Insulin receptor tyrosine kinase-catalyzed phosphorylation of 422(aP2) protein. Substrate activation by long-chain fatty acid.

It was established previously that the 15-kDa protein phosphorylated in 3T3-L1 adipocytes treated with insulin and phenylarsine oxide is O-phospho-Tyr19 422(aP2) protein, a fatty acid-binding protein. To assess its capacity to serve as substrate of the insulin receptor tyrosine kinase in vitro, native 422(aP2) protein was isolated from 3T3-L1 adipocytes and purified to homogeneity. Receptor-catalyzed phosphorylation of 422(aP2) protein on Tyr19 was markedly activated when long-chain fatty acid, e.g. oleic acid, is bound to the protein. Fatty acid had no effect on autophosphorylation of the insulin receptor by its intrinsic tyrosine kinase. Both saturated (C14, C16, and C18) and unsaturated (all cis-delta 9 C16, -delta 9 C18, and -delta 9,12 C18, -delta 9,12,15 C18, and -delta 5,8,11,14 C20) fatty acids caused substrate activation. The Km for 422(aP2) protein was greatly reduced (from 170 to 3 microM) by oleic acid with little or no effect on Vmax. Upon binding fatty acid to 422(aP2) protein the susceptibility of Tyr19 and Tyr128 to iodination by the lactoperoxidase method increased greatly. These results indicate that upon binding fatty acid, 422(aP2) protein undergoes a conformational change whereby Tyr19, which lies within a consensus-type sequence for tyrosine kinase substrates, becomes accessible for phosphorylation by the insulin receptor tyrosine kinase and to iodination by lactoperoxidase.

It was established previously that the 15-kDa protein phosphorylated in 3T3-Ll adipocytes treated with insulin and phenylarsine oxide is  protein, a fatty acid-binding protein.
To assess its capacity to serve as substrate of the insulin receptor tyrosine kinase in vitro, native 422(aP2) protein was isolated from 3T3-Ll adipocytes and purified to homogeneity.
Receptor-catalyzed phosphorylation of 422(aP2) protein on Tyr" was markedly activated when long-chain fatty acid, e.g. oleic acid, is bound to the protein.
The K,,, for 422(aP2) protein was greatly reduced (from 170 to 3 NM) by oleic acid with little or no effect on V,,,. Upon binding fatty acid to 422(aP2) protein the susceptibility of Tyr19 and Tyr"* to iodination by the lactoperoxidase method increased greatly.
These results indicate that upon binding fatty acid, 422(aP2) protein undergoes a conformational change whereby Tyr", which lies within a consensustype sequence for tyrosine kinase substrates, becomes accessible for phosphorylation by the insulin receptor tyrosine kinase and to iodination by lactoperoxidase.
The major pathways for energy storage and mobilization in animal cells are under strict hormonal control, most notably by insulin and glucagon. A large body of evidence indicates that the pleiotropic response to insulin, which includes the stimulation of glucose and fatty acid uptake, glycogenesis, and lipogenesis, is initiated by the binding of the hormone to its specific cell surface receptors on the plasma membrane (1). This insulin receptor interaction stimulates autophosphorylation of specific tyrosine residues in the P-subunit of the receptor (l -6) and thereby the activity of the receptor's intrinsic tyrosine kinase toward protein substrates ( [6][7][8][9]. The importance of the receptor's tyrosine kinase activity was demonstrated by experiments in which insulin action was inhibited by site-directed mutagenesis of either a critical lysine residue in the putative ATP-binding site (9)(10)(11)  tyrosines within the critical autophosphorylation site (12).
