Characterization of phosphodiesterase catalytic sites by means of cyclic nucleotide derivatives

Interactions between cyclic nucleotide phosphodiesterase 11 catalytic site and substrates or tadalafil and role of a critical Gln-869 hydrogen bond

Poor understanding of the topography of cyclic nucleotide (CN) phosphodiesterase (PDE) catalytic sites compromises development of potent, selective inhibitors for therapeutic use. In the X-ray crystal structures of the catalytic domains of some PDEs, an invariant glutamine hydrogen bonds with groups at C6 and N1 or N7 on catalytic products or analogous positions of some inhibitors, inferring similar bonds with CNs (Nature 425:98-102, 2003; J Mol Biol 337:355-365, 2004; Mol Cell 15:279-286, 2004). A site-directed mutant (Q869A) lacking this invariant Gln in cGMP-/cAMP-hydrolyzing PDE11 had unaltered catalytic activity and affinity for sildenafil; but cGMP/cAMP or tadalafil affinity was reduced approximately 50- or 140-fold, respectively, and calculated free energy of binding suggested one hydrogen bond for each. A cGMP analog lacking the C6 oxygen had approximately 80-fold weakened affinity, modifications at N(2), N7, or 2'-OH diminished affinity approximately 16-fold, and analogs with groups appended at N1 had only 2- to 6-fold weakened affinity. Analogs with C8 substitutions were ineffective inhibitors, suggesting that cGMP binds in the anti conformation. Calculated decline in free energy of binding was consistent with that for one hydrogen bond only in the analog lacking binding potential at C6. In conclusion, Gln-869 interacts strongly with cGMP/cAMP and tadalafil, but not with sildenafil; interactions with CN analogs suggest a hydrogen bond only between Gln-869 and the C6 substituent. The results define interactions between the PDE11 catalytic site and substrates/inhibitors and advance potential for inhibitor design.

Cyclic nucleotide phosphodiesterases: relating structure and function

Cyclic nucleotide phosphodiesterases (PDEs) comprise a superfamily of metallophosphohydrolases that specifically cleave the 3',5'-cyclic phosphate moiety of cAMP and/or cGMP to produce the corresponding 5'-nucleotide. PDEs are critical determinants for modulation of cellular levels of cAMP and/or cGMP by many stimuli. Eleven families of PDEs with varying selectivities for cAMP or cGMP have been identified in mammalian tissues. Within these families, multiple isoforms are expressed either as products of different genes or as products of the same gene through alternative splicing. Regulation of PDEs is important for controlling myriad physiological functions, including the visual response, smooth muscle relaxation, platelet aggregation, fluid homeostasis, immune responses, and cardiac contractility. PDEs are critically involved in feedback control of cellular cAMP and cGMP levels. Activities of the various PDEs are highly regulated by a panoply of processes, including phosphorylation events, interaction with small molecules such as cGMP or phosphatidic acid, subcellular localization, and association with specific protein partners. The PDE superfamily continues to be a major target for pharmacological intervention in a number of medically important maladies.

Identification of key amino acids in a conserved cGMP-binding site of cGMP-binding phosphodiesterases. A putative NKXnD motif for cGMP binding

cGMP-binding phosphodiesterases contain two kinetically distinct cGMP-binding sites (a and b), and each site contains a conserved N(K/R)XnFX3DE sequence. N276A, K277A, K277R, D289A, and E290A mutants in the N276KX7FX3DE290 sequence of site a (higher affinity site) of bovine cGMP-binding, cGMP-specific phosphodiesterase (cGB-PDE or PDE5A) were expressed in High Five cells and purified. The cGMP-binding affinities of three mutants [K277A (Kd approximately 12 microM), D289A (Kd approximately 24 microM), and N276A (Kd approximately 60 microM)] were decreased in comparison with wild-type enzyme (Kd = 1.3 microM), which suggested an important role for Asn276, Lys277, and Asp289 in cGMP binding. These residues could be presented as a putative NKXnD motif, and their functions were predicted based on analogy with the canonical NKXD motif in GTP-binding proteins. No marked differences in catalytic functions such as specific activity, Km for cGMP, and IC50 for zaprinast or 3-isobutyl-1-methylxanthine were found among wild-type and mutant cGB-PDEs. This suggested that cGMP binding to site a does not influence the catalytic properties of cGB-PDE.

