Isozyme-dependent sensitivity of adenylyl cyclases to P-site-mediated inhibition by adenine nucleosides and nucleoside 3'-polyphosphates.

Recombinant adenylyl cyclase isozyme Types I, II, VI, VII, and three splice variants of Type VIII were compared for their sensitivity to P-site-mediated inhibition by several adenine nucleoside derivatives and by the family of recently synthesized adenine nucleoside 3'-polyphosphates (Désaubry, L., Shoshani, I., and Johnson, R. A. (1996) J. Biol. Chem. 271, 14028-14034). Inhibitory potencies were dependent on isozyme type, the mode of activation of the respective isozymes, and on P-site ligand. For the nucleoside derivatives potency typically followed the order 2',5'-dideoxyadenosine (2',5'-ddAdo) > beta-adenosine > 9-(cyclopentyl)-adenine (9-CP-Ade) >/= 9-(tetrahydrofuryl)-adenine (9-THF-Ade; SQ 22,536), with the exception of Type II adenylyl cyclase, which was essentially insensitive to inhibition by 9-CP-Ade. For the adenine nucleoside 3'-polyphosphates inhibitory potency followed the order Ado < 2'-dAdo < 2',5'-ddAdo and 3'-mono- < 3'-di- < 3'-triphosphate. Differences in potency of these ligands were noted between isozymes. The most potent ligand was 2',5'-dd-3'-ATP with IC50 values of 40-300 nM. The data demonstrate isozyme selectivity for some ligands, suggesting the possibility of isozyme-selective inhibitors to take advantage of differences in P-site domains among adenylyl cyclase isozymes. Differential expression of adenylyl cyclase isozymes may dictate the physiological sensitivity and hence importance of this regulatory mechanism in different cells or tissues.

Adenylyl cyclase is potently and directly inhibited by analogs of adenosine via a domain referred to as the P-site from its requirement for an intact purine moiety (1)(2)(3)(4). Domains for catalysis and inhibition have been distinguished by use of enzyme purification (5,6), inhibition kinetics (7,8), site-specific covalent ligands (9), and selective amino acid substitutions (10). These data suggest that the P-site is distinct from, yet homologous to and interacting with, the catalytic domain. The observation that purified native and recombinant Type I ad-enylyl cyclases are inhibited by P-site ligands, although exhibiting decreased sensitivity to inhibition (4 -6, 11), establishes the locus of the P-site on the enzyme per se and that inhibition is not via cell surface receptors or G-proteins. P-site-mediated inhibition has been characterized pharmacologically (1,2,4,(12)(13)(14)(15)(16). Inhibition requires an intact adenine moiety, and potency of inhibition is increased substantially for deoxyribose and especially 3Ј-phosphorylribose adenine nucleosides. Inhibitory potency follows the order: 3Јmono-Ͻ 3Ј-di Ͻ 3Ј-triphosphate and adenosine (Ado) Ͻ 2Јdeoxy (d) 1 -Ado Ͻ 2Ј,5Ј-ddAdo, with 2Ј,5Ј-dd-3Ј-ATP being the most potent ligand and exhibiting an IC 50 ϳ40 nM (15). In addition, tolerance for large substitutions at the 3Ј-position and for other ribose modifications has been demonstrated (1,2,4).
We reported previously that levels of 2Ј-d-3Ј-AMP and 3Ј-AMP varied considerably in different tissues and were dependent on the metabolic state of the animal (17). Moreover, sensitivity of adenylyl cyclases to inhibition by these nucleotides varied among rat tissues. We suggested that diversity in P-sitemediated inhibition depended on the adenylyl cyclase isozyme and that this diversity may well dictate differences in pharmacological and possibly physiological influence of this inhibition in the respective tissues (17). This is also suggested from studies in which P-site agonists, e.g. 2Ј,5Ј-ddAdo or 9-THF-Ade, have been used to alter function in a variety of cell systems, e.g. (14, 18 -23). Although effects on cell function and on cellular cAMP levels have been uniformly consistent with P-site inhibition of adenylyl cyclase, potencies of a given compound differed among systems, and unexpected potencies of some ligands were noted in others. Since it was not possible in those earlier studies to establish which adenylyl cyclase isozyme(s) was(were) present, it is possible that variations in expected behavior may well have been due to differences in levels of expression of the several isozymes.
