Transmembrane Signaling by the Human Insulin Receptor Kinase RELATIONSHIP BETWEEN INTRAMOLECULAR p SUBUNIT tram- AND cis-AUTOPHOSPHORYLATION AND SUBSTRATE KINASE ACTIVATION*

To examine the role of intramolecular @ subunit trans- and cis-autophosphorylation in signal transduc-tion, the vaccinia virus/bacteriophage T7 expression system was used to generate insulin holoreceptors composed of a kinase-defective half-receptor precursor (abAIK or (XBAm.ACT) and a kinase-active half-receptor precursor (a@ACT or a#?wT). In the aBAm-a@ACT hybrid insulin receptor, insulin stimulated a 20-fold increase in intramolecular B subunit trans-phosphorylation, whereas cis-phosphorylation increased only 3-fold over the basal state. Similarly, in the (Y@WT-(YBAm.ACT hybrid insulin receptor, insulin stimulated trans-phos-phorylation approximately 30-fold and cis-phos- phorylation only %fold over the basal state. Although cis-phosphorylation of the kinase-functional aB half- receptor was observed within these hybrid receptor species, this was not sufficient to stimulate exogenous substrate kinase activity. These data demonstrate that insulin primarily activates an intramolecular /I subunit trans-phosphorylation reaction within the insulin ho- loreceptor and suggest that this reaction is necessary for activation of the holoreceptor. Furthermore, our results suggest a molecular basis for the dominant- negative phenotype observed in insulin-resistant patients possessing one kinase-defective insulin receptor allele. virus vTF7-3 for 1 h followed by transfection with 100 Fg of the insulin receptor expression plasmid pTM1- WT. The cells were solubilized in detergent 20 h later, and the supernatants were incubated with either the insulin receptor-specific monoclonal antibody 83-7 or the IGF-1 receptor-specific monoclonal antibody cuIR3. Following immunoabsorption, the resulting superna- tants were then assayed for '2'I-insulin and 12'II-IGF-1 binding activity as described under "Experimental Procedures." This represents the average of duplicate assays independently performed three times. buffer containing 500 mM dithiothreitol and resolved on a 7.5% SDS-polyacrylamide gel. This is a representative autoradi- ograph which was independently performed five times. . N and heterologous (Y~~WT-LY~~A/K.A\CT and partially purified by Bio-Gel A1.5m gel filtration chro- matography as described under Procedures.” Homologous (Y@WT-(Y~~WT (34 fmol, a of homologous CY~~WT- (Y~~WT and homologous &4,K.4CT-aj3A/K,4m respec-tively, heterologous (Y~~VJT-CY~~A/K,ACT Q / K , c T Y Proce-dures.” This is a representative experiment of duplicate determina- tions independently performed two times.

To examine the role of intramolecular @ subunit trans-and cis-autophosphorylation in signal transduction, the vaccinia virus/bacteriophage T7 expression system was used to generate insulin holoreceptors composed of a kinase-defective half-receptor precursor (abAIK or (XBAm.ACT) and a kinase-active half-receptor precursor (a@ACT or a#?wT). In the a B A m -a @ A C T hybrid insulin receptor, insulin stimulated a 20-fold increase in intramolecular B subunit trans-phosphorylation, whereas cis-phosphorylation increased only 3-fold over the basal state. Similarly, in the (Y@WT-(YBAm.ACT hybrid insulin receptor, insulin stimulated trans-phosphorylation approximately 30-fold and cis-phosphorylation only %fold over the basal state. Although cis-phosphorylation of the kinase-functional aB halfreceptor was observed within these hybrid receptor species, this was not sufficient to stimulate exogenous substrate kinase activity. These data demonstrate that insulin primarily activates an intramolecular / I subunit trans-phosphorylation reaction within the insulin holoreceptor and suggest that this reaction is necessary for activation of the holoreceptor. Furthermore, our results suggest a molecular basis for the dominantnegative phenotype observed in insulin-resistant patients possessing one kinase-defective insulin receptor allele. The insulin receptor is minimally composed of two extracellular a subunits and two transmembrane p subunits which are disulfide-linked into a functional a2p2 heterotetrameric complex (1, 2). The a subunit contains the high affinity insulin binding domain and the cytoplasmic portion of the / 3 subunit functions as a tyrosine-specific protein kinase. In response to insulin binding, the insulin receptor p subunit undergoes autophosphorylation at several tyrosine residues.
The major p subunit autophosphorylation sites have been localized to tyrosine residues 1158,1162,1163,1328, and 1334 (3-5), using the nomenclature of Ebina et ul. (6). Insulinstimulated p subunit autophosphorylation occurs sequentially with Tris-phosphorylation of the 1160 region correlating with maximal substrate kinase activity (4, 7, 8). These initial *This work was supported by Research Grants DK33823 and DK25295 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. events occur within a single insulin holoreceptor in an intramolecular autophosphorylation cascade (9-12); however, intermolecular cross-phosphorylation between holoreceptors has also been reported to occur under certain conditions (13-16).
