Cyclic Nucleotide Phosphodiesterase PDE1C1 in Human Cardiac Myocytes*

Isoforms in the PDE1 family of cyclic nucleotide phosphodiesterases were recently found to comprise a significant portion of the cGMP-inhibited cAMP hydrolytic activity in human hearts. We examined the expression of PDE1 isoforms in human myocardium, characterized their catalytic activity, and quantified their contribution to cAMP hydrolytic and cGMP hydrolytic activity in subcellular fractions of this tissue. Western blotting with isoform-selective anti-PDE1 monoclonal antibodies showed PDE1C1 to be the principal isoform expressed in human myocardium. Immunohistochemical analysis showed that PDE1C1 is distributed along the Z-lines and M-lines of cardiac myocytes in a striated pattern that differs from that of the other major dual-specificity cyclic nucleotide phosphodiesterase in human myocardium, PDE3A. Most of the PDE1C1 activity was recovered in soluble fractions of human myocardium. It binds both cAMP and cGMP with Km values of ∼1 μm and hydrolyzes both substrates with similar catalytic rates. PDE1C1 activity in subcellular fractions was quantified using a new PDE1-selective inhibitor, IC295. At substrate concentrations of 0.1 μm, PDE1C1 constitutes the great majority of cAMP hydrolytic and cGMP hydrolytic activity in soluble fractions and the majority of cGMP hydrolytic activity in microsomal fractions, whereas PDE3 constitutes the majority of cAMP hydrolytic activity in microsomal fractions. These results indicate that PDE1C1 is expressed at high levels in human cardiac myocytes with an intracellular distribution distinct from that of PDE3A and that it may have a role in the integration of cGMP-, cAMP- and Ca2+-mediated signaling in these cells.

PDE1 3 cyclic nucleotide phosphodiesterases are dual-specificity enzymes that bind and hydrolyze cAMP and cGMP in a mutually competitive manner (for review, see Ref. 1). Their activity is increased by their binding to Ca 2ϩ and calmodulin, a feature unique to the PDE1 family of cyclic nucleotide phosphodiesterases. Three PDE1 genes, PDE1A, PDE1B, and PDE1C, have been identified, and several protein isoforms are generated from these genes by alternative splicing. PDE1A and PDE1B isoforms have significantly higher affinities for cGMP than for cAMP, whereas PDE1C isoforms have similar affinities for both substrates (2)(3)(4). The modulation of the catalytic activity of PDE1 isoforms by Ca 2ϩ -dependent stimulation and their susceptibility to mutually competitive inhibition by cAMP and cGMP suggest that these enzymes may be points of interaction among several signaling pathways.
Cyclic nucleotide phosphodiesterases are particularly important in cardiac muscle. In humans, alterations in cAMP metabolism and cAMP-mediated signaling leading to decreases in intracellular cAMP content and in the phosphorylation of some substrates of cAMP-dependent protein kinase are prominent features of the pathophysiology of heart failure, and inhibition of PDE3 cAMP hydrolytic activity has a major role in its treatment (5). Changes in cGMP-mediated signaling in cardiac disease have not been characterized as extensively in humans, but studies in animals suggest that inhibition of PDE5 cGMP hydrolytic activity may have anti-hypertrophic and anti-apoptotic actions (6). PDE1 cyclic nucleotide phosphodiesterases have also been found in cardiac muscle in a number of species, including humans (7)(8)(9)(10)(11)(12)(13). As to the cellular expression of PDE1 in cardiac muscle, this has been examined, to our knowledge, only in rat hearts, where PDE1 activity was reported to be absent from cardiac myocytes (12).
In experiments in which we quantified the contribution of PDE3 isoforms to cyclic nucleotide hydrolytic activity in human myocardium, we found a large amount of Ca 2ϩ /calmodulinstimulated cAMP hydrolytic activity in soluble fractions of this tissue (14). Whether this activity was present in cardiac myocytes or in nonmuscle cells in the myocardium was unclear. In the experiments described below, we used monoclonal antibodies to identify PDE1C1 as the predominant PDE1 isoform expressed in human myocardium and to demonstrate its intracellular distribution along the Z-and M-lines of cardiac myocytes in a striated pattern distinct from that of PDE3. Assays in the absence and presence of the PDE1-selective inhibitor IC295 and the PDE3-selective inhibitor cilostazol showed that PDE1C1 constitutes the great majority of cAMP hydrolytic and cGMP hydrolytic activity in soluble fractions and the majority of cGMP hydrolytic activity in microsomal fractions, whereas PDE3 constitutes the majority of cAMP hydrolytic activity in microsomal fractions.

