Covalent Labeling of Adenylyl Cyclase Cytosolic Domains with γ-Methylimidazole-2′,5′-dideoxy-[γ-32P]3′-ATP and the Mechanism for P-site-mediated Inhibition

A truncated first cytosolic domain of type V adenylyl cyclase (VC1) and a truncated second cytosolic domain of type II adenylyl cyclase (IIC2) were used alone and in the readily reversible complex (VC1·IIC2) to evaluate interactions with each other and with reversible and irreversible P-site ligands. Enzyme activity was used to assess formation and dissolution of VC1·IIC2. The data suggest that binding of 2′,5′-dideoxy-3′-ATP to VC1 and IIC2 prevented formation of VC1·IIC2 and that 2′,5′-dideoxy-3′-ATP dissociation occurred slowly. To enable configuration specific cross-linking to the catalytic site, 2′,5′-dideoxyadenosine 3′-[γ-(1-methylimidazole)-triphosphate] (γ-MetIm-2′, 5′-dd-3′-ATP) and 2′,5′-dd-adenosine 3′-(γ-azidoanilido)-triphosphate (γ-azidoanilido-2′,5′-dd-3′-ATP) were synthesized, the former also as its γ-32P-labeled analog. γ-Azidoanilido-2′,5′-dd-3′-ATP exhibited an inhibitory potency comparable with that of 2′,5′-dd-3′-ATP. γ-MetIm-2′,5′-dd-[γ-32P]3′-ATP labeled the individual VC1 and IIC2 domains comparably and covalently to ∼20% within 1 h. Formation of VC1·IIC2 resulted in reduced labeling of VC1 but enhanced labeling of IIC2. The data imply that formation of the catalytically active VC1·IIC2 complex affects the interaction of each domain with the 2′,5′-dd-3′-ATP, the binding of which also affects the interaction between the two cytosolic domains, leading to a pseudo-irreversible inhibition.

The proposed topology of mammalian adenylyl cyclases suggests a tandem repeat of a membrane-spanning region and a cytosolic domain, with the two cytosolic domains (C 1 and C 2 ) sharing significant sequence homology (1). When the C 1 domain of the type V isozyme (VC 1 ) and the C 2 domain of the type II isozyme (IIC 2 ) are expressed separately in Escherichia coli and then recombined, a functional adenylyl cyclase is formed upon their association in solution (2). This truncated chimeric construct exhibits stimulation by G s ␣ or forskolin and inhibition by P-site ligands, 1 properties that are characteristic of the native enzyme (2)(3)(4)(5). Available evidence suggests that both catalysis and inhibition by P-site ligands occur at a common site within a cleft formed at the interface of the two cytosolic domains (6,7). The noncompetitive inhibition seen by P-site ligands (8) is viewed as occurring via a post-transition state dead-end mechanism (5), although it is unclear how this is effected, especially in view of the competitive inhibition observed with comparably potent substrate analogs that act at the same locus but apparently on the pretransition state of the enzyme (9,10). The difference in inhibitory mechanisms induced by the ATP isomers suggested oscillation of adenylyl cyclase between two conformational states but only one of which could bind P-site ligands.
In earlier studies we demonstrated through direct photo cross-linking of VC 1 and IIC 2 cytosolic domains with 5Ј-[␣-32 P]ATP and 2Ј-d-3Ј-[ 32 P]AMP that each cytosolic domain was capable of binding both substrate and P-site inhibitor (11). The data suggested that the individual cytosolic domains exhibit distinct conformational states and that these are not in rapid equilibrium, thereby allowing independent binding of substrate and inhibitor with the individual domains (11). These observations suggested the possibility that the binding of P-site ligands to C 1 and/or C 2 may substantially alter the interaction between the domains. Inasmuch as one of the most potent ligands, 2Ј,5Ј-dd-3Ј-ATP 2 (IC 50 ϳ40 nM; (12)) may occupy binding domains for both products, cAMP and inorganic pyrophosphate and is readily prepared as a 32 P-labeled ligand (13), we undertook a study of the interaction of this ligand with VC 1 and IIC 2 alone and in combination. To circumvent the poor quantum efficiency of direct photo cross-linking of labeled adenosine derivatives, which we noted in the earlier studies (11, 14 -16), chemically activated derivatives of 2Ј,5Ј-dd-3Ј-ATP were synthesized and used for these studies.
