GLUT-4 Phosphorylation and Its Intrinsic Activity MECHANISM OF Ca2+-INDUCED INHIBITION OF INSULIN-STIMULATED GLUCOSE TRANSPORT*

In this study, we examined the influence of high levels of cytosolic calcium on phosphorylation status and function of GLUT-4 in isolated rat adipocytes. Intracellular calcium was elevated by exposing adipocytes to either extracellular ATP (1.6 mM) or thapsi- gargin (100 nM). Both agents increased cytosolic calcium 2-3-fold. While basal glucose uptake was unaf- fected, both ATP and thapsigargin reduced insulin-stimulated glucose transport by 40-70% ( p e 0.05). Neither ATP nor thapsigargin affected GLUT-4 content or its translocation from the low density micro- somes to the plasma membrane (PM). In contrast, GLUT-4 immunoprecipitated from the PM of adipocytes exposed to either ATP or thapsigargin was phos- phorylated to a greater extent than the GLUT-4 isolated from control cells. ATP and thapsigargin also abolished insulin-stimulated dephosphorylation of GLUT-4. At the same time, GLUT-4 intrinsic activity was significantly reduced in adipocytes with high levels of cytosolic calcium ( p c 0.05). Preincubation of adipocytes with cAMP antagonist, RpcAMP M), and calcium channel blocker, nitrendipine (30 p ~ ) , improved the ability of insulin to dephosphorylate GLUT-4 and restored insulin-stimulated GLUT-4 intrinsic

thase and . Sustained levels of [Ca2+Ii in the insulin target cells inhibit phosphoserine phosphatase-1 activity, at least in part, by the phosphorylation and activation of the inhibitor 1 (11). This effect of high [Ca2+Ii appears to be mediated largely via a CAMP-dependent pathway (11). Inhibition of phosphoserine phosphatase-1 activity results in an impaired dephosphorylation of several insulin-sensitive enzymes and proteins, such as glycogen synthase and 10). Previous studies from this and other laboratories (12-15) as well as the accompanying article (16) indicate that phosphorylation of GLUT-4 decreases its intrinsic activity.
In this study, we attempted to dissect further the mechanism of high [Ca2+Ii-induced insulin resistance. We used the following two different approaches to elevate [Ca2+Ii in normal rat adipocytes: extracellular ATP and thapsigargin, a potent inhibitor of Ca2+-ATPase (17). We evaluated the effects of these agents on glucose transport, phosphorylation, and distribution of GLUT-4, as well as on GLUT-4 intrinsic activity.

Materials
Porcine insulin was a gift from Eli Lilly Co.
In some experiments, a cAMP antagonist, RpcAMP M ) , or a calcium channel blocker, nitrendlpine (30 PM), were added 10 min prior to the addition of ATP or thapsigargin. At the end of the incubation period, adipocytes were processed for the measurements of [Ca"], CAMP, basal and insulin-stimulated 2-deoxyglucose transport, quantitation of GLUT-4 content, and phosphorylation status in the plasma membrane (PM) and low density microsomal (LDM) fractions as well as GLUT-4 intrinsic activity in the plasma membranes.
Measurement of cAMP Leuel.-Intracellular cAMP was extracted from control and experimental adipocytes with ice-cold 5% trichloroacetic acid. Trichloroacetic acid was removed by successive extractions with ether. The levels were then measured by non-acetylation CAMP. radioimmunoassay (19) with a specific polyclonal antibody against Measurement of 2-Deoxyglucose Transport-The procedure was described in our previous publications (3,15). In brief, the cells were incubated with and without 25 ng/ml insulin for 30 min a t 37 "C. Glucose transport was initiated by the addition of 50 p1 of [3H]2deoxyglucose/ml (in 2 mM 2-deoxyglucose). At the end of 3 min of incubation, 300-pl aliquots of cell suspension were added to the Eppendorf tubes containing silicone oil, and the cells were separated from the medium by rapid centrifugation. The cell pellet was cut off and counted in a liquid scintillation counter.
