Substrate phosphorylation catalyzed by the insulin receptor tyrosine kinase. Kinetic correlation to autophosphorylation of specific sites in the beta subunit.

The kinetics of insulin-stimulated autophosphorylation of specific tyrosines in the beta subunit of the mouse insulin receptor and activation of receptor kinase-catalyzed phosphorylation of a model substrate were compared. The deduced amino acid sequence of the mouse proreceptor was determined to locate tyrosine-containing tryptic peptides. Receptor was first incubated with unlabeled ATP to occupy nonrelevant autophosphorylation sites, after which [32P]autophosphorylation at relevant sites and attendant activation of substrate phosphorylation were initiated with [gamma-32P]ATP and insulin. Activation of substrate phosphorylation underwent an initial lag of 10-20 s during which there was substantial 32P-autophosphorylation of tryptic phosphopeptides p2 and p3, but not p1. Following the lag, incorporation of 32P into p1 and the activation of substrate phosphorylation increased abruptly and exhibited identical kinetics. The addition of substrate to the receptor prior to ATP inhibits insulin-stimulated autophosphorylation, and consequently substrate phosphorylation. Insulin-stimulated autophosphorylation of the receptor in the presence of substrate inhibited primarily the incorporation of 32P into p1 and drastically inhibited substrate phosphorylation. From Edman radiosequencing of 32P-labeled p1, p2, and p3 and the amino acid sequence of the mouse receptor, the location of each phosphopeptide within the beta subunit was determined. Further characterization of these phosphopeptides revealed that p1 and p2 represent the triply and doubly phosphorylated forms, respectively, of the region within the tyrosine kinase domain containing tyrosines 1148, 1152, and 1153. The doubly phosphorylated forms contain phosphotyrosines either at positions 1148 and 1152/1153 or positions 1152 and 1153. These results indicate that insulin stimulates sequential autophosphorylation of tyrosines 1148, 1152 and 1153, and that the transition from the doubly to the triply phosphorylated forms is primarily responsible for the activation of substrate phosphorylation.


Substrate Phosphorylation Catalyzed by the Insulin Receptor
The kinetics of insulin-stimulated autophosphorylation of specific tyrosines in the j 3 subunit of the mouse insulin receptor and activation of receptor kinase-catalyzed phosphorylation of a model substrate were compared. The deduced amino acid sequence of the mouse proreceptor was determined to locate tyrosine-containing tryptic peptides. Receptor was first incubated with unlabeled ATP to occupy nonrelevant autophosphorylation sites, after which [32P]autophosphorylation at relevant sites and attendant activation of substrate phosphorylation were initiated with [ T -~~P I A T P and insulin. Activation of substrate phosphorylation underwent an initial lag of 10-20 s during which there was substantial 32P-autophosphorylation of tryptic phosphopeptides p2 and p3, but not pl. Following the lag, incorporation of 32P into p l and the activation of substrate phosphorylation increased abruptly and exhibited identical kinetics. The addition of substrate to the receptor prior to ATP inhibits insulin-stimulated autophosphorylation, and consequently substrate phosphorylation. Insulin-stimulated autophosphorylation of the receptor in the presence of substrate inhibited primarily the incorporation of 32P into p l and drastically inhibited substrate phosphorylation. From Edman radiosequencing of 32P-labeled p l , p2, and p3 and the amino acid sequence of the mouse receptor, the location of each phosphopeptide within the B subunit was determined. Further characterization of these phosphopeptides revealed that p l and p2 represent the triply and doubly phosphorylated forms, respectively, of the region within the tyrosine kinase domain containing tyrosines 1148, 1152, and 1153. The doubly phosphorylated forms contain phosphotyrosines either at positions 1148 and 115211153 or positions 1152 and 1153. These results indicate that insulin stimulates sequential autophosphorylation of tyrosines 1148, 1152 and 1153, and that the transition from the doubly to the triply phosphorylated forms is primarily responsible for the activation of substrate phosphorylation.
Upon binding to its specific cell-surface receptor, insulin * This work was supported by Research Grant NIDDK-14574 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 "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) 505 149.
tutes of Health. initiates a pleiotropic cellular response, notably the activation of energy storage processes including glucose uptake, glycogenesis, and lipogenesis (1). The insulin receptor is a transmembrane allosteric enzyme composed of two types of subunits, i.e. CY and 0 subunits, that are stabilized in a 0-01-01-0 tetrameric structure by intersubunit disulfide bonds (2,3). The CY subunit which is entirely extracellular contains the insulin-binding site, while the 0 subunit which spans the plasma membrane houses a cytoplasmic tyrosine kinase catalytic domain (4, 5 ) . Studies with purified insulin receptor preparations (6)(7)(8) and with intact cells (9,10) have shown that the interaction of insulin with its binding site in the 01 subunit stimulates autophosphorylation of multiple tyrosyl groups within the intracellular kinase domain.
The mechanism by which the insulin-induced allosteric signal is transmitted across the plasma membrane is unknown. Nevertheless, there is substantial evidence (11)(12)(13)(14) that insulin-stimulated autophosphorylation of the subunit causes activation of kinase-catalyzed phosphorylation of model protein substrates (and presumably cellular protein substrates). Thus, Rosen et al. (11) first demonstrated that the lag in insulin-activated substrate phosphorylation (initiated with ATP) is eliminated by prior incubation of the receptor with insulin and ATP. Since the receptor underwent autophosphorylation (on tyrosine) during preliminary incubation, it appeared that insulin-stimulated autophosphorylation might be an essential step in the activation process. This interpretation was substantiated by Yu and Czech (12) who showed that removal of the phosphoryl groups (added during autophosphorylation) with alkaline phosphatase reversed the activation of substrate phosphorylation.
Compelling evidence that insulin-stimulated autophosphorylation induces activation of substrate phosphorylation has been obtained using protein substrates as inhibitors of insulin-stimulated autophosphorylation (15)(16)(17)(18). For example, it was shown (15,16) that RCAM-lysozyme,' which is an excellent model substrate of the insulin receptor tyrosine kinase (K, -10 pM), is also a potent inhibitor of insulin-stimulated autophosphorylation (Kt -1 @M). Thus, RCAM-lysozyme, added prior to ATP and insulin, totally blocks insulin-stimdated autophosphorylation and as a consequence, blocks insulin-stimulated substrate phosphorylation (15,16). By taking advantage of these properties of RCAM-lysozyme, the dependence of substrate phosphorylation capacity upon frac- The abbreviations used are: RCAM-lysozyme, reduced and carboxamidomethylated-lysozyme; DTT, dithiothreitol; Hepes, 4-(2-hydroxyethylbl-piperazineethansulfonic acid SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; IGF-1, insulin-like growth factor-1; HPLC, high performance liquid chromatography; MIR, mouse 3T3-Ll adipocyte insulin proreceptor; HIR, human insulin proreceptor. 21557 h u h Receptor Autoand Substrate ~h~s p h o r y~~t w n tional autophosphorylation (stimulated by insulin) of the receptor was demonstrated. It was also determined that maximal insulin-stimulated autophosphorylation and substrate phosphorylation capacity were achieved before all autophosphorylation sites had been occupied. This finding revealed that not all sites of autophosphorylation are involved in the activation process and prompted efforts to identify the responsible sites.
