Phosphorylation and activation of a cAMP-specific phosphodiesterase by the cAMP-dependent protein kinase. Involvement of serine 54 in the enzyme activation.

A cAMP-specific phosphodiesterase (PDE4D3) is activated in rat thyroid cells by TSH through a cAMP-dependent phosphorylation (Sette, C., Iona, S., and Conti, M. (1994) J. Biol. Chem. 269, 9245-9252). This short term activation may be involved in the termination of the hormonal stimulation and/or in the induction of desensitization. Here, we have further characterized the protein kinase A (PKA)-dependent phosphorylation of this PDE4D3 variant and identified the phosphorylation site involved in the PDE activation. The PKA-dependent incorporation of phosphate in the partially purified, recombinant rat PDE4D3 followed a time course similar to that of activation. Half-maximal activation of the enzyme was obtained with 0.6 μM ATP and 30 nM of the catalytic subunit of PKA. Phosphorylation altered the Vmax of the PDE without affecting the Km for cAMP. Phosphorylation also modified the Mg2+ requirements and the pattern of inhibition by rolipram. Cyanogen bromide cleavage of the 32P-labeled rat PDE4D3 yielded two or three major phosphopeptide bands, providing a first indication that the enzyme may be phosphorylated at multiple sites in a cell-free system. Site-directed mutagenesis was performed on the serine residues present at the amino terminus of this PDE in the context of preferred motifs for PKA phosphorylation. The PKA-dependent incorporation of 32P was reduced to the largest extent in mutants with both Ser13 → Ala and Ser54 → Ala substitutions, confirming the presence of more than one phosphorylation site in rat PDE4D3. While substitution of serine 13 with alanine did not affect the activation by PKA, substitution of Ser54 completely suppressed the kinase activation. Similar conclusions were reached with wild type and mutated PDE4D3 proteins expressed in MA-10 cells, where the endogenous PKA was activated by dibutyryl cAMP. Again, the PDE with the Ser54 → Ala substitution could not be activated by the endogenous PKA in the intact cell. These findings support the hypothesis that the PDE4D3 variant contains a regulatory domain target for phosphorylation at the amino terminus of the protein and that Ser54 in this domain plays a crucial role in activation.

The intracellular concentrations of the cyclic nucleotides cAMP and cGMP are determined by their rates of synthesis and degradation. The hormonal regulation of cyclic nucleotide synthesis by adenylyl and guanylyl cyclases was the first to be recognized, and most of the steps involved have been elucidated (reviewed in Refs. 2 and 3). The exact role of cyclic nucleotide degradation by phosphodiesterases (PDEs) 1 during hormonal stimulation is less well understood (reviewed in Ref. 4). A large number of PDE forms are present in a cell and may be activated by hormones via different mechanisms. These isoenzymes may be divided into seven classes (PDE1 to PDE7) according to their substrate specificity, selectivity for different drugs, and sequence homologies (reviewed in Refs. 1 and 4)). The presence of several genes that encode more than one protein variant within any given family (1) adds to the complexity of the PDE system, again implying differences in function and regulation during hormone action.
Recent studies from different laboratories have indicated that phosphorylation may play a major role in the regulation of PDEs. Phosphorylation of isoforms of most PDE classes has been reported, and, in some cases, phosphorylation was associated with changes in the hydrolytic activity (reviewed in Ref. 4). The brain 63-and 61-kDa calmodulin-dependent PDE isoenzymes (PDE1) are phosphorylated in vitro by calmodulin-dependent kinase II (5)(6)(7)(8) and PKA (5,9), respectively. Phosphorylation by either kinase causes a decrease in the affinity of the PDE for the Ca 2ϩ -calmodulin complex. A purified cGMP-stimulated PDE (PDE2) from bovine brain is a substrate for PKA in vitro (10). Phosphorylation does not affect the kinetic properties of the enzyme, and the physiological function of this modification is unknown (10). The cGMP-inhibited PDE (PDE3) is phosphorylated and activated in response to stimulation by insulin, prostaglandins, or ␤-adrenergic agonists in rat adipocytes (11,12), hepatocytes (13,14), and platelets (15,16). Although phosphatidylinositol 3-kinase may play a role (17), the kinase and the exact mechanism mediating the insulin-dependent phosphorylation and activation of the PDE3 (18) are still unknown. ␤-Adrenergic agonists and prostaglandins activate a cGI-PDE through a PKA-dependent phosphorylation (11), and Ser 427 in a cGI-PDE (19) may be a site for phosphorylation/activation. The cGMP-binding cGMP-PDE (PDE5) is phosphorylated in vitro by the cGMP-dependent kinase and PKA, with the former kinase having higher affinity for the PDE (20). This PDE is also phosphorylated in primary culture of rat vascular muscle cells in response to stimulation with atrial natriuretic factor (21). The function of this phosphorylation is not clear, even though a recent report showed that phosphorylation of a PDE5 isoform by PKA produces activation of the enzyme (22). Finally, it has been shown that the ␣, ␤, and ␥ subunits of the photoreceptor PDE (PDE6) are also phosphoproteins (23)(24)(25)(26). Phosphorylation of the ␥ inhibitory subunits by PKC increases their affinity for the ␣ and ␤ catalytic subunits and could play a role in the adaptation to light (26).
