Subcellular Localization of Rolipram-sensitive, cAMP-specific Phosphodiesterases

Biochemical and immunofluorescence analyses revealed that phosphodiesterase variants encoded by thePDE4D gene are targeted to discrete subcellular structures. In quiescent FRTL-5 thyroid cells, the rolipram-sensitive phosphodiesterase (PDE) activity (cAMP-PDE) was recovered both in the soluble and particulate fractions of the homogenate. Although an immunoreactive 93-kDa PDE (PDE4D3) variant was recovered in both compartments, a 105-kDa variant with the properties of PDE4D4 was recovered mostly in the particulate fraction. The PDE4D3 form was readily solubilized with nonionic detergents. Conversely, the PDE4D4 form required buffers containing ionic detergents for extraction, suggesting that different mechanisms target these variants to insoluble structures. A 15-min stimulation with thyroid-stimulating hormone (TSH) led to an activation of the cAMP-PDE in both compartments and was correlated with a shift in electrophoretic mobility of the PDE4D3 polypeptide. Long term incubation with TSH caused an increase of the PDE activity in the soluble fraction and the appearance of a 68-kDa immunoreactive polypeptide with the properties of PDE4D2. Immunofluorescence analysis showed, in addition to diffuse staining, a signal localized on regions adjacent to the plasma membrane on cytoskeletal structures and in a perinuclear region of quiescent cells. Long term incubation with TSH caused an increase in the immunofluorescence signal in the soluble compartment. These data demonstrate that three PDE4D splicing variants are targeted to discrete subcellular compartments and that hormones cause the activation of these isoforms in a temporally and spatially dependent manner.

With the discovery of modular binding domains in proteins, it has been established that protein-protein interactions and compartmentalization play a crucial role in the transduction of growth factor signals (1)(2)(3). Binding and activation of the receptors for these growth factors trigger a cascade of phosphorylation/dephosphorylation reactions that in turn controls association/dissociation of oligomeric structures. These macromolecular complexes direct the signals to specific cellular compartments, i.e. plasma membrane, cytoskeletal structures, or the nucleus. Disruption of these complexes by blocking pro-tein-protein interactions causes an impairment of signaling (2).
Cyclic nucleotides are small second messengers that can diffuse throughout the cell. Synthesis of cAMP occurs at the level of the plasma membrane where receptors and adenylyl cyclases are localized. Under physiological conditions, diffusion from the site of synthesis may be minimal (4) so that compartmentalization of the different components of this transduction machinery and proximity effects may dictate the specificity of the cAMP signaling. This hypothesis is supported by ample evidence that the intracellular effector for cAMP, protein kinase A (PKA), 1 is targeted to different subcellular structures. The regulatory subunit RII␤ of PKA associates with PKAanchoring proteins, which are proteins associated with particulate structures (5)(6)(7). PKA-anchoring proteins serve to anchor the holoenzyme close to its putative substrates on cellular components such as microtubules, Golgi apparatus, or membranes (8 -10). PKA function appears to rely on PKA-anchoring protein targeting; microinjection of a synthetic peptide that inhibits the binding of the RII subunit of PKA to PKA-anchoring proteins prevents the PKA-mediated regulation of ␣-amino-3-hydroxy 5-methyl-4-isoxazole propionic acid glutamate receptor-gated ion channel in hippocampal neurons (11) and the PKA-mediated potentiation of skeletal muscle L-type Ca 2ϩ channels (12).
In addition to PKA, targeting of the phosphodiesterases (PDEs), the enzymes that degrade intracellular cAMP, may play an important role in compartmentalization of the cAMP signal and of phosphorylation. At present there is scant evidence on the mechanism and significance of subcellular localization of PDEs. Of the enzymes that specifically hydrolyze cAMP, forms that are derived from the PDE4A gene have been shown to be either membrane-bound (13,14) or particulate (15) and are localized in discrete structures of the olfactory neurons (16,17). In view of the finding that PDE inhibitors increase the diffusion of cAMP from the site of synthesis (4), PDEs may control local cAMP levels, the access of cAMP to the regulatory subunit of PKA, and therefore, the state of activation of PKAs.
