Distinct Mechanisms of Regulation by Ca2+/Calmodulin of Type 1 and 8 Adenylyl Cyclases Support Their Different Physiological Roles*

Nine membrane-bound mammalian adenylyl cyclases (ACs) have been identified. Type 1 and 8 ACs (AC1 and AC8), which are both expressed in the brain and are stimulated by Ca2+/calmodulin (CaM), have discrete neuronal functions. Although the Ca2+ sensitivity of AC1 is higher than that of AC8, precisely how these two ACs are regulated by Ca2+/CaM remains elusive, and the basis for their diverse physiological roles is quite unknown. Distinct localization of the CaM binding domains within the two enzymes may be essential to differential regulation of the ACs by Ca2+/CaM. In this study we compare in detail the regulation of AC1 and AC8 by Ca2+/CaM both in vivo and in vitro and explore the different role of each Ca2+-binding lobe of CaM in regulating the two enzymes. We also assess the relative dependence of AC1 and AC8 on capacitative Ca2+ entry. Finally, in real-time fluorescence resonance energy transfer-based imaging experiments, we examine the effects of dynamic Ca2+ events on the production of cAMP in cells expressing AC1 and AC8. Our data demonstrate distinct patterns of regulation and Ca2+ dependence of AC1 and AC8, which seems to emanate from their mode of regulation by CaM. Such distinctive properties may contribute significantly to the divergent physiological roles in which these ACs have been implicated.

Nine membrane-bound mammalian adenylyl cyclases (ACs), 2 AC1-AC9, have been identified (1). They possess a common predicted structure (2) 3 and are stimulated by forsko-lin (FSK; except AC9) and G s␣ , although they are distributed and regulated differently (1,3,4). Four ACs are regulated by physiological concentrations of Ca 2ϩ and thereby provide a critical link between the Ca 2ϩ -and cAMP-signaling pathways (3,5); AC5 and AC6 are directly inhibited by Ca 2ϩ , whereas AC1 and AC8 are stimulated by Ca 2ϩ in a calmodulin (CaM)dependent manner (5). AC3 is also regulated by CaM in vitro, although this requires supramicromolar concentration of Ca 2ϩ (6), and in vivo AC3 is inhibited by Ca 2ϩ via CaM kinase II (7), unlike AC1 and AC8.
AC1 is closely related in sequence to the Ca 2ϩ /CaM-stimulable rutabaga AC from Drosophila, which is important in Drosophila learning tasks (8 -10). AC1 and the other Ca 2ϩ /CaMstimulable mammalian AC, AC8, have also been implicated in learning and memory (11). As a means of establishing their proposed roles, single and/or double AC1 and AC8 knockout mice have been generated. Mouse models have demonstrated that Ca 2ϩ /CaM-stimulable ACs are involved in long-term potentiation and long-term memory (12). However, despite the general view that AC1 and AC8 can behave similarly, discrete physiological actions of each isoform are becoming apparent. Recent studies by Zhuo's group demonstrated that AC1 specifically participates in N-methyl-D-aspartic acid receptor-induced neuronal excitotoxicity (13) and an increase in GluR1 synthesis induced by blocking AMPA receptors (14). Furthermore, Nicol and colleagues (15,16) showed a contribution of AC1, but not AC8, in axon terminal refinement in the retina. On the other hand, AC8 specifically was seen to be responsible for retrieval from adaptive presynaptic silencing (17) and the acquiring of new spatial information (18). These differences in physiological roles must reflect not only differences in their distributions but also presumably in their regulatory properties. Both enzymes are expressed in brain; AC1 is neuro-specific, whereas the expression of AC8 is more widespread (1,12). Within the central nervous system, AC1 is abundant in the hippocampus, the cerebral cortex, and the granule cells of the cerebellum, whereas AC8 has a high expression level in the thalamus and the cerebral cortex (19). Studies of mouse brain revealed that AC1 is distributed pre-synaptically and AC8 post-synaptically (18,20).
Although physiological differences in the roles of these two enzymes are suggested from the studies outlined above, the regulatory mechanisms that might underlie these differences are not. AC1 is more sensitive to Ca 2ϩ than is AC8 in vitro (21,22), yet details on how these two enzymes are regulated by Ca 2ϩ /CaM are sparse. In non-excitable cells, a Ca 2ϩ elevation caused by capacitative Ca 2ϩ entry (CCE), the mode of Ca 2ϩ entry triggered by emptying Ca 2ϩ from internal stores (23), preferentially stimulates AC1 and AC8 (21). Although stimulation of AC8 by CCE has been shown to be at least partially dependent on its localization at lipid rafts (24), whether AC1 is also targeted to this region of plasma membranes has never been addressed. In addition, CaM regulation of AC1 and AC8 has not been compared in detail, although CaM appears to bind to different domains of the two enzymes. AC8 utilizes two CaM binding domains: a classic amphipathic "1-5-8-14" motif at the N terminus and an IQ-like motif in the C2b domain (25). A recent study indicates that CaM pre-associates with the N terminus of AC8, where it becomes fully saturated upon a Ca 2ϩ rise, and activates the enzyme via a C-terminally mediated relief of auto-inhibitory mechanisms (26). By contrast, only residues 495-522 of the C1b region of AC1 have been shown to bind CaM in a Ca 2ϩ -dependent manner (27,28). With the presence of only one CaM binding domain in AC1, a simpler mechanism of CaM regulation might be expected.
CaM mediates the regulation of numerous Ca 2ϩ -dependent processes in eukaryotic cells (29). The protein possesses N-and C-terminal lobes, both of which contain two Ca 2ϩ binding EF hands (EF1 and EF2 in the N lobe, and EF3 and EF4 in the C lobe (30)). Mutations in the EF hands have been valuable for investigating the interaction of CaM with its targets. Alanine substitutions in the EF12 pair or EF34 pair have generated CaM 12 and CaM 34 to investigate the independent function of the C and N lobes of CaM, respectively (31,32).
Against the background of the distinct physiological roles carried out by AC1 and AC8, we performed a detailed comparison of the two enzymes expressed in HEK 293 cells. Their sensitivity to Ca 2ϩ /CaM was compared both in vitro and in vivo; the possibility that they might be expressed in different domains of the plasma membrane was addressed; and putative lobe-specific CaM regulation was assessed using Ca 2ϩ -binding mutants of CaM. Single cell measurements using a FRET-based cAMP sensor were performed to compare the kinetic responses of the enzymes to physiological elevations of [Ca 2ϩ ] i . The results demonstrate superficial similarities in the regulation of AC1 and AC8 but critical disparities in their mechanism of activation by the lobes of CaM and in the speed and pattern of their responsiveness to [Ca 2ϩ ] i . These discrete behaviors provide a physiological opportunity for different outcomes to elevation of [Ca 2ϩ ] i , depending on the AC that is expressed in particular contexts.