In rat thyroid cells, TSH induces activation of a cAMP-PDE (PDE4D) (27). The activation is mediated by a cAMP-dependent phosphorylation and can be reproduced in a cell-free system by incubation of the native or recombinant PDE4D3 with the catalytic subunit of PKA (27,28). The PDE4D gene encodes three or more mRNA variants that differ in the 5Ј region (28). PDE4D1 (encoded protein of 72 kDa) and PDE4D2 (67-68 kDa) differ in the presence (PDE4D1) or removal (PDE4D2) of a short intron sequence (29) in the mRNA, and are regulated by cAMP at the level of transcription or mRNA stability (28,30). Hormones that raise intracellular cAMP levels in Sertoli and thyroid cells induce the expression of these two PDE4D variants (28,30,31). PDE4D3 mRNA is instead constitutively expressed in thyroid cells (28,32). This mRNA codes for protein with an additional 132-amino acid domain at the amino terminus, which, according to our hypothesis, renders the protein sensitive to PKA activation (28). In this study we characterized the phosphorylation and activation of the PDE4D3 protein and investigated the role of this regulatory domain by site-directed mutagenesis.
PCR Amplification and Product Purification-PCR was performed using Taq polymerase and a DNA thermal cycler (Perkin-Elmer). Amplification was performed using conditions provided by Perkin-Elmer. In a final volume of 100 l, the reaction mixture contained 10 mM Tris-Cl, pH 8.3, 50 mM KCl, 1.5 mM MgCl 2 , 0.01% (w/v) gelatin, 200 M each dNTP, 0.5 M each primer, 1-5 ng of template (pCMV5-rat PDE4D3 cDNA), and 1 unit of Taq polymerase. Reactants were overlaid with 100 l of mineral oil and subjected to 15 cycles of denaturation (1 min, 94°C), annealing (2 min, 55°C), and extension (1 min, 72°C). The PCR products were separated by electrophoresis on a 1.5% agarose gel, followed by staining with ethidium bromide, and visualized with UV light. DNA bands of the correct size were excised and purified from the gel with the Pharmacia Band Prep Kit following the instructions of the manufacturer. For the Ser 13 3 Ala mutation, oligomers A and H were used in a first round of PCR. The amplified fragment was purified and used as a "megaprimer" in a second round of PCR together with oligomer I (megaprimer method (33)). The product of the second amplification was analyzed by restriction digestion and agarose gel electrophoresis. For the Ser 54 3 Ala mutation, oligomers B and H and oligomers C and I were used in a first round of PCR to synthesize two fragments overlapping at the site of mutation. After purification, the two PCR-amplified fragments were mixed together, allowed to anneal, and extended (overlapping PCR method (34)). These fragments were then used as a template for a second round of PCR with oligomers H and I. The same strategy was used for Ser 74 3 Ala and Ser 119 3 Ala. Oligomers D and E and oligomers F and G were used for Ser 74 and Ser 119 , respectively. The double mutant Ser 13 /Ser 54 3 Ala was obtained by the overlapping PCR method with primers B/H and C/I using mutant Ser 13 3 Ala PDE4D3 cDNA as a template. Details of these PCR-based mutagenesis strategies have been described elsewhere (33,34).
Eukaryotic Expression Vector and Mutation Constructs-To express wild type and mutant PDE4D3 proteins in MA-10 Leydig tumor cells, the eukaryotic expression vector pCMV5 was used. For the mutation constructs, PCR fragments containing Ser 13 3 Ala, Ser 74 3 Ala, or Ser 119 3 Ala substitutions were digested with EcoRI and StuI, and the PCR fragments containing either Ser 54 3 Ala or Ser 13 /Ser 54 3 Ala were digested with EcoRI and CfoI. The fragments were purified as described above. The rat PDE4D3 cDNA was digested either with EcoRI and StuI or with EcoRI and CfoI. The fragments containing the sequence downstream from the mutated fragments were purified as described above and ligated to the mutated fragments and to the pCMV5 vector that had been digested with EcoRI followed by dephosphorylation. The correctness of the constructs was verified by sequencing.
To express the recombinant PDEs in Sf9 insect cells, the wild type and mutated PDE4D3 cDNAs were excised from the pCMV5 constructs by EcoRI digestion and ligated to the eukaryotic expression vector pSYN XIV VI ϩ X/3. An ATG internal to the pSYN XIV VI ϩ X/3 vector was used in order to increase the efficiency of expression of the recombinant proteins. Therefore, recombinant proteins were expressed as fusion proteins containing six additional amino acids (Met-Gly-Ser-Ser-His-Gly) at the amino terminus.
