Zinc Inhibition of cAMP Signaling*

Zn2+ is required as either a catalytic or structural component for a large number of enzymes and thus contributes to a variety of important biological processes. We report here that low micromolar concentrations of Zn2+ inhibited hormone- or forskolin-stimulated cAMP production in N18TG2 neuroblastoma cells. Similarly, low concentrations inhibited hormone- and forskolin-stimulated adenylyl cyclase (AC) activity in membrane preparations and did so primarily by altering theV max of the enzyme. Zn2+ also inhibited recombinant isoforms, indicating that this reflects a direct interaction with the enzyme. The IC50 for Zn2+inhibition was ∼1–2 μm with a Hill coefficient of 1.33. The dose-response curve for Zn2+ inhibition was identical for AC1, AC5, and AC6 as well as for the C441R mutant of AC5 whose defect appears to be in one of the catalytic metal binding sites. However, AC2 displayed a distinct dose-response curve. These data in combination with the findings that Zn2+ inhibition was not competitive with Mg2+ or Mg2+/ATP suggest that the inhibitory Zn2+ binding site is distinct from the metal binding sites involved in catalysis. The prestimulated enzyme was found to be less susceptible to Zn2+ inhibition, suggesting that the ability of Zn2+ to inhibit AC could be significantly influenced by the coincidence timing of the input signals to the enzyme.

In general, Zn 2ϩ is required as either a catalytic or structural component for a large number of enzymes and thus contributes to a wide variety of important biological processes including gene expression, replication, hormonal storage and release, neurotransmission, and memory. Zn 2ϩ is also critical for the structural integrity of cells, influencing membrane stability and cytoskeletal organization (reviewed in Refs. [1][2][3]. In this light, it is not surprising that dietary Zn 2ϩ deprivation is lethal in mice and in man has been associated with a variety of abnormalities related to growth, sexual maturation, and wound healing (4). The complexity of Zn 2ϩ actions is further appreciated by the fact that the consequences of excess Zn 2ϩ are also severe being associated with a number of neurodegenerative disorders including Parkinson's and Alzheimer's disease and epilepsy (1)(2)(3). Toward understanding the paradoxical functions of Zn 2ϩ , attention has focused on defining the critical candidates that mediate its effects and whether they can reflect the physiological, pharmacological, or toxicological effects of the metal.
In the brain, Zn 2ϩ along with iron is the most concentrated metal (1)(2)(3)(4). Significant levels of chelatable histochemically reactive Zn 2ϩ are present in a subset of glutamatergic neurons in which Zn 2ϩ appears to be localized to synaptic vesicles (5)(6)(7). These Zn 2ϩ -containing neurons are primarily located in the hippocampus (mossy fibers), striatum, and neocortex. The concentrations of Zn 2ϩ within these vesicles have been estimated to be as high as millimolar levels (3). Neuronal firing results in the release of both glutamate and Zn 2ϩ into the synaptic cleft (4 -7). Intense firing can result in Zn 2ϩ concentrations of several hundred micromolars (4). This estimate is based upon the accumulation of Zn 2ϩ in the perfusate of hippocampal slices. Thus, the actual localized concentrations of the metal may be significantly greater. The fate of neuronally released Zn 2ϩ is not totally clear. It is likely that some of the metal is taken back up by the presynaptic neurons and reconcentrated into vesicles (8). The movement of Zn 2ϩ from presynaptic to postsynaptic cell has been documented. In general, this translocation is associated with the neurotoxic effects of Zn 2ϩ (4). For example, excessive firing of mossy fibers of the hippocampus leads to selective Zn 2ϩ uptake and the destruction of the CA1 neurons that are innervated by these bundles. However, the fact that Zn 2ϩ translocation from presynaptic to postsynaptic cells occurs allows for the possibility that this metal is also used in the central nervous system as a transcellular messenger (1)(2)(3).
