Identification of the Insulin Receptor Tyrosine Residues Undergoing Insulin-stimulated Phosphorylation in Intact Rat Hepatoma Cells*

Tyr(P)-containing proteins were purified from ex- tracts of insulin-treated rat hepatoma cells (H4-II-E-C3) by antiphosphotyrosine immunoaffinity chroma- tography. Two major insulin-stimulated, Tyr(P) proteins were recovered: an M, 95,000 protein (identified as the insulin receptor B subunit by its immunoprecip- itation by a patient-derived anti-insulin receptor serum and several anti-insulin receptor (peptide) anti- sera) and an M, 180,000 protein (which was unreactive with all anti-insulin receptor antibodies). After purification and tryptic digestion of the M, 95,000 protein, tryptic peptides containing Tyr(P) were purified by sequential antiphosphotyrosine immunoaffinity, reversed-phase, anion-exchange chro- matography. The partial amino acid sequence obtained by gas- and solid-phase Edman degradation was com- pared to the amino acid sequence of the intracellular extension of the rat insulin receptor deduced from the genomic sequence. Approximately 80% of all 8 subunit [32P]Tyr(P) resides on two tryptic peptides: 50-60% of [32P]Tyr(P) is found on the tryptic peptide Asp-Ile-Tyr-Glu-Thr-Asp-Tyr-Tyr-Arg

The insulin-stimulated tyrosine phosphorylation of the insulin receptor in intact rat hepatoma cells thus involves at least 6 of the 13 tyrosine residues located on the subunit intracellular extension. These tyrosines are clustered in several domains in a distribution virtually identical to that previously found for partially purified human insulin receptor autophosphorylated in vitro in the presence of insulin. This multisite * This work was supported in part by National Institutes of Health Grant AM17776 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. The insulin receptor (IR)' is a ligand-activated, tyrosinespecific protein kinase (1,2) which phosphorylates tyrosine residues on its own p subunit (3)(4)(5)(6)(7) and subsequently on other proteins (8)(9)(10). The crucial importance of IR f? subunit autophosphorylation/tyrosine kinase to the signaling function of the receptor has been established by several recent reports which demonstrate that interference with the receptor kinase function, either by introduction of inhibitory monoclonal antibodies into intact cells (11,12) or via expression of genetically engineered receptors whose ATP-binding site has been inactivated (13,14), prevents all of the cellular responses to insulin examined thus far (augmented glucose uptake, ribosomal $6 serine-protein kinase, glycogen deposition, and thymidine incorporation). The autophosphorylation and substrate phosphorylation reactions catalyzed by the IR have been extensively characterized with partially purified receptor and model substrates. A stimulatory, probably obligatory role for subunit autophosphorylation in kinase activation has been established (15-18); and the majority of the multiple tyrosine residues which participate in p subunit autophosphorylation in vitro have been identified (20, 21). By contrast, less complete information is available concerning the tyrosine-specific phosphorylation of insulin receptors in situ in intact cells (22)(23)(24)(25)(26)(27). Herein, we show that in response to insulin, the IR of intact rat hepatoma (H4) cells undergoes phosphorylation of at least 6 tyrosine residues, located on three domains of the IR , f 3 subunit. The pattern of autophosphorylation sites found in insulin-stimulated H4 hepatoma cells corresponds very closely, qualitatively and quantitatively, to that observed previously for partially purified human IR autophosphorylated in vitro in the presence of comparable insulin concentrations and saturating levels of ATP (21).

32P Labeling of H4 Hepatoma Cells and Affinity Purification of 32P-
H4-II-E-C3 rat hepatoma cells (American Tissue Culture Collection, CRL 1548) grown on 10-cm plates to late log phase in Swim's S-77 medium with 20% horse serum and 5% fetal bovine serum were  were labeled with 32P for 2 h and then exposed to insulin (1 p~) or carrier for 5 min. The plates were rinsed and solubilized, and a supernatant after centrifugation at 10' X g for 60 min was subjected to immunoaffinity chromatography on antiphosphotyrosine monoclonal antibody 1G2, as described under "Materials and Methods." The phenyl phosphate eluates were subjected to SDS-PAGE; the autoradiograph of the dried gel is shown. Lanes 1 and 2, eluates from 32P-labeled extracts of control ( l a n e 1 ) and insulin-treated ( l a n e 2) cells; lanes 3 and 4, hapten eluates of the second immunoaffinity adsorption of the flow-through fraction from the first antibody column. Note that less than 10% additional 