Kinetic Properties and Sites of Autophosphorylation of the Partially Purified Insulin Receptor from Hepatoma Cells*

Autophosphorylation of the insulin receptor was studied using a glycoprotein fraction solubilized and purified partially from the rat hepatoma cell line, Fao. Incubation of this receptor preparation with [T-~”P] ATP, Mn2+

Autophosphorylation of the insulin receptor was studied using a glycoprotein fraction solubilized and purified partially from the rat hepatoma cell line, Fao.  20-fold with no effect on the K , for ATP. Mn2+ stimulated autophosphorylation by decreasing the K,,, of the kinase for ATP, whereas Mg2+ had no effect. Dilution of the insulin receptor over a 10-fold concentration range did not decrease the rate of autophosphorylation suggesting that it may occur by an intramolecular mechanism. When the phosphorylated &subunit of the insulin receptor was digested with trypsin, at least 5 phosphopeptides could be separated by high performance liquid chromatography on a pBondapak C,S reverse-phase column. Insulin stimulated the phosphorylation of all sites. These phosphate acceptor sites varied in their rate and degree of phosphorylation. Phosphopeptides pp4 and pp5 were phosphorylated very rapidly and reached steady state within 20 s, whereas phosphorylation of ppl and pp2 required several minutes to reach steady state.
Tyrosine phosphorylation may be involved in the regulation of cellular growth and metabolism. In addition to the insulin receptor, the receptors for epidermal growth factor (17)(18)(19)(20)(21) and platelet-derived growth factor (22)(23)(24) appear to be hormone-sensitive tyrosine kinases. Cellular transformation induced by infection with the Rous sarcoma virus also seems to be mediated through the tyrosine kinase activity of its gene product, pp60""" (25). The physiologic role of autophosphorylation at tyrosine residues is unknown, although it may regulate the catalytic activity of tyrosine kinases (26) as has been suggested for some serine and threonine kinases (27).
In this report, we characterize the autophosphorylation reaction of the insulin receptor partially purified from the hepatoma cell Fao. We find that the reaction is under dual regulation. Insulin stimulates autophosphorylation by increasing the V,,,,,, whereas Mn2+ activates this reaction by decreasing the Km for ATP. Autophosphorylation of the insulin receptor kinase appears to occur at several distinct sites in vitro as determined by the separation of tryptic phosphopeptides using reverse-phase HPLC.' These sites show different time courses suggesting that autophosphorylation of one site may stimulate the autophosphorylation of other sites. Autophosphorylation of the (3-subunit of the insulin receptor at tyrosine residues may be the first molecular event to signal intracellularly the binding of insulin on the external surface of the plasma membrane.

EXPERIMENTAL PROCEDURES AND RESULTS3
T h e Time Course of Insulin Receptor Autophosphorylation-An autoradiogram showing the time course of in vitro auto-Sites of Insulin Receptor Phosphorylation phosphorylation of the p-subunit of the insulin receptorkinase is shown in Fig. 4. In the absence of insulin, autophosphorylation occurred slowly. Insulin (100 nM) stimulated the incorporation of :v2P from [y-'"PIATP into the &subunit of the receptor. Within 30 s, autophosphorylation of the insulinstimulated receptor reached 50% of maximum and a steady state value was reached after about 10 min (Fig. 5A) incorporated per insulin binding site estimated from a Scatchard plot (Fig. 1). Only tryptic peptides pp4 and pp5 were detected after 5 s of incubation with [y-'"P]ATP and reached apparent steady state levels after only 20 s (Fig. 6). These rapid sites of' autophosphorylation contrasted with the slower sites (ppl and pp2) that were detectable only after longer incubation intervals and reached steady state in 5-10 min (Fig. 6). The phosphorylation of pp2 appeared to lag slightly, whereas ppl lagged markedly relative to the phosphorylation of pp4 and pp5. Although the occurrence of multiple phosphopeptides could be due to incomplete digestion of the p-subunit by trypsin, our kinetic results are consistent with the notion that the phosphopeptides resolved by HPLC are unique sites of autophosphorylation. Additional phosphopeptides eluted after 40 min appeared to be phosphorylated relatively slowly The Effect of Receptor Concentration on Autophosphorylation of the @-Subunit-When concentrations of all other components in the reaction mixture were unchanged, autophosphorylation of the insulin receptor was proportional to the amount of protein used in the assay (Fig. 7 , left). The slope of the line has a value of 95 pmol of '"P per mg of protein per min which agrees with the value shown in Fig. 5A at the corresponding time interval. These results suggest that nonspecific hydrolysis of ATP which could interfere with the kinetic studies did not occur to a significant extent with increasing protein concentrations because ATPase activity should cause a nonlinear relation between protein concentration and autophosphorylation. These results also suggest that the decreasing ratio of protein to Triton X-100 does not affect autophosphorylation. Furthermore, when the protein concentration was constant at 40 pg/pl and the Triton X-100 concentration was varied between of 0.02% and 0.1%, the rate of autophosphorylation measured during a 30-s time interval was not affected (data not shown). Autophosphorylation of the &subunit of the insulin receptor could occur by an intramolecular reaction in which the catalytic domain and the phosphorylation sites are located on the same molecular aggregate or by an intermolecular reaction in which the catalytic domains and phosphorylation sites reside on separate receptor molecules. The velocity of phosphorylation by the former mechanism should be independent of the concentration of the receptor, whereas the concentration of receptor should affect the rate of phosphorylation which occurs by an intermolecular reaction (37, 38). When the amount of lectin-purified protein added to the reaction mixture was maintained a t a constant level, but its concentration was decreased by increasing the volume of the phosphorylation reaction mixture, no significant effect on the approximate initial rate of autophosphorylation measured during a 1-min interval was observed (Fig. 7 , right). Therefore, autophosphorylation of the insulin receptor may occur by an intramolecular reaction, a result which suggests that only receptors occupied with insulin will undergo stimualted autophosphorylation.
