The insulin receptor activation process involves localized conformational changes.

The molecular process by which insulin binding to the receptor alpha-subunit induces activation of the receptor beta-subunit with ensuing substrate phosphorylation remains unclear. In this study, we aimed at approaching this molecular mechanism of signal transduction and at delineating the cytoplasmic domains implied in this process. To do this, we used antipeptide antibodies to the following sequences of the receptor beta-subunit: (i) positions 962-972 in the juxtamembrane domain, (ii) positions 1247-1261 at the end of the kinase domain, and (iii) positions 1294-1317 and (iv) positions 1309-1326, both in the receptor C terminus. We have previously shown that insulin binding to its receptor induces a conformational change in the beta-subunit C terminus. Here, we demonstrate that receptor autophosphorylation induces an additional conformational change. This process appears to be distinct from the one produced by ligand binding and can be detected in at least three different beta-subunit regions: the juxtamembrane domain, the kinase domain, and the C terminus. Hence, the cytoplasmic part of the receptor beta-subunit appears to undergo an extended conformational change upon autophosphorylation. By contrast, the insulin-induced change does not affect the juxtamembrane domain 962-972 nor the kinase domain 1247-1261 and may be limited to the receptor C terminus. Further, we show that the hormone-dependent conformational change is maintained in a kinase-deficient receptor due to a mutation at lysine 1018. Therefore, during receptor activation, the ligand-induced change could precede ATP binding and receptor autophosphorylation. We propose that insulin binding leads to a transient receptor form that may allow ATP binding and, subsequently, autophosphorylation. The second conformational change could unmask substrate-binding sites and stabilize the receptor in an active conformation.

The molecular process by which insulin binding to the receptor a-subunit induces activation of the receptor @-subunit with ensuing substrate phosphorylation remains unclear. In this study, we aimed at approaching this molecular mechanism of signal transduction and at delineating the cytoplasmic domains implied in this process. To do this, we used antipeptide antibodies to the following sequences of the receptor @-subunit: (i) positions 962-972 in the juxtamembrane domain, (ii) positions 1247-1261 at the end of the kinase domain, and (iii) positions 1294-1317 and (iv) positions 1309-1326, both in the receptor C terminus. We have previously shown that insulin binding to its receptor induces a conformational change in the @-subunit C terminus. Here, we demonstrate that receptor autophosphorylation induces an additional conformational change. This process appears to be distinct from the one produced by ligand binding and can be detected in at least three different @-subunit regions: the juxtamembrane domain, the kinase domain, and the C terminus. Hence, the cytoplasmic part of the receptor /3subunit appears to undergo an extended conformational change upon autophosphorylation. By contrast, the insulin-induced change does not affect the juxtamembrane domain 962-972 nor the kinase domain 1247-1261 and may be limited to the receptor C terminus. Further, we show that the hormone-dependent conformational change is maintained in a kinase-deficient receptor due to a mutation at lysine 1018. Therefore, during receptor activation, the ligand-induced change could precede ATP binding and receptor autophosphorylation. We propose that insulin binding leads to a transient receptor form that may allow ATP binding and, subsequently, autophosphorylation. The second conformational change could unmask substratebinding sites and stabilize the receptor in an active conformation.
Insulin regulates cellular metabolism and growth through activation of its specific receptor. This is a heterotetrameric glycoprotein consisting of two extracellular a-subunits con-* This project was supported by Bayer-Pharma (Grant 89038) and by the Institut National de la SantB et de la Recherche MBdicale. 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.
$ Supported by a fellowship from the Fondation pour la Recherche MBdicale (France).
J Supported by a "Poste Vert" from the Institut National de la SantB et de la Recherche MBdicale.
7 To whom correspondence should be addressed. Tel.: 33-93-85-16-54: Fax: 33-93-92-07- 13. taining the ligand-binding site and two transmembrane @subunits carrying tyrosine kinase activity in its cytoplasmic domain (1,2). Insulin binding enhances this enzymic function with resulting receptor autophosphorylation and phosphorylation of several cellular substrates (3)(4)(5). Over the last few years, considerable evidence has accumulated in support of the idea that the receptor kinase activity is essential for all known early and late cellular responses evoked by insulin (6)(7)(8). However, the molecular mechanism by which the hormone activates the kinase remains to be determined. It has been shown that ligand binding alters the conformation of the receptor a-subunit and modifies the relative position of these two a-subunits within the a& tetramer (9, 10). Yet, a comprehensive scheme of receptor activation is still missing and we are left with several unanswered questions pertaining to the nature of the transmembrane signal and to the mechanism by which an extracellular perturbation affects a cytoplasmic enzymic activity.
In this context, an early key observation was made by Herrera and Rosen (ll), who showed that receptor autophosphorylation is associated with a conformational change in the @-subunit kinase domain in the area of the major autophosphorylation sites (Tyr 1146, 1150, and 1151). This phenomenon appears to be common to tyrosine kinase receptors, as it has been detected also in the receptors for platelet-derived growth factor (12, 13) and colony-stimulating factor-1 (14). For all these receptors, the changes were found to be strictly dependent on kinase activity and/or autophosphorylation, since they disappeared in kinase-deficient receptors (12,14,15). At least for the insulin receptor, conformational events unrelated to phosphorylation can occur. Indeed, we have shown that insulin binding induces a change in the receptor C terminus, which is independent of autophosphorylation (16). Although the general prescient feeling is that conformational changes play an important role in receptor functioning, several key issues remain unresolved. In trying to address some of these, we have used antipeptide antibodies to various @-subunit domains. In this work, we were able to distinguish between insulin-induced and phosphorylation-dependent conformational changes of the receptor, and we demonstrate that the insulin receptor may exist in at least three conformations. Moreover, we provide evidence for the idea that these changes are occurring at the level of different cytoplasmic domains of the receptor @-subunit. Finally, we show that the hormoneinduced process is preserved in a kinase-deficient receptor.