The immediate targets of intracellular Zn 2ϩ are unclear as are the physiological consequences of its actions on specific targets. In neuronal cells, targets such as glyceraldehyde 3Јphosphate dehydrogenase, NAD-catabolizing activities such as poly(ADP-ribose) synthetase, or the production of reactive oxygen species have been proposed as mediators of Zn 2ϩ neurotoxicity (for review, see Ref. 9 and references therein). However, Zn 2ϩ deprivation is also cytotoxic, and there are numerous examples where culturing cells in Zn 2ϩ -deficient medium results in cell death, whereas higher external concentrations of Zn 2ϩ appear as antiapoptotic (for review, see Ref. 10 and references therein). In that case, Ca 2ϩ /Mg 2ϩ -dependent endonucleases, caspases, Bcl2/Bax ratios, and cytoskeletal components are some of the proposed targets that Zn 2ϩ modulates with a protective outcome.
In light of the importance of cAMP as a second messenger, we have examined whether Zn 2ϩ could elicit its effects by altering adenylyl cyclase (AC) 1 activity. Biochemical studies of mutant ACs and recent crystallographic studies of a truncated soluble form of AC have indicated that the enzyme possesses two catalytic metal binding sites (11,12). X-ray crystallography studies distinguished these two sites by their preferential occupancy by either Zn 2ϩ or Mn 2ϩ . Although both sites are presumed to be occupied by Mg 2ϩ in vivo (12), these observations allow for the possibility that Zn 2ϩ could influence cAMP production particularly in the brain where Zn 2ϩ concentrations are significant. In the studies reported here, we have examined the effects of Zn 2ϩ on cAMP signaling in N18TG2 neuroblastoma cells. We have determined that low micromolar concentrations of Zn 2ϩ inhibit hormone-and forskolin-stimulated cAMP accumulation directly by inhibiting AC. Furthermore, we have characterized the potent inhibitory effects of Zn 2ϩ on AC in both isolated N18TG2 membranes and membranes isolated from Sf9 cells expressing recombinant isoforms.

EXPERIMENTAL PROCEDURES
Cell Culture Conditions-N18TG2 cells were grown in Dulbecco's modified Eagle's/Ham's F-12 (1:1) medium containing 10% heat-inactivated donor bovine calf serum and penicillin/streptomycin. The medium was changed 16 -24 h prior to experimentation, and cells were confluent at the time of experimentation. Cells were dissociated from flasks by gentle trituration with phosphate-buffered saline containing 0.6 mM EDTA and resuspended to 2 ϫ 10 6 cells/ml in Gey's balanced salt solution without calcium but containing 0.1 mg/ml fatty acid-free bovine serum albumin, 10 mM Na-Hepes, pH 7.4, and the phosphodiesterase (PDE) inhibitor Ro20-1724 at 0.1 mM. Cells were incubated in that buffer in a shaking 37°C water bath for increasing time periods during which different concentrations of Zn 2ϩ were added.
cAMP Assay-To determine cAMP levels in intact N18TG2 cells, we treated cells with or without stimulatory agents for a maximum of 4 min, a time within the linear range of stimulation, and then lysed them by boiling in sodium acetate, pH 4.5 (13). cAMP levels in the soluble fraction were measured using a modified Gilman assay (14). All assay points were taken in triplicate and assayed in triplicate. Triplicates generally agreed within 5% (13). Protein determinations were done according to Bradford (15).
AC Assay-Membrane-bound AC activity was assayed by incubating N18TG2 and Hi5 membranes for 20 min or Sf9 membranes for 10 min at 30°C in a HEPES-buffered mixture containing Mg 2ϩ , ATP, an ATPregenerating system, cAMP, and PDE inhibitors (0.1 mM Ro20-1724 or isobutylmethylxanthine, respectively) and monitoring the conversion of [␣-32 P]ATP to [ 32 P]cAMP (16 -18). Unless indicated otherwise, ATP and MgCl 2 were added at 0.5 and 10 mM, respectively, and EDTA was not added to the reaction mixture. When PGE 1 -stimulated AC activity in N18TG2 membranes was monitored, 10 Ϫ4 M GTP was added. All samples were assayed in triplicate in agreement generally within 5%. The data are presented as the mean Ϯ S.E. Enzyme activity is proportional to the concentration of membranes and the time of incubation. Unless indicated otherwise, when Zn 2ϩ was added, it was present in the reaction mixture at the start of the incubation. The concentrations of Zn 2ϩ added do not interfere with the energy-regenerating system. Because AC activity in Sf9 membranes is not dependent upon that system, it could be eliminated without altering the dose-response curves for Zn 2ϩ inhibition. It is also noted that the concentrations of Zn 2ϩ were sufficiently low as not to alter the concentration of Mg-ATP available as the substrate. This finding was evaluated using the free WinmaxC software developed by C. Patton at Stanford (Palo Alto, CA).
