An Important Functional Role of the N Terminus Domain of Type VI Adenylyl Cyclase in G (cid:1) i -mediated Inhibition* □ S

We show herein that removal of the first 86 amino acids (aa) of the N terminus (designated N) of type VI adenylyl cyclase (ACVI) caused the resultant ACVI mutant (ACVI- (cid:2) A87) to be more greatly inhibited by a G (cid:1) i coupled receptor or activated G (cid:1) i protein. Moreover, in vitro binding of the full-length N and C1a domain (des-ignated C1a), which interacts with G (cid:1) i , was detected. A truncated N terminus (aa 1–86) also interacted with C1a, suggesting that the C1a-interacting region is located within aa 1–86. Mutation analyses further revealed that N might interact with C1a in the region (aa 434–505) where G (cid:1) i is bound. Mutations of two residues (Leu-472 and Val-476) located in this N-binding region of C1a suppressed the interaction between recombinant N and C1a and markedly reduced G (cid:1) i -mediated inhibition of ACVI- (cid:2) A87. Further biochemical analyses of the effect of internal mutations of Leu-472/Val-476 on G (cid:1) i -medi- ated inhibition of wild-type ACVI and ACVI- (cid:2) A87 suggested that N modulates the G (cid:1) i -mediated inhibition of ACVI via binding to C1a when the level of G (cid:1) i is The PCR product was then subcloned into the pcDNA3.1/V5-His-TOPO vector using the TA expression kit (Invitro- gen). Mutations of L472A and V476A were confirmed by DNA sequenc-ing. The resultant construct was digested with SacI and Bsu36I and subcloned into the SacI/Bsu36I-digested pVL1393-ACVI construct to create the pVL1393-ACVI-L472A-V476A mutant construct. The eu- karyotic expression construct (pcDNA3-ACVI-L472A-V476A) of the ACVI-L472A/V476A mutant protein was constructed by subcloning a BamHI/EcoRI fragment encoding ACVI-L472A-V476A from pVL1393-ACVI-L472A-V476A into the corresponding sites of pcDNA3. The ACVI- (cid:5) A87-L472A-V476A mutant was created by subcloning the NheI/ Bsu36I-digested pcDNA3-ACVI-L472A-V476A fragment containing the mutated residues (L472A and V476A) into the NheI/Bsu36I-digested pcDNA3-ACVI- (cid:5) A87 construct. The ACVI-N5 mutant was also created by the two-step PCR tech-nique using two ACV-specific primers (5 (cid:3) -GGATCCATGTCCGGCTC- CAAAAGC-3 (cid:3) and 5 (cid:3) -GTGAGGCTGCTCTGGTTCAGGCGGAA-3 (cid:3) ) and pVL1393-ACV (23) as the DNA template and two ACVI-specific primers (5 (cid:3) -AACCAGAGCAGCCTCACGCTG-3 (cid:3) and 5 (cid:3) -CTCCGCAGCCAGCTT-GTCGAA-3 (cid:3) ) and pVL1393-ACVI the DNA

The mammalian adenylyl cyclase (AC) 1 superfamily consists of nine membrane-bound isoforms. All of these possess three large cytosolic domains (designated N, C1a, and C2 domains; Fig. 1A), which are separated by two sets of six-transmembrane domains (1)(2)(3). The C1a and C2 domains among the nine AC members are highly homologous (with 50 -90% similarity in amino acids). In addition, the C1a and C2 domains of each AC share ϳ50% similarity and form the catalytic core complex, which can be stimulated by forskolin or G␣ s proteins (4 -6). Crystallographic analysis of the catalytic complex consisting of the C1a domain, the C2 domain, and the GTP␥S-bound G␣ s protein revealed that forskolin and/or G␣ s stimulate ACs by enhancing the interaction between the C1a and C2 domains and by stabilizing the C1a-C2 catalytic core complex (7).
Although most ACs can be activated by G␣ s and forskolin, regulation of each AC isozyme differs. Studies of the morevariable N and C1b domains reveal that these two domains may play important regulatory roles. For example, the C1b domain has been implicated in the regulation of AC isozymes mediated by Ca 2ϩ /calmodulin, calcineurin, or protein kinase A (8 -11). In addition, the C1b domain of ACV and ACVII has been shown to modulate G␣ s -evoked activation by interacting with catalytic core complexes (12,13). Others and we have shown that another variable region, the N-terminal domain, also significantly contributes to the regulation of AC activity (14,15). Specifically, the N-terminal domain of ACVI (amino acids 1-160) plays an important role in the protein kinase C (PKC)-mediated inhibition and phosphorylation of ACVI (14). Removal of the first 86 amino acids (aa) of ACVI reduced the inhibitory effect of PKC on ACVI activity without affecting the basic enzymatic properties, including the affinities toward its substrate and two stimuli (forskolin and G␣ s protein (14)). The N-terminal domain of ACVI therefore functions as a regulatory domain. Further biochemical analyses revealed that at least four PKC phosphorylation sites (Ser-10, Ser-568, Ser-674, and Thr-931), located in the three large cytosolic domains of ACVI, significantly contribute to PKC-mediated inhibition of ACVI (16). Intramolecular interactions among the N, C1a/b, and C2 domains of ACVI therefore appear to be important for regulation of ACVI activity.
