Identification and characterization of a widely expressed form of adenylyl cyclase.

A novel mammalian adenylyl cyclase was identified by reverse transcription-polymerase chain reaction amplification using degenerate primers based on a conserved region of previously described adenylyl cyclases (Premont, R. T. (1994) Methods Enzymol. 238, 116-127). The full-length cDNA sequence obtained from mouse brain predicts a 1353-amino acid protein possessing a 12-membrane span topology, and containing two regions of high similarity with the catalytic domains of adenylyl cyclases. Comparison of this novel adenylyl cyclase with the eight previously described mammalian enzymes indicates that this type 9 adenylyl cyclase sequence is the most divergent, defining a sixth distinct subclass of mammalian adenylyl cyclases. The AC9 gene has been localized to human chromosome band 16p13.3-13.2. The 8.5-kb mRNA encoding the type 9 adenylyl cyclase is widely distributed, being readily detected in all tissues tested, and is found at very high levels in skeletal muscle and brain. AC9 mRNA is found throughout rat brain but is particularly abundant in hippocampus, cerebellum, and neocortex. An antiserum directed against the carboxyl terminus of the type 9 adenylyl cyclase detects native and expressed recombinant AC9 protein in tissue and cell membranes. Levels of the AC9 protein are highest in mouse brain membranes. Characterization of expressed recombinant AC9 reveals that the protein is a functional adenylyl cyclase that is stimulated by Mg2+, forskolin, and mutationally activated Gsα. AC9 activity is not affected by Ca2+/calmodulin or by G protein βγ-subunits. Thus AC9 represents a functional G protein-regulated adenylyl cyclase found in brain and in most somatic tissues.

Adenylyl cyclases (EC 4.6.1.1) convert intracellular ATP into cyclic-3Ј,5Ј-adenosine monophosphate (cAMP). In most tissues, the activity of adenylyl cyclase is controlled by hormones through the action of specific receptors coupled to heterotrimeric G proteins. 1 Many hormones stimulate adenylyl cyclase activity by binding to G protein-coupled receptors on the cell surface, which then activate the stimulatory G protein, G s (1,2). Activation of G s involves the receptor-dependent exchange of bound GDP for GTP by the G s ␣ subunit and the dissociation of this activated ␣ s subunit from the ␤␥ dimer. The GTP-bound G s ␣ subunit then associates with adenylyl cyclase and greatly accelerates the rate of cAMP synthesis. Similarly, hormonal inhibition of adenylyl cyclase involves receptor activation of the inhibitory G i proteins. Binding of GTP-bound G i ␣ subunit to adenylyl cyclase reduces the rate of cAMP synthesis (3,4).
Recent studies have demonstrated a great diversity of G protein-regulated adenylyl cyclases in mammalian tissues. Eight distinct adenylyl cyclase cDNAs have been identified and cloned from mammalian tissues (5,6). These adenylyl cyclases can be categorized into five distinct classes based on sequence and functional similarities (7,8). Type 1 adenylyl cyclase is found mainly in brain and is stimulated by Ca 2ϩ /calmodulin and inhibited by G protein ␤␥-subunits, both independent of G s activation (5,6). The highly similar types 2, 4, and 7 adenylyl cyclases form the largest known subfamily of adenylyl cyclases. Type 2 is found in brain and lung, while types 4 and 7 are more widely distributed (5, 6, 9 -11). Types 2 and 4 share the property of being highly activated by G protein ␤␥-subunits, but only in the presence of simultaneous G s ␣ activation (12,13). Regulation of type 7 by ␤␥-subunits has not been reported. Type 3 adenylyl cyclase is highly abundant in olfactory neuroepithelium and is thought to mediate olfactory receptor responses (14). The type 3 enzyme is stimulated by Ca 2ϩ /calmodulin but is unaffected by G protein ␤␥-subunits (5,6). The highly similar type 5 and 6 adenylyl cyclases form a twomember adenylyl cyclase subfamily. Both enzymes are widely expressed at low levels, with very high expression of type 5 in striatum and heart and of type 6 in brain and heart (7,8,15,16). Both are reported to be inhibited by low levels of Ca 2ϩ , but to be unaffected by G protein ␤␥-subunits (7,17). Type 8 adenylyl cyclase is reported to be found only in brain and to be stimulated by Ca 2ϩ /calmodulin (18 -20).
