Functions of the N-terminal region of cyclic nucleotide phosphodiesterase 3 (PDE 3) isoforms.

The N-terminal portion of phosphodiesterase (PDE) 3 was arbitrarily divided into region 1 (amino acids 1-300), which contains a large hydrophobic domain with six predicted transmembrane helices, and region 2 (amino acids 301-500), with a smaller hydrophobic domain ( approximately 50 residues). To analyze these regions, full-length human (H)PDE3A and mouse (M)PDE3B and a series of N-terminal truncated mutants were synthesized in Sf9 cells. Activities of HPDE3A, H3A-Delta189, MPDE3B, and M3B-Delta196, which retained all or part of the hydrophobic domain in region 1, were recovered almost entirely in particulate fractions. H3A-Delta321 and M3B-Delta302, containing region 2, were recovered essentially equally in particulate and cytosolic fractions. H3A-Delta397 and H3A-Delta457, lacking both hydrophobic domains, were predominantly cytosolic. H3A-Delta510 and M3B-Delta604, lacking both regions 1 and 2, were virtually completely cytosolic. M3B-Delta196 eluted as a large aggregated complex during gel filtration. With removal of greater amounts of N-terminal sequence, aggregation of PDE3 decreased, and H3A-Delta607, H3A-Delta721, and M3B-Delta604 eluted as dimers. Truncated HPDE3A proteins were more sensitive than full-length HPDE3A to inhibition by lixazinone. These results suggest that the hydrophobic domains in regions 1 and 2 contain structural determinants important for association of PDE3 with intracellular membranes, as well for self-association or aggregation during gel filtration and sensitivity to a specific inhibitor.

Regulation of the intracellular concentration of cAMP is achieved through control of its synthesis by adenylyl cyclases and its degradation by cyclic nucleotide phosphodiesterases (PDEs). 1 Ten different, but structurally related, PDE gene families (PDE1-10) have been identified (1)(2)(3)(4). PDE3 isoforms are characterized by their high affinities for both cAMP and cGMP as well as their sensitivity to cilostamide and other drugs that increase myocardial contractility, inhibit platelet aggregation, and relax airway and vascular smooth muscle (5). The two PDE3 isoforms, PDE3A and PDE3B, are products of distinct but related genes and are differentially expressed and regulated in a variety of cells and tissues. PDE3B, for example, is found in adipocytes and pancreatic ␤-cells; PDE3A, in platelets; and PDE3A and PDE3B in vascular smooth muscle (1, 6 -13). All mammalian PDEs have a similar structural organization with a conserved catalytic domain in the C-terminal half of the molecules and divergent N-terminal regulatory domains. These N-terminal regions contain structural elements that confer the specific regulatory characteristics of the different PDE families, such as the calmodulin-binding domain in PDE1, two cyclic nucleotide-binding regions in PDE2 and membrane-targeting domains in PDE4 (1). Although there is little sequence identity in the deduced N-terminal portions of PDE3A and PDE3B isoforms (7,11), both contain hydrophobic membraneassociation domains of ϳ200 aa (from amino acids ϳ50 -250) with six predicted helical transmembrane segments, followed by several consensus sequences (RRXS) for phosphorylation by protein kinase A and some additional hydrophobic sequences (from ϳaa 300 -400). In intact rat adipocytes, phosphorylation of Ser-302 of PDE3B in response to insulin and agents that increase cAMP is associated with activation of the enzyme (14). PDE3 isoforms are found in different intracellular locations, being predominantly membrane-associated in adipocytes (1,(15)(16)(17), cytosolic in platelets (9,10), and cytosolic as well as associated with the sarcoplasmic reticulum in myocardium (18 -20). Recently, we showed that activities of full-length PDE3A and PDE3B mutants were recovered predominantly in particulate fractions of NIH-3006 fibroblasts and Sf9 insect cells, whereas N-terminal-truncated PDE3A mutants were cytosolic, suggesting that the N-terminal portion of PDE3 isoforms might be important in their association with cellular membranes (21). This study was designed to define the role of the N-terminal hydrophobic domains in the targeting of PDE3 isoforms to membrane structures and to characterize other possible effects of those regions on the self-association of PDE3 isoforms and their sensitivity to specific PDE3 inhibitors.

