Soluble adenylyl cyclase from Spodoptera frugiperda (Sf9) cells. Purification and biochemical characterization.

An insect ovarian cell, Spodoptera frugiperda (Sf9), has been widely used to express recombinant proteins, including adenylyl cyclase, as a host cell in the baculovirus expression system. We report the presence and characterization of a soluble adenylyl cyclase (sAC) distinct from a membrane-bound form of adenylyl cyclase (mAC) that is also present in Sf9 cells. sAC was purified 3,500-fold to near homogeneity; a single band at 25 kDa on SDS-polyacrylamide gel electrophoresis correlated well with adenylyl cyclase catalytic activity. The purified enzyme had a catalytic activity of 0.1 μmol/min·mg and the Km of 0.55 mM for the substrate ATP. In contrast to mAC, sAC was heat-stable. Enzymatic activity of sAC was not stimulated by forskolin and was inhibited by salts at high concentrations. sAC utilized both manganese- and magnesium-ATP as substrate. Di- or triphosphate-containing nucleotides, such as GTP and GDP, as well as pyrophosphate, noncompetitively inhibited sAC. Our data suggest that the physical and biochemical characteristics of sAC are different from those of mAC in Sf9 cells as well as from those of other known forms of adenylyl cyclase in animal cells; sAC in Sf9 cells may constitute a new member of adenylyl cyclase found in animals.

An insect ovarian cell, Spodoptera frugiperda (Sf9), has been widely used to express recombinant proteins, including adenylyl cyclase, as a host cell in the baculovirus expression system. We report the presence and characterization of a soluble adenylyl cyclase (sAC) distinct from a membrane-bound form of adenylyl cyclase (mAC) that is also present in Sf9 cells. sAC was purified 3,500-fold to near homogeneity; a single band at 25 kDa on SDS-polyacrylamide gel electrophoresis correlated well with adenylyl cyclase catalytic activity. The purified enzyme had a catalytic activity of 0.1 mol/min⅐mg and the K m of 0.55 mM for the substrate ATP. In contrast to mAC, sAC was heat-stable. Enzymatic activity of sAC was not stimulated by forskolin and was inhibited by salts at high concentrations. sAC utilized both manganese-and magnesium-ATP as substrate. Di-or triphosphate-containing nucleotides, such as GTP and GDP, as well as pyrophosphate, noncompetitively inhibited sAC. Our data suggest that the physical and biochemical characteristics of sAC are different from those of mAC in Sf9 cells as well as from those of other known forms of adenylyl cyclase in animal cells; sAC in Sf9 cells may constitute a new member of adenylyl cyclase found in animals.
It is well known that cAMP plays a widespread role in the control of gene expression and the integration of hormonal stimulation (1). Adenylyl cyclase is widely expressed in many organisms from bacteria to animals (2,3). Although this enzyme invariably converts ATP to cAMP, its other biochemical and physical properties vary among species. In bacteria, adenylyl cyclase exists within the bacterial body to integrate metabolic function (the enterobacterial form). In some bacteria, such as Bacillus anthracis (4), adenylyl cyclase is secreted as a toxin and is activated by calmodulin in the host cell, facilitating the bacterial invasion of the target animal cell (the calmodulinactivated toxin form). In animals, on the other hand, adenylyl cyclase exists in a membrane-bound form to integrate hormonal stimulation. Typically, this membrane-bound adenylyl cyclase has tandem repeats of a six-transmembrane structure and a cytoplasmic catalytic domain similar to a transporter or ion channel (5). Regulation of this membrane-bound form of adenylyl cyclase by the receptor/G protein and the production of cAMP as a second messenger within the cell are the two common features conserved in both non-vertebrates and vertebrates. A calmodulin-sensitive form of adenylyl cyclase from Drosophila, unlike the bacterial form, shares the same membrane topology and the amino acid sequence in the catalytic domain with a calmodulin-sensitive form of adenylyl cyclase found in mammalian brains (type I). This Drosophila adenylyl cyclase is also capable of coupling to the endogenous mammalian G protein when overexpressed in mammalian cells (6). Thus, the properties of adenylyl cyclase in animals, including its structure and its functional interaction with other signal components, are well conserved between non-vertebrates and vertebrates.