3ZP-labeled proteins are recovered in this second pass. E, immunoprecipitation of insulinstimulated hepatoma cell [32P]Tyr(P)-containing proteins by antiinsulin receptor antibodies. An extract of insulin-treated, 32P-labeled H4 cells was subjected to antiphosphotyrosine immunoaffinity chromatography. Aliquots (1 X) of the hapten eluate (lanes 1 and 8 = 1 X; lane 2 = 0.1 X; see A , lane 2) were incubated with the following antisera: lane 3 a rabbit antiserum raised against the synthetic peptide RDIYETDYYRK, which corresponds to residues 1143-1153 of the amino acid sequence of the human IR precursor (the corresponding segment of the rat IR has an identical sequence); lane 4, a rabbit antiserum raised against the synthetic peptide KRSYEEHIPY, which corresponds to residues 1313-1322 of the amino acid sequence of the human IR precursor (the corresponding segment in the rat IR has the sequence KRTYDEHIPY); lane 5, a rabbit antiserum raised against the synthetic peptide KTRPEDFRD, which corresponds to residues 40-48 of the amino acid sequence of the human IR precursor (the corresponding sequence in the rat IR is unknown); lane 6, a human antiserum (B2) which contains antibodies reactive with the human I R lane 7, normal rabbit serum. After overnight incubation at 0 "C, the immune complexes were adsorbed to heat-killed staphylococci (Pansorbin), washed five times in 1% Triton X-100, 50 mM Tris-HC1 (pH 7.4), and eluted into 1% SDS, 40 mM DTT by boiling for 5 min. The autoradiograph of the stained, dried SDS-PAGE gel is shown. C, phosphoamino acid analysis of the insulin-stimulated 32P-labeled M, 180,000 protein. An extract of insulin-treated, 32Plabeled H4 cells was subjected to antiphosphotyrosine immunoaffinity chromatography. A portion of the hapten eluate was subjected to SDS-PAGE; the gel was stained/destained with Coomassie Blue and dried. The region containing the 32P-labeled M, 180,000 protein was mmol) for 2 h, except where indicated. Thereafter, insulin (1 p~) was added to one set of plates; after 5 min, the medium was aspirated, and the monolayer was washed once with ice-cold 0.15 M NaCl, 0.01 M NaPi (pH 7.4). All procedures thereafter were at 0-4 "C. A solubilization buffer (0.75 ml/plate) was added 1% (w/v) Triton X-100, 2 mM Na3V04, 0.1 M NaF, 2 mM EGTA, 10 mM EDTA, 50 mM Tris-HCl (pH 7.4), 100 kallikrein-inactivating units/ml Trasylol, 20 phf pepstatin, 20 p~ leupeptin, and 0.1 mM diisopropyl fluorophosphate. The plates were scraped with a rubber policeman, and the extracts were combined and centrifuged at lo5 X g for 60 min. The supernatant was filtered through glass wool and cycled twice over a column containing antiphosphotyrosine monoclonal antibody 1G2 coupled to Sepharose 4B (15 mg of IgG/ml of settled beads; 10-50 p1 of settled beads/lO-cm plate). The column was washed successively with 100 volumes of 0.2% Triton X-100, 50 mM Tris-HC1 (pH 7.4), 0.25 M NaC1, followed by 10 volumes of the same buffer minus NaCl and finally 5 column volumes of 0.02% Triton X-100,5 mM Tris-HC1 (pH 7.4). Elution was performed with Triton 0.02% X-100, 5 mM Tris-HCl (pH 7.4), 40 mM phenyl phosphate either in 1 ml/fraction or batchwise with 3 column volumes. The column was regenerated by washing with 10 volumes of 0.2% Triton X-100,50 mM Tris-HC1 (pH 7.4), 1.0 M NaCl and re-equilibrated with the same buffer without NaCl.
Prior to anion-exchange chromatography, pooled peaks from the C1/C8 column were briefly evaporated to remove acetonitrile, diluted 2-fold or more with the starting buffer (0.02 M triethanolamine HCl (pH 7)), and neutralized with NaOH. Anion-exchange chromatography was carried out on a Pharmacia LKB Biotechnology Inc. Mono Q column (HR 5/5) with NaCl gradient elution.
Digestion of tryptic 32P-peptides with Staphylococcus aureus V8 protease (20 pg/ml) was carried out in 50 mM ammonium bicarbonate (pH 7.8) at 37 "C for 1-20 h. Tryptic peptides eluted from the Mono Q column were digested directly; the reactions were terminated by addition of diisopropyl fluorophosphate to 1 mM. The digests were analyzed by anion-exchange chromatography on a Pharmacia LKB Biotechnology Inc. Mono Q column (HR 5/5) as described above.
Protein sequencing of [32P]Tyr(P)-containing peptides was carried out on an Applied Biosystems Protein Sequencer 470A equipped with PTH analyzer 120A. Coupling of [32P]Tyr(P) peptides to aminobenzylpolystyrene beads, manual Edman degradation of the coupled peptides, and identification of PTH-[32P]Tyr(P) by thin-layer electrophoresis at pH 1.9 were carried out as described before (21).
Peptides were synthesized by the stepwise solid-phase method (28,29) in a Applied Biosystems 430A peptide synthesizer and purified by reverse-phase HPLC on Cl8 Vydac columns.
