The Role of Insulin Receptor Autophosphorylation in Signal Transduction*

We have examined the role of autophosphorylation in insulin signal transmission by oligonucleotide dl-rected mutagenesis of seven potential tyrosine autophosphorylation sites in the human insulin receptor. Chinese hamster ovary cells transfected with these receptors were analyzed for insulin stimulated 2-deox-yglucose uptake, thymidine incorporation, endogenous substrate phosphorylation, and in vitro kinase activity. We transduction. 960 capabilities

The effects of insulin are mediated through a transmembrane receptor, a glycosylated tetramer composed of two extracellular a subunits linked by disulfide bonds to two transmembrane /3 subunits. After insulin binds to the cy subunits, the cytoplasmic protein tyrosine kinase encoded by the /3 subunit is activated (1). The tyrosine kinase activity was abolished by the mutation of the lysine (at position 1018) in the ATP-binding domain. This mutation rendered the receptor biologically inactive; the receptor no longer mediated insulin stimulated glucose uptake, thymidine incorporation, S6 kinase activation, receptor down-regulation, or any insulin mediated response examined (2)(3)(4)(5)(6). These studies demonstrated that the protein tyrosine kinase activity was essential for mediating the effects of insulin. The activated kinase phosphorylates itself exclusively on tyrosine residues (7-9). This autophosphorylation enhances protein tyrosine kinase activity toward exogenous substrates and renders the kinase active in the absence of insulin (10). The major sites of autophos-* This work was supported by the National Institutes of Health Grant DK35158 (to 0. M. R.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisenent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ TO whom correspondence should be addressed: Program in Molecular Biology, Sloan Kettering Inst., Cornell Graduate School of Fax: 212-639-2861.
phorylation, tyrosine residues 1146, 1150, 1151, 1316, and 1322, were identified by protein microsequencing of phosphorylated tryptic peptides (12,13). Phosphorylation of tyrosine residues 1146, 1150, and 1151 was correlated with the activation of the human insulin receptor (hIR)' protein tyrosine kinase (14,15). Tyrosine residues 1146 and 1150 have been mutated to phenylalanine residues individually (16,17), and tyrosine residue 1150 has been mutated in combination with tyrosine residue 1151 (17). Mutations in this region cause aberrant kinase activity and result in partially defective receptors. The mutation Y1146F impairs the mitogenic, but .not metabolic responses, whereas the mutation Y1150/1151F impairs the metabolic, but not the mitogenic responses (16)(17)(18). Removal of 43 amino acids at the carboxyl terminus (including tyrosine residues 1316 and 1322) resulted in a receptor that was defective for metabolic responses.
Interestingly, this receptor appeared to augment mitogenic responses (19). In contrast to these partially defective receptors, mutation of the ATP binding domain (Lys-1018 to Ala) abolished kinase activity and created a biologically inactive receptor (2, 3).
Mutation of tyrosine residue 960 in the juxtamembrane domain of the hIR was reported to eliminate both metabolic and mitogenic responses (20). Although some studies did not detect autophosphorylation sites in the juxtamembrane domain (12,13,20,21), others have evidence that this region is involved in kinase activation (22) and that tyrosine residues in this region may be autophosphorylated (14,23).
In order to analyze the role of autophosphorylation in the transmission of insulin mediated responses, seven putative autophosphorylation sites of the hIR were mutated to phenylalanine residues. The simultaneous mutation of the three tyrosine residues at positions 1146, 1150, and 1151 resulted in a receptor that was unable to mediate any of the insulin stimulated responses examined. In contrast to a previous study (20), we found that the mutation of tyrosine residue 960 in the juxtamembrane domain resulted in a receptor with reduced responsiveness to insulin and normal kinase activity in uitro and in uiuo. Finally, mutations of the tyrosine residues at positions 953, 1316, and 1322 had no discernable effect on the ability of the receptor to mediate insulin responses.

