Insulin-sensitive phosphorylation of serine 1293/1294 on the human insulin receptor by a tightly associated serine kinase.

In these studies we demonstrate that insulin stimulates both tyrosine and serine phosphorylation of the insulin receptor after its partial purification on wheat germ-agarose, and after affinity purification on insulin-agarose. Analysis of the serine phosphate incorporated into partially purified or highly purified insulin receptor suggests that an insulin-sensitive serine kinase (IRSK) copurifies with the insulin receptor. Following trypsin digestion, reversed-phase high pressure liquid chromatography (HPLC) analysis of the phosphorylated, affinity-purified insulin receptor preparation reveals phosphopeptide profiles similar to those of trypsin-digested receptors immunoprecipitated from 32P-labeled fibroblasts overexpressing the human insulin receptor. The major insulin-stimulated HPLC phosphopeptide peak from insulin receptors labeled in intact cells contains a hydrophilic phosphoserine-containing peptide which rapidly elutes from a C18 column. HPLC and two-dimensional separation indicate that the same phosphopeptide is obtained when affinity-purified insulin receptors are phosphorylated by IRSK. The serine containing tryptic peptide within the cytoplasmic domain of the human insulin receptor predicted to elute most rapidly upon HPLC had the sequence SSHCQR corresponding to residues 1293-1298. A synthetic peptide containing this sequence is phosphorylated by the insulin receptor/IRSK preparation. After alkylation and trypsin digestion, the synthetic phosphopeptide comigrates with the alkylated, tryptic phosphopeptide derived from insulin receptor phosphorylated in vitro by IRSK. We propose that serine 1293 or 1294 of the human insulin receptor is a major site(s) phosphorylated on the insulin receptor in intact cells and is phosphorylated by IRSK. Furthermore, insulin added directly to affinity-purified insulin receptor/IRSK preparations stimulates the phosphorylation of synthetic peptides corresponding to this receptor phosphorylation site and another containing threonine 1336. Kemptide phosphorylation is not stimulated by insulin under these conditions. No phosphorylation of peptide substrates for Ca2+/calmodulin-dependent protein kinase, protein kinase C, casein kinase II, or cGMP-dependent protein kinase by IRSK is detected. These data indicate that IRSK exhibits specificity for the insulin receptor and may be activated by the insulin receptor tyrosine kinase in an insulin-dependent manner.

In these studies we demonstrate that insulin stimulates both tyrosine and serine phosphorylation of the insulin receptor after its partial purification on wheat germ-agarose, and after affinity purification on insulin-agarose.
Analysis of the serine phosphate incorporated into partially purified or highly purified insulin receptor suggests that an insulin-sensitive serine kinase ( Serine phosphorylation of the insulin receptor appears to play a role in modulating insulin receptor function. Phorbol esters promote serine phosphorylation of the insulin receptor in intact rat hepatoma cells (1,2) and human IM-9 lymphocytes (3). Phorbol ester-induced phosphorylation also decreases insulin receptor tyrosine kinase activity (2). The serine phosphorylation of insulin receptors correlates with the ability of phorbol esters to inhibit insulin activation of glycogen synthase and tyrosine aminotransferase activity (1). Since tyrosine kinase activity appears to be required for the insulin receptor to signal (4-6), a cellular mechanism which decreases tyrosine kinase activity (e.g. serine phosphorylation of the insulin receptor @ subunit) may also regulate receptor action on intracellular processes.
Purified serine kinases are able to phosphorylate and modulate the kinase activity of purified insulin receptor. Insulin receptor can be phosphorylated in uitro' by protein kinase C (7) and cyclic AMP-dependent protein kinase (8). High concentrations of each enzyme are required for stoichiometric serine phosphorylation of the insulin receptor. Receptor phosphorylation catalyzed by either of these kinases reduces the tyrosine kinase activity of the insulin receptor toward exogenous substrates. However, insulin does not enhance the ability of these serine/threonine kinases to catalyze phosphorylation of the insulin receptor. Thus, the physiological significance of insulin receptor phosphorylation by these kinases may be limited to insulin-independent mechanisms.
