Peptide-activator of cyclic nucleotide phosphodiesterases in the rat and human uterus

The presence of the cyclic nucleotide phosphodiesterase peptide-activator previously isolated from the calf myometrium was studied in uterine tissue of rats at different functional states (in puberty, estrus and diestrus stages, pregnancy) and in human myometrium and myoma. The peptide isolated from calf myometrium, immature rat myometrium and uterine tissue in myoma caused a 3-4-fold stimulation of the phosphodiesterase activity.     

Evidence for convertible forms of soluble uterine cyclic nucleotide phosphodiesterase

The cyclic nucleotide phosphodiesterase (3':5'-cyclic nucleotide 5'-nucleotidohydrolase, EC 3.1.4.17) systems of many tissues show multiple physical and kinetic forms. In contrast, the soluble rat uterine phosphodiesterase exists as a single enzyme form with non-linear Lineweaver-Burk kinetics for cyclic AMP (app. Km of approx. 3 and 20 microM) and linear kinetics for cyclic GMP (app. Km of approx. 3 microM) since the two hydrolytic activities are not separated by a variety of techniques. In uterine cytosolic fractions, cyclic AMP is a non-competitive inhibitor of cyclic GMP hydrolysis (Ki approx. 32 microM). Also, cyclic GMP is a non-competitive inhibitor of cyclic AMP hydrolysis (Ki approx 16 microM) at low cyclic GMP/cyclic AMP substrate ratios. However, cyclic GMP acts as a competitive inhibitor of cyclic AMP phosphodiesterase (Ki approx 34 microM) at high cyclic GMP/cyclic AMP substrate ratios. When a single hydrolytic form of uterine phosphodiesterase, separated initially by DEAE anion-exchange chromatography, is treated with trypsin (0.5 microgram/ml for 2 min) and rechromatographed on DEAE-Sephacel, two major forms of phosphodiesterase are revealed. One form elutes at 0.3 M NaOAc- and displays anomalous kinetics for cyclic AMP hydrolysis (app. Km of 2 and 20 microM) and linear kinetics for cyclic GMP (app. Km approx. 5 microM), kinetic profiles which are similar to those of the uterine cytosolic preparations. A second form of phosphodiesterase elutes at 0.6 M NaOAc- and displays a higher apparent affinity for cyclic AMP (app. Km approx. 1.5 mu) without appreciable cyclic GMP hydrolytic activity. These data provide kinetic and structural evidence that uterine phosphodiesterase contains distinct catalytic sites for cyclic AMP and cyclic GMP. Moreover, they provide further documentation that the multiple forms of cyclic nucleotide phosphodiesterase in mammalian tissues may be conversions from a single enzyme species.     

CYCLIC NUCLEOTIDE METABOLISM IN ELECTROPLAX OF ELECTROPHORUS ELECTRICUS

Abstract— In order to describe the regulation of cyclic nucleotide metabolism in a cholinergic tissue, the properties of cyclic nucleotide phosphodiesterase were determined in electroplax of Electrophorus electricus and compared to those of mammalian brain. Electroplax phosphodiesterase was Mg²⁺ -dependent. localized in the soluble fraction and displayed normal linear Lineweaver-Burk kinetics ( K _m: cyclic AMP. 1.4 μ m ; cyclic GMP, 0.54 μ m ). No low affinity (i.e. high K _m) activity was detected. These results were correlated with comparatively low tissue levels of cyclic AMP (67 pmol/g) and cyclic GMP (3.2 pmol/g). Attempts were made to detect calcium-dependent phosphodiesterase because of the presence of large amounts of the calcium-dependent regulator protein (CDR) in electroplax, as this protein has been shown to activate brain phosphodiesterase. Assay with EGTA under a variety of conditions revealed that no calcium-dependent activity could be detected. Preparation of CDR-deficient phosphodiesterase also failed to produce calcium-dependent activity. Assay of phosphodiesterase in other cholinergic tissues revealed calcium-dependent activity in Electrophorus muscle and rat diaphragm but not in Torpedo electroplax. The results suggest that calcium-dependent activity is not a significant portion of phosphodiesterase in electroplax and indicate alternate roles for CDR in electric tissue.     

Activation of mammalian cyclic AMP phosphodiesterases by trypsin.