Incubation of Adipocytes with f2Pli"Adipocytes prepared from 30-40 control rats were resuspended in low Pi medium containing 145 mM NaCl, 5.4 mM KCl, 1.4 mM CaC12, 1.4 mM MgS04, 30 mg/ml bovine serum albumin, and 10 mM HEPES, pH 7.4, and incubated with [3ZP]orthophosphate (0.3 mCi/ml) for 2-3 h at 37 "C to achieve steady state labeling (20). The 32P-labeled cells were divided into several aliquots and incubated in duplicate with buffer alone (controls), ATP (1.6 mM), ATP + RpcAMP M), thapsigargin (100 n M ) , or thapsigargin + RpcAMP M) for 30 min followed by the treatment with and without insulin (25 ng/ml) for an additional 30 min. At the end of the incubation, the cells were centrifuged a t 800 X g'for 15 s, and the medium was aspirated. The cells were rinsed at 37 "C in HES buffer containing phosphatase and protease inhibitors (10, 13) and homogenized a t 0 "C in HES buffer using a glass homogenizer and a Teflon pestle (six to eight strokes) at 1500 revolutions/minute. HES buffer contained 10 mM HEPES, 255 p M sucrose, 1 mM EDTA, 50 mM sodium fluoride, 200 p~ sodium orthovanadate, 50 mM sodium pyrophosphate, 10 pg/ml leupeptin, 10 pg/ ml aprotinin, 1 mM benzamidine, 1 mM phenylmethylsulfonyl fluothe cells were homogenized in the above buffer containing 1% SDS ride, 10 pg/ml soya trypsin inhibitor, pH 7.4. In some experiments, t o ensure rapid denaturation of proteins and complete extraction of the 32P-labeled GLUT-4. Subcellular Fractionation of 32P-Labeled Adipocytes-The cell homogenates were used to isolate LDM and PM fractions according to the method of James et al. (12). The rapid fractionation procedure reduced the possibility of changes occurring in the phosphorylation state of the GLUT-4. PM and LDM fractions were resuspended in homogenization buffer and assayed for protein content using bicinchroninic acid (21). 5"Nucleotidase activity was measured as the marker of P M enrichment. ATP or thapsigargin treatment did not alter the protein content or distribution of the 5'-nucleotidase activity.
Immunoprecipitation of 32P-Labeled GLUT-4 and Electrophoretic Analysis-GLUT-4 was immunoprecipitated using the monoclonal antibody IF8 (12). SDS-solubilized PM, LDM, and cell homogenate5 (100 pl containing 50-100 pg of protein) were added to 900 pl of HES buffer containing 1% Triton X-100, 100 mM NaC1, protease, and phosphatase inhibitors. Samples were incubated for 30 min a t 22 "C and centrifuged for 5 min a t 13,000 X g. Supernatants were used for immunoprecipitation with IF8. The monoclonal antibody was first immobilized by incubation a t 22 "C with goat anti-mouse IgG (binding capacity 0.4 mg IgG/ml packed beads) coupled to agarose beads (1 pg of anti-mouse IgG, 1 pg of IF8), and resuspended in HES buffer containing 0.1% bovine serum albumin. After 60 min of incubation with constant shaking, the beads were pelleted by centrifugation and washed three times with HES buffer containing 1% Triton X-100, 100 mM NaCl, protease, and phosphatase inhibitors. Beads (-100 pl) were added to solubilized samples and incubation was continued for 60 min at 22 "C with constant shaking. The samples were centrifuged at 13,000 X g for 10 s, and the pellets containing immune complexes were washed with HES containing 1% Triton X-100. SDS sample buffer (30-50 p1) was added to elute glucose transporters. Samples were incubated for 10 min a t 37 "C, and electrophoresis was performed in duplicate using 10% SDS-polyacrylamide resolving gel (12, 13). After electrophoresis, one gel was dried and exposed to Kodak XAR-5 film a t -70 "C, while the second gel was blotted and probed with polyclonal GLUT-4 antibody to quantitate GLUT-4 content as described below. 32P was quantitated by optical density scanning of autoradiograms, and the results were compared with the "cut-and-count" technique. Low molecular weight standards from Bio-Rad were always run with the sample to estimate the molecular weight of Immunoblotting was performed to estimate the relative amounts of GLUT-4 in control and experimental preparations. Proteins were electrophoretically transferred from SDS gels to nitrocellulose sheets or immunolite membranes supplied by Bio-Rad according to the manufacturer's instructions. To identify GLUT-4, the membranes were incubated with the polyclonal antibody R820 (12, 13). Antibody binding was detected by chemiluminescence technique.
