Control of the translational regulators PHAS-I and PHAS-II by insulin and cAMP in 3T3-L1 adipocytes.

The eukaryotic initiation factor 4E (eIF-4E)-binding proteins PHAS-I and PHAS-II were found to have overlapping but different patterns of expression in tissues. Both PHAS proteins were expressed in 3T3-L1 adipocytes, in which insulin stimulated their phosphorylation, promoted dissociation of PHAS·eIF-4E complexes, and decreased the ability of both to bind exogenous eIF-4E. The effects of insulin were attenuated by rapamycin and wortmannin, two agents that block activation of p70S6K. Unlike PHAS-I, PHAS-II was readily phosphorylated by cAMP-dependent protein kinase in vitro; however, the effects of insulin on both PHAS proteins were attenuated by agents that increase intracellular cAMP, by cAMP derivatives, and by phosphodiesterase inhibitors. These agents also markedly inhibited the activation of p70S6K. In summary, our results indicate that PHAS-I and -II are controlled by the mammalian target of rapamycin and p70S6K signaling pathway and that in 3T3-L1 adipocytes this pathway is inhibited by increased cAMP.

The stimulation of protein synthesis by insulin occurs in a wide variety of cell types and happens within minutes of a rise in the concentration of the hormone (1,2). Two distinct signaling pathways (3), one leading to activation of MAP 1 kinase and the other to p70 S6K , are among those that mediate insulin action. MAP kinase is a downstream element of the Ras-signaling pathway (4), whereas p70 S6K is downstream of mammalian target of rapamycin (mTOR) in a pathway that has not been fully defined (5). Rapamycin potently inhibits mTOR function in cells and blocks the activation of p70 S6K by a variety of hormones and growth factors (5).
MAP kinase rapidly phosphorylates Ser 64 in PHAS-I (18) and abolishes binding of PHAS-I to eIF-4E in vitro (11). Although Ser 64 is phosphorylated when adipocytes are incubated with insulin (18), MAP kinase does not appear to mediate the action of insulin on PHAS-I, as blocking MAP kinase activation by using an inhibitor of MEK activation (19), PD 098059, did not block the effects of insulin on phosphorylation of PHAS-I in either adipocytes (11) or skeletal muscle (15). Rapamycin attenuates the actions of insulin and growth factors on increasing the phosphorylation of PHAS-I (15, 17, 20 -22).
Major questions exist regarding the effect of cAMP on the phosphorylation of PHAS-I and p70 S6K . Increasing cAMP has been reported to increase PHAS-I phosphorylation in rat adipocytes (23). In contrast, increasing cAMP with forskolin in rat aortic smooth muscle cells increased the electrophoretic mobility of PHAS-I, supporting the conclusion that cAMP promotes dephosphorylation of PHAS-I (21). Forskolin also inhibited activation of p70 S6K in smooth muscle cells (21), and incubating lymphoid cells with the combination of forskolin and IBMX inhibited the activation of p70 S6K by interleukin 2 (24). However, in Swiss 3T3 cells, neither forskolin nor 8-bromo-cAMP decreased serum-stimulated p70 S6K activity, and the inhibition of p70 S6K by methylxanthine phosphodiesterase inhibitors was reported to occur without significant increases in cAMP (25)(26)(27). Based on these findings, the interpretation that increased cAMP inhibits activation of p70 S6K has been questioned (25)(26)(27).
Control of the phosphorylation of PHAS-II (4EBP-2), the predicted amino acid sequence of which is approximately 60% identical to that of PHAS-I (8, 12), has not been characterized. The amino acid sequences of PHAS-I and -II differ in ways that suggest that the two proteins are regulated differently (9). For example, PHAS-II contains a consensus site for phosphorylation by PKA (28), which does not phosphorylate PHAS-I (8,18,29).
In view of the potential importance that phosphorylation of PHAS proteins and p70 S6K have for regulating protein synthesis and/or mitogenesis, it is important to resolve issues relating to their control by cAMP. The present experiments were performed to compare the regulation, phosphorylation, and pattern of expression of PHAS-I and -II and to investigate the role of the p70 S6K pathway in the actions of hormones and agents that act by increasing cAMP in 3T3-L1 adipocytes.