G␣ i -mediated inhibition is a major regulatory feature of the AC superfamily. Among AC members, only ACI, ACV, and ACVI can effectively be inhibited by G␣ i proteins (17,18). Based on results obtained from an in vitro binding assay, the C1a domain was shown to bind myristoylated G␣ i proteins and form stable complexes (19,20). Mutagenesis of full-length ACV in the ␣2 and ␣3 helices of the C1a domain revealed several residues important for the inhibition by G␣ i proteins. It was postulated that G␣ i may exert its inhibitory effect through binding to the C1a domain at the site just opposite the G␣ sbinding site on the C2 domain and, subsequently, causes reduced interaction between the C1 and C2 domains (21). In the present study, we present evidence to suggest that the Nterminal domain of ACVI directly interacts with the C1a domain and may play an important role in the regulation of ACVI by G␣ i proteins.

EXPERIMENTAL PROCEDURES
Materials-Quinpirole (Quin) was obtained from Research Biochemicals (Natick, MA). Forskolin, cAMP, and ATP were obtained from * This work was supported by grants from the National Science Council (Grants NSC89-2320-B001-011 and NSC90-2320-B001-009), from the National Health Research Institutes (Grant NHRI-EX92-9203NI), and from Academia Sinica, Taipei, Taiwan, Republic of China. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The online version of this article (available at http://www.jbc.org) contains one supplementary figure (Fig. S1) and one supplementary table (Table S1).
¶ To whom correspondence should be addressed. Cell Culture and Transfection-The CHOP cell line, which expresses the polyoma large T antigen, was originally derived from a Chinese hamster ovary cell line and was a generous gift from Dr. J. W. Dennis (Samuel Lunenfeld Research Institute, Toronto, Ontario, Canada) (22). These cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 200 g/ml Geneticin in 10% CO 2 and 90% air. The cDNA of rat ACVI was kindly provided by Dr. R. Iyengar (Dept. of Pharmacology, Mount Sinai School of Medicine, New York) (23). Coding regions of rat ACVI variants were subcloned into an expression construct (pCEP4 or pcDNA3, Invitrogen, Carlsbad, CA). Expression of ACVI in transfected cells was driven by the human cytomegalovirus immediate-early gene enhancer-promoter. The expression construct of the short form dopamine D2 receptor (D2s-R (24)) was kindly provided by Dr. J.-C. Chen (Chang-Gung University, Taoyuan, Taiwan). For transient transfection experiments, CHOP cells were transfected with the desired construct using the DEAE-dextran method (25). Cells were harvested 72 h post-transfection for analysis.
SDS-PAGE and Western Blotting-Membrane fractions were separated on 8 -15% separating gels according to the method of Laemmli (26). For Western blot analyses, gels were transferred to polyvinylidene difluoride membranes following electrophoresis that were blocked with 5% skim milk in phosphate-buffered saline (PBS), and then incubated with the desired antiserum at 4°C overnight. The polyclonal antibody, AC6D, was raised against aa 987-1187 (the C2 domain) of ACVI (27). Typically, a 1:5000 dilution was used for AC6D unless stated otherwise. To determine if the recombinant G␣ i1 proteins were myristoylated, 9% SDS-PAGE gels containing 4 M urea were employed (28) as were Western blot analyses using an anti-G␣ i1 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) at a 1:200 dilution. Then the membranes were incubated with peroxidase-conjugated donkey anti-rabbit IgG (1:5000 dilution, Amersham Biosciences) for 1 h at room temperature and washed three times with PBS. The immunoreactive bands were stained using a light-emitting non-radioactive method (ECL, Amersham Biosciences).
Adenylyl Cyclase Assay-AC activity was assayed as previously described (29). Briefly, cells were sonicated using a W-380 sonicator (Ultrasonics, Farmingdale, NY) at a setting of 20% output power for a total of 90 s. The homogenate was centrifuged at 50,000 ϫ g for 30 min to collect the P1 membrane fractions. Unless specifically addressed, the AC activity assay was performed at 37°C for 10 min in a 400-l reaction mixture containing 1 mM ATP, 100 mM NaCl, 0.4 unit of adenosine deaminase, 50 mM HEPES, 0.5 mM 3-isobutyl-1-methylxanthine, 6 mM MgCl 2 , 1 M GTP, 0.2 mM EGTA, and 20 g of membrane protein.
Reactions were stopped by the addition of 0.6 ml of 10% trichloroacetic acid. The cAMP formed was isolated by Dowex chromatography (Sigma) and assayed by radioimmunoassay. The enzyme activity was linear for up to 30 min with membrane proteins up to 40 g. The ACVI activity was determined as the difference between the cyclase activities measured in the membrane fractions collected from CHOP cells transfected with the indicated ACVI cDNA versus an empty vector (pcDNA3). For CHOP cells expressing wild-type ACVI, endogenous cyclase activities represented ϳ40% of the total activity. The absolute values of ACVI activity observed in different transient transfection experiments might have slightly varied due to the passages of cells. Nevertheless, the overall pattern of AC regulation was consistent among experiments.
