Mammalian ACs (adenylyl cyclases) are integrating effector molecules in signal transduction regulated by a plethora of molecules in either an additive, synergistic or antagonistic manner. Out of nine different isoforms, each AC subtype uses an individual set of regulators. In the present study, we have used chimaeric constructs, point mutations and peptide competition studies with ACs to show a common mechanism of multiple contact sites for the regulatory molecules Gβγ and calmodulin. Despite their chemical, structural and functional variety and different target motifs on AC, Gβγ and calmodulin share a two-site-interaction mechanism with Gαs and forskolin to modulate AC activity. Forskolin and Gαs are known to interact with both cytosolic domains of AC, from inside the catalytic cleft as well as at the periphery. An individual interaction site located at C1 of the specifically regulated AC subtype had been ascribed for both Gβγ and calmodulin. In the present study we now show for these two regulators of AC that a second isoform- and regulator-specific contact site in C2 is necessary to render enzyme activity susceptible to Gβγ or calmodulin modulation. In addition to the PFAHL motif in C1b of ACII, Gβγ contacts the KF loop in C2, whereas calmodulin requires not only the Ca2+-independent AC28 region in C1b but also a Ca2+-dependent domain in C2a of ACI containing the VLG loop to stimulate this AC isoform.
- binding region
- heterotrimeric G-protein
- signal transfer region
ACs (adenylyl cyclases) catalyse the conversion of ATP into the universal second messenger cAMP. Class III ACs comprise a family of structurally similar enzymes  with a catalytic centre composed of two pseudosymmetric domains, C1 and C2. Mammalian ACs contain those domains on a single polypeptide chain that is folded into two membrane regions each built up by six transmembrane helices; C1 follows transmembrane region M1 and preceds transmembrane region M2; C2 follows transmembrane region M2 (Figure 1A). On the basis of sequence similarity both cytosolic domains C1 and C2 comprise subdomains Ca and Cb. The C1a subdomain shares approx. 60% identity with C2a at the amino acid level, and both subdomains heterodimerize to form the pseudosymmetrical catalytic core . Mammalian ACs are represented by at least nine different isoforms (ACI–ACIX) that have been cloned and analysed . They are grouped into three subclasses  according to their regulatory molecules: ACI represents the prototype of CaM (calmodulin)-stimulated ACs, ACII belongs to the subgroup of ACs that are stimulated by Gβγ (Gβγ complex of the heterotrimeric G-proteins), and ACV and ACVI are inhibited by submicromolar concentrations of Ca2+. Gαs (α subunit of the stimulatory heterotrimeric G-protein) and forskolin are common stimulators of all AC subtypes, whereas inhibition by Gαi (α subunit of the inhibitory heterotrimeric G-protein) is restricted to ACI as well as ACV and ACVI .
Based on crystal structures of the catalytic domains of ACs, binding sites for forskolin and Gαs are known. One molecule of forskolin binds to the C1+C2 heterodimer  at sites that are distinct from that for Gαs . Contacts between forskolin and both catalytic halves of AC occur at multiple amino acids (Lys896, Ile940, Gly941 and Ser942 in C2 of ACII; Phe394, Trp507, Val511 and Tyr443 in C1 of ACV). When bound to the domain interface, forskolin stabilizes the interaction between C1 and C2. Two structural elements of Gαs form the interface, primarily through contacts with C2. The most prominent interaction is the insertion of the Gαs switch II helix (residues 225–240) into the groove formed by α2′ and the α3′–β4′ loop of AC. The second contact surface is formed by the α3–β5 loop of Gαs, which interacts with both C1 and C2 .
