Insulin-regulated Glucose Uptake in Rat Adipocytes Is Mediated by ...

6 downloads 0 Views 5MB Size Report
(LDM) of rat adipocytes (13). This antibody recognizes a form of glucose transporter of M, 43,000, unique to insulin- sensitive tissues, which shows a marked ...
THEJOURNALOF BIOLOGICAL CHEMISTRY

Vol. 264, No. 21, Issue of‘July 25, PP. 1235&12363,1989 Printed in U.S.A.

0 1989 by The American Society for Biochemistry and Molecular Biology, Inc.

Insulin-regulated Glucose Uptake in Rat Adipocytes Is Mediated by Two Transporter Isoforms Present inat Least Two Vesicle Populations* (Received for publication, April 13, 1989)

Antonio ZorzanoS, Wells Wilkinson, NatalioKotliar, GaliniThoidis, Brian E. Wadzinkskis, Arnold E. Ruohog, and PaulF. PilchB From the Departmentof Biochemistry, Boston University Schoolof Medicine, Boston, Massachusetts 02118 and the §Department of Pharmacology, University of Wisconsin Medical School, Madison, Wisconsin53706

We have recently described a monoclonal antibody location to the plasma membrane (1, 2). Examples of this (1F8) that recognizes a form of glucose transporter phenomenon include vasopressin-induced insertion of water unique to fat and muscle (James, D. E., Brown, R., channels in the bladder (2) and the insulin-dependent inNavarro, J., and Pilch, P. F. (1988)Nature 333,183- crease in plasma membrane glucose transporters in fat and 185),tissues that respond acutely to insulin by mark- muscle (1, 3-6). Insulin also causes rapid translocation to the edly increasing their glucose uptake. Here, we report cell surface of receptors for insulin-like growth factor I1 (IGFthat rat adipocytes possess two immunologically dis- 11)’ (7, 8) and transferrin (9, 10). Intracellular vesicles entinct glucose-transporters: one recognizedby 1F8, and riched in glucose transporters have been purified about 10one reactive with antibodies raised against the human erythrocyte glucose transporter. Immunoadsorption fold and partially characterized (11, 12), but it has not been of the proteins experiments indicate that these glucose transporters determined if the insulin-dependent movement or different vesicle described above results from the same reside in different vesicle populations and that both pools. Moreover, recent biochemical, immunological, and getransporter isoforms translocatefromintracellular netic evidence has provided compelling evidence that facilisites to the plasma membrane in response to insulin. tative glucose transport is carried out by a family of transThe insulin-regulatable transporter residesa unique in vesicle that comprises 3% or less of the low density porter isoforms whose expression can be highly tissue specific microsomes of fat cells and hasa limited protein com- (13-26). Insulin-responsivetissuesexpresstwoapparently position that does not include thebulk of another immunologically distinct transporters(25,26): one found only translocatable protein, the insulin-like growth factor in muscle andfat (13-16), andonecorrespondingtothe I1 receptor. Immunoprecipitation with 1F8 of micro- erythrocyte/Hep G2/rat brain species (17, 18).These studies somal glucose transporters photoaffinity labeled with raise the question as to whether the two transporter isoforms [3H]cytochalasin B brings down 90%of the label. Sim- reside in the same intracellular vesicles and to what quantiilarly, immunoprecipitation with 1F8of glucose trans- tative extent each contributes to the total insulin-mediated portersfrom insulin-stimulatedplasmamembranes increase in plasma membrane glucose transporters. photolabeled with 3-[’251]iodo-4-azidophenethylamWehave raisedamonoclonal antibody, 1F8, against a ido-7-O-succinyldeacetyl-forskolin, another transprotein from a purified vesicle preparation enriched in glucose porter-selective reagent, results in 75%of the labeled the low density microsomes transporter localized in the immunoprecipitate. Thus, transporters, derivingfrom insulin action involvesthe combined effect of translo- (LDM) of rat adipocytes (13). This antibody recognizes a cation from at least two vesicle pools each containing form of glucose transporter of M , 43,000, unique to insulinsensitive tissues, which shows a marked translocation from different glucose transporters. The 1F8-reactive in to insulin, and transporter comprises the majority of the total trans- the LDM to the plasma membrane response porter pool that is responsible for the insulin-induced we havecalled this protein the insulin-regulatable glucose increase in glucose transporter number. transporter (13). Recent evidence has confirmed that antibody 1F8 recognizes the product of a unique glucose transporter gene whose expression is limited to muscle and fat (14-16), the tissues that mediate the insulin-dependent clearance of Certain hormones promote the translocation of exocytic glucose from theblood (27). One of the studies cited (16) has vesicles containing membrane proteins from an intracellular demonstrated that the mRNA-deriving from cloned cDNA for the insulin-regulatable transporter, when microinjected * This work was supported in part by Grant DK-30425 from the into Xenopus oocytes, results in enhanced glucose uptake. United States Public Health Service and a grant from the Juvenile This clone (16), independently isolated from rat (14) and Diabetes Foundation, International (to P. F. P.). The costs of publi- human (15) sources, is recognized by monoclonal antibody cation of this article were defrayed in part by the payment of page recognized by 1F8 charges. This article must therefore be hereby marked “aduertise- 1F8, thus unequivocally linking the protein rnent” in accordance with 18 U.S.C. Section 1734 solely to indicate to glucose transport function. Here, we employ immunoadthis fact. $ Current address: Departamento de Bioquimica y Fisiologia, Facultad de Biologia, Universidad de Barcelona, Avda Diagonal 645, Barcelona 08071, Spain. ‘11 Recipient of Research Career Development Award DK-01352 from the United States Public Health Service. To whom correspondence should be addressed.

