Identification of the Insulin-Regulated Interaction of Phosphodiesterase 3B With 14-3-3 β Protein

The cDNA encoding the regulatory region of mouse PDE3B (amino acid [aa] sequence between 223 and 636) (mPDE3B[aa 223–636]) was ligated, in frame, to the DNA-binding domain of GAL4 in a pGBT9 vector. The yeast strain PJ69-4A, which has HIS3, ADE2, and lacZ reporters, was transformed with mPDE3B (aa 223–636) in pGBT9 using the LiAc method (13). As the first screen, the PJ69-4A strain transformed with mPDE3B (aa 223–636) in the pGBT9 plasmid was used to screen the 3T3L1 adipocyte cDNA library in pGAD GH. Approximately 3.2 × 10⁶ primary transformants were obtained from the 3T3L1 adipocyte cDNA library. These transformants were first selected on SD-(−)His(−)Trp(−)Leu plates containing 20 mmol/l 3-aminotriazole at 30°C for 7 days, and then the positive clones were selected on SD-(−)Ade(−)Trp(−)Leu plates using the ADE2 reporter of the PJ69-4A yeast strain to confirm the specificity of the positive clones. As another confirmation of protein-protein interactions involving mPDE3B, the lacZ reporter of the PJ69-4A yeast strain was examined in His⁺ Ade⁺ Trp⁺ Leu⁺ clones by measuring β-galactosidase (β-gal) activity using a filter assay. Plasmid DNA from positive yeast clones, which were both His⁺ Ade⁺ Trp⁺ Leu⁺ and β-gal activity positive, was transformed into the Escherichia coli (HB101) by electroporation. Positive plasmid clones derived from the 3T3L1 adipocyte cDNA library were isolated. As the second screen, PJ69-4A was transformed simultaneously with each isolated positive plasmid and mPDE3B (aa 223–636) in pGBT9. His⁺ Ade⁺ Trp⁺ Leu⁺ and β-gal activity positive clones were selected again, as described above, except that β-gal activity was measured using a liquid assay. The DNA sequence of each clone was determined using an ABI 310 instrument, and nucleotide sequences from each positive clone were compared with those entered in the GenBank database using the BLAST sequence search and comparison program.

The cDNA encoding FLAG-tagged full-length or truncated forms of mouse PDE3B was ligated, in frame, to pET 3a vector. The FLAG-tagged full-length or truncated forms of mouse PDE3B, represented as mPDE3B (aa 1–322), mPDE3B (aa 223–636), mPDE3B (aa 621–1,100), or mPDE3B (full-length), were labeled with [³⁵S]methionine (Amersham Pharmacia, Buckinghamshire, U.K.) using the T7 TNT Quick-coupled Transcription/Translation System (Promega, Madison, WI). The BL21 bacterial strain transformed with pGEX 3X-14-3-3 β was grown in LB media, and expression of the chimeric proteins was induced with 0.4 mmol/l isopropyl-β-thiogalactoside at 37°C for 4 h. Chimeric proteins were affinity-purified using glutathione resin (Novagen, Darmstadt, Germany) according to the manufacturer’s protocol. The radiolabeled PDE3B proteins and either glutathione S-transferase (GST)·14-3-3 β- or GST-glutathione resin complex was resuspended in interaction buffer (10 mmol/l Tris-HCl [pH 8.0], 50 mmol/l NaCl, 5 mmol/l EDTA, 0.1% Triton X-100, 1 mmol/l phenylmethylsulfonyl fluoride (PMSF), 1 mmol/l DTT) and incubated for 30 min at room temperature with continuous gentle mixing. A sample representing 10% of the input-labeled proteins and the entire pull-down complexes were subjected to SDS-PAGE. The gel was dried, and ³⁵S-labeled proteins were visualized using a BAS 3000 (Fuji, Tokyo, Japan) image analyzer.

Isolated adipocytes were prepared from epididymal fat tissue of Sprague-Dawley rats using the collagenase method. The cells were incubated with or without insulin (3 nmol/l) in Krebs Ringer Albumin HEPES buffer at 37°C for 10 min. In some experiments, adipocytes were preincubated with okadaic acid (1 μmol/l) or wortmannin (100 nmol/l) at 37°C for 20 or 30 min, respectively. The cells were then homogenized in buffer A (50 mmol/l Tris HCl [pH 7.5], 1 mmol/l EDTA, 1 mmol/l EGTA, 50 mmol/l NaF, 5 mmol/l sodium pyrophosphate, 10 mmol/l sodium β-glycerophosphate, 0.1 mmol/l PMSF, 1 μg/ml aprotinin, 1 μg/ml pepstatin, 1 μg/ml leupeptin, 1 μmol/l microcystin, 0.5 mmol/l sodium orthovanadate) in a Dounce tissue grinder. After brief centrifugation of the homogenates, the supernatant without the fat cake was centrifuged (100,000g) for 1 h at 4°C, and the membrane (P-2) and cytosol fractions were obtained. P-2 was then solubilized with 1% triton X-100 in buffer A and centrifuged (100,000g) for 1 h at 4°C, and the supernatant (solubilized P-2) was obtained.