Expression of Recombinant PDE4D3 Enzymes-To express the recombinant PDE4D3 wild type and mutants in MA-10 cells, the pCMV5 vectors containing the respective cDNAs were transfected by the CaPO 4 method (37). Cells were harvested 24 h after transfection in a homogenization buffer containing 20 mM Tris-Cl, pH 8.0, 1 mM EDTA, 0.2 mM EGTA, 50 mM NaF, 10 mM sodium pyrophosphate, 50 mM benzamidine, 0.5 g/ml leupeptin, 0.7 g/ml pepstatin, 4 g/ml aprotinin, 10 g/ml soybean trypsin inhibitor, and 2 mM phenylmethylsulfonyl fluoride. Cells were then homogenized and centrifuged for 10 min at 14,000 ϫ g. The PDE activity was measured in the homogenates or in the soluble extracts. Protein concentration was measured according to Bradford (38). To express recombinant PDE4D3 wild type in Sf9 insect cells, the rat PDE4D3 cDNA was subcloned in the expression vector pSYN XIV VIϩX/3. The plasmid was cotransfected with BaculoGold virus DNA (PharMingen) by lipofection. The recombinant virus stock was made from a single plaque isolated after 6 days. The virus was amplified by infection of 10 ml of growing Sf9 cells in a shaking incubator at 27°C for 5 days. At the end of the incubation, the infected cells were separated by centrifugation at 1,500 ϫ g for 10 min, and the medium containing the virus was collected and stored at 4°C in the dark. To prepare recombinant PDE4D3, 30 -50 ml of Sf9 cells at a density of 0.8 -1.0 ϫ 10 6 cells/ml were infected with recombinant baculovirus and grown for 3 days in Sf900 medium (Life Technologies, Inc.) containing 1% heatinactivated fetal calf serum, 50 g/ml Gentamicin, and 4% feed stock. After 3 days, cells were collected by centrifugation and resuspended in lysis buffer containing 40 mM Tris-Cl, pH 8.0, 1 mM EDTA, 0.2 mM EGTA, 50 mM benzamidine, 0.5 g/ml leupeptin, 0.7 g/ml pepstatin, 4 g/ml aprotinin, 10 g/ml soybean trypsin inhibitor, and 2 mM phenylmethylsulfonyl fluoride. Cells were homogenized and centrifuged for 10 min at 14,000 ϫ g, at 4°C. Soluble extracts were centrifuged for 1 h at 100,000 ϫ g, at 4°C. The cytosolic fraction was then diluted in 200 mM sodium acetate, pH 6.5, and loaded onto a DEAE ion exchange HPLC column equilibrated with 200 mM sodium acetate, pH 6.5, at a flow rate of 1 ml/min. After washing the column with 2-3 volumes of the same buffer, bound proteins were eluted with a 200 -750 mM sodium acetate pH 6.5 linear gradient. PDE activity was assayed in each fraction, and fractions containing PDE activity were diluted to 33% ethylene glycol and stored at Ϫ20°C for further studies. To determine the specific activity of the wild type and the double mutant PDE, the fractions of the DEAE containing the peak of activity were pooled and injected on a TSK-Phenyl 5PW-HPLC column equilibrated at 70 mM sodium acetate, pH 6.5. Proteins were then eluted with a gradient of ethylene glycol (0 -35%) and assayed for PDE activity. Details of the purification have already been reported elsewhere (39). The activities of these preparations were measured as detailed below. Protein content was measured either by the Lowry procedure or by quantitation of the Coomassie Blue-stained PDE band, using bovine serum albumin as the standard.
PDE Assay-PDE activity was measured using 1 M cAMP as substrate, according to the method of Thompson and Appleman (40) and as detailed previously (41). Samples were assayed in a total volume of 200 l of reaction mixture including 40 mM Tris-HCl (pH 8.0), 1 mM MgCl 2 , 1.25 mM 2-mercaptoethanol, 1 M cAMP, 0.14 mg of bovine serum albumin, and [ 3 H]cAMP (0.1 Ci/tube). In some experiments serial dilutions (1 nm to 10 M) of rolipram were added to the reaction mixture. After incubation at 34°C for 5-15 min, the reaction was terminated by adding an equal volume of 40 mM Tris-Cl, pH 7.5, containing 10 mM EDTA, followed by heat denaturation for exactly 1 min at 100°C. To each reaction tube 50 g of C. atrox snake venom was added, and the incubation was continued at 34°C for 20 min. The reaction products were separated by anion exchange chromatography on AG1-X8 resin, and the amount of radiolabeled adenosine collected was quantitated by scintillation counting.
