Insulin Action on Activity and Cell Surface Disposition of Human HepG2 Glucose Transporters Expressed in Chinese Hamster Ovary Cells*

DNA encoding a facilitative glucose transporter isolated from line (HepG2) subcloned metal-inducible mammalian expression human II promoter

2-Deoxy-D-glucose uptake was increased 2-7-fold in transfected cell lines. Polyclonal antisera directed against purified red blood cell glucose transporter were raised in several rabbits. Antiserum from one rabbit, 6, was found to bind to the surface of intact red cells but not to inside-out red cell ghosts. Using this a-antiserum in intact cell-binding assays, 1.6-g-fold increases in cell surface expression of the human glucose transporter were measured in CHO-Kl cell lines transfected with the transporter expression vector. Measurements of total cellular glucose transporter immunoreactive protein using anti-HepG2 transporter C-terminal peptide serum, cell surface glucose transporter protein using &antiserum and 2-deoxyglucose uptake revealed proportional relationships among these parameters in transfected cell lines expressing different levels of transporter protein. Insulin increased 2-deoxyglucose uptake 40% in control CHO-Kl cells and in CHO-Kl cells expressing modest levels of the human glucose transporter protein. However, stimulation of sugar uptake by insulin was only 10% in cells overexpressing human glucose transporter protein g-fold, and no effect of insulin on sugar uptake was detected in several cell lines expressing very high levels (12-17- from a HepG2l cell line several years ago. Since that time, other glucose transporter cDNAs have been isolated from rat and rabbit brain (Birnbaum et al., 1986;, human and rat liver (Fukomoto et al., 1988;Thorens et al., 1988), human fetal skeletal muscle (Kayano et al., 1988), rat adipocyte and muscle (James et al., 1989;Birnbaum, 1989;Charron et at., 1989), human muscle (Fukumoto et al., 1989), and mouse adipocyte (Kaestner et al., 1989). Amino acid sequences deduced from these cDNA sequences have revealed four distinct isotypes of glucose transporter. 1) A glucose transporter denoted as GLUTl' is prevalent in HepG2 cells, erythrocytes, and brain. The GLUT1 protein has been identified in human, rat, and mouse (Sogin and Hinkle, 1980;Oka et al., 1988;Joost et al., 1988;Blok et al., 1988;Calderhead and Lienhard, 1988;James et al., 1989;Zorzano et al., 1989). 2) A glucose transporter denoted as GLUT2 is prevalent in liver, kidney, and intestine. The GLUT2 protein has been identified in rat (Thorens et al., 1988). 3) A glucose transporter denoted as GLUT3 prevalent in human fetal skeletal muscle. 4) A glucose transporter denoted as GLUT4 is prevalent in adipose and muscle tissues. The GLUT4 protein has been identified in rat, human, and mouse (James et ai., 1988(James et ai., , 1989Zorzano et al., 1989;Fukumoto et al., 1989). The sequence identity between members of one isotype in different tissues and species ranges from approximately 90 to 97%, whereas the identity between glucose transporter proteins from different isotypes, even within a given species, ranges from 50 to 65%. These preliminary classifications are not meant to imply that the identical  (1989).

5794
Insulin Regulation of Glucose Transporters proteins will be the only ones expressed in those tissues. It seems quite probable that additional glucose transporter isoforms will be identified in the near future.
Several important questions about mammalian glucose transporters relate to the mechanisms that regulate the functions of these proteins. Do hormones and other agents that modulate cellular sugar uptake regulate all of the different isotypes of glucose transporters present in a single cell? Are signaling pathways that regulate different transporter isotypes divergent? One of the major regulators of hexose uptake is insulin, which stimulates facilitative glucose transporter activity in skeletal muscle and in adipose tissues (Simpson and Cushman, 1986). These effects contribute to the lowering of blood glucose in intact animals by this hormone. The signal transduction pathway or mechanism of activation of glucose transport by insulin has not yet been elucidated at the molecular level. Cushman and Wardzala (1980) and Suzuki and Kono (1980) originally proposed the hypothesis that insulin regulates glucose transport by stimulating the translocation of glucose transporter protein from an intracellular membrane pool to the plasma membrane in responsive adipocytes. This hypothesis was based on measurements of cytochalasin B binding to glucose transporters and by reconstitution of transport activity from the different membrane fractions. Additional evidence for insulin-stimulated translocation was obtained by labeling glucose transporters covalently in intact cells (Oka and Czech, 1984;Holman et al., 1988;Calderhead and Lienhard, 1988) and by immunocytochemical analysis using electron microscopy (Blok et al., 1988).