Other experiments have revealed that insulin action can be blocked by introducing antibodies into cells that inhibit insulin-induced activation of the tyrosine kinase (13). Other studies have suggested that several monoclonal antibodies directed against the extracellular domain of the receptor are insulin mimetic without activating the tyrosine kinase activity of the receptor (14,15). A recent examination (16), however, using a more sensitive assay revealed that these same antibodies do stimulate kinase activity and that this stimulation correlates with their ability to elicit a biological response. Several years ago we identified a 15-kDa protein (~~15) that is phosphorylated on tyrosine in 3T3-Ll adipocytes treated with insulin and phenylarsine oxide (PAO),' an agent that forms stable ring complexes with vicinal or neighboring dithiols (17). Previous studies had shown that PA0 blocks insulin-stimulated glucose uptake without affecting insulin binding to its receptor, insulin receptor autophosphorylation, or the receptor's capacity to catalyze protein substrate phosphorylation (17-19). The site of PA0 action that leads to accumulation of pp15 may be a tyrosine-specific phosphatase (20,21). Preliminary experiments have confirmed the existence of two membrane-bound enzymes which catalyze the dephosphorylation of pp15 (22). Both enzymes are inhibited by PA0 and have been extensively purified in our laboratory from 3T3-Ll adipocytes. The formation of pp15 is insulinspecific, epidermal growth factor, platelet-derived growth factor, and insulin-like growth factor 1 being inactive despite the occurrence of receptors for these factors in 3T3-Ll adipocytes 07).
The cellular function of pp15 is not known. Our earlier experiments suggested that pp15 may serve an intermediary role in insulin-activated glucose uptake (17,21). The reciprocal effects of PAO, i.e. the accumulation of pp15 and the inhibition of glucose uptake, are both rapidly reversed by the vicinal dithiol, 2,3-dimercaptopropanol but not by the monothiol, 2-mercaptoethanol. The temporal relationship between insulin receptor autophosphorylation, p15 phosphorylation, and insulin-activated glucose uptake is also consistent with an intermediary role for ppl5.
Recently, pp15 was purified to homogeneity from . Amino acid sequence analysis of phosphotyrosine-containing peptides generated by tryptic cleavage of pp15 revealed that pp15 is the phosphorylated product of 422(aP2) protein, i.e.

21076
Fatty Acids Activate Phosphorylation of 422(aP2) Protein By coincidence, the full-length cDNA corresponding to 422(aP2) protein had already been cloned and sequenced in this laboratory (24). By its striking amino acid sequence homology to myelin P2 protein (69%), 422(aP2) was recognized as a member of a family of related 15-kDa proteins known as fatty acid-binding proteins (FABP) (25)(26)(27). These proteins have no known enzymatic activity but appear to bind straight chain lipophilic ligands such as fatty acids or retinoids (25,28,29). Even so their natural ligands and the biological consequences of binding, if any, are not known. When studied crystallographically, relatively diverse members of the family have been shown to conform to a distinctive tertiary structure described as a "/3-clam" with two parallel planes of perpendicular strands of P-pleated sheet forming a ligand cavity closed by a short helix-turn-helix "door" (27,30,see Fig. 15). Of most significance, it was shown in the structure of bovine myelin P2 protein that the tyrosine residue analogous to Tyrlg of 422(aP2) is oriented inside the ligand cavity, a state apparently incompatible with access to the active site of a macromolecular catalyst. Consistent with the "sequestered tyrosine" model, our early attempts to phosphorylate 422(aP2) with the insulin receptor, or to iodinate 422(aP2) by standard methods resulted in unexpectedly low yields. On this basis we began to explore the possibility of ligand-induced conformational activation of 422(aP2) protein as a substrate for the insulin receptor. EXPERIMENTAL PROCEDURES'

RESULTS
Purification of 422faP2) Protein-Since earlier studies (17,21,23) indicated that 422(aP2) protein is phosphorylated by the insulin receptor tyrosine kinase in the intact 3T3-Ll adipocytes, it was important to characterize this enzymatic phosphorylation reaction in vitro with purified components. As our previous method of purifying the phosphorylated form of 422(aP2) protein (0-  protein) involved steps that denature the protein with urea and SDS (23), a less drastic procedure was sought to obtain native homogeneous 422(aP2) protein. Fully differentiated 3T3-Ll adipocytes in monolayer culture were used as starting material for the purification.