Knowledge of the complete amino acid sequence of the human insulin receptor (4,5) greatly facilitated identification of tyrosyl autophosphory~ation sites within the p subunit.
Thus, it became possible to locate candidate autophosphorylation sites based on consensus amino acid sequences at tyrosyl residues phosphorylated by other tyrosine-specific kinases. Moreover, by analysis of ["P]phosphopeptides generated by proteolytic cleavage of the 32P-labeled receptor @ subunit at sites predicted by the primary sequence, it has been determined that at least 6 tyrosyl residues undergo phosphorylation during insulin-stimulated autophospho~lation (19). Most of these tyrosines are located in two segments of the p subunit. One segment is located within the putative catalytic tyrosine kinase domain and contains tyrosines 1146, 1150, and 1151' clustered in a small region, while the other segment contains tyrosines 1316 and 1322 near the COOHterminal end of the p subunit (19). A number of investigations (see "Discussion" and Refs. 20-23) have implicated autophosphorylation at tyrosines 1146,1150, and 1151 in the activation of protein substrate phosphorylation catalyzed by the receptor.
In the present study, we have identified the earliest autophosphorylation events, i.e, between 0 and 60 s, folIowing the addition of insulin to insulin receptor from 3T3-Ll adipocytes under conditions that dissociate these events from those that occur in the absence of insulin. Insulin-stimulated autophosphorylation at specific sites in the @ subunit was closely correlated kinetically with the activation of model substrate (RCAM-lysozyme) phosphorylation. Likewise, inhibition of autophosphorylation of these sites by RCAM-lysozyme was closely correlated with the loss of substrate phosphorylation capacity. To facilitate the identification of these sites within the p subunit, the deduced amino acid sequence of the insulin receptor from mouse 3T3-Ll adipocytes was determined. Our results provide strong kinetic evidence that insulin-induced activation of substrate phosphorylation involves obligatory autophosphorylation of neighboring tyrosines 1148,1152, and 1153 in the mouse insulin receptor (equivalent to tyrosines 1146, 1150, and 1151 in the human insulin receptor). Moreover, partial phosphorylation of this domain (i.e. tyrosines 1148 and 1152/1153 or 1152 and 1153) precedes full phosphorylation and activation of substrate phosphorylation.

Amino Acid Sequence of the Insulin Receptor of Mouse 3?3-
LI Adipocytes-To identify the site(s) of insulin-activated autophosphorylation within the fi subunit of the insulin receptor of mouse 3T3-Ll adipocytes, it was necessary to know the amino acid sequence of receptor from this source. Therefore, cDNAs which encode the proreceptor were isolated and se- The numbering system used in that reported for the human insulin receptor by Ullrich et al. (4).
:1 Portions of this paper (including "Experimental Procedures" and Figs. 1-5 and 13-15) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that i s available from Waverly Press.
____ quenced. A cDNA library prepared with mRNA from differentiated 3T3-Ll adipocytes (26) was screened both with a 3.5kilobase human insulin receptor probe and a 700-base pair mouse insulin receptor probe. Twenty-one independent cDNA clones, representing different overlapping segments of the translated region of the mRNA, were isolated and the inserts sequenced from their 5' and 3' ends. In addition, the inserts from three of these clones (XIRc-2, XIRc-9, and XIRc-11), which together comprise the entire translated sequence of the proreceptor, were ligated at common restriction sites to generate two pBluescript clones: PET-IR, which contains the entire coding sequence and p2-9, which is the fusion product of the inserts of XIRc-2 and XIRc-9. PET-IR and a group of 5' exonuclease 111 deletion plasmids generated from plasmid p2-9 were sequenced using vector primers or synthetic oligonucleotide primers derived from sequences of several of the cDNA clones (see Fig. 4, Miniprint). The sequences obtained from these clones were assembled to generate the full-len~h nucleotide-coding sequence (see Fig. 5, Miniprint) for the mouse insulin receptor cDNA.
The deduced amino acid sequence of the mouse 3T3-Ll adipocyte insulin proreceptor is shown in Fig. 6 (see Miniprint), along with the corresponding sequence of the human proreceptor. As in the case of the human proreceptor, the mouse proreceptor contains a 27-residue signal peptide? This is followed by 1345 residues, compared with 1343 residues in the human proreceptor, which give rise to the a and fi subunits of the mature receptor (4). The two additional amino acids (threonine and proline) in the mouse proreceptor are inserted in the a subunit following a glutamine residue at position 546, thereby shifting the numbering beyond this point in the mouse proreceptor by +2 amino acid residues with respect to the human proreceptor. The overall amino acid sequence identity between the two receptors is 95%, with the a and / I subunits exhibiting 97% and 94% identity, respectively. The positions of the 36 cysteine residues in the a subunit are totally conserved in both receptors, while 9 of 10 cysteines in the @ subunit of the human receptor are conserved in the mouse receptor with the exception of cysteine 971, which is substituted by a serine residue. The number of relative positions of potential N-linked glycosylation sites are the same in both receptors, and the sequence of the COOH terminus of the a subunit, which contains the putative proteolytic cleavage site and gives rise to the cy and fi subunits, is totally conserved.
The transmembrane sequences of the subunit in both receptors are identical and the intracellular domain, which contains the tyrosine kinase catalytic site, is highly conserved exhibiting 95% amino acid sequence identity. The intracellular segment of the @ subunit, which begins at arginine 943 (941 in the human proreceptor) contains a consensus ATP-binding site (Gly-X-Gly-X-X-Gly; residues 993-998), 50-amino acid residues from the transmembrane domain. The subunit also contains a lysine residue at position 1020 that is equivalent to lysine 1018 in the human proreceptor which has been shown to be essential for tyrosine kinase activity (31). The only region of significant sequence divergence between the mouse and the human receptors is at the NH2 terminus of the @ subunit encompassing residues 728-755 (presumably an extracellular region), where the amino acid sequence identity is 57%.
Since autophosphorylation of the insulin receptor occurs only on tyrosine residues within the cytoplasmic domain of the p subunit, it was important to locate the position of each ' The alignment of the mouse and human proreceptors was made assuming that the first NHn-terminal residue at the mature a subunit is the same (histidine) in both cases.

