Effect of phenylarsine oxide on insulin-dependent protein phosphorylation and glucose transport in 3T3-L1 adipocytes.

We have reported previously that phenylarsine oxide (PAO) blocks insulin-stimulated glucose transport in 3T3-L1 adipocytes (Frost, S. C., and Lane, M. D. (1985) J. Biol. Chem. 260, 2646-2652). As shown in the present study, the locus of inhibition is post-receptor. Insulin stimulated the extent of receptor autophosphorylation in solution and in the intact cell by approximately 4-fold. PAO had no effect on this activity. Using reduced and carboxamidomethylated lysozyme as a substrate for the tyrosine-specific receptor, insulin stimulated the rate of receptor kinase-catalyzed substrate phosphorylation by 2-fold; PAO had no effect on this stimulation. However, the insulin-stimulated, serine-specific phosphorylation of two endogenous phosphoproteins (pp24 and pp240) in the intact cell was blocked by 25 microM PAO. These complementary in situ and in vitro studies demonstrate that the inhibition by PAO must be distal to the insulin receptor's protein tyrosine kinase activity.


Effect of Phenylarsine Oxide on Insulin-dependent Protein Phosphorylation and Glucose Transport in 3T3-
PA0 had no effect on this activity. Using reduced and carboxamidomethylated lysozyme as a substrate for the tyrosine-specific receptor, insulin stimulated the rate of receptor kinase-catalyzed substrate phosphorylation by 2-fold; PA0 had no effect on this stimulation. However, the insulin-stimulated, serine-specific phosphorylation of two endogenous phosphoproteins (pp24 and pp240) in the intact cell was blocked by 25 I .~M PAO. These complementary in situ and in vitro studies demonstrate that the inhibition by PA0 must be distal to the insulin receptor's protein tyrosine kinase activity.
The insulin receptor is a heterodimer composed of two identical insulin-binding subunits (the a-subunit, 135 kDa) convalently attached by disulfide bonds to two identical kinase subunits (the p-subunit, 92 kDa). It is known that insulin stimulates not only the extent of autophosphorylation of the @-subunit but also the catalytic activity toward artificial substrates (1)(2)(3)(4)(5). Thus the receptor, itself, could provide the initial steps in a cascade leading eventually to activation of the glucose transport system. These steps would include insulin binding, autophosphorylation of the receptor, and receptor-mediated phosphorylation of an endogenous substrate.
To hypothesize that autophosphorylation signals an insulin-sensitive system, a temporal connection must exist between the receptor and activated system. Using the stimulation of glucose transport, we have shown previously that, in the presence of insulin, the @-subunit of the insulin receptor in 3T3-Ll adipocytes reached maximal autophosphorylation on tyrosine prior to the initiation of activation of 3-O-methylglucose uptake in situ (2). Furthermore, when cells were exposed to insulin, the receptor isolated from these cells exhibited increased catalytic activity toward an artificial sub-strate compared to receptor isolated from control cells (2). Although clearly not proof that receptor phosphorylation signals transport activation, two points were established. First, the essential temporal connection between receptor phosphorylation and transport activation is now documented in an insulin-sensitive intact cell. Second, these results show that the receptor itself is indeed activated by insulin, as demonstrated by the enhanced catalytic activity.
