Insulin regulation of the two glucose transporters in 3T3-L1 adipocytes.

The amounts of the brain type and muscle type glucose transporters (designated Glut 1 and 4, respectively) in 3T3-L1 adipocytes have been determined by quantitative immunoblotting with antibodies against their carboxyl-terminal peptides. There are about 950,000 and 280,000 copies of Glut 1 and 4, respectively, per cell. Insulin caused the translocation of both types of transporters from an intracellular location to the plasma membrane. The insulin-elicited increase in cell surface transporters was assessed by labeling the surface transporters with a newly developed, membrane-impermeant, photoaffinity labeling reagent for glucose transporters. The increases in Glut 1 and 4 averaged 6.5- and 17-fold, respectively, whereas there was a 21-fold in hexose transport. These results indicate that the translocation of Glut 4 could largely account for the insulin effect on transport rate, but only if the intrinsic activity of Glut 4 is much higher than that of Glut 1. The two transporters are colocalized intracellularly: vesicles (average diameter 72 nm) isolated from the intracellular membranes by immunoadsorption with antibodies against Glut 1 contained 95% of the Glut 4 and, conversely, vesicles isolated with antibodies against Glut 4 contained 85% of the Glut 1.

The amounts of the brain type and muscle type glucose transporters (designated Glut 1 and 4,respectively) in 3T3 The increases in Glut 1 and 4 averaged 6.5-and 17-fold, respectively, whereas there was a al-fold in hexose transport. These results indicate that the translocation of Glut 4 could largely account for the insulin effect on transport rate, but only if the intrinsic activity of Glut 4 is much higher than that of Glut 1. The two transporters are colocalized intracellularly: vesicles (average diameter 72 nm) isolated from the intracellular membranes by immunoadsorption with antibodies against Glut 1 contained 95% of the Glut 4 and, conversely, vesicles isolated with antibodies against Glut 4 contained 85% of the Glut 1.
Insulin stimulates hexose transport in 3T3-Ll adipocytes rapidly and typically by lo-fold or more (1,Z). Because of this large effect, this cultured ceil line has served as a model for investigations of the basis for insulin action on transport. It has been known for a number of years that 3T3-Ll adipocytes contain a glucose transporter of the type also found in brain and human erythrocytes (hereafter referred to as Glut 1)' (3). Within the past year, it was discovered that these cells also possess a second glucose transporter, of the type found in several tissues (muscle, fat) where transport is acutely stimulated by insulin (hereafter referred to as Glut 4) (4-7). We have previously shown that a pool of the Glut 1 in 3T3-Ll adipocytes is located in intracellular membranes and that insulin causes the translocation of a portion of this intracellular Glut 1 to the plasma membrane. However, the increase in Glut 1 at the cell surface in response to insulin, which we have assessed by quantitative immunoelectron microscopy (8), by labeling of surface Glut 1 with galactose oxidase and [3H]borohydride (2), and by the relative Glut 1 content of a plasma membrane fraction (9), is only 2-3-fold, much less than the increase in the hexose transport rate. We have also described the isolation of membranous vesicles that contain the insulin-responsive intracellular Glut 1 and the characterization of these vesicles (10).
The present study was undertaken to define the role of Glut 4 in insulin-stimulated hexose transport in 3T3-Ll adipocytes. Through the use of a newly developed immunological method, the absolute amounts of the two transporters in 3T3-Ll adipocytes has been determined.
By means of a novel membrane-impermeant, photoactivated affinity-labeling reagent for glucose transporters (26), and also by subcellular fractionation, the insulin-elicited increases in both Glut 1 and Glut 4 at the cell surface have been assessed. Finally, the vesicles containing the intracellular Glut 1 have been shown to contain also the intracellular Glut 4.
Only a fraction of the adsorbed transporters (in each case about 30% of the total in the unadsorbed supernatant) was released from the protein A-Sepharose into the SDS-urea sample buffer. In two experiments the relative yields of both Glut 1 and of Glut 4 in these SDS-urea samples derived from basal and insulin-treated cells were compared and found to be the same. In some experiments, which are not reported in Table I, the amount of label found in Glut 1 and 4 was about one-third the values given in the table. The basis for the low label was traced to the loss of the transporters from the SDS-urea samples by adsorption onto the walls of colored plastic microfuge tubes to which they had been transferred. To avoid this problem, we now leave the SDS-urea sample over the protein A-Sepharose in the clear 1. The assay volume was 0.5 ml of KRP, to conserve ATB-BMPA.