The PDE4D form has been implicated in homeostatic mechanisms regulating cAMP levels in hormone-targeted cells (18 -21). A splicing variant derived from the PDE4D gene, PDE4D3, is the most abundant form present in quiescent FRTL-5 cells; this form is a substrate for PKA and is rapidly activated by TSH through phosphorylation (18,22,23). The expression of an additional variant, PDE4D2, is induced by an increase in in-tracellular cAMP, and the accumulation of this enzyme contributes to cell desensitization (22). Although not studied in these thyroid cells, the existence of other PDE4D variants has been demonstrated both by cloning and Western blot analysis in brain (24 -26). 2 Here we have addressed the question of the subcellular localization of different forms derived from the PDE4D gene. We examined the distribution pattern of the PDE4D cAMP-PDEs in quiescent FRTL-5 cells using biochemical and immunocytochemical approaches. Furthermore, we have compared the pattern of localization after short and long term stimulation with TSH. The data generated show that TSH regulates, in a temporally and spatially dependent manner, the PDE activity expressed in these thyroid cells.

EXPERIMENTAL PROCEDURES
Materials-Coon's modified F-12 medium, bovine insulin, human transferrin, culture-grade bovine TSH, and Crotalus atrox snake venom were purchased from Sigma. Purified bovine TSH (Lot AFP-5555B) for the stimulation of FRTL-5 cells was obtained from the National Hormone and Pituitary Program, NIDDK, NIH; Pansorbin cells were purchased from Calbiochem; Immobilon was from Millipore Corp. (Bedford, MA); [2, H]cAMP (20 -50 Ci/mmol) was from NEN Life Science Products; AG 1-X8 resin was from Bio-Rad; and ECL Western blot detection kit was from Amersham Pharmacia Biotech. Rolipram was provided by Schering AG. Fluorescein-conjugated antimouse IgGs and Vectashieldmounting medium were purchased from Vector Laboratories, Inc. (Burlingame, CA). Unless otherwise noted, all chemicals were the purest grade available from Sigma.
Preparation of Cell Extracts-FRTL-5 cells were cultured in Coon's F-12 medium supplemented with 5% calf serum, TSH (1 milliunit/ml), insulin (10 mg/ml), and transferrin (5 mg/ml). Cells were routinely cultured in 75-cm 2 flasks (Corning) at 37°C in an atmosphere of 95% air, 5% CO 2 in a humidified incubator. For Western blot and enzymatic analyses, FRTL-5 cells were seeded in 90-mm dishes (Corning) containing the above-described Coon's F-12 medium. Five to seven days later, when 80 -90% confluent, the cells were rinsed twice with Hanks' balanced salt solution and cultured for an additional 24 h in Coon's F-12 medium containing 0.1% bovine serum albumin and transferrin (5 mg/ml) to induce quiescence. When required by experimental protocol, the cells were stimulated with 10 nM bovine TSH in quiescence medium and incubated in a humidified incubator for the indicated period of time; at the end of the treatment cells were placed on ice, washed quickly with cold PBS, and harvested following the procedure described below.
PDE Assay-PDE activity was assayed using 1 M cAMP as substrate according to the method of Thompson et al. (27). Samples were assayed at 34°C in a final volume of 200 l; the solution consisted of 40 mM Tris-HCl, pH 8.0, 10 mM MgCl 2 , 5 mM 2-mercaptoethanol, 0.1 mg/ml bovine serum albumin, and 1 M [ 3 H]cAMP (approximately 0.1 ϫ 10 6 cpm/tube; 30 Ci/mmol). To identify the cAMP-specific phosphodiesterase activity, rolipram, a specific inhibitor of the cAMP-PDEs (28), was added to the incubation mixture at a final concentration of 10 M; the PDE activity measured in the presence of rolipram was subtracted from the total activity to quantitate the cAMP-PDE activity.