BHK fibroblasts contain two forms of cyclic AMP phosphodiesterase 3':5'-cyclic nucleotide 5'-nucleotidohydrolase EC 3.1.4.17) as analyzed by linear sucrose gradient fractionation; a 3.6-S form (peak I) and a 6.7-S form (peak II). Peak I is specific for cyclic AMP as substrate and displays Michaelis-Menten kinetics with an apparent Km of 2--3 micrometer. Peak II hydrolyzes cyclic GMP and displays anomalous kinetics for cyclic AMP hydrolysis. The activity of isolated peak II for cyclic AMP is increased by storage at 4 degrees C, treatment with trypsin, or treatment with rat brain and BHK fibroblast activator proteins. The activity of isolated peak I is unaffected by these conditions. Linear sucrose gradient fractionation demonstrates that activation of peak II by trypsin leads to the formation of a 3.6-S cyclic AMP-specific enzyme form, possibly peak I. In contrast to BHK fibroblasts (and most other mammalian tissues), rat uterus contains only one form of cyclic nucleotide phosphodiesterase on linear sucrose gradients, a 7-S form capable of hydrolyzing both cyclic AMP and cyclic GMP. Treatment of rat uterine supernatant with trypsin leads to the appearance of a 4-S, cyclic AMP-specific form with properties similar to that of BHK peak I. These data suggest that the kinetically complex, higher molecular weight cyclic nucleotide phosphodiesterases may consist of more than one catalytically active site and that multiple forms of the enzyme arise through dissociative mechanisms, possibly as a means of in vivo regulation.     

Purification, characterization and production of rabbit antibodies to rat liver particulate, high-affinity, cyclic AMP phosphodiesterase

The cyclic nucleotide phosphodiesterase (EC 3.4.16) activities of a rat liver particulate fraction were analyzed after solubilization by detergent or by freeze-thawing. Analysis of the two extracts by DEAE-cellulose chromatography revealed that they contain different complements of phosphodiesterase activities. The detergent-solubilized extract contained a cyclic GMP phosphodiesterase, a low affinity cyclic nucleotide phosphodiesterase whose hydrolysis of cyclic AMP was activated by cyclic GMP and a high affinity cyclic AMP phosphodiesterase. The freeze-thaw extract contained a cyclic GMP phosphodiesterase and two high affinity cyclic AMP phosphodiesterase, but no low affinity cyclic nucleotide phosphodiesterase. The cyclic AMP phosphodiesterase activities from the freeze-thaw extract and from the detergent extract all had negatively cooperative kinetics. One of the cyclic AMP phosphodiesterases from the freeze-thaw extract (form A) was insensitive to inhibition by cyclic GMP; the other freeze-thaw solubilized cyclic AMP phosphodiesterase (form B) and the detergent-solubilized cyclic AMP phosphodiesterase were strongly inhibited by cyclic GMP. The B enzyme appeared to be converted into the A enzyme when the particulate fraction was stored for prolonged periods at -20 degrees C. The B form was purified extensively, using DEAE-cellulose, a guanine-Sepharose column and gel filtration. The enzyme retained its negatively cooperative kinetics and high affinity for both cyclic AMP and cyclic GMP throughout the purification, although catalytic activity was always much greater for cyclic AMP. Rabbit antiserum was raised against the purified B enzyme and tested via a precipitin reaction against other forms of phosphodiesterase. The antiserum cross-reacted with the A enzyme and the detergent-solubilized cyclic AMP phosphodiesterase from rat liver. It did not react with the calmodulin-activated cyclic GMP phosphodiesterase of rat brain, the soluble low affinity cyclic nucleotide phosphodiesterase of rat liver or a commercial phosphodiesterase preparation from bovine heart. These results suggest a possible interrelationship between the high affinity cyclic nucleotide phosphodiesterase of rat liver.     

Inhibition of cyclic nucleotide phosphodiesterase activity by an endogenous factor.

