Inhibition by Calcium of Mammalian Adenylyl Cyclases*

Ca2+ regulates mammalian adenylyl cyclases in a type-specific manner. Stimulatory regulation is moderately well understood. By contrast, even the concentration range over which Ca2+ inhibits adenylyl cyclases AC5 and AC6 is not unambiguously defined; even less so is the mechanism of inhibition. In the present study, we compared the regulation of Ca2+-stimulable and Ca2+-inhibitable adenylyl cyclases expressed in Sf9 cells with tissues that predominantly express these activities in the mouse brain. Soluble forms of AC5 containing either intact or truncated major cytosolic domains were also examined. All adenylyl cyclases, except AC2 and the soluble forms of AC5, displayed biphasic Ca2+ responses, suggesting the presence of two Ca2+ sites of high (∼0.2 μm) and low affinity (∼0.1 mm). With a high affinity, Ca2+ (i) stimulated AC1 and cerebellar adenylyl cyclases, (ii) inhibited AC6 and striatal adenylyl cyclase, and (iii) was without effect on AC2. With a low affinity, Ca2+inhibited all adenylyl cyclases, including AC1, AC2, AC6, and both soluble forms of AC5. The mechanism of both high and low affinity inhibition was revealed to be competition for a stimulatory Mg2+ site(s). A remarkable selectivity for Ca2+was displayed by the high affinity site, with a K i value of ∼0.2 μm, in the face of a 5000-fold excess of Mg2+. The present results show that high and low affinity inhibition by Ca2+ can be clearly distinguished and that the inhibition occurs type-specifically in discrete adenylyl cyclases. Distinction between these sites is essential, or quite spurious inferences may be drawn on the nature or location of high affinity binding sites in the Ca2+-inhibitable adenylyl cyclases.

Profound physiological significance derives from the regulation of adenylyl cyclase by Ca 2ϩ , which provides a confluence of two major signaling pathways. For instance, Ca 2ϩ stimulation of adenylyl cyclase has been implicated in learning-memory functions (1)(2)(3). Compelling evidence for this assertion comes from studies with Aplysia (4) and with Drosophila mutants (5) and more recently mice that have had adenylyl cyclase genes deleted (6). On the other hand, Ca 2ϩ inhibition of adenylyl cyclase has been proposed to contribute to oscillations and/or pacemaking in cardiac tissue (7) and the maintenance of endothelial cell permeability (8). A central issue in implicating the Ca 2ϩ responsiveness of an adenylyl cyclase in a physiological process concerns the correspondence between the concentra-tions of Ca 2ϩ achieved in response to physiological stimuli versus those required to regulate these adenylyl cyclases in vitro. 1 This issue seems moot with Ca 2ϩ -stimulable adenylyl cyclases, which are stimulated by Ca 2ϩ in vitro and are also stimulated in response to various physiological means of elevating Ca 2ϩ in vivo (2,7,9). Such correlations are mutually supportive. The situation with Ca 2ϩ -inhibitable adenylyl cyclases is more complex. The averaged cytosolic concentration of Ca 2ϩ achieved in intact cells upon triggering either capacitative-or voltage-gated Ca 2ϩ entry (ϳ1 M) corresponds to the concentrations reported to be effective at inhibiting these adenylyl cyclases in some in vitro studies (10). However, a wide array of in vitro inhibitory sensitivities have been reported from various sources (10 -25). For instance, both monophasic (11-13, 21, 23) and biphasic (10, 14 -19, 24) inhibition by Ca 2ϩ have been reported, spanning sub-to supramicromolar concentration ranges.
Some of the discrepancies may emanate from somewhat uncontrolled assay conditions, e.g. with respect to pH or free EGTA concentrations, which will confound estimates of free Ca 2ϩ (26). However, mixed populations of adenylyl cyclases in selected tissues may also confound analysis. The nine isoforms of mammalian adenylyl cyclases that have been cloned to date (7,27) possess distinct characteristics including their sensitivity to Ca 2ϩ , which allows them to be classified as (i) Ca 2ϩstimulated (AC1, AC8, and, possibly, AC3), (ii) Ca 2ϩ -insensitive (AC2, AC4, and AC7), and (iii) Ca 2ϩ -inhibited (AC5 and AC6; Refs. 7, 28, and 29). Heterogeneous preparations of such adenylyl cyclases could give rise to a variety of Ca 2ϩ concentration responses. There is also the issue of low affinity Ca 2ϩ inhibition, which appears to be a property of all adenylyl cyclases. This inhibition is believed to reflect competition by Ca 2ϩ for an allosteric regulatory site for Mg 2ϩ (16, 17, 20 -22), although the precise mechanism is unknown. This low affinity inhibition by Ca 2ϩ can be confounded with high affinity inhibition, particularly if Ca 2ϩ concentrations are not rigorously established or controlled (26).