The ATB-BMPA was added 5 min before the initiation of uptake.
Since the nonspecific association of 2-deoxyglucose with the monolayers was found to increase slightly at higher concentrations of ATB-BMPA, the uptake was assayed in the presence of cytochalasin B at several different concentrations of ATB-BMPA across the range of interest, and these values were subtracted from the corresponding values of the rate in the presence of ATB-BMPA alone. Uptake by insulin-stimulated cells was again for 5 min, but because the inhibited rates for basal cells were low, uptake by these was for 10 min. Subcellular Fraction&ions-The following method was employed to assess translocation of the transporters. Plates (10 cm) of 3T3-Ll adipocytes in serum-free Dulbecco's modified Eagle's medium at 37 "C were treated with 160 nM insulin for 15 min or left in the basal state. Subsequent operations were done with basal and insulin-treated cells, in parallel at 4 "C. The cells were washed twice with 250 mM sucrose, 20 mM Hepes, 1 mM EDTA, pH 7.4, and then scraped into 1.5 ml of this buffer containing protease inhibitors (0.1 mM phenylmethanesulfonyl fluoride, 1 pM pepstatin A, and 10 PM L-trons-epoxysuccinylleucylamido-(3-methyl)butane).
The cells from two plates were homogenized together in a 55-ml Wheaton tissue grinder ( 1 and 4.3T3-Ll adipocytes were solubilized in 2% C,,ER with protease inhibitors and the Glut 1 and 4 were immunoprecipitated from separate portions of the CIPER supernatant with the affinity purified antibodies against their carboxyl-terminal peptides, as described under "Experimental Procedures." As a control, the antibodies alone were adsorbed from the solubilization buffer onto protein A-Sepharose. The immunoadsorbates were solubilized in SDS sample buffer and blotted for Glut 1 (part A) and Glut 4 (part B 3T3-Ll adipocytes (35-mm plates) on day 9 after differentiation were treated with 1 FM insulin or not for 15 min and then solubilized in SDS sample buffer, as described under "Experimental Procedures." Samples were then immunoblotted with antibodies against the carboxyl-terminal peptide of Glut 1 (part A) or Glut 4 (part I?). Part A: lanes 1-5, human erythrocyte membranes containing 16, 8, 6, 4, and 2 ng of Glut 1, respectively; lanes 6-7, 2.5 and 1.25% of a 35mm plate of basal cells; lanes 8-9, 2.5 and 1.25% of a 35-mm plate of insulin-treated cells. Part B: lanes I-5, rat adipocyte low density microsomes containing 12.5, 6.3, 5, 2.5 and 1.3 ng of Glut 4, respectively; lanes 6-7, 5 and 2.5% of 1 35-mm plate of basal cells; lanes 8-9, 5 and 2.5% of a 35-mm plate of insulin-treated cells.
Electron Microscopy-Thin section electron microscopy of the intracellular vesicles containing Glut 1 and 4 adsorbed to Staph A cells was carried out as described previously (17).

RESULTS
immunoprecipitate with antibodies against Glut 1 and only Glut 4 was detected in the immunoprecipitate with antibodies against Glut 4 ( Fig. 1). Thus, the antibodies against one transporter neither immunoprecipitated nor immunoblotted the other.

