fragments of the transporter arising from tryptic digestion of the membrane-bound protein. This in turn enabled the epitopes for a number of anti-transporter monoclonal antibodies to be located within either the central cytoplasmic loop or the C-terminal region of the protein. Of those monoclonal antibodies which inhibited cytochalasin B binding to the protein, all but one were found to have epitopes within the central region of the sequence. In conjunction with the results of the anti-peptide antibody studies, these findings indicate the importance of this part of the protein for transporter function.     

Glycosylation of the human erythrocyte glucose transporter: a minimum structure is required for glucose transport activity

The involvement of the carbohydrate moiety of the human erythrocyte glucose transporter in glucose transport activity was previously demonstrated (Feugeas et al. (1990) Biochim. Biophys. Acta 1030, 60-64): N-glycanase treatment of the transport glycoprotein reconstituted in proteoliposomes resulted in a dramatic decrease of the Vmax. In this study, kinetic measurements of glucose equilibrium influx confirm our previous results. In order to investigate that a minimum glycosidic structure is required to maintain glucose transport activity, proteoliposomes were respectively treated with either sialidase, or sialidase and endo-beta-galactosidase, or a pool of exo-glycosidases which allows the release of all the sugar residues, except the proximal N-acetylglucosamine. Kinetic measurements of zero-trans influx made on sialidase- and (sialidase + endo-beta-galactosidase)-treated proteoliposomes did not reveal any significant changes in the glucose transport activity. On the contrary, treatment of the same proteoliposomes by a pool of exoglycosidases led to a complete abolition of activity, suggesting that a minimum glycosidic structure is required for glucose transport activity.     

Transport function and subcellular distribution of purified human erythrocyte glucose transporter reconstituted into rat adipocytes

In order to delineate the insulin-independent (constitutive) and inssulin-dependent regulations of the plasma membrane glucose transporter concentrations in rat adipocytes, we introduced purified human erythrocyte GLUT-1 (HEGT) into rat adipocytes by poly(ethylene glycol)-induced vesicle-cell fusion and its transport function and subcellular distribution in the host cell were measured. HEGT in adipocytes catalysed 3-O-methylglucose equilibrium exchange with a turnover number that is indistinguishable from that of the basal adipocyte transporters. However, insulin did not stimulate significantly the HEGT function in adipocytes where it stimulated the native transporter function by 7-8-fold. The steady state distribution and the transmembrane orientation assays revealed that more than 85% of the HEGT that were inserted in the physiological, cytoplasmic side-in orientation at the adipocytes plasma membrane were moved into low-density microsomes (LDM), while 90% of the HEGT that were inserted in the wrong, cytoplasmic side-out orientation were retained in the plasma membrane. Furthermore, more than 70% of the LDM-associated HEGT were found in a small subset of LDM that also contained 80% of the LDM-associated GLUT-4, the insulin-regulatable, native adipocyte glucose transporter. However, insulin did not cause redistribution of HEGT from LDM to the plasma membrane under the condition where it recruited GLUT-4 from LDM to increase the plasma membrane GLUT-4 content 4–5-fold. These results demonstrate that the erythrocyte GLUT-1 introduced in adipocytes transports glucose with an intrinsic activity similar to that of the adipocyte GLUT-1 and/or GLUT-4, and enters the constitutive GLUT-4 translocation pathway of the host cell provided it is in physiological transmembrane orientation, but fails to enter the insulin-dependent GLUT-4 recruitment pathway. We suggested that the adipocyte plasma membrane glucose transporter concentration is constitutively kept low by a mechanism where a cell-specific constitutent interacts with a cytoplasmic domain common to GLUT-1 and GLUT-4, while the insulin-dependent recruitment requires a cytoplasmic domain specific to GLUT-4.     

Fourier transform infrared spectroscopic study of the structure and conformational changes of the human erythrocyte glucose transporter

Fourier transform infrared spectroscopy has been used to study the secondary structure of the human erythrocyte glucose transporter after purification and reconstitution in erythrocyte lipids. The spectra indicate that the glucose transporter contains, in addition to the predominant alpha-helical structure, an appreciable amount of beta-structure and random coil conformation. A study of the time dependency of H-2H exchange revealed that more than 80% of the polypeptide backbone is readily accessible to the solvent. This result indicates that a portion of the intramembrane-spanning region of the membrane protein is exposed to the solvent, suggesting the existence of an intraprotein aqueous channel. The residual (10-20%) portion of the protein which exchanges slowly includes some alpha-helical structure, probably situated in a hydrophobic environment inside the membrane. The infrared spectra of transporter preparations were also examined after incubation with substrate and substrate analogues. Compared with the spectra recorded under conditions in which the "inward-facing" form predominates, a small but reproducible shift in the bands assigned to alpha-helical and beta-strand structures is observed after incubation with 4,6-O-ethylidene-D-glucose, which largely fixes the transporter in the "outward-facing" conformation. An increase of temperature, which is known to increase the proportion of transporter in the outward-facing conformation, results in a similar shift in this alpha-helical absorption band.     