22653
Oligonucleotide-directed mutagenesis was carried out as described by Kunkel et al. (24). Chemically synthesized oligonucleotides (MSKCC core facility) used for the mutagenesis are as follows, mismatches are underlined Y953F, 5' GGC ACT GAG ATA CTC AGG GTT TGA A G A A G C G A A A A G G G G C C C C A G C G G C C C A T C T G G C T G CCT CTT 3' (resulting% a n a p a I site a t bp 2977 in the hIR cDNA); AGA AGC GAA AAG CGG TCC C A G m 3 ' (EcoRI site a t bp 3604); RY-3, 5'GAG CAG ACC CTT GCC CCC TTT CCG GAA Deoxyuridine-containing single-stranded DNA t e m p h e s were grown in Escherichia coli strain CJ236 (&-lung-), and mutagenesis was accomplished in uitro as described in Ref. 24. Mutated templates were selected in E. coli strain DH5a. Phage were screened for the presence of mutations by restriction digestion of the plasmid replicative form I DNA and then verified by single-stranded DNA sequencing (Sequenase, U. S. Biochemical Corp.). A 1373-bp BglI-SpeI fragment was isolated from the mutated M13 replicative form I DNA and subcloned into pCVSVHIRc (2). The presence of the mutation was verified by restriction digestion of the expression vector (each mutation introduced a new restriction site). All cDNAs, including the K1018A and the wild type hIR (2), were cotransfected with pSV3-neo (10 and 0.5 pg, respectively) into CHO cells obtained from ATCC. Neomycinresistant cells were selected in DME supplemented with glutamine, 10% fetal calf serum, and 500 pg/ml G418 (GIBCO). Positive colonies were identified by Dynabead (Dynal Inc.) rosetting. The Dynabeads were coated with sheep anti-mouse antibody and mouse monoclonal antibody 5D9 (a gift from Dr. R. Roth, Stanford University) which recognized the extracellular domain of the hlR. Clonal cell lines were isolated with cloning rings. Cells were maintained in DME containing glutamine and 10% fetal calf serum. Preparation and manipulation of DNA was done as described in Maniatis et al. (25).
Metabolic Labeling-Confluent monolayers (approximately 1 X lo7 cells) were incubated in 3 ml of cysteine and methionine-free DME, 0.5% dialyzed BSA for 1 h. Cells were labeled with ["'S]cysteine and [,""S]methionine (200 pCi/ml, Trans%-label'", ICN, 1026 Ci/mmol methionine) for 16 h in 3 ml of DME (Cys/Met-free) with 10% fetal calf serum and glutamine. Cells were washed with PBS and then prepared for immunoprecipitation by solubilization in 0.5 ml of S buffer (50 mM HEPES, pH 7.4, 5 mM EDTA, 5 mM EGTA, 2% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 50 pg/ml leupeptin, 10 pg/ml aprotinin, and 2 pg/ml pepstatin). After incubation a t 4 "C for 2 h, cell extracts were clarified by centrifugation a t 4 "C for 20 min at 100,000 X g. Insulin receptors were immunoprecipitated from the supernatants with 2 pg of mouse monoclonal antibody CII 25.3 (which recognizes the extracellular N subunit of the hIR, Ref. 26) covalently cross-linked to protein A-Sepharose CL4B (monoclonal antibody-protein-A complex was a gift from J. C. Ahn of this laboratory). Immune complexes were washed with 60 volumes of 50 mM HEPES, pH 7.4, 0.1% Triton X-100, and 150 mM NaCl and prepared for SDS-PAGE (8). The gel was treated for 30 min with ENHANCE0 (Du Pont-New England Nuclear). Radiolabeled receptors were detected following autoradiography at -70 "C for 16 h.
Assays of Receptor Protein Tyrosine Kinase Activity-Approximately 5 X 10" cells were solubilized in 5 ml of S buffer and prepared as described above. The high speed (100,000 X g ) supernatants were applied to wheat germ agglutinin (WGA)-agarose (Vector Laboratories), and receptors were partially purified as described previously by Petruzzelli et al. (8). Briefly, 0.3 ml of WGA-agarose was washed with 12 ml of 50 mM HEPES, 150 mM NaCI, and 0.1% Triton-X-100 (WGA wash buffer). High speed supernatants were applied to the WGA agarose, and the column was washed with 12 ml of WGA wash buffer. Proteins were eluted from the column in 0.5 ml of WGA wash buffer containing 0.5 M N-acetylglucosamine (Sigma). Determination of insulin binding activity in the WGA eluates is described below. Protein tyrosine kinase assays were performed as described Ref. 9. Each 60-pl reaction contained 20 mM HEPES, pH 7.45, 150 mM NaC1, 0.1% Triton X-100, 5 mM MgCI2, 5 mM MnCL, 20 p M sodium orthovanadate, 20 p~ [y-""PI ATP at 8 cpm/fmol (Du Pont-New England Nuclear, 3000 Ci/mmol), 1 mM dithiothreitol, 100 pg/ml BSA with or without 30 nM insulin (porcine, Lilly). I n vitro protein tyrosine kinase activity toward exogenous substrates was assayed by addition of 5 pg of histone H2B (in 5 pl, Worthington) to each 6O-pl reaction. Autophosphorylation assays were performed in 60 pl using WGA-agarose eluates containing 15-20 fmol of insulin binding activ-eJM Y-2,5' ATC ACT GGC ACT GAG