Amounts of Glut 1 and 4 in 3T3-Ll Adipocytes-The
For the assay of Glut 1, the standards were human erythamounts of Glut 1 and 4 were determined by quantitative rocyte membranes, in which Glut 1 is known to constitute immunoblotting with antibodies specific for the carboxyl-5.2% of the protein by weight (14). The antibodies against terminal peptide of each. Since the carboxyl-terminal 20 the carboxyl-terminal peptide of Glut 1 should react equally amino acids of Glut 1 and 4 are the same at only four positions, well with the human and mouse Glut 1, since the sequence of which except for a dipeptide, are separated in the sequence this peptide is the same in the Glut 1s of these two species (5, (5), the antibodies against this region of the two transporters 18). For Glut 4, the standards were rat adipocyte low density should not cross-react. In agreement with this expectation, microsomes. The total transporter content of these was dewhen Glut 1 and 4 were immunoprecipitated from 3T3-Ll termined by cytochalasin B binding. Although these microadipocytes and then each immunoprecipitate was immunosomes contain both Glut 1 and Glut 4, there is evidence that blotted for both transporters, only Glut 1 was detected in the at least 90% of the transporter in these is Glut 4 (19,20 have confirmed the low percentage of Glut 1 by determining the Glut 1 content of the microsomes through immunoblotting, with human erythrocyte membranes as standards (Glut 1 8% of the total, data not shown). In this case also, the antibodies against the carboxyl-terminal peptide should react equally as strongly with the standard and the 3T3-Ll transporter, since the sequence of the carboxyl-terminal peptide of Glut 4 is the same in rat and mouse (4,5). The relative amounts of Glut 1 and 4 on days 8 through 12 after the initiation of differentiation were also determined by immunoblotting. Glut 1 was constant over this period, whereas Glut 4 increased moderately; its relative value was 1.0 (day 8), 1.4 (day 9), and 1.8 (days 10-12) (data not shown). Thus, the expected ratio of Glut 1 to Glut 4 in cells on days 10 to 12 is 215/(45 X 1.8/1.4) or 3.8. In another experiment of the type presented in Fig. 1, performed 6 months later with cells on day 11, the amounts of Glut 1 and 4 were found to be 150 and 50 ng/35-mm plate, respectively, and consequently the ratio was 3.0. Thus, although both transporters are present in 3T3-Ll adipocytes, Glut 1 is the predominant one. Because of the increase in Glut 4 between day 8 and 10, the studies described below were carried out with adipocytes at days lo-12 after differentiation.
The data in Fig. 1 also show that the amounts of Glut 1 and 4 detected in cells where transport was fully stimulated by insulin were the same as in basal cells, and that insulin treatment does not alter the electrophoretic mobility of either transporter.
In one experiment, the electrophoretic mobilities of Glut 1 and 4 on SDS-gel electrophoresis were directly compared by immunoblotting adjacent lanes from the same blot. The mobilities of the two transporters were the same and corresponded to a molecular weight of 55,000. Trandocation of the Glucose Transporters: Labeling of Surface Transporters-Holman and associates (26) have recently developed an improved membrane-impermeant, photoactivated affinity labeling reagent for glucose transporters. This compound, ATB-BMPA, is a bis(D-mannose) derivative in which the two mannoses are linked by a propyl bridge between the oxygens at carbons 4 and in which the diazirine, 4-( l-azi-2,2,2-trifluorethyl)benzamido, is linked to the middle carbon of the propyl bridge. Photolysis of this compound generates a highly reactive carbene through the loss of Nz from the diazirine functional group. 3T3-Ll adipocytes were labeled with [3H]ATB-BMPA, and the glucose transporters were isolated by immunoprecipitation and gel electrophoresis. Fig. 3 shows the results of a representative experiment. Both Glut 1 and 4 were labeled, and the extent of labeling of each was markedly increased by insulin treatment of the cells. The data in Table I summarizes the results from the experiment in Fig. 3 and four other experiments of this type. The -fold increases in labeling of Glut 1 and 4 due to insulin averaged 6.5 and 17.1, respectively; the corresponding value for stimulation of hexose transport in these experiments was 21. It should be noted that in two of the experiments the labeling of Glut 4 in basal cells was quite low, and consequently the -fold increase can only be taken as an estimate.
In a control experiment, designed to verify that only transporters at the cell surface were labeled by ATB-BMPA, basal and insulin-treated cells were labeled and homogenized, and then the plasma membranes were separated from the intracellular membranes by centrifugation at 16,000 X g for 20 min (see "Experimental Procedures"). Glut 1 and 4 in each fraction were isolated by immunoprecipitation and gel electrophoresis. No radioactivity was found in the Glut 1 or 4 in the intracellular membranes from basal cells, and the label in the transporters in this fraction from insulin-treated cells was 4% or less of that in transporters of the plasma membrane-containing fraction (data not shown). However, this intracellular membrane fraction contained a substantial portion of both transporters, as determined by immunoblotting. As we have previously reported, the percentages of total Glut 1 in this fraction from basal and insulin-treated cells are about 50 and 25%, respectively (10). The same distribution was found for Glut 4 (data not shown).
Interpretation of the increase in ATB-BMPA labeling in response to insulin as due to translocation involves the assumption that insulin does not alter the intrinsic reactivity of the transporter at the cell surface with this reagent (see Two Glucose Transporters in 3T3-Ll Adipocytes Cell surface Glut 1 and 4 in basal and insulin-treated 3T3-Ll adipocytes were photoafiinity labeled with ATB-BMPA, and the rates of hexose transport were measured on separate 35-mm plates, as described under "Experimental Procedures." In each experiment both procedures were performed in duplicate on the same day with the same plating of cells. The values for the counts per minute (cpm) in Glut 1 and 4 are the actual values for the duplicates from peaks of the type shown in Fig..  ' Ratio of the value for insulin-treated cells to that for basal cells. * In these experiments, the insulin stimulation was at 27 "C and the labeling was performed on cells chilled to 4 "C. ' Only a single determination for Glut 1 insulin in this experiment. d ND, not determined.