Obviously, domain association is a prerequisite for AC activity. However, C1+C2 assembly alone is clearly not sufficient to achieve the high level of AC activity displayed in the presence of forskolin, Gαs or both activators . Hence, it is assumed that the binding of forskolin or Gαs facilitates cAMP synthesis by altering the conformation of the active site. The three-dimensional structure suggests one way such activation could occur: the β2′–β3′ loop of C2 contacts Lys436 and Leu438 of the β2–β3 loop of C1, thereby linking residues that form the forskolin-binding site with structural elements carrying residues important for catalysis. In this model, the regulator contacts allosteric sites in C1 and C2, thereby rearranging residues at the C1/C2 cleft that positively impacts cAMP catalysis.
Besides forskolin and Gαs, no other crystal structures are solved for AC complexed to regulatory proteins. Binding of Gαi was suggested to take place within the cleft formed by the α2 and α3 helices of C1  analogous to but pseudosymmetric with the Gαs binding in C2. For both, CaM and Gβγ, a single regulator-specific contact site was deduced so far: for CaM binding, an AC sequence in C1b (a stretch of 28 amino acids forming the AC28 region) with high affinity for CaM had been identified . The AC28 region comprises a hydrophobic sequence containing basic amino acids as expected for CaM effector sites, and is located centrally in the cyclase. The corresponding peptide efficiently interferes with the stimulation of AC by CaM, but exhibits a higher affinity for CaM binding (2 nM) than does AC (Kd for CaM-mediated AC stimulation 15–20 nM; [9,10]). Gβγ is a conditional regulator of ACs, i.e. both activation of ACII and inhibition of ACI are best observed at the pre-stimulated ACs . Recently, the PFAHL motif in the variable C1b domain of ACII was shown to be indispensable for Gβγ stimulation of the ACII .
In the present study we show that Gβγ- and also CaM-regulation of AC isoforms required a second regulator-specific site in the other catalytic domain, C2. For Gβγ the KF loop in ACII between the α2′ helix and the β2′ sheet was identified, and for CaM the VLG loop at the C2a/C2b boundary of ACI. Although the general AC stimulators forskolin and Gαs contact conserved amino acids in C1 and C2, motifs of the isoform-specific regulators Gβγ and CaM were not conserved, but rather showed subgroup specificity according to the different regulatory patterns of ACs.
Generation of AC constructs
ACI with a N-terminal Myc-tag was generated by PCR using bovine cDNA (GenBank® accession number P19754) encoding ACI as a template and primer Myc–IM1C1 (5′-GGAGGAACTAGTACCATGGAACAAAAACTGATATCGGAAGAAGACCTCGCGGGGGCGCCGCGCGGCCGAGGC-3′). The PCR product was ligated into pBSKII+ (Stratagene) using SpeI and HindIII. Activity and regulation of this ACI (denoted ACI wt) was indistinguishable from untagged wild-type ACI and was used as a control throughout the present study. The generation of constructs (Figure 1B) encoding N- and C-terminal halves of ACI (I-M1C1 and I-M2C2), ACII (II-M2C2), and ACV (V-M1C1) was performed by PCR-based mutagenesis and has been described previously . Construction of ACII.IC1b has been described previously . For generation of ACI.IIC2 the oligonucleotides 5′-GCCGTCTCAGCAGAGTGAATATTACTGTAGGTTAG-3′ and 5′-CCAAGCTTATTCAGGATGCCAAGTTGCTCTG-3′ were applied in an analogous strategy using the endogenous restriction sites Bsu36I and HindIII. The ACI-deletion mutant ACI.