Expression of a Gas/Gai Chimera That Constitutively Activates Cyclic AMP Synthesis*

A chimeric Ga subunit cDNA, referred to as Gasli(38), was constructed containing the complete 5’-untrans-lated region of Gas, the first 356 codons of the rat Gas and the last 36 codons and 428 base pairs of the 3’-untranslated region of the rat Gai cDNA. Transient expression of the Gamli(38) protein in COS cells allowed detection of a chimeric protein which was recognized by antibodies generated against an internal Ga* sequence as well as antibodies recognizing the carboxyl terminus of Gaiz. Chinese hamster ovary cell clones stably expressing the chimeric G-protein a subunit transcript (G0./~(38)) demonstrated 1.5-2.5-fold constitutively elevated cyclic AMP levels and a 3-4-fold increase in the activity ratio of cyclic AMP-dependent protein kinase, although expression of the chimeric polypeptide could not be demonstrated presumably because of low expression of the mutant as. Expression of the rat Ga. transcript yielded clones that were similar to wild-type Chinese hamster ovary cells in regard to cyclic AMP levels and protein kinase activity. In the presence of methyl isobutylxanthine, a cyclic AMP phosphodiesterase inhibitor, cyclic AMP levels in clones expressing the Ga./i(38) transcript were 10-15-fold higher than Ga. expressing clones. Adenylyl cyclase activation by guanosine 5’-0-(thiotriphosphate) (GTPyS)

A chimeric Ga subunit cDNA, referred to as Gasli(38), was constructed containing the complete 5'-untranslated region of Gas, the first 356 codons of the rat Gas and the last 36 codons and 428 base pairs of the 3'untranslated region of the rat Gai cDNA. Transient expression of the Gamli(38) protein in COS cells allowed detection of a chimeric protein which was recognized by antibodies generated against an internal Ga* sequence as well as antibodies recognizing the carboxyl terminus of Gaiz. Chinese hamster ovary cell clones stably expressing the chimeric G-protein a subunit transcript (G0./~(38)) demonstrated 1.5-2.5-fold constitutively elevated cyclic AMP levels and a 3-4-fold increase in the activity ratio of cyclic AMP-dependent protein kinase, although expression of the chimeric polypeptide could not be demonstrated presumably because of low expression of the mutant as. Expression of the rat Ga. transcript yielded clones that were similar to wild-type Chinese hamster ovary cells in regard to cyclic AMP levels and protein kinase activity. In the presence of methyl isobutylxanthine, a cyclic AMP phosphodiesterase inhibitor, cyclic AMP levels in clones expressing the Ga./i(38) transcript were 10-15fold higher than Ga. expressing clones. Adenylyl cyclase activation by guanosine 5'-0-(thiotriphosphate) (GTPyS) in membranes from clones expressing the Ga.,i(38) transcript demonstrated a diminished lag time for maximal activation, indicating an increased relative GDP dissociation rate for the chimeric Ga subunit and an increase in total adenylyl cyclase activity relative to wild-type Gas expressing clones. Cholate extracts from membranes of GaSli(38) expressing clones, when mixed with cyc-549 membranes, reconstituted an increased GTPyS-stimulated adenylyl cyclase activity and a diminished lag time for maximal activation compared to cholate extracts prepared from Gas-expressing clones. The Gas,i(38) construct confers a dominant constitutive activation of adenylyl cyclase when expressed in cells in the presence of a background of wild-type Gas.
* This work was supported by National Institutes of Health Grants GM30324 and DK37871. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertkement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Established Investigator of the American Heart Association.
8 To whom correspondence should be addressed.
GTP-binding regulatory proteins (G-proteins)' are composed of a, ,f3, and y subunits. These heterotrimers couple receptor activation to control of membrane-associated enzymes and ion channels (1-4). Specific receptors, upon agonist binding, catalyze the exchange of GDP for GTP at the guanine nucleotide-binding site in the a-subunit (Ga) of an associated G-protein (5). The GTP-liganded a subunit then activates the appropriate effector enzyme or channel.
