Cholera Toxin Induces CAMP-independent Degradation of G,*

Cholera toxin stimulates adenylyl cyclase by catalyzing ADP-ribosylation of the a chain (a.) of G., a guanine nucleotide binding regulatory protein. In a rat pituitary cell line, GH3, the toxin-induced increase in GTP- dependent adenylyl cyclase activity is maximal at 1 h; adenylyl cyclase remains elevated for at least 32 h. Surprisingly, cholera toxin also induces a 74-95% decrease in the amount of immunoreactive as in the same cells, as assessed on immunoblots probed with either of two antisera directed against separate a. peptide se- quences. The decrease in immunoreactive as, which begins after 1 h of toxin treatment and is complete by 8 h, is accompanied by a comparable decrease in the amount of biochemically active as, as assessed by its ability to complement the biochemical defect of as- deficient 549 cyc- membranes. Cholera toxin induces similar

Cholera toxin stimulates adenylyl cyclase by catalyzing ADP-ribosylation of the a chain (a.) of G., a guanine nucleotide binding regulatory protein. In a rat pituitary cell line, GH3, the toxin-induced increase in GTPdependent adenylyl cyclase activity is maximal at 1 h; adenylyl cyclase remains elevated for at least 32 h. Surprisingly, cholera toxin also induces a 74-95% decrease in the amount of immunoreactive as in the same cells, as assessed on immunoblots probed with either of two antisera directed against separate a . peptide sequences. The decrease in immunoreactive as, which begins after 1 h of toxin treatment and is complete by 8 h, is accompanied by a comparable decrease in the amount of biochemically active as, as assessed by its ability to complement the biochemical defect of asdeficient 549 cyc-membranes. Cholera toxin induces similar decreases in a. in wild type 549 lymphoma cells, in S49 kin-mutants, which lack CAMP-dependent protein kinase, and in 549 H21a mutants, in which a . is unable to assume an active conformation upon binding GTP. The toxin-induced decrease in a . is somewhat temperature-dependent, but is not blocked by agents that increase lysosomal pH or by colchicine, which promotes breakdown of microtubules. a . in detergent-solubilized GH3 membranes is susceptible to proteolysis by an endogenous protease; this susceptibility is markedly increased in membranes from cells previously exposed to cholera toxin for 1 h. Taken together, these results suggest that cholera toxin-induced covalent modification of a . marks the protein for accelerated degradation. In addition, the persistence of elevated GTP-dependent adenylyl cyclase activity despite loss of a substantial fraction of as suggests that the amount of a . membranes is greater than the amount necessary for maximal activation of cAMP synthesis by cholera toxin.
The G proteins, a family of membrane-bound guanine nucleotide binding proteins, play key roles in transducing hormonal and sensory signals (1-3). These proteins are heterotrimers, composed of a (39-52 kDa), p (35)(36), and y (8-10 kDa) chains. The proteins are distinguished principally by their structurally distinct a chains, which bind and hydrolyze GTP; the more highly conserved p and y chains serve to attach the G proteins to the cytoplasmic face of the plasma membrane and to present the a chain to the receptor. The * This work was supported in part by Grants GM-27800 and GM-28310 from the National Institutes of Health and grants from the March of Dimes. 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.
$To whom correspondence and reprint requests should be addressed. mechanism of signal transduction by these proteins is best characterized for G., the stimulatory regulator of adenylyl cyclase, and retinal transducin, which mediates stimulation of cGMP phosphodiesterase by photorhodopsin (1-3). In both cases, hormone-or light-activated receptors promote binding of GTP by the G protein a chain, which in turn stimulates the appropriate effector enzyme; stimulation of the effector is terminated when the a chain hydrolyzes its bound GTP.
