Inhibition of adenylyl cyclase by acyclic nucleoside phosphonate antiviral agents.

Acyclic derivatives of adenine, known as highly effective nucleotide analogs with broad spectrum antiviral activity, were evaluated for potential cross-reactivity with adenylyl cyclases, a family of membrane-bound enzymes that share putative topologies at their catalytic sites with oligonucleotide polymerases and reverse transcriptases. A series of derivatives of 9-(2-phosphonylmethoxyethyl)adenine (PMEA) inhibited a preparation of adenylyl cyclase derived from rat brain with IC(50) values that ranged from 66 microM (PMEA) to 175 nM for its diphosphate derivative (PMEApp) and mimics of it. PMEApp mimics included PMEAp(NH)p, PMEAp(CH(2))p, PMEAp(CX(2))p (X = fluorine, chlorine, or bromine), PMEAp(CHX)pp, and PMEAp(C(OH)CH(3)pp. The data suggest that inhibition of adenylyl cyclases may contribute to the therapeutic action of some of these or similar compounds or constitute part of their side effects in therapeutic settings.

Adenylyl cyclases are a family of membrane-bound enzymes central to one of the most important signal transduction systems and influence regulation of cell function in virtually all cells. The putative membrane topology of adenylyl cyclases conforms to a repeated sequence of a six-membrane spanning region followed by a cytosolic domain (1). The two cytosolic domains (C 1 and C 2 ) are homologous and contain regions that are highly conserved among adenylyl cyclase isozymes (1). The cleft formed by interaction of C 1 and C 2 contains the enzyme catalytic site (2,3), the topology of which resembles aspects of the palm domain of DNA polymerases and human immunodeficiency virus (HIV) 1 reverse transcriptase (4 -7). Moreover, aspects of the catalytic mechanisms of adenylyl cyclases and enzymes involved in polymerization of oligonucleotides are similar as well. Each involves nucleoside triphosphate as substrate and divalent cation-dependent catalysis that includes attack involving the substrate 3Ј-OH group, to catalyze either chain elongation of a primer oligonucleotide or formation of the 3Ј:5Ј-cyclic phosphate, with pyrophosphate as a leaving group. Adenine nucleoside 3Ј-polyphosphates are among the most potent inhibitors of adenylyl cyclases (8 -11) and of these 2Ј-d-3Ј-ATP has been known for some time to bind to DNA polymerases (12).
Acyclic nucleoside phosphonates belong to a class of highly effective nucleotide analogs with broad spectrum antiviral activity (13)(14)(15)(16). Of these 9-(2-phosphonylmethoxyethyl)adenine (PMEA) has demonstrated antiviral activity against HIV and different types of DNA viruses (13)(14)(15)(16), differentiation-inducing activity (17,18), and anti-tumor activity (18,19). A related acyclic adenine derivative, PMPA, also is active against HIV and several members of the herpesvirus family (15,20). Prodrug forms of these compounds provide transient protection of the phosphonate charge, allowing them to enter cells where they undergo deprotection and subsequent phosphorylation to their active inhibitory forms (19 -23). Reported here is an evaluation of PMEA and various analogs of PMEApp as inhibitors of adenylyl cyclase.

EXPERIMENTAL PROCEDURES
Assay of Adenylyl Cyclase-Adenylyl cyclase was prepared as a detergent extract from rat brain as described previously (24,25). IC 50 values were determined graphically from logistic regression plots of adenylyl cyclase inhibition curves.