Lactic Dehydrogenase Assay-The lactic dehydrogenase activity was assayed following the protocol of Kornberg (29). In brief, lactic dehydrogenase activity was derived by measuring the oxidation of DPNH at 340 nm wavelength during the reduction of pyruvate to lactate. Optical density values were determined at 30 s intervals for 3 min, and the rate of reduction was calculated.
Antiserum Generation-The generation of the polyclonal PDE4 antibody and of the monoclonal antibodies specific for PDE4D has been previously described (21,25,30,31). The monoclonal antibody F34-8F4 specific for the PDE4D was a generous gift from Dr. John Cheng (Pfizer Inc, Groton, CT). This antibody cross-reacted with the recombinant PDE4D variants but not with PDE4A or PDE4B proteins (data not shown).
Ion Exchange Chromatography of the Soluble and Particulate PDE4D3-Quiescent FRTL-5 cells were stimulated with 10 nM TSH for 15 min as described previously. The cells were homogenized in isotonic buffer containing 10 mM Mg 2ϩ and centrifuged at 20,000 ϫ g at 4°C for 15 min, yielding a soluble and particulate fraction. The particulate fraction was washed twice and resuspended in an equal volume of hypotonic buffer containing 1 mM EDTA for 10 min on ice. The fraction was then centrifuged at 100,000 ϫ g for 15 min, and the supernatant was diluted to a final concentration of 200 mM sodium acetate, pH 6.5. The sample was then loaded onto a DEAE ion exchange HPLC column (Waters 5PW ion exchange column, 7.5 ϫ 0.75 cm) equilibrated with 200 mM sodium acetate, pH 6.5, at a flow rate of 1 ml/min. The buffer contained 50 mM NaF, 1 mM EDTA, 0.2 mM EGTA, 5 mM ␤-mercaptoethanol, and protease inhibitors at the same concentration used for the homogenization buffer. After washing the column with 5 volumes of the same buffer, bound proteins were eluted with a linear gradient of 200 -750 mM sodium acetate, pH 6.5. An aliquot of each fraction was used for the PDE activity assay; the remainder was precipitated by 5% trichloroacetic acid solution in H 2 O and resuspended in 25 l of 1 ϫ Laemmli sample buffer (32) for SDS-PAGE and Western blot analysis. The soluble fraction was subjected to DEAE ion exchange chromatography following an identical procedure.
Immunofluorescence-FRTL-5 cells were sparsely seeded in 8-well glass culture slides (Nunc Inc. Naperville, IL) and maintained for 3 to 4 days in regular, supplemented Coon's F-12 medium (described above). The cells were made quiescent for 24 h following the procedure already described. For TSH stimulation studies, cells were incubated with 10 nM bovine TSH in Coon's quiescence medium for either 15 min or 24 h. The cells were then washed three times with PBS and fixed in 4% paraformaldehyde at room temperature for 10 min. The cells were washed three times in PBS, permeabilized with 0.2% Triton X-100 in PBS for 30 min, then incubated in 3% normal goat serum in PBS for 1 h as a blocking step. After blocking, the cells were incubated overnight at 4°C in a humidified chamber with the primary antibody or the primary antibody preabsorbed with the antigen. The following day, the cells were washed three times in PBS and incubated with fluorescein-conjugated antimouse IgGs (1:400) for 1 h at room temperature. Cells were then washed multiple times (Ͼ5) in PBS and cover-slipped using Vectashield mounting medium.
In some experiments, the soluble and solublilized particulate fractions were immunoprecipitated using fixed Staphylococcus aureus cells (Pansorbin) according to the method of McPhee with minor modifications (33). Samples were immunoprecipitated by Pansorbin previously incubated with normal rabbit sera, anti-PDE4 antisera, or antisera specific for PDE4A. The bound PDE was eluted from the pellet by incubation with 1% SDS in PBS at room temperature.