GAA TTC AGG GTT TGA ity (approximately 30 pg of protein as determined by the method described in Ref. 27). The partially purified receptor preparations were incubated in a reaction mixture with or without insulin for 20 min on ice after which autophosphorylation was assayed for 5 min at 30 "C. The reactions were terminated by diluting the reaction mixture to 500 pI with EDTA and sodium orthovanadate (final concentrations 10 and 1 mM, respectively). Immunoprecipitations were carried out from the diluted reaction with the monoclonal antibody CII 25.3protein A-Sepharose complex, and immune complexes were washed and prepared for SDS-PAGE as described above. Histone kinase activity was determined by incubating approximately 20 fmol of insulin binding activity (WGA eluate) in a reaction mixture with or without insulin (30 nM) for 20 min on ice; receptors were autophospborylated by incubation a t 30 "C for 5 min, then returned to ice, histone H2B (5 pg in 5 11) was added, and assays were completed at 30 "C for 5 min. Reactions were terminated by the addition of Laemli sample buffer containing SDS, dithiothreitol, and EDTA (2%, 100 mM, and 50 mM, respectively) and then subjected to 20% SDS-PAGE (28). Polyacrylamide gels were subjected to alkali treatment (1 N NaOH, 55 "C for 1 h). "P incorporation into the ( 3 subunit of the hIR and histone was resistant to this treatment, indicating phosphorylation on tyrosine residues (29). Quantitation of ?' P incorporation was accomplished with a Molecular Dynamics Phosphor Imager (laboratory of Dr. J. Massague, Sloan Kettering Institute).
Inhibition of Autophosphorylation by Antibody P4-Assays were performed as described by Herrera et al. (22). 15-20 fmol of insulin binding activity (WGA eluate) were preincubated (5 h at 4 "C) with antibody P4 or antibody P5 (2 p l ) , with or without 4 pg of peptide 4 (peptide 4 was the antigen to which antibody P4 was raised). After preincubation, reaction mixture with or without insulin (30 nM) was added. Autophosphorylation assays were carried out and analyzed as described above.
Insulin Binding and Internalization-Insulin binding activity in the WGA eluate was determined by incubating 10 pl of WGA eluate wit.h 1 ml of binding solution (DME, 0.5% dialyzed BSA, 50 mM HEPES, pH 7.45, and 12'II-insulin (150,000 cpm/ml, 1450 cpm/fmol, iodinated by chloramine T , see Ref. 2) for 4 h at 4 "C. Nonspecific binding was determined in the presence of 1 p M unlabeled insulin (<IO% of specific binding). Insulin binding activity was quantitated by immunoprecipitation (12 h, 4 "C) with monoclonal antibody CII 25.3 (this antibody does not interfere with insulin binding, Ref. 2). Immune complexes were washed three times with 1 ml of ice-cold PBS, and the associated radioactivity was quantitated in a Beckman y counter.
Cell surface insulin binding assays were performed in triplicate on subconfluent monolayers of cells (approximately 5 X lo5 cells/well, six-well plates). Cells were incubated in 1 ml of binding solution (described above) in the presence of 0-1 p~ unlabeled insulin for 4 h a t 4 "C. Cells were washed three times with 5 ml of ice-cold PBS, solubilized in 1 ml of 0.1% SDS, and the radioactivity in the lysates was quantitated in a Beckman y counter. Scatchard analysis of insulin binding data was carried out as described in Ref. 6.
Internalization experiments were performed essentially as given in Ref.
6. Approximately 1 x lo6 cells/well (six-well plate) were incubated in 1 ml of binding solution (150,000 cpm/ml '2'I-insulin) for 4 h a t 4 "C. Cells were warmed to 37 "C for 30 min to permit receptormediated endocytosis of the radiolabeled ligand. The cells were washed twice either with ice-cold PBS or with ice-cold 0.3 M sodium acetate, pH 4.5, containing 0.15 M NaCI. The cells were washed two more times with ice-cold PBS and then solubilized in 1 ml of 0.1% SDS. The radioactivity present in the lysates was quantitated in a Beckman y counter. The radioactivity associated with the cells washed at pH 7.4 in PBS represented the total cell-associated ligand (total cpm). The radioactivity associated with cells washed at pH 4.5 in sodium acetate was designated the internalized (acid-resistant cpm) '261-insulin. The acid-resistant radioactivity was normalized to total cell-associated radioactivity for each cell line. This value (acid resistant cpm/total cpm X 100) is designated "percent internalized and is an indication of the receptor's ability to mediate endocytosis of the ligand. Nonspecific binding and endocytosis was determined in the presence of 1 p~ unlabeled insulin; these values (less than 10% of the acid-resistant and total cell-associated radioactivity) were determined for both washing conditions. All assays were carried out in quadruplicate.