In addition to the above correlations, studies from Morgan et al. (6) and Ellis et al. (7) further support a role for the tyrosine-specific kinase activity. Using a monoclonal antibody which inhibits insulin receptor kinase activity, Roth's group (6) have shown that microinjection of this antibody into Xenopus oocytes blocks the ability of insulin to stimulate oocyte maturation. Studies from Rutter's group (7) used the technique of site-directed mutagenesis of the human insulin receptor. Human placental insulin receptor cDNA was transfected into Chinese hamster ovary cells and then assayed for insulin-stimulated glucose transport. Although the enhancement was small compared to 3T3-Ll adipocytes, a 2-fold stimulation was noted. When a mutant receptor cDNA was transfected, one with point mutations in which tyrosine 1162 or 1163 was replaced with phenylalanine, insulin could no longer stimulate the rate of glucose transport. These data together strongly support a functional role for the receptor tyrosine kinase in insulin action. If receptor kinase plays a role in signal transduction, the next question to ask is whether the receptor signals the transporter by direct interaction or through intermediate steps. Previous studies with 3T3-Ll adipocytes led to the discovery that a trivalent arsenical, phenylarsine oxide (PAO),' blocked insulin-stimulated glucose transport but not basal transport (8). We suggested at that time that PA0 was affecting a post-binding site because neither insulin association to cells at 37 "C nor insulin binding in solution were altered by the presence of PAO. In addition, the K,,, of the transporter for glucose was not affected by PAO, suggesting that the arsenical was not interfering with sugar transport, directly. We report, here, that PA0 does not alter receptor phosphorylation or catalytic activity of the receptor kinase toward an artificial substrate. It does, however, perturb the ability of selective endogenous phosphoproteins to exhibit insulin sensitivity.

Effect of Phenylarsine
Oxide on Insulin-dependent Events 3-0-Methylglucose Uptake"ST3-Ll fibroblasts were grown and differentiated on 15-mm plastic coverslips set in 24-well Lindbro culture plates (5 X 10" cells/disc). Before experimentation, the cells were washed three times with 1.0 ml of Krebs-Ringer phosphate buffer, 37 "C. The coverslips were transferred to a vertical holder, made according to the design of Norton and Munck (9). Cells were preincubated with or without 20 p~ P A 0 and/or 1 p~ insulin for 10 min. The holder was then lowered into a radioactive solution containing 50 p M 3-0-[methyl-'4C]glucose (4.8 mCi/mmol) with hormone and/or inhibitor additions when appropriate for 1 min. The coverslips were then passed through a series of three washes of phosphatebuffered saline (25 "C) containing 0.3 mM phloretin to terminate sugar transport. With each new solution that the cells were exposed to, the holder was quickly raised and lowered five times, resulting in rapid equilibration of the buffer over the surface of the cells (9). The coverslips were transferred to 15-ml glass scintillation vials containing 1.0 ml of 0.1% sodium dodecyl sulfate. After the experiment was completed, 10 ml of scintillation fluor was added and vials were shaken until the solution cleared. Results were corrected for trapped intercellular radioactivity (5%) using 20 p~ cytochalasin B, which competitively blocks glucose uptake. With 1.0-min uptake intervals, 3-0-methylglucose had just reached 100% of the equilibrium value in the presence of insulin but only 25% in the absence of hormone (8). The data are expressed as the average of quadruplicate determinations.
Labeling with ~2P]0rthophosphate in Cell Culture-The labeling procedure for determining phosphorylated insulin receptor and phosphorylated cellular proteins has been presented in detail elsewhere (2). Briefly, monolayers containing approximately 5 X lofi cells were washedseveral times with phosphate-free Dulbecco's modified Eagle's medium. The monolayers were then overlayed with this same medium containing carrier-free :"Pi (500 pCi/ml of medium). After 2 h in a 37 "C. COP incubator, the nucleotide pool was fully equilibrated although the protein pool was not. For this reason, addition of effectors was made such that the final exposure to ?'Pi was identical with each experimental point. Identification of the @-Subunit of the Insulin Receptor and Insulinsensitive Phosphoproteins in Situ-When the experimental protocol was complete, monolayers were washed three times with ice-cold phosphate-buffered saline over a 6-s period. Cells were extracted by addition of 2 ml of lysis buffer containing 25 mM Tris, 192 mM glycine, 25 mM EDTA, 5 mM EGTA, pH 8.3, with 0.1% SDS. The plates were washed with 3 ml of this same buffer only containing 1.7% Triton X-100. The combined lysates were homogenized by hand in a glass homogenizer with a Teflon pestle. The homogenates were centrifuged a t 4 "C for 45 min a t 40,000 X g in a Beckman 70.1 fixed angle rotor. Four ml was removed from under the lipid layer for exposure to 50 pl of WGA-Sepharose (10). An aliquot of the remaining 1 ml was run on an SDS-polyacrylamide gel for the identification of insulin-sensitive phosphoproteins (see below).