Eluted samples were diluted in 4ϫ sample buffer. The samples were boiled for 5 min, and proteins were separated on an 8% SDS-polyacrylamide gel. The proteins were transferred to an Immobilon membrane, and nonspecific binding sites were blocked by incubating the membrane overnight in 10% bovine serum albumin (w/v) dissolved in TBS-T solution (0.2% Tween 20, 20 mM Tris-HCl, 14 mM NaCl, pH 7.6). The following day the membrane was incubated in antisera in TBS-T for 1 h at the following dilutions: K116, 1:2,000 (v/v); M3S1, 1:250 (v/v); F34-8F4, 1:1,700 (v/v). After incubation with the primary antibody, the membranes were washed extensively for 1. in TBS-T for at least 1.5 h, bound antibodies were detected using a luminescence method (ECL, Amersham) or a Bio-Rad Immunostar detection kit and recorded after exposure to XAR-5 x-ray film. Exposure time varied between 30 and 60 s.

Fractionation of cAMP-PDE Activity Present in FRTL-5 Cell
Extracts by Differential Centrifugation-When homogenates prepared with an isotonic buffer from quiescent FRTL-5 cells were fractionated by differential centrifugation, rolipram-sensitive PDE activity was recovered in both soluble and particulate fractions (Table I). After a 20,000 ϫ g centrifugation, 65% of the PDE activity was recovered in the supernatant and 35% in the pellet fraction of these cells (Table I). Further centrifugation of the 20,000 ϫ g supernatant at 100,000 ϫ g for 1 h demonstrated that negligible amounts of rolipram-sensitive PDE activity is recovered in the high speed microsomal pellet, and the remaining activity is in the soluble fraction (Table I).
The possibility of contamination of the 20,000 ϫ g pellet fraction with soluble proteins was determined by assaying the soluble enzyme lactate dehydrogenase. Minimal contamination (less than 5%) of the pellet fraction by lactic dehydrogenase was observed (see below); therefore, the PDE activity recovered in the pellet fraction could not be accounted for by incomplete fractionation or entrapment of soluble proteins in vesicles during homogenization. This subcellular distribution of the PDE activity was minimally affected when hypotonic homogenization buffers were used (data not shown). Since less than 5% of the PDE activity was recovered in the 100,000 ϫ g pellet, all the following experiments were conducted on the 20,000 ϫ g supernatant and pellet fractions.
The particulate PDE activity had kinetic properties similar to those of the soluble enzyme with an apparent K m for cAMP of 1 M. Rolipram inhibited the PDE activity of the particulate fraction with an IC 50 of 0.39 Ϯ 0.2 M, whereas the soluble activity was inhibited with an IC 50 of 0.048 Ϯ 0.03 M (mean Ϯ S.E. n ϭ 3). Differences in rolipram sensitivity between the soluble and particulate cAMP-PDEs have been previously reported (15,34).