It is not yet understood how insulin regulates the membrane distribution of glucose transporters, nor is it clear whether the intrinsic activity of the translocated glucose transporter proteins is also increased in response to insulin. Analysis of protein-immunoblotting data obtained with antibodies directed against specific transporter isotypes suggests that insulin regulates both GLUT1 and GLUT4 in rat (Joost et al., 1988;James et al., 1988James et al., , 1989Birnbaum, 1989;Zorzano et al., 1989) and mouse (Blok et al., 1988;James et al., 1989) adipocytes. However, insulin regulation of GLUT1 has not been observed in a number of other cells, including brain, human erythrocytes, and cultured HepG2 cells. Recent work by Oka and co-workers indicates that rabbit GLUT1 expressed in CHO fibroblasts is stimulated by insulin (Asano et al., 1989).
Similarly, insulin causes translocation of human GLUT1 expressed in differentiated 3T3-Ll cells (Gould et al., 1989), but the regulatory mechanisms underlying hormonal sensitivity are unknown.
The present work describes our recent progress toward the development of new CHO-Kl fibroblast phenotypes and novel antibody reagents that allow us to address some of these problems. Complementary DNA encoding a facilitative glucose transporter was cloned from an insulin-unresponsive HepG2 cell cDNA library and then subcloned into a mammalian expression vector, pLEN. CHO-Kl fibroblasts transfected with the HepGP transporter expression vector pLENGT exhibited high levels of expression of HepG2 glucose transporter protein and activity. A novel anti-glucose transporter antiserum that recognizes GLUT1 exposed on the exofacial surface of intact human erythrocytes and CHO fibroblasts was developed and used to assess cell surface expression of human GLUTl.
We report here that insulin stimulates human GLUT1 expressed at low but not at high levels in hormonally responsive CHO fibroblasts and that this stimulation occurs without a concomitant increase in cell surface immunoreactive GLUT1 transporters.
These nrobe seauences were derived from the published HepGP glucose &ansport& cDNA sequence of Muekler et al. (1985). Approximately 100 putative positive plaques were identified from 1.5 X lo6 plaques screened by hybridization to a mixture of the three "P-labeled orobes. Individual nhaee colonies were cloned from these plaques by limiting dilution, and tie DNA was then digested with the restriction enzyme EcoRI. The cDNA fragments were then resolved by electrophoresis on 1% agarose gels, transferred to nitrocellulose, and hybridized to the individual 5' and 3' probes. None of the isolated clones contained full-length coding cDNAs for the HepGP glucose transporter. The majority of our clones hybridized to the 3' but not the 5' probes.
A few clones did hybridize to the 5' but not the 3' probes, and the sizes of these 5' and 3' cDNA fragments suggested to us that the internal EcoRI cleavage site in the HepG2 glucose transporter mRNA (Muekler et al., 1985)   . a-Serum and &serum were diluted 5,000. and 4,000-fold, respectively, in 500 ~1 of phosphate buffer  and incubated for 30 min at 20 "C with varying amounts of purified glucose transport protein (O.l-lo4 ng). The glucose carrier was then pelleted by centrifugation at 45,000 X g for 40 min, and 200 ~1 of the clear supernatant was applied to wells in duplicate.
The plate was then processed (Carruthers and  using peroxidase-conjugated goat anti-rabbit IgG and peroxidase substrates as the reporter system. The reaction was arrested using 1% oxalic acid and the plate read using a Dynatech MR700 plate reader. A parallel experiment was performed in which 104-10"' human or rat erythrocytes were substituted for purified carrier in the preincubation step. Bindingofdnti-glucose Transporter Antibodies to Intact Red Cells-Protein A-Sepharose (CL-4B, Sigma; 25 pg of protein A/10 pl of Sepharose beads) was preincubated in 1 ml of saline containing 10 ~1 of a-serum or 8 ~1 of &serum for 30 min at 37 "C. The beads were then washed three times in 10 volumes of ice-cold saline. One aliquot of beads that had been treated with &serum was then incubated with 40 pg of purified glucose transporter in 1 ml of saline (0.73 pM carrier) for 30 min at 37 "C and washed as above. The Sepharose beads were finally incubated with human red cells (1 X lo7 cells in saline) in 1 ml of saline (final red cell glucose transporter concentration = 2.5 nM) for 30 min at 37 "C. Samples were placed on microscope slides, examined by phase-contrast microscopy at a magnification of x 400, and photographed.