To date, a large number of G-proteins have been described including, but not limited to, G,, three Gis, GO, and transducin (GT) (1). These proteins are distinguished most clearly by the differences in amino acid sequences among the Gcu subunits. Although highly conserved, the G-proteins appear to interact with very different receptor and effector molecules. Two lines of evidence indicate that the Ga carboxyl-terminus composes a part of the receptor-binding site or is a domain which is essential for controlling receptor activation of GDP/GTP exchange. First, pertussis toxin ADP-ribosylates a cysteine residue four amino acids from the carboxyl terminus in Gi, Go, and GT and uncouples the G-protein from the appropriate receptor (1, 6). Second, the unc 549 mouse lymphoma cell mutant, whose G, cannot be activated by @-adrenergic or prostaglandin E, receptors, has an arginine to proline mutation six amino acids from the carboxyl terminus of Ga. (7,8). Thus, both ADP-ribosylation by pertussis toxin and an amino acid substitution within the carboxyl-terminal domain of the Gaa subunit uncouple the G-protein from receptor activation.
Based on these findings, we predicted that appropriate genetic manipulation of the Ga carboxyl-terminal sequence might result in activation rather than inactivation of the Gprotein. In other words, appropriate changes in amino acid sequence at the Ga carboxyl terminus could possibly mimic, at least in part, the conformational changes in the G-protein that occur upon interaction with agonist-activated receptor.
We have approached this problem by substituting the Gas carboxyl terminus with sequences predicted to have significantly different secondary structure from Ga. but which are known to be functional at the carboxyl terminus of other Gaproteins. In this report, we describe a Go, chimeric construct in which the last 38-carboxyl-terminal amino acids are replaced by the last 36-carboxyl-terminal residues of a Gai protein. When this chimeric transcript, referred to as Ga,h(38), is expressed in cells, cyclic AMP levels are persist-' The abbreviations used are: G-protein, GTP-binding regulatory protein; G., adenylyl cyclase stimulatory G-protein; Gi, adenylyl cyclase inhibitory G-protein; Go, brain G-protein; GT, retinal G-protein; Go,,i(38), chimeric G-protein between G. and Gi; CHO, Chinese hamster ovary cells; bp, base pairs; GTPrS, guanosine B'-O-(thiotriphosphate); kb, kilobase pairs; SDS, sodium dodecyl sulfate; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; EGTA, [ethylenebis(oxyethylenenitri1o)ltetraacetic acid. 5687 ently elevated, resulting from a dominant, constitutively active GB-like activity.

MATERIALS AND METHODS
Construction of Plasmids-Expression vector pCW1-neo is a pUC13 derivative containing the SV40 enhancer, replication origin, and early promoter (EcoRI-Hind111 fragment) and the SV40 splicing and polyadenylation sequences (HindIII-BamHI fragment) from pKO-neo (gift from D. Hanahan, Cold Spring Harbor Laboratory). In addition, the plasmid carries the Tn5 gene for neomycin resistance under the control of MoLTR from Homer 6 (9). Insertion into the unique HindIII cloning site allows the expression of cloned cDNA under the control of the SV40 early promoter. In the expression plasmid pGa., the 1.56-kb HindIII-Hind111 fragment spanning the 5"untranslated region, the entire coding region, and 192 bp of the 3"untranslated region of the rat Ga. cDNA (10) was inserted into pCW1-neo. The orientation of the insert was determined by restriction analysis. In constructing pGa./i(38), the rat Ga. cDNA was truncated with Aha11 and blunt-ended with the Klenow fragment of DNA polymerase. The corresponding 3' region of the rat G% was obtained as a 640-bp BglII-Hind111 fragment, and the BglII end was filled to form a blunt end with the Klenow fragment. The chimeric cDNA was obtained by ligating the 1.26-kb HindIII-Aha11 Ga. fragment and the 640-bp BglII-Hind111 Ga, fragment into pUC12 linearized with HindIII. The identity of the chimeric cDNA was verified by restriction analysis and DNA sequencing (11). The 1.9-kb HindIII-HindIII cDNA was then inserted into the HindIII cloning site of pCW1-neo to produce the expression plasmid pGa.li(38).
DNA-mediated Gene Transfer and Cell Culture-CHO-K1 cells were grown in F-12 medium supplemented with 10% fetal calf serum and 100 units/ml penicillin and streptomycin. Transfections were performed using protoplast fusion (12). Twenty-four h after transfection the cells were placed in growth medium plus 500 pg/ml geneticin (G418). G418-resistant clones were isolated using glass cyliners, subcloned at least one time, and propagated in selective medium for further analysis. Clones Gcu./i(38) 2 and 3 were isolated from the same transfection while GaBIi(38) 9 was isolated from an independent fusion. The Ga8 and GaaIi(38) clones are representative of the phenotype characterized for expression of these constructs in 4 Ga. and 12 Ga,,i(38) independent clones as determined by Southern analysis (13).