Although the mechanism of insulin-enhanced serine phosphorylation of the insulin receptor in intact cells is not known, it may require prior tyrosine autophosphorylation of the receptor. Increases in tyrosine phosphate on the insulin receptor appear to precede the appearance of phosphoserine in insulin-treated cells (9). In addition, mutant insulin receptors which lack a lysine critical for ATP binding and tyrosine kinase activity also fail to undergo insulin-sensitive serine phosphorylation (10). Several previous investigations (11)(12)(13) have demonstrated insulin-sensitive serine kinase activity which elutes with the insulin receptor from wheat germagarose. However, those studies did not ascertain whether the insulin-responsive serine kinase activity was associated with the insulin receptor or bound independently to the lectin.
As an initial step toward understanding the mechanism of insulin-stimulated serine phosphorylation we utilized an affinity-purified insulin receptor preparation to compare the pattern of insulin receptor phosphorylation in vitro with the pattern of receptor phosphorylation observed in intact cells. In this paper, we demonstrate that an insulin-sensitive serine Insulin Receptor-associated SePine Kinase kinase (IRSK)' is tightly associated with insulin receptors purified on insulin-agarose as well as wheat germ affinity resins. Furthermore, IRSK phosphorylates the affinity-purified insulin receptor in an insulin-responsive manner on sites that are also phosphorylated in the intact cell. We identify a site of serine phosphorylation on the insulin receptor phosphorylated by IRSK in vitro, and in intact cells as serine 1293: or 1294, or both. IRSK in affinity-purified insulin receptor preparations phosphorylates synthetic peptides identical to this and other phosphorylation sites on the insulin receptor in an insulin-dependent manner, suggesting IRSK is activated by insulin addition to receptor preparations. at a right angle to the direction of electrophoresis in l-butanol/pyridine/acetic acid/water (15:10:3:12) as described previously (1). After chromatography the plates were dried and exposed to x-ray film to localize '"P-labeled peptides. Fig. 1 demonstrates insulin-sensitive phosphorylation of the insulin receptor fi subunit from human placenta after partial purification on wheat germ lectin (Fig. lA, lanes 1 and 2) or after sequential wheat germ lectin and insulin-agarose affinity purification (Fig. lA, lanes 3 and 4). Phosphoamino acid analysis of the excised fi subunit reveals that insulin increases serine as well as tyrosine phosphorylation.

RESULTS
In the presence of insulin approximately 0.03-0.06 mol of phosphate are incorporated into serine per mol of our insulin receptor preparation in 1 h at 22 "C compared to 0.5 mol of phosphate incorporated into tyrosine. These low stoichiometries of receptor phosphorylation result from the low ATP concentrations (5 FM) in each reaction required to maximize the specific activity of the [r-"'P]ATP and thus the "P incorporation into the /3 subunit. Experiments examining the time course of phosphorylation indicate that phosphoserine increases proportionately with the amount of phosphotyrosine incorporated (data not shown).
Experiments were conducted to test whether IRSK activity copurified with the receptor. Insulin-stimulated phosphate incorporation into serine and tyrosine on the receptor /3 subunit was determined relative to the amount of protein or 'Z"I-insulin binding activity present in wheat germ-agarose eluates and in insulin-agarose eluates (Table I). IRSK activity, assessed as the rate of receptor @ subunit phosphorylation, increases nearly 300-fold/pg of protein when insulin receptor is purified from wheat germ eluate on insulin-agarose. change. These data indicate that the insulin-sensitive serine kinase is being concentrated during receptor purification. Tyrosine phosphorylation of the fi subunit increases at least 1500-fold when insulin receptor is purified from wheat germ eluate on insulin-agarose (Table I). After purification on insulin-agarose the ratio of tyrosine phosphate on the /? subunit to '"ii-insulin binding activity increases at least d-fold. This observation may indicate the presence of tyrosine phosphatase activity or an inhibitor of the insulin receptor tyrosine kinase in wheat germ eluates which is removed by purification on insulin-agarose.
