The Insulin-elicited 60-kDa Phosphotyrosine Protein in Rat Adipocytes Is Associated with Phosphatidylinositol 3-Kinase”

Insulin stimulates the tyrosine phosphorylation of a 60-kDa protein (pp60) in rat adipocytes. After insulin treatment of these cells, pp60, as well as the 160-kDa insulin receptor substrate-1 (IRS-1), were found to be associated with the enzyme phosphatidylinositol 3-ki- nase (PtdIns-3-kinase) in separate complexes. By contrast, pp60 was not detected in insulin-treated mouse 3T3-Ll adipocytes, which contain abundant IRS-1. PtdIns-3-kinase complex. The pp60. PtdIns 3-kinase complex was located in both the soluble and membrane fractions of the rat adipocytes. Fusion proteins con- taining the isolated src homology 2 domains from the 85-kDa subunit of PtdIns-3-kinase bound to pp60 in lysates of insulin-treated rat adipocytes. This finding indicates that the most likely mode of association of pp60 with PtdIns-3-kinase is through binding of phos- photyrosine residues in pp60 to these domains. By immunoaffinity chromatography on a monoclonal an- tibody against phosphotyrosine, pp60 was purified in high percentage yield from insulin-stimulated rat adi- pocytes, but the low amount of the protein obtained (about 3 ng from the adipocytes of one rat) precluded sequence analysis. concen-tration


The Insulin-elicited 60-kDa Phosphotyrosine Protein in Rat Adipocytes Is Associated with Phosphatidylinositol 3-Kinase"
(Received for publication, October 26, 1992) Brian E. Lavan  Insulin stimulates the tyrosine phosphorylation of a 60-kDa protein (pp60) in rat adipocytes. After insulin treatment of these cells, pp60, as well as the 160-kDa insulin receptor substrate-1 (IRS-1), were found to be associated with the enzyme phosphatidylinositol 3-kinase (PtdIns-3-kinase) in separate complexes. By contrast, pp60 was not detected in insulin-treated mouse 3T3-Ll adipocytes, which contain abundant IRS-1. PtdIns-3-kinase complex. The pp60. PtdIns 3-kinase complex was located in both the soluble and membrane fractions of the rat adipocytes. Fusion proteins containing the isolated src homology 2 domains from the 85-kDa subunit of PtdIns-3-kinase bound to pp60 in lysates of insulin-treated rat adipocytes. This finding indicates that the most likely mode of association of pp60 with PtdIns-3-kinase is through binding of phosphotyrosine residues in pp60 to these domains. By immunoaffinity chromatography on a monoclonal antibody against phosphotyrosine, pp60 was purified in high percentage yield from insulin-stimulated rat adipocytes, but the low amount of the protein obtained (about 3 ng from the adipocytes of one rat) precluded sequence analysis.
The cellular effects of insulin are mediated through the insulin receptor, a heterotetrameric protein consisting of two LY and two p subunits located in the plasma membrane. The ct subunits are located entirely extracellularly and constitute the binding site for insulin, while the p subunits are transmembrane. Upon insulin binding, a protein tyrosine kinase intrinsic to the / 3 subunit is activated, leading to phosphorylation of this subunit on tyrosine residues. Initiation of many, if not all, the effects of insulin require activation of the insulin receptor as a tyrosine kinase (1)(2)(3). In addition to the p subunit of the insulin receptor, several other phosphotyrosyl polypeptides appear after insulin challenge of various cell types (4). Tyrosine phosphorylation of such proteins is likely to be part of the pathway for signal transmission from the insulin receptor to insulin-sensitive effector systems. Identification and characterization of insulin-elicited Tyr(P)' proteins is there-* This work was supported in part by a postdoctoral fellowship from the Juvenile Diabetes Foundation International (to B. E. L.) and National Institutes of Health Grant DK 42816. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In this regard, we and others have recently purified a widely distributed, insulin-elicited Tyr(P) polypeptide of 160-185 kDa, now referred to as IRS-1 (insulin receptor substrate-1) (5,6). Highly related cDNA clones encoding this protein in rat liver (7), human hepatocellular carcinoma (8), and mouse 3T3-Ll adipocytes? have been isolated. The full role played by IRS-1 in insulin signaling is unknown, but one function appears to be to couple the insulin receptor to activation of the enzyme PtdIns-3-kinase. The Tyr(P) form of IRS-1, generated by phosphorylation by the activated insulin receptor, associates with PtdIns-3-kinase (7, 9, lo), and this association stimulates the activity of the enzyme (10). PtdIns-3-kinase consists of a 110-kDa catalytic subunit and an 85-kDa regulatory subunit which contains two src homology-2 (SH2) domains (11)(12)(13)(14)(15). SH2 domains are domains of about 100 amino acids that bind strongly to Tyr(P) in specific peptide sequences in signaling proteins (reviewed in Ref. 16). Two types of evidence indicate that the association of IRS-1 with PtdIns-3-kinase involves binding to the SH2 domain of the latter. First, we have shown that the isolated SH2 domains of PtdIns-3-kinase bind the Tyr(P) form of IRS-1 (9). Second, IRS-1 has nine potential tyrosine phosphorylation sites that lie in the sequence Y X X M , a consensus sequence for binding to PtdIns-3-kinase (7, 17).
