Direct photoaffinity labeling of individual cytosolic domains of adenylyl cyclase by [32P]2'-deoxy-3'-AMP and [alpha-32P]5'-ATP.

The susceptibility of purines to form a covalent attachment with proteins upon exposure to UV irradiation was applied to adenylyl cyclase by use of [32P]2'-d-3'-AMP, a dead-end inhibitor that binds to the post-transition configuration of the enzyme. [32P]2'-d-3'-AMP was synthesized enzymatically. It and [alpha-32P]5'-ATP were used for direct photocross-linking to individually expressed cytosolic domains of adenylyl cyclase. Both the C1 domain of the type V isozyme (VC1) and the C2 domain of the type II isozyme (IIC2) were labeled, whether alone or combined, upon photolysis of [32P]2'-d-3'-AMP in the presence of acetone. Labeling of VC1 and IIC2 was greatly enhanced in the presence of PPi, was almost completely suppressed by 50 microM 2',5'-dideoxy-3'-ATP, the most potent reported P-site inhibitor of adenylyl cyclases, but was partially suppressed by 1 mM 3'-IMP, a ligand that does not inhibit the enzyme via the P-site. Neither 3':5'-cAMP nor 5'-ATP had a major effect on labeling by [32P]2'-d-3'-AMP. Direct cross-linking of VC1 with [alpha-32P]5'-ATP was substantially suppressed by 2', 5'-dideoxy-3'-ATP and partially suppressed by 2'-d-3'-AMP, whereas cross-linking of IIC2 was less affected by the 3'-triphosphate. The data imply that either cytosolic domain can interact directly with either substrate or P-site ligand and that subunit interaction modifies the susceptibility of each domain to UV-induced covalent modification by either [alpha-32P]5'-ATP or [32P]2'-d-3'-AMP.

The susceptibility of purines to form a covalent attachment with proteins upon exposure to UV irradiation was applied to adenylyl cyclase by use of [ 32 P]2-d-3-AMP, a dead-end inhibitor that binds to the posttransition configuration of the enzyme. [ 32 P]2-d-3-AMP was synthesized enzymatically. It and [␣-32 P]5-ATP were used for direct photocross-linking to individually expressed cytosolic domains of adenylyl cyclase. Both the C 1 domain of the type V isozyme (VC 1 ) and the C 2 domain of the type II isozyme (IIC 2 ) were labeled, whether alone or combined, upon photolysis of [ 32 P]2d-3-AMP in the presence of acetone. Labeling of VC 1 and IIC 2 was greatly enhanced in the presence of PP i, was almost completely suppressed by 50 M 2,5-dideoxy-3-ATP, the most potent reported P-site inhibitor of adenylyl cyclases, but was partially suppressed by 1 mM 3-

IMP, a ligand that does not inhibit the enzyme via the P-site. Neither 3:5-cAMP nor 5-ATP had a major effect on labeling by [ 32 P]2-d-3-AMP. Direct cross-linking of VC 1 with [␣-32 P]5-ATP was substantially suppressed by 2,5-dideoxy-3-ATP and partially suppressed by 2-d-3-AMP, whereas cross-linking of IIC 2 was less affected by the 3-triphosphate. The data imply that either cytosolic domain can interact directly with either substrate or P-site ligand and that subunit interaction modifies the susceptibility of each domain to UV-induced covalent modification by either [␣-32 P]5-ATP or [ 32 P]2-d-3-AMP.
