site was sensitive to a guanine-type cyclic nucleotide structure. Comparing the two ... Our data have suggested that the activating site is cGMP specific and distinct from the ... Enzymes. Cyclic-nucleotide phosphodiesterase or 3',5'-cyclic-.
Eur. J. Biochem. 136, 571 - 575 (1983) 0 FEBS 1983
Characterization of phosphodiesterase catalytic sites by means of cyclic nucleotide derivatives Dominique COUCHIE, Georg PETRIDIS, Bernd JASTORFF, and Christophe ERNEUX Institute of Interdisciplinary Research, School of Medicine, Free University of Brussels and Fachbereich Biologie-Chemie, Universitat Bremen (Received July 26, 1983) - EJB 83 0811 Cyclic nucleotide derivatives have been used as a tool to characterize distinct catalytic sites on phosphodiesterase enzyme forms : the cGMP-stimulated enzyme from rat liver and the calmodulin-sensitive enzyme from rat or bovine brain. Under appropriate assay conditions, the analogues showed linear competitive inhibition with respect to cAMP (adenosine 3’,5’-monophosphate) as substrate. The inhibition sequence of the fully activated cGMP-stimulated phosphodiesterase was identical to the inhibition sequence of the desensitized enzyme, i. e. the enzyme which has lost its ability to be stimulated by cGMP. The inhibition pattern could, therefore, not be attributed to competition with cGMP at an allosteric-activating site. Also, the inhibition sequence of the calmodulin-sensitive phosphodiesterase was maintained whether activity was basal or fully stimulated by calmodulin. When cAMP and cGMP, with identical chemical ligands substituted at the same position, were compared as inhibitors of the calmodulin-sensitive phosphodiesterase, the cGMP analogues were always the more potent suggesting that, for that enzyme, the catalytic site was sensitive to a guanine-type cyclic nucleotide structure. Comparing the two phosphodiesterases, it was possible to establish both similar and specific inhibitor potencies of cyclic nucleotide derivatives. In particular, the two enzymes exhibited large differences in analogue specificity modified at C-6,6-chloropurine 3‘,5‘-monophosphate or purine 3‘,5’-monophosphate. Cyclic-nucleotide phosphodiesterases exist in multiple forms distinct from each other on the basis of their substrate specificity, physical properties or control mechanisms (for reviews, see [I, 21). In mammalian tissues, two of the major forms of the phosphodiesterases can be extensively purified respectively from brain and adrenal medulla. The brain phosphodiesterase is specifically characterized by its stimulation by calmodulin in the presence of Ca2+ [3, 41. The apparent activation constant for Caz+ at saturating calmodulin is about 0.1 pM . It is called the calmodulinsensitive phosphodiesterase [ 1, 21. The adrenal medulla phosphodiesterase is characterized by positively cooperative kinetic behavior with respect to both substrates cGMP and cAMP . At micromolar concentrations, cAMP hydrolysis is stimulated by cGMP . The samephosphodiesterase can be purified from other tissues including rat liver [7, 81. It is called the cGMPstimulated phosphodiesterase [l, 21. In a previous study of the rat liver cGMP-stimulated phosphodiesterase, we have used multiple analogues of the cyclic nucleotides and have compared their stimulatory effect with cGMP on cAMP hydrolysis . Our data have suggested that the activating site is cGMP specific and distinct from the catalytic site . In this report, we compare the catalytic sites of two purified phosphodiesterase forms : the calmoddinsensitive and the cGMP-stimulated forms. The K , of both forms is lower for cGMP than for cAMP [l, 21. In order to
establish the distinct identities of the two phosphodiesterases, we attempted to demonstrate differences between catalytic sites by means of analogues of cGMP and CAMP. The analogues were chosen to identify quite intrinsic molecular interactions at the binding site of cyclic nucleotides 19 - 111. Comparing the two phosphodiesterases, we observed both similar and specific inhibitor potencies of the analogues. Some of these data have been reported elsewhere in preliminary form [12, 131. MATERIALS AND METHODS Phosphodiesterase preparation