Materials-[␣-32 P]ATP and [ 3 H]cAMP were purchased from ICN and PerkinElmer Life Sciences, respectively. Tissue culture medium was purchased from Biowhittaker (Walkersville, MD). Heat-inactivated donor bovine calf serum was from JHR Biosciences (Lenexa, KS). All other reagents were from Sigma.

RESULTS
Zn 2ϩ Inhibits cAMP Accumulation in N18TG2 Neuroblastoma Cells-N18TG2 cells were incubated in the absence or presence of 300 M Zn 2ϩ for 2 h, and cAMP accumulation in response to forskolin and PGE 1 stimulation was monitored. As seen in Fig. 1, the incubation of cells with Zn 2ϩ resulted in a dramatic attenuation of cAMP accumulation in response to both stimuli. It is unlikely that the low levels of cAMP reflect a stimulation of a PDE by Zn 2ϩ , because cells were preincubated throughout the experiment with PDE inhibitors. Moreover, Zn 2ϩ had no effect on nonstimulated (basal) cAMP levels, also arguing against a Zn 2ϩ effect on PDE activity. Because forskolin binds to and directly activates AC, the inhibition of cAMP accumulation would not likely reflect a primary effect on hormone receptors or G s . When cells were preincubated with the heavy metal ionophore, pyrithione, a greater inhibition by Zn 2ϩ was observed, suggesting that internalization of Zn 2ϩ is necessary for the inhibition of cAMP production (Fig. 1).
The extracellular levels of Zn 2ϩ necessary to observe this inhibition were examined (Fig. 2). Little effect on forskolinstimulated cAMP accumulation was seen when cells were preincubated with 1 or 10 M ZnCl 2 for 2 h. Preincubation with 25 M ZnCl 2 resulted in an approximate 25-30% attenuation of the forskolin response, whereas 150 M resulted in an approximate 60 -70% inhibition. When cells were incubated with Zn 2ϩ and 5 M pyrithione, concentrations of Zn 2ϩ as low as 10 M could significantly inhibit cellular cAMP accumulation in response to forskolin, and complete inhibition was observed at ϳ25 M ZnCl 2 . Thus, the relatively high extracellular concentrations of Zn 2ϩ appear necessary to allow for the accumulation of sufficient intracellular levels. This is not unusual. For example, high concentrations of Zn 2ϩ will suppress apoptosis in model cell culture systems, but concentrations approaching serum levels of Zn 2ϩ (15-25 M) will do so if pyrithione is present to facilitate uptake (19). In the experiment shown in this figure, pyrithione alone had a slight inhibitory action on forskolin-stimulated cAMP production. This effect was inconsistent and relatively minor such as not to alter the interpretation of the data. Fig. 3 indicates that when cells were preincubated with 100 M ZnCl 2 , a time period of more than 60 min was necessary to observe significant inhibition of forskolin-stimulated cAMP accumulation. The extent of inhibition increased with time and appeared maximum after 120 min. However, when pyrithione was present, the inhibition could be observed even after 30 min of incubation. In that case, maximum inhibition seemed to require between 60 and 90 min of incubation. The time dependence of the effects of Zn 2ϩ is consistent with an uptake process that appears to be a limiting factor in eliciting the inhibition of cAMP accumulation. It is generally accepted that the transfer of metals across the plasma membranes involves metal complexes that influence the efficiency with which the metal is delivered or transferred (8, 21-23). Thus, we evaluated the inhibition of cAMP accumulation by Zn 2ϩ when the metal was presented as ZnCl 2 or as a complex of zinc ascorbate. As shown in Fig. 4, Zn 2ϩ became a more potent inhibitor of forskolin-stimulated cAMP accumulation when ascorbate was present. In that case, significant inhibition was observed with 25 M Zn 2ϩ and appeared almost maximum with 75 M. In contrast, zinc citrate and ZnSO 4 behaved in a fashion similar to ZnCl 2 (data not shown).