Although these aspects of insulin receptor structure and function have been firmly established, several molecular details have eluded description. For example, the mechanism by which insulin binding to the extracellular a subunit(s) activates the intracellular p subunit(s) tyrosine kinase domain has not been clearly defined. Previous studies have demonstrated that specific ap-ap subunit interactions within the insulin holoreceptor are necessary for insulin-dependent activation of p subunit autophosphorylation and substrate kinase activity (12, 17-20). Furthermore, mutations or proteolysis which inactivate the kinase activity of only one / 3 subunit within an azp2 insulin holoreceptor impairs the substrate kinase activity of the functional wild type p subunit (21, 22). These data suggest that two functional p subunits are required for substrate kinase activation. Insulin binding may enable each p subunit to phosphorylate itself in an intramolecular cis reaction, or insulin may stimulate one p subunit to phosphorylate the adjacent ,8 subunit in an intramolecular trans reaction. Whether cis-phosphorylation, trans-phosphorylation, or both are required for substrate kinase activation of the insulin holoreceptor has not yet been determined. Recently, evidence for an intramolecular cis reaction has been presented based upon the autophosphorylation of partially proteolyzed and immobilized insulin half-receptors (23). In contrast, characterization of wild type/mutant hybrid insulin receptors assembled in vitro has demonstrated an intramolecular trans-autophosphorylation pathway (22).
To address this apparent contradiction and to examine the events that occur in an intact, heterotetrameric receptor during insulin-stimulated transmembrane signaling, we have adapted the vaccinia virus/bacteriophage T7 transient expression system to generate homologous or heterologous (hybrid) insulin receptors in cultured cells (24-26). These receptor species were then used to examine the relationship between intramolecular /3 subunit trunsand cis-autophosphorylation and substrate kinase activation.

EXPERIMENTAL PROCEDURES
Materials-The recombinant vaccinia virus encoding T7 RNA polymerase, vTF7-3, and the vaccinia expression vector, pTM-1, were a kind gift of Dr. Bernard Moss (National Institutes of Health). HeLa and BHK-21 fibroblast cell lines were obtained from the American Type Tissue Culture Collection, and a 3T3-F442A cell line was obtained from Dr. Christin Carter-Su (University of Michigan). A plasmid containing the entire cDNA sequence of the human insulin receptor including exon 11 was constructed as described previously (27). Monoclonal antibodies directed against the human insulin receptor (83-7 and CT-1) and a monoclonal antibody directed against the 19521 human IGF-1 receptor (aIR-3) were gifts of Dr. Kenneth Siddle (University of Cambridge) and Dr. Steven Jacobs (Burroughs Wellcome), respectively. Porcine insulin was a gift from Dr. Ronald Chance (Eli Lilly); recombinant IGF-1 was purchased from Toyobo Biochemicals. "'I-Insulin and 1251-IGF-l were prepared by the Diabetes and Endocrinology Research Center at The University of Iowa. Tran:"S-label and [y-"PIATP were purchased from ICN and Du Pont-New England Nuclear, respectively. Methionine-free Dulbecco's minimal essential media and Lipofectin were purchased from GIBCO-Bethesda Research Laboratories. The insulin receptor peptide fragment (Thr-Lys,' amino acids 1154-1165, TRDIYET-DYYRK) was a generous gift of Dr. David Coy (Tulane University, New Orleans). Poly GLU:TYR (4:l) was purchased from Sigma.
Plasmid Construction and Mutagenesis-The "Altered Sites in vitro Mutagenesis System" (Promega) was used to generate DNA encoding two previously described mutant insulin receptors; A/K (alanine for lysine substitution at amino acid 1030) and ACT (carboxyl-terminal 43-amino acid truncation, alanine to termination codon at position 1313) (28,29). Briefly, an insulin receptor BamHI cDNA fragment (nucleotides 2017-4400) was cloned from pMTACE into pSelect, and single-stranded DNA was prepared. The mutagenic oligonucleotides, complementary to coding strand sequence, were 5"ACTCGTT- ,, 50 mM Hepes, pH 7.6) plus 0.1% bovine serum albumin for 16 h at 4 "C or for 1 h at 23 "C. The binding reaction was terminated by the serial addition of 0.5 ml of 0.1% y-globulin in 10 mM NazHP04, pH 7.4, and 1 ml of 25% polyethylene glycol. The samples were centrifuged at 12,000 X g, and the pellets were washed with 1.0 ml of 10% polyethylene glycol and counted for bound "51-insulin. Nonspecific binding was determined as the amount of 12511-insulin bound in the presence of 1 p~ unlabeled insulin. For measurement of total cell surface ligand binding, cell monolayers were washed once with Krebs-Ringer-Hepes buffer, 0.1% bovine serum albumin and incubated with 0.125 nM '*'I-insulin for 16 h at 4 "C or 1 h at 23 "C. The incubation medium was aspirated followed by 3 X 1-ml washes with ice-cold phosphate-buffered saline (5 mM Na2HP04, 150 mM NaCl, pH 7.4). Cells were solubilized in 0.1 M NaOH for 60 min at 37 'C then counted for cell-associated radioactivity. Nonspecific binding was determined as the amount of Iz5I- The abbreviations used are: Thr-Lys, synthetic insulin receptor peptide (amino acids 1154-1165, TRDIYETDYYRK); poly GLU:TYR, synthetic 4:l co-polymer of glutamic acid:tyrosine; A/K, alanine for lysine substitution at position 1030; ACT, alanine to termination codon at position 1313; A/K.ACT, double mutant containing both the A/K substitution and carboxyl-terminal truncation; WT, wild type; BHK-21, baby hamster kidney; Hepes, 4-(2-hydrox-yethy1)-1-piperazineethanesulfonic acid SDS, sodium dodecyl sulfate.