Derivatives of 1-beta-D-ribofuranosylbenzimidazole 3',5'-phosphate that mimic the actions of adenosine 3',5'-phosphate (cAMP) and guanosine 3',5'-phosphate (cGMP)

A series of new analogues of 1-beta-D-ribofuranosylbenzimidazole 3',5'-phosphate (cBIMP) has been designed according to the properties predicted by the MNDO method, and synthesised from substituted benzimidazoles. Dipole vectors and HOMO and LUMO energies for each benzimidazole base were calculated by the MNDO method and the lipophilicities of the cBIMP derivatives were determined. In general, the cBIMP derivatives activate cAMP-dependent protein kinases I and II and preferentially bind to site B, especially for the type II kinase, with 2-trifluoromethyl-cBIMP and 5,6-difluoro-cBIMP exhibiting the highest site selectivity. Each cBIMP derivative can stimulate cGMP-stimulated cyclic phosphodiesterase (cGS-PDE), with 5,6-dimethyl-cBIMP being as potent as cGMP, and also inhibit cGMP-inhibited phosphodiesterase (cGI-PDE). Only the 2-trifluoromethyl-cBIMP and the Rp-phosphorothioates (cBIMPS) (equatorial P = S) were resistant to hydrolysis by cPDE. The Sp-phosphorothioates were hydrolysed slowly, if at all. In addition to exhibiting a high lipophilicity, the most active compounds for the induction of apoptosis and inhibition of proliferation were also resistant to cPDE (Sp-5,6-dichloro-cBIMPS) and/or were potent activators of cAMP-dependent protein kinase (5,6-dichloro-cBIMP).

Interaction of the two structural domains of calmodulin with mature and immature rat brain microtubules

The inhibitory effect of calmodulin on the assembly of mature and immature rat brain microtubules was compared with that of the two major structural domains of this protein, the COOH-terminal fragment (amino acids 78-148) and the NH2-terminal fragment (amino acids 1-77), to determine the calmodulin structural domain responsible for the inhibitory effect on microtubule assembly. Microtubules prepared during the early stages of brain development, i.e., during intensive neurite outgrowth, are more sensitive to inhibition by the Ca2(+)-calmodulin complex than those obtained from adult brain. Significant inhibition of immature microtubule assembly was observed with both fragments in the absence of Ca2+, but the effects were more important when Ca2+ was present. With adult brain microtubules, the two fragments remained without effect on assembly in the absence of Ca2+, whereas some inhibition was seen in its presence but only with the COOH-terminal polypeptide. Under all these conditions, the COOH-terminal fragment was always more active than the NH2-terminal fragment on microtubule polymerization, albeit to a lesser extent than native calmodulin.

Involvement of a plasma membrane phosphodiesterase in the negative control of cyclic AMP levels by vasopressin in rat hepatocytes

Vasopressin has been shown previously to lower the glucagon-induced increase of cyclic AMP levels in isolated rat hepatocytes by way of an enhanced phosphodiesterase (EC 3.1.4.17) activity. Five phosphodiesterase inhibitors were tested for their ability to prevent vasopressin from lowering cyclic AMP levels in intact hepatocytes and for their inhibitory effect in vitro on soluble and particulate phosphodiesterase activities partially purified from hepatocytes. Three soluble activities have been separated by DEAE-cellulose chromatography: a phosphodiesterase hydrolyzing both cyclic AMP and cyclic GMP, a form stimulated by cyclic GMP and a cyclic AMP-specific activity. The absence of any statistically significant correlation between the in vivo (in intact cells) and the in vitro (on isolated phosphodiesterases) potencies of the inhibitors does not support a role for the cytosolic phosphodiesterases in mediating the vasopressin-induced decrease in cyclic AMP levels. No statistically significant correlation was observed between the inhibition of the vasopressin effect on cyclic AMP accumulation and the inhibition of phosphodiesterase activity either associated with the native plasma membranes or solubilized from these membranes with 0.4 M NaCl. In contrast, a statistically significant correlation was observed between the degree of inhibition of the vasopressin effect in the intact cells and the degree of inhibition of the intrinsic phosphodiesterase still associated with the plasma membranes after high-salt treatment. These data indicate that a phosphodiesterase activity integral to the plasma membrane is very likely involved in the negative control of cyclic AMP levels by vasopressin.

Ca2+/calmodulin-sensitive inositol 1,4,5-trisphosphate 3-kinase in rat and bovine brain tissues

Inositol 1,4,5-trisphosphate (Ins P3) 3-kinase catalyzes the ATP-dependent phosphorylation of Ins P3 to Inositol 1,3,4,5-tetrakisphosphate (Ins P4). Ca2+/calmodulin (CaM)-sensitivity of Ins P3 3-kinase was measured in the crude soluble fraction from rat brain and different anatomic regions of bovine brain. Kinase activity was inhibited in the presence of EGTA (free Ca2+ below 1 nM) as compared to Ca2+ (10 microM free Ca2+) or Ca2+ (10 microM free Ca2+) and CaM (1 microM). Ca2+-sensitivity was also seen for the cAMP phosphodiesterase measured under the same assay conditions, but was not for the Ins P3 5-phosphatase. DEAE-cellulose chromatography of the soluble fraction of rat brain or bovine cerebellum resolved a Ca2+/CaM-sensitive Ins P3 3-kinase (maximal stimulation at 1 microM Ins P3 substrate level was 2.0-3.0 fold).