EXPERIMENTAL PROCEDURES
Materials-FSK and thapsigargin (TG) were purchased from Merck (Nottingham, UK). [2-3 H]Adenine, [2, H]cAMP, ECL Western blotting analysis system, Hyperfilm TM , and horseradish peroxidase-conjugated goat anti-rabbit IgG were obtained from GE Healthcare (Little Chalfont, UK). [␣-32 P]ATP was from PerkinElmer Life Sciences. Lipofectamine TM 2000 transfection reagent, fura-2/AM, fura-2 free acid, fura-FF free acid, and pluronic F-127 were from Invitrogen. Horseradish peroxidase-conjugated goat anti-mouse IgG was from Promega (Madison, WI). Polyclonal caveolin antibody was obtained from BD Transduction Laboratories, and monoclonal FLAG antibody was from Stratagene. M13 skeletal muscle myosin light chain kinase peptide (M13) was from Cambridge Bioscience (Cambridge, UK). CaM mutant constructs were a gift from J. H. Caldwell (University of Colorado Health Sciences Center, Denver, CO), and cDNA encoding the Epac1-based fluorescent cAMP sensor (Epac1-camps) was a gift from M. J. Lohse (Institute of Pharmacology and Toxicology, Würzburg, Germany). All other agents were purchased from Sigma unless stated otherwise.
Purification of Rat His 6 CaM-Wild-type and mutant rat His 6 -CaM were produced following cloning of CaM cDNA into the pQE30 vector (Qiagen), and propagation in the XL10 Gold strain of Escherichia coli. His 6 -CaM was purified using TALON resin (BD Clontech) immobilized metal ion affinity chromatography.
Cell Culture and Transfection of HEK 293 Cells-HEK 293 cells (European Collection of Cell Cultures, Porton Down, UK) were grown in minimum essential medium with Earle's salts, supplemented with 10% (v/v) fetal bovine serum, 2 mM L-glutamine, 50 g/ml penicillin, 50 g/ml streptomycin, and 100 g/ml neomycin. The cells were maintained at 37°C in a humidified atmosphere of 95% air and 5% CO 2 . To establish stable cell lines expressing AC8, AC8M1 (AC8 ⌬1-106 ), AC1, AC1M1, AC1F503A, or vector (pcDNA3.1; pcD), HEK 293 cells were transfected with 2 g of cDNA according to the calcium phosphate method described previously (33). Two days after transfection, cells were selected by growth in medium containing 800 g/ml G-418 for 4 days, and subsequently maintained in medium containing 400 g/ml G-418. For transient transfection with Epac1-camps, pcD or various forms of CaM cDNA, cells stably expressing AC1 or AC8 were plated onto 25-mm poly-L-lysine-coated coverslips or 100-mm dishes at 60% confluence 1 day prior to transfection with 1 g of cDNA using Lipofectamine TM 2000 according to the manufacturer's instructions. Cells were used 2 days after transfection.
Measurement of AC Activity-Determination of adenylyl cyclase activity in vitro was performed as described previously (34) with some modifications. Adenylyl cyclase activity of crude membranes isolated from transfected HEK 293 cells (35) was measured in the presence of the following components: 12 mM phosphocreatine, 2.5 units of creatine phosphokinase, 0.1 mM cAMP, 1.4 mM MgCl 2 , 0.1 mM ATP, 0.04 mM GTP, 0.5 mM IBMX, 0.5 Ci of [␣-32 P]ATP, and 10 M FSK. Wild-type or mutant CaM was included where indicated. Free Ca 2ϩ concentrations were established from a series of CaCl 2 solutions buffered with 200 M EGTA, based on the BAD4 program (36) and confirmed by spectrofluorometric measurements with fura-2 and fura-FF as described previously (26). Membranes were washed twice with assay buffer (800 M EGTA, 0.25% bovine serum albumin, 40 mM Tris, pH 7.4) to remove Ca 2ϩ -bound CaM before starting the assay. Where indicated, membranes were incubated with 1 M M13 for 10 min at 4°C prior to washing with assay buffer. The reaction mixture (final volume, 100 l) was incubated at 30°C for 20 min, and the reactions were terminated with 0.5% (w/v) sodium lauryl sulfate. [ 3 H]cAMP (ϳ6000 cpm) was added as a recovery marker, and the [ 32 P]cAMP formed was quantitated using a sequential chromatography technique described previously (37). Data points are presented as mean activities Ϯ S.D. of triplicate determinations. Where indicated, the AC activity of cells transfected with empty vector was subtracted from the data.
[Ca 2ϩ ] i Measurements in Cell Populations-[Ca 2ϩ ] i was measured in populations of HEK 293 cells as described previously (38). Briefly, cells were loaded with 2 M fura-2/AM plus 0.02% pluronic F-127 for 40 min at room temperature, washed twice, and then aliquoted into samples containing 4 ϫ 10 6 cells. The cells were finally resuspended in 3 ml of nominally Ca 2ϩfree Krebs buffer (120 mM NaCl, 4.75 mM KCl, 1.44 mM MgSO 4 , 11 mM glucose, 25 mM HEPES, pH 7.4), and used for [Ca 2ϩ ] i measurements in a PerkinElmer Life Sciences LS50B spectrofluorometer. Fluorescence emission ratios at 340 nm/380 nm were measured and converted to [Ca 2ϩ ] i values using the standard formula (39).
Measurement of cAMP Accumulation-cAMP accumulation in intact cells was measured as described previously (40), with some modifications. Transfected HEK 293 cells were incubated in minimum essential medium with [2-3 H]adenine (1.5 Ci/well in 24-well plates) at 37°C for 90 min to label the ATP pool. Assays were carried out in Krebs buffer supplemented with bovine serum albumin (1 mg/ml) at 30°C and terminated by the addition of ice-cold 5% (w/v final concentration) trichloroacetic acid. Unlabeled cAMP (50 l, 20 mM), ATP (10 l, 65 mM), and [␣-32 P]ATP (ϳ6000 cpm) were added to monitor recovery of cAMP and ATP. The [ 3 H]ATP and [ 3 H]cAMP contents were quantified using the sequential chromatography technique described above (see "Measurement of Adenylyl Cyclase Activity"). Accumulation of cAMP is expressed as the conversion of [ 3 H]ATP into [ 3 H]cAMP. Results are presented as mean Ϯ S.D. of triplicate determinations.