EXPERIMENTAL PROCEDURES
Monoclonal Antibodies-A fusion protein containing truncated human recombinant PDE1C3 and glutathione S-transferase was used to generate PDE1C-selective monoclonal antibodies. Briefly, the nucleotide sequence corresponding to amino acids 150 -709 of PDE1C3 (2) was subcloned into pGEX-2T (Amersham Biosciences) and expressed in XL-1 Blue Escherichia coli (Stratagene, La Jolla, CA) as described by the manufacturer. A 90-kDa protein corresponding to the glutathione S-transferase-PDE1C3 fusion protein was purified from bacterial lysates by SDS-PAGE and electroelution. This fusion protein was used as the immunogen to generate mouse monoclonal antibodies 114C and 114G using standard procedures (15). A PDE1A monoclonal antibody (367B) was generated against amino acids 140 -677 of PDE1A3 using the techniques described above. The PDE1B antibody 114B has been described previously (16).
A 3Ј-deletion series of PDE1C3 was generated using Erase-a-Base (Promega, Madison, WI) and used to define the linear epitopes for 114C and 114G. Two blocking peptides, corresponding to the PDE1C3 amino acid sequences AKSQAEEGAS (p43; 114G epitope) and KKEAEEKARL (p44; 114C epitope), were synthesized at Peptide Innovations (UCB-Bioproducts, Raleigh, NC). Blocking experiments were performed by preincubating the antibodies in 50-fold molar excess of specific or nonspecific epitope peptide prior to immunostaining. The PDE3A-specific mouse monoclonal antibody 325T was generated from a gel-purified glutathione S-transferase fusion protein containing amino acids 614 -1139 of human PDE3A.
Western Blot Analysis-Extracts of human heart were prepared from frozen tissue obtained from either the National Disease Research Interchange (NDRI, Philadelphia, PA) or the Cooperative Human Tissue Network (CHTN, Bethesda, MD). Briefly, frozen tissue was pulverized in the presence of liquid nitrogen to a fine powder using a mortar and pestle. SDS sample buffer (125 mM Tris, pH 8.8, 2% SDS, 20% glycerol, 100 mM DTT, and 0.01% bromphenol blue) at 95°C was immediately added to the frozen powder, and the sample was boiled for 5 min. Samples containing 4 -7 pmol/min of cAMP hydrolytic activity, assayed as described below at 0.1 M cAMP, were loaded onto 10% SDS-polyacrylamide gels, electrophoresed, and transferred electrophoretically to Immobilon P (Millipore) membranes for Western blot analysis. Primary antibodies were used at a concentration of 0.6 g/ml in blocking solution A (Tris-buffered saline, 5% powdered milk, 0.1% Tween). Peroxidase-conjugated, goat or rabbit anti-mouse secondary antibodies (Bio-Rad) were used at a 1:10,000 dilution in blocking solution A followed by detection using the Renaissance Enhanced Luminal Western blot Chemiluminescence Reagent (PerkinElmer Life Sciences). In other experiments, Western analysis was performed on soluble and microsomal fractions of human myocardium prepared as described below.
Immunofluorescence-Samples of human cardiac tissue frozen in Tissue-Tek II O.C.T. Compound (Miles, Elkhart, IN) were obtained from NDRI. The tissue was cut into 6-m sections, fixed with acetone for 10 min at 4°C, and air dried. All subsequent steps were carried out at room temperature. The sections were incubated in blocking solution B (phosphatebuffered saline, 10% human serum, 10% normal goat serum, and 1% bovine serum albumin) for 30 min. The tissue sections were subsequently incubated with primary antibody diluted in blocking solution B for 1-2 h. The anti-PDE1C antibodies 114C and 114G were used at 25 g/ml. Antibodies to various cardiomyocyte proteins were also used including anti-desmin (rabbit polyclonal antibody, 1:50 dilution; Biomedia Corp), anti-PDE3A (mouse monoclonal antibody, 3 g/ml), and antimyosin (mouse monoclonal antibody, 3 g/ml; ICOS Corp.). The secondary antibody was either a biotin-conjugated goat anti-mouse or goat anti-rabbit antibody (Jackson Immuno-Research Laboratories, Inc., West Grove, PA) diluted 1:400 in blocking solution B. Streptavidin-Cy3 conjugate (Jackson ImmunoResearch Laboratories) was used for final detection at a dilution of 1:3000 in 1% bovine serum albumin/phosphatebuffered saline. For double labeling, sections were first stained with 114C and streptavidin-Cy3 as described above followed by incubation with the desmin antibody. Detection was done with a fluorescein isothiocyanate-conjugated goat anti-rabbit antibody (Jackson ImmunoResearch Laboratories). Confocal microscopy was performed using the Bio-Rad 600 confocal microscope.