Zinc Inhibits AC Activity in N18TG2 Membranes-Plasma membranes purified by sucrose gradient centrifugation from N18TG2 cells were assayed for hormone-stimulated AC activity in the presence of varied concentrations of ZnCl 2 . As shown in Fig. 5A, micromolar concentrations of Zn 2ϩ effectively inhib-ited AC activity such that at 60 M ZnCl 2 , little PGE 1 -stimulated enzyme activity could be detected. The IC 50 for ZnCl 2 was ϳ8 -9 M. Similar results were obtained using ZnSO 4 . Because the assay for AC is performed in the presence of a PDE inhibitor and PDE activity is monitored by the tracer [ 3 H]cAMP present in the assay, it is clear that the decreased levels of cAMP do not reflect the activation of PDE by Zn 2ϩ . Low micromolar concentrations of Zn 2ϩ also effectively inhibited forskolin-stimulated AC activity (Fig. 5B). However, in this case, significant forskolin-stimulated activity could still be observed when 60 M ZnCl 2 was present. The data, analyzed as a simple two-site binding model, suggested that two classes of binding sites could be present, one with an apparent IC 50 of ϳ2-3 M and another greater than 10-fold.
Kinetic Analyses of the Effects of Zn 2ϩ -We first determined the effects of Zn 2ϩ on forskolin-stimulated AC activity measured at increasing concentrations of substrate. The results obtained when AC activity in N18TG2 membranes was assayed in the presence or absence of 10 M Zn 2ϩ are depicted in Fig.  6A. An analysis of the rate of substrate utilization indicated that the K m slightly increased from 0.23 Ϯ 0.02 to 0.39 Ϯ 0.09 mM in the presence of Zn 2ϩ . The V max of the activity in N18TG2 membranes showed a 2-fold decrease in the presence of Zn 2ϩ going from 645 Ϯ 19 pmol/min/mg to 318 Ϯ 25 pmol/min/mg. Current data indicate that the active site of AC contains two metal ion binding sites referred to as A and B, which can be selectively occupied by Zn 2ϩ and Mn 2ϩ , respectively (12). To assess whether the inhibitory Zn 2ϩ site is indeed the metal site A, we also assayed forskolin-stimulated activity using a constant ATP concentration and varying Mg 2ϩ concentrations in the absence or presence of Zn 2ϩ . As depicted in Fig. 6B, the presence of 40 M ZnCl 2 resulted in an approximate 4-fold decrease in V max . The respective V max values were 258 Ϯ 11 pmol/min/mg and 73 Ϯ 10 pmol/min/mg. K m values were 0.6 Ϯ 0.6 M and 1.3 Ϯ 0.12 M when AC activity was measured in the absence and presence of 40 M Zn 2ϩ , respectively. With 10 M ZnCl 2 , we observed an approximate 2-fold decrease in both the apparent K m and V max for the Mg 2ϩ dependence of AC activity (data not shown). The sum of the data suggests that the major effect of Zn 2ϩ on AC activity is to reduce the rate of enzyme conversion of substrate to product.