Subcellular Localization of the PDE4D Variants in Quiescent FRTL-5 Cells-Previous mRNA and protein analysis from our laboratory had shown that the PDE4D3 variant is the predominant form in the soluble fraction of quiescent FRTL-5 cells (18). To investigate the properties of the cAMP-PDEs present in the particulate fraction of these cells, the 20,000 ϫ g pellet and supernatant were fractionated by SDS-PAGE, and the PDEs were detected by Western blot with nonselective polyclonal antibodies (K116) or antibodies selective for PDE4D (M3S1 and F34-8F4). An immunoreactive polypeptide of 93 Ϯ 1.0 kDa was detected in both the soluble and particulate frac-tions with the three antibodies ( Fig. 1), and the migration of this protein was identical to that of the recombinant PDE4D3 (data not shown). Using densitometric analysis of the immunoreactive bands it was estimated that 73.6 Ϯ 8.5% (n ϭ 5) and 26.4 Ϯ 8.5% (n ϭ 5) of the PDE4D3 polypeptide was recovered in the soluble and particulate fractions, respectively. Similar results were obtained when the soluble fraction was immunoprecipitated with the K116 antibody and analyzed by Western blot using the PDE4D-specific antibody M3S1 (Fig. 2). An additional immunoreactive polypeptide of 105 kDa was observed in the particulate fraction of FRTL-5. Although the monoclonal antibodies gave a stronger signal (Fig. 1), this polypeptide cross-reacted with all three antibodies and had an electrophoretic mobility identical to that of the recombinant PDE4D4 (data not shown). This 105-kDa polypeptide could not be detected after extraction with Triton and immunoprecipitation of the particulate fraction of FRTL-5 cells (Fig. 2). A densitometric analysis indicated that the 105-kDa polypeptide is more than 90% particulate under these homogenization conditions. Some variability was observed in the relative abundance of PDE4D3 and PDE4D4. Although the differences in antibody affinity precludes exact quantitations, in some experiments PDE4D4 was more abundant than PDE4D3, whereas in others PDE4D3 was more prominent. No systematic studies were performed to relate PDE4D4 expression to the culture conditions.
The nonselective polyclonal antibody K116 recognized an additional polypeptide of 74 kDa present mostly in the soluble fraction of FRTL-5 cells. Although this protein may correspond to a PDE4 other than PDE4D isoenzymes, its identity was not further investigated.
Previous analysis of the PDE4D mRNAs expressed in FRTL-5 cells had indicated the presence of PDE4D1, PDE4D2, and PDE4D3 variants (22). To determine whether a PDE4D4 mRNA is also expressed in FRTL-5 cells, reverse transcription polymerase chain reaction was used to analyze the mRNA from quiescent FRTL-5 cells. A polymerase chain reaction product of the correct size hybridizing to a specific probe was obtained with primers specific for the 5Ј end of the PDE4D4 variant (data not shown). Thus, the mRNA analysis confirmed that the PDE4D4 variant is expressed in FRTL-5 cells.
Effect of Different Homogenization and Solubilization Conditions on the Particulate PDE4D-The PDE4D3 polypeptide TABLE I PDE activity recovered in the fractions from differential centrifugation of the extract from quiescent FRTL-5 cells Quiescent FRTL-5 cells were homogenized in isotonic buffer with Mg 2ϩ and centrifuged at 20,000 ϫ g for 15 min at 4°C, yielding a 20,000 ϫ g supernatant and pellet. The supernatant was centrifuged again at 100,000 ϫ g for 1 h at 4°C, yielding a 100,000 ϫ g supernatant and pellet. PDE activity was assayed as described under "Experimental Procedures" and is expressed as pmol/min/plate of cells. recovered in the particulate fraction was readily solubilized with Triton X-100 and chelating agents and could be immunoprecipitated with PDE4-nonselective and PDE4D-selective antibodies ( Fig. 2A). Conversely, only traces of the PDE4D4 form were recovered after solubilization with nonionic detergents (Fig. 2A). Only when the cells were homogenized in RIPA buffer containing 0.5% SDS could most of the PDE4D4 be recovered in the soluble fraction (Fig. 2B). However, small amounts of PDE4D4 remained in the pellet even after extraction in RIPA buffer (Fig. 2B).
It was also observed that a portion of the PDE activity recovered in the pellet could be released by treatment with EDTA in a hypotonic buffer. This released cAMP-PDE activity was further analyzed by immunoprecipitation or ion exchange chromatography. Only the PDE4D-selective antibody (M3S1) and the PDE4nonselective antibody (K116) immunoprecipitated the PDE activity solubilized with this procedure (data not shown). Upon DEAE ion exchange chromatography, the PDE activity migrated as a single major peak with a retention time identical to that of the soluble cAMP-PDE (Fig. 3). Western blot analysis of the fractions containing the peak of activity showed the presence of an immunoreactive protein of 93 kDa (Fig. 3). Thus, these data demonstrate that the solubilized PDE is the PDE4D3 form.