Rabbit N-and &Antisera Binding to Intracellular and Extracellular Domains of the Glucose Transporter Protein in Human Erythrocyte Membranes-Intact human red cells, sealed red cell ghosts, leaky red cell ghosts, and sealed inside-out red cell membrane vesicles were prepared as described by Carruthers and Helgerson (1989). These membrane preparations (in 1 ml of saline) were then incubated in the presence of rabbit a-or &serum for 1 h (10 ~l/lO ~1 of packed red cells (1 X 10' cells) or 10 ~1/60 pg of membrane protein) at 20 "C. These conditions were determined in parallel experiments to allow both equilibrium and saturated IgG binding. The cells and membranes were washed five times in 10 ml of ice-cold saline and then incubated with "'I- with surface areas equal to those of the excised bands were cut from the same filters immediately above and below each glucose transporter protein band. These control blanks were analyzed in the y-counter, the cpm were divided by 2, and the specific antibody binding was determined by subtracting these nonspecific control values.
Results from the assays were normalized to cpm/mg of total membrane protein for each plate of cells.

Development of a Novel Antitransporter Antiserum That Recognizes Exofacial
Epitopes-Erythrocyte glucose transporter protein was purified from human red cells and injected into rabbits as described under "Experimental Procedures." Preimmune and immune sera harvested from these animals were analyzed for their utility in competition ELISA and protein immunoblots and for their ability to bind to intact human red blood cells. Two of the rabbit antisera, a-and 6-, were found to be useful in these immunoassays of the erythrocyte glucose transporter GLUTl.
Human erythrocyte proteins were solubilized in SDS sample buffer, resolved on 10% acrylamide gels, transferred to nitrocellulose filters, and immunoblotted with either a-or &antiserum.
As can be seen in Fig. 1, both of these polyclonal sera recognize a broad band of erythrocyte protein with an apparent molecular weight of 45,000-60,000.
Identical results were observed using purified erythrocyte glucose transporter protein (data not shown). The LY-and s-antisera were then tested by competition ELISA for their ability to recognize purified human erythrocyte glucose transporter protein. Fig. 2A demonstrates that both N-and d-antisera bind to nondenatured glucose transporter protein to a similar extent and that purified erythrocyte glucose transporter competes equally for 01-and d-antibody binding to the solid phase glucose transporter protein. In a parallel experiment, intact erythrocytes were substituted for purified glucose transporter protein and incubated with o(-or &antiserum prior to the addition of the antisera to microtiter wells. Results from this experiment are shown in Fig. 2B. Intact human (circles) and rat (triangles) red blood cells depleted the 6-but not the a-serum, of anti-human erythrocyte glucose transporter immunoglobulins. This unexpected finding suggested that all of the anti-glucose transporter antibodies in the &serum and few or no anti-glucose transporter antibodies in the a-serum were binding to extracellular epitopes or domains of the erythrocyte glucose transporter I I These results are very close to those obtained in measurements of D-glucose-inhibited cytochalasin B binding to human (Helgerson and Carruthers, 1987) and rat (Helgerson and Carruthers, 1989) erythrocytes (1.5 X 10" and 800 glucose transporter proteins/ cell, respectively).