Despite the identification of these eight adenylyl cyclase enzymes, the distributions and/or functional properties of these various adenylyl cyclases have been inconsistent with any particular enzyme being the "general" adenylyl cyclase found in most somatic tissues. Most of the enzymes identified by cloning appear to have limited tissue distributions (types 1, 2, 3, and 8), to have rather low level expression in somatic tissues (types * This work was supported by National Institutes of Health Grant HL 16037 to R. J. Lefkowitz and by the Howard Hughes Medical Institute. 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. The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM

-7)
, or to perform specialized functions that were essentially unknown prior to characterization of one or another of these cloned adenylyl cyclases (such as G protein ␤␥-subunit stimulation of the members of the type 2 subfamily). A novel ninth adenylyl cyclase has been identified, which is shown to have high levels of mRNA and protein in a wide variety of tissues. The primary structure of the type 9 adenylyl cyclase 2 is quite distinct from members of the other five known adenylyl cyclase subfamilies and thus constitutes a sixth subfamily of mammalian adenylyl cyclases.

MATERIALS AND METHODS
DNA Amplification-Adenylyl cyclase fragments were amplified using the polymerase chain reaction, using degenerate oligonucleotide primers based on conserved regions of the second intracellular domain of mammalian adenylyl cyclases (5Ј-CCGCAGCTCGAGAARATHAA-RACIRTIGG, encoding KIKTIG, and 5Ј-CCGGGACTCGAGACRTTI-ACIGTITTICCCCA, encoding WGNTVN), 3 as described previously (21). Briefly, first-strand cDNA was prepared from mouse tissue poly(A) RNAs using SuperScript Moloney murine leukemia virus reverse transcriptase (Life Technologies, Inc.). 10 ng of cDNA was subjected to amplification in 100-l reactions containing 1 ϫ polymerase buffer, 1.5 mM MgCl 2 , 200 M each dNTP, 500 nM each oligonucleotide primer, and 2.5 units of Taq DNA polymerase (Promega). Reactions with all components except polymerase were heated 5 min at 95°C and cooled to 72°C for the addition of polymerase. Reactions were cycled 35 times for 1 min at 95°C, 1 min at 55°C, and 3 min at 72°C followed by a final 10 min at 72°C. Products were separated on 4% NuSieve (FMC Bioproducts) agarose gels, and appropriate size bands were excised and subcloned either in pBS II (Stratagene) after XhoI digestion (underlined sites in primers) or directly TA-subcloned into pCRII (Invitrogen). Amplification with these primers yielded a novel 288-bp adenylyl cyclaselike sequence from mouse brain cDNA (21).
Library Screening-An oligo(dT)-primed mouse brain cDNA library in Uni-ZAP II was obtained from Stratagene. 1.2 ϫ 10 6 phage were screened by hybridization (22) to a mixture of mouse adenylyl cyclase type 7, 8, and 9 probes obtained by PCR (21). The 44 positively hybridizing phage were then classified by amplification from the virus T3 primer to antisense primers specific to individual mouse adenylyl cyclase isoforms or by hybridization of PCR-amplified insert DNAs (T3 to T7 primers) to individual adenylyl cyclase probes. Eight phage, representing five independent clones, were determined to encode portions of type 9 adenylyl cyclase. Insert DNAs were recovered into pBS by in vivo excision using ExAssist helper phage according to Stratagene's protocol.
A directional, specifically primed mouse brain cDNA library was constructed in the ZAP-Express vector (Stratagene). Mouse brain total RNA was extracted using RNazol (Tel-Test), and poly(A) RNA was purified using OligoTex-dT beads (Qiagen). Five g of poly(A) RNA was used as template for first-strand cDNA synthesis using me 7 -dCTP and Moloney murine leukemia virus reverse transcriptase (ZAP cDNA synthesis kit, Stratagene) and an antisense primer immediately after the stop codon of the type 9 adenylyl cyclase cDNA (5Ј-ACACACCTCGAG-CACCTTGGCAGCTCTCGGCGCT). A separate reaction was performed using an antisense primer specific for the 3Ј-untranslated region immediately after the stop codon of the mouse type 8 adenylyl cyclase (5Ј-ACACACCTCGAGAAGAAACACAAAGAAAATGCTT). Secondstrand cDNAs were ligated to EcoRI adaptors, digested with XhoI to cut the underlined sites in the original primers, and size-selected for length greater than 2 kb on Sephacryl S-500 columns (cDNA sizing column, Life Technologies). The two specifically primed cDNA mixtures were pooled and ligated with EcoRI/XhoI-digested ZAP-Express viral DNA (Stratagene). 2.5 ϫ 10 5 phage of this unamplified library were screened by hybridization to the 1.1-kb EcoRI/BamHI fragment of type 9 clone 22.1, and 66 positive phage were identified. Initial positive phage were sorted for length by amplification of insert 5Ј ends from virus T3 primer to an antisense adenylyl cyclase type 9 primer (5Ј-GCTCTGGATCTT-GGTGCG). Specific bands were identified on Southern blots by hybrid-ization to an end-labeled internal primer (5Ј-GAAGAAGGTGAGGATA-AGCTC). Three clones with the longest 5Ј sequences, from 2 to 3 kb longer than the previously known sequence, were plaque-purified. Insert DNAs were recovered in the pBK-CMV vector by in vivo excision with the ExAssist helper phage.