Progress of the purification was monitored by a radioimmunoassay which utilized polyclonal antibodies directed against a synthetic peptide corresponding to the 12 COOH-terminal amino acids of 422(aP2) protein (see "Experimental Procedures"). The first two steps of the purification procedure involved perforating the plasma membranes of cells with dilute buffered digitonin followed by ammonium sulfate fractionation of the released cytosolic proteins (32). Since 422(aP2) protein, like pp15 (23), is localized in the cytosol, cell lysis with digitonin releases 422(aP2) protein leaving the cytoskeletal, organellar proteins, and triacylglycerol droplets within the lysed cell monolayer (32). Ammonium sulfate fractionation of the digitonin supernate revealed that, whereas pp15 is precipitated by 60% saturated ammonium sulfate, 422(aP2) protein remains soluble in 70% saturated ammonium sulfate. This step resulted in a g-fold purification of 422(aPZ) protein (Table I). After reducing the volume and ammonium sulfate concentration of the supernate by ultrafiltration (Amicon YM5 filter) and dialysis, 422(aP2) protein was further purified ' Portions of this paper (including "Experimental Procedures," Figs. 1-7, and Tables I-III) 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 is available from Waverly Press.
by HPLC (MONO S cation exchange column) using a NaCl gradient. The elution profile shown in Fig. 1 reveals one major and several minor protein peaks. The major and one minor peak (peaks 1 and 2, respectively) react with antibodies directed against the COOH-terminal sequence of 422(aP2) protein. One-dimensional SDS-PAGE and two-dimensional (nonequilibrium isoelectric focusing SDS/PAGE) gel electrophoresis (Fig. 2) carried out on the major protein peak showed a single Coomassie Blue-staining protein. Consistent with the properties of pp15 (0-phospho-  protein), this protein migrates with an apparent molecular mass of 15 kDa, but with a more basic p1 (about 8.5) than pp15 (p1 = 6.3). Two-dimensional gel analysis of the minor peak (peak 2) (results not shown) reveals that it is composed of approximately equal amounts of two 15-kDa proteins that may be associated as a heterodimer.
One of the two proteins migrates identically to the major protein in peak 1 by two-dimensional gel analysis (nonequilibrium isoelectric focusing SDS/PAGE), while the other protein has a more basic p1. Both proteins react by Western blot analysis with antibodies against the COOH terminus of 422(aP2) protein (data not shown) and thus, are at least partially related. The more basic protein found in peak 2 is probably a minor isoform of 422(aP2) protein. The protein in the major peak 1, presumably 422(aP2) protein, was further characterized.
Identification and Characterization of Peak 1 Protein as 422(aP2) Protein-Several studies were conducted to verify that the protein (peak 1) purified to homogeneity was in fact 422(aP2) protein. It was shown above ( Fig. 1) that the purified protein reacts with antibodies directed against the COOH terminus of 422(aP2) protein. In addition, Western blot analysis (results not shown) revealed that the purified protein migrates upon two-dimensional (nonequilibrium isoelectric focusing SDS/PAGE) gel electrophoresis as a protein with an apparent molecular mass of 15 kDa and a basic p1 (approximately 8.5). The amino acid composition of the purified protein was found to be virtually identical to that predicted by the deduced amino acid sequence from 422(aP2) cDNA (Table II).
The purified protein was subjected to limited proteolytic digestion with trypsin after which the tryptic fragments were purified by reverse-phase HPLC on a Cq column. The tryptic peptide profile ( Fig. 3) is similar to that reported by Matarese and Bernlohr (25) using a Cu reverse-phase column. Three of the major tryptic peptides (A-C in Fig. 3) were subjected to gas-phase amino acid sequencing. The sequences of the three peptides matched exactly those of tryptic peptides predicted from the amino acid sequence from 422(aP2) cDNA (Table  III). It is of interest that Peptide C corresponds to the tryptic peptide of pp15 which is phosphorylated in the intact cell (23). Based on these findings we conclude that the purified protein (peak 1) is 422(aP2) protein.