COS cell transfections were performed using DEAE-dextran as described by Ausubel et al. (14). Clones were screened for protein expression 65-80 h after the transfection.
Northern Blot Analysis-Appropriate cell clones harvested by trypsinization and pelleted by centrifugation were resuspended in 1 ml of 300 mM sodium acetate, pH 5.5, 10 mM EDTA, 1.0 mg/ml heparin, and 1% SDS and briefly vortexed. An equal volume of buffered phenol was added and the tube shaken vigorously. The sample was heated to 55 'C for 5 min, set on ice for a further 5 min and then centrifuged at 3000 X g for 5 min. The aqueous phase was recovered and reextracted two times with an equal volume of phenol-chloroform. The RNA was precipitated by the addition of 2.5 volumes of ethanol. Total RNA was then analyzed by Northern analysis (13).
Immunoblot Analysis-Clones were collected by scraping and homogenized using 20 strokes of a Dounce homogenizer. Nuclei were removed by centrifugation at 500 X g for 5 min and the supernatant collected and centrifuged at 49,000 X g for 30 min. The membrane pellet was resuspended and solubilized in SDS and proteins were resolved by SDS-polyacrylamide gel electrophoresis on 12% acrylamide gels. Proteins were transferred overnight to nitrocellulose using 150 mA constant current. Filters were blocked with yglobulin-free horse serum and then blotted with rabbit antiserum raised against the carboxyl-terminal peptide sequence of transducin (15), Gas (15), or the DATPEPGEDPRVTRAK sequence within Ga, (gift from H. Bourne, University of California, San Francisco). Blots were washed and then incubated with '251-protein A, washed five times and then autoradiographed as detailed previously (16).
Assay of Cyclic AMP-dependent Protein Kinase in Permeabilized Cells-Wild-type and transfected cells were seeded (1-5 X lo3 cells/ well) into 96-well dishes and maintained for 2-3 days in growth medium. Growth medium was aspirated, and the attached cells were rinsed with a buffered salt solution consisting of 137 mM NaCl, 5.4 mM KC1, 0.3 mM sodium phosphate, 0.4 mM potassium phosphate, 1 mg/ml glucose, and 20 mM HEPES, pH 7.2, at 30 "C. To permeabilize the cells and initiate protein kinase assays, the buffer solution was aspirated and replaced with 40 pl of the buffered salt solution supple-mented with 50-100 pg/ml digitonin, 10 mM MgC12, 25 mM P-glycerophosphate, 100 p~ [y3'P]ATP (5000 cpm/pmol), and 100 pM kemptide (LRRASLG). The kinase assays were allowed to proceed for 10 min at 30 "C and then terminated by addition of 10 p1 of 25% trichloroacetic acid. Phosphorylated kemptide was quantified by spotting aliquots (40 pl) of the acidified kinase reaction mixtures on 2 X 2-cm phosphocellulose strips (Whatman, P81) and washing batchwise in 75 mM H3P04 as described by Roskoski (17). Total cyclic AMPdependent protein kinase activity was defined as the difference between reactions containing 1 pm of CAMP to maximally activate cyclic AMP-dependent protein kinase and reactions containing 25 pg/ml IP-20 to completely inhibit cyclic AMP-dependent protein kinase. The synthetic peptide IP-20 corresponds to residues 5-24 of the heat stable inhibitor of cyclic AMP-dependent protein kinase (18). Basal kinase activity was defined as the kemptide kinase activity inhibited by 25 pg/ml IP-20. IP-20-insensitive kemptide phosphorylation was 10-15% of the activity measured in the presence of 1 p M CAMP. Adenylyl Cyclase Activity Measurements-CHO clones were grown to 75% confluence, harvested, and resuspended in lysis buffer (20 mM Tris-HC1, pH 8.0, 2 mM MgClZ, 1 mM EDTA, 1 mM P-mercaptoethanol, and 0.02 units/ml aprotinin). Cells were ruptured by nitrogen cavitation (700 psi, 20 min on ice). Nuclei were removed by centrifugation at 1,500 X g and 4 "C for 5 min, and membranes were then pelleted by centrifugation at 31,000 X g for 1 h, resuspended in lysis buffer and used for adenylyl cyclase assays. Aliquots of membranes (25-50 pg) were added to 80 p1 of reaction mix that contained 50 mM Na-HEPES, pH 8.0, 5 mM MgCl,, 0.2 mM EGTA, 1 mM @-mercaptoethanol, 0.1 mg/ml bovine serum albumin, 10 mM creatine phosphate, 10 units/ml creatine phosphokinase, 0.4 mM [CC-~'P]ATP (20 cpm/pmol), and either 100 p~ GTP, 100 p M GTP-yS, 1 mM sodium fluoride, or H20. Reactions were incubated for various times at 30 "C, stopped by addition of 1 ml of 1% SDS and 32P-cyclic AMP purified by chromatography on Dowex-50 and Alumina.