Non-detergent Isolation and Immunoblotting of Raft and Non-raft Membranes-Raft and non-raft membranes were separated by a procedure that exploited their different buoyancies as previously described (41), using sonication with sodium bicarbonate (42). Raft and non-raft membranes were resolved using 7.5 and 12% SDS-polyacrylamide gels and transferred to a polyvinylidene difluoride membrane, as described previously (42). Anti-FLAG antibody (1:1000) or anti-caveolin polyclonal antibody (1:5000) was used as a primary antibody, and goat anti-mouse IgG conjugated to horseradish peroxidase (1:3000) or goat anti-rabbit IgG conjugated to horseradish peroxidase (1:5000) was used as a secondary antibody, respectively. Membranes were then treated with ECL reagent and exposed to Hyperfilm TM .
Single Cell Measurement of [Ca 2ϩ ] i -[Ca 2ϩ ] i in single cells was measured as described previously (43). Briefly, cells were plated onto poly-L-lysine-coated coverslips 1 day before experiments and loaded with 2 M fura-2/AM plus 0.02% Pluronic F-127 for 40 min at room temperature in HBS buffer (140 mM NaCl, 4 mM KCl, 0.2 mM MgCl 2 , 11 mM D-glucose, 10 mM HEPES, pH 7.4) with 1 mM CaCl 2 . After loading, cells were washed, further incubated in HBS buffer to de-esterify intracellular AM esters, and then imaged using a CoolSNAP-HQ CCD camera (Photometrics, Tucson, AZ) and monochromator system (Cairn Research, Kent, UK) attached to a Nikon TMD microscope (ϫ40 objective). Emission images (D510/80M) at 340 nm and 380 nm excitation were collected every second with MetaFluor software (Molecular Devices, Downingtown, PA). Data are plotted as 340/380 nm ratio changes (⌬340/380) relative to the fluorescence ratio at 0 min.

Single Cell Measurement of cAMP in AC1 and AC8 Expressing Cells Using Epac1-based Fluorescent cAMP Sensor-
[cAMP] i in single cells expressing AC1 or AC8 was measured by fluorescent imaging of Epac1-camps as described previously (43). Briefly, images were captured using an Andor Ixonϩ camera and an Optosplit (505DC) to separate CFP (470 nm) and YFP (535 nm) emission images (Cairn Research). For dual emission-ratio imaging, cells were excited at 436 nm using a monochromator (Cairn Research) and 51017 filter set (Chroma, Rockingham, VT) attached to a Nikon eclipse TE2000-S microscope (ϫ40 objective). Emission images at 470 nm and 535 nm were collected every 3 s (250-ms integration time). Captured images were background-subtracted and analyzed using Metamorph imaging software (Molecular Devices). Fluorescence resonance energy transfer data are plotted as changes in 470 nm versus 535 nm (⌬CFP/YFP) emission ratio relative to the fluorescence ratio at 0 min, for each individual cell.
Curve-fitting and Statistical Analysis-Sigmoidal dose-response curves, linear regressions, and kinetic parameters were obtained using GraphPad Prism Version 4 (GraphPad Software Inc., La Jolla, CA). Results are shown as mean Ϯ S.D. of at least three individual experiments, or mean Ϯ S.E. The initial rates of increase in [cAMP] i in single cell measurements were determined by obtaining slopes of linear regressions fitted over a 20-s period, starting 30 s after the addition of TG or carbachol (CCh). Statistical significance was assessed by using one-way analysis of variance followed by Newman-Keuls multiple comparisons tests, where p Ͻ 0.05 was considered significant.

AC1 and AC8 Differ in Their
Regulation by Ca 2ϩ -To establish a baseline for comparing the regulation of AC1 and AC8 by Ca 2؉ /Calmodulin Regulation of AC1 and AC8 FEBRUARY 13, 2009 • VOLUME 284 • NUMBER 7 Ca 2ϩ , their responses were first examined in vitro. Prior to each assay, membranes prepared from HEK 293 cells stably expressing AC1 or AC8 were washed in EGTA to remove Ca 2ϩ -dependent bound CaM. AC1 and AC8 were stimulated by 10 M FSK and a range of [Ca 2ϩ ] free in the absence (control) or presence of exogenous CaM (1 M). Inclusion of 10 M FSK enhanced the activities of both enzymes, allowing more sensitive assays, although this did not affect the individual kinetics of regulation by Ca 2ϩ /CaM (data not shown). Membranes from HEK 293 cells stably expressing empty vector (pcD) showed Ca 2ϩ inhibition (supplemental Fig. S1) due to the background of the Ca 2ϩ -inhibitable AC6 in these cells (44). To eliminate the inhibitory component, endogenous AC activity was subtracted from activities measured in membranes from cells expressing AC1 or AC8 at each Ca 2ϩ concentration (Fig. 1, A and B). AC1 and AC8 were both robustly stimulated by submicromolar concentrations of Ca 2ϩ (Fig. 1, A and B). (Their sensitivities to Ca 2ϩ were unaffected by the background subtraction (data not shown).) A reduced Ca 2ϩ stimulation in AC1 and AC8 was seen in the absence of added CaM, indicating that both enzymes can be regulated by CaM that was not removed by the wash with EGTA.
To verify that the observed residual Ca 2ϩ stimulation was due to endogenous CaM, the synthetic peptide M13 was used as a "CaM sponge." M13 is a 26-residue peptide derived from a CaM binding domain of skeletal muscle myosin light chain kinase (45), which has a very high affinity for CaM (K d ϳ 0.2 nM (46)). When membranes were incubated with 1 M M13 prior to washing with EGTA, the residual Ca 2ϩ stimulation in the absence of added CaM was completely eliminated (Fig. 1, A and  B), showing that endogenous CaM was responsible for Ca 2ϩ stimulation in the absence of added CaM.
Although basal activities varied depending on expression levels, the differences in Ca 2ϩ sensitivity between AC1 and AC8 were consistent. AC8 was stimulated by Ca 2ϩ by 5-to 10-fold in the presence of CaM and 2.7-to 4.2-fold in the absence of added CaM (n ϭ 11, Fig. 1B). AC1, in contrast, was much less stimulated by Ca 2ϩ ; Ca 2ϩ stimulation of AC1 was 1.5-to 3-fold and 1.1-to 1.5-fold, in the presence and absence of added CaM, respectively (n ϭ 9, Fig. 1A). Although AC1 is less stimulable, AC1 is more sensitive to Ca 2ϩ than AC8. AC1 was stimulated by Ca 2ϩ with an EC 50 of 0.15 M in the presence of added CaM (Fig. 1, A and E), whereas the equivalent value for AC8 was 0.56 M (Fig. 1, B and E).