Recombinant AC Is Inhibited by Zn 2ϩ -N18TG2 cells appear to express AC6 as their predominant activity (18). To determine whether AC itself is the target of Zn 2ϩ inhibition, we examined recombinant AC6. We also examined the other member of this isoform family, AC5, which is expressed primarily in the striatum, a site of relatively high neuronal levels of Zn 2ϩ (reviewed in Refs. 1-3 and 24 -26). Low micromolar concentrations of Zn 2ϩ effectively inhibited the forskolin-stimulated activity of both isoforms (Fig. 7A). The data were identical to those obtained when ZnSO 4 or ZnCl 2 plus ascorbate was evaluated (data not shown). The fact that Zn 2ϩ inhibits the recombinant enzyme indicates that this reflects a direct interaction with the enzyme. The inhibition of forskolin-stimulated AC5 and AC6 activity was dramatic (100% at 10 M) occurring over a narrow range of Zn 2ϩ concentrations. An analysis of the data by Hill plot revealed a coefficient of 1.33, indicating that the inhibition was a cooperative process. The IC 50 for ZnCl 2 inhibition was ϳ1-2 M, a value similar to that of the apparent higher affinity site seen when forskolin-stimulated AC activity in N18TG2 membranes was monitored. No evidence for the lower affinity site present in N18TG2 membranes was obtained in our experiments using recombinant enzyme. This would suggest that the apparent lower affinity site seen using N18TG2 membranes is not inherent to the AC enzyme but may reflect the participation of additional AC regulators present in those membranes. Because of the lower specific activity of the AC6 preparations, additional analyses were performed using AC5.
The Inhibitory Zn 2ϩ Site Does Not Appear to Be the Catalytic Metal Binding Site-The inability of Zn 2ϩ to dramatically alter the K m for Mg 2ϩ or Mg 2ϩ /ATP would suggest that Zn 2ϩ is not competitively inhibiting the binding of Mg 2ϩ and, therefore, is not binding to metal site A. To further address this issue, we assessed the dose-response curve for Zn 2ϩ inhibition of AC5 and of the activity in N18TG2 membranes using a range of Mg 2ϩ concentrations varying from 3 to 40 mM. Concentrations of Mg 2ϩ Ͼ50 mM were found to be partially inhibitory and were therefore not investigated. The reasons for this inhibition are currently unclear. The dose-response curve obtained when 40 mM MgCl 2 was added to the assay for recombinant AC5 is representative of the results and is shown in Fig. 7B. A comparison with the data in Fig. 7A indicates that the dose-response curve was unaffected by the increased concentration of Mg 2ϩ present. Neither the IC 50 nor the cooperative nature of the inhibition by Zn 2ϩ was altered. Reducing the Mg 2ϩ concentration to 3 mM also did not change the enzyme sensitivity to MgCl 2 was added at 10 mM. The activity seen with 0.1 M ZnCl 2 was equivalent to that seen when no Zn 2ϩ was added. The results are representative of three experiments. The data are presented as the percent of maximum activity as the specific activity of the AC6 preparation is ϳ15-fold lower than that of AC5. B, the dose-response curve for Zn 2ϩ inhibition of AC5 was examined when activity was measured in the presence of 40 mM MgCl 2. The data are representative of two experiments. Zn 2ϩ (data not shown). Such findings are consistent with the premise that Zn 2ϩ does not bind to a catalytic metal binding site unless Zn 2ϩ binds to such a site in a manner that appears kinetically irreversible.
Several experiments were performed to address the reversible nature of Zn 2ϩ inhibition. In the experiment shown in Fig.  8A, we monitored the time course of AC5 inhibition by 15 M Zn 2ϩ . As shown, the rate of forskolin-stimulated enzyme activity in the presence of Zn 2ϩ was reduced but linear over the time frame of the experiment, suggestive of a reversible reaction (27,28). The reversibility of the inhibition was further evaluated by adding EDTA to the assay after the first 10 min of incubation with Zn 2ϩ . In that case, the rate of forskolin-stimulated cAMP synthesis was substantially recovered. The addition of EDTA to control samples did not alter the rate of cAMP formation. Similar observations were made when AC activity in N18TG2 membranes was monitored (data not shown). The data indicate that the inhibition of AC by Zn 2ϩ is reversible.
To address whether the site of Zn 2ϩ inhibition is located at the other metal site, the preferential binding site of Mn 2ϩ (site B), we assessed the effects of Mn 2ϩ on the efficacy of Zn 2ϩ inhibition of AC5. The data depicted in Fig. 8B shows that 0.1, 1, and 10 mM Mn 2ϩ had no effect on the IC 50 for Zn 2ϩ . Note that Mn 2ϩ is a potent activator of AC activity, and thus higher maximal activities accompany increasing Mn 2ϩ concentrations. The data support the premise that the site of Zn 2ϩ inhibition is distinct from the catalytic metal binding sites.