Ideally, well controlled, direct comparisons of expressed adenylyl cyclase isoforms should resolve these ambiguities. Indeed, in membranes from transfected HEK 293 cells, AC1 and AC8 were stimulated by submicromolar concentrations of Ca 2ϩ , while they displayed inhibition in response to higher concentrations, mirroring what happens in most brain tissues (9). However, again, the situation with the Ca 2ϩ -inhibitable adenylyl cyclases is less clear. A monophasic inhibition by Ca 2ϩ spanning sub-to supramicromolar concentrations was associated with the first description of canine AC5 (25). A second report on rabbit AC5 indicated an inhibition by very low submicromolar (suprananomolar) Ca 2ϩ concentrations (30). In the first description of AC6, its activity in transfected cell mem-branes was inhibited by submicromolar concentrations of Ca 2ϩ (31), although higher concentrations were not explored. Recently, a chimeric, soluble form of AC5, comprising its fused cytosolic (C1 and C2) domains was reported to display high affinity inhibition by Ca 2ϩ , which was lost upon truncation of the C1 region (32). Furthermore, AC2, which is classified as a Ca 2ϩ -insensitive adenylyl cyclase, displays inhibition in response to high concentrations of Ca 2ϩ . It therefore seems essential to clearly define the effective concentration ranges for Ca 2ϩ inhibition of individual adenylyl cyclases.
Consequently, in the present study, we compared the effects of a comprehensive range of Ca 2ϩ concentrations on adenylyl cyclase activity from a variety of sources. Adenylyl cyclase activity in membranes of Sf9 cells expressing AC1, AC2, and AC6 was compared with adenylyl cyclase activity of two selected brain tissues that express respectively AC5 (the striatum; Refs. 33 and 34) and AC1 (the cerebellum; Ref. 35). Also, the full-length and truncated forms of the AC5 C1/C2 chimera were assessed. Unequivocal evidence is provided demonstrating that, with a high affinity, Ca 2ϩ inhibits AC6 (and striatal AC5), stimulates AC1 (and cerebellar AC1), and does not modulate AC2. In the supramicromolar concentration range, all adenylyl cyclases tested showed the same low affinity inhibition by Ca 2ϩ . Further analyses reveal that both the selective high affinity inhibition occurring in AC6 (and AC5) and the low affinity inhibition occurring in all adenylyl cyclase isoforms involve competitive mechanisms with Mg 2ϩ activation of the enzyme. As for the two soluble forms of AC5, our results show that both chimeras display only low affinity inhibition and that the site of high (and low) affinity inhibition must be sought by other experimental approaches.  (36,37). For construction of recombinant baculovirus encoding AC6, the entire coding region of mouse AC6 was engineered by polymerase chain reaction into the baculovirus vector, pBlueBacHis2, between KpnI and SalI restriction sites. The culture of Sf9 cells and the production, cloning, and amplification of recombinant baculovirus were performed according to the method of Summers and Smith (38). Sf9 cells were usually infected with 1 plaque-forming unit/cell of baculovirus and were harvested 48 h after infection.