Λ1057 was constructed using 5′-CTACACATCACCCGGGTCCAGTG-3′ and 5′-CGAAAGCTTAGAAGTATGTCAGCATC-3, and ACI. Λ1094 was constructed using 5′-CTACACATCACCCGGGTCCAGTG-3′ and 5′-CGAAAGCTTAGGGGTGACCCGC-3′; cloning was performed using the ACI-endogenous sites XmaI and HindIII. QuikChange® site-directed mutagenesis using PfuTurbo Cx Hotstart DNA Polymerase (Stratagene) was used according to the manufacturer's instructions to generate NAAIRS (asparagyl-alanyl-alanyl-isoleucyl-arginyl-serine) and AA (alanyl-alanine) mutants. ACII.Λ928 was generated as described previously  using 5′-TGATGATCTGCTTTCTAATGCTGCTATACGATCGGTTGAAAAGATCAAG-3′ and the corresponding reverse complementary oligonucleotide as primers. The ACII.AAxxx mutants were generated using primers: 5′-GCTGACTTTGATGATGCTGCTTCTAAGCCAAAGTTC-3′ (ACII.AA925), 5′-TTTGATGATCTGCTTGCTGCTCCAAAGTTCAGTGGT-3′ (ACII.AA927), 5′-CTGCTTTCTAAGCCAGCTGCTAGTGGTGTTGAAAAG-3′ (ACII.AA930), and 5′-TCTAAGCCAAAGTTCGCTGCTGTTGAAAAGATCAAG-3′ (ACII.AA932) in combination with their respective reverse complementary pendants. ACI-Gαi mutants were generated using primers 5′-CCACGTTGCCCAGCACGCCCTAATGTCCAACCCTCG-3′(ACI.F852A), 5′-GCACTTCCTAATG-TCCGCCCCTCGCAACATGGACC-3′ (ACI.N856A), 5′-CCTAATGTCCAACCCTGCCAACATGGA-CCTGTATTACC-3′, (ACI.R858A), and 5′-CCTAATGTCCAACCCTCGCGCCATGGACCTGTATT- ACC-3′ (ACI.N859A).
The synthetic peptides were: pAC28, NH2-IKPAKRMKFKTVCYLLVQLMHCRKMFKA-COOH, derived from sequences located in IC1b; pVLG, NH2-TEEVHRLLRRGSYRFVCRGKV-COOH, derived from sequences located in IC2a; pAAG, NH2-TEEAHRLARRGSYRFVCRGKV-COOH, identical with pVLG except for two alanine mutations at the CaM-critical residues V and L (underlined); and pTT, NH2-TQPKTDHAHCCVEMGLDMIDT-COOH, derived from sequences located in IC1a.
Preparation of AC
All AC constructs were expressed in Sf9 cells using the baculovirus expression system; purified plasma membranes were the sources for AC assays. Baculovirus encoding the AC was generated from the pFastBac1-AC construct in Sf9 insect cells (Invitrogen). Cells (106/ml) were then infected with the baculovirus (1 plaque-forming unit/cell), harvested 48–52 h later, and lysed by nitrogen cavitation. After removal of nuclei by centrifugation (1000 g for 5 min at 2 °C), membranes were collected, washed and resuspended.
Regulators of AC
Gβγ was expressed in Sf9 insect cells using baculoviruses encoding Gβ1 and Gγ2, detergent extracted and purified by affinity chromatography as described using the Gγ-attached N-terminal His-tag .
Gαs was expressed in bacteria and purified using the C-terminal attached His-tag . Purified Gαs was activated in vitro with GTP[S] (guanosine 5′-[γ-thio]triphosphate) by incubation for 30 min at 30 °C followed by 1 h on ice. The activated Gαs-GTP[S] was purified by subsequent size-exclusion chromatography to remove the free nucleotide. The concentration of the activated Gαs was determined by inclusion of radiolabelled [35S]GTP[S] during activation. CaM was purchased from Calbiochem.
Equimolar amounts of dansyl chloride and N-hydroxysuccinimide were mixed in dimethylformamide and activated by the addition of 1 mol of triethylamine. Activated dansyl ester was incubated with 1/4 mol of CaM at pH 7.5 overnight at room temperature (21 °C) before the formed dansyl-CaM was purified by size-exclusion chromatography. This procedure was dissimilar to protocols used by others [16a,16b] and Sigma previously, but allowed the preferred dansylation of just the N-terminal amino acid of CaM, rather than an uncontrolled number of internal hydroxyl or amino groups.