insulin bound in the presence of 1 p~ unlabeled insulin.
Metabolic Labeling of Expressed Insulin Receptors-Cell monolayers (100-or 150-mm dish) were infected and transfected as described above. Five h after transfection, media was replaced with hypertonic (190 mM NaCl) methionine-free, serum-free Dulbecco's modified Eagle's medium for 30 min. Fresh media containing 100 pCi of Tran3%-label was then added to the dishes. The reaction was chased after 5 h with the addition of 0.5 ml of Ham's F-12 medium and 0.5 ml of fetal bovine serum, and the cells were harvested 12-16 h later. Receptors were then immunoprecipitated with the specified antibodies and resolved on 7.5% SDS-polyacrylamide gels or 3-10% gradient SDS-polyacrylamide gels under reducing or nonreducing conditions, respectively, as indicated in the text. Receptor bands on the resulting autoradiograms were quantified by scanning laser densitometry.
Purification of Expressed Insulin Receptors-Monolayers (four to six 100-mm dishes) were infected with vTF7-3, transfected with receptor expression plasmids, and detergent-solubilized as described above. Insoluble material was removed by centrifugation, and the supernatant was applied to Bio-Gel A1.5m gel filtration columns (22). The peak insulin binding fractions were pooled, and receptors were further purified by immunoprecipitation with one of the insulin receptor-specific monoclonal antibodies, 83-7 or CT-1, coupled to Affi-Gel 10 as indicated in the text.
Kinase Assays-Partially purified receptors (approximately 80 fmol of receptor/reaction) were incubated with 100 nM insulin, 50 mM Hepes, pH 7.4, 0.05% Triton X-100 for 1 h at 23 "C. Autophosphorylation was initiated by incubation with 3 mM manganese acetate, 1 mM CTP, and 5 p~ [y-32P]ATP (2000 pCi/nmol) (16) or by incubation with 10 mM MnClZ, 10 mM MgClZ, 100 p M [y-3ZP]ATP (3 pCi/ nmol) (22) for the indicated times. The reaction was terminated with the addition of 5 mM ATP, 5 mM EDTA, 100 mM sodium fluoride, and 10 mM sodium pyrophosphate. Receptors were then immunoprecipitated with the indicated antibody and resolved on SDS-polyacrylamide gels (7.5%) under reducing conditions. For poly GLU:TYR substrate kinase reactions, the immunoprecipitated receptors were incubated with 10 mM MnCl,, 10 mM MgCIz, 50 mM Hepes, pH 7.4, 0.1% Triton X-100 in the presence or absence of 100 nM insulin for 1 h. Autophosphorylation was initiated by a 6-min incubation with [y-32P]ATP (100 p~ ATP, 3 pCi/nmol) before 2 mg/ml poly GLU:TYR was added. Substrate phosphorylation reactions were terminated after 30 min by precipitation onto Whatman 3 MM filter paper with 10% (w/v) trichloroacetic acid. The filters were washed extensively with 10% trichloroacetic acid followed by liquid scintillation counting. For Thr-Lys peptide substrate kinase assays (311, the immunoprecipitated receptors were incubated with 10 mM MnCL, 10 mM MgCl,, 50 mM Hepes, pH 7.4, 0.1% Triton x-100 in the presence or absence of 100 nM insulin for 1 h. Autophosphorylation was initiated by the addition of [Y-~*P]ATP (100 p~ ATP, 3 pCi/ nmol), and the reactions were incubated at 23 "C for various times followed by the addition of 1 mM Thr-Lys for 20 min. The substrate kinase reactions were terminated with 20 pl of 1% bovine serum albumin plus 50 p1 of 10% trichloroacetic acid. Precipitated protein was removed by centrifugation, and the supernatant was spotted onto Whatman P81 phosphocellulose filters. The filters were washed with 4 liters of 75 mM phosphoric acid for 24 h followed by liquid scintillation counting.