Cyclic nucleotide phosphodiesterase activity of several tissues of rat is inhibited by an endogenous factor isolated from rat adipocytes following exposure of these cells to agents that raise intracellular cyclic AMP levels. The inhibitory action was demonstrated with varying cAMP concentrations from 0.1-400 muM. Enzyme from 10,000 X g supernatant of epididymal adipose tissue was inhibited approximately 2-3 fold more than the plasma membrane of adipocytes by a given concentration of the feedback regulator. Kinetic analysis of cAMP phosphodiesterase of plasma membrane showed that feedback regulator (8.8 U/ml) inhibited the Vmax 48%. The maximum inhibition of phosphodiesterase by feedback regulator (20 U/ml) was about 80%. The apparent Km for cAMP was increased. The ability of phosphodiesterase from several tissues of rat (10,000 X g supernatant) to hydrolyze cAMP and cGMP was tested. Feedback regulator inhibited cGMP hydrolysis in cardiac muscle and 5 other tissues 23-92% more than it inhibited the hydrolysis of cAMP. The physiological significance of this inhibitory effect can begin to be clarified when the feedback regulator is purified to homogeneity and characterized.     

The Ca2+ -activated protease (calpain) modulates Ca2+/calmodulin dependent activity of smooth muscle myosin light chain kinase

The Ca2+ -activated neutral protease can proteolyze both Ca2+ -dependent cyclic nucleotide phosphodiesterase and smooth muscle myosin light chain kinase. Ca2+ -dependent cyclic nucleotide phosphodiesterase from rat brain was converted to the Ca2+ -independent active form by Ca2+ -activated protease. The proteolytic effects on myosin light chain kinase of Ca2+-activated protease differed in the presence and absence of the Ca2+-calmodulin (CaM) complex. In the presence of bound CaM, myosin light chain kinase (130k dalton) was degradated to a major fragment of 62 kDa, which had Ca2+/CaM-dependent enzyme and CaM-binding activity. When digestion occurred in the absence of bound CaM, myosin light chain kinase cleaved to a fragment of 60 kDa. This peptide had no enzymatic activity in the presence or absence of the Ca2+-CaM complex. Available evidence suggests that the Ca2+-activated proteases may recognize the conformational change of smooth muscle myosin light chain kinase induced by Ca2+-CaM complex.     

Characterization of soluble uterine cyclic nucleotide phosphodiesterase.

Soluble cyclic nucleotide phosphodiesterase of rat uterus displays distinct structural and regulatory properties. Like phosphodiesterases from many mammalian sources the soluble uterine enzyme system exhibits nonlinear Lineweaver--Burk kinetics with cyclic adenosine 3':5'-monophosphate (cAMP) as substrate (apparent Kms congruent to 3 and 20 micron) and linear kinetics with cyclic guanosine 3':5'-monophosphate (cGMP) as substrate (apparent Km congruent to 3 micron). Unlike most other mammalian phosphodiesterases, however, numerous separation procedures reveal only a single form of uterine phosphodiesterase which catalyzes the hydrolysis of both cAMP and cGMP. A single form of the enzyme is observed upon sucrose gradient centrifugation (7.9 S), agarose gel filtration, and DEAE-cellulose chromatography at either pH 8.0 OR 6.0. Heat denaturation (50 degrees C) of soluble uterine phosphodiesterase causes the loss of both cAMP and cGMP hydrolytic activities at the same rate. Isoelectric focusing reveals major (pI = 5.2) and minor forms (pI = 5.8) of phosphodiesterase which both catalyze the hydrolysis of the two cyclic nucleotide substrates. In vivo administration of estradiol produces identical decreases in the activities of cAMP and cGMP phosphodiesterase. These results raise the possibility that the uterus contains a single form of soluble phosphodiesterase which catalyzes the hydrolysis of both cAMP and cGMP.     

Effect of andrenalectomy on cyclic 3',5'-guanosine monophosphate metabolism of rat liver and other tissues

Adrenalectomy increased guanyl cyclase and cyclic GMP phosphodiesterase activities in liver and other rat tissues. Liver guanyl cyclase activities from adrenalectomized rats were increased above those of normal controls according to kinetic analysis, gel filtration, ion-exchange chromatography, discontinuous sucrose gradient fractionation, sulfhydryl inhibition, and secretin activation. The effects of adrenal insufficiency on hepatic guanyl cyclase and cyclic GMP phosphodiesterase were prevented by cortisone acetate administration. Immunoassay of liver and skeletal muscle cyclic GMP after adrenalectomy showed markedly decreased levels in liver, but increased levels in skeletal muscle. In liver and other tissues, basal adenyl cyclase and cyclic AMP phosphodiesterase activities were unaffected by adrenalectomy. Hepatic levels of cyclic AMP were also unchanged by adrenalectomy. Hypophysectomy raised guanyl cyclase activity in liver but had no effect on liver cyclic GMP phosphodiesterase activity. These alterations are discussed in relation to possible glucocorticoid regulation of cyclic GMP metabolism.     