After mixing for 90 min at room temperature, the WGA-Sepharose was collected by centrifugation and washed five times with 1.0 ml of the lysis buffer containing 1.0% Triton X-100 and 0.04% SDS. Glycoproteins were eluted with the addition of Laemmli sample buffer containing 1.5 M N-acetylglucosamine and 2% SDS for 30 min. A two-dimensional system was used to separate the receptor from other glycoproteins, as described in detail earlier (10). Briefly, the entire sample was overlayed on a nonreducing tube gel (5% total acrylamide separating gel). The second dimension was performed on a reducing slab gel (8% total acrylamide separating gel). The second dimension was dried on Whatman No. 3MM paper, and the @-subunit of the insulin receptor was visualized by autoradiography after an 18-h exposure. The radioactive spot was cut from the gel, rehydrated with 0.1 ml of water, and counted in 10 ml of Fluorosol after heating for 2 h a t 60 "C.
Identification of cellular phosphoproteins was performed in one of two ways. Fifty pl of extract was mixed with SDS sample buffer and run on a 12.5% SDS reducing gel. Molecular weight standards included myosin, M, 200,000; phosphorylase @, M, 97,000 bovine albumin, M, 68,000; ovalbumin, M, 45,000; n-chymotrypsin, M, 25,000; lysozyme, M. 14,400. The gel was dried on Whatman No. 3MM paper and phosphoproteins identified by autoradiography. Additionally, the segments were cut, rehydrated, and counted as above.
In Vitro Phosphorylation of the Insulin Receptor and Artificial Substrate-Insulin receptor from 3T3-Ll adipocytes was partially purified through the WGA-Sepharose chromatography step according to Kohanski and Lane (10). The receptor (approximately 0.1-0.5 pmol of insulin-binding sites) was incubated in the presence or absence of 1 p~ insulin, 100 p~ pol-ypeptide substrate (reduced and carboxamidomethylated lysozyme), and 100 p~ phenylarsine oxide in a buffer containing 20 p~ [y-'"PIATP (70 dpm/fmol), 50 mM Hepes, 5 mM Mn(CH:,C0P)2 at pH 6.9 containing 0.1% Triton X-100 in a total reaction volume of 220 pl. Reactions were initiated with the addition of labeled ATP, and an aliquot was removed and quenched after the appropriate time with the addition of Laemmli sample buffer concentrated 3-fold. Separation of the autophosphorylated insulin receptor @-subunit (92 kDa) from RCAM-lysozyme (14.7 kDa) was achieved by SDS-PAGE using a linear gradient from 8 to 15% total acrylamide (2.5% bisacrylamide). The labeled proteins were identified by autoradiography, and radioactivity was determined by liquid scintillation counting. Phosphoamino Acid Analysis-Excised gel segments were soaked for 2 days in 10 ml of water to remove excess salts. The segments were homogenized in 1 ml of 0.1 M ammonium bicarbonate, pH 8.0, containing 50 pg of trypsin. Digestion proceeded a t 37 "C for 20 h, followed by a second addition of trypsin and incubation for an additional 8 h. The residue was removed and the supernatant lyophilized three times. Hydrolysis with 6 N HCI was performed for 100 min a t 108 'C in glass tubes evacuated to 50 millitorrs. High-voltage electrophoresis was performed by the procedure of Cooper et al. (1 1 ). Phosphoamino acid standards were visualized by ninhydrin and labeled phosphoamino acids by autoradiography.
MateriaLs-3T3-Ll fibroblasts were the gift of Dr. Carl Grunfeld (Dept. of Medicine, University of California, San Francisco). 3-0-[methyl-'4C]Glucose, ["'Plorthophosphate, and [y-R2P]ATP were purchased from Amersham Corp.; wheat germ agglutinin was from Behring Diagnostic; trypsin was from Worthington; phenylarsine oxide was from Aldrich. All other chemicals were of the highest quality available.

