Inhibition of Adenylyl Cyclase by a Family of Newly Synthesized Adenine Nucleoside 3 (cid:42) -Polyphosphates*

The synthesis of a number of adenine nucleoside 3 (cid:42) polyphosphates has been devised via a phosphotriester approach that combines the method of alkoxide activation with the use of 2,2,2-tribromoethyl phosphoromor- pholinochloridate as a phosphorylating agent. The family of compounds included 3 (cid:42) ADP, 3 (cid:42) ATP, 2 (cid:42) -deoxy-3 (cid:42) ADP, 2 (cid:42) -deoxy-3 (cid:42) ATP, 2 (cid:42) ,5 (cid:42) -dideoxy-3 (cid:42) ADP, and 2 (cid:42) ,5 (cid:42) -dideoxy-3 (cid:42) ATP. Potency as inhibitors of adenylyl cyclases followed the order: 3 (cid:42) -mono- < 3 (cid:42) -di- < 3 (cid:42) triphosphate and adenosine (Ado) < 2 (cid:42) -d-Ado < 2 (cid:42) ,5 (cid:42) -dd-Ado derivatives, with 2 (cid:42) ,5 (cid:42) -dideoxy-3 (cid:42) ATP exhibiting an IC 50 of (cid:59) 40 n M . This order was maintained with purified and recombinant forms of the type I enzyme. The nucleoside 3 (cid:42) -phosphates caused noncompetitive inhibition of the type I adenylyl cyclase from bovine brain, consistent with inhibition via the

The 3Ј:5Ј-cAMP 1 signaling pathway can be regulated pharmacologically by many drugs that are of particular value in the treatment of various diseases, and therefore there is much current interest in identifying new agents acting on this pathway. Regulation of this pathway can be achieved through changes in the activities of 3Ј:5Ј-cAMP-phosphodiesterases (1), 3Ј:5Ј-cAMP-dependent protein kinases (2), or adenylyl cyclases.
Adenylyl cyclase is a family of membrane-bound enzymes that catalyze the formation of 3Ј:5Ј-cAMP from 5ЈATP and is regulated by numerous neurotransmitters and hormones via cell surface receptors and guanine nucleotide-dependent regulatory proteins (G-proteins). In an isozyme-dependent manner, enzyme activity is also regulated directly, by forskolin, by Ca 2ϩ / calmodulin, and by certain adenosine derivatives via a site that is distinct from the catalytic site. Whereas numerous drugs have been developed that act on the cyclic nucleotide phosphodiesterases, those that act directly on adenylyl cyclases have been less well explored; the main class of such pharmacological agents comprises forskolin and its analogs (3).
The domain through which inhibition of mammalian adenylyl cyclases occurs with adenosine derivatives, excepting the enzyme from sperm, is referred to as the P-site from an evident requirement for a purine moiety (4 -11). Key structural requirements for P-site ligands include a requirement for an intact adenine moiety, enhanced inhibitory potency with 2Јdeoxy-and especially 2Ј,5Ј-dideoxy-ribosyl moieties, and a notably strong preference for a 3Ј-phosphate (4 -8). The most potent ligands have been 2Ј,5Ј-dideoxyadenosine 3Ј-monophosphate (2Ј, 5Ј-dd-3ЈAMP), the naturally occurring 3ЈAMP, and 2Ј-d-3ЈAMP (8,12), and recently we described the effects of the 2Ј,5Ј-dd-3ЈADP and 2Ј,5Ј-dd-3ЈATP on brain adenylyl cyclases (13). Noncompetitive inhibition kinetics with these ligands (13)(14)(15) and irreversible inactivation studies with P-site-selective covalent affinity probes (16) 2 are consistent with inhibition occurring at a site that is distinct from the catalytic site. As a part of our effort (17) to investigate the structure and properties of this inhibitory site on adenylyl cyclases, we describe here a short and efficient synthesis of a number of adenosine 3Јpolyphosphates and the potent inhibition of adenylyl cyclase that these 3Ј-nucleotides exert. Although the biology and biochemistry of nucleoside 3Ј-polyphosphates are essentially unexplored in eukaryotic systems, inhibition of adenylyl cyclases by this class of compound may well imply roles as intracellular regulators of this transmembrane signaling system.

