Two Cytoplasmic Domains of Mammalian Adenylyl Cyclase Form a G- and Forskolin-activated Enzyme in Vitro

Mammalian adenylyl cyclases have two homologous cytoplasmic domains (C and C). The first cytoplasmic domain of type I enzyme (IC) and the second cytoplasmic domain of type II enzyme (IIC-Δ3, a construct in which 36 N-terminal amino acids of the C region are deleted) were expressed and purified to homogeneity. Alone, each had no adenylyl cyclase activity; however, mixing of the two domains in vitro resulted in G- and forskolin-activated enzyme activity. The turnover number for G- and forskolin-stimulated enzyme activity of the complex between IC and IIC-Δ3 was 8.2 s. The concentration of IIC-Δ3 to achieve half-maximal activation of IC was 0.8 and 1.3 μM when stimulated by forskolin and G, respectively. The concentration of IIC-Δ3 needed to complex with IC was reduced 10-fold (0.08 μM) when the enzyme was activated by both forskolin and G, suggesting that G and forskolin increased the affinity of the two cytoplasmic domains for each other.

The enzymatic activity of adenylyl cyclase is the key step in regulating the intracellular cAMP concentration upon stimulation of a variety of hormones, neurotransmitters, and other regulatory molecules. There are at least nine distinct mammalian adenylyl cyclases which have a similar structure (Fig. 1A) (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11). This includes two intensely hydrophobic domains (M 1 and M 2 ) and two ϳ40-kDa cytoplasmic domains (C 1 and C 2 ). The C 1 and C 2 domains contain sequences (C 1a and C 2a ) that are similar to each other and to other adenylyl and guanylyl cyclases (12,13). Each isoform of adenylyl cyclase has its own distinct tissue distribution and unique regulatory properties, providing modes for different cells to respond diversely to similar stimuli (12,14).
Membrane-bound adenylyl cyclases are expressed in small quantities, and the enzyme is labile and difficult to manipulate in detergent-containing solutions. To facilitate biochemical and structural analysis, a soluble adenylyl cyclase has been constructed by linking the C 1a and C 2a domains of type I and type II adenylyl cyclases, respectively (15). The resulting protein is sensitive to activation by G s␣ 1 and forskolin and to inhibition by P-site inhibitors, indicating the essential roles of C 1a and C 2a domains for catalysis and regulation. In this paper, we describe the expression and purification of the C 1a and C 2a domains of type I and type II adenylyl cyclase, respectively. Alone, each has no adenylyl cyclase activity; however, mixing of the two domains in vitro results in G s␣ -and forskolin-activated enzyme activity.
Expression of Soluble Adenylyl Cyclase-Escherichia coli cells with a plasmid were grown in Luria's broth containing ampicillin (50 g/ml) at 30°C and IPTG to 0.1 mM was added when the culture reached an A 600 of 0.4. Cells were harvested at suitable times, centrifuged at 6,000 ϫ g at 4°C, and frozen. Frozen cells was thawed in 1/10 culture volume of T 20 ␤ 5 P 0.1 (20 mM Tris HCl, pH 8.0, 5 mM ␤-mercaptoethanol, 0.1 mM phenylmethylsulfonyl fluoride), lysozyme to 0.1 mg/ml was added, the cells were sonicated briefly, and the supernatant after centrifugation (150,000 ϫ g for 30 min, 4°C) was saved. The concentration of proteins was determined using Bradford reagent (Bio-Rad) and bovine serum albumin as standard (16). The proteins were separated by electrophoresis on 11% SDS-PAGE and immunoblot was performed using the ECL system (Amersham). Ascites fluid of hybridoma 12CA5 was raised and collected as described (17).