The abbreviations used are: IGF-11, insulin-like growth factor 11; LDM, low density microsomes; PM, plasma membraneb);PBS, phosphate-buffered saline; SDS, sodium dodecyl sulfate; PAGE,polyacrylamide gel electrophoresis; IAPS, 3-iodo-4-azidophenethylamido7-0-succinyldeacetyl; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid.

12358

Insulin-regulated Glucose Transport sorption protocols with monoclonal antibody 1F8 to partially characterize the natureof glucose transporter-containing vesicles and to assess the relative quantitative contribution of the two transporters presentin fat cells to theoverall insulindependent translocation to the cell surface. EXPERIMENTALPROCEDURES

12359

therefore varied from about 1,500 to 10,000 cpm/100 pg of membrane protein, also depending on the membranefraction employed. Photolabeling fat cell microsomes with tritiated cytochalasin B was performed as we described previously (13) with further details given in the legend to Fig. 3. Materials-Electrophoresis chemicals were purchased from BioRad. Other routine chemicals were obtained from Sigma. Iodinated goat anti-mouse antibody and protein A were obtained from Du PontNew England Nuclear. Nitrocellulose was obtained from Schleicher & Schuell, and Immobilon was obtained from Millipore. Collagenase for fat cell preparations was purchased from CooperBiomedical Inc. and prestained molecular weight markers from Bethesda Research Laboratories.

