Cell, Vol. 70, 411-418,

August

7, 1992, Copyright

0 1992 by Cell Press

Phospholipase C-PI Is a GTPase-Activating for Gglll, Its Physiologic Regulator Gabriel Berstein,’ Jonathan L. Blank,t Deok-Young Jhon,* John H. Exton,t Sue Goo Rhee,S and Elliott ht. Ross’ Department of Pharmacology University of Texas Southwestern Medical Center Dallas, Texas 752359041 t Howard Hughes Medical Institute Department of Molecular Physiology and Biophysics Vanderbilt University School of Medicine Nashville, Tennessee 37232-2195 *Laboratory of Biochemistry National Heart, Lung and Blood Institute National Institutes of Health Bethesda, Maryland 20892 l

Summary Purified Ml muscarlnlc cholinergic receptor and G,,M were coreconstituted in llpid vesicles. Addition of purifled phospholipase C-p1 (PLC-pl) further stimulated the receptor-promoted steady-state GTPase activity of Gqlll up to PO-fold. Stimulation depended upon receptor-medlated GTP-GDP exchange. Addition of PLCpl caused a rapid burst of hydrolysls of GWllbound GTP that was at least IO-fold faster than In its absence. Thus, PLC-(31 stlmulates hydrolysis of GWI1bound GTP and acts as a GTPase-activating protein (GAP) for its physiologic regulator, G,,,,,. GTPasestimulating actlvlty was specific both for PLCpl and Gull. Such GAP activity by an effector coupled to a trimeric G protein can reconcile slow GTP hydrolysis by pure G protelns In vitro with fast physiologic deactivation of G protein-mediated signaling. Introduction G proteins convey information from cell surface receptors to intracellular effector proteins by following a controlled cycle of activation and deactivation. G proteins are activated by binding GTP. They are deactivated when they hydrolyze bound GTP to GDP; GDP does not cause activation. Hormone-liganded receptors promote the activation of G proteins by catalyzing the binding of GTP, both directly and by accelerating the release of bound GDP. Receptors have no effect on GTP hydrolysis, and the deactivation of G proteins by GTP hydrolysis has been widely assumed to be unregulated (for reviews see Bourne et al., 1990, 1991; Kaziro et al., 1991; Ross, 1989). The deactivation of monomeric, low molecular weight GTP-binding proteins, such as ras and rho, is promoted by GTPase-activating proteins (GAPS; Trahey and McCormick, 1987; reviewed by Bollag and McCormick, 1991; McCormick, 1989; Bourne et al., 1990). GAPS stimulate the hydrolysis of bound GTP and thus accelerate deactivation about lOO-fold. Without GAPS, the GTPase activities of the small GTP-binding proteins are negligible. Because