RESULTS
The adenylyl cyclase extracted from rat brain has served as a basis for comparison of the effects of numerous adenine nucleosides and nucleoside phosphates (8 -11). PMEA inhibited this enzyme with an IC 50 of approximately 66 M (Table I). Whereas this was notably higher than the previously reported inhibition by 3Ј-AMP or 2Ј-d-3Ј-AMP (IC 50 ϳ8.9 and 1.2 M, respectively), it was below that for 5Ј-AMP or 2Ј-d-5Ј-AMP (IC 50 ϳ150 and Ͼ300 M, respectively) (8), reflecting an intermediate capacity of this acyclic phosphonate to interact with this adenylyl cyclase. Because adenine nucleoside 3Ј-polyphosphates exhibited progressively lower IC 50 values as 3Ј-phosphate groups were added (Table I) (9 -11), and because it is the diphosphate derivative of PMEA that effects its antiviral action in intact cells (21,30), the diphosphate form of PMEA was also tested. PMEApp and the corresponding imidodiphosphate derivative, PMEAp(NH)p, exhibited comparable potencies (IC 50 ϳ170 and ϳ180 nM, respectively). Although PMEApp and * This work was supported by National Institutes of Health research Grant DK 38828 and an Innovative Technology Grant from the Center for Biotechnology (to R. A. J.). Early aspects of the synthetic work were supported by a fellowship from the Deutsche Forschungsgemeinschaft (to W. L.). We are grateful to the Alexander-von-Humboldt-Stiftung for the Feodor Lynen Fellowship awarded (to W. L.). 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.
PMEAp(NH)p were not as potent as 2Ј-d-3Ј-ATP or 2Ј,5Ј-dd-3Ј-ATP (cf . Table I), they fall within ranges seen with this class of inhibitors of this enzyme family and suggest that they interact with the enzyme catalytic cleft in a similar orientation.
Adenylyl cyclases catalyze formation of cAMP from 5Ј-ATP and also from 5Ј-APP(NH)P and 5Ј-APP(CH 2 )P but with reduced catalytic efficiency (24). The presumption has been that because of their changed electronic structures the imidodiphosphate and methylenediphosphate moieties do not form as good a leaving group as does inorganic pyrophosphate. Consequently, it was not surprising that the ␤␥-methylene diphosphate derivative PMEAp(CH 2 )p (IC 50 ϳ5 M) exhibited a reduced inhibitory potency compared with the unmodified PMEApp ( Fig. 2 and Table I). Halogen substitutions within the ␤␥-methylene diphosphate caused noticeable but modest improvements in potency (Fig. 3) as did the substitution of the -CH 2 -group with -C(CH 3 OH)- (Table I). The most potent of these was PMEAp(CF 2 )p (IC 50 ϳ1.5 M; Fig. 3 and Table I). The effects on the cyclase of these substitutions were as expected from the altered size and electronic character of the pyrophosphate leaving group in these PMEApp mimics. DISCUSSION The topology of the adenylyl cyclase catalytic site and aspects of its catalytic reaction resemble those of oligonucleotide polymerases and guanylyl cyclases (3, 4 -7, 12, 31-33). These similarities in catalytic mechanism and structure among distinctly different enzyme families suggest that agents acting within the catalytic cleft of one may interact also with the other. This has been borne out in earlier studies with these enzyme families with some ligands and was extended here with the inhibition of adenylyl cyclases by a class of 9-substituted adenine acyclic phosphonate derivatives (Table I). PMEA, its diphosphate derivative, PMEApp, and mimics of PMEApp inhibited adenylyl cyclase with potencies (IC 50 Ͻ200 nM) that were reminiscent of those of 2Ј,5Ј-dd-3Ј-ATP (IC 50 ϳ40 nM) (9, 10) and ␤-L-2Ј,3Ј-dd-5Ј-ATP (IC 50 ϳ24 nM; K d 16 nM) (35). These latter two nucleotides are the most potent nucleoside triphosphate inhibitors of this enzyme, but inhibit by different mechanisms. Inhibition by 2Ј,5Ј-dd-3Ј-ATP is of the post-transition state of the enzyme through a noncompetitive, dead-end mechanism (9 -11), whereas that by ␤-L-2Ј,3Ј-dd-5Ј-ATP is of the pretransition state and is competitive (35). This implies that the catalytically competent and the 3Ј-nucleotide-inhibited configurations of the enzyme differ.
In addition to adenylyl cyclases, PMEApp inhibits viral re-   2. Inhibition of rat brain adenylyl cyclase by several acyclic adenine derivatives. Enzyme was prepared and assayed as described under "Experimental Procedures". The rat brain extract was assayed with 0.1 mM 5Ј-ATP and 5 mM MnCl 2 . Average initial velocity was, in nmol/(min⅐mg protein): 3.96 Ϯ 0.18 (n ϭ 4). verse transcriptases (13)(14)(15)(16). It is presumably this mechanism through which the oral prodrug derivative of PMEA (adefovir dipivoxil) effects the therapeutic and potent inhibition of retroviruses and hepatitis B virus (19,21,30), but it is less certain that this is the means by which this drug also enhances differentiation and exhibits anti-tumor activity (17)(18)(19). This bis(S-acyl-2-thioethyl) ester derivative of PMEA undergoes deprotection by cellular carboxyesterase(s) and subsequent phosphorylation to PMEApp to yield the active inhibitor of reverse transcriptases (21,30). The data presented here, though, suggest that inhibition of adenylyl cyclase may contribute to either the therapeutic action or to the side effects of this drug. This caveat would also apply to analogously acting antiviral compounds in this class, such as PMPA (15,18,20,23). Given the central role that adenylyl cyclases play in the regulation of cell metabolism, function, and development, drugs that affect this enzyme family could be expected to elicit a variety of effects on cells. A pertinent example is the acceleration in differentiation in preadipocytes that was noted with adenine nucleoside inhibitors of adenylyl cyclase (34). Moreover, because there are at least nine adenylyl cyclase isozymes that are differentially expressed in cells and tissues, the effects of such drugs may also elicit tissue and cell-dependent effects.
Although this cross-reactivity of inhibitors likely extends to many of the enzymes for which nucleoside triphosphates are substrates and pyrophosphate is leaving group (3, 4 -7, 12, 31-33), the identification of enzymes with which these antiviral agents interact and the knowledge of their structures and respective inhibitory profiles should allow more specific ligands to be designed. First, the similarity in catalytic mechanism and homologous structure among guanylyl and adenylyl cyclases (1,31,32) would suggest that guanine-based acyclic phosphonate derivatives would inhibit guanylyl cyclases and may be useful in reducing cellular cGMP levels. Second, if inhibition of adenylyl cyclase contributes to the side effects of the antiviral drugs, it can be readily designed around simply by use of some other base or by almost any modification of the adenine moiety, e.g. 3-deaza-, or 7-deaza-adenine (8 -11, 35, 36). These changes would yield agents that do not affect adenylyl cyclase but which may yet inhibit oligonucleotide polymerases. Third, if therapeutic effects of adenine acyclic phosphonate derivatives are in fact brought about by inhibition of adenylyl cyclases or if inhibition of adenylyl cyclase yields the appropriate pharmacologic or therapeutic action, e.g. differentiation (34), more specific inhibitors could be developed that take advantage of our knowledge of the structures of this enzyme and its inhibitors. Some of the known potent inhibitors of this enzyme may be used or form the basis for the development of more selective ligands. The most effective agents might well be transition-state inhibitors, and adenine nucleoside 3Ј-polyphosphates are the closest to these of the known inhibitors of this enzyme family. Their usefulness for such purposes should be enhanced substantially by the use of transient protecting groups that will allow these compounds to enter cells and find application in intact cell systems. Other enzymes with nucleoside triphosphate as sub-strate and pyrophosphate as leaving group may be similarly and more specifically targeted.