Estimation of Glucose Transporter Intrinsic Actiuity-Control, ATP-, or thapsigargin-treated adipocytes were divided into 2 aliquots and incubated with and without insulin (25 ng/ml), followed by the subcellular fractionation as described in the accompanying article (16). P M fractions were reconstituted in transport buffer and used for the assay of intrinsic activity using ['4C]glucose as described by Weber et al. (22).
Protein Assay-The amount of protein was determined by the method of Bradford (23) or by bicinchroninic acid (21).
Statistics-The results are presented as mean f S.E. of four to six individual experiments performed in duplicate. Paired or unpaired Student's t test was used to compare the mean values among the groups as indicated.

RESULTS
In previous studies, we employed either sequential depolarization of cells with 40 mM K' or treatment with parathyroid hormone (PTH) (20 ng/ml) to maintain elevations in [Ca2+]; above 200 nM (3,11,12). To prove the specificity of [Ca2+]; effects and to investigate the mechanisms of its action, it was important to induce and maintain [Ca2+]; a t elevated levels by other mechanisms. In this study, we used extracellular ATP or thapsigargin to induce sustained elevations in [Ca2+];. ATP elevated [Ca2+]; by increasing mobilization from intracellular pools as well as by increasing [Ca"]; influx through the action of plasma membrane (Ca2+-Mg2+)-ATPase that acts as a Ca2+ pump (24, 25). Thapsigargin increases [Ca2+]; by inhibiting the intracellular Ca2'-ATPase (17, 26).
Incubation of isolated rat adipocytes with either ATP (1.6 mM) or thapsigargin (100 nM) increased [Ca2+]; 2-3-fold (Table I). These increases were not affected by either a calcium channel blocker, nitrendipine (30 PM), or a cAMP antagonist, RpcAMP ( M). Exposure of adipocytes to the extracellular ATP also resulted in a %fold increase in the intracellular cAMP levels (Table I). In contrast, thapsigargin did not affect cellular cAMP levels.
The presence of 1.6 mM ATP did not alter the basal glucose uptake. In contrast, ATP reduced insulin-stimulated glucose transport by 70% (Table 11). The IDb0 for ATP was 0.4 mM. Despite the inherent instability of ATP and its likely susceptibility to tissue ATPases, this inhibition was effective after TARI.E II / < / / w l o/A7'1' and !hnpsignrgin on in.su/in-slimuIn!cd glucose tronsporl Adiporvtes were exposed to ATP (1.6 mM) or thapsigargin (a = 25 nM: 1) = 100 nM) in the presence or in the almnce of either nitrentlipine o r I1prAMP. Insulin (25 ng/ml) WAS added and the cells were incuhatetl lor 30 min followed hv glucose transport assav. Details are given under "Experimental I'rocedures." Results are the mean t S.E. of six experiments performed in duplirate. * p < 0.05 ucrsus control. ** p < 0.05 twrsus ATP. p < 0.05 ocrsus t hapsigargin (25 nM). $5 p < 0.05 Lvr.sus thapsigargin (100 nM).

Experiment
Insulin-stimulated glucose transport Similar results were ohtained in three different experiments. exposure of cells for only a few minutes and persisted after removal of A T P from the medium (data not shown). Nitrendipine was ineffective in preventing the ATP-induced inhibition of the insulin-stimulated glucose uptake, whereas RpcAMP restored insulin's effect by 60%. Neither nitrendipine nor RpcAMP affected the basal glucose uptake (not shown).
Treatment of adipocytes with thapsigargin also resuked in a 38% decrease in insulin-stimulated glucose uptake ( Table   11). The half-maximal effect of thapsigargin was observed a t 10 nM. Although t.hapsigargin did not increase cellular CAMP cont.ent (Table I), the presence of RpcAMP completely prevented the t,hapsigargin-induced inhibition of glucose transport (Table 11).