Extreme 5Ј end sequences were obtained using the 5Ј-RACE protocol (23) using mouse brain 5Ј-RACE-ready cDNA (Clontech). Sequences were amplified in two steps from nested antisense AC9 primers and an anchor primer ligated to the 3Ј end of the first-strand cDNA following Clontech's protocol, and product bands visible on agarose gels were TA-subcloned into pCRII (Invitrogen). Two complete rounds of 5Ј-RACE reactions yielded 500 bp of additional AC9 sequence, including the candidate start codon preceded by an in-frame stop codon. 5Ј-RACE sequences were verified by polymerase chain reaction amplification from novel sequences to known AC9 sequences and subcloning and sequencing of DNA products of the predicted size. The complete open reading frame was reconstructed in pBK-CMV from three overlapping clones, including one of these 5Ј-RACE verification clones, using unique restriction sites.
DNA inserts in pBS, pBK-CMV, or pCRII plasmids were sequenced by chain termination using [␣-32 P]dATP and Sequenase version 2 T7 DNA polymerase (U.S. Biochemicals), from vector-and sequence-specific oligonucleotide primers. The sequences of both strands of selected clones were determined. DNA and protein sequences were analyzed using the GeneWorks program (Intelligenetics). Pairwise amino acid similarity was calculated using the GAP program of the GCG DNA analysis package (24). DNA data base searches were performed using the NCBI e-mail server using the FASTA program (25).
Northern Blotting-Mouse poly(A) RNAs transferred to a nylon filter (mouse MTN, Clontech) were probed by hybridization to the 1.2-kb EcoRI/XhoI fragment of clone 22.1 (bases 3278 -4522) encoding the C2 domain and the 3Ј-untranslated region and to the 0.9-kb BamHI fragment of the full-length AC9, encoding the amino-terminal coding sequence through membrane span 6. DNA was radiolabeled with [␣-32 P]dCTP by random priming (NEBlot kit, New England Biolabs). The filter was hybridized overnight at 42°C in 5 ϫ SSPE, 10 ϫ Denhardt's, 1% SDS, 100 g/ml boiled calf thymus DNA, 50% formamide and washed for 1 h at room temperature in 1 ϫ SSC buffer and for 15 min at 60°C in 0.2 ϫ SSC buffer (22).
In Situ mRNA Hybridization-A 1.2-kb EcoRI-XhoI fragment of AC9 clone 22.1 in pBlueScript was linearized with XhoI or EcoR I and transcribed in the presence of [␣-33 P]UTP and T3 or T7 RNA polymerase to prepare sense or antisense cRNA probes, respectively. Probes were hydrolyzed by treatment with 0.1 M sodium carbonate (pH 8.2) for 20 min at 60°C to reduce the length to 150 -200 bp. In situ hybridization was performed essentially as described previously (26). Wistar rats (180 -200 g) were sacrificed by decapitation, and the brains were frozen in dry ice-isopentane and stored at Ϫ80°C. Sections (14 M) were cut on a cryostat, thaw-mounted on gelatin-coated slides, and fixed with 4% paraformaldehyde. After acetylation with acetic anhydride (0.25%), sections were dehydrated and defatted by an ethanol/chloroform series: 5 min in each of 70, 90, and 100% ethanol, 10 min in chloroform, and 5 min in 100% ethanol, followed by air drying. Slides were preincubated in a humid chamber at 37°C for 6 -18 h with 150 l of prehybridization buffer (50% formamide, 4 ϫ SSC, 2 ϫ Denhardt's reagent, 10 mM Tris-HCl (pH 7.4), 1 mM EDTA, 0.1% SDS, 100 g/ml denatured salmon sperm DNA, and 100 g/ml yeast tRNA). Hybridization was performed at 55°C overnight in 100 l of prehybridization buffer containing 10% dextran sulfate and 10 7 cpm/ml cRNA probe. Slides were rinsed with 1 ϫ SSC buffer and incubated for 30 min at 37°C with 20 g/ml RNase A and 50 units/ml RNase T 1 in 100 mM Tris-HCl (pH 8.0), 500 mM NaCl, 1 mM EDTA. Slides were washed with 1 ϫ SSC for 30 min at 60°C, and three times with 0.1 ϫ SSC for 30 min at 60°C. Slides were exposed to BioMax film (Kodak) for 10 days.