Metabolic Labeling of PDE4D3 with [ 32 P]Orthophosphate-FRTL-5 cells were seeded in 90-mm dishes (Corning) and cultured as described above. Cells were made quiescent by replacing the serum and the hormones in the medium with 0.1% bovine serum albumin for 24 h. After induction of quiescence, the medium was replaced with phosphate-free minimal essential medium containing 20 mM Hepes, pH 7.4, and carrier-free [ 32 P]orthophosphate (0.2-0.3 mCi/ml), and cells were incubated for an additional 2 h. During the last 15 min of incubation, part of the cells were treated with 10 nM TSH. At the end of the treatment, cells were washed 3 times with Hanks' balanced salt solution, harvested in homogenization buffer, and homogenized. Supernatants from a 10-min centrifugation at 14,000 ϫ g were immunoprecipitated with K116 anti-PDE4 antibody. The immunoprecipitation was carried out as detailed below with the following modification. Pansorbin was first incubated with unlabeled soluble extract from quiescent FRTL-5 cells for 1 h in order to reduce the nonspecific binding of labeled proteins and then adsorbed to K116 for an additional 60 min at 4°C. After three washes, the immobilized antibody was incubated with labeled FRTL-5 cell extracts for 90 min at 4°C. After recovering bound protein from the immunoprecipitation pellet and boiling, proteins were separated on a 10% SDS-PAGE and blotted onto an Immobilon membrane. The membrane was dried and exposed for autoradiography. Immunoprecipitation-Pansorbin was washed twice with PBS containing 0.1% bovine serum albumin and then incubated with K116 antibody usually diluted at 1:50. After a 60-min incubation at 4°C, the unbound antibody was removed by one wash with 20 mM Tris-HCl, pH 7.8, 0.5 M NaCl and two washes in the same buffer without NaCl. The immobilized antibody was mixed with the crude enzyme extracts and incubated for 90 min at 4°C with occasional mixing. At the end of the incubation, the complexes were separated from free protein by centrifugation at maximum speed in a microcentrifuge at 4°C. Pellets containing immunoadsorbed proteins were first washed with buffer containing 80 mM Tris-Cl, pH 8.0, 0.7 M NaCl, 0.1% SDS, and 0.4% Triton X-100 and then eluted with 1% SDS in PBS. The eluted protein was diluted with concentrated SDS-PAGE sample buffer (4 ϫ) and separated by SDS-PAGE.
Phosphopeptide Analysis-Partially purified rat PDE4D3 was incubated for 15 min with the catalytic subunit of PKA (0.1 M) in the presence of 0.1 mM [␥-32 P]ATP (0.5 Ci/l) at 30°C. At the end of the incubation, the reaction was stopped by the addition of SDS-PAGE sample buffer, and the sample was boiled for 5 min at 100°C. Proteins were separated by 8% SDS-PAGE and blotted onto nitrocellulose membrane. The membrane was exposed to autoradiography, and the phosphorylated rat PDE4D3 band was excised from the membrane. The membrane was rehydrated in water and then incubated for 90 min with 100 mg/ml cyanogen bromide in 70% formic acid at room temperature. At the end of the incubation, the samples were centrifuged for 5 min at 14,000 ϫ g, and the supernatant was collected and lyophilized. The pellet was resuspended in SDS-PAGE sample buffer and boiled for 5 min at 100°C. Peptides were separated by SDS-Tricine-Tris 16% PAGE and blotted onto Immobilon membrane. The membrane was dried and exposed to autoradiography. In some experiments the SDS-Tricine gel was dried and directly used for autoradiography. Omission of the Immobilon transfer step yielded the same phosphopeptides but higher background.
Time Course of the Activation of PDE4D3 by PKA-Recombinant PDE4D3 proteins were diluted in a reaction buffer containing 40 mM Tris-Cl, pH 7.4, 2 mM magnesium acetate, 0.1 mM ATP, 1.25 mM 2-mercaptoethanol, 0.1 mg/ml bovine serum albumin and incubated at 30°C for increasing intervals of time in the absence or presence of the catalytic subunit of protein kinase A (0.1 M) (Promega). When samples were assayed for PDE activity, 1 M [ 3 H]cAMP (0.1 Ci) was used as substrate. When 32 P incorporation was measured, 0.1 mM [␥-32 P]ATP (0.5 Ci/l) was added to the reaction. The final volume in each tube was 2 ml, 200-l aliquots of the reaction mixtures were withdrawn at different times, and the reaction was terminated as described for the PDE assay. For the ATP dose-dependent activation of PDE4D3 by PKA (0.1 M), the reaction buffer was modified to contain 20 mM magnesium acetate and serial dilutions of ATP (0 -0.4 mM). For the PKA dose-dependent activation, the reaction buffer was modified to contain 20 mM magnesium acetate, 0.1 mM ATP, and serial dilutions of the catalytic subunit of PKA (0 -1 M). In these two last experiments, after a 15-min incubation at 30°C, samples were diluted in 40 mM Tris-Cl, pH 8, containing 0.1% bovine serum albumin, and the PDE activity was measured as described above. 32 P incorporation was measured by densitometry of the autoradiogram using the NIH Image software.