The Effect of the ATP Concentration on the Sites of Autophosphorylation of the Insulin Receptor-The velocity curves of insulin-stimulated autophosphorylation were sigmoidal with respect to the ATP concentration (Fig. 8). Mn", within the concentration interval of 0-5 mM, was an essential activator of autophosphorylation, whereas higher concentrations of Mn2+ were inhibitory (Fig. 9). These results are consistent with the complex kinetic behavior reported for metal-activated enzymes (42)(43)(44) or may arise from a regulatory phenomenon involving activation of the receptor by autophosphorylation. However, this kinetic analysis should be viewed cautiously because autophosphorylation of the insulin receptor occurs a t multiple sites with distinct time courses and apparently by an intramolecular reaction.
HPLC tryptic peptide maps of the phosphorylated &subunit obtained a t various ATP concentrations after the incubation with 100 nM insulin are shown in Fig. 10, top. Phosphorylation of pp2 through pp5 was detected in this experiment, whereas the 1-min interval was not long enough to observe ppl. The improved resolution was due presumably to the collection of fractions during 24-s time intervals (compare with Fig. 5, B-H). In some experiments, pp2 was separated into 2 peaks (2a and 26), but for the present study, these peaks were analyzed together. Plotting the total radioactivity associated with each peptide fragment against the ATP concentration yielded kinetic curves describing the phosphorylation of pp2, pp3, pp4, and pp5 ( Fig. 10, middle). Like the phosphorylation of the entire receptor, sigmoidal curves were obtained for pp2, pp3, and pp4. However, the curve for pp5 was hyperbolic. Although the phosphorylation of pp4 and pp5 has probably reached steady state during the 1-min reaction (Fig. 6), the different shape of the curves for pp4 and pp5 supports our conclusion that these phosphopeptides represent distinct phosphorylation sites on the p-subunit.
As emphasized by the relative curves in Fig. 10, bottom, the phosphorylation of pp5 precedes the phosphorylation of the other peptides. These results, in addition to the time courses shown before ( Fig. 6), suggest that autophosphorylation of pp5 may facilitate subsequent phosphorylation of the receptor at the

Kinetics and Sites
of Insulin Receptor Phosphorylation other sites and possibly other substrates (26). The Effect of Insulin Concentration on the Phosphorylation of the Insulin Receptor-Insulin stimulated autophosphorylation of the @-subunit about 20-fold; half-maximal stimulation occurred at about 5 nM insulin with a slight inhibition detected above 100 nM (Fig. l l , top). Analysis of a series of sigmoidal velocity curves obtained at various insulin concentrations suggests that insulin stimulated autophosphorylation by increasing the V,,, of the reaction with no effect on the K,,, for ATP (Table I).