EXPERIMENTAL PROCEDURES
Construction of MPDE3B N-terminal Deletion Mutants-An ϳ3.6-kb cDNA fragment that had been cloned into the XhoI site of pBluescript and contained the full-length open reading frame of mouse (M)PDE3B 2 was used as template to generate mutants lacking different amounts of N-terminal sequence. The presence of unique AlwnI and DraIII restriction sites in MPDE3B cDNA was utilized in production of the Nterminal truncations. To prepare M3B-⌬196 and M3B-⌬302 constructs (lacking, respectively, 196 and 302 aa), a common reverse primer, 5Ј-ACATTACCGCTCGAGAGAGAGCCAGCAGACACTGGTACA (corresponding to nt 1523-1500), was used. This primer contains the unique * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Western Immunoblots-Proteins in Sf9 cell lysates were separated by SDS-polyacrylamide gel electrophoresis in 10% gels (120 min, 100 V) and transferred to nitrocellulose membranes (90 min, 100 V) which were incubated overnight with TNA buffer (0.15 M NaCl, 10 mM Tris-HCl, pH 7.4) containing 50 mg/ml BSA and 5 mg/ml ovalbumin. Membranes were then incubated for 2 h at room temperature in TNA buffer containing 10 mg/ml BSA and anti-platelet cGI-PDE antibodies (10) for HPDE3A mutants, or, for MPDE3B mutants, either anti-rat PDE3B C terminus antibodies (21) or anti-FLAG M2 monoclonal antibodies (IB13010, Kodak). Membranes were washed once with TNA containing 0.05% Nonidet P-40 and three times (5 min each) with phosphatebuffered saline. Membranes were incubated for 1 h with TNA buffer containing 10 mg/ml BSA and horseradish peroxidase-conjugated goat anti-rabbit IgG or mouse anti-rabbit IgG (Promega), then washed twice with TNA plus 0.05% Nonidet P-40, three times with phosphate-buffered saline, and evaluated by chemiluminescence (Amersham Pharmacia Biotech).
Inhibition Studies-For IC 50 determinations, lysates of Sf9 cells containing different recombinant proteins were assayed as described above (with 0.1 M [ 3 H]cAMP) without or with 5-6 concentrations of the indicated inhibitors, including lixazinone, cilostamide, trequinsin, and rolipram (final concentration of 0.05% dimethyl sulfoxide (v/v) in all samples). For some experiments, 1% Nonidet P-40 was added to the Sf9 cell lysates for 30 min before homogenization and centrifugation (5 min, 10,000 rpm). Activity in the absence of inhibitors was taken as 100%, and IC 50 values were calculated from graphical analysis of inhibition curves.
Protein Assay-Protein content was determined by the Bio-Rad protein assay kit or the bicinchoninic acid Protein Assay Reagent Kit (Pierce) for experiments in which detergents were present, with BSA as standard.
To assess whether the association of PDE3 recombinants with particulate fractions exhibited characteristics of integral or peripheral proteins, Sf9 cells were disrupted in buffer containing 0.5 M NaBr (Fig. 4, B and E) or 0.5 M NaBr plus 1% C 13 E 12 (Fig. 4, C and F). Very little of the activity of HPDE3A, H3A-⌬189, MPDE3B, or M3B-⌬196 was solubilized in the presence of salt alone (Fig. 4, B and E). Particulate H3A-⌬321 and M3B-⌬302, however, were more readily solubilized by NaBr alone than were the other recombinants (Fig. 4, B and E). When cells were solubilized with 0.5 M NaBr and 1% C 13 E 12 (Fig. 4, C and F), activities of the full-length and deletion recombinants were recovered almost completely in soluble fractions (70 -100% of total activity). These results are consistent with the idea that the transmembrane segments in region 1 are important for strong association of PDE3 with, or insertion into, membranes. Recombinants that contained all or part of the hydrophobic domain in region 1 were, like integral proteins, solubilized with salt plus detergent. Region 2 does contain structural determinants that allow targeting to, or association with, membranes but this association reflects that of a peripheral, non-integral membrane protein and is disrupted by homogenization and salt extraction. The small hydrophobic domain in region 2 is most likely primarily responsible for the salt-sensitive association of H3A-⌬321 and M3B-⌬302 with membranes (Fig. 4, B and E). Activities of H3A-⌬397 and H3A-⌬457 (which lack this domain) are predominantly (ϳ70 -80%) cytosolic and the particulate activities of these mutants are little affected by detergent and/or salt (Fig. 4, B, C, E, and F). Except for H3A-⌬510 and M3B-⌬604, some activity of the other recombinants was resistant to extraction with detergent and/or salt (Fig. 4, B, C, E, and F). Whether this represents an artifact of overexpression, retention in a subcellular organelle, or a different type of association with particulate structures is not known. Removal of an additional ϳ50 aa from H3A-⌬457, i.e. in H3A-⌬510, disrupts this interaction. After removal of both regions 1 and 2, i.e. in H3A-⌬510 and M3B-⌬604, PDE3 isoforms are virtually completely cytosolic.