The rat liver cGMP-stimulated phosphodiesterase was prepared as in . The bovine brain calmodulin-sensitive phosphodiesterase was isolated as follows : all buffers, except that used for affinity chromatography, contained the following protease inhibitors : 75 mg/l PhMeS02F (solubilized first in 1.5 ml dimethyl sulfoxide), 10 mM benzamidine and 5 pM leupeptin. Bovine brain was collected in a local slaughterhouse just after killing and immediately put in the following ice-cold buffer medium: 20 mM Tris/HCl (pH 7 . 9 , 2 mM MgCI2, 7 mM 2-mercaptoethanol. In general 150 g tissue were homogenized with a motor-driven teflon/glass homogenizer (1 g brain tissue/5 ml buffer medium). This preparation was filtered through two layers of cheesecloth and centrifuged first at 2400 x g for 20 min then at 40000 x g for 60 min. The supernatant was fractionated by DEAE-cellulose chromatogAbbreviations. dcAMP, 2-deoxyadenosine 3’,5‘-monophosphate; dcGMP, 2’-deoxyguanosine 3‘,5‘-monophosphate; PhMeS02F, raphy (6 x 20 cm) equilibrated with the following buffer: 20 mM Tris/HCl (pH 7 . 9 , 7 mM 2-mercaptoethanol, 0.05 M phenylmethylsulfonyl fluoride. Enzymes. Cyclic-nucleotide phosphodiesterase or 3’,5’-cyclic- (NHJ2S04. The column was washed with 1 volume of the nucleotide 5’-nucleotidohydrolase (EC 184.108.40.206); 5’-nucleotidase or equilibrating buffer, followed by 1 volume of the same buffer 5’-ribonucleotide phosphohydrolase (EC 220.127.116.11). made 0.08 M (NH,),SO,. The enzyme was then eluted with
572 1.5 volumes of the same buffer made 0.15 M (NH4),SO4. The enzyme was immediately concentrated by ammonium sulfate precipitation (436 g/l) m the presence of 1 mM EDTA. The pellet was resuspended in a minimum volume of 40 mM Tris/HCl (pH 7.9, 50 mM NaCl, 3 mM MgCl,, 7 m M 2-mercaptoethanol and dialyzed overnight against the same buffer. Affinity chromatography on a (0.9 x 12 cm) column of calmodulin-Sepharose was carried out as described in [I 41. Active fractions from the EGTA eluate were pooled and stored in the presence of 4 mg bovine serum albumin/ml at - 80 “C. Assay procedures
Assay of cyclic-nucleotide phosphodiesterase activity was performed as described in detail in . Whether the incubation mixture of the first step of the assay is made 10 % ethanol o r not, none of the analogues tested interfered with the effectiveness of the 5’-nucleotidase step. All the results are means of triplicate determinations. Protein concentration was measured by the method of Peterson [I51 using bovine serum albumin as standard.
Table 1. Inhibition sequence of cyclic-nucleotide phosphodiesterases The inhibition sequences were obtained from K,values (pM). Ki of each compound was determined from substrate/velocityrelationships with cAMP as substrate. Saturated calmodulin was added to the assay of the calmodulin-sensitive phosphodiesterase, whereas activity of the cGMP-stimulated form was measured in the presence of 3 pM cGMP (effector). The derivatives were added in the 1-200 pM range. At 200 pM, benzimidazole 3’,5’-monophosphate is ineffective as an inhibitor of the calmodulin-sensitive phosphodiesterase (n.d., not detectable). Analogue number refers to Fig. 1. Results are expressed as means f SEM Enzyme
Analogue Analogue number
Calmodulin- 14 sensitive 9 13 10
15 11 17
[8-3H]cAMP (specific activity 30 Ci/mmol) and [fb3H]cGMP(specific activity 21 Ci/mmol) were obtained from the Amersham International and purified on Dowex-50 cationexchange resin columns . 5’-Nucleotidase (Crotalus atrox venom) and PhMeS0,F were purchased from Sigma Chemical Company (St Louis). Leupeptin was supplied by Peptide Institute (Osaka). QAE-Sephadex A-25 and cyanogenbromide-activated Sepharose 4 B were from Pharmacia (Uppsala). DEAE-cellulose was from Whatman. Fatty-acidpoor bovine serum albumin was purchased from Serva (Heidelberg). The analogue compounds 1, 2, 4, 7, 8, 9, 10, 12, 15, 16 (see Fig. 1) were purchased from Boehringer (Mannheim). Compound 3 was a generous gift of D r R. Hanze (Upjohn Co.). Compounds 5,6,11,13,28,19 were synthesized as described [17- 211. The compound 14 was synthesized in the Bremen Laboratory, according to the procedures described for the correspondent CAMP-derivative [I 7, 22, 231.
