Functional Labeling of Insulin Receptor Subunits in Live Cells

Both receptor subunits were functionally labeled in order to provide methods allowing, in live cells and in broken cell systems, concomitant evaluation of the in- sulin receptor dual function, hormone binding, and kinase activity. In cell-free systems, insulin receptors were labeled on their a-subunit with ‘251-photoreactive insulin, and on their &subunit by autophosphorylation. Thereafter, phosphorylated receptors were separated from the complete set of receptors by means of anti- phosphotyrosine antibodies. Using this approach, subpopulation of receptors was found which had bound insulin, but which were not phosphorylated. Under nonreducing conditions, receptors appeared in three oligomeric species identified as a2D, and a2. Mainly the a2B2 receptor species was to be phosphorylated, while insulin was bound to a&, and a2 forms. cells, receptors used. Receptors were first labeled with thionine. the led to receptor autophosphorylation The total was precipitated whereas bodies, the phosphorylated

Both receptor subunits were functionally labeled in order to provide methods allowing, in live cells and in broken cell systems, concomitant evaluation of the insulin receptor dual function, hormone binding, and kinase activity. In cell-free systems, insulin receptors were labeled on their a-subunit with '251-photoreactive insulin, and on their &subunit by autophosphorylation. Thereafter, phosphorylated receptors were separated from the complete set of receptors by means of antiphosphotyrosine antibodies. Using this approach, a subpopulation of receptors was found which had bound insulin, but which were not phosphorylated. Under nonreducing conditions, receptors appeared in three oligomeric species identified as a2B2, a2D, and a2. Mainly the a2B2 receptor species was found to be phosphorylated, while insulin was bound to a&, ma@, and a2 forms.
In live cells, biosynthetic labeling of insulin receptors was used. Receptors were first labeled with [3SS]methionine. Subsequently, the addition of insulin led to receptor autophosphorylation by virtue of the endogenous ATP pool. The total amount of [3SS]methioninelabeled receptors was precipitated with antireceptor antibodies, whereas with anti-phosphotyrosine antibodies, only the phosphorylated receptors were isolated. Using this approach we made the two following key findings: (1) Both receptor species, a& and a&, are present in live cells and in comparable amounts. This indicates that the a2fl form is not a degradation product of the a2/32 form artificially generated during receptor preparation.
Insulin receptors are composed of a-and @-subunits. Both subunits possess a distinct function: the a-subunit contains the insulin binding site, a feature which has been evidenced by chemical cross-linking of insulin to its receptor (1) and by photoaffinity labeling ( 2 , 3). The P-subunit displays an insulin-stimulatable tyrosine kinase activity, which is thought to play a key role in hormonal signaling (4-6). This activity was shown to be intrinsic to the receptor (7-9), as confirmed by the subsequent identification of an ATP binding site consensus sequence in the receptor &subunit (10,11). Receptor * This work was supported by grants from Institut National de la Sante et de la Recherche Medicale, France, University of Nice, Fondation pour la Recherche Mkdicale, by Bayer Pharma (France), and from Deutsche Forschungsgemeinschaft (SFB 113 Diabetesforschung Dusseldorf). 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. kinase activity appears to be essential for transmission of the insulin signal. Thus, in various physiopathological states associated with modifications of insulin action, the receptor tyrosine kinase is altered in parallel (12)(13)(14). More recently, conclusive evidence for the important role of the receptor kinase was presented by the demonstration that cells transfected with receptors mutated in the ATP binding site completely lose the ability to transmit an insulin response both for metabolic and mitogenic effects (15,16).
When analyzed under nonreducing conditions which preserve disulfide bonds between the receptor subunits, or nondenaturing conditions which maintain native protein conformations, insulin receptors appear as multiple species (17)(18)(19)(20)(21). Some of those species were identified as being partially degraded proteolytic products, devoid of autophosphorylating activity (20, 21). Since most of those studies were performed with purified receptor preparations, the proteolytic forms could have been generated during the receptor isolation procedure; their "natural" occurrence in intact cells has not been addressed by the authors. In various pathophysiological conditions, receptor kinase activity has been related to insulin binding activity and parallel alterations in hormone action and kinase functioning have been taken as indications for a role of the receptor kinase in hormone signaling. The validity of these studies relies entirely on the assumption that insulin receptor forms with malfunctioning kinase are not artificially generated during the receptor extraction procedure.