The elevation of Ca 2ϩ in intact cells also regulates Ca 2ϩsensitive ACs. AC8 is stimulated robustly by CCE, but not by ionophore-mediated Ca 2ϩ release or arachidonic acid-or oleylarachidonylglycerol-mediated Ca 2ϩ entry (21,35,47). To compare the Ca 2ϩ regulation of AC1 and AC8 in vivo, cAMP accumulation in response to CCE was measured in cell populations expressing either AC1 or AC8. Ca 2ϩ was emptied from ER stores by passive depletion with the sarcoplasmic-endoplasmic reticulum calcium ATPase pump inhibitor, TG (100 nM) in Ca 2ϩ -free buffer; CCE was subsequently evoked by the addition of extracellular Ca 2ϩ ([Ca 2ϩ ] ex ) after 4 min (Fig. 1C). CCE achieved values of 0.15-0.55 M [Ca 2ϩ ] i (Fig. 1C). CCE in cells expressing AC1 or AC8 was not significantly different from that in untransfected cells (data not shown). Under the same condi-tions, cAMP accumulation in cell populations was measured over a 1-min period beginning with the addition of 10 M FSK and the indicated [Ca 2ϩ ] ex , following 4-min pretreatment with TG, in the presence of 100 M EGTA and 100 M IBMX, phosphodiesterase inhibitor. Cells expressing vector (pcD) did not show Ca 2ϩ stimulation (Fig. 1D), underscoring the absence of endogenous Ca 2ϩ -stimulable ACs in HEK 293 cells. As with the in vitro experiments, Ca 2ϩ stimulation of AC8 was much greater than that of AC1; AC8 was stimulated by 4-to 8-fold (n ϭ 25), whereas AC1 was stimulated only by 1.8-to 3-fold (n ϭ 11) by 8 mM [Ca 2ϩ ] ex (Fig. 1D). However, the apparent Ca 2ϩ sensitivity of AC1 was greater than AC8. Ca 2ϩ stimulation of AC1 was maximal at 2 mM [Ca 2ϩ ] ex (half-maximum concentration, 0.59 mM, Fig. 1E). In contrast, Ca 2ϩ stimulation of AC8 did not reach a plateau unless [Ca 2ϩ ] ex was increased to 8 mM or more (half-maximum concentration, Ն3.71 mM, Fig. 1E). Fig. 1E (Fig. 1E). AC8 was not further stimulated when 10 mM Ca 2ϩ was added, possibly due to saturation of CCE (data not shown). Thus, although both AC1 and AC8 are stimulated by Ca 2ϩ , they clearly differ in their sensitivity to Ca 2ϩ both in vitro and in vivo.
Ca 2ϩ Regulation and Localization of AC1 Mutant-Ca 2ϩsensitive AC5, AC6, and AC8 are targeted to cholesterol-and sphingolipid-rich regions of the plasma membrane, called lipid rafts (24, 42, 48 -50). In contrast, the Ca 2ϩ -insensitive AC7 is not targeted to this domain (24,42). Inhibition of AC6 by CCE was compromised by cholesterol depletion but fully restored upon cholesterol repletion (48). Thus, specific raft localization seems to be a key requirement for the CCE regulation of Ca 2ϩsensitive ACs. To investigate whether the Ca 2ϩ -stimulable AC1 shares this property, the subcellular localization of AC1 and a key mutant was examined. The N terminus of AC8 is not critical for targeting of AC8 to lipid rafts, but it is important for Ca 2ϩ stimulation in vivo (24). Deletion of the N terminus of AC8 resulted in reduced stimulation by CCE (24). The N terminus of AC8 also interacts with protein phosphatase 2A (51), demonstrating a function other than CaM binding. Whether the N terminus of AC1 has a similar role has not been addressed, although the amino acid sequence is quite dissimilar to that of AC8. Tang and coworkers made an N-terminally truncated AC1, AC1 ⌬1-52 , which showed only 10% activity compared with the wild-type AC1, despite having a higher expression level (52). In the present experiments, a less extensively truncated AC1 construct, AC1 ⌬1-35 (AC1M1), was generated to provide a direct comparison with AC8M1, which also lacks a similar proportion (the first 106 amino acids) of the N terminus.
Caveolar and non-caveolar membranes from HEK 293 cells stably expressing FLAG-tagged AC1 and AC1M1 were separated by density gradient fractionation as a function of their increased buoyancy in sucrose gradients. To avoid any detergent-induced artifacts, membranes were disrupted by sonica-tion, instead of separating detergent-insoluble membranes using Triton X-100. Ten fractions were collected from top to bottom, and then sucrose and protein concentrations of each fraction were determined (Fig. 2B). Following fractionation, Western blot immunoreactivity for FLAG (for AC1) and the lipid raft marker, caveolin of each fraction were examined. Most of the FLAG and caveolin immunoreactivities were enriched at fractions 3 and 4 (15-25% sucrose, Fig. 2, A and B). This result revealed that, like AC8, AC1 (and AC1M1) co-localizes with caveolin, and that the deletion of the early N-terminal domain did not prevent the targeting of AC1 to lipid rafts.
The first part of the N terminus of AC1 was not essential for targeting of AC1, but would it impact on Ca 2ϩ stimulation in vivo? To address this question, cAMP accumulation in intact cells was measured in response to CCE. The robust stimulation by 4 mM [Ca 2ϩ ] ex observed in AC8 was greatly reduced in AC8M1 (Ͻ2-fold stimulation by the same [Ca 2ϩ ] ex , Fig. 2C). However, unlike AC8, deletion of the first part of the N terminus of AC1 did not affect its activity or Ca 2ϩ stimulation. Both AC1 and AC1M1 showed 1.8-to 3-fold stimulation by 2 mM [Ca 2ϩ ] ex with the same half-maximum values (Fig. 2D). When membranes were isolated from cells expressing AC8 or AC8M1, they showed robust Ca 2ϩ stimulation. Although AC8 has a CaM binding domain at the N terminus, AC8M1 was fully active in the presence of exogenous CaM, and its Ca 2ϩ sensitivity was similar to that of AC8 (Fig. 2E). Crude membranes from cells expressing AC1M1 also showed stimulation by Ca 2ϩ /CaM, which was comparable to AC1 (Fig. 2F). Both AC8M1 and AC1M1 showed full activation by Ca 2ϩ in the presence of exogenous CaM, demonstrating that the N-terminal region is not essential for Ca 2ϩ /CaM stimulation of AC8 or AC1 in vitro. However, the N terminus is important for AC8, but not AC1, activity in vivo.