Mammalian adenylyl cyclases (ATP pyrophosphate-lyase (cyclizing); EC 4.6.1.1) comprise a family of proteins with similar domain organization. The enzyme contains a short and variable amino terminus followed by a tandem repeat of transmembrane and cytosolic domains. The cytosolic domains, referred to as C 1 and C 2 , are both required for catalysis (1). Truncated forms of C 1 and C 2 have been chimerically linked or separately expressed and used in the development of soluble forms of the enzyme (1)(2)(3). Both catalytic activity and regulatory properties can be reconstituted by a simple mixture of the two cytosolic domains of the enzyme after their independent synthesis in Escherichia coli (4). When recombined in solution, the independently expressed truncated C 1 domain from the type V adenylyl cyclase (VC 1 ) and the truncated C 2 domain of type II adenylyl cyclase (IIC 2 ), together, demonstrate many of the features of native or wild type enzyme, including stimula-tion by G s ␣ and forskolin and inhibition by 2Ј-d-3Ј-AMP 1 and 2Ј-dAdo (5,6). Thus, these preparations are well suited for studies aimed at identification of domains involved in specific aspects of regulation of this important enzyme family.
An intriguing aspect of adenylyl cyclases is their inhibition by a class of compounds collectively referred to as P-site ligands. The P-site is so designated because of its requirement for an intact purine in the inhibitory ligand. Naturally occurring P-site ligands include, for example Ado, 2Ј-dAdo, 3Ј-AMP, and 2Ј-d-3Ј-AMP (7), and the most potent ligands in this class are 2Ј-deoxy-and 2Ј,5Ј-dideoxyadenosine 3Ј-polyphosphates (8). Inhibition of adenylyl cyclases by this class of compounds is either uncompetitive or noncompetitive with respect to ATP, depending on reaction conditions (9). The nature of inhibition and results from experiments with covalent ligands targeted specifically to the P-site (10,11) give rise to questions about the number, specificity, and localization of nucleotide binding configurations on the enzyme per se. In an approach to these questions, we synthesized [ 32 P]2Ј-d-3Ј-AMP and have used it for photoaffinity labeling of VC 1 and IIC 2 to evaluate distribution of nucleotide binding sites between domains of mammalian adenylyl cyclase and to estimate specificity of nucleotide binding configurations on the enzyme.

EXPERIMENTAL PROCEDURES
Materials-TdNT buffer, alkaline phosphatase, and glycogen were from Promega. Oligonucleotide primer (pdA) 9 and terminal nucleotidyltransferase were from Sigma, and [␣-32 P]2Ј-d-5Ј-ATP (3000 Ci/mmol) was from International Chemical and Nuclear Corp. Calf spleen phosphodiesterase (2 units/mg) was from Boehringer Mannhem. Photolysis was conducted in a Rayonet mini-reactor from Southern New England Ultraviolet Company. VC 1 and IIC 2 Domains of Adenylyl Cyclase-Recombinat VC 1 and IIC 2 were expressed in E. coli and were purified as described (4). Purified proteins were electrophoretically homogeneous with apparent molecular masses of 30 and 26 kDa for VC 1 and IIC 2 , respectively.
Synthesis of [ 32 P]2Ј-d-3Ј-AMP-[ 32 P]2Ј-d-3Ј-dAMP was prepared from enzymatic hydrolysis of 32 P-labeled poly(dA). 32 P-Labeled poly(dA) was prepared by repetitive additions of [ 32 P]2Ј-d-5Ј-AMP moieties to the 3Ј-end of (pdA) 9 in a reaction catalyzed by terminal nucleotidyltransferase. The reaction mixture contained TdNT buffer, 0.05 A 260 units of (pdA) 9 , 35 units of terminal nucleotidyltransferase, 1 to 2 mCi of [␣-32 P]2Ј-d-5Ј-ATP (3000 Ci/mmol) that had been previously lyophilized in a volume of 100 l. After incubation at 30°C for 4 h or overnight, the mixture was treated for 5 min with 1 unit of alkaline phosphatase to remove residual 2Ј-d-5Ј-ATP and to dephosphorylate the 5Ј-end of the oligonucleotide. Without this treatment, overall yields of [ 32 P]2Ј-d-3Ј-AMP were poor and variable. Alkaline phosphatase was inactivated by placing the reaction tube in boiling water for 5 min. 32 P-Labeled * This research was supported by National Institutes of Health Grant DK38828 (to R. A. J.). 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.