"Discussion"). As a partial test of this assumption, we have determined the affinity of the transporters for ATB-BMPA from its effect as a reversible inhibitor of transport. The results in Fig. 4 show that unphotolyzed ATB-BMPA inhibited 2-deoxyglucose transport in both basal and insulintreated 3T3-Ll adipocytes. The data is adequately fit by the assumption that inhibition is due to binding to a single site, with a dissociation constant of 530 pM in basal cells and 420 ~.LM in insulin-treated cells. Thus, insulin does not significantly alter the apparent affinity of the transporters for ATB-BMPA. This value for the dissociation constant is similar to the value of 300 PM found for the binding of ATB-BMPA to Glut 1 in human erythrocyte membranes at 20 "C (26). It might have been expected that Glut 1 and Glut 4 would have different affinities for ATB-BMPA, and therefore that the plots in Fig. 3 would be curved. Two possible explanations for the linear plots are that Glut 1 and 4 have very similar affinities for ATB-BMPA and/or that the contribution of one of the two to the transport rate is a relatively small fraction of the total (see "Discussion").
Translocation of the Glucose Transporters: Subcellular Fractionation-In order to examine the translocation of the glucose transporters by subcellular fractionation, we employed the procedure developed by Kono and associates (21,22) for rat adipocytes in which the less rapidly sedimenting intracellular membranes are separated from the plasma membranes on a sucrose gradient. The results of such an experiment are presented in Fig. 5. The fractions show a strong peak of Glut I and Glut 4 centered at fraction 10 (about 17% sucrose) and a broad shallow peak centered in fractions 446 (about 30-26% sucrose), which are intracellular and plasma membrane-containing fractions, respectively (22). Insulin treatment caused a portion of both transporters to shift from the intracellular to the plasma membrane fraction. The data in Fig. 4 were quantitated, and the relative amounts of Glut 1 and 4 in each fraction were expressed as a percent of the total. Insulin treatment led to increases in Glut 1 and 4 in fractions 4-6 that averaged 2.0-and 4.2-fold, respectively, with corresponding decreases in the transporters in fraction 10 to 44% and 64% of the basal value, respectively.
In response to insulin the distribution of Glut 1 between fractions 1-8 plus the pellet and fractions 9-13 shifted from 38:62 to 68:32; the corresponding shift for Glut 4 was from 15:85 to 38:62. A repetition of the entire experiment shown in Fig. 4 gave the same results.
Colocalization of Glut 1 and 4 in Vesicles-Previously, we have described a method for the isolation of the intracellular vesicles containing Glut 1 from the bulk of membranes in the 16,000 X g supernatant from 3T3-Ll adipocytes (10). This consists of immunoadsorption of these vesicles onto Staph A cells coated with antibodies against the carboxyl-terminal peptide of Glut 1. The procedure is quite selective, since only 10 pg of membrane (as protein) out of the 300 pg of membranes present in the 16,000 X g supernatant are adsorbed. In order to determine whether Glut 4 is located in the same vesicles as Glut 1, we immunoadsorbed the Glut l-containing vesicles and then immunoblotted both the unadsorbed membranes and the adsorbed ones for Glut 4 (as well as for Glut 1). Since it seemed possible that even if all the Glut 4 were located in vesicles containing Glut 1, these might only constitute a fraction of the Glut l-containing vesicles, the complementary experiment in which membranes were adsorbed with antibodfies against the carboxyl-terminal peptide of Glut 4 on Staph A cells was also performed.
The results of one experiment of this type are presented in Fig. 6. As expected, the antibodies against the carboxyl terminus of Glut 1 adsorbed the membranes containing Glut 1 (part A, lanes 5-7), whereas control antibodies did not (part A, lanes Z-4). Virtually all the Glut 4 was associated with the vesicles containing Glut 1 (part B,. Similarly, as expected, the antibodies against the carboxyl terminus of Glut 4 adsorbed the membranes containing Glut 4 (part B, lanes 8-lo), whereas control antibodies did not (part B, lanes 2-4). Most of the Glut 1 was associated with vesicles also containing Glut 4 (part A, lanes 8-10). In nine experiments of this type in which the adsorption of Glut 4 by Staph A cells coated with the antibodies against Glut 1 was determined, the extent of adsorption of Glut 4 averaged 95% of the total in the 16,000 x g supernatant.
