HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (4) Citing Articles via Google Scholar Google Scholar Articles by Willoughby, D. Articles by Cooper, D. M. F. Search for Related Content PubMed PubMed Citation Articles by Willoughby, D. Articles by Cooper, D. M. F.

Organization and Ca²⁺ Regulation of Adenylyl Cyclases in cAMP Microdomains

Department of Pharmacology, University of Cambridge, Cambridge, United Kingdom

ABSTRACT

I. INTRODUCTION

II. ADENYLYL CYCLASES

A. General Structure of ACs

B. Are the ACs cAMP Exporters?

C. Voltage Sensitivity

D. Dimerization

E. Catalytic Core

F. AC Distribution

G. Soluble AC

III. TYPE-SPECIFIC REGULATION

A. G Proteins

1. Gsalpha

2. Gialpha

3. Gbetagamma

4. Selectivity in G protein subunit requirements

B. PKA

C. PKC

D. Calmodulin Kinase

E. Receptor Tyrosine Kinase

F. Calcineurin

G. RGS Proteins

IV. REGULATION BY CALCIUM IN VITRO

A. Stimulation by Ca2+

1. AC1

2. AC8

3. AC3

B. Inhibition by Ca2+

1. AC5 and AC6

V. REGULATION BY CALCIUM IN VIVO

A. Ca2+ Homeostasis in Nonexcitable Cells

B. Dependence of Ca2+-Sensitive ACs on CCE Over Release in Nonexcitable Cells

C. Dependence of Ca2+-Sensitive ACs on CCE Rather Than Other Forms of Ca2+ Entry in Nonexcitable Cells

1. Ionophore-mediated entry

2. Arachidonic acid-mediated entry

3. OAG-mediated entry

D. Close Apposition of ACs With CCE Channels

1. BAPTA versus EGTA

2. Measurements with aequorin

3. Comparisons between the efficacy of Ca2+, Ba2+, and Sr2+

4. Selective regulation of Ca2+-sensitive ACs by CCE in endogenous sources

E. Regulation by Calcium in Excitable Cells

F. What Is the Nature of the Ca2+ Signal to Which the ACs Respond In Vivo?

G. Role of CaM Recruitment in the Regulation of AC8 by Ca2+

VI. COMPARTMENTALIZATION OF THE ADENYLYL CYCLASES

A. Lipid Rafts

B. Lipid Rafts and ACs

C. Components of the Microdomain

1. NHE1

2. PDEs

3. Calmodulin

4. TRPC channels

VII. DYNAMIC AND LOCAL CHANGES IN cAMP

A. Methodologies for Single-Cell cAMP Measurements

1. PKA-based probes

2. Epac-based probes

3. CNGCs

B. Local Dynamics of cAMP Signals Detected With Real-Time Probes

1. Neuronal studies

2. Studies in cardiac myocytes

3. Studies in HEK293 cells

C. Oscillations in Intracellular cAMP

VIII. ORGANIZATION OF cAMP SIGNALING VIA DIRECT PROTEIN-PROTEIN INTERACTIONS

A. AKAP-Based Signaling Complexes

1. Perinuclear mAKAP/PDE4D3 signaling complex in cardiac myocytes

2. Subplasmalemmal gravin/PDE4D signaling complex in HEK293 cells

3. AKAP79 and AC5/6 interactions in cortical tissue and nonexcitable cells

4. AC and L-type Ca2+ channel-associated complex in rat forebrain and ventricular myocytes

B. Other AC-Associated Proteins

IX. PHYSIOLOGICAL SYSTEMS

A. Specialized Actions of Ca2+-Inhibitable AC6

1. Role of AC6 in cardiac tissue

2. Function of AC6 in endothelial cells

3. Differential expression of AC5 and AC6 in the kidney

B. Physiological Role of AC3 in Olfaction

C. Functions of the Ca2+/CaM-Stimulated ACs

1. Role in synaptic function

2. Role in the pancreas

X. CONCLUSIONS AND FUTURE ISSUES

GRANTS

ACKNOWLEDGMENTS

REFERENCES

	ABSTRACT

Top Next References

	I. INTRODUCTION

Top Previous Next References

 
cAMP is the prototypical second messenger, which impacts on every aspect of the life of the cell, from differentiation and development through to cell death. The converse is probably also true, i.e., every major cellular event will have direct or indirect consequences for cAMP signaling. Production of cAMP results from the activity of the adenylyl cyclase (AC) family of enzymes. Not so long ago, AC was viewed as a single entity that was regulated by hormones in a stimulatory or inhibitory manner through the mediation of stimulatory and inhibitory G proteins (159). The discovery of multiple AC isoforms (9 membrane-bound and 1 soluble) rapidly revealed that not only was this regulation by G proteins diverse, but also an array of regulatory properties were displayed that permitted numerous points of interaction with a range of major signaling pathways (373). This potential for diverse interplay with other key regulatory molecules is accompanied by an elegant spatiotemporal organization of cAMP signals to maintain tight control of cAMP-dependent events. Details are now emerging of directed targeting of the ACs to various microdomains of the cell and their close association with, or integration into, macromolecular complexes containing other essential signaling proteins and their modulators. These proteins include cAMP effectors such as protein kinase A (PKA) and specific cAMP phosphodiesterases (PDEs), which are themselves subject to refined targeting and modulation. A further level of sophistication comes from the fact that many of the ACs are regulated by intracellular Ca²⁺ changes, which can be highly complex in space and time. A major consequence of such elaborate microdomain organization, and the complexity of the signals that influence enzyme activity, is the likelihood that cAMP signals are not simple, and this may be an essential element of their regulatory influence.