Recombinant G␣ s and G␣ i Protein Expression-The expression construct (pQE60/H6-G␣ S ) of the G␣ s protein was a generous gift from Dr. W.-J. Tang (University of Chicago, Chicago, IL). The hexahistidine (H6)-tagged G␣ s protein was expressed in Escherichia coli and was purified using Ni-NTA His⅐Bind® resin (Novagen, Madison, WI) as described elsewhere (30). To stimulate AC activity, the indicated concentration of G␣ s protein was activated by GTP␥S (20 M) in a 100-l reaction mixture containing 10 mM MgCl 2 , 20 mM Tris, 1 mM EDTA, and 1 mM DTT for 20 min at 20°C. The expression construct (pET11D-H6-G␣ i1 ) of the G␣ i protein was constructed by subcloning an XhoI/BamHI fragment encoding full-length human G␣ i1 from the pCDNA3.1 ϩ -G␣ i1 construct (Guthrie cDNA Resource Center, Sayre, PA) into the corresponding sites of a H6-tagged pET11d vector (Novagen). The G␣ i protein with an H6 tag fused to its N terminus was expressed in E. coli harboring a yeast protein, N-myristoyltransferase (28), and was purified using Ni-NTA His⅐Bind® resin (Novagen) as described elsewhere (32). The expression construct of N-myristoyltransferase was a generous gift from Dr. Tohru Kozasa (Dept. of Pharmacology, University of Illinois at Chicago, IL). To inhibit AC activity, the indicated concentration of G␣ i protein was activated by GTP␥S (400 M) in the presence of 50 mM NaHepes (pH 8.0), 5 mM MgSO 4 , 1 mM EDTA, and 2 mM dithiothreitol at 30°C for 2 h.  (7). VIC2 is shown in dark gray. The N-interacting peptide (aa 434 -505) of VIC1a is shown in black. The region of VIC1a, which is not involved in the interaction with the N terminus of ACVI, is shown in light gray. Residues located on the ␣2 (Glu-411, Met-414, and Thr-415) and ␣3 (Leu-472 and Val-476) helices of C1a, which might be involved in the interaction with G␣ i based on a previous mutational analysis of ACV (19), are indicated.
DNA fragments encoding the C1a and N domains of ACVI were produced using the PCR technique. DNA amplification was carried out in a solution containing 10 mM Tris-HCl, 50 mM KCl, 1.5 mM MgCl 2 , 0.001% (w/v) gelatin, 0.2 mg of the desired primers, 0.2 mM of each deoxynucleoside triphosphate (dNTP), a DNA template, and 2 units of DynaZyme TM thermostable DNA polymerase (Finnzymes, Espoo, Finland) per 50 l of reaction solution. The reactions proceeded for 40 PCR cycles. Primers utilized to create domains of ACVI (23) are listed in the section of Supplementary Materials (Table S1). The amplified DNA fragments were ligated to the C terminus of the H6 tag in the pET11d vector (Novagen; N 1-160 , N 1-86 , and C1a-2), or subcloned into the pcDNA3.1/V5-His-TOPO vector using a TA expression kit (Invitrogen; C1a-1, C1a-3, C1a-4, and C1a-5). Nucleotide sequences in all constructs used were confirmed by DNA sequencing.
In Vitro Binding-The recombinant domains of ACVI were synthesized in vitro by the TNT®T7-coupled transcription/translation system (Promega) using [ 35 S]methionine (Amersham Bioscience) following the manufacturer's protocol. Percent incorporation was determined by trichloroacetic acid precipitation. In vitro binding assays were performed by incubating the TNT lysates containing the [ 35 S]methionine-labeled recombinant proteins (8 -12 fmol) in 50 l of reaction buffer (1 mM DTT, 150 mM NaCl, 20 mM Tris, pH 8.0) for 1 h at 30°C. The complex was purified immunologically using antiserum against the N terminus of ACVI (AC6N (34)), analyzed using 12.5% or 15% SDS-PAGE, and visualized by autoradiography. For the interaction between G␣ i and the C1a domain, recombinant H6-tagged GTP␥S-G␣ i1 protein (10 g) purified as described above was first bound to the Ni-NTA His⅐Bind® resin (10 l, Novagen) following the manufacturer's protocol, and then incubated with the TNT lysates containing the [ 35 S]methionine-labeled C1a-1 proteins (12 fmol) in 200 l of binding buffer (5 mM MgCl 2 , 100 mM NaCl, and 20 mM Hepes, pH 8.0) for 2 h at 4°C. The complexes bound on the Ni-NTA His⅐Bind® resins were washed with wash buffer (2 mM MgCl 2 , 200 mM NaCl, 0.5% Triton, and 20 mM Hepes, pH 8.0) three times and eluted with 20 l of the sample treatment buffer. The complexes were analyzed using 12.5% SDS-PAGE. Following electrophoresis, gels were cut into half based on the molecular weight markers. The halves of gels, which contained proteins larger than 30 kDa were transferred to polyvinylidene fluoride membranes, and levels of G␣ i1 were analyzed using the Western blot analysis described above. The other halves of the gels, which contained proteins smaller than 30 kDa, were fixed and autoradiographed as described above.