Verification That pp15 Is Phosphorylated 422CaP2) Protein-Our previous investigation showed that the amino acid sequence of the phosphotyrosine-containing tryptic peptide derived from ppl5 corresponded to an amino acid sequence near the NH2 terminus of 422(aP2) protein (23). Nevertheless, it was important to verify the identity of pp15 as the phosphorylation product of 422(aP2) protein by independent means. This was accomplished by two approaches. First, as shown in Fig. 4 [32P]pp15 in the cytosol from 32Pilabeled 3T3-Ll adipocytes (treated with insulin, PAO, and vanadate to maximize the accumulation of pp15) was specifically immunoprecipitated almost quantitatively (86%) by an affinity purified antibody directed against a synthetic peptide corresponding to the COOH-terminal sequence of 422(aP2) Fatty Acids Activate Phosphorylation of 422(aP2) Protein protein. Immunoprecipitation of [32P]pp15 was almost completely blocked by the synthetic peptide antigen (Fig. 4). The fact that this antibody, which is directed against an epitope different from the phosphorylation site on 422(aP2) protein, efficiently precipitates pp15 supports the view that this protein is a cellular substrate of the insulin receptor tyrosine kinase. The two minor phosphoproteins that were also specifically immunoprecipitated (Fig. 4) are probably phosphorylated isoforms of 422(aP2) protein.
Second, purified 422(aP2) protein was phosphorylated in uitro by the tyrosine kinase of the insulin receptor isolated and purified from membranes of 3T3-Ll adipocytes. The receptor was first activated by autophosphorylation for 5 min in the presence of insulin and [Y-~*P]ATP as previously described (6) after which purified 422(aP2) protein was added and its phosphorylation monitored. After quenching the reaction, the proteins were separated by two-dimensional gel electrophoresis and 32P-labeled proteins were visualized by autoradiography (Fig. 5A). The migration of [32P]pp15 isolated from 3T3-Ll adipocytes is shown for comparison ( Fig.  5B). A mixture containing equal amounts of 32P from both ["*P]pp15 and phosphorylated [32P]422(aP2) protein was also analyzed on a companion gel (Fig. 5C). Since pp15 and phosphorylated 422(aP2) protein migrate identically, it is evident that they possess similar p1 values and subunit molecular weights. Interestingly, although the 422(aP2) protein used as substrate for the receptor kinase was apparently homogeneous by staining with Coomassie Blue, a small satellite phosphoprotein (in addition to pp15) appeared upon radiography. A similar satellite [32P]phosphoprotein spot (which was removed by the purification procedure) is found when extracts of intact 3T3-Ll adipocytes labeled with 32Pi and treated with insulin and PA0 are subjected to immunoprecipitation using anti-422(aP2) protein antibodies (Fig. 4).
[32P]pp15 and 32P-phosphorylated 422(aP2) protein were first isolated from two-dimensional polyacrylamide gels and then exhaustively digested with trypsin. As illustrated in Fig. 7, the HPLC elution profiles of the [32P]phosphopeptides were virtually identical. Although ppl5 is phosphorylated on a single tyrosine, cleavage by trypsin is incomplete, yielding three phosphopeptides ( Fig. 7 and Ref. 23). The three negatively charged amino acids, i.e. phosphotyrosine and 2 aspartic acid residues, adjacent to Lys*' in pp15 and phosphorylated 422(aP2) protein (See Table III, peptide C), appear to inhibit cleavage by trypsin. The same tryptic site is readily cleaved, however, in unphosphorylated 422(aP2) protein (see Table III, peaks C and A). Sequence analysis of the three phosphopeptides (23) revealed that the first peak is a limit digestion while peaks 2 and 3 are derived from incomplete digestion.
Activation by Long-chain Fatty Acid of Insulin Receptor Tyrosine Kim-se-catalyzed Phosphmylution of 422(aP2) Protein-As demonstrated above the insulin receptor tyrosine kinase catalyzes the phosphorylation (on tyrosine) of purified 422(aP2) protein in uitro. The rate of phosphorylation of this protein by the isolated receptor was, however, relatively slow. Since 422(aP2) protein is structurally similar to a family of fatty acid-binding proteins (25)(26)(27) and is itself known to bind .