6 18 2
19 5 cGMP stimulated 14
9 18 11 6
RESULTS AND DISCUSSION Kinetics of phosphodiesterase enzymes in the presence of cyclic nucleotide derivatives
Kinetics of the liver cGMP-stimulated phosphodiesterase have been studied with respect to cAMP as substrate and in the presence of cGMP as effector [24,25]. At 3 pM cGMP shifts the cooperative kinetic behavior of the enzyme to normal Michaelis-Menten behavior [7, 81. Under such conditions the analogues of Table 1 yielded linear competitive inhibition with respect to CAMP. The kinetic data are shown for 5’-amino-5‘deoxyadenosine 3‘,5’-monophosphate (compound 5 in Fig. 1). A Ki value of 3 pM is calculated from the slopes of the primary reciprocal plot (Fig. 2A inset) so that potency for inhibition can be compared (Table 1). In another series of experiments, we have studied the desensitized cGMP-stimulated phosphodiesterase. cAMP hydrolysis, measured in the presence of 10% ethanol, was no longer activated by cGMP ( and data not shown). Under such assay conditions, we obtained an inhibition pattern with a sequence identical to that obtained from Ki determinations (Table 1). We concluded that the inhibition
17 15 13 19
5’-amino-5’-deoxycGMP cGMP 3’-amino-3’-deoxycGMP cIMP dcGMP 7-deaza-cGMP 2‘-0-(2,4-dinitrophenoxy)-cGMP 5’-amino-5’-deoxycAMP 3’-amino-3’-deoxyCAMP purine 3‘,5’-monophosphate 6-chloropurine 3’3’monophosphate benzimidazole 3’3mono phosphate 5’-amino-5‘-deoxycAMP 5’-amino-5’-deoxycGMP 6-chloropurine 3’,5’monophosphate cIMP cGMP purine 3’,5’-monophosphate 7-deaza-cGMP 3’-amino-3’-deoxycAMP 2’-0-(2,4-dinitrophenoxy)-cGMP dcGMP 3’-amino-3’-deoxycGMP benzimidazole 3’3’monophosphate
PM 0.67 f.0.05 3.3 f 0.2 4.4 f 0.8 5.6 f 0.4 9.4 0.4 10.0 f 0.8 15.5 f 0.4 22.0 f 1.3 159.0 & 7.9 175.6 f13.2 297.0 i 11.2 n. d. 3.0 f 0.1 4.3 f 0.1 20.7 f 0.5 24.9 f 1.3 25.8 f 1.1 44.2 f 6.5 48.9 f 3.3 54.3 f. 3.6 60.5 f 3.9 63.2 f 5.3 145.9 f12.9 275.0 f 12.4
pattern of the cGMP-stimulated enzyme could not be attributed to direct competition with cGMP at the activating site but rather resulted from competitive inhibition at the catalytic site. The same approach was applied to the bovine brain calmodulin-sensitive phosphodiesterase. Activity was measured in the presence of an excess of C a Z + and calmodulin where linear patterns on double-reciprocal plots were obtained as a function of CAMP ([I41 and Fig. 2B). As previously stated, the analogues that have been tested showed linear competitive inhibition (kinetic data shown for 7-deazaguanosine 3‘,5‘monophosphate, Fig. 2B) and a sequence of inhibitors was
C A M P
Fig. 1. Chemical structures of analogues of cAMP and cGMP. Compound numbers defined in Table 1 are given in parenthesis Calmodulin-sensitive enzyme
2 1 6 IAnaIoguel (pM\
Fig. 2. ( A ) Inhibitory effect of 5'-amino-5'-deoxyadenosinejl',S-monophosphate (compound 5 ) on CAMP hydrolysis by the cGMP-stimulated raf liver phosphodiesterase (double-reciprocal plot). ( B ) inhibitory effect of 7-deazaguanosine 3',5'-monophosphate (compound 1 1 ) on cAMP hydrolysis by the calmodulin-sensitive bovine brainphosphodiesterase. (A) Activity was measured in the 5- 100 pM range for cAMP in the presence of 3 pM cGMP (effector). (B) Activity was measured in the presence of a saturated concentration of calmodulin. cAMP concentration ranges shown are 10- 250 pM. The insets show the replots of the slopes of the primary plots as a function of inhibitor concentrations. Results are means
established (Table 1). The same sequence was obtained whether the calmodulin-sensitive phosphodiesterase was purified from rat or bovine tissue. The sequence was also unmodified when phosphodiesterase activity was measured in the absence of calmodulin, i. e. at basal activity (data not shown). Comparison between the two phosphodiesterase forms