RESULTS
Expression of Human Insulin Receptors-To evaluate the level of human insulin receptor expression achieved by the vaccinia virusfiacteriophage T7 system, monolayer HeLa cell cultures were first infected with recombinant vaccinia virus, vTF7-3, which encodes for the bacteriophage T7 RNA polymerase. After viral infection, the cells were transfected with various amounts of the plasmid pTM1-WT, which directs transcription of the wild type human insulin receptor from a T7 promoter (Fig. L4). HeLa cells that were infected with virus then transfected with as little as 0.5 kg of pTM1-WT plasmid DNA demonstrated specific, cell-surface lZ6I-insulin binding, whereas virally infected but untransfected HeLa cells had no detectable cell surface '251-insulin binding activity. The level of the expressed human insulin receptor increased linearly as the plasmid concentration increased and became Gg), and cell surface "'I-insulin binding activity was determined at the indicated times following transfection. saturated at 2 pg DNA/well. Cell surface 12sI-insulin binding activity slightly decreased as the amount of plasmid DNA increased from 5 to 25 pg. The appearance of cell-surface insulin binding activity was extremely rapid following transfection with pTM1-WT (Fig. 1B). '251-Insulin binding was detected as early as 2 h after transfection and increased linearly over the next 6-h period. Steady-state levels of insulin binding were observed 12-16 h after transfection and resulted in approximately 5 X IO6 to 5 X IO7 insulin receptors/cell (data not shown). Wild type insulin receptor produced by the vaccinia virus/bacteriophage T 7 system was essentially identical to insulin receptor purified from human placenta membranes with regard to heterotetramer assembly and transport to the plasma membrane, mature sialic acid glycosylation, high affinity insulin binding, intramolecular autophosphorylation, and insulin-stimulated tyrosine kinase activation (data not shown).
Several groups have demonstrated that cultured cell lines expressing both insulin and IGF-1 homologous receptor complexes also express various amounts of hybrid insulin/IGF-1 receptors (32)(33)(34)(35). Since HeLa cells contain endogenous IGF-1 receptors, we examined whether virally expressed insulin receptors could form hybrids with pre-existing IGF-1 receptors. HeLa cells were infected with vTF7-3, transfected with the wild type insulin receptor expression plasmid, solubilized, and immunoprecipitated with either 83-7, an insulin receptorspecific monoclonal antibody, or aIR3, an IGF-1 receptorspecific monoclonal antibody (Table I). '251-Insulin binding activity was not detected in control HeLa cells and therefore could not be immunoabsorbed with either monoclonal antibody. In contrast, lZ5I-IGF-1 binding in control (infected only) HeLa cells was specifically immunoabsorbed by aIR3 (89%) but not by 83-7 (8%). Expression of wild type insulin receptors resulted in insulin binding activity which was specifically immunoabsorbed by 83-7 (87%) but not by aIR3 (6%). Similarly, '*'I-IGF-l binding was only immunoabsorbed by the IGF-1 receptor monoclonal antibody (85%) and not by the insulin receptor monoclonal antibody (5%). Immunodepletion of insulin binding activity exclusively by 83-7 and IGF-1 binding activity exclusively by aIR3 demonstrated that hybrid receptor formation between the transiently expressed insulin receptors and the endogenous IGF-1 receptors did not occur under these conditions.

Assembly of hybrid insulinlIGF1 receptors in vivo does not occur between virally expressed insulin receptors and endogenous IGF-1 receptors in HeLa cells
Confluent HeLa cell monolayers (1 X lo6 cells) were infected with the recombinant vaccinia virus vTF7-3 for 1 h followed by transfection with 100 Fg of the insulin receptor expression plasmid pTM1-WT. The cells were solubilized in detergent 20 h later, and the supernatants were incubated with either the insulin receptor-specific monoclonal antibody 83-7 or the IGF-1 receptor-specific monoclonal antibody cuIR3. Following immunoabsorption, the resulting supernatants were then assayed for '2'I-insulin and 12'II-IGF-1 binding activity as described under "Experimental Procedures." This represents the average of duplicate assays independently performed three times.  2 ) . The three high molecular weight bands represent previously described, partially reduced forms of the insulin receptor that result from boiling the immunoprecipitates in the presence of SDS in order to dissociate the insulin receptor from the antibody resin (36). These data demonstrate that the insulin receptor primarily exists in cells as an a& heterotetrameric state with little, if any, free ap precursors.
To determine the subunit composition of these high molecular weight receptor species, receptors were prepared in the same manner but resolved on 7.5% SDS-polyacrylamide gels under reducing conditions (Fig. 2B). Immunoprecipitation of homologous A/K insulin receptors with the 83-7 monoclonal antibody demonstrated the specific labeling of the M, 95,000 p subunit, the M, 130,000 a subunit, and the M , 190,000 precursor bands (Fig. 2B, lane 4 ) . Similarly, immunoprecipitation of homologous ACT insulin receptors with 83-7 demonstrated the presence of the receptor precursor, a and p subunits (Fig. 2B, lane 5 ) . The ACT receptor p subunit displayed increased mobility compared to the A/K receptor p subunit reflecting its M, 5,000 carboxyl-terminal truncation.