Preparation of Recombinant Protein-Full-length recombinant human PDE1C1 was expressed in Sf9 cells using standard protocols. PDE1C1 was purified from a 5-g pellet of Sf9 cells using the following procedure. Cells were resuspended in lysis buffer (50 mM MOPS-Na, pH 7.4, 10 M ZnSO 4 , 0.1 mM CaCl 2 , 1 mM DTT, 2 mM benzamidine HCl, 5 g/ml each pepstatin, leupeptin, and aprotinin, and 20 g/ml each of calpain I and calpain II) and lysed using a French press (11,000 p.s.i. in a 40,000 PSI French Pressure Cell with a 1-inch diameter piston (SLM-AMINCO)). The lysed cell extract was centrifuged at 100,000 ϫ g for 45 min. MgCl 2 was added to the supernatant to a final concentration of 10 mM. The supernatant was filtered through a 0.2-m syringe filter with a glass fiber pre-filter and subsequently loaded onto a Cibracron Blue Sepharose (CB400) dye-exclusion column that had been pretreated with polyvinyl pyrrolidone to cover nonspecific binding sites as described previously (17). The extract was circulated five to six times through the column over several hours at a flow rate of 1 ml/min (30 cm/h). The column was subsequently washed with 5 column volumes of buffer A (50 mM MOPS-Na, pH 7.4, 10 M ZnSO 4 , 5 mM MgCl 2 , 0.1 mM CaCl 2 , 1 mM DTT, 2 mM benzamidine HCl) followed by 5 column volumes of 5 mM 5Ј-AMP in buffer A. The column was then washed with 5 volumes of buffer C (50 mM MOPS-Na, pH 7.4, 10 M ZnSO 4 , 0.1 mM CaCl 2 , 1 mM DTT, 2 mM benzamidine HCl), and PDE1C1 activity was eluted in buffer C with 1 mM cAMP. Eluted protein was dialyzed overnight against buffer X (50 mM MOPS-Na, pH 7.4, 10 M ZnSO 4 , 100 mM NaCl, 1 mM CaCl 2 , 1 mM DTT, 1 mM benzamidine HCl). rtPDE3A1 (formerly designated rtPDE3A-136) was expressed in Sf9 cells and prepared as previously described (14). Other rtPDE constructs were expressed either in yeast, using the expression vector YepC-PADH2d, or in Sf9 cells, using a baculovirus expression system. The extracts were prepared by suspending the cells in a buffer containing 50 mM MOPS, pH 7.2, 1 mM dithiothreitol, 2 mM benzamidine, 5 M ZnSO 4 , 20 g/ml calpain inhibitors I and II, and 5 g/ml each of leupeptin, pepstatin, and aprotinin. The mixture was sonicated twice for 30 s in a Branson Model 450 Sonifier, and the cells were lysed in a French press cell at 20,000 p.s.i. (4°C). The lysate was sedimented 100,000 ϫ g for 45 min. For most proteins, the supernatant was recovered and filtered with a 0.2-m unit containing a glass fiber pre-filter. For rtPDE3A and rtPDE3B, the pellet was washed once in lysis buffer, suspended in lysis buffer with a Dounce homogenizer, and sonicated for 30 s. The extracts were stored at Ϫ80°C until use.
Preparation of Human Retinal Extracts-Eye globes were obtained from organ donors for whom no suitable recipients were identified at the time of organ harvest. Retinas were dissected from eye globes, washed once in phosphate-buffered saline and frozen rapidly on dry ice for transport. One frozen retina was ground with a mortar and pestle, thawed in 5 ml of phosphate-buffered saline, and centrifuged at 100,000 ϫ g for 30 min. The supernatant (800 l) was treated with 100 l of packed trypsin-agarose beads (Sigma) for 30 min at room temperature. The beads were removed by low-speed centrifugation, and the supernatant was diluted 1:1 with 80% glycerol for storage at Ϫ80°C until use.