RESULTS
One important regulation for ACVI, which can be observed for only two other AC isozymes (ACI and ACV), is the inhibition mediated by G␣ i proteins (17,18). Because the N-terminal domain is highly variable among AC isozymes, we set out to investigate the role of the N-terminal domain of ACVI in G␣ imediated inhibition. The wild-type ACVI or an N terminustruncated ACVI mutant, which lacks the first 1-86 aa and exhibits a smaller size in the SDS-PAGE/Western blot analysis (ACVI-⌬A87; Fig. 2A (14)), was co-expressed with a G␣ i -coupled short form D2 dopamine receptor (D2s-R) in CHOP cells. AC assays were performed in the presence of forskolin (100 M) at the indicated concentration of a D2s-R-selective agonist (quinpirole, Quin). As shown in Fig. 2B, inhibition of ACVI-⌬A87 by D2s-R stimulation was more significant than that of wild-type ACVI at high concentrations of Quin. This observation indicates that partial truncation of the N-terminal domain might cause ACVI to be more sensitive to G␣ i -mediated inhibition. To verify this hypothesis, we next examined the inhibitory effect of myristoylated recombinant G␣ i protein on the G␣ s -evoked activities of wild-type ACVI and ACVI-⌬A87. Myristoylation of the recombinant G␣ i1 protein produced in the presence of N-myristoyltransferase (28) was verified by its faster mobility in 9% SDS-PAGE gels containing 4 M urea when compared with those of the unmodified G␣ i1 proteins prepared in the absence of N-myristoyltransferase (Fig. 3A). Consistent with the above observation, inhibition of ACVI-⌬A87 by the myristoylated G␣ i was much more significant than that of wild-type ACVI (Fig. 3B). No significant difference in the IC 50 values of the activated/myristoylated G␣ i for the wild-type and ACVI-⌬A87 was observed (16.6 and 21.5 nM, respectively). However, the extent of maximal inhibition of ACVI-⌬A87 by the activated/myristoylated G␣ i was markedly higher than that of wild-type ACVI (71.5% and 46.4% for ACVI-⌬A87 and the wild-type ACVI, respectively) (Fig. 3B), further supporting a functional role of the N-terminal domain of ACVI in G␣ i -evoked inhibition.
We next replaced the N-terminal domain of ACVI by that of type V adenylyl cyclase (ACV), the closest isozyme to ACVI in the AC superfamily. Except for the N-terminal domain, the C1a and C2 domains of ACV and ACVI are highly homologous (23). Using a two-step PCR-based mutagenesis method, we replaced the N terminus (aa 1-160) of rat ACVI with the N terminus (aa 1-242) of rat ACV. The resultant mutant (designated ACVI-N5) exhibited a larger size than that of wild-type ACVI because of the longer N terminus it possessed (Fig. 4A). In line with our previous finding (14), which suggested that the N terminus of ACVI does not contribute to the catalytic core complex, major enzymatic properties (i.e. V max and EC 50 values of forskolin and G␣ s protein) of ACVI-N5 were very similar to those of wild-type ACVI (Table I). In contrast, when co-expressed with D2s-R in CHOP cells, D2s-R-mediated inhibition of ACVI-N5 was much less effective at both low (10 nM) and high (10 M) doses of a D2s-R agonist (Quin; Fig. 4B). Because replacement of the N terminus domain markedly altered its response to G␣ i -mediated inhibition of ACVI, these results strengthened our hypothesis that the N terminus of ACVI plays an important role in G␣ i -mediated inhibition.
Dessauer and colleagues (19) previously reported that the G␣ i protein interacts with the C1a domain of ACV. It is therefore possible that the N terminus of ACVI might modulate G␣ i -mediated inhibition by interacting directly with the C1a domain. We first prepared a C1a domain of ACVI (aa 364 -587, designated C1a-1) which comprised the potential G␣ i binding domain based on a previous study of ACV (19) using an in vitro TNT translation system. As shown in Fig. 5A, interaction be-tween C1a-1 and the activated/myristoylated G␣ i protein, which harbored an H6 tag was demonstrated by an Ni-NTA resin pull-down assay. Thus, the C1a domain of ACVI, similar to ACV, bound to the G␣ i protein. Moreover, recombinant C1a domains synthesized using this in vitro translation system appeared to exhibit the proper conformation. We next performed the in vitro binding analyses of the recombinant N (aa 1-160, designated N 1-160 ) and the C1a-1 domain. Using an antibody that specifically interacts with the N-terminal domain of ACVI (34), the C1a-1 fragment could be co-immunoprecipitated with N 1-160 (Fig. 5B). Note that a smaller degradation protein sometimes appeared in the in vitro-translated C1a-1 preparation (Fig. 5B); we therefore constructed another C1a domain (aa 364 -575, designated C1a-2) based on a previous study of ACV (7). The C1a-2 protein comprises the predicted G␣ i -interacting domain of ACVI as does the C1a-1 protein and an H6 tag fused to its N terminus. As shown in Fig. 5C, C1a-2 was very stable and could also be co-immunoprecipitated with N 1-160 as could C1a-1. The H6 tag did not appear to affect the interaction between the N and C1a domains in that both C1a-1 and C1a-2 proteins could be co-immunoprecipitated by N 1-160 . The C1a-2 protein therefore was used in the following experiment.
To determine the interaction region on the N terminus with the C1a domain, a truncated N-terminal domain (aa 1-86, designated N 1-86 ) was produced for the in vitro binding assay. As shown in Fig. 5D, N 1-86 was also effectively co-immunoprecipitated with the recombinant C1a-2 domain as was the N 1-160 fragment. Under the conditions used, ϳ10% of the C1a-2 protein in the reaction was pulled down with both N 1-160 and N 1-86 . The C1a-interacting region of the N terminus might therefore reside in the region containing aa 1-86.
We further performed experiments to determine where on the C1a domain the N-terminal domain binds. Three shorter recombinant proteins comprising different portions of the C1a-2 protein (aa 364 -505, 399 -540, and 434 -575, designated C1a-3, C1a-4, and C1a-5, respectively) were prepared in vitro. As shown in Fig. 6, the N 1-160 protein could immunoprecipitate all three of these recombinant C1a proteins tested. The overlapping region (aa 434 -505) of these three recombinant C1a proteins thus appeared to contain the structural determinants for the binding of the N-terminal domain.