Solubilized P-2 (200 μg protein) from rat adipocytes was incubated with GST glutathione resin or GST·14-3-3 β glutathione resin at 4°C overnight with continuous mixing. The resin-bound complexes were collected by centrifugation, subjected to SDS-PAGE, and then electrotransferred to a polyvinylidene fluoride membrane. Immunoblotting was performed using an anti-PDE3B antibody raised against a synthetic rat PDE3B peptide (6). The amount of the added GST fusion proteins was assessed by amido black staining of the electrotransferred membrane.

Solubilized P-2 (200 μg protein) obtained from rat adipocytes incubated with or without 3 nmol/l insulin at 37°C for 10 min was incubated with an anti-14-3-3 β antibody (Santa Cruz Biotechnology, Santa Cruz, CA) at 4°C overnight, followed by the addition of protein G-agarose for last 2 h. The agarose-bound immune complex was collected by centrifugation, and washed twice with each of 1% Triton X-100 in 50 mmol/l HEPES pH 7.5 and 0.1% Triton X-100 in 50 mmol/l HEPES at pH 7.5. The complexes were then subjected to SDS-PAGE. The amount of PDE3B in the immune complex was quantified by immunoblotting using an anti-PDE3B antibody. In the experiment of peptide inhibition, solubilized P-2 was incubated with the synthesized peptide (1 mmol/l) at 4°C overnight before immunoprecipitation. For peptide inhibition experiments, Ser-phosphorylated (pS279, pS302, and pS427) or nonphosphorylated (S279, S302, and S427) peptides, each of which consists of 15 amino acids corresponding to the sequence around Ser-279, Ser-302, and Ser-427 in rat PDE3B, respectively, were synthesized (Fujiya, Tokyo, Japan). The peptide sequences were PVIRPRRR[pS]SCVSLG(pS279), SGKMFRRP[pS]LPCISR(pS302), and TAQLRRS[pS]GASGLLT(pS427), and their nonphosphorylated serine-containing counterparts (S279, S302, and S427, respectively) had the same surrounding sequences.

PDE activity was assayed as described previously (14). Briefly, samples were incubated with 0.1 μmol/l [³H]cAMP (Amersham Pharmacia) for 5 min at 30°C, and the formed [³H]AMP was degraded by 5′-nucleotidase of snake venom to [³H]adenosine. The latter was isolated by chromatography on AG 1-X2 (Bio-Rad, Hercules, CA), and the radioactivity was measured by liquid scintillation counter.

We used the yeast two-hybrid system for the detection of PDE3B interacting proteins. A 3T3L1 adipocyte cDNA library (15) was screened using mouse PDE3B as the bait. The cDNA encoding the regulatory region of mouse PDE3B (aa sequence between 223 and 636) (mPDE3B [aa 223–636]) was ligated, in frame, to the DNA-binding domain of GAL4 in a pGBT9 vector. Approximately 3.2 × 10⁶ independent clones were screened to obtain 40 positive clones, 5 of which were identified as 14-3-3 isoforms by DNA sequencing after a second screening. One of these corresponded to isoform 14-3-3 β and was selected for further study.

The GST pull-down assay was used to confirm an in vitro interaction between mouse PDE3B and 14-3-3 β. The FLAG-tagged full-length or truncated forms of mouse PDE3B were ³⁵S-labeled using an in vitro transcription/translation system. Each of the labeled PDE3B proteins was incubated with glutathione-resin bound GST·14-3-3 β. GST·14-3-3 β interacted with mPDE3B (aa 1–322), mPDE3B (aa 223–636), and mPDE3B (full-length) but not with mPDE3B (aa 621–1,100). GST alone did not interact with any of these proteins (Fig. 1). The region of mouse PDE3B that interacts with 14-3-3 β is thus located between aa 223 and 322. It is interesting that this region of mouse PDE3B contains the two reported insulin-dependent phosphorylation sites, namely mouse Ser-273 (rat Ser-279) (8) and rat Ser-302 (7).