Anti-peptide antibodies were elicited as follows. Synthetic peptide (1 mg), mixed with an equal weight of methylated BSA (Sigma detected by autoradiography, excised, and eluted three times into 0.5% SDS by boiling for 30 min each time. The protein, in the presence of carrier BSA (0.5 mg/ml), was precipitated by addition of 4 volumes of acetone at -20 "C and, after 30 min, was collected by centrifugation a t -20 "C (1.2 X lo4 X g for 20 min). The pellet was rinsed, dried with Nz, and subjected to partial acid hydrolysis (6 N HCl, 110 "C, 2 h). The residue was redissolved in water and subjected to two-dimensional thin-layer electrophoresis as described (6) in the presence of carrier phosphoamino acids. The autoradiograph of the plate is shown. (1 p~, even-numbered lanes) or carrier (odd-numbered lanes) was added. After 5 min, the medium was aspirated, the cells were rinsed once in ice-cold phosphate-buffered saline, and 0.75 ml of extraction buffer (see "Materials and Methods") was added, as described for Fig.  1. The extracts from the control and insulin-treated plates were separately pooled. An aliquot of each extract was removed; SDS (to 2%) and DTT (to 40 mM) were added, and the samples were boiled for 5 min (lanes 1 and 2). Another aliquot was maintained at 0 "C until the immunoaffinity purification step described below was completed (approximately 3 h), where upon SDS/DTT was added (lanes   3 and 4 ) . The remainder of the extract was centrifuged at lo5 X g for 60 min, and the supernatants from the control and insulin-treated cells were subjected to purification on wheat germ agglutinin-sepharose (lanes [5][6][7][8] or antiphosphotyrosine monoclonal antibody columns (lanes [9][10][11][12]. For wheat germ agglutinin-Sepharose, the supernatant was adsorbed to the column by recycling three times; after washing, the column was eluted with 0.2% Triton X-100,50 mM Tris-HCI (pH 7.4), 0.3 M N-acetylglucosamine. SDS/DTT was added to aliquots of the nonadsorbed fractions (lane 5 and 6) and the N-acetylglucosamine eluate (lanes 7 and 8). For immunoaffinity chromatography, the supernatants were adsorbed to columns of antiphosphotyrosine monoclonal antibody 1G2, coupled to Sepharose 4B, washed, and eluted with 40 mM phenyl phosphate as described under "Materials and Methods" and for Fig. L4. SDS/DTT was added to aliquots of the nonadsorbed flow-through fraction (lanes 9 and 10) and phenyl phosphate eluates (lanes 11 and 12). These fractions were then subjected to SDS-PAGE; the proteins were transferred by a semi-dry apparatus to nitrocellulose paper. After blocking with BSA, the blot A1009), was emulsified in complete Freund's adjuvant (to a final volume of 2.2 ml); 0.5 ml was injected intradermally into New Zealand White rabbits at multiple sites on days 1,8, and 15. A fresh emulsion was used each time. The rabbits were bled at 8-10 weeks and at intervals thereafter. Sera were screened by immunoprecipitation of partially purified 32P-labeled human IR autophosphorylated in vitro; immunoprecipitation was carried out as described (6).
Polyclonal antiphosphotyrosine antibodies were prepared by immunization of rabbits with a t/v-Ab1 fusion protein prepared as described (30) by the schedule outlined in Ref. 31. The subfraction of antibodies with specificity for Tyr(P) was purified by affinity chromatography according to Ross et al. (32). The human serum reactive with the insulin receptor (B2) was a gift from Simeon Taylor (National Institutes of Health).
Monoclonal antiphosphotyrosine antibody 1G2 is a high affinity IgG1, K monoclonal antibody that demonstrates high specificity for the phenyl phosphate moiety of phosphotyrosine proteins? The antibody was affinity-purified and coupled to CNBr-activated Sepharose 4B as described previously (33) to a final concentration of 15 mg of antibody/ml of settled beads. We have previously shown that 1 ml of beads completely adsorbs at least 2 nmol of 32P-labeled IR (21).