Cellular Cyclic AMP Determinutiom-Levels of intracellular cyclic AMP were determined with the cyclic AMP [lZ5I] Assay System from Amersham Corp. using the manufacturers protocol. Cells were incubated for 10 min in fresh Dulbecco's minimum essential medium, 10 mM Na-HEPES, pH 7.5, in the presence or absence of 1 mM isobutylmethylxanthine. The medium was then rapidly aspirated, the cells rinsed with phosphate-buffered saline, and then quick frozen by placing the dish in a bath of liquid nitrogen. Cyclic AMP was extracted by scraping the cells in 65% ethanol, vortexing, then centrifuging to remove protein. The ethanol supernatants were dried and used for cyclic AMP determinations and the pellets solubilized in NaOH for protein analysis.

RESULTS AND DISCUSSION
Expression vector pCW1-neo and expression plasmids pGaB and pGa,/i(38) that were used to transfect COS and wild-type CHO cells are illustrated in Fig. 1. pGa. contains the complete 5"untranslated region, the entire coding region, and 192 bp of the 3"untranslated region of the cDNA coding for the 52-kDa rat Gas subunit (10). pGaSli(38) also contains the 5'untranslated sequence but only the first 356 codons of the rat Gas cDNA. The remaining 3' end of pGa,/i(38) is constructed from the last 36 codons and 428 bp of the 3"untranslated region of the rat Gai cDNA (10).
The aligned carboxyl-terminal amino acid sequences for Gas, Gai, and GaSli(38) are shown in Fig. 2. For Ga, and Ga./i(38) the first 356 residues are identical. The construction of the Gc~./~(38) chimera resulted in a two amino acid (His-Tyr) deletion from Gas at the switch site. This is followed by the last 36 amino acids of Gai. Of these 36 amino acids, there are 17 nonconserved residues between Gai and Ga,. Within this same region, Gai and Gas have, respectively, 8 and 13 charged residues with a net charge of +2 for Gas and -1 for Gai. Charge distributions of the predicted amphipathic helices in this region of the Ga polypeptides are also significantly different (19). Finally, Gai and GaeIi(38) terminate with a cysteine followed by three uncharged residues which is the pertussis toxin-catalyzed ADP-ribosylation site in Gi, GO, and GT (1, 6) but is absent in Ga, (1, 6). The expression vector pCW1-neo is a pUC13 derivative containing the Tn5 gene (neomycin resistance), the SV40 enhancer, replication origin and early promoter (SV40EP), and the SV40 splice and polyadenylation sequences (SV4OPA). Insertion of cDNA into the unique HindIII-cloning site permits expression under the control of the SV40 early promoter. pGa. was obtained by inserting the 1.56-kb HindIII-Hind111 fragment containing the rat Ga. cDNA into the HindIII site of pCW1-neo. pGa.li (38)  with the carboxyl-terminal 36-amino acid residues of the chimeric Ga, Gn./i(38), and Gni. The amino acids in the boxes represent conserved residues and conservative changes between Ga., Gai,and Ga.li(38). The deletion of 2 amino acid residues in Ga.li(38) and Gai is represented by --. Panel B shows the structure of the chimeric Ga.li(38), with nucleotide sequence and amino acid residues at the switch site. The chimera contains the first 356-amino acid residues of Ga. and the last 36-amino acid residues of Gai as indicated.