Materials-Forskolin
Preparation of Membranes from Sf9 Cells-Membranes were prepared from Sf9 cells expressing individual isoforms of adenylyl cyclase, as described previously (39). The cell suspensions were centrifuged, washed with Phillip's buffer containing protease inhibitors (20 g/ml soybean trypsin inhibitor, 4 g/ml leupeptin, 12 units/ml kallikrein inactivator, 4 g/ml antipain, 52.4 g/ml benzamidine, 52.3 g/ml phenylmethylsulfonyl fluoride, and 2 g/ml pepstatin A). Following centrifugation and subsequent lysis of cells in hypo-osmotic buffer containing protease inhibitors, samples were sheared mechanically by homogenization with a Wheaton Dounce homogenizer with 25 strokes and passage through a 22-gauge needle 10 times. The lysates were centrifuged at 270 ϫ g for 10 min. The crude membranes were prepared by pelleting this supernatant at 20,000 ϫ g. Purified membranes were prepared by fractionation of the lysates using a continuous gradient of 5-50% sucrose in lysis buffer. The plasma membranes banding between 30 and 42% sucrose were removed and washed. The membranes were resuspended in buffer containing 40 mM Tris, pH 7.4, 800 M EGTA, and 0.25% bovine serum albumin, at a final protein concentration of 0.5-2.5 mg/ml (as determined by the method of Lowry et al. (40) using bovine serum albumin as a standard). The preparations were stored in liquid nitrogen.
Preparation of Soluble Forms of AC5-The plasmids encoding the soluble forms of AC5 comprising either the C1 or C1a domain linked to the C2 domain (VC1C2 and VC1aC2, respectively) were kindly provided by Dr. T. B. Patel (University of Kentucky, Lexington; Ref. 32). The proteins were expressed in the Escherichia coli strain TP2000, which is incapable of producing cAMP (42). The expression of VC1C2 and VC1aC2, cell lysis and assays were performed as described previously (32).
Preparation of Mouse Cerebellum and Striatum Membranes-Male mice of the C57bl/6 strain were decapitated. The brain was removed and placed on dry ice. The cerebellum was detached and homogenized immediately. A transverse section of the forebrain was made at the level of the optic chiasma, and the dorsal striatum were dissected based on the method described by Glowinski and Iversen (43). The tissues were homogenized and centrifuged at 1000 ϫ g for 10 min in 20 volumes of a cold Tris-sucrose (50 mM, 10%) buffer, pH 7.4, containing 0.8 mM EGTA and a mixture of protease inhibitors as described above. The supernatant was centrifuged at 15,000 ϫ g for 10 min, and the resulting pellet was washed and centrifuged three times before final suspension in a volume of Tris buffer (50 mM) sufficient to give a concentration of proteins in the range of 1 mg/ml as determined by the method of Lowry (40). Aliquots were immediately frozen in liquid nitrogen until use. Before use for the adenylyl cyclase assay, the samples were further washed and centrifuged two times in 2 ml of a Hepes buffer (50 mM), pH 7.4, containing 240 M EGTA and 0.25% bovine serum albumin.
Adenylyl Cyclase Activity Measurements-The adenylyl cyclase activity of infected Sf9 cells or mouse brain tissues (striatum and cerebellum) was measured in the presence of the following components: 12 mM phosphocreatine, 2.5 units of creatine phosphokinase, 0.1 mM cAMP, 1 mM MgCl 2 , 0.1 mM ATP, 0.04 mM GTP, 0.5 mM isobutylmethylxanthine, 70 mM HEPES buffer, pH 7.4, and 1 Ci of [␣-32 P]ATP. (MgCl 2 and ATP concentrations were varied in some experiments, as indicated). Because we were inhibiting adenylyl cyclase activity to low levels (at high [Ca 2ϩ ], particularly at low [Mg 2ϩ ]), maximizing these values was desirable to generate robust kinetic parameters; consequently, basal activity was enhanced with forskolin for all sources except cerebellum and striatum, which displayed elevated basal activities. Forskolin stimulation was employed in some experiments from these latter sources and yielded unchanged parameters (not shown). Calmodulin (1 M) was included for the assay of cells expressing AC1 and cerebellar membranes. Free Ca 2ϩ concentrations were established from a series of CaCl 2 solutions buffered with 60 M EGTA in the assay and were calculated as described previously in detail (44). The reaction mixture (final volume, 100 l) was incubated at 30°C for 30 min. Reactions were terminated with sodium lauryl sulfate (0.5%) containing 1.5 mM cAMP and 22 mM ATP. [ 3 H]cAMP (ϳ10,000 cpm) was added as a recovery marker, and the [ 32 P]cAMP formed was quantified as described previously (45). Data points are presented as mean activities of triplicate determinations (error bars are indicated when they are larger than the data point symbols in Figs. [1][2][3][4][5]. For each assay, at least two adenylyl cyclases were compared, and a cerebellum Ca 2ϩ doseresponse curve was determined as an internal control to monitor the accuracy of the Ca 2ϩ buffering system. Assays for Effect of Ca 2ϩ on Mg 2ϩ or MgATP 2Ϫ Requirement-These assays were conducted as described above except that concentrations of Mg 2ϩ or ATP varied. Free Mg 2ϩ concentrations were established from a series of MgCl 2 solutions buffered with 60 M EGTA and were calculated as described previously (44). In order to investigate the respective effects of occupancy of high and low affinity regulatory sites for Ca 2ϩ , two selected concentrations of Ca 2ϩ (2.75 and 86 M) were included in these assays. After calculation, the concentrations of free Ca 2ϩ actually ranged from 2.59 to 3.43 M and from 81.6 to 90.1 M, respectively, in the presence of the lowest and the highest concentration of Mg 2ϩ or ATP.