21077
interaction might alter its properties as a substrate. Moreover, since pp15 accumulates in insulin-treated 3T3-Ll adipocytes only when PA0 is present, it was also important to determine whether PA0 per se interacts directly with 422(aP2) protein to render it a better substrate for the insulin receptor tyrosine kinase. PA0 is an agent that forms relatively stable complexes with vicinal or neighboring dithiols (47). To assess the effect of fatty acid on the properties of 422(aP2) protein as substrate of the insulin receptor tyrosine kinase, 422(aP2) protein was first incubated with oleic acid (10~1 molar ratio of fatty acid to protein) to allow the fatty acid to bind. Then insulin receptor, previously autophosphorylated for 5 min either in the presence or absence of insulin, was added to the 422(aP2) protein-oleic acid complex, and protein substrate phosphorylation was followed for 5 min. As shown in Fig. 8A, oleic acid introduced into the reaction mixture with 422(aP2) protein at the time substrate phosphorylation was initiated had no effect on the extent of basal or insulin-stimulated autophosphorylation of the receptor's P-subunit. In other experiments not shown, it was also demonstrated that oleic acid had no effect on receptor autophosphorylation even when the receptor had been preincubated with fatty acid prior to initiating autophosphorylation. Oleic acid did, however, dramatically increase the rate of phosphorylation of 422(aP2) protein by receptor autophosphorylated both in the presence and absence of insulin (Fig. 8B).
The effect of oleic acid is entirely dependent upon insulin receptor since, in the absence of receptor, phosphorylation of 422(aP2) protein did not occur (results not shown). It was also shown (Fig. 6, A and 8 A, liposomes containing variable amounts of [3H]oleic acid were incubated with 10 pM 422(aP2) protein for 1.5 h at 37 "C. The amount of oleate bound to 422(aP2) protein was then determined as described under "Experimental Procedures." Inset, Scatchard analysis of binding isotherm. B, ['HI oleic acid was added to 422(aP2) protein in ethanol and incubated 1.5 h at 37 "C. Free fatty acid was then removed with ice-cold HAP-dextran and the labeled oleate in the supernate determined as described under "Experimental Procedures." Inset, Scatchard analysis. acid in ethanol for the activation of substrate phosphorylation.
Since oleic acid markedly alters the properties of 422(aP2) protein as substrate of the receptor kinase, it was of interest to determine 1) whether purified 422(aP2) protein contained bound fatty acid, and 2) the affinity with which exogeneous fatty acid binds to the protein. To determine whether the 422(aP2) protein preparation contained fatty acid, the purified protein was extracted with a chloroform-methanol mixture and any fatty acids in the extract were converted to their methyl esters and analyzed by gas chromatography (GC). GC analysis revealed that the purified 422(aP2) protein preparation was devoid of detectable fatty acids. Control experiments in which oleic acid was added exogeneously to purified 422(aP2) (at a 1:l molar ratio of oleic acid to protein), followed by incubation at 37 "C for 1.5 h, extraction and methylation of the fatty acid and analysis by GC, demonstrated that the fatty acid could be quantitatively extracted, methylated, and analyzed by GC. By this method myristic, palmitic, stearic, oleic, linoleic, linolenic, and arachidonic acids could have been detected at a fatty acid to protein molar ratio of <0.003:1.0. We conclude that our purified 422(aP2) protein preparation is devoid of bound fatty acids.
The capacity of purified 422(aP2) protein to bind [3H]oleic acid was investigated using two different methods (the liposome delivery and hydrophobic affinity matrix methods) for delivering labeled fatty acid to the protein and for removing unbound free fatty acid following its equilibration with the protein (28,42). For the liposome delivery method, liposomes containing egg lecithin and cholesterol in a 3:l molar ratio and   properties of 422(aP2) protein described above are in agreement with those reported by Matarese and Bernlohr (25). The time course (not shown) of phosphorylation of 422(aP2) protein by the insulin receptor tyrosine kinase in the absence or presence of oleic acid (lO:l, oleic acid/422(aP2) protein molar ratio) revealed that substrate phosphorylation was linear for 5 min. Thus, all substrate phosphorylation reactions were carried out for 5 min unless otherwise stated.
The dependence of phosphorylation rate on concentration of 422(aP2) protein was investigated in the absence and presence of oleic acid (Fig. 10). Lineweaver-Burk analysis of these data (Fig. 10, inset) indicates that the addition of oleic acid, and presumably its binding to the protein, caused a decrease of the Km for 422(aP2) protein in the phosphorylation reaction from 170 to 3 pM, but had little effect on the V znax-Insulin had no effect on the K,,, for 422(aP2) protein but increased V,,,,, 3-4-fold (result not shown).
