Control of endogenous phosphorylation of the major cAMP-dependent protein kinase substrate in adipocytes by insulin and beta-adrenergic stimulation.

In isolated, 32Pi-loaded, rat adipocytes, we have examined phosphorylation of the major cAMP-dependent protein kinase (A-kinase) substrate, a protein that appears to be associated with the lipid storage droplet and migrates in sodium dodecyl sulfate-polyacrylamide gel electrophoresis as a 65-67-kDa doublet. In control cells, a strong phosphorylation signal is detected as the (+/- cAMP) A-kinase activity ratio ranges from approximately 0.1 to approximately 0.3-0.4 with increasing isoproterenol concentrations. By contrast, insulin-treated cells exhibiting A-kinase activity ratios over the range of 0.1-0.25 contain less 32P in the 65-67-kDa protein than control cells exhibiting identical A-kinase activity ratios. At higher activity ratios (greater than 0.3), this reduction in phosphorylation of the 65-67-kDa protein by insulin disappears. It is concluded that insulin stimulates a phosphatase activity that acts on the 65-67-kDa protein. Insulin actions aside, these studies reveal two interesting phenomena. 1) Whereas elevated, steady-state A-kinase activities are established rapidly (1-2 min) upon isoproterenol stimulation, phosphorylation of the 65-67-kDa substrate proceeds through a burst, followed by a decline to a steady-state level by 10-12 min. An "adaptation" mechanism, providing for a constant response to a constant stimulus, may underlie this lack of parallelism between the time course of phosphorylation and A-kinase activity. 2) Removal of [32Pi] orthophosphate immediately before isoproterenol stimulation leads to a rapid (t approximately 10 min) loss in labeling of the 65-67-kDa protein, whereas the phosphorylation state of other phosphoproteins are not changed. These data suggest that elevation of A-kinase activity leads to a rapid exchange of external Pi with an ATP pool that is used by A-kinase.

In isolated, 32Pi-loaded, rat adipocytes, we have examined phosphorylation of the major cAMPdependent protein kinase (A-kinase) substrate, a protein that appears to be associated with the lipid storage droplet and migrates in sodium a0aecyi sulfate-polyacrylamide gel electrophoresis as a 65-67-kDa doublet. In control cells, a strong phosphorylation signal is detected as the (2 CAMP) A-kinase activity ratio ranges from ~0.1 to ~0.3-0.4 with increasing isoproterenol concentrations.
By contrast, insulin-treated cells exhibiting A-kinase activity ratios over the range of 0. l-0 25 . contain less 32P in the 65-67-kDa protein than control cells exhibiting identical A-kinase activity ratios. At higher activity ratios (>0.3), this reduction in phosphorylation of the 65-67-kDa protein by insulin disappears.
It is concluded that insulin stimulates a phosphatase activity that acts on the 65-67-kDa protein.
Insulin actions aside, these studies reveal two interesting phenomena.
1) Whereas elevated, steady-state A-kinase activities are established rapidly (1-2 min) upon isoproterenol stimulation, phosphorylation of the 65-67-kDa substrate proceeds through a burst, followed by a decline to a steady-state level by lo-12 min. An "adaptation" mechanism, providing for a constant response to a constant stimulus, may underlie this lack of parallelism between the time course of phosphorylation and A-kinase activity.
2) Removal of [32Pi] orthophosphate immediately before isoproterenol stimulation leads to a rapid (t = 10 min) loss in labeling of the 65-67-kDa protein, whereas the phosphorylation state of other phosphoproteins are not changed. These data suggest that elevation of A-kinase activity leads to a rapid exchange of external Pi with an ATP pool that is used by A-kinase.
Previously, by simultaneously monitoring steady-state levels of CAMP-dependent protein kinase (A-kinase)' activity * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Recipient of National Research Service Award F32 DK07685. To whom correspondence should be addressed.