Purification of Soluble Adenylyl Cyclase-All steps of the purification were performed at 4°C in a cold room. Supernatant of E. coli lysate (from 4 liters, harvested 4 h after IPTG induction) was applied directly to a 20-ml Ni-NTA column (Qiagen) that was equilibrated with T 20 ␤ 5 P 0.1 containing 100 mM NaCl. The Ni-NTA column was washed with 100 ml of T 20 ␤ 5 P 0.1 containing 500 mM NaCl and 100 ml of T 20 ␤ 5 P 0.1 N 100 containing 20 mM imidazole (pH 7.0). The column was then eluted with 100 ml of T 20 ␤ 5 P 0.1 N 100 containing 150 mM imidazole (pH 7.0). The eluate was concentrated by ultrafiltration (Amicon, positive pressure ultrafiltration device, PM 10 membrane) and then diluted 2-fold with T 20 E 1 D 1 (20 mM Tris-HCl (pH 8.0), 1 mM EDTA, and 1 mM * This work was supported by the Cancer Research Foundation and Brain Research Foundation. 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.
Gel Filtration-Purified IC 1 , IIC 2 -⌬3, or mixed IC 1 and IIC 2 -⌬3 (200 l) were applied to a Pharmacia Superdex 200 HR 10/30 gel filtration column; the flow rate was 0.3 ml/min and 0.3-ml fractions were collected. For Fig. 5A, T 20 E 1 D 1 N 500 was used in sample dilution and in equilibrating and running the column. For Fig. 5B (to mimic assay condition), samples were incubated in 200 l of T 20 D 1 with 10 mM MgCl 2 , 1 mM ATP, and 100 M forskolin at 30°C for 2 min; forskolin (100 M, 200 l) was applied to a Pharmacia Superdex 200 HR 10/30 gel filtration column that had been equilibrated with T 20 D 1 N 100 containing 10 mM MgCl 2 and 1 mM ATP, and samples were applied immediately after application of forskolin. Adenylyl cyclase activity was measured in the presence of 100 M forskolin. As a control, a lysate of E. coli containing IC 1 IIC 2 (300 g in T 20 ␤ 5 P 0.1 N 100 ) was applied under the same conditions.

RESULTS AND DISCUSSION
Expression of IC 1 IIC 2 , IC 1 , and IIC 2 -A soluble adenylyl cyclase was constructed by linking the conserved cytoplasmic domains from type I (C 1 ) and type II (C 2 ) adenylyl cyclases (15). The resulting protein, IC 1 IIC 2 , could be activated by G s␣ and forskolin, and the activated enzyme could be inhibited by P-site inhibitors. IC 1 IIC 2 was tagged with hexa-histidine and the HA1 epitope at the N terminus to facilitate the detection and purification (Fig. 1A). Hexo-histidine allowed affinity purification using immobilized metal affinity chromatography (e.g. using Ni-NTA resin), and the HA1 epitope permitted the detection of recombinant protein using monoclonal antibody made by hybridoma, 12CA5 (20,22). High speed supernatant of lysates from both E. coli BL21DE3 and SG22094 cells that expressed IC 1 IIC 2 had increased forskolin-stimulated adenylyl cyclase activities, and the enzyme activities were higher (4 -120-fold) in lysates of cells that were harvested after 8 -19 h of IPTG induction than after only 2-4 h of IPTG induction (Fig.  1B) (23). Using monoclonal antibody 12CA5 and anti-peptide antiserum C2-1077 (detecting the N and C terminus of IC 1 IIC 2 , respectively), the expected 60-kDa protein was detected (Fig.  1B). While the lysates from the later times after IPTG induction had higher forskolin-stimulated enzyme activity, they did not have increased amounts of 60-kDa full-length protein. This suggested that the majority of adenylyl cyclase activity from IC 1 IIC 2 was proteolyzed after 4 -19 h of IPTG induction and that the proteolyzed product was catalytically active. Expression of IC 1 IIC 2 in protease-deficient strains, BL21DE3 (Lon Ϫ ), SG22094 (Lon Ϫ , Clp Ϫ ), or SG21163 (Lon Ϫ , htpR) did not enhance the accumulation of full-length 60-kDa protein (Fig. 1B, data not shown for SG21163) (23,24).