In an attempt to provide a better approach to study the insulin receptor role, we have used functional labeling of receptor subunits combined with discriminating receptor immunoprecipitation. Thus, receptor subunits were labeled first by a-subunit tagging with iodinated hormone and then by @subunit autophosphorylation; subsequently, phosphoreceptors were extracted with phosphotyrosine antibodies. Using this approach we found that both in live cells and cell-free systems, the a& insulin receptor species was the prevalently autophosphorylated form and likely represents the signaling receptor.

EXPERIMENTAL PROCEDURES
M a t e r i~l s " N a '~~1 was from CEA (France). [y3'P]ATP (triethylammonium salt; 3000 Ci/mmol) was from the Radiochemical Centre (Amersham Bucks, United Kingdom). Wheat germ agglutinin-agarose was from ICN (Bucks, U.K.). Antibodies to insuIin receptor (serum from patient B5 or B7) was kindly provided by Dr. P. Gorden (National Institutes of Health, Bethesda, MD). Rat fibroblast cell line transfected with human insulin receptor cDNA and expressing IO6 receptors/cell was a gift of Dr. A. Uilrich (Genentech, South San Francisco, CA). Antibodies to insulin were from Miles (Paris, France). Anti-phosphotyrosine antibodies were obtained from a rabbit injected with phosphotyrosine coupled to human IgG, and the serum was affinity purified on a phosphotyrosine-agarose column (22). All re-agents for SDS'-polyacrylamide gel electrophoresis (SDS-PAGE) were from Bio-Rad or from Serva (Heidelberg, Federal Republic of Germany).
Preparation of Partially Purified Insulin Receptors-Insulin receptors were prepared as described previously (13,23) from skeletal muscle, hepatocytes, or from cells transfected with human insulin receptor cDNA (24). Briefly, tissues or cells were homogenized and solubilized in Hepes buffer (50 mM, pH 7.6), NaCl (150 mM), 1% Triton X-100, and protease inhibitors for 90 min at 4 "C by continuous stirring and centrifuged at 150,000 X g at 4 "C for 90 min. The supernatants were applied to a wheat germ agglutinin-agarose column and recycled three times. Following washing, bound glycoproteins were desorbed with 0.3 M N-acetylglucosamine in Hepes buffer (30 mM, pH 7.6), NaCl (30 mM), 0.1% Triton X-100 and stored at -80 "c until use. This preparation will be referred to as partially purified insulin receptors.
Photoaffinity Labeling and Autophosphorylation of Insulin Receptors-The photoreactive insulin analog, B2-(2-nitro,4-azidophenylacetyl)-desPheB1-insulin, was prepared as described previously (25) following an improved procedure' and iodinated in the dark to a specific activity of 200-250 pCi/Fg using the chloramine-T method (3). This photoreactive insulin is a full agonist with a 30% decrease in affinity for its receptors compared to native hormone (3). Partially purified insulin receptors were incubated for 3 h at 15 "C in the dark with photoreactive '251-labeled insulin (5 X lo-' M), conditions that permitted steady state binding. Samples were then put on ice and irradiated for 5 min under a mercury lamp (Philips HPK 125 W/L) a t a 10 cm distance. The light was passed through a glass filter (WG 345, thickness 3 mm, Schott Glaswerke, Mayence, F.R.G.) which suppresses the short UV emissions. This procedure was used in order to preserve the integrity of the kinase activity while insulin covalent labeling efficacy was maximal (data not shown). Following irradiation, phosphorylation was initiated with [-y-32P]ATP (15 p~) , MnCI2 (4 mM), and MgCI, (8 mM). After 15 min at 20 "C, the reaction was stopped by adding an ice-cold stopping solution containing NaF (80 mM) and EDTA (30 mM). Samples were immunoprecipitated overnight at 4 "C with antibodies to the insulin receptor, to insulin, and to phosphotyrosine as indicated in the figure legends. After precipitation with protein A, the pellets containing the immunoadsorbed proteins were washed three times with 1 ml of ice-cold Hepes buffer (30 mM, pH 7.6), NaCl (30 mM), and 0.1% Triton X-100. They were solubilized in 3% (w/v) boiling SDS solution containing 10% glycerol (v/v) and 0.01% bromphenol blue (w/v) without (nonreducing conditions) or with (reducing conditions) 2% @-mercaptoethanol (v/v). Analysis of samples were performed by one dimensional SDS-PAGE with a 5% or a 7.5% acrylamide resolving gel (26). In some experiments, following this first immunoprecipitation, the supernatants were subjected to a second immunoprecipitation with antireceptor antibodies and analyzed as described. The gels were stained, dried and autoradiographed by exposing them to Kodak X-Omat film. The M , values of the standards used were: myosin, 200,000; p-galactosidase, 116,000; phosphorylase b, 94,000; bovine serum albumin, 66,000; ovalbumin 45,000; carbonic anhydrase, 30,000; soybean trypsin inhibitor 20,000.