Statistics-Nonlinear regression was used to fit a competition curve with either one or two components to the data (Graphpad, Inplot4). Goodness of fit was quantified by the least-squares method (F-test) and was deemed better suited to one model when p was Ͻ0.05.
Paired Student's t tests (two-tailed) were performed to compare the means of K m or K a values determined with different concentrations of Ca 2ϩ .

Species-specific Effects of Ca 2ϩ on Adenylyl Cyclase Activity-
The effects of a broad range of Ca 2ϩ concentrations on adenylyl cyclase activity were compared in membranes from Sf9 cells expressing AC6 and striatum (which predominantly expresses AC5; Ref. 33). Clearly, Ca 2ϩ elicits a biphasic decline in activity with increasing concentration (Fig. 1). In both cases, curvefitting analyses revealed that the inhibition curves always fitted significantly better to a two-component model (p Ͻ 0.05), for which the effective concentrations giving 50% of the response (EC 50 ) were in the ranges of submicromolar concentrations and submillimolar concentrations, respectively (Table I). By contrast, the effects of Ca 2ϩ on the adenylyl cyclase activity in Sf9 cells expressing AC1 and in cerebellum membranes were bidirectional. Stimulation occurred within the range of submicromolar concentrations of Ca 2ϩ , whereas inhibition occurred at submillimolar concentrations (Fig. 1). Curve fitting of these data indicated that stimulation occurred with a K a in the submicromolar range and inhibition in the supramicromolar range (Table I). In Sf9 cells expressing AC2, Ca 2ϩ also inhibited adenylyl cyclase activity (Fig. 1). However, curve-fitting analyses of these data revealed only one component, for which the EC 50 was in the low affinity (submillimolar) concentration range (Table I).
A recent report suggested that the C1b region of AC5 might mediate high affinity inhibition by Ca 2ϩ (32). Consequently, the Ca 2ϩ sensitivity of the full-length and truncated soluble forms of AC5 (VC1C2 and VC1aC2, respectively) were examined. In both cases, Ca 2ϩ produced an inhibition similar to that measured in the cells expressing AC2 (Fig. 1). Curve fitting analyses showed that, as in material expressing AC2, the inhibition was monophasic and, as with AC2, the EC 50 was in the low affinity concentration range.