To appreciate the questions and opportunities provided by current insights into cAMP signaling, this review first focuses on recent advances in our understanding of the structure and dynamic regulation of the ACs, with particular emphasis on the molecular and cellular specializations developed for the regulation of the Ca²⁺-sensitive ACs. We describe recently published data on the regulation of these ACs by Ca²⁺ in vitro and begin to assemble a picture to detail how their regulation in vivo depends upon the "context" in which the ACs are found. We will expand on what is known of the compartmentalization of ACs and cAMP signals to specific regions of the cell, and their association with numerous regulatory and scaffold proteins. The use of improved real-time biosensors to monitor cAMP changes in single live cells is described in some detail. We discuss how such methodological advances are beginning to open the door not only on the consequences, but also the mechanisms underlying such compartments of dynamic cAMP signals. We conclude this review by looking at some examples of discrete regulation and compartmentalization of the ACs in specific physiological situations and attempt to anticipate future directions in unraveling the diversity and complexity of cAMP signaling. The scope of this article touches on a large number of areas related to cAMP signaling, some of which have already been covered in recent reviews elsewhere. The reader will be directed to these sources when extra background seems appropriate.

	II. ADENYLYL CYCLASES

Top Previous Next References

 
A. General Structure of ACs

Following the purification, partial sequencing, and subsequent cloning of the first AC in 1989, a structure was revealed that comprised 12 transmembrane-spanning (TM) domains, two homologous apparent ATP-binding regions and extensive NH₂ and COOH termini (215). The remarkable complexity of the structure for AC1 went some way towards explaining previous difficulties associated with stabilizing and purifying the enzyme. Over subsequent years, eight more species of AC were cloned (21, 60, 144, 150, 205, 216, 309, 397, 422), all of which displayed this broad structure first described for AC1 (Fig. 1).

View larger version (19K): [in this window] [in a new window]   FIG. 1. Adenylyl cyclase (AC) structure. The ACs can be divided into 5 major domains: the NH2 terminus, the first transmembrane cluster (TM1, blue cylinders), the first cytoplasmic loop comprised of C1a (red) and C1b (black), the second transmembrane cluster (TM2, blue cylinders) with extracellular N-glycosylation sites, and the second cytoplasmic loop comprising C2a (orange) and C2b (black). C1a and C2a are highly conserved catalytic ATP-binding regions, which dimerize to form the catalytic site. C1b and C2b domains are less conserved. Comparison of this general AC structure with that of multidrug resistance transporters shows a remarkably similar topology in terms of the transmembrane architecture and cytosolic organization (see Ref. 112).

B. Are the ACs cAMP Exporters?

In the early days of cAMP research, the issue of exporting the cyclic nucleotide from the cell was considered by Davoren and Sutherland (108) as a possible means of controlling intracellular [cAMP]. A sizable body of studies provided evidence to confirm that cAMP extrusion did occur (122, 407) and that it was a unidirectional, ATP-dependent process with activity that correlated with intracellular [cAMP] (51). It was concluded that the energy-dependent cAMP extrusion, which was inhibited by probenecid [subsequently identified as a nonspecific anion transport inhibitor (112)], was independent of the activities of either ACs or PDEs (51, 122, 407). Indeed, in many cell types, extrusion of cAMP is only detectable after prolonged and high stimulation of AC activity. However, in the kidney cAMP extrusion is quite significant, to the extent that urinary cAMP levels largely reflect the activity of renal ACs. This situation is particularly notable as a measure of hyperparathyroidism in which parathyroid hormone exerts its effects on renal tubule Ca²⁺ reabsorption by stimulating AC activity (15, 103).

The potential role of the ACs in cAMP extrusion was revisited when the structure of the first AC was revealed. Krupinski and colleagues (215) noted its remarkable similarity to classic ATP-driven transporters and speculated that the 12 TM domains coupled with the two putative ATP binding sites (see Fig. 1) might reflect an ability of the AC to extrude cAMP. These studies were driven further by the knowledge that exported cAMP acted at extracellular receptors to mediate critical developmental functions in the slime mould Dictyostelium discoidium (164, 326, 327). However, deletion of a 12-TM domain AC (ACA) from Dictyostelium, with the retention of a simple AC enzyme (ACG) possessing only one TM span, did not perturb cAMP secretion, which again strongly suggested that the ACs themselves were not involved in exporting cAMP (303). The possibility that ACs were involved in cAMP extrusion was further dissipated by the lack of selective inhibitor of cAMP export, coupled with the identification of a plethora of alternate transporters that were capable of extruding cAMP (for example, the multidrug resistance proteins, as recently reviewed by Refs. 71, 112, 178).

C. Voltage Sensitivity

The ACs could be considered superficially to resemble ion channels due to the complexity of their membrane-spanning organization. Although the mammalian ACs do not possess "signature" S4 voltage sensors (116) or a canonical potassium pore loop often associated with ion channels, there have been some grounds to support the possibility that the ACs could function as ion channels from studies on the unicellular organism Paramecium. Synthesis of cAMP increased upon membrane depolarization of Paramecium; in addition, a purified AC protein from Paramecium placed in artificial lipid bilayers conducted K⁺ (342). Subsequently, an AC from Paramecium was cloned that displayed a clear voltage-sensor pore loop in tandem with a cytosolic AC homology domain (399). This structure was also observed in a number of other lower organisms, such as Plasmodium and Tetrahymena (399).

The precedent set by the Paramecium study may have contributed to the enthusiasm for an apparent voltage-sensitive AC in rat cerebellar granule cells. A prolonged (30 min) depolarization of these cells, in the absence of extracellular Ca²⁺, led to an increase in cAMP accumulation (316). A later study showed that this effect was in fact due to Na⁺ entry through L-type Ca²⁺ channels, as a consequence of membrane depolarization;¹ the effect could be mimicked by Na⁺ entry through gramicidin-generated pores where the membrane potential had been dissipated (96). There have since been a number of cases where the depolarization of cells, in the absence of extracellular Ca²⁺, can bring about unexpected effects that are more typically associated with Ca²⁺ influx. For example, in astrocytoma cells, modest increases in extracellular [K⁺] can stimulate inositol 1,4,5-trisphosphate (IP₃)-mediated Ca²⁺ release from the endoplasmic reticulum (ER), independently of Ca²⁺ influx (304). The effect can be explained in part by depolarization-dependent activation of G_q-coupled purinergic receptors, but it is largely independent of changes in membrane potential (243, 244).