Bacterial toxins induce characteristic changes in the function of G proteins by catalyzing covalent modification of their a chains (1-3). Cholera toxin catalyzes transfer of ADP-ribose from NAD' to the a chains of G. and transducin, while pertussis toxin attaches ADP-ribose to a different amino acid residue on the a chains of transducin, Gi, and Go. ADPribosylation of the a chain of G, (a.) by cholera toxin stabilizes the GTP-bound conformation of a. and decreases its intrinsic GTPase activity, thereby producing increased stimulation of adenylyl cyclase and elevated intracellular cAMP (1-4). Most cells contain two distinguishable forms of aa, 52 and 45 kDa in apparent molecular weight, which are produced by alternative splicing of primary transcripts of a single aB gene (5, 6).
Despite rapid recent advances in the molecular characterization of G protein structure and function, we know remarkably little about the processes that determine their relative abundance in cells. In many cases, including that of hormonestimulated adenylyl cyclase, the stoichiometric relation among concentrations of G proteins, receptors, and effectors remains poorly defined. In the course of investigating effects of cholera toxin in a rat pituitary cell line, GH3, we came upon a surprising observation that sheds light on these issues: Several hours of exposure to cholera toxin caused a 74-95% decrease in the amount of immunoreactive as in GH3 cells, while GTP-dependent adenylyl cyclase activity remained elevated. Here we report experiments that explore the biochemical mechanism and implications of this phenomenon.

EXPERIMENTAL PROCEDURES
Materials-Cholera toxin was obtained from List Biological (Campbell, CA). Radioactive ATP, NAD', and cAMP were obtained from Du Pont-New England Nuclear. '251-Protein A was supplied by Amersham. All other reagents were purchased from Sigma.
Cell Culture-GH3 cells, obtained from the Cell Culture Facility, University of California (San Francisco, CA), were propagated in Dulbecco's modified Eagle's-H21 medium with 10% fetal calf serum, as described (7). Wild type and mutant S49 mouse lymphoma cells were propagated in the same medium, supplemented with 10% horse serum, as described (8).
5352 500 X g for 5 min, and the supernatant fraction (hereafter called the postnuclear supernatant fraction) was then centrifuged a t 16,000 X g for 10 min. The resulting pellet (membrane particulate fraction) was resuspended in 50 mM Tris.HCI (pH 7.5) containing 1 mM PMSF. PMSF was omitted from the preparation in experiments designed to test effects of endogenous proteases. Aliquots of membranes were stored a t -70 "C before use. Protein content was measured by the method of Lowry et al. (9).
Adenylyl Cyclase and G, Actiuities-Adenylyl cyclase was measured by the method of Salomon et al. (lo), slightly modified as described (7). For assays of G. activity, GH3 membranes (0.1 mg/ml) were extracted with 0.2% (w/v) Lubrol ( l l ) , and G. activity in the extract was assessed by its ability to complement the G. deficiency of S49 mouse lymphoma cyc-membranes in vitro, exactly as described (7,12,13). Briefly, the Lubrol extract (0-20 p l ) from GH3 membranes (donor extract) was added to cyc-membranes (200 pg) and incubated a t 30 ' C for 40 min in the presence of GTP or Gpp(NH)p (each a t 100 p~) .
Then, [32P]ATP was added, and adenylyl cyclase activity was measured for 20 additional min in a total volume of 100 p l . In order to maintain constant concentrations of detergent, protein, and lipid in reactions containing less than 20 pl of donor extract, these reactions were supplemented with appropriate volumes (20-0 p l ) of donor extract that had been inactivated by heating a t 90 "C for 10 min. In these experiments, unless otherwise stated, the following protease inhibitors were used in combination to retard degradation of a . in the extract: PMSF (1 mM) and 1 PM concentration of each of the following: leupeptin, antipain, pepstatin A, and chymostatin.
Toxin-catalyzed Radiolabeling-Particulate fractions from control cells or cells treated for varying times with cholera toxin were subjected to cholera toxin-catalyzed ADP-ribosylation, using radioactive NAD+, exactly as described (14). After gel electrophoresis, autoradiograms were quantitated by scanning with a Hoeffer densitometer. Relative intensities of the radioactive bands were assessed by comparison to an autoradiogram of serially diluted control GH3 membranes (i.e. [32P]ADP-ribosylated membranes from cells not previously exposed to cholera toxin).