Hydrolysis of cyclic nucleotides by a purified cGMP-stimulated phosphodiesterase: structural requirements for hydrolysis

Hydrolysis of cyclic AMP and cyclic GMP analogues by a purified cGMP-stimulated phosphodiesterase from bovine adrenal tissue was investigated by reversed-phase HPLC. The results indicate that both a negative charge and an equatorial oxygen atom located at the cyclic phosphate residue are absolute requirements for the process of hydrolysis. Other substituents only gradually decreased the apparent hydrolytic activity. C-8-substituted derivatives were generally poor substrates due to the limited ability of these compounds to rotate freely around the glycosidic bond. While C-6- and 0-2'-substituted analogues carrying bulky substituents were also poorly hydrolysed, all other derivatives, including different C-2-, C-6-, 0-3'- and 0-5'-modified cyclic nucleotides, were good substrates. We consistently observed that cyclic GMP and cyclic GMP analogues were better hydrolysed than the corresponding cyclic AMP analogues. Hydrolysis was correlated with neither the hydrogen bond donor/acceptor abilities nor the hydrophobicity of selected cyclic nucleotide analogues. Based on quantum-chemical calculations of the size and direction of the dipole moments of different purine bases, we propose that the polarization of inducible amino acid side-chains within the binding site is involved in the differential binding of adenine-derived and guanine-derived nucleotides. However, the size of the dipole moment alone is not sufficient to explain the observed cGMP-preference. Rather, the direction of the polarization power relative to the other molecular structures involved in binding and hydrolysis seems to be the molecular mechanism by which the enzyme is able to discriminate between cAMP- and cGMP-like structures.

Modulation of cyclic AMP action in the dog thyroid by its agonist and antagonist Sp- and Rp-adenosine 3',5'-monophosphorothioate

The diastereoisomeric forms of adenosine 3',5'-monophosphorothioate, Sp-cAMPS and Rp-cAMPS, have been shown to mimic and to inhibit activation of protein kinase type I and type II by cyclic AMP. In the present work, Sp-cAMPS mimicked thyrotropin (TSH) action on thyroid hormone secretion and protein iodination in dog thyroid slices, whereas Rp-cAMPS antagonized those effects. The phosphorothioates have been tested as inhibitors or activators of the three major phosphodiesterases: the Ca2+/calmodulin-sensitive form, the cyclic GMP-stimulated form and the cyclic AMP-specific enzyme. At 100 microM Sp-cAMPS inhibited the three enzyme activities. In contrast, Rp-cAMPS failed to stimulate activity of the three enzymes. From a comparison of the biological properties of Sp- and Rp-cAMPS and 3-isobutyl-1-methyl xanthine, it is suggested that one site of action of the phosphorothioates is on the cyclic AMP-dependent protein kinases, i.e. the effects of Sp-cAMPS and Rp-cAMPS observed in intact cell can be ascribed to the agonistic and antagonistic effects on the cyclic AMP-dependent protein kinases. However, partial inhibition of phosphodiesterase activities by the phosphorothioates cannot be excluded.

Specificity of cGMP binding to a purified cGMP-stimulated phosphodiesterase from bovine adrenal tissue

The binding of [3H]cGMP (guanosine 3',5'-monophosphate) to purified bovine adrenal cGMP-stimulated phosphodiesterase was measured by Millipore filtration on cellulose ester filter. [3H]cGMP-binding activity was enhanced when the assay was terminated in buffer containing 70% of saturated ammonium sulfate to dilute the enzyme and wash the filters. The cGMP-binding activity was co-purified with the phosphodiesterase activity. The binding of [3H]cGMP to purified enzyme was measured in the presence or absence of the phosphodiesterase inhibitor, 1-methyl-3-isobutylxanthine. 1-Methyl-3-isobutylxanthine showed linear competitive inhibition with respect to cGMP as substrate in the phosphodiesterase reaction but stimulated the [3H]cGMP-binding activity in the binding assay. The stimulatory effect appeared not to be the result of preservation from [3H]cGMP hydrolysis; no cGMP phosphodiesterase activity has been measured under the cGMP-binding assay conditions, in the absence or presence of the inhibitor. Half-maximal stimulation by 1-methyl-3-isobutylxanthine occurred in the 5-7 microM concentration range. The specificity of binding of [3H]cGMP was investigated by adding increasing concentration of unlabeled analogs of cAMP (adenosine 3',5'-monophosphate) and cGMP. The binding of [3H]cGMP (50 nM) was displaced by unlabeled cGMP and cAMP with the following potency: 50% displacement was reached at the 0.1 microM cGMP range and only at a fiftyfold higher cAMP concentration. Our data with comparative series of analogs (e.g. 5'-amino-5'-deoxyguanosine 3',5'-monophosphate and 3'-amino-3'-deoxyguanosine 3',5'-monophosphate) showed that the potencies of stimulation of cAMP phosphodiesterase activity parallels displacement curves or [3H]cGMP binding to purified enzyme with no correlation with phosphodiesterase inhibition sequences. Those experiments suggest that the cGMP-binding activity is directly related to the non-catalytic (allosteric) cGMP-binding site.