Based on the available crystallographic structure of the catalytic complex of ACs (7), we built a computer model of the catalytic core composed of the C1a and C2 domains of ACVI (Fig. 1B). Dessauer and colleagues (19) reported that G␣ i binds to the C1a domain at the cleft formed by the ␣2 and ␣3 helices. Mutations of several residues located in these two helices caused a significant reduction in the affinity between G␣ i and   FIG. 2. Truncation of the N terminus of ACVI alters its response to a G␣ i -coupled receptor. A, membrane proteins (100 g per lane) from CHOP cells expressing the wild-type (WT) or the N terminus-truncated ACVI mutant (ACVI-⌬A87) were subjected to Western blot analysis using AC6D. The thick and thin arrows indicate the wild-type and truncated ACVI proteins, respectively. B, membrane fractions collected from CHOP cells transiently transfected with D2s-R plus ACVI-WT or ACVI-⌬A87 were used for the AC activity assay. AC activity evoked by forskolin (100 M) was measured in the presence of a D2s-selective agonist (Quin) at the indicated concentrations. Values are expressed as percentages of forskolin-evoked ACVI activity measured in the absence of Quin (574.6 Ϯ 105.7 pmol/(min⅐mg) for wild-type ACVI and 673.6 Ϯ 117.3 pmol/(min⅐mg) for ACVI-⌬A87), and represent the mean Ϯ S.E. of four independent experiments. Differences between the Quin-mediated inhibition of forskolin-evoked AC activities of WT and those of ACVI-⌬A87 at 3 and 10 M are statistically significant: *, p Ͻ 0.05, compared with WT at the indicated concentration (by Student's t test).
ACV. Our results demonstrated that the N terminus binding region of the C1a domain (aa 434 -505) containing the ␣3 helix, which is located on the surface of the catalytic core complex (Fig. 1B), is likely accessible to the N-terminal domain of ACVI. Amino acid alignment analysis with ACV suggested that two ACV-equivalent residues (Leu-472 and Val-476) in the ␣3 helix of the C1a domain of ACVI might be important for interaction with the G␣ i protein (Supplementary Materials, Fig. S1). To test whether Leu-472 and Val-476 on the ␣3 helix might also be important for the interaction of the N and C1a domains, we created a recombinant C1a-2 mutant carrying mutations at Leu-472 and Val-476 (designated mC1a-2). As observed by pull-down assays, mC1a-2 did not interact with the N terminus as did the recombinant wild-type C1a-2 protein (Fig. 7). Thus, Leu-472 and Val-476 located in the ␣3 helix of the C1a domain of ACVI might play a critical role in the interaction with the N-terminal domain.
To examine whether Leu-472 and Val-476 were important for interacting with the G␣ i protein, we mutated Leu-472 and Val-476 into alanine in the full-length ACVI and the N terminus-truncated ACVI mutant (ACVI-⌬A87). The resultant mutants (designated ACVI-L472A-V476A and ACVI-⌬A87-L472A-V476A, respectively) were then subjected to regulation by myristoylated G␣ i . With the N terminus being truncated, which would free the C1a domain from binding to the N terminus, the Leu-472/Val-476 mutation markedly reduced the in-hibition evoked by G␣ i at all dosages examined (ACVI-⌬A87-L472A-V476A versus ACVI-⌬A87, Fig. 8). With an intact N terminus, the effect of the Leu-472/Val-476 mutation (ACVI -L472A-V476A versus wild type, Fig. 8) depended on G␣ i dosages: little or no effect at low dosages (22 and 44 nM, ϳ1to 2-fold of the IC 50 values of ACVI) and moderate reduction at high dosages (175 and 350 nM, ϳ10to 20-fold of the IC 50 values of ACVI). In combination, truncation of the N terminus did not alter the inhibition of low dosage G␣ i on a mutant for which a potential G␣ i -binding site involving Leu-472 and Val-476 had been mutated, but as the dosage of G␣ i increased substantially, increased inhibition became evident (ACVI-⌬A87-L472A-V476A versus ACVI -L472A-V476A, Fig. 8). DISCUSSION In the present study, we investigated the contribution of the N-terminal domain of ACVI to G␣ i -mediated inhibition by determining 1) the effect of G␣ i -mediated regulation on various ACVI mutants of a truncated N terminus and/or amino acid mutations of the C1a domain at its N-and G␣ i -binding site and 2) the in vitro interaction of the recombinant N terminus and the C1a domain with which the G␣ i protein binds. Enzymatic analyses revealed that the absence of aa 1-86 created a conformation of ACVI, which enabled the resultant mutant to be more severely inhibited by a G␣ i -coupled receptor (D2s-R; Fig.  2) or the myristoylated recombinant G␣ i protein (Fig. 3). Sub-  (28) in bacterial BL21-DE3 were subjected to electrophoresis on a 9% SDS-polyacrylamide gel containing 4 M urea, followed by Western blot analyses using an anti-G␣ i antibody as described under "Experimental Procedures." The thick and thin arrows indicate the lipid-modified and unmodified G␣ i1 proteins, respectively. B, membrane fractions collected from CHOP cells transiently transfected with ACVI-WT or ACVI-⌬A87 were used for the AC activity assay. AC activity evoked by the activated G␣ s protein (0.19 M) was measured in the presence of activated/myristoylated recombinant G␣ i1 protein (G␣ i1 ) at the indicated concentrations. Values are expressed as percentages of G␣ s -evoked ACVI activity measured in the absence of G␣ i1 (146.2 Ϯ 30.3 pmol/(min⅐mg) for wild-type (WT) ACVI and 108.7 Ϯ 21.2 pmol/(min⅐mg) for ACVI-⌬A87) and represent