PKA-mediated Phosphorylation of Rat PDE4D3-
The activity of a cAMP-PDE is stimulated in the intact thyroid cell by TSH (32) and in a cell-free system by incubation with the catalytic subunit of PKA (28). Together with PDE activation, short-term treatment of quiescent FRTL-5 thyroid cells with TSH induced an increase in 32 P incorporation in a polypeptide of 93-97 kDa (32) (Fig. 1A). When recombinant PDE4D3 was incubated in a cell-free system with the catalytic subunit of PKA (Fig. 1B), a polypeptide with electrophoretic properties identical to the phosphoprotein derived from intact cells was phosphorylated. Similar results were obtained whether the recombinant PDE was derived from expression in mammalian cells or in insect cells as a fusion protein (see below). To investigate the phosphorylation/activation of the enzyme by PKA in more detail, the recombinant PDE4D3 was partially purified by a DEAE ion exchange HPLC column as described under "Experimental Procedures." Incubation of PDE4D3 with the catalytic subunit of PKA produced a 2-3-fold increase of phosphodiesterase activity (Fig. 2A). In Fig. 2, B and C, the time courses of activation and the incorporation of 32 P in the enzyme are reported. In the absence of the kinase, the activity of PDE4D3 was constant over 15 min, and a trace amount of 32 P incorporation was detected (Fig. 2C). The activation of PDE4D3 by PKA was completely suppressed both by the addition of a PKA inhibitor and by the omission of ATP from the assay (data not shown). Incubation of rat PDE4D3 with similar concentrations of cGMP-dependent protein kinase or protein kinase C, under appropriate conditions (see "Experimental Procedures"), neither produced changes in the phosphodiesterase activity nor caused any detectable 32 1 M). When the PDE activity was measured (A), 1 M [ 3 H]cAMP was added to the reaction, and the activity was assayed as described under "Experimental Procedures." When the 32 P incorporation was measured (B and C), 0.1 mM [␥-32 P]ATP was used as substrate for the kinase. The reaction was stopped by the addition of SDS-PAGE sample buffer, and the samples were separated by SDS-PAGE and blotted onto Immobilon membrane. The membrane was first exposed to autoradiography and then analyzed by Western blot (C). The 32 P incorporation was analyzed by densitometry of the autoradiography and normalized for densitometric values derived from the Western blot (B). Data are the mean of two separate experiments.
ing that PDE4D3 is a substrate for PKA in the intact cell, where the concentration of the kinase ranges between 0.2 and 0.7 M (42). It was also observed that phosphorylation causes a decrease in the Mg 2ϩ concentration required for cAMP hydrolysis (Fig. 3).
Rolipram Inhibition of Phosphorylated Rat PDE4D3-PDE4 isoenzymes are selectively inhibited by the antidepressant rolipram (1,43). To investigate the effect of phosphorylation on rolipram inhibition, recombinant rat PDE4D3 was incubated for 15 min in the absence or presence of PKA (0.1 M), and the PDE activity was measured in the presence of increasing concentrations of rolipram (Fig. 4A). In the absence of PKA, the inhibition profile of the PDE4D3 activity showed a biphasic curve, with a high affinity inhibition at 10 Ϫ9 M rolipram and a low affinity inhibition at 10 Ϫ6 M rolipram. The addition of the kinase produced a stimulation of the PDE activity that is inhibited by rolipram with a high affinity (Fig. 4A). To determine whether the presence of two rolipram affinity states was an artifact induced by the partial purification of the enzyme, a similar experiment was performed using FRTL-5 cells. These cells were chosen because the predominant PDE activity expressed can be activated by short term incubation of quiescent cells with TSH (32). A similar inhibition profile was observed with soluble extracts from quiescent and TSH-stimulated FRTL-5 cells (Fig. 4B). TSH treatment induced activation only of the PDE that displayed a high affinity for rolipram, reproducing the effect obtained by PKA in a cell-free system (Fig. 4, A and B).
Site-directed Mutagenesis of PDE4D3-The amino-terminal region of PDE4D3 contains 19 Ser and 6 Thr residues (Fig. 5). The sequence surrounding Ser 13 (RRHSW) and Ser 54 (RRESF) is identical to the PKA preferred motif (RRX(S*/T*)X) found in many physiological substrates of the kinase (44) (Fig. 5). Two additional Ser in the amino terminus of PDE4D3 (Ser 74 and Ser 119 ) are preceded by basic residues. These sequences are similar to the RX(S*/T*) motif, which has been shown to be the site of PKA phosphorylation for some substrates, at least in vitro (44). In Fig. 5 the predicted cyanogen bromide cleavage sites in this region of PDE4D3 are also indicated. Two peptides with predicted sizes of 3.5 and 4.1 contain Ser 13 and Ser 54 . An additional peptide of 5.5 kDa contains serines 74 and 119. A phosphopeptide analysis of rat PDE4D3 cleaved with cyanogen bromide yielded two major phosphopeptide bands of 5.3 and 4.5 kDa (Fig. 6). In several experiments, the 4.5-kDa band could be resolved into two components (data not shown). This experi-ment provided a first indication that PDE4D3 is phosphorylated at multiple sites under these cell-free conditions.