A 30-kDa proteolytic fragment of IC 1 IIC 2 was detected using antiserum C2-1077, suggesting that there is a prominent cleavage at the junction between IC 1 and IIC 2 . To investigate whether the complex of IC 1 and IIC 2 was part of a catalytically active species of the proteolyzed IC 1 IIC 2 , HA1 and hexo-histidine-tagged IC 1 and IIC 2 were expressed separately. Using the monoclonal antibody from 12CA5, the 30-and 31-kDa proteins were detected in high speed supernatants of lysates from E. coli ACII ϭ type II enzyme. Below are shown the constructs used in this work. The first four include, at their amino termini, a hexahistidine and a HA1 epitope tag with a short linker to create the EcoRI and NcoI restriction sites. B, adenylyl cyclase activities and protein expression of the IC 1 IIC 2 construct in protease-deficient E. coli strains, BL21DE3 and SG22094. Samples (10 l) were taken for enzyme assay and immunoblot assay with either 12CA5 or C2-1077 antibodies at the indicated times after IPTG induction. C, adenylyl cyclase activities for a mixture of 10 l of a bacterial lysate (from BL21DE3 cells) containing the components at the top that were obtained after the indicated hours of induction with IPTG and 10 l of an lysate (from BL21-DE3 cells) containing either IIC 2 or IC 1 . The immunoblot shows the amount of the component at the top after the indicated hours of induction by IPTG. p.i. ϭ post-IPTG induction. Adenylyl cyclase activity is shown as nmol⅐min Ϫ1 ⅐mg Ϫ1 . Data are representative of two experiments. that expressed IC 1 or IIC 2 (expected molecular mass as 27 and 31 kDa, respectively), indicating that the IC 1 and IIC 2 protein were stable, soluble proteins. Adenylyl cyclase activities of E. coli lysates that expressed either IC 1 or IIC 2 were not different from those of lysates of E. coli that carried the control vector (ϳ0.01 nmol⅐min Ϫ1 ⅐mg Ϫ1 ). However, mixing of the lysates, each expressing one of these constructs, resulted in high enzyme activity (2-9 nmol⅐min Ϫ1 ⅐mg Ϫ1 , Fig. 1C). The enzyme activity correlated generally with expression (monitored by immunoblot) of IC 1 and IIC 2 from cells (Fig. 1C).
Purification of IC 1 and IIC 2 -IC 1 could be purified by Ni-NTA ( Fig. 2A). The enriched IC 1 could be further purified using a FPLC Mono Q column and Superdex 200 gel filtration. The 30-kDa protein was the adenylyl cyclase (indicated by arrow and verified based on enzyme activity and immunoblot, Fig. 2, B and C). A 29-kDa protein was a major contaminant. The recovery of forskolin-stimulated adenylyl cyclase activity was only 5%, and the yield was 50 g from each liter of E. coli culture.
The same procedure did not succeed in the purification of IIC 2 . The majority of the adenylyl cyclase activity from IIC 2 did not bind to Ni-NTA, probably due to proteolysis or masking of the hexo-histidine tag. When lysates containing IC 1 were mixed with lysates containing IIC 2 and applied to Ni-NTA column, most of adenylyl cyclase did not bind to the column (data not shown). This indicated that the binding between IC 1 and IIC 2 was not strong enough for copurification of IC 1 and IIC 2 .
To purify IIC 2 , we used IIC 2 -⌬3, a construct that deleted 36 N-terminal amino acids of IIC 2 , residues that are not conserved among mammalian adenylyl cyclases. HA1 and hexo-histidinetagged IIC 2 -⌬3 protein was expressed as a soluble protein, based on immunoblot, and formed G s␣ -and forskolin-regulated adenylyl cyclase when mixed with lysate containing IC 1 in vitro (Fig. 1C). IIC 2 -⌬3 could be purified by Ni-NTA and, after subsequent chromatography on FPLC-Mono Q, 95% pure protein (29 kDa) was obtained (Fig. 2B). Its identity was confirmed by immunoblot (Fig. 2C). The recovery of adenylyl cyclase activity was about 35%, and the yields for IIC 2 -⌬3 proteins were 2 mg from each liter of E. coli culture.
