Chronic Treatment with Insulin Selectively Down-regulates Cell-surface GLUT4 Glucose Transporters in 3T3-Ll Adipocytes*

A new method for photoaffinity labeling of glucose transporters has been used to compare the effects of glucose-starvation, acute-insulin, and chronic-insulin treatments on the cell-surface glucose transporters in 3T3-Ll adipocytes. Starvation alone increased the cell-surface levels of GLUTl and GLUT4 by =4- and =2-fold, respectively. As shown by Calderhead, D, M., Kitagawa, K., Tanner, L. T., Holman, G. D., and Lien-hard, G. E. (1990) J. Biol. Chern. 265, 13800-13808) acute-insulin treatment increased cell-surface GLUTl and GLUT4 by =5- and =15-fold respectively. In con- trast to this, chronic-insulin treatment gave a further 3-4-fold increase in both cell-surface and total cellular GLUT1, but availability of GLUT4 at the cell-surface was down-regulated to half the level found in the acute treatment but with no change in the total cellular level. This effect occurred in starved and non-starved cells and suggests that starvation, acute-insulin, and chronic-insulin treatments regulate glucose transporter availability through independent mechanisms. The down-regulation of GLUT4 reached a maximally reduced cell-surface level in 6 h while the rise in GLUTl reached a maximum after 24-48 h. The rise in GLUTl appeared to compensate for the decline in cell-surface GLUT4 as glucose transport activity was further increased during the long term treatment with insulin. The down-regulation of

Differentiated 3T3-Ll adipocytes are a good model system for investigating long term regulatory effects on glucose transport. They respond to acute-insulin treatment with increases in glucose transport of 10-20-fold above basal levels (1,2). It has recently been shown that differentiation of these cells from fibroblasts is associated with an increase in cellular mRNA and protein for the GLUT4 isoform (3,4). Calderhead et al. (2) have shown that the cell-surface labeling of GLUT4 increases 15-fold above basal levels and that this increase is similar to, but slightly less than, the stimulation of glucose transport activity. The GLUTl isoform is also abundant in differentiated 3T3-Ll cells (1,2,5), and this isoform increases 3-fold (1) to 5-fold (2) above basal levels in response to acute * This work was supported by grants from the Medical Research Council (United Kingdom) and the British Diabetic Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
insulin. In contrast to this, chronic-insulin treatment has been shown (5,6) to produce an 4-fold further increase in the mRNA and total protein for the GLUTl isoform. However, no change was observed in the total cellular mRNA or protein for the GLUT4 isoform. The redistribution of these transporters to the cell-surface following the chronic treatment is examined here.
The relationship between glucose starvation and insulin treatments as modifiers of glucose transport has been examined in many cultured cells (6)(7)(8)(9)(10)(11) including 3T3-Ll adipocytes (9)(10)(11) and in whole animal studies (12,13). Starved rats show decreases in adipocyte GLUT4 which resemble those changes which occur in streptozotocin-treated diabetic rats. In both cases circulating insulin decreases and so separating the possible independent effects of insulin and starvation is difficult. An advantage of using cultured cells for studies on regulation of this type is that glucose-starvation and insulin treatment can be independently varied. A disadvantage of using cultured cells is that they generally contain high levels of the GLUTl transporter which is only present at low levels in isolated adipose and muscle tissue. Thus, for the cultured cells a resolution of the separate changes occurring in GLUTl and GLUT4 is required. It is particularly important to measure the cell-surface availability of these isoforms under the different regimes of starvation and acuteor chronic-insulin treatment. We have therefore studied these changes here by cell-surface labeling the glucose transporters using the impermeant bis-mannose photolabel, 2-N-(4-( l-azi-2, 2, 2-trifluoroethyl)benzoyl-1,3-bis(~-mannos-4-yloxy)-2propylamine (ATB-BMPA)' and then immunoprecipitating the labeled transporters with anti-GLUT1 or anti-GLUT4 antibodies (2,14,15).