The efficiency of the covalent labeling of photoreactive insulin was measured as follows: after incubation of the partially purified insulin receptors with labeled photoreactive insulin and UV irradiation, half of the samples were treated for 30 min at 20 "C at pH 5.0 to dissociate noncovalently bound insulin. Then, insulin receptors were separated from unbound hormone using polyethylene glycol (27). The efficiency of the covalent labeling was estimated from the ratio of counts present in the precipitates treated or not at acid pH.
Two-dimensional electrophoresis was performed as described (18). Following one-dimensional electrophoresis in 5% polyacrylamide under nonreducing conditions, the entire lane was cut out, rinsed in 0.125 M Tris-HC1 (pH 6.8), 0.1% SDS at room temperature. The lane was placed atop a second SDS gel in 0.125 M Tris-HC1 (pH 6.8) and overlaid with 0.1 M DTT. Electrophoresis was then performed into a resolving gel with an acrylamide concentration of 7.5%.
Cell Surface Labeling of Insulin Receptors-Cells transfected with human insulin receptor cDNA were grown to confluence in 150-mm Petri dishes, washed twice with phosphate-buffered saline, removed gently with a rubber scraper, and resuspended in 2.5 ml of phosphatebuffered saline containing 6 units of lactoperoxidase, 20 units of glucose oxidase and 2 mCi of Na'"1. At times 0, 10, and 20 min, 140 pl of 1 M glucose was added. At 30 min, the reaction was stopped by three washes in phosphate-buffered saline; cells were solubilized and the cell extracts immunoprecipitated as described for the biosynthetic labeling.
Hepatocytes were incubated at 15 "C for 3 h in the dark with " ' 1photoreactive insulin without or with unlabeled insulin M). Cells were irradiated as previously described, solubilized, and extracts exposed to normal serum or antibodies to insulin receptor.

RESULTS
Discriminating Immunoprecipitation of Insulin Receptors following Functional Labeling-Taking advantage of the distinct biological properties of the insulin receptor a-and psubunits, a selective labeling of each subunit was obtained. The receptor a-subunit, which contains the hormone binding site, was tagged with photolabeled 'Z51-insulin, while the @subunit was labeled by means of its autophosphorylation. To know whether all receptors were labeled on both subunits, samples were immunoprecipitated with the following antibodies: 1) antibodies to insulin receptors, which precipitate all insulin receptor species; 2) antibodies to phosphotyrosine residues, which recognize receptors having a tyrosine residue phosphorylated; and 3) antibodies to insulin, which immunoprecipitate hormone-receptor complexes, where insulin is covalently bound. Both the immunoprecipitates and the supernatants were analyzed by SDS-PAGE under reducing conditions. As shown in Fig. 1, antireceptor antibodies precipitated two bands, a 130,000-Da protein corresponding to the a-subunit (labeled with 'zsI-insulin) and a 95,000-Da subunit identified as the receptor @-subunit (labeled with "P). All the insulin receptors present in the samples (lane B or E ) were immunoprecipitated, since no radioactivity remained in the corresponding supernatants (lanes B' and E'). The opposite situation was obtained when immunoprecipitation was performed with control serum (lanes A and D), all the receptors remaining in the supernatants (lanes A' and D'). When samples were exposed to antibodies against phosphotyrosine, all the phosphorylated forms of the receptor were precipitated since the labeling of the @-subunit was similar to that obtained with antireceptor antibodies (lane C compared to lane B ) and no labeled 95,000-Da band remained in the supernatant (lane C'). By contrast, a significant amount of the labeled 130,000-Da band was still present in this supernatant, indicating that some receptor population with covalently bound insulin was not phosphorylated. When antibodies to insulin were used, nearly all the radioactivity present in the a-subunit was recovered in the precipitate (lane F), while most of the labeled @-subunits were recovered in the supernatant. This is expected, since the efficiency of the covalent labeling was found to be approximately 30% (results obtained with three different insulin receptor preparations).  (lanes C, D, G and H), or together with "'1-photoreactive insulin and [y- "PIATP (lanes B and F ) . Samples were precipitated with antibodies to insulin receptor (lanes A, B, D, E, F, and H) or antibodies to phosphotyrosine (lanes C and C) as described in Fig. 1, and analyzed by SDS- PAGE in reducing (lanes A-D) or nonreducing (lanes E-H) conditions. OR, origin.