The possibility that Ser 13 and Ser 54 are indeed phosphorylated and involved in the PDE4D3 activation by PKA was further tested by site-directed mutagenesis. Serine-to-alanine substitutions were inserted by PCR amplification with oligomers containing a single base change (as described under "Experimental Procedures"). The insertion of the mutations in PDE4D3 was confirmed by sequencing of the cDNAs. The wild type and mutant PDE4D3 enzymes were then expressed in MA-10 cells by transient transfection or by infection in insect cells. Comparable results were obtained with proteins derived from the two expression systems. To test the degree of phosphorylation of the PDE4D3 wild type and mutant proteins, phosphorylation was carried out in the presence of [␥-32 P]ATP (Fig. 7). Substitution of Ser 13 or Ser 54 caused a reduction in 32 P incorporation in PDE4D3 when compared with the wild type enzyme (Ser 13 3 Ala, 48.6 Ϯ 3.9% of wild type, n ϭ 5; Ser 54 3 Ala, 58.2 Ϯ 10.0 of wild type, n ϭ 4). The protein with double substitutions at Ser 13 and Ser 54 incorporated much less 32 P than the wild type PDE (20.1 Ϯ 8.7 of wild type, n ϭ 3). Under these experimental conditions of maximal phosphorylation, it was estimated that approximately 2 mol of phosphate/mol of PDE were incorporated in the wild type PDE, while the double mutant incorporated less than 0.4 mol/mol of PDE. This experiment indicated that Ser 13 and Ser 54 were the preferential sites of phosphorylation but that additional sites were also used in this cell-free model.
The effect of the different mutations on the PKA-dependent activation of the PDE was next tested in a cell-free system. The Ser 13 3 Ala substitution did not affect the activation by PKA (Fig. 8). By contrast, mutation Ser 54 3 Ala completely suppressed the effect of the kinase (Fig. 8). The double mutation of Ser 13 /Ser 54 3 Ala in the conserved consensus sites behaved in a manner identical to the Ser 54 3 Ala mutant (Fig. 8). Since the PDE4D3 sequence contains two additional serines (Ser 74 and Ser 119 ), which reside in the less conserved consensus for PKA (Fig. 5), phosphorylation of these residues may occur under the conditions used, as suggested by the phosphorylation studies and the phosphopeptide map analysis. To determine whether phosphorylation of these two additional sites is important for activation of PDE4D3 by PKA, site-directed mutagenesis of the two residues was performed. Neither substitution Ser 74 3 Ala nor substitution Ser 119 3 Ala affected the activa-tion of PDE4D3 by the kinase (data not shown). This indicated that, even if phosphorylated, these two residues do not play a crucial role in PDE activation.
The lack of stimulation of the PDE bearing the Ser 54 3 Ala substitution may be the consequence of a change in conformation of the protein rendering it constitutively activated. This constitutive activation unrelated to phosphorylation would prevent or mask a further activation by PKA. If this were the case, the specific activity of PDE4D3 Ser 13 /Ser 54 3 Ala would be 2-3-fold higher than that of wild type PDE4D3. To investigate this possibility, the recombinant proteins expressed as fusion proteins in the baculovirus-Sf9 cell system were purified by two chromatographic steps to 90 -95% homogeneity as determined by SDS-PAGE and Coomassie Blue staining (data not shown). In the two experiments performed, the specific activities of wild type and mutated proteins were comparable (PDE4D3 wild type 11.4 Ϯ 2.6 mol/min ϫ mg; the double mutant PDE4D3 6.64 Ϯ 0.17 mol/min ϫ mg), indicating that the double mutation does not yield a constitutively activated enzyme.
The whole of these data then indicate that, under the cellfree conditions used, PKA phosphorylates PDE4D3 on several residues. Ser 54 was the only residue involved in the kinase activation of the enzyme.
Activation of PDE4D3 Mutants in Intact MA-10 Cells-To determine whether Ser 54 is also important for activation of the rat PDE4D3 in the intact cell, MA-10 cells were transfected with rat PDE4D3 wild type or with rat PDE4D3 mutant cDNAs. After transfection, cells were incubated in the presence or absence of Bt 2 cAMP to activate the endogenous PKA, and the PDE activity was measured in the cell homogenates. Bt 2 cAMP treatment induced a 50% stimulation of PDE activity in cells transfected with PDE4D3 wild type cDNA (Table I). Similar activation was observed in cells transfected with mutant Ser 13 3 Ala (46%), whereas in cells transfected with mutant Ser 54 3 Ala or Ser 13 /Ser 54 3 Ala, the activation was negligible (2-9%). (Table I). These data indicate that Ser 54 of PDE4D3 is a residue necessary for PKA-dependent activation also in the intact cell. . At the end of the incubation, phosphorylated PDE4D3 was separated by SDS-PAGE and blotted onto nitrocellulose membrane, and migration was detected by autoradiography. The excised band was then incubated for 90 min with 100 mg/ml cyanogen bromide in 70% formic acid. At the end of the incubation, the sample was collected by centrifugation at 14,000 ϫ g for 5 min, lyophilized, and resuspended in SDS-PAGE sample buffer. The digested peptides were separated by SDS-Tricine-Tris-HCl, 16% polyacrylamide gel and blotted onto Immobilon membrane, and the membrane was exposed for autoradiography. system and in the intact cell. This finding strongly suggests that this PDE is a substrate for PKA in vivo and that TSH activates this PDE via a PKA-dependent phosphorylation. Furthermore, the identification of this residue at the amino terminus of this variant opens the possibility that this region of the protein encodes a regulatory domain. The presence of this domain distinguishes this protein from the previously described variants (PDE4D1 and PDE4D2) derived from the same gene, and explains why this variant is the only one activated by a PKA-dependent phosphorylation (28).