The Effect of Short and Long Term TSH Stimulation on cAMP-PDE Localization-We have previously shown that two modes of PDE regulation are activated upon stimulation of FRTL-5 cells with TSH (18,22,23). A short term phosphorylation of PDE4D3 is followed by a long term accumulation of the PDE4D2 variant. In view of the presence of both soluble and particulate PDE4D forms, the PDE activation in the soluble and particulate fractions after short term TSH stimulation was further investigated (Fig. 4). After 0, 2.5, 5, 10, and 15 min of TSH stimulation, cells were harvested, and the homogenate was separated into soluble and particulate fractions. The PDE activity in the soluble fraction increased at 1-2 min of TSH stimulation and reached a maximum after 5 min, and the activity declined thereafter. The particulate PDE activity was increased with some delay, reaching a maximum at 10 min of TSH stimulation (Fig. 4).
Previous data from our laboratory had indicated that the TSH stimulation of PDE correlates with the PKA-dependent phosphorylation of PDE4D3 (23). This phosphorylation is associated with a shift in the mobility of the PDE4D3 polypeptide. To determine whether the time course of phosphorylation of PDE4D3 is compatible with the activation of the cAMP-PDE in the two compartments, the mobility of PDE4D3 was studied at different times after TSH stimulation. The results of this Western blot analysis are reported in Fig. 5. A shift in the mobility of the soluble PDE4D3 polypeptide was observed as early as 2.5 min after TSH stimulation and was still present after 15 min. The mobility of the particulate PDE4D3 was also affected by the TSH stimulation, although some delay in the stimulation was again apparent.
To determine the effect of TSH on the long term activation of the PDE4D variants present in the soluble and particulate fraction, PDE activity was measured after 24-h stimulation with TSH, and these figures were compared with the data obtained after 15-min stimulation. Consistent with the previous findings, an increase in PDE activity in the soluble fraction could be observed after 15 min (Fig. 6, top panel). A 24-h incubation with hormones caused an additional increase in PDE activity in the soluble fraction of the cell. The PDE activity in the particulate fraction increased approximately 2-fold after a 15-min TSH stimulation, but no significant further increase was observed after 24 h (Fig. 6, top panel). This pattern of activation following TSH stimulation is due to the cAMP-PDE family of PDEs, since the distribution of rolipraminsensitive PDE activity did not change significantly (Fig. 6,  middle panel). In addition, the difference in recovery of the PDE in the two fractions is not due to inadequate fractionation procedures, since less than 4% of soluble lactic dehydrogenase activity could be observed in the pellet and remained constant under the different experimental conditions (Fig. 6, bottom  panel). Western blot analysis was performed on the 20,000 ϫ g pellet and supernatant 24 h after TSH stimulation (Fig. 7). As for the PDE activity studies, groups of quiescent cells or cells treated for 15 min with TSH were used as a control. After 24 h of stimulation with hormone, a polypeptide of 68 kDa appeared in the soluble fraction of the cells (Fig. 7). This polypeptide cross-reacted with all three antibodies (data not shown) and had a mobility identical to that of the recombinant PDE4D2 (data not shown). Confirming our previous observation, a shift in the mobility of PDE4D3 was evident in both the soluble and particulate fraction at 15 min and 24 h after hormone stimulation (Fig. 7). Although the mobility shift was complete after 15 min, the PDE4D3 migrated as a doublet after a 24-h stimulation (Fig. 7). Furthermore, an increase in the intensity of the PDE4D3 polypeptide was sometimes observed (Fig. 7). Interestingly, no clear shift in mobility could be observed for the PDE4D4 variant. Similar results were obtained when the PDE4-nonselective polyclonal antibody was used as a primary antibody (data not shown).