RESULTS
Insulin receptors were solubilized from 32P-labeled, control and insulin-treated H4 cells into buffers containing Triton X-100 and phosphatase inhibitors. Over 90% of cellular insulin binding activity is recovered in the supernatant after centrifugation at lo5 X g for 60 min. Insulin receptors containing ["P]Tyr(P) were partially purified by immunoaffinity chromatography on columns of monoclonal antiphosphotyrosine antibodies coupled to Sepharose 4B (Fig. 1). The hapten eluate of these columns contains two major insulin-stimulated 32P-labeled polypeptides which migrate at M , 95 Application of the flow-through from the first antiphosphotyrosine monoclonal antibody column into a second, identical column gave (10% additional recovery of these [32P]Tyr(P)containing proteins (Fig. lA). The M , 95,000 32P-labeled polypeptide is the p subunit of the rat I R it is immunoprecipitated by a human (patient) serum which contains polyclonal antibodies reactive with the human IR, as well as by several rabbit anti-peptide antibodies raised against synthetic peptides corresponding to human IR precursor residues 40-48 ( a subunit), 1143-1153 (0 subunit), and 1313-1322 ( p subunit) (the latter two peptides encompass 5 tyrosine residues which correspond to two-thirds of all Tyr(P) in the human IR subjected to maximal autophosphorylation in vitro). By contrast, the M , 180,000 insulin-stimulated [32P]Tyr(P)-containing protein is unreactive with all of these anti-receptor antibodies (Fig. 1B). The recovery of [32P]Tyr(P)-containing proteins in this onestep affinity purification was estimated by immunoblot analysis of nitrocellulose-immobilized hepatoma H4 proteins using a rabbit polyclonal antiphosphotyrosine antibody distinct from that employed in the immunoaffinity chromatography (Fig. 2). These studies indicate that: 1) insulin substantially A. R. Frackelton  serum-starved for 72 h and labeled with carrier-free 32Pi (1 mCi/ plate) for 2 h, were incubated with 1 PM insulin for 5 min. The cells were rinsed with cold (0-4 "C) phosphate-buffered saline and solubilized as described under "Materials and Methods," and the ["PI Tyr(P)-containing proteins were immunoaffinity-purified using immobilized antiphosphotyrosine monoclonal antibodies as described under "Materials and Methods." The 32P-labeled M, 95,000 b ' subunit ( Fig. 1, A and B ) was isolated by SDS gel electrophoresis. The data from 2 of 12 preparations are shown. Steps 1-4 were at 0-4 "C; steps 5 and 6 were at 23 "C. Analysis of step 4 (preparation B) by SDS/ urea-polyacrylamide gradient gel electrophoresis is shown in Fig. 3A.
augments the immunoreactive Tyr(P) content of M , 95,000 and 180,000 proteins in serum-starved rat hepatoma (H4) cells; 2) the increment in Tyr(P) is relatively stable after cell disruption in the extraction buffer a t 0-4 "C for 3 h; 3) all of the rat IR (which contains Tyr(P)), but none of the M , 180,000 protein (which contains Tyr(P)), is adsorbed to a wheat germ agglutinin-Sepharose column; 4) all rat insulin receptors (which contain Tyr(P)) and approximately 50% of the M , 180,000 Tyr(P)-containing protein are adsorbed by the monoclonal antiphosphotyrosine antibody column and are recovered in the phenyl phosphate eluate. Thus, antiphosphotyrosine immunoaffinity chromatography, followed by SDS-PAGE/gel elution, provides essentially complete recovery of the H4 cell insulin receptors whose / 3 subunit contain ["PI 32P-Labeled B subunit (SDS-denatured and acetone-precipitated) was subjected to complete digestion with trypsin. The Tyr(P). digest was partially purified by immunoaffinity chromatography on monoclonal antiphosphotyrosine antibodies coupled to Sepharose 4B. Approximately equal fractions (40-50'36 each) of the applied "P were recovered in the phenyl phosphate eluate and nonadsorbed flow-through (Table I); reapplications of the flow-through fraction to a second antibody column did not give further adsorption of 32P (not shown).
Analysis of the unfractionated tryptic digest by SDS/urea gradient-PAGE (Fig. 3A) revealed that 70-75% of 32P migrates near the front, corresponding to peptides with M , 5 2000; whereas 25-30% of the "P-peptide migrates at M , 4000-5000. The bulk (-80%) of these M , 4000-5000 "P-labeled tryptic peptides are recovered in the flow-through of the antiphosphotyrosine antibody column, where they constitute 50% of all "P-labeled peptide in this fraction. By contrast, 90% of the 32P-labeled peptides in the hapten eluate migrate at M , 5 (Table I,   Recovery of 3*P was 90%. D, Peak 1 (---) and peak 2 (-) from B were each run separately on the same Mono Q column under identical conditions as for C. Recovery of '*P was 64% for peak 1 and 87% for peak 2. Note that A and C reflect steps 5 and 6 from preparation A in Table I The [32P]Tyr(P)-containing peptides were further purified on reverse-phase (Cl/C8) chromatography (Fig. 4, A and B ) .
Elution with a steep gradient of acetonitrile yielded 6545% of applied counts/minute in a slightly asymmetrical peak, with the remainder of the 3*P (ranging in various cell preparations from 10 to 30% of applied counts/minute) emerging later in a broad, heterogeneous zone (Fig. 4A). The initial major C1/C8 peak could be partially subfractionated on a shallow gradient of acetonitrile to 12% (Fig. 4B) into two major peaks (peaks 1 and 2) and several minor peaks (unnamed) in a pattern very similar to that observed previously for the tryptic peptides containing [32PP)Tyr(P) derived from 32P-labeled human IR autophosphorylated in vitro (cf. Fig. 2B in Ref. 21). However, a more incisive subfractionation of the array of peptides in the major C1/C8 peak is achieved by anion-exchange chromatography (Fig. 4, C and D). When the initial Cl/C8 peak is applied to a Mono Q anion-exchange column (Fig. 4C), over 90% of applied 32P is recovered in four peaks: two minor peaks eluting at 0.17 and 0.34 M NaCl, respectively (the 0.34 M NaCl peak is the predominant component of Cl/C8 peak 1 shown in Fig. 4B), and two major peaks eluting at 0.22 and 0.29 M NaC1, respectively (both derived from Cl/C8 peak 2 indicated above; cf. Fig. 40). The chromatographic profile observed on anion-exchange chromatography is virtually indistinguishable from that previously described for [32P]Tyr(P)-containing tryptic peptides derived   Table 11).