An alternative method for testing the ability of the 6antiserum to bind to an extracellular domain(s) on the glucose transporter protein was to immunoprecipitate intact red blood cells bound to protein A-Sepharose beads. In this experiment, protein A-Sepharose beads were incubated with either (Yserum or &serum preincubated with purified red cell glucose carrier, washed extensively, and then incubated with intact human red blood cells. These beads were then washed, placed on microscope slides, and examined by phase-contrast microscopy. Results are shown in Fig. 3  Protein A-Sepharose beads were incubated in 1 ml of saline containing 10 ~1 of rabbit o-antiserum or 8 ~1 of rabbit a-antiserum for 30 min at 37 "C. The beads were then washed in ice-cold saline. One aliquot of beads that had been first treated with rabbit &antiserum was additionally incubated with 40 rg of purified glucose transporter for 30 min at 37 "C and then washed as above. The Sepharose beads were finally incubated with human red cells (1 x 10' cells in saline) for 30 min at 37 "C. Samples were placed on microscope slides, examined by phasecontrast microscopy at a magnification of x400, and photographed. Results are shown for protein A-Sepharose beads treated with 01antiserum, a-antiserum, and h-antiserum plus purified glucose carrier. The results shown here are typical of two experiments performed using separate batches of red cells and protein A-Sepharose. Intact human red cells, sealed red cell ghosts, leaky red cell ghosts, and sealed inside-out red cell membrane vesicles were incubated in the presence of (Y-or &serum for 1 h at 20 "C (10 wl/l X 10' packed red cells or 10 ~1/60 pg of membrane protein). These conditions were determined in preliminary experiments to reflect both equilibrium and saturated IgG binding (not shown). The cells and membranes were washed in ice-cold saline and then incubated with I" I-protein A for 1 h at 20 "C. Parallel experiments demonstrated that protein A binding reached equilibrium by this time and that protein A binding was not limited by the concentration of protein A in solution (not shown). The membranes were washed five times in 10 ml of ice-cold saline, resuspended in 200 ~1 of saline, and 60-~1 aliquots counted in triplicate. The experiment was repeated using two separate membrane and red cell preparations. The results are shown as mean + standard deviation. Preimmune serum control determinations resulted in background activities of 343-398 cpm.
intact red cells, sealed red cell ghosts, sealed inside-out red cell vesicles, and leaky red cell ghosts. Antibody binding was measured by 1251-protein A binding to the various antiseratreated membrane preparations, as shown in Fig. 4. The conditions used in this experiment were determined in parallel experiments to allow both equilibrium and saturated IgG binding (not shown). The cr-serum bound only to a glucose transporter domain or domains located inside red cells, as evidenced by "'1 binding to sealed inside-out ghosts and by the lack of lZ51 binding to sealed ghosts and intact cells. In contrast, the d-serum recognized only an extracellular domain(s) on the red cell glucose transporter protein, as evidenced by the "'1 binding to sealed ghosts and intact cells but not to the sealed inside-out ghosts. Note that in leaky red cell ghosts, both intracellular and extracellular glucose trans- Human erythrocyte ghosts and control and transfected CHO-Kl cell membranes were prepared as described under "Experimental Procedures." The membrane proteins were solubilized in sample buffer at room temperature for 30 min, resolved by SDS-PAGE (10% polyacrylamide gels), and transferred electrophoretically to nitrocellulose at 200 mA for 3 h. A, nitrocellulose was incubated with anti-GLUT1 C-terminal peptide antiserum (l/1000 dilution), washed extensively, then incubated with ""I-protein A for 1 h at room temperature. The immunoreactive proteins were visualized by autoradiography. Lanes I,9,and 10  porter domains are accessible to the antibodies and that the leaky ghosts bound roughly equal amounts of (Y-and &immunoglobulins. Similar results were obtained when fluorescein-conjugated goat anti-rabbit IgG was substituted for ?protein A as the reporter molecule (not shown).

Generation of CHO-Kl Phenotypes
That Express Human GLUT1 Protein-CHO-Kl control cells and CHO-Kl cells transfected with a mammalian expression vector containing the HepG2 glucose transporter cDNA pLENGT were grown in culture, and the total membrane proteins from these cells were harvested as described under "Experimental Procedures." Expression of total immunoreactive GLUT1 protein was assayed by protein immunoblot analysis using an anti-GLUT1 C-terminal peptide serum (Fig. 5, A and B) and a mouse polyclonal antiserum that is specific for human GLUT1 protein" (Fig. SC). for 48 h as described in detail under "Experimental Procedures." Cells were washed three times with phosphate-buffered saline and serum starved for 2 h at 37 "C. Cell monolayers were washed again once and then poisoned for 20 min in buffer containing 10 mM sodium azide and 20 mM 2-deoxyglucose at 37 "C. Following this incubation, the cells were washed with phosphate-buffered saline, incubated with either rabbit preimmune serum or &antiserum (diluted from l/250 to l/2000) for 2 h at room temperature. The cells were then washed with phosphate-buffered saline and incubated for 1 h with ""I-protein A at room temperature. The wells were washed three times with phosphate-buffered saline, and the bound "'I-protein A was solubilized in 0.1% SDS and counted in a y-counter. Nonspecific binding was subtracted from total binding for each cell line. Each assay point is the average of three determinations.