In the course of investigating the association of IRS-1 with PtdIns-3-kinase (9), we examined whether other insulin-elicited Tyr(P) proteins have the same property. Insulin treatment of rat adipocytes elicits the rapid tyrosine phosphorylation of a protein of approximately 60 kDa (hereafter referred to as pp60), as well as that of IRS-1 and the p subunit of the insulin receptor (18-23). Estimated from the intensity of its signal on Tyr(P) immunoblots, pp60 is a major insulin-elicited Tyr(P) protein, comparable to IRS-1 and the insulin receptor subunit. To investigate associations of pp60 with PtdIns-3kinase we have used antibodies to the 85-kDa subunit to immunoprecipitate the enzyme and associated proteins. We show that in rat adipocytes the Tyr(P) forms of both pp60 and IRS-1 complex tightly, but independently, with PtdIns-3-kinase and evidence is presented that the association occurs via the SH2 domains of the kinase. In addition, we describe an efficient procedure for the purification of pp60 from insulin-stimulated rat adipocytes.  Honnor et al. (24). Adipocytes were washed in the above buffer, resuspended to a cell density of 1.2 X lo6 cells/ml, and stimulated with 100 nM insulin for 5 min or left in the basal state. Cells were subsequently washed three times in albumin-free buffer; 100 nM insulin was included in the wash buffer for the insulinstimulated cells to maintain a stimulated state.

Isolation of Rat Adipocytes and Preparation
To prepare the soluble, membrane, and cytoskeleton fractions, adipocytes from eight rats (about 20 mg total protein) were homogenized at room temperature in 7.5 ml of Buffer A (50 mM Hepes, 150 mM NaCl, 1 mM EGTA, 100 mM sodium fluoride, 10 mM sodium pyrophosphate, 10% glycerol, 1.5 mM MgCl,, 1 mM sodium orthovanadate, 10 pg/ml aprotinin, 10 p M leupeptin, 1 mM PMSF, at pH 7.0) using 10 up-anddown strokes in a glass homogenizer with a Teflon pestle driven at 1250 rpm. The extract was subsequently centrifuged at 116,000 X g. , for 1 h at 4 "C. The soluble fraction (infranatant) was retained, and the fat cake discarded. The pellet was resuspended in 7.5 ml of 1% Triton X-100 in Buffer A and recentrifuged as above to yield the solubilized membrane preparation. The Triton X-100-insoluble material was solubilized in 7.5 ml of SDS sample buffer to give the cytoskeleton fraction.
For immunoprecipitation experiments, nondenatured and denatured lysates were prepared from the adipocytes isolated from eight rats as follows. Nondenaturing detergent solubilization was accomplished by mixing 17 ml of 1% Triton X-100 in Buffer A directly with the cells at room temperature. For denatured extracts, the adipocytes were lysed by vigorous mixing with 1.3 ml of 4% SDS in 100 mM Hepes, 300 mM NaCI, 80 p M leupeptin, 80 pg/ml aprotinin, 8 mM PMSF, 8 mM sodium orthovanadate, and 20 mM dithiothreitol, at pH 7.0. The mixture was held at 100 "C for 5 min and then centrifuged at 4800 X g , .
for 15 min to separate the SDS-solubilized material (infranatant) from the fat cake. The DNA in the infranatant was then sheared by passage through syringe needles, and the sulfhydryl groups were alkylated by the addition of 30 mM N-ethylmaleimide. Finally, the extract was diluted in 15 ml of 1.2% Triton X-100 in 60 mM Hepes, 175 mM NaCI, 1.2 mM EGTA, 120 mM sodium fluoride, 12 mM sodium pyrophosphate, 12% glycerol, 2 mM MgCl,, 0.5 mM sodium orthovanadate, 5 pg/ml aprotinin, 5 pM leupeptin, 0.5 mM PMSF, at pH 7.0. Both the nondenatured and denatured extracts were centrifuged at 116,000 X g, , for 1 h, and the supernatant, after passage through a Millipore filter (type Millex GV 0.22 pm), was used for immunoprecipitation.
In the purification of pp60, adipocytes from 30 rats were treated with 100 nM insulin for 5 min, to elicit the tyrosine phosphorylation of pp60. The adipocytes were lysed with 5 ml of an SDS lysis buffer (4% SDS, 200 mM NaCl, 60 mM Tris-HCI, 20 mM dithiothreitol, 8 mM sodium orthovanadate, 80 pM leupeptin, 80 pg/ml aprotinin, 8 pg/ml pepstatin A, 80 p~ EP475, at pH 7.6) and held at 100 for 5 min. The lysate was separated from the fat, sheared, and alkylated with N-ethylmaleimide as detailed above. Finally, the extract was diluted by the addition of 60 ml of 1.2% Cl2ES in 30 mM Tris-HCl,80 mM NaCI, 0.5 mM sodium orthovanadate, 5 p M leupeptin, 5 pg/ml aprotinin, 0.5 pg/ml pepstatin A, 5 p M EP475, at pH 7.6. The lysate was then centrifuged, and the supernatant filtered as above.