Effect of Ca 2ϩ on Mg 2ϩ Activation of Adenylyl Cyclase-For a considerable period, it has been believed that low affinity inhibition of adenylyl cyclase activity by Ca 2ϩ reflected competition for a Mg 2ϩ activation site (11, 16 -18, 20 -22). However, this issue has not been addressed in the light of current information on adenylyl cyclase diversity. Consequently, the effects of occupancy of high affinity and low affinity Ca 2ϩ inhibition sites on the activation by Mg 2ϩ of various adenylyl cyclases were investigated . The effects of two concentrations of free Ca 2ϩ , corresponding to the plateau region of the first inhibitory phase (2.75 M; cf. Fig. 1) and within the second phase (86 M; cf. Fig.  1) of the Ca 2ϩ inhibition curves, on the activation by Mg 2ϩ of AC1, AC2, AC6, and striatal adenylyl cyclase were determined. Concentration-activity curves were generated using varying concentrations of Mg 2ϩ (Fig. 2). The results showed that Mg 2ϩ increased adenylyl cyclase activity concentration-dependently, as expected. The two selected concentrations of Ca 2ϩ inhibited the Mg 2ϩ stimulation in membranes expressing AC6. Similar effects were observed in striatal membranes. In contrast, the stimulation by Mg 2ϩ was inhibited only by the high Ca 2ϩ concentration (86 M) in membranes expressing AC2. Similar assays were also performed with AC1 membranes. Only the effect of the high concentration of Ca 2ϩ was investigated (in the absence of calmodulin), since at low concentrations AC1 would be stimulated by Ca 2ϩ . Indeed, in this condition, Ca 2ϩ also reduced the activation produced by Mg 2ϩ in AC1. Reciprocal plots of velocity versus Mg 2ϩ concentration are shown in Fig. 3. The linear regression analyses performed on these data indicate that, in AC6 and striatal membranes, both 2.75 and 86 M Ca 2ϩ inhibited Mg 2ϩ activation competitively; i.e. the intercept (1/V max ) was unchanged by Ca 2ϩ , whereas the slope (K a /V max ) increased (Fig. 3, Table II). The inhibition detected in AC2 was also competitive at the high concentration of Ca 2ϩ (Fig. 3, Table  II). Finally, Ca 2ϩ also demonstrated competitive inhibition at 86 M Ca 2ϩ in AC1 membranes. The mean K a for Mg 2ϩ was about 1 mM in all of the adenylyl cyclases examined in the absence of Ca 2ϩ (Table II). However, in the presence of 2.75 M Ca 2ϩ , this value was significantly increased (by about 2-fold), with Ca 2ϩ -inhibitable adenylyl cyclases. This increase reached 6 -8-fold and occurred in all adenylyl cyclase isoforms when 86 M Ca 2ϩ was included in the assays (Table II).
Effects of Ca 2ϩ on the MgATP 2Ϫ Requirement in AC6, AC2, AC1, or Striatum-Since Mg 2ϩ not only activates adenylyl cyclase, but also participates in the formation of the cyclase substrate, MgATP 2Ϫ , we investigated the effects of inhibition by Ca 2ϩ (3.41 and 89 M) on the relationship between activity and substrate (MgATP 2Ϫ ) concentration. In these experiments, the concentration of free Mg 2ϩ was held constant (10 mM) so that it could be determined whether the antagonistic effect of Ca 2ϩ on Mg 2ϩ activation reported in Fig. 2 (and Table I Fig. 5. Linear regression analyses of these data indicated that the low concentration of Ca 2ϩ produced a noncompetitive inhibition on the substrate MgATP 2Ϫ activation for both AC6 and striatum. These plots were largely parallel, although there was also a slight indication of a K m effect. Finally, noncompetitive patterns of inhibition were revealed in all of the adenylyl cyclase preparations examined when the high concentration of Ca 2ϩ was included in the assays. As indicated in Table II, in all cases, the mean K m value was significantly decreased (p Ͻ 0.05) with respect to controls in the presence of  89 M Ca 2ϩ , although the effect on V max was most prominent.
Overall, these analyses are consistent with the predominant effect of Ca 2ϩ being to antagonize stimulation by Mg 2ϩ , along with a minor noncompetitive effect on substrate utilization.

DISCUSSION
The stimulation of adenylyl cyclases AC1 and AC8 by Ca 2ϩ is understood to occur in the submicromolar concentration range and to be mediated by calmodulin (9, 46 -48). In addition, the calmodulin binding sites are known for both AC1 and AC8 (49,50). By contrast, even the concentration ranges over which Ca 2ϩ inhibits adenylyl cyclases are not unambiguously understood; even less so is the mechanism of inhibition. There have been a variety of studies reporting monophasic inhibition (11-13, 21, 23) as well as some showing evidence for two distinct binding sites of low and high affinity for Ca 2ϩ , giving rise to biphasic inhibition (14 -19). When AC5 and AC6 were identified, they were characterized as being Ca 2ϩ -inhibited adenylyl cyclase isoforms (25,31). When expressed in intact cells, these cyclases are inhibited by capacitative Ca 2ϩ entry (51). The mRNAs of these species are also expressed in tissues, such as the heart, in which the adenylyl cyclase activity is inhibited by Ca 2ϩ (52,53). To resolve the confusion that attends Ca 2ϩ inhibition of adenylyl cyclase, examination of material expressing a single isoform would seem to be a rational procedure to understand the Ca 2ϩ regulation of adenylyl cyclase. Consequently, in the present study we first compared the regulation of a presumably Ca 2ϩ -stimulable and a Ca 2ϩ -inhibitable adenylyl cyclase expressed in Sf9 cells with tissues that predominantly express these activities in the mouse brain. It was gratifying to note that the response to Ca 2ϩ of Sf9 cell membranes expressing AC1 and AC6 was virtually identical to the response of cerebellar and striatal membranes, respectively, which are a major source of the analogous adenylyl cyclase mRNAs. Such findings indicate that no co-factor or post-translational modification is lacking between moths and mice to alter adenylyl cyclase responsiveness. Consequently, it seems reasonable to characterize further the Ca 2ϩ -regulation in membranes from either source.