brane proteins with either a GLUT1 C-terminal peptide antiserum (Fig. 5B) or an antiserum specific for human GLUT1 (Fig. 5C) confirmed that it was the human glucose transporter protein that was overexpressed (Fig. 5C, lanes 4-9) and not host glucose transporter protein. Note that no CHO host glucose transporter protein was detected by the mouse serum (Fig. 5C, lanes 2 and 3), but the CHO GLUT1 protein was clearly detected by the C-terminal peptide antiserum (Fig. 5B,  lanes 2 and 3). This experiment also illustrates the low level of expression of human GLUT1 in the CHO GT26 cells (Fig.  5, B and C, lane 9). Additionally, no rat GLUT4 protein was detected in the CHO-Kl cells by immunoblot analysis with the monoclonal antibody lF8 or with anti-rat GLUT4 Cterminal peptide serum (not shown)! Quantitative immunoblot analysis indicates that overexpression of the human GLUT1 protein in these cell lines ranges from 2-fold (CHO-GT26) to 17-fold (CHO-GT3). Assuming equivalent binding of the C-terminal peptide antibodies to the denatured erythrocyte ghost glucose transporter and to the expressed HepG2 glucose transporter protein (Gould et al., 1989) and a M, = 55,000 for each of these proteins (Fig.  5), and assuming that 2% of the erythrocyte ghost protein preparation is GLUT1 protein (Helgerson and Carruthers, 1987), the calculated numbers of GLUT1 proteins/cell in the control CHO-Kl and in CHO-GT3 cells are approximately 7.3 X lo4 and 1.2 X 106, respectively.
Quantitative Comparisons among Total Cellular Expressed GLUT1 Protein, Cell Surface GLUT1 Protein, and 2-Deoxyglucose Uptake-Experiments were conducted to determine whether the &antiserum-binding assay could detect the expressed human glucose transporter protein on the surface of CHO-GT cells (Fig. 6). Cell surface expression of GLUT1 was measured in CHO-Kl control, CHO-GTl, and CHO-GT3 cells, using a range of dilutions of d-serum from 250-to 2000fold. Specific, saturable d-antibody binding was observed for each of the three cell lines tested. At these antibody dilutions, overexpression of cell surface GLUT1 was observed in cell lines CHO-GTl and CHO-GT3 relative to the CHO-Kl control. The magnitude of lz51-protein A binding in the cell culture wells was dependent upon a-antibody dilution (Fig. 6 and upon cell density (not illustrated).
The results appeared to correlate well with levels of expression of total immunoreactive GLUT1 protein measured by immunoblot analysis in these cells.
The relationships among levels of expression of total immunoreactive GLUT1 protein, cell surface s-antiserum binding, and 2-deoxyglucose uptake activity were measured in various transfected cell lines (Fig. 7). CHO-Kl control and CHO-GT cell lines were plated on the same day in paired 24well culture dishes, grown in culture for 48 h, and assayed for b-serum binding and 2-deoxyglucose uptake, as described under "Experimental Procedures." Total immunoreactive GLUT1 was determined by quantitative immunoblot analysis using anti-GLUT1 C-terminal peptide serum. The results of these three assays are plotted as fold increases over values measured in control cells (CHO-Kl = 1). Levels of expression of total immunoreactive GLUT1 protein ranged from approximately 2-fold over those of control in cell line CHO-GT26 to 17-fold in cell line CHO-GT3, with two other cells lines expressing intermediate levels of GLUTl. CHO-GT26 cells expressed 1.6-fold greater 2-deoxyglucose uptake activity and 1.6-fold greater cell surface antibody binding than CHO-Kl controls, and similar good correlations between levels of expression of sugar uptake and intact cell h-antibody binding were observed in the CHO-GT cell lines expressing much higher levels of human GLUT1 protein (Fig. 7). It was also observed that increasing levels of expression of total immunoreactive glucose transporter protein in different cell lines resulted in increasing levels of expression of 2-deoxyglucose uptake and cell surface antibody binding to intact cells. Note that at low levels of human GLUT1 expression, 