Ultrogel AcA34 and TSK-G3000SW HPLC Size Exclusion Chromatography-This laboratory has previously reported (21) that rat (R) recombinant PDE3B and native rat adipocyte PDE3B eluted at or close to the void volume of Sephacryl S-300. M3B-⌬196 activity also was associated with highly aggregated complexes that eluted as a single peak near the void volume of Interestingly, M3B-⌬288 (not shown) and M3B-⌬302 (Fig. 7E) eluted as double peaks; the higher molecular weight material eluted near the void volume. As shown in Fig. 8, H3A-⌬607, H3A-⌬721, and M3B-⌬604 (not shown) eluted in positions consistent with dimers; other recombinants eluted with M r values much greater than those predicted for homodimers calculated from their deduced sequences (Fig. 8B). DISCUSSION As reported (11,21,22,(25)(26)(27), PDE3 mutants with Nterminal deletions were catalytically active and sensitive to inhibition by specific PDE3 inhibitors, confirming that the N termini of PDE3A and PDE3B are not required for catalytic activity. HPDE3A mutants from which 608 or 679 N-terminal aa had been removed were active when expressed in yeast (22,25) or Escherichia coli (11). Although H3A-⌬686 (27) and H3A-⌬721 (22) were reported to be inactive when expressed in yeast, H3A-⌬721 was active (and was inhibited by cilostamide and lixazinone) in Sf9 cells, but with specific activity much lower than those of all other mutants. These results suggest that the earlier reports by others might indicate that the region between aa 680 and 721 contributes significantly to (but is not absolutely essential for) catalytic activity. Specific activities of the smaller PDE3 proteins (H3A-⌬510, H3A-⌬607, and M3B-⌬604) appeared to be greater than those of the full-length molecules. This could be related to differences in levels of expression or proportion of active enzyme. It is also possible that sequences in the N-terminal region are directly autoinhibitory or inhibit PDE3 activity indirectly through interactions with other proteins. Others have suggested that deletion of sequences in the N-terminal domain increases activity of PDE3A (22,25) and PDE4 (28 -31), but it is not certain that the N-terminal region of PDE3 isoforms contains an autoinhibitory domain.

FIG. 4. Subcellular distribution of HPDE3A (A-C) or MPDE3B (D-F) recombinants in Sf9 cells.
Sf9 cells were harvested, washed, and sonicated in 1 ml of homogenization buffer alone (A and D), plus 0.5 M NaBr (B and E) or plus 0.5 M NaBr and 1% C 13 E 12 (C and F) as described under "Experimental Procedures." After centrifugation (100,000 ϫ g, 60 min), PDE3 activity and protein content were measured in particulate and cytosolic fractions. Activities (picomoles of cAMP hydrolyzed/min) are expressed as a percentage of the sum of PDE3 activities (that activity inhibited by cilostamide) recovered in the two fractions. Data are mean Ϯ S.E. of values from three to four independent infections. Approximately 92-97% of total lysate PDE3 activities were recovered in the particulate and cytosolic fractions.