The dependence of the rate of phosphorylation of 422(aP2) protein on oleic acid concentration was also investigated. 422(aP2) protein was incubated at 37 "C with the fatty acid for 1.5 h prior to initiating the phosphorylation reaction. The rate of 422(aP2) protein phosphorylation increased markedly with oleic acid concentration before reaching a maximal rate near a 6:l molar ratio of fatty acid to 422(aP2) protein (Fig.  11).
Homologous series of long-chain saturated and unsaturated fatty acids were tested for their ability to activate the phosphorylation of 422(aP2) protein. Fatty acids were added to 422(aP2) protein at a 1O:l molar ratio of fatty acid to protein, incubated 1.5 h at 37 "C, and then added to preactivated receptor for 5 min. All fatty acids tested dramatically increased 422(aP2) phosphorylation (Fig. 12), although unsaturated fatty acids generally had a greater activating effect than saturated fatty acids.
Effect of Oleic Acid on the Phosphorylation of 422CaP2) Protein Catalyzed by the Cytoplasmic Domain of the Insulin Receptor-The insulin receptor preparation used in the substrate phosphorylation studies described above was extracted from membranes, purified, and assayed in the presence of the Such factors could account for the relatively limited period (-5 min) during which the phosphorylation of oleic acid-loaded substrate remained linear after the reaction was initiated with detergent-containing receptor (see above). It was, in fact, demonstrated that the stimulating effect of oleic acid on substrate phosphorylation decreased progressively with time of preincubation of the 422(aP2) protein-oleic acid mixture with 0.1% Triton X-100 prior to initiating substrate phosphorylation with receptor. Thus, after a 60-min or overnight preincubation with detergent, the activating effect of oleic acid was 30 and 15%, respectively, than that observed without detergent preincubation.
To avoid and thereby assess the effect of sequestration of fatty acid by detergent, substrate phosphorylation experiments were also conducted in the absence of detergent by using a truncated form of the human insulin receptor (31,43) which contains only the soluble 45-kDa cytoplasmic tyrosine kinase domain. The cytoplasmic kinase domain is constitutively highly active in the absence of detergent both with respect to autophosphorylation and catalysis of substrate phosphorylation (31,43). The oleic acid concentration dependence of phosphorylation of 422(aP2) protein catalyzed by the detergent-free truncated receptor kinase domain (Fig. 13) was found to be similar to that catalyzed by the wild-type receptor (Fig. 11). The phosphorylation of 422(aP2) protein by both receptor kinases is markedly activated by fatty acid achieving a maximum rate at an oleic acid/422(aP2) protein molar ratio of 6:l. The close agreement of these results indicates that substrate activation by oleic acid is not compromised by the presence of detergent when the phosphorylation assay is brief (5 min). However, removal and sequestration of fatty acid bound to 422(aP2) protein by detergent micelles reduces substrate activation when the phosphorylation assay exceeds 5 min. Thus, linear rates of substrate phosphorylation persist longer in the absence of detergent (Fig. 13).
Effect of Bound Fatty Acid on the Accessibility to Iodinution of Tyrlg and Tyr'*' WI 422CaP2) Protein at increasing molar ratios of 1:l (W), 2:l (a), 4:l (A), 6:l (A), and 10~1 (0) as described in Fig. 11 together with recent crystallographic data on a structurally similar protein, myelin P2 protein (27), strongly suggest that Tyrlg in native 422(aP2) protein without bound fatty acid is inaccessible to the active site of the insulin receptor tyrosine kinase. Moreover, it appears that upon binding fatty acid, a conformational change occurs whereby Tyr" becomes more readily accessible to the kinase active site. To test this hypothesis by an independent method, the accessibility of the 2 tyrosine residues, Tyrlg and Tyr"', in 422(aP2) protein to lactoperoxidase-catalyzed iodination was tested in the presence and absence of excess fatty acid. Although the mechanism of oxidative iodination of tyrosine is not known, it is believed to involve the enzymatic generation of a small metastable reactive species, containing at most a few atoms, which diffuses to the ultimate sites of labeling (48). For this reason, we expected Tyr19 to show a less stringent fatty acid requirement for iodination than for insulin receptor mediated phosphorylation. As shown in Fig. 14A iodination of 422(aP2) protein in the absence of fatty acid reached a plateau after about 3 min. Ethanol alone had no effect on iodination.