EXPERIMENTAL PROCEDURES
Incubation of 3T3-L1 Adipocytes-Conditions for culturing 3T3-L1 adipocytes, for incubating the cells with 32 P i , and for preparing extracts were as described previously (17). To measure [ 35 S]methionine incorporation into protein, adipocytes were incubated at 37°C in Dulbecco's modified Eagle's medium for 3 h before the medium was replaced with methionine-free Dulbecco's modified Eagle's medium containing insulin and other additions as indicated. The cells were incubated for 1 h before the medium was replaced with medium containing the same additions plus 10 M [ 35 S]methionine (20 Ci/ml). After 1 h the cells were rinsed twice with phosphate-buffered saline (145 mM NaCl, 5 mM KCl, and 10 mM sodium phosphate, pH 7.4), and extracts were prepared (17). Samples (40 g of protein) were subjected to SDS-PAGE, and after drying the gel the total amounts of 35 S in proteins having apparent M r values between approximately 20,000 and 200,000 were determined by phosphorimaging.
Antibodies-Antibodies to PHAS-I, PHAS-II, and eIF-4E were generated by immunizing rabbits with hemocyanin conjugated to peptides (CSSPEDKRAGGEESQFE, CDRKHAVGDDAQFE, and CTATKSGST-TKNRFV, respectively) corresponding to stretches of amino acids in the COOH-terminal regions of the respective proteins. Peptide conjugation, immunization of rabbits, and purification of antibodies by using peptide affinity columns were performed as described previously (17). PHAS-II antiserum (R693) was generated by immunizing rabbits with recombinant PHAS-II as described previously for preparation of PHAS-I antibodies (17). A peptide identical to that described previously (30), except containing an NH 2 -terminal cysteine residue (to facilitate coupling to hemocyanin), was used to generate antibodies to p70 S6K .
Immunoprecipitations and Electrophoretic Analyses of PHAS Proteins-Immunoprecipitations were performed as described previously (17). Samples of extracts and immunoprecipitated proteins were subjected to SDS-PAGE (31) before proteins were electrophoretically transferred to polyvinylidene difluoride membranes (Immobilon-P, Millipore). For immunoblotting membranes were incubated with antibodies (1 g/ml for affinity-purified antibodies or 500-fold dilutions of sera) and washed as described previously (17). Antibody binding was detected by enhanced chemiluminescence using alkaline phosphatase conjugated to goat anti-rabbit IgG. Far Western analyses to evaluate eIF-4E binding to PHAS proteins were performed as described previously (12) by incubating blots with 32 P-labeled FLAG-4E, a fusion protein containing glutathionine S-transferase, a FLAG epitope, a PKA phosphorylation site, and eIF-4E.
Cloning of 3T3L1 PHAS-II cDNA-A 264-base pair fragment encoding a portion of mouse PHAS-II was generated by using the polymerase chain reaction with reverse-transcribed fetal mouse cDNA as template and degenerate primers (forward, GACTACAGCACCACNCC; reverse, AATGTCCATCTCAAARTC) based on the nucleotide sequence of human 4EBP-2 cDNA (12). The polymerase chain reaction product was inserted into pCRII by TA cloning (Invitrogen). The fragment containing PHAS-II cDNA was excised with EcoRI, purified, and then 32 Plabeled by using [␣-32 P]CTP and Klenow with GACTACAGGACCAC-NCC as the primer. The 32 P-labeled fragment was used to probe a 3T3-L1 adipocyte cDNA library (in ZapII, provided by Dr. Fred Fiedorek, University of North Carolina). Hybridization and wash conditions were as described previously (8). All clones isolated (12 of approximately 10 6 plaques) were cDNAs that had sequences identical to regions of the PHAS-II sequence, which was derived separately from two independent clones. 2 The coding region is found in a 441-base pair open reading frame. The predicted amino acid sequence is 95% identical to that of the human PHAS-II counterpart 4EBP-2 (12).