A volume of 800 μl of dansyl-CaM (80 nM) was adjusted to the indicated concentrations of Ca2+ ions, peptides and/or EGTA; dilution effects were maintained below 3%. Fluorescence emission spectra were recorded from λ=450 nm to λ=550 nm (λex=334 nm, 20 nm/min) at 27 °C with a Luminescence Spectrometer LS50B (PerkinElmer).
AC activity was determined based on the conversion of [α-32P]ATP into [32P]cAMP with subsequent purification of cAMP by sequential chromatography using cation exchange (Dowex 50) and neutral alumina resins. All samples contained pyruvate kinase and phosphoenolpyruvate as an ATP-regenerating system and Ro 20-1724 as an inhibitor of cAMP-specific phosphodiesterases. Assays were performed for 7–10 min at 30 °C in a final volume of 100 μl with the indicated amounts of recombinant Sf9 membranes in the presence of 10 mM MgCl2 and 0.5 mM ATP.
To determine CaM regulation, AC-containing membranes were washed with 1 mM EGTA to remove Sf9 endogenous CaM, then AC activity was determined in the presence of 100 μM CaCl2. After a 2 min pre-incubation at 30 °C the AC assay was started by addition of the substrate and stopped after another 7 min incubation period. For peptide competition studies, peptides were pre-incubated for 60 min on ice in the presence of CaM.
CaM kinase assays were performed according to the manufacturer's protocol (SignaTECT from Promega).
Membrane proteins were quantified by dye-binding using acidic Coomassie Brilliant Blue (Bio-Rad) with BSA as a standard. Immunodetection of AC constructs was performed with commercially available antibodies: anti-c-Myc (9E10; Santa Cruz Biotechnology), anti-ACII (C20; Santa Cruz Biotechnology) and anti-HA (haemagglutinin; 12CA5; Roche).
RESULTS AND DISCUSSION
A second CaM site in ACI
Amino acids 495–522 (denoted as the AC28 region) in the C1b domain of ACI have been shown to be necessary for stimulation of this AC subtype by CaM and were represented by the synthetic peptide pAC28 . To test whether the AC28 region in the context of the intact C1b domain of ACI was able to transfer CaM stimulation to the CaM-insensitive ACII, we generated the ACII.IC1b chimaera (see Figure 1). The results shown in Figure 2 revealed that the presence of the AC28 region was not sufficient to turn ACII into a CaM-stimulated ACII.IC1b. Another chimaera indicated a second site located in C2 necessary for CaM to stimulate the catalytic activity of type I AC. This second chimaera, ACI.IIC2, was a type I-like AC with IC2 substituted by the homologous C2 domain of the CaM-insensitive ACII. ACI.IIC2 was no longer regulated by CaM, although it was still stimulated by Gαs and forskolin; indeed, Gαs- and forskolin-stimulated activities of ACI.IIC2 by far exceeded those of parent wild-type ACs, type I and type II (results not shown).
In order to localize the missing CaM motif in either C2a or C2b of ACI, two deletion mutants of ACI were generated (see Figure 1). In ACI.Λ1094 the 44 C-terminal residues of the C2b domain were removed; ACI.Λ1057 was devoid of all amino acids assigned to C2b. Both deletion mutants were active and stimulated by Ca2+/CaM as well as ACI (Figure 3). These results provided evidence that the C2a of the ACI domain was necessary for CaM to stimulate ACI.
In order to define the amino acids in C2a responsible for mediating the CaM-stimulatory signal to ACI, the catalytic C2a subdomain was checked for regions matching the rules for putative CaM-interaction sites described by Rhoads et al. . Although no universal CaM-binding motif had been defined and CaM-interaction sites were difficult to predict, we could identify one stretch of 14 amino acids in the C-terminus of the C2a subdomain of ACI that obeyed the 1-5-8 rule (Figure 4A). This domain contained hydrophobic residues at positions 1, 5 and 8 (valine, leucine and glycine) and was therefore called the VLG loop. The VLG homologous region in the ACII crystal showed a helix-loop-helix structure and was located in the periphery of the catalytic heterodimer (Figure 4B).