Functions of the N-terminal Region of PDE3 Isoforms
first to indicate that lixazinone, a known PDE3 inhibitor (32,33), is a more potent inhibitor of the short forms of PDE3A than of PDE3B. This observation may be of potential functional and clinical significance since PDE3A isoforms with M r values similar to that predicted for recombinant H3A-⌬189, i.e. "short" forms of PDE3A, are present in the cytosolic fractions of platelets and vascular smooth muscle (10,34,35). Although the 4-fold difference in IC 50 values for lixazinone inhibition of short forms of PDE3A and full-length PDE3B might not be sufficient to inhibit PDE3A selectively in intact cells or to be of therapeutic value, it does raise the possibility of designing more specific lixazinone-like PDE3A inhibitors. This could be of practical importance given the cytosolic location of PDE3A isoforms in platelets and vascular smooth muscle and the different cellular and tissue distributions of PDE3A and PDE3B, i.e. PDE3A in platelets, PDE3B in adipocytes, pancreatic ␤-cells, and lymphocytes (1), and both PDE3A and PDE3B in vascular smooth muscle (12,13). Detergent induced the same decrease in IC 50 for inhibition by lixazinone of particulate HPDE3A (Fig. 6) as did removal of the N-terminal region (Fig. 5A, Table II). This suggests that conformational changes brought about by insertion of PDE3 into, or association with, cellular membrane structures may be involved in the different sensitivities of full-length and trun-cated HPDE3A proteins to lixazinone. It is not known whether such conformational changes might alter binding of lixazinone to distinct sites on PDE3A, analogous to the high and low-   affinity binding sites for rolipram on PDE4 (36 -38). Others (25) have reported that endogenous (ϳ105-110-kDa) platelet PDE3A (10,34,35) was inhibited by lixazinone with an IC 50 value of 0.053 M, similar to that of the (106-kDa) H3A-⌬189 mutant. In that study (25), recombinant HPDE3A truncated mutants (lacking the first 561 or 661 aa) containing an Nterminal His 6 tag, however, exhibited higher IC 50 values for inhibition by lixazinone than did the endogenous platelet PDE3A (25) or the truncated HPDE3A mutants described in this report. The reason(s) for these differences is not clear. It is possible that the presence of multiple histidine residues at the N terminus, or expression in E. coli, resulted in an increase of the IC 50 value to that of the full-length HPDE3A expressed in Sf9 cells.
HPDE3A, H3A-⌬189, MPDE3B, and M3B-⌬196 (Fig. 4) and, as previously reported, rat (R)PDE3B (21) were recovered almost completely in particulate fractions of Sf9 cells. Thus, the removal of 3-4 putative transmembrane segments from region 1 of H3A-⌬189 or M3B-⌬196 did not alter their association with particulate fractions in Sf9 cells. The helical segments in region 1 are most likely responsible for insertion of PDE3 into, or strong association with, membranes, since recombinant proteins that contained transmembrane helical sequences (HPDE3A, H3A-⌬189, MPDE3B, M3B-⌬196) required salt and detergent for solubilization (Fig. 4). Complete deletion of region 1 of H3A-⌬321 and M3B-⌬302 resulted in the almost equal division of their activities between cytosolic and particulate fractions (Fig. 4). Recovery of a significant portion of the activity of H3A-⌬321 and M3B-⌬302 in particulate fractions of Sf9 cells suggests that either the hydrophobic domain (aa 340 -390 and 320 -370 in HPDE3A and MPDE3B, respectively) in region 2 or other targeting sequences in that region contribute to the interaction of H3A-⌬321 and M3B-⌬302 with Sf9 cell particulate fractions. Particulate H3A-⌬321 and M3B-⌬302, however, were almost completely solubilized by salt alone without detergent (Fig. 4), indicating that although sequences following aa 302 contain information for targeting to membranous structures, the molecules associated as non-integral membrane proteins.
The small hydrophobic domain may be primarily responsible for this salt-sensitive association, since activities of H3A-⌬397 and H3A-⌬457 (which lack this domain) are predominantly (70 -80%) cytosolic and their particulate activities are little affected by detergent and/or salt. Recently, intracellular localization of epitope-tagged PDE3 recombinants in COS-7 cells by immunofluorescence indicated that HPDE3A, H3A-⌬189, and MPDE3B were associated with the endoplasmic reticulum. H3A-⌬397 was located predominantly in the cytoplasm and the fraction of H3A-⌬397 that associated with the endoplasmic reticulum was perinuclear. The intracellular distribution of immunofluorescent H3A-⌬397 was quite different from that of H3A-⌬189 and M3B-⌬302, which, like HPDE3A, were distributed throughout the extended endoplasmic reticulum. 3 Thus, results of studies in Sf9 cells and COS-7 cells suggest that, although the hydrophobic domain in region 2, i.e. in H3A-⌬321 and M3B-⌬302, does contain information important for targeting, this association with membrane structures is readily disrupted by homogenization and/or salt extraction. Removal of this hydrophobic domain in region 2, however, does alter localization since H3A-⌬397 and H3A-457 do not anchor efficiently to membranes and are recovered primarily in the cytosol. It is not known whether the perinuclear localization of H3A-⌬397 in COS-7 cells is an artifact of overexpression, or if the sequence between aa 457 and 510, i.e. in H3A-⌬457 (Fig. 4), targets HPDE3A to this intracellular location. H3A-⌬510 and M3B-⌬604, which lack region 1 as well as region 2, were virtually completely cytosolic in Sf9 cells (Fig. 4). Results of our studies in Sf9 cells are consistent with the immunofluorescence studies of Shakur et al., 3 in which the helical segments in region 1 appeared to direct the protein to the endoplasmic reticulum and the removal of regions 1 and 2 allowed movement of epitope-tagged H3A-⌬510 and M3B-⌬604 into the cytoplasm of COS-7 cells.