Adenylyl Cyclase Assay-Adenylyl cyclase was assayed at 30°C in a 10-min reaction in 100 l of reaction mixture containing 50 mM HEPES buffer, pH 7.5, 1 mM MnCl 2 , 100 M forskolin, 0.5 mM [␣-32 P]5Ј-ATP (2 ϫ 10 13 cpm/mol by Cherenkov radiation), and 50 nM VC 1 and 50 nM IIC 2 . The reaction was started by the addition of [␣-32 P]5Ј-ATP and was terminated by the addition of zinc acetate and sodium carbonate. [ 32 P]cAMP was purified by sequential chromatography on Dowex 50 and alumina as described previously (12). Acetone was included in the assay mixture as indicated at 0.5% v/v to 10% v/v.
Photoaffinity Labeling-Labeling of adenylyl cyclase subunits VC 1 and IIC 2 by [␣-32 P]5Ј-ATP or [ 32 P]2Ј-d-3Ј-AMP was achieved with a 5-min exposure to 254 nm UV light at room temperature in a Rayonet mini-reactor. The 25-l reaction volume contained 50 mM HEPES, pH 7.5, 5 mM MnCl 2 , 100 M forskolin, 0.5% acetone, 100 M [␣-32 P]5Ј-ATP (25 Ci/mmol) or 100 M [ 32 P]2Ј-d-3Ј-AMP (1 to 2 ϫ 10 16 cpm/mmol, by Cherenkov radiation), 5 M of VC 1 , and/or 5 M of IIC 2 . The reaction was initiated by the addition of VC 1 and/or IIC 2 to the rest of the reaction mixture at 0°C. The resulting mixtures were transferred immediately to a Parafilm support in the mini-reactor with 254-nm lamps on. After irradiation, the reaction mixture was transferred to a tube containing 0.1% SDS, 0.1 M dithiothreitol, and 5% glycerol, and this was placed in a boiling water bath for 3 min. Proteins and unreacted nucleotides were separated on an 11% polyacrylamide SDS gel. Protein was visualized by silver staining as described (13). The dried gel was exposed to a phosphoimager screen for 3 to 5 h, and protein labeling was quantified by PhosphorImager and ImageQuant software (from Molecular Dynamics). Alternatively, dried gels were exposed at Ϫ65°C to X-Omat imaging film (from Kodak) for 12 h with an intensifying screen. To quantify the incorporation of 32 P-labeled ligand into proteins, bands corresponding to VC 1 and IIC 2 were cut from dried gels and counted in a liquid scintillation counter.

RESULTS
ATP Photolysis and Effects of Acetone-Simple exposure of 5Ј-ATP in water to high intensity UV light was not sufficient to induce photolysis ( Fig. 1, panel B). No meaningful changes in the UV spectra of ATP were observed after 60 min of irradiation. However, in the presence of 0.5% (v/v) acetone, UV light induced time-dependent photoactivation of the adenine ring of 5Ј-ATP ( Fig. 1, panel A). Half-time for photolysis of 5Ј-ATP was approximately 10 min, and photolysis of 2Ј-d-3Ј-AMP in acetone followed a similar time course (not shown). Consequently, in subsequent experiments, a 5-min exposure with either nucleotide was used for protein labeling. These results are consistent with the idea that energy is transferred from the UVexcited acetone molecules to the adenine ring.
Whereas low concentrations of acetone facilitated photoactivation of 5Ј-ATP or 2Ј-d-3Ј-AMP (above), these concentrations were essentially without effect on adenylyl cyclase (Fig. 2). Acetone at 0.5% to 1% (v/v) did not affect adenylyl cyclase appreciably, but, as expected, higher concentrations inactivated the enzyme. The concentration eliciting a 50% reduction in activity of purified and recombined VC 1 and IIC 2 was approximately 3% acetone. By comparison, crude enzyme extracted from rat brain by simple detergent dispersion was unaffected by concentrations of acetone as high as 10% (not shown). Consequently, in subsequent experiments on photoaffinity labeling of VC 1 and IIC 2 , 0.5% acetone was used.