To evaluate the potential physiological relevance of the IRSK activity associated with affinity-purified receptors, HPLC tryptic phosphopeptide maps of insulin receptor /3 subunit phosphorylated in vitro were compared to HPLC maps of insulin receptor p subunits isolated from "'P-labeled cells. Fig. 2  insulin-sensitive labeled phosphopeptide peaks from insulin receptor p subunits phosphorylated in oitro have the same mobility as tryptic phosphopeptides generated from insulin receptors labeled in intact cells. The relative proportion of ,"P in the phosphopeptide peaks varies between /3 subunits from affinity-purified insulin receptors and receptor preparations labeled in uiuo, as previously found (18). The recovery of '?P from the HPLC column ranged from 70 to 84%. Thus, we cannot unequivocally exclude the possibility that additional, hydrophobic phosphopeptides fail to elute from the CM column. Phosphoamino acid analysis revealed that the initial peak contained only free "'P. Peak 2 contains the major labeled phosphopeptide(s) on insulin receptor @ subunit isolated from intact cells, whereas peak 3 is the predominant phosphopeptide peak labeled in the affinity-purified insulin receptor preparation in vitro (Fig. 2). Phosphoamino acid analysis of the major phosphopeptide peak in insulin receptor immunoprecipitated from intact cells, HPLC peak 2, shows that it consists primarily of an insulinstimulatedphosphoserine-containingpeptide or peptides (Fig.  3). After phosphorylation in uitro, HPLC tryptic peptide maps of the affinity-purified insulin receptor preparation yield the same peak 2 (Fig. 2), and serine is the primary phosphorylated  Wheat germ agarose eluate (9.6 pg) and insulin agarose eluate (140 ng) were assayed for IRSK and insulin receptor kinase activity by measurement of "P incorporation on to serine and tyrosine residues of the insulin receptor 0 subunit. Phosphate incorporation was calculated from "P-labeled phosphoamino acids released during partial hydrolysis of the isolated insulin receptor @ subunit. 32P incorporation was normalized relative to the total protein assayed or to the amount of insulin receptor present as determined by specific '?-insulin bound at a '251-insuIin concentration of 0.9 nM. IRSK activity amino acid (Fig. 3). However, the phosphopeptides in HPLC peak 2 from insulin receptors phosphorylated in uitro contain a smaller fraction of the total phosphate incorporated into the insulin receptor /3 subunit compared to those derived from insulin receptor labeled in uiuo (Fig. 2).
Two-dimensional peptide mapping of the phosphopeptides in HPLC peak 2 reveals the presence of two phosphopeptides when either insulin receptor phosphorylated in uiuo (Fig. 4) or in vitro (Fig. 5) is analyzed. Phosphoamino acid analysis of the individual phosphopeptides within HPLC peak 2 from insulin receptor phosphorylated in vitro demonstrates that one contains phosphotyrosine while the other is a phosphoserine-containing peptide (Fig. 5). Rapid elution on HPLC (Fig. 2), strong migration upon thin-layer electrophoresis, and poor migration upon thin layer chromatography (Fig. 5) suggest that the phosphoserine-containingpeptide in HPLC peak 2 is strongly hydrophilic.
Phosphothreonine and phosphoserine were identified in HPLC phosphopeptide peak 5 from insulin receptor p subunit phosphorylated in intact cells. In contrast, HPLC peak 5 contains primarily phosphotyrosine and a small amount of phosphothreonine when generated from insulin receptor phosphorylated in vitro ( Fig. 5 and data not shown). HPLC maps of receptor p subunit phosphorylated in intact cells lack phosphotyrosine-containing peptides in peak 5. Two-dimensional analysis of HPLC peak 5 from @ subunit labeled in uiuo contains a single phosphopeptide with the chromatographic properties similar to the phosphothreonine-containing peptide labeled in HPLC peak 5 from affinity-purified and phosphorylated insulin receptor (Fig. 4). The same phosphothreonine-containing tryptic peptide appears to be phosphorylated on the insulin receptor p subunit in uivo in response to phorbol ester addition to intact cells, or after phosphorylation of purified insulin receptor with protein kinase C (19). Phosphoserine-containing peptides derived from insulin receptor phosphorylated in vitro are also present in HPLC peaks 4,6, and 7 (Fig. 5). Two-dimensional analysis of HPLC peak 4 reveals a complex pattern of phosphopeptides. HPLC peak 4B was composed of seven of the nine phosphopeptides observed in peak 4A, including the phosphoserine-containing peptide labeled "b" (data not shown). Phosphopeptides in HPLC peak 4 from receptor labeled in uivo have a distribution similar to peak 4 derived from affinity-purified insulin receptor, but with significant differences in the relative intensity of labeling (Fig. 4). Two-dimensional analysis of peak 6 dem- onstrates a similar pattern of phosphopeptides derived from insulin receptor phosphorylated either in uitro (Fig. 5) or in intact cells (Fig. 6). Phosphoamino acid analysis of the phosphopeptides in HPLC peak 6 from insulin receptor labeled in vitro indicates the presence of tyrosine in all three peptides.