Preparation of Subcellular Fractions of Human Myocardium-Human myocardium from the left ventricular free wall was obtained from the hearts of organ donors for whom no suitable recipients were identified at the time of organ harvest. Soluble and microsomal fractions were prepared by homogenization and differential sedimentation using a protocol adapted from previously published methods (18). Tissue was homogenized for two 10-s cycles in a Kinematica GmbH homogenizer, setting 10, in 5 volumes of 0.29 M sucrose, 3 mM NaN 3 , and 10 mM MOPS (pH 7.0, 4°C) ("sucrose buffer"), to which were added 1 mM dithiothreitol, 1 mM benzamidine, 2 mM EGTA, 0.8 mM phenylmethylsulfonyl fluoride, and 1 g/ml each of pepstatin A, leupeptin and anitpain. (In some experiments, EGTA was omitted from and 1.0 mM CaCl 2 added to the sucrose buffer.) Following two 10-min sedimentations at 5,000 ϫ g to remove debris, the supernatant was sedimented for 60 min at 30,000 ϫ g. The supernatant was used as the "soluble fraction." The pellet was resuspended by hand homogenization (glass-glass) in 2 volumes (relative to starting material) of sucrose buffer containing 0.6 M KCl (pH 7.0, 4°C). Following resedimentation for 40 min at 113,000 ϫ g, the pellet was suspended in sucrose buffer using a Teflon glass homogenizer and stored at Ϫ80°C until use as the "microsomal fraction." (In some experiments, 0.6 M KCl was omitted from the sucrose buffer used to wash the pellet.) Each preparation was made using tissue from at least three different hearts. Protein yields (mg of protein/g wet weight of starting tissue, calculated using the Bradford method with bovine serum albumin as the standard) were 22.6 Ϯ 4.3 for the soluble fraction and 1.16 Ϯ 0.30 for microsomal fractions.
Assay of Cyclic Nucleotide Hydrolytic Activity-Cyclic nucleotide hydrolytic activity was quantified at 30°C using the twostep snake venom method with [ (19). The amount of protein used per assay and the incubation times were adjusted to ensure that substrate hydrolysis did not exceed 20% of the total cyclic nucleotide. EGTA, CaCl 2 , and calmodulin (provided by Donald K. Blumenthal, University of Utah) were included as indicated. Values for K m and V max for cAMP hydrolytic and cGMP hydrolytic activity were calculated by nonlinear regression using the equation PDE1 and PDE3 activities in the absence of Ca 2ϩ and calmodulin were measured using the new PDE1 selective inhibitor IC295 (described below) and the PDE3 selective inhibitor cilostazol (Otsuka) (20). To minimize possible inhibition of other PDE families, these measurements were made at concentrations of drug that inhibited cyclic nucleotide hydrolytic activity by rtPDE1C1 and rtPDE3A1 between 30 and 50%. PDE1 activity was calculated by dividing the amount of activity inhibited by IC295 in subcellular fractions by the fractional inhibition of rtPDE1C1 activity at the same IC295 concentration; thus, if the concentration of IC295 used in the experiment inhibited rtPDE1C1 activity by 35%, the activity inhibited by the drug in subcellular fractions was divided by 0.35 to determine the amount of PDE1C1 in these fractions. Similarly, PDE3 activity was calculated by dividing the amount of activity inhibited by cilostazol by the fractional inhibition of rtPDE3A1 activity at the same cilostazol concentration (14).

Expression of PDE1C1 in Human Cardiac
Myocytes-We examined the expression of PDE1 isoforms in human myocardium. Monoclonal antibodies specific for PDE1C were generated using a truncated recombinant form of the human enzyme. Two of these antibodies were characterized extensively: mAb 114C recognizes amino acids 537-544 of PDE1C3, corresponding to the sequence EKARLAAE, whereas mAb 114G recognizes amino acids 557-565, corresponding to the sequence EEGASGKAE. Both epitopes lie within a region of PDE1C common to all of its splice variants but absent from PDE1A and PDE1B. These antibodies were used to analyze Western blots of protein extracts from whole human heart and of subcellular fractions prepared from human left ventricular myocardium by homogenization and differential sedimentation (Fig. 1). mAb 114C and mAb 114G both reacted with a single protein band in heart extracts whose apparent M r on SDS-PAGE, 72,000 -75,000, is consistent with its identity as PDE1C1. For comparison, Western blots were also performed using monoclonal antibodies to PDE1A and PDE1B. These antibodies did not react with proteins in the human heart.