Structural basis of human erythrocyte glucose transporter function in reconstituted system. Hydrogen exchange

Hydrogen exchange kinetic behavior of human erythrocyte glucose transporter protein in vesicles was studied in the absence and in the presence of D-glucose or a well known inhibitor, cytochalasin B. This is to detect a proposed channel of water penetrating into the protein through which the sugar molecule passes and to monitor any conformational changes induced by the substrate or inhibitor. Analyses of the kinetic data revealed several classes of hydrogens which exchange with readily distinguishable rates. Of 660 hydrogens detected per transporter, approximately 30% exchanged with rates generally characterized as those of free amide hydrogens indicating they are interfaced to solvent water. Since the transporter is known to be embedded deep in the hydrophobic area of the membrane with minimum exposure to the outside of the membrane lipid bilayer, a significant portion of these free amide hydrogens must be at the purported channel rather than outside of the membrane. D-Glucose and cytochalasin B affected the exchange kinetics of these presumably channel-associated free amide hydrogens rather differently. D-Glucose reduced the apparent rate constants, but not the total number. Cytochalasin B on the other hand reduced the total number to one-half without significantly changing the apparent rate constants. The remaining 70% of the labeled hydrogens exchanged with much slower rates which vary 10-10,000-fold, indicating that they are internally structured peptide amide and side chain hydrogens. Both D-glucose and cytochalasin B further reduced the rates of these hydrogens, indicating a global stabilization of the protein structure.     

Hydrophobic photolabeling in membranes: the human erythrocyte glucose transporter

Human erythrocyte membranes were labeled with a hydrophobic photoactivable reagent, 2-[3H]Diazofluorene. Electrophoretic analysis of the protein fraction showed that several membrane spanning proteins like Band 3 (the anion transporter), Band 4.5 (the glucose transporter), and the sialoglycoproteins PAS 1, 2, and 3 have been labeled. To isolate the diazofluorene-labeled glucose transporter, the membrane preparation was solubilized with Triton X-100 and passed through a DEAE-cellulose column. The flow-through fraction was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Radioactive analysis of the gel indicated that besides the Band 4.5, two more proteins corresponding to the Band 3 and Band 6 regions also coelute with the glucose transporter in the flow-through fraction. On the other hand, use of n-octyl glucoside gave a relatively better preparation. The 2-[3H]DAF-labeled glucose transporter isolated by the latter method on tryptic digestion indicated that the Mr 18,000 fragment corresponding to the C-terminal transmembrane fragment is labeled.     

Microheterogeneity of the carbohydrate moiety of the human erythrocyte glucose transporter

The carbohydrate moiety of the human erythrocyte glucose transporter was isolated using two independent methods: hydrazinolysis andN-glycanase treatment. The major structure observed was constituted of complex-type carbohydrate chains carrying repetitive units ofN-acetyllactosamine. This structure exhibited microheterogeneity: a broad variability in the number of repetitive units, presence of branched structures and substitution by fucosyl residues. Moreover, significant amounts of bi-antennary and hybrid structures were present.     

Glycosylation of the human erythrocyte glucose transporter is essential for glucose transport activity

The human erythrocyte glucose transporter is a fully integrated membrane glycoprotein having only one N-linked carbohydrate chain on the extracellular part of the molecule. Several authors have suggested the involvement of the carbohydrate moiety in glucose transport, but not definitive results have been published to date. Using transport glycoproteins reconstituted in proteoliposomes, kinetic studies of zero-trans influx were performed before and after N-glycanase treatment of the proteoliposomes: this enzymatic treatment results in a 50% decrease of the Vmax. The orientation of transport glycoproteins in the lipid bilayer of liposomes was investigated and it appears that about half of the reconstituted transporter molecules are oriented properly. Finally, it could be concluded that the release of the carbohydrate moiety from the transport glycoproteins leads to the loss of their transport activity.     

Structural basis of human erythrocyte glucose transporter function: pH effects on intrinsic fluorescence.

The effects of pH on the intrinsic fluorescence of purified human erythrocyte glucose transporter (HEGT) were studied to deduce the structure and the ligand-induced dynamics of this protein. D-Glucose increases tryptophan fluorescence of HEGT at a 320-nm peak with a concomitant reduction in a 350-nm peak, suggesting that glucose shifts a tryptophan residue from a polar to a nonpolar environment. Cytochalasin B or forskolin, on the other hand, only produces a reduction at the 350-nm peak. The pH titration of the intrinsic fluorescence of HEGT revealed that at least two tryptophan residues are quenched, one with a pKa of 5.5, the other with a pKa of 8.2, indicating involvement of histidine and cysteine protonation, respectively. D-Glucose abolishes both of these quenchings. Cytochalasin B or forskolin, on the other hand, abolishes the histidine quenching but not the cysteine quenching and induces a new pH quenching with a pKa of about 4, implicating involvement of a carboxyl group. These results, together with the known primary structure and the transmembrane disposition of this protein, predict the dynamic interactions between Trp388 and His337, Trp412 and Cys347, and Trp412 and Glu380, depending on liganded state of HEGT, and suggest the importance of the transmembrane helices 9, 10, and 11 in transport function.     