Antisera-Antibody U10 was raised (17) against a synthetic peptide corresponding to a sequence near the carboxyl terminus of a. (residues 323-338, DATPEPGEDPRVTRAK). Antiserum A569, generously provided by Susan Mumby and Alfred G. Gilman, is directed against a conserved sequence located near the amino terminus of G protein a chains (residues 47-61 of a . , GAGESGKSTIVKQMK); it detects a broad spectrum of G protein a chains (8).

RESULTS
After exposure to cholera toxin (1 pg/ml) for 8 h, GHs cells were homogenized, and fractions separated by centrifugation were analyzed by immunoblotting with antibodies against G protein a chains (Fig. 1). Cholera toxin caused a marked reduction in the 52-kDa form of as, as assessed by an antiserum (UlO), directed against a sequence near the carboxyl terminus of a . (Fig. 1, lane 2 versus lane I); the effect of cholera toxin was less marked on the 45-kDa form of as, which is much less prominent in GHs membranes than the 52-kDa form (7). The possibility that cholera toxin treatment induced transfer of a . from membranes to cytosol is ruled out by the observation that the toxin-induced decrease was also seen in postnuclear supernatant fractions (Fig. 1, lane 4 versus lane  3), which contain both membranes and cytosol. Parallel decreases in immunoreactive an, in both membranes and postnuclear supernatant fractions, were also seen with an antibody (A569) directed against a peptide sequence located near the amino terminus of G protein a chains (Fig.  1, lanes 5-8). The parallel results with two antibodies directed against spatially separate portions of the a. polypeptide make it quite unlikely that toxin-catalyzed ADP-ribosylation somehow prevented these antibodies from detecting as. In addition, the A569 antibody showed that cholera toxin caused no change in the amount of a broad band migrating a t approximately 40 kDa, which corresponds to the a chains of Gi and Go. Antibody U10 recognized a 58-kDa band in postnuclear supernatant fractions (Fig. 1, lanes 3 and 4); this band was unaffected by cholera toxin and was not detected by antibody A569; it is almost certainly not a G protein a chain.
Cholera toxin is known to activate adenylyl cyclase in a time-and concentration-dependent manner. In our experiments, cholera toxin in culture medium a t concentrations ranging from 100 ng to 3 pg per ml caused quantitatively similar elevations of adenylyl cyclase and decreases in immunoreactive a . a t 8-16 h (data not shown). Fig. 2 shows the time course of changes induced by a single concentration of toxin (1 pg per ml). The toxin caused a progressive decrease in immunoreactive a . , beginning after the first hour of incubation and complete by 8 h (Fig. 2, open squares). By this time, densitometric analysis of immunoblots showed a 92% reduction in the 52-kDa form of as. Visual inspection of the autoradiograms (Fig. 2, bottom) shows that the 45-kDa form was also decreased; the small size of the 45-kDa peak prevented its reliable quantitation by densitometry. In other experiments, exposure of GH3 cells to cholera toxin for 8 h led to decreases in immunoreactive a . that ranged from 74 to 95%.
In contrast to the decrease in immunoreactive as, cholera toxin induced a 4-to 5-fold increase in GTP-stimulated adenylyl cyclase activity, which reached a maximum a t about 1 h and persisted for 32 h (Fig. 2, solid squares).
To assess the extent of ADP-ribosylation of as in intact cells exposed to cholera toxin, we measured the ability of the toxin to transfer radiolabel from ["PINAD' to the 52-kDa a . polypeptide in particulate fractions prepared from cells ex- posed for various times to cholera toxin in culture; only a . molecules that were not ADP-ribosylated by the toxin in culture would be susceptible to subsequent radiolabeling catalyzed by cholera toxin in particular fractions. By this measure (Fig. 2, open circles), only a small portion ( 4 5 % ) of a .
remained unmodified by toxin after 1 h; at this time, activation of adenylyl cyclase had reached its maximum, and immunoreactive a . had not yet begun to fall. Toxin-catalyzed ADP-ribosylation of a . in intact cells preceded disappearance of the immunoreactive polypeptide by 1.5-2 h. The decrease in immunoreactive as was maximal within 8 h of culture; by this time, the amount of a . accessible to radiolabeling by [32P] NAD+ in uitro approached zero.