Expression of AC9 -The reconstructed full-length type 9 adenylyl cyclase cDNA in pBK-CMV was excised using SpeI and NotI, and subcloned into the pVL1393 baculovirus shuttle vector digested with XbaI and NotI. pVL-AC9 was cotransfected with BaculoGold virus DNA (PharMingen) into Sf9 cells to obtain a recombinant AC9 baculovirus stock. The full-length mouse adenylyl cyclase type 8 4 was subcloned into pVL1393 in the same manner to produce recombinant AC8 baculovirus.
Sf9 cells (1.5 ϫ 10 6 /ml) grown in Grace's insect media were infected with recombinant baculoviruses at a MOI of 5 and grown in suspension culture for an additional 48 -50 h at 27°C. Cells were sedimented, 2 Due to previous uncertainty about the relationship between mouse AC7 (9, 15) and a human AC (10,52), which was originally referred to as AC9 but subsequently has been shown to be the human AC7 homolog (11), earlier reports of the sequence described herein (21, 37, 53) referred to it as "AC10." 3 The nucleotide degeneracies are as follows: R represents A or G; H represents A, C, or T; I represents inosine. 4 R. T. Premont, unpublished results.
washed with phosphate-buffered saline, and resuspended in 20 mM HEPES, pH 7.2, 5 mM EDTA, 20 mM NaCl, and a mixture of protease inhibitors (2 g/ml aprotinin, 10 g/ml benzamidine, 4 g/ml leupeptin, 1 g/ml pepstatin, and 100 M phenylmethylsulfonyl fluoride) (lysis buffer). Lysates were frozen in liquid nitrogen and stored at Ϫ80°C. Thawed cell lysates were homogenized with a Dounce homogenizer and spun at 1000 ϫ g to remove debris. Membranes were then pelleted at 20,000 ϫ g, washed with 400 mM NaCl in lysis buffer, and resuspended at 1-5 mg/ml in 20 mM HEPES, pH 7.2, 2 mM EDTA, 200 mM sucrose, 2 mM dithiothreitol, and protease inhibitors. Membrane aliquots were frozen in liquid nitrogen and stored at Ϫ80°C. For expression in CMT cells, the mouse AC9 cDNA and the human AC8 cDNA (20) were inserted into pcDNAI-Amp (Invitrogen). CMT cells (27) were transfected by a modified DEAE-dextran procedure (28) and grown for 72 h prior to membrane preparation. Cells were washed twice with phosphate-buffered saline and lysed in 2 ml of 10 mM Tris-HCl, pH 8.0, 0.1 mM EGTA, 0.5 mM phenylmethylsulfonyl fluoride, 5 mM ␤-mercaptoethanol. The lysate was spun at 100 ϫ g for 5 min, and the supernatant was respun at 17,000 ϫ g for 30 min. The pelleted membranes were washed twice with homogenization buffer and stored in aliquots in the same buffer frozen at Ϫ80°C.
Antisera and Immunoblotting-The carboxyl-terminal portion of type 9 adenylyl cyclase expressed as a fusion with glutathione Stransferase was used to raise polyclonal antisera in rabbits as described (29). The cDNA fragment encoding the last 110 amino acids, which contains no similarity to any other known form of adenylyl cyclase, was amplified by polymerase chain reaction using the specific oligonucleotide primers 5Ј-ACCTACGGATCCCCAAAGTGCACGGACAAT and 5Ј-CTCGGCGAATTCTCACACACTCTTTGAGAC. The amplified DNA was digested with BamHI and EcoRI (underlined sites) and inserted into the pGEX-2T vector (Pharmacia/LKB). NM522 strain Escherichia coli bearing the pGEX-AC9CT plasmid were grown to an A 600 of 0.5, and expression of the fusion protein was induced by the addition of 500 M isopropyl-1-thio-␤-D-galactopyranoside and growth for 2 h at 30°C. The fusion protein was purified from cell extracts using glutathione-agarose and eluted using 5 mM glutathione. The purified GST-AC9CT protein was injected into two rabbits to raise specific antisera as described (29).