To evaluate this diversity in response, we are reporting the first investigation of P-site-mediated inhibition of several recombinant types of adenylyl cyclase. This study compares potency of several P-site ligands to inhibit Type I (24), Type II (25), Type VI (26), Type VII (27), and three splice variants of Type VIII (28,29) adenylyl cyclases. These adenylyl cyclase isozymes were selected because they represent forms of the enzyme exhibiting distinct regulatory characteristics (30). For example, all are activated by G s ␣ but Types I and VIII are activated also by Ca 2ϩ /calmodulin. When Type I enzyme is activated by G s ␣ it is inhibited by G␤␥ (11,31,32), whereas when Type II adenylyl cyclase is activated by G s ␣ its activity is increased further by G␤␥ (31,32). The Type VI enzyme responds to G s ␣ (26) but is not activated by Ca 2ϩ /calmodulin or G␤␥.

Preparation and Assay of Adenylyl Cyclases from Rat and Bovine
Brain-Both detergent-solubilized and particulate preparations of adenylyl cyclase from rat and bovine brains were prepared and assayed as described previously (4,5,8,9,17). Enzyme assays were conducted in duplicate or triplicate. IC 50 values were derived from inhibition curves comprising six to eight concentrations of inhibitor in two to eight experiments for each condition.
Preparation of Expressed Recombinant Adenylyl Cyclases-Membranes and membrane extracts were prepared as described previously (25,26,32) from fall army worm ovarian (Sf9) cells infected with baculovirus encoding Type I, Type II, or Type VI adenylyl cyclase or vector alone (thyroid peroxidase controls). Cells were lysed by N 2 cavitation, and the particulate fraction was collected by centrifugation. The membranes from Sf9 cells encoding Type I adenylyl cyclase (32) were resuspended in a buffer containing 20 mM HEPES, pH 8, 1 mM EDTA, 6% sucrose, 5 mM MgCl 2 , and 1 mM dithiothreitol. Membranes from Sf9 cells encoding Type II, Type VI, and thyroid peroxidase (25,26) were resuspended in a buffer containing 20 mM HEPES, pH 8, 1 mM EDTA, 2 mM dithiothreitol, 200 mM sucrose, 1 mM phenylmethylsulfonyl fluoride, and 2 g/ml each of leupeptin and aprotinin. Detailed procedures for the preparation of Type II and Type VI enzyme have been described (33). All membranes were diluted at least 10-fold with a buffer containing 20 mM HEPES, pH 8, and 5% glycerol before being diluted further into the adenylyl cyclase assay containing the same buffer.
Enzyme Type VII (27) and the splice variants of Type VIII (29) were expressed in HEK 293 cells (ATCC CRL 1573) as described previously. HEK cells expressing Type VII enzyme were washed in phosphatebuffered saline (prepared from Sigma tablets P-4417), scraped from dished into a small portion of the buffered saline, collected by centrifugation, and then stored at Ϫ80°C. Membranes were prepared following lysing of the cells by N 2 cavitation as described previously (28).