As shown in Fig.  14B almost all of the radioactivity incorporated in the absence of fatty acid was attributed to Tyr". When fatty acid is included at a lo-fold molar excess over 422(aP2) protein, Tyr" is iodinated at a similar rate, but the reaction continues for at least 20 min, resulting in a greater total incorporation of radiolabel.
More striking, in the presence of fatty acid, Tyr"' is labeled at a rate far greater than that observed in the absence of fatty acid. These results strongly support our hypothesis of a fatty acid-induced conformational change of 422(aP2) protein and provide a second sensitive criterion with which to define a native ligand-free conformational state of 422 (aP2)

DISCUSSION
There is compelling evidence that the tyrosine kinase activity of the insulin receptor is important in signal transmission initiated by the binding of insulin (1). Previously, we reported (17,21) the accumulation of a 15-kDa protein (~~15) phosphorylated on tyrosine in 3T3-Ll adipocytes treated with insulin and PAO. Insulin appeared to activate receptor-catalyzed phosphorylation of p15 while PA0 blocked the turnover of the pp15 tyrosine phosphoryl group (17,(20)(21)(22), thereby leading to its accumulation.
Our recent discovery (23) that pp15 is  protein initiated our attempts to reconstitute the phosphorylation reaction in vitro using purified proteins. Although there is no known catalytic activity for any fatty acid-binding protein, our strategy for maintaining the native state of 422(aP2) protein was to employ purification methods which are generally regarded as mild with respect to maintenance of enzyme activities. Our procedure differs from that of Matarese and Bernlohr (25) in that we used ammonium sulfate fractionation instead of gel filtration, delipidation with Lipidex-1000, and thiol affinity chromatography.
We obtained purified protein free of endogeneous fatty acid in a yield (38%) probably similar to that reported by these authors.
Quantitative analysis of 3T3-Ll adipocytes with immunochemical methods showed that 422(aP2) protein is abundant, comprising about 3% of cellular protein.
By far the vast majority of this protein is accounted for by the form we have purified.
Interestingly, two minor 15-kDa phosphoproteins were identified (Fig. 4)  In the absence of fatty acid, Tyr" is sequestered from the insulin receptor in the ligand (fatty acid-)-binding cavity of the P-pleated sheet (thin lines) domain. In this state iodination occurs exclusively, but to a low extent, on Tyrlg. This drawing is adapted from the crystal structure of bovine myelin P2, a closely related homologue of 422(aP2) protein, reported by Jones et al. (27).
In the presence of fatty acid, Tyr" becomes accessible to the insulin receptor protein tyrosine kinase by an unknown mechanism, here represented through a hinge motion of the a-helical (bold lines) domain with respect to the P-pleated sheet domain. In this state, iodination of both Tyr" and Tyrlz8 occurs.  into three immunologically indistinguishable forms that differ in their composition of bound fatty acid. Bound fatty acid does not appear to explain the heterogeneity in 422(aP2) protein, since incubation with a molar excess of oleic acid does not affect its ~1." A similar finding has been reported (51) for two immunologically identical FABPs from rat mammary tissue separated by isoelectric focusing or DEAE-HPLC. Interestingly, these FABPS, like 422(aP2) protein, are quite homologous to heart FABP. Although the origin and significance of the isoforms are unknown, there is no question that the major form of 422(aP2) protein gives rise to the major phosphorylated species that we have identified as pp15 in vivo and in vitro, Phosphorylation of purified 422(aP2) protein in vitro by the 3T3-Ll insulin receptor tyrosine kinase was dependent on the addition of fatty acid. Kinetic analysis showed that this effect was largely explained by a reduction in the apparent & for 422(aP2) protein from 170 to 3 pM with only an insignificant change in V,,,,, (Fig. 10). Fatty acid did not affect the autophosphorylation of the insulin receptor. Although all fatty acids tested gave some stimulation of phosphorylation, there was a slight preference for unsaturated fatty acids (Fig.  12). The stimulatory activity of saturated fatty acids militates against the possibility that the stimulating effect is caused by traces of oxidative degradation products that accumulate in preparations of unsaturated lipids. The mechanism by which fatty acid activates 422(aP2) protein as substrate of the insulin