We used mAb 114C and 114G to determine the cellular and intracellular distribution of PDE1C1 in human myocardium. Immunostaining with mAb 114C demonstrated that PDE1C1 is distributed in a striated pattern within cardiac myocytes ( Fig.  2A), with little if any staining of nonmuscle cells (Fig. 2B). This pattern was also observed with mAb 114G (Fig. 2C). At higher magnification, PDE1C1 was seen in alternating major and minor bands in cardiac myocytes (Fig. 2D). Immunostaining with mAb 114C was blocked by addition of a peptide encoding the epitope to that it binds (Fig. 2E), but not by addition of an unrelated peptide (Fig. 2F).
We compared the intracellular distribution of PDE1C1 to that of other proteins in cardiac myocytes, including PDE3A, desmin, and myosin (Fig. 3). All four proteins have a striated staining pattern within cardiac myocytes; desmin is also seen in high density in the intercalated discs. Unlike PDE1C1, however, none of the other proteins were seen in alternating major and minor bands. The intracellular distribution of PDE1C1 was characterized in greater detail through double labeling with mAb 114C and a polyclonal anti-desmin antibody (Fig. 4). Overlay of the anti-PDE1C and anti-desmin images shows colocalization of PDE1C1 and desmin in the major bands, consistent with its presence in the Z-lines of cardiac myocytes. PDE1C1 is also found in the thinner bands midway between the desmin staining regions, consistent with its presence in the M-lines of cardiac myocytes.
PDE1C1 Activity in Subcellular Fractions of Human Myocardium-We characterized the catalytic activity of PDE1C1 and quantified its contribution to cyclic nucleotide hydrolytic activity in soluble and microsomal fractions of human left ventricular myocardium. This activity is stimulated in the presence of Ca 2ϩ /calmodulin in a manner that results in an increase in V max without a significant change in K m (1). We therefore measured cyclic nucleotide hydrolysis at 0.1 M cAMP or cGMP in the absence and presence of Ca 2ϩ /calmodulin. In soluble fractions, approximately half of the cAMP hydrolytic and cGMP hydrolytic activities were Ca 2ϩ /calmodulin-dependent (Fig. 5). In microsomal frac- . Localization of PDE1C in human myocardium by immunofluorescence labeling. A, cryosections (6 m) of human heart were incubated with anti-PDE1C mAb 114C, then with a biotinylated anti-mouse secondary antibody and streptavidin Cy3, and then imaged by confocal microscopy. B, cryosections were imaged by Nomarski. Arrows indicate nonmuscle cells not stained in A. C, immunostaining by mAb 114G. D, immunostaining with mAb 114C as in A but at higher magnification. Arrows indicate major and minor bands. E, immunostaining by mAb 114C is blocked by inclusion of peptide p44, whose sequence corresponds to its mapped epitope. F, immunostaining by mAb 114C is not blocked by inclusion of peptide p43, encoding sequence from elsewhere in PDE1C1. All images are at 600-fold optical magnification; D has an additional digital magnification of 5-fold, for a total magnification of 3,000-fold; F has an additional digital magnification of 1.5-fold, for a total magnification of 900-fold.

PDE1C1 in Human Cardiac Myocytes
tions, in contrast, a minimal fraction of the cAMP hydrolytic activity and ϳ30% of the cGMP hydrolytic activity were Ca 2ϩ /calmodulin-dependent.
By multiplying the specific activities in soluble and microsomal fractions by the protein yields per g of tissue, we determined that the Ca 2ϩ /calmodulin-stimulated cAMP hydrolytic and cGMP hydrolytic activity in the soluble fractions represented ϳ99% of the total Ca 2ϩ /calmodulin-stimulated cAMP hydrolytic and cGMP hydrolytic activity recovered from the myocardium. To test whether this was the result of the dissociation of PDE1C1 from contractile elements by homogenization in the presence of EGTA, which would block Ca 2ϩ -dependent interactions, we homogenized tissue in the presence of 1.0 mM CaCl 2 before preparing subcellular fractions, with the same result. We also considered that the content of PDE1 activity in microsomal fractions could have been reduced by our washing of the microsomes in 0.6 M KCl. To test this, we repeated the experiment in microsomes washed in zero salt buffer. The result was again the same.