the mean Ϯ S.E. of four independent experiments. Differences between the G␣ i1 -mediated inhibition of G␣ s -evoked AC activities of WT and those of ACVI-⌬A87 are statistically significant: *, p Ͻ 0.05, compared with WT at the indicated concentration (by Student's t test). stituting the full-length N terminus domain of ACVI with that of ACV also markedly altered the inhibition evoked by D2s-R without affecting other enzymatic properties of ACVI ( Fig. 4 and Table I). In vitro binding demonstrated that the N domain binds its catalytic core complex at the C1a domain. In addition, the interacting regions located on the N and C1a domains comprised aa 1-86 and 434 -505, respectively (Figs. 5D and 6). Computer modeling suggested that this N terminus-interacting region on the C1a domain contains part of the potential G␣ i -interacting region (aa 452-478). It is important to note that the C1a-interacting region (aa 1-86) of the N terminus delineated by the in vitro binding assay is consistent with the observation that the N-truncated mutant (ACVI-⌬A87) was more sensitive to G␣ i -mediated inhibition, presumably due to the increased accessibility of the G␣ i -interacting site on the C1a domain. More interestingly, simultaneous mutation of two residues (Leu-472 and Val-476) reduced the inhibition of the N-truncated ACVI variant (ACVI-⌬A87) evoked by G␣ i (Fig. 8) and markedly hampered the binding of the N terminus to the C1a domain (Fig. 7). These two residues thus play an important role in G␣ i -mediated inhibition in the absence of an intact N terminus and apparently also act as a modulator of the binding between the N and C1a domains. Mutations of Leu-472 and Val-476 in the full-length ACVI might not cause significant effects on G␣ i -mediated regulation, because such mutations could elicit dual effects: inactivation of a G␣ i -interacting site and dissociation of the N and C1a domains to enable the inactivated G␣ i -interacting site to be more accessible to G␣ i . In line with this model, we observed no difference in G␣ i -mediated inhibition of the enzyme between mutants ACVI-⌬A87-L472A-V476A and ACVI -L472A-V476A at low dosages (i.e. around the IC 50 value) of G␣ i proteins (Fig. 8), presumably because truncation of the N terminus mainly exposed the G␣ i -interacting site of the C1a domain, which had already been inactivated by the mutations of Leu-472 and Val-476. Together with the observations that, at low dosages of G␣ i , the G␣ i inhibition of wild type (ACVI) was significantly reduced by the truncation of the N domain (ACVI-⌬A87) but essentially was not changed by the Leu-472/Val-476 mutation (ACVI-L472A-V476A), we concluded that binding of N to C1a contributed to inhibition of ACVI evoked by low levels of G␣ i . The N terminus domain also plays a critical role in G␣ i -mediated inhibition at high dosages of G␣ i , because its removal enhanced the inhibition of the full-length mutant (ACVI-L472A-V476A). The molecular mechanism underlying such an action of the N terminus at high levels of G␣ i (i.e. ϳ10to 20-fold of the IC 50 value) is currently unclear. Because interaction of N and C1a domains was not observed with the mutations at Leu-472/Val-476 as assessed by the pull-down assay (Fig. 7), N and C1a domains in ACVI-L472A-V476A were likely to be dissociated. In the absence of the binding of N terminus to C1a, truncation of the N terminus retained the ability to enhance G␣ i -mediated inhibition at high dosages of G␣ i , suggesting that binding to C1a might not be relevant to how the N terminus suppressed the inhibition of ACVI by high dosages of G␣ i . It is possible that the N terminus might interact with an additional G␣ i -interacting site other than that on the C1a domain. Alternatively, the pull-down immunoprecipitation assays utilized in the present study might not be sensitive enough to detect weak interactions. We cannot rule out the possibility that a weak interaction between N and C1a domains persisted in ACVI-L472A-V476A and moderately hindered the accessibility of the G␣ i -interacting site of the C1a domain at high dosages of G␣ i . Collectively, data presented herein suggest a novel function of the N terminus of ACVI in G␣ i -mediated regulation. At least part of the action of the N terminus was mediated by binding to the ␣3 helix of the C1a domain, which subsequently might hinder the accessibility of the cleft formed by the ␣2 and ␣3 helices of the C1a domain (Fig. 1B) to G␣ i at low levels of G␣ i protein. FIG. 4. Altering the extent of inhibition mediated by a G␣ icoupled receptor through replacing the N terminus of ACVI with that of ACV. A, membrane proteins (100 g per lane) from CHOP cells expressing the wild-type (WT) or the ACVI mutant containing the N terminus of ACV (ACVI-N5) were subjected to Western blot analysis using AC6D. The thick and thin arrows indicate wild-type and ACVI-N5 proteins, respectively. B, membrane fractions collected from CHOP cells transiently transfected with D2s-R plus ACVI-WT or ACVI-N5 were used for the AC activity assay. AC activity evoked by forskolin (100 M) was measured in the presence of a D2s-selective agonist (Quin) at the indicated concentrations. Values are expressed as percentages of forskolin-evoked ACVI activity measured in the absence of Quin and represent the mean Ϯ S.E. of three independent experiments. a, p Ͻ 0.5. Specific comparisons