The High Molecular Weight Insulin Receptor Species Is the Major Phosphorylated Form in Partially Purified Receptor
Preparations-We next wanted to know whether all the oligomeric receptor forms were equally phosphorylated. In Fig.  2, we compared insulin receptors labeled with "'I-photoreactive insulin or with 32P only or together with '2'I-photoreactive insulin and 32P, SDS-PAGE being performed both under nonreducing or reducing conditions. When receptors were labeled with '251-insulin alone and samples analyzed under reducing conditions, one band was obtained with 130,000 Da, which we identified as the insulin receptor a-subunit (lane A ) . Under nonreducing conditions (lane E ) , two major bands with M, higher than 300,000 and a minor one with Mr approximately 230,000 were visualized, which correspond to oligomeric receptor forms. Those bands were specific, since no labeling could be found in the presence of an excess of native insulin or when samples were immunoprecipitated with control serum (data not shown). When receptors tagged with 12'I-insulin were subjected to phosphorylation with [-y-"P] ATP before a similar analysis, a labeled band appeared at but only the highest one was heavily labeled. This experiment also shows that the different oligomeric phosphorylated forms of the receptor were similarly identified by anti-phosphotyrosine and antireceptor antibodies.

The High Molecular Weight Insulin Receptor Species Is the Major Phosphorylated Form in Live
Cells-We next extended our study to live cells (HIR cells) using another "double labeling" of insulin receptors. Cells were first labeled biosynthetically with [3sSS]methionine for 12 h, and then exposed to insulin for 10 min to allow autophosphorylation of insulin receptors by endogenous ATP. They were then solubilized and the cell extracts subjected to immunoprecipitation with antibodies to insulin receptor or to phosphotyrosine. As shown in Fig. 3, lanes A  cells with antireceptor antibodies three bands were found under reducing conditions: the heavily labeled receptor a-and @-subunits and the less intensely labeled insulin receptor precursor with an apparent molecular mass of 200,000 Da (23). In cells not exposed to insulin, those bands were absent when samples were immunoprecipitated with antiphosphotyrosine antibodies (Fig. 3, lane B ) . By contrast, when cells have been exposed to insulin for 10 min, a large proportion of insulin receptors were phosphorylated on tyrosine residues as shown by the appearance of 95,000-and 130,000-Da receptor subunits following precipitation with anti-phosphotyrosine antibodies (Fig. 3, lane D). More important, when the same samples were analyzed under nonreducing conditions, only the highest band was clearly visible when samples were precipitated with anti-phosphotyrosine antibodies (Fig. 3, lane   H), while all the oligomeric forms were present in samples precipitated with antireceptor antibodies (Fig. 3, lanes E and GI. The precise molecular composition of the oligomeric receptor forms is difficult to define, since in this range molecular weight determinations are not reliable for large hydrophobic glycoproteins. Indeed, the intermediary form seen under nonreducing conditions has been reported to be either a 2 @ (17) or an@@', where @' would represent a proteolytically derived @subunit fragment (21). To discriminate between those two possibilities, we performed the next analysis. Immunoprecipitates of ["S]methionine labeled receptor were subjected to SDS-PAGE under nonreducing conditions. The lane of interest was cut out and placed horizontally atop a second gel and subjected to electrophoresis in the presence of 0.1 M DTT to reduce disulfide bonds (Fig. 4). Under these conditions, the two highest bands were separated into two bands with M , 95,000 and 130,000 corresponding to a-and @-subunits, respectively. For the two bands, the relative amounts of radioactivity in the two subunits were different. Thus, from the highest band, we obtain under reducing conditions twice as Disulfide bond reduction was carried out during the electrophoresis. When the radioactivity of the bands corresponding to the a-and &subunits was measured, the following counts/min were obtained for, respectively, a2&, a2& a2, and free @subunits: in the a-subunit, 3680,3527, 3120, and 30; in the 8-subunit, 7110,4376,540,9500. OR, origin. much radioactivity in the 0as in the a-subunit. This was also observed in the gel shown in Fig. 3, lane A. Since the aand the @-subunits contain, respectively, 9 and 20 methionine residues (10,11), these results indicate that the composition of the highest form is a2p2. By contrast, the intermediate form gave rise to two bands (the a-and @-subunits) containing the same amount of counts, indicating that this form corresponds to a?@. As expected, the lowest form migrated under reducing conditions as a-subunits only, and is therefore identified as a 2 . It should be noted that the amount of the two species, a& and a2@, was similar. Some free @-subunits were also visible.