Production of Antipeptide Antibodies
Peptides corresponding to four insulin proreceptor sequences were synthesized: (i) the sequence 962-972 in the juxtamembrane portion of the receptor (3-subunit, (ii) the sequence 1247-1261 located between the kinase and the C terminus domains, and (iii) the sequences 1294-1317 and (iv) 1309-1326 in the receptor C terminus, containing one and two autophosphorylation sites, respectively (tyrosines 1316 and 1322). In this study, we used the numbering system published by Ullrich et al. (2).
Peptides were coupled to keyhole limpet hemocyanin and injected into rabbits as described before (17). Antipeptides detected by enzyme-linked immunosorbent assay were partially purified by chromatography on protein A-Sepharose. Elution was performed with 0.1 M glycine, 0.5 M NaC1, pH 2, and immediate neutralization with 1 M Tris, pH 8. Immunoglobulins were dialyzed against 20 mM Hepes, 60 mM NaCl, pH 7.5.

Insulin Receptor Purification
We used rat embryo fibroblasts transfected with an expression plasmid encoding either the human insulin receptor or a kinasedeficient human receptor mutated on lysine residue 1018. These cell lines were a gift from Dr. A. Ullrich, Max-Planck Institute h r Biochemie, Munchen, Germany.
The kinase-deficient human receptors were purified by affinity chromatography using monoclonal antibodies specific for human insulin receptors bound to Sepharose. This avoids possible complications due to the presence of endogenous (i.e. murine) native receptors. The receptors were eluted using 1.5 M MgCl2, 120 mM sodium borate, pH 7.5. Elution was immediately followed by a 7-fold dilution in 30 mM Hepes, 30 mM NaC1, 0.1% bovine serum albumin, 0.1% Triton X-100, pH 7.5 (final volume, 60 ml). A second chromatography on wheat germ agglutinin was performed to eliminate the elution buffer and to concentrate the purified receptors. Insulin-occupied 35S-Labeled Receptors-Receptors were treated with M insulin for 3 h at 15 "C. The receptors were diluted during the immunoprecipitation experiment, but the insulin concentration was maintained constant over the entire procedure.