To show the involvement of the VLG loop in CaM-stimulation of ACI, two amino acids at key positions 1 and 5 of the motif, Val1027 and Leu1031, were mutated to an alanine residue. The resulting mutant was not catalytically active under basal, Gαs- or forskolin-stimulated conditions (results not shown). The fact that point mutations within the VLG region were sufficient to completely abolish the catalytic activity of the enzyme led to the conclusion that this region was crucial for the integrity of the catalytic core. Its impact on the enzymatic function of AC already became obvious when truncation immediately C-terminal to the VLG region resulted in a mutant (ACI.Λ1057) with reduced catalytic activity (see Figure 3). To date the crystal structure of the IC2a domain is not known. All structural data rely on sequence alignments of IC2a with the IIC2a subdomain the structure of which had been resolved in complex with VC1a . In ACII, lacking a C2b subdomain, the VLG-homologous sequence is assigned to the C-terminal part of the enzyme (see Figure 4B) and has not been reported to be involved in catalysis. In contrast, the VLG motif of ACI is followed by the variable C2b subdomain, thereby forming the C2a/C2b boundary that might be crucial for the correct orientation of the two subdomains and thereby crucial for the integrity of the catalytic core.
As we had identified the key amino acids of the VLG loop, Val1027 and Leu1031, to be indispensable for the catalytic activity of ACI, peptide studies were applied to show indirect evidence for the involvement of that loop in CaM regulation. The peptide pVLG, comprising all amino acids of the VLG loop, was synthesized and analysed for direct CaM interaction and for competition with the CaM effectors ACI and the cyclase-unrelated CaM-dependent kinase II.
The peptide corresponding to the VLG loop did bind CaM as shown by fluorescence changes of dansyl-CaM (Figure 5A). Whereas the fluorescence intensity stayed unaltered in the presence of a non-binding peptide (pTT), increasing concentrations of pVLG led to increasing intensity of the dansyl-CaM fluorescence. Similar fluorescence changes were obtained using the peptide pAC28 covering the amino acids of C1b, already known to be necessary for ACI stimulation by CaM. Obviously, the concentration of 80 nM dansyl-CaM used in this experiment exceeded the Kd of pAC28 and pVLG for CaM as shown by the linear relation between fluorescence change and peptide concentration below the equimolar ratio peptide/CaM indicating a 1:1 binding stochiometry. A minimum of 80 nM dansyl-CaM was necessary to obtain precise fluorescence data. When the key amino acids valine and leucine of pVLG were changed to alanine residues (pAAG), fluorescence changes caused by the peptide were reduced, pointing to a sequence dependence of the VLG region for binding to CaM.
The peptide pVLG did not only bind CaM, but obviously occupied the effector-interacting epitope of CaM, as the CaM–pVLG complex could no longer stimulate the CaM effectors ACI (Figure 5C) and CaM-dependent kinase II (Figure 5D). Again, the importance of amino acids Val1027 and Leu1031 in ACI (positions 1 and 5 in the VLG motif respectively) for CaM binding became evident as pAAG competed for CaM less efficiently than pVLG with both effectors.
Surprisingly, ACI inhibition was not exclusively mediated by peptide binding to CaM, but also by a direct peptide effect on ACI as depicted in the insert of Figure 5(C). Nevertheless, the calculated peptide net effect on CaM stimulation, corrected by the peptide effect on basal AC activity, was still more pronounced for pVLG than for pAAG. The phenomenon that pVLG and pAAG also inhibited the basal activity of ACI corroborated our earlier hypothesis, which evolved from the inactive point mutations in ACI, that integrity of the VLG-region was a prerequisite for overall enzymatic activity of ACI.