We have also used this technique to examine changes in the cell-surface transporter distribution that are associated with the insulin resistance which has been shown to occur following a chronic-insulin treatment (16,17 Cell Culture-3T3-Ll fibroblasts were cultured in DMEM and differentiated to adipocytes as described (1,19,20). Before use in 2deoxy-D-glucose transport assays or in cell-surface labeling experiments the cells were subjected to a standard washing and refeeding procedure. Cells were first washed with phosphate-buffered saline (154 mM NaCl, 12.5 mM sodium phosphate, pH 7.4) and were then The abbreviations used are: ATB-BMPA, 2-N-4-(1-azi-2, 2, 2trifluoroethyl)benzoy1-1,3-bis-(~-mannos-4-yloxy)-2-propylamine; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid MES, 2-(N-morpholino)-ethanesulfonic acid ClzEs, nonaethyleneglycol dodecyl ether; SDS, sodium dodecyl sulfate; PAGE, polyacrylamine gel electrophoresis; DMEM, Dulbecco's modified Eagle's medium. incubated for 2 h in serum-free medium containing 25 mM D-glucose. This was followed by three washes in Kreb's-Ringer-HEPES buffer (KRH buffer,136 mM NaCl,4.7 mM KCl,1.25 mM CaC12,1.25 mM MgS04,10 mM HEPES,pH 7.4) and where appropriate with a subsequent treatment with 100 nM porcine monocomponent insulin for 30 min at 37 "C in 1 ml of KRH buffer (an acute-insulin treatment). In the case of the glucose-starvation treatment, fully differentiated cells were incubated in RPMI containing 25 mM D-xylose to replace glucose. In the case of the chronic-insulin treatment, fully differentiated cells were either incubated for 24 h in normal medium (DMEM with 25 mM D-glucose) or glucose-starvation medium (RPMI containing 25 mM D-xylose instead of D-glucose) with 500 nM porcine monocomponent insulin. In the cases where cells were glucose-starved and chronically insulin treated, the treatment was reversed using the standard washing and refeeding procedure (above). For non-starved cells which were chronically insulin treated, the treatment was reversed by incubating cells for 1 h in Kreb's-Ringer-MES buffer (of the same composition as KRH buffer but with MES instead of HEPES) pH 6.0, and containing 25 mM D-glucose, followed by three further washes in this buffer in addition to the standard washing procedure (above).
2-Deoxy-~-Glucose Transport Assays-2-Deoxy-D-glucose uptake rates were determined as described (1,19). The 3T3-Ll cell monolayers in 35-mm dishes were incubated with 50 p M 2-de0xy-[l-~H]~glucose in 1 ml of KRH buffer at 37 "C for 5 min. Cells were then rapidly washed three times in KRH buffer at 0-4 "C, and then the cell associated radioactivity was extracted and counted. The transport data were corrected for a background and for nonspecific uptake which was determined by incubation of cells in the presence of 30 p~ cytochalasin B.
Insulin Binding-Tracer insulin binding was assayed as described by Van Putten et al. (18) with some modifications. Cells in 35-mm dishes were incubated in serum-free DMEM for 2 h, washed three times in KRH buffer, and incubated at 20 "C in 1 ml of KRH containing 70 fmol of m0no-A14-'~~I-insulin, 2% serum albumin, bacitracin (10 mg/ml) and where appropriate with 5 nM insulin. The dishes were shaken (65 oscillations/min) for 3 h. After washing three times in KRH, 1 ml of ice-cold 10% trichloracetic acid was added and the insoluble protein precipitated and counted for radioactivity. Specific '2sI-insulin binding was calculated as the difference between the '""Iinsulin which was bound to the cells in the presence and absence of added unlabeled 5 nM insulin. The nonspecific binding was 25% of the total '"I-insulin binding. To determine insulin binding following chronic-insulin treatment, cells were subjected to the washing and refeeding procedures described under "Cell Culture" above.