EXPERIMENTAL PROCEDURES
Preparation and Assay of Adenylyl Cyclase-Both detergent-solubilized and particulate preparations of adenylyl cyclase from rat and bovine brains were prepared and assayed as described previously (8,15,16). Lubrol-PX was filtered through alumina (Neutral, AG7, from Bio-Rad) to remove peroxides. Bovine brain adenylyl cyclase was purified as described by Pfeuffer et al. (18). Inhibition kinetics were determined on enzyme assayed with concentrations of divalent cation fixed in excess of the 5ЈATP concentration as described previously (19).
Quantification of Nucleotides by HPLC-Nucleotides were quantified after HPLC as areas under peaks determined with a Waters 996 photodiode array detector and the accompanying Millennium software (v.2.10). Chromatography was on a DEAE-5PW column (Altex, 5 m, 7.5 ϫ 75 mm) developed with sequential step gradients of triethylam-* This work was supported by National Institutes of Health Research Grant DK38828 (to R. A. J.) and an award from the Philippe Foundation (to L. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Materials-Recombinant type I enzyme was generously supplied by Drs. R.Taussig and A. G. Gilman. Membranes and membrane extracts were from fall army worm ovarian (Sf9) cells infected with a type I adenylyl cyclase encoding baculovirus. THF was distilled from benzophenone ketyl before use. NMR spectra were recorded with a Bruker AC250 at 250 MHz for proton spectra and at 101 MHz for 31 P spectra, with a 85% solution of H 3 PO 4 as external standard. FAB mass spectra were recorded with a glycerol matrix.
2Ј,5Ј-Dideoxyadenosine 3Ј-O-[(2,2,2-Tribromoethyl)-morpholinophosphonate] 3a-To a suspension of 2Ј,5Ј-dideoxyadenosine 1a (20) (1.18 g, 5 mmol) in a mixture of THF (60 ml) and pyridine (10 ml) was added dropwise at room temperature a 0.9 M solution of tert-butyl-magnesium chloride (5.55 ml, 5 mmol) in THF. After stirring for 5 min, 2,2,2tribromoethyl phosphoromorpholinochloridate 2 (2.25 g, 5 mmol) was added, and stirring was continued for 1.5 h. The mixture was then concentrated in vacuo, diluted with EtOAc (30 ml), quenched with water (60 ml), and then extracted with EtOAc (30 ml ϫ 2). The combined organic layers were washed with a saturated solution of NaCl, were dried over Na 2 SO 4 , and then were evaporated to give a gum. The crude product was subjected to silica gel chromatography. Elution with CH 2 Cl 2 : Et 2 O (95:5) followed by CH 2 Cl 2 :MeOH (95:5) afforded a mixture of the two diastereoisomers of the adduct 3a (1. 2Ј,5Ј-Dideoxyadenosine 3Ј-triphosphate 4a-Phosphotriester 3a (0.89 g, 1.5 mmol) was added to a solution of pyridine (30 ml) containing activated zinc (22) (0.15 g) and bis(tri-n-butyl-ammonium)-pyrophosphate (15 mmol) under the exclusion of moisture. The mixture was stirred at room temperature during 2 days. The reaction mixture was then centrifuged, and the supernatant fraction was evaporated in vacuo and purified by QAE-Sephadex (HCO 3 Ϫ form) with a linear gradient of triethylammonium bicarbonate (0.01-0.4 M). The appropriate fractions were lyophilized and then coevaporated several times with methanol, yielding 1.16 mmol (77%) of 2Ј,5Ј-dideoxyadenosine 3Ј-triphosphate 4a. The nucleotide 4a was isolated as its sodium salt by the addition of 1 M sodium iodide in acetone to a methanol solution of the triethylammonium nucleotide. The precipitate was centrifuged and washed three times with cold acetone and dried in vacuo giving the sodium salt of 2Ј,5Ј-dideoxyadenosine 3Ј-triphosphate. No impurities were noted on ion exchange HPLC: 1  2Ј-Deoxyadenosine 3Ј-Triphosphate 4b-Phosphotriester 3b (2.37 g, 2.5 mmol) was added to a solution of pyridine (30 ml) containing activated zinc (0.25 g) and bis(tri-n-butylammonium)-pyrophosphate (25 mmol), under the exclusion of moisture. The mixture was stirred at room temperature during 2 days. Then the reaction mixture was centrifuged, and the supernatant was evaporated in vacuo and treated with 80% acetic acid at room temperature for 30 min. The medium was then neutralized with a cold solution of 0.5 M NaHCO 3 , diluted to 2 liters, filtered, and then purified by chromatography on QAE-Sephadex as above, yielding 0.55 mmol (22%) of 2Ј-deoxyadenosine 3Ј-triphosphate 4b. This nucleotide was also isolated as its sodium salt as above.

