Purification and Characterization of Cyclic GMP-stimulated Cyclic Nucleotide Phosphodiesterase from Calf Liver

Cyclic GMP-stimulated cyclic nucleotide phosphodiesterase purified >13,000-fold to apparent homogeneity from calf liver exhibited a single protein band (M. 102,000) on polyacrylamide gel electrophoresis under denaturing conditions. Enzyme activity comigrated with the single protein peak on anaIyticaI polyacrylamide gel electrophoresis, sucrose density gradient centrifugation, and gel filtration. From the sedimentation coefficient of 6.9 S and Stokes radius of 67 A, an M, of 201,000 and frictional ratio (f/fo) of 1.7 were calculated, suggesting that the native enzyme is a nonspherical dimer of similar, if not identical, peptides. The effectiveness of Mg2+, Mn2+, and Coz+ in supporting catalytic activity depended on the concentration of cGMP and cAMP present as substrate or effector. Over a wide range of substrate concentrations, optimal concentrations for M 8 + , MnZ+, and Co2+ were about 10, 1, and 0.2 mM, respectively. At concentrations higher than optimal, Mg2+ inhibited activity somewhat; inhibition by Co2+ (and in some instances by Mn2+) was virtually complete. At low substrate concentrations, activity with optimal Mn2+ was equal to or greater than that with Co2+ and always greater than that with M8+. With 20.5 PM cGMP or 20 to 300 PM cAMP and for CAMP-stimulated cGMP or cGMP-stimulated cAMP hydrolysis, activity with M8+ > Mn2+ > Co2+. In the presence of Mg2+, the purified enzyme hydrolyzed cGMP and cAMP with kinetics suggestive of positive cooperativity. Apparent K , values were 15 and 33 PM, and maximal velocities were 200 and 170 pmol/min/mg of protein, respectively. Substitution of Mn2+ for Mg2+ increased apparent K,,, and reduced Vma, for cGMP with little effect on K , or V,,, for CAMP. Co2+ increased K,,, and reduced V,,, for both. cGMP stimulated cAMP hydrolysis -32-fold in the presence of Mg2+, much less with Mn2’ or Co”. In the presence of Mg2+, Mn2+ and Co2+ at concentrations that increased activity when present singly inhibited cGMPstimulated cAMP hydrolysis. It appears that divalent cations as well as cyclic nucleotides affect cooperative interactions of this enzyme. Whereas Co2* effects were observed in the presence of either cyclic nucleotide, Mn2+ effects were especially prominent when cGMP was present (either as substrate or effector).

stimulated cAMP hydrolysis -32-fold in the presence of Mg2+, much less with Mn2' or Co". In the presence of Mg2+, Mn2+ and Co2+ at concentrations that increased activity when present singly inhibited cGMPstimulated cAMP hydrolysis. It appears that divalent cations as well as cyclic nucleotides affect cooperative interactions of this enzyme. Whereas Co2* effects were observed in the presence of either cyclic nucleotide, Mn2+ effects were especially prominent when cGMP was present (either as substrate or effector).
* A preliminary report of some of this work has appeared in abstract form (1982) (Fed. Proc. 41,760). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduettisemnt" in accordance with 18 U.S.C.
Section 1734 solely to indicate this fact. Intracellular concentrations of cAMP and cGMP are controlled through regulation of synthesis by adenylate cyclase (EC 4.6.1.1) and guanylate cyclase (EC 4.6.1.2), respectively, and degradation by cyclic nucleotide phosphodiesterases (EC 3.1.4.17). Multiple forms of phosphodiesterases with distinct physical, catalytic, and regulatory properties are known (for review, see Refs. 1 and 2). One form, referred to as the cGMPstimulated cyclic nucleotide phosphodiesterase, has been found in several tissues (3-12) and recently purified from bovine heart and adrenal (13). The enzyme is also present in cultured HTC hepatoma cells, where its activity is decreased by dexamethasone (14,15). The enzyme from rat liver hydrolyzed cGMP, cIMP, and cAMP with kinetics suggestive of positive cooperativity. cGMP, at concentrations that increased hydrolysis of cAMP and cIMP, accelerated the rate of inactivation of the enzyme by chymotrypsin ( l l ) , suggesting that on interaction with cGMP the enzyme undergoes conformational changes that are associated with an increase in the rate of cAMP hydrolysis and enhanced susceptibility to proteolytic cleavage in a domain relevant to catalytic function (11). We report here the purification to apparent homogeneity of a cGMP-stimulated cyclic nucleotide phosphodiesterase from calf liver along with a description of some of its hydrodynamic properties and the complex effects of divalent cations on catalytic activity.