Assay of Adenylyl Cyclase-Adenylyl cyclase assays followed previously described procedures (19). Reactions were in a medium containing 50 mM HEPES, pH 7.5, 5 mM MnCl 2 , 100 M forskolin, 0.5 mM [␣-32 P]5Ј-ATP (10 -60 cpm/pmol, by Cherenkov radiation), 1 M VC 1 and 1 M IIC 2 , in a volume of 30 l for 5 min at 30°C. Reactions were started by the addition of VC 1 ⅐IIC 2 complex and were terminated by the addition of zinc acetate and sodium carbonate. [ 32 P]cAMP was purified by sequential chromatography on Dowex 50 and Al 2 O 3 and was quantified by Cherenkov radiation in a liquid scintillation spectrometer.
Determinations of Stability of the VC 1 ⅐IIC 2 Complex-Adenylyl cyclase activity was used as a measure of the formation of the active VC 1 ⅐IIC 2 complex. Association of the VC 1 ⅐IIC 2 complex was achieved in a reaction mixture containing 50 mM HEPES, pH 7.5, 5 mM MnCl 2 , 0.1 mM forskolin, 2 M VC 1, and 2 M IIC 2, in a volume of 30 l for 2 min at 22°C. VC 1 ⅐IIC 2 complex formation was evaluated in the absence or presence of 10 M 2Ј,5Ј-dd-3Ј-ATP, 1 mM 5Ј-ATP, or in the absence of nucleotides and then after sequential treatment either with 10 M 2Ј,5Ј-dd-3Ј-ATP followed by 1 mM 5Ј-ATP, or with 1 mM 5Ј-ATP followed by 10 M 2Ј,5Ј-dd-3Ј-ATP, for 2 min at 22°C per each ligand. Dissociation of the VC 1 ⅐IIC 2 complex was initiated by dilution of the enzymereaction mixture 20-fold with 50 mM HEPES, pH 7.5, 5 mM MnCl 2, and 100 M forskolin. At various times after dilution of the enzyme-reaction mixture, 30 l portions were added to 3 l of 10 mM [␣-32 P]5Ј-ATP (10 -20 cpm/pmol, by Cherenkov radiation). To estimate adenylyl cyclase activity before dilution, portions of the enzyme were taken directly from the complex formation mixture and were added to an adenylyl cyclase assay mixture containing 50 mM HEPES, pH 7.5, 5 mM MnCl 2 , 100 M forskolin, and 0.9 mM [␣-32 P]5Ј-ATP (10 -20 cpm/pmol, by Cherenkov radiation). Adenylyl cyclase activity was then determined as above.

Effects of 5Ј-ATP and 2Ј,5Ј-dd-3Ј-ATP on the Dissociation of Adenylyl Cyclase Cytosolic Domains-
The association of C 1 and C 2 cytosolic domains of adenylyl cyclase yields the catalytically competent VC 1 ⅐IIC 2 complex in a reversible process with an apparent K d ϳ1 M (2). Because the individual VC 1 and IIC 2 domains do not exhibit meaningful catalytic activity (2), adenylyl cyclase activity could be used to estimate the amount of the functionally active VC 1 ⅐IIC 2 complex. This is demonstrated in the readily reversible nature of the active complex upon 20-fold dilution of the preformed VC 1 ⅐IIC 2 complex (Fig. 2, control plot). The dilution was from 2 M enzyme to a concentration of each cytosolic domain that was an order of magnitude below the K d and resulted in a dissociation that occurred rapidly with Ͼ50% inactivation occurring by the time of the earliest sampling at 15 s. When VC 1 and IIC 2 were added separately to a solution already containing 10 M 2Ј,5Ј-dd-3Ј-ATP (plot 2) the resulting adenylyl cyclase activity was low and remained low after dilution. This suggested that the preassociation of inhibitor with enzyme prevented the formation of a functionally active VC 1 ⅐IIC 2 complex. However, when 10 M 2Ј,5Ј-dd-3Ј-ATP was added to a preformed VC 1 ⅐IIC 2 complex just before dilution (Fig. 2, plot 3), initially measured activity was reduced to approximately 15% of the value seen without the inhibitor (compare values at 0 min for control and plot 3). Upon dilution there may have been a small increase in activity in the 15 s sample, conceivably due to dissociation of weakly bound inhibitor, but thereafter the decrease in adenylyl cyclase activity occurred at a significantly lower rate than was seen with control enzyme in the absence of 2Ј,5Ј-dd-3Ј-ATP. Even after 60 min the activity of the enzyme exposed to inhibitor remained higher than that of the control enzyme (Fig. 2, plot 3). A similar behavior was observed if, just before dilution, the preformed VC 1 ⅐IIC 2 complex was treated sequentially with 2Ј,5Ј-dd-3Ј-ATP and then 5Ј-ATP (Fig. 2, plot 4), suggesting that even 1 mM 5Ј-ATP did not accelerate dissociation of the inhibitor. Thus, treatment of the VC 1 ⅐IIC 2 complex with 2Ј,5Ј-dd-3Ј-ATP followed by 5Ј-ATP neither activated the enzyme nor affected the rate of subunit dissociation, when compared with enzyme treated with 2Ј,5Ј-dd-3Ј-ATP alone (Fig. 2, plots 3 and 4).