The results shown in Fig. 2 pose a paradox. Despite the disappearance of 92% of immunoreactive a . , the elevation of GTP-dependent adenylyl cyclase remained essentially unchanged. We therefore asked (Fig. 3) whether cholera toxin treatment induced a decrease in the amount of biochemically active as, in parallel with the decrease in immunoreactive a . . Biochemical activity of a. was measured by virtue of its ability to complement the a . deficiency of S49 cyc-membranes (11,19). In order to assure that a . activity would be limiting, the complementation assay used low amounts of donor (GH3) extract and high concentrations of cyc-membranes (see Fig.  3 legend). Consequently, GTP-stimulated adenylyl cyclase activity in the reconstituted mixture increased almost linearly with increasing amounts of GH3 extract (Fig. 3). Comparing extracts from cells treated with cholera toxin for 1 ucrsus 8 h, the biochemical activity of a. in the latter extract was reduced by 69% (Fig. 3). The results were similar when 5"guanylyl imidodiphosphate (Gpp(NH)p) was substituted for GTP. The 69% reduction in biochemical activity of a . correlated well with the 74% reduction in immunoreactive 52-kDa a . in the GH3 donor membranes used in the same experiment (assessed by densitometric analysis of the immunoblot shown in the inset of Fig. 3). Additional cyc-reconstitution experiments, not shown, revealed similar toxin-induced reductions in the These data suggest that the amounts of ADP-ribosylated a . differ in cells exposed to toxin for 1 uersus 8 h, but that the intrinsic ability of a . to stimulate adenylyl cyclase is not changed. The paradoxical observation that adenylyl cyclase activities are similar in GH3 membranes that contain very different amounts of ADP-ribosylated and biochemically active a . (Fig. 2, 1 uersus 8 h) suggests that maximal cAMP synthesis by such membranes is limited by their content of catalytic adenylyl cyclase rather than by the amount of active as. In other words, membranes from GH3 cells exposed to toxin for 1 h contain "spare" a . , substantially in excess over the amount needed to produce maximal stimulation of adenylyl cyclase. Accordingly, subsequent disappearance of a . in toxin-treated cells would not be expected to reduce adenylyl cyclase activity until a . becomes limiting; this does not appear to occur even after a greater than 90% reduction in a . (see Fig. 2).
The cholera toxin-induced decrease in immunoreactive a . is not limited to GH3 cells. Exposure of wild type and three mutant S49 mouse lymphoma cell lines to the toxin for 8 h resulted in a substantial decrease in membrane an, as assessed with the U10 antiserum (Fig. 4A)

. The mutants included cells with altered a. (urn and H21a) (8) and a cell line lacking CAMP-dependent protein kinase (kin-) (20).
In addition, agents that increase or mimic cAMP (Fig. 4B,   lanes 3 and 4 ) did not reproduce the action of cholera toxin in GH3 cells. These results, along with those in S49 H21a and kin-cells, conclusively rule out cAMP as a mediator of the toxin-induced decrease in immunoreactive as.