Characterization of Purified IC 1 and IIC 2 -⌬3-Purified IC 1 and IIC 2 -⌬3 proteins by themselves had little enzyme activity ( Table I). Mixing of IC 1 and IIC 2 -⌬3 proteins resulted in G s␣and forskolin-stimulated activity (Table I, Fig. 3, A and B). As it did for IC 1 IIC 2 , GTP␥S-G s␣ acted synergistically with forskolin in activating mixed IC 1 and IIC 2 -⌬3, while 2Ј-d-3Ј-AMP inhibited the activity (Fig. 3, C and D). NaCl inhibited the activity of mixed IC 1 and IIC 2 -⌬3 (IC 50 ϭ 300 mM, not shown). The highest turnover number for adenylyl cyclase activity of mixed IC 1 and IIC 2 -⌬3 (when activated by G s␣ and forskolin simultaneously) was 8.2 s Ϫ1 , similar to rates of the purified native and recombinant type I adenylyl cyclase (19,25,26). Thus, the purified soluble adenylyl cyclase has the proper catalytic and regulatory properties and could be used as a model system for the biochemical and structural analysis of mammalian adenylyl cyclase.
Increased concentrations of IIC 2 -⌬3 markedly increased adenylyl cyclase activity when added to a fixed amount of IC 1 (6 nM) (Fig. 4A). The half-saturable concentration (EC 50 ) of IIC 2 -⌬3 for forskolin-and G s␣ -GTP␥S-activated activity was 0.8 and 1.3 M, respectively. When G s␣ and forskolin were used together, EC 50 of IIC 2 -⌬3 fell about 10-fold to 0.08 M. This suggested that the synergistic effects of G s␣ -GTP␥S and forskolin on enzyme activity reflected an increase in the affinity of IC 1 and IIC 2 -⌬3 for each other.
Purified IC 1 or IIC 2 -⌬3 were subjected to gel filtration on Pharmacia Superdex 200 using T 20 E 1 D 1 N 500 as the buffer. A major peak of adenylyl cyclase activity consistent with a globular 30-kDa protein was observed, half of the size for the enzyme activity of lysates containing IC 1 IIC 2 (Fig. 5A). The molecular size did not shift for the mixture of IC 1 and IIC 2 -⌬3, presumably due to the low affinity between two molecules (Fig. 5A).
To investigate whether a complex of IC 1 and IIC 2 -⌬3 could be detected, gel filtration was performed in the presence of forskolin under the conditions for the enzyme assay (forskolin, 1 mM ATP, and 10 mM MgCl 2 and the minimal concentration (100 mM) of NaCl; Fig. 5B). A major peak of adenylyl cyclase activity consistent with a globular 30-kDa protein was observed when the purified IIC 2 -⌬3 was applied alone, half of the size for the enzyme activity of lysates containing IC 1 IIC 2 . When IC 1 alone was applied, a major peak of adenylyl cyclase activity consistent with a globular 55 kDa was observed; the shift from 30-to 55-kDa protein was due to the lower concentration of NaCl. This suggested that IC 1 might exist as a dimer or as a nonglobular protein at lower salt concentrations (100 mM). When the mixture of IC 1 (1 g) and IIC 2 -⌬3 (50 g) was tested, a peak of adenylyl cyclase activity consistent with 45-kDa proteins was observed. The shift in elution profile suggested that IC 1 and IIC 2 -⌬3 did interact. The low apparent molecular mass (45 kDa instead of the expected 60 kDa) could be accounted for by dissociation of the complex of IC 1 and IIC 2 -⌬3 and/or an unusual shape of the complex.