Photoaffinity  (Fig. 3). 32 P-Labeled ligand was incorporated into 5% of VC 1, as determined by excision of gel slices and counting in a scintillation counter. By this method the extent of labeling of VC 1 was found to be 3.5 times greater than that of IIC 2. The addition of 1 mM 2Ј-d-3Ј-AMP or 50 M 2Ј,5Ј-dd-3Ј-ATP resulted in protection of both VC 1 and IIC 2 domains from covalent labeling by [␣-32 P]5Ј-ATP (Fig. 3). Efficiencies of VC 1 or IIC 2 protection by 2Ј-d-3Ј-AMP and 2Ј,5Ј-dd-3Ј-ATP were estimated by Phospho-rImager techniques. Because isotope decay events (cpm) are directly proportional to arbitrary PhosphorImager units (over 5 orders of magnitude), the ratio of densities of any two bands will directly reflect the ratio of isotope incorporated into the respective proteins. Consequently, the relative protecting effect of a 3Ј-nucleotide ligand is simply measured by comparison of band densities when the density with no added ligand is taken as 100% for each cytosolic domain. A summary of such data obtained from two experiments is given in Table I. It was not possible to attempt photoaffinity labeling of a mixture of VC 1 and IIC 2 by [␣-32 P]5Ј-ATP per se. When VC 1 and IIC 2 are combined at concentrations sufficient to allow complex formation, [␣-32 P]5Ј-ATP is rapidly and completely converted to [ 32 P]cAMP.
Photoaffinity Labeling of Adenylyl Cyclase Domains by [ 32 P]2Ј-d-3Ј-AMP-Irradiation of adenylyl cyclase cytosolic domains with UV light in the presence of 100 M [ 32 P]2Ј-d-3Ј-AMP resulted in covalent modification of both VC 1 and IIC 2 , whether exposed individually or in the form of a VC 1 ⅐IIC 2 complex (Fig. 4). The extent of covalent modification of IIC 2 was 2%, as determined by scintillation counting of excised gels, and the ratio of 32 P incorporated into VC 1 and IIC 2 was 0.7.
Photoaffinity labeling of VC 1 , IIC 2 , or VC 1 ⅐IIC 2 by 100 M [ 32 P]2Ј-d-3Ј-AMP occurred in the presence of a 10-fold molar excesses of either 5Ј-ATP (1 mM) or cAMP (1 mM), respectively, substrate and product of the cyclase reaction (Fig. 4). Although both nucleotides suppressed incorporation of 32 P into VC 1 ϳ 50%, this was less evident for incorporation of 32 P into IIC 2 but was visually most evident when the enzyme was in the VC 1 ⅐IIC 2 complex (Fig. 4, lanes 7-9). By comparison, a 10-fold molar excess of 3Ј-IMP (1 mM) afforded partial protection against 32 P incorporation into IIC 2 but afforded no protection of VC 1 (Fig. 5, lanes 2 and 5). 3Ј-IMP gave partial protection of each subunit when the VC 1 ⅐IIC 2 complex was irradiated in the presence of [ 32 P]2Ј-d-3Ј-AMP (Fig. 5, lane 8). In striking contrast, at a concentration of only 50 M, 2Ј,5Ј-dd-3Ј-ATP resulted in essentially complete protection of both VC 1 and IIC 2 , individually or in complex, when labeled by 100 M [ 32 P]2Ј-d-3Ј-AMP (Fig. 5, lanes 3, 6, and 9). Data from several experiments are summarized in Table II.