These data suggest that the control plot reflects the rapid dissociation of the VC 1 ⅐IIC 2 complex, whereas the slow rate of activity loss observed in plot 3 might well reflect the slow rate of dissociation of 2Ј,5Ј-dd-3Ј-ATP from the VC 1 ⅐IIC 2 enzyme complex, because activity was higher than that seen when the VC 1 ⅐IIC 2 complex was formed in the presence of inhibitor (plot 2). Alternatively, both inhibitor and substrate may stabilize the enzyme, and the shallow slopes of plots 3 and 4 may reflect enzyme inactivation due to other factors, e.g. thermal inactivation. The data (Fig. 2, plot 2) imply that binding of 2Ј,5Ј-dd-3Ј-ATP to the individual VC 1 and IIC 2 domains prevents the formation of a functionally active VC 1 ⅐IIC 2 complex. That is, the binding of inhibitor keeps the enzyme domains apart or in an inappropriately associated complex.
The behavior of the VC 1 ⅐IIC 2 complex when treated with 5Ј-ATP (Fig. 3) was different from that above. The dissociation data of control VC 1 ⅐IIC 2 enzyme formed in the absence of nucleotides (Fig. 3, control plot) are the same as shown in Fig. 2, control plot. If the enzyme was treated with 5Ј-ATP just before dilution, though, there was a notable 2.5-fold greater initial activity (compare initial values in plots 1 and 3) and the halftime of complex dissociation was about 10 min (Fig. 3, plot 1). The apparent increase in enzyme activity was not due to an effect of carryover 5Ј-ATP into the assay mixture because this would represent at maximum an 11% increase in the 900 M substrate concentration used in these assays and the K m for 5Ј-ATP for this chimeric construct has been reported to be in the range of 72-220 M in assays with Mn 2ϩ as cation (4). It is more likely that substrate and/or products formed during the 2-min pretreatment period stabilized the VC 1 ⅐IIC 2 enzyme complex or facilitated its formation. Under these reaction conditions 50% of the substrate was converted into cAMP and PP i during the treatment of the VC 1 ⅐IIC 2 complex with 5Ј-ATP for 2 min at 22°C (not shown). This implies that the exchange of 5Ј-ATP, cAMP, and PP i between adenylyl cyclase and solution is a rapid process. Otherwise, saturation of the active site with nonradioactive substrate and products would have effectively restricted the formation of [ 32 P]cAMP from 32 P-labeled substrate in the subsequent adenylyl cyclase assay. Because these reactions were conducted with 2 M for each VC 1 and IIC 2 and the K d ϳ1 M, approximately half the concentration of each cytosolic domain will be bound in the VC 1 ⅐IIC 2 complex. The  2. Effect of 2,5-dd-3-ATP on the dissociation of the VC 1 ⅐IIC 2 complex. Adenylyl cyclase cytosolic domains VC 1 and IIC 2 (2 M each) were preassociated to form the VC 1 ⅐IIC 2 complex and then allowed to dissociate upon dilution as described under "Experimental Procedures." The rate of cAMP formation was normalized to enzyme concentration, and activities are expressed as cAMP formed/min. Values shown are averages from either two or three separate experiments. The first (zero time) point on each curve was taken immediately before dilution and the second point (15 s) was taken as soon as possible after the 20-fold dilution. Time courses are of adenylyl cyclase activity of enzyme that had been treated as follows: Control, the VC 1 ⅐IIC 2 complex was formed in the absence of nucleotides; plot 2, VC 1 was added to a solution of 10 M 2Ј,5Ј-dd-3Ј-ATP and incubated for 2 min at room temperature, then IIC 2 was added and incubated for an additional 2 min before zero time sampling and then dilution; plot 3, the preformed VC 1 ⅐IIC 2 complex was prepared in the absence of nucleotides (with a 2-min incubation at room temperature) and then 10 M 2Ј,5Ј-dd-3Ј-ATP was added for an additional 2 min at room temperature before zero time sampling and dilution; plot 4, the preformed VC 1 ⅐IIC 2 complex was prepared in the absence of nucleotides (as for plot 3), but then before dilution was treated sequentially with 10 M 2Ј,5Ј-dd-3Ј-ATP and 1 mM 5Ј-ATP for 2 min at room temperature per each ligand. Inset, enlarged depiction of plots 2 and 3.