Effect of CaM Mutants on Ca 2ϩ Stimulation of AC8, AC8M1, AC1, and AC1 F503A-The amphipathic CaM binding domain at the N terminus of AC8, which is absent in AC8M1, plays a key role in the stimulation of AC8 by Ca 2ϩ in vivo. Although AC8M1 showed full activation by Ca 2ϩ in the presence of added CaM in vitro (Fig. 2E), this was strictly dependent on added CaM, unlike AC8, which showed ϳ3-fold stimulation without exogenous CaM (Fig. 1B (26)). To determine how the CaM binding domain of AC1 affects its Ca 2ϩ /CaM stimulation in comparison with AC8M1, another mutant of AC1 was generated. A phenylalanine residue in the C1b domain was previously suggested to be critical for CaM binding based on an F503R mutation (53). Because arginine is a positively charged, basic amino acid, we produced a construct with a milder mutation to neutrally charged alanine at this residue (AC1F503A).
A further facet of the regulation of ACs by Ca 2ϩ /CaM is the potentially separate roles of the two lobes of CaM. Ca 2ϩ stimulation of AC8 is affected by partially liganded Ca 2ϩ /CaM (26). Gao et al. showed that AC1 could be activated by CaM mutants, which were occupied by three Ca 2ϩ ions due to a single mutation at EF1, EF2, EF3, or EF4 (54). Ca 2ϩ binds cooperatively within each lobe of CaM (55); however, the effect on AC1 of CaM with two Ca 2ϩ bound at one lobe has never been addressed. Here, we used lobe-specific EF-hand mutants of FIGURE 2. Subcellular localization and Ca 2؉ regulation of AC1 and its mutant. A, fractions (lanes 2-8) from HEK 293 cells stably expressing FLAG-tagged AC1 and AC1M1 were separated using a non-detergent method, resolved by SDS-PAGE, and then immunoblotted with antibodies raised against FLAG and the lipid-raft marker, caveolin (CaV). B, the percentage of protein content as compared with the total recovered protein (E) and percentage of sucrose (OE) from each fraction are indicated. C and D, in vivo cAMP accumulation was measured in cell populations expressing AC8 and AC8M1 (C) or AC1 and AC1M1 (D) in response to increasing CCE, as described under "Experimental Procedures." Data are normalized to cAMP accumulation at zero Ca 2ϩ and plotted as mean Ϯ S.D. of at least six independent experiments. E and F, in vitro adenylyl cyclase activity was measured in crude membranes prepared from cells expressing AC8 and AC8M1 (D, F and Ⅺ, respectively), or AC1 and AC1M1 (F, F and Ⅺ, respectively), in the presence of 1 M exogenous CaM. Data are plotted as mean Ϯ S.D. and are representative of at least five separate experiments.
CaM to compare the contribution of each CaM lobe in regulating AC1 and AC8. The CaM mutant, CaM 1234 , does not bind Ca 2ϩ due to point mutations of aspartate residues in all four EF hands, whereas CaM 12 (N-lobe mutant) and CaM 34 (C-lobe mutant) retain partial Ca 2ϩ sensitivity and can bind certain target proteins (32). To observe a clear effect of exogenous CaM, M13 treatment was employed. Activities of AC8, AC8M1, AC1, and AC1F503A were measured in crude membranes washed with EGTA (supplemental Fig. S3, A, C, E, and G, respectively), or in crude membranes incubated with M13 prior to the EGTA wash (supplemental Fig. S3, B, D, F, and H,  respectively), with a range of [Ca 2ϩ ] free in the absence or presence of different forms of exogenous CaM (1 M). Although the activities without Ca 2ϩ or exogenous CaM were reduced due to the removal of endogenous CaM, incubation of membranes with M13 did not alter their regulation by Ca 2ϩ , when 1 M CaM WT was present. AC activities were also measured with increasing concentrations of CaM in the presence of 1 M free Ca 2ϩ for AC8 and AC8M1 (Fig. 3, A-D) or 0.3 M free Ca 2ϩ for AC1 (Fig. 3, E  and F). CaM WT stimulated AC activities in a dose-dependent manner, and CaM 1234 (even up to 10 M) did not affect activities.
AC8 was stimulated 1.8-fold with 10 M CaM WT , and the stimulation was increased to 4-fold with an EC 50 of 0.12 M, when membranes were incubated with M13 prior to the assay (n ϭ 3, p Ͻ 0.001, Fig. 3B and supplemental Fig. S4B). Ca 2ϩ stimulation of AC8 after EGTA wash was unchanged in the presence of CaM 34 or CaM 1234 , but significantly inhibited when CaM 12 was present (supplemental Fig. S3A), as shown previously (26). This inhibition was concentration-dependent; AC8 showed a 25-30% reduction with 10 M CaM 12 in the absence of M13 pretreatment (n ϭ 4, p Ͻ 0.001, Fig.  3A and supplemental Fig. S4A). However, when endogenous CaM was completely removed from AC8 by M13 incubation, the inhibition by CaM 12 was no longer observed (supplemental Fig. S3B), even when [CaM 12 ] was increased to 10 M ( Fig. 3B and supplemental Fig. S4B), demonstrating that the inhibition was due to competition with endogenous CaM. Interestingly, AC8 displayed a slight stimulation by Ca 2ϩ in the presence of 1 M CaM 34 (ϳ1.7-fold), after endogenous CaM was removed (supplemental Fig.  S3B). To investigate the possibility that CaM 34 can stimulate AC8, we increased the concentration up to 10 M and found this further activated AC8 (Fig. 3A and  supplemental Fig. S4A). This stimulation was enhanced from 1.3-to 2.3-fold when endogenous CaM was removed by M13 washing (n ϭ 3, p Ͻ 0.001, Fig. 3B and supplemental Fig. S4B).
As shown previously, AC8M1 was stimulated by Ca 2ϩ only slightly in the absence of CaM (supplemental Fig. S3C (26)). AC8M1 did not show Ca 2ϩ stimulation to any great extent in the presence of 1 M CaM mutants (supplemental Fig. S3C), which reaffirms the idea that CaM 12 competes with endogenous CaM at the N terminus in intact AC8. When endogenous CaM was removed by incubation with M13, AC8M1 was regulated by Ca 2ϩ /CaM very similarly to AC8; AC8M1 was fully stimulated by Ca 2ϩ with CaM WT , stimulated ϳ1.7-fold with CaM 34 , and not stimulated by Ca 2ϩ in the absence of exogenous CaM, or in the presence of CaM 12 or CaM 1234 (supplemental Fig. S3D). The CaM regulation of AC8M1 was comparable to that of AC8 with M13 pretreatment, although the EC 50 value for CaM WT was slightly increased (0.27 M, n ϭ 3, Fig. 3C and supplemental Fig. S4C). These data demonstrate that the endogenous CaM, which is resistant to EGTA washing, is tethered at the N terminus of AC8, and a high concentration of CaM 34 activates AC8 through the C terminus. Although -fold stimulation and the EC 50 value for CaM WT were increased, the overall profile of regulation by CaM of AC8M1 activity after M13 pretreatment was similar to that without M13 (Fig. 3D and  supplemental Fig. S4D). The CaM 34 -mediated Ca 2ϩ stimulation of AC8 and AC8M1 suggests that the N lobe of CaM is more important than the C lobe in the binding to and activation of AC8.