In transient COS cell transfections, the expression of both Go, and Gqi(38) proteins could be detected by Western blotting. Fig. 3 shows that the expression of the 52-kDa polypeptide of Gas encoded by the cDNAs is enhanced by transfection of COS cells with either pGa. or pGa./i(38).
Densitometry of the 42-and 52-kDa bands indicated an increase in the 52-kDa band relative to the 42-kDa band of approximately %fold for Ga. and 2-fold for Ga8/i(38)-transfected COS cells relative to control or Gwp-transfected cells. Panel B of Fig. 3 shows that an antiserum which recognizes the carboxyl terminus of Gaia (14) also recognized the  and 7); pGa.li(38) (lanes 4 and 8).
GaSli(38) 52-kDa polypeptide, in addition to the 41-kDa Gai peptide, but not Gas. For reasons that are unclear, the antibody recognizing the Gai carboxyl terminus does not react particularly well with the GaSli(38) polypeptide. This may reflect differences in the conformation of the 52-kDa chimeric polypeptide relative to the 41-kDa Gai polypeptide. Neverthe- The endogenous Gai mRNA in CHO cells is detected as two distinct messages of size 1.9 and 2.4 kb (panel C ) .

TABLE I Cyclic AMP and cyclic AMP-dependent protein kinuse activity in
CHO clones expressing Ga.li(38) Stable G418-resistant CHO clones expressing the pGab or pGa./i(38) cDNA constructs were assayed for basal cyclic AMP levels and cyclic AMP-dependent protein kinase activity as described under "Materials and Methods." Phenotype of each clone has remained stable for greater than 50 cell doublings. A cyclic AMP-dependent protein kinase activity ratio greater than 0.3 does not linearly correlate with cyclic AMP levels. This results in the same apparent kinase activity for Ga.li(38) clones 2, 3, and 9 even though the basal cyclic AMP levels are 1.5-2-fold higher in clone 9 relative to clones 2 and 3. Results represent two independent cyclic AMP measurements, and protein kinase activity is representative of two to three experiments for each clone. less, the immunoblots with the Ga, and Gai antibodies confirm the chimeric nature of the Ga./i(38) polypeptide.

Cyclic AMP-dependent protein kinase
To characterize the phenotype resulting from the expression of the chimeric GaSli(38) construct, stable transfectants were isolated in CHO cells. The expected size of the plasmidexpressed Ga. and GaSli(38) mRNAs are 2.5 and 3.0 kb, respectively, whereas the endogenous Ga, mRNA is 1.8 kb. Therefore, it was possible to screen G418-resistant clones by Northern analysis and detect the expression of the plasmid expressed Ga transcripts. Fig. 4 shows the Northern analysis of CHO cell clones isolated after transfection with pGa., pGa./i(38), or a pCW1-neo control construct and probed using Gas cDNA. While wild-type CHO cells transfected with 1; . ,,,.,,,$ , , 1 , .~ ,,,,,,,, _ _ _ , _ _ _ , pCW1-neo express only endogenous Ga, mRNA, clones transfected with pGas or pGa.li(38) express the predicted plasmid encoded messages. Fig. 4 also shows the same RNA preparations probed with the BglII-Hind111 fragment of a Gw cDNA (lo), encoding the last 108 bases of the coding region and the entire 3"untranslated region. The Ga,/i(38) transcript was detected in addition to the endogenous Gai transcripts, verifying the chimeric nature of the 3.0-kb mRNA. Each clone analyzed varied one from another, within a %fold range, in the level of expression of Ga, or Ga./i(38) mRNA. Southern analysis of genomic DNA verified the independence of each Ga, and Ga,/i(38) 2, 3, and 9 clones (not shown). In G418-resistant clones expressing the Ga8/i(38) transcript, basal intracellular cyclic AMP is elevated, and cyclic AMPdependent protein kinase is significantly activated relative to other CHO clones (Table I). The Ga,/i(38) clones vary in basal cyclic AMP levels but are all 1.5-2.5-fold higher than wildtype CHO, Ga. clones, and pCW1-neo clones. Similarly, the CAMP-dependent protein kinase activity ratio is elevated 3-4-fold over Gas, wild-type, and pCW1-neo controls. In the Ga.li(38) clones, elevated cyclic AMP and kinase activity has been a stable phenotype for more than 150 cell doublings. No Gas, Gain, or pCW1-neo clone has demonstrated this phenotype, indicating that the constitutively elevated cyclic AMP levels in Ga./i(38) clones is a consequence of expression of this transcript. In addition, we have now transfected other cell lines with the Ga,/i(38) construct and isolated stable G418-resistant clones with elevated cyclic AMP levels and protein kinase activity.' In the presence of the phosphodiesterase inhibitor, methyl isobutylxanthine, cyclic AMP levels in clone Ga,/i(38)9 are elevated 10-15-fold higher than those observed in control clones (Fig. 5). The greater steady-state cyclic AMP level achieved in the presence of methyl isobutylxanthine is indicative of an enhanced synthesis of cyclic AMP in the Ga./i(38)9 clone relative to wild type. The concentration curve for methyl isobutylxanthine is the same for the Ga.li(38) expressing clone and wild-type cells, indicating the phospho-* S. Soparkar and G. L. Johnson, manuscript in preparation.