D. Dimerization

Evidence has been reviewed elsewhere that ACs may exist in higher-order states than simple monomers (91). In brief, early sedimentation and target size analysis studies indicated that dimerization or tetramerization of ACs could occur (276, 325). In the case of AC8, dimerization occurs as a result of interactions between the TM sections (Fig. 2).

View larger version (52K): [in this window] [in a new window]   FIG. 2. Dimerization of AC8 molecules via their transmembrane domains. AC8 associates internally by interactions between transmembrane clusters TM1 and TM2, or externally via interactions between two TM2 domains. [From Cooper and Crossthwaite (91).]

Further support for AC dimerization comes from kinetic modeling studies which suggest that dimerization is required for the activation of purified ACs by G_s (72). Although a growing body of evidence suggests that ACs can form dimers and that these dimers may be of regulatory significance, we cannot exclude the possibility that even higher-order associations may occur. Such propensity for oligomerization would provide further similarity between the ACs and the ABC transporters. Indeed, oligomerization is a level of organization seen in several of the ABC transporters, and it is considered to play an important role in their ER processing and membrane insertion (52). The similarity between ACs and ABC transporters might even be fancifully expanded to include heterooligomerization with other regulatory elements. For instance, an analogy with the sulfonylurea receptors (SUR1 and SUR2A/B) from the ABC transporter family can be considered. The SUR proteins, like the ACs, possess two ATP binding motifs (e.g., Refs. 52, 61, 255). In this case, a heteromultimeric protein assembly comprised of four regulatory SUR subunits and four inwardly rectifying ATP-activated K⁺ channel (K_ATP) subunits has been identified (13, 256). The structure suggests that SUR1 directly interacts with Kir6.2 (431). Both the SUR and K_ATP subunits possess ER retention motifs, which are complementarily obscured to ensure their coexpression at the plasma membrane. Adjacent SUR1 proteins appear to interact and form a large docking platform for cytosolic proteins, which can add to the complexity and function of this complex. We are inclined to speculate that the Ca²⁺-regulated ACs might form similar, large heteromultimeric complexes to provide an intimate association with Ca²⁺ entry channel elements.²

E. Catalytic Core

A remarkable and puzzling feature of the ACs noted upon their first structural characterization was their possession of two ATP binding domains.³ These domains are highly homologous and complementary both within single ACs and among different mammalian AC isoforms. When the domains were expressed separately in Escherichia coli and purified, they had to dimerize for full AC activity; upon their recombination, AC activity was revealed, which could be regulated by G_s and forskolin (376). Functional chimeric molecules can be formed between separate AC1 C1a and AC2 C2a domains, as well as between AC5 C1a and AC2 C2a (368, 376). Even an AC2 C2a dimer has been generated that displays some catalytic activity (376, 432). A peptide-linked, but soluble, AC1 C1a and AC2 C2a have also been created with full activity, when purified from E. coli (115, 368). Indeed, even in mammalian cells, two half AC molecules that comprise the respective first or second TM cassettes with their associated cytosolic domains, which lack activity when expressed on their own, can be coexpressed to display full activity (174). Such observations suggest great avidity for the halves of AC molecules, both in their TM and cytosolic domains. Seebacher et al. (345) had shown that for retention of functional activity, TM domains could not be swapped between ACs, even if the cognate catalytic domains were retained in chimeric constructs.

The independence of the cytosolic domains from the TM domains, at least in the neutral in vitro environment, allowed the crystal structure of the catalytic domain to be determined. The first structure was that of a C2a dimer (432), which was shortly followed by the structure of an AC2 C1a/AC5 C2a dimer (376). Although no linear sequence homology is observed, the overall structure is remarkably similar to the catalytic site of DNA polymerases. Thus, like DNA polymerases, ACs possess a -domain, which also possesses two aspartate residues at characteristic locations that coordinate two Mg²⁺. These mediate the attack of the 3'-OH on the -phosphate of ATP, promoting the cyclization of cAMP and release of PP_i. Both C1a and C2a domains possess this palm motif, but only the C1a domain possesses the catalytic aspartates (376, 377, 432).

Two metal binding sites are identified: one, "site A," is a tetrahedral coordination sphere, which comprises an oxygen from the -phosphate of ATP, a carboxylate oxygen from each of the aspartates, and a water molecule; the second metal binding site, "B," is octahedrally coordinated by an oxygen from each of the aspartates, an oxygen from each of the , , and , unesterified oxygens of ATP, and a backbone carbonyl group of a nearby isoleucine (I397). In the crystallization study, by way of confirming the liganding atoms proposed above, Zn²⁺ bound preferentially to site A, while Mn²⁺ bound to site B (377; see Fig. 3). Although Mg²⁺ is expected to be the physiologically relevant cation in these sites, Mn²⁺and Zn²⁺ are used in crystallization studies to provide a more stable representation of these states. Both Mn²⁺and Zn²⁺ have similar atomic radii (and identical charge) with differences in their binding properties presumably reflecting their ability to accommodate four and six (or eight) ligands, respectively.

View larger version (58K): [in this window] [in a new window]   FIG. 3. Crystal structure showing complexes of ATP analog inhibitors with the catalytic core of AC. A: the C1a:C2a catalytic core of AC. Forskolin (FSK) and ATP, which bind between the C1a (tan) and C2a (mauve) domains, are shown as stick models. The switch II helix of Gs, which forms much of the interface with AC, is shown in red. B: model of ATP bound to specific amino acids in the active site. Amino acids are labeled according to their position in canine type V AC for C1a and rat type II AC for C2a. C, left: the active site of the AC·LddATP·Mn complex. Superimposed is electron density from a 2.8-Å resolution FoMg – Fc omit map (blue wire cage). Structural elements donated by the C1a and C2a domains of AC are shown in tan and mauve, respectively. In the right panel(shown close up), the green and purple wire cages represent electron density contoured at 5 for 3.0 Å FoZn – FoMg and 2.8 Å FoMn – FoMg omit maps, respectively, demonstrating that Zn2+ preferentially binds at site A and Mn2+ at site B. D: the active site of AC·ATPS-Rp·Mn (left) and a close up view of its thiotriphosphate (right). Superimposed is electron density from a 3.0 Å Fo – Fc omit map contoured at 2.5. A 7 peak marks the position of metal B and is modeled as Mn2+. A 2 peak marks the position of metal A and is modeled as Mg2+, although it could also be a weakly bound Mn2+. [From Tesmer et al. (377).]