The influence of Inorganic Pyrophosphate-Data from both dead-end inhibition kinetics and equilibrium binding studies with 3 H-2Ј-dAdo to VC 1 ⅐IIC 2 in complex with G s ␣ suggested that inorganic pyrophosphate should enhance or be required for binding of P-site ligands (14). Consequently, the effect of both adenylyl cyclase products, cAMP and PP i , alone and in combination, were evaluated for their possible effects on photoaffinity cross-linking with [ 32 P]2Ј-d-3Ј-AMP (Fig. 6). 2 As pre-dicted, PP i (1 mM) substantially enhanced labeling of VC 1 , IIC 2 , or VC 1 ⅐IIC 2 by [ 32 P]2Ј-d-3Ј-AMP (Table II), and cAMP afforded protection that was compatible with its effect in the absence of PP i .

DISCUSSION
Direct photocross-linking of nucleotides with proteins has been widely used for identification of nucleotide binding sites (15,16). However, protein labeling by adenosine derivatives is complicated by instability of the glycoside bond of the UVactivated nucleoside (17) and low quantum yield of purine photoactivation (ϳ3 ϫ 10 Ϫ4 ) (18). This is somewhat circumvented by the use of a chromophore to allow indirect activation of adenine by facilitating energy transfer from UV light to the adenine ring (19). Acetone has been used successfully in this capacity as a sensitizer (19), and site-specific modification of nucleotide binding sites in proteins has been demonstrated with 5Ј-ATP and acetone (20 -22). The effect of acetone to enhance adenine nucleotide photolysis was also clearly evident here (Fig. 1). Even so, covalent modification in such experiments may not exceed a few percent of the protein, limited due either to UV-induced cleavage of the glycosidic bond of adenosine, in our case causing a loss of the reporter 32 P label, or to nonspecific interactions of the labeled ligand with buffer components.
Studies of adenylyl cyclases have included enzyme kinetics, 2 An important technical aspects of these studies is that inorganic pyrophosphate and Mn 2ϩ , which we used in all experiments as our activating ligand, form a complex. Hence, in such experiments PP i must be added immediately before the reaction is initiated. Otherwise precipitates may form, and the effects of PP i will not be evident.   reversibly binding ligands, and covalent modifications of wild type and mutated enzyme. Inhibition kinetics of adenylyl cyclase forward and reverse reactions are consistent with an interaction of P-site ligands with the post-transition state of the enzyme (E°) as dead-end inhibitors (7,9,14,23). By this mechanism, adenine nucleosides and adenosine 3Ј-monophosphates interact with the cAMP binding site and before metal⅐PP i leaving (6, 14 ,23). These analyses would place both substrate 5Ј-ATP and P-site ligands at a common site believed to be formed at the interface of the C 1 and C 2 domains during the reaction cycle (6,14,24,25).
Kinetics analyses involve evaluations that include both ligand binding and enzyme conformational changes, whereas direct measurements of ligand binding assess bimolecular interactions, albeit with a set of possible enzyme configurations. Because substrate and inhibitor share a common binding site, although of different configurations, one might expect some competition between P-site ligands and substrate. However, this was not observed in equilibrium binding studies with [ 3 H]2Ј-dAdo nor with [ 3 H]5Ј-AP(CH 2 )PP (6,14). 2Ј-d-3Ј-dAMP neither displaced nor competed with binding. Binding of these ligands with this chimeric truncated construct of the C 1 ⅐C 2 complex required not only the formation of an active tertiary structure but also catalysis. Because conformational states of an enzyme are in equilibrium and a certain percentage of the enzyme will be in the different configurations, reversible binding techniques may not have been able to detect associations that might be observable with labeled covalent probes.