Syntheses-
The synthesis of nucleoside 3Ј-polyphosphates has received considerably less attention than have those of the corresponding 5Ј-polyphosphates because the former are more difficult to synthesize and because their biological roles are not as well described (23)(24)(25)(26)(27)(28)(29)(30)(31). To our knowledge the only reported chemical synthesis of 2Ј-deoxy-nucleoside 3Ј-triphosphates was that by Josse and Moffatt (28) and proceeded by homologation of the corresponding nucleoside 3Ј-monophosphate by use of phosphoromorpholidate. Other nucleoside 3Ј-polyphosphates have also been prepared by homologation of the respective 3Ј-monophosphates (23)(24)(25)(26)29). The major drawback with this method is that the synthesis of the phosphoromorpholidates requires easy accessibility to the corresponding nucleoside 3Јmonophosphate. Van Boom and colleagues prepared nucleoside 3Ј-diphosphates in the thymidine and guanosine series by a phosphotriester approach that involved the use of an activated morpholinylphosphonate (27,30,31). Unfortunately, such a procedure cannot be applied when the base of the nucleotide is adenine, the 6-amino group being more reactive toward phosphorylating agents than is the 3Ј-hydroxyl of the nucleoside moiety (32,33). Recently Uchiyama et al. (33) reported that a facile 3Ј-O-phosphorylation of N-unprotected nucleosides could be achieved via magnesium alkoxide formation. We wish to report here that sequential treatment of N-unprotected 2Јdeoxyadenosine derivatives 1 (20,21) with an equimolar amount of tert-butylmagnesium chloride and tribromoethyl phosphoromorpholinochloridate 2 (34) in THF at 20°C afforded the phosphotriesters 3 in good yields (57 and 82%) (Scheme 1). The phosphotriester derivative 3a was reacted with bis-(tri-nbutylammonium) pyrophosphate in the presence of activated zinc (22), affording 2Ј,5Ј-dd-3ЈATP 4a in 77% yield. 2Ј-d-3ЈATP 4b was prepared similarly from 3b with an additional detritylation step in 22% yield. This method can also be applied to the synthesis of nucleoside 3Ј-diphosphates, since 2Ј,5Ј-dd-3ЈADP 5a and 2Ј-d-3ЈADP 5b also were prepared by reaction of 3a with tributylammonium phosphate in 54% and 49% yield, respectively.
In conclusion, the utility and efficiency of the method of the alkoxide activation combined with the use of the phosphorylating agent 2 has been demonstrated by its application to the syntheses of various adenosine 3Ј-polyphosphates in only two steps, avoiding additional steps generally required for such syntheses (protection and deprotection of the exocyclic amino group and condensation of morpholine with the nucleoside monophosphate).