The cholera toxin-induced decrease in a. was partially temperature-dependent (Fig. 4C). Immunoreactive 52-kDa a . was measured in GH3 cells that were exposed to toxin for 1 h  H21a (lanes 3 and 4 ) , kin-(lanes 5 and 6 ) , and  unc (lanes 7 and 8) a t 37 "C and then incubated in the same medium for a further 7 h at 37, 4, or 18 "C (Fig. 4C, lanes 3, 4, and 5, respectively). The toxin-induced decrease in a . was diminished at 4 "C (30% decrease) and 18 "C (70% decrease), as compared to 37 "C (>go% decrease). This difference was not due to ineffective ADP-ribosylation, because, regardless of the temperature during the last 7 h of incubation, most of the a. in the membranes was in the ADP-ribosylated form, as assessed by its unavailability for radiolabeling in uitro by cholera toxin and ["PI NAD' (results not shown). The relatively weak temperature dependence of the cholera toxin-induced decrease in a . suggests that endocytosis may not be involved. Because lysosomes are involved in ligand-induced down-regulation of cell surface receptors (21,22), we asked whether lysosomotropic agents could block the cholera toxin-induced decrease of as. Neither ammonium chloride (10 mM) nor chloroquine (500 p~) altered the effect of cholera toxin (Fig. 4C, lanes 6 and 7  uersus lane 3). Similarly, colchicine (3 p~) , which blocks assembly of microtubules and can prevent redistribution of organelles (23), also had no effect (Fig. 4C, lane 8 uersus lane   3).

W T H a C T -+ -+ -+ -
Taken together, these experiments provide no evidence for possible roles of movement and fusion of organelles, including lysosomes, in the cholera toxin effect. The experiments shown in Fig. 5, however, do provide a strong hint that cholera toxin treatment increases the susceptibility of a, to cleavage by an endogenous protease and also suggest that accessibility of a . to the protease in uitro requires the presence of a detergent. In the first experiment (Fig. 5A), membranes were prepared from GHa cells that had been incubated with or without cholera toxin for 1 h. The membranes were then incubated with 0.2% Lubrol, a nonionic detergent, for 8 h at 4 or 37 "C in the presence or absence of a mixture of protease inhibitors. The presence of Lubrol caused the 52-and 45-kDa an bands to migrate as a fuzzy broad band in SDS, but did not obscure  with (lanes 2, 4, and 6 ) or without (lanes I , 3, and 5) cholera toxin were lysed by Dounce homogenization, and membranes were prepared (see "Experimental Procedures"). Lubrol (0.296, w/v) was added to the membranes, and the mixture was incubated for an additional 8 h in the presence (lanes 1 and 2) or absence (lanes 3-6) of a mixture of protease inhibitors (+ or -PI; see "Experimental Procedures") a t 4 "C (lanes 1-4) or 37 "C (lanes 5 and 6 ) . R, postnuclear supernatant fractions (200 pg) were obtained from GH3 cells treated for 1 h with (lanes 8 and 10) or without (lanes 7 and 9 ) cholera toxin. These fractions were frozen (lanes 7 and 8 ) or incubated a t 37 "C for an additional 8 h (lanes 9 and 10). Numbers at the right and left margins indicate sizes of molecular weight markers (X lo-?. Arrows indicate the positions of the 52-and 45-kDa forms of as. Antiserum U10 was used in all immunoblots. the key result: a . from cholera toxin-treated cells was cleaved to a 40-kDa immunoreactive species a t 4 "C in the absence of protease inhibitors ( l a n e 4); under identical conditions, a, from control cells (not exposed to cholera toxin) was not detectably cleaved ( l a n e 3). The conversion of a . from 52 to 40 kDa was prevented by protease inhibitors ( l a n e 2). The addition of 100 p~ Gpp(NH)p to extracts from control cells did not accelerate degradation of as (data not shown), indicating that the degradation seen in extracts from toxintreated cells was not due to an increased susceptibility to proteases that might be induced by activation of G. . Incubation of detergent extracts for 8 h at 37 "C led to the appearance of a -38-40-kDa immunoreactive band, even in extracts of cells that had not been exposed to cholera toxin (Fig. 5A, lanes 5 and 6). The degradation of a . at 37 "C was more extensive, however, in extracts from cholera toxintreated cells (lam 6 uersus lane 5). These observations suggest that modification of a. by cholera toxin increases the protein's susceptibility to a protease that can also act, although less effectively, on unmodified as. Although the identity of the protease is unknown, it is probably a serine or cysteine protease, because chymostatin, antipain, leupeptin, and PMSF each was able individually to block degradation of a . in detergent extracts of toxin-treated cells (result not shown).