Adenosine 3Ј-Diphosphate (3Ј-ADP) and Adenosine 3Ј-Triphosphate (3Ј-ATP)-3Ј-ADP was synthesized by a modification of the methods described by Mitchel et al. (35) and Sheridan et al. (36) (Fig. 1). The principal modifications are the use of sonication to facilitate the reaction, which otherwise would have been allowed to continue for 150 h (35), and the use of anion exchange chromatographic techniques to separate the several polyphosphorylated derivatives of adenosine obtained. This modified procedure represents a one-step preparation of nucleoside 3Ј-di-and 3Ј-triphosphates from the corresponding monophosphate and is characterized by a short reaction time and convenient reaction conditions. A suspension of 3Ј-AMP (1 g, 2.8 mmol) and phosphoramidic acid (1.63 g, 16.8 mmol) in 50 ml of formamide was sonicated under an argon atmosphere overnight. The temperature was maintained at 20 -25°C by putting the ultrasonication bath in a cold box at 4°C. The medium was cooled to 5°C, diluted with 2 liters of cold water (5°C), and neutralized with triethylamine (2.8 ml, 20 mmol). Reaction products were purified on QAE-Sephadex (HCO 3 Ϫ form) with a linear gradient of triethylammonium bicarbonate (0.01-0.3 M) followed by an isocratic elution with 0.3 M triethylammonium bicarbonate and then another linear gradient with triethylammonium bicarbonate (0.3-1 M) (Fig. 2). The appropriate fractions were lyophilized and then coevaporated several times with methanol, yielding 380 mol (14%) of 3Ј-ADP and 14.6 mol 3Ј-ATP. Both nucleotides were isolated as their respective sodium salts by precipitation in 1 M sodium iodide in acetone from the methanol solution of the triethylammonium nucleotide. The precipitate was centrifuged and washed three times with cold acetone and dried in vacuo giving the sodium salts of 3Ј-ADP and 3Ј-ATP. No impurities were noted  Fig. 1 was subjected to high performance anion exchange chromatography, as described under "Experimental Procedures." Elution was with a discontinuous gradient of triethylammonium bicarbonate (Et 3 NH 2 CO 3 ), as indicated. Materials-[␣-32 P]ATP was purchased from ICN Pharmaceuticals. Lubrol-PX was filtered through alumina (Neutral, AG7, from Bio-Rad) to remove peroxides. Dimethyl sulfoxide was redistilled under partial vacuum over CaH 2 . 2Ј,5Ј-ddAdo and the 3Ј-polyphosphates of 2Ј,5Ј-ddAdo and of 2Ј-dAdo were synthesized by methods reported previously from this laboratory (9,15,16). 9-THF-Ade (SQ 22,536) was a gift from Salvatore Lucania, Bristol-Myers-Squibb, Pharmacology Research Institute, P.O. Box 4000, Princeton, NJ 08543.

RESULTS
Activation of Adenylyl Cyclases and Sensitivity to P-site Ligands-Previously reported data suggested tissue-dependent differences in inhibitory potency of P-site ligands and also showed that activation of the enzyme by various agents resulted in different sensitivity of the enzyme to inhibition (17). To ascertain whether these differences would be reflected in specific isozymes, the potency of 2Ј-d-3Ј-AMP was evaluated on recombinant bovine Type I, Type II, and Type VI adenylyl cyclases ( Fig. 3 and Table I). IC 50 values for inhibition of the rat brain 2 enzyme by 2Ј-d-3Ј-AMP are shown for comparison. As expected, various stimulatory agents resulted in substantially different activities, with the greatest activity being uniformly obtained when assays were conducted with 5 mM Mn 2ϩ and 100 M forskolin. Reaction velocities with Mg 2ϩ and forskolin were roughly similar to those noted with Mn 2ϩ alone, and membranes assayed with Mg 2ϩ exhibited the lowest activities. Furthermore, the Type I enzyme was most potently inhibited, whether activation was with Mn 2ϩ (IC 50 ϳ4 M) or Mn 2ϩ and forskolin (IC 50 ϳ6 M). This greater sensitivity of the Type I enzyme is consistent with previously published data with enzyme derived from a number of tissues (1-4, 13, 14, 17). The small increase in IC 50 upon addition of forskolin is also consistent with earlier experiments with the bovine brain enzyme (5) but is in contrast with behavior of hepatic adenylyl cyclase (7) and the Type VI enzyme. Although forskolin stimulated each adenylyl cyclase, its effect to enhance potency of P-site ligands varied among isozymes. Hence, potency is uncoupled from enzyme activity, suggesting that the processes of catalysis and inhibition might not be causally linked.