The availability of [ 32 P]2Ј-d-3Ј-AMP, a more potent and easily detected ligand than [ 3 H]2Ј-dAdo, allowed these interactions to be addressed by a different technique. The data presented here clearly show that by use of direct photo crosslinking, both [␣-32 P]5Ј-ATP and [ 32 P]2Ј-d-3Ј-AMP labeled both VC 1 and IIC 2 (Figs. [3][4][5][6], indicating that both ligands can interact with either VC 1 or IIC 2 . This is consistent with the high level of sequence homology between C 1 and C 2 (26) in all mammalian adenylyl cyclases and is not inconsistent with the idea that the two cytosolic domains are required for catalysis and P-site-mediated inhibition (1-6). Because photoactivation induces covalent cross-linking through the adenine ring, the fact that both VC 1 and IIC 2 are labeled by [ 32 P]2Ј-d-3Ј-AMP also in the catalytically competent complex of VC 1 ⅐IIC 2 implies contact of the adenine ring of this ligand with both VC 1 and IIC 2 . Because cross-linking results in an irreversible linkage, it will select for and lock in those enzyme configurations with which the ligands interact. This is quite different from the results one can obtain from either enzyme kinetics or reversible binding studies. It was therefore not surprising that a measurable but weak competition between 5Ј-ATP and 2Ј-d-3Ј-AMP was noted. Both bind to the enzyme under similar conditions. Either will displace binding of the other, but only partially (Tables I and II), and that at 10-fold greater concentrations of the competing ligand. These observations are consistent with the steady-state kinetic behavior of the soluble enzyme (6,9,14) but not inconsistent with the lack of competition of 2Ј-d-3Ј-AMP in the equilibrium binding studies with [ 3 H]5Ј-AP(CH 2 )PP, which suggested that 2Ј-d-3Ј-AMP binds after 5Ј-ATP but before the return of the enzyme to its initial state. Shown here is that [ 32 P]2Ј-d-3Ј-AMP can bind to either free enzyme (E ϩ I 3 E⅐I) or to the post-transition state of the   enzyme in the absence of pyrophosphate (E°ϩ I 3 E°⅐I). PP i , which shifts the equilibrium of enzyme to a state that accepts P-site ligands, enhanced labeling of the individual domains 3to 5-fold and labeling of the VC 1 ⅐IIC 2 complex 7-to 9-fold (Table  II). This effect of PP i is fully consistent with the proposed model for inhibition by P-site ligands (6,14,23), and it is consistent with the effect of PP i to enhance binding of [ 3 H]2Ј-dAdo to the VC 1 ⅐IIC 2 complex (14). By comparison, cAMP afforded less protection of VC 1 and IIC 2 from labeling by [ 32 P]2Ј-d-3Ј-AMP than might have been expected (Figs. 4 and 6 and Table II). This may be explained by the low apparent affinity of cAMP for adenylyl cyclase (K m cAMP ϳ16 mM in the reverse reaction). The use of cAMP concentrations sufficient to interact with the enzyme effectively, e.g. 10 mM, could result in a "quenching" effect, absorbing UV light used to activate the nucleotides. Overall the cross-linking observed with both [ 32 P]2Ј-d-3Ј-AMP and [␣-32 P]5Ј-ATP to both both VC 1 and IIC 2 and the weak competitive behavior are consistent with reaction kinetics and with the idea that the two cytosolic domains are required for catalysis and P-site-mediated inhibition (1-6, 14, 24, 25).
The reduced labeling by [ 32 P]2Ј-d-3Ј-AMP noted upon association of VC 1 with IIC 2 , particularly with regard to effects of competing ligands , suggests either that interaction of the cytosolic domains affects access of ligand to those sites to which covalent linkages are formed or that the association forms a preferential configuration to which the ligands have differential access. That is, association of C 1 and C 2 domains changes during the catalytic cycle, exhibiting more than one conformational state. These are represented minimally by two configurations, one that is a catalytically competent configuration and one that occurs as a result of P-site-mediated inhibition. These two configurations are in equilibrium, but the equilibrium between them is not rapid. Differential protection by competing ligands of cross-linking with [␣-32 P]5Ј-ATP or [ 32 P]2Ј-d-3Ј-AMP would result from these two distinct enzyme configurations with which these nucleotides interact.