Gas-phase microsequencing of tryptic peptides containing p2P]Tyr(P) derived from rat insulin autophnsphnrylated in intact rat hepatoma cells
Comparison of these partial sequences to the amino acid sequence of the intracellular extension of the rat IR p subunit The tryptic peptides are numbered from the transmembrane region toward the carboxyl terminus.
(deduced from the rat genomic ~equence)~ permitted unequivocal identification of the 0.29 M NaCl peak as tryptic peptide 4 (Table 111) and the 0.22 M NaCl peak as tryptic peptide 8 (Table 111) aureus V8 protease (Fig. 6 ) and manual Edman degradation of the 32P-labeled tryptic peptides coupled to solid-phase supports (Fig. 7). Inspection of the amino acid sequence of the intracellular extension of the rat IR @ subunit, as deduced from genomic sequence, indicates that there are 13 tyrosines on this segment, distributed on eight potential tryptic peptides (Table  111). Two of these peptides are 38 residues in length, and it is likely that the M, 4000-5000 32P-labeled tryptic peptide eluting later in the C1/C8 gradient (Figs. 4A and 5) is one or both of these peptides). Conversely, since all of the 32P-labeled peptides in the initial Cl/C8 peak (Fig. 4A) (and therefore the Mono Q peaks in Fig. 4 (C and D) derived therefrom) are entirely M, 5 2000 on SDS/urea gradient gel electrophoresis, peptides 1 and 5 cannot contribute to this array. Each of the potential tyrosine-containing tryptic peptides shown in Table  I11 contains 1 or more glutamic acid residues and should be susceptible to cleavage by S. aureus V8 protease. Thus, the [32P]Tyr(P)-containing peptide eluting at 0.29 M NaCl (identified above as DIYETDYYR, peptide 4, Table 111) was cleaved completely by S. aureus V8 protease into two 32Plabeled peptide products, each containing essentially equal amounts of 32P, eluting at 0.24 and 0.13 M NaCl on Mono Q anion-exchange chromatography (Fig. 6B) Fig. 4A were subjected to SDS/urea-polyacrylamide gradient gel electrophoresis (as described for Fig. 3A). The autoradiograph of the frozen gel is shown. The heavy dark line at the bottom of the autoradiograph is primarily an edge artifact.
Taken together, the V8 cleavage and manual Edman degradation indicate that the 0.29 M NaCl [32P]Tyr(P)-containing peptide is the double phosphorylated form of DIYETDYYR, predominantly in the form with Tyr(P) at residues 3 and 7 from the amino terminus, probably admixed with the species containing Tyr(P) at residues 3 and 8. The double phosphorylated variant with Tyr(P) at residues 7 and 8 is absent (since two, not three, 32P-fragments are observed on S. aureus V8 digestion of the 0.29 M NaCl tryptic peptide). The identity of the minor Mono Q peak eluting at 0.34 M NaCl is established as tryptic peptide 4 by the results of solidphase manual Edman degradation; a peak of PTH-[32P] Tyr(P) is observed at cycle 3 from the amino terminus (Fig.  7, upper). Only tryptic peptide 4 contains a tyrosine at this location (Table 111). A second, broad peak of PTH-[32P] Tyr(P) release is observed in the later cycles of this degradation, indicating phosphorylation at 5 r ' and/or Tyr"; quantitative assignment of [3ZP]Tyr(P) at these residues is established by the results of S. aureus V8 protease cleavage of the 0.34 M NaCl 32P-labeled peptide. Two new 3ZP-labeled V8 fragments are observed, eluting at 0.21 and 0.24 M NaCl and containing 32P in a ratio of 2:1, respectively (Fig. 6A). The 0.24 M NaCl S. aureus V8 fragment of the 0.34 M NaCl tryptic peptide coelutes with one of the V8 fragments derived from the 0.29 M tryptic NaCl peptide, identified above as the double phosphorylated form of tryptic peptide 4 (compare Fig. 6, A to B). This V8 fragment (eluting at 0.24 M NaCl) is likely the peptide Asp-Ile-Tyr(P)-Glu, the phosphorylated amino-terminal V8 fragment of tryptic peptide 4. Thus, the 0.21 M NaCl S. aureus V8 fragment of the 0.34 M NaCl tryptic peptide (Fig. 6A) represents the carboxyl-terminal V8 fragment of tryptic peptide 4 (TDYYR). Inasmuch as this carboxyl-terminal V8 fragment, which contains 2 tyrosines, also contains twice as much 32P as the amino-terminal V8 fragment (  Fig. 4C) were each digested with S. aureus V8 protease (50 pg/ml) for 20 h in 50 mM ammonium bicarbonate (pH 7.8) at 37 "C. The V8 protease digests were applied to the Mono Q column (HR 5/51) and eluted under identical conditions as described for Fig. 4 (C and D).