Cell Culture and Preparation of Extracts from 3T3-Ll Adipocytes-3T3-Ll adipocytes were carried as fibroblasts and differentiated into adipocytes as previously described (25). 10-cm plates of adipocytes at 10-12 days post differentiation were incubated for 2 h in serum-free medium and treated with 300 nM insulin for 3 min or left in the basal state.
To prepare soluble, membrane, and cytoskeleton fractions, cells from a plate (about 10' cells with 5 mg of total protein) were scraped at room temperature into 2 ml of Buffer B (20 mM Tris-HC1,140 mM NaCI, 30 mM sodium pyrophosphate, 1 mM sodium orthovanadate, 10 pg/ml aprotinin, 1 pg/ml pepstatin A, 10 pM EP475 and 1 mM PMSF, at pH 7.5) and homogenized in a glass homogenizer with a Teflon pestle by 25 hand-driven up-and-down strokes at room temperature. The homogenate was centrifuged at 105,000 X g. , for 1 h at 4 "C, and the infranatant removed. After discarding the fat cake, the membrane pellet was resuspended in 2 ml of 1.5% ClzEs in Buffer B and recentrifuged to prepare a solubilized membrane extract. The ClZE8-insoluble material was solubilized in 2 ml of SDS sample buffer and was used as the cytoskeleton fraction.
Whole cell lysates were prepared under nondenaturing or denaturing conditions for use in immunoprecipitation experiments as follows. For lysis under nondenaturing conditions, the cells on one 10-cm plate were scraped at room temperature into 6 ml of 1.5% ClzEs in Buffer B and vortexed briefly. Denatured cell lysates were prepared by scraping the cells on one 10-cm plate into 0.8 ml of 2% SDS in Buffer B plus 10 mM dithiothreitol, vortexing briefly, and holding the mixture at 100 "C for 5 min. The DNA in the lysate was sheared, and the sulfhydryl groups were alkylated, as described above for the denatured lysate of rat adipocytes. Subsequently, the extract was diluted with 5.2 ml of 1.7% ClzEa in Buffer B. Both the nondenatured and denatured extracts were then centrifuged and filtered as described above for the lysate from rat adipocytes.
Antibodies-Affinity purified rabbit antibodies against Tyr(P) were prepared as described elsewhere (26). The mouse monoclonal antibody against Tyr(P), designated 4G10, was a generous gift of Dr. Brian Druker (Dana Farber Cancer Institute, Boston, MA). The antibodies against IRS-1 were either the affinity purified rabbit antibodies against the carboxyl-terminal peptide, described in Lamphere and Lienhard (41) or the rabbit antiserum against an internal peptide (peptide 3) described in Keller et al. (5). Rabbit antiserum against the 85-kDa subunit of PtdIns-3-kinase was purchased from Upstate Biotechnology, Lake Placid, NY.
Immunoprecipitations and Immunoblotting-Immunoprecipitations of IRS-1, p85, and Tyr(P) proteins from the extracts of rat adipocytes were performed as follows. The extracts were incubated with primary antibody for 30 min (soluble and membrane fractions) or 2 h (nondenatured and denatured lysates) and the immune complexes collected by mixing for 1 h (soluble and membrane fractions) or 2 h (nondenatured and denatured lysates) with 20 pl of protein A-Sepharose. The immunoprecipitates were washed three times with the solubilization buffer. The immunoprecipitations of these proteins from the lysates of 3T3-Ll adipocytes were performed in a similar way. The amounts of the antibodies used are given in the figure legends. These amounts were sufficient to adsorb at least 50% of the IRS-1, 70% of the p85, and 90% of the Tyr(P) proteins, as assessed by immunoblotting of the initial lysate and the lysate after immuby adding 60 pl of SDS sample buffer and heating at 65 "C for 10 noadsorption (data not shown). Immune complexes were dissociated min. After removal of the beads, the SDS sample was heated at 100 "C for 3 min. The SDS sample buffer consisted of 4% SDS, 20 mM dithiothreitol, 1 mM EDTA, 10% glycerol, 100 mM Tris-HC1, at pH 6.8 with protease inhibitors (10 p~ EP475, 10 p~ leupeptin, 10 pg/ ml aprotinin, 1 pg/ml pepstatin A, 1 mM PMSF) and phosphatase inhibitors (1 mM sodium orthovanadate, 10 mM phenyl phosphate).
Samples were separated by SDS-gel electrophoresis on small 10% acrylamide slab gels and transferred electrophoretically to Immobilon-P (Millipore, Bedford, MA) at 400 mA for 3 h. The transfer buffer consisted of 25 mM Tris, 190 mM glycine, 20% methanol, 0.005% SDS. Proteins were immunoblotted as follows: Tyr(P) proteins, with the affinity-purified rabbit anti-Tyr(P) as detailed in Lavan et al. (9); IRS-1, with the affinity purified rabbit antibodies against the carboxyl-terminal peptide, at 4 pg/ml; p85, with the anti-p85 antiserum at a dilution of 1/500. Tyr(P) immunoblots were blocked with 30 mg/ ml bovine serum albumin, and the primary and secondary antibodies were in 2 mg/ml albumin; the IRS-1 and p85 blots were blocked in 5% Carnation nonfat dry milk, and the antibodies were in 1% milk. The buffer used for blocking and for the washes was 20 mM Tris-HCI, 150 mM NaC1,0.3% Tween-20, at pH 7.6. Blots of SDS samples of immunoadsorbed proteins were developed with horseradish peroxidase conjugated to protein A (Bio-Rad); the other blots were developed with horseradish peroxidase conjugated to goat anti-rabbit immunoglobulins (Bio-Rad). The enhanced chemiluminescence reagent (Amersham Corp.) was then used to detect immunoreactive proteins.