Cyclic AMP phosphodiesterase and cyclic GMP phosphodiesterase activities of rat mammary tissue.

Cyclic nucleotide phosphodiesterase activity in mammary tissue from rats in midlactation was resolved by DEAE-cellulose chromatography into three functionally distinct fractions: a Ca2+/calmodulin-stimulated cyclic GMP phosphodiesterase, a cyclic GMP-stimulated low-affinity cyclic nucleotide phosphodiesterase, and a high-affinity cyclic AMP-specific phosphodiesterase. The absolute activities and relative proportions of high- and low-affinity enzymes resemble those found, for example, in liver, as distinct from those in excitable tissues. Three functional characteristics are described which are peculiar to mammary-tissue phosphodiesterases. Firstly, the concentration of free Ca2+ required to achieve half-maximal activation of the Ca2+/calmodulin-stimulated phosphodiesterase is somewhat higher than for the analogous enzyme in other tissues; secondly, the activity of this enzyme towards cyclic AMP relative to that towards cyclic GMP is unusually low, and thirdly, the low-affinity cyclic nucleotide phosphodiesterase is inhibited by low concentrations of free Ca2+.     

Lack of altered cyclic nucleotide phosphodiesterase activity in the aorta and heart of the spontaneously hypertensive rat

DEAE-cellulose chromatography demonstrated that the levels of the individual cyclic nucleotide phosphodiesterase were unchanged in the aorta and heart of the spontaneously hypertensive rat as compared with the normotensive control rat. Three peaks of cyclic nucleotide phosphodiesterase activity were observed in both heart and aorta. Peak I enzyme hydrolyzed predominantly cyclic GMP while peak III enzyme hydrolyzed predominantly cyclic AMP. Peak II enzyme was less specific but hydrolyzed more cyclic GMP than cyclic AMP The levels of phosphodiesterase activator in aorta and the responsiveness of peaks I and II from aorta and heart to activator were unchanged in the hypertensive rat. Therefore the decrease in cyclic AMP levels observed by others in aorta and heart of the spontaneously hypertensive rat were probably not due to altered phosphodiesterase activity.     

Cyclic nucleotide metabolism in rat colonic epithelial cells with different proliferative activities

Cyclic nucleotide metabolism was examined in rat distal colonic epithelial cells with different proliferative activities. Lower crypt cells had DNA synthetic rates 7-10-fold higher than surface cells. Without a phosphodiesterase inhibitor proliferative cells had reduced basal cyclic AMP-, cyclic GMP-, and cyclic AMP-dependent protein kinase activity ratios, as well as blunted cyclic AMP responses to prostaglandin E2 and vasoactive intestinal peptide compared to superficial cells. In the presence of 3-isobutyl-1-methylxanthine, basal cyclic AMP and responses to prostaglandin E2 and vasoactive intestinal peptide of proliferative cells exceeded values in superficial cells. This correlated with higher membrane adenylate cyclase activity in the proliferative cells. By contrast, particulate and soluble guanylate cyclase activities of superficial cells were higher than in proliferative cells. The apparent high Km soluble and particulate cyclic AMP and cyclic GMP phosphodiesterase activities of proliferative cells were 4-7-fold higher than those in superficial cells. Moreover, the apparent low Km soluble activity was absent in superficial cells. Thus, an altered rate of nucleotide degradation may mediate reduced cyclic AMP and cyclic GMP in proliferative versus superficial cells. Dibutyryl cyclic AMP, prostaglandin E2 or vasoactive intestinal peptide inhibited [3H]thymidine incorporation into DNA of colonic segments. Thus, reduced cyclic AMP in lower crypt cells may be a determinant of their greater proliferative activity.     