Owing to readout-dependent differences in assay sensitivity, higher CaM concentrations were applied for peptide competitions with ACI than for dansyl-CaM binding (200 nM compared with 80 nM; see Figure 5C compared with 5A). No differences between pAC28 and pVLG were detected in the fluorescence assay as 80 nM dansyl-CaM was well above the Kd of pAC28 binding to CaM . In contrast, the AC readout with the coupled reaction scheme, involving CaM binding and AC interaction at two sites (see below), revealed for pAC28 (IC50 500 nM) higher affinity in binding CaM than for pVLG (IC50 10 μM; results not shown). This would point to a minor role of the VLG motif in transmitting CaM stimulation to ACI compared with the AC28 region. However, despite its relatively high affinity for CaM, C1b harbouring the AC28 region could not be the relevant CaM-sensitive region of ACI for the following reasons: (i) in the context of the chimaera ACII.IC1b the AC28 region did not transmit the stimulatory Ca2+/CaM signal on to a Ca2+/CaM-insensitive AC isoform; and (ii) although ACI-activity modulation by CaM is known to be strongly Ca2+-dependent , pAC28 bound to CaM in a Ca2+-independent manner. Figure 5(B) shows that pAC28 binding was also detected in the absence of free calcium ions (‘+EGTA’), a hitherto unappreciated phenomenon. In contrast, pVLG derived from the newly detected second CaM site in ACI exhibited Ca2+-dependent binding to CaM (see Figure 5B). Therefore CaM regulation became both highly potent by the AC28 region and Ca2+-sensitive by the VLG region, establishing a molecular basis by which ACI activity can be changed during cytosolic calcium oscillations .
A similar scenario has already been described for the second CaM-stimulated AC isoform, ACVIII, with the N-terminus pre-assembling CaM and the C2 domain enabling both Ca2+-sensitivity and the stimulatory action of CaM . ACI and ACVIII do not provide identical signal transfer motifs for CaM, but their mechanistic modulation is based on the same two-site scheme. The pre-recruitment of CaM by the AC28 region and the N-terminus of ACI and ACVIII respectively, is highly beneficial, if not essential, in the intact cell. In living cells, the majority of the total of 10 μM CaM is sequestered, tethered or compartmentalized, so the concentration of potential CaM-binding proteins in the cytosol or cytoplasm facing membranes exceeds that of the free, available CaM (approx. 45 nM) about 2-fold . In the resting cell, CaM tethering is important to advance complex formation between the three partners Ca2+, CaM and effector. Following an increase in intracellular calcium concentration, the AC-pre-associated CaM might change its conformation and activate the AC involving the catalytic AC domain C2a, supposedly by stabilizing structural elements such as the VLG loop. Similar activation schemes have already been described for the two AC exotoxins of Bordetella pertussis and Bacillus anthracis. CaM in an extended conformation binds to several loops close to the catalytic centre of the bacterial AC and activates the enzyme through subtle changes in the surroundings of the active cleft .
A second Gβγ site in ACII
The structural basis of ACII regulation by Gβγ showed a strong analogy to stimulation of ACI by CaM. Also for Gβγ, an obligatory motif in C1b had been described to observe ACII stimulation. Like the AC28 region in ACI, amino acids of the PFAHL motif in C1b of ACII had been identified as an essential Gβγ signal transfer site of ACII because substitution of IIC1b or just the PFAHL motif by IC1b or irrelevant amino acids respectively, completely abolished Gβγ stimulation . In the present study, we showed that besides the PFAHL motif on C1b, domain C2 of ACII was also important in mediating Gβγ stimulation to sensitive ACs.