The carboxyl-terminal-specific monoclonal antibody CT-1 also immunoprecipitated the precursor, a subunit, and p subunit of the expressed A/K insulin receptor (Fig. 2B, lane   I). Again, as expected, CT-1 did not immunoprecipitate the ACT insulin receptor expressed in cells transfected with the ACT plasmid DNA (Fig. 2B, lane 2 ) . However, co-expression of both A/K and ACT insulin receptors resulted in the immunoprecipitation of the M , 90,000 ACT p subunit by the  5 ) . Receptors were eluted from the antibody resins by boiling in Laemmli sample buffer containing 500 mM dithiothreitol and resolved on a 7.5% SDS-polyacrylamide gel. This is a representative autoradiograph which was independently performed five times.
CT-1 antibody (Fig. 2B, lane 3). These data demonstrate that the (~( 3~~~ half-receptor can assemble with the (Y@A/K halfreceptor to form a hybrid (Y(3A/K-(Y@&-T heterotetrameric receptor complex. In a similar analysis, a double mutant (Y(3A/K.&T half-receptor was shown to assemble with the ( Y (~W T halfreceptor to form a hybrid (Y@WT-(Y@A/K.ACT heterotetrameric receptor complex (data not shown). Since the stoichiometry of @ subunits within a heterotetramer is 1:1, the amount of hybrid receptor immunoprecipitated by CT-1 can be calculated based upon the molar ratio of MI 90,000 (3 subunit compared to total (3 subunit protein. For example, in Fig. 2B the ratio of MI 90,000 to M , 95,000 (3 subunit is approximately 1:3 indicating that 50% of the insulin receptors immunoprecipitated by the CT-1 antibody were in a hybrid complex (e.g. Based upon these results, we next examined the formation of (YfiWT-(YPA/K.ACT hybrid receptors as a function of the relative expression (mole fraction) of the two receptor species (Fig.   3). When the mole fraction of (YBA/K.ACT was 0.5, i.e. equal O@A/K-&A/K + a @ A C T d A / K = 3 a @ A / K + l a @ A C T ) * expression of A/K.ACT and WT receptor subtypes, hybrid receptors accounted for approximately 50% of the total receptor population immunoprecipitated by CT-1. As the mole fraction of (Y@A/K,XT increased, the percentage of hybrid receptors present in the CT-1 immunoprecipitate increased. Thus, expression of one insulin receptor precursor (approximately 3-10-fold) over another receptor precursor resulted in the near quantitative association of the less abundant species into a hybrid receptor complex. Similarly, hybrid insulin/IGF-1 receptors could be randomly assembled and were quantitatively driven with a 3-10-fold excess of one receptor species over the other (data not shown). Identical results were obtained using BHK-81, HeLa, or 3T3-F442A fibroblasts (data not shown).

Intramolecular cis-Versus trans-Autophosphorylation-To
determine the relative magnitude of intramolecular cis-and trans-autophosphorylation during insulin-stimulated transmembrane signaling, (Y(3A/K-(Y(3JCT hybrid insulin receptors were obtained by infection with vTF7-3 and co-transfection with equal amounts of the pTM1-A/K and pTM1-ACT insulin receptor expression plasmids (see Fig. 4). Hybrid (Y(~A/K-( Y @~C T heterotetramers and homologous (YPA/K-(Y@A/K heterotetramers were then autophosphorylated in the presence or absence of insulin and selectively immunoprecipitated with the carboxyl-terminal-specific CT-1 monoclonal antibody (Fig. 5). Consistent with previously reported data, the homologous A/K insulin holoreceptors were completely devoid of (3 subunit autophosphorylation activity both in the absence (Fig.   5A, lanes 11 and 12) and presence (Fig. 5A, lanes 13 and 14) of insulin. In contrast, (3 subunit autophosphorylation was observed in the ff@A/K-(YPACT hybrid receptors in a time-dependent manner. In the absence of insulin, autophosphorylation of both the M , 95,000 A/K and M, 90,000 ACT @ subunits was essentially identical with respect to both rate and extent of phosphate incorporation (Fig. 5A, lanes 1-5). However, the M , 95,000 A/K (3 subunit was preferentially autophosphorylated in the presence of insulin (Fig. 5A also displayed a time-dependent and insulin-stimulated autophosphorylation. The relative autophosphorylation rates of the A/K and ACT (? subunits within a(?A/K-a(?AcT hybrid heterotetrameric complexes were determined by scanning laser densitometry (Fig. 5B). In the absence of insulin, the initial rate of both cis-and trans-autophosphorylation was similar, whereas in the presence of insulin the initial rate of trans-autophosphorylation of the M , 95,000 A/K (? subunit was increased approximately 20-fold. In contrast, insulin stimulated the initial rate of cis-phosphorylation of the M , 90,000 ACT (? subunit only approximately %fold. Essentially identical results were obtained in both solubilized and solid-phase (immobilized immunoprecipitate) kinase assays as well as with hybrid receptors expressed in HeLa, BHK-21, and 3T3-F442A fibroblasts (data not shown).