After 80 min, 1 pCi of ['LH]2-deoxyglucose was added with unlabeled 2-deoxyglucose (in 50 pl, Amersham Corp., 42 Ci/mmol) to a final concentration of 0.1 mM. 2-deoxyglucose uptake was assayed a t 37 "C for 10 min and then terminated by washing cells three times with icecold PBS. Cells were solubilized in 1 ml of 0.1% SDS, and the radioactivity was measured using Ecoscint (ICN).
All assays were carried out in triplicate.
'I'hymidine Incorporation-Subconfluent monolayers (1 X lO"cells/ well, six-well plate) were incubated 24 h in DME with glutamine and 0 . 5 9 dialyzed HSA to achieve quiescence. Insulin was added at various concentrations (1 pM to 1 pM). 16 h later the cells were labeled for 2 h at 37 "C with 1 pCi of ["Hlthymidine (in 50 pl of DME, Du Pont-New England Nuclear, 20 Ci/mmol). The cells were washed three times with ice-cold PRS and solubilized in 1 ml of 0.1% SDS. Cell lysates were incubated 5 min on ice with 1 ml of 20% trichloroacetic acid. The acid-precipitable counts were collected on Enzo glass fiber filters (FL-05-25). The filters were washed with 20 ml of 5% t.richloroacetic acid, dried, and the radioactivity was quantitated using Ecoscint (ICN). All assays were performed in triplicate.

RESULTS
The nomenclature and positions of the tyrosine to phenylalanine mutations in the @ subunit of the hIR are shown schematically in Fig. 1. The 2 juxtamembrane tyrosine residues, 953 and 960, were mutated independently (Y953F, Y960F)' or together (JM Y-2). Tyrosine residues 1146, 1150, and 1151, in the regulatory domain, were simultaneously mutated to phenylalanine residues; this mutation is referred to as R Y-3. Finally, the C Y-2 receptor has tyrosine residues 1316 and 1322 in the carboxyl terminus mutated to phenylalanine residues. K1018A is a lysine to alanine mutation at residue 1018 in the ATP binding domain (2). The cell line referred to as Y960F (W) or Y960F (MW) was a generous gift 'This mutated receptor, Y960F, contains a second mutation in which serine residue 962 was changed to a threonine residue. from Dr. M. F. White, Harvard University. Y960F cDNA (obtained from Dr. White) was subsequently transfected into CHO cells in our laboratory; this Y960F cell line is referred to as Y960F (R). All assays are done with the Y960F (R) cell line; exceptions are indicated.
Expression of Mutant Receptors-CHO cells were stably transfected with wild type and mutated hIR cDNAs. Clonal cell lines were examined for expression of the mutated receptors. Extracts were prepared from cells metabolically labeled with Trans""S-label'". The radiolabeled insulin receptors were immunoprecipitated with one of four antibodies (Fig. 2). Antibodies P2, P4, and P5 (indicated 2, 4, and 5 above the lanes) are antipeptide antibodies directed against amino acids 1143-1162, 952-967, and 1328-1343, respectively (22, 31). Monoclonal antibody CII 25.3 (indicated M above the lane) is directed against the extracellular domain of the hIR (26). In cell lines transfected with wild type or mutated cDNAs, all four antibodies specifically immunoprecipitated two proteins with M , values of 135,000 and 95,000, corresponding to the N and p subunits of the hIR, respectively. Monoclonal antibody CII 25.3, which recognizes only the human and not the hamster receptor (26), efficiently precipitated receptors from all cell lines except the untransfected CHO cell line (Fig. 2, lane  I). Antibody P2 precipitated receptors least efficiently, this may be explained by the preference of antibody P2 for the autophosphorylated form of the receptor (14). The mutations did not alter the affinity of the antibodies for the hIR variants, although some of the antibodies were directed against regions that were altered by the mutations ( e g . Ab P4 is directed against amino acids 952-967, and J M Y-2 is mutated at residues 953 and 960). The rate of insulin receptor biosynthesis was examined by labeling the cells metabolically with ["?3]methionine and ['"S]cysteine for 15 min, followed by incubation with unlabeled amino acids for varying lengths of time (pulse-chase analysis). All mutated receptors were synthesized at approximately the same rate. Each receptor was initially made as a 200-kDa precursor that was subsequently processed into a mature receptor comprised of a 135-kDa N subunit and a 95-kDa @ subunit (data not shown).