Preparation of Recombinant PHAS Proteins and PHAS⅐eIF-4E Complexes-cDNA containing the coding region of mouse PHAS-II was generated by polymerase chain reaction and inserted between the NcoI and BamHI sites of pET-14b (Novagen). The entire insert was sequenced to ensure that no errors had been introduced. To express the protein, Escherichia coli [BL21(DE 3 ) strain containing the plasmid pLysS; Novagen] was transformed with pET-PHAS-II. Conditions for growth and induction were as described previously for recombinant PHAS-I (18). PHAS-I and -II were purified by taking advantage of their high affinity for eIF-4E. Complexes were generated by allowing eIF-4E to renature in the presence of excess PHAS proteins. This was accomplished by slight modification of the eIF-4E purification method (32) as described previously for the purification of [His 6 ]PHAS-I⅐eIF-4E complexes (17).

Phosphorylation of PHAS-I and -II in Vitro-Recombinant
PHAS proteins (2 g) were incubated at 30°C in a solution (20 l) containing 10 mM MgCl 2 , 100 M [␥-32 P]ATP, 50 mM ␤-glycerol phosphate (pH 7.0), and either the catalytic subunit of bovine heart PKA or recombinant MAP kinase as described previously (17). After the appropriate time of incubation, the reactions were terminated by adding 20 l of SDS sample buffer. Phosphate incorporated into PHAS-I and -II was determined from the 32 P contents of the respective proteins after samples (20 l) were subjected to SDS-PAGE.
Measurements of Protein Kinase Activities in Cell Extracts-p70 S6K activity was measured in an immune complex assay using intact 40 S ribosomes as substrate as described previously (30). The activities of the ERK-1 and ERK-2 isoforms of MAP kinase activity were measured using [␥-32 P]ATP and myelin basic protein as substrates in the gel renaturation method described by Wang and Erikson (33).
Other Materials-PD 098059 was provided from Parke-Davis. SQ20006 and SQ20009 were supplied by Bristol Myers-Squibb. Epidermal growth factor was purchased from Boehringer Mannheim.
[␥-32 P]ATP and Na 32 P i were from DuPont NEN. Rapamycin was obtained from Calbiochem. Cyclic nucleotide derivates, theophylline, IBMX, forskolin, wortmannin, phorbol 12-myristate 13-acetate, and most commonly used chemicals were from Sigma. Oligonucleotides and peptides were provided by the Washington University Diabetes Center.

PHAS-I and -II-PHAS-I in tissue extracts migrates in SDS-
PAGE as multiple species having apparent M r values (20,000 -24,000) much larger than its actual M r of approximately 12,500 (8) (Fig. 1). PHAS-II exhibited higher electrophoretic mobility than PHAS-I, more closely approximating that of its predicted M r of 12,900. 2 The relative amounts of PHAS-I and PHAS-II present in tissues differed significantly (Fig. 1). Liver and kidney, which contained relatively low levels of PHAS-I, were found to express the highest levels of PHAS-II. On the other hand, adipose tissue, which contained the highest levels of PHAS-I, contained relatively little PHAS-II.
Relative Rates of Phosphorylation of PHAS Proteins In Vitro-To compare the phosphorylation of PHAS-I and -II, recombinant proteins were incubated with [␥-32 P]ATP and the ERK-2 isoform of MAP kinase. The initial rate of phosphorylation of PHAS-II was lower than that of PHAS-I, although with FIG. 1. Tissue distribution of PHAS-I and -II. Isolated rat adipocytes (W. Adipocyte) and the following rat tissues were homogenized: diaphragm (Sk. Muscle), heart, lung, liver, kidney, thymus, spleen, small intestine (S. Intestine), testis, epididymal adipose tissue (W. Adipose), and intrascapular brown adipose (B. Adipose). Homogenates were centrifuged at 10,000 ϫ g for 30 min. Extract protein concentrations were adjusted to 4 mg/ml before samples were incubated at 100°C for 10 min. After centrifugation at 10,000 ϫ g for 30 min, extract samples (50 l) were subjected to SDS-PAGE, and immunoblots were prepared with the peptide antibodies to either PHAS-I or PHAS-II. The positions of the standard proteins, soy bean trypsin inhibitor (SBTI) and cytochrome C (Cyto C), are indicated. times of incubation greater than 30 min, more phosphate was incorporated into PHAS-II than PHAS-I ( Fig. 2A). That MAP kinase would be found to phosphorylate PHAS-II was not unexpected, as the equivalent to Ser 64 , the major site in PHAS-I phosphorylated by MAP kinase (11,18), is found at position 65 in PHAS-II, and the amino acid sequence, -Arg-Asn-Ser-Pro-, is conserved in the two proteins. 2 However, a Cys is found in the N-3 position from the Ser in PHAS-I, and Arg is present in the corresponding position in PHAS-II. 2 Thus, Ser 65 in PHAS-II represents an ideal consensus site (Arg-Arg-Xaa-Ser/Thr-) for phosphorylation by PKA (28). Incubating PHAS-II with the catalytic subunit of PKA resulted in the rapid phosphorylation of PHAS-II to a level that approached a stoichiometry of 0.9 mol of phosphate/mol after 1 h (Fig. 2B). Under the same conditions, less than 0.01 mol/mol was incorporated into PHAS-I. These findings suggested that increasing cAMP might selectively increase PHAS-II phosphorylation.