During HPLC on TSK-G3000SW columns, PDE3 molecules containing transmembrane helical sequences, and/or the hydrophobic domain in region 2, i.e. M3B-⌬196 and M3B-⌬302, eluted as large aggregates. Removal of the hydrophobic domains in region 1 and 2, i.e. H3A-⌬397, reduced the sizes of these complexes. The functional significance of aggregated complexes formed by both native PDE3 and PDE3 recombinant proteins (21,39, this work), as well as PDE4B2B (40) and PDE4D1 (30) is unknown. Removal of larger amounts of the N-terminal portion of PDE3 changed the gel filtration behavior of the proteins and resulted in a significant change in the state of aggregation of the recombinants. M3B-⌬604, H3A-⌬607, and H3A-⌬721 did not elute as aggregated complexes, but as dimers. The structural determinants involved in dimer formation have not been identified. Our data suggest that the large N-terminal hydrophobic domain in region 1 plays a role in the insertion PDE3 into, or strong association with, cellular membranes and formation of large aggregates, while the smaller hydrophobic region (between aa 300 -400) and other sequences in region 2 contain information for targeting to and salt-sensitive association with membranes. These sequences induce formation of elongated (not globular) complexes, self-association of PDE3 molecules, or interactions of PDE3 recombinants with  (3) and Bio-Rad standards versus elution volume on HPLC gel filtration chromatography. The regression curve (r 2 ϭ 0.927) was produced by analysis (using Cricket-Graph) of elution volumes of molecular mass standards (q), (thyroglobulin (670 kDa), ␥-globulin (158 kDa), ovalbumin (44 kDa), myoglobin (17 kDa), and vitamin B-12 (1.3 kDa)). Position of standards are based on 10 HPLC analyses with essentially identical results. B, the observed M r (q) of the N-terminal deletion mutants (calculated from the regression curve in Fig. 7A and elution position of each protein in Fig. 6) was plotted as a function of the predicted M r of PDE3 mutant homodimers (calculated from their deduced aa sequences). (-) represents M r values (based on gel filtration studies) that would be are identical to predicted MW values of PDE mutant dimers (calculated from deduced aa sequences). other proteins. Consistent with our results, it had been suggested that the N-terminal region of PDE4D1 is responsible for formation of large aggregates, and removal of the upstream conserved region 2 (UCR2) domain from the N-terminal region was associated with formation of dimers (30).
It is probably relevant that between the large hydrophobic membrane-association domain in region 1 and the smaller hydrophobic domain in region 2 of PDE3B lie consensus sites for phosphorylation by PKB/Akt and PKA, thought to be important enzymes in regulation of phosphorylation and activation of PDE3B in adipocytes (1,41,42) and FDCP2 myeloid cells (43). One might speculate that the transmembrane helical segments of PDE3 anchor the enzyme at specific locations and bring PDE3 in proximity to other membrane-associated signaling molecules or complexes including protein kinase A-anchoring proteins, multifunctional kinases, phosphatases, and perhaps other regulatory proteins (44). The structural determinants in region 1 and region 2, which apparently permit different types of interactions with membranes, might also be important in interactions of PDE3 isoforms with different types of signaling partners, scaffold proteins, regulatory molecules, etc. Current studies are designed to identify the specific cellular location(s) of PDE3 isoforms and associated signaling complexes that regulate their function. Acknowledgments-We thank Dr. Martha Vaughan for critical reading of the manuscript, Dr. John Colicelli for providing the N-terminal deletion constructs of HPDE3A in pADNA54 vector, Ronald Adamik and Linda Stevens for technical advice, and Carol Kosh for secretarial assistance.