Adsorption of Insulin-elicited Tyr(P) Proteins with the SH.2 DOmains of PtdIns-3-kinase-The amino and carboxyl SH2 domains of the 85-kDa subunit of PtdIns-3-kinase (a form) were expressed as glutathione S-transferase fusion proteins in Escherichia coli and isolated on glutathione-Sepharose beads, as described previously (9, 27). Beads (5 pl containing 50 pg of fusion protein) were mixed for 4 h at 4 "C with 0.5 mi of denatured lysate prepared from rat adipocytes according to the procedure used in the purification of pp60 (see above). The beads were washed four times with 20 mM Tris-HCL150 mM NaCI, 1 mM sodium orthovanadate, 10 p M leupeptin, 10 pg/ml aprotinin, 1 pg/ml pepstatin A, 10 p~ EP475, at pH 7.6, containing successively lower concentrations of detergents (1% Cd$/O.25% SDS, 1% C12E8, 0.1% C12E8, and 0.015% CI2E8). The bound proteins were eluted with SDS sample buffer containing 8 M urea at 60 "c for 10 min and subjected to immunoblotting.
Purificution ofpp60-Lysate (70 ml) prepared from the adipocytes of 30 rats (see above) was thawed rapidly and clarified by centrifugation (48,000 X gmu) for 20 min. The Tyr(P) polypeptides were immunoadsorbed at 4 "C by passing the lysate at 0.2 ml/min through a 1-ml column of goat IgG agarose in tandem with a 0.5-ml column of the monoclonal anti-Tyr(P) (4G10) immobilized on agarose (1 mg/ ml; Upstate Biotechnology, Lake Placid, NY). Once all the extract was applied, the goat IgG column was disconnected and the 4G10 washed sequentially at 0.5 ml/min with 10 ml of 1% CIzER in Buffer C (20 mM Tris-HCI, 150 mM NaCI, 5 mM EDTA, 1 mM sodium orthovanadate, 2 pg/ml aprotinin, 2 p~ leupeptin, 0.2 pg/ml pepstatin A; pH 7.4) and 10 ml of 0.1% CI2Es in Buffer C. Finally, the column was washed extensively a t 0.1 ml/min with 80 ml of 0.015% ClzER in Buffer C. Tyr(P)-containing polypeptides were eluted with 40 mM phenyl phosphate in the final wash buffer, by running 1.5 bed volumes of elution buffer onto the column, stopping elution for 1 h, and then proceeding with elution a t 0.5 ml/min. Eluted proteins were collected in 2-ml fractions in low protein adsorption tubes (4.0 ml Minisorp tubes from Nunc, Naperville, IL).
T o estimate the yield of pp60 and other Tyr(P) polypeptides, samples containing several dilutions of the original lysate, immunodepleted lysate, and a pooled mix of the peak fractions (typically three to four fractions), were separated by SDS-gel electrophoresis on a large 10% acrylamide slab gel, transferred to nitrocellulose (200 mA for 16 h) and immunoblotted for Tyr(P) with detection by '*'Ilabeled protein A (ICN, Costa Mesa, CA) a t 0.08 pCi/ml. Quantitation of the radiolabel on the immmunoblots was done by cutting out the area of the nitrocellulose corresponding to the band on the autoradiogram and measuring the radioactivity in a y-counter; the label in an appropriate blank area of the blot was used to correct for background.
T o visualize proteins by colloidal gold staining, proteins in peak fractions were concentrated by trichloroacetic acid precipitation. Sodium deoxycholate a t 1.5 mg/ml was added as carrier, and precipitation achieved by the addition of trichloroacetic acid to a concentration of 10%. After 30 min on ice, the pellet was collected by centrifugation for 5 min a t 16,000 X gmax. The pellet was dissolved in 200 pl of SDS sample buffer without dithiothreitol, and the sample was neutralized with Tris base. Samples were separated as described for Tyr(P) immunoblotting and the proteins on nitrocellulose stained with colloidal gold (Bio-Rad).