Insulin Binding and Receptor Number-The affinity of the receptors for insulin was assessed by Scatchard analysis of equilibrium "'I-insulin binding data (Table I). All receptors displayed dissociation constants ( K d ) for insulin between 0.72 and 0.36 nM (Table I,  Arroum indicate positions of the l:l.',-kDa CY subunit and the 9.5-kDa 6 subunit. The autoradiogram was exposed at -70 "C for I6 h.

TABLE I
Scatchard analwis of insulin binding data Equilibrium insulin binding assays were done as indicated under "Experimental Procedures." Scatchard analysis of the insulin binding data produced dissociation constants ( K , ) for insulin and its receptor.
Scatchard analysis also produced the approximate number cell surface receptors. All assays were carried out in triplicate. The values for the dissociation constants and cell surface receptor number were generated onlv for the hieh affinitv bindine sites. approximately 2000 endogenous hamster receptors/cell (6). Cells transfected with wild type and mutated insulin receptor cDNAs expressed 120,000-300,000 receptors/cell (Table I,  column 3). Thus, all cell lines transfected with hIR cDNAs expressed high levels of receptors with normal insulin binding affinities. Insulin-dependent Protein Tyrosine Kinase Activity-The protein tyrosine kinase activity of the mutated receptors was measured in vitro as insulin-dependent :I2P incorporation into the / 3 subunit of the hIR or '"P incorporation into the exogenous substrate histone H2B. Quantitation of the '"P incorporation was accomplished using a Molecular Dynamics Phosphor Imager and was normalized to femptomoles of insulin binding activity. Insulin addition resulted in a 3-to 7-fold increase in the autophosphorylation of the wild type, Y953F, Y960F, J M Y-2, and CY-2 receptors (Fig. 3). In the case of the R Y-3 receptor, insulin addition resulted in a 1.3-fold stimulation of autophosphorylation (Fig. 3). Basal levels of R Y-3 receptor autophosphorylation activity were %fold greater than basal levels observed for the wild type receptor (Fig. 3). We observed the histone kinase activity of the mutated receptors to parallel the autophosphorylation activity (Table 11). Insulin addition caused 3-to 7-fold stimulation of histone kinase activity in the wild type, Y953F, Y960F, J M Y-2, and CY-2 receptors. Insulin addition caused a 1.7-fold increase in the histone kinase activity of the R Y-3 receptors. Interest-

TABLE I1
Insulin-stimulated histone kinase activity of mutated insulin receptors Insulin receptor (20 fmol of insulin binding activity, approximately 30 pg of protein) partially purified by WGA-agarose chromatography was incubated in reaction mixture (see "Experimental Procedures") with [y-'"PJATP in the presence or absence of 30 nM insulin. Receptors were autophosphorylated for 5 min a t 30 "C followed by the addition of histone H2R ( 5 pg). Histone kinase activity was assayed for an additional 5 min a t 30 "C, and reactions were terminated and subjected to 20% SDS-PAGE. Polyacrylamide gels were subjected to alkali treatment (see "Experimental Procedures"). The alkali stable "T' incorporated into the histone was quantitated using a Molecular Dynamics Phosphor Imager.
The units for the values below are arbitrary and normalized to fmol of insulin binding activity. ingly, basal levels of histone kinase activity were the same for both the R Y-3 receptor and the wild type receptor. The R Y-3 receptor displayed elevated in vitro basal autophosphorylation activity (Fig. 3); however, there was no coincident increase in the basal levels of in vitro histone kinase activity (Table 11).
Antibody P4, which is directed against amino acids 952-967, was shown to inhibit insulin stimulated autophosphorylation of wild type insulin receptors (22). T o further characterize the mutated receptors, the partially purified receptors were preincubated with either antibody P4 or antibody P5 (antibody P5 does not affect kinase activity). Antibody P4 dramatically inhibited the insulin stimulated autophosphorylation of the wild type, Y953F, Y960F, J M Y-2, and CY-2 receptors (10-20-fold, insulin and Ab 5 versus insulin and Ab 4, Fig. 4) but weakly inhibited the activity of the R Y-3 receptor (2.7-fold, Fig. 4). The inhibitory effect of antibody P4 was specifically relieved by preincubation with an excess of the synthetic peptide (peptide 4) used to raise antibody P4 (Fig. 4). Thus, we conclude that all of the mutated receptors, except R Y-3, had essentially normal in vitro protein tyrosine kinase activity as judged by their autophosphorylation activity, histone kinase activity, and inhibition of autophosphorylation by antibody P4. In contrast, the RY-3 receptor was observed to phosphorylate itself but not histone in the absence of insulin; the kinase activity of this receptor was minimally stimulated by insulin, and antibody P4 did not inhibit autophosphorylation of this receptor.