Cilostamide, an inhibitor of PDE3, was used to confirm that the PDE activity in the solubilized membrane fraction of rat adipocytes (solubilized P-2) reflects PDE3B activity. PDE activity was increased 2.5-fold compared with the basal activity by 3 nmol/l insulin (Fig. 2A). Approximately 90% of both basal and insulin-induced PDE activity was inhibited by 1 μmol/l cilostamide (Fig. 2A, hatched bars). Because PDE3B is known to be a major isoform of PDE3 in adipocytes, the PDE activity measured primarily reflects PDE3B activity.

GST·14-3-3 β was used as the bait to examine whether this chimeric protein can interact with endogenous PDE3B in rat adipocytes and to determine whether this interaction is insulin-inducible. Glutathione resin-bound GST·14-3-3 β was incubated with the solubilized P-2 from rat adipocytes treated with or without 3 nmol/l insulin. This resin-bound complex was then subjected to immunoblot analysis with anti-PDE3B antibody raised against a synthetic rat PDE3B peptide (6). GST·14-3-3 β but not GST alone was associated with endogenous PDE3B (Fig. 2B, top). This association was increased in response to insulin. Accordingly, the PDE3B activity in the GST·14-3-3 β resin-bound complex was increased in response to insulin (Fig. 2B, bottom). The amount of PDE3B protein in the solubilized P-2 before GST pull-down was the same in the presence or absence of insulin (as shown in Fig. 2A, top), suggesting that the insulin-stimulated form of endogenous PDE3B associates more effectively with 14-3-3 β than the enzyme from untreated cells.

We next examined whether endogenous PDE3B associates with endogenous 14-3-3 β in vivo, in rat adipocytes. The solubilized P-2 of rat adipocytes, treated with or without insulin, was immunoprecipitated with the anti-14-3-3 β antibody. The protein G-agarose-bound immune complex was then subjected to immunoblot analysis with anti-PDE3B antibody. The PDE3B protein was only weakly associated with 14-3-3 β in the basal state, but insulin markedly increased this association (Fig. 2C, top). PDE3B activity in the agarose-bound immune complex was increased approximately fourfold by insulin (Fig. 2C, bottom). The reverse experiment was performed to confirm further this association. The solubilized P-2 was first immunoprecipitated with the anti-PDE3B antibody, and the agarose-bound immune complex was then immunoblotted with the anti-14-3-3 β antibody. An insulin-induced increase of the association of the two proteins was consistently detected (data not shown). These in vivo results indicate that PDE3B associates with 14-3-3 β and that this interaction is enhanced by insulin in rat adipocytes.

Insulin promotes serine phosphorylation and the activation of PDE3B in adipocytes (16). Members of the 14-3-3 protein family are known to bind to a specific phosphoserine-containing motif in various proteins (10, 17). We thus examined whether the insulin-induced association of endogenous PDE3B with endogenous 14-3-3 β is phosphorylation-dependent. Treatment of adipocytes with 1 μmol/l okadaic acid, a potent, cell-permeable Ser/Thr phosphatase inhibitor, increased the association of PDE3B with 14-3-3 β to the same level as that achieved with insulin (Fig. 3A, top). Pretreatment with 1 μmol/l okadaic acid did not further enhance the effect of insulin on this association. Likewise, PDE3B activity in the solubilized P-2 was increased to the same extent by okadaic acid and insulin (Fig. 3A, bottom). Okadaic acid and insulin are reported to exert similar effects on PDE3B phosphorylation (18). In addition, when the solubilized P-2 was prepared in buffer lacking phosphatase inhibitors, both the protein-protein association and insulin-induced PDE3B activation were abolished (data not shown). That the PDE3B synthesized by the reticulocyte lysate system interacts with 14-3-3 β in the absence of insulin (Fig. 1) suggests that this protein is phosphorylated. Indeed, certain proteins synthesized in the system are phosphorylated (19). In general, there seems to be a positive correlation between insulin-induced phosphorylation of PDE3B, activation of the enzyme, and the association of 14-3-3 β with the enzyme. Isoproterenol, another inducer of PDE3B phosphorylation and activation, also promotes the interaction between PDE3B and 14-3-3β (data not shown).

We have previously reported that 100 nmol/l wortmannin, a relatively specific inhibitor of PI3-K, inhibits insulin-dependent phosphorylation and activation of PDE3B in rat adipocytes (6). We next tested whether wortmannin would prevent the insulin-induced association of endogenous PDE3B with endogenous 14-3-3 β in rat adipocytes. Pretreatment of adipocytes with 100 nmol/l wortmannin completely inhibited the insulin-induced activation of PDE3B and its association with 14-3-3 β (Fig. 3B). In addition, pretreatment of adipocytes with 100 μmol/l Ly294002, another PI3-K inhibitor, had the same effect (data not shown). These results suggest that PI3-K-dependent phosphorylation of PDE3B by insulin is required for the association of 14-3-3 β with PDE3B.