The present studies clearly indicate that Ca 2ϩ inhibits AC6 (and AC5) over precisely the same concentration range as it stimulates AC1. Specifically, our results demonstrate clearly that, in both adenylyl cyclases, the effects of Ca 2ϩ progress in a biphasic manner, strongly suggesting interactions with two distinct regulatory binding sites of low and high affinity. Inter- action with sites of high affinity for Ca 2ϩ (ϳ0.15 M) confer opposite regulation in AC6 (inhibition) and AC1 (stimulation), whereas interaction with sites of low affinity (ϳ0.06 mM) inhibits both isoforms. The effects of Ca 2ϩ on striatal adenylyl cyclase activity were identical to those detected in AC6. Previous reports had found that the basal adenylyl cyclase activity in the striatum was higher but less stimulated by Ca 2ϩ , or insensitive to Ca 2ϩ , as compared with other brain areas, in the presence of calmodulin (54). More recently, the adenylyl cyclase activity of the rat striatum was shown to exhibit a daily oscillation with a peak occurring around 10:00 -12:00 a.m., during which Ca 2ϩ inhibition was selectively manifest (55). Accordingly, we dissected striatum during this critical phase, and in agreement with the latter findings, our data showed that Ca 2ϩ potently inhibits the striatal adenylyl cyclase activity. Finally, comparisons with AC2 demonstrated that, in sharp contrast, this adenylyl cyclase was insensitive to low concentrations of Ca 2ϩ . Nevertheless, like the other adenylyl cyclases examined here, AC2 was inhibited when Ca 2ϩ reached submillimolar levels. Together, these findings demonstrate that the various isoforms of adenylyl cyclases behave differently when Ca 2ϩ interacts at high affinity binding sites, whereas they respond similarly when Ca 2ϩ binds low affinity sites.
Next, we compared the mechanisms that yielded high affinity inhibition in AC6 and striatal adenylyl cyclase, compared with the low affinity inhibition that is common to all adenylyl cyclases. Previous studies of adenylyl cyclase activity in various tissues had proposed that high concentrations of Ca 2ϩ reduced activity depending on the substrate concentration in a noncompetitive manner (11,12,16,17,21,56). When competition between Ca 2ϩ and Mg 2ϩ was considered, more contradictions were encountered, even between studies carried out on the same tissues. Specifically, some studies showed that Ca 2ϩ inhibition of adenylyl cyclase occurred noncompetitively with Mg 2ϩ (11,19,23), while others found that high concentrations of Ca 2ϩ compete with Mg 2ϩ , whereas lower concentrations do not (56). In other studies, it was proposed that Ca 2ϩ inhibits adenylyl cyclase by competing with Mg 2ϩ and suggested that a common metal site bound both ions (16, 17, 20 -22). Here, we show that high affinity inhibition by Ca 2ϩ of either AC6 or striatal adenylyl cyclase does not involve competition with the MgATP 2Ϫ substrate. Instead, high affinity inhibition reflects competition by low concentrations of Ca 2ϩ for the Mg 2ϩ activation in both AC6 and striatal adenylyl cyclase, so that the affinity for Mg 2ϩ is reduced selectively in these isoforms. On the other hand, low affinity inhibition produced by Ca 2ϩ in AC6, AC1, and AC2, as well as in the striatal adenylyl cyclase, also stems largely from a competitive action on the Mg 2ϩ activation of the enzyme, with a minor effect on substrate utilization.