We characterized Ca 2ϩ /calmodulin-stimulated activity in soluble fractions in terms of K m and V max , and we compared these values to those obtained for rtPDE1C1 (Table 1). For native PDE1C1 and rtPDE1C1, K m values for cAMP and cGMP were similar, with native and recombinant enzymes both hav-   FIGURE 4. Labeling of human myocardium tissue with mAb 114C and anti-desmin polyclonal antibody. A, cryosections of human heart were incubated with mAb 114C, then with a biotinylated anti-mouse secondary antibody and streptavidin Cy3. Major bands alternate with minor bands (arrows). B, the cryosection was incubated with rabbit anti-desmin polyclonal antibody, then with fluorescein isothiocyanate-conjugated goat anti-rabbit secondary antibody. Major bands and intercalated discs are seen (arrows). C, the images from A and B are merged, showing duallabeled major bands and single-labeled minor bands. All images are at 600-fold optical magnification and 3.6-fold digital magnification, for a total magnification of 2,160-fold. NOVEMBER 9, 2007 • VOLUME 282 • NUMBER 45

PDE1C1 in Human Cardiac Myocytes
ing slightly higher affinities for cAMP. For the native enzyme, V max values for cAMP and cGMP were comparable with one another; this was also true for rtPDE1C1. We also quantified the Ca 2ϩ /calmodulin-independent component of PDE1C1 activity in human myocardium. To do this, we identified a compound, IC295, that inhibits the cAMP hydrolytic and cGMP hydrolytic activity of PDE1 isoforms with high potency and has a high degree of selectivity for PDE1 relative to other cAMP hydrolyzing and cGMP hydrolyzing enzymes (Table 2). At a substrate concentration of 0.1 M, IC295 inhibited both the cAMP hydrolytic and cGMP hydrolytic activity of rtPDE1C1 with an IC 50 of 0.04 M (Fig. 6); under these conditions, IC295 had no effect on the cAMP hydrolytic or cGMP hydrolytic activity of rtPDE3A1. PDE1 activity was therefore determined by quantifying IC295-sensitive activity at these concentrations of inhibitor and substrate in soluble and microsomal fractions in the absence of Ca 2ϩ /calmodulin. PDE1 constituted over half of the Ca 2ϩ /calmodulin-independent cAMP hydrolytic activity and the great majority of Ca 2ϩ /calmodulin-independent cGMP hydrolytic activity in the soluble fractions (Fig. 5). In microsomal fractions, in contrast, PDE1 constituted a small fraction of the Ca 2ϩ /calmodulin-independent cAMP hydrolytic activity and approximately half of the Ca 2ϩ /calmodulin-independent cGMP hydrolytic activity. When the values for Ca 2ϩ /calmodulin-stimulated and Ca 2ϩ / calmodulin-independent PDE1 activities were added, PDE1C1 was seen to constitute the great majority of the cAMP hydrolytic activity in soluble fractions but a minority of the cAMP hydrolytic activity in microsomal fractions (Table 3). PDE1C1 also constitutes the great majority of the cGMP hydrolytic activity in the soluble fractions and over half of the cGMP hydrolytic activity in the microsomal fractions.
For comparison, we quantified the contribution of PDE3 to cyclic nucleotide hydrolytic activity in these fractions by meas-uring cAMP and cGMP hydrolysis at 0.1 M in the absence and presence of the PDE3-selective inhibitor cilostazol at its IC 50 (0.007 M for cAMP and 0.117 M for cGMP) (14). We noted major differences regarding the contribution of PDE1 and PDE3 to cyclic nucleotide hydrolytic activity in the different subcellular fractions. Whereas the specific activity of PDE1 was considerably higher in soluble fractions than in microsomal fractions, and was roughly comparable with respect to cAMP and cGMP, the specific activity of PDE3 was similar in magnitude in soluble and microsomal fractions, and was significantly higher for cAMP than for cGMP ( Fig. 5 and Table 3). Moreover, whereas PDE3 constituted a minority of the cAMP hydrolytic activity in soluble fractions at 0.1 M substrate, it constituted the majority of the cAMP hydrolytic activity in microsomal fractions (Table 3). PDE3 constituted a minor component of the cGMP hydrolytic activity in soluble fractions but almost half of the cGMP hydrolytic activity in microsomal fractions. These results are reflected in the large differences in the ratios of PDE1 and PDE3 activities in the soluble and microsomal fractions (Table 3).