The presence of multiple conformation states is also supported by the facts (i) that diffusion of 2Ј-d-3Ј-ATP into crystals of IC 1 ⅐IIC 2 ⅐G s ␣ resulted in structure disordering and (ii) that formation of crystals with 2Ј-d-3Ј-ATP was unsuccessful (25). The VC 1 ⅐IIC 2 complex with 2Ј-d-3Ј-AMP and PP i is not well ordered, and Tesmer et al. (25) suggest that this is because of incomplete occupancy or incomplete formation of the binding pocket because of constraints imposed by the crystalline lattice. Whereas the locus of 2Ј-d-3Ј-AMP and PP i in structures of VC 1 ⅐IIC 2 are clear and suggestions have been made as to the locus for 5Ј-ATP in models derived from these (27), no crystal structure of a C 1 ⅐C 2 ⅐ATP complex is available, whether with 5Ј-ATP, 3Ј-ATP, or a derivative of either (24 -26). The rank order for inhibitory potency of P-site ligands (2Ј-dAdo Ͻ 2Ј-d-3Ј-AMP Ͻ 2Ј-d-3Ј-ADP Ͻ 2Ј-d-3Ј-ATP Ͻ 2Ј-d-3Ј-A4P) (7,28), the effect of PP i to enhance binding of [ 3 H]2Ј-dAdo (14), and the effect of PP i to enhance covalent cross-linking of 2Ј-d-3Ј-AMP ( Fig. 6 and Table II) clearly demonstrate the important contribution of 3Ј-(␤, ␥, and ␦)-phosphates to the association of these ligands with adenylyl cyclase. Thus it was not surprising that the relatively low concentration of 50 M 2Ј,5Ј-dd-3Ј-ATP caused substantial protection against labeling of VC 1 or IIC 2 by 100 M [ 32 P]2Ј-d-3Ј-AMP and caused virtually complete protection against labeling of the catalytically competent VC 1 ⅐IIC 2 complex ( Fig. 5 and Table II). This contrasts sharply with the much less effective protection afforded by 10-fold greater concentration of either 5Ј-ATP or 3Ј-IMP (each 1 mM; Figs. 4 and 5). In the VC 1 ⅐IIC 2 complex 3Ј-IMP afforded only 50 -60% protection of IIC 2 and only 30% protection of VC 1 and afforded no protection of VC 1 alone (Fig. 5 and Table II).
Taken together the data strongly argue that the site of covalent cross-linking by [ 32 P]2Ј-d-3Ј-AMP has the same characteristics as that through which P-site-mediated inhibition of native adenylyl cyclases occurs. It is selective for an intact adenine moiety, and it recognizes the ribosyl 3Ј-polyphosphate moiety (cf. Refs. 7,8,28,29). Effective inhibition by or binding of adenine nucleosides or adenine nucleoside-3Ј-monophosphates requires the presence of PP i , whereas inhibition by the adenosine-3Ј-tri-or 3Ј-tetraphosphates (7, 28) does not. P-sitedirected ligand does not compete with 5Ј-ATP for binding with adenylyl cyclase (9), and P-site-targeted covalent modification occurs in the presence of 5Ј-ATP (11). The catalytically competent form of the enzyme exists in two conformational states, one with which substrate, metal-5Ј-ATP interacts, and a posttransition state in which product is released and with which P-site ligands interact. It is this latter form that [ 32 P]2Ј-d-3Ј-AMP covalently modifies and the labeling of which is enhanced by pyrophosphate. Each cytosolic domain evidently contains elements of a nucleotide binding site, but an effective catalytic cleft must form only when they associate. Because the two states of the enzyme are not in rapid equilibrium, the opportunity is provided for independent binding of ligands to substrate and P-site configurations of the enzyme.