In four experiments where the adsorption of Glut 1 by Staph A cells coated with the antibodies against Glut 4 was examined, the degree of adsorption of Glut 1 averaged 85% of the total.
The morphology of the membranes containing Glut 1 and 4 isolated by immunoadsorption on the Staph A cells was examined by electron microscopy. A typical field is shown in fields examined in the same way. Virtually no vesicles were found bound to Staph A cells coated with irrelevant antibodies (Fig. 7B).

DISCUSSION
Previous studies have shown that 3T3-Ll adipocytes contain both Glut 1 and 4 but did not determine the relative or absolute amounts of these two transporters (4)(5)(6)(7). The results herein show that Glut 1 is the predominant transporter.
Since there are about 2.1 x lo6 tells/35-mm plate, the amounts of Glut 1 and 4 are about 950,000 and 280,00O/cell. This ratio of the two transporters is in sharp contrast to that in rat adipocytes, where the ratio of Glut 1 to Glut 4, as determined by immunoblotting of total cell membranes, is about 0.08.' This difference between the predominant transporter type found in the cultured cells and the tissue cells is not limited to adipocytes. We have recently found that three muscle cell ' D. M. Calderhead, E. M. Gibbs, and G. E. Lienhard, unpublished results. lines express Glut 1, but not Glut 4, whereas Glut 4 is present in rat heart and skeletal muscle at three to four times the level of Glut 1 (24).
In an effort to measure the increase in the amounts of the two transporters at the cell surface in response to insulin, we employed the membrane-impermeant, photoactivated affinity labeling reagent, ATB-BMPA.
The use of this reagent involves the assumption that the intrinsic reactivity of each transporter with the reagent is not altered by insulin. The finding that insulin does not change the apparent affinity of the transporters for the reagent, as measured by its Ki value for inhibition of transport, supports this assumption, but does not prove it. Nevertheless, it seems likely that increases in ATB-BMPA labeling reflect increases in the amounts of the transporter at the cell surface. The results show that the -fold increase in Glut 4 (average 17-fold) was considerably greater than the -fold increase in Glut 1 (average 6-fold). These results are similar to those found upon labeling rat adipocytes with ATB-BMPA, where insulin increased the label in Glut 1 and 4 by 5-and 20-fold respectively.3 The redistribution of the two transporters from the intracellular to the plasma membrane-containing fractions of the sucrose gradient in response to insulin provides further evidence for the translocation of each transporter. The fact that the -fold increases in the plasma membrane-containing fractions are considerably less than those found by the ATB-BMPA labeling method (2-and 4-fold, respectively, for Glut 1 and 4) may be due to the contamination of these fractions with intracellular membranes containing the transporters. Immunoelectron microscopic studies suggest that the translocatable intracellular Glut 1 and 4 are located in the truns-Golgi network and possibly also the endosomes (see below). However, there is no known marker for the intracellular membranes containing the insulin-responsive transporters that can be used to assess the contamination of the plasma membrane fractions (10). In a recent study with rat adipocytes, the plasma membrane fraction was isolated from basal and insulin-treated cells and analyzed for its Glut 1 and 4 contents by immunoblotting (20). Insulin treatment increased Glut 1 by 1.6-fold and Glut 4 by &fold.
The question of whether the entire increase in the rate of hexose transport in response to insulin is due to translocation of Glut 1 and 4 to the plasma membrane involves the following considerations.
With the concentration of 2-deoxyglucose ([S]) much less than its half-saturation constant, the basal and insulin stimulated rates of transport (ub and ui) are   1-3, 5, 6, 8, 9), and the vesicles adsorbed to the Staph A cells were released into the same volume of SDS sample buffer (lanes 4, 7, and IO). Part A shows the immunoblot for Glut 1; part B is for Glut 4. The loads/lane as a percent of the total from a lo-cm plate were part A, lanes 3, 6, 9 (3%); other lanes (6%); part B, lanes 3, 6, 9 (1.5%); other lanes (3%).