Cyclase activity was measured as described under "Materials and Methods"
in the presence of 100 p~ GTP (O), 100 pM GTP-yS (*), or 10 mM sodium fluo-

ride (A) for membranes isolated from
Ga.li (38) clones 9 ( A ) , 2 ( B ) , 3 (C Mlnutes diesterase is not dramatically altered in its properties in Ga8li(38)9 relative to wild type. Cellular export of cyclic AMP was similar in wild-type and Ga.li(38) expressing clones (not shown), indicating that changes in cyclic AMP extrusion was not responsible for the dramatic effect of methyl isobutylxanthine in the chimera-expressing clones. The most likely explanation is a compensatory increase in cyclic AMP phosphodiesterase Ga.li(38) clones. Similar compensatory changes in phosphodiesterase levels have been described in other cell types (20). Nonetheless, even in the absence of phosphodiesterase inhibitors, the Ga.li(38) construct is capable of constitutively elevating cyclic AMP levels and activating cyclic AMP-dependent protein kinase in the presence of a background of wild-type Gas.
Activation of adenylyl cyclase by guanine nucleotides in membranes isolated from pGaB/i(38)-transfected clones differs in two significant ways relative to preparations from wildtype, pGa., and pCW1-neo clones (Fig. 6). First, membranes from three independent Ga.li(38) clones demonstrate a significant decrease in the time required to achieve maximal adenylyl cyclase activation by GTPyS. This diminished lag time is somewhat variable but reproducible among Ga,/i(38) clones 2, 3, and 9 relative to Gas wild-type or pCW1-neo clones. In general, Ga./i(38)9 has the highest level of chimeric mRNA and shortest lag. In contrast, Ga,/i(38)3 has the lowest chimeric mRNA level of the three clones, relative to the endogenous Gas mRNA, and has less of a decreased lag than either clones Ga./i(38)9 or 2 compared with the two control clones, Ga, or pCW1-neo.
The second change in guanine nucleotide-activated adenylyl cyclase observed with expression of the Ga.li(38) chimera was that GTPyS and fluoride-activated adenylyl cyclase activities were reproducibly greater in membranes from the Ga.li(38) clones compared with Gas or pCW1-neo clones. This observation has been found with several membrane preparations where caution was taken to prepare membranes from different clones on the same day and when cells were seeded and harvested at similar densities. This finding is consistent with the Ga8,i(38) product increasing adenylyl cyclase activity and thus, elevating cyclic AMP levels relative to wild-type or Ga. clones. It is interesting to note that clone Ga./i(38)9, which has the shortest lag in GTPyS activation of adenylyl cyclase and the highest basal cyclic AMP levels, also has the highest activity in the presence of GTP, consistent with a more active adenylyl cyclase system as a result of expressing the Ga.li(38) transcript.
In an attempt to normalize for differences in clonal expression of Ga./i(38), the ratio of GTPySIGTP-stimulated adenylyl cyclase activity was plotted as shown in panel F, Fig. 6.
The GTPySIGTP ratio indicates the relative adenylyl cyclase activation intrinsic for the two guanine nucleotides. This comparison between clones emphasizes the more rapid activation of adenylyl cyclase in response to GTPyS with GaS/i(38)-expressing CHO clones relative to Ga. or pCW1neo clones. The simplest interpretation of the decreased lag is that, relative to wild-type Ga., the Ga.li(38) construct has a more rapid GDP dissociation rate allowing faster GTPyS binding and activation. However, the greater apparent intrinsic adenylyl cyclase activity in clones expressing Ga.li(38) cannot be readily explained simply by enhanced GDP dissociation.