Recent deletion analyses have suggested the existence of a Mg 2ϩ binding site essential for catalysis and distinct from the ATP-bound cation (57). Along with insight gained from crystallographic studies (58), the latter findings led these authors to conclude that one conserved aspartate residue (corresponding to Asp-396 in the C1a region of canine AC5) plays a critical role in coordinating catalytic Mg 2ϩ ions. Specifically, this Mg 2ϩ was predicted to facilitate the nucleophilic attack of the 3Ј-hydroxyl group of ATP on the ␣-phosphate. It is conceivable that Ca 2ϩ competes with Mg 2ϩ at this site, which is largely conserved among all adenylyl cyclases, to yield low affinity inhibition. By contrast, the site of high affinity inhibition by Ca 2ϩ of AC5 and AC6 would be expected to be mediated by a region unique to AC5 and AC6.
In the latter context, Scholich et al. (32) reported recently that Ca 2ϩ inhibited a soluble form of AC5, composed of the C1 and C2 regions of the enzyme, but not a shorter form lacking the unconserved C1b domain. These findings were taken to suggest that the 112-amino acid C1b region in AC5 mediated high affinity Ca 2ϩ inhibition of the enzyme activity. In agreement with this study, we also found that Ca 2ϩ would inhibit the C1C2 protein. However, since we were conducting our experiments in parallel with Ca 2ϩ -stimulated cerebellar adenylyl cyclase, as an internal control of estimated Ca 2ϩ concentrations, it was clear that the inhibition occurred with submillimolar concentrations of Ca 2ϩ , not with submicromolar concentrations. Furthermore, in our experiments, the truncated form of C1C2, namely the C1aC2 construct, displayed the same low affinity inhibition by Ca 2ϩ . Based on these results, it seems unlikely that the Ca 2ϩ site responsible for high affinity inhibition resides within the cytosolic domains of AC5. However, it is possible that the soluble chimera expressed in Escherichia coli does not adopt the structural configuration of the native ACV. Alternatively, the absence of other domains might explain the absence of high affinity inhibition in the VC1C2 protein. In any case, it is reasonable to conclude that further studies are needed to clarify the structural features conferring Ca 2ϩ inhibition on adenylyl cyclases.
The present study has clearly separated the high and low affinity effects of Ca 2ϩ . With a high affinity, Ca 2ϩ stimulates AC1 and cerebellar membrane adenylyl cyclase in a manner that requires dissociable calmodulin. Again, with a high affinity, Ca 2ϩ inhibits AC6 expressed in Sf9 cells and striatal adenylyl cyclase independently of added calmodulin. These low concentrations of Ca 2ϩ do not affect AC2 expressed in Sf9 cells. With a low affinity, Ca 2ϩ inhibits all adenylyl cyclases, including AC1, AC2, AC6, striatal AC5, and soluble forms of AC5, intact or with the C1b region deleted. Kinetically, the mechanism of both high and low inhibition by Ca 2ϩ appears similar and appears to involve competitive inhibition for a stimulatory Mg 2ϩ site(s). The inhibition does not involve competitive mechanisms for the substrate MgATP 2Ϫ . A K i value of ϳ0.2 M for the high affinity inhibitory site, in the face of a 5000-fold excess of Mg 2ϩ , suggests a remarkable selectivity for Ca 2ϩ . This selectivity for Ca 2ϩ over Mg 2ϩ is analogous with the specificity of E-F-hand-containing proteins for Ca 2ϩ (59). However, the nature or location of this site remains elusive. No E-F hand or other obvious Ca 2ϩ -binding motif is present in AC5 or AC6 (31). In addition, earlier biochemical studies appeared to eliminate the participation of calmodulin in the inhibition; for instance, neither calmodulin antagonists nor EGTA washing modulated the effect (60), although this does not exclude the possibility that a tightly bound calmodulin is involved (as with phosphorylase kinase). On the other hand, the low affinity site appears to represent a more conventional competition between Ca 2ϩ and Mg 2ϩ , which, as with a number of metabolic enzymes, shows an approximately 10-fold difference in effective concentrations of the two cations (62,63). Given that averaged, cytosolic concentrations of Ca 2ϩ rarely reach 50 M, except transiently around the mouths of ion channels, it is difficult to imagine that this inhibition is of any physiological utility. On the other hand, it is important to take pains to distinguish this inhibition from high affinity inhibition, or quite spurious inferences may be drawn on the nature or location of high affinity binding sites.