Inhibition of Adenylyl Cyclases-The adenosine 3Ј-polyphosphates and the deoxy derivatives (Fig. 1) are a family of inhibitors of adenylyl cyclase in which potency increased with the number of 3Ј-phosphates and with the removal of the 2Ј-OH and 5Ј-OH groups (Fig. 2). IC 50 values for inhibition of rat brain adenylyl cyclase by the three series of adenosine 3Ј-phosphates are presented in Table I and compared with those of the parent nucleosides and with three cyclic nucleotides. The cyclic nucleotides were poor inhibitors. For each parent nucleoside, inhibitory potency increased with the number of 3Ј-phosphates. Inhibitory potency of the 3Ј-phosphates followed the order Ado Ͻ 2Јd-Ado Ͻ 2Ј,5Ј-dd-Ado. The greatest effects were those caused by the removal of the 2Ј-hydroxyl group and by the sequential addition of phosphates at the 3Ј position, each increasing potency many-fold. The combined effects of the removal of the 2Јand 5Ј-hydroxyl groups and the addition of the 3Ј-polyphosphate are incorporated in 2Ј,5Ј-dd-3ЈATP, which exhibited an IC 50 ϳ40 nM (Table I and Ref. 13). This potency is almost 2 orders of magnitude greater than that of previous ligands and makes 2Ј,5Ј-dd-3ЈATP the most potent nonprotein regulator of adenylyl cyclases thus far described. It approaches the potency of the stimulatory effect of rG s ␣ on the type I adenylyl cyclase and the stimulatory and inhibitory effects of G␤␥ on ␣ s -activated-types II and I, respectively (35). The observation that adenosine and 2Ј-deoxyadenosine 3Ј-polyphosphates are also very potent inhibitors of adenylyl cyclase is important in that these 3Ј-nucleotides are known to occur naturally, whereas the 2Ј,5Јdideoxyadenosine 3Ј-polyphosphates are not. Consequently, the adenosine and 2Ј-deoxyadenosine 3Ј-polyphosphates are potentially important and unexplored intracellular regulatory nucleotides. The rank order presented in Table I was maintained with the purified native bovine type I and with the recombinant wild type I adenylyl cyclases, though with these enzymes each of the 3Ј-nucleotides was noticeably less potent than with the enzyme in the cruder detergent-dispersed rat brain preparation (13). This is consistent with the loss of potency of 2Ј-d-3ЈAMP and 2Ј,5Ј-dd-Ado we had previously noted upon purification of the bovine brain adenylyl cyclase (8). This may reflect the loss of membrane phospholipid important for tertiary structure or loss of calmodulin in the purified preparations, because Ca 2ϩ /calmodulin enhances sensitivity to P-site ligands (36).

FIG. 2.
Inhibition of rat brain adenylyl cyclase by several adenosine 3-phosphates. All 3Ј-nucleotides were added as their sodium salts. Activities were determined on detergent-dispersed enzyme from rat brain with 100 M ATP, 5 mM MnCl 2 , and 100 M forskolin. consistent with inhibition occurring at the P-site (15). Not surprisingly, the 2Ј-deoxyadenosine 3Ј-polyphosphates also caused noncompetitive inhibition (Fig. 3). Thus, all of the adenine 3Ј-nucleotides synthesized and tested here constitute a class of substantially more potent P-site inhibitors of adenylyl cyclases. 3Ј-Nucleotide Stability-Typical adenylyl cyclase preparations contain a number of phosphohydrolyses and substantial degradation of the adenine nucleoside 3Ј-polyphosphates might be expected (37,38). However, when these 3Ј-nucleotides were incubated with a detergent-dispersed adenylyl cyclase from rat brain under conditions known to elicit excellent sensitivity to P-site-mediated inhibition, with the exception of 3ЈATP, there was no significant breakdown of any of the adenosine or 2Јdeoxyadenosine 3Ј-polyphosphates or of 2Ј,5Ј-dd-3ЈATP (Table  II). 3 Notably, there was some conversion of 3ЈADP to 3ЈATP (as well as to breakdown products of 3ЈATP) and of 2Ј,5Ј-dd-3ЈADP to 2Ј,5Ј-dd-3ЈATP. This phosphorylation of 3ЈADP and 2Ј,5Ј-dd-3ЈADP was found to be due to contaminating kinase(s) in the incubation and did not occur in the absence of enzyme. 3ЈATP breakdown led to the formation of 3ЈAMP and 2ЈAMP in variable ratios, and this occurred also in the absence of enzyme. The hydrolysis of 3ЈATP implies its potency (IC 50 ϳ2 M) was likely an underestimation. In the absence of the adenylyl cyclase preparation, the principal product of 3ЈATP hydrolysis was 2Ј:3Ј-cAMP (Table III) and was found to be catalyzed by divalent cation alone. It occurred with Mn 2ϩ , Mg 2ϩ , and to a slightly lesser extent with Ca 2ϩ (Table III). Because the inclusion of the adenylyl cyclase preparation led to the formation of 3ЈAMP or 2ЈAMP (cf. Table II) but to no measurable 2Ј:3Ј-cAMP, the sequence 3ЈATP 3 2Ј:3Ј-cAMP 3 3ЈAMP ϩ 2ЈAMP is suggested. The latter step would be catalyzed by enzymes in the rat brain extract. This reaction sequence was obviously not possible with the 2Ј-deoxyribosides, and with them significant breakdown did not occur and the consequences of the little breakdown of the 3Ј-polyphosphates that does occur could be ignored.