As a direct measure of G-protein activity, cholate extracts from membranes of Ga. and GaS/i(38)-expressing clones were mixed with cyc-S49 membranes, which lack Ga. (21), and assayed for GTP and GTPyS-stimulated adenylyl cyclase activity (22). Fig.  7 shows that cholate extracts from Ga8/i(38)9 membranes reconstituted a GTP+-activated adenylyl cyclase activity with both a greater maximal activity and decreased lag time required to reach maximal adenylyl cyclase activation compared with a Gwexpressing clone. Similar findings were observed with other Gaeli(38)-expressing clones when activities were compared in the reconstitution assay with cyc-membranes relative to either Gas-expressing clones or wild-type CHO cells. Fig. 8 shows that the enhanced adenylyl cyclase activity observed with extracts from Ga,li(38)9 is linear with increas- Membranes (8-10 mg/ml) from CHO clones Ga, and Ga.h(38)9 were solubilized on ice in 1% sodium cholate and varying amounts (0.07, 0.72, 1.8, and 7.2 pg of protein) of the 100,000 X g supernatants were mixed with 36 pg of cyc-S49 membranes and assayed for adenylyl cyclase activity in the presence of 100 p~ GTP-yS as described under "Materials and Methods." Reconstitution with heat-inactivated cycmembranes gave activities similar to background indicating that catalytic adenylyl cyclase was not contributed by the cholate extract in the assay. Values are means of duplicate determinations which varied by less than 10% and representative of two independent membrane preparations and experiments.
ing protein concentration, indicating a direct relationship between G-protein addition in the reconstitution assay and the enhanced activity observed in the GaSli(38)9 clone relative to clone Ga,. Furthermore, Fig. 9 shows that at equal relative activities in the reconstitution assay the lag time to reach maximal GTPyS-activated adenylyl cyclase activity is about 3.2 min with cholate extracts from Ga.li(38)9 and 6.5 min with the extract from the Gas clone. The inset of Fig. 9 also shows that the adenylyl cyclase activity measured with GTP in the reconstitution assay is greater with extracts from GaSli(38)9 relative to the Gas clone. Thus, similar kinetic properties of adenylyl cyclase activation are observed with either intact membranes from clones expressing plasmidderived Ga.li(38) or Ga. transcripts and with cholate extracts from these membranes using the reconstitution assay with S49 cyc-membranes. The reconstitution assay utilizing cyc-membranes is a direct measure of G-protein activity in the cholate extracts of donor membranes. The phenotype observed with cells expressing the Ga.li(38) transcript is easily observed in the Gprotein regulation of adenylyl cyclase. Plasmid expression of a normal Ga. transcript had no effect on the phenotype of CHO cells. Thus, the utility of the Ga.li(38) construct is that it is dominant among a background of wild-type Ga, in  were reconstituted with 36 pg of cyc-S49 membranes and the time course of GTPyS (100 p~) activation of adenylyl cyclase was determined. The protein content in the cholate extracts was adjusted while maintaining a concentration of 1% cholate so that the GTP-ySstimulated V,, for adenylyl cyclase-reconstituted activity would be approximately equal for Ga. and Ga./i(38)9. The inset shows the activation of adenylyl cyclase in the same assay but in the presence of 100 p~ GTP showing an increased rate of adenylyl cyclase stimulation with GTP that was similar to that observed with GTP-yS.
constitutively elevating cyclic AMP synthesis in cells.