Immunoblotting was performed as described previously (29). Mouse tissue membranes were prepared by homogenization for 30 s in lysis buffer in a Polytron and centrifugation at 40,000 ϫ g for 30 min after a low speed spin to remove unlysed debris. Pellets were washed one time in lysis buffer, resuspended in the same buffer with a syringe, and stored frozen at Ϫ80°C. Twenty-five g of tissue membrane proteins or 5 g of Sf9 cell membrane proteins were separated by SDS-polyacrylamide gel electrophoresis on 6% gels, and electrophoretically transferred to nitrocellulose. Blots were incubated with primary antiserum (1:2000) in 3% bovine serum albumin in phosphate-buffered saline buffer for 1 h, washed in TBS containing Tween-20, and incubated with secondary antibody conjugated to alkaline phosphatase (Pierce). Blots were washed and developed using Western Blue substrate (Promega). Alternatively, blots were incubated with secondary antibody conjugated to horseradish peroxidase and developed using ECL (Amersham Corp.). Antigen-blocked antiserum was prepared by incubation of 50 l of primary antiserum with 50 g of GST-AC9CT antigen protein overnight at 4°C before dilution in bovine serum albumin/phosphate-buffered saline buffer. For deglycosylation studies, 50 g of membrane proteins were incubated for 4 h at 37°C in lysis buffer in the presence of 1 unit of peptide:N-glycosidase F (Genzyme) prior to electrophoresis and blotting.
Human Chromosomal Mapping by in Situ Hybridization-The 1.2-kb EcoRI-XhoI fragment of mouse AC9 clone 22.1 was labeled with [ 3 H]dCTP to 1.7 ϫ 10 8 dpm/g. In situ hybridization was carried out using chromosome preparations obtained from phytohemagglutininstimulated human lymphocytes cultured for 72 h. 5-bromodeoxyuridine (60 g/ml) was added for the final 7 h of culture to improve chromo-somal banding. Radiolabeled probe (25 ng/ml) was hybridized to metaphase spreads as described previously (35). Slides were washed and coated with nuclear track emulsion (Kodak NTB2) and exposed for 18 days at 4°C. To avoid slipping of silver grains during banding, developed chromosome spreads were stained in buffered Giemsa and photographed. R-banding was then performed by the fluorochrome-photolysis-Giemsa method and rephotographed for analysis.

RESULTS
Polymerase chain reaction using degenerate oligonucleotide primers based on conserved sequences of the second intracellular domain of previously characterized adenylyl cyclases was used to amplify adenylyl cyclase-like sequences from mouse brain cDNA. A 288-bp fragment was found to encode a previously unknown sequence with high similarity to adenylyl cyclases (21).
Using this sequence as a probe, several overlapping cDNA clones totaling 4.6 kb were obtained from commercial and specifically primed mouse brain cDNA libraries and from 5Ј-RACE amplifications (Fig. 1A). The DNA sequence obtained from sequencing both strands of several overlapping clones was used to determine the deduced amino acid sequence shown in Fig.  1B. The cDNA sequence terminates in a poly(A) tail and contains a 4062-bp open reading frame defined by a consensus initiator methionine codon (36) preceded by an in-frame stop codon. The open reading frame encodes a 1353-amino acid protein, which we refer to as adenylyl cyclase type 9, 2 with a calculated mass of 151 kDa and pI of 6.9. Computer hydropathy predictions indicate the potential for 12 transmembrane spans in two groups of six, as do previously characterized adenylyl cyclases. Two regions, after the first and the second set of membrane spans (residues 297-772 and 997-1353), have high similarity to the conserved C1 and C2 intracellular domains of other adenylyl cyclases. These domains have been suggested to contain the intracellular cyclase catalytic site (1, 6). The original PCR probe sequence is found in the predicted location and proper reading frame in the C2 conserved domain. The sequences of the membrane-spanning regions are quite divergent compared with other adenylyl cyclases. Two consensus Asnlinked glycosylation sites are found between predicted membrane spans 11 and 12 at residues 955 and 964. A pair of predicted protein kinase A sites are located at serines 373 and 374. Many potential protein kinase C sites are found throughout the sequence.