Overall recovery of 32P recovery was 73% were coupled through their free carboxyl groups to aminobenzylpolystyrene beads, and manual Edman degradation was carried out as described (21). The cleavage products at each cycle were dried and redissolved in H20, and a constant aliquot was subjected to thin-layer electrophoresis at pH 1.9 on cellulose sheets. Carrier PTH-Tyr(P) was detected by fluorescence quenching, and 32P by autoradiography, which is shown. The migrations of PTH-[32P]Tyr(P) and "Pi are indicated by arrows. The 32P observed near the origin is 32P-peptide, released nonspecifically or due to peptide coupling via side chain carboxyl residues.  (21), which also has the same sequence DI-YETDYYR, elutes at pH 7 from Mono Q at 0.34 M NaCl. The peak of 32P-peptide eluting from Mono Q at 0.22 M NaCl exhibited, on gas-phase microsequencing (Table 11), PTH-Glu at cycle 4 (present in tryptic peptides 4 and 8 ) and PTH-Ile at cycle 6 (present only in tryptic peptide 8). Solidphase manual Edman degradation (Fig. 7, lower) reveals a clear-cut PTH-[32P]Tyr(P) at cycle 2; tyrosine is found at residue 2 only in peptide 8. These data establish tryptic peptide 8 as the major component of the 0.22 M NaCl peak. S. aureus V8 protease cleavage of the 0.22 M peptide yields two 32P-labeled fragments, indicating phosphorylation of tyrosine at residue 8 (carboxyl-terminal to the putative V8 cleavage site at residue 4), as well as at residue 2. Residue 8 is not, however, recovered cleanly in the manual Edman procedure (Fig. 7, lower). It is noteworthy that the double phosphorylated form of the tryptic peptide from the human insulin receptor (SYEEHIPYTHMNGGK) which corresponds to rat insulin receptor tryptic peptide 8 (TYDEHIP-YTHMNGGK) also elutes at pH 7 from Mono Q at 0.22 M NaCl (21). The minor peak of 32P-peptide eluting at 0.17 M NaCl (Fig.  4C) is unidentified. Recovery of PTH-[3ZP]Tyr(P) on manual Edman degradation was too low for unequivocal assignment. S. aureus V8 protease digestion of two different preparations of the 0.17 M NaCl 32P-peptide gave only a single, new 32Plabeled fragment, eluting at 0.09 M NaCl (Fig. 6 D ) . We surmise that the 0.17 M NaCl peak is a monophosphorylated form of tryptic peptide 4 (Tyr(P)) at residue 7 or 8) or more likely tryptic peptide 8 (Tyr(P) at residue 8), or both.

DISCUSSION
The present studies demonstrate that when serum-starved rat hepatoma cells are exposed to supramaximal concentrations of insulin for 5 min, the insulin receptor, initially devoid of immunoreactive Tyr(P) (Figs. U t and M ) , undergoes in-sulin-stimulated autophosphorylation on at least 6 tyrosine residues, clustered on three separate domains of the p subunit intracellular extension. One domain (peptide 4, Table 11), defined by the tryptic peptide DIYETDYYR, is identical to human IR precursor residues 1144-1152 (1) and is situated in the "tyrosine kinase" domain of the p subunit; residue 1150 is homologous to the major in uitro autophosphorylation site of the Rous sarcoma virus-transforming antigen. This peptide is recovered predominantly in a double phosphorylated form (Mono Q peak eluting at 0.29 M NaC1; the amount of ["PI Tyr(P) at residue 3 = the amount found at residues 7 and 8 combined) and to, a lesser extent, as a triple phosphorylated species (Mono Q peak eluting at 0.34 M NaCl). Taken together, these isoforms of peptide 4 (cf.  (Table 111), recovered predominantly in a double phosphorylated form; this is a region through which the insulin receptor shares virtually no similarity in amino acid sequence with the epidermal (34) and platelet-derived (35) growth factor receptor tyrosine kinases and is the IR p subunit region least similar to the insulin-like growth factor-I receptor @ subunit (36). A tentative estimate of the relative abundance of 32P-peptides 4 and 8 can be obtained from data such as that in Table I and Fig. 6; consideration of the distribution of [32P]Tyr(P)-containing tryptic peptides recovered from seven different cell harvests suggests that peptide 4 is recovered in roughly 2-fold molar excess (in terms of peptide, not phosphorus) over peptide 8. Finally a third, probably minor domain (-10% of total / 3 subunit [32P]Tyr(P)) is encompassed by a tryptic peptide of M, 4000-5000, probably peptide 1 or 5 (Table 111). Together, these three peptides encompass 80-90% of all rat IR @ subunit [32P]Tyr(P).