D Y Y R K G G K G L L P Y R W H A P E S L K D G V F T T~~D~S F G W L W E I T S U E O P Y O G L S N E O V L K F V H D G G Y L D O P D N C P E R V T D~C W O F N P N~~F~I Y N L~D O L H P S F P E V S F F H S E E N~E S E E L E~F E D~~L D R S S H C
tyrosine within this domain. The intracellular domain of the (3 subunit of the mouse receptor contains 13 tyrosine residues, 1 2 of which exhibit positional identity to those in the human receptor. Tyrosine 1075 in the human receptor is substituted by histidine (at position 1077) in the mouse receptor, while tyrosine 1268 in the mouse receptor substitutes for histidine 1266 in the human receptor. The conserved tyrosines in the mouse receptor include all tyrosines considered as candidates for autophosphorylation sites in the human insulin receptor (19). In the mouse proreceptor, these tyrosines are found at positions 962, 1148, 1152, 1153, 1318, and 1324 which are equivalent to tyrosines 960, 1146, 1150, 1151, 1316, and 1322 in the human insulin proreceptor.
Effect of Insulin on the Kinetics of Receptor Autophosphorylation and Catalysis of Substrate Phosphorylation-Preliminary experiments were conducted (see "Experimental Procedures") to verify that the purified insulin receptor and its isolated p subunit were suitable for use in experiments to identify specific autophosphorylation sites required for the activation of protein substrate phosphorylation. It was established 1) that autophosphorylation of the 95-kDa (3 subunit is markedly activated by insulin ( Fig. 1, Miniprint); 2) that the insulin receptor preparation does not contain active IGF-1 receptor, 3) that the K, for insulin is the same (2-4 nM) for both autophosphorylation and RCAM-lysozyme phosphorylation; 4) that following autophosphorylation, the "P-labeled (3 subunit of the insulin receptor can be quantitatively recovered by immunoprecipitation with two different types of antibodies directed against the insulin receptor; 5) that the receptor preparation is free of an ATPase or a phosphotyrosine protein phosphatase capable of dephosphorylating autophosphorylated [32P]phospho-insulin receptor; 6) that [3'P]phosphoryl groups (on tyrosyl residues) of the (3 subunit do not turn over or undergo transfer to substrate during the course of autophosphorylation or substrate phosphorylation; 7) that the autophosphorylated ''P-B subunit, isolated as a homogeneous polypeptide (Fig. 3, Miniprint) by two-dimensional nonreducing/reducing SDS-PAGE, contains only [32P]phosphotyrosine with no detectable [32P]phospho-serine or -threonine (Fig. 2, Miniprint Supplement); and 8) that the model substrate, RCAM-lysozyme, is phosphorylated by the receptor exclusively on tyrosine residues (Fig. 3). On the basis of these results we concluded that the insulin receptor preparation was suitable for use in the autophosphorylation site analysis experiments described below.
When autophosphorylation of the receptor is conducted in the presence of insulin, there is an 8-to 10-fold stimulation in rate during the first 10 min of reaction relative to the basal rate in the absence of insulin (Fig. 7 A ) . As is evident from the progress curve, autophosphorylation in the presence of insulin reaches an apparent maximum after approximately 30 min ( Fig. 7 A ) , whereas the maximal capacity of the receptor kinase to catalyze phosphorylation of RCAM-lysozyme is achieved within 5 min of autophosphorylation (Fig. 7 B ) . Since insulin-stimulated autophosphorylation achieved only 60% of its maximum in 5 min, it is clear that full activation of the receptor tyrosine kinase does not require the maximal extent of autophosphorylation. These results are consistent with kinetic evidence obtained earlier in this laboratory (15) for activating and nonactivating components of autophosphorylation. Thus, at least half of the maximal activation of substrate phosphorylation occurs within the first minute of insulin-stimulated autophosphorylation, at which time autophosphorylation has reached approximately 37% of its maximum (see Fig. 7A).
The results illustrated in Fig. 7 show that in the absence of insulin, autophosphorylation approaches a plateau by 60 min having reached about 40% of the level achieved with insulin present (Fig. 7 A ) . In contrast, the rate of substrate phosphorylation in the absence of insulin plateaued within about 15 min at which point it had achieved about 20% of the maximal insulin-stimulated rate. These findings validate previous suggestions (15) that the nature of autophosphorylation differs in the absence versus the presence of insulin. When the addition of insulin is delayed until autophosphorylation has proceeded for 15 min (without insulin), there is an immediate stimulation of autophosphorylation (Fig. 7A, at arrow) and an associated activation of substrate phosphorylation (Fig. 7B, at arrow). The progress curves for both activities and the half-times for reaching their maximal levels are similar to those obtained when insulin was added at zero time (Fig. 7).
Autophosphorylation at Specific Sites in the p Subunit of the Receptor in the Absence and Presence of Insulin-As an initial step in the identification of autophosphorylation sites required for the activation of catalysis of substrate phosphorylation, insulin receptor was incubated with [y3'P]ATP in the absence or presence of insulin after which the 32Plabeled 8 subunit was isolated by two-dimensional nonreducing/reducing SDS-PAGE (see "Experimental Procedures" and Fig. 3). The gel segments containing the 32P-labeled / 3 subunit were exhaustively digested with trypsin and the resulting [32P]phosphopeptides were separated by Cs reversephase HPLC (Fig. 8). In this experiment, the phosphorylation reaction was allowed to proceed for 15 min since insulinstimulated substrate phosphorylation reaches its maximum during this period of time (Fig. 7B).
In both the basal and insulin-stimulated cases, at least six discrete [32P]phosphopeptide peaks (hereafter referred to as "sites") were detected and designated p l , p2, p3, p4, p5, and p6, based on their order of elution during HPLC (Fig. 8). The first four peaks eluted, p l , p2, p3, and p4, have retention times of 13, 16, 20, and 29 min, respectively. Following the elution of p4, there is a variable region of decreasing radioactivity where two small peaks are occasionally detected (not shown). Peak p5 is eluted as a broad, heterogeneous peak, with a retention time of 54-58 min, while the last peak detected, p6, is eluted after 65 min. Although insulin-stimulated autophosphorylation is observed to some extent at all sites, the major stimulation occurs at sites p l , p2, and p3. It is apparent, however, that a significant fraction of the phosphorylation at all sites in the presence of insulin, represents insulin-independent autophosphorylation (see Fig. 8 were applied to a C, Synchropak reverse-phase column developed in 0.1% phosphoric acid/triethylamine, pH 6. Elution was initiated 5 min after injection with a gradient of isopropyl alcohol/acetonitrile 1:l (dotted line) and a flow rate of 1.5 ml/min (see "Experimental Procedures"). Recovery of the 3ZP-label applied was 85% in both cases. The radioactive peaks were designated pl, p2, p3, p4, p5, and p6, according to their order of elution.
panel). It should be emphasized that stimulation by insulin does not cause autophosphorylation at any "new" sites. The identification of the specific sites responsible for insulinactivated substrate phosphorylation will be described in the following section. Size analysis of pl, p2, and p3 by HPLC size-exclusion chromatography and SDS-PAGE in 10-20% gradient gels revealed that these sites represent small phosphopeptides with apparent M, values of 1,000-2,000 (results not shown). A similar analysis of the other HPLC peaks, p5 and p6, was considerably more difficult due to their insolubility and consequent poor recovery. However, the latter sites appear to correspond to heterogeneous mixtures of peptides with apparent M , values in the 2,000-30,000 range, which probably represent incompletely digested @ subunit.