The 1353-amino acid size of AC9 is larger than the other eight mammalian adenylyl cyclases, with much of the extra size found in a relatively long amino-terminal region (117 amino acids), the longest nonconserved (C1b) region of the C1 loop (residues 566 -772, 207 amino acids), and the longest carboxyl-terminal nonconserved (C2b) region (residues 1244 -1353, 110 amino acids) among mammalian adenylyl cyclases. A search of the GenBank TM data base identified an AC9-like sequence from Xenopus (accession number Z46958), which is most similar to mouse AC9. 5 Comparison of the deduced amino acid sequence of AC9 with known ACs indicates that AC9 appears essentially equidistant from other AC types (from 52.8 to 57.1% similar) and more divergent than any other known mammalian adenylyl cyclase. Analysis of conserved amino acid residues among the mammalian adenylyl cyclases highlights this divergence, as AC9 has unique amino acid residues at 42 positions where the other eight adenylyl cyclases are identical. 128 residues remain identical among all nine adenylyl cyclases. By comparison, type 3 enzyme, which appears most divergent among the first eight adenylyl cyclases, has 23 differences from the overall consensus of the other seven sequences. Three such variant residues in AC9 in the otherwise conserved motif 5 J. Olate, unpublished results. WQYDVW (FKFDVW in AC9) are probably responsible for the failure to identify this form of the enzyme using first conserved domain PCR primers (15).
The tissue distribution of the mRNA encoding the type 9 adenylyl cyclase was determined by Northern blotting using mouse tissue poly(A) RNAs and both amino-terminal (data not shown) and carboxyl-terminal probes (Fig. 2). An 8.5-kb band was found in all tissues examined, with highest levels in skeletal muscle and intermediate amounts in brain, heart, lung, liver, and kidney. These are very high levels of mRNA for an adenylyl cyclase, as a 4-h exposure of a blot containing 2 g of poly(A) RNA gives a readily apparent signal (Fig. 2). Low levels of this 8.5-kb mRNA were also evident in testes and spleen in a 20-h exposure (data not shown). The AC9 mRNA is also abundant in mouse white and brown adipose tissue (37). The 8.5-kb size of this mRNA is nearly 4 kb longer than the size of the overlapping clones that were obtained. The clones with the longest 3Ј-untranslated region all terminate in a common poly(A) tail almost 400 bases downstream from the stop codon. The difference in size between the native mRNA and the cDNA clones obtained is most probably due to a long 5Ј-untranslated region, such as that found in AC8 (20).
The distribution of mRNA encoding the type 9 adenylyl cyclase in brain was examined by in situ mRNA hybridization. The mouse AC9 probe detected an 8.5-kb band in both mouse and rat (data not shown) brain poly(A) RNA. Using the mouse antisense cRNA probe, a discrete distribution of AC9 mRNA was observed in coronal sections of rat brain (Fig. 3). Highest levels of labeling were observed in hippocampus, particularly in the pyramidal cell layer of Ammon's horn and the polymorph layer of dentate gyrus. The cerebellar granular cell layer was also highly labeled. In neocortex, labeling was prominent in four clear laminae, with high labeling of layers II and III and layers V and VI but little labeling of layers I and IV. Piriform cortex was also highly labeled. A moderate signal was seen in striatum, brainstem, and amygdala. Labeling was weak or absent in other brain regions, such as thalamus and hypothalamus. The sense cRNA probe did not label any region of rat brain (data not shown).
Expression of the AC9 protein was induced by infection of Sf9 cells with recombinant AC9 baculovirus. Immunoblot analysis of Sf9 cell membranes using antiserum raised against a glutathione S-transferase fusion protein containing the last 110 amino acids of AC9 revealed the appearance of a 140-kDa protein in AC9-infected cells, which is not present in uninfected cells (Fig. 4). The specificity of this antiserum was demonstrated by blotting with preimmune serum, or antiserum pretreated by the addition of GST-fusion protein antigen, each of which fails to detect the 140-kDa protein. The size of this immunoreactive band expressed in Sf9 cells was unaffected by peptide:N-glycosidase F deglycosylation (data not shown).
To determine the tissue distribution of the AC9 protein, mouse tissue membrane proteins were electrophoresed and probed with AC9 antiserum. In Fig. 5, the antiserum readily recognized proteins of 140 -160 kDa in membranes from brain, spleen, lung, liver, testis, and pancreas, indicating a high expression of the AC9 protein in these tissues. Lower amounts of immunoreactivity are also apparent in membranes from heart, skeletal muscle, and kidney. Immunoreactivity was blocked by pretreatment of the antisera with GST fusion protein antigen. Peptide:N-glycosidase F deglycosylation of tissue membrane proteins prior to immunoblotting reduced the AC9 immunoreactive bands to 140 kDa in all tissues, indicating that the size differences apparent among tissues are due to differential glycosylation of the same size protein backbone. This size is in agreement with the predicted mass deduced from the nucleotide sequence. However, the pattern of AC9 immunoreactivity among various tissues appears quite different from the pattern of AC9 mRNA expression, such that testis with low mRNA levels has high immunoreactivity, while skeletal muscle has high mRNA but low immunoreactivity.