Although these data are consistent with insulin activation of trans-autophosphorylation, the ACT insulin receptor (? subunit lacks two carboxyl-terminal (Y1328 and Y1334) tyrosine autophosphorylation acceptor sites. T o assess the potential contribution of these tyrosine residues to cis-autophosphorylation, the kinase properties of another hybrid insulin receptor, a(?wT-a(?A/K.AcT, were examined (Fig. 6 ) . Expression of the CY(?A/K.ACT precursor in a 3-10-fold molar excess over the o(?wT precursor resulted in the quantitative incorporation of the expressed wild type a(? half-receptor into a hybrid receptor complex (Pig. 3). In this hybrid insulin receptor complex, incorporation of radiolabeled phosphate into the M , 90,000 A/K.ACT (? subunit represents intramolecular trans-autophosphorylation, whereas labeling of the M , 95,000 wild type (? subunit represents the cis reaction component (see Fig. 4). In the absence of insulin, autophosphorylation of the (Y(?WT-LY(?A/K.ACT hybrid receptors resulted in the labeling of both the A/K.ACT ( M I 90,000) (? subunit as well as the wild type ( M , 95,000) (? subunit (Fig. 6A, lanes 1-5).
The basal level of cis-autophosphorylation of the wild type (? subunit slightly exceeded that of trans-autophosphorylation of the A/K.ACT (? subunit; however, insulin markedly enhanced trans-autophosphorylation of the A/K.ACT (? subunit with only a small effect on the wild type (? subunit cisautophosphorylation (Fig. 6A, lanes 6-10). As observed for the LY(?A/K-cY(?A\c.I. hybrid insulin receptor samples, autophos-  , lanes 11-14) as described under "Experimental Procedures." The resultant insulin receptors were partially purified by detergent solubilization and Bio-Gel A1.5m gel filtration chromatography, The samples were then subjected to autophosphorylation for various times in the absence (lanes 1-5, 11, and 12) and presence (lanes 6-10, 13, and 1 4 ) of 100 nM insulin. The homologous NPAIK-NPAIK and heterologous N P A / K -N P x T insulin receptors were then specifically immunoprecipitated with the carboxyl-terminal CT-1 monoclonal antibody, and the pellets were subjected to SDS-polyacrylamide gel electrophoresis and autoradiography. R, determination of the relative initial rates of P subunit cis-and trans-autophosphorylation. The autoradiographic hands corresponding to P subunit autophosphorylation within an N(~A/K-NS, KT hybrid receptor (Fig. 5A phorylation of the M , 190,000 insulin receptor precursor also occurred in a time-and insulin-dependent manner (Fig. 6 A ) .
As controls, the homologous A/K.ACT insulin holoreceptors were virtually devoid of (? subunit autophosphorylation activity in the presence of insulin at every time point examined (Fig. 6A, lanes 11-15). Quantitation of the (? subunit intensity on the autoradiogram within (Y(?wT-(Y(?A/K.AcT hybrid receptors by scanning laser densitometry revealed that insulin stimulated the initial rates of trans-autophosphorylation (MI 90,000) 30-fold and cis-autophosphorylation (Mr 95,000) 3fold (Fig. 6 B ) . Identical results were obtained when the (Y(?wT-(Y(?A/K.AcT hybrid insulin receptors were purified from either HeLa, 3T3-442A, or BHK-21 fibroblasts and when the kinase assays were performed with either solubilized or immobilized receptor preparations (data not shown).
Since the pattern of p subunit autophosphorylation within phorylation of one holoreceptor species by another holoreceptor, probably does not occur under these experimental conditions. To directly examine whether cross-phosphorylation occurred in these assays, wild type heterotetrameric insulin receptors were mixed with the truncated kinase-inactive heterotetrameric insulin receptors prior to autophosphorylation (Fig. 7). In the presence of insulin, the homologous (YPWT-CY&T wild type insulin receptors displayed M, 95,000 P subunit autophosphorylation (Fig. 7, lane 1 ), whereas the homologous (Y/~A/K.ACT-(Y@A/K.ACT insulin receptors were completely kinaseinactive (Fig. 7, lane 4 ) . In contrast, autophosphorylation of the (YPwT-(YPA/K.AcT hybrid insulin receptor resulted in the expected labeling of the M, 90,000 p subunit (Fig. 7, lane 3 ) indicative of an intramolecular truns reaction (Fig. 6). Coincubation of homologous (YPWT-CU(~WT insulin receptor with the homologous (YPA/K.ACT-(YPA,K.ACT insulin receptor demonstrated no measurable labeling of the M, 90,000 A/K.ACT / 3 subunit (Fig. 7, lane 2). Cross-phosphorylation of the M, 90,000 A/K.ACT P subunit by the wild type heterotetrameric insulin receptor was not evident even after prolonged exposure of the gel (data not shown).