After a 1-min incubation with 50 PM [y3*]ATP, insulin stimulated the incorporation of 32P into pp2-pp5 of the insulin receptor (Fig. 11, middle). A dose response curve for each phosphopeptide measured between 1 and 1000 nM insulin is shown in Fig. 11, bottom. Summation of the radioactivity and pp5 (0) identified in Fig. 5 were quantitated by determining the approximate radioactivity associated with each peak. Where baseline resolution was not complete, the radioactivity was divided equally between each peak. fold). Phosphorylation of ppl was too slow to detect after a pp2 (9.1-fold) > pp4 (7.7-fold) > pp3 (6.7-fold) > pp5 (4.4-  After a 1-min incubation, the reaction was stopped, the proteins were separated by SDS-PAGE, and the @-subunit was localized by autoradiography. The gel fragments corresponding to these regions were excised and the radioactivity was determined by scintillation counting. The kinetic parameters f S.E. were determined by fitting the data to Equation 2.    After 1 min, the reaction was terminated by heating to 100 "C for 2 min and the proteins were separated by SDS-PAGE. The phosphorylated P-subunit of the insulin receptor was identified by autoradiography. The gel fragments containing the p-subunit were incubated with trypsin for 24 h. Top, the phosphopeptides obtained at each ATP concentration were separated by HPLC and the relevant regions of the chromatograms are shown. The acetonitrile gradient used was 0% (5 min) and 0%-70% (75 min). Middle, the radioactivity associated with pp2 (A), pp3 (V), pp4 (D), andpp5 (0) was determined and plotted against the ATP concentration. Bottom, a replot was constructed to show the relative 32P incorporation into peptides pp2-pp5 normalized to the maximum phosphorylation obtained for each peptide at 60 PM ATP. The @-subunit was separated by PAGE and its position was identified by autoradiography. The associated radioactivity was quantitated by scintillation counting. Top, 0, the insulin dose response curve is plotted relative to the maximum velocity of phosphorylation which occurred at 100 nM insulin (82 pmol of 32P/mg of protein/min). Middle, the gel fragments obtained from the experiments with 1 nM, 10 nM, 100 nM, and 1000 nM insulin were digested with trypsin for 24 h. The phosphopeptides, pp2-pp5, were separated by HPLC and the corresponding chromatograms of the relevant regions are shown. The acetonitrile gradient used was 0% (5 min) and 0%-70% (75 min). Bottom, the radioactivity associated with pp2 (a), pp3 (El), pp4 (A), and pp5 (0) are plotted as a function of the insulin concentration. For comparison, the insulin dose response curve obtained by adding the radioactivity associated with each peak is compared to the curve obtained from the original gel fragments (far kft, i 3 ) . (Fig. 6). When the insulin concentration was raised to 1000 nM, the phosphorylation of pp2 was inhibited by approximately 50%. Reconstruction of the relative dose response curve from the individual peptides suggests that the decreased incorporation of 32P into the intact P-subunit obtained at 1000 nM insulin was due entirely to the inhibition  Table 11. Right, K,,, (MnZ+)/Vmax, calculated from the values of the kinetic parameters in Table 11, is plotted against the coyresponding reciprocal of the ATP concentration.

1-min incubation
of phosphorylation of pp2 (Fig. 11, top). The Relation between ATP and Mn2+ for Autophosphorylation-In contrast to the effect of insulin on autophosphorylation which occurs by increasing the V, , , (Table I), the sigmoidal velocity curves shown in Fig. 12 (left) indicate that Mn2+ stimulated the in vitro autophosphorylation reaction by decreasing the concentration of ATP necessary to obtain the half-maximal velocity. Although the concentration interval of ATP was not saturating at 0.5 and 1 mM MnZ+, it appeared that the reaction velocities were approaching a similar maximum in all three curves. Therefore, within the concentration interval of Mn'+ that stimulated autophosphorylation (Fig.  9), the major effect of this cation was to decrease the apparent K,,, for ATP.
The relation between the initial velocity of autophosphorylation and Mn'+ measured a t various ATP concentrations is shown in Fig. 12, center. [Mn"] between 0.5 and 5.0 mM activated the receptor-kinase at all ATP levels tested; however, the concentration of Mn2+ required to activate the kinase decreased as the ATP concentration increased. Quantitative analysis of these hyperbolic velocity curves yielded the kinetic constants listed in Table 11. The V,,,, increased sigmoidally with the ATP concentration to a limiting value of 160 5 10 pmol/mg of protein/min which represents the maximum initial velocity of autophosphorylation at saturating concentrations of MnZ+, ATP, and insulin. The corresponding K , for ATP under these conditions was 19 _+ 1 p M .
The K,, for Mn"+ decreased as the ATP concentration increased (Table 11). Fig. 12, right, shows that the ratio, K,,, (Mn")/V,,,,, which was calculated from the kinetic parameters in Table I1 and plotted against the reciprocal of the ATP concentration, was linear and extrapolated to the origin (slope f S.E. = 782 _+ 27 pmol/fiM2/mg of protein/min). This result suggests that the K,, for Mnz+ approached zero as the ATP concentration increased to a saturating level. Therefore, in the intact cell where the ATP levels are relatively high and Mn2+ levels are relatively low (45), the receptor kinase should be active (1,2).