The phosphoserine-containing peptides within HPLC peak 7 from affinity-purified insulin receptor (Fig. 5) have no clear counterpart within the corresponding HPLC fractions from receptor labeled in vivo (Fig. 4). These hydrophobic phosphopeptides appear not to be incompletely cleaved tryptic products, because further trypsin digestion failed to change their mobility (data not shown). These data suggest that the phosphoserine-containing peptides in peak 7 are unique and unrelated to the phosphoserine-containing peptides in peak 2 and peak 4.
The hydrophilic phosphoserine-containing peptide in HPLC peak 2 isolated from insulin receptors phosphorylated in vitro (Fig. 5) corresponds to the major phosphopeptide isolated from insulin receptor phosphorylated in vivo (Fig. 4), and as such may be an important site of receptor regulation. We therefore sought to identify this peptide. The mobilities of potential tryptic fragments from the cytoplasmic domain of the insulin receptor were predicted with an equation for elution by HPLC using a mobile phase of acetonitrile and trifluoroacetic acid (20). The most rapidly eluting tryptic peptide containing at least one serine residue was predicted to be SSHCQR, corresponding to residues 1293-1298 of the human insulin receptor. Previous experiments with HPLC peak 2 of in vitro phosphorylated insulin receptor indicated that the phosphoserine-containing peptide (Fig. 5) was insensitive to both V8 protease and chymotrypsin (data not shown). The tryptic fragment corresponding to residues 1293-1298 of the human insulin receptor would also be insensitive to these proteases. To determine if this region of the insulin receptor is phosphorylated by IRSK, the synthetic peptide VPLDRSSHCQREEAG corresponding to residues 1288-1302 of the human insulin receptor was synthesized and phosphorylated by the affinity-purified insulin receptor preparation. After alkylation and trypsin digestion, the phosphorylated synthetic peptide was observed to have the same migration on two-dimensional analysis as the phosphoserine-containing peptide from HPLC peak 2 (Fig. 6). Furthermore, when equal amounts of radioactivity from receptor-derived and trypsindigested synthetic phosphopeptides were combined and analyzed, two-dimensional analysis yielded a single spot of radioactivity (Fig. 6).
The data presented in Fig. 6 indicate that serine 1293 or 1294 or both are sites of insulin-stimulated phosphorylation on the insulin receptor. In other studies, we have identified the phosphothreonine-containing phosphopeptide in HPLC peak 5 as residues 1334-1339 (19). Insulin receptor threonine 1336 is the phosphorylated residue in this latter peptide when receptor is phosphorylated in intact cells (19), or by IRSK (Fig. 5). The identification of these receptor substrate sites for IRSK were used to test whether insulin-sensitive serine/ threonine phosphorylation reflects a receptor conformation change or actual activation of IRSK activity. Fig. 7  in viva. HPLC phosphopeptide peaks (numbers designated at left) of phosphorylated insulin receptor p subunits immunoprecipitated from intact cells were lyophilized, reconstituted in 8 ~1 of 30% formic acid, and spotted on thin-layer cellulose plates. The phosphopeptides were separated by electrophoresis in 30% formic acid for 90 min at 400 V in one dimension and by chromatography at right angles to the direction of electrophoresis in n-butanol/pyridine/acetic acid/ water (15:10:3:12).