The possibility of an an@@' composition for the intermediate unreduced form appears unlikely, since no labeled phosphoprotein could be observed at a molecular weight of 45,000 (even after a long exposure of the autoradiogram, data not shown). However, the methionine residues are mainly located in the cytoplasmic tail of the insulin receptor @-subunit (10, l l ) , and thus the 8' fragment would not be heavily labeled with methionine. Therefore, we performed cell surface iodination, which labels the insulin receptor at the level of the extracellular portion of the @-subunit (which would give rise to the @' fragment) and the entire a-subunit. Using this labeling technique, we found under nonreducing conditions the three major molecular species, which we identified previously using biosynthetic labeling as a2B2 (the highest), a2/3 (intermediate), and a2 (the lowest) (Fig. 5). Some free 0subunits were also visible. The absence of a polypeptide with M , 45,000 under reducing conditions confirms that the intermediate form corresponds to a28 and not to a2@@'.
To verify that the a2@ and az receptor structures are not anomalies resulting from abnormal synthesis or degradation in the particular transfected cell line used in most of the experiments reported above, normal cells were studied in the experiments illustrated in Fig. 6. First, when hepatocytes were labeled with '"I-photoreactive insulin, the a2P2, a2@, a2 insulin receptor species were observed under nonreducing conditions (Fig. 6, lane B ) , this labeling being specific since it was not found when the labeling was performed in the presence of an excess of unlabeled insulin ( l a n e A ) . Second receptors from skeletal muscle were phosphorylated and precipitated with antireceptor antibodies, the species was markedly labeled, and minute amounts of a2/3 and free 8subunits were also visible (Fig. 6, lane D). No labeled proteins were detected with a control serum (lane C).

DISCUSSION
In our experiments with partially purified preparations, the insulin receptor a-subunit was labeled with 1251-photoreactive hormone before autophosphorylation with [y-32P]ATP. Subsequently, by the use of discriminating immunoprecipitations it was possible to extract, with an anti-insulin antibody, receptor species occupied by the hormone and, with antiphosphotyrosine antibodies, phosphorylated insulin receptors. Our experiments show that a population of insulin receptors carrying covalently bound insulin is not phosphorylated on tyrosine residues. These results suggest that either those insulin receptors were dephosphorylated or, as shown by O'Hare and Pilch (21,28), that some receptor forms with intact hormone binding capacity have lost autophosphorylating ability. Analyses of samples in nonreducing conditions were in favor of the second hypothesis, since it was mainly the highest molecular weight species, identified as a2& that was autophosphorylated. By contrast, the intermediate species, recognized as a2@, was able to bind insulin but was not autophosphorylated.
The experiments discussed so far were performed using partially purified, solubilized insulin receptors, which were phosphorylated in vitro following exposure to photoreactive insulin and UV irradiation. To prevent artefactual generation of some oligomeric forms, we have used a short (5 min) solubilization procedure in SDS, since it has been shown that longer incubations give rise to an increased amount of inter-mediate oligomeric forms (29). For the same reason, freshly prepared insulin receptor preparations were routinely used, since storage at -70 "C increases the appearance of reduced insulin receptor forms (18).