A direct demonstration of this putative insulin-stimulated dephosphorylation of hormone-sensitive lipase requires isolation and biochemical dissection of the enzyme to determine the phosphate content of the "regulatory" phosphorylation site within the enzyme (2). As an alternative approach, we sought other, more abundant A-kinase substrates in fat cells, in order to probe the question of insulin-stimulated dephosphorylations.
The goal was to compare directly the phosphorylation state of such proteins with the steady-state levels of A-kinase activity in order to determine if insulin stimulated the removal of phosphates by mechanisms independent of its ability to decrease CAMP, and therefore, lower A-kinase activity. In this paper we examine the phosphorylation state of the major adipocyte substrate for A-kinase, a 65-67-kDa protein that is apparently associated with the lipid storage droplet. Out findings indicate that insulin decreases the phosphorylation of this protein by a CAMP-independent process.
EXPERIMENTAL PROCEDURES*

Identification
of Major Adipocyte Phosphoproteins and Akinme Substrates-For reasons described in detail previously (4,5), in order to obtain a strong phosphorylation signal upon elevation of A-kinase with isoproterenol, control cells were incubated with adenosine or the potent adenosine receptor agonist, PTA (10). Fig 1 shows an autoradiograph of 32Pproteins in the supernatant, membrane, and fat "cake" fractions which were separated by SDS-PAGE.
Clearly, the most dramatic changes upon isoproterenol stimulation were seen in proteins associated with the fat fraction, especially those at 84, 65-67, 45, and 36 kDa. Among this group, there was copious incorporation of radioactivity into the protein(s) that migrate at 65-67 kDa, the species we selected as a model The overall distribution of proteins, as determined by silver staining, among supernatant, membranes, and fat cake fractions did not mirror the distribution of radiolabeled proteins (data not shown). Despite the preponderance of phosphoproteins in the fat, especially in the stimulated cells, less than 20% of the stainable proteins were found in this fraction. Some of the fat-associated phosphoproteins may be membrane proteins which adhered to or were trapped within the fat cake, as evidenced by their distribution between the gross membrane and fat cake fractions, such as the 75 and 56-kDa species that are prominent phosphoproteins in both resting and stimulated cells. However, several of the major phosphoproteins appeared to be located exclusively in the fat fraction, including those at 45, 65-67-, and 84-kDa protein, which is hormone-sensitive lipase (11,12). Similarly, the 62-kDa protein in unstimulated cells was found only in the fat cake. Rapid Termination of Cell Incubations and Recovery of ,"P-Proteins-The reasons for terminating incubations of "'Ploaded cells by acid precipitation and the methodology for separating cellular proteins from exogenous BSA are described under "Experimental Procedures." Fig. 2 presents a comparison of the phosphoproteins harvested after extraction of BSA with the phosphoproteins present in whole adipocytes that had been incubated briefly in a BSA-free medium. The results demonstrate that most, if not all, of the major detectable phosphoproteins were recovered with this procedure, and the dramatic increase in phosphorylation of the 65-67.kDa protein seen in the extracts of trichloroacetic acid precipitates reflected that seen in whole cells washed free of BSA. Scanning densitometry (data not shown) revealed that >80% of all radiophosphate incorporated into all adipocyte proteins was found in this single, fat-associated ( Fig. 1)  .,_ and suspended in KRH without BSA at a cell concentration of 50% (v/v).
After a brief (2-3 min) incubation with either PIA or isoproterenol, the adipocytes were washed once in media containing the respective inhibitory or stimulatory ligands and centrifuged briefly to paik the cells. Ten ;l of packed ceils were removed and added to 200 ul of SDS-PAGE samnle buffer containing 20% SDS, a high concentration necessary to solubilize the lipid in the fat cells. The right side of the autoradiograph (Extracted Cells) shows ""P-proteins from trichloroacetic acid precipitates of 2% (v/v) cell incubations after extraction of BSA as described under "Experimental Conditions." Equivalent aliquots of extracted cells and whole cells were subjected to SDS-PAGE in 7.5% polyacrylamide gels. protein ( Fig. l), disappeared upon stimulation with isoproterenol (Fig. 2).