To understand the regulation of the AC9 enzyme, adenylyl cyclase activity was determined in membranes prepared from Sf9 cells infected with AC9 baculovirus in the presence of various adenylyl cyclase activators and inhibitors. The effects of G protein subunits were determined in Fig. 6A. Compared with membranes prepared from uninfected Sf9 cells, AC9-expressing membranes contain 4-fold increased basal (5 mM Mg 2ϩ -stimulated) adenylyl cyclase activity. Mutationally activated recombinant Q227L G␣ s * at 5 nM increased adenylyl cyclase activity of AC9 by 4-fold over the AC9 basal activity after subtracting native G␣ s *-stimulated activity. The ␤␥-subunits of G proteins purified from bovine brain, at 100 nM, had a only slight depressing effect on AC9 activity (as well as native Sf9 AC activity), in the absence or presence of 5 nM G␣ s *. Forskolin (10 or 100 M) failed to elicit increased activity from AC9 expressed in Sf9 cells (Table I) 1, 3, and  5) and AC9 baculovirus-infected cell membrane protein (lanes 2, 4, and 6) were separated by SDS-polyacrylamide gel electrophoresis on a 6% gel. The transferred proteins were probed with preimmune serum (lanes 1 and 2), immune serum (lanes 3 and 4), and antigen-blocked immune serum (lanes 5 and 6). The arrow indicates the 140-kDa immunoreactive product.
FIG. 5. Tissue distribution of mouse AC9 protein. 25 g of the indicated mouse tissue membrane proteins, either untreated or following treatment with peptide:N-glycanase F (PNGase F) to remove Asnlinked carbohydrates, were separated by SDS-polyacrylamide gel electrophoresis on a 6% gel. The transferred proteins were probed with AC9 immune serum (immune) or with AC9 antiserum pretreated with GST-AC9CT antigen protein (antigen-blocked). The arrows indicate the specific immunoreactive products.
CMT cells, AC9 was stimulated 2-fold by 5 M or 50 M forskolin, indicating that AC9 is activated by forskolin poorly compared with both native Sf9 and CMT cell adenylyl cyclase(s) or overexpressed AC8 (Table I). In Fig. 6B, the role of Ca 2ϩ on regulation of AC9 was assessed. The addition of 1 mM EGTA, 1 M Ca 2ϩ , or 1 M calmodulin plus 1 M Ca 2ϩ had no apparent effect on the activity of AC9 in infected Sf9 cell membranes. In contrast, Sf9 expressed AC8 was dramatically stimulated by Ca 2ϩ /calmodulin, but was returned to basal activity by the further addition of EGTA. In Fig. 6C, the ability of a hormone receptor to activate AC9 activity was determined. PGE 2 highly stimulated AC9 activity in membranes prepared from CMT cells transfected with the AC9 cDNA. Thus, AC9 is an active adenylyl cyclase that is stimulated by hormone receptor, recombinant G␣ s , and forskolin but unaffected by G protein ␤␥-subunits, Ca 2ϩ , or Ca 2ϩ /calmodulin. Southern blotting of human genomic DNA using a probe from the second conserved intracellular domain of AC9 was consistent with the presence of only a single AC9 gene in the human genome (data not shown). The location of the human AC9 gene was determined by chromosomal in situ hybridization (Fig. 7). In the 150 metaphase cells examined, there were 205 silver grains associated with chromosomes. Of these, 63 grains (30.7%) were associated with chromosome 16. Along This location is close to loci for polycystic kidney disease, phosphodiesterase IB, multidrug resistance-associated protein, and the ␣-globin gene cluster (38). DISCUSSION The adenylyl cyclase family in mammals is comprised of at least nine distinct isoforms encoded by distinct genes. These genes are widely scattered throughout the human genome (39). The localization of the AC9 gene to 16p13 identifies chromosome 16 as the only chromosome to contain two known AC genes, as AC7 has been mapped to the other arm of chromosome 16 to 16q12-13 (10,40). The wide dispersal of genes encoding adenylyl cyclases attests to the passage of time since the individual AC subtypes first arose. The maintenance of these diverse ACs as distinct genes since that time indicates the importance of the distinctive character of each adenylyl cyclase form, either in regulatory properties or expression patterns.