Competition of 35S-Labeled Phosphoreceptor Immunoprecipitation by Unlabeled Native or Phosphorylated Receptors
Increasing amounts (ranging from 0.1 to 6 pmol as determined by Scatchard analysis) of unlabeled insulin receptors, phosphorylated or not, were incubated for 3 h at 15 "C with the antipeptide to sequence 1294-1317 at a final concentration of 100 pg/ml. The 35S-labeled receptors were phosphorylated and added to each sample at a constant concentration (200 fmol). The incubation was continued for 3 h at 15 "C and, after addition of protein A-Sepharose for 1 h at 4 "C, immune pellets were washed twice with Hepes/NaCl containing 0.1% Triton X-100. Pellets were resuspended in 1 ml of scintillation solution for counting.
Immunoprecipitation of Different 35S-Labeled Receptor Forms "S-Labeled receptors were used in a native, phosphorylated, or insulin-occupied state. These receptors (200 fmol/sample) were incubated for 2 h at 4 "C with increasing concentrations of partially purified antipeptides in a final volume of 50 pl. After the addition of protein A-Sepharose, the immune pellets were treated as described above.
Immunoprecipitation of Receptors Bound to '251-Insulin Insulin was labeled according to the chloramine-T method at a specific radioactivity of 200 pCi/pg (19). Kinase-deficient insulin receptors mutated at residue 1018 were incubated with iodinated insulin (0.05 pCi/200 fmol of receptor) for 3 h at 15 "C. Antipeptides were then added at increasing concentrations for 2 more hours. Protein A-Sepharose was used to immunoprecipitate the complex, and pellets were counted in a y-counter.

Immunoprecipitation of Receptors Activated in Intact Cells
Cells were labeled with [35S]methionine (0.5 mCi/l5-cm dish) and then cultured with medium containing 0.2% bovine serum albumin for 2 h at 37 "C. Insulin (10" M) or buffer alone were added to the cells for 5 min at 37 "C. Stimulation was stopped by three washes with 50 mM Hepes, 150 mM NaCl, 10 mM EDTA, 10 mM N&P207, 2 mM sodium orthovanadate, 100 mM NaF, 10% glycerol, pH 7.5. Cells were scraped using the same buffer plus protease inhibitors, and homogenization was performed by 15 passages through a 26-gauge syringe. After centrifugation at 15,000 X g during 30 min at 4 "C, pellets containing the membrane fraction were solubilized in 1% Triton X-100 for 30 min at 4 "C. After another centrifugation step, supernatants were immunoprecipitated with different antibodies previously adsorbed on protein A-Sepharose. The antibodies were used at optimal concentrations, i.e. concentrations giving the maximal difference between the receptor forms in vitro. Pellets were washed five times, and proteins were analyzed by SDS-polyacrylamide gel electrophoresis. A fraction of each supernatant was kept to determine by '251-insulin binding the receptor number in the stimulated and unstimulated conditions.

RESULTS
To determine whether @-subunit domains other than the C terminus are implied in the hormone-induced conformational change (16), we used polyclonal antipeptides against (i) the juxtamembrane domain corresponding to amino acids 962-972 and (ii) the intermediate region between the kinase and C terminus domains, i.e. the sequence corresponding to amino acids 1247-1261. We measured the ability of these antipeptides to distinguish between native and insulin-bound receptors. To do this, we used [35S]methionine-labeled receptors in immunoprecipitation experiments with increasing antibody concentrations. Results obtained with the antipeptide to sequence 962-972 are shown in Fig. 1. No striking difference was observed with the two receptor forms, indicating that these antibodies did not discriminate between occupied and nonoccupied receptors. We also found that the antipeptide to receptor sequence 1247-1261 did not immunoprecipitate the native receptors nor the insulin-bound receptors (data not shown). Thus, the insulin-induced change could be limited to the C terminus, or at least was not detected in other domains.
A possible role of this change could be to render the receptor