cGMP-stimulated Cyclic Nucleotide Phosphodiesterase
phosphodiesterase activity was assayed essentially as described (16) in a total volume of 0.3 ml containing 15 pmol of Hepes, pH 7.5, 15 pmol of NaCI, 2.5 pmol of MgCl2, 15 nmol of EDTA, 150 pg of bovine serum albumin, and 150 pmol of [3H]~AMP or [3H]~GMP (20,000-23,000 cpm). After incubation at 30 "C for 5-15 min, assays were terminated with addition of 0.1 ml of 0.25 N HCI containing 4 mM CAMP (or cGMP) and 5 mM 5'-AMP (or 5"GMP). After neutralization with 0.1 ml of 0.25 N NaOH in 250 mM Tris, pH 8.0, 0.19 mg of 5'-nucleotidase (c. atror venom) in 0.1 ml of 100 mM Tris, pH 8.0, was added, and the mixture was incubated for 20 min at 30 "C. A portion (0.5 ml) was applied to a column (0.7 X 2.5 cm) of DM-Sephadex A-25 followed by 3.2 ml of water. The eluate containing [3H]nucleoside was collected in a counting vial for radioassay. Activities reported represent verifkd initial rates with <20% of substrate hydrolyzed.

Colculatwn of Free Metal Concentrations-Free metal concentrations
were calculated according to Bartfai (18)

K P H~~ = W~H
where a~ is a function of pH and the H+-EDTA stability constants. The stability constants (KpH7.5) for the metal-EDTA chelates at pH 7.5 were 0.81 X 106 M" for Mg-EDTA, 1.82 X 10" M" for Mn-EDTA, and 3.39 X IOL3 M" for Co-EDTA. Using the appropriate Kp~,,& value, concentrations of free metal in the presence of 50 p~ EDTA were calculated (18). The values for free metal concentrations listed in Figs. 5 and 6 must be considered as approximations, since the stability constants on which they are based were not determined under assay conditions. Protein Determination-Protein was assayed using Coomassie brilliant blue G-250 (19) with bovine serum albumin as standard.