Kinetic Correlation between the Activation of Substrate Phosphorylation and Autophosphorylution of Specific Sites in the Receptor's / 3 Subunit-The progression of autophosphorylation of specific sites on the receptor's p subunit was followed kinetically during the period of greatest increase in insulin-stimulated substrate phosphorylation, i.e. during the first 60 s following insulin addition (Fig. 7B). During this initial phase of the activation process, 50-60% of the maximal extent of activation of substrate phosphorylation occurs.
To minimize 32P-labeling of the autophosphorylation sites least relevant to the activation of substrate phosphorylation by insulin, the receptor was first preincubated for 15 min with unlabeled ATP in the absence of insulin. Under these conditions, the sites of autophosphorylation which contribute least to the activation (by insulin) of substrate phosphorylation  not (0, dashed line), and the initial rate of substrate (100 @M RCAM-lysozyme) phosphorylation was measured at the times indicated (between 0 and 60 s) as described under "Experimental Procedures." It should be noted that further autophosphorylation, and hence activation, is blocked from the time of addition of RCAM-lysozyme. Data are from three independent experiments.  peaks p l ,   p2, p3, p5, and p6 (indicated in the upper left corner of each panel), is plotted as a function of the time elapsed from the addition of insulin and [y-32P]ATP following autophosphorylation with unlabeled ATP in the absence of insulin for 15 min. The relative extent of activation of substrate (RCAM-lysozyme) phosphorylation capacity at each time is shown for comDarison (doshed line. ripht-hand axis). The data are from Fig. 9 and two identical experiments as described in Fig. 10.

I I
phosphorylation capacity following insulin addition are shown in Fig. 9 (solid line). Upon the addition of insulin, there is an initial lag of 10-20 s in the activation of RCAM-lysozyme phosphorylation, which then increases abruptly during the next 40 se5 By 60 s, the initial rate of substrate phosphorylation has reached nearly 60% of its maximal level. In contrast, if insulin is not added there is virtually no change in the initial rate of RCAM-lysozyme phosphorylation during the 60-s incubation with [y3'P]ATP (Fig. 9, dashed line).
The autophospho~lation of specific sites on the #3 subunit of the receptor was monitored concomitantly fo~lowing the addition of insulin and [T-~'P]ATP in an experiment identical to that described above (Fig. 9) in which the activation of substrate phosphorylation was analyzed. The progress of autophosphorylation of the major autophosphorylation sites, analyzed by HPLC mapping of the J2P-labeled tryptic phosphopeptides of the 32P-p subunit during the first 60 s following the addition of insulin and [y3*P]ATP, is shown in Fig. 10. During the first 10 s after insulin addition, a low level of 32P is detected in at least five sites, pl, p2, p3, p5, and p6 (phosphorylation of p4 was not observed).
[32P]Phosphorylation of p2 and p3 is markedly increased during the initial 30 s, while p l is barely detectable at 20 s, becomes significant after 30 s and is the major site by 60 s. Site 5 (p5) does not appear to be a major insulin-stimulated site relative to the other sites. While p6 is significant at all times in the analysis, its role in the activation of the receptor kinase (along with that of p5) is questionable in light of experiments to be described below.
The temporal relationship between the extent of autophosphorylation of each site and the extent of activation of substrate phosphorylation, is illustrated in Fig. 11. It is evident To rule out the possibility that the lag phase was due to a mixing effect, an experiment was conducted in which the kinetics of receptorcatalyzed R~AM-~ysozyme phosphorylation was followed at. closely spaced time intervals (s). The rate of 3zP inco~oration into RCAMlysozyme was found to be completely linear from 5 to 300 s. Moreover, when the receptor was allowed to equilibrate with 1 p~ insulin before initiating autophosphorylation at zero time with [ Y -~~P ] A T P (;.e. without a previous 15-min incubation with unlabeled ATP), the activation of substrate phosphorylation also underwent a 10-20 s lag, although in this case the contribution of the "basal" component was significant.
that the phosphorylation of site p l (first panel) correlates best with the rate of activation of the receptor. However, it should be noted that phosphorylation at site p2 occurs faster than the activation of substrate phosphorylation during the first 30 s (Fig. 11, secondpanel) preceding the phosphorylation of site pl. The phosphorylation of p2 and p3 plateaus early and remains constant for the next 30 s when the rate of activation of substrate phosphorylation is increasing rapidly (from 24 to 58% of maximal activation; see also Fig. 9).
Finally, the kinetics of phosphorylation of p5 and p6 clearly depart from that of activation of substrate phosphorylation, although in the case of p6 it is rather difficult to obtain an unequivocal trend due to the great variability observed for this peak. However, in the experiments presented below, the phosphorylation of p5 and p6 are shown to occur independent of activation of substrate phosphorylation.
Based on these findings, it is evident that insulin-induced activation of the receptor kinase (for substrate phosphorylation) most closely correlates with a u~p h o s p h o~l a t i o n of site pl. It should be noted, however, that phosphorylation at other sites (p2, p3, and p6) precedes the phosphorylation at site pl, raising the possibility that activation of the receptor kinase may be the result of a series of concerted autophosphorylation events required for phosphorylation at site pl.
Effect of Inhibiting Autophosphorylation at Sites p l and p 2 on Insulin-stimulated Substrate Phosphorylation-To further explore the role of autophosphorylation sites p l and p2 in the activation of substrate phosphorylation, the known inhibitory effect of substrate (Le. RCAM-lysozyme) on autophosphorylation (15) was investigated. We previously showed that RCAM-lysozyme is both an excellent substrate for the fully activated (autophosphorylated) insulin receptor and a potent inhibitor of insulin-stimulated autophosphorylation (15). Thus, the addition of RCAM-lysozyme prior to (or with) insulin blocks autophosphorylation and, therefore, the activation of substrate phospho~lation.
The concentration dependence of inhibition of insulinstimulated autophosphorylation and substrate phosphorylation by RCAM-lysozyme is shown in Fig. 12, A and B, respectively. In both cases, autophosphorylation was carried out for 5 min in the presence of insulin and increasing levels (0-100 FM) of RCAM-lysozyme. In Fig. 9B, substrate phosphoryla- These inhibition studies provided a means by which the effects of RCAM-lysozyme on site-specific autophosphorylation and substrate phosphorylation could be correlated. After a preliminary incubation of the receptor with unlabeled ATP for 15 min (as described above, Figs. 9 and lo), insulin and [y3'P]ATP were added with or without 2 p~ RCAM-lysozyme, the K, concentration for overall autophosphorylation.