At present, the molecular basis for the G~u./~(38) phenotype is unclear. We have no evidence that the difference in 3'untranslated regions of pGa. and pGa.li(38) influence the phenotype of the cells. It also appears that the Ga.li (38) polypeptide is expressed at low levels in CHO cells and has eluded detection with currently available antibodies. The pGa./i(38) construct, however, was shown to code for a chimeric protein in transient COS cell transfections. Several characteristics of the Ga.li(38) phenotype can be defined. First, increased activation of adenylyl cyclase is observed for both GTP and GTPyS. Second, expression of GaSli(38) results in the activation of adenylyl cyclase and is not the result of FIG. 10. Immunoblot analysis of steady-state levels of endogenous G-proteins in Ga. and Ga.,1(38)9 relative to wildtype CHO cells. Cholate extracts of membranes from wild-type (lones 1, 4, and 7), Ga. (lunes 2, 5, and 8) and Ga.,;(38)9 (lanes 3, 6, and 9) CHO cells were prepared and 100 pg of extract protein resolved by SDS-polyacrylamide gel electrophoresis on 12.5% acrylamide gels. Proteins were transferred to nitrocellulose and immunoblotted as described under "Materials and Methods" with antisera recognizing the @-subunit (lanes Id), the Gai carboxyl terminus (lanes 4-6). and the Ga. carboxyl terminus (lanes 7-9). As shown in Fig. 3, the G q carboxyl-terminal antibody appears to poorly recognize Ga./i(38), and the chimera is not readily visualized in these blots. Filters were exposed for 8, 12, 24, and 48 h, and each lane was scanned by densitometry three times and averaged for the value of the appropriate 8, ai, or a. band. Densitometric values were averaged for the 12and 24-h exposures. Assigning an arbitrary unit of 1.0 for 8, ai, and a. in wild-type CHO cells, Gas cells expressed 3,3, and 1.7 units and Ga.,i(38)9 cells expressed 1.8, 3, and 1.3 units of 8, ai, and as, respectively. The determinations are representative of three different experiments with different wild-type, Ga., and Ga.li(38) CHO clones. phosphodiesterase inhibition or altered cellular cyclic AMP extrusion. Third, we have found that pertussis toxin treatment of Ga.li(38) expressing clones did not influence the properties of adenylyl cyclase activation in these clones relative to wild-type cells. Nor does pertussis toxin treatment of CHO cells mimic expression of the Ga.li(38) transcript? The constitutive elevation of cyclic AMP with Ga.li(38) expression does not, therefore, appear to be due to an inhibition of Gi, but appears to be due to a constitutively active Ga,-like activity. Because of the low level of expression in CHO cells it is not clear whether Ga,/i(38) directly interacts with catalytic adenylyl cyclase or influences the regulation of the endogenous Ga, polypeptide. We have been unable to express Ga.li(38) in S49 cells in an attempt to answer this question, presumably because of the constitutively active nature of this construct, and the fact that S49 cells are killed by elevated cyclic AMP levels. Using other constructs such as pGa,, we have been able to successfully transfect S49 cells and express functional Ga. proteins suggesting expression of Ga.li(38) is lethal to cyc-S49 cells. However, the same phenotype as described for Gaa/i(38)/CHO cells has now been defined in other cell types such as CCL39 fibroblasts: indicating the constitutive activation of cyclic AMP synthesis is a characteristic of the Ga,li(38) chimeric construct.
Immunoblot analysis, using antisera raised against peptides corresponding to the carboxyl terminus of Gai and Ga, and an antiserum raised against / 3, indicated the phenotype observed with expression of Ga.li(38) cannot be explained by an alteration in the expression of endogenous CQ, as, or subunits. Fig. 10 shows that both Ga. and Ga.li(38) cause similar changes in the expression of G-protein subunits. Relative to wild type, expression of a . and a.li(38) cDNAs increased /3 subunit levels 3-and 1.8-fold, respectively. Similarly, as and ai subunits were increased 1.7-and %fold for Gas and 1.3and %fold for Ga./i(38)-expressing clones. Because Ga, expression has little measurable influence on cyclic AMP levels in CHO cells, and both Ga. and Ga.li(38) influence a,, ai, and / 3 subunit expression similarly, these changes do not appear to induce the phenotype of Ga,/i(38) activation of adenylyl cyclase.
To our knowledge, this is the first demonstrated strategy for the constitutive activation of cyclic AMP synthesis by the stable expression of a chimeric mutant Gas cDNA. Bacterial expression and reconstitution of the chimeric protein with the Bysubunit complex and adenylyl cyclase will be required to understand the molecular mechanism for the Ga.li(38) effect on cyclic AMP synthesis. Given the conservation in sequence among the Ga subunits of the various G-proteins (19) similar constructs may give similar constitutively active Gi or Go phenotypes. This should provide an important new strategy for studying the regulation of cell function by Gproteins.