Phosphopeptides
were localized by autoradiography. Two-dimensional maps were exposed to x-ray film for 6 h (peak 2) or 23 h (peaks 4A, 5, 6/7). ate synthetic peptides containing putative insulin receptor serine phosphorylation site 1293/1294, as well as threonine phosphorylation site 1336 (19) in an insulin-stimulated manner. Experiments without peptide and the insulin receptor preparation demonstrate that additional radioactive spots observed after two dimensional analysis are due to contaminants from the [y-32P]ATP which are occasionally retained during purification of the phosphopeptide (data not shown). HPLC phosphopeptide peaks (numbers designated at left) from affinity-purified insulin receptor (J' subunits phosphorylated in vitro were analyzed as described in Fig. 4. The region corresponding to each phosphopeptide from insulin receptor labeled in oitro was cut from each plate. Phosphopeptides were eluted in 30% formic acid, dried, and analyzed for phosphoamino acid content as described under "Experimental Procedures." Two-dimensional maps were exposed to x-ray film for 5 h. The primary phosphoamino acid of each in vitro labeled phosphopeptide is designated in the right column of panels. S, phosphoserine; T, phosphothreonine; Y, phosphotyrosine. (1.4 + 0.2-fold, n = 4) when insulin is added to the affinitypurified insulin receptor preparation (not shown). Furthermore, IRSK is not inhibited by a peptide inhibitor (21) of CAMP-dependent protein kinase.q No phosphorylation of peptide substrates for multifunctional calmodulin-dependent protein kinase (KKRPQRATSNVFS), casein kinase II (RREEE-TEEE), protein kinase C (KRTLRR), or cGMP-dependent protein kinase (RKRSRKE) by the insulin receptor/IRSK preparation could be detected. DISCUSSION Results presented here demonstrate that a serine kinase (IRSK) capable of phosphorylating the insulin receptor in an ' R. E. Lewis  insulin-dependent manner remains tightly associated with the receptor during sequential purification on wheat germ and insulin affinity columns (Fig. 1, Table I). Furthermore, serine phosphorylated peptides derived from the insulin receptor p subunit phosphorylated by IRSK in uitro demonstrates simi-lar migration on HPLC (Fig. Z), and two-dimensional analysis (Fig. 5) as phosphopeptides from insulin receptor phosphorylated in intact cells (Fig. 4). One tryptic phosphopeptide derived from insulin receptor phosphorylated by IRSK appears to be SSHCQR corresponding to residues 1293-1298 of the deduced sequence of the human insulin receptor cDNA. A synthetic peptide identical to receptor residues 1288-1302 is phosphorylated on serine by the insulin receptor/IRSK preparation. After alkylation and trypsin digestion, the phosphorylated synthetic peptide comigrates with the phosphoserine-containing peptide in HPLC peak 2 released by trypsin from purified insulin receptor p subunits phosphorylated in vitro (Fig. 6). However, the difficulties inherent in radiosequence analysis of a peptide with consecutive serines at its amino terminus preclude the unambiguous assignment of serine 1293 or serine 1294 or both as phosphorylation sites. We conclude that the tryptic phosphopeptide released from purified insulin receptor corresponds to residues 1293-1298 of the human insulin receptor, and that serine 1293 and/or serine 1294 is the major site of insulin-dependent phosphorylation by IRSK. The copurification of IRSK with a highly purified preparation of insulin receptor (Table I) is suggestive of an important and specific interaction of this kinase activity with the insulin receptor. This conclusion is supported by the observation that several sites phosphorylated on the insulin receptor in intact cells appear to also be phosphorylated by IRSK in u&o. Previous reports have demonstrated that an insulinsensitive serine kinase activity was associated with human insulin receptors partially purified from placental membranes on wheat germ-agarose (11)(12)(13). One investigation (12) emphasized the necessity to perform phosphorylation assays at 22 "C, to purify insulin receptor on wheat germ-agarose in the absence of NaCl, and to prepare membranes from tissue prior to solubilization in order to maximize the association of the serine kinase activity with the insulin receptor. Our purification scheme is largely consistent with this method, but utilizes 100 mM NaCl during part of the membrane preparation, 1.0 M NaCl during final purification on insulin-agarose yet still retains IRSK activity. The difference between the purification scheme used here and the procedure used by others (12) may indicate that NaCl interferes with the ability of IRSK to bind directly to the lectin. However, NaCl may not interfere with the association of IRSK to insulin receptor bound on the insulin-agarose column. Alternatively, the difference in purification procedures may indicate the presence of more than one serine kinase activity that can phosphorylate the insulin receptor.