Protein

the GTP-binding a subunits of heterotrimeric G proteins are structurally similar and functionally homologous to the small GTP-binding proteins, the existence of GAPS for G proteins has been a subject for speculation. Some data suggest that there is a cellular mechanism to accelerate deactivation of G proteins. Hydrolysis of GTP bound to isolated G proteins is slow when measured in vitro. The tH for hydrolysis is 10-20 s for most G proteins (Brandt and Ross, 1988; Higashijimaet al., 1987; reviewed in Gilman, 1987), 50 s for Gql, (a purified mixture of closely related G, and G,,; Berstein et al., 1992; Strathmann and Simon, 1990), and >7 min for G, (Casey et al., 1990). For Gs-stimulated adenylylcyclase, the rate of deactivation after hormonal stimulation (Cassel et al., 1977) agrees reasonably well with the measured rate of GTP hydrolysis by purified and reconstituted G, (Brandt and Ross, 1988). However, the rates of deactivation of other G protein-regulated effecters are considerably faster than the rates at which their isolated G proteins hydrolyze bound GTP in vitro. For example, isolated transducin hydrolyzes bound GTP with a tH of about 15 s (Navon and Fung, 1984), but the deactivation of transducin-stimulated cGMP phosphodiesterase in rod outer segments takes less than 1 s (Dratz et al., 1987; Vuong and Chabre, 1990,199l; Arshavsky et al., 1991). Similarly, isolated Gi hydrolyzes GTP as slowly as does purified transducin (Higashijima et al., 1987), but a Grregulated K+ channel in cardiac myocytes is deactivated about 100 times as fast upon removal of agonist (Breitwieser and Szabo, 1988). Such discrepancies between slow GTP hydrolytic rates in vitro and rapid deactivation rates in cells argue that GAPS for G proteins must exist. G proteins do appear to have the potential to hydrolyze GTP rapidly. For example, G, that has been removed from detergent solution hydrolyzes bound GTP at least 20-fold faster than G, in detergent micelles or reconstituted into phospholipid vesicles (Brandt and Ross, 1985). It has been suggested that effector proteins such as adenylylcyclase, phospholipase C-81 (PLC-pl), or cyclic GMP phosphodiesterase may act as GAPS for the G proteins that regulate them. Direct evidence for this idea has been limited, though. In vitro studies of the GqrI-mediated activation of PLC-81 by the Ml muscarinic cholinergic receptor suggested that PLC-51 may contribute to the deactivation of Gqlll (Berstein et al., 1992). Similarly, biphasic GTP hydrolysis by transducin in retinal rod outer segments appeared related to the presence of cyclic GMP phosphodiesterase because the rapid phase of hydrolysis was inhibited by cyclic GMP and because the amount of rapid hydrolysis was roughly stoichiometric with the phosphodiesterase (Arshavsky et al., 1991). Genetic evidence suggested that rasGAP binds to ras at a site that is needed to convey a signal to the still unknown ras effector (Adari et al., 1988; Cales et al., 1988). This finding prompted speculation that rasGAP is an effector protein for ras and, by extension, that other GAPS are effecters for ras-like monomeric GTP-binding proteins (Bollag and McCormick,

Cell 412

1991). Other data, however, argue that rasGAP is primarily a negative regulatory component in the ras signaling system (Tanaka et al., 1990; Hall, 1990 for review). To address the possibility that effecters may function as GAPS for G proteins, we have examined the effect of PLC-51 on GTP hydrolysis by GwI1, the G protein that stimulates PLC-81 in response to Ml muscarinic cholinergic receptors. PLC-51 stimulates the hydrolysis of GTP bound to Gqlll by 50-fold. PLC-51 is thus a GAP for the G protein that mediates its activation.

PLCpl Accelerates GTP Hydrolysis by Gqlll GdrI, which mediates activation of PLC-51 by Ml muscarinic cholinergic receptors, displays negligible steadystate GTPase activity by itself because it binds GTP and releases GDP very slowly (Berstein et al., 1992). When G+, and Ml receptor are coreconstituted in lipid vesicles, nucleotide exchange and steady-state GTPase are accelerated upon addition of agonist. GTPase activity remains relatively low, however, in part becauseof the slow intrinsic rate of hydrolysis of GTP bound to GUI,. The rate constant for the hydrolytic step, kcat, is only about 0.8 min-‘, which corresponds to a half-life for GTP-activated GM,, of about 1 min (Berstein et al., 1992). The receptor-stimulated GTPase activity of GUI, wasfurther stimulated lo- to 20-fold by the addition of recombinant PLC-j31 purified from HeLacells(Figure 1). Theextent of stimulation by PLC-81 depended on the concentrations of PLC-81, receptor, and agonist, but has reproducibly fallen within this range (lo- to 20-fold) in a large number of experiments using three preparations of PLC-51 and four preparations of Gq/,, . Stimulation of steady-state activity was half-maximal at about 2 nM PLC-51 in three experi-

-0 0.02

0.2 [PLC-@l]