Purification of cGMP-stimulated Cyclic Nucleotide
Phosphodiesterase from Calf Liver Data from a representative preparation are shown in Table I.
The specific activity of cGMP-stimulated cyclic nucleotide phosphodiesterase in calf (<4 months old) liver supernatant was twice that in the Supernatant from adult bovine liver.
Step I: Extraction-Fresh calf livers from a local abattoir were transported to the laboratory in ice. All procedures were performed in a cold room at 4 "C unless otherwise stated. Liver (about 6 kg) was homogenized in 2 volumes (w/v) of ice-cold Buffer I (50 mM Hepes, pH 7.5,0.25 M sucrose, 1 mM EDTA, 1 mM NaN3, 2 p~ leupeptin, 1 p g / d of soybean trypsin inhibitor, 0.4 mM PMSF, 1 p M pepstatin A, 10 mM NaF, 1 mM DTT) containing 50 mM NaCl for 60 s in a Waring commercial blender. After adjusting the pH to 6.0 with 1 N acetic acid, the homogenate was centrifuged (23,300 X g, 30 min). The supernatant was filtered through glass wool and adjusted to pH 7.3 with 1 M Hepes, pH 10.
Step 2: DEAE-cellulose (DE52) Chromatography-The extract was divided into four portions, each of which was loaded onto 1.1 liter of DE-52 previously equilibrated with Buffer I containing 50 mM NaCl on a fritted disc funnel. The gel was washed with 2 bed volumes of the same buffer and 5 bed volumes of Buffer I containing 125 mM NaCl. Enzyme activity was eluted with 2.5 bed volumes of Buffer I containing 280 mM NaCl.
Step 4: DEAE-Sepharose CL-GB Chromatography-The dialyzed enzyme was applied to 800 ml of DEAE-Sepharose CL-6B in a glass fritted disc funnel previously equilibrated with Buffer I1 containing 50 mM NaCl. The gel was washed with 2 bed volumes of the same buffer and 5 bed volumes of Buffer I1 containing 125 mM NaC1. The enzyme was then eluted with 2.5 bed volumes of Buffer I1 containing 250 mM NaCl and concentrated to 100 ml with a Pellicon cassette system (Millipore).
Chromatography on DE-52 ( Step 2) separated cGMP-stimulated cyclic nucleotide phosphodiesterase activity from essentially all of the calmodulin-activated phosphodiesterase. Steps 3 and 4 removed most of the CAMP-specific activity, presumably the "low K,,," CAMP phosphodiesterase (20).
Step 5: C?-H2N(CHJ2NH-cAMP (C?-cAMP)-Agarose Chromatography-The concentrated enzyme from DEAE-Sepharose CL-GB was diluted 5-fold with Buffer I1 (final concentration, 50 mM NaC1) and applied to 150 ml of C'-cAMP-agarose (in a glass fritted disc funnel) previously equilibrated with Buffer I1 containing 50 mM NaCl (flow rate, 250 ml/h). The gel was washed with 2 bed volumes of the same buffer. The material that did not bind to the Cs-CAMP-agarose and the wash fraction were pooled and concentrated to 100 ml using a PeUicon cassette system.
Although phosphodiesterase activity was not retained on columns of C8-CAMP-agarose (Table I, Step 5), other proteins, which would otherwise have reduced the purity of the final preparation by at least 50%, did bind to the matrix (data not shown). If the CS-CAMP-agarose column (Step 5) was not employed, these proteins were major contaminants in the purified phosphodiesterase. With the introduction of chromatography on DEAE-Sepharose CL-GB (Step 4) into the purification procedure, chromatography on C8-CAMP-agarose as described (Step 5) effectively removed contaminating cyclic nucleotide-binding proteins.

TABLE I
Purification of a cyclic GMP-stimulated cyclic nucleotide phosphodiesterase from calf liver Data are from one preparation (6.6 kg of calf liver) representative of seven others. Step