We then examined t,he effects of ATP and thapsigargin on GLUT-4 cellular distribution, phosphorylation, and its intrinsic act,ivit.y. Neither ATP nor thapsigargin affected GLUT-4 content or it.s translocation from the LDM to the PM fraction ( Fig. 1 ) . In contrast, studies wit,h '"P-labeled subcellular fractions demonstrated that GLUT-4 immunoprecipit,at.ed from t h e P M of adipocytes exposed to either ATP (Fig. 2 A ) or thapsigargin (Fig. 2R) was phosphorylated to a great,er extent than the GLUT-4 isolated from control cells. This increase  Table 111. ATP and thapsigargin increased GLUT-4 specific activity (expressed as the amount of :'"P/unit of protein determined hy Western blotting) by 30-407;) in the basal and insulin-stimulated states. RpcAMP partially prevented ATP and thapsigargin's effect on GLUT-4 phosphorylation ( Table 111). J'hosphorvlation of the LDM GLUT-4 was also increased by either A'J'J' or thapsigargin.
Since our concurrent studies indicated that enhanced phosphorylation of GLUT-4 diminished its intrinsic activity (16). we examined GLUT-4 intrinsic activity in adipocytes treated with either ATP or thapsigargin. Adipocytes were incul~atetl with either ATP or thapsigargin for 30 min followed hy a treatment with and without insulin (2.5 ng/ml). At the end of the incubation time, the PM vesicles were isolated from the control and experimental cells for the measurements of I)-[l"C]glucose uptake. The ability of GLUT-4 to transport ["C] glucose into the I" vesicles was taken as a measure of its intrinsic activity. ATP and thapsigargin did not alter ( ; T A C T -4 intrinsic activity in the absence of insulin, but hoth agents blocked the insulin-stimulated increases in the intrinsic activity of GLUT-4 (Figs. 3 and 4). Preincuhation of adipocytes with RpcAMP (10P M ) or with RpcAMl' and nitrendipine (30 p~) restored insulin-stimulated intrinsic activity of GLUT-4 to normal (Fig. 4).

DISCI~SSION
The major finding of this study is that two dissimilar agents that raise [Ca"], hv independent mechanisms inhihitd insulin-stimulated glucose transport in normal rat adipocytes, increased phosphorylation of (;LU'J'-4, and inhibited ( ; l , l T -4 intrinsic activity. There was no effect of increasing [Ca'.], on GLUT-4 intracellular localization or translocation t o the plasma membrane in response to insulin. These ohservntions lend further support to our previous reports that high levels of [Ca"], render insulin target cells resistant to insulin (2. 11. 12).
These metabolic effects appear t o be specific :In(! causally related to high levels of [Ca"A],. In the present studies, two agents which increased [Ca"], by different mechanisms induced similar metabolic ahnorrnalities.
In previous experi-Cell Calcium, GLUT-4 Phosphorylation, and Intrinsic Activity 3355   in control, ATP-, and thapsigargin-treated adipocytes Autoradiograms of 3ZP-labeled GLUT-4 and Western blots of GLUT-4 protein were scanned for optical density and areas beneath the peaks corresponding to GLUT-4 were determined. Specific activities in arbitrary units were calculated by dividing the values of 32P peaks by those of protein peaks. To compare results from different experiments, specific activity of control LDM was assigned a value of 1, and the rest of the data was expressed relative to control LDM specific activity. Results are the mean f S.E. of four independent experiments. * p < 0.05 versus no insulin. ** p < 0.05 uersus corresponding control. *** p < 0.05 uersus ATP alone. 3 p < 0.05 uersus cont.ro1. Sf p < 0.05 uersus thapsigargin alone.
GLUT-4 specific activity (arbitrary units) Adipocytes were incubated without (control), with ATP (1.6 mM), or thapsigargin (100 nM) for 30 min then exposed to insulin (25 ng/ml) for an additional 30 min. PM vesicles were isolated and 2-deoxyglucose transport measured by pulsing 30 pg of PM protein with I 4 Clabeled 2-deoxyglucose for 3 s as described under "Experimental Procedures." Results represent the mean & S.E. of 6-14 independent experiments. i, < 0.001 uersus control basal glucose uptake. "p < 0.05 uersus insulin-stimulated controls. RpcAMP was effective in preventing thapsigargin's action, while levels of cAMP remained normal, suggests that calcium's action upon GLUT-4 phosphorylation and activity is likely to be mediated via a CAMP-dependent pathway. The precise mechanism of the convergent effects of Ca2+ and cAMP is incompletely understood. A similar convergence was described recently by Sheng et al. (27) and Ginty et al. (28). These authors demonstrated that Ca2+ phosphorylates transcription factor CREB (CAMP response element-binding protein) at the same site as does CAMP, but presumably via different kinases. Alternatively, high levels of [Ca2++li may activate a number of Ca2+-dependent kinases such as calciumcalmodulin-dependent kinase (23), calcium-phospholipid-dependent kinase (29), etc. Partial restoration of insulin action by the cAMP antagonist may suggest an involvement of the Ca2+-dependent kinases that are not inhibited by RpcAMP. Further work is needed to identify the mechanism of crosstalk between Ca2+ and cAMP in their effects on phosphoserine phosphatase-1 phosphorylation and activity.