Immunofluorescence in FRTL-5 Cells-The fluorescent immunocytochemistry of intact, quiescent FRTL-5 cells demonstrated a nonrandom distribution of PDE4D (Fig. 8). A diffuse fluorescence throughout the cytoplasm of the cells was present, and no specific fluorescence was evident in the nucleus of the cells. Strong fluorescence was also localized in a region adjacent to the plasma membrane in filamentous structures and in a perinuclear region (panel A), which was blocked when the anti- body was preabsorbed with the fusion protein (panel B). The staining around the perinuclear region was asymmetrical, suggesting a staining of the Golgi apparatus or the cell-organizing center. The observed filamentous structures do not correspond to microtubules, since staining with antitubulin antibodies gave a different pattern of staining (data not shown). These structures may represent other cytoskeletal components of the cell.
After 15 min of TSH stimulation, there was no major change in the pattern of immunofluorescence, although a reinforcement of the staining around the nucleus and in the centrosomal region could be observed (panels C and D). However, after 24 h of TSH stimulation, a significant increase in the diffuse staining throughout the cytoplasm was observed in all experiments performed (panels E and F). Experiments performed with a different antibody, K116, showed a similar increase in diffuse cytoplasmic staining 24 h after hormone stimulation (data not shown).
Several controls were performed to determine the specificity of the staining. Under the conditions used, the secondary antibody did not show appreciable staining. As previously mentioned, the pattern of staining was blocked when the monoclonal antibody was preadsorbed to the fusion protein (Fig. 8). Instead, the staining was not blocked when an unrelated protein or GST was used for the preadsorption (data not shown). When antibodies against PDE4A and PDE4B were used, a nonspecific or different pattern of staining was observed (data not shown).

DISCUSSION
The PDE4D gene contains at least four transcriptional units that encode proteins composed of a common catalytic domain but different regulatory domains (26,(35)(36)(37). The long forms, including the PDE4D3, PDE4D4, and PDE4D5 variants, possess phosphorylation sites at the amino terminus absent in the short forms, PDE4D1 and PDE4D2 (19,22). Although hormones cause a short term activation of the long PDE4D3 form, the de novo synthesis of PDE4D1/PDE4D2 variants follow a sustained increase in cAMP in the cell. This long term accumulation is the result of a cAMP-dependent activation of transcription from an intronic promoter (38), with consequent accumulation of PDE4D mRNA and protein (21,39). These modes of regulation of the short and long forms of PDE4D have provided the first explanation for the presence of different splicing variants (36). The data reported in the present study identify an additional property that distinguishes short and long splic-ing variants of PDE4D and that further explains the existence of these variants. Two long forms, PDE4D3 and PDE4D4, are recovered in both the soluble and particulate fractions of FRTL-5 cells. After short term incubation with the hormone, TSH causes the preferential activation of the PDE4D3 enzymes in the soluble and in the particulate fraction. Conversely, the PDE4D2 variant, which accumulates after long term treatments with TSH, is mostly soluble. Thus, the PDE4D gene produces splicing forms that are activated at different times in different compartments of the cell during hormonal action.
Our data indicate that variants derived for the PDE4D gene are present in both the soluble and particulate fraction of FRTL-5 in quiescent cells. Although the soluble activity is mostly due to the PDE4D3 variant, both PDE4D3 and PDE4D4 contribute to the activity recovered in the particulate fraction of FRTL-5 cells. That these forms are bound to particulate structures is also indicated by the immunofluorescence analysis of quiescent FRTL-5 cells. Two antibodies that recognize PDE4D show staining of the cortical region below the plasma membrane, the Golgi/centrosome and the filamentous structures. Similar staining of a region below the plasma membrane has been shown when a PDE4A4 product of the human PDE4A gene is transiently expressed in COS7 cells (15). Thus, our data on native PDE4D forms lend further support to the view that some PDE isoforms are present on cytoskeletal structures in the cortical region of the cell.