Isolation of Cells and Membrane Preparation-Rat adipocytes were obtained from the epididymal fat of male Sprague-Dawley rats (175200 g) by collagenase digestion essentially according to the Rodbell 13). Thefat cell procedure (28)as we describedpreviously(12, membrane fractionation procedure for the isolation of plasma membranes (PM) and LDM was performed according to the published RESULTS procedures of Simpson and co-workers (2, 29) which we have also employed previously for this purpose (12, 13). T o determine the relationship of the two glucose transImmunoadsorption Protocols-Antibody 1F8was purifiedover protein A-agarose and was coupled to cyanogen bromide-activated Seph- porter isoforms and their membrane environment, we used arose to 3 mg/ml of resin accordingto themanufacturer’s (Pharmacia antibody 1F8 covalently linked to agarose beads to immuLKB BiotechnologyInc.) instructions. Membrane proteins a t 1-2 noadsorb either detergent-solubilized fat cell LDM protein mg/ml(25-400 pg of total protein;see figure legends)were solubilized (Fig. 1A) or theintact-LDM (Fig. 1B). SDS-PAGEwas then in phosphate-buffered saline (PBS) containing 0.2% Triton X-100, performed on the protein(s) bound to agarose the bead pellets 0.2% SDS, 0.1% deoxycholate, and 0.1% bovine albumin, pH 7.4 and ( P ) and on the nonadsorbed supernatants (S) followed by were then incubatedwith 30 p1 of lF8-agarose overnighta t 4 “C. The immunoblotting.Using 1F8 astheblotting antibody, the agarose beads were collected by a 5-s spin in a microcentrifuge, the immunoadsorption protocol is quantitatively complete with supernatant was carefullycollected, andthebeads werewashed several times in the abovebuffer.Afterwashing, the beads were all the 1F8-reacting transporter being present in the pellet incubated in electrophoresis sample buffer (30) (100-200 p l ) , incu- (Fig. lA, left panel). However, when the anti-red cell transbated for 5 min a t 95 “C, and cooled, and transporter was obtained porter antibody (aGT) is used for immunoblotting, >90% of by microcentrifugation to yield the pellet fraction for electrophoresis. the signal is in the nonadsorbed supernatant fraction (Fig. The same immunoadsorptionprotocol as above was appliedto intact lA, right panel). The small signal in the agarose pellet (right microsomal membrane vesicles from fatand muscle which were cell antibody may recognize suspended in PBS in the absence of detergents. The vesicles that panel) indicates that the anti-red the 1F8-reactive species to a slight degree, and this is now bound to the immobilized antibody were washed in PBS, and the supernatant and bound material were collected for electrophoresis. Immunoblotting and EIectrophoresis Conditions-Electrophoresis was performed according to Laemmli (30) with various sample preparations and electrophoresis conditions as described in the figure legends. Proteins were transferred to nitrocellulose or Immobilon (see IF8 aGT figure legends) in buffer consisting of 20% methanol, 192 mM glycine, 25 mM Tris, pH 8.3. Following transfer, the filters were blocked with 5% nonfat dry milk in PBS for 1 h a t 37 “C andwere incubated with antibody for sametime at the sametemperature.Transfer was confirmed by Coomassie staining of the gel after the electroblot. Throughout, antibody1F8was purified by protein A chromatography before use and was used a t 5-10 pg/ml in 1% nonfat dry milk in PBS for immunoblotting. Detection of antibody-antigen complexes was effected with goat anti-mouse ‘251-antibody (Du Pont-New England Nuclear) and autoradiography. Rabbit antiserum raised against the purified humanerythrocyte glucose transporter(a gift of Dr. C. P S P S Carter-Su, University of Michigan) was used directly a t a 1:250 dilution and was incubated with transferred protein overnight a t room temperature. Detectionof the immunecomplex with the rabbit antibody was accomplished using lZ5I-protein A(Du Pont-New England Nuclear). Withbothantibodies,standard curves were constructed for immunoblotting by varying the amount of membrane IF8 aGT protein. These curves gave linear signals from 2 to 200 pg of total membrane protein. The anti-IGF-I1 receptor antibody was a gift of Dr. R. Baxter, Royal Prince Alfred Hospital, Sidney, Australia and was used in immunoblots ina fashion identical to thepolyclonal antiglucose transporter antibody. 0 .*. Transporter Photolabeling-Carrier-free 3-[’251]iodo-4-azidophen43 ethylamido-7-0-succinyldeacetyl-forskolin (IAPS-forskolin) was prepared as described (31). Fat cell membranes were isolated as noted above and suspended a t 1-2 mg/ml in 30 mM Hepes buffer, pH 7.6. One mg each was incubated for 30 min with 6 nM IAPS-forskolin in P S P S the presence and absence of 100 p~ cytochalasin B in 15-ml Corex centrifuge tubes kept on ice and in the dark. After this incubation, FIG.1. Rat adipocytes contain two immunologically distinct the membranes were diluted 10-fold in Hepes f cytochalasin B and glucose transporters that are present in different membrane spun a t 40,000 X g for 15 min. They were resuspended to theoriginal populations. Immunoadsorption of solubilized membrane (panel A, volume of Hepes and photolyzed a t 20 cm for 2 min with a focused P ) and membranevesicles (panel B, P)was accomplished as described 200-watt mercury lamp (Oriel Corp). 8-Mercaptoethanol was added under “ExperimentalProcedures.” Starting with 200 pg of membrane to 1% as a free radical scavenger, and the membranes were again protein, one-fourth of the supernatant ( S ) (50 pg) and 0.25 volume diluted 10-fold, washed, and then processed as described above for of the dispersed beads ( P fraction) were applied to each gel, and these immunoadsorption ordirectly run on SDS-PAGE. Sincewe used the were electrophoresed and immunoblotted, also as described under forskolin derivative a t various times after its synthesis, its specific “Experimental Procedures.” This experiment was performed three activity varied from experiment to experiment. Incorporation of label times on independent membrane preparationswith identical results.