Transport function and subcellular distribution of purified human erythrocyte glucose transporter reconstituted into rat adipocytes.

In order to delineate the insulin-independent (constitutive) and insulin-dependent regulations of the plasma membrane glucose transporter concentrations in rat adipocytes, we introduced purified human erythrocyte GLUT-1 (HEGT) into rat adipocytes by poly(ethylene glycol)-induced vesicle-cell fusion and its transport function and subcellular distribution in the host cell were measured. HEGT in adipocytes catalysed 3-O-methylglucose equilibrium exchange with a turnover number that is indistinguishable from that of the basal adipocyte transporters. However, insulin did not stimulate significantly the HEGT function in adipocytes where it stimulated the native transporter function by 7-8-fold. The steady state distribution and the transmembrane orientation assays revealed that more than 85% of the HEGT that were inserted in the physiological, cytoplasmic side-in orientation at the adipocytes plasma membrane were moved into low-density microsomes (LDM), while 90% of the HEGT that were inserted in the wrong, cytoplasmic side-out orientation were retained in the plasma membrane. Furthermore, more than 70% of the LDM-associated HEGT were found in a small subset of LDM that also contained 80% of the LDM-associated GLUT-4, the insulin-regulatable, native adipocyte glucose transporter. However, insulin did not cause redistribution of HEGT from LDM to the plasma membrane under the condition where it recruited GLUT-4 from LDM to increase the plasma membrane GLUT-4 content 4-5-fold. These results demonstrate that the erythrocyte GLUT-1 introduced in adipocytes transports glucose with an intrinsic activity similar to that of the adipocyte GLUT-1 and/or GLUT-4, and enters the constitutive GLUT-4 translocation pathway of the host cell provided it is in physiological transmembrane orientation, but fails to enter the insulin-dependent GLUT-4 recruitment pathway. We suggested that the adipocyte plasma membrane glucose transporter concentration is constitutively kept low by a mechanism where a cell-specific constituent interacts with a cytoplasmic domain common to GLUT-1 and GLUT-4, while the insulin-dependent recruitment requires a cytoplasmic domain specific to GLUT-4.     

Further characterization and chemical purity assessment of the human erythrocyte glucose transporter preparation

Chemical and functional purity of the human erythrocyte glucose transporter preparation obtained by DEAE column chromatography after octyl glucoside solubilization was assessed. The cytochalasin B binding capacity of the preparation indicates that the preparation is 60-85% functional glucose transporter. Gel filtration chromatography on TSK 250 column separates this preparation into at least three major peptide fractions, namely, P0, P1 and P2, with apparent Mr of approx. 80 000, 43 000 and 17 000, respectively. When the preparation is photolabelled with [3H]cytochalasin B prior to the separation only P0 and P1 are labelled. Exposure of the preparation to octyl glucoside or to ultraviolet light irradiation results in an increase in P0 in a time-dependent manner with a concomitant and proportional reduction in P1, without affecting P2 appreciably. For individual preparations, relative abundance of P0 and P1 vary widely in a reciprocal fashion, while that of P2 is practically fixed at approx. 10% of the total protein. The specific activity of cytochalasin B binding of each preparation correlates linearly with the relative abundance of P1 of the preparation, which gives a calculated specific binding activity of 22 nmol/mg protein for this fraction. These results indicate that P1 and P0 are native and denatured transporter, respectively, while P2 is contaminating protein impurities. These results demonstrate that the glucose transporter preparation contains approx. 10% of nontransporter protein impurities, with a varying amount (up to 30%) of denatured transporter, and that the transporter free of the chemical impurities and the denatured transporter can be obtained by a gel filtration chromatography of this preparation.     

Endoglycosidase f cleaves the oligosaccharides from the glucose transporter of the human erythrocyte

The glucose transporter from human erythrocytes is a heterogeneously glycosylated protein that runs as a very broad band of average apparent Mr 55 000 upon sodium dodecyl sulfate polyacrylamide gel electrophoresis. When the purified preparation of transporter, solubilized in Triton X-100, was treated with endoglycosidase F, much of it ran as a sharp band of Mr 46 000 upon electrophoresis. Moreover, endoglycosidase F released 80% of the radioactivity in a preparation of the transporter labeled in its oligosaccharides with galactose oxidase and tritiated borohydride, and almost none of the remaining radioactivity was located in the Mr 46 000 band. These results suggest that endoglycosidase F can release virtually all of the carbohydrate linked to the transporter polypeptide. A quantitative analysis of the gels was complicated by partial aggregation of polypeptides that occurs due to prolonged incubation in Triton X-100, but at least 65% of the protein in the preparation of purified transporter is the 46 kDa polypeptide. The extracellular domain of the transporter is very resistant to proteolysis; no cleavage occurred upon treatment of intact erythrocytes with seven different proteases at high concentration.