Because possible products of nucleoside 3Ј-polyphosphate hydrolysis would include inorganic pyrophosphate (PP i ) or inorganic triphosphate (PPP i ), we determined the potency and nature of inhibition of adenylyl cyclase by these agents (Fig. 4). PP i (IC 50 ϭ ϳ1.4 mM) was less potent than either PPP i or PPPP i (IC 50 ϭ ϳ55 M for both). For PP i kinetics of inhibition was mixed (15), and for PPP i inhibition was competitive with respect to substrate. Thus, inhibition by the adenosine 3Јpolyphosphates conformed to established behavior of classical P-site ligands, and inhibition could not be accounted for by either potency or inhibition kinetics of the minimal hydrolytic products formed in the adenylyl cyclase incubations.
Alternative Substrate-Related to the question of 3Ј-nucleotide stability was the possibility that 3ЈATP or 2Ј-d-3ЈATP might also serve as substrates of adenylyl cyclase for the formation of 3Ј:5Ј-cAMP or 2Ј-d-3Ј:5Ј-cAMP, respectively. In three experiments with purified type I adenylyl cyclase from bovine brain, with substrate concentrations ranging from 100 to 500 M, no 3Ј:5Ј-cAMP was detected from 3ЈATP nor 2Ј-d-3Ј:5Ј-cAMP from 2Ј-d-3ЈATP. Because an optically based detection system was used, it remains possible that by use of a radioactively labeled substrate and the greater inherent sensitivity it would allow, some small amount of 3Ј:5Ј-cAMP or 2Ј-d-3Ј:5Ј-cAMP might have been detected. This caveat notwithstanding, the data argue that the adenylyl cyclases catalyze the hydrolysis of pyrophosphate and the formation of the cyclic 3Ј:5Јdiester solely from 5ЈATP at the catalytic site and that this does not also occur from the adenosine 3Ј-triphosphates at the P-site. DISCUSSION Presented here are the synthesis of a new family of adenosine 3Ј-polyphosphates and the effects of these 3Ј-nucleotides on adenylyl cyclase. The characteristics of inhibition conformed 3 These experiments on nucleotide stability were conducted in the absence of 5ЈATP, which is present in adenylyl cyclase reactions. 5ЈATP was omitted because it partially comigrates with the 3Ј-triphosphates we tested, making the quantification of nucleoside 3Ј-triphosphates more difficult. In other experiments we found no evidence that 5ЈATP altered the small appearance of 3Ј-di-and 3Ј-monophosphate breakdown products, implying that 5ЈATP did not affect the hydrolysis of the 3Ј-triphosphates.

TABLE II
Stability of adenosine 3Ј-polyphosphates during reaction with detergent-extracted adenylyl cyclase from rat brain Values are percentages normalized from the areas of the eluted peaks from anion exchange HPLC for each nucleotide added. Nucleotides were exposed for 15 min at 30°C with normal adenylyl cyclase reaction mixture, including 50 mM triethanolamine ⅐ HCl, pH 7.5, 5 mM MnCl 2 , 100 M forskolin, 1 mM 3-isobutyl-1-methylxanthine, 1 mM dithiothreitol, 1 mg of bovine serum albumin/ml, 0.1% (w/v) Lubrol-PX, 2 mM creatine phosphate, and 100 g of creatine kinase/ml, but absent 5ЈATP. 3 Extracted rat brain protein was 21 g/tube. to those of ligands acting on adenylyl cyclases via the P-site. For the parent nucleosides and for their respective 3Ј-phosphorylated derivatives, potency increased with removal of ribosylhydroxyl groups and exhibited the rank order of Ado Ͻ 2Ј-d-Ado Ͻ 2Ј,5Ј-dd-Ado. These compounds exhibited increased inhibitory potency with the successive addition of phosphates to the 3Ј-ribose position. Although the most potent inhibitor was 2Ј,5Ј-dd-3ЈATP (IC 50 ϳ40 nM), it is important that the adenosine and 2Ј-deoxyadenosine 3Ј-polyphosphates also exhibited potency in the micro-to submicromolar range, respectively. Of these 3Ј-nucleotides it is probable that all but the 2Ј,5Ј-dideoxyadenosine derivatives are naturally occurring.