Additional support for this premise was obtained in our studies of the C441R mutant of AC5 (11). This mutant AC5 displays a dramatic decrease in catalysis with only a slight change (2-fold) in the ability to bind substrate. The mutation is located at the residue adjacent to Asp-440, one of the critical aspartates that participate in coordinating metal binding (12). The close proximity of Cys-441 led the authors to suggest that a mutation in this residue results in a conformational change that disrupts the orientation of Asp-440 and thus the metal binding pocket (11). As shown in Fig. 9A, the C441R mutant behaves in a fashion identical to wild-type AC5 in response to ZnCl 2 inhibition of forskolin-stimulated activity (IC 50 of 1.8 M). Thus, this mutation did not supplant the Zn 2ϩ inhibition of AC5. The Effects of Zn 2ϩ Are Isoform-specific-To examine the effects of Zn 2ϩ on other AC isoforms, we focused our attention on recombinant AC1 as representative of the AC1, AC3, and AC8 isoform family, and AC2 as representative of the AC2, AC4, and AC7 family (24 -26). AC1 and AC2 are of particular interest, because like AC5 they are expressed in regions of the brain with relatively high vesicular levels of Zn 2ϩ (1-3, 24 -26). As illustrated in Fig. 9B, recombinant AC1 was effectively inhibited over a narrow range of Zn 2ϩ concentrations, displaying an IC 50 of 1.4 M, similar to that of AC5. The dose-response curve of AC2, however, was distinct from the other isoforms (Fig. 9C). AC2 was not inhibited by the low micromolar con- In the latter case, an aliquot was removed after the initial 7 min of incubation, and EDTA was added to achieve a final concentration of 0.1 mM EDTA (f). The assay was continued, and cAMP production was monitored at the indicated time. The rates of enzyme activity were: control, 1.9 nmol/min/mg; Zn 2ϩ before EDTA addition, 0.66 nmol/min/mg; Zn 2ϩ after EDTA addition, 1. centrations of ZnCl 2 that were effective against AC1 and AC5. It appeared that such concentrations were actually somewhat stimulatory to the enzyme. Only at concentrations of ZnCl 2 of Ͼ10 M did we observe an inhibitory effect. This would suggest that the high affinity inhibitory site seen in AC1 and AC5 is different in AC2. Because the residues surrounding the metal binding sites are strictly conserved across all isoforms, these data would be consistent with the premise that Zn 2ϩ binds to a site distinct from the catalytic metal binding sites.
The Effects of Zn 2ϩ Are Attenuated When It Is Added to an Active Enzyme-In the experiment shown in Fig. 10A, hormonestimulated AC activity of N18TG2 membranes was inhibited ϳ60 -70% when assayed in the presence of 10 or 40 M ZnCl 2 (Fig. 10A, last two bars). However, if Zn 2ϩ was added 5 min after hormone activation of AC was initiated, the inhibition was attenuated (being almost negligible with 10 M ZnCl 2 and only ϳ20% with 40 M ZnCl 2 ). Similarly, when Zn 2ϩ was added after the assay of forskolin-stimulated AC had been initiated, the inhibition was also attenuated (Fig. 10B). The calculation of the possible chelation of Zn 2ϩ by ATP in the presence of Mg 2ϩ using WinmaxC indicated that minimal changes in the free concentration of Zn 2ϩ would be expected. This finding rules out the possibility that ATP is simply chelating Zn 2ϩ . The data suggest that Zn 2ϩ is a less potent inhibitor of AC activity when the enzyme has achieved an activated state. DISCUSSION We have demonstrated that the incubation of neuroblastoma cells with Zn 2ϩ attenuates their ability to synthesize cAMP in response to hormone or forskolin stimulation. This appears to reflect a reversible inhibition of AC and thus would be compatible with a homeostatic role for this regulation. This would also be consistent with the fact that little or no loss in cell viability (trypan blue exclusion) upon incubation with Zn 2ϩ was observed during the time course of our experiments. 