Major differences are found between the phosphodiesterases in comparing Kivalues of several derivatives (Table 1) : (a) 6-chloropurine 3',5'-monophosphate shows more selectivity for inhibition of the cGMP-stimulated form (Ki = 20.7 & 0.5 FM) than for the calmodulin-sensitive form (Ki = 297.0
11.2 pM), (b) 3'-amino-3'-deoxy-cGMP is about equipotent with cGMP as inhibitor of the calmodulin-sensitive enzyme but is much less potent than cGMP as inhibitor of the cGMPstimulated enzyme, (c) the order of potency for inhibition of the two 3'-amido derivatives of cAMP and cGMP (respectively compounds 6 and 13) is discordant for the two enzymes etc. In comparing the two sequences of inhibitors, it is obvious that the sequence is specific for each phosphodiesterase. Other analogues of the substrate are equally recognized without any selectivity of binding to a given enzyme (Table 1) : (a) both forms are strongly inhibited by 5'-amino-5'-deoxycGMP, which suggests that the 5'-oxygen atom is not involved in the binding of cGMP to the catalytic site of either phos-
574 phodiesterase, (b) 5'-amido derivatives of cGMP or cAMP always compete better with the hydrolysis of cAMP than the corresponding 3'-amido isomers, (c) benzimidazole 3 ' 3 monophosphate is the least potent inhibitor of both phosphodiesterases although it is well recognized and bound to the CAMP-dependent protein kinase [lI]. We conclude that the phosphodiesterase enzymes may contain very similar domains in their catalytic sites so that some chemical interactions, e. g. those involved in the binding of the cyclic phosphate-ribose moiety of cGMP or cAMP are conserved. The CAMPdependent protein kinases exist in two isoenzyme forms. They also share conserved protein structures at the CAMP-binding sites: the overall pattern of analogue specificity for site I and 2 of type I is similar to that of type I1 . Specificity of cGMP binding at the catalytic site of the calmodulin-sensitive phosphodiesterase
A specificity towards cGMP at the active site of the calmodulin-sensitive phosphodiesterase is established as follows: cAMP and cGMP substituted at the same position (C-8, 0-2, 0-3' and 0-5')with identical chemical structures are compared as competitive inhibitors of cAMP hydrolysis. The guanine-type compounds are more potent inhibitors than the adenine-type compounds. Such a relationship does not apply for the cGMP-stimulated enzyme. The data are shown for 3'- and 5'-amido derivatives of cGMP and CAMP:3'-amino-3'deoxy-cGMP is much potent than 3'-amino-3'-deoxy-cAMP as an inhibitor of the calmodulin-sensitive phosphodiesterase. It is not if the comparison is made with the cGMP-stimulated enzyme (Table 1). Similar conclusions are obtained in comparing 8-bromo-cGMP with 8-bromo-CAMP and dcGMP with dcAMP (data not shown). The reasons of specificity for cGMP at the catalytic site of the calmodulin-sensitive enzyme are not known. Possibilities of hydrogen-bond interaction at each individual atom position are different when comparing cGMP and cAMP and could at least partially account for the specificity . In hydrogen bonding, the 6-amino group in cAMP can act as a donor, while the carbonyl group at the same position in cGMP is clearly an acceptor. Actually the two phosphodiesterases exhibit large differences in analogue specificity modified at C-6. The loss of the carbonyl function (at C-6) of cIMP markedly affects the selectivity of derivatives : 6-chloropurine 3',5'-monophosphate and purine 3',5'-monophosphate are excluded from the catalytic site of the calmodulin-sensitive enzyme whereas 6-chloropurine 3'3'monophosphate is as potent as cIMP as an inhibitor of the cGMP-stimulated form (Table 1).