The AC9 sequence described here is quite divergent from the other eight known ACs. Previously, adenylyl cyclases had been categorized into five subfamilies based on sequence and functional similarities (7,8). Based on sequence comparisons, AC9 defines a sixth subfamily of mammalian adenylyl cyclases. Functional analysis of AC9 indicates that it is distinct from other known adenylyl cyclase subfamilies as well. AC9 activity is unaffected by G protein ␤␥-subunits in the presence or absence of G␣ s stimulation, unlike AC1 and AC2 subfamily members (12). AC9 appears poorly responsive to the diterpine for-skolin. Forskolin has been shown to stimulate mammalian adenylyl cyclase types 1-8, although the extent of stimulation varies among these enzymes from 2-fold (for AC2) to 50-fold (for AC5 and AC6) (41,42). AC9 appears to be among the least forskolin-sensitive of the known adenylyl cyclases, with no apparent stimulation of the Sf9 cell-expressed enzyme assayed in the absence of GTP. AC9 activity is not stimulated by Ca 2ϩ / calmodulin, unlike AC1, AC3, and AC8 (43), nor is AC9 inhibited by Ca 2ϩ as are AC5 and AC6 (17,43). Thus AC9 appears to have a distinct set of regulatory properties from other known adenylyl cyclases, supporting its assignment as a member of a distinct AC subfamily.
The regulation of AC9 by G␣ s , but not by more specialized mechanisms, supports the contention that AC9 represents a major contributor to the adenylyl cyclase activity in many somatic tissues, which generally lack calcium-or ␤␥-subunitregulated AC activity. Further, the low forskolin responsiveness of AC9 is also consistent with the relatively low degree of forskolin stimulation (2-3-fold) observed with native adenylyl cyclase(s) in membranes from many tissues (44,45). The widely expressed AC6, with a 40 -50-fold stimulation by forskolin (41,42), cannot be a major AC isoform present in tissues with much lower forskolin stimulation. Other regulatory modes, such as by inhibitory G i proteins (3,4) or by protein phosphorylation by protein kinase A (46,47) or protein kinase C (3,48,49), remain to be tested.
The widespread and generally high expression of the AC9 mRNA in somatic tissues also argues for a general role of AC9 activity throughout the body. Immunoblotting also indicates a widespread distribution for the AC9 protein. However, there does not appear to be a good correlation between expression levels of the AC9 mRNA and the AC9 protein among various tissues. The highest protein levels are apparent in brain membranes, which have high mRNA. Skeletal muscle, with the highest mRNA level, has only a fraction of the AC9 protein observed in brain. Testis has low mRNA levels but relatively high protein expression. Since the protein and mRNA distributions do not precisely correspond, it is possible that the stability of the mRNA or protein differs among tissues. Similar discrepancy between adenylyl cyclase subtype mRNA and activity distribution has been described recently in hypothalamus (50). There are no published reports comparing the mRNA and protein levels in various tissues for other "widely expressed" adenylyl cyclases. Within brain, AC9 mRNA is highly abundant in hippocampus, cortex, and cerebellum. This distribution is similar to the mRNA distributions of other ACs generally abundant in brain, Ϫfold simulation was calculated as stimulated activity attributed to overexpressed adenylyl cyclase type 9 (or 8) divided by basal (MgCl 2 ) activity attributable to the overexpressed adenylyl cyclase form. such as AC1, AC2, and AC8 (16,26,43). Highly distinct regulatory properties are found within this set of adenylyl cyclase isoforms, so it will be of interest to know whether individual cells within the brain may express some or all of these AC subtypes. It is not understood how several distinctive ACs may contribute to the overall adenylyl cyclase activity in a complex temporal signaling environment such as brain. In addition, the spatial aspects of how individual AC subtypes may be distributed along neuron plasma membranes, and the functional consequences of this distribution, are also unknown. In this regard, brain adenylyl cyclase protein recently has been shown to be highly enriched in neuronal synapses using a pan-AC antiserum (51).
Assignment of the precise physiological role of AC9, as well as other ACs, will require the development of methods to inhibit the function of individual enzyme subtypes as well as the determination of the relative levels of AC subtype proteins in tissues and in individual cells. Within brain, regional and cellular maps of individual AC protein distributions will be vital to understanding the integrative role of multiple adenylyl cyclases in neuronal function.