between the Quin-treated and the control group of each ACVI variant were performed using the Dunnett method. b, p Ͻ 0.5. Specific comparisons between the WT and the ACVI-N5 mutants under the indicated conditions were performed using the Dunnett method. The N-terminal domains of AC isozymes are highly variable and are generally considered regulatory domains. For example, the N-terminal domain of ACVIII has been implicated in calmodulin binding and the Ca 2ϩ -dependent activation of ACVIII (15). We previously showed that the N terminus of ACVI is crucial for PKC-mediated inhibition (14,16). Removal of the most-N-terminal portion (aa 1-86) of ACVI did not affect its general enzymatic properties (14). In contrast, as demonstrated in the present study, truncation of this region markedly enhanced the regulation mediated by G␣ i (Figs. 2 and 3). The N-terminal domain of ACV did not exert a significant effect on the enzymatic properties or on certain regulatory modes (such as the inhibition mediated by G␣ i or calcium (12,35)). Replacement of the N terminus of ACVI with that of ACV did not affect the enzymatic properties of ACVI either (Table I) but markedly altered the inhibition mediated by D2s-R (Fig. 4). This finding further strengthens our hypothesis that, in addition to its involvement in PKC-mediated inhibition, the N terminus of ACVI might also play a neutralizing role in G␣ i -mediated inhibition. Note that replacement of the N terminus of ACVI with an irrelevant domain bearing no sequence homology can only be expected to alter its function (i.e. G␣ i -mediated regulation) but would not necessarily create the same enhancing effect as that observed in the N-truncated ACVI mutant (ACVI-⌬A87, Fig. 2). One possible explanation for such a finding is that the N terminus of ACV is much longer than that of ACVI (242 versus 160 aa) and thus might impose a steric hindrance over the G␣ i -interacting site on the C1a domain and reduce the inhibition evoked by D2s-R.
Ample evidence has previously been shown that different G␣ i -coupled GPCRs exhibit distinct abilities to activate G␣ i proteins and thus lead to different extents of G␣ i -dependent inhibition of AC (36 -38). In response to stimulation by different GPCRs, G␣ i protein might inhibit only certain, but not all, G␣ i -sensitive AC isozymes. For example, the D3 dopamine receptor significantly inhibits the activity of ACV but not that of ACVI, which belongs to the same subfamily as ACV. On the contrary, the extents of D2-R-mediated inhibition are similar for ACV and ACVI (39). The mechanism underlying such specificity is not fully understood. One possible explanation suggested by those authors is that the actions of G␣ i protein on ACVI and ACV may differ. Alternatively, stimulation of differ- For in vitro binding analysis, 12 fmol of the N 1-160 and 8 fmol of the C1a-1 proteins were incubated for 60 min at 30°C to allow complex formation. Immunoprecipitation was performed using an antiserum against the N terminus of ACVI (AC6N (34)). The protein complexes were separated by 15% SDS-PAGE, dried, and visualized by autoradiography. The thick and thin arrows indicate C1a-1 and N 1-160 , respectively. The star indicates a degradation product of C1a-1. C, in vitro binding analyses of the N 1-160 and C1a-2 (aa 364 -587, with an H6 tag fused to its N terminus domain) were performed and analyzed as in B, except that the protein complexes were separated by 12.5% SDS-PAGE. The thick and thin arrows indicate C1a-2 and N 1-160 , respectively. D, in vitro binding analyses of N 1-86 and C1a-2 were carried out and quantified as in B. The thick and thin arrows, respectively, indicate C1a-2 and N 1-86 . Ni-NTA, Ni-NTA His⅐Bind® resin. ent GPCRs might cause specific conformational changes in G␣ i and, subsequently, lead to selective regulation of AC isozymes. As discussed above, although the transmembrane regions and catalytic core complexes of AC isozymes are similar, they diverge in their N-terminal domains. This raises the possibility that the N terminus of AC isozymes might influence their response to G␣ i by modulating the G␣ i -binding site on the catalytic core complex. The association of its N terminus and the C1/C2 catalytic complex endows ACVI with a fine-tuned, regulatory mode for G␣ i -mediated inhibition. One interesting observation in the data presented herein is that the difference in G␣ i -evoked inhibition between ACVI-⌬A87 and the wildtype ACVI was more significant when the enzyme was stimulated by G␣ s (Fig. 3) rather than by forskolin (Fig. 2). Such a difference might have resulted from the observation that G␣ i proteins are more effective in suppressing G␣ s -activated AC activity than forskolin-stimulated AC activity as previously reported (18). Alternatively, this difference might have been caused by the G␣ i -coupling specificity of D2s-R utilized herein. Specifically, in at least two different cell lines, the D2s-R has been shown to inhibit forskolin-stimulated ACs preferentially through G␣ i 2, but not G␣ i1 (40,41) . The endogenous G␣ i isoform employed by the D2s-R in CHOP cells to mediate the inhibition of ACVI therefore might not be G␣ i1 . It is plausible that the regulatory mode of ACVI by G␣ i1 might differ slightly from those by other G␣ i isoforms, which would contribute to the lower effectiveness of N terminus truncation of ACVI as seen in Fig. 2B.