In broken cell systems, the two functions of the insulin receptor (kinase activity and hormone binding) do not present the same sensitivity to proteolysis, kinase function being more labile than binding function (20,21,30). To verify that the different high molecular weight insulin receptor forms correspond to native receptor species and that they did not lose their autophosphorylating properties during preparation, we performed a second series of experiments in live cells expressing a high number of insulin receptors after transfection with human insulin receptor cDNA (HIR cells). In this case, insulin receptors were biosynthetically labeled with [35S]methionine, and phosphorylation occurred "naturally" by means of the endogenous unlabeled ATP pool. The phosphorylated insulin receptor species could subsequently be separated from the entire receptor pool using anti-phosphotyrosine antibodies. This approach permitted us to study receptor autophosphorylation in live cells, thus excluding the possibility that the receptor kinase activity could have been partly destroyed during purification. In these experiments, anti-phosphotyrosine antibodies precipitated only the a2P2 insulin receptor form, despite the fact that there was a nearly equal amount of a z~2 and a2B insulin receptor species as shown by methionine labeling. If the a2/3 form were a degraded product of the a2P2 species, it should also have been precipitated by the antiphosphotyrosine antibody since the remaining 8-subunit should be phosphorylated. Therefore, these observations indicate that a2p insulin receptors exist in live cells, and that they are not artificially induced degradation products of the a2P2 receptor, and that they were not autophosphorylated. It should be noted that although the a28 receptor does not appear to undergo autophosphorylation we cannot exclude that this species can phosphorylate cellular proteins or initiate intracellular signals. Our recent demonstration (31) that antibodies to intracellular receptor domains stimulate the receptor substrate phosphorylation capacity without modifying receptor autophosphorylation makes this certainly a reasonable possibility.
The precise molecular composition of the oligomeric receptor species appears to vary depending upon the tissue and the receptor extraction procedure. In some studies, intermediary receptor forms appear as either a& or a2PP', in which 8' would be a truncated &subunit (20,21,28,30). In our preparations, detectable degradation of the 8-subunit does not seem to occur, since we could not find a labeled band in the 45-kDa region either with methionine labeling or with cell surface iodination. The variation between our results and those reported by others (20,21,28,30) could be explained by the different preparations used. In our study insulin receptors were solubilized directly from intact tissues without using an intermediary step of membrane preparation. In contrast, in the studies mentioned, insulin receptors were purified by a longer preparation procedure, which consisted of placental membrane preparation, solubilization, Sephacryl 400 chromatography, wheat germ agglutinin chromatography, and purification of the different forms on mono Q chromatography.
Note that using cell surface labeling, we found a significant amount of a polypeptide which we identified as a2 based on its tagging with l2'1-photoreactive insulin and its subunit composition. Knowing that the a-subunit is not a transmembrane glycoprotein (10, ll), the occurrence of a a2 species associated with the cell surface must imply that the a-subunits are withheld by the transmembrane P-subunits through noncovalent interactions.
In most studies on possible alterations of the insulin receptor kinase, quantitation of kinase activity is normalized to hormone binding capacity (12-14,32). In light of the different sensitivity to degradation of the two insulin receptor functions, the validity of this mode of expression could be questioned. The results reported here show clearly that a subpopulation of insulin receptors is able to bind insulin without being autophosphorylated. Furthermore, this does not seem to be due to artefacts in insulin receptor preparations, since we found an identical receptor labeling pattern using live cells. The precise mechanism underlying the appearance of the difference receptor species is not known. However, provided that receptor degradation products are not generated during the experimental procedure, it remains valid to correlate kinase activity to insulin binding capacity. A lower kinase activity expressed per binding unit reflects either a higher number of an@ compared to a2P2 oligomeric forms or a decreased intrinsic activity of the azP forms. Both situations lead to a decrease in the amount of "functional" receptors, and could contribute to a reduced insulin action. In this context, it would be of interest to define physiological and/or pathological factors which are able to interfere with the relative abundance of a& and a& receptor forms. Another possibility which has also to be considered is that the aZP receptor form impairs the signaling of the a& receptor. This situation would be reminiscent of the inhibition of normal receptor function by kinase deficient insulin receptors (33).
Finally, the results reported in this paper, that in live cells only the a& oligomeric form is capable of autophosphorylation, add further support to the idea that the a&aP interaction is critical for receptor activation and autophosphorylation. In purified preparations, the dithiothreitol reduction of the tetrameric receptor into CUP dimers is accompanied by a disappearance of insulin-dependent autophosphorylation, and those dimers need to reassociate, but not necessarily covalently, to express insulin-activated kinase activity (34-36). Our data are in accord with these findings and indicate that despite an equal hormone binding to the receptor a-subunits, the presence of only one P-subunit in the a@ receptor is not sufficient to induce autophosphorylation.