2 The concentrations of Zn 2ϩ that are effective in attenuating the cell response to hormone or forskolin between 25 and 150 M may be encountered in vivo (1)(2)(3)(4)(5)19). However, the extracellular concentrations needed to inhibit cAMP accumulation in N18TG2 cells could be significantly reduced under conditions that facilitated cellular uptake of the metal. One condition was the use of the heavy metal ionophore, pyrithione, whereas another condition more apt to reflect a physiological relevant condition was the addition of ascorbate. It is not uncommon that the form of metal chelation influences the efficiency of metal transport (8,(21)(22)(23). Ascorbate is of particular interest, because it accumulates in the brain (being maintained at relatively high (1-2 mM) concentrations (29,30)). It is further concentrated in synaptic vesicles of glutamatergic neurons and released into the extracellular space as a result of neuronal activity (29,30). Therefore, ascorbate has the potential to ligate Zn 2ϩ , rendering the metal a more potent neuromodulator. It is also noted that we are probably manifesting the effects of a passive influx of Zn 2ϩ . Passive influx through the neuronal membrane has been demonstrated to occur when extracellular levels of Zn 2ϩ are elevated, e.g. when Zn 2ϩ is added to the cell culture medium or when Zn 2ϩ is released from presynaptic vesicles (31). However, experiments using cultured neuronal cells have also demonstrated that Zn 2ϩ uptake can be stimulated upon activation of voltage-gated Ca 2ϩ or NMDA channels, for example (1)(2)(3)32). Thus, physiological conditions that stimulate Zn 2ϩ uptake will potentially facilitate a more effective inhibition of AC.
That Zn 2ϩ uptake is required to observe the inhibition of AC would indicate that inhibition does not result from the nonspecific binding of Zn 2ϩ either to the plasma membrane or to the exofacial residues of the AC. That Zn 2ϩ inhibition of AC1, AC5, and AC6 appears identical could argue that amino acid residues of the transmembrane domains, which show limited homology, are not involved but that Zn 2ϩ binds to the cytosolic CI and/or CII regions, which exhibit considerable homology. This assumption is substantiated by the fact that a soluble AC construct composed of the CI region of AC5 and the CII region of AC2 is inhibited by low micromolar concentrations of Zn 2ϩ (12) and that this reflects a decrease in V max of the enzyme. 3 It is of interest to note that the IC 50 of the soluble construct is ϳ15 M, an intermediate value among that of the contributing isoforms. Although this value may reflect the loss of regions outside of the domains in question, it may also reflect the contribution of both subunits in defining the Zn 2ϩ binding site.
In our attempts to determine the mechanism by which Zn 2ϩ inhibits AC, we have made a number of observations to support the premise that it does not function at the catalytic metal binding sites. We wish to emphasize that our findings do not preclude Zn 2ϩ binding to site A, a possibility indicated by crystallographic studies. Rather, the biochemical effects of Zn 2ϩ binding to this "novel" inhibitory site may be of an affinity such that we are not in a position to observe enzymatically the effects of Zn 2ϩ binding to alternative sites. We also note that Zn 2ϩ does not generally inhibit two-metal ion-requiring enzymes as are ACs (34). Perhaps upon the disruption of the Zn 2ϩ site that inhibits AC activity, we will observe that Zn 2ϩ functions more typically, supporting or enhancing AC activity when it occupies a catalytic metal binding site. In that light, it is tempting to speculate that for AC2, which has a lower affinity Zn 2ϩ inhibitory site (IC 50 ϭ 25 M), the stimulation of activity observed at the lower Zn 2ϩ concentrations reflects the binding of Zn 2ϩ to site A. Such questions are currently being explored.
This still leaves the question of how Zn 2ϩ binding to this novel site leads to an attenuation of AC activity. The doseresponse curve indicates that this is a cooperative process, reflecting either the binding of two (more than one) Zn 2ϩ molecules or a Zn 2ϩ -induced protein-protein interaction. Although we can only speculate at this point, several observations could 2 T. Y. Hudson, unpublished observations. 3 R. K. Sunahara, unpublished observations.