1.9-fold increases in glucose transporter protein result in 1.6-fold increases in sugar uptake, whereas 9-12-fold increases in GLUT1 protein expression result in 4-5-fold increases in sugar uptake. Similar relationships between 2-deoxyglucose uptake by CHO-Kl, CHO-GTl, and CHO-GT3 were observed when uptake was measured at 2 mM substrate (not shown),4 suggesting that hexokinase activity was not saturated in these cells under our experimental conditions. ing, as described in detail under "Experimental Procedures." Cells were serum starved for 2 h, and sugar uptake was measured following incubation of the cells with 10m7 M insulin for 20 min at 37 "C. As shown in Fig. 8, insulin stimulated 2deoxyglucose uptake 40% in both CHO-Kl control cells and in CHO-GT26 cells that express human GLUT1 at levels about 2-fold over controls. Insulin stimulated 2-deoxyglucose uptake 10% in CHO-GTl cells, and CHO-GT3 cells were insensitive to insulin in these experiments. These latter two cell lines exhibited much higher levels of human GLUT1 protein (9-and 17-fold over controls, respectively, Fig. 7). The absolute increase in the rate of deoxyglucose uptake due to insulin in CHO-GTl cells were approximately equal to that measured in the CHO-GT26 cells (Fig. SA, legend). Insulin failed to detectably stimulate &antibody binding to any of the cell lines tested in these experiments.
In the absence of insulin, increases of 1.6-and 5-fold in s-antibody binding to CHO-GT26 cells and CHO-GTl cells, respectively, were observed compared with control cells, consistent with the increases in basal sugar uptake measured in those cells. In all h-antibody-binding determinations (six determinations in each experiment) the standard errors of the mean were always ~3.5%. These data demonstrate that insulin-stimulated glucose transport activity in control and transfected CHO cells is not accompanied by increases in immunoreactive GLUT1 at the cell surface.

DISCUSSION
The results presented in this report demonstrate the development of a novel anti-GLUT1 antiserum that recognizes one or more extracellular epitopes on the GLUT1 protein. Intracellular epitopes on GLUT1 are not recognized by the antiserum. This rabbit s-antiserum preparation is unique in that most antisera raised against the erythrocyte glucose transporter previously appear to recognize intracellular domains on the red cell glucose transporter.
The rabbit a-antiserum described in the present work is an example of this type of antiserum.
Both LY-and b-antisera recognize denatured GLUT1 protein in immunoblots (Fig. l), and both antisera recognize similar amounts of nondenatured red cell glucose transporter measured by competitive ELISA and by antibody binding to leaky red cell ghosts (Figs. 2 and 4). However, 6but not a-antiserum binds to intact red cells and sealed right side-out red cell ghosts, and the &antiserum binding to the glucose transporter extracellular domain(s) was blocked by prior incubation of the antibody preparation with purified human erythrocyte glucose transporter protein. Additionally, 6-but not a-antiserum immobilized onto protein A-Sepharose beads was able to immunoprecipitate intact red cells (Fig. 3). An antibody reagent that may recognize exofacial GLUT1 epitopes has been described previously by Burdett and Klip (1988). However, the binding of that antibody to intact red cells was very low, only 0.25% of the total GLUT1 protein/ cell. Rabbit &antiserum appears to bind quantitatively, and it was used to calculate the number of cell surface glucose transporter proteins on human and rat red cells, resulting in an estimate of 1.4 X lo5 and 680 glucose transporter proteins/ cell, respectively.
These data confirm independent measurements of red cell glucose transporter proteins determined by D-glucose-inhibited cytochalasin B binding to human (Helgerson and Carruthers, 1987) and rat  erythrocytes.
The GLUT1 protein was examined in control CHO fibroblasts and in CHO fibroblasts transfected with the human GLUT1 expression vector and pLENGT.