FIG. 3.
Effect of 5-ATP on the dissociation of the VC 1 ⅐IIC 2 complex. Experimental conditions were as described under "Experimental Procedures" and in the legend to Fig. 2. The rate of cAMP formation was normalized to enzyme concentration and activities are expressed as cAMP formed/min. Values shown are averages from three separate experiments. Time courses are of adenylyl cyclase activity upon the dilution of the VC 1 ⅐IIC 2 complex: Plot 1, formed in the absence of nucleotides and then treated before a 20-fold dilution with 1 mM 5Ј-ATP for 2 min at 22°C; plot 2, formed in the absence of nucleotides, but then treated sequentially before dilution with 1 mM 5Ј-ATP and 10 M 2Ј,5Ј-dd-3Ј-ATP, 2 min at 22°C per each ligand; control, formed in the absence of nucleotides (the same data as shown in Fig. 2, control).
high substrate concentration should facilitate association of VC 1 and IIC 2 yielding a larger percentage of the enzyme in the catalytically active complex and result in greater measured activity.
Comparison of these effects of 5Ј-ATP and 2Ј,5Ј-dd-3Ј-ATP would suggest that binding of the inhibitor is much stronger than of substrate. This is fully consistent with the observation that IC 50 values for inhibition by 2Ј,5Ј-dd-3Ј-ATP is three orders of magnitude lower than the K m value for metal-5Ј-ATP substrate, seen with native as well as recombinant adenylyl cyclase (12,20). The data show further that 2Ј,5Ј-dd-3Ј-ATP and 5Ј-ATP do not compete for the same enzyme configuration, consistent with the noncompetitive inhibition typical of P-site ligands.
Covalent Labeling of VC1 and IIC 2 with ␥-MetIm-2Ј,5Ј-dd-[␥-32 P]3Ј-ATP-To verify that 2Ј,5Ј-dd-3Ј-ATP could, in fact, bind to VC 1 and IIC 2 domains independently, and thereby alter their association to the catalytically competent VC 1 ⅐IIC 2 form, cross-linking ligands were synthesized and used. Substitution of 2Ј,5Ј-dd-3Ј-ATP at the ␥-phosphate with an azidoanilido group did not meaningfully affect the potency of the 3Ј-nucleotide to inhibit adenylyl cyclase (Fig. 4). Although irradiation of VC 1 , IIC 2 , or VC 1 ⅐IIC 2 with UV light at 300 nm in presence of ␥-azidoanilodo-2Ј,5Ј-dd-[␥-32 P]3Ј-ATP resulted in covalent modification of both subunits (not shown), the level of labeling was less than 1%. This was measurably better incorporation of label than we noted in the earlier use of direct UV irradiation in the presence of acetone as a sensitizer (11). Because a more reactive reagent would be preferable, ␥-MetIm-2Ј,5Ј-dd-3Ј-ATP was synthesized in both unlabeled and ␥-32 P-labeled forms (cf. see "Experimental Procedures"). However, due to the high rate of hydrolysis in aqueous solutions of ␥-MetIm-2Ј,5Ј-dd-3Ј-ATP to the unsubstituted 2Ј,5Ј-dd-3Ј-ATP, a meaningful comparison of its inhibitory potency with either 2Ј,5Ј-dd-3Ј-ATP or the azidoanilido derivative was not possible. The fact that the azidoanilido group did not impair inhibitory potency, an observation that was consistent with the established tolerance of adenylyl cyclase to large substitutions at the 3Ј-ribosyl position (21), suggested that the comparably sized ( Fig. 1) but more reactive ␥-methylimidazole derivative may also be useful for covalent labeling of the enzyme.