A

”@

B



Insulin-regulated Glucose Transport

12360

understandable from the recent cDNA data predicting the sequence of the insulin-regulatable transporter(14-16). This sequence shows that there is considerable homology (65%) of the insulin-regulatable transporter with the erythroidspecies monoclonal 1F8 appearsto (17, 18). On theotherhand, recognize a carboxyl-terminal sequence unique to the insulinregulatable transporter (14). In Fig. 1B, the same immunoadsorption-immunoblotting protocol was employed using intact LDMs rather than soluble protein, and the same qualitative results were obtained. That is, 80% of the vesicles containing the 1F8-reactive protein were retained by the 1F8-agarose conjugate (Fig. l B , left panel), and >90% of the transporter reactivewith the anti-red cell antibodyisnotin vesicles containing the fat/muscle type transporter (Fig. l B , right panel). Thus, thetwo fat cell glucose transporter typesreside in two (or more) intracellular compartments. To determine if both glucose transporters translocate in response to insulin, we performed immunoblotting as shown in Fig. 2. Fat cell membranes were fractionated from untreated and insulin-treated cells and were divided in two equal portions for SDS-PAGE and immunoblotting with the two antibodies described above. In the top panel, a n 8-fold insulindependent increase in the 1F8-reactive transporter observed is in the plasma membrane (170 cpm basal to 1370 cpm with insulin). With aGT, a 1.6-fold increase in the anti-red cellreactive material is seen in the PM due to insulin (1437 cpm basal to2314 with insulin). Themicrosomal fractions show a corresponding decreasein transporters due to insulin measas ured by both antibodies (1F8, from 2633 to 1827 cpm; aGT, from 1824 to 687 cpm). It should be noted that the relative signalsfor thetwoantiseraarenot comparable, possibly because of different antibody-antigen affinities and because the signal generated useddifferent iodinateprobes of different specific activities. In five independent preparationsof fat cell membranes in which we compared the total immunoblotting signal in the fractions from the basal state with the total

PMILDM

Ab

..mm

1F8

+ +

Insulin

FIG. 2. The two adipocyte transporter isoforms both translocate from the LDM to the PM in response to insulin. Adipocyte membrane fractions wereprepared as described (1, 29) from insulin-treated (2)or untreated cells. Equal portions of each membrane fraction (25 p g ) were applied to both gels, and these were run on 10% acrylamide gels. Protein from the gels was transferred to nitrocellulose and immunoblotted with 1F8 (top panel) or with the anti-red cell antibody ( A b ) (nCT,bottom panel). This experiment has been performed >10 times with qualitatively similar results and 5 times where all fractions were quantitated (see “Results”).

signal from the insulin-stimulated state, the sums differed by 3-15% indicatingthatimmunoblottingis a quantitatively reliable method for assessing translocation. Because antibodies 1F8 andaGT are recognizing different transporter isoforms (Fig. l), the datafrom Fig. 2 provide a clear demonstration that insulin-activatedglucose transport involves the translocation of two transporter species that reside in different intracellularvesicles. It is noteworthy that a very the glucose transporter recognized by 1F8 is present to small degree inthe plasma membrane of resting fatcells (Fig. 2, top; see also Refs. 13 and 14) and isdramatically increased in plasma membranes from insulin-treated cells. In contrast, the glucose transporter recognized by the anti-erythrocyte antibody gives a strong signal in the basal PM which is not dramatically increased by insulin, a result seen by others using similar antibodies against the red cell transporter (24, 31-34; see below). Thus, the data from Figs. 1 and 2 demonstrate that the total increase in plasma membrane glucose transporters would be due to translocationof two transporter typesfrom a t least two intracellular pools. The apparent discrepancy between the insulin-dependentincrease in the numberof cytochalasin B-binding sites (maximum, 7-fold) (34) uersus the approximately 2-fold increase in plasma membrane immunoreactivity using anti-red cell antibody seen in a number of studies (25, 32-35) can be resolved if the 1F8-reactiveglucose transporter comprises most of the total transporter present in fat cell microsomes. T o quantitate the relative number of 1F8-reactive transporters, we immunoprecipitated fat cell low density microsomes after they had been photoaffinity labeled with [3H] cytochalasin B. The K d for cytochalasin binding to the transporter(s) of fat cells (29) is indistinguishable from its & for binding to the erythrocyte transporter (36,37). Photocoupling of cytochalasin B to the transporter is thought to occur via (38), photoactivation of an aromaticresidue in the transporter probably in the 10th and 11th transmembrane segments (39), in which these residues are identical in the sequence of the insulin-regulatable and erythrocyte/Hep G2/rat brainspecies (14-18). Thus, we assume thatphotocoupling of cytochalasin B to these two transporter isoforms occurs with equal efficiency and isa valid method for quantitatively comparing the two species in fat cells. Shown in Fig. 3 are the immunoprecipitate and supernatant, following electrophoresis, wherethe labeled transporter in eachfraction wasquantitated, and90% of thetransporteris immunoprecipitablewith 1F8 under conditions in which immunoadsorption of this transporter form was complete as indicated by immunoblotting (see Fig 1A). Since the translocation of transporters from the LDM to the PM accountsfor essentially all the transporter redistribution due to insulin(above and Ref. 1)and most (90%)of the LDM transporters areimmunoprecipitable with 1F8, the data in Fig. 3 support the notion that theinsulin-regulatable transporter is the quantitatively predominant form responsible for the observed increase in total cell surface transporters. Further evidence for this notion is provided in Fig. 4. A newly developed probe for glucose transporters is the photoaffinity label IAPS-forskolin (31).We photolabeled fat cell membrane fractionswith this reagent and rangels on the photolabeled transporter before and after immunoadsorption of the labeled protein with 1F8-linked agarose as shown in Fig. 4. As reported previously, the photolabeling is specific for the glucose transporter because it is inhibited by cytochalasin B (Fig. 4A, lane 1 uersus lune 2). In the experimentshown in Fig. 4, B and C, the PM fractions from insulin-treated and basal adipocytes were each labeled with IAPS-forskolin, and