At least five sites of autophosphorylation can be identified by reverse-phase HPLC after trypsin digestion of the /?subunit. The phosphorylation of each site is stimulated by insulin. The autophosphorylation of two of these sites (pp4 and pp5) is complete after only 20 s of incubation with [y-"'PIATP and Mnn+. At steady state, about 2 molecules of "' P are incorporated per insulin binding site as estimated from a Scatchard plot. Assuming that the a-and @-subunit are associated in a stoichiometric ratio of 1:l (46) and each asubunit binds insulin (30), then about 2 molecules of :'2P are covalently bound to the @-subunit at steady state. However, these data do not imply phosphorylation of the /?-subunit to an integral molar ratio because our results indicate that more than two phosphorylation sites exist on the molecule. This calculation of stoichiometry must be viewed cautiously for the following reasons: uncertainty remains about the subunit composition and stoichiometry of the intact insulin receptor (36); some denaturation of the a-or P-subunits may occur during receptor isolation (30); and the intact receptor may not bind more than 1 insulin molecule (48). Furthermore, the phosphorylation stoichiometry remains provisional because the amount of endogenous phosphorylation of the receptor isolated from the Fao cell is unknown. This variable may explain why all the sites of autophosphorylation do not reach the same level as well as why the amount of phosphorylation varies between preparations of insulin receptor. Insulin stimulates autophosphorylation by increasing the V,,,;,, of :"P incorporation 20-fold. All five major sites of phosphorylation in the @-subunit were stimulated by insulin. We have previously shown with the purified receptor kinase that insulin stimulates the V,,, for substrate phosphorylation as measured with a synthetic tyrosine-containing peptide (8). These results suggest that insulin binding activates the recep-tor kinase by causing a conformational change in the catalytic domain. However, it is possible that autophosphorylation is stimulated by an insulin-induced change which makes an important site of autophosphorylation more accessible to the catalytic domain. The resulting covalent modification may subsequently enhance the activity of the receptor-kinase toward other sites of autophosphorylation and exogenous substrates (26).
The kinetics of autophosphorylation for the various sites present in the @-subunit of the insulin receptor is not identical. Phosphorylation of pp4 and pp5 are very rapid, whereas ppl and pp2 display a definite lag in the rate of phospho:ylation. It is possible that the insulin-induced autophosphorylation of pp4 and pp5 stimulates the subsequent autophosphorylation of ppl and pp2. Autophosphorylaticn of the 6-subunit of the insulin receptor is slightly inhibited by concentrations of insulin greater than 100 n M (4,7).' Our results suggest that this inhibition is due to a 50% decrease in the phosphorylation of pp2. It may be possible to establish a relation between the sites of autophosphorylation in the /3-subunit and activation of the kinase with respect to substrate phosphorylation (26). T o show a relation between the physiological insulin response and phosphorylation of the receptor at specific sites, it will be important to correlate the autophosphorylation sites observed i n vitro with those observed i n vivo.
In addition to insulin, regulation of the kinase may also occur by changing the concentration of divalent cations. Metal ions activate many enzymatic reactions by combining with both the enzyme and the substrate. Prototype examples of this reaction are the protein kinases which require ATP and MgS+, the active substrate being MgATP2-(49). Divalent metal ion activation of insulin receptor kinase autophosphorylation appears to be restricted in vitro to Mn'+ (6). This stimulation occurs predominantly by decreasing the K, for ATP and is observed at concentrations of Mn2+ in excess of that necessary to form the presumed substrate, MnATP". For example, a t a Mn" concentration of 0.5 mM, assuming a dissociation constant for MnATP2-of 10 p~ (50), the ATP calculated to be present as MnATP2-is between 97% and 98%. This relative distribution remains constant over the concentration interval of ATP used in these experiments. The increase in autophosphorylation observed at a constant concentration of ATP by increasing the concentration of Mn" from 0.5 to 5 mM, therefore, occurs without a major increase in the fraction of MnATP2-. Since autophosphorylation of the receptor kinase is linear with respect to the concentration of protein used in the assay, it is unlikely that the solubilized receptor preparation nonspecifically complexes Mn2+. Thus, activation of the insulin receptor kinase presumably requires binding of free Mn'+ to a specific site on the kinase, in addition to chelation of ATP4-by Mn2+. Since Mn2+ and M$+ form similar complexes with ATP (49, 50), the lack of any significant effect of Mgz+ in the presence or absence of Mn" also argues against the possibility that the stimulation of the kinase by Mn'+ occurs exclusively through chelation of ATP4-, or by decreasing the concentration of uncomplexed inhibitory derivatives of ATP, such as HATP3or ADP"-. Our results are analogous to the selective stimulation by divalent cations on the hepatic adenylate cyclase system (51,52). Since Mn2+ can play a dual role in activating the insulin receptor kinase, it is not surprising that a variety of complex kinetic curves are observed (42,43). The sigmoidal relation between the velocity of autophosphorylation uersw ATP concentration may reflect a complicated kinetic mechanism involving Mn", allosteric effects, or activation by autophosphorylation.