DISCUSSION
In previous studies, we identified enzymes in the PDE3 family of cyclic nucleotide phosphodiesterases, a family characterized by its high affinity for both cAMP and cGMP, which are mutually competitive substrates, as the principal membrane-associated cAMP hydrolytic activity in human myocardium (14,21,22). Those studies and others suggested that isoforms of PDE1,

Inhibition of PDE isoforms by IC295
All recombinant proteins were expressed as full-length constructs except for PDE10A1, which was expressed as amino acids 14 -779. PDE6A activity was quantified in extracts prepared from human retinas. Each IC 50 value is derived by nonlinear regression from a 10-point inhibitor concentration curve.

Isoform
Gene

PDE1C1 in Human Cardiac Myocytes
a second family of dual-specificity phosphodiesterases, are also present at high levels in this tissue (13,14). In the experiments reported here, we identified PDE1C1 as the predominant and perhaps sole isoform of PDE1 in human myocardium. It is expressed largely, if not exclusively, in cardiac myocytes, where it is seen in a pattern of alternating major and minor bands consistent with its distribution in the Z-and M-lines. These findings contrast with the results of an earlier study in rat myocardium, in which PDE1 activity was found in extracts of whole myocardium but not in preparations from isolated myocytes; the authors inferred that PDE1 in rat myocardium was expressed in nonmuscle cells (12). Whether these differences between rats and humans are true inter-species differences or result from differences in the approaches used is unclear, as the study in rat myocardium did not include an attempt to localize PDE1 in situ with anti-PDE1 antibodies. The fact that PDE1C1 was seen in a distinct striated pattern within cardiac myocytes but was recovered in soluble fractions raises the possibility of its being localized intracellularly through interactions with other proteins that are disrupted during tissue homogenization. To our knowledge, however, calmodulin is the only known PDE1-binding protein, and the fact that the recovery of PDE1 in the soluble fractions was not diminished when CaCl 2 was present during tissue homogenization suggests that its intracellular localization is not calmodulin-dependent. Furthermore, the specific activity of PDE1 in microsomal fractions was not diminished by high salt washing, raising some questions as to the role of protein-protein interactions in its intracellular localization. An alternative possibility is that the apparent localization of PDE1C1 to the Zand M-lines instead reflects its exclusion from organized structures in other intracellular compartments.
Our experiments showed that PDE1C and PDE3A are found in different intracellular patterns in cardiac myocytes, and that the contribution of PDE1C1 to cAMP hydrolytic and cGMP hydrolytic activity in subcellular fractions of human myocardium differs from that of PDE3. Whereas PDE3 is the principal cAMP hydrolytic activity in cardiac microsomes at 0.1 M substrate, PDE1C1 comprises the majority of the cAMP hydrolytic activity in soluble fractions. (Because the K m for cAMP is severalfold higher in the case of PDE1, its contribution relative to that of PDE3 would be increased at higher cAMP concentrations.) Furthermore, whereas PDE3 constitutes a significant portion of the cGMP hydrolytic activity in microsomal fractions at a substrate concentration of 0.1 M, PDE1 constitutes a comparable portion of the microsomal cGMP hydrolytic activity in these fractions and nearly all of the cGMP hydrolytic activity in soluble fractions.