The germ cells of animals, however, contain an additional form of adenylyl cyclase that displays physical and functional properties markedly different from those of the membranebound form (7)(8)(9). This adenylyl cyclase is not stimulated by G protein or forskolin. It is active only in the presence of manganese, but not magnesium, and is unaffected by fluoride or gonadotropic hormones (10 -12). Thus this adenylyl cyclase is regulated differently from the hormonally regulated adenylyl cyclase found in somatic cells; its physical properties are similar to those of adenylyl cyclase found in bacteria, an indication that the adenylyl cyclase in germ cells may be an ancestral form of mammalian adenylyl cyclase. However, attempts to characterize this soluble adenylyl cyclase have been hampered by its limited availability in germ tissues; in the absence of a cultured cell line to study this adenylyl cyclase, sperm has so far provided the only source.
Sf9 cell, an ovarian cell from Spodoptera frugiperda, or fall armyworm, has been widely used as a host cell of baculovirus to express recombinant proteins. We and others have successfully overexpressed the membrane-bound form of mammalian adenylyl cyclase in this cell (13)(14)(15)(16)(17)(18). Sf9 cells, like other animal cells, have an endogenous membrane-bound form of adenylyl cyclase that is stimulated by GTP and forskolin (16). We have noticed that there is an additional form of adenylyl cyclase that is exclusively expressed in the cytosolic fraction of this cell.
In this study, we attempt to answer two questions. First, what are the physical properties of this soluble adenylyl cyclase in Sf9 cells as compared with those of the membrane-bound form of adenylyl cyclase? Second, how are the biochemical characteristics of this adenylyl cyclase different? Our data suggest that the adenylyl cyclase found in Sf9 cells is similar to the bacterial form of adenylyl cyclase, as well as to the form of adenylyl cyclase found in mammalian germ cells; it may constitute a new member of the adenylyl cyclase family found in animals.
Cells were grown in Grace medium containing 4% (v/v) fetal bovine serum, penicillin (100 g/ml), and streptomycin (100 g/ml). For harvesting, cells were washed three times with ice-cold phosphate-buffered saline and stored at Ϫ70°C until use.
Adenylyl Cyclase Assay-We performed adenylyl cyclase assays by incubation at 30°C in a reaction buffer containing 20 mM Hepes (pH 8.0), 5 mM MgCl2, 0.1 mM cAMP, 1 mM creatine phosphate, 8 units/ml creatine phosphokinase, and 0.1 mM [␣-32 P]ATP (about 1 Ci/assay tube), unless otherwise specified. After 20 min, the reaction was terminated by the addition of 2% SDS. We measured the production of cAMP as described previously (19). The protein concentration was determined either by the method described by Bradford (20) or by staining with Amido Black (21) using bovine serum albumin as a standard.
Purification of Soluble Adenylyl Cyclase-All purification steps were performed at 4°C. We thawed frozen cells (about 5 ϫ 10 9 ) in 100 ml of a homogenization buffer containing 20 mM Tris/HCl (pH 7.5), 1 mM EDTA, 1 mM EGTA, 2 mM dithiothreitol, and a protease inhibitor mixture (10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, 50 units of egg white trypsin inhibitor, 20 g/ml L-1-tosylamido-2-phenylethyl chloromethyl ketone, 20 g/ml 1-chloro-3-tosylamido-7-amino-2heptanone, and 2 g/ml aprotinin). The cells were lysed by nitrogen cavitation (800 p.s.i., 30 min). Cavitated cells were centrifuged at 500 ϫ g to remove cell debris, and the supernatant was further centrifuged at 100,000 ϫ g for 1 h. The resulting pellet was resuspended in the homogenization buffer and used as "the membrane fraction." The supernatant was passed through a 0.45-m filter membrane and was used as "the cytosolic fraction." We also purified adenylyl cyclase protein from this fraction.
The cytosolic fraction was loaded onto a DEAE-MemSep-1500 column (Millipore) at 10 ml/min using the ConSep LC100 system (Millipore). The column was washed with 50 ml of buffer A (20 mM Tris/HCl, pH 7.5, 1 mM EDTA, 5 mM ␤-mercaptoethanol, and 0.5 mM phenylmethylsulfonyl fluoride) containing 20 mM NaCl, followed by elution with 15-min gradient from 20 to 200 mM NaCl in buffer A. The adenylyl cyclase activity typically eluted at 100 mM NaCl (Fig. 1A, top). These fractions with high adenylyl cyclase catalytic activity were concentrated to 2-3 ml by use of a Centricon 30 (Amicon).