The effect of CaM mutants on Ca 2ϩ stimulation of AC1 was quite different. None of the CaM mutants affected the Ca 2ϩ stimulation of AC1. AC1 was stimulated by Ca 2ϩ in the absence of exogenous CaM by ϳ1.2-fold, and this stimulation was neither enhanced nor reduced by CaM mutants (supplemental Fig.  S3E). The Ca 2ϩ inhibition of AC1 observed in the absence of exogenous CaM after M13 incubation was again unaffected by CaM mutants (supplemental Fig. S3F). Unlike AC8, neither CaM 12 nor CaM 34 affected AC1 activity compared with CaM 1234 , even when endogenous CaM was washed away (Fig. 3,  E and F, and supplemental Fig. S4, E and F). CaM WT stimulated AC1 activity 1.8-fold with an EC 50 value of 0.27 M (n ϭ 3, Fig.  3F and supplemental Fig. S4F). The apparent stimulation of AC1 by 10 M CaM 12 when membranes were incubated with M13 was not statistically significantly different compared with the activity by 10 M CaM 1234 or in the absence of exogenous CaM (Fig. 3F and supplemental Fig. S4F).
The activity of AC1F503A was neither stimulated nor inhibited by Ca 2ϩ in the presence of CaM WT ; however, it showed inhibition by Ca 2ϩ in the absence of CaM and in the presence of CaM mutants (supplemental Fig. S3G). This profile was unchanged by M13 incubation (supplemental Fig. S3H). These data establish that, although the residue Phe-503 is important for AC1 to be fully stimulated by Ca 2ϩ /CaM, AC1F503A can still interact with and be activated weakly by fully loaded CaM.
To prove that the mutated C1b domain of AC1 could still interact with CaM, pulldown experiments were conducted between GST-AC1C1b, GST-AC1C1b F503A , and CaM. In these experiments both constructs bound CaM, although the interaction was weaker with the point mutation in the C1b domain (supplemental Fig. S5). The weak activation of AC1F503A by CaM WT and no other CaM mutant (supplemental Fig. S3, G and H) supports the mechanism whereby only fully liganded CaM activates AC1. Fig. 4A summarizes the effect of CaM on Ca 2ϩ stimulation of AC8, AC8M1, AC1, and ACF503A, compared with that in the absence of exogenous CaM, and schematic diagrams show interactions between different forms of CaM and AC8 (Fig. 4B) and AC1 (Fig. 4C). These results demonstrate a distinct mechanism of CaM regulation in AC8 and AC1. Although both AC8 and AC1 require Ca 2ϩ -saturated CaM for stimulation, AC8, but not AC1, is partially activated by CaM with two Ca 2ϩ bound at the N lobe.
The effect of CaM mutants on Ca 2ϩ stimulation of AC8 and AC1 was then compared in vivo. An empty vector (pcD), wildtype, or mutant forms of CaM were transfected in cells expressing AC8 or AC1, and cAMP accumulation in response to CCE was monitored in cell populations. CaM WT did not enhance the Ca 2ϩ stimulation of AC8 or AC1, presumably because endogenous cellular CaM levels were adequate (56) to stimulate the overexpressed ACs. In vivo measurements also showed that unlike AC8, which was strongly inhibited by CaM 12 (Fig. 5A), AC1 activity was not affected by CaM mutants (Fig. 5B). Along with the in vitro data, these in vivo data underscore a clear difference in the regulation by CaM of AC8 and AC1.
Regulation of AC1 and AC8 by Physiological [Ca 2ϩ ] i Induced by TG and CCh-The biochemical differences in the regulation of AC1 and AC8 noted up to this point hint that these differences may be exploited in a physiological context. Consequently we compared the temporal responsiveness of the two enzymes to physiological elevations in [Ca 2ϩ ] i . Inositol 1,4,5trisphosphate-dependent Ca 2ϩ increase caused by the muscarinic receptor agonist, CCh (1 mM), stimulates AC8 by 1.7-fold, which is significant, but much lower than the stimulation by CCE (21). We wished to examine whether AC1 showed any selectivity toward regulation by CCE. In Ca 2ϩ -free media, various concentrations of CCh were added to release Ca 2ϩ from intracellular stores in HEK 293 cells. Values of 80 -300 nM [Ca 2ϩ ] i were achieved (Fig. 6A). The Ca 2ϩ influxes caused by CCh in cells expressing AC1 and AC8 were comparable to those in untransfected cells (data not shown). In the same procedure, cAMP accumulation in HEK 293 cells expressing AC1 or AC8 was measured over a 1-min period beginning with the addition of 10 M FSK and various concentration of CCh in the presence of 100 M EGTA and 100 M IBMX. Although AC8 was stimulated robustly by CCE (Fig. 1D), stimulation by CChmediated Ca 2ϩ release was only up to 35% of the stimulation brought about by CCE resulting from 4 mM Ca 2ϩ (Fig. 6B), which gave a comparable increase in [Ca 2ϩ ] i . (The stimulation caused by CCh was likely due to Ca 2ϩ release, because this stimulation was abolished when the Ca 2ϩ store was depleted by TG prior to the addition of CCh.) This demonstrates that AC8 is stimulated by CCE much more efficiently than by CCh-mediated Ca 2ϩ release. By contrast, AC1 showed a 2-fold stimulation by CCh-mediated Ca 2ϩ release, which was the same as the maximal stimulation achieved by CCE (Fig. 6B). Thus AC1 shows no particular selectivity for CCE over Ca 2ϩ release. The kinetics of cAMP accumulation was very similar in AC1 and AC8; the maximum stimulation by CCh was achieved in both by 20 M CCh with a half-maximum concentration of ϳ6 M CCh (Fig. 6B).