Although the tertiary structure of adenylyl cyclases is not known, the deduced primary sequence suggests a membrane topology exhibiting a repeated structure of six-membrane spanning regions followed by a large cytosolic domain (39). The two cytosolic domains (C1 and C2) are homologous with each other and with the established catalytic domain of guanylyl cyclases (40), supporting the idea that each contains a nucleotide binding region. It is easy to speculate that each of these two cytosolic domains binds one of the nucleotides, one site for catalysis (5ЈATP) and a distinct site for inhibition (3ЈATP), but this is uncertain. It is known that expression of either C1 or C2 domain alone is insufficient to catalyze effectively the formation of 3Ј:5Ј-cAMP and that co-expression of independently vectored cytosolic domains substantially improves the catalytic efficiency (41). In fact, the chimeric protein created by linking the C1 domain of the type I enzyme with the C2 domain of the type II enzyme yielded a soluble enzyme also capable of catalysis and of inhibition by 2Ј-d-3ЈAMP (44). These observations suggest that both C1 and C2 participate in or must interact for catalysis and possibly also for inhibition by 3Ј-nucleotides. However, it remains to be established whether C1 and C2 each interact separately with 5ЈATP or the adenosine 3Ј-polyphosphates, or whether C1 and C2 act together to form one or two nucleotide binding sites, possibly at the interface of C1 and C2. To date results from truncation, deletion, and site-directed mutation studies have not allowed clear identification of two binding domains (39,(41)(42)(43)(44).
Available evidence suggests that for most adenylyl cyclases, catalysis and inhibition by P-site ligands occur at distinct sites. This conclusion is supported principally by studies of irreversible inactivation with P-site targeted ligands and by enzyme kinetics (15,16). Inactivation by 2Ј,5Ј-dd-3ЈFSBAdo (16) occurred in the presence of 5ЈATP but not in the presence of 2Ј,5Ј-dd-Ado, a modestly potent P-site ligand. We suggested that this was consistent with P-site ligands binding at a site distinct from and independent of that of the substrate. This is nominally supported by inhibition kinetics, which have consistently exhibited straightforward noncompetitive behavior (4 -8, 13-15), except for enzyme activated by pretreatment with GTP␥S⅐␣ s , when uncompetitive inhibition was noted (15). Both uncompetitive and noncompetitive inhibition typically imply inhibitor binding to the enzyme in the presence of substrate and at a site distinct from the catalytic site (45), but other mechanisms may lead also to this apparent kinetic behavior as we have discussed previously (15). The principal argument is that P-site ligand may form an inactive metal-PP i -adenosine complex at the catalytic site through the binding of the P-site ligand at the region occupied by product 3Ј:5Ј-cAMP (15,46). Though consistent with noncompetitive and even uncompetitive inhibition, this argument is at odds with the observation that PP i did not affect the apparent K i for 2Ј-d-3ЈAMP (15). The nucleoside 3Ј-triphosphates, being substantially more potent inhibitors, present a different circumstance in that the inhibitor contains three phosphates, and the possible formation of a dead-end enzyme-inhibitor complex subsequent to enzyme activation cannot be dismissed. Examples of this exist with Ordered Bi Bi reaction mechanisms and in the example of FIG. 4. Inhibition of rat brain adenylyl cyclase by inorganic polyphosphates. Upper panel, enzyme velocities with polyphosphate relative to control velocities. Activities were determined as for Fig. 2, with MnCl 2 fixed at 5 mM in excess of the ATP concentrations. å, inorganic ammonium tetraphosphate; f, inorganic sodium triphosphate; q, inorganic sodium pyrophosphate. Lower panel, double-reciprocal plot for inhibition of rat brain adenylyl cyclase by inorganic manganese triphosphate. aldehyde reductase crystallographic evidence places inhibitor in the catalytic domain, at the alcohol product site (47,48). Adenylyl cyclases also conform to a bireactant sequential reaction mechanism with divalent cation and cation-ATP as substrates and some isozymes exhibit ordered rather than random behavior (19,49). Enzyme is activated by divalent cation, either by Mn 2ϩ directly or indirectly by activated Gs, 4 and becomes more susceptible to inhibition by P-site ligands (4 -8, 10 -12, 14, 15, 19, 36, 37, 46, 49). Thus, key characteristics of catalysis and inhibition by P-site ligands are the roles of divalent cation and the similar nucleotide requirements for both processes. Each