ATE-BMPA Photolabeling-Plates of differentiated cells (35 mm) were washed in KRH buffer and were then incubated in 200 pCi of ATB-[B"H]BMPA for 2 min and irradiated for 1 min in a Rayonet RPR-100 photochemical reactor with RPR-300-nm lamps as described (2). To estimate relative changes in total cellular transporter, the differentiated cells were scraped from 35-mm dishes and homog-

-
enized in 500 pl of KRH buffer. The suspension was then mixed with 200 pCi of label in 35-mm dishes and irradiated as described for the cell-surface labeling. The irradiated cells or homogenates were then washed three times in KRH buffer and solubilized in 1.5 ml of detergent buffer containing 2% &Eg, 5 mM sodium phosphate, 5 mM EDTA, pH 7.2, with the proteinase inhibitors antipain, aprotinin, pepstatin, and leupeptin, each at 1 pg/ml. Following centrifugation at 20,000 X gmax for 20 min, the supernatants were subjected to sequential immunoprecipitation with 8 pl of protein A-sepharose coupled to either affinity purified anti-GLUT1 (20 pg) or anti-GLUT4 (10 pg) antibodies. These antibodies were raised against C-terminal peptides as described (14). After incubation for 1.5-2 h at 0-4 "C and washing of the immunoprecipitates three times with 0.2% CI2E9 in detergent buffer, the labeled glucose transporters were released from the antibody complexes with SDS-urea electrophoresis sample buffer and subjected to electrophoresis on 10% acrylamide gels. The radioactivity on the gels was estimated by cutting and counting gel slices. The radioactivity in transporter peaks was corrected for a background which was based on the average radioactivity of the slices on either side of the peak (1). Fig. 1 shows the acute response to insulin in non-starved 3T3-Ll cells. The ATB-BMPA was used to label intact cells, and the GLUTl and GLUT4 immunoprecipitates were resolved on SDS-polyacrylamide gels. The acute (30 min) treatment with insulin increased the labeling of GLUTl and GLUT4 in this experiment to =3-and ZlO-fold (~3 . 5 -and ~12-fold from five experiments) of the level found with basal cells. Also shown in Fig. 1 is the effect of exposure of the cells to insulin for 24 h (a chronic treatment). This resulted in a further 4-fold increase in GLUTl to give a total stimulation which was 13-fold above basal GLUTl levels (~14-fold from three experiments). In contrast to this, the cell-surface levels of GLUT4 were reduced to 53% of that found in the acute treatment (from three experiments). In addition we have confirmed ( 5 ) (by measuring ATB-BMPA photolabeling of transporters in cell homogenates) that the total amount of GLUTl rises 6-fold while the total amount of GLUT4 rises only slightly (by ~1.4-fold) following the chronic-insulin treatment.

RESULTS
The time course for the up-regulation of GLUTl and the down-regulation of cell-surface GLUT4 has been examined. Fig. 2 shows that a decrease in GLUT4 occurred after 4-6 h while GLUTl steadily increased over 24 h and remained at this elevated level for a further 24 h. The stimulation of 2-Y In P n 2*ol  x In ? slice slic GLUT4 Down-regulation in 3T3-Ll Adipocytes deoxy-D-glucose transport was increased by the chronic-insulin treatment, and Fig. 2 shows that this rise paralleled the increase in GLUT1.
Since long term regulation of glucose transport activity is known to be glucose sensitive (7)(8)(9)(10)(11)(12)(13)  acute and chronic responses to insulin in cells which were maintained in glucose-deprived media. This starvation regime resulted in a 3-fold rise in the basal level of 2-deoxy-~-glucose transport which as shown in Fig. 3 is associated with a -4fold increase in cell surface GLUTl and a -2-fold increase in cell-surface GLUT4 above basal levels obtained in nonstarved cells from the same plating. As shown in this figure, the starvation regime reduces the fold change in insulin stimulation mainly by raising basal levels of cell-surface transporters. Chronic treatment of starved cells with insulin increased GLUTl =4-fold and decreases GLUT4 to 49% (from five experiments) of the levels found with the acute treatment. In both starved and non-starved cells, the ratio of GLUTl/GLUT4 changed from 1.1:l in acute-insulin treatment to =9:1 following chronic-insulin treatment.
In Fig. 4 a and b, the 2-deoxy-D-glucose transport rates in non-starved and glucose-starved cells are compared with the cell-surface levels of photolabeled GLUTl and GLUT4. The results presented in this figure are the mean and S.E. from three to seven independent experiments. In non-starved cells acute-insulin treatment increased the 2-deoxy-~-glucose transport by =14-fold compared with basal levels. Chronicinsulin treatment of both non-starved and glucose-starved cells resulted in only an additional 40% increase in glucose transport activity above that found following the acute treatment. The very large 4-fold increase in GLUTl compensates for a -50% down-regulation of cell-surface GLUT4.