The distribution of rat IR p subunit [32P]Tyr(P) provided by this analysis is in effect a "snap shot" of the autophosphorylation reaction in situ. We have examined in detail the effect of a supramaximal dose of insulin at one time point; analysis of overall insulin-stimulated p subunit tyrosine phosphorylation in intact H4 cells, by immunoblot with polyclonal antiphosphotyrosine IgG, indicates that the conditions chosen give maximal steady-state tyrosine phosphorylation (not shown). However, we have no information on the temporal progression of site-specific phosphorylation. Thus, our estimates of the relative abundances of various [32P]Tyr(P) residues must be tempered by the possibility that the continued activity of tyrosine phosphatase in the intact cell may have altered the relative occupancy of various tyrosines in a differential manner and conceivably resulted in the complete dephosphorylation of certain residues within the 5-min period of insulin stimulation. However, it should be emphasized that the distribution of [32PP]Tyr(P) on the rat IR p subunit in intact hepatoma cells exposed to a supramaximal dose of insulin for 5 min is virtually identical to the pattern observed for partially purified human IR autophosphorylated in vitro in the presence of comparable concentrations of insulin and saturating concentrations of MnATP (21). In each case, 6 or more phosphotyrosines, clustered on three separate domains, are observed; and the specific tyrosines participating appear to be the same. During in vitro autophosphorylation, tyrosine phosphatase was fully suppressed, and the distribution of [32P]Tyr(P) observed reflected the autokinase reaction only.
The similarity in the distribution of [32P]Tyr(P) on the IR p subunit purified from insulin-stimulated hepatoma cells, as compared to IR autophosphorylated in vitro, suggests that tyrosine phosphatase action probably did not contribute significantly to the pattern of rat IR [32P]Tyr(P) observed in the intact cell. It seems likely that the high intensity of the insulin stimulus employed in these studies greatly favored rephosphorylation of tyrosine residues undergoing dephosphorylation. These considerations, coupled with the high recoveries at every step ( Fig. 2 and Table I), indicate that the distribution of [32P]Tyr(P) observed reflects with fair accuracy the products of the insulin-stimulated IR intramolecular autophosphorylation per se as it occurs within the cell.
The immunoblotting data presented in Fig. 2 provide some evidence that tyrosine phosphatase action was effectively inhibited after cell disruption and during the immunoaffinity purification of the IRs which bear Tyr(P). Fig. 2 also verifies that affinity chromatography of crude cell extracts on columns of antiphosphotyrosine monoclonal antibody 1G2 adsorbs all of the IR which bear Tyr(P), which are then recovered in the hapten (phenyl phosphate) eluate. We have not attempted to determine whether the cell extracts contain 32P-labeled IR which did not bind to the antiphosphotyrosine monoclonal antibody column; previous observations (23) suggest that such a population of IR phosphorylated exclusively on Ser/Thr residues may be present after insulin stimulation. Moreover, we have no definitive information on the fractional recovery of total cellular IR achieved by antiphosphotyrosine immunoaffinity chromatography.
In analyzing the hapten eluates of the monoclonal antiphosphotyrosine immunoaffinity columns, a striking concordance is observed between the array of 32P-proteins eluted when extracts of 32P-labeled, insulin-treated hepatoma cells are applied (Fig. 1) as compared to the proteins labeled by lZ5I-protein A when corresponding nonradioactive eluates are immunoblotted with an independently generated, polyclonal anti-Tyr(P) antibody (Fig. 2). In addition to the / 3 (Mr 95,000) subunit of the insulin receptor, only one other major insulinstimulated Tyr(P)-containing protein is detected in these extracts, at M, 180,000, first detected by White et al. (37).
The immunological evidence that this protein contains Tyr(P) (Figs. 1 and 2) is verified by two-dimensional thin-layer electrophoresis of an acid hydrolysate (Fig. IC). The M, 180,000 protein is clearly distinguishable from the IR by its lack of cross-reactivity with several anti-receptor antibodies (including two directed at synthetic peptides which encompass the major human IR ,6 subunit autophosphorylation sites); it is also less efficiently adsorbed by the monoclonal antiphosphotyrosine antibody. The identity of this protein is as yet unknown.