CONCLUSIONS Phosphodiesterases exist in multiple forms that recent evidence of peptide mapping shows as probably distinct molecular entities . In this study, we have compared the catalytic sites of two purified mammalian phosphodiesterases: the calmodulin-sensitive and the cGMP-stimulated forms. Analogues (e. g. 6-chloropurine 3',5'-monophosphate) are found that bind preferentially to a given enzyme, which suggests a specific type of interactions between the cyclic nucleotide substrates and the two phosphodiesterases. Nevertheless, the two catalytic sites are found to be to some extent similar in that the potency for inhibition of several analogues does not vary from one enzyme form to another. One example was given by comparing S-amino-S'-deoxy-cGMP
with cGMP. We have suggested that the S'-oxygen in the ribose moiety is not involved in hydrogen bonding at either catalytic site of the phosphodiesterases. This contrasts with the CAMPdependent protein kinase type I : as proven with the same analogues, hydrogen bonds to the 5' and 3'-oxygen contribute to the binding of the ribose moiety to one of the CAMP-binding sites [lo, 111. Phosphodiesterases with similar activities towards cAMP and cGMP as substrate are often compared as homologous proteins : e. g. the properties of a phosphodiesterase purified from S49 mouse lymphoma cells were found to resemble the characteristics of an enzyme previously and independently isolated from lymphocytes and leukemic cells . The present data emphasize the potential use of cyclic nucleotide derivatives as a tool to compare multiple forms of phosphodiesterase and to map their active sites for comparison with other enzymes involved in cyclic nucleotide metabolism or action. Work realized under contract of the Ministkre de la Politique Scientifique (Actions Concertees). C. Erneux is Charge de Recherches au Fonds National de la Recherche Scientifique. G. Petridis and B. Jastorff are supported by the grant (Ja 246/4-3) of the Deutsche Forschungsgerneinschaftand by the Fond der Chemischen Industrie. We are grateful to Dr J. E. Dumont for criticism and to Dr Morr and Dr Seela for their participation in the synthetic work. The authors would like to thank Mrs C. Moreau for technical assistance and Mrs D. Leemans for typing the manuscript.