The initial cloning of AC1 and AC2 led to a flurry of AC cloning in the early 1990s, which resulted in the identification of nine mammalian species (21, 60, 144, 150, 192, 205, 215, 216, 309, 397, 422). They all showed the same gross topological design with highly conserved catalytic domains, displaying up to 65% amino acid sequence homology. However, significant differences in sequence were observed between the noncatalytic cytosolic domains, which later turned out to accommodate different regulatory features (Fig. 4). Cloning the same AC in different animal species also revealed differences in some unconserved regions. For example, the NH₂ terminus of the canine AC5 is 19 amino acids longer than that of the comparable murine or rabbit species.

View larger version (68K): [in this window] [in a new window]   FIG. 4. AC isoform sequences are presented in cylindrical schematic form and grouped according to their regulatory properties. ACs 1, 3, and 8 are stimulated by Ca2+ acting via calmodulin (see later); ACs 2, 4, and 7 are not susceptible to Ca2+ regulation; ACs 5 and 6 are directly inhibited by Ca2+; AC9 is unique in its calcineurin-mediated inhibition by Ca2+. The relative positions of distinct domains have been determined by alignment of all known ACs and the positions of sequences predicted to form membrane-spanning helices. Basic domains are indicated by color: NH2 terminus (Nt), green; TM1 and TM2, gray; C1a, yellow; C1b, red; C2a, light blue; C2b, dark blue. C1a and C1b comprise the first, and C2a and C2b form the second, large cytosolic domain. It can be clearly appreciated that although the catalytic domains are conserved in terms of size, significant differences occur between the less-conserved domains. [Modified from Hanoune et al. (181).]

G. Soluble AC

A soluble AC (sAC) was originally identified in seminiferous tubules and sperm from various animals (42). This enzyme preferentially utilizes MnATP, rather than MgATP, as substrate. Unlike membrane-bound ACs, it is insensitive to G protein regulation and displays a lower affinity for ATP (K_m 1 mM, compared with a K_m 100 µM for the membrane-bound ACs). The enzyme was purified and cloned from rat testis, and its catalytic domain was found to resemble that of ACs expressed in various microorganisms, such as cyanobacteria (53). Uniquely among ACs, the sAC enzyme is activated by bicarbonate ions in vivo and in vitro in a pH-independent manner (69) and may mediate the effect of bicarbonate ions on sperm motility. Indeed, if the sAC is genetically deleted from mice, their sperm mobility is inhibited and the animals are infertile (132). Although the initial isolation of sAC was from sperm, sAC is also expressed in other bicarbonate-responsive tissues such as the kidney, which suggests that bicarbonate regulation of cAMP signaling may play a more widespread role in biological systems (69), and perhaps, the modest formation of cAMP of which this AC is capable may be of significance within discrete cellular domains (438). One elegant demonstration of this possibility was recently published by Zippin et al. (439) who showed that the perinuclear pool of sAC may provide a pathway for CREB transcription following an increase in intracellular bicarbonate levels, such as might arise during metabolic activity.

	III. TYPE-SPECIFIC REGULATION

Top Previous Next References

 
The discovery of several AC isoforms to generate a family of AC enzymes was rapidly succeeded by evidence to reveal diversity in their regulation by G proteins, plus a multiplicity of regulatory options for individual AC molecules (373). Although the ACs can be loosely subclassified in terms of their responsiveness or not to Ca²⁺ (see Table 1 and sects. IV and V), an outline is first presented here of the regulatory options available to specific AC isoforms, beginning with the complex options utilized by G proteins.⁴

A. G Proteins

1. G_s

All of the cloned ACs can be stimulated by G_s (reviewed in Ref. 365). Although no gross differences in sensitivity between the different species are observed during in vitro reconstitution assays, some apparent differences have emerged when the ACs are expressed in cell lines and then stimulated by endogenous G_s-coupled hormone receptors (e.g., Ref. 422). However, it has been argued that selective expression of specific G protein-coupled receptors (GPCRs) and ACs in particular subcellular domains, or limitations in the amounts of AC relative to the G protein, could give rise to apparent differences in sensitivity to G_s stimulation that might not be borne out in extremely well-controlled experimental situations (365). This argument underscores a central point of this review, which is that it is the overall cellular context rather than the absolute affinities of proteins for each other that determine their regulation. Later in the review, such cellular context issues are explored in more detail, but for the moment our discussion centers on insights from in vitro experiments.

2. G_i

In vitro assays reveal clear and absolute differences in how, and whether, different AC species are regulated by G_i. AC1, AC3, AC5, AC6, AC8, and AC9 are all inhibited by G_i, whereas AC2, AC4, and AC7 are not (90, 180, 365; see Table 1). It is perhaps no coincidence that inhibitory receptors that couple to G_i inhibition of AC are absent from tissues where AC2, AC4, and AC7 are prominent, e.g., lung and liver (406).

3. G

G exerts type-specific effects on AC species. AC1 and AC8 are inhibited (361, 369, 370), whereas AC2, AC4, and AC7 are stimulated (365, 369). AC3, AC5, AC6, and AC9 are insensitive to direct regulation by G (309, 369). These effects are seen in vitro and in transfected cells upon stimulation of appropriate receptors. Thus, in HEK293 (human embryonic kidney) cells expressing AC7, dopamine via a coexpressed D2 receptor stimulates cAMP accumulation as a result of liberating -subunits from endogenous G_i proteins; however, when AC5 is expressed in these experiments, cAMP accumulation is inhibited by dopamine as a consequence of liberating G_i (423).