Substrate Kinase Activity of Wild TypelMutant Hybrid Insulin Receptors-In both the (Y/~A/K-(YPACT and the ( Y~W T -(Y@A/K.A~ hybrid insulin receptors, the predominant insulinstimulated event was an intramolecular trum-phosphorylation of the kinase-inactive / 3 subunit by the kinase-active subunit. Although cis-autophosphorylation of the kinase-active B subunit occurred to a lesser extent and with slower kinetics than trans-autophosphorylation, the possibility existed that this reaction could result in substrate kinase activation. We therefore examined whether this cis-autophosphorylation could activate the exogenous substrate kinase activity (poly GLUTYR phosphorylation) of the hybrid insulin receptor species (Fig. 8). As previously reported for in vitro assembled hybrid insulin receptors (22), the in vivo assembled (YPA/K-(YPACT hybrid receptors were substrate kinase-inactive even in the presence of saturating insulin concentrations (Fig. 8A). Similarly, the (YPWT-(YPA/K.A.CT hybrid insulin receptors were essentially substrate kinase-inactive ( Fig. 8A). As expected, homologous (YPWT-(YPWT and ( Y~A c T -(YPACT insulin receptors displayed a 3-5-fold stimulation of substrate kinase activity in the presence of insulin, whereas the homologous CYPA/K-(Y/~A/K insulin receptors were completely devoid of substrate kinase activity ( Fig. 8A and data not shown). Identical results were observed using the Thr-Lys synthetic peptide as a substrate for each homologous and heterologous receptor species (data not shown). To determine if the slow cis component of autophosphorylation could activate CYPWT-(YPA/K,ACT hybrid insulin receptors, receptors were incubated with ATP for various times in the presence or absence of insulin before the Thr-Lys peptide substrate was added to the reaction. Preincubating homologous (Y/~wT-(Y&T insulin receptors with insulin for 1 h followed by ATP for 2 min was sufficient for maximal stimulation of substrate phosphorylation (Fig. 8B). In contrast, even after a 60-min preincubation with ATP, the (Y&T-cYPA/K,AcT hybrid insulin receptors display a relatively low level of substrate kinase activity which was unaffected by insulin. These data demonstrate that the insulin-stimulated cis-autophosphorylation within an a& heterotetrameric holoreceptor does not result in substrate kinase activation.

DISCUSSION
Recently, several studies have demonstrated the presence of hybrid insulin/IGF-1 receptors in human placenta as well as various cultured cell lines (32)(33)(34)(35). The proportion of homologous receptors compared to heterologous hybrid receptors identified in these studies appears to be variable and may reflect the relative expression of insulin and IGF-1 receptor precursors. The presence of variable amounts of homologous and heterologous insulin and IGF-1 receptors severely complicates the interpretation of studies examining insulin receptor structure-function relationships as well as studies exam-ining the biological signaling by both insulin and IGF-1.
To exclusively generate homologous insulin receptors or defined combinations of homologous and hybrid receptors, we utilized the vaccinia virus/bacteriophage T 7 expression system. Because host protein synthesis is inhibited during vaccinia virus infection, formation of hybrid receptors between endogenous host receptor precursors and transfected receptor precursors is not observed (Table I). Hybrid receptors are generated, however, by co-transfecting cells with two different insulin receptor expression plasmids (Fig. 2). Moreover, the proportion of homologous receptors and hybrid receptors expressed on the cell surface can be manipulated by altering the molar ratios of the two plasmids used in the transfection (Fig.  3). We have found that the general properties of transfected, wild type insulin receptors generated by the vaccinia virus/ bacteriophage T7 expression system are identical to those of the native insulin receptors isolated from human placenta and various cell lines. This includes a normal itinerary of receptor subunit processing, assembly, and transport to the plasma membrane. Although the level of insulin receptor precursor protein is substantially greater in the vaccinia virus/ bacteriophage T7 expression system than that typically observed in cells expressing normal amounts of insulin receptor, it is comparable to that observed in stable cell lines expressing high levels of insulin receptor and in transiently transfected cells expressing high levels of insulin receptor from an SR-a promoter (37). Interestingly, the M , 190,000 aP precursor band was not apparent on nonreducing, SDS-polyacrylamide gels but was observed on reducing, SDS-polyacrylamide gels. These data suggest that Class I disulfide bonds form early in receptor processing, and noncovalent aP dimers are transient intermediates that do not appreciably accumulate in the endoplasmic reticulum. In any event, the mature cell surface u2PZ insulin holoreceptors produced by the vaccinia virus/bacteriophage T 7 expression system display typical high affinity curvilinear ligand binding as well as insulinstimulated intramolecular autophosphorylation and exogenous substrate kinase activity.