AC1 showed a higher Ca 2ϩ sensitivity and lower -fold stimulation than AC8, when stimulated by TG-evoked CCE for 1 min (Fig. 1D). To compare the kinetics of their regulation at more physiological Ca 2ϩ concentrations, we examined the time course of cAMP accumulation in response to a sub-maximal Ca 2ϩ rise evoked by TG or CCh. In the presence of 1 mM Ca 2ϩ , 100 nM TG and 10 M CCh gave a comparable Ca 2ϩ increase (up to ϳ200 nM; Fig. 6, C and D, respectively), which were caused by both Ca 2ϩ release from the internal stores and CCE (57). CCh produced a rapid inositol 1,4,5-trisphosphatedependent Ca 2ϩ increase followed by sustained CCE, whereas TG gave a slower, sustained Ca 2ϩ rise due to its passive depletion of ER stores and subsequent CCE (Fig. 6, D and C, respectively). We then applied these conditions to measure cAMP generation in AC1 and AC8 cell populations over 200 s after the addition of TG or CCh. Cells were pretreated with low dose of FSK (100 nM) for 60 s to enhance AC activity. AC1 and AC8 responded similarly to TG-induced Ca 2ϩ increase; they were both stimulated about 3-fold and showed a similar slow, gradual increase in cAMP (Fig. 6E). By contrast, the [cAMP] i rise induced by CCh activated AC1 and AC8 differently. AC8 responded much more rapidly than AC1 (Fig. 6F), although both responses were faster than those evoked by TG. Ca 2ϩ release mediated by CCh stimulated AC1 fully but not AC8 (Fig. 6B), demonstrating that the faster response of AC8 was not due to enhanced responsiveness of AC8 to CCh-mediated Ca 2ϩ release. This result demonstrated, perhaps paradoxically, that AC1, the more Ca 2ϩ -sensitive AC isoform, was activated more slowly than AC8.
To examine the kinetics of the regulation in a continuous and more discerning setting, single cell experiments were performed. First, [Ca 2ϩ ] i changes were established in cells loaded with fura-2/AM. In the presence of 1 mM Ca 2ϩ , 100 nM TG or 10 M CCh was added at 2 min to permit an increase in [Ca 2ϩ ] i ; this increase was reduced when the buffer was switched to Ca 2ϩ -free at 7 min. Cells treated with TG showed smooth, homogenous [Ca 2ϩ ] i responses (data are representative of 10 cells from 6 coverslips (n ϭ 600), Fig. 7A), whereas [Ca 2ϩ ] i responses following CCh addition were heterogeneous (data are representative of 10 cells from 12 coverslips (n ϭ 1200), Fig.  7B). Addition of CCh caused a rapid increase in [Ca 2ϩ ] i , which was observed in all cells followed by a plateau phase in ϳ65% of cells, and ϳ35% cells displayed Ca 2ϩ oscillations, as shown previously (43). The [Ca 2ϩ ] i responses in cells expressing AC1 or AC8 did not differ from those in untransfected HEK 293 cells (data not shown). Parallel experiments were carried out to assess cAMP levels in HEK 293 cells expressing AC1 and AC8, using a genetically encoded cAMP sensor, Epac1-camps. Because of its high temporal resolution, the Epac1-camps fluorescent probe has been used as a sensor for rapid changes in intracellular cAMP levels caused by phosphodiesterase activity (58,59) and AC activity (43,60). Epac1-camps was transiently transfected into cells stably expressing AC1 or AC8, and the changes in CFP/YFP emission ratio of the probe were measured. As in cell population measurements, AC1 and AC8 were primed by a low dose of FSK (100 nM), which did not change the fluorescent signals in buffer containing 1 mM Ca 2ϩ . AC1 and AC8 responded analogously, and slowly, to TG-induced Ca 2ϩ rises (Fig. 7, C and E). The initial rates of increase in cAMP upon the addition of TG did not differ significantly with AC1 and AC8 (0.041 Ϯ 0.003 min Ϫ1 (n ϭ 39) and 0.054 Ϯ 0.010 min Ϫ1 (n ϭ 24), respectively, Fig. 7G). Upon addition of 10 M CCh in 1 mM Ca 2ϩ buffer, cAMP levels in both AC1 and AC8 expressing cells increased rapidly, compared with TG. However, AC8 cells showed a heterogeneous, often oscillatory cAMP response upon CCh addition (Fig. 7F), whereas by contrast, AC1 cells all exhibited a steady increase in [cAMP] i , which was sustained until the buffer was switched to Ca 2ϩ -free buffer (Fig. 7D). This kinetics reflects the slower activation of AC1. The initial rate of increase in cAMP of AC1 was 0.082 Ϯ 0.007 min Ϫ1 (n ϭ 65), whereas that of AC8 was 0.313 Ϯ 0.022 min Ϫ1 (n ϭ 71), upon the addition of CCh (Fig. 7G). These single cell data confirmed that the rise in cAMP levels caused by CCh-mediated Ca 2ϩ rise was significantly faster in cells expressing AC8 compared with AC1.

DISCUSSION
Behavioral, developmental, and electrophysiological studies in mice have clearly established that AC1 and AC8 play distinct physiological roles; however, few biochemical or cell biological insights are available to provide potential mechanisms for these differences. The expression levels of AC1 and AC8 vary depending on the developmental stage and region of the brain (19,20). AC1 seems to act both pre-synaptically (16,61,62) and post-synaptically (63,64). In mouse brain, AC1 is primarily localized at the postsynaptic density and AC8 at the presynaptic active zone (18,20). Conti et al. suggested that the presynaptic role of AC1 and postsynaptic role of AC8 might arise in extrasynaptic membranes, where they are both expressed (20). Furthermore, the expression level of AC8 in cerebellum decreased significantly from postnatal day 7 to 14, whereas that of AC1 was constant (20). The different localization and expression pattern may reflect the discrete roles of these enzymes, but their individual regulatory properties might also explain how they can be utilized in particular contexts. Consequently, we used HEK 293 cells as a model system to compare and assess the properties of these enzymes against a constant background.
AC1 and AC8 are both stimulated by Ca 2ϩ /CaM in vitro and CCE in vivo, but AC1 is significantly more sensitive to Ca 2ϩ ; thus AC1 could reach maximum stimulation with much lower concentrations of [Ca 2ϩ ] i , or lesser activation of Ca 2ϩ channels. In intact non-excitable cells, both AC1 and AC8 were not stimulated by Ca 2ϩ influx caused by ionophores, but they were stimulated robustly by CCE, due to an apparent co-localization with CCE channels (21). A component of the dependence of AC8 on CCE is its residence in lipid raft domains of the plasma membrane. The present study finds that AC1 is also targeted to lipid rafts, adding to the sense that all Ca 2ϩ -sensitive ACs are concentrated in such domains (65) and underscoring the regulatory importance of their localization in such environments. Another difference between AC1 and AC8 is that AC1 but not AC8 is maximally stimulated by CCh-mediated Ca 2ϩ release. This difference could reflect the higher sensitivity to Ca 2ϩ of AC1, which might in turn allow AC1 to be localized somewhat more distally from CCE channels than AC8. In this regard it is also notable that the N terminus of AC8 is required for its sensitivity to CCE, whereas this is not the case for AC1. Could this reflect a potential direct interaction between the N terminus of AC8, but not of AC1, and the CCE apparatus?