"'I-Znsulin Internalization-The ability of the mutated receptors to mediate endocytosis of radiolabeled insulin was measured by loading cell surface receptors with l2"1-insulin for 4 h a t 4 "C. The cells were warmed to 37 "C for 30 min, allowing receptor mediated endocytosis to occur. Cells were washed with PBS (total counts/min) or with 0.3 M sodium acetate, pH 4.5, and 150 mM NaCl (acid-resistant counts/ min) which released the radiolabeled insulin from cell surface receptors (6). The acid-resistant radioactivity is taken to represent the number of receptors internalized during the 30min incubation a t 37 "C and was normalized to total cellassociated radioactivity. 40-50% of the total cell-associated radioactivity was resistant to the acid treatment in the wild type, Y953F, Y960F, and C Y-2 cell lines (Table 111). 35% of the total cell-associated radioactivity was resistant to acid treatment in cells expressing the J M Y-2 receptor. In cells with 2 pg of monoclonal antibody CII 25.3 and subjected to 7.5% SDS PAGE. The autoradiogram was exposed for 12 h a t -70 'C.

TABLE 111 Internalization of mutant insulin receptors
insulin for 4 h a t 4 "C, and internalization was assayed as described under "Experimental Procedures." Cells were warmed to 37 "C for 30 min to allow receptor mediated endocytosis to occur. Cells (5 X 10") were washed with PBS (total 'Z51-insulin counts/min) or with 0.3 M sodium acetate pH 4.5, 150 mM NaCl (acid-resistant "'I-insulin counts/min). The acid-resistant radioactivity was normalized to total cell-associated radioactivity for each cell line. This value (acid resista n t radioactivity/total radioactivity X 100) is termed percent internalized and is an indication of the receptor's ability to mediate endocytosis of ligand. Nonspecific binding and endocytosis was assayed in the presence of 1 p~ insulin; these "nonspecific" counts were subtracted from the "specific" counts (assayed in the absence of unlabeled insulin). All assays were carried out in quadruplicate. The results shown here are representative of two to four separate experiments. expressing the K1018A and the R Y-3 receptors only 18% of the total cell-associated radioactivity was resistant to the acidtreatment. The percent internalized values for the wild type and the K1018A cell lines agree with previous reports (30). Kinetic analysis showed no difference in the rate of internalization of the Y953F, Y960F (W), Y960F (R), and C Y-2 receptors (data not shown). The R Y-3 receptor was not internalized a t any (15, 30, 45, or 60 min) time point (data not shown). In summary, all mutated receptors mediated essentially normal levels of insulin internalization except the K1018A and the R Y-3 receptors, which permit minimal levels of insulin internalization.

Cells expressing the mutated receptors were incubated with
Biological Actiuity-The stimulation of 2-deoxyglucose uptake by insulin was examined by treating the cell lines with various concentrations of insulin for 30 min. The amount of [~'H]2-deoxyglucose incorporated into the cells during a 10min period was expressed as percent of maximal incorporation. Previous studies (2-6) demonstrated that high levels of wild type receptor expression increased the sensitivity of CHO cells to insulin, whereas high levels of K1018A receptor expression did not. Untransfected parental CHO cells and cells expressing the K1018A mutant (data not shown) were half-maximally stimulated a t 40 nM insulin (Fig. 5). Cells expressing wild type, Y953F, and C Y-2 receptors were halfmaximally stimulated by approximately 1 PM of insulin (Fig.  5, panel A). Cells expressing the Y960F (R) and JM Y-2 receptors were half-maximally stimulated by approximately 0.5 nM insulin (Fig. 5, panel I?). Cells expressing the R Y-3 receptor were half-maximally stimulated a t 14 nM, approximately the same concentration at which half-maximal stimulation occurred for untransfected parental CHO cells (Fig.  5 , panel C) and cells expressing K1018A receptors (data not shown).
The stimulation of ["Hlthymidine incorporation by insulin was measured after exposing the cells to insulin for 16 h. Cells expressing wild type, Y953F, and CY-2 receptors were halfmaximally stimulated by approximately 1 PM insulin (Fig. 6,  panel A). Expression of similar numbers of Y960F and JM Y-2 receptors resulted in half-maximal stimulation by 0.15 and 0.01 nM insulin (Fig. 6,panelB). Parental CHO cells, K1018A, and R Y-3-expressing cells were stimulated half-maximally by approximately 1 nM insulin (Fig. 6, panel C).