REFERENCES 1. Vaughan, M., Danello, M. A., Manganiello, V. C. & Strewler, G. J. (1981) Adv. Cyclic Nucleotide Res. 14, 263-271. 2. Beavo, J. A,, Hansen, R. S., Harrison, S. A., Hurwitz, R. L., Martins, T. J. &Mumby, M. C. (1982) Mol. Cell. Endocrinol. 28,
387-410. 3. Morrill, M. E., Thompson, S. T. & Stellwagen, E. (1979) J . B i d . Chem. 254, 4371 - 4374. 4. Sharma, R. K., Wang, T. H., Wirch, E. & Wang, J. H. (1980) J. Biol. Chem. 255, 5916- 5923. 5. Huang, C. Y., Chan, V., Chock, P. B., Wang, J. H. & Sharma, R. K . (1981) Proc. Natl Acad. Sci. USA, 78, 871-874. 6. Martins, T. J., Mumby, M. C. & Beavo, J. A. (1982) J. Biol. Chem. 257, 1973- 1982. 7. Beavo, J. A,, Hardman, J. G. & Sutherland, E. W. (1971) J. Biol. Chem. 246, 3841- 3846. 8. Terasaki, W. L. &Appleman, M. M. (1975) Metabolism, 24,311 -
319. 9. Erneux, C., Couchie, D., Dumont, J. E., Baraniak, J., Stec, W. J., Garcia Abbad, E., Petridis, G. & Jastorff, B. (1981) Eur. J . Biochem. 115, 503 - 510. 10. Jastorff, B., Hoppe, J. & Morr, M. (1979) Eur. J . Biochem. 101, 555-561. 11. de Wit, R. J. W., Hoppe, J., Stec, W. J., Baraniak, J. & Jastorff, B. (1982) Eur. J. Biochem. 122, 95-99. 12. Couchie, D., Erneux, C. & Jastorff, B. (1982) Arch. Int. Physiol. Biochem. 90, 178- 179. 13. Erneux, C., Couchie, D., Dumont, J. E. & Jastorff, B. (1983) Adv. Cyclic Nucleotide Res. 16, in the press. 14. Klee, C. B., Crouch, T. H. & Krinks, M. H. (1979) Biochemistry, 18, 722-729. 15. Peterson, G. L. (1977) Anal Biochem. 83, 346-356. 16. Hardman, J. G. & Sutherland, E. W. (1969) J. Biol. Chem. 244, 6363 - 6370. 17. Jastorff, B. & Freist, W. (1974) Bioorg. Chem. 3, 103-113. 18. Morr, M., Kula, M. R., Roesler, G. & Jastorff, B. (1974) Angew. Chem. 86, 308.
575 19. Yagura, T.S.,Kazimierczuk, Z., Shugar, D. & Miller, J. P. (1980) Biochem. Biophys. Res. Commun. 97, 731-743. 20. Thrau-Thi, Q.H.,Franzen, D. & Seela, F. (1982) Angew. Chem. SUPPI. 945 - 952. 21. Morr, M.(1982) Liebigs. Ann. Chem. 666-614. 22. Simonson, L. P., Bar, H. P., Jastorff, B. & Hitkari, R. (1975) Biochim. Biophys. Acta, 379, 114- 124. 23. Jastorff, B. & Krebs, T. (1972) Chem. Ber. 105, 3192-3202. 24. Couchie, D.,Erneux, C. & Dumont, J. E. (1981) Biochem. J . 199, 441 - 446.
25. Erneux, C.,Miot, F., Boeynaems, J. M. & Dumont, J. E. (1982) FEBS Lett. 142, 251 -254. 26. Corbin, J. D.,Rannels, S. R., Flockhart, D. A., Robinson-Steiner, A. M., Tigani, M. C., Doskeland, S. O., Suda, R. H., Suda, R. & Miller, J. P. (1982) Eur. J. Biochem. 125, 259-266. 27. Takemoto, D.J., Hansen, J., Takemoto, L. J. & Houslay, M. D. (1982) J . Biol. Chem. 257, 14597- 14599. Walker, N. & Bourne, H. R. (1982) J. Biol. Chem. 28. Brothers, V.M., 257, 9349 -9355.
D. Couchie and C. Emeux, Institut de Recherche Interdisciplinaire en Biologie Humaine et Nuclkaire, Faculte de Medecine et de Pharrnacie de l'Universit6 Libre de Bruxelles, Campus HBpital Erasme, Route de Lennik 808, B-1070 Bruxelles, Belgium
G. Petridis and B. Jastorff, Fachbereich Biologie-Chemie, Universitat Bremen, NW2, LoebenerstraBe, D-2800 Bremen 33, Federal Republic of Germany