At this concentration, insulin-stimulated substrate phosphorylation is inhibited by 85% (see Fig. 12B). After 60 s, the reaction was terminated and an autophosphorylation site analysis was performed. As shown in Fig. 12C, RCAM-lysozyme inhibits the phosphorylation of p l by 82%, and p2 by 31% (see quantitation in legend of Fig. 12C). In contrast, the phosphorylation of p3, p5, and p6 were minimally inhibited. These results indicate that p l and possibly p2, are the major autophosphorylation sites involved in insulin-activated substrate phosphorylation, whereas p3, p5, and p6 do not appear to play an essential role in this process. These findings are consistent with the results described above (see Figs. 9 and 10).

Assignment of the Positions of Autophosphorylation Sites p l , p2, and p3 to Specific Loci within the 0 Subunit of the
Receptor-From the primary structure of the insulin receptor of 3T3-U adipocytes (see Fig. 6), the amino acid sequences of all tyrosine-containing peptides that would be generated by complete digestion of the p subunit with trypsin can be predicted (Table I). Three of these tryptic peptides, T1, T3, and T5, can be eliminated as candidates for identity with p l , p2, and p3 on the basis of molecular weight. As indicated above pl, p2, and p3 have molecular weights (Mr = 1000-2000) much lower than those calculated for T1, T3, and T5.
Since the quantities available of the tryptic phosphopeptides corresponding to autophosphorylation sites pl, p2, and p3 were insufficient for direct amino acid sequencing, an alternative approach was used. Given that phosphotyrosine is the only phosphoamino acid present in the p subunit of receptor that has undergone insulin-stimulated autophosphorylation (Fig. 2), it was possible to determine the position(s) of [32P]phosphotyr~~ine in autophosphorylation sites p l , p2, and p3 by Edman radiosequencing. With this information and knowledge of the complete amino acid sequence of the 3T3-Ll adipocyte receptor (Fig. 6), the specific positions of these autophosphorylation sites within the / 3 subunit were deduced. Fig. 13 (Miniprint) shows the level of 32P release (derived from [32P]phosphotyrosine) at each cycle in gas-phase amino acid sequencing runs on each of the three [32P]phosphopeptides, i.e. pl, p2, and p3. It should be noted that the phosphotyrosine phenylthiohydantoin-derivative formed in the gasphase sequenator is not efficiently released due to its insolubility in the extraction solvent which gives rise to the leaching of 32P observed in subsequent cycles (19). The results are expressed relative to the total amount of 32P radioactivity recovered in 20 cycles of Edman degradation. The patterns of

Predicted ~y~s~~-c o n~a~n i n~ tryptic peptides from the c y t~p~m i c domain of the /3 subunit of the mouse insulin receptor
The tyrosine-containing peptides that would be generated upon complete tryptic digestion of the cytoplasmic domain of the /3 subunit are shown (numbered T1 through TS) in order of their appearance in the primary structure of 3T3-LI adipocyte insulin receptor (see Fig. 6 (Table I).
Similarly, site p2 also gave rise to the release of radioactivity at cycle 3, with an additional smaller release in cycle 7. This suggested that p2 and p l represent the same peptide, T4, phosphorylated only at two of its 3 tyrosine residues (see next section). This peptide represents a highly conserved region within the tyrosine kinase domain which contains three tyrosine residues (at positions 1148, 1152, and 1153) and is identical to its counterpart in the human insulin receptor. Edman radiosequencing of p3 gave rise to increases in the release of radioactivity in cycles 2 and 3, with subsequent increases of release in cycles 8 and 9. This pattern corresponds to the predicted positions of tyrosines in peptide T8 (Table  I), which represents another autophosphorylation site near the COOH terminus of the p subunit. This site contains 2 tyrosine residues at positions 1318 and 1324, which correspond to positions 1316 and 1322 in the human insulin receptor. In this case, alternative sites of digestion by trypsin at the NH2 terminus would be expected to generate two products which differ with respect to the presence of an additional amino-terminal arginine, thus shifting the relative position of the first tyrosine residue (see below).
Further Characterization of Autophosphorylation Sites p l , p2, and p3"The results of Edman radiosequencing of p l and p2 suggested that both tryptic phosphopeptides represent the same domain of autophosphorylation (the region containing tyrosines 1148/1152/1153), which is phosphorylated to different extents. This could occur if the multiple tyrosyl residues within this domain underwent sequential phosphorylation. It was important, therefore, to determine whether p l and p2 indeed represent the same phosphopeptide that is phosphorylated to different extents.
The initial resolution of pl, p2, and p3 by reverse-phase HPLC appeared to resolve domains of autophosphorylation but might not have yielded homogeneous phosphopeptides; therefore, each of the three phosphopeptides was subjected to further fractionation by ion-exchange chromatography. Charge heterogeneity within a phosphopeptide could not only result from differences in number of p h o s p h o~l groups but also from differences in the number of the COOH terminal basic amino acid residues arising from differential tryptic cleavage of pairs of basic amino acids (see Table I).
Fractions from the C, reverse-phase HPLC column containing each peak (pl and p2) were separately pooled and subjected to anion-exchange chromatography on a Mono Q FPLC column (Pharmacia LKB Biotechnology Inc.). Both p l and p2 gave rise to two [32P]phosphopeptides, designated plA and plB or p2A and p2B, as shown in Fig. 14 (Miniprint). AS expected, both plA and plB were eluted after p2A or p2B, consistent with their higher phosphoryl group content (compare retention times in Fig. 1 4 plA, 30 min and plB, 34 min uersus p2A, 21 min and pZB, 26 min).
As shown in Fig. 15 (Miniprint), secondarypeptide mapping of these phosphopeptides by thin layer electrophoresis indicated that plA and plB correspond to two distinct homogeneous species which are considerably more acidic than p2A and p2B (compare lanes 1 and 2 with lanes 5 and 6 in Fig.   15). As expected from their order of elution from the Mono Q column, plB is more acidic than plA, to about the same extent as p2B is more acidic than p2A.
Since the amino acid sequence of the pl/p2 tryptic peptide which contains tyrosines 1148/1152/1153 includes only a single glutamic acid at position 1147 (interposed between tyrosine 1148 and tyrosines 115211153; see Table I, peptide  * The arrowheads indicate tyrosine residues that are phosphorylated. e The monophosphorylated character of p2A is inferred (see "Results" for details).