Phosphorylation studies in a cell-free system indicate that PDE4D3 is a good substrate for PKA. Phosphorylation and activation occur at PKA concentrations well within the physiological range found in the cell. Several sites are phosphorylated in this cell-free system. Two of them have been identified as Ser 13 and Ser 54 , since substitutions at the two sites cause a decrease in phosphate incorporation in the PDE protein. The finding that the double mutant is still phosphorylated by PKA, albeit at a much lower level, indicates that additional phosphorylation sites may be present in PDE4D3. Whether phosphorylation at these additional sites occurs also in vivo in the intact cell remains to be determined. Of the four potential phosphorylation sites studied, only mutation of Ser 54  incubation, proteins were precipitated with 7% trichloroacetic acid and 0.01% deoxycholic acid followed by centrifugation for 10 min at 14,000 ϫ g. Pellets were resuspended in SDS-PAGE sample buffer, and the resuspended proteins were separated by SDS-PAGE and blotted onto Immobilon membrane. The membrane was first exposed for autoradiography and then subjected to Western blot using an anti-PDE4D3 monoclonal antibody (M3S1).  dibutyryl cAMP. Thus, PKA activation in the intact cell phosphorylates PDE on Ser 54 , in turn producing its activation. An alternative explanation of the mutagenesis studies is that the Ser to Ala substitution produces a conformational change and a constitutively active enzyme that cannot be further activated by PKA. However, this hypothesis is in conflict with the finding that the specific activity of this mutant is comparable with or lower than that of the wild type enzyme. It is then unlikely that the introduced mutations cause a constitutive activation of the PDE. Another possibility that cannot, at present, be excluded is that the introduced mutation causes a change in conformation of the protein that prevents phosphorylation at a site other than Ser 54 , thus disrupting the activation. Phosphopeptide mapping after metabolic labeling of the native enzyme in intact FRTL-5 cells and sequencing will be required to answer this question. In addition, it will be important to determine whether Ser 54 is the only phosphorylation site involved in TSH activation in the intact cell.
In all experiments performed, phosphorylation of the cAMP-PDE produced a 2-3-fold increase in activity. It is worth noting that, although small, this activation is quantitatively similar to the activation of a cGI-PDE by insulin or isoprenaline (45). Under basal conditions, i.e. without kinase treatment, substantial activity of the isoenzyme was detected. Measurements of the activity of this enzyme under conditions of complete dephosphorylation need to be performed to clarify this point. However, it should be noted that the PDE with substitution of the two major phosphorylation sites has substantial hydrolytic activity. Since it is assumed that the double mutants are recovered in the dephosphorylated state, the possibility needs to be entertained that PDE phosphorylation modifies the activity but does not produce a complete transition from an inactive to an active state. This possibility is at odds with the finding that rolipram, a specific inhibitor of PDE4, has no effect on cAMP levels in the intact FRTL5 cell under basal conditions (46), an indication that this enzyme is mostly inactive under basal conditions. A factor to be considered in the interpretation of the data is that the PDE assay used does not reflect the conditions in the intact cell. For instance, the Mg 2ϩ concentration used in the PDE assay is about 10-fold higher than the concentration found in the cell. Indeed, if activation is measured at lower Mg 2ϩ concentrations, basal activity is reduced, thus increasing the degree of activation brought about by the kinase treatment. It is also possible that the cell-free system used here is lacking additional regulatory components present in the cell. For instance, phosphorylation may affect the interaction of a putative inhibitor or activator with the PDE. These regulatory molecules may affect the conformation and the activation of the enzyme (see below).
It remains possible, however, that the cell requires only minor changes in PDE activity and that these small changes can greatly affect the response of a target cell to hormones. It has been reported that hCG-induced steroidogenesis in MA-10 Leydig tumor cells transfected with a cAMP-PDE was reduced by 80%, although only a 2-3-fold increase in PDE activity could be measured in extracts from these cells (46). Furthermore, FRTL-5 thyroid cells infected with a constitutively active Gs␣ protein showed a 10-fold increase in adenylyl cyclase activity but only a 40 -50% increase in PDE activity compared with wild type FRTL-5 cells (46,47). Since less than a 2-fold increase in intracellular cAMP levels was found in these cells, the small increase in PDE detected is sufficient to counteract the large increase in cyclase activity (46).