RESULTS
Insulin-elicited Tyr(P) Proteins in Rat and 3T3-Ll Adipocytes-The distribution of insulin-elicited Tyr(P) proteins between soluble, membrane, and cytoskeleton fractions was examined in both rat and 3T3-Ll adipocytes by immunoblotting these fractions from basal and insulin-treated cells with antibodies against Tyr(P) (Fig. 1). Insulin-elicited Tyr(P) proteins of 160 and 95 kDa were observed in both cell types, the former being largely soluble and the latter largely membrane bound. In contrast, a 60-kDa Tyr(P) protein, found in both the soluble and membrane fractions, was observed in rat adipocytes but not in 3T3-Ll adipocytes (compare  (9,18), and most probably, the extracellular signal regulated protein kinases 1 and 2 (28), respectively. The identity of the 60-kDa protein (pp60) which has been found in insulin-treated rat adipocytes in other studies (see introduction), is unknown. Low levels of the insulin receptor /3 subunit were detected in the cytoskeleton fraction of both rat and 3T3-Ll adipocytes (Fig. 1, A and B, lanes 5 and 6).
Association of Insulin-elicited Tyr(P) Polypeptides with PtdIns-3-kinase in Rat Adipocytes-In order to determine whether PtdIns-3-kinase was associated with any of these Tyr(P) proteins, the approach of immunoadsorption and immunoblotting was adopted. Lysates of basal and insulintreated cells were immunoadsorbed with antibodies against IRS-1, the 85-kDa subunit of PtdIns-3-kinase (p85), and Tyr(P), and the immunoadsorbates were immunoblotted with all three antibodies. The lysates were prepared under both nondenaturing conditions (solubilization in the nonionic detergent, Triton X-100) and denaturing conditions (solubilization in hot SDS/dithiothreitol, followed by N-ethylmaleimide and an excess of Triton X-loo), in order to maintain or disrupt protein-protein associations, respectively. This method allowed us to assess whether a precipitating polypeptide was present in the immunoprecipitate as a result of association with the antigen protein or simply as a consequence of crossreactivity with the precipitating antibodies. Since proteinprotein interactions should be disrupted in extracts prepared under denaturing conditions, the amount of a polypeptide precipitated as a consequence of association should be reduced under these conditions. pp60, detected by anti-Tyr(P) immunoblotting, was highly enriched in the PtdIns-3-kinase immunoprecipitate from the nondenatured lysate of insulin-treated adipocytes (Fig. 2 A , compare lane 2 with 8). Co-immunoprecipitation of the pp60 with PtdIns-3-kinase was completely abolished when the lysate was prepared under denaturing conditions (Fig. 2 A , compare lane 18 with 8). In contrast, the ability of anti-p85 to precipitate p85 was found not to differ between the nondenatured and denatured lysates, since similar levels of PtdIns-3-kinase were recovered from each (Fig. 2B, lanes 7 and 8  versus 17 and 18). These observations indicate that pp60 coimmunoprecipitated with PtdIns-3-kinase as a result of association, rather than as a result of cross-reaction of anti-p85 with pp60. It was not possible to assess whether pp60 was also associated with PtdIns-3-kinase under basal conditions because there are no antibodies against the pp60 protein and because it contained little or no Tyr(P) prior to insulin stimulation (Fig. lA, lanes 1 and 3, and Fig. 2 A , lane 9).
As expected (see the introduction) insulin treatment of rat adipocytes promoted the association of PtdIns-3-kinase with IRS-1. Under basal conditions the two were associated to a small extent and the extent of this association was markedly increased with insulin treatment (Fig. 2 A , lanes 7 and 8; Fig.   2B, lanes 5 and 6; and Fig. 2C, lanes 7 and 8). The increased association paralleled the increased tyrosine phosphorylation of IRS-1 ( Fig. 2A, lanes 5 and 6). This association was substantially reduced in lysates prepared under denaturing conditions ( Fig. 2A, lane 18 versus 8; Fig. 2B, lane 16 versus 6; and Fig. 2C, lane 18 versus 8 ) . The observation that some slight association of the Tyr(P) form of IRS-1 with p85 occurred after SDS denaturation at 100 "C ( Fig. 2A, lane 18) may be explained by partial renaturation of the SH2 domains of p85. If both pp60 and IRS-1 were associated in a ternary complex with PtdIns-3-kinase, then some pp60 would be expected in the anti-IRS-1 immunoprecipitates. The absence of any ( Fig.  2A, lanes 5 and 6) indicates that these two proteins form separate complexes with PtdIns-3-kinase.
The results in Fig. 2 also show that the 85-kDa subunit of PtdIns-3-kinase was not itself tyrosine phosphorylated in response to insulin. First, although p85 appeared in the anti-Tyr(P) immunoprecipitate from the nondenatured cell lysate, it was not found in that from the denatured cell lysate (Fig.  2B, compare lane 10 with 20). The efficiency of precipitation by the anti-Tyr(P) antibody was not significantly different between nondenatured and denatured extracts ( Fig. 2A, compare lane 10 with 20). Second, there was no detectable insulinelicited Tyr(P) protein of 85 kDa in total cell lysates ( Fig. 2A,  lanes 2 and 12) or in anti-p85 or anti-Tyr(P) immunoprecipitates (Fig. 2 A , lanes 8, 10, 18, and 20).
The data in Fig. 2A also suggest that there is no strong association between the Tyr(P) form of the insulin receptor and either IRS-1 or p85 under the conditions employed. The immunoprecipitates with anti-IRS-1 and anti-p85 exhibit little or no signal for the insulin receptor /3 subunit, whereas the immunoprecipitate with anti-Tyr(P) shows a strong signal ( Fig. 2A, lanes 6 and 8 versus 10).