In membranes, immunoreactive a. is remarkably stable, even in the absence of protease inhibitors, as long as detergent is not present. After an 8-h incubation at 4 or even 37 "C (Fig.  5B), a. was not cleaved in broken cell preparations derived from either control or toxin-treated cells. This experiment (Fig. 5B) utilized postnuclear supernatant fractions, rather than particulate fractions, in order to provide optimal accessibility of membrane-bound a, to potential degradative enzymes in the cytoplasm. Addition of ATP and an ATPregenerating system in such incubations did not affect the results (not shown).
Because ADP-ribose attached to as appears to trigger its disappearance, it is reasonable to ask whether ADP-ribose attached to ai chains ( a subunits of Gi) causes these proteins to disappear also. This was not the case. Treatment of GHs cells with pertussis toxin had no effect on the amount of immunoreactive ai detected with the A569 antiserum (result not shown).

DISCUSSION
The mechanism by which cholera toxin activates adenylyl cyclase is well established (see Refs. 1-3 for review). The toxin catalyzes ADP-ribosylation of the a chain of G,, thereby stabilizing the protein in its active form; the resulting increase in active a. leads to increased adenylyl cyclase activity. Here we describe a previously unsuspected additional effect. Cholera toxin induces a delayed, progressive disappearance of a, from toxin-treated cells. The new finding raises two questions. 1) What mechanism is responsible for the toxin-induced loss of a,? 2) How can the disappearance of a. be reconciled with continued cholera toxin-induced elevation of adenylyl cyclase?
With respect to the first question, our experiments do not establish a precise mechanism by which cholera toxin induces disappearance of a., but do conclusively rule out several possibilities. For example, the decrease in a. cannot be attributed to masking of a specific a. epitope, because: (a) loss of immunoreactive protein was confirmed by two antisera, directed against widely separated peptide sequences in a. (Fig.  1); in addition, (b) the biochemical activity of a. decreased in parallel with immunoreactive protein (Fig. 3). Moreover, failure of forskolin and 8-bromo-CAMP to reproduce the effect of cholera toxin indicates that elevated cAMP is not responsible for the decrease in as.
Toxin-induced decreases of a. in membranes of mutant S49 mouse lymphoma cells rule out other hypotheses. The H21a mutation in as, which substitutes alanine for glycine at position 226 of a8 (24), prevents stimulation of adenylyl cyclase in response to cholera toxin and other agents that act via aB, although the mutant polypeptide is still a substrate for ADPribosylation by the toxin (8); the H21a mutation prevents as from assuming the "active" GTP-dependent conformation (24). Accordingly, susceptibility of a* in H21a cells to a decrease induced by the toxin makes it quite unlikely that this effect of cholera toxin requires stimulation of adenylyl cyclase or even an "active" conformation of as. Similarly, the S49 kinmutation (20) inactivates CAMP-dependent protein kinase activity; the toxin's ability to reduce immunoreactive as in kincells indicates that CAMP-dependent protein kinase is neither a mediator of nor a permissive participant in the effect. Finally, the unc mutation of a. uncouples G. from interactions with receptors (8, 25,26); susceptibility of as in unc cells to the toxin effect rules out the possibility that the decrease requires interaction with GB-coupled receptors.
The simplest explanation of the data is that the cholera toxin-induced covalent modification of a, somehow marks the protein for proteolytic degradation. Toxin-induced degradation of as is consistent with our failure to find a shift of aB from membranes to cytoplasm, with the relative rapidity of the effect, which was completed between 1 and 8 h of toxin treatment, and with observations (not shown) that cycloheximide (5 wg/ml) neither induced a decrease in ae (even after 16 h) nor altered the toxin-induced decrease in as.