Resolution of 65-67-kDa Phosphoprotein
Doublet-For studies in which the phosphorylation state of proteins in metabolically regulated cells (see below) was investigated, all cell incubations were terminated and processed by the extraction method described above. Quantitation of the phosphorylation state of the 65-67-kDa protein was performed by scanning densitometry of autoradiographs.
Although not immediately apparent in the overexposed autoradiographs in the above figures, there was a clear doublet of phosphoproteins in the 65-67 kDa region. To test if the 65-and 67-kDa bands were a single protein whose migration in gels changed as a result of increased phosphate incorporation, a partially purified fraction of adipocyte protein from unlabeled cells was phosphorylated with exogenous A-kinase. As seen in Fig. 3A, upon phosphorylation of this fraction with a tracer concentration of [-y-"LP]ATP, the 65-kDa phosphoprotein was detected initially. However, within 3 min of addition of an excess of non-radioactive ATP, the sharp phosphoprotein band at 65 kDa became diffuse over the 65-67 kDa region, and with fat cells were subjected to finely graded increases in isoproterenol concentration in order to produce finely graded increases in the A-kinase activity ratio. As found previously by this laboratory under identical incubation conditions (5), glycerol release approached a maximum as the activity ratio approached approximately 0.4 in the non-radioactive cells (data not shown). Similarly, phosphorylation of the 65-67-kDa polypeptide in the parallel, ,"P-loaded cells was acutely sensitive to A-kinase activity ratios (Fig. 4A), and in numerous experiments, incorporation of "P ranged from nil to maximal as the A-kinase activity ratio ranged from ~0.1 to approximately 0.3-0.4 in the absence of insulin. As with the experiment in Fig. 2, increased incorporation of .jLP into the 65-67-kDa species was accompanied by a parallel loss of the 62-kDa phosphoprotein ( Fig. 4.4). The reciprocal relationship between phosphorylation of the two species, 62 and 65-67 kDa, was maintained in incubations in which cells stimulated by removal of adenosine (4, 5)  In order to view possible insulin effects not related to decreases in A-kinase actvity, autoradiographs and corresponding densitometric scans of the 65-67-kDa protein from control and insulin-replete incubations are aligned according to A-kinase activity (Fig. 4A). Note that since insulin decreases CAMP and thus, A-kinase activity, slightly higher isoproterenol concentrations were required in the insulinreplete incubations than in control incubations in order to achieve a given A-kinase activity ratio (see legend to Fig. 4). continued incubation nearly all of the radioactivity was found at the 67 kDa region (see densitometric scans, Fig. 3B). A reasonable interpretation of this experiment is that progressive phosphate incorporation after 7 min of incubation, not visible because of the lowered ATP specific radioactivity, altered the migration of the 65kDa phosphoprotein such that We showed previously (5) that under our incubation conditions, for any given isoproterenol concentration, an elevated A-kinase activity is established within l-2 min and remains constant for up to 25 min. However, in the 'lP1 replete incubations (Fig. 5A), labeling of the 65-67-kDa protein proceeded through an initial spike, followed by a decline, and steady-state phosphorylation of the protein was not established until 8-12 min. Isolated adipocytes were loaded with %'Pi at 10 &i/ml and maintained in the presence of '"Pi during the incubation with stimulatory and inhibitory ligands, such as isoproterenol and insulin. Cells were incubated with receptor ligands for 20 min. Incubations of radioactive cells were terminated with trichloroacetic acid, and the proteins were extracted as described under "Experimental Conditions." Non-radioactive cells were incubated in parallel for determinations of A-kinase activity ratios and glycerol content as described under "Experimental Procedures." Panel A shows autoradiographs of the 65-67 kDa region of SDS-PAGE gels of extracts of "'P-loaded fat cells that were subjected to a range of isoproterenol concentrations as indicated below, as were the nonradioactive cells shown in panel A. The '"P-loaded cells were incubated with the P-adrenergic agonist both in the absence (Control) and presence (+ Insulin) of 2000 microunits/ml of insulin. The 65-67-kDa band is noted with the arrow at the right of the autoradiograph strips. The arrow at the