Insulin is reported to promote the phosphorylation of mouse Ser-273 (rat PDE3B on Ser-279) (8) or rat PDE3B on Ser-302 (7), and cAMP-dependent protein kinase (PKA) results in phosphorylation of rat PDE3B on Ser-427 (20). We therefore examined whether phosphorylation of any of these residues is required for the interaction between endogenous PDE3B and endogenous 14-3-3 β. Τhe effects of synthetic 15 amino acid peptides of rat PDE3B, containing either phosphorylated serine (pS279, pS302, or pS427) or nonphosphorylated serine (S279, S302, or S427) were tested on the protein-protein interaction. The pS279 and pS302 peptides both inhibited the insulin-induced association of PDE3B with 14-3-3 β in rat adipocytes (Fig. 3C). These effects were dose-dependent (data not shown). The corresponding nonphosphorylated peptides (S279 and S302) had no effect. Neither pS427 nor S427 had an effect. Thus, the two insulin-regulated phosphorylation sites, Ser-279 and Ser-302, are involved in the insulin-induced association of 14-3-3 β with PDE3B. β-Adrenergic agonists, as well as insulin, phosphorylate and activate PDE3B (21,22), but the exact residues involved remain controversial. Rahn et al. (7), using ³²P-labeled rat adipocytes, reported that insulin and isoproterenol both phosphorylate Ser-302 of rat PDE3B. Kitamura et al. (8), using mutagenesis analysis in 3T3L1 adipocytes, reported that insulin phosphorylates Ser-273 of mouse PDE3B (rat Ser-279), whereas isoproterenol phosphorylates the mouse enzyme on Ser-296 (rat Ser-302). The present data suggest that both Ser-279 and Ser-302 in rat PDE3B are involved in the insulin-induced interaction between 14-3-3 β and PDE3B. However, the addition of phosphorylated peptides (pS279 and pS302) to the solubilized P-2, which results in the dissociation of 14-3-3 β from PDE3B, caused only a slight decrease of insulin-induced PDE3B activity (data not shown). Thus, 14-3-3 β may not directly affect PDE3B activity, although the 14-3-3 protein is reported to regulate directly the phosphorylation and activity of Raf-1 and tyrosine hydroxylase (23–25).

14-3-3 proteins interact with conserved phosphorylated motifs in other proteins. In these motifs, an Arg residue at the -3 or -4 position relative to the phosphoserine is crucial, and Pro at +2 is also important (26). The sequences around Ser-279 and Ser-302 in rat PDE3B both contain Arg at -3 and -4 and the sequence around Ser-302 also includes Pro at +2. PDE3B therefore contains motifs with the potential of interacting with 14-3-3 proteins. In addition, the sequence around Ser-279 on rat PDE3B is consistent with the motif RX_1–2SX_2–3S found in Cbl and PKCμ, both of which bind to 14-3-3 proteins (27,28).

In summary, we have demonstrated an interaction between 14-3-3 and PDE3B, a key enzyme in the metabolic action of insulin. This interaction occurs in vivo and in vitro, is insulin-inducible, and involves the phosphorylation on serine residues 279 and 302 in rat PDE3B. 14-3-3 proteins exist as dimers and can simultaneously bind two ligands. The reports that IRS-1 associates with 14-3-3 β raise the possibility that the latter serves as a scaffolding protein to link IRS-1 to PDE3B. It is not clear whether the interaction between these molecules is involved in basal or insulin-mediated antilipolysis. Additional experiments are required to clarify these points.

This work was supported by a Grant-in-Aid for Scientific Research on Priority Areas (C) “Medical Genome Science” from the Ministry of Education, Culture, Sports, Science and Technology of Japan (No. 12204007), a Grant-in-Aid for Scientific Research (C) from the Japanese Society for the Promotion of Science (Nos. 11671122 and 11671124), and a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan by the Research Fellowships of the Japan Society for the Promotion of Science for Young Scientists (No. 60206-00).

We thank Drs. Philip James, Alan Saltiel, and Masato Kasuga for providing PJ69-4A, 3T3L1 adipocyte cDNA libraries, and the mouse PDE3B cDNA, respectively. We also thank Drs. Akira Sakai and Minako Kakui for technical advice in the use of two-hybrid screening.