Cell-permeable P-site Ligands-One goal of these studies has been to identify P-site ligands that could be used pharmacolog- 2 Earlier comparisons of P-site ligands by our laboratory were mostly with the adenylyl cyclases isolated from either rat or bovine brain (4, 5, 8,9). Although these tissues express several adenylyl cyclases, the Type I isozyme predominates with the isolation techniques we used (11,24).

FIG. 4. Differential inhibition of adenylyl cyclase isozymes by 9-CP-Ade.
Activities were determined with 100 M ATP, 5 mM MnCl 2 , and 100 M forskolin for Types I, II, and VI, but 500 M ATP was used for Type VIIIA because of its higher K m for 5Ј-ATP.  ically on intact cells or tissues and exhibit isozyme-selectivity and possibly tissue selectivity for inhibition of adenylyl cyclase. That such tissue selectivity may occur is suggested from studies with a number of tissues and with the several recombinant isozymes presented above. Structure-activity relationships for P-site ligands have indicated a tolerance for ribose substitutions and replacements. Several of the membrane-permeable compounds that have been used by many investigators in studies with intact cells and tissues, and these include: 9-(␤-darabinofuranosyl)adenine (Ara-Ade; IC 50 ϳ100 M) (1), 9-THF-Ade (IC 50 ϳ10 M; SQ 22-536), and 9-CP-Ade (IC 50 ϳ20 M) (13). These compounds were tested with several recombinant adenylyl cyclases, and significant differences in sensitivity were noted (Table II). The general order of sensitivity to P-sitemediated inhibition was Type I Ͼ Type VI Ϸ Type VIIIA Ͼ Type II. For the classical P-site ligands, 2Ј-d-3Ј-AMP and 2Ј,5Ј-ddAdo there were smaller differences in ligand potency among enzyme Types I, VI, and VIIIA than were noted with the ribose-substituted ligands. For these ligands, each was significantly less potent with the Type II enzyme than with the other isozymes. An interesting compound was 9-CP-Ade, which was essentially inactive with the Type II enzyme (Fig. 4). There was greater than 1 order of magnitude difference in potency of 9-CP-Ade for the Type II adenylyl cyclase (IC 50 Ͼ1 mM) compared with Type I (IC 50 ϳ109 M) and Type VIIIA (IC 50 ϳ117 M) (Table II and Fig. 4). This selectivity ratio of Ͼ10 is likely an underestimate since in some experiments the IC 50 for 9-CP-Ade was Ͼ4 mM, giving a selectivity ratio Ͼ40. 3Ј-Polyphosphates-Although the above data indicate some isozyme selectivity for the ribose-modified set of P-site ligands, less pronounced differences in sensitivity were noted with the adenine nucleoside 3Ј-polyphosphates (Table III). Typically, potency followed the order Ado Ͻ 2Ј-dAdo Ͻ 2Ј,5Ј-ddAdo and for the 3Ј-phosphates: 3Ј-mono-Ͻ 3Ј-di-Ͻ 3Ј-triphosphate (cf. Refs. 15 and 16). A comparison is shown in Fig. 5 for inhibition of the Type VI adenylyl cyclase. Although some differences were evident among the isozymes scanned in this study, e.g. Type II enzyme was least sensitive to P-site ligands, in general the more potent the compound the less selectivity was noted. For some isozymes the impact of the removal of the hydroxyl moieties was greater that with others. For example, the IC 50 values for the 2Ј-deoxy-and 2Ј,5Ј-dideoxynucleotides for enzyme Types VII and VIIIA are little different, whereas for the Type II enzyme the difference is 1 order of magnitude (cf . Table  III). Analogously, the effect on inhibitory potency of the sequential addition of phosphates to the 3Ј-position differed among these isozymes. For the 2Ј-deoxy series the largest difference in IC 50 values for the Type II enzyme was between the 3Ј-di-and 3Ј-triphosphate derivatives, whereas for the Type VI enzyme it was between the 3Ј-mono-and 3Ј-diphosphate derivatives (Table III and Fig. 5). For enzyme Types VI, VII, and all three splice variants of Type VIII, IC 50 values were little affected by the addition of the third 3Ј-phosphate group. The data suggest that inhibition by this class of ligand is a universal feature of adenylyl cyclase isozymes but that the structure of the respective isozymes influences potency of ligands, whether with ribosyl 2Ј-and 5Ј-hydroxyl or 3Ј-mono-, 3Ј-di-, or 3Ј-triphosphate groups.