Since pp60 was located in both the soluble and membrane fractions of rat adipocytes (Fig. lA), we determined whether the protein in both locations complexed with PtdIns-3-kinase. Soluble and membrane fractions were prepared from basal and insulin-treated rat adipocytes, and the membrane fraction was then solubilized with Triton X-100. These fractions were then immunoprecipitated with antibodies against IRS-1, p85, and Tyr(P) and immunoblotted for the same (Fig. 3). Association between pp60 and PtdIns-3-kinase occurred in both the soluble and membrane fractions (Fig. 3A, lanes 8 and 18). The association between IRS-1 and PtdIns-3-kinase occurred almost entirely in the soluble fraction (Fig. 3A, lane 8 Fig. 3C, lane 8 versus 18). This finding is consistent with PtdIns-3-kinase and IRS-1 being mainly cytosolic proteins in rat adipocytes (Fig. 3B, lanes 1  and 2 versus 11 and 12; Fig. 3C, lanes 1 and 2 versus 11 and  12). Interestingly, insulin caused some translocation of PtdIns-3-kinase to the membrane fraction (Fig. 3B, lane 11  versus 12 and 17 versus 18). The extent of translocation was small (Fig. 3B, compare lune 12 with 2 and lane 18 with 8 ) , but it was also observed in a replicate experiment.

Association of Insulin-elicited Tyr(P) Polypeptides with PtdIns-3-kinase in 3T3-Ll
Adipocytes-In order to compare further the rat and 3T3-Ll adipocytes, immunoprecipitation and immunoblotting with antibodies against IRS-1, p85, and Tyr(P) were also carried out with nondenatured and denatured lysates of 3T3-Ll adipocytes (Fig. 4). With the nondenatured lysate, anti-p85 immunoprecipitates contained the Tyr(P) form of IRS-1 but not of the insulin receptor /3 subunit (Fig. 4A, lane 8). The band in Fig. 44, lane 8, with an electrophoretic mobility slightly less than that of the insulin receptor p subunit is probably an antibody-derived band, since it was present in the basal immunoprecipitate (Fig. 4 A , lane Lanes with immunoprecipitates contain samples derived from 5% of the cells on a 10-cm plate. The autoradiograms presented are representative ones of an experiment performed three times. 7) and the control immunoprecipitates (Fig. 4A, lunes 3 and some p85 (Fig. 4B, lune 5 ) ; the extent of association was 4 ) . As was the case with the lysates of rat adipocytes, the significantly increased as the result of the increased tyrosine association of PtdIns-3-kinase with the Tyr(P) form of IRS-phosphorylation of IRS-1 in response to insulin (Fig. 4R, lune 1 was drastically reduced in lysates prepared under denaturing 6 ) . In the complimentary immunoprecipitation, IRS-1 was conditions (Fig. 4A, lune 18 uersus 8 ) . Association of IRS-1 barely detectable in the anti-p85 immunoprecipitate from with p85 occurred to some extent in the basal state, since the basal cells, while a larger amount was present in this immuanti-IRS-1 immunoprecipitate from basal cells contained noprecipitate after insulin stimulation (Fig. 4C, lunes 7 and with Phosphatidylinositol 3-Kinase 8). These associations were absent in extracts prepared under denaturing conditions (Fig. 4B, lanes 15 and 16 versus 5 and   6; Fig. 4C, lanes 17 and 18 versus 7 and 8). This finding again indicates that co-immunoprecipitation is due to association rather than to cross reactivity of the antisera.
As was the case in the rat adipocytes, PtdIns-3-kinase was present in the anti-Tyr(P) immunoprecipitate after insulin treatment, but this association was abolished in denatured extracts (Fig. 4B, lane 10 versus 20). There was no detectable insulin-elicited Tyr(P) protein of 85 kDa either in total cell lysates (Fig. 4 A , lanes 2 and 12) or in the anti-p85 or the anti-Tyr(P) immunoprecipitates (Fig. 4 A , lanes 8, 10, 18, and 20).
Thus, insulin did not cause the tyrosine phosphorylation of p85 in 3T3-Ll adipocytes.
It is important to note that in sharp contrast to the results with rat adipocytes, pp60 was not detected in either the anti-p85 or the anti-Tyr(P) immunoprecipitates of 3T3-Ll adipocytes (compare Fig. 4 A , lanes 8,10, and 20, with Fig. 2A, lanes   8, 10, and 20). Since immunoprecipitation followed by immunoblotting is a procedure that concentrates pp60 and therefore markedly enhances its signal ( Fig. 2 A , compare the pp60 signals in lanes 2 and 12 with those in lanes 8, 10, and 20), the absence of pp60 in these immunoprecipitates is further evidence that the amount of this protein is very low in 3T3-L1 adipocytes relative to rat adipocytes.