Incubation of VC 1 , IIC 2 , or VC 1 ⅐IIC 2 with ␥-MetIm-2Ј,5Ј-dd-[␥-32 P]3Ј-ATP resulted in covalent labeling of both cytosolic domains (Figs. 5-7). 32 P-Incorporation increased with time and approximately 20% of IIC 2 became labeled after a 60-min exposure to this ligand (Fig. 5). VC 1 was labeled with a compa-rable time course (not shown) and efficacy (Figs. 6 and 7, cf.  lanes 1 and 4), as calculated from PhosphorImager data. Formation of a functional VC 1 ⅐IIC 2 complex resulted in reduced 32 P-incorporation into VC 1 and enhanced 32 P-incorporation into IIC 2 (Figs. 6 and 7, lane 7). Quantitatively, in the VC 1 ⅐IIC 2 complex 32 P-incorporation into VC 1 decreased approximately 50% from that of VC 1 , labeling of IIC 2 effectively doubled, and resulted in 32 P-incorporation into VC 1 being only 25% that into IIC 2 . As expected, 2Ј,5Ј-dd-3Ј-ATP completely blocked 32 P-incorporation into VC 1 and IIC 2 , whether alone or in the VC 1 ⅐IIC 2 complex (Figs. 6 and 7, lanes 3, 6, and 9), whereas there was marginal protection afforded by either 5Ј-ATP (Fig. 6) or 5Ј-AP(CH 2 )PP (Fig. 7). The concentrations of 5Ј-ATP or 5Ј-AP(CH 2 )PP used were 100-fold greater than that of ␥-MetIm-2Ј,5Ј-dd-3Ј-ATP, but afforded protection of VC 1 and IIC 2 only 10 -30% (Table I). Protection afforded by 5Ј-AP(CH 2 )PP was less effective with IIC 2 (ϳ10%) than with VC 1 (ϳ30%), regardless of VC 1 ⅐IIC 2 complex formation. In contrast, the protective effect of 5Ј-ATP depended on subdomain association. 5Ј-ATP caused ϳ30% reduction in labeling of the individual VC 1 and IIC 2 proteins, whereas formation of the VC 1 ⅐IIC 2 complex resulted in enhanced protection of VC 1 protection (ϳ33% with VC 1 to ϳ75% for VC 1 in the VC 1 ⅐IIC 2 complex) and of IIC 2 (ϳ27% with IIC 2 to ϳ60% in the VC 1 ⅐IIC 2 complex). Considering that 50% of the 5Ј-ATP was converted to cAMP and PP i during incubation with the enzyme, it is possible that the protection afforded by 5Ј-ATP was actually due to the products of the enzyme reaction.
The labeling data are in agreement with those obtained by direct photo cross-linking of VC 1 and IIC 2 with [ 32 P]2Ј-d-3Ј-AMP (11) and corroborate the conclusion that P-site ligands can covalently cross-link with the cytosolic domains of adenylyl cyclase, whether alone or in complex. However, the asymmetric labeling of the subunits noted here (Figs. 6 and 7) was not observed in the earlier studies (11). This difference likely lies in the fact that the earlier experiments used the nucleoside 3Јmonophosphate, whereas these used the nucleoside 3Ј-triphosphate. First, the polyphosphate moiety may measurably influence protein-protein interactions between VC 1 and IIC 2 ; second, the site at which VC 1 is labeled in the vicinity of the ligand ␥-phosphate may be restricted or reconfigured by formation of the VC 1 ⅐IIC 2 complex, whereas the site at which IIC 2 is labeled becomes more accessible. In contrast, the adenine moiety, through which direct photo cross-linking occurs, would contact both VC 1 and IIC 2 . Moreover, it is within a domain that is shared by P-site ligand and substrate and is not significantly altered by formation of the VC 1 ⅐IIC 2 complex (7).