Figure 1. PLC-61 ity of G.w

Accelerates

2

20

200

(red)

the Receptor-Stimulated

GTPaseActiv-

Ml muscarinic receptors and G4(,, were reconstituted into lipid vesicles and the vesicles were mixed with increasing amounts of either purified recombinant PLC-61 (open circles) or PLC-61 purified from bovine brain (closed circles). GTP hydrolysis was assayed at 60% for 15 min. All assays contained 1 mM carbachol. Half-maximal stimulation was obtained at 1.9 nM recombinant PLC-61. The concentration of receptor was 0.35 nM (open circle) or 0.5 nM (closed circle). The concentration of receptor-coupled GWjI, determined by receptor-stimulated [%IGTPyS binding, was 0.80 nM (open circle) or 1 .O nM (closed circle) in the absence of PLC-61 and 1.2 nM (open circle) or 1.4 nM (closed circle) in the presence of 67 nM (open circle) or 22 nM (closed circle) PLC-61. The concentration of Gqlrt according to the assay of bound GDP was either 2.0 nM (open circle) or 3.3 nM (closed circle).

ments performed as described in the legend to Figure 1. This concentration is greater than the concentration of GWI and is close to the EC,,, with which GTPyS-activated Gqltl stimulates PLC-pl(0.4nM; Smrckaet al., 1991; Blank et al., 1991; Berstein et al., 1992). This may therefore be the KD for the binding of the two proteins to each other. Note that maximal stimulation of GTPase was achieved at PLC-pl concentrations well above the concentration of GW (- 1 nM in Figure 1, for example). Heating the PLC-81 preparation at 100% for 5 min abolished its effect. Natural PLC-51 purified from bovine brain increased the steady-state GTPase activity of Ml muscarinic receptorG,,,, vesicles as well as did the recombinant enzyme (Figure 1). At concentrations up to 20 nM, the stimulatory activities of natural and recombinant PLC-51 were identical. Because only limited amounts of natural bovine PLC-51 are available, saturation could not be demonstrated. The concentration of agonist-liganded receptor had a profound effect on the PLC-j31-mediated increase in the steady-state GTPase rate. Carbachol caused substantial stimulation in the presence or absence of PLC-51, but stimulation by PLC-81 increased from a-fold in the absence of carbachol to 20-fold with increasing concentrations of the agonist. Vesicles that contained Gqlll but no receptor did not display detectable GTPase activity in the presence or absence of PLC-51. Thus, agonist-liganded receptor and PLC-81 are synergistic in their stimulation of the GTPase activity of G,,,,. About lo-fold more carbachol was needed to halfmaximally stimulate the GTPase reaction in the presence of PLC-51 than in its absence (- 15 PM versus - 1.5 PM; Figure 2A). This behavior indicates a greater dependence of the rate on agonist-liganded receptor in the presence of PLC-pl , consistent with the idea that receptor-catalyzed GTP-GDP exchange is the primary rate-limiting step in the presence of PLC-51, but that hydrolysis of bound GTP is rate limiting in its absence. Because the absolute extent of activation in the presence of PLC-51 was so much greater than in its absence, however, other explanations for this shift in agonist potency remain plausible. The effects of carbachol were identical in the presence of natural and recombinant PLC-51. Stimulation of GTPase by carbachol was blocked by the muscarinic antagonist atropine in the presenceor absence of PLC-51. Addition of atropine to the reaction mixture at steady-state inhibited further GTP hydrolysis immediately (Figure 2C), supporting the coordinate nature of the stimulation of steady-state GTPase activity of GUI, by both PLCpl and activated receptor. About lo-fold more atropine was required to inhibit steady-state GTPase activity in the absence of PLC-81 than in its presence (Figure 28) as is consistent with the higher ECso for agonist in the presence of PLC-51. Calculated Ki’s for atropine were about equal in the presence or absence of PLC-81(- 0.8 nM with PLCPl7 -0.4 nM without) and are consistent with the KD for this antagonist (G. B., unpublished data). Because both recombinant and natural bovine brain PLC-81 stimulate the GTPase activity of GuytI and because both preparations are highly purified, as judged by silver staining of polyacrylamide gels (Lee et al., 1987; Ryu et

Phospholipase 413

C-g1 Is a GAP for G,

.iz 20

(0)

3

2

-1

,....A_"

;: ._.'