We next examined whether the down-regulation of the GLUT4 glucose transporter in both starved and non-starved cells is associated with any lesion in the cell's ability to respond to a subsequent acute-insulin treatment (Figs. 4 and 5). In Fig. 4, a and b, the 2-deoxy-~-glucose transport rates are compared with the labeling of both GLUTl and GLUT4. In all cases the second challenge with insulin resulted in blunted responses. Fig. 5 shows that the effect of starvation alone is readily reversed by refeeding with 25 mM D-glucose. The starved and refed cells showed normal rises in cell-surface GLUTl (panel a) and GLUT4 (panel c ) in response to acuteinsulin treatment. However, starved cells which had been  (panel b and d ) . The transport stimulations in response to the second insulin treatment were also found to be reduced below those found in a normal acute insulin response (Fig.  4b), but the increases in transport appeared to be greater than could be accounted for by any increase in cell-surface transporters. Thus, these cells have developed striking insulin resistance mainly in the ability to recruit additional transporters to the cell-surface. In addition, it is clear from this experiment that starvation alone is not responsible for this resistance. We have confirmed (17) that insulin binding is unchanged following chronic-insulin treatment; the tracer insulin binding was 89% (for starved cells) and 88% (for cells which had been subjected to the starvation and chronic insulin and washing regime) of that of normal cells.
In non-starved chronically insulin-treated cells, the transport stimulation due to chronic insulin is more persistent and is more difficult to reverse. The resistance to further insulin stimulation thus has to be measured against an already high transport rate. The persistence of the elevated transport activity appears to be due to a failure to reverse the high cellsurface level of GLUTl transporters. As is the case in starved cells, the non-starved cells show a larger proportionate increase in 2-deoxy-~-glucose transport activity in response to additional insulin than can be accounted for by any additional increase in cell-surface transporters. As in non-starved cells, the main source of the resistance appears to be in a failure to recruit additional transporters to the plasma membrane.

DISCUSSION
The bis-mannose-photolabeling technique has been shown to provide a means of studying changes in cell-surface levels of glucose transporters without the need to subcellular fractionate cells (2,14). The acute effect of insulin on cell-surface labeled transporters has been studied in 3T3-Ll cells (2) and in rat adipocytes (14). In both systems insulin stimulates the cell-surface labeling of GLUT4 by -15-fold while GLUTl is increased by only -5-fold. Translocation of GLUT4 to the cell-surface can therefore account for a large proportion of the stimulatory effect of insulin on glucose transport activity in these cells. However, the stimulation of transport activity is generally slightly greater (-2-fold) than the translocation of either transporter so that intrinsic activation of these transporters may also occur as suggested (21-24).
As well as providing a means of measuring changes in each transporter, the technique can also give an estimate of the relative abundance of the two transporter isoforms. These latter comparisons are of course only valid if the efficiencies of labeling and immunoprecipitation of the two transporters are equal. Western blotting has shown that erythrocytes contain only GLUTl (25) and that in rat adipocytes over 90% of the total transporter is the GLUT4 isoform (2,(26)(27)(28). The affinities of GLUTl (in erythrocytes) and GLUT4 (the predominant isoform in rat adipocytes) for ATB-BMPA have been shown, from transport inhibition experiments, to be -250-300 ~L M (14, 15). We have measured the K, for ATB-BMPA inhibition of 3-O-methyl-~-glucose transport in 3T3-L1 cells and find that this is -250 PM following both acuteand chronic-insulin treatments. In addition, we have carried out binding experiments at a range of ATB-BMPA concentrations and find that the K d of GLUTl and GLUT4 are equal and are -200 p~ following both acute-and chronic-insulin treatments.2 The efficiencies of labeling of these glucose transporters in erythrocytes and adipocytes has been estimated to be -2500 dpm/pmol transporter/mCi ml" of . These comparisons suggest that the labeling efficiencies for GLUTl and GLUT4 are roughly equal. In addition, we have shown that the immunoprecipitations of erythrocyte GLUTl and adipocyte GLUT4 are about 70-80% efficient (14,15). Thus, the proportion of label recovered in the GLUT4 and GLUTl immunoprecipitates is likely to be a good approximation of the relative abundance of the two isoforms. In insulin-stimulated rat adipocytes, our estimate (14) of the ratio of GLUTl/GLUT4 is 1:lO which agrees well with the Western blotting data of total cell membranes (2). However, in insulin-stimulated 3T3-Ll cells the relative abundance of the GLUTl/GLUT4 as estimated from cell-surface labeling with ATB-BMPA is 1.6:l (2) while Western blotting of total cell membranes has shown a ratio of -3:l (2). However, Western blotting data is also subject to quantitative uncertainties such as the reliability of the protein standards. The transporter contents of these protein standards have to be equated with the number of cytochalasin B-binding sites which may also be uncertain since the affinity of GLUT4 for cytochalasin B has not been directly measured. Thus, both techniques may give only an approximation of the relative transporter abundance but can give a reliable estimate of changes in the levels of each transporter.