The putative degradative pathway triggered by ADP-ribosylation of as has not been identified. as in particulate fractions is susceptible to an endogenous protease only in the presence of added detergent, and this protease degrades the protein in extracts from cholera toxin-treated cells to a greater extent than in control extracts (Fig. 5). The requirement for detergent makes it attractive to imagine that toxin-catalyzed ADP-ribosylation in intact cells marks a. for degradation by a protease from which the protein is normally separated by a membrane. This notion is neither supported nor conclusively refuted by other data. Indeed, the 40-kDa as degradation product seen in detergent extracts is not seen in intact cells treated with cholera toxin; thus, the endogenous protease that cleaved a. in detergent extracts is not necessarily responsible for toxin-induced degradation of a. in intact cells. Furthermore, the weak temperature dependence of the toxin effect (Fig. 4C) and the lack of effect of lysosomotropic agents argue, albeit weakly, that neither endocytosis nor maintenance of an acidic pH in lysosomes is required for degradation of as.
In summary, the data suggest that attachment of ADPribose to as somehow increases the protein's susceptibility to proteolytic degradation, but do not identify the degradative pathway involved.
The elevation of adenylyl cyclase induced by cholera toxin persists for at least 32 h, despite a substantial decrease in as, which is completed much earlier, by 8 h (Fig. 2). The demonstration that toxin reduces the amount of biochemically active a. (Fig. 3) rules out the possibility that the toxininduced decrease in immunoreactive protein was selective for a subpopulation of inactive as molecules. Taken together, the evidence suggests that a8 in toxin-treated cells is spare, at least early in the response (e.g. at 1 h), i.e. that the concentration of a. is substantially higher than is required for maximal activation of adenylyl cyclase. Consequently, if we assume that GTP-dependent activation of the enzyme requires a one-to-one complex of a. and adenylyl cyclase, the concentration of a, in GH3 membranes must be severalfold greater than that of the catalytic protein, adenylyl cyclase. To our knowledge the relative concentrations of a. and adenylyl cyclase have not been defined in any mammalian cell.
The inference that as is present in excess of adenylyl cyclase appears to contradict a number of observations indicating that G, limits the maximal activity of adenylyl cyclase. Thus, glucocorticoids induced an increase in G. activity and a. mRNA in GH, cells, changes which were associated with a parallel increase in adenylyl cyclase activity (7). In ethanoltreated neuroblastoma cells, a decrease in ae protein and mRNA was associated with decreased hormone-sensitive adenylyl cyclase (27). In addition, inherited deficiency of as, caused by loss of an autosomal allele, produces the characteristic resistance to parathyroid hormone and other hormones that is seen in pseudohypoparathyroidism, Type I (28, 29). Taken together, the evidence suggests that both increases and decreases in the expression of as produce parallel changes in adenylyl cyclase activity and implies that the amount of available G, is rate-limiting for stimulation of cAMP synthesis.
This implication can be reconciled with our present results if we take account of the possibility that ADP-ribosylated a. may be more spare than the unmodified protein. In membranes of control cells, i.e. cells not exposed to cholera toxin, the transience and reversibility of each as-adenylyl cyclase interaction could allow the concentration of a,-adenylyl cyclase complexes to depend on the concentration of free (uncomplexed) a,; thus, in control cells, the amount of a. can be in molar excess, relative to adenylyl cyclase, but may still limit maximal stimulation of cAMP synthesis. In membranes from toxin-treated cells, however, each a,-adenylyl cyclase complex would be expected to persist in the active (CAMPsynthesizing) form for minutes rather than seconds, because cholera toxin-catalyzed ADP-ribosylation stabilizes the active GTP-bound form of as (1-4); the relative irreversibility of the a,-adenylyl cyclase interaction would make toxin-modified as a much more potent stimulator of adenylyl cyclase.
Critical tests of these notions will require measurements of adenylyl cyclase in lipid vesicles that incorporate known amounts of G, components and adenylyl cyclase or experimental manipulations that alter the relative expression of as, adenylyl cyclase, and other components of the adenylyl cyclase system in intact cells.