The concentrated sample was loaded onto a Sephacryl-200 HR column (1 ϫ 80 cm, Pharmacia Biotech Inc.) equilibrated with buffer A containing 150 mM NaCl. Fractions were collected at a flow rate of 0.1 ml/min. Fractions with adenylyl cyclase activity were pooled in a final KPO 4 concentration of 20 mM.
The concentrated samples were applied again onto a Sephacryl-200 HR column under the same conditions described above. The fractions with high adenylyl cyclase catalytic activity were pooled (Fig. 1A, middle) and applied onto an ATP-Sepharose column (1 ϫ 8 cm, Sigma) equilibrated with buffer C (10 mM Tris/HCl, pH 7.5, 0.1 mM EDTA, 2 mM MgCl 2 , 1 mM ␤-mercaptoethanol, and 0.1 mM phenylmethylsulfonyl fluoride). The column was washed with 20 ml of buffer C at a flow rate of 5 ml/min, and eluted with 20-min gradient from 0 to 2 mM ATP in buffer C (Fig. 1A, bottom). Fractions with high catalytic activity were pooled (2.5 ml), and ATP was repeatedly removed at different dilutions and concentrations by use of a Centricon 30. The active fractions were pooled and stocked at Ϫ70°C.

RESULTS AND DISCUSSION
Cytosolic Adenylyl Cyclase in Sf9 Cells-We first noticed the presence of adenylyl cyclase catalytic activity in the cytosol of Sf9 cells when we purified a recombinant membrane-bound form of adenylyl cyclase (14). This catalytic activity ceased during the purification process, indicating that the physical properties of this putative adenylyl cyclase were different from those of the membrane-bound form of adenylyl cyclase.
We further investigated the adenylyl cyclase catalytic activity present in the cytosol of Sf9 cells. We used the cytosolic fraction obtained by centrifugation at 100,000 ϫ g for 1 h, followed by filtration through a 0.45-m filter membrane. The pellet was used as the membrane-bound form of adenylyl cy-clase in Sf9 cells (mAC). 1 Table I summarizes the characterization of adenylyl cyclase from the cytosolic fraction (sAC). sAC had a basal activity 2-fold greater than that of mAC. sAC was insensitive to GTP␥S (ϳ100 M) or forskolin (ϳ100 M) stimulation that in the same conditions increased the catalytic activity of mAC 3-4-fold. When the membrane and cytosolic fractions were mixed and assayed in the presence of GTP␥S (100 M), the enhancement of catalytic activity was similar to that of the membrane fraction alone (data not shown), a finding that suggests membrane-associated GTP-binding proteins do not affect the catalytic activity of sAC. sAC was also insensitive to the purified GTP␥S-Gs␣. The K m for the substrate ATP was different; sAC had a 4-fold greater K m value and thus a lower affinity for ATP. With a relatively high intracellular concentration of ATP (ϳ10 Ϫ3 M), we estimate that more than 80% of all catalytic activity takes place in the cytosolic fraction of Sf9 cells under nonstimulated conditions. It is possible, although unlikely, given our findings, that the cytoplasmic adenylyl cyclase is released from mAC by proteolysis. Several examples of this mechanism are known (22)(23)(24). However, we did not find more of the enzyme in the cytosol when there was a delay in processing the cells even though the delay left more time for proteolysis to take place than when cell disruption and centrifugation occurred without delay. The specific activity of sAC was not altered by the methods of cell disruption (sonication, nitrogen cavitation, or homogenization with a Polytron). When mammalian mAC was overexpressed, the catalytic activity in the cytosolic fraction did not increase over that in nonoverexpressing cells. In addition, High Five cells, another insect germ cell line derived from Trichoplusia ni egg cell homogenates, possessed a similar adenylyl cyclase catalytic activity in the cytosol (data not shown). Our characterization indicated that sAC is a cytoplasmic enzyme distinct from mAC. We therefore decided to purify and further characterize this novel soluble adenylyl cyclase from Sf9 cells.