4. Selectivity in G protein subunit requirements

Given the plethora of G protein subunits, i.e., 15, 5, and 13, coupled with nine membrane-bound ACs, there is potential for an extremely large variety of combinations of G protein subunit and AC interactions, so that the issue of selectivity among potential partners arises (406). This subject again raises the conflict between in vitro and in vivo assessments of affinity versus compatibility. Typical of a rigorous in vitro approach was a study by Taussig et al. (374) in which a variety of purified G protein subunits were reconstituted with recombinant AC1, AC2, AC5, or AC6 that had been purified from Sf9 cell membranes expressing high levels of individual ACs. Concentration response curves were obtained and compared, and based on the findings, the following selectivities were observed: all four ACs were activated by G_s; AC2 was not inhibited by G_i or G_o but was stimulated by G; AC1 was inhibited by G, G_i, and G_o with selectivity of G > G_i > G_o; whilst AC5 and AC6 were unaffected by G_o but potently inhibited by G_i. Interestingly, there were no differences in the abilities of G_i-1, G_i-2, and G_i-3 to inhibit AC5, AC6, and AC1. The value of such studies is the fact that affinities are obtained in a reasonably well-defined milieu; however, it is not known whether such precise mixtures of components ever encounter each other in live cells. Deriving from that comment, an insightful, though still limited, approach would be to determine the ACs and G protein subunits that are expressed in a particular single cell. Although heroic, the exercise might be beneficial in at least a few situations to yield a sense of whether in vitro sensitivities were ever exploited, or whether other factors in an in vivo context changed the desirability of particular combinations. Such a study was performed for olfactory neurons and led to the discovery that the G protein G_olf must stimulate AC3 (199). These two proteins colocalize in the distal segments of olfactory cilia where they play an essential role in the early stages of olfactory transduction (252). Thus certainly in that context, whatever other options might have been preferable from a currently informed biochemical viewpoint, that mixture was selected during evolution. One early approach to assess the selectivity of ACs for specific G proteins in an in situ context was by the use of microinjected antisense oligonucleotides against individual G protein subunits in single-cell experiments, where modulation by a GPCR linked to AC modified the activity of L-type Ca²⁺ channels (238). Another approach has been the use of tissue-targeted transgenic mice (195). This study showed that G_i-2 mediated the effect of carbachol on isoprenaline stimulation of cardiac L-type Ca²⁺ currents. However, a number of variable factors, e.g., unexpected developmental abnormalities, the pluripotency of G proteins, and their involvement in processes other than AC regulation, as well as compensatory changes in other tissues render this approach difficult to apply lightly. A more recent, direct approach to identify protein interacting partners is the use of interference RNA (RNAi) to knock-down specific proteins of interest. The rationale is that if a particular response can be abrogated by the selective elimination of a particular protein, and ideally restored by expression of an ortholog of the knocked out protein, then that target protein is most likely involved in the original context.

Very recently, an extensive assessment of G protein interaction has been made using RNAi techniques. The authors, wishing to answer the apparently simple question of which G protein subunits were associated with particular AC regulation, performed selective RNAi targeted knockdowns of various G protein -, -, and -subunits. However, instead of merely checking that their intended targets were knocked down, they also examined the consequences of such knockdown on the expression of other G protein subunits and ACs, in addition to the effect on AC responsiveness (214). What they found was a dramatic interdependence and a far-ranging series of consequences for their knockdowns, not only in terms of compensatory or apparently unrelated changes in other G protein subunits, but also changes in cognate receptors and effectors. A striking example from their study illustrates the problem; simultaneous knockdown of G₁ and G₂ in HeLa cells resulted in a nontranscriptionally mediated increase in G₄, which was subsequently able to maintain G_s activation of two AC isoforms whose expression levels had also unexpectedly increased as a consequence of the knockdown. Thus, although severe impairment was anticipated, an enhanced response to agonist was actually encountered. This perturbation also resulted in a substantial decrease in G_i protein (despite a large increase in mRNA for G_i-1), with unknown effects on G subunits. Given the multiple roles of G proteins in numerous signaling pathways, cautionary measures are required to avoid naive application and interpretation of such approaches (214). Thus awareness and addressing of potential pitfalls by independent experimental strategies may be required to establish with any confidence the composition of specific signaling complexes, including those involving the ACs.

B. PKA

PKA-mediated phosphorylation inhibits the activity of AC5 and AC6 (30, 180). This effect was proposed to be part of the hormone-induced desensitization of AC activity (308) and was confirmed by the Ishikawa lab, which showed that PKA directly phosphorylates AC5, resulting in reduced enzyme activity (194). It was later shown that phosphorylation of AC6 by PKA, in vitro, mediated a similar decrease in activity of the AC in response to G_s or forskolin (68). Mutation of the serine at position 674 to an alanine in AC6 prevented the phosphorylation-dependent inhibition both in vitro and in vivo (28, 68). This amino acid sits within the C1b catalytic site of AC6 in a region involved in G_s stimulation. Other PKA phosphorylation sites are yet to be identified, although AC6 contains at least 14 putative sites for PKA interaction (68, 194). In AC5, the targeting of serine at position 676 by PKA to reduce AC activity has been demonstrated in a recent study revealing that PKA-dependent phosphorylation of the ACs is optimized by the localization of PKA in a signaling complex incorporating both AKAP79 (an A-kinase anchoring protein) and AC5/6 (25; see below).

C. PKC

PKC activation stimulates the G_i- and Ca²⁺-insensitive ACs (AC2, AC4, and AC7) (90, 338, 366) and inhibits AC9 (101). Initial reports were established by activation of PKC using phorbol esters. Thus all ACs are potentially subject to regulation as a consequence of activation of the phospholipase C (PLC) pathway, either via IP₃-mediated Ca²⁺ elevation (and subsequent regulation of Ca²⁺-sensitive ACs, see below) or via production of diacylglycerol (and subsequent PKC activation of Ca²⁺-insensitive ACs). However, AC5 and AC6 are potentially regulated by both Ca²⁺ and PKC (30, 90, 180). Although several studies provide support for PKC-regulated AC activity in vitro, the significance of this effect in the intact cell is not well established. The importance of similar regulation in vivo is, however, suggested by the fact that individual ACs are selective for specific PKC isoforms. For instance, work by Kawabe and colleagues (208, 209) revealed preferential stimulation of AC5 by PKC- over PKC-. In contrast, AC6 exhibited reduced activity following activation of numerous PKC isoforms (, , and ) in osteoblastic cells (76). Along similar lines, in human erythroleukemia cells, ethanol was found to potentiate AC7 activity through a PKC--mediated phosphorylation (277, 366).