In a first step, AC was cut into halves comprising M1C1 and M2C2 (see Figure 1). Both halves together lined up to the bisected ACI (AChalvesI+I) and ACII (AChalvesII+II), and, after coexpression, functionally rebuilt the Gβγ-inhibited ACI (Figure 6) and Gβγ-stimulated ACII . This proved that the Gβγ-regulatory pattern was preserved even in the bisected AC. However, in a second step, IIM2C2 was co-expressed with IM1C1, the N-terminal half of the Gβγ-inhibited AC type I, or VM1C1, the N-terminal half of the Gβγ-insensitive canine ACV . Both resulting bisected chimaeras AChalvesI+II and AChalvesV+II respectively, were still stimulated by Gβγ (1.5–2-fold) indicating a Gβγ-stimulating feature of IIM2C2. This stimulatory effect was even observed when only IIC2 was introduced into an ACI background (ACI.IIC2) pointing to a second Gβγ site located in C2 of ACII. The inverse chimaera ACII.IC2 was catalytically inactive under basal, Gαs- and forskolin-stimulated conditions (results not shown).
To define amino acids in C2 responsible for mediating the Gβγ-stimulatory signal to ACII, the catalytic C2 domain was screened by NAAIRS substitutions. The hexapeptide NAAIRS is a flexible linker adopting various secondary structures in different proteins. It therefore had been applied in different substitution experiments, for example leading to the identification of the PFAHL motif in the C1b domain of the ACII . Although all amino acids that were known from the literature to be involved in catalysis, in Gαs or in forskolin interaction were excluded from this screen, 50% of the generated mutants were not catalytically active; the remaining eight mutants that functionally could be tested were all stimulated by Gβγ (results not shown).
In the following screen we substituted only two instead of six amino acids. The AA substitutions were introduced into a short loop (named the KF loop according to its central amino acids) in IIC2 that showed significant similarity for all three Gβγ-stimulated ACs (types II, IV and VII; Figure 7A), but none for the Gβγ-inhibited ACI or the Gβγ-insensitive canine ACV. Additionally the loop appeared an ideal target for regulators like Gβγ as it is located dorsal to the catalytic cleft, thereby supposed to be easily accessible for regulators (Figure 7B). Moreover, the KF loop connects the two major structural elements of C2a that form the interface sites with C1a, i.e. the β2′ sheet of C2a that stacks on top of the β4–β5 loop of C1a, and the α2′ helix of C2a that contacts the N-terminal segment of β2 and the C-terminal end of β4 of C1a . Consequently, a conformational change of the KF loop, caused by the docking of Gβγ, should have a direct effect on the AC activity.
The consecutive substitution of two amino acids from this nine-residue motif resulted in four mutants (Figure 7C) that were catalytically active (Figure 7D). In contrast, NAAIRS substitution in this region starting at amino acid residue 928 had generated the catalytically inactive ACII.Λ928 mutant. One reason might be the detrimental substitution of Pro929 as a structurally relevant amino acid in the NAAIRS screen, whereas Pro929 was not exchanged in the AA screen. The potential impact of Pro929 on intrinsic enzyme activity was underlined by the diminished activity of ACII.AA930 with the two amino acids substituted adjacent to Pro929.
Figure 8 depicts the Gβγ responses of the four ACII AA mutants. Substitution of the KF pair by two alanine residues resulted in ACII.AA930 that was no longer stimulated by Gβγ. This mutant defined Lys930 and Phe931 of ACII as key residues for Gβγ-stimulation. AA substitutions preceding the central KF spot and adjacent to Pro929 (ACII.AA927) showed diminished Gβγ stimulation, whereas substitutions more distant from the KF pair did not significantly change Gβγ regulation of the resulting mutants ACII.AA925 or ACII.AA932. These results provided clear evidence that the KF loop was an essential motif in ACII to observe Gβγ-mediated stimulation.
The QEHA motif (amino acids 956–982) was another region in IIC2 that had been described in the past to interact with the Gβγ-complex ; however, domain swaps and substitutions in that region revealed that, in contrast with the PFAHL motif and KF loop, the QEHA motif was not necessary as a mediator of the stimulatory effect of Gβγ to ACII . Hence, we assumed that the QEHA motif as a neutral Gβγ tether provided local concentration of Gβγ, without playing a role in the regulatory signal transfer to the AC.