Association of pp60 with the SH2 Domains of PtdIns-3kinase-In order to assess whether the association of pp60 with PtdIns-3-kinase could be mimicked by the SH2 domains of its 85-kDa subunit alone, we employed glutathione Stransferase fusion proteins containing the amino-and carboxyl-terminal SH2 domains of p85. Insulin-stimulated lysates of rat adipocytes were prepared under denaturing conditions, in order to dissociate endogenously associated PtdIns-3-kinase, and were then adsorbed with the glutathione Stransferase fusion proteins bound to glutathione-Sepharose. The original lysate, depleted lysate, and adsorbates were subsequently blotted for Tyr(P) (Fig. 5 ) . pp60  sate, lysates after adsorption (depleted lysates), and the adsorbates were immunoblotted for Tyr(P). The lanes with original and depleted lysates contained samples equivalent to 7% of the lysate from the adipocytes from one rat, whereas those with the adsorbates were derived from 20% of the lysate from one rat. A second experiment of this type, which included both basal and insulin lysates, gave similar results with the insulin lysate, and no Tyr(P) pp60 was found in the SH2 adsorbates of the basal lysate. (Fig. 5, lanes 6 and 7). No association of these proteins was observed with the glutathione S-transferase alone (Fig. 5, lane  5 ) . These associations were also observed with the SH2 domains of PtdIns-3-kinase present as TrpE fusion proteins in an experiment where the adsorption was carried out with approximately 50 times less fusion protein/assay compared to the amount of the glutathione S-transferase fusion proteins (data not shown; see Ref. 9 for method).
Purification of pp60 from Rat Adipocytes-To purify pp60 from rat adipocytes we used immunoaffinity chromatography on anti-Tyr(P). Insulin-treated rat adipocytes were lysed in hot SDS/dithiothreitol, alkylated with N-ethylmaleimide, and finally diluted in an excess of nonionic detergent. This lysate was passed over a column of an immobilized monoclonal anti-Tyr(P) antibody. The column was washed and adsorbed Tyr(P) proteins eluted with phenyl phosphate. Fig. 6 shows the profile of eluted Tyr(P) proteins, as determined both by anti-Tyr(P) immunoblotting and by colloidal gold staining for protein. The purified material contained four major Tyr(P) bands (Fig. 6, left panel). We identified three of these Tyr(P) proteins as IRS-1, the / 3 subunit of the insulin receptor, and pp60 by comparing their electrophoretic mobilities with those of the Tyr(P) bands in the total cell lysate from insulintreated adipocytes. The fourth band, which migrates directly underneath pp60, most probably is a constitutive Tyr(P) protein. When the purification was performed on a small scale with lysate from both basal and insulin-treated adipocytes, this protein was also present in the phenyl phosphate eluate from the basal lysate, whereas as expected IRS-1, the / 3 subunit of the insulin receptor, and pp60 were not (data not shown). When the same proteins eluted with phenyl phosphate were visualized with a sensitive colloidal gold stain (Fig.  6, right panel), bands corresponding to IRS-1, the insulin receptor /3 subunit, pp60, and the constitutive Tyr(P) protein were observed. On the basis of its identical electrophoretic mobility in both blots and also its rather broad appearance, the colloidal gold band corresponding to pp60 is likely to be purified pp60.
The recoveries of Tyr(P) proteins were estimated by quantitative immunoblotting for Tyr(P). Samples of total lysate, depleted lysate from the flow-through of the anti-Tyr(P) immunoaffinity column, and the purified Tyr(P) proteins were each separated a t several loads on a single gel and immunoblotted with the combination of polyclonal anti-Tyr(P) and ""I-protein A. In a preparation that started with the adipocytes from 30 rats, 75% of the pp60 from the total cell lysate was adsorbed by the anti-Tyr(P) column, and subsequently 50% of the adsorbed pp60 was recovered in the Tyr(P) proteins eluted with phenyl phosphate. Thus the overall yield of pp60 was 37%. A similar yield was obtained for the /3 subunit of the insulin receptor (21%). A rough estimate of the amount of pp60 purified by this procedure can be made by visual comparison of its intensity upon colloidal gold staining with the intensities of various amounts of the protein standards (Fig. 6, right panel). On this basis, 3 ng of pp60 were purified from the adipocytes of one rat by the combination of immunoaffinity chromatography and preparative scale polyacrylamide electrophoresis. Due to the low abundance of pp60, it was not feasible to purify enough to carry out sequence analysis on tryptic peptides, a procedure that typically requires about 100 pmol (6 pg) (29).

DISCUSSION
Our results confirm those of earlier studies showing that with rat adipocytes insulin induces the tyrosine phosphorylation of a 60-kDa protein, in addition to the insulin receptor and IRS-1 (18-23). A major new finding of this study is that the Tyr(P) form of pp60, like that of IRS-1, is associated with PtdIns-3-kinase. The primary evidence for this is that under nondenaturing conditions, pp60 was immunoprecipitated by an antiserum against the 85-kDa subunit of PtdIns-3-kinase.