Insulin-regulated Glucose Transport 28

43

500 3 68 V

V

V

h

-!

v

R

300

-0 0

.:

200

0 c

>r 0

I

0..

i\

m

5

Pellet . . 0 Supernalant

0-•

400

100

72 0 10

14

18

22

26

3430

Slice Number FIG. 3. Antibody 1F8 immunoadsorbs90%of adipocyte glucose transporters photoaffinitylabeled with [‘H]cytochalasin B. Adipocyte LDMs were photolabeled with 1 p~ [‘H]cytochalasin B just as we described previously (13).The membranes were solubilized and adsorbed onto 1F8-conjugated agarose as described in Fig. 1. Thesupernatant was precipitatedin 10% trichloroacetic acid, resuspended in sample buffer, and both fractions were subjected to SDS-PAGE. Following electrophoresis, the gel was sliced,solubilized, and countedexactly as we described previously (13).As a control, we verified that the immunoprecipitation of the 1F8-reactive transporter was complete as determined by immunoblotting as in Fig. 1A. This experiment was done threetimes.

A

B

+

+

C

-

PM

?

12361

four separate experimentsemploying three independent membrane preparations, 72 f 5% of the forskolin label was immunoprecipitable. Incontrast, only 28%of the label was immunoprecipitable from the plasma membrane of resting cells (Fig. 4C)in whichmore incorporation of label into nontransporter species occurred due to the relative scarcity of totaltransporterinthis fraction. Thus, the forskolinlabeling protocols confirm the postulate that 1F8 identifies the predominant transporter isoform present on the fat cell surface following insulin exposure (see “Discussion”). The protein yield typically obtained in immunoadsorption protocols (n = 6 ) such as those shown in Fig. 1B for the transporter-containing vesicles was 3.0% of the LDM, thus suggesting that these vesicles might be pure or highly enriched. We performed silver staining of vesicles immunoadsorbed from insulin-treated and untreated adipocyte LDMs as shown in Fig. 5. A small number of bands is seen which specifically diminish (wedges) in response to insulin as exM, of about 45,000 which pected,including onewithan migrates in the region of the glucose transporter. However, the purified erythrocyte glucose transporter stainsvery poorly with protein reagents, and we are not sure if the stained case, in proteinin Fig. 5 is actuallytransporter.Inany contrast to the 1F8-adsorbed vesicles, the vesicle population that was not adsorbed contains numerous proteins, and few, if any, of these show obvious insulin regulation. Thus, the vesicles containing theinsulin-regulatable glucose transporter contain a limited number of proteins, and the composition of these differs markedly from the bulk of the LDM proteins. As noted in theIntroduction, cellular insulinexposure enhances thecell surface expression of other membrane proteins including the IGF-I1receptor (7, 8). If the vesicles

B(-

-t=

-+

INS

204-

97+

-

+

P

S

P

S

66-

FIG.4. Photocoupling of [‘”I]IAPS-forskolin to rat adipocyte glucose transporters. The conditions for the IAPS-forskolin (6 nM) photolabeling of plasma membranes from fat cells are described under “Experimental Procedures.” In panel A, the photolabeled membranes from insulin-stimulated cells incubated in thepresence of 100 p~ cytochalasin B (lane 2) or not (lane 2 ) were electrophoresed directly. In panels B and C, the plasma membranes from insulin-treated ( E ) or untreated (C)cells were solubilized and incubated with 1F8-linked agarosebeads asdescribed under “Experimental Procedures.” Following centrifugation and washing of the beadantibody-antigen complex, a portion corresponding to an equal fraction of the beads ( P ) and the supernatant ( S ) was solubilized for SDS-PAGE run on a 10% acrylamide gel. Detection of transporter was by autoradiography, and transporters were quantitated as described under “Results.”