left of the autoradiograph strips denotes the 62.kDa band. Densitometric tracings over the 65 and 67 kDa region, as well as the area of each tracing in arbitrary absorbance units, are shown directly below the autoradiographs. Also shown below each scan is the A-kinase activity ratio for the identical incubation in the non-radioactive cells. The autoradiographs and densitometric scans are aligned to match incubations with and without insulin according to A-kinase activity ratios. Since insulin lowers CAMP and, thus, A-kinase activity ratios, slightly higher isoproterenol concentrations were necessary to achieve approximately equivalent A-kinase activities in the presence than in the absence of insulin. Panel B, derived from the data in panel A, is a plot of the combined densitometric scanning area of the 65-67 kDa region of the autoradiograph uersus the A-kinase activity ratios for incubations conducted in the absence (o---O) and presence (U) of insulin. Isoproterenol concentrations (UM), reading from left to right, were (-insulin) 1, 2, 4, 6, 9, 12, 24, 32, 64, and 128; (+insulin) 1, 2, 12, 18, 24, 32, 64, 128, and 1000. The arrow on the horizontal axis denotes the A-kinase activity ratio and corresponding area of the 65-67-kDa band for cells incubated without isoproterenol or insulin but in the presence of 200 nM adenosine. of phosphorylation over the incubation time course. Although ity ratio, steady-state phosphorylation of the 65-67 kDa prothe early time course of labeling was not studied in detail, a ceeds through a "burst," or overshoot, prior to declining to a number of different experiments revealed that the steadysteady-state level. state levels of phosphorylation achieved by 15-20 min were In the incubations conducted without readdition of "Pi, generally 20-40% lower than peak levels seen at 3-8 min. (Fig. 5B) 20% lower than in the "'Pi replete incubations. Remarkably, however, over 8-16 min after introduction of isoproterenol, there was a marked, time-dependent decrease (t = 10 min) in the radioactivity incorporated into the 6%67-kDa protein upon removal of the radioisotope from the medium (Fig. 5A). Note that the only difference between Fig. 5, A and B, is the presence or absence of 32Pi. Thus, the time-dependent loss of "'P incorporation into the 6%67-kDa protein occurred under conditions of active phosphorylation, presumably with ATP of rapidly decreasing specific radioactivity.
By contrast, over the relatively brief time course of these experiments, the presence or absence of radiophosphate during the second incubation had little or no effect on the phosphorylation state of other major phosphoproteins, all of which are not A-kinase substrates (data not shown). DISCUSSION A major goal of the present investigation was to examine the effects of insulin on the phosphorylation state of A-kinase substrates in intact adipocytes. Based on the arguments developed below, we conclude that insulin promotes the dephosphorylation of the major adipocyte A-kinase substrate by a mechanism separate from insulin-mediated lowering of CAMP and A-kinase activity.
As a model A-kinase substrate we chose the protein that migrates at 65-67 kDa on SDS-PAGE, by far the most abundant phosphoprotein in adipocytes in which A-kinase is elevated. Remarkably, the vast majority of 32P phosphates transferred to all adipocyte proteins by A-kinase are inserted into this single, as yet unidentified, protein. Indeed, as suggested by Garrison (14), we have found that examining phosphorylation changes of this protein is the most sensitive indicator of hormone action in intact cells. The protein is multiply phosphorylated by A-kinase, both in vitro and in uivo, resulting in reduced migration during SDS-PAGE, i.e. apparent molecular mass shift from 65 to 67 kDa. Preliminary analysis of proteolytic digests of the 65-and 67-kDa species provides further evidence that these bands are phosphorylation variants of the same protein (data not shown). Phosphorylation of the 65-67-kDa doublet is correlated closely with increasing A-kinase activity ratios over the range of 0.1 to approximately 0.35, the same range over which lipolysis is increased. HOWever, the 65-67 kDa is not hormone-sensitive lipase, which is an 84-kDa A-kinase substrate (11,12), and polyclonal antibodies against the lipase do not recognize the 65-67-kDa protein.3 With few exceptions (15, 16), this abundant phosphoprotein has been overlooked, primarily because it partitions into the fat cake, a fraction frequently discarded in adipocyte studies.