T4), cleavage after this residue with staphylococcal V8 protease should generate two tyrosine-bearing peptide products. Determination of the extents of phosphorylation of these peptides, relative to each other, should indicate whether one, two, or all three tyrosines are phosphorylated. As expected, subdigestion of plA and plB with V8 protease yielded two discrete phosphopeptides, the most acidic of which exhibited the same electrophoretic mobility (see lanes 3 and 4 in Fig.  15). The other phosphopeptide product in each case behaved as a less acidic species than the corresponding parental phosphopeptide. Quantitation of the radioactivity associated with each phosphopeptide product revealed that their 32P contents were in a 2:l ratio, where of the radioactivity corresponds to the least acidic phosphopeptide (see quantitation in legend to Fig. 15). Thus, this phosphopeptide corresponds to the COOH-terminal fragment bearing two phosphorylated tyrosines (positions 1152 and 1153) while the most acidic phosphopeptide represents the NH,-terminal fragment containing one phosphorylated tyrosine (position 1148). Taken together with the results of the Edman radiosequencing experiments, we conclude that pl represents a triply tyrosine-phosphorylated form of the peptide containing tyrosines 1148/1152/ 1153. The occurrence of subspecies plA and plB is due to variable digestion by trypsin at the COOH terminus of this peptide, which would be expected to generate two products which differ by an additional lysine at the COOH terminus (Table I), and hence, differ in charge. It should be noted that limited subdigestion of plA with carboxypeptidase B, generates a phosphopeptide with the same electrophoretic mobility as plB, confirming that these phosphopeptides differ only by the presence of an additional COOH-terminal basic residue, in this case lysine (results not shown).
For the analysis of p2 by digestion with V8 protease, phosphopeptide p2B was used since it constituted the major component (by a factor of 2.5-fold) after purification by Mono Q anion-exchange chromatography (see Fig. 14). When p2B was digested with V8 protease, three widely separated phospho-peptides were produced whose 32P contents were in the ratio 1:0.75:1 in descending order of mobility, as shown in Fig. 15 (lune 8 ) . The most acidic phosphopeptide had the same mobility as the most acidic V8 protease digestion product of plA or plB, and represents the NH2-terminal fragment of the autophosphorylation domain, which contains tyrosine 1148. The intermediate-mobility [32P]phosphopeptide product behaved similarly to the least acidic V8 digestion product of plA, comprising the COOH-terminal fragment of the same domain and bearing two [32P]phosphotyrosines (1152 and 1153). Finally, the least acidic phosphopeptide represented a new species not observed above which exhibited the lowest electrophoretic mobility of all the phosphopeptides described. We believe that this phosphopeptide corresponds to the COOH-terminal fragment of the autophosphorylation domain, which contains only one phosphorylated tyrosine (either 1152 OT 1153; see Table 11).
Our interpretation of these results is that the major species of p2, i.e. p2B, corresponds to the same peptide as pl, but is composed of a mixture of two doubly phosphorylated forms, one involving tyrosines 1148 and 1152 or 1153, the other involving tyrosines 1152 and 1153. In support of this interpretation is the fact that the ratio of phosphorylation (32P) of the most acidic to the least acidic [32P]phosphopeptide V8 protease digestion product is 1:l. This indicates that these products arose from the original peptide phosphorylated at the least acidic V8 digestion product of plA suggests that the species comprising p2B represents the doubly phosphorylated counterpart of plA (but not plB), bearing two basic amino acid residues at the COOH terminus (arginine and lysine; see Table I). If this is the case, the only alternative explanation to account for the less acidic character of p2A and p2B is that the former represents a monophosphorylated form of the tyrosine 1148/1152/1153 domain, bearing two basic amino acid residues (arginine and lysine) at the COOH terminus, which would render it more basic than p2B. In contrast, the more acidic p2B represents a heterogeneous mixture of doubly phosphorylated forms of the same domain, as described above. Thus, limited subdigestion of p2A with carboxypeptidase B (Fig. 15, lane 7) removes one COOH-terminal basic amino acid residue (ie. lysine, giving rise to a phosphopeptide with one phosphoryl group and only one basic amino acid (arginine) at the COOH terminus, shifting its mobility to that of p2B, which contains two phosphoryl groups and two basic COOH-terminal amino acid residues. Since p2A is a minor component of p2, further attempts were not made to verify its possible monophosphorylated character.
Finally, based on the ratio of the 32P contents of the V8 protease digestion products of p2B, the level of the doubly phosphorylated form involving tyrosine 1148 and 1152 or 1153 is approximately 2.7-fold higher (2/0.75) than the doubly phosphorylated form involving vicinal tyrosines 1152 and 1153. Taken together, these results suggest that activation of substrate phosphorylation, caused by insulin-stimulated autophosphorylation of the receptor, involves the random addition of three phosphoryl groups to tyrosines 1148, 1152 and 1153 in the p subunit.
A similar analysis was performed for the autophosphorylation domain represented by tryptic phosphopeptide p3, which contains tyrosines 1318 and 1324 near the carboxyl terminus of the p subunit (see Table I). As shown in Fig. 14, further fractionation of p3 by anion-exchange chromatography gives rise to two [32P]phosphopeptides designated p3A and p3B, where p3A represents the major component by a factor of >3fold (Fig. 14). When analyzed by thin-layer electrophoresis, these phosphopeptides behaved as homogeneous species differing in charge with p3A being less acidic than p3B (Fig. 15, lane 9 and IO), consistent with their order of elution from the Mono Q column (Fig. 14). Limited subdigestion of p3A with carboxypeptidase B shifted its mobility to that of p3B (Fig.  25, lane I I ) , suggesting that these species differ by one basic amino acid residue at the COOH terminus. It should be noted, however, that in this particular case charge heterogeneity might also derive from variable tryptic digestion at the aminoterminal end of this domain, where there are two potential sites of tryptic cleavage (lysine 1315 and arginine 1316; see Table I).
Digestion of p3A with V8 protease generated two [32P] phosphopeptide products in an approximate ratio of 1:l (Fig.  15, lune 12). The most acidic phosphopeptide product exhibited an electrophoretic mobility higher than its parental peptide (p3A) and corresponds to the amino-terminal fragment of the autophosphorylation domain containing a phosphorylated tyrosine at position 1316. The other product barely migrated from the origin and corresponds to the carboxylterminal fragment of the same domain, phosphorylated at tyrosine 1324.7 Were p2A the result of tryptic cleavage at arginine 1316, the amino-terminal fragment generated by V8 These assignments were made on the basis of the calculated mass/charge ratio of each V8 protease digestion product, as compared to the relative mobility of known phosphopeptides with a fixed number of phosphoryl groups. protease digestion would have the sequence Thr-(P)Tyr-Asp-Glu. This phosphopeptide would be expected to have a very similar charge/mass ratio (and presumably a comparably electrophoretic mobility) to the corresponding amino-terminal V8 protease digestion fragment from the tyrosine 1148/1152/ 1153 domain, whose sequence is Glu-Ile-(P)Tyr-Glu. However, the acidic V8 protease digestion product from p3A migrated as a less acidic species (Fig. 15, lane 12) than the corresponding product from plA, plB, or p2B (see above). These observations suggest that p3A results from tryptic digestion after lysine 1315 (and not arginine 1316), thus generating a product which contains an additional arginine residue at the NH2 terminus (thereby rendering it more basic and reducing its electrophoretic mobility).