Insulin stimulated tyrosine phosphorylation of the hIR @ subunit and pp185, a well characterized endogenous substrate (20), was analyzed in uiuo. These results paralleled the results for insulin stimulation of thymidine incorporation and 2deoxyglucose uptake (Fig. 7). Cells were treated with varying concentrations of insulin. The extracts were immunoblotted and reacted with a monoclonal anti-phosphotyrosine antibody pY20. Fig. 7 shows the tyrosine phosphorylation of the hIR @ subunit and pp185 at 1 nM insulin in cells which expressed wild type, Y935F, Y960F(MW), and C Y-2 receptors. Cells expressing the J M Y-2 receptor show tyrosine phosphorylation of the hIR @ subunit and pp185 a t 10 nM insulin. The cells expressing the K1018A and RY-3 receptor show tyrosine phosphorylation of pp185 a t 1 p~ insulin; however, these cells do not show tyrosine phosphorylation of the hIR @ subunit at that insulin concentration.
In summary, for all in uiuo responses examined, cell lines which expressed the Y953F and the C Y-2 receptors responded to insulin with the same sensitivity as cells which expressed the wild type receptors. Cell lines which expressed the JM Y-2 and Y960F receptors were 10-fold less sensitive than cells lines which expressed the wild type receptor; however, they were 10-fold more sensitive to insulin than untransfected CHO cells or cells expressing the K1018A receptors. Expression of high levels of the R Y-3 receptors did not enhance the sensitivity of transfected cells to insulin.

DISCUSSION
Autophosphorylation of the insulin receptor activates the intrinsic protein tyrosine kinase and renders the kinase independent of insulin (1). In order to determine the role of autophosphorylation in insulin action, we mutated seven putative sites of tyrosine autophosphorylation to phenylalanine residues. Cell lines transfected with the mutated cDNAs were isolated and shown to express similar numbers of insulin receptors. The mutations did not affect the affinity of the receptor for insulin or the rate of receptor synthesis. The mutated receptors fell into three groups: those that showed wild type signaling (Y953F and CY-2), those that showed reduced signaling (Y960F and JM Y-2), and those that did not signal (R Y-3).
The mutation of tyrosine residues 1316 and 1322 at the carboxyl terminus (C Y-2) did not alter any of the insulin- blasts. They observe these receptors to have increased sensitivity to insulin for mitogenic signaling, yet t.hey are defective for metabolic signaling (19). Additionally, Takata ~t al. (33) have shown that mutation of tyrosine residues 1316 and 1322 (identical to our CY-2 mutation) had no effect on metabolic signaling, yet augmented mitogenic signaling in Rat 1 fibroblasts. I t seems that the carboxyl terminus of the insulin receptor plays a role in signal transduction in Rat 1 fibroblasts but not CHO cells. I t would be interesting to determine the function of the carboxyl-terminal region in a human cell line.
Autophosphorylation of residues 1316 and 1322 may be important for aspects of insulin action not examined in our study, since these tyrosine residues are phosphorylated in vivo (21, 34). Of note, tyrosine residue 1322 of t h e h I R is located in a putative consensus sequence involved in binding phosphatidylinositol 3-kinase (Pt,dIns %kinase) ( 3 5 ) . T h e association of tyrosine kinases with PtdIns %kinase was correlated with the ability to mediate transformation ( 3 5 ) . PtdIns 3kinase was shown to associate with the autophosphorylated form of several tyrosine kinases, and autophosphorylation was essential for this association ( 3 5 ) . T h e C Y-2 receptor, however, is able to activate and associate with PtdIns

3-
kinase as well as the wild t-ype receptor.:' I t seems that these tyrosine residues are not necessary for the association or activation of Ptdlns 3-kinase. This agrees with functional studies presented here.
Antipeptide antibody P4, directed against the juxtamembrane region, inhibits insulin-stimulated autophosphorylation. This suggests that this region plays a role in kinase activation (22). I t was postulated that antibody P4 mediated inhibition by blocking autophosphorylation of tyrosine residue 953 or 960. However, we find that receptors mutated at those residues (Y953F, YSGOF, and J M Y-2) are still inhibited by ant,ibody P4. This suggests that the inhibitory effects of antibody P4 are not mediated by blocking phosphorylation of those residues. In addit.ion, t,he fact t.hat the Y960F and the J M Y-2 mut.ations did not affect the ability of the receptor to be inhibited by the antibody emphasizes their similarity to wild t-ype receptors with regard to kinase activity. Mutation of tyrosine residue 953 caused a slight decrease in insulin internalization; however, this did not affect the ability of the Y953F receptor to transduce insulin mediated responses. In cont.rast to the mutation of tyrosine residue 953, which had no demonstrable effect on the receptor, mutation of tyrosine residue 960 resulted in a receptor with reduced signaling capabilities (Ref. 20  these results suggest that the signaling defects of the .J.M Y-2 receptor (Y953/960F) and the Y960F (Y960F/S962T) receptor are due to the mutation at tyrosine residue 960. Since mutation of tyrosine residue 960 reduced the signaling capabilities of the receptor without altering kinase activity, this region must. have a role in signal transmission that is independent of its role in kinase activation. Furthermore. this demonstrates that normal signding by the insulin receptor involves processes other than kinase activity. However, it is important to emphasize that signaling by the Y96OF and ,134 Y-2 receptors was reduced, not abolished. We observe signaling to be abolished only by mutations that interfere with kinase activity (K1018A and R Y-3).