The vaccinia virus/bacteriophage T7 expression system was used to generate and characterize the functional properties of hybrid insulin receptors in order to address the molecular basis of insulin-stimulated transmembrane signaling with regard to the specific interactions between the @ subunit cytoplasmic domains. In previous studies, we and others have reported that isolated aP insulin half-receptors and partially proteolyzed a2@@' receptors are kinase-inactive species (12,(17)(18)(19)(20)(21). However, the aP half-receptors regain kinase activity upon reassociation into an a& heterotetrameric state (17,22). In addition, in vitro assembled hybrid receptors composed of a kinase-active aP half-receptor with a kinase-inactive ab half-receptor are completely devoid of substrate kinase activity but display at least partial insulin-stimulated autophosphorylation (22). Similarly, Cobb et al. (38) have observed that the kinase activity of the baculovirus-expressed 6 subunit cytoplasmic domain occurs via an intermolecular association. These data suggest a requirement for interaction between @ subunits within an a& heterotetramer and suggest the possible trans-phosphorylation of one /3 subunit by the adjacent @ subunit as a means of activating the holoreceptor.
In contrast, it has also been reported that the baculovirusexpressed /I subunit kinase domain remains active as a monomer (39,40). Furthermore, partial proteolysis of the a subunits within an a& holoreceptor followed by isolation of the resultant half-receptors was reported to generate a monomeric, kinase-active, a'@ half-receptor species (23). These results were interpreted as evidence for an intramolecular cis-autophosphorylation reaction within the native a& insulin holoreceptor. The reasons for these discrepant results are not apparent at the present time.
Nevertheless, the data presented in this study clearly demonstrate that insulin initially activates an intramolecular trans-autophosphorylation reaction in which the kinase-active ,6 subunit phosphorylates the adjacent kinase-inactive p subunit within a wild type/mutant a2p2 heterotetrameric insulin holoreceptor complex. Although intermolecular crossphosphorylation (one holoreceptor phosphorylating a second holoreceptor) has been observed under certain conditions (13-16), in the two different wild type/mutant hybrid receptor species examined in this study, autophosphorylation occurred predominantly on the kinase-inactive p subunit both in solution and in solid-phase assays. Since the solid-phase assay prevents interaction between holoreceptors, these data suggest cross-phosphorylation does not occur in this system. In addition, when kinase-active holoreceptors were mixed with truncated, kinase-inactive holoreceptors, phosphorylation of the truncated, kinase-inactive / 3 subunit was not observed (Fig. 7). Thus, these data are consistent with previous studies that demonstrate autophosphorylation is an intramolecular reaction (9-12) and establish that the primary action of insulin is to activate intramolecular fi subunit trans-autophosphorylation within a single holoreceptor.
Previous studies have identified the major /3 subunit autophosphorylation sites as tyrosine residues 1158, 1162, 1163, 1328, and 1334 (3-5). Tyrosine phosphorylation of the 1158, 1162, and 1163 sites were found to correlate with the activation of exogenous substrate kinase activity (4, 7, 8). In contrast, phosphorylation of the carboxyl-terminal sites (1328 and 1334) has been reported to be important in mitogenic signaling (41,42). In both of the hybrid receptor species examined, autophosphorylation of the kinase-inactive / 3 subunit occurred at similar rates even though the M , 90,000 A/K.ACT ( 3 subunit lacks the carboxyl-terminal 1316 and 1322 acceptor sites. These data indicate that the primary sites of intramolecular trans-autophosphorylation occur in the 1160 region. Furthermore, the relatively small amount of intramolecular cis-autophosphorylation observed on the kinase-active p subunit was not sufficient to activate substrate kinase activity (Fig. 8). These data suggest that the cisautophosphorylation reaction occurs on other tyrosine residues which do not play a significant role in kinase activation.
The activation of the human insulin receptor by intramolecular trans-autophosphorylation suggests a molecular basis for the dominant-negative phenotype observed in insulinresistant patients possessing one kinase-defective insulin receptor allele (43). If the formation of hybrid receptors in such patients occurs randomly with equal expression from both insulin receptor alleles, then as much as 75% of the total population of insulin receptors would be biologically inactive. Formation of hybrid insulin/IGF-1 receptors has been documented in vivo and in vitro and could also contribute to the defects in IGF-1 responsiveness observed in insulin-resistant patients (44).
In summary, subunit interactions between insulin receptor a p half-receptors are necessary for both high-affinity insulin binding and tyrosine kinase activity. The rapid insulin-stimulated phosphorylation that occurs within intact crzfiz insulin holoreceptors predominantly results from an intramolecular trans-autophosphorylation reaction in which one p subunit phosphorylates the adjacent fi subunit. These data further suggest that substrate kinase activation of the insulin holoreceptor requires sequential or simultaneous intramolecular trans-autophosphorylation reactions which results in the quantitative tyrosine phosphorylation of both p subunits.