FIG. 10. The effects of Zn 2؉ on activated AC. PGE 1 -stimulated (10 Ϫ5 M) (A) and forskolin-stimulated (10 Ϫ4 M) (B) AC activity was measured in N18TG2 membranes in the absence (solid bars) or presence (last two bars in 20-min group) of 10 or 40 M ZnCl 2 . Alternatively, Zn 2ϩ was added to the assay of hormone-or forskolin-stimulated activity 5 min after the assay had been initiated (open and hatched bars). Cyclic AMP synthesis was determined after an additional 5 and 15 min, resulting in a total incubation time of 10 and 20 min, respectively. The data are representative of three experiments. favor the latter interpretation. Based on current crystallographic data, the physical relationship of CI to CII is altered upon occupancy of the substrate binding site. Subsequent association of AC with G␣ s or forskolin is believed to allosterically influence the interaction to enhance the catalytic capacity of the enzyme (20,33). Thus, there is precedence for the regulation of AC via changes in protein-protein interactions. We have also determined that a preassociation of AC with forskolin or G␣ s protects the enzyme from the inhibitory effects of Zn 2ϩ . We interpret such findings to suggest that both activators induce a conformational change(s) in the enzyme that minimizes the effects of Zn 2ϩ . This interpretation as opposed to one evoking steric hindrance of the Zn 2ϩ binding site is favored, because forskolin and G␣ s bind to adjacent but distinct regions of AC and thus would obscure different residues of the enzyme (12,33).
That the G␣ s ⅐AC complex would display an altered sensitivity to Zn 2ϩ could explain the biphasic dose-response curve for forskolin-stimulated AC activity in N18TG2 membranes (Fig.  5B) compared with that of the recombinant enzyme (Fig. 7A). It can be expected that a population of enzyme in N18TG2 membranes is associated in equilibrium with G␣ and as such exhibits a decreased sensitivity to Zn 2ϩ . A greater population of enzyme would be associated with G␣ s when PGE 1 -stimulated AC activity in N18TG2 membranes is monitored, thus explaining the higher IC 50 value for Zn 2ϩ inhibition (Fig. 5A). Having shown that AC1, AC5, and AC6 display an IC 50 of ϳ1-2 M while AC2 inhibition occurs with an IC 50 of ϳ20 -25 M, we could also account for the data obtained using N18TG2 membranes if a population of AC isoforms were present. Our Northern analysis of AC expression in N18TG2 cells did not detect the mRNA for AC2 (18), but we did not probe for the presence of the other members of this family. Although other isoforms may be present, they are not typically predominant in neuronal cells, and biochemical evidence for their presence in N18TG2 cells has not been reported. The aforementioned observations also suggest the coincidence timing requirements to effect an inhibition of AC by Zn 2ϩ . Cyclic AMP synthesis would be relatively refractory to an influx of Zn 2ϩ should that occur subsequent to hormonal stimulation of the enzyme. It would also appear that the consequences of Zn 2ϩ relative to cAMP synthesis will not only depend upon the timing of the input signals but will reflect the nature of the AC isoform predominant in a particular cell or tissue. In regions of the brain in which AC2 is expressed, the stimulation of cAMP synthesis could still occur in the presence of Zn 2ϩ and may actually be enhanced under conditions that effectively limit cAMP synthesis supported by either AC1 or AC5.
In this study, we have described a novel regulation of AC by Zn 2ϩ , whereby the presence of this metal reversibly inhibits its activity. Such a regulation raises the possibility that in the re-gions of the brain like the hippocampus where Zn 2ϩ is believed to function as either a neuromodulator or neurotoxic agent, it could do so in part by altering cAMP levels. In addition to defining a previously unappreciated mechanism of Zn 2ϩ action, these experiments suggest a novel target for mediating the physiological effects of Zn 2ϩ in the central nervous system as well as in other tissues in which Zn 2ϩ concentrations may be significant.