Insulin Receptor
Conformational Changes capable of binding ATP. Therefore, it would take place before nucleotide binding in the chronology of events involved in the molecular activation process. We hypothesized that a kinasedeficient receptor, mutated at the ATP-binding site (replacement of lysine 1018 by alanine), should undergo the ligandinduced modification. To examine this possibility, we tested the two a n t i 4 terminus antipeptides to sequences 1294-1317 and 1309-1326 in experiments with a kinase-deficient receptor, in the presence or absence of insulin. As shown in Fig. 2, the antibodies immunoprecipitated hormone-occupied receptors to a lesser extent as compared with native receptors, and at optimal antibody concentrations, the difference reached 40-50% of maximal immunoprecipitation. Thus, the kinasedeficient receptor remains capable of sensing insulin-induced conformational changes. Further, this result reinforces the notion that the observed structural change is independent of receptor autophosphorylation or ATP binding per se.
In previous studies, however, phosphorylation-induced changes have been described for the insulin receptor (11,20). Therefore, we were interested in seeing whether the phosphorylation-and insulin-dependent events we observed were the same or different ones. To answer this question, 35Slabeled receptors in their native or autophosphorylated form were immunoprecipitated by increasing concentrations of antipeptide antibodies. Fig. 3 shows the results obtained with antibodies to receptor sequence 1247-1261 (panel A ) and 962-972 (panel B ) . At every concentration tested, both antibodies immunoprecipitated the phosphoreceptor better than the native one. This indicates that the receptor epitopes recognized by these antibodies are uncovered upon autophos-  phorylation. This is particularly striking for the antipeptide to sequence 1247-1261, since this antipeptide fails to interact with native receptors. Hence, the &subunit is likely to participate not only in an insulin-induced conformational change but also in a phosphorylation-dependent one.
In our previous study (16), we were not able to demonstrate in the C terminus domain any conformational change due to receptor autophosphorylation. Here, we used phosphorylated 35S-labeled receptors in competition experiments with unlabeled receptors phosphorylated or not (Fig. 4). Using this approach, we found now that the antipeptide to sequence 1294-1317 recognized native and phosphorylated receptors differently. The shift observed between the two curves indicated clearly that the overall affinity of our antipeptide was distinct for the two receptor forms. Moreover, a fraction of approximately 25% native receptors was not recognized by the antibody. Thus, the C-terminal domain appears to be more accessible to the antipeptide to sequence 1294-1317 when the receptor is phosphorylated. In summary, during its autophosphorylation, the insulin receptor appears to undergo an important conformational change that might span the whole cytoplasmic part of the 8-subunit.
Finally, to ensure that our in vitro observation may be physiologically relevant, we performed immunoprecipitation experiments using receptors activated in intact cells. In these experiments, cells were labeled with [35S]methionine and then stimulated or not by insulin for 5 min. Receptors from solubilized cells were precipitated by different antipeptide antibodies and a monoclonal antibody specific for the human insulin receptor. Fig. 5 shows an autoradiogram of such a typical experiment. We observed that each antipeptide recognized the activated receptor better than the nonactivated one, whereas immunoprecipitation with the monoclonal antibody confirmed that we had similar amounts of the two receptors. As the stimulation took place in intact cells, the activated receptors correspond for the greatest part to phosphorylated receptors, indicating that ligand-induced in vivo autophosphorylation is accompanied by a conformational change.