These differences in the distribution of PDE1 and PDE3 activities are interesting in the context of the compartmentation of cAMP-mediated signaling in cardiac myocytes, through which cAMP content is regulated selectively in spatially distinct intracellular compartments in response to different adenylyl cyclase coupled-receptor agonists (23)(24)(25)(26)(27)(28)(29). Cyclic nucleotide phosphodiesterases contribute to this compartmentation by restricting receptor-mediated increases in cAMP content to specific intracellular locations (30). In rodents, PDE3 has a greater role in regulating forskolin-induced increases in intra-cellular cAMP content, whereas PDE2 and PDE4 have greater roles in regulating ␤-adrenergic and glucagon receptor-mediated increases (31)(32)(33)(34). These differences in the actions of these phosphodiesterases reflect differences in their intracellular targeting. Viewed from this perspective, the differences in the contribution of PDE1 and PDE3 to cAMP hydrolytic activity in soluble and microsomal fractions of human myocardium suggest that these enzymes may regulate the phosphorylation of different PKA substrates. The contribution of PDE1 to the compartmentation of cAMP-mediated signaling has until now remained relatively unexplored, owing in part to a previous dearth of suitable PDE1-selective inhibitors and in part to the belief that this enzyme is generally not expressed in cardiac myocytes (12). A recent computational model for the compartmentation of cAMP metabolism in cardiac myocytes, for example, omitted PDE1 from the analysis (35). Our findings indicate this model is not likely to be applicable to the compartmentation of cAMP-mediated signaling in human myocardium. Our identification of IC295 as a PDE1-selective inhibitor may help in elucidating the role of PDE1 in cAMP-mediated signaling in species, including humans, in which the enzyme is expressed in cardiac myocytes.
Our finding that PDE1C1 constitutes the great majority of cGMP hydrolytic activity in human myocardium suggests that it is also likely to have a role in cGMP-mediated signaling. Increases in cGMP content in response to activators of membrane-bound guanylyl cyclases are associated with positive chronotropic and inotropic effects, whereas increases in cGMP content in response to activators of soluble guanylyl cyclases are associated with negative inotropic effects (36 -41). These differences probably reflect a compartmentation of cGMP metabolism in cardiac myocytes in which phosphodiesterases are involved. In adult rat ventricular myocytes, for example, PDE2 inhibitors potentiate increases in cGMP synthesized by soluble and membrane-bound guanylyl cyclases, whereas PDE5 inhibitors only potentiate increases in cGMP synthesized by soluble guanylyl cyclases (42). Efforts to determine the role of PDE1C1 in cGMP-mediated signaling in cardiac myocytes may be facilitated through the use of IC295 or other PDE1 selective inhibitors.
The fact that the cAMP hydrolytic and cGMP hydrolytic activities of PDE1C1 are mutually competitive raises the possibility of a role for PDE1C1 in the integration of cAMP-and cGMP-mediated signaling. Such a role has been established for other cyclic nucleotide phosphodiesterases found in cardiac muscle. Allosteric stimulation of the cAMP hydrolytic activity of PDE2 by cGMP leads to a reduction in cAMP content that attenuates several ␤-adrenergic receptor-mediated events in atrial and ventricular myocytes (43)(44)(45)(46)(47). In contrast, competitive inhibition of the cAMP hydrolytic activity of PDE3 by cGMP in atrial and ventricular myocytes leads to an increase in cAMP content that potentiates delayed-rectifier K ϩ currents and L-type Ca 2ϩ currents (48 -50). PDE1C1 may be a second phosphodiesterase through which cGMP can amplify cAMPmediated signaling in cardiac myocytes. The similar values for K m and V max for the cAMP hydrolytic and cGMP hydrolytic activities of PDE1C1 raise the novel possibility that cAMP may increase cGMP content in these cells by inhibiting the cGMP PDE1C1 in Human Cardiac Myocytes NOVEMBER 9, 2007 • VOLUME 282 • NUMBER 45 hydrolytic activity of PDE1C1. The expression of PDE1C1 may also allow both cAMP-and cGMP-mediated signaling in cardiac myocytes to be modulated by Ca 2ϩ . Identifying the possible role of PDE1C1 in the integration of these signaling pathways is an important area for further study.
The presence of PDE1C1 in human myocardium may have therapeutic relevance. Inhibitors of cAMP hydrolysis by PDE3 have been used as inotropic agents in the treatment of heart failure; they have beneficial effects in the short term, but their chronic use has been associated with an increase in mortality (5). The effects of PDE1 inhibition on cAMP content in intracellular compartments of cardiac myocytes are likely to differ from those of PDE3 inhibition, perhaps with more favorable sequelae. In animal models, inhibition of cGMP hydrolysis through PDE5 inhibition has been shown to prevent and reverse hypertrophic responses to pressure overload and ␤-adrenergic receptor stimulation and to reduce infarct area size and myocyte apoptosis in injured myocardium (51)(52)(53)(54)(55)(56)(57)(58)(59). PDE1 inhibition might elicit similar benefits. Characterization of the effects of PDE1 inhibition in models of heart disease may lead to insight into new therapeutic approaches.