A study by Levin and Reed (223) has identified a small unconserved region near the COOH terminus as the site of action of PKC-mediated upregulation of AC2. In contrast, Chern's group (218) demonstrated that NH₂-terminal deletion of AC6 (amino acids 1–86) or mutation of S10A reduced the PKC-mediated inhibition and phosphorylation of the AC when expressed in Sf-21 insect cells. Further studies showed that PKC-mediated inhibition of AC6 was ablated by mutating two of the three potential phosphorylation sites in the Nt and C1 regions (S10, S568, and S674) or a single mutation in the C2 region (T931) (227). More recent studies using HEK293 cells stably expressing AC6 demonstrate a potentiation of epidermal growth factor (EGF)-evoked AC activity when cells are treated with phorbol esters, which suggests an even more complex relationship, wherein a PKC-dependent upregulation of AC6 in the intact cell can be mediated via the direct interaction of AC6 with a protein downstream of PKC, Raf1 (28).

D. Calmodulin Kinase

AC3 seems unique in its sensitivity to calmodulin (CaM) kinase. Early work provided evidence that AC3 was inhibited by CaM kinase II in vivo (398), which yielded an additional link between Ca²⁺ and the ACs, alongside the well-described direct Ca²⁺- or Ca²⁺/CaM-mediated regulation of several AC isoforms (see sects. IV and V). In HEK293 cells stably expressing AC3, increases in intracellular Ca²⁺ concentration ([Ca²⁺]_i) induced by ionophore or carbachol inhibited glucagon- and isoprenaline-mediated AC activity. This effect was antagonized by an inhibitor of CaM kinase, while coexpression of constitutively activated CaM kinase II completely inhibited G_s-mediated AC3 activity (398). Mutation of a CaM kinase II consensus site (S1076 to A1076) in AC3 suggested direct phosphorylation of the AC (401). Later studies suggest that CaM kinase II inhibition of AC3 may contribute to the termination or adaptation of olfactory signaling (222, 402). This inhibition of AC3 by CaM kinase II is also thought to underlie the long-lasting inhibition of AC activity observed following Ca²⁺ release from ER stores in the A7r5 smooth muscle cell line (128).

E. Receptor Tyrosine Kinase

A number of reports of modulation of ACs by receptor tyrosine kinases (RTK) are mentioned below. However, it should be noted that in the life of a cell, the responsiveness to an RTK may occur in a very narrow time frame and/or a very discrete cellular domain, e.g., at a membrane ruffle, somewhere where movement or cell cycle stage-dependent changes are occurring, and a very local change in cAMP levels is of significance. Consequently, it should not be surprising if the effects that have been observed in the literature are rather modest, which may merely reflect the lack of temporal or contextual appropriateness for the proposed measurements, rather than the overall significance of the effect. In many of the quoted examples that follow, ACs have been expressed heterologously and are definitely out of context.

One elegant example of a contextual effect of a growth factor on AC is provided by the case of ephrin and AC1. The expression of ephrin-A5 (a membrane-bound guidance molecule) and its receptor (EphA) play a critical role in defining the protrusion of axonal fibers in the developing retina (63, 248). AC1 is necessary to enact a retraction response of the retinal axons to ephrin-A5 during the refinement of the retinotopic map (278). This requirement for AC1 in mediating the actions of ephrin-A5 was demonstrated in retinal axons from AC1 knockout mice. These axons maintained their profusion but had lost their selectivity in branching towards their targets and inhibitory response to ephrin (278).

EGF increases AC activity and elevates cAMP accumulation in cells expressing AC5, while not affecting overexpressed AC1, AC2, or AC6 (70). The EGF stimulation of AC5 required G_s, a finding that is consistent with previous data concerning EGF effects on cardiac AC activity (70, 271). In studies using recombinant AC5, such activation of the enzyme was found to be due to the phosphorylation of G_s, by EGF, at one or more tyrosine residues (306). The stimulation of AC5 by EGF is required for activation of a K_Ca1.1 channel in rat vascular smooth muscle and the subsequent upregulation of genes that are critical for cell proliferation (193). RTKs also modulate the activity of AC6. Insulin-like growth factor I (IGF-I) activation of RTK and tyrosine phosphatase inhibition (using sodium orthovanadate) has been shown to phosphorylate several serine residues within AC6 and increase enzyme activity. Mutation of either S603 and S608 or S744, S746, S750, and S754 (which occur in the C1 region) attenuated both the RTK-mediated enhancement of enzyme phosphorylation, as well as the sensitization of function (367). Studies showing that AC6 phosphorylation is similarly increased by Raf1 (a component of the ERK signaling pathway), and that AC6 and Raf1 coimmunoprecipitate, suggest that pathways downstream of the RTK may potentiate AC signaling (28, 30, 118, 367). This latter effect also accounts for the PKC-dependent upregulation of AC6 (28).

F. Calcineurin

The role of calcineurin in AC9 function has been the subject of some controversy, with suggestions of either inhibition of AC9 activity (11, 295, 296) or no effect of the Ca²⁺-dependent protein phosphatase on overexpressed AC9 activity (309). Work by Antoni and colleagues (10, 11) revealed inhibition of AC9 by the same range of intracellular Ca²⁺ changes that stimulated AC1. In the case of AC9, however, the Ca²⁺-dependent inhibition of cAMP production was clearly attenuated by calcineurin blockers. High AC9 mRNA levels in the brain and potential regulation of this AC by Ca²⁺/calcineurin is proposed to account for some of the diverse actions of the protein phosphatase in brain function (11, 66).