cGMP-stimulated Cyclic Nucleotide Phosphodiesterase
tography-The concentrated enzyme from Step 5 was diluted 5fold with Buffer I1 (final concentration, 10 mM NaCl) and adjusted to pH 6.0 with 0.5 N acetic acid. After centrifugation at 23,300 X g for 30 min, the supernatant was applied to a column (2.5 x 3.0 cm) of Nfi-CAMP-agarose equilibrated with Buffer I11 (50 mM Hepes, pH 6.0,l mM EDTA, 1 mM NaNs, 2 p~ leupeptin, Analytical polyacrylamide gel electrophoresis of purified cGMP-stimulated cyclic nucleotide phosphodiesterase. Samples of enzyme (30 &gel) were applied to two cylindrical (0.6 X 8 cm) polyacrylamide gels (2.5% stacking gel, 5% resolving gel) and subjected to electrophoresis according to Davis (29) a t 4 "C for 6 h a t 200 V. One g e l was stained for protein with Coomassie brilliant blue R-250 and Am was monitored by Zeineh soft laser scanning densitometer (---). The second gel was manually divided into 2.0-mm slices which were extracted with 0.5 ml of 50 mM Hepes (pH 7.3) containing 1 mM EDTA, 1 mM NaN3, 10 mM NaF, 2 mM D m , 0.4 mM PMSF, 1 p M pepstatin A, 2 p M leupeptin, 1 pg/ml of soybean trypsin inhibitor, and 50 mM NaCI. After 12 h a t 4 "C, portions were assayed for phosphodiesterase activity with 0.5 p~ [3H]cAMP in the presence of 1 p~ cGMP (M). Activity (23% of that applied) was recovered only in fractions 15-21.    (1) were precipitated, solubilized, and subjected to SDS-gel elec-volumes of Buffer I11 (to remove NaCl). The enzyme was eluted with 3 bed volumes of Buffer I11 containing 500 PM cGMP.
Columns of C"-CAMP-and N'-cAMP-agarose were regenerated by washing with 5 bed volumes of 6 M urea, pH 4.5, containing 1 M NaCl and 10 mM EDTA (which eluted strongly adsorbed proteins not removed by salt or cGMP) and 10 bed volumes of water before equilibration with appropriate buffer solution. Columns were used repeatedly with no apparent loss of capacity to remove contaminants or retain enzyme activity.
cGMP-stimulated phosphodiesterase activity was not retained by C8-CAMP-agarose at pH 7.3 or pH 6.0 nor by N6-CAMPagarose at pH 7.3. It was retained on N'-cAMP-agarose at pH 6.0 and eluted at pH 6.0 with 500 PM cGMP in the absence of NaCl at 4 "C. Elution with 50 PM cGMP at pH 7.3 or with 500 PM cAMP at pH 6.0 or 7.3 yielded preparations of lower purity (data not shown).
Step 7 : Concentration of Enzyme and Removal of G M P -Enzyme from Step 6 was adjusted to pH 7.0 with 1 M Hepes, pH 7.8, and applied to a column (0.8 x 1.8 cm) of DEAE-Sepharose CL-GB equilibrated with Buffer IV (50 mM Hepes, pH 7.5, 1 mM NaN3, 0.4 mM PMSF, 1 pM pepstatin A, 10 mM NaF, 1 mM DTT) containing 50 mM NaCl. The column was washed with 30 bed volumes of Buffer IV containing 120 mM NaCl to remove cGMP. Enzyme was eluted with 2.0 ml of Buffer IV containing 250 mM NaCl and extensively dialyzed against Buffer IV containing 50 mM NaC1.
Using this procedure, the enzyme was purified 13,000to 20,000-fold from calf liver supernatant with a yield of 5-10% (Table I). Assuming that cGMP-stimulated activity can be accurately assessed in crude extracts, it may be calculated that calf liver contains 2.5-3.0 mg of soluble cGMP-stimulated cyclic nucleotide phosphodiesterase per kg of tissue.

Physicochemical Properties of the Purified Enzyme
The purified enzyme exhibited a single protein band (M, = 102,000) on SDS-polyacrylamide slab gel electrophoresis (Fig.  1). Phosphodiesterase activity was associated with a single protein band after analytical polyacrylamide disc gel electrophoresis (Fig. 2). After sucrose density gradient centrifugation (Fig. 3) or chromatography on Sepharose CL-GB (Fig. 4), the fractions that hydrolyzed ["HIcGMP contained a single protein-staining band     Table 11, the sedimentation coeffkient of the purified enzyme was tstimated at 6.9 S (Fig. 3, inset) and the Stokes radius at 67 A (Fig. 4, inset). Assuming a partial specific volume of 0.74 g/ml, the estimated molecular weight was 201,000 and the frictional coefficient 1.7. The enzyme preparation was stable for at least four months at -70 "C in the presence of 15% glycerol and ovalbumin, 5 mg/ml.

Effects of Divalent Cations on Kinetic Properties
The At all substrate concentrations of cAMP (Fig. 5) or cGMP (Fig. 6) tested, the optimal free M P concentration was -10 mM; with higher concentrations, activity was somewhat lower. At all substrate concentrations (Figs. 5 and 6), Mn2+ was effective at lower concentrations than was M$' ; at higher substrate concentrations of cAMP (Fig. 5 ) or cGMP (Fig. 6), optimal concentrations of Mn2+ were lower than those of M e . At 500 PM cAMP (Fig. 5 0 ) or cGMP (Fig. 6D), activity with low (-10 nM free metal) Mn2+ was 80-90% that at optimal Mn2+. Increas-