receptor kinase appears to involve changes in the tertiary structure of the protein. In Fig. 15 we interpret our observations in a two-state model for 422(aP2) protein based in part on the crystal structure of myelin P2 protein (27), a close structural homologue. It is noteworthy that the crystal structure for myelin P2 protein showed a continuous band of electron density within the ligand binding cavity that was interpreted to be bound fatty acid. We suggest that fatty acid induces a conformational change, here portrayed as a hinge motion between the o(helical and @-pleated sheet domains of the protein. Thereby, Tyrlg is brought out of the fatty acid-binding pocket permitting it to undergo phosphorylation by the insulin receptor or enhancing its iodination. The same conformational change ' R. C. Hresko, unpublished results. renders Tyri's accessible to iodination. This model is supported by kinetic data. Fatty acid activation of 422(aP2) protein results from a large decrease in its K, as a substrate for the insulin receptor tyrosine kinase with no significant change in V,,,,, (Fig. 10). In this light, the activation process can be thought of as an increase in the effective substrate concentration.
We visualize that the two states depicted in Fig. 15 are in equilibrium and that fatty acid shifts the equilibrium toward the state in which Tyrlg is exposed. The existence of this equilibrium is supported by the time course of iodination.
In the absence of fatty acid the small fraction of protein substrate in the "active state" is rapidly consumed during the brief linear phase of iodination (Fig. 14). The plateau that follows the linear phase suggests that the rate of spontaneous transition from the inactive state to the active state is slow. When fatty acid is present the linear phase of iodination persists much longer indicating an increased fraction of the protein substrate in the active state.
Our work is the first reported instance of a ligand-induced conformational change in a fatty acid-binding protein leading to an altered function. No other fatty acid-binding protein is known to undergo tyrosine phosphorylation.
Despite these considerations, several other fatty acid-binding proteins show extensive primary sequence homology and presumably also have tertiary structures similar to 422(aP2) protein. Four of these proteins, bovine myelin P2 (52), rat heart FABP (53), rat cellular retinoid-binding protein (54,55), and bovine mammary-derived growth inhibitor protein (26) possess a conserved tyrosine residue analogous to Ty?' of 422(aP2) protein in the context of a consensus protein tyrosine kinase recognition motif (56). The potential for these related proteins to undergo ligand-induced conformational changes or substrate activation for tyrosine phosphorylation has not been explored. The significance of fatty acid-induced substrate activation of 422(aP2) protein in the  adipocyte may relate to the anti-lipolytic function of insulin. It has long been known (57,58) that insulin inhibits lipolysis in adipocytes stimulated by P-adrenergic and other lipolytic hormones. These hormones activate the triacylglycerol lipase of adipocytes through phosphorylation of the enzyme by the CAMP-dependent protein kinase (58, 59). Lipolysis is also subject to feedback inhibition by free fatty acids which accumulate in adipocytes under various physiological circumstances (60). Recent evidence from Bernlohr's laboratory (25,(44)(45)(46) has shown convincingly that 422(aP2) protein mediates intracellular fatty acid transport in 3T3-Ll adipocytes. In this connection our results reveal that loading 422(aP2) protein with fatty acid renders the protein susceptible to phosphorylation by the insulin receptor tyrosine kinase. Conceivably, phosphorylation of 422(aP2) protein might decrease its capacity to bind/ transport fatty acids and lead to their accumulation, thereby inhibiting lipolysis. Thus, insulin might serve its anti-lipolytic role both by altering the phosphorylation state of the lipase and by causing the accumulation of free fatty acids. The possible role of phosphorylation of 422(aP2) protein by these mechanisms is currently under investigation with intact 3T3-Ll adipocytes.
The role of fatty acid-binding proteins in biology has long been debated. The putative consequences of fatty acid binding have been limited to protection of cells from toxic high local concentrations of fatty acid. Our observations suggest the existence of dynamic conformational changes that may transduce intracellular signals in a wide range of biologic systems.