G. RGS Proteins

An example of the likely major gaps that still remain in our knowledge of AC signaling complexes are provided by the intracellular regulator of G protein signaling (RGS) proteins. RGS proteins were first identified as the GTPase-activating proteins (GAPs) for the G protein -subunits, G_i and G_q, but not for G_s. This class of proteins has since been found to regulate numerous GPCRs and their effector molecules, including the ACs (reviewed in Ref. 2). In olfactory epithelial membranes, odorant-stimulated cAMP production, which occurs via the G_s family member G_olf was reduced by the addition of recombinant RGS1, -2, and -3, with RGS2 having the largest inhibition (RGS4 and -5 had no effect). Similarly, Sf9 cells expressing AC3 displayed reduced cAMP production when stimulated with either forskolin or the nonhydrolyzable GTP analog guanosine 5'-O-(3-thiotriphosphate) (GTPS). Unexpectedly, RGS2 reduced odorant-elicited cAMP production, not by acting on G_olf but by directly inhibiting the activity of AC3, the predominant AC isoform in olfactory neurons (352). RGS2 also inhibited AC5 and AC6 when expressed in HEK293 cells, apparently via a direct interaction with the C1 domain (330, 333). Recent imaging studies in live cells reinforce the binding of RGS2 to a receptor/G protein/effector signaling complex to regulate AC activity (329). The full significance of RGS proteins in AC signaling remains to be determined.

	IV. REGULATION BY CALCIUM IN VITRO

Top Previous Next References

 
A. Stimulation by Ca²⁺

AC1 and AC8 are clearly stimulated by Ca²⁺ in a CaM-dependent manner. Almost coincident with the discovery of CaM as an activator of PDE (77, 201) came the finding that Ca²⁺/CaM could stimulate AC from brain membranes (48). Purification studies later showed that a Ca²⁺-stimulable form of AC could be separated from a Ca²⁺-insensitive form (404). Cloning and expression of AC1 from brain (215) led to the demonstration that this species was potently stimulated by Ca²⁺/CaM in membranes (370). AC8 was discovered a few years later, as part of a PCR screening for the diversity of ACs that existed in mammalian cDNA libraries (216). Because of the abundance of AC8 mRNA in brain, Ca²⁺ simulation was also explored. The expressed protein was Ca²⁺/CaM stimulable, although with a lower sensitivity to Ca²⁺ than AC1 (60). Interestingly, the identification of a second Ca²⁺/CaM-stimulated AC in brain had been anticipated, since the hypothalamus displayed high levels of Ca²⁺-stimulated AC activity but no AC1 mRNA (263). AC8 turned out to be expressed well in the hypothalamus (60).

1. AC1

AC1 is stimulated in a CaM-dependent manner by Ca²⁺ with an apparent K_d of 0.1 µM (136, 418). An examination of the linear amino acid sequence of AC1 revealed a number of potential CaM binding sites of the "classical" amphipathic helix design. Indeed, dansylated CaM bound to synthetic peptides corresponding to these sequences (390). A further study showed that mutation of AC1 in one putative CaM binding site (F503A) eliminated the Ca²⁺ stimulation of the enzyme in vitro (418). This binding site is located close to the catalytic domain of AC1, within the C1b domain; however, information is not yet available on how the activation occurs.

2. AC8

AC8 is stimulated by Ca²⁺/CaM, with an apparent K_d of 0.5 µM (60, 136). Examination of the amino acid sequence of AC8 suggested two types of candidate CaM binding sites: an amphipathic helix at the NH₂ terminus and an apparent IQ motif at the COOH terminus. Deletion, mutagenesis, and pull-down studies revealed that both potential sites were utilized (173, 354).

An intriguing mechanism for Ca²⁺/CaM regulation of AC8 has now been suggested, as depicted in Figure 5. A common mode of enzyme activation by CaM is the relief of autoinhibition (185); that is, under basal conditions, the catalytic core of an enzyme is sterically hindered by a regulatory domain within the enzyme, which contains a pseudo-substrate sequence and a CaM binding domain (CaMBD). CaM binding induces a conformational change within the enzyme that subsequently removes the inhibition, exposing the catalytic site to the substrate. In many cases, such as myosin light-chain kinase, CaM kinase II, and the plasma membrane Ca²⁺-ATPase (PMCA), CaM acts through a single CaMBD in the target enzyme, and these enzymes are not associated with CaM when inactive (4, 62, 80, 179, 250). In contrast, AC8 presents a variation on this theme, as it is apparently autoinhibited at resting Ca²⁺, an effect that can be relieved upon CaM interaction with the COOH terminus of AC8. However, the CaM utilized is not derived directly from the cytosol. The CaMBD at the NH₂ terminus of AC8 appears to recruit CaM before activation. The preassociated CaM can then relieve autoinhibition, following a stimulating elevation in [Ca²⁺]_i and subsequent binding of a fully liganded CaM to the COOH-terminal CaMBD. Although several proteins, particularly ion channels, share this ability to preassociate with CaM, preassociation has not previously been linked to relief of autoinhibition by Ca²⁺. AC8, therefore, represents a novel mode of activation, combining two established CaM interactions via coordination of two CaMBDs (351).

View larger version (29K): [in this window] [in a new window]   FIG. 5. Model of Ca2+/calmodulin (CaM) activation of AC8. The domains of AC8 are indicated as follows: transmembrane domains (light blue), NH2 terminus (green), C1a (red), C1b (orange), C2a (purple), and C2b (blue). The lobes of CaM are either Ca2+ free (white arc) or Ca2+ loaded (orange circle). Top panel: under resting Ca2+ conditions, CaM is partially liganded, loaded at the C lobe, which binds to the NH2-terminal amphipathic helix CaM binding domain (CaMBD) of AC8. Middle panel: transition state following Ca2+ entry. Ca2+ binds to the N lobe of CaM, which subsequently binds to the COOH-terminal IQ-like CaMBD. Bottom panel: activated AC8. Both lobes of Ca2+/CaM bind to the COOH-terminal CaMBD triggering relief of autoinhibition.