The finding that the K,, for Mn2+ approached zero as the [ATP] increased suggests that Mn2+ binding precedes Mn-ATP'-binding by an equilibrium-ordered mechanism (42,44). Another characteristic of this mechanism is that the v,,, for the reaction is independent of the concentration of Mn2+ which seems to be the case for the receptor kinase. According to this hypothesis, the insulin receptor kinase will not bind ATP in the absence of a metal activator, presumably Mn2+; however, the high level of ATP found in the intact cell (27) can drive the reaction to completion even in the presence of a very low level of the metal cation (42). Thus, the receptor kinase is active even when the activator concentration is low, provided that the substrate concentration is high. Autophosphorylation of the partially purified insulin receptor is apparently independent of the protein concentration used in the assay. If a linear relation exists between the receptor kinase concentration and the initial rate of receptor phosphorylation, the 10-fold dilution of protein used in our experiments should be adequate to detect it. Insulin receptors could form aggregates in detergent solution which cause an apparent concentration independence of autophosphorylation; however, this possibility is unlikely because phosphorylation of the insulin receptor is independent of the Triton X-100 concentration. Owing to phosphatases which are present in the receptor preparation, dilution may cause fortuitously an equal decrease of autophosphorylation and dephosphorylation so that no net change is observed; however, this is unlikely because the in vitro dephosphorylation reaction is slow relative to the phosphorylation reaction.' The mechanism by which an intramolecular autophosphorylation reaction could occur at multiple sites is not clear.
Like the insulin receptor, other protein kinases undergo autophosphorylation a t multiple sites by what appears to be an intramolecular reaction (27, 37, 38). The gene product of the Rous sarcoma virus, pp60"-"", phosphorylates itself at Tyr in the COOH-terminal portion of the molecule in vitro (53). I n vivo, pp60""" is phosphorylated at serine and tyrosine residues (54). This finding is similar to the insulin receptor (1, 2 ) . Recently, the major site of tyrosine autophosphorylation in pp60'~"" was altered by constructing a mutant protein in which Tyr-416 was changed to a phenylalanine residue (55). This mutation had no effect on the transforming properties relative to wild type pp60""" and no effect on the tyrosine kinase activity measured with exogenous substrates. However, another site of tyrosine autophosphorylation was detected in the COOH-terminal portion of the mutant molecule, so tyrosine autophosphorylation at this position may still be important for the biological activity of the kinase. Multiple sites of tyrosine phosphorylation in the epidermal growth factor receptor have been reported (56,57), but their effect on the activity of the kinase is unknown. Autophosphorylation at serine residues of the regulatory subunit of the CAMP-dependent protein kinase from bovine cardiac muscle decreases the rate at which the active catalytic subunit reassociates with the regulatory subunit to form the less active holoenzyme (38, 58, 59). Therefore, this reaction appears to have a regulatory role. The catalytic activity of other serine kinases is also regulated by autophosphorylation (27).
The rapid autophosphorylation of the &subunit suggests that the insulin receptor may translate the extracellular regulatory signal arising from insulin binding into an intracellular signal through the stimulation of tyrosine phosphorylation at the inner face of the plasma membrane. This regulatory signal, initiated by insulin binding, may be transmitted across the plasma membrane by stimulation of substrate phosphorylation of exogenous proteins at tyrosine residues. Alternatively, autophosphorylation itself may be a sufficient intracellular molecular signal to initiate the physiological Sites of Insulin Receptor Phosphorylation responses of insulin. In this model, signal transmission could occur without any additional phosphotransfer reactions if the phosphorylated receptor is modified in its ability to interact with relevant cytoplasmic or membrane-bound proteins.
In conclusion, we have shown that the insulin receptor has characteristic kinetic properties that may be important for insulin action. Both insulin and divalent cations affect the rate of autophosphorylation and may be important physiologic regulators of the membrane-bound kinase, but these act through different mechanisms. Autophosphorylation of the insulin receptor is a very rapid intramolecular reaction that may provide the link between extracellular insulin binding and a recognizable intracellular covalent modification.