cGMP-stimulated Cyclic Nucleotide
Phosphodiesterase 12531 Co" (  ing Mn2+ above 10 mM decreased activity. co2+ was relatively more effective at lower substrate concentrations. At 2 nM cAMP (Fig. 5A), maximal activity with Co2+ or Mn2+ was almost twice that with optimal M$+; at 2 nM cGMP (Fig. 6A), activity with optimal Co2+ was equal to that of M e and only slightly less than that with optimal Mn2+. Except with the highest concentration of substrate (Figs. 5 0 and 6D) where activity with very low free Co2+ was 60-70% of that at optimal Co2+, this ion activated over a relatively narrow concentration range. Increasing Co2+ above its optimal concentration led to virtually complete inhibition (Figs. 5 and 6). With cAMP as substrate (Fig. 5, A, B, and D), maximal activity with MnZ+ was greater than or equal to that with Co2+ or Mg2+ except when cGMP was present with 0.5 p~ [3H]cAMP (Fig. 5C). In this instance, the order for maximal activity was Mg2+ > Mn2+ > Co2+. The same order was observed with 0.5 or 500 I.~M [3H]cGMP as substrate (Fig. 6, C and D) or with 2 nM ["HIcGMP plus 40 p~ cAMP (Fig. 6B). With 2 nM [3H]cGMP (Fig. 6A), hydrolysis with optimal Mn2+ was slightly greater than with Mg2+ or Co2+, which were similar.
Effects of substrate concentration on velocity in assays with M$+, Mn2+, or Co2+ are shown in Figs. 9 (cGMP) and 10 (CAMP). Kinetic constants derived from these data are summarized in Table 11. Hill coefficients ranged from 1.2 to 1.8 for cAMP and 1.2 to 1.5 for cGMP in the presence of the different cations, suggesting positive homotropic cooperativity for hydrolysis of both nucleotides (Figs. 9 and 10, Table 11).
Hydrolysis of 0.5 p~ [3H]cAMP was maximally stimulated, by 2 to 4 p~ cGMP, 32-fold with M e , 9-fold with Mn2+, and 6-fold with Co2+ (Fig. 11). The order of activity with 0.025 p~ cGMP was Mn'+ > Co2+ > Mg2+ which was the same order seen in the absence of cGMP. The apparent activation constant for cGMP was -0.5 p~ with M F and somewhat lower in the presence of MnZ+ (0.1 pM) or co2+ (0.2 p~) (Fig. 11). With 2 or 40 nM [3H]cGMP as substrate, activity was maximally increased by 20 to 50 PM cAMP higher concentrations inhibited (Fig. 12). Hydrolysis of 0.5 PM [3H]cGMP was not stimulated by cAMP but was decreased with concentrations in excess of 10 p~ (Fig.  12).