Our data do not contain the information needed to calculate to expect that the amount of Glut 1 in the plasma membrane a value for the ratio, Ui/Ub, that can be compared with the is greater than that of Glut 4 in both the basal and insulin ratio of the experimentally determined values, which ranged state. First, on the assumption that the two transporters are from lo-to 36-fold in various platings of the cells. However, equally susceptible to labeling with ATB-BMPA, the ratio of one implication can be drawn. The average -fold increase in Glut 1 to 4 in the plasma membrane averaged 4.6 and 1.6 for surface  as assessed by ATB-BMPA labeling, the basal and insulin state, respectively (data in Table I). approached that of the -fold stimulation of transport (21-Second, on the basis of the distribution of the transporters fold), whereas the -fold increase in surface Glut 1 was consid-between intracellular and plasma membrane fractions on the erably less (6-fold Vesicles were immunoadsorbed from the 16,000 x g supernatant of basal 3T3-Ll adipocytes on Staph A cells coated either with antibodies against the carboxyl terminus of Glut 1 (part A) or with irrelevant rabbit IgG @art B). Adsorption was with 4 pg of antibodies on 2 ~1 of Staph A cells/ml supernatant. The Staph A pellets were processed for electron microscopy as described under "Experimental Procedures." translocation to account for the full insulin effect, the intrinsic activity of Glut 4 (k4/&) must be considerably larger than that of Glut 1 (h,/&).
95% of the Glut 4 in the intracellular membranes of the 16,000 X g supernatant was present in vesicles that also contained Glut 1; conversely, 85% of the Glut 1 was present in vesicles that contained Glut 4. Consequently, these transporters were largely colocalized, with only a small fraction of the Glut 1 in separate vesicles. Previously, we have reported that Glut l-containing vesicles in the 16,000 x g supernatant from basal cells on 35mm plate contain about 2.5 and 2.4 rg of protein and phospholipid, respectively (10). Since 50% of the Glut 1 and 4 are in the 16,000 x g supernatant, it can be calculated that Glut 1 and 4 are 3.7 and 1.0% of the protein in these vesicles. Also, the number of transporters of each type in the average 72-nm vesicle can be roughly estimated from the amounts of the transporters in the 16,000 x g supernatant and the number of transporter-containing vesicles. The latter can be estimated from the amount of vesicle phospholipid and the expected molecular weight of the phospholipid in a 72-nm vesicle (see Ref. 10 for details). The values are 24 Glut 1 and 7 Glut 4 molecules/vesicle.
Our finding on the colocalization of intracellular Glut 1 and 4 contrasts with the results of a study with rat adipocytes where it was found that less than 10% of the Glut 1 is the vesicles isolated with antibodies against Glut 4 (20). Thus, 3T3-Ll adipocytes are less effective than rat adipocytes at intracellular segregation of the two types of transporters. Possibly this is because of the larger amount of Glut 1 in 3T3-Ll adipocytes. It should be noted that since as few as one molecule of transporter/vesicle may be required for immunoadsorption of a vesicle, it is not possible to decide whether the transporter-containing vesicles in 3T3-Ll adipocytes are heterogeneous in the sense that some contain a much higher ratio of Glut 1 to 4 than others. The best approach to examine this issue will be to determine simultaneously the cellular distribution of both transporters in basal and insulin-treated cells by immunocytochemistry on cell sections, through detection with protein A-gold particles of two sizes. Previously, it has been demonstrated by immunoelectron microscopy that the insulin-responsive intracellular pool of Glut 1 is located in tubulovesicles in the trans-Golgi region of the 3T3-Ll adipocytes (8). A recent similar study with brown fat tissue has shown that the insulin-responsive Glut 4 is also located in the truns-Golgi area, as well as near large endocytic vacuoles, and preliminary results with basal 3T3-Ll adipocytes show concentration of Glut 4 in the trans-Golgi network." In conclusion, the two types of glucose transporters in 3T3-Ll adipocytes are qualitatively similar in their response to insulin and in their subcellular distribution. However, thefold increase in Glut 4 at the cell surface in response to insulin is several times larger than that of Glut 1. The subcellular fractionation data suggest that the basis of this difference may lie in a greater tendency of Glut 4 to be located intracellularly in the basal state. The two types of transporters differ in 37% of their amino acid sequences (5). In the future it may be possible to identify particular sequences accounting for differences in insulin responsiveness through expression of chimaeric transporters in 3T3-Ll adipocytes (25). Finally, we note that mRNA encoding a third isoform of glucose transporter (Glut 3) has been detected in human fat tissue (27). When antibodies to this form become available it will be of interest to determine whether Glut 3 is also expressed in 3T3-Ll adipocytes and, if so, whether it translocates in response to insulin.