Final purification of [32P]Tyr(P)-containing p subunit is obtained by SDS-PAGE with high recovery on elution and acetone precipitation. The strategy employed herein for identification of the Tyr(P) autophosphorylation sites is adapted from our recent studies characterizing the Tyr(P) residues on the partially purified human IR autophosphorylated in uitro (21). In essence, the tryptic peptides containing Tyr(P) are first purified by antiphosphotyrosine immunoaffinity chromatography; after further separation, partial amino acid sequence of the purified [32P]Tyr(P) tryptic peptides is obtained and compared with the amino acid sequence deduced from the nucleotide sequence. Finally, assignment of 3zP to individual tyrosine residues is accomplished by a combination of subdigestion of the tryptic peptides with S. aureus V8 protease and identification of PTH-[32P]Tyr(P) during sequential Edman degradation of 32P-peptides coupled to solid supports. The major technical problems encountered are due to the very low mass of peptide available for microsequence analysis, despite the excellent recoveries at each step in the purification. The low levels (<20 pmol) of PTH-derivatives recovered on gas-phase sequencing are nevertheless sufficient for rigorous identification of individual residues; and the partial Insulin Receptor Autophosphorylation in Cells sequence obtained, coupled with the amino acid sequence deduced from the nucleotide sequence, permits unequivocal identification of tryptic peptides. This information is augmented by the direct identification of PTH-[32P]Tyr(P) at specific cycles on solid-phase sequencing. The possibility of coeluting peptides cannot be eliminated by the data from gasphase sequencing, but is addressed effectively by the more sensitive analysis of PTH-[32P]Tyr(P) provided by solidphase sequencing. Two circumstances greatly facilitated this analysis. First, there is very great similarity between the tyrosine-containing peptides in the human and rat insulin receptor @ subunits. Lewis et uL4 recently completed the sequence of a segment of rat genome encompassing the intracellular extension of the rat IR p subunit. They found the amino acid sequence (carboxyl-terminal to the residue equivalent to human insulin proreceptor residue 943 (1)) to be 95% identical to the corresponding regions of the human IR. With regard to possible autophosphorylation sites, the intracellular extensions of the rat IR /3 subunit contain 13 tyrosines, distributed on eight potential tryptic peptides. Seven of these eight peptides (encompassing 12 of the 13 tyrosines) are identical in length, charge, and location of the tyrosines from the amino terminus and are 95% identical in amino acid sequence to tyrosinecontaining tryptic peptides derived from the human IR @ subunit intracellular extension. One tyrosine-containing tryptic peptide is completely dissimilar in that the tyrosine at 1075 in the human IR is replaced by a histidine in the rat IR, whereas the histidine at 1266 in the human IR is a tyrosine in the rat IR. Thus, exclusive of this one peptide (which contains a tyrosine not among those shown to be phosphorylated in vitro in the human IR, Ref. 21), the rat and human IR /3 subunits share virtually identical tyrosine-containing tryptic peptides. This provides the basis for the very similar chromatographic profiles of tryptic digests of [32P]Tyr(P) human IR (21) and [3ZP]Tyr(P) rat IR peptides. A second useful feature is the negligible amounts of [3zP]Ser/[32P] Thr(P) on the [32P]Tyr(P)-containing tryptic peptides (Fig.  3B) derived from the rat IR autophosphorylated in the intact cell. This circumstance increased the accuracy of estimates of relative distribution of [32P]Tyr(P), as well as assignment to specific tyrosine residues.
In conclusion, in intact rat hepatoma cells, insulin stimulates the autophosphorylation of at least six of the 13 tyrosine residues on the intracellular extension of the IR subunit.
The Tyr(P) residues are clustered and distributed on separate domains, including a peptide which contains 3 Tyr(P) residues within 6 residues, located in the tyrosine kinase segment, and a second major peptide segment near the carboxyl terminus, which contains 2 Tyr(P) residues. This distribution corresponds closely to that observed previously for the partially purified human IR autophosphorylated in uitro in the presence of insulin and MnATP. Whereas much evidence indicates that overall 6 subunit autophosphorylation is critical to kinase activation and to intracellular signaling, the functional role of each phosphorylated domain cannot be rigorously defined at present. For example, the data of Ellis et al. (38) indicated that replacement of the tyrosine corresponding to residue 7 on peptide 4 (or both residues 7 and 8) with phenylalanine by site-directed mutagenesis gave mutant insulin receptor species with a diminished (Phe7 + Phe' = abolished) insulin-stimulated component of p subunit autophosphorylation (despite substantial basal autophosphorylation) and substrate phosphorylation when examined in uitro with cell-free preparations of insulin receptor. Yet, when expressed in the intact cell, these two mutant IRs (Chinese hamster ovary YF1 and YF3) exhibited persistent insulin-stimulated p subunit tyrosine phosphorylation and continued (albeit impaired) ability to signal cellular activation in response to insulin. These rather confounding results are probably attributable, in part, to the complexity dictated by the multiplicity of tyrosines participating in the autophosphorylation reaction shown in this report and, in part, to the likelihood that mutant insulin receptor may interact with and potentially be transphosphorylated by endogenous wild-type insulin receptors, especially in the intact cell. Additional mutagenesis (preferably with expression in cells lacking endogenous insulin receptor) and further characterization of the kinetics of autophosphorylation in conjunction with kinase activation and biologic signaling will be necessary for the elucidation of the operation of this complex cascade of regulatory tyrosine phosphorylation.