The molecular basis underlying the regulation of ACVI by G␣ i has heretofore not been extensively investigated. Based on a mutational analysis of ACV (19), the basic G␣ i -interacting region of ACVI might reside in the cleft formed by the ␣2 and ␣3 helices of the C1a domain. Our mutation analyses and computer modeling revealed that the N terminus might interact with C1a at the ␣3 helix. In addition, mutations of Leu-472 and Val-476 located in the ␣3 helix caused dual effects, including a reduction in G␣ i -mediated inhibition in the absence of an intact N terminus (Fig. 8) and dissociation of the N and C1a domains (Fig. 7). These two residues (Leu-472 and Val-476) thus appeared to be important for the interaction between the C1a and N domains, and between the C1a domain and G␣ i . Nevertheless, the exact binding sites on the C1a domain for the N terminus, and G␣ i might not overlap, because the difference in the G␣ i -mediated inhibition between wild-type ACVI and the N terminus-truncated ACVI variant (ACVI-⌬A87) could not be reversed by increasing the concentration of the D2 agonist (Quin) or the activated/myristoylated G␣ i (Figs. 2 and 3). Binding of the N terminus to the C1a domain thus appears to negatively regulate the inhibition evoked by low levels of G␣ i in a non-competitive manner. We were unable to further characterize the binding of the N terminus domain and G␣ i to the C1a FIG. 6. Amino acids 434 -505 comprising the N terminus-interacting domain. Recombinant N terminus (aa 1-160, N 1-160 ) and three C1a fragments (aa 364 -505, 399 -540, and 434 -575 for C1a-3, C1a-4, and C1a-5, respectively) were produced using the in vitro TNT system in the presence of [ 35 S]methionine. For the in vitro binding analysis, 12 fmol of the N 1-160 protein and 12 fmol of the indicated C1a-1 variant were incubated for 60 min at 30°C to allow complex formation. Immunoprecipitation was performed using an antiserum against the N terminus of ACVI (AC6N (34)). The protein complexes were separated by 15% SDS-PAGE and visualized by autoradiography. The thick and thin arrows indicate the C1a-1 variant and N 1-160 protein, respectively. FIG. 7. Mutation of the potential G␣ i -interacting residues (Leu-472 and Val-476) in the ␣3 helix hampered the binding of the C1a domain to the N terminus. The recombinant N 1-160 protein and the indicated C1a-2 fragments were produced using the in vitro TNT system in the presence of [ 35 S]methionine. In vitro binding analysis was carried out as described in Fig. 5. Note that, unlike C1a-2, mC1a-2 does not contain an H6 tag fused to its N terminus domain, and therefore is smaller than C1a-2. The thick and thin arrows indicate C1a-2 and mC1a-2, respectively. The arrowhead indicates the N terminus. The asterisk indicates a minor degradation product of mC1a-2.
FIG. 8. Effect of mutations of Leu-472 and Val-476 on the G␣ imediated inhibition of wild-type ACVI and the N-truncated ACVI variant. Membrane fractions collected from CHOP cells transiently transfected with the indicated ACVI variant were used for the AC activity assay. AC activity evoked by the activated recombinant G␣ s protein (95 nM) was measured in the absence or presence of the activated/myristoylated recombinant G␣ i1 protein (G␣ i1 ) at the indicated concentrations. Values are expressed as percentages of G␣ s -evoked ACVI activity measured in the absence of G␣ i1 (264.3 Ϯ 48.9 pmol/ (min⅐mg) for wild-type (WT) ACVI; 231.0 Ϯ 83.0 pmol/(min⅐mg) for ACVI-⌬A87; 333.6 Ϯ 61.8 pmol/(min⅐mg) for ACVI-L472A-V476A; and 268.5 Ϯ 102.2 pmol/(min⅐mg) for ACVI-⌬A87-L472A-V476A) and represent the mean Ϯ S.E. of 4 -6 independent experiments. Statistical significance was evaluated by one-way analysis of variance followed by the Student-Newman-Keuls method. a, p Ͻ 0.05. Specific comparisons between variants carrying the intact and the truncated N terminus (i.e. ACVI wild-type versus ACVI-⌬A87, or ACVI-L472A-V476A versus ACVI-⌬A87-L472A-V476A) are shown. b, p Ͻ 0.05. Specific comparisons between variants with and without mutations at Leu-472/Val-476 (i.e. ACVI wild-type versus ACVI-L472A-V476A, or ACVI-⌬A87 versus ACVI-⌬A87-L472A-V476A) are shown. domain in vitro due to the extremely low expression and instability of the recombinant N terminus protein (data not shown). Consistently, sequence analyses (PONDR, Molecular Kinetics, Pullman, WA) predicted that the C1a-interacting region (aa 1-86) on the N terminus is likely to be intrinsically unstructured (42). The potentially highly disordered structure of the N terminus also suggests that it is likely to form a complex with other peptides (e.g. the C1a domain) to stabilize its structure. It remains to be determined whether the interaction between the N and the C1a domains can be regulated. It is possible that the N terminus of ACVI might play a fine-tuning role and disassociate from the catalytic core complex under certain regulating condition, which subsequently leads to increased accessibility of the G␣ i protein to its interacting cleft on the C1a domain. Exposure of an active site of an enzyme covered by its regulatory domain upon stimulation has many precedent examples, including protein kinase C (PKC). Activation of conventional PKC causes a conformational change that leads to exposure of the kinase domain originally covered by its regulatory domain in the resting stage (31).
In summary, we provide evidence to demonstrate that the N terminus of ACVI interacts with its catalytic core and plays an important role in the regulation of G␣ i -mediated inhibition. These findings further attribute the functional role of the Nterminal domain to the heterogeneity of the AC superfamily and add additional dimensions to the specificity of G␣ i -mediated inhibition.