If insulin were to decrease the phosphorylation of A-kinase substrates by merely decreasing the A-kinase activity, the relationship between A-kinase activity and phosphorylation state would be identical in the presence and absence of insulin. However, this is clearly not the case (Fig. 4B). In the presence of insulin, at intermediate A-kinase activity ratios (approximately 0.15-0.25), there is substantially less phosphate in the 65-67-kDa protein than in the absence of insulin. Moreover, this difference in phosphorylation between the insulin-replete and insulin-deficient condition disappears as the A-kinase activity ratio is increased to approximately 0.4. Previously, with highly similar findings on the relationship between Akinase activity and glycerol release in the presence and absence of insulin, we surmised that insulin stimulated the dephosphorylation, and thus the inactivation, of hormonesensitive lipase (1). Recently, Stralfors and Honnor (17) confirmed our earlier speculation (1) with their finding that insulin reduced the phosphate content of hormone-sensitive lipase by a mechanism not explained by the CAMP-lowering effect of insulin. Our findings with the 65-67-kDa protein indicate that this insulin effect might apply generally to Akinase substrates. Moreover, as was argued previously for the effects of insulin on lipolysis (I), the ability of increasing Akinase activity to overcome the insulin-induced reduction in 65-67-kDa phosphorylation suggests that insulin activates a phosphatase. Thus, at the intermediate A-kinase activity levels, phosphatase activity prevails, whereas further increases in A-kinase activity overcome the effects of the phosphatase.
Although increased phosphatase activity has been invoked frequently to explain some physiological effects of insulin (see Ref. 18 for review), direct demonstrations of increased protein phosphatase activity are few, including early observations with fat cells (19), and recent studies with 3T3 cells (20,21). Additional support is derived from findings on insulin-mediated changes in the phosphorylation state of phosphatase inhibitors (22,23). Further insulin has been shown to decrease the phosphorylation of a number of proteins at sites not subject to phosphorylation by A-kinase, and thus not attributable to decreases in CAMP. In most cases, phosphatase activity has been the preferred explanation for these insulininduced dephosphorylations of, among others, glycogen synthase (24), phosphofructokinase-2 (25), acetyl-CoA-carboxylase (26), pyruvate dehydrogenase (27), and glycogen phosphorylase (28). However, changes in phosphorylation may result from changes in kinases, phosphatases, or both. In none of the above examples was it possible to measure simultaneously the active state of the kinase responsible for phosphorylating those sites which were dephosphorylated by insulin, and thus conclusions must be tempered by this lack of information. Indeed, an insulin-mediated decrease in multifunctional protein kinase has been demonstrated by Ramakrishna a 3. J. Egan, A. S. Greenberg, M.-K. Chang, and C. Londos, unpublished observations. and Benjamin (29). In the studies described herein, cell incubations were terminated in a manner which maintained both the endogenous activation state of A-kinase as well as the concomitant phosphorylation state of the major cell target of A-kinase action. As such, we have had the distinct advantage of knowing the intrinsic activation state of A-kinase under all conditions for which insulin promotes dephosphorylation of this major 65-67-kDa phosphoprotein. Insulin actions aside, there are two intriguing findings in these studies on the temporal relationships between A-kinase activity and phosphorylation of the 65-67-kDa protein. First, a gradient of increasing isoproterenol concentrations results in a gradient of increasing steady-state A-kinase activities, matched by a gradient of increasing steady-state protein phosphorylation states. Under our cell incubation conditions, isoproterenol-stimulated increases in A-kinase activity achieve steady-state by l-2 min and remain elevated over the time course of these experiments (4, 5), a finding confirmed by others (30). Similarly, phosphorylation of the 65-67-kDa protein occurs rapidly, albeit with an initial "overshoot" (see below). Given the rapid elevations of both A-kinase and phosphorylation, one might expect, in the absence of a compensatory phosphatase, all levels of stimulation to lead ultimately to maximal phosphorylation.