,:'

,:' ..I'

,I' ,:' ,:'

_/

,:'

.yo

0 .

15

10

5

0

0

10.'

1o-6

10~5

1o-4

[Carbachol]

B I.,

,

0)

1o-3 1o-2

(M)

rff

3

[Atropine]

(M)

C 500,

/--t

L

/’ l/&&y 1 4

8 12 Time(mln)

i 16

20

Figure 2. Stimulation of Steady-State GTPase by PLC-pl Depends on Carbachol and Is Blocked by Atropine Ml muscarinic receptor-G,,, vesicles were assayed for GTPase activity in the presence or absence of 25 nM recombinant PLC-fll as described in Experimental Procedures. For (A) and (B), activity was measured for 15 min in the presence of the concentrations of carbachol or atropine shown on the abscissa. Different scales are used for the activities measured in the presence (closed circles) or absence (open circles) of PLC-bl (A) Stimulation by carbachol. Relative stimulation of activity by PLC-pl (dotted line) was 18-fold at saturating carbachol and 1.7-fold in the absence of carbachol. (6) Blockade of carbachol stimulation by atropine. Carbachol was present in all assays at 300 PM. (C)Onset of inhibition byatropine. Activitywas assayed in the presence of 300 uM carbachol. with (triangles) or without (circles) PLC+l. Atropine (50 )rM) was added at 9 min (closed symbols only). [“S]GTPyS binding activity (fmol) in the absence or presence of PLCbl , respectively, was: 28.8 and 40 (A), 24 and 30 (B), and 19 and 26 (C). The amount of receptor was 8.9 fmol (A), 10.6 fmol (B), and 8.3 fmol (C).

al., 1987a; Park et al., 1992), the observed GAP activity is apparently intrinsic to PLC-pl . However, we were able to detect minor nucleoside triphosphatase activity and substoichiometric guanine nucleotide binding in the recombi-

Figure 3. Effect of Monoclonal Anti-PLC-Bl Antibodies on PLC-fllStimulated GTPase Purified recombinant PLC-Bl (closed bars; 0.1 ug) or buffer(open bars) was incubated with 2 pg of either the indicated monoclonal antibodies (Suh et al., 1988)or bovine serum albumin (“Buffer”) for 30 min at room temperature followed by 30 min on ice. The mixtures were then added to reconstituted Ml receptor-G,,, vesicles and GTPase activity was assayed in the presence of 1 mM carbachol as described in Experimental Procedures. Vesicles contained 8.8 fmol of receptor and 20 fmol of [%]GTPyS binding activity.

nant PLC-pl preparation (see Experimental Procedures), which suggests that minor contaminants may be present. We therefore verified the identity of the GAP activity using monoclonal antibodies specific for PLC-Bl (Suh et al., 1988). Antibodies K-32-2 or L-54-4 both inhibited the GAP activity by 80%-90% (Figure 3). These antibodies recognize nearby or identical epitopes in the C-terminal portion of PLC-pl (Suh et al., 1988; unpublished data). Antibody K-92-3 inihibited the GAP activity less, and the other antibodies were increasingly weak inhibitors. This rank order of inhibitory potency is identical to that with which these six antibodies inhibit the stimulation of PLC-pl by AIF4-activated aq (D. Park and S. G. R., unpublished data). None of the antibodies inhibited carbachol-stimulated GTPase in the absence of PLC-01. These data further support the identity of PLC-01 and the GAP for Gqlll. The GAP activity of PLC-Bl is specific for GUI,. There was no effect of added PLC-01 on the steady-state GTPase activity of either G, reconstituted with P-adrenergic receptors or G, reconstituted with M2 muscarinic cholinergic receptors (Table 1). This is consistent with the inability of G, to stimulate PLC-pl (Smrcka et al., 1991).