We have also obtained this result with two monoclonal antibodies against the a form of p85, the epitopes for which are located in the bcr-like and SH3 domains, as well as with an antiserum specific for the carboxyl-terminal peptide of the 110-kDa subunit of the enzyme (all kindly provided by Dr. M. Waterfield, Ludwig Institute for Cancer Research, London; data not shown). Consistent with this finding, the isolated SH2 domains of PtdIns-3-kinase, expressed as the glutathione S-transferase fusion proteins, efficiently bound the Tyr(P) form of pp60. This observation also indicates that the association of pp60 with PtdIns-3-kinase most likely occurs through the binding of its phosphotyrosine phosphorylation sites to these domains in the kinase. This mode of association would also account for the absence of a ternary complex of PtdIns-3-kinase, pp60, and IRS-1, since IRS-1 also probably binds to PtdIns-3-kinase via the SH2 domains (9). Previous studies have shown that insulin treatment of rat adipocytes causes the appearance of PtdIns-3-kinase activity in anti-Tyr(P) immunoprecipitates (30,31,42). A priori the basis of this effect could either be tyrosine phosphorylation of PtdIns-3-kinase itself or association of PtdIns-3-kinase with Tyr(P) proteins. Our study strongly supports the latter. In addition to pp60, PtdIns-3-kinase was also found in association with the Tyr(P) form of IRS-1. Moreover, no tyrosine phosphorylation of p85 was detected by anti-Tyr(P) immunoblotting of the lysates or of the anti-Tyr(P) or anti-p85 immunoprecipitates.
We found that all of the PtdIns-3-kinase complex with IRS-1, and most with pp60, was located in the soluble fraction, although in the latter case some complex was observed in the membrane fraction. This agrees with the results of a recent study, in which PtdIns-3-kinase activity immunoprecipitated with anti-Tyr(P) was located in both the soluble and membrane fractions of insulin-treated rat adipocytes (42). In a separate study, however, all of the PtdIns-3-kinase activity immunoprecipitated with anti-Tyr(P) was located in the membrane fraction (31). A possible explanation for these discrepancies may lie in the different buffers used for homogenization of the cells. Ours and that in (42) contained relatively high concentrations of salts (300 mM), whereas that in (31) contained only 25 m M salts.
The functional consequence of the association of pp60 with PtdsIns-3-kinase remains to be determined. In the case of IRS-1, it has recently been demonstrated that association of its Tyr(P) form with PtdIns-3-kinase activates the kinase about %fold (10). Since both pp60 and IRS-1 probably associate via the SH2 domains of the kinase, a reasonable expectation is that association with pp60 is also activating. The rapid appearance of PtdIns 3,4-and 3,4,5-phosphates seen upon exposure of adipocytes to insulin (32) may thus involve signaling through both pp60 and IRS-1.
The identity of pp60 is currently unknown. There is considerable evidence that this is a unique protein, rather than a proteolytic fragment of IRS-1 or the p subunit of the insulin receptor. First, Mooney and Bordwell (23) have labeled rat adipocytes with "Pi, isolated these three Tyr(P) proteins by immunoadsorption with anti-Tyr(P) followed by gel electrophoresis, and generated one-dimensional peptide maps with both V8 proteinase and chymotrypsin. Most of the bands in with Phosphatidylinositol 3-Kinase the maps of pp60 did not correspond to bands in those of IRS-1 or the p subunit. Second, our finding that pp60, but not the insulin receptor, is associated with PtdIns-3-kinase also indicates that pp60 is not a fragment of the insulin receptor. Also, since pp60 is largely a soluble protein, it would have to be derived from the cytoplasmic domain of the p subunit of the receptor, but the mobility of the entire domain on gel electrophoresis corresponds to only 48 kDa (33). Finally, we have immunoblotted the mixture of Tyr(P) polypeptides purified from rat adipocytes (see Fig. 6) with antibodies against the carboxyl-terminal peptide of the insulin receptor ((34), kindly provided by Dr. Robert Smith, Joslin Diabetes Center, Boston, MA) and with antibodies against peptides corresponding to amino acid residues 764-777 and 1222-1235 (carboxyl terminus) in rat IRS-1 (7). In each case the antibodies reacted with the expected polypeptide but did not react with pp60 (data not shown). In combination with the report that antibodies against a peptide corresponding to amino acid residues 489-503 of rat IRS-1 do not react with pp60 (35), the result with our antibodies against IRS-1 peptides also eliminates IRS-1 as the precursor of pp60.
Although we were able to purify pp60 in high percentage yield from rat adipocytes, the actual amount was too low to obtain the sequence of peptides. Unfortunately, rat adipocytes are the most abundant source of this protein known to date. This Tyr(P) protein was not detected in the mouse 3T3-Ll adipocytes, and in another study it was not detected in human adipocytes (36). Also, it has not been observed in the other major insulin-sensitive cell types, liver (6, 37, 38) and muscle (33,41). The lack of detection of this protein in other cell types may be accounted for by low abundance or alternatively because of limited tyrosine phosphorylation or rapid dephosphorylation. Further characterization of this insulin-elicited Tyr(P) protein will require discovery of a richer source or an approach different from purification. In this regard, we have noted that pp60 is similar in size to that of the tyrosine kinases of the src family (39). However, an antibody against the carboxyl-terminal peptide common to src, yes, and fYn (40) (kindly provided by Dr. Sara Courtneidge, EMBL, Heidelberg) did not immunoblot purified pp60 under conditions where a rat brain homogenate gave a strong signal (data not shown). This result indicates that pp60 is not one of these kinases.