Purification of an insulin-sensitive cyclic AMP phosphodiesterase from rat liver

A low-Km cyclic nucleotide phosphodiesterase solubilised from rat liver membranes by mild proteolysis with chymotrypsin has been purified to apparent homogeneity. The purification included chromatography on cellulose phosphate, Ecteola-cellulose, hydroxyapatite, a theophylline affinity matrix and HPLC on a DEAE-substituted column. The purified enzyme has linear kinetic plots with a Km of 0.24 microM and a Vmax of 6.2 mumol mg-1 min-1 with cyclic AMP as a substrate. It also hydrolyses cyclic GMP with a Km of 0.17 microM and a Vmax which is about a third of that with cyclic AMP. Cyclic GMP is also a competitive inhibitor of cyclic AMP hydrolysis with a Ki of 0.18 microM. The proteolytically solubilised enzyme has a subunit molecular mass of 73 kDa by SDS gel electrophoresis and of 130 kDa by HPLC size-exclusion chromatography, suggesting that it exists as a dimer. A partially purified preparation of this enzyme was used to raise antiserum in a sheep. The antiserum immunoprecipitated activity from liver and adipose tissue of rat and mouse. It had little activity against phosphodiesterase from other rat tissues or other species. Insulin-activated phosphodiesterase from both adipocytes and hepatocytes was immunoprecipitated by the antiserum suggesting that the purified enzyme was an insulin-sensitive phosphodiesterase.     

Effects of phosphodiesterase inhibitors, imidazole and phosphate on cyclic CMP phosphodiesterase are different from those on cyclic AMP and cyclic GMP phosphodiesterases.

The effects of various agents on the newly identified cyclic CMP phosphodiesterase (C-PDE) in crude extracts of a number of rat tissues and on the enzyme partially purified from the rat liver were examined. Papaverine and 1-methyl-3-isobutylxanthine were without effects on C-PDE at concentrations that inhibited up to 90% of cyclic AMP phosphodiesterase (A-PDE) and cyclic GMP phosphodiesterase (G-PDE) activities. When assayed using 1 micron substrates, theophylline inhibited C-PDE to a lesser extent than A-PDE and G-PDE. 2'-Deoxy cyclic AMP (specific A-PDE inhibitor) and 2'-deoxy cyclic GMP (specific G-PDE inhibitor) were relatively poor and non-specific inhibitors for C-PDE. Imidazole, while augmenting the high Km A-PDE and G-PDE from the liver but not from the heart, was without effect on the liver C-PDE but stimulated the heart C-PDE. Potassium phosphate was more specific in inhibiting C-PDE than A-PDE and G-PDE. The present findings suggest that C-PDE represents a potential site of specific pharmacological regulations, and that C-PDE may be a separate enzyme distinguishable from the purine cyclic nucleotide class of phosphodiesterases.     

Adenyl cyclase and cyclic nucleotide phosphodiesterase activities in murine muscular dystrophy.

Adenyl cyclase and cyclic nucleotide phosphodiesterase activities were assayed in homogenates of hind leg skeletal muscle from dystrophic and normal mice. Adenyl cyclase activity was stimulated 2.5 times by epinephrine and 6 times by fluoride over the basal activity in both dystrophic and normal mice. The activity of adenyl cyclase from dystrophic muscle of mice was significantly higher than that of normal mice under all the conditions tested (i.e. basal, epinephrine and fluoride). Cyclic nucleotide phosphodiesterase from skeletal muscle of mice has two Km's (2.1 and 11 mumol/l) which suggests the existence of either two forms of enzyme or a single enzyme with negative cooperativity. The activity of this enzyme was significantly elevated in the skeletal muscle of dystrophic mice compared to the normal controls. The available evidence suggests that the same cyclic nucleotide phosphodiesterase is responsible for the hydrolysis of both cyclic AMP and cyclic GMP.     

A new peptide (1150Da) selectively activates the calcium-calmodulin sensitive isoform of cyclic nucleotide phosphodiesterase from human myometrium.

The cyclic nucleotide phosphodiesterase enzymatic system is examined in extracts of human myometrium and four individual phosphodiesterase isoforms have been isolated and characterized. A new thermostable peptide, recently purified in rat and calf myometrium, is able to stimulate up to 55-fold, the calcium-calmodulin dependent phosphodiesterase isoform. Activation of cAMP hydrolysis is by far the most marked with a 55-fold maximal stimulation at a concentration of 0.1 microM peptide and a IC50 value estimated at 30nM. For cGMP hydrolysis, the maximal effect (x25) obtained at 40nM peptide is lesser and the IC50 value is in the 10nM range. Furthermore, we verified that classical calmodulin antagonists such as calmidazolium or trifluoroperazine did not change stimulation of the calcium-calmodulin phosphodiesterase by the peptide, indicating that the myometrial peptide is different from calmodulin. To our knowledge, this is the first evidence for such a strong and selective stimulation of one isoform of the phosphodiesterase enzymatic system by a natural peptide.     