RESULTS AND DISCUSSION
To define the steps in a metabolic cascade, an inhibitor of that process is particularly useful, as has been demonstrated multiple times in the various metabolic pathways. We have demonstrated that a trivalent arsenical, phenylarsine oxide Insulin receptor was partially purified from Triton X-100-extracted 3T3-Ll membranes by affinity chromatography on WGA-Sepharose. The effect of P A 0 on in vitro autophosphorylation was tested by incubating the receptor preparation with 100 p~ P A 0 for 10 min. Insulin, when present at 1 p~, was added with [y-"PIATP a t time 0; this stimulation is equivalent to preincubation of the receptor with insulin (10) and more closely mimics the in situ situation of preincubating cells with P A 0 before exposure to insulin. Reactions were terminated a t 5-min intervals by the addition of 3-fold concentrated Laemmli sample buffer containing 20 mM dithiothreitol and 20 mM EDTA. The "P-labeled @-subunit of the insulin receptor was resolved on a single one-dimensional polyacrylamide gel.  and insulin. In addition. the gel segments at 9 2 kDa were excised and counted (see text for quantitation).

P M ) without affecting insulin association under physiological conditions (8).
T h e discovery that the insulin receptor undergoes autophosphorvlation on the &subunit (1) has provided an experimental approach to investigating the mechanism of signal transmission. With mounting evidence that receptor kinase plays a part in insulin action, it was important for us to determine whether the site of action of P A 0 was at the tyrosine kinase of the receptor. We have therefore studied this activity both in uitro and in situ in the absence and presence of PAO. Effect of P A 0 on the 'I:vrosine Kinase Actiuitv of the Insulin Receptor in Vitro-The in vitro protein kinase assay allows us to examine the cat.alvtic activity of the kinase, under defined conditions, both toward itself and artificial subst,rates in lieu of an endogenous substrate. The results are shown in Figs. 1 and 2. Insulin, at 1 PM, stimulated the incorporation ot'[y-'"P] from ATP into the &subunit ofthe soluble receptor, seen as an increase in the rate as well as in the extent of autophosphorvlation (Fig. 1). PAO, at. 100 p~, had no effect non-reduclng 3T:LLl adiporvtes equilihated with "IJ, for 2 h were exposed or not to 25 p M PA0 for 1 0 min followed hv incuhation in t,he presence or ahsence of 1 p M insulin for an additional 10 min. After harvesting, glycoproteins were collected hv chromatography with WGA-Sepharose. eluted with N-acetvlglucosamine. and separated by two-dimensional SDS-polvacrylamide gel electrophoresis (2). on either the basal or insulin-stimulated autophosphorylation rate. These studies were performed in the absence ot' added substrate so that the only protein phosphorylated in the assav mix was the $subunit, itself, shown in the inset of Fig. 1.
Since autophosphorylation did not, appear to be altered by PAO, we tested the effect of PA0 on the catalytic activity of the receptor toward an artificial substrate. RCAM-lysozyme has been shown to be a n excellent substrate for the insulin receptor with a K,, in the low micromolar range and phosphorylated strictly on tvrosine (4). As shown in Fig. 2, A and H , in the presence of partially purified insulin receptor, insulin pretreatment stimulated the rate of incorporation of:"P from [y-:"P]ATP into RCAM-lysozyme by about. 2-fold. PAO. at 100 p~, did not affect the rate of phosphorylation by the receptor kinase either in t,he basal or insulin-stimulated state. Note that autophosphorylation of the receptor preincubated with insulin (in the presence or absence of PAO) was stimulated 2-fold during the first 2 min of incubation with RCAMlysozyme ( Fig. 2 A )  This fractional autophosphorylation correlat.es well with the 2-fold increase in the rate of subst.rate phosphorylation. Together, these results suggest t,hat the protein kinase activity ofthe receptor in solution, he it toward itself or toward other tyrosine-containing substrates, is not altered by PAO.' Effect of P A 0 o n Insulin-serzsitioe Phosphoproteins in Intact 3T3-Ll Adipocytes-The second approach for defining the '' I t was necessary to establish that, P A 0 was accessible to the receptor kinase in this assav system, Le. that PA0 was not inactivated hv incorporation into triton micelles. Under receptor kinase assav conditions (minus the receptor protein), PA0 was treated with 2,3dimercaptopropanol, a vicinal dithiol reagent, and the free sulfhvdryls in dimercaptopropanol were detected hv the Ellman procedure (12). Dimercaptopropanol reacted stoichiometrically with PA0 under these condit,ions, demonstrating that PA0 was ahle to react, with vicinal sulfhydryls. The conclusion stands, then, that PA0 does not hlock the tvrosine kinase activitv of the insulin receptor in solution.