Tordjman et al. ( 5 , l l ) have examined the long term effects of starvation and of chronic-insulin treatment on glucose transport regulation in 3T3-Ll cells by Western blotting of the GLUTl and GLUT4 isoforms. The increases in ATB-BMPA-labeled cell-surface GLUTl (by -4-fold) and GLUT4 (by =2-fold) following starvation that we have observed here agree with their results (11). Tordjman et al. (5) and Hanique et al. (6) showed that chronic-insulin treatment increased total cellular GLUTl by -4-fold but did not change total cellular GLUT4. We have confirmed this in the cells used in our study by photolabeling the total cell homogenates with ATB-BMPA (see "Results") and by photolabeling transporters in digitonin-permeabilized cells with ATB-BMPA.' However, Clancy and Czech (24) have recently reported a reduction in total cellular GLUT4 following chronic-insulin treatment. We have extended these studies on total cellular transporter by showing that GLUT4 at the cell-surface is down-regulated following the chronic-insulin treatment while levels of GLUTl at the surface are markedly increased. The down-regulation occurs before the increase in GLUTl at the surface reaches a maximum. Over this time course glucose transport activity is also increased. This suggests that the increase in GLUTl compensates for the decrease in GLUT4. The effect occurs in non-starved and glucose-starved cells and suggests that the two transporters are independently regulated by glucose and insulin.
The chronic-insulin treatment has been shown here and by Rosen et al. (17) to cause a persistent lesion in the cells ability to increase glucose transport activity in response to further acute-insulin treatment. We show here that the treatment is particularly associated with a resistance of the cells to recruit further GLUT4 to the cell-surface following the additional insulin treatment. 2-Deoxy-D-glucose transport activity has been shown to be less resistant than transporter recruitment to the second insulin challenge. This suggests that, in addition to the well established role of transporter translocation in activation of transport (29, 30), intrinsic activation of any transporters already present in the plasma membrane may GLUT4 Down-regulation in 3T3-Ll Adipocytes 11731 also play an important role in regulation of the glucose transport rate (21-24). Thus, it is possible that when insulin is removed there is some intrinsic deactivation of GLUT1 transporters that are not removed from the surface. These cellsurface transporters may be reactivated in the subsequent insulin treatment. However, the most serious lesion that occurs following chronic-insulin treatment of starved and non-starved cells appears to be a failure to increase GLUT4 to normal cell-surface levels in a subsequent acute-insulin challenge.
Three types of insulin resistance in glucose transport have been recognized (31-38). These are resistance at the insulin receptor level (31-33), resistance in the signaling pathway (34, 35), and resistance due to a depleted pool of glucose transporters (12,13,(36)(37)(38). We have shown that the downregulation of cell-surface GLUT4 and the subsequent resistance of this isoform to respond to acute-insulin treatment is neither at the insulin receptor nor due to depleted total reserves of transporter. Instead, chronic-insulin treatment results in an insulin resistance that is most likely due to a signaling defect between the occupied insulin receptor and the pathway which stimulates glucose transporter translocation to the cell-surface.