Comparison of the HPLC tryptic phosphopeptide maps generated from insulin receptors phosphorylated in intact cells, or in vitro reveals a marked similarity in the distribution of the tryptic phosphopeptides generated from each receptor preparation, but a significant difference in the level of phosphate incorporated into the various phosphopeptides. However, two-dimensional analysis of individual HPLC peaks clearly demonstrates the similar migration of a serine phosphorylation site on affinity-purified insulin receptors labeled by IRSK and the major phosphorylation site on insulin receptor phosphorylated in intact cells ( Fig. 4 and Fig. 5, peak 2). Tyrosine phosphorylation is predominant in the affinitypurified insulin receptor preparation. Differences in the pattern of phosphorylation in viuo and in uitro have been previously described for the insulin receptor (18) and epidermal growth factor receptor (22). These differences in the stoichiometry of phosphorylation of certain sites may be due to changes in receptor conformation after purification or to the presence of different amounts of endogenous phosphate remaining on insulin receptors during isolation of the two preparations.
Serines 1293 and 1294 reside on a hydrophilic stretch of amino acids with predicted fi turn structure (23). Similar regions are believed to be common locations for phosphorylation sites because they may reside on the outer surface of globular proteins (24). The tryptic peptide from this region appears to be strongly phosphorylated on insulin receptors isolated from phorbol ester-treated cells even though protein kinase C fails to phosphorylate this site on the receptor in vitro (19). Serines 1293 and 1294 are not surrounded by the density of positively charged amino acids typically associated with a protein kinase C phosphorylation site (25), although one arginine is located adjacent to the phosphorylation site at position 1293. These observations suggest the hypothesis that in intact cells protein kinase C may activate IRSK which then phosphorylates the insulin receptor. The inability of the insulin receptor/IRSK preparation to detectably phosphorylate a protein kinase C substrate indicates that IRSK is distinct from protein kinase C.
An important issue is the mechanism by which insulin increases serine/threonine phosphorylation of the insulin receptor by IRSK. One possibility is that the binding of insulin alters the conformation of the receptor cytoplasmic domain to make it a better substrate for IRSK. Alternatively, IRSK may be activated by the receptor. The observation that insulin addition to the insulin receptor/IRSK preparation significantly enhances the serinelthreonine phosphorylation of receptor-specific peptides (Fig. 7) supports the notion that IRSK is activated by the insulin receptor. The inability of the insulin receptor/IRSK preparation to phosphorylate peptide substrates to other known kinases in an insulin-dependent manner distinguishes IRSK from these enzymes. Whether IRSK is a substrate for the insulin receptor tyrosine kinase is not known, although recent investigations suggest that an active insulin receptor tyrosine kinase is required for insulinsensitive serine phosphorylation of partially purified insulin receptor by IRSK (26). Recent studies (27) have demonstrated that the proto-oncogene product Raf-1 is tightly associated with the ligand-activated platelet-derived growth factor receptor. Furthermore, Raf-1 kinase activity is elevated following platelet-derived growth factor receptor-mediated tyrosine phosphorylation of Raf-l (27). We do not detect any phosphorylated bands with a mass near that of Raf-1 (74 kDa) following electrophoresis of our phosphorylated, affinity-purified insulin receptor preparation. In addition insulin does not stimulate the phosphorylation of Raf-I in quiescent Balb 3T3 cells (28). These observations argue against the identity of IRSK as Raf-1. It will be important in future studies to determine whether IRSK is phosphorylated on tyrosine by the insulin receptor and, if phosphorylation on tyrosine occurs, whether or not such a phosphorylation alters IRSK activity or IRSK association with the insulin receptor.
The major receptor serine phosphorylation site 1293/1294 is located an equal distance from the end of the tyrosine kinase domain and from the carboxyl terminus of the insulin receptor /3 subunit (29, 30). Comparison of the predicted amino acid sequences of the human insulin and IGF-I receptors (29-31) reveals poor conservation of the amino acids in the IGF-I receptor corresponding to residues 1292-1303 in the insulin receptor. Only the serine corresponding to serine 1293 of the insulin receptor is retained from this stretch of amino acids in the IGF-I receptor. Future studies will be required to determine if this region of the IGF-I receptor is also phosphorylated and whether phosphorylation of these Znsulin Receptor-assoc residues is important in regulating receptor signaling.