It is probable that PDE4D3 and PDE4D4 are present in different compartments and/or interact with particulate structures via distinct mechanisms. Although PDE4D3 is readily solubilized by Triton X-100 and chelating agents, PDE4D4 is mostly solubilized only after treatment of the particulate fraction with ionic detergents. A possible explanation of this divergent behavior of the two forms is that PDE4D3 is interacting predominantly with membranous structures, whereas PDE4D4 binds preferentially cytoskeletal structures that are solubilized poorly by Triton X-100. Both localizations were observed by immunofluorescence, since the monoclonal antibody used does not distinguish between PDE4D3 and PDE4D4. It is interesting to note that the polyclonal antibody K116 stained the Golgi/centrosome structures but gave no clear cytoskeletal pattern of staining. Since this antibody does not recognize PDE4D4 well in Western blot analysis, the incomplete overlap with the monoclonal staining would be consistent with the view that PDE4D3 is the membrane-bound form. Form-selective antibodies will be required to further clarify the exact localization of PDE4D3 and PDE4D4.
After short term stimulation with TSH, an increase in rolipram-sensitive PDE activity was detected in both the soluble and particulate fractions. That PDE4D3 is the target for the activation is indicated by the shift in electrophoretic mobility of this variant in both fractions. Interestingly, although the shift in mobility was readily apparent for PDE4D3, no change in mobility could be observed with the PDE4D4 polypeptide. This observation may indicate that PDE4D4 is either not a good target or is not accessible to the catalytic subunit of PKA. In line with this view is the observation that the PDE activated by TSH in the particulate fraction could be readily solubilized by Triton X-100 or metal chelators, whereas PDE4D4 is not readily solubilized by these treatments. Further studies are nevertheless required to clarify whether PDE4D4 can be activated by PKA, since the putative PKA phosphorylation sites present in PDE4D3 are also present in PDE4D4 (24).
The long term stimulation of the FRTL-5 cells with TSH provides the unique opportunity to compare the subcellular localization of native long and short forms of PDE4D within the same cell while avoiding potential artifacts associated with overexpression of a protein. The PDE4D2 variant, whose accumulation is induced by long term hormonal treatment, is recovered predominantly in the soluble compartment of the FRTL-5 cell. This conclusion is supported by the long term increase in PDE activity in the soluble fraction, by the Western blot analysis where the 68-kDa protein is mostly soluble, and by the immunofluorescence data. In addition, the recovery of PDE4D2 in the soluble fraction and of PDE4D4 in the particulate fraction rules out the possibility of cross-contamination during cell fractionation.
The PDE4D2 variant is a naturally occurring truncated form of PDE4D3 and PDE4D4 (36). On the basis of the structural differences of these variants and their subcellular distribution, it is tempting to speculate that the targeting of PDE4D3 and PDE4D4 is mediated by domains present at the amino terminus of these proteins, domains that are absent in the soluble PDE4D2. It is interesting to note that the recovery of this native short form in the cytosol is consistent with the observation of Houslay and co-workers (40) that an engineered Met-26 form of PDE4A1 is exclusively soluble when expressed in COS cells. Also consistent with the present data is the observation that PDE4D3 and PDE4D4 are recovered in the particulate fraction of the cortex and cerebellum (25).
In conclusion, our data demonstrate that hormones such as TSH have the ability to control cAMP hydrolysis in target cells in a temporally and spatially dependent manner. Short term stimulation causes the activation of distinct pools of PDE4D3 followed by a delayed accumulation of a PDE4D2 in a soluble compartment. Other signals may control the activity of PDE4D4, which is most likely found in a different compartment. Although their exact physiological significance needs further investigation, these findings are consistent with the concept that the diffusion of the cAMP signal is tightly controlled in the cell. Activation of PDEs in distinct subcellular compartments supports the idea that cAMP concentration can be differentially regulated in different regions of the cell. This, in turn, affects the state of activation of PKA and thereby controls the phosphorylation of the substrate in a spatially determined manner.