DISCUSSION
There is an increasing body of evidence that the molecular mechanism leading to receptor kinase activation upon insulin binding implies conformational changes in the receptor Bsubunit. From the published data, three series of facts emerge indicating that the insulin receptor undergoes structural changes upon the following distinct events: (i) insulin binding (16); (ii) ATP binding in the presence of ions, an effect potentiated by the hormone (15); and (iii) receptor autophosphorylation (11,20). Whereas the insulin-induced conformational change was demonstrated in the C-terminal part of the @subunit, phosphorylation and ATP-induced changes have been localized in the kinase domain, near, or including the major autophosphorylation sites of tyrosine residues 1146, 1150, and 1151.
In an attempt to integrate these findings into a comprehensive molecular model and to delineate more precisely the Bsubunit parts involved in this process, we used a series of antipeptide antibodies to detect insulin-and phosphorylationdependent changes. In our previous work (16) tides directed to the receptor C-terminal sequences 1309-1326 and 1294-1317 were shown to identify an insulin-induced conformational change. Our results did not allow us to conclude anything about the existence of additional changes due to receptor phosphorylation. To address this possibility, we took advantage of receptors biosynthetically labeled with [35S] methionine and of antipeptide antibodies to several receptor domains. As a whole, our data now provide direct evidence that the insulin receptor may exist in distinct conformations following hormone binding and autophosphorylation. We would like to draw the following key conclusions. First, in contrast to the phosphorylation-induced change that can be detected at least in three regions (the juxtamembrane domain, the kinase domain, and the C terminus), the insulin-dependent process may be limited to the C terminus. This may explain results previously published by Maddux and Goldfine (15), showing that insulin-binding alone did not modify the interaction between the receptor and their antibody directed against a domain near the tyrosines 1150 and 1151. Second, the two events are clearly distinct for the following reasons: (i) they were not detected in the same domains, (ii) antipeptides to domain 1247-1261 did not interact with insulinoccupied receptors, but did immunoprecipitate the phosphorylated receptors, and (iii) antipeptide to domain 1294-1317 recognized the phosphoreceptor better than the native one ( Fig. 4) and also recognized the native receptor better than the insulin-occupied one (16).
Phosphorylation-induced changes have been demonstrated in several receptors with tyrosine kinase activity, for example in the platelet-derived growth factor receptor C-terminal domain (12,13), in the colony-stimulating factor-l receptor juxtamembrane domain (14), and in the insulin receptor kinase domain (11,15,20). However, it is difficult to explain why despite the close structural similarity between the platelet-derived growth factor receptor and the colony-stimulating factor-1 receptor, the first one undergoes a conformational change at its C terminus, whereas such an alteration was detected in the juxtamembrane domain of the colony-stimulating factor-1 receptor. Considering our results, it could be possible that all tyrosine kinase receptors are subjected to a widespread process covering the entire cytoplasmic domain. The observation of a ligand-induced conformational change, which is independent of autophosphorylation, leads us to suggest that an intermediate receptor form may exist during the activation process. This preactivated state may be evanescent in cells due to the presence of ATP and ions that lead to immediate phosphorylation. We have shown that kinasedeficient mutated receptors retain the capacity to undergo insulin-induced changes and to shift to the intermediate, preactivated but nonactive conformation. Our results add further weight to the concept that before ATP binding or autophosphorylation, an event involving a conformational change takes place. In addition, this result indicates that the kinase-deficient insulin receptor is not totally inert, but is capable of transmitting ligand-induced information across the cell membrane. This adds even more decisive evidence to the idea that the inability of this kinase-deficient receptor to generate a biological response is due to its enzymic deficiency and not to a globally altered conformation as compared with the kinase active receptor nor to a defect in the transmembrane activation mechanism.
Based on our observations and on previously published data, we would like to propose the following schematic model for the insulin receptor activation. Insulin binding to the receptor a-subunit induces a conformational change in the extracellular domain (9, IO), and a modification in the inter-action between the two receptor halves (21, 22). These changes are transmitted to the /%subunit down to its C terminus, leading to a short-lived, preactivated receptor that becomes competent to bind ATP. The ensuing receptor autophosphorylation induces a second conformational change, distinct from the first one, which affects the major part of the cytoplasmic domain. Although the precise role of these phenomena remains to be determined, we would like to suggest that they could lead to unmasking of the receptor catalytic domain and/or of binding sites for cellular proteins, allowing enzyme-substrate interactions.