Using specific antibody reagents, we were able to monitor total cellular and cell surface host (hamster) GLUT1 in control cells, total cellular and cell surface hamster plus human GLUT1 in transfected cells, and total human GLUT1 in transfected cells. Host and heterologously expressed human GLUT1 proteins were detected by antipeptide antiserum directed against GLUT1 C-terminal 12 amino acids and by the rabbit polyclonal antiserum (6) described above (Fig. 5, A and B). The expressed human GLUT1 was also detected by a mouse polyclonal antiserum raised against purified human erythrocyte glucose transporter. This latter antiserum is highly selective for human GLUT& and it did not react with the rodent GLUT1 (Fig. 5C). James and co-workers (1988) developed a monoclonal antibody, lF8, by immunizing mice with partially purified low density microsomal proteins from insulin-treated rat adipocytes. Their antibody is specific for GLUT4 protein, prevalent in fat and muscle, and it does not recognize GLUT1 from rat, mouse, or human tissues4 (James et al., 1888(James et al., , 1989Zorzano et al., 1989). In our studies, no GLUT4 protein was detected in CHO-Kl cells using the antiadipocyte glucose transporter monoclonal antibody or an anti-rat GLUT4 Cterminal peptide serum (James et al., 1989).
The specific antibody reagents described above were used to determine whether 1) increased expression of GLUT1 in transfected CHO cells leads to increased numbers of cell surface GLUT1 proteins; and 2) the expressed GLUT1 transporters are functional.
We observed excellent correlations in numerous cell lines among overexpression of human GLUT1 5800 Insulin Regulation of Glucose Transporters total immunoreactive protein, cell surface GLUT1 protein, and 2-deoxyglucose uptake (Fig. 7). These data demonstrate that the expressed human GLUT1 protein is processed and exported to the surface of the transfected fibroblasts and that these cell surface carrier proteins are capable of transporting deoxyglucose across the plasma membranes of these cells.
The pLENGT mammalian expression vector contains metal-inducible human metallothionein gene II-promoting sequences, and zinc and calcium were expected to induce expression of GLUT1 in cells transfected with this vector. As expected, transporter expression from this vector in 3T3-Ll cells exhibited very low constitutive expression, and high levels of zinc (75-125 pM) were required to achieve 4-&fold overexpression of glucose transporter protein.3 However, in the transfected CHO cells used in the present study, expression of the transporter protein was constitutively very high in the absence of the added metals. Addition of zinc or cadmium had little or no further effect on GLUT1 expression. Similarly, Oka and co-workers observed constitutively high levels of rabbit GLUT1 in CHO cells transfected with an expression vector containing mouse metallothionein gene Ipromoting sequences (Asano et al., 1989). We have no documented explanation for this phenomenon at present. It is unclear whether the insulin sensitivity of sugar uptake in various types of cells, including muscle, adipocyte, and fibroblasts, is dependent upon glucose transporter protein primary structure, cell-specific regulatory machinery, or both. Insulin stimulation of glucose uptake in adipocytes occurs in association with recruitment of GLUT4 from intracellular membranes to membranes at the cell surface (James et al., 1988;Zorzano et al., 1989). Only 5-10% of the glucose transporters present in these cells are GLUT1 proteins Zorzano et al., 1989). Insulin stimulation of glucose uptake has also been observed in CHO (Fig. 5) and 3T3-Ll fibroblasts3 (James et al., 1989) that contain GLUT1 but not GLUT4. However, insulin stimulation of glucose uptake has not been observed in a number of other GLUTl-containing cells including brain, human erythrocytes, and cultured HepG2 cells. Recent data by Oka and co-workers indicate that when GLUT1 isolated from rabbit brain (97.5% identity with the HepG2 glucose transporter ) is expressed in CHO fibroblasts, both the CHO control and the expressed rabbit GLUT1 deoxyglucose uptake activities are stimulated approximately 40% by insulin (Asano et al., 1989).
Similarly, insulin caused translocation of human GLUT1 expressed in differentiated 3T3-Ll cells (Gould et al., 1989), but the GLUT1 contribution to insulin-stimulated sugar uptake was not examined. We found that 100 nM insulin increased 2-deoxyglucose uptake approximately 40% in control CHO-Kl cells and in CHO-Kl cells expressing the human GLUT1 protein 2-fold over endogenous host GLUT1 protein (Fig. 8A). These results strongly suggest that a human glucose transporter protein, GLUTl, which is not responsive to insulin in HepG2 cells, is regulated by insulin when expressed at low levels in Chinese hamster ovary cells. Thus, specific insulin-responsive cellular processes rather than transporter isotype structure appear to be involved in transporter regulation by this hormone.