Complementation of IC 1 and IIC 2 -⌬3 by the Halves of Adenylyl Cyclases-Although IC 1 and IIC 2 did form a complex with adenylyl cyclase activity, we failed to detect enzyme activity when two cytoplasmic domains from type I enzyme (IC 1 and IC 2 ) were used (linked or a mixture of IC 1 and IC 2 ) (not shown). To investigate the relative affinity of IC 1 for either IC 2 or IIC 2 , we tested the ability of IC 1 to complement the carboxyl-terminal half of type I or type II enzyme (IM 2 C 2 and IIM 2 C 2 ) (Fig.  4C). As reported previously, the truncation mutants of type I and type II adenylyl cyclases that consisted of either the aminoterminal half (INM 1 C 1 -(1-570) and INM 1 C 1 -(1-484)) or the carboxyl-terminal half (IM 2 C 2 and IIM 2 C 2 ) of the protein had no detectable adenylyl cyclase activity when expressed alone; however, coexpression of the amino-and the carboxyl-terminal  (17 kDa). A, total activity values of the purified IC 1 (5 g), IIC 2 -⌬3 (5 g), mixed IC 1 (0.5 g) and IIC 2 -⌬3 (5 g), and an extract containing IC 1 IIC 2 (300 g) were 80, 183, 47, and 9.9 nmol⅐min Ϫ1 , respectively. B, total activity values of the purified IC 1 (5 g), IIC 2 -⌬3 (5 g), mixed IC 1 (1 g) and IIC 2 -⌬3 (50 g), and an extract containing IC 1 IIC 2 (300 g) were 82.5, 66.2, 115.8, and 6.37 nmol⅐min Ϫ1 , respectively. Data are representative of two experiments. halves resulted in G s␣ -and forskolin-regulated adenylyl cyclase activity (Fig. 1A) (21). IC 1 and IIC 2 -⌬3 each had no adenylyl cyclase activity alone, and Sf9 cell membranes containing the amino-or carboxyl-terminal halves of adenylyl cyclase had enzymatic activities that were similar to the control cell membranes (that containing ␤-galactosidase) (Fig. 4, B and C). When IIC 2 -⌬3 was added to Sf9 cell membranes containing either INM 1 C 1 -(1-570) or INM 1 C 1 -(1-484), there was up to a 40-fold increase in forskolin-stimulated adenylyl cyclase activity (Fig. 4B). In contrast, there was no increase in enzyme activity when IIC 2 -⌬3 was mixed with Sf9 cell membranes containing IM 2 C 2 . When IC 1 was reconstituted with Sf9 cell membranes containing IIM 2 C 2 , there was up to a 20-fold increase in forskolin-stimulated adenylyl cyclase activity (Fig.  4C). However, only a 3-4-fold increase in enzyme activity was observed when IC 1 was reconstituted with Sf9 cell membranes containing IM 2 C 2 . The EC 50 of IC 1 to reconstitute adenylyl cyclase activity of IIM 2 C 2 was at least 10-fold lower than that of IM 2 C 2 .
There has been considerable speculation about the roles of the transmembrane domains of adenylyl cyclases (12). The transmembrane domains target adenylyl cyclase to the plasma membrane for interaction with, and thereby regulation by, G proteins. Our studies indicate that the two cytoplasmic domains of mammalian adenylyl cyclases do not appear to have high affinity for each other. EC 50 for IIC 2 to complex with IC 1 is 0.8 and 1.3 M in forskolin-and G s␣ -stimulated activity, respectively.
Affinity between two natural linked cytoplasmic domains (IC 1 and IC 2 ) is at least 10-fold less than that between IC 1 and IIC 2 . Thus, the transmembrane domain (M 2 ) could link and facilitate the interaction of the two cytoplasmic domains by creating a high local concentration. It remains to be determined whether the transmembrane domains have additional functions, such as altering the interaction between two cytoplasmic domains for regulations or serving as pore structures.