Phosphorylation of PHAS-I and -II in 3T3-L1 Adipocytes-3T3-L1 adipocytes were found to express both PHAS-I and -II, and these cells were used to investigate the effects of cAMP on the phosphorylation of the two proteins. Adipocytes were incubated with 32 P i and then treated with forskolin or isoproterenol, two agents that increase cAMP by activating adenylate cyclase. 32 P-Labeled PHAS proteins were immunoprecipitated and subjected to SDS-PAGE. As previously noted (17), two 32 P-labeled bands corresponding to the ␤and ␥-phosphorylated forms of PHAS-I were detected by autoradiography (Fig.  3). PHAS-II appeared predominantly as a single 32 P-labeled band. Treating cells with insulin, epidermal growth factor, or phorbol 12-myristate 13-acetate markedly increased the 32 P contents of both PHAS-I and PHAS-II (Fig. 3). The effects of insulin on the phosphorylation of PHAS-I and -II were abolished by wortmannin, an inhibitor of phosphatidylinositol 3Јhydroxykinase (34). Forskolin and isoproterenol decreased the 32 P content of both PHAS-I and -II and attenuated the effects of insulin on increasing the phosphorylation of the two proteins (Fig. 3).
To investigate further the regulation of PHAS-I by agents that increase intracellular cAMP, PHAS-I immunoblots were prepared. Forskolin and isoproterenol increased the relative proportion of PHAS-I found in the nonphosphorylated ␣ form (Fig. 4A). The half-maximum effect of isoproterenol on decreasing the phosphorylation of PHAS-I was observed at a concentration of approximately 50 nM, which is similar to the concen-tration required to produce a half-maximum effect on increasing cAMP in these cells (Fig. 5). The cAMP derivatives N 6 ,2Ј-O-dibutyryl-cAMP, CPT-cAMP, and 8-bromo-cAMP also increased the proportion of PHAS-I in the ␣ form, providing additional evidence that cAMP promotes dephosphorylation of PHAS-I (Fig. 4A).
The effects of agents on the ability of PHAS-I to bind to eIF-4E was investigated by determining the amount of PHAS-I that copurified with eIF-4E when the initiation factor was  ). Extracts were prepared, and PHAS-I⅐eIF-4E complexes were partially purified by affinity chromatography using m 7 GTP-Sepharose. Samples were subjected to SDS-PAGE and transferred to nylon membranes. Immunoblots of the extract samples (A) or the partially purified PHAS-I⅐eIF-4E complexes (B) were prepared using PHAS-I antibodies. To assess the ability of the adipocyte PHAS-I to bind exogenous eIF-4E, membranes were incubated with 32 P-labeled FLAG-4E. An autoradiogram was prepared after washing to remove the unbound ligand (C). partially purified by affinity chromatography with m 7 GTP-Sepharose. PHAS-I ␣ was the predominant form of PHAS-I recovered under these conditions (Fig. 4B), consistent with the interpretation that the nonphosphorylated form of the protein binds most tightly to eIF-4E. As previously observed (11,12) increasing the phosphorylation of PHAS-I with insulin markedly decreased the amount of PHAS-I that copurified with eIF-4E, indicative of dissociation of the PHAS-I⅐eIF-4E complex. In contrast, incubating cells with cAMP derivatives or with forskolin or isoproterenol markedly increased the amount of PHAS-I recovered, indicating that increased cAMP results in an increase in the amount of PHAS-I bound to eIF-4E. These agents also markedly increased binding of PHAS-I to an exogenous 32 P-labeled FLAG-4E probe (Fig. 4C), supporting the conclusion that it is the covalent modification of PHAS-I that accounts for the increase in the PHAS-I⅐eIF-4E complex.