Biochemical events associated with activation of smooth muscle contraction

Biochemical events associated with activation of smooth muscle contraction were studied in neurally stimulated bovine tracheal smooth muscle. A latency period of 500 ms preceded increases in isometric force and myosin light chain phosphorylation. However, stimulation resulted in the rapid hydrolysis of inositol phospholipids as demonstrated by increases in inositol phosphates by 500 ms. Inositol trisphosphate increased 2-fold with no significant change in inositol tetrakisphosphate. The apparent activation state of myosin light chain kinase was assessed indirectly through measurements of the fractional activation of a second calmodulin-dependent enzyme, cyclic nucleotide phosphodiesterase. The fractional activation of cyclic nucleotide phosphodiesterase increased after neural stimulation to a maximal extent by 500 ms and remained at this level for at least 4 s. The monophosphorylation of myosin light chain increased after 500 ms and reached a maximum value by 2 s. Diphosphorylation also occurred but to a much lesser extent. Fractional activation of cyclic nucleotide phosphodiesterase and myosin light chain phosphorylation both decreased after 10 min continuous stimulation, although the force response remained at a maximal level. These observations demonstrate that inositol trisphosphate formation and activation of cyclic nucleotide phosphodiesterase (and hence most likely myosin light chain kinase) by calmodulin precede myosin light chain phosphorylation and that these events are sufficiently rapid to mediate the contractile response of neurally stimulated tracheal smooth muscle.     

Dibutyryl cyclic AMP increases phosphodiesterase activity in the rat heart

The influence of increasing the in vivo concentration of cyclic AMP on the activity of cyclic nucleotide phosphodiesterase (PDE) in rat heart was investigated. One, three, and five hourly injections of 5.0 mg dibutyryl (Bt2) cyclic AMP significantly increased the activity of PDE in the supernatant fraction of rat heart using 1.0 microM cyclic AMP as the assay substrate concentration. When 100 microM cyclic AMP was used in the assay reaction, increases in enzymes activity were seen following five and eight nucleotide injections. The nucleotide-induced increase in PDE activity was dose dependent. When the five-injection protocol was used, PDE activity remained elevated for at least 4 h, while activity had returned to control levels within this time when two hourly injections were used. The nucleotide stimulation of PDE activity was blocked by cycloheximide. Five hourly infections of Bt2 cyclic AMP increased PDE activity in the liver and fast-twitch red muscle. A reduction in PDE activity in fast-twitch white muscle was seen following nucleotide injections. These findings are consistent with the hypothesis that prolonged elevations in the intracellular concentration of cyclic AMP cause an elevation in myocardial PDE activity. The increased activity seems to be the result of protein synthesis. These data suggest that cyclic AMP contributes significantly in regulating its own metabolism in the rat heart.     

Effect of some smooth muscle relaxant drugs on calcium-related phenomena

Some smooth muscle relaxant drugs devoid of anticholinergic action have been tested for their interaction with calmodulin, calmodulin-stimulated cyclic nucleotide phosphodiesterase activity, and uterine membrane binding sites for nitrendipine and adenosine. The myolytic activity of octylonium bromide and pinaverium bromide may be due to their interaction with calmodulin-dependent systems. Trimebutine maleate does not bind either to calmodulin or to nitrendipine and adenosine receptors. Tiropramide has no effect on calmodulin-dependent systems and on Ca2+ channels but it shows a competition for the A2-type adenosine receptors.     

The isoenzyme selectivity of AH 21-132 as an inhibitor of cyclic nucleotide phosphodiesterase activity.

The smooth muscle relaxant, AH 21-132, was tested for its inhibitory effect on the cyclic nucleotide phosphodiesterase (PDE) activities fractionated from guinea-pig cardiac ventricle and bovine trachealis muscle. Both tissues yielded significant PDE-I and PDE-II activities. The cardiac ventricle also contained a significant amount of PDE-III whilst the trachealis contained PDE-IV. AH 21-132 inhibited PDE-III and PDE-IV selectively (Ki values 0.30-0.55 microM) compared with PDE-I and PDE-II (Ki values 20-140 microM).