The third group contains the R Y-3 mutation which rendered the receptor biologically inactive. We suggest that the RY-3 mutant is defective because it is unahle to phosphorylate the critical tyrosine residues needed to activate the kinase toward endogenous substrates and that this activation is necessary for signaling to occur. In support of this h?rpothesis, we observe diminished histone kinase activity in the H Y-3 receptor. Autophosphorylation of the receptor has been dernonstrated to increase the kinase activity toward exogenous substrates ( l o ) , and phosphorylation of tyrosine residues 1146, 1150, and 1151 was shown to correlate with this activation (14, 15). We believe it is unlikely that the R Y-3 receptor is defective because it fails to internalize. Data from other laboratories (36) suggest that internalization-deficient receptors still mediate insulin stimulated glucose uptake and thymidine incorporation. I t is interesting that two signals appear to be involved in insulin receptor internalization; one signal involves the juxtamembrane region, since deletions in this region (30, 37) result in receptors unable to internalize. The second signal requires either kinase activity ( 6 ) or autophosphorylation on tyrosine residues 1146. 1150, and 11.51 o r both (Ref. 38 and data presented here).
Both t.he R Y-8 receptor and the K1018A receptor (mutated in the ATP binding domain) are biologically inactive; however, only the R Y-.? receptors have elevated basal in ritro autophosphorylation activity (2, 3). Curiously. the 13 1'4 receptor has no concomitant increase in basal autophosphorylation in uiuo. Similar findings were reported by Ellis et al.
(17) and Zhang et al. (39) using receptors mutated at two or three of these tyrosine residues (1146, 1150 and 1151). High levels of basal in vitro autophosphorylation suggest that the kinase has become insulin-independent. This may be caused by a conformational change generated by the mutation of tyrosine residues 1146, 1150, and 1151. In support of this hypothesis, the basal in vitro autophosphorylation activity of the mutated (RY-3) receptor was not efficiently inhibited by antibody P4, suggesting an altered receptor structure. Antibody P4 prevents the insulin-stimulated activation of unphosphorylated receptors, but it does not inhibit kinase activity if receptors are autophosphorylated prior to incubation with antibody (22). We cannot rule out the possibility that the mutations in RY-3, which cause increased basal autophosphorylation, produce a constitutively active receptor. This constitutive activity may lead to a down-regulation of subsequent signaling events, thereby creating an insulin insensitive cell. If down-regulation involved the activation of a tyrosine phosphatase, the difference between basal autophosphorylation levels in uiuo and in vitro could be explained.
Autophosphorylation of receptor tyrosine kinases has a t least two functions; the first is to modulate kinase activity, and the second is to enhance the affinity for substrates of the kinase (Refs. 35 and 40 and references therein). Autophosphorylation has been demonstrated to enhance the affinity of the platelet-derived growth factor receptor for PtdIns 3-kinase and the epidermal growth factor receptor for phospholipase C--y (Ref. 35 and references therein).Autophosphorylation sites that enhance affinity for kinase substrates have yet to be defined in the insulin receptor. However, the insulin receptor does have autophosphorylation sites (tyrosine residues 1146, 1150, and 1151) that modulate kinase activity. Could these same sites enhance interactions with substrates? It is interesting to note that the mutation of tyrosine residues 1146, 1150, and 1151 interferes with PtdIns 3-kinase association and activation.3 In conclusion, mutation of tyrosine residues 953 or 1316 and 1322 did not alter the ability of the hIR to transduce the insulin signal, whereas mutation of tyrosine residue 960 reduced, but did not abolish, the signaling capabilities of the receptor. Mutation of tyrosine residues 1146, 1150, and 1151 resulted in a biologically inactive receptor, indicating that the insulin receptor can be functionally inactivated not only by mutation of the ATP binding domain but also by removal of key autophosphorylation sites.