Methylxanthines are a commonly used class of agents that increase intracellular cAMP by inhibiting phosphodiesterase activity (35). Effects of incubating 3T3-L1 adipocytes with four such agents are shown in Fig. 6. SQ20006, SQ20009, theophylline, and IBMX increased the electrophoretic mobility of PHAS-I (Fig. 6A), increased the amount of PHAS-I bound to endogenous eIF-4E (Fig. 6B), and increased the affinity of PHAS-I for exogenous eIF-4E (Fig. 6C). The effects of the agents were observed in both the absence and presence of insulin.
The electrophoretic mobility of PHAS-II was not detectably changed in response to insulin or agents that increase cAMP. 3 Thus, measuring mobility shifts by immunoblotting was not useful for assessing changes in the phosphorylation of PHAS-II. Also, because of the relatively low levels of PHAS-II, it was not practical to use the m 7 GTP-Sepharose method to detect PHAS-II⅐eIF-4E complexes. For these reasons far Western analyses proved most useful for assessing changes in PHAS-II that affect its binding to eIF-4E. To increase the signal intensity, experiments were performed in which PHAS-II was concentrated by immunoprecipitation prior to SDS-PAGE (Fig. 7). Insulin decreased 32 P-labeled FLAG-4E binding. In the presence of the hormone, binding of the probe to PHAS-II was increased by rapamycin (Fig. 7) and wortmannin. The effects of insulin were also attenuated by isoproterenol and forskolin, by the phosphodiesterase inhibitors, theophylline and IBMX, and by the cAMP-derivatives, CPT-cAMP, N 6 ,2Ј-O-dibutyryl-cAMP, and 8-Br cAMP. In contrast, inhibiting MAP kinase activation with PD 098059 did not attenuate the effect of insulin on decreasing binding of PHAS-II to 32 P-labeled FLAG-4E (Fig. 7). 4 cAMP-dependent and -independent Control of p70S6K-To investigate the role of the p70S6K pathway in mediating the effects of forskolin and cAMP derivatives on PHAS-I, the effect of the agents on the electrophoretic mobility of p70 S6K (Fig. 8A) and on the activity of the kinase (Fig. 8B) were determined. Insulin promoted a shift to lower mobility species (Fig. 8A), which represent more highly phosphorylated forms of p70 S6K (36). Insulin-stimulated phosphorylation was associated with a dramatic increase in p70 S6K activity, as indicated by the increased incorporation of 32 P into ribosomal protein S6 when 3 T.-A. Lin and J. C. Lawrence, Jr., unpublished observations. 4 In three experiments binding of 32 P-labeled FLAG-4E to PHAS-II was quantitated by phosphorimaging. Results (expressed as percentages of control) were: PD 098059 alone, 105 Ϯ 20%; insulin alone, 16 Ϯ 8%; and insulin plus PD 098059, 21 Ϯ 6%.  7. Binding of 32 P-labeled FLAG-4E to PHAS-II immunoprecipitated from 3T3-L1 adipocytes. Duplicate cultures were incubated at 37°C for 40 min with 10 nM rapamycin, 50 M PD 098059, 25 M forskolin, 1 mM SQ20006, 0.5 mM CPT-cAMP, and 1 M isoproterenol before extracts were prepared. One-half of the cells were incubated with 20 nM insulin for the last 10 min. PHAS-II was immunoprecipitated from extracts (300 l) by using R693 and subjected to SDS-PAGE. After transferring protein to membranes, 32 P-labeled FLAG-4E binding was measured. An autoradiogram is presented.