Purification and Identification of sAC- Fig. 1A shows a purification profile described under "Experimental Procedures" that is summarized in Table II. We obtained good yields in the first three steps with a 30-fold purification, whereas in the last three steps, in which a further 100-fold purification was achieved, we obtained rather poor yields. The overall recovery of the enzymatic activity after ATP-Sepharose affinity chromatography was 2%. We typically obtained 1-2 g of protein from 500 mg of cytosol proteins. Fig. 1B shows a SDS-PAGE analysis of the final product with a Coomassie staining, as well as an example of how intensity of the different protein bands correlates with adenylyl cyclase catalytic activity at each step of the purification process. The fractions from the ATP-Sepharose column with the highest activity showed a single band of molecular mass ϭ 25 kDa. The intensity of this band always and exclusively correlated with adenylyl cyclase catalytic activity; we therefore concluded that this band of molecular mass ϭ 25 kDa is likely to encode sAC. Based upon the intensity of stain-ing, we estimate the purity of the final product to be at least 90%. Thus, a 3500-fold purification was required to obtain enzyme that is Ͼ90% pure. Taking our purification data from the profile, we calculate that adenylyl cyclase represents 0.025% of the total cytosolic protein in Sf9 cells. It is unlikely that we underestimated the purification as a result of inactivation during purification because sAC was, unlike adenylyl cyclase in Escherichia coli (25), very stable. Indeed, even when the crude cytosolic fraction was left at room temperature overnight, a significant portion of adenylyl cyclase catalytic activity was retained. We did not obtain more than 100% recovery in any step, which indicates that serious overestimation is also unlikely.
Molecular Mass of sAC-To measure the molecular mass of sAC in its native condition, partially purified fractions, after undergoing DEAE-MemSep chromatography, were loaded on a Sephacryl-200 column and the elution profile was obtained using protein molecular mass standards. The activity eluted in a single peak corresponding to a molecular mass of approximately 40 kDa (Fig. 2).
The apparent size difference between SDS-PAGE and gel chromatography may be due to the structural nature of sAC, the formation of dimer in native condition, or a protein-matrix interaction in the gel. This size (40 kDa) was unchanged when fractions from different purification steps, such as the initial crude cytosolic fraction and eluate from the hydroxyapatite column, were similarly analyzed by gel filtration, which indicates that the size difference is not due to degradation of sAC during purification. When eluate from the hydroxyapatite column was subjected to a microfiltration membrane, Centricon 30, with the molecular mass cutoff of 30 kDa, the adenylyl cyclase catalytic activity was completely retained, a finding that suggests the size of sAC in native condition is greater than 30 kDa.
The size of sAC in Sf9 cells is apparently different from that of sAC in mammalian testis, which shows a molecular mass ϭ 52 kDa on SDS-PAGE and 50 kDa by gel filtration (26). sAC in Sf9 cells is much smaller than many bacterial adenylyl cyclases (80 -200 kDa) and membrane-bound forms of adenylyl cyclase (100 -120 kDa) (3). Thus, sAC in Sf9 cells is likely the smallest adenylyl cyclase so far identified in animal cells, although we await the results of a future cloning study to determine the exact size. The small size of sAC in Sf9 cells is similar to that found in Rhizobium meliloti (20 kDa) (27).
Effects of Salts-Both NaCl and KCl inhibited the catalytic activity of sAC at high concentrations (Fig. 3); the potency of inhibition was similar for both salts at the IC 50 value of 500 mM. This inhibition was reversible; catalytic activity recovered completely and rapidly after the salts were extracted. Similar salt-mediated inhibition was reported in the adenylyl cyclase of E. coli. (25).
We also examined the hypothesis that salt inhibition was due to disassociation of the putative homodimer (40 kDa) into monomers (25 kDa). Fractions from 5 to 150 kDa were obtained by gel filtration (Sephacryl-200) in the presence of 500 mM NaCl. When diluted to 50 mM NaCl, the 40-kDa fraction recovered adenylyl cyclase catalytic activity, whereas the 25 kDa fraction did not. Thus it is very unlikely that the inhibition of sAC catalytic activity by salts is the result of disassociation.
Effects of ␤-Mercaptoethanol-In contrast to the potent inhibitory effect of salts, no change in the catalytic activity of sAC was not detected in the presence of ␤-mercaptoethanol at 0 -50 mM (Fig. 4). Therefore, disulfide links do not seem important in maintaining its catalytic activity.