,Y//rc! o/phmvlarsinr oxidr on in.su/in-.srnsitivc phosphoproteins
Intact :3T3-IAl monolayers were pre-equilibrated with ['T'jorthophosphate for 2 h and then exposed or not to 2.5 P M P A 0 for 10 min. Monolayers were then stimulated or not with 1 W M insulin for 10 min. Following detergent extraction of the monolayers, the :"I'-laheled proteins were resolved hv nonreducing/reducing two-dimensional SIX-PACE of the W(;A-Sepharose-adsorbed proteins (insulin receptor @-subunit and pp240; Fig. 3) or hy onedimensional reducing SDS-PAGE of non-adsorbed proteins (pp18 and pp24: Fig. 4). Quantitation of radiolahel in the rehydrated gel segments was by liquid scintillation counting; data were normalized to total radioactivity per 4 X 10" cells according to the aliquot size as a fraction of the total extract. The predominant phosphoamino acids were phosphotyrosine (d-subunit) and phosphoserine (pp240, pp24, and pp18). Details of3-0-methylglucose uptake 9875 are given in the text. An aliquot of extract not adsorbed to WGA-Sepharose from Fig. 3 was analyzed by SDS-polyacrylamide gel electrophoresis. Thus. when present, 26 W M P A 0 was inruhated with cells for 10 min after ".P equilihration followed by incubation with 1 P M insulin for 10 min. In A , the proteins were visualized hy autoradiography and the gel slices rehydrated and counted for radioactivity (quantitation in text). Shown in A is the autoradiogram of extracts from control cells, insulin-treatedcells, and PA0 and insulintreated cells. To determine the time frame over which pp24 and pp18 were phosphorylated, cells were labeled for 2 h fnllowed hy addition of 1 p M insulin for varying times ( H ) . After washing in phosphate-buffered saline, the cells were extracted as descrihed under "Experimental Procedures" and an aliquot was analyzed by SDS-polyacrylamide gel electrophoresis. The dosed circkra represent the radioactivity from the higher molecular weight protein (pp24), the closrd trinnp/es represent the radioactivity from the lower molecular weight protein (pp18). The inset shows the radioactive hands of pp24 and pp18 as visualized by autoradiography. for 10 min. Following a IO-min incubation with or without insulin, the cells were extracted and assayed for " T , incorporated into the /hubunit. The autoradiogram of a two-dimensional SDS-PAGE is presented in Fig. 3 and the data are summarized in Table I. As described earlier (2).
insulin increased the extent of phosphorylated receptor by about 4-fold (Fig. 3, panels A and R ) specifically on t,yrosine. The only form of the receptor observed in the nonreducing first dimension exhibited an apparent M, of 350,000 from which the 95-kDa If-subunit was generated in the reducing second dimension. In two separate experiments, P A 0 had no effect on the incorporation of label into the @subunit in either the basal or insulin-stimulated receptor (PAO-treated " A s shown by Frost and Lane (8). 25 W M P A 0 is the minimal concent.ration required to elicit maximal inhihition of insulin-stimulated glucose transport. ~~ basal receptor not. shown; insulin-stimulat.ed, or PAO-and insulin-treated receptor shown in Fig. 3, panels R and C, respectively). Note also t.hat PA0 did not affect the mobility of the receptor in the first dimension and thus does not appear to alter the structural integrity ofthe holoreceptor.