The CaM binding domains of AC1 and AC8, and their interactions with CaM are clearly distinct. AC8 has CaM binding domains at the N terminus (an amphipathic helix) and C2b domain (an IQ-like motif), whereas AC1 apparently has only one CaM binding domain (an amphipathic helix) at the C1b domain. The present study has clarified differences in the mechanism of activation by CaM of the two ACs and added to the literature on differential roles of the N and C lobes of CaM in regulating different CaM targets. CaM pre-associates with the N terminus of AC8 (26). Mutant CaM that could not bind Ca 2ϩ exerted no effect on wild-type AC8 activity. However, activity was decreased in the presence of an N-lobe mutant of CaM both in vitro and in vivo, which suggested an interaction between the C lobe of CaM and AC8. When endogenous CaM was removed by M13 incubation or deletion of the N terminus, inhibition was no longer observed, illustrating that CaM with two Ca 2ϩ bound at the C lobe competes with endogenous CaM at the N terminus of AC8. In elegant studies, Yue's group has shown that Ca 2ϩdependent facilitation of the P/Q Ca 2ϩ -channel opening is initiated by Ca 2ϩ binding to the C lobe of CaM, which detected primarily a local Ca 2ϩ rise; Ca 2ϩ inactivation was induced by the N lobe of CaM, which sensed a global Ca 2ϩ rise (32). The present study showed a distinct effect of the N lobe and the C lobe of CaM on Ca 2ϩ stimulation of AC8. Unlike the N-lobe mutant, a C-lobe mutant of CaM stimulated AC8 and AC8M1 activities, and the stimulation was increased when membranes were washed with M13, indicating that N-lobe Ca 2ϩ -occupied CaM interacts more weakly than fully liganded CaM, but it can bind at the C2b domain of AC8. In contrast to AC8, neither N-lobe nor C-lobe mutants of CaM affected AC1 activity. Thus, none of these partially Ca 2ϩ -occupied CaM states regulates AC1; therefore, unlike AC8, only fully loaded CaM regulates AC1. Thus, a fundamentally different mode of activation by Ca 2ϩ /CaM is suggested for the two enzymes.
Not only did the sensitivity of AC1 and AC8 to Ca 2ϩ differ, the kinetics of this regulation also diverged. Unexpectedly (given that AC1 was more sensitive to Ca 2ϩ ), the response of AC1 was significantly slower both in population and in single cell measurements. This kinetics might reflect the different mechanism of regulation by CaM of AC1 and AC8 noted above. Although the affinity of CaM for Ca 2ϩ is higher in the C lobe than the N lobe (55), Ca 2ϩ may bind the N lobe of CaM first due to its higher on-rate (and off-rate) for Ca 2ϩ binding (66,67), in response to an initial fast [Ca 2ϩ ] i rise, as would be observed nearer to a Ca 2ϩ -entry site. Indeed, a study by Johnson et al. showed that the N lobe, but not the C lobe of CaM is occupied maximally during a fast Ca 2ϩ transient (68). An interaction between partially liganded CaM and a target protein then increases the affinity of the other lobe for Ca 2ϩ and may decrease off-rates (69). Thus, because AC8 can interact with partially loaded CaM giving rise to such a cooperative mechanism, its speed of activation by Ca 2ϩ influx may be faster than AC1.
Single cell studies revealed further differences between AC8 and AC1. AC8 gives rise to oscillations in cAMP in response to Ca 2ϩ transients (43), but quite surprisingly AC1 activity yielded only a steady cAMP increase in response to CCh-evoked Ca 2ϩ oscillation. Two possible reasons for non-oscillatory response of AC1 are suggested; first, the "sub-maximal" Ca 2ϩ rise induced by 10 M CCh in the presence of 1 mM Ca 2ϩ is sufficient to activate maximally the more Ca 2ϩ -sensitive AC1 and, second, the frequency of Ca 2ϩ oscillations under this condition was too fast for AC1 to respond to, possibly reflecting a slower on/off-rate for Ca 2ϩ by CaM bound to AC1. Biophysical measurements of Ca 2ϩ dissociation rates from CaM in the absence or presence of peptides representing the CaM binding domains of AC1 or AC8 could clarify these possibilities. Whether the oscillatory behavior of AC8 or its faster response to rapid [Ca 2ϩ ] i rises would be of more significance in its neuronal setting remains to be addressed.
These real-time observations anticipate different functional roles for the two ACs. Earlier studies had shown that Ca 2ϩ transients yielded distinct transcriptional consequences in B lymphocytes than did equivalent but steady Ca 2ϩ rises (70). Later these authors showed that oscillations are more efficient at activating the nuclear transcription factor of activated T-cells than a steady Ca 2ϩ rise (71). They also showed that different transcription factors are activated depending on the frequency of Ca 2ϩ oscillations (71). The possibility that equivalent Ca 2ϩ transients can give rise to either cAMP transients or a steady rise in cAMP depending on the AC target, combined with the role of cAMP in mediating stimulation of CREB transcription factors (72), might anticipate different transcriptional consequences of Ca 2ϩ activation of the two enzymes.
Overall, a series of major differences arise between what might be considered AC homologues. The dissimilarity in sensitivity to Ca 2ϩ both in vitro and in vivo was a first observation; this coupled with a distinct relative sensitivity to Ca 2ϩ release versus Ca 2ϩ entry hinted at more significant mechanistic differences. This was confirmed when obviously disparate mechanisms of activation by Ca 2ϩ /CaM were observed, both in terms of where CaM binds to the ACs and in the different roles played by the two lobes of CaM. Finally, strikingly dissimilar kinetics in individual cell analysis in response to physiological elevation of [Ca 2ϩ ] i underscore how these regulatory differences can be translated into diverse outcomes for cellular cAMP, which in turn might have quite distinct results for cellular cAMP targets. One might also speculate that, because of their dissimilar amino acid sequences, especially at the N and C termini, the complement of putative binding partners of AC1 and AC8 could also vary, such that the creation and regulation of their immediate environment would be distinct. Thus this relatively simple regulatory motif of stimulation by Ca 2ϩ /CaM can be seen to provide the potential range of physiological roles for these ACs that are inferred from knockout studies.