The PDE4 isoenzymes are selectively inhibited by the antidepressant rolipram. The experiments reported in this study indicated that freshly isolated and partially purified rat PDE4D3 is inhibited by rolipram with low (EC 50 ϭ 1 M) and high affinity (EC 50 ϭ 10 nM). Phosphorylation of the enzyme causes an increase exclusively in the activity inhibited by rolipram with a high affinity. These data are similar to what has been observed with the PKA activation of the human PDE4D3 and inhibition by analogs of nitraquazone (48). It has been previously shown that a rolipram high affinity (K d ϭ 1-10 nM) binding site copurifies with cAMP-PDEs from brain (49) and from recombinant yeast extracts expressing a cAMP-PDE (50). The affinity of this binding site is considerably higher than the affinity of rolipram for the catalytic site, as estimated from inhibition of the activity (K i ϭ 0.5-1 M). To reconcile these differences, it can be hypothesized that PDE4D3 is recovered in two states. One conformation allows binding of rolipram with a high affinity and inhibition of the activity at low nanomolar concentrations. The PDE in this conformation can be activated by phosphorylation. The second state has low affinity binding for rolipram, and the catalytic activity is inhibited by micromolar concentrations of rolipram. This second state does not allow activation by phosphorylation. The physiological relevance of these two conformation states is uncertain, but the finding that some biological responses are inhibited by rolipram at very low concentrations suggests that a significant portion of these enzymes may be present in the cell in a high affinity conformation.
The PDE4D locus encoding the cAMP-PDE object of our study contains several transcriptional units under the control of different promoters (4). This feature and the occurrence of alternate splicing of the mRNAs indicate that distinct protein variants (28,29) are derived from this gene. Several PDE proteins have indeed been identified immunologically as products of this gene. PDE4D1 is a short protein of 72 kDa with an amino terminus different from that of the long 93-kDa PDE4D3. The PDE4D2 (68 kDa) is derived from alternate splicing of the mRNA and encodes a truncated PDE (28). The presence of one or more additional variants has been inferred from cDNA cloning and from mRNA or Western blot analysis (48,51). The observation that different regulatory mechanisms control these proteins provides a physiological explanation for the existence of these different forms. The expression of PDE4D1 and PDE4D2 is regulated by cAMP at the level of transcription and/or messenger stabilization in endocrine cells like the Sertoli cell (30,31). Similar regulations have been demonstrated for FRTL-5 thyroid cells (32), inflammatory cells (52,53,54), glioma cells (31), and skeletal muscle cells (55). The experiments described here strengthen the hypothesis that the variant PDE4D3 is activated by a different mechanism. Hormones that act through the cAMP-dependent pathway produce a rapid activation of this PDE form, and this activation is mediated by a PKA-dependent phosphorylation. Although this form may be phosphorylated at multiple sites at the amino terminus of the protein, our data indicate that the phosphorylation at Ser 54 plays a crucial role in the PKA-dependent activation both in a cell-free system and in the intact cell. Since the other two variants studied, PDE4D1 and PDE4D2, are not good substrates for PKA and cannot be activated by phosphorylation (28), we propose that the longer PDE4D3 form contains a regulatory domain absent in the other variants. The identification of Ser 54 as the most likely target for this phosphorylation confirms this view. Thus, the presence of different variants may be the result of the inclusion of different regulatory domains that modify the function of the protein. This points to the modularity of the PDE protein and suggests that all PDEs may have similar arrangements of regulatory and catalytic domains. Our observation made on these cAMP-PDEs is in complete agreement with early observations made on the CaM-PDE and the cGS-PDE (56,57), where regulatory domains are present at the amino terminus of these proteins.
Activation by phosphorylation is not unique to the rat PDE4D3 because human (48) and mouse recombinant PDE4D3 can also be activated by phosphorylation. The same consensus around Ser 54 is present in variants derived from the PDE4A, PDE4B, and PDE4C genes (51). On the basis of these structural similarities, one would predict that all variants containing that domain should be good substrates for PKA and that the phosphorylation should cause an activation of these additional forms. In preliminary experiments performed, PDED4A and PDE4B variants containing this domain could not be activated under conditions in which PDE4D3 activity is stimulated. The reason for this is unclear. It is possible that conditions for the PKA-dependent activation are not optimal to detect activation of these PDE4 variants derived from different genes. Alternatively, the large differences at the carboxyl terminus may influence the phosphorylation and activation of these two other variants. Finally, other kinases may be involved in the phosphorylation and activation of PDE4A, PDE4B, and PDE4C.
In summary, our results demonstrate that PKA phosphorylation activates rat PDE4D3 and the activation can be suppressed by substitution of Ser 54 with Ala. It is proposed that the N terminus 132-amino acid domain of PDE4D3 is a regulatory domain that allows modulation of the activity of the protein by phosphorylation. The phosphorylated/activated PDE4D3 may be the preferred target of the antidepressant drug rolipram in the intact cell. These data, therefore, suggest that PDE4D3 plays a crucial role in the hormonal control of intracellular cAMP levels. The exact physiological significance of this regulation requires further investigation. Phosphorylation and activation of this PDE variant may occur at the same time as adenylyl cyclase activation to maintain cAMP in a narrow range of concentrations and to decrease the time required to attain a new steady state. In addition, this regulation may be a component of the mechanisms that terminate the hormone stimulation and produce cell adaptation or desensitization.