We conclude that p3 represents an autophosphorylation domain near the COOH terminus of the @ subunit containing phosphorylated tyrosines 1318 and 1324. Hence, p3 would be expected to give rise to two 32P-labeled V8 protease digestion products in a ratio of 1:l. The species p3A and p3B represent the same peptide with differing charge characteristics probably resulting from an additional amino-terminal arginine (in p3A) generated by incomplete tryptic digestion. This interpretation is consistent with the Edman radiosequence analysis of p3, where the higher release of [32PJphosphotyrosine at cycles 3 and 9 (with respect to cycles 2 and 8; see Fig. 13), could reflect the higher abundancy observed for p3A (see Fig. 14). Table I1 summarizes the deduced structures and phosphorylation states of all the phosphopeptides described above.

DISCUSSION
There is now compelling evidence that insulin-stimulated autophosphorylation of the insulin receptor's p subunit is an essential step in the activation of substrate phosphorylation catalyzed by its intrinsic tyrosine kinase (6)(7)(8)(9)(10). Autophosphorylation has been found to occur at multiple sites in the 0 subunit and involves at least six tyrosyl groups (19,23). Several approaches have been employed to characterize the specific site(s) required for insulin-induced activation of substrate phosphorylation. First, mutation of the insulin receptor at tyrosines 2150 and 1151 result in a reduction of insulinstimulated autophosphorylation and substrate phosphorylation, both in vitro and in intact cells (20). Second, antibodies raised against a synthetic peptide corresponding to the region in the human receptor containing tyrosines 1146, 1150, and 1151, immunoprecipitated a CNBr-cleavage fragment of the p subunit whose autophosphorylation correlated with insulininduced activation of substrate phosphorylation (21). Third, an antiphosphotyrosine antibody inhibits insulin-activated substrate phosphorylation by interrupting autophosphorylation of the tyrosine 1146/1150/1151 site at the doubly phosphorylated stage (22). Finally, preferential autophosphorylation at this site occurs during the activation of substrate phosphorylation induced by insulin, although autophosphorylation at tyrosine 953 and/or 960 also correlates well with the activation process (23).
In the present investigation, a kinetic approach was used to follow autophosphorylation at specific sites in the mouse insulin receptor during the initial, acute phase ( i e . 0-60 S) of insulin-induced activation of substrate phosphorylation. Since basal autophosphorylation (in the absence of insulin) contributes substantially to overall autophosphorylation in the presence of insulin without materially activating substrate phosphorylation (Fig. 7), special precautions were taken to minimize its contribution.
The tryptic phosphopeptide patterns (analyzed by HPLC) from the insulin receptor autophosphorylated in the absence or presence of insulin were found to be qualitatively similar (Fig. 8), suggesting that in both cases common sites of autophosphorylation are involved. However, since the partial activation of substrate phosphorylation (-20%) that occurs in the absence of insulin plateaus by 10-15 min of basal autophosphorylation (Fig. ?), the contribution of this component can be minimized by preincubating the receptor with unlabeled ATP for 15 min before adding insulin and [Y-~~PIATP (Fig. 10). This step ensured prior occupancy of "basal" sites of autophosphorylation (with unlabeled phosphoryl groups) which contribute the least to substrate phosphorylation, thereby permitting us to simultaneously track insulin-induced activation of substrate phosphorylation and autophosphorylation at the sites in the p subunit most relevant to the activation process. The identification of the relevant autophosphory~ation sites in the 0 subunit was facilitated by having available the complete amino acid sequence of the mouse 3T3-Ll adipocyte insulin receptor (Fig. 6). The major predicted structural features of this receptor are similar to those of the human insulin receptor, as might be expected from the high degree of amino acid sequence identity (-95%) between the two receptors.
The intracellular domain of the 0 subunit possesses a Gly-X-Gly-X-X-Gly sequence (residues 993-998) upstream from a lysine residue at position 1020 (1018 in the human receptor) which is typical of the ATP-binding site in most kinases (32).
In addition, at least 5 of the 13 tyrosine residues in the cytoplasmic domain of the ,9 subunit in the mouse receptor correspond to autophosphorylation sites identified in the human (19) and rat insulin receptors (22). As in the human receptor, these residues in the mouse receptor are ~s t r i b u t e d between two distinct sites in the /3 subunit: tyrosines 1148, 1152, and 1153 (1146, 1150, and 1151 in the human insulin receptor) are clustered in a region considered to be part of the tyrosine kinase domain, while tyrosines 1318 and 1324 (1316 and 1322 in the human receptor) are located near the COOH terminus of the 0 subunit. This information, taken together with the results of Edman radiosequencing (Fig. 13) and secondary proteolytic digestion of tryptic ~3zP]phosphopeptides ( Fig. 15) allowed us to locate the relevant phosphotyrosines within the @ subunit of the receptor from mouse 3T3-L1 adipocytes. Activation of the receptor by insulin is rapid. Thus, within 1 min after the addition of insulin, substrate phosphorylation is activated >50% of its maximal rate (Fig. 7 B ) . This defined the narrow time window during which autophosphorylation occurs at sites most relevant to the activation process. Within the first 10-20 s of insulin addition, there is a substantial lag phase in the activation of substrate phosphorylation (Fig. 9), During this period, however, significant labeling (by [-y-"P] ATP) of tryptic phosphopeptides p2 and p3, but not p l , is observed (Figs. 10 and 11). Following this lag, Le. between 20 and 60 s, activation of substrate phosphorylation increases abruptly following virtually identical kinetics to the labeling of phosphopeptide pl, while the rate of labeling of p2 and p3 have virtually reached a plateau (Fig. 11). Structura~ characterization of these phosphopeptides revealed that pl and p2 are both derived from the tyrosine 1148/1152/1153 domain but differ in their extent of phosphorylation. Thus, p l corresponds to the triply phosphorylated form of this domain, while p2 corresponds to the doubly phosphorylated form (Table 11). Finally, p3 is derived from the tyrosine 1318/1324 domain, which lies near the COOH terminus of the subunit (Table 11). In contrast to pl, the kinetics of 32P-labeling of p3 do not correlate with the insulin-induced activation of substrate phosphorylation (Fig. 11).