During the course of the preparation of this manuscript Gao et al.  published data regarding the Gβγ-stimulation of the human ACV and ACVI whereas canine ACV was Gβγ insensitive. Interestingly, the human ACV amino acid sequence is identical with canine ACV in the PFAHL motif as well as the KF loop and dissimilar to the orthologous ACVI as well as ACII sequences. Furthermore, Gao et al.  identified a Gβγ-interaction site at the N-terminus of human ACV and human ACVI. We concluded that the PFAHL and the KF motif were specific Gβγ-regulatory hot spots for type II-like ACs (including ACIV and ACVII, all providing identical motifs), whereas other AC subgroups provided different motifs to result in analogous modulation.
In summary, the results of the present study show a minimum of two sites being necessary for AC isoforms to be regulated by Gβγ or Ca2+/CaM. Analogously to the already described contact sites for Gαs and forskolin [6,19–22], they are located on both halves of AC. For the general stimulators (Gαs and forskolin) interaction sites have been described at the highly conserved C1a and C2a domains, whereas the isoform-specific regulators (Gβγ, Ca2+/CaM) used sites in C2a plus the less conserved C1b region specifying the individual isoform regulation (see Figure 1A). A similar scenario was also observed concerning the Gαi-mediated inhibition of the isoform ACI (S. Diel and C. Kleuss, unpublished work). The substitution of the cytosolic domain IC2, resulting in the chimaera ACI.IIC2, was sufficient to completely abolish Gαi inhibition. This clearly indicated that besides the Gαi-binding site in the C1a domain , the C2 domain was also essential for Gαi inhibition on ACI. Preliminary data based on single alanine substitutions in full length ACI to define a point-symmetric residue to the Gαs-specific Phe379 (ACV numbering; see above) in IC2 for the Gαi-interaction showed that neither Phe852 or Asn856, or Arg858, or Asn859 were involved in Gαi inhibition (results not shown).
Two schemes for the regulatory signal transfer via motifs on both AC halves are conceivable: either a concerted action takes place where binding of the multifaceted regulator to AC and modulation of AC activity occurs almost simultaneously at the C1- and C2-contact sites of AC thereby rearranging the catalytic C1/C2-interface for more effective (activation) or diminished (inhibition) cAMP catalysis. Alternatively, a sequential mechanism takes place. First, the regulator binds with reasonably high affinity at a general binding region of AC, potentially on the variable C1b domain; secondly, the pre-assembled regulator induces a signal transfer cascade within the AC involving the second regulatory site, potentially on the catalytic domain C2a. The sequential model offers the advantage that the second regulatory region like the VLG motif may afford relatively low affinities for the regulator because in the pre-assembled complex the local concentration of the regulator exceeds overall cytosolic concentrations by orders of magnitude. At least for CaM regulation of ACI and ACVIII, the sequential mechanistic model fitted best to the data and seemed plausible with respect to multiple effector molecules in the cell.
In general, the existence of multiple AC hot spots for one regulator appears to be a key feature of ACs in their function as multifaceted detectors and integrators of various regulatory signals that are then translated into one common, easily-decoded, intracellular signal, the cAMP molecule.
We thank Dr Alfred G. Gilman (UT Southwestern Medical Center at Dallas, Pharmacology, Dallas, TX, U.S.A.) for cDNAs encoding ACI and ACII, and baculoviruses encoding Gβ1 (M13236), tagged Gγ2 (K02199) and β-galactosidase. We appreciate the work of Kathrin Klass who provided the dansyl-CaM. The work was initially supported by the Deutsche Forschungsgemeinschaft (DFG KL773/5).
Abbreviations: AA, alanyl-alanine; AC, adenylyl cyclase; CaM, calmodulin; GTP[S], guanosine 5′-[γ-thio]triphosphate; NAAIRS, asparagyl-alanyl-alanyl-isoleucyl-arginyl-serine
- © The Authors Journal compilation © 2008 Biochemical Society