DISCUSSION
To begin establishing a basis for the known tissue dependence for inhibition of adenylyl cyclases by P-site ligands, the efficacy of several ribose-modified and 3Ј-polyphosphorylated adenosines has been tested on several recombinant forms of this enzyme family. The rationale for the study is that the physiological importance of this regulatory mechanism in a given tissue may be reflected in the sensitivity of the adenylyl cyclase expressed in that tissue. The degree of inhibition would depend on levels of P-site ligand, and these may change as a  result of changes in cell function. Our additional consideration has been that ligands may be identified exhibiting selectivity toward individual isozymes. The influence of P-site inhibition in a given tissue may also be influenced by factors affecting activity of adenylyl cyclases. Although the data did not conform to the notion that sensitivity to inhibition by P-site ligands increased as enzyme activity increased (7), they indicated that different means of activation of adenylyl cyclase led to differences in the IC 50 values for inhibition (cf. Table I). For example, enzyme activity with Mn 2ϩ ϩ forskolin was uniformly greater than that with Mn 2ϩ alone, whereas the presence of forskolin led to increased sensitivity to inhibition by 2Ј-d-3Ј-AMP (Table I) or 2Ј,5Ј-ddAdo (not shown) with the Type VI enzyme but decreased sensitivity with Types I and II enzyme. Arguably the enzyme conformation resulting from activation by Mn 2ϩ must differ from those resulting from Mn 2ϩ ϩ forskolin, Mg 2ϩ , GTP⅐␣ s , GTP⅐␣ i , ␤␥, and appropriate combinations of these. Different conformations will exhibit greater or lesser sensitivity to inhibition by these agents, and this influence will be isozyme-dependent and may affect the physiological role of this inhibition.
Potency of P-site ligands obviously depends also on ligand structure. This was particularly evident with the ribose-modified adenosine derivatives as the comparison was extended to several isozymes (Table II). For example, Type II adenylyl cyclase, unique in its regulation by G-protein subunits and susceptibility to phosphorylation by protein kinase C (31)(32)(33), was insensitive to inhibition by 9-CP-Ade and was poorly sensitive to 9-THF-Ade. The differences in P-site potency among the isozymes were more acutely noted with ribose-modified adenine nucleosides than with the 3Ј-nucleotides. The addition of the 3Ј-phosphates contributes substantial binding energy (10 kcal/phosphate; Ref. 37) and may minimize the effects of differences in isozyme structure on ligand binding; that is, the differences among isozymes, noted by IC 50 values for 9-CP-Ade (Fig. 4), may be masked by the increased binding affinities of the nucleoside 3Ј-polyphosphates (Fig. 5). An avenue for the development of isozyme-selective ligands may be to exploit the differences established here with the ribose-modified adenine nucleosides and the dramatic inhibitory potency of the adenine nucleoside 3Ј-polyphosphates, thereby also aiding investigation of enzyme structure and mechanisms of catalysis and inhibition.