DISCUSSION
The enzyme described here seems similar to the cyclic GMP-stimulated cyclic nucleotide phosphodiesterase identified in rat liver supernatant and particulate fractions by a number of investigators (3, 7 , 11, E ) , although the relationship between soluble and particulate forms remains to be established. Martins et al. (13) recently reported the purification of this enzyme from bovine heart and adrenal by chromatography on cGMP-epoxy-activated Sepharose. They found that the enzyme was not retained by N6-H2N(CH2)2-cAMP attached to CNBr-activated Sepharose. The enzyme described here, however, could be bound to N6-H2N(CH2)2-CAMP-agarose at pH 6.0 (not at pH 7.3) and eluted at pH 6.0 with cGMP, suggesting that cAMP binding involves hydrogen bond(s) and/or conformational changes in the enzyme influenced by pH. Although the enzyme was not retained by C8-CAMP-agarose, several CAMP-binding proteins were selectively adsorbed and efficiently separated from the cGMPstimulated phosphodiesterase in this way. Binding of the cGMP-stimulated phosphodiesterase to N6but not C8-CAMP-agarose and the retention of presumed regulatory subunits of protein kinase by C8-CAMP-agarose were consistent with the structural requirements for activation of the two enzymes by analogs of CAMP. Derivatives of cAMP modified at the N6but not the C8-position activated the cGMPstimulated phosphodiesterase (11) and C8 derivatives activated protein kinase (21). Different phosphodiesterases differ in their ability to interact with immobilized cyclic nucleotide derivatives. The calmodulin-activated and "CAMP-specific'' phosphodiesterases from calf liver were not retained on either C8or N6-cAMP-agarose.2 Helfman et al. ( 2 2 ) reported that cCMP phosphodiesterase activity was bound to C8-cAMPagarose.
The purified enzyme exhibited a single protein band ( M , = 102,000) on SDS gels. Enzyme activity comigrated with the single protein peak on electrophoresis under nondenaturing conditions, sucrose density gradient centrifugation, and gel filtration. From the sedimentation coefficient and Stokes radius, an Mr of 201,000 and frictional ratio of 1.7 were estimated. Although the frictional coefficient depends on the extent of hydration of a protein as well as its shape (axial ratio), such a relatively high ratio usually cannot be explained on the basis of hydration alone (23). It appears that the native enzyme exists as a nonspherical dimer of similar, if not identical, subunits. The high frictional ratio may account for the higher molecular weight (size) estimated for the phosphodiesterase on the basis of gel filtration (7, 151 or electrophoresis under nondenaturing conditions (13) alone.
The calf liver phosphodiesterase, like the purified heart enzyme (13) and partially purified preparations from several sources (3,5-7,10,11,24), hydrolyzed cAMP and cGMP with kinetics suggestive of positive cooperativity. Maximal velocities with both substrates were higher than any reported for this enzyme. As found with partially purified preparations (11,24), cGMP appears to be preferred as both substrate and activator of the purified enzyme. The activity of a purified cGMP phosphodiesterase from guinea pig lung (25) and of crude preparations from bovine heart (26) was increased by M$+, MnZ+, or Co2+. With the enzyme described here, there were marked differences in the effects of Mg2+, Mn2+, and Co2+ depending on the concentrations of cAMP and cGMP present as substrate and/or effector. Over a wide range of nucleotide concentrations, maximally effective concentrations of M F , Mn2+, and Co2+ were about 10, 1, and 0.2 mM, respectively. At low substrate concentrations ( 2 nM cGMP or 40.5 p~ CAMP) with optimal concentrations of each ion, activity with Mn2+ was similar to or greater than with Coz+ and always exceeded that with M$+. With >0.5 p M cGMP or 20 to 300 p~ cAMP and with maximal cGMP stimulation of CAMP hydrolysis or cAMP stimulation of cGMP hydrolysis, the order of activity was Mg2+ > Mn2+ > Co2+. In the presence of 10 mM Mg2+, Mn2+ and Coz+ increased activity under conditions where, if ions were present singly, the order of activity would have been Mn2+ > Co2+ > M$+ and decreased activity where the order would have been M$+ > Mn2+ > Co'+.
With >300 PM CAMP, Mn2+ and Mg2+ were equally effective, i.e. maximal velocities with cAMP as substrate were the Same with Mn2' and M e . Substitution of Mn2+ for Mg", which had little or no effect on the apparent K , for CAMP, increased the apparent K,,, for cGMP and decreased Vmax. In fact, with Mn2+ maximal velocities and affinities were very similar for cAMP and cGMP (Table 11) were especially prominent in the presence of cGMP, perhaps because cGMP-induced changes in conformation and activity were reduced by Mn2+ or rendered the enzyme less responsive to Mn2+. Substitution of Co2+ for M$+, on the other hand, increased the apparent K , for cAMP and cGMP and reduced maximal velocities with both.
These differences in effects of the metals on maximal velocities were not apparently related to substrate-induced alterations in concentration dependence of metal activation, since, at high substrate concentrations, very low concentrations of Mn2+ and Co2+ were as nearly effective as optimal concentrations. Some of the "inhibitory" effects (relative to M P ) of Mn2+ on hydrolysis of cGMP and of Co2+ on hydrolysis of cAMP and cGMP were probably related to effects on the affinities for these substrates; however, the metals must also have other effects. Except at high substrate concentrations, activation by Coz+ occurred over a narrow concentration range; high concentrations of Co2+ and in some circumstances of Mn2+ produced almost complete inhibition. Even MgZ+ at concentrations above optimal decreased activity. Although concentrations of Mn2+ and Co2+ well within the physiological range (27, 28) can support catalytic activity of the cGMPstimulated phosphodiesterase, much higher concentrations are required to affect enzyme activity in the presence of Mg". Thus, the physiological significance of the effects of these ions remains to be established. Nevertheless, the different effects of these ions on cooperative properties of the enzyme and on the complex interactions of cAMP and cGMP at what are presumably distinct regulatory and catalytic sites (24) might be useful in studies designed to elucidate the molecular mechanisms for regulation of the activity of this phosphodiesterase.