43-

Adsorbed

NonAdsorbed

FIG.5. The insulin-regulatable glucose transporter resides in a unique intracellular vesicle. Membrane vesicles (400 pg of

protein) from the LDM of untreated and insulin-treated adipocytes were incubated with immobilized antibody 1F8 as in Fig. 1R. After the incubation, the entire adsorbed fraction was applied to a 7.5% acrylamide gel (no reducing agent in the sample buffer) along with 10 pg of the nonadsorbed fraction. The gel was then silver stained as these were solubilized and immunoadsorbed with 1F8-linked described (50).The lane on the left ( E )indicates a silver stain of 1F8agarose beads.An equal portion of the immunoprecipitate( P ) linked beads notexposed to membrane. The position of the molecular and the supernatant (S) was electrophoresed, and following weight markers is shown on the left, and the wedges indicate specifiautoradiography (depicted) thegels were cut and counted. In cally stained bands.

Insulin-regulated Glucose Transport

12362

aIGF-II R

200

Insulin

-+P

+ S

FIG.6. Most of the adipocyte IGF-I1 receptors (aZGF-ZZR) are not inthe same vesicles as the insulin-regulatableglucose transporter. Fat cell low density microsomes (200 p g ) from adipocytes exposed to insulin (t)or not (-) were isolated from the cells as described under“Experimental Procedures.” These microsomes were exposed to antibody as in Fig. l B , and equal portions of the immunoadsorbed pellets (P)(one-fourth) and supernatants ( S ) (50 pg) were electrophoresed on a 5-12% acrylamide gradient gel. IGF-I1 receptors were detected by immunoblotting as described under “Experimental Procedures.” This experiment was performed twice. Another 50-pg portion of the supernatant and one-fourth of the beads were blotted with antibody 1F8, which verified that immunoadsorption of 280% of the insulin-regulatable transporter hadoccurred (data not shown).

containingtheinsulin-regulatabletransporterrepresent a unique vesicle type, then one might expect that these would not be enriched in IGF-I1 receptors, and the data in Fig. 6 show that these receptors arein fact found predominantly in the supernatant fraction following vesicle adsorption with antibody 1F8. We show that the IGF-I1 receptor content is depletedin the adsorbed and nonadsorbed LDM fractions following cellular insulin exposure, and others have shown that this movement is exclusively to the PM (7,8). DISCUSSION

red cell antibody (11)and based on membrane fractionation assayed by cytochalasin B binding (12). Given the data presented here, we believe that thesevesicles described previously do not correspond to those that mediate most of the insulininduced increase in cell surface glucose transporters. Rather, it seems likely thatthere may be anotherpopulation of vesicles which mediates the movement of IGF-I1 receptors and the erythroidglucose transporter. The vesicles that contain these two proteins overlap on sucrose gradients and on agarose gels, but theydo not coincide (12,40), thus indicating the existence of vesicle heterogeneity. Neither the IGF-I1 receptor not the erythrocyte/Hep G2/rat brain transporter is present to an appreciable degree in the vesicles adsorbed by 1F8 (Figs. 1B and6). A possible explanation for these observations is that proteins such as the IGF-I1 receptor, transferrin receptor, and red cell/Hep G2/rat brain transporter may exist in a heterogeneous vesicle population whose movement to the cell surface is part of the generalphenomenon of insulin-inducedexocytosisseen infat cells (41). They all exhibit a similar 2-fold increasedexpression in isolated plasma membranesdue to insulin,a much smaller effect than observed for the insulin-regulatable species (Refs. 13 and 14 and Fig. 2). Recently, two reports appeared in which the Hep G2 (42) and rabbit brain (43) glucose transporter were transfected into 3T3-Ll andChinese hamster ovary cells, respectively. A modest insulin-dependent increase in transport (43) or in translocation of the transfected transporter (42) was observed which was markedly smaller than thatobserved for the insulin-regulatable transporter isoform in fatcells. Again, it seems likely that the insulin-dependent regulation of these transfected transporters is more characteristic of the response described above for the erythrocyte/Hep G2lrat brain carrier rather than of that seen for the 1F8-immunoreactive species. However, the signalingmechanism by which any of these vesicles respond to insulin and translocate could be similar in all cases. Four lines of evidence support the notion that the insulinregulatable transporter recognized by 1F8 is the predominant transporter isoform responsible for the insulin-induced increase in cell surface glucose carriers. First, the translocation of the 1F8-reactive isoform from the LDM to the PM correlates tightly with the translocationof cytochalasin B binding (13), whereas translocation of the erythrocyte isoform does not (35). Second, labelingof the microsomal transporters with labeled cytochalasin B followed by immunoprecipitation with 1F8 brings down 90% of the labeled transporter (Fig. 3), whereas the converse experiment employing an anti-erythrocyte antibody brings down 3% of the label (26). Third, immunoprecipitation of plasma membrane transporters from insulin-activated adipocytes labeled with IAPS-forskolin brings down about 75% of the label (Fig. 4). Fourth, induction of diabetes in ratsby streptozotocin injection results in a loss of 280% of the 1F8 immunoreactive transporter*andits cognatemRNA3 under conditions that markedlydiminish only the insulin-activated componentof cellular glucose uptake and have little, if any, effect on the basal component (44). It is the erythrocyte/Hep G2/rat brain transporter isoform that comprises thedominant species in the plasma membrane of resting cells (Figs. 2 and 4). The question remains as to whether the translocation process accounts for all of the observed increase in glucose transport induced by