DISCUSSION
The data presented here demonstrate that the interaction of 2Ј,5Ј-dd-3Ј-ATP with the adenylyl cyclase formed from the association of the VC 1 and IIC 2 cytosolic domains is a process that is not in rapid equilibrium. This would explain the lack of competition between 2Ј,5Ј-dd-3Ј-ATP and substrate under steady state kinetic conditions. Moreover, it suggests that binding of this ligand is a pseudo-irreversible process and the noncompetitive inhibition observed with it (12,22) likely occurs by a somewhat different mechanism than does the post-transition state, dead-end inhibition previously observed with 2Ј-d-3Ј-AMP, which requires bound metal-PP i for inhibition (5,8,11). The lack of competition between substrate and 2Ј,5Ј-dd-3Ј-ATP was clearly evident both in the experiments on associationdissociation of VC 1 and IIC 2 (Figs. 2 and 3) and in the experiments involving cross-linking and labeling with ␥-MetIm-2Ј,5Јdd-[␥-32 P]3Ј-ATP (Figs. 5-7). Also evident was the importance of the order of addition of substrate or protecting ligand and the nucleoside 3Ј-triphosphate. Rates of dissociation of the preformed VC 1 ⅐IIC 2 complex differed depending on the nucleotides present and the order of their addition (Figs. 2 and 3). Furthermore, no protection against covalent labeling of VC 1 or IIC 2 by ␥-MetIm-2Ј,5Ј-dd-[␥-32 P]3Ј-ATP was afforded by 5Ј-ATP, 5Ј-AP(CH 2 )PP, or 2Ј,5Ј-dd-3Ј-ATP when the proteins were added to a mixture of the covalent affinity ligand and protecting ligand. This suggested either that ␥-MetIm-2Ј,5Ј-dd-3Ј-ATP binds with the adenylyl cyclase subdomains faster than the other nucleotides or that it cross-links with a different site. The latter possibility was excluded by the complete protection observed when the VC 1 , IIC 2 , or the VC 1 ⅐IIC 2 complex were pretreated with 2Ј,5Ј-dd-3Ј-ATP before addition of the covalent affinity ligand (Figs. 6 and 7). The observation that the protective effect of 5Ј-ATP was substantially better with VC 1 ⅐IIC 2 than with VC 1 or IIC 2 individually could be due either to the presence of products (cAMP and PP i ) or to an effect on the assembly of the VC 1 ⅐IIC 2 complex. This latter process may be promoted by substrate binding and result in stabilization of a "closed" configuration in which IIC 2 effectively blocks polyphosphate binding to VC 1 , as contrasted with an "open" configuration in which VC 1 is more accessible to ␥-MetIm-2Ј,5Ј-dd-[␥-32 P]3Ј-ATP. In general, these observations corroborate those we made earlier (11) by direct photo cross-linking of [ 32 P]2Ј-d-3Ј-AMP with VC 1 , IIC 2 , and the VC 1 ⅐IIC 2 complex. Each cytosolic domain was observed to react individually with [ 32 P]2Ј-d-3Ј-AMP, but complex formation also affected the interaction of each domain with it (11). This was consistent with the enzyme being distributed between two conformational states that are not in rapid equilibrium, one binding 5Ј-ATP and the other binding 2Ј-d-3Ј-AMP and PP i .
No structural data are available for adenylyl cyclases in complex with any adenine nucleoside 3Ј-triphosphate. The data presented here show that both VC 1 and IIC 2 domains interact with 2Ј,5Ј-dd-3Ј-ATP. This interaction affects the association of the major cytosolic domains and the resulting conformation of the enzyme; the formation of a functionally active VC 1 ⅐IIC 2 complex is prevented. From comparisons with structural data obtained with 2Ј-d-3Ј-AMP and PP i (7) and with analogs of 5Ј-ATP (22), it would be expected that the polyphosphate group of 2Ј,5Ј-dd-3Ј-ATP would interact with both VC 1 and IIC 2 domains. The terminal ␤-and ␥-phosphates should interact with the C 1 domain through divalent cation and with the C 2 domain through Arg-1029 and/or Lys-1065 (7,22), amino acids susceptible to covalent interaction with ␥-MetIm-2Ј,5Ј-dd-3Ј-ATP. Labeling of VC 1 and IIC 2 domains by the 32 P-labeled derivative occurred with similar efficiency, but after stimulation by Mn 2ϩ and forskolin and the formation of VC 1 ⅐IIC 2 , the IIC 2 domain was labeled preferentially. Furthermore, treatment of VC 1 with 2Ј,5Ј-dd-3Ј-ATP prevented formation of catalytically competent VC 1 ⅐IIC 2 . If one assumes that the dissociated open state of VC 1 and IIC 2 accurately models the inactive state of native holoenzyme, binding of 2Ј,5Ј-dd-3Ј-ATP would prevent the conformational changes promoted by enzyme interactions with activated G s ␣ and/or forskolin. The 2Ј,5Ј-dd-3Ј-ATP would hold the enzyme in an inactive configuration. This is conceptually different from conclusions drawn from earlier experiments with 2Ј-d-3Ј-AMP and 2Ј,5Ј-dd-adenosine, inhibition by either of which relies on product PP i being in the catalytic site. That is, inhibition was preferentially of stimulated enzyme. The requirement of enzyme activity to generate the pyrophosphate necessary for inhibition is obviated by 2Ј-d-3Ј-ATP or 2Ј,5Ј-dd-3Ј-ATP, because these ligands by their nature already contain the PP i group and can inhibit basal as well as stimulated forms of the enzyme.