The situation with AC3 is less clear. AC3 is grouped with the Ca²⁺/CaM-stimulated ACs because of one early report stating that it could be stimulated by high (submillimolar) concentrations of Ca²⁺ in the presence of forskolin or GppNHP, a nonhydrolyzable GTP analog, in a CaM-dependent manner (82). As AC3 had been cloned from an olfactory cDNA library and was expressed in olfactory neurons, and AC1 was found in central neurons, the possibility was entertained that AC3 would be similarly stimulated by Ca²⁺/CaM. However, follow-up studies suggested that AC3 was either inhibited directly by Ca²⁺ or inhibited via CaM kinase II (393). In one direct comparison between AC1, AC3, and AC8 expressed in HEK293 cells, only AC3 was not stimulated by physiologically relevant [Ca²⁺] (136). No follow-up study since that time has suggested any explanation for the initial reported stimulatory Ca²⁺ effect. Unlike AC1 and AC8, there is no experimental evidence, or likely amino acid sequence, indicating that AC3 binds CaM. In terms of sequence, AC3 is an outlier from all of the other AC isoforms (216, 365), and it may be appropriate to place it in a separate category, where its robust inhibition by CaM kinase may be the major effect of Ca²⁺ (see sect. IIID).

B. Inhibition by Ca²⁺

All ACs are inhibited by supramicromolar concentrations of Ca²⁺(Fig. 6). This is most likely due to Ca²⁺ competing for what one presumes to be a low-affinity "site A" metal binding site in the catalytic domain (see sect. IIE). Although all ACs are inhibited by very high [Ca²⁺], only two (AC5 and AC6) are inhibited by what are considered more physiological Ca²⁺ changes.

View larger version (10K): [in this window] [in a new window]   FIG. 6. Response of the ACs to Ca2+. Representation of the range of responses evoked by Ca2+ in tissues expressing Ca2+-stimulable, Ca2+-insensitive, and Ca2+-inhibitable ACs. Comparisons are made in the presence and absence of exogenous CaM. Note that regardless of the response to Ca2+ at submicromolar concentrations, all classes of AC exhibit inhibition at supramicromolar Ca2+ concentrations. [Modified from Cooper (90).]

Although these ACs differ significantly from each other in their NH₂-terminal domains, they are strikingly similar in their catalytic domains sharing 93% amino acid identity (365). As mentioned above, although AC activity from all tissues displays a low-affinity inhibition by Ca²⁺, early studies revealed that tissues such as cardiac, renal, anterior pituitary, and various cell lines also displayed a high-affinity, submicromolar inhibition by Ca²⁺ (49, 58, 87). Due to specific interest in determining the nature of such high-affinity inhibition by Ca²⁺, AC6 was first cloned from NCB-20 cells, a cell line in which the inhibitory process had been characterized in some detail (39, 422). At the same time, the potential therapeutic interest of cardiac AC triggered the cloning of AC5 from a cardiac cDNA library (192).⁵ AC5 mRNA was expressed at high levels in cardiac and striatal tissue. AC6 was also expressed in these tissues, although at lower mRNA levels, but it was also more widely distributed in other cell types including renal, endothelial, and blood cells (216). Expression of AC5 and AC6 in vitro revealed the expected high-affinity inhibition by Ca²⁺ (176, 188, 422), displaying a K_d for Ca²⁺ of 0.2 µM.

Unlike AC1 and AC8, high-affinity inhibition by Ca²⁺ does not require readily dissociable CaM. In addition, no readily identifiable motif such as an EF hand is displayed anywhere within the sequence of these Ca²⁺-inhibitable ACs. A kinetic characterization of high-affinity inhibition indicated that it was a competition by Ca²⁺ for the activation of AC5 and AC6 by Mg²⁺, which was intrinsic to the catalytic process (176). Thus inhibition was maximal at low levels of activation, but antagonized by higher levels of activation. A more detailed structural study used mutants that showed a range of susceptibilities to activation by Mg²⁺; the efficacy of Ca²⁺ at causing inhibition was a direct reflection of their susceptibility to be activated. Thus poorly activated mutant ACs were poorly inhibited by Ca²⁺, even though Ca²⁺ still competed for the activation. In addition, chimeras were constructed between the Ca²⁺-inhibitable AC5 and the Ca²⁺-insensitive AC2. Fully active ACs were generated whose ability to be inhibited by low [Ca²⁺] relied on them retaining the C1a catalytic domain of AC5, rather than the analogous domain of AC2 (188).

From the crystal structure of AC which is available, and which fortuitously happens to utilize the catalytic C1a domain of AC5, it is possible to speculate that the site of high-affinity inhibition by Ca²⁺ is the "B" site described earlier. Although no EF hand is shown in the catalytic domain, the octahedral coordination capability of the "B" site aspartates and the phosphoryl oxygen can readily accommodate, and even select for, the eight liganding form of Ca²⁺ (as it readily binds Mn²⁺) so it would not be unexpected that a structure might be obtained containing Ca²⁺ at that site (see Fig. 7).

View larger version (48K): [in this window] [in a new window]   FIG. 7. Structure of the catalytic domain of AC showing sites for high-affinity Ca2+ inhibition. Left panel: complex of AC5C1a and AC2C2a showing forskolin binding site and active site of AC chimera. Forskolin, shown in red, binds in the cleft that contains the active site. The ATP analog [2',3'-dideoxy 5'-ATP (DAD), in yellow] and two Mg2+ (represented as orange spheres) are bound in the AC active site. Right panel: a ribbon representation of the active site and location of the four active site mutations. The two Mg2+ are again displayed as orange spheres, whereas DAD and the amino acids that are mutated are shown in stick representation. Each residue is colored by atom (carbon, yellow; nitrogen, blue; oxygen, red; sulfur, pink). Each of the four mutations occurs proximate to the Mg2+ binding sites. Cys-441 and Tyr-442 are the two residues just after Asp-440, which is of prime importance to Mg2+ binding and AC activity. The Arg-434 and Phe-423 are located more distally from the active site. Both residues contact Tyr-442, suggesting that mutations of these amino acids might transmit their effects on Mg2+ activation through this residue. [From Hu et al. (188).]

	V. REGULATION BY CALCIUM IN VIVO

Top Previous Next