Although an endogenous substrate (tyrosine-specific) for the insulin receptor has not yet been conclusively identified, several candidate proteins have been detected (13, 14). In addition, various proteins have been noted whose insulinstimulated phosphorylation occurs on serine (15-1 7). We have identified several such proteins in 3T3-Ll adipocytes, as shown in Figs. 3 and 4 and the results summarized in Table   I. One such protein is a glycoprotein (pp240) observed in the WGA-Sepharose eluate along with the insulin receptor. Shown in Fig. 3 is the migration of this phosphoprotein, in relation to the @-subunit of the receptor, in the second dimension of a reducing SDS-PAGE (see "Experimental Procedures"). Insulin stimulat.ed the extent of phosphorylation exclusively on serine residues (data not shown) by about 7 -Effect of Phenylarsine Oxide on Insulin-dependent Events fold (Fig. 3, panels A and B ) . Pretreatment of the cells with 25 ~L M PA0 blocked this insulin-stimulated event (Fig. 3,

panel C).
A smaller molecular weight protein (pp24) has also been identified which exhibits insulin-sensitivity ( Fig. 4 and Table  I). This protein is not adsorbed to WGA-Sepharose. As shown, insulin stimulated the incorporation of 32P by about 2-fold, specifically on serine. PA0 inhibited the insulin-stimulated phosphorylation of this protein as well (Fig. 44). Interestingly, the insulin-stimulated phosphorylation of this protein is timedependent (tlh = 1.7 min) (Fig. @), occurring in the "window" between receptor phosphorylation (txh = 8 s) and transport activation (tlh = 2.5 min) (2). This suggests that pp24 may be an intermediate in an insulin-sensitive cascade. A similar insulin-sensitive phosphoprotein has been identified by Blackshear et al. (17), although the time course of activation is extended, suggesting parallel activation with transport. Like the inhibition of insulin-sensitive phosphorylation of pp240 and pp24, insulin-stimulated glucose transport is inhibited by PA0 (Table I), confirming earlier results. It should be emphasized that PA0 reduced the insulin-sensitive component of pp240, pp24, and glucose transport to basal values; the basal values, themselves, were not affected by PA0 (data not shown). Unlike pp240 and pp24, the phosphorylation state of other proteins, altered by the presence of insulin, is not changed due to the presence of PAO. For example, the phosphorylation state of a smaller molecular weight protein (pp18) is reduced in the presence of insulin and not reversed by PA0 (Fig. 4 ) . This suggests that PA0 does not generally and nonspecifically disrupt the response pathways of insulin in whole cells and indicates a divergence in the signaling system at some point along the (hypothetical) cascade since PA0 blocks only one branch of the path.
Independent studies by Cushman and Wardzala (18) and Suzuki and Kono (19) provide evidence that the glucose transporter is recruited from an intracellular storage site to the plasma membrane in the presence of insulin. It would be attractive if phosphorylation of the transporter were the signal that initiated this recruitment. However, recent studies by Gibbs et al. (20) show that insulin does not alter the extremely low phosphorylation state of the transporter. So even though the insulin-stimulated kinase activity seems to be important for signaling the transport system, the transporter itself does not appear to be phosphorylated.
Thus, a variety of experiments show positive correlation between the receptor kinase and acceleration of the glucose transport rate. Our results suggest that PAO, which inhibits the activation of glucose transport by insulin, does not affect the function of the insulin receptor, itself, as determined by insulin binding and phosphorylation. PA0 appears to interact at a post-receptor point, but not at the transporter, uncoupling the activated receptor from the glucose transporter system and several insulin-sensitive phosphoproteins. Interestingly, it has been shown recently in 3T3-Ll adipocytes, that a phosphotyrosyl-containing protein (pp15) accumulates in the presence of insulin and PA0 (21). The accumulation of this phosphoprotein and the decrease in the phosphoseryl proteins pp240 and pp24 are consistent with PA0 inhibition at an intervening step.
These data, then, demonstrate two important facts. First, the receptor does not appear to communicate directly with the transporter; at least one additional step must exist between the receptor and the glucose transport system. Second, by localizing the site of PA0 action we may be able to identify some of the components of the insulin "cascade." Thus, PA0 is not just a pharmacological agent but may itself be a useful tool for the examination of the mechanism of some of insulin's actions inside the cell.