immune complexes containing p70 S6K were incubated with 40 S ribosomes and [␥-32 P]ATP. Forskolin, isoproterenol, and all three cyclic nucleotide derivatives tested markedly attenuated the effects of insulin on p70 S6K electrophoretic mobility (Fig.  8A) and on p70 S6K activity (Fig. 8B). No effects of the agents were apparent in the absence of insulin, possibly because of the very low basal activity of p70 S6K (Fig. 8B). Methylxanthines also increased the electrophoretic mobility of p70 S6K (Fig. 9A) and decreased kinase activity (Fig. 9B) Effects of Insulin and cAMP on Protein Synthesis-To determine whether the effects of insulin and agents that increase cAMP on the phosphorylation of p70 S6K and PHAS-I and -II were correlated with effects on protein synthesis, the effect of the agents on [ 35 S]methionine incorporation into protein were investigated. Insulin increased synthesis of 35 S-labeled protein by approximately 2-fold (Fig. 10). Accumulation of 35 S-labeled protein was markedly inhibited by forskolin, isoproterenol, CPT-cAMP, and theophylline, both in the presence and absence of insulin. Of the different agents, forskolin and theophylline were most effective, particularly in the presence of insulin. Rapamycin also decreased protein synthesis, but its effect was more pronounced in cells incubated with insulin. DISCUSSION PHAS-I and -II were found in a wide variety of tissues, consistent with the important role that these proteins appear to have in regulating mRNA translation initiation. The relative contribution of PHAS-I and -II to the control of protein synthesis presumably depends on the cell type. PHAS-I is almost certainly more important in adipocytes, in which little PHAS-II was found (Fig. 1), but PHAS-II is likely to have a significant role in liver and kidney, which express relatively high levels of PHAS-II. 3T3-L1 adipocytes expressed both PHAS-I and -II and proved useful for comparing the regulation of the two proteins. Important conclusions are that insulin stimulates the phosphorylation of both PHAS-I and -II and that increasing cAMP promotes dephosphorylation of both PHAS proteins, thereby increasing their binding to eIF-4E. The predicted consequences of these effects are consistent with the stimulation and inhibition of protein synthesis observed with insulin and increased cAMP, respectively (Fig. 10).
Based on the extent of homology, both similarities and differences in the regulation of PHAS-I and -II might be expected. The Ser that is the site in PHAS-I most rapidly phosphorylated by MAP kinase in vitro (11,18) and that is a major site of insulin-stimulated phosphorylation in cells (18) is conserved in PHAS-II. MAP kinase readily phosphorylated recombinant PHAS-II (Fig. 2); however, inhibiting MAP kinase activation by using PD 098059 did not attenuate the effect of insulin on decreasing the ability of PHAS-II to bind 32 P-labeled FLAG-4E (Fig. 7). 4 Thus, as concluded previously with respect to the regulation of PHAS-I (17), MAP kinase does not appear to be the major mediator of the effects of insulin on increasing phosphorylation of PHAS-II.
Even though PHAS-II was rapidly phosphorylated by PKA in vitro, agents known to activate PKA in cells not only decreased the 32 P content of PHAS-II in 32 P-labeled adipocytes (Fig. 3) but also increased binding of PHAS-II to the FLAG-4E probe (Figs. 4, 6, and 7). Likewise, PHAS-I was dephosphorylated, and the amount of the protein bound to eIF-4E was increased by agonists that increase cAMP (forskolin and isoproterenol; Figs. 4 and 5), cAMP derivatives (CPT-cAMP, N 6 ,2Ј-O-dibutyryl-cAMP, and 8-Br cAMP; Fig. 4), and phosphodiesterase inhibitors (isobutylmethylxanthine, theophylline, and SQ 20009; Fig. 6). Thus, PHAS-I and -II were regulated in a similar manner in response to increased cAMP in 3T3-L1 adipocytes.
Increasing cAMP in 32 P-labeled adipocytes has been reported to increase phosphorylation of a M r ϳ22,000 protein (29) concluded to be PHAS-I (23). Other investigators have not observed increased 32 P-labeling of the M r 22,000 protein in response to increased cAMP in either 3T3-L1 (37) or rat adipocytes (9, 38, 39). The reason for this discrepancy is not FIG. 8. Effects of forskolin, isoproterenol, and cAMP derivatives on p70 S6K activity in 3T3-L1 adipocytes. Cells were incubated as described in the legend to Fig. 4. Samples of the extracts were subjected to SDS-PAGE, and an immunoblot was prepared using antibodies to p70 S6K (A). B, to assess p70 S6K activity, the kinase was immunoprecipitated from extracts, and the immune complexes were incubated with [␥-32 P]ATP and 40 S ribosomal subunits. After 20 min at 30°C, samples were subjected to SDS-PAGE, and an autoradiogram of the dried gel was prepared. The amount of 32 P incorporated into ribosomal protein S6 provides the index of p70 S6K activity.