The datawe show here indicate that the glucose transporter that comprises most of the total transporterexpressed in fat cells (Figs. 3 and 4) and thatis translocatable in response to insulin (Fig. 2) resides in an apparently unique membrane vesicle population (Figs. 5 and 6). These unique transportercontaining vesicles are highly enriched and comprise 3% of the LDM protein as determined by quantitating the total protein yield inimmunoadsorptionexperimentssuchas shown in Figs. 1B and 5. The vesicles have a limited protein composition and contain very little IGF-I1 receptor (Fig. 6). Two facts keep us from concluding that these vesicles are homogeneous aspresently isolated. First,thereare some proteins that appear with equal intensity before and after insulin exposure (Fig. 5 , adsorbed uersus nonadsorbed). Second, some of these vesicles could derive from the biosynthetic pathway for membrane proteins, i.e. the Golgi or the endoplasmic reticulum. In any case, assuming that there is one cytochalasin B-binding site/transporter, that the number of these cytochalasin-binding sites is approximately 70 pmol/ mg of protein (62 pmol/mg, our data, Ref. 12; and 81 pmol/ mg, Ref. 29) and an M , of 55,000 (14-16), then the immunoaffinity-purified vesicles of Fig. 5 contain a minimum of 13% of their protein as transporter. Sucha concentration of asingle functionalprotein is highly unusualthusfurther supporting our contention that the vesicles containing the insulin-regulatable glucose transporterrepresent aunique cellular organelle or compartment. We (12) andothers (11) havedescribed previously the * Berger, J., Biswas, C., Vicario, P. P., Strout, H. V., Saperstein, purification of vesicles enriched inglucose transporters of the R., and Pilch, P. F. (1989) Nature, in press. erythroid type. About a 10-fold purification was achieved Sivitz, W. I., Desautel, S. L., Kayano, T., Bell, G. I., and Pessin, based on immunoaffinity protocols using a polyclonal anti- J. E. (1989) Nature, in press.