These findings show that the tyrosine 1148/1152/1153 domain undergoes rapid insulin-stimulated stepwise autophosphorylation. In a preliminary step preceding the acute onset of activation of substrate phosphorylation, the doubly phosphorylated form of this domain (p2) accumulates (Fig. 10). In a subsequent step, this form undergoes further autophosphorylation giving rise to the triply phosphory~ated form (~1 ) . It is only the formation of the triply phosphorylated tyrosine 1148/1152/53 domain, however, that correlates with the kinetics of activation of receptor-catalyzed substrate phosphorylation. We believe that this event corresponds to the critical transition shown by White et al. ( 2 2 ) to be interrupted by an antiphosphotyrosine antibody. Our results provide direct kinetic proof that activation of substrate phosphorylation is proportional to accumulation of the triply (and not the doubly) phosphory~ated state of the tyrosine ~148/1152/1153 domain.
Despite the evident precursor-product relationship between the doubly and triply phosphorylated tyrosine 1148/1152/ 1153 form of the receptor, the fraction of doubly phosphorylated form is still substantially detectable (Fig. 8) even after the receptor has become fully activated (Fig. ?B), i.e. after 15 min of autophosphory~ation in the presence of insulin. This finding indicates that the doubly-phosphorylated form is not completely converted into the triply phosphorylated form, suggesting that these two forms (represented by p l and p2) are both present at the end point of activation by insulin. Given the tetrameric (@-a-a-@) structure of the insulin receptor and the intramolecular character of autophosphorylation (17), these observations may be explained by an asymmetric mode of au~phosphory~ation in which the a&* tetramer has one ,9 subunit with two phosphoryl groups in the tyrosine 11~8/1~52/1153 domain, while the other p subunit has three phosphoryl groups. In this view, double phosphorylation of the tyrosine 1148/1152/1153 domain occurs first in one /? subunit and somehow triggers full phosphorylation of the corresponding domain in the contralateral 0 subunit, resulting in activation of substrate phosphorylation. It should be noted that White et al. (22) have suggested that there may be a control point at the transition between these two forms in the intact cell. Moreover, our V8 protease digestion analysis of the tryptic phosphopeptide p2 indicates that the doubly phosphorylated state of the tyrosine 1148/1152/1153 domain can assume two different isoforms, one phosphorylated at tyrosines 1148 and either 1152 or 1153 and another phosphorylated at vicinal tyrosines 1152 and 1153. The occurrence of these variants probably reflects the fact that this domain undergoes multiple phosphorylation in a nonordered sequence of tyrosine phos~horylat~o~. A~~hough the relative levels of these forms were not followed kinetically during the acute phase of the insulin-induced activation of the receptor, a t least one, and possibly both forms accumulate rapidly during the early lag phase of the activation process ( Fig. 9) because phosphopeptide p2 is detected at all time points (Fig. 10). Thus, acute activation of catalysis may not reflect uis a uis autophosphorylation of a specific tyrosine residue in the tyrosine 1148/1152/1153 domain but rather may be the result of the comphte autophosphorylation of all three tyrosines.
Additional evidence implicating the triply phosphorylated tyrosine 1148/1152/1153 domain in the activation of substrate phosphorylation was obtained through studies of the inhibition of autophosphorylation with the model substrate, RCAM-lysozyme. Experiments presented in this paper and previously (15,16) indicate that the insulin-induced component of autophosphorylation and thus, of substrate phosphorylation are selectively blocked by RCAM-lysozyme in a concentration-dependent manner (see Fig. 12, A and B ) . The major impact of RCAM-lysozyme upon the sites of autophosphorylation is the drastic inhibition of accumulation of the triply phosphorylated form ( p l , Fig. 12C), while blocking the activation of substrate phosphorylation to essentially the same extent (82 uersus 85%, respectively). In contrast, the accumulation of the doubly phosphorylated form is only partially inhibited, i.e. by 30%, and autophosphorylation at tyrosines 1318 and 1324 in the COOH-terminal domain of the /3 subunit is not affected. Thus, RCAM-lysozyme disrupts insulin-stimulated autophosphorylation within the tyrosine 1148/1152/1153 domain as reflected by a dramatic shift in the ratio of the triply to the doubly phosphorylated form (from pl/p2 = 1.3 to 0.3, in the absence and presence of RCAM-lysozyme, respectively). Thus, the nearly complete inhibition of the insulin-induced activation of substrate phosphorylation is closely correlated with a blockade of the formation of the triply phosphorylated form of this domain.
Two major observations indicate that autophosphorylation at tyrosines 1318 and 1324 (p3) in the COOH-terminal domain of the /3 subunit, does not play an essential role in the acute activation of catalysis induced by insulin, as previously suggested by Herrera and Rosen (21). First, the kinetics of 32P labeling of p3 and of the activation of substrate phosphorylation, are not appropriately correlated during the acute phase of activation by insulin (Fig. 11). Second, inhibition of insulinstimulated autophosphorylation and substrate phosphorylation with RCAM-lysozyme did not affect the incorporation of 32P into p3 (Fig. 12C). It should be noted, however, that autophosphorylation within this domain, like that in the doubly phosphorylated form of the tyrosine 1148/1152/1153 domain (p2), also precedes the acute onset of insulin-induced activation of substrate phosphorylation, rapidly reaching a plateau by 20-30 s (Fig. 11). Thus, although autophosphorylation of the tyrosine 1318/1324 domain is not kinetically correlated with the activation of substrate phosphorylation, a possible prerequisite role for this process cannot be excluded. Nevertheless, this possibility seems unlikely in view of the fact that proteolytic removal of this domain does not affect the tyrosine kinase function of the receptor (33).
Autophosphorylation at tyrosines 953 and/or 960 in the human insulin receptor has also been found to correlate with the extent of kinase activation (23). The equivalent of these tyrosines in the mouse receptor are represented by tryptic peptide T1 (see Table I), containing tyrosines 955 and 962 with a predicted M , of approximately 4000. Due to the heterogeneous composition of the fraction represented by HPLC peak 6 (p6), including [32P]phosphopeptides with M, values of 2,000-30,000 (results not shown), it is possible that tryptic phosphopeptide T1 is a constituent of p6. As stated above (see "Results"), the great variability in the phosphorylation of p6 did not permit us to unequivocally correlate it with the activation of substrate phosphorylation, although it is barely inhibited by RCAM-lysozyme (i.e. by 15-20%). Thus, it is still possible that preliminary autophosphorylation in this domain is required for the subsequent activation of catalysis. In fact, Herrera et al. (34) have suggested a critical role for this site in the activation process because autophosphorylation and substrate phosphorylation are prevented with an antibody directed against this region of the /3 subunit, but only when added before initiating autophosphorylation with ATP. However, a recent report by White et al. (35) indicates that substitution of tyrosine 960 by site-directed mutagenesis had no effect upon the tyrosine kinase function of the receptor. Additional studies will be required to ascertain the role, if any, of autophosphorylation at tyrosines 955 and/or 962 in the activation process. a -1