Sustained levels of [Ca*+]i not only increased GLUT-4 phosphorylation in the basal state but also prevented the insulin-induced dephosphorylation of GLUT-4 occurring under normal circumstances (10, 15). Insulin promotes rapid translocation of GLUT-4 from the LDM pool to the PM (10, 12, 14, 30-32) and dephosphorylates GLUT-4 in both compartments (10, 13, 14). It is possible that insulin induces translocation of GLUT-4 and thereby simply dilutes the pool of the phosphorylated PM transporter. This possibility appears to be unlikely, however, since insulin dephosphorylates both PM and LDM . It appears that phosphorylation of GLUT-4 with either calcium or cAMP does not affect the ability of insulin to translocate glucose transporters to the plasma membrane (1). However, phosphorylated GLUT-4 has a lower intrinsic activity than the GLUT-4 dephosphorylated normally by insulin. The influence of GLUT-4 phosphorylation on its reinternalization (13,14) remains to be determined. The mechanism whereby high levels of [Ca2+Ii inhibit dephosphorylation of GLUT-4 remains incompletely understood. We have previously shown that high levels of [Ca2+]i inhibit the overall phosphoserine phosphatase activity in adipocytes and skeletal muscle (11). This inhibition can occur in one of the two possible ways. In one sequence, high levels of [Ca'+], stimulate phosphorylation and activation of the inhibitor 1 (11). This endogenous inhibitor in its active form binds the catalytic subunit of phosphoserine phosphatase-1 and thus reduces its enzymatic activity (33,34). The effect of calcium on inhibitor 1 appears to be mediated via CAMPdependent pathways (11). Alternatively, high levels of [Ca2+li may promote direct phosphorylation of phosphoserine phosphatase-1 at the CAMP-sensitive sites (sites 1, 2, and 3). Phosphorylation of site 2 of the regulatory subunit of phosphoserine phosphatase-1 leads to dissociation of the catalytic subunit from the regulatory subunit of this enzyme (35). The enzyme loses its activity when the catalytic subunit released into the cytosol is bound by the activated inhibitor 1 (33, 34). This mechanism of phosphoserine phosphatase-1 inactivation by CAMP has been demonstrated in skeletal muscle in the elegant experiments of Hubbard and Cohen (33,34) and Dent et al. (35). Whether or not a similar mechanism is operational in adipocytes is not known. Of note, the role of [Caztli in the phosphorylation of the phosphoserine phosphatase-1 regulatory subunit has not yet been demonstrated.
The present study confirms our previous observations that phosphorylation of GLUT-4 does not interfere with its translocation from the intracellular pool to the plasma membrane (1). In contrast, phosphorylated GLUT-4 appears to exhibit reduced intrinsic activity. If the intrinsic activity of GLUT-4 is increased by insulin-stimulated dephosphorylation (via phosphorylation and activation of phosphoserine phosphatase-1), then an inhibitory effect of high [Ca2+]i on phosphoserine phosphatase-1 may be responsible for the diminished GLUT-4 intrinsic activity in adipocytes treated with ATP or thapsigargin.
In summary, high levels of [Ca2+Ii in insulin target cells stimulate phosphorylation of GLUT-4 either directly or indirectly by inhibiting its dephosphorylation. Increased phosphorylation of GLUT-4 interferes with its intrinsic activity. This inhibition of GLUT-4 intrinsic activity results in diminution of insulin-stimulated glucose transport in these cells. Thus, we conclude that high levels of [Ca2++Ii induce insulin resistance at the post-receptor steps of insulin action by interfering with normal dephosphorylation of insulin-sensitive substrates.