Heat Stability-sAC was resistant to heat inactivation (Fig.  5). After 1-h incubation at 37°C, the catalytic activity of sAC decreased by 10%, whereas that of mAC decreased 40%; at 55°C for 1 h, mAC was totally inactivated while sAC was mostly (Ͼ60%) alive.
Effects of Divalent Cations-The presence of manganese was not an absolute requirement to maintain the catalytic activity of sAC in Sf9 cells (Fig. 6). However, the increase of catalytic activity of sAC was manganese and magnesium concentrationdependent and reached its maximum at 1 mM for both cations. The maximal activity was higher by 30% with magnesium than manganese. This property is similar to that of mAC (28), ad-  and membrane fraction (closed bar) of Sf9 cells (5 g) were preincubated at 37 or 55°C for 60 min, followed by adenylyl cyclase assays. The value of adenylyl cyclase activity is shown as a percentage of control that was preincubated at 0°C. When the purified sAC was used, the results were similar. Means Ϯ S.E. from four experiments are shown. enylyl cyclase from E. coli (25), and adenylyl cyclase from yeast (29), in contrast with the sAC found in mammalian testis, which is stimulated by manganese, but not by magnesium (9 -11).
Effects of P-site Inhibitors-Adenosine and its analog, deoxyadenosine, inhibited the catalytic activity of mAC, but not of sAC (Fig. 7). At 100 M adenosine or deoxyadenosine, 40 -50% of catalytic activity was inhibited in mAC while no inhibition was seen in sAC. This finding suggests that sAC, unlike mAC, lacks P-site regulatory sites, although it remains possible that activated sAC is sensitive to P-site inhibition. This is in contrast to mammalian testicular sAC, which is readily inhibited by P-site inhibitors upon stimulation with forskolin (30).
Effects of Other Nucleotides-The inhibition of catalytic activity of sAC by GTP at higher concentrations (Ͼ0.1 mM) was dose-dependent. Kinetics studies revealed that this inhibition by GTP was noncompetitive (Fig. 8) with the K i of 0.98 mM.
This inhibition was not unique to GTP; di-or triphosphate analogs of ATP or its related reagents also inhibited sAC. The assays were performed using partially purified adenylyl cyclase in the absence of the ATP regeneration system. At 5 mM, the order of inhibition potency was pyrophosphate Ͼ Ͼ GTP Ͼ NADP ϭ GDP ϭ ADP (Table III). AMP, cAMP, and adenosine showed no inhibition. Notably, pyrophosphate, but not monophosphate, inhibited sAC. This finding is similar to that obtained from the adenylyl cyclase in E. coli; PP i releasing from the enzyme-product complex is slow and a rate-limiting step (25).
Other Effectors-We also examined the effects of calcium/ calmodulin, protein kinase C, and nitroprusside; none showed an effect on sAC activity (data not shown). Similarly, no guanylyl cyclase-like activity was detected when cGMP formation was measured by the use of the purified enzyme (data not shown).
The above data suggest that, having such a high enzyme catalytic activity and without any apparent stimulators, sAC is regulated, unlike mAC, in an inhibitory manner, rather than in a stimulatory manner. Candidates for the putative physiological regulator of sAC may include nucleotides and salts, as shown in our study. sAC in Sf9 cells is clearly distinct in physical properties from mAC in Sf9 cells, and has more in

TABLE III
Effects of nucleotides on adenylyl cyclase activity Catalytic activities of the purified soluble adenylyl cyclase from the second Sephacryl 200 chromatography (1-5 g) were measured in the presence of various nucleotides (5 mM). The reaction buffer contained 0.2 mM ATP and 5 mM MgCl 2 but not the ATP regeneration system. The value of adenylyl cyclase activity is shown as a percentage of control in the absence of nucleotide. common with sAC found in mammalian testis. The enzyme catalytic activity was, like that of sAC in testis, stable and nonresponsive to forskolin (26). On the other hand, several biochemical properties of sAC in Sf9 cells, such as pyrophosphate inhibition, sensitivity to salts, and probable lack of P-site inhibition, are similar to those of bacterial adenylyl cyclase. These data suggest that sAC in Sf9 cells is a form of adenylyl cyclase that shares some properties with both bacterial adenylyl cyclase and the mammalian sAC found in germ cells. sAC from Sf9 cells may be a novel animal form of adenylyl cyclase; it may also serve as a good experimental model to study how the cAMP signal is regulated nonhormonally in animal cells.