Clearly, this does not occur (Fig. 5A). Thus, it may be assumed that the cells have the capacity to adjust, temporally, to prolonged stimulation. It should be emphasized that this is not classical desensitization at the receptor level, since A-kinase activity, presumably a reflection of the steady-state CAMP concentration, is unchanged over the time course of these experiments.
The second interesting kinetic phenomenon in response to isoproterenol is the initial spike of phosphorylation prior to establishment, by 12-16 min, of an elevated steady-state level of phosphorylation.
Concomitant with an unchanging steadystate A-kinase activity level, there is an initial overshoot of "'P incorporation into the A-kinase substrate. This phenomenon might be explained by a rapid, but temporally limited, inhibition of a phosphatase, which would account for the initial burst of phosphorylation.
For example, transient phosphorylation and modification of a phosphatase inhibitor might occur (31). Alternatively, the declining phase of the spike might represent a delayed activation of a phosphatase. The simultaneous activation of both kinase and phosphatase activities would provide the cell a strategy for producing graded responses to graded stimuli and for generating a constant response to a constant concentration of stimulant. By mechanisms as yet unknown, it would appear that the fat cell "adapts" to constant stimulation to produce a constant response, akin to the response repertoire of unicellular organisms (32).
Finally, a further surprising finding is the rapid turnover of "P content of the 65-67-kDa protein upon removal of radioactive orthophosphate from the medium (Fig. 5B), a phenomenon not seen with other proteins that are not Akinase substrates. This finding, coupled with the maintenance of steady-state labeling of the 65-67-kDa species in the 32P replete medium (cf. Fig. 5, A and B), indicates that the phosphates in this protein are rapidly turning over. The enigma is the requirement for the continued presence of 32Pi in the medium to achieve steady-state 32P incorporation, especially since the half-time for disappearance of radioactivity in the protein (-10 min) is considerably shorter than the half-time required to achieve steady-state labeling of cellular ATP in adipocytes, usually ~40 min (24,33).4 These data "Under our incubation conditions, the time required to achieve steady-state '*P incorporation into [y-32P]ATP was in excellent agreement with data in the cited references (N. B. Garty and C. Londos, unpublished observations).
suggest that phosphate of rapidly decreasing specific radioactivity is introduced into the protein upon removal of radiophosphate from the cell incubation medium. Loading of cells with 32Pi is performed under unstimulated conditions, whereas the above mentioned "unloading" of radiophosphate from protein occurs under isoproterenol-stimulated conditions. The precipitous decline in radiophosphate content of the 65-67-kDa protein may result from a rapid equilibration of the radioactive ATP pool with the external non-radioactive orthophosphate upon stimulation with isoproterenol. Preliminary experiments indicate that isoproterenol stimulation does not lead to a drastic lowering of the specific radioactivity of the total cellular ATP pool labeled in the gamma position.4 Perhaps a small fraction of the total ATP pool serves as the substrate for A-kinase, as suggested by the findings of Mayer and Krebs (34), who observed that phosphates incorporated into phosphorylase kinase in 32P-leaded skeletal muscle were of higher specific radioactivity than phosphates in total cellular ATP. Whatever the underlying reason for this phenomenon, these data indicate that optimal radiolabeling of proteins requires the continued presence of 32Pi in the medium during stimulation of cells with hormones.