FIG. 9. Effect of cAMP phosphodiesterase inhibitors on p70 S6K in 3T3-L1 adipocytes. Cells were incubated as described in the legend to Fig. 6. Immunoblots were prepared using p70 S6K antibodies (A). p70 S6K was immunoprecipitated, and ribosomal protein S6 kinase activity was measured in an immune complex assay (B). clear; however, as multiple phosphorylated heat-and acidstable proteins having this molecular weight are present in adipocytes (9), the possibility that elevating cAMP increases the phosphorylation of an M r ϳ 22,000 protein that is not PHAS-I might be considered. In a recent study isoproterenol markedly increased 32 P labeling of the M r ϳ22,000 protein without decreasing the electrophoretic mobility of PHAS-I measured in the same extract by immunoblotting (23). This observation is not supportive of phosphorylation of PHAS-I, the mobility of which is markedly decreased by phosphorylation of the appropriate sites.
Results obtained thus far are consistent with the interpretation that elements of the mTOR signaling pathway mediate the actions of insulin on increasing the phosphorylation of both PHAS-I and PHAS-II. The effects of insulin on both PHAS proteins were attenuated by rapamycin ( Fig. 7 and Ref. 17). Rapamycin acts via an intracellular receptor, FKBP12, and it is the rapamycin-FKBP12 complex that binds to mTOR (5). The immunosuppressant FK506 competes with rapamycin for binding to FKBP12, but the FK506⅐FKBP12 complex does not inhibit mTOR signaling (5). We previously demonstrated that FK506 abolished the effects of rapamycin on PHAS-I (22). The phosphorylation of PHAS-I and -II was also increased by phorbol 12-myristate 13-acetate and epidermal growth factor (Fig.  3), which activate p70 S6K , a downstream element of the mTOR pathway (5).
Increasing cAMP with forskolin was previously shown to be associated with inactivation of p70 S6K in aortic smooth muscle cells (21,40) and lymphoid cells (24). In Swiss 3T3 cells methylxanthines, such as SQ20006 and theophylline, produced marked inhibition of p70 S6K , whereas the kinase was not inhibited by the agonist that increased cAMP or by cAMP derivatives (25)(26)(27). It is possible that an action unrelated to inhibition of cAMP phosphodiesterase also contributed to the effects of the phosphodiesterase inhibitors on p70 S6K activity in 3T3-L1 adipocytes. However, the findings that insulin-stimulated p70 S6K activity was inhibited by agonists that increased cAMP as well as by cAMP derivatives (Fig. 8) provide strong evidence that increasing cAMP attenuates the activation of p70 S6K in 3T3-L1 adipocytes.
As p70 S6K phosphorylated neither PHAS-I nor PHAS-II, 3 other elements in the mTOR pathway must be involved in phosphorylating the PHAS proteins. Interestingly, mTOR itself has limited homology with the catalytic subunit of DNA-dependent protein kinase (41). mTOR has the potential to act as a protein kinase, as it may undergo autophosphorylation on Ser (42), and it is interesting to speculate that mTOR is a PHAS kinase. However, it should be stressed that phosphorylation of an exogenous substrate by mTOR still has not been described. It might be noted that wortmannin would be expected to inhibit mTOR protein kinase, as it has been recently shown to bind mTOR and inhibit the autophosphorylation reaction (43). An implication is that the effect of wortmannin on decreasing phosphorylation of PHAS-I and -II (Fig. 3) might be due to inhibition of mTOR instead of inhibition of phosphatidylinositol 3Ј-hydroxykinase. Additional experiments are needed to investigate the possibilities that mTOR functions as a PHAS kinase and as a site of action of methylxanthines and agents that increase cAMP.