mechanism requires divalent cation; each is adenine-specific; for each the 2Ј-deoxyadenosine derivative is a better ligand; and for each the nucleoside triphosphate is either required (catalysis) or preferred (inhibition). Thus, we suggest that catalysis and inhibition likely occur at very similar if not the same sites. Second messengers are typically viewed as mediating the effects of extracellular signals to regulate specific intracellular activities, thereby serving to coordinate functions of individual cells with events elsewhere in the organism. By contrast, adenosine 3Ј-polyphosphates, formed from precursors within cells, may be viewed as cell-initiated censors or attenuators, effectively minimizing the influence of extracellular signals, specifically of those that activate the adenylyl cyclase-3Ј:5Ј-cAMPprotein kinase A cascade. Levels of adenosine 3Јpolyphosphates may conceivably be modulated by intracellular events or possibly other extracellular signals via cascades in which the formation of these attenuators is a critical function, that is, to shut down the 3Ј:5Ј-cAMP-protein kinase A pathway, thereby allowing other cell functions to occur. Because there are few if any adenine nucleotides that interact with only one protein or protein type, we suspect that adenosine 3Ј-polyphosphates will be found to interact with a number of proteins or enzymes other than adenylyl cyclases and will thereby induce a family of physiological responses. Important in this conjecture, though, is whether adenosine 3Ј-polyphosphates occur in mammalian cells and to what extent their levels may change and be regulated. Although the natural occurrence of ribo-and 2Ј-deoxyribo-nucleoside 3Ј-diand 3Ј-triphosphates per se have been known for several decades, they have received relatively little attention in mammalian systems (50 -58). Nucleoside 3Ј-polyphosphates have been of interest in the regulation or prokaryotes because guanosine 3Ј-diphosphate was identified during early investigations of guanosine 5Ј-di-and 5Ј-triphosphate-3Ј-diphosphate (magic spot; ppGpp 5 and pppGpp) (50 -54). The analogous adenosine polyphosphates, pppAppp 5 , pppApp, and ppApp, were also reported to be synthesized by ribosomes from sporulating but not vegetative bacteria (55,56) and as part of the stringent response (57). This has suggested that in these organisms adenosine 3Ј,5Ј-bis-polyphosphates may be involved in the regulation of development. For guanosine 3Ј-polyphosphates, the likely precursors are the respective 3Ј,5Ј-bis-polyphosphates. By analogy and given the prevalence of 5Ј-nucleotidases, the likelihood that adenosine 3Ј-polyphosphates may also occur naturally is suggested. Although these more rigorous studies have been done in bacterial systems, the reported presence of both guanosine and adenosine 3Ј,5Ј-bis-polyphosphates in cultured mammalian cells suggests that they may be involved in comparable responses also in these more complex systems (58). The concentration of pppAppp was estimated in WI-38 cells to be 6 -7% that of 5ЈGTP, or in the low micromolar range, and similar results were obtained with Chinese hamster ovary and BHK-21 cells (58). This could result in 3ЈATP levels consistent with the IC 50 for inhibition of adenylyl cyclase. Thus, an increase in the levels of adenosine or 2Ј-deoxyadenosine 3Јpolyphosphates would precipitate P-site-mediated inhibition of adenylyl cyclase, lowered cellular levels of 3Ј:5Ј-cAMP, and all the downstream effects this would cause. The effectiveness of this form of regulation may differ among tissues, reflecting differential expression of adenylyl cyclase isozymes and their respective sensitivity to P-site ligands. 6 Cell-permeable P-site ligands induce a number of responses in cells and tissues. We and others have used 2Ј,5Ј-dd-Ado with epididymal fat cells (9), isolated hepatocytes (59), primary cultures of thyroid follicles (60), dorsal root ganglion neurons (61), bone organ cultures (62), platelets (6), cortical collecting tubules (63), and cultured mouse fibroblasts (64) to name but a few. End points included changes in water conductance (63), action potential after-hyperpolarization (61), PTH-stimulated bone resorption (62), glycerol production (9), altered enzyme activities (59), DNA synthesis and cell growth (60), and cell differentiation (64). The effects of 2Ј,5Ј-dd-Ado on cell function and on cellular 3Ј:5Ј-cAMP levels have been uniformly consistent with P-site-mediated inhibition of adenylyl cyclase. It is not the effect of 2Ј,5Ј-dd-Ado per se that is important, though, but rather the natural intracellular regulatory processes and ligands that it may mimic.