Insulin-regulated Glucose Transport

12363

2. Wade, J. B. (1986) Annu. Reu. Physiol. 4 8 , 213-223

insulin by increasing the number of transporters, or if other 3. Cushman, S. W., and Wardzala, L. J. (1980) J. Biol. Chem. 2 5 5 , 47584762 factors are involved. 4. Suzuki, K., and Kono, T. (1980) Proc. Natl. Acad. Sci. U. S. A. 7 7 , 2542The most careful estimates of the increase in cell surface 2545 5. Oka, Y., and Czech, M. P. (1984) J. Biol. Chem. 259,8125-8133 transporters report a 7-fold increase in cytochalasin B-bind6. Blok, J., Gibbs, E. M., Lienhard, G. E., Slot, J. W., and Geuze, H. J. J. ing sites (35). Since both transporterisoforms bind cytochal(1988) J. Cell Biol. 106.69-76 7. Wardzala, L. J., Simpson,I. A,, Rechler, M. M., and Cushman,S. W. (1984) asin B with indistinguishable affinity (29, 36, 37), this estiJ . Biol. Chem. 259,8378-8383 mate is likely to be the most accurate oneavailable. Our own 8. Oka, Y., Mottola, C., Oppenheimer, C. L., and Czech, M. P. (1984) Proc. Natl. A c d . Sci. U. S. A. 81,4028-4032 estimates of translocation based on forskolin labeling (also 9. Davis,R.J.,Corvera, S., and Czech, M. P. (1986) J. Biol. Chem. 2 6 1 , assuming that this compound labels the two fat cell trans8708-8711 porters with equal efficiency) varies from 4- to 6-fold (data 10. Tanner, L. I., and Lienhard, G. E. (1987) J. Biol. Chem 262,8975-8980 11. Biber, J. W., and Lienhard, G. E. (1986) J. Biol. Chem. 261,16180-16184 not shown). Taken together, these data do not account for 12. James, D. E., Lederman, L., and Pilch, P. F. (1987) J. Biol. Chem. 2 6 2 , 11817-11824 the 20-30-fold increase that can be seen in cellular glucose D. E., Brown, R., Navarro, J., and Pilch, P. F. (1988) Nature 3 3 3 , uptake due to insulin. Our data (Figs. 2 and 4) also suggest 13. James, 183-185 14. James, D. E., Strube, M., and Mueckler, M. (1989) Nature 338,83-87 that the presence of appreciable transporterinthebasal H., Kayano, T., Buse, J. B., Edwards, Y., Pilch, P. F., Bell, G. plasma membrane may not be an artifact of contamination 15. Fukumoto, I., and Seino, S. (1989) J. Biol. Chem. 264,7776-7779 16. Birnbaum, M. J. (1989) Cell 57,305-315 since it is the erythrocyte form that appears to present in the M., Caruso, C., Baldwin, S. A,, Panico, M., Blemch, I., Morris, basal state toa much greater degree than the insulin-regulat- 17. Mueckler, H. R., Allard, W. J., Lienhard, G. E., and Lodish, H. F. (1985) Science 229,941-945 able form. This would appear to be the normal state of the 18. Birnhaum, M. J., Haspel, H. C., and Rosen, 0. M. (1986) Proc. Natl. Acad. basal plasma membrane unless contamination of the basal Sci. U. S. A. 83,5784-5788 19. Taylor, L. P., and Holman, G.D. (1981) Biochim. Biophys. Acta 642,325plasma membrane with internal membranes containing the 335 erythrocyte glucose transporter isoformoccurs to a much 20. Axelrod, J. D., and Pilch, P. F. (1983) Biochemistry 22,2222-2227 larger extent than contamination with membranes enriched 21. Horuk, R., Matthaei, S., Olefsky, J. M., Baly, D. L., Cushman, S. W., and Simpson, I. A. (1985) J . Biol. Chem. 261,1823-1828 in the insulin-regulatablespecies. 22. Fukumoto, H., Seino,S., Imura, S., Seino, Y., Eddy, R. L., Fukushima, Y., Byers, M. G., Shows, Y. B., and Bell, G. I. (1988) Proc. Natl. Acad. Sci. Thus, itseems that translocationmay not be the only means U. S. A. 85,5434-5438 by which insulin stimulatesglucose transport, a concept that 23. Thorens, B., Sarkar, H. K., Kabach, H. R., and Lodish, H. F. (1988) Cell 55,281-290 has been suggested by others on thebasis of the discrepancy Kayano, T., Fukumoto, H., Eddy, R. L., Fan, Y.-S., Byers, M. G., Shows, 24. between apparent transporter number and transport rate (45) T. B., and Bell, G. I. (1988) J. Biol. Chem. 263,15245-15248 25. Wang, C. (1987) J. Biol. Chem. 262,15689-15695 as well as on the basisof data showing diminished transport Y., Asano, T., Sbibasaki, Y., Kasuga, M., Kanazawa, Y., and Takaku, rates following exposure of adipocytes to isoproterenol with 26. Oka, F. (1988) J. Biol. Chem. 263,13432-13439 no change in transporter distribution (46,47). It hasbeen also 27 , James, D. E., Jenkins, A. B., and Kraegen, E. W. (1985) Am. J. Physiol. 248. E567-F,574 demonstrated that the rate of glucose transport activity re- 28 Rodbei,M. (1964)J. Biol. Chem. 239,375-380 29 Simpson, I. A,, Yver, D. R., Hissin, P. J., Wardzala, L. J., Karnielli, E., constituted from the total membrane fraction from insulinSalans, L. B., and Cushman, S. W. (1983) Biochim. Biophys. Acta 7 6 3 , treated cells exceedsthat reconstitutedfrom resting cells (48). 393-407 30. Laemmli, U. K. (1970) Nature 227,680-685 Three explanations for these data come to mind. First, the B. E., Shanahan, M. F., and Ruoho, A. E. (1987) J. Biol. Chem. insulin-regulatable transporter isoform may have an intrin- 31. Wadzinski, 262,17683-17689 sically higher transport activity since it is only 65% homolo- 32. Wheeler, T. J., Simpson, I. A., Sogin, D. C., Hinkle, P. C., and Cushman, S. W. (1982) Biochem. Biophys. Res. Comm. 105,89-95 gous in primary sequence with the erythrocyte/Hep G2/rat 33. Shanahan, M. F., Olson, S. A., Weber, M. J., Lienhard, G. E., and Gorga, J. C. (1982) Biochern. Biophys. Res. Comm. 107,38-43 brain isoform (14-16). Second,theinteraction of the cell H. C., Rosenfeld, M. G., and Rosen, 0. M. (1988) J. Biol. Chem. surface-expressed insulin-regulatable transporter with intra- 34. Haspel, 263.398-404 ,----35. Jo;