Table 1. GAP Activity

of PLC-pl GTPase

Is Specific Activity

for G@,,

(fmollmin)

Agonist

Antagonist

Receptor

G

Control

+ PLC-Bl

Control

+ PLC-pl

Ml muscarinic M2 muscarinic b-adrenergic

q/l1 0 s

1.44 13.9 18.3

9.88 12.5 18.1

ND

ND

2.2 2.0

2.1 2.9

Receptors and G proteins were reconstituted as described in Experimental Procedures, mixed with PLC-pl or buffer as shown, and assayed for steady-state GTPase activity in the presence of agonist (1 mM carbachol for muscarinic receptors, 10 &I (-) isoproterenol for P-adrenergic receptor) or antagonist (50 PM atropine or 1 WM (- ) propranolol). ND: not detectable over background.

Cell 414

-0

2

4

6

Time (mln)

Figure 4. PLC-31 Promotes Rapid Hydrolysis of GTP Bound to Gql,, Reconstituted Ml receptor-GW,, vesicles (33 fmol receptor) were incubated for 60 s at 30°C with 50 nM [y-“PjGTP in the presence of carbachol as described in Experimental Procedures. A quenching mixture of GTP and atropine was then added. Four seconds later, defined arbitrarily as zero time on the figure, either 3.3 pmol of PLC-51 (closed circles) or buffer (open circles) was added and GTP hydrolysis was measured at the indicated times. The zero time point was sampled before addition of PLC-31 and the next three data points are at 4, 7, and 10 s. These data are representative of three similarly performed experiments and other assays designed to detect hydrolysis at short times.

Mechanism of GTPase Activation by PLC-pl The initial characterization of the PLC-pl-stimulated GTPase reaction indicated that PLC-pl accelerates steady-state GTP hydrolysis beyond the previously measured rate of hydrolysis of bound GTP (Berstein et al., 1992). PLC-91 must therefore increase the hydrolysis reaction itself, as is the case for the GAPS that act on small monomeric GTP-binding proteins. To examine the effect of PLC-pi on hydrolysis itself, we measured the rate of GTP hydrolysis during a single catalytic turnover (Figure 4). Gqlll was allowed to bind [Y-~P]GTP by incubating receptor-G4/,, vesicles and nucieotide in the presence of carbachol. Binding was terminated by adding a mixture of excess unlabeled GTP (to prevent further [Y-~P]GTP binding) and atropine (to prevent receptor-promoted GTP dissociation). PLC-51 (or buffer) was added immediately, and hydrolysis of bound GTP was then monitored as the formation of [32P]orthophosphate. In the absence of PLC-@l , GTP hydrolysis followed firstorder kinetics and displayed a rate constant of 0.9 min-‘, in agreement with previous measurements (Berstein et al., 1992). After addition of PLC-91, there was an initial rapid burst of GTP hydrolysis that was complete within 4 s, the earliest time point (Figure 4). This corresponds to a rate constant for GTP hydrolysis of at least 40 min-I, 50-fold faster than that measured in the absence of PLC-pl Similar behavior has been observed in five separate experiments, and the magnitude of the burst has varied from 20% to 40% of total [Y-~P]GTP hydrolysis. The burst is followed by a slower phase, similar to that observed without PLC-91, that completes the hydrolysis of the GUI,bound [Y-~P]GTP. The burst was not observed when PLC51 was heated at 100°C for 5 min. This behavior suggests that PLC-(31 binds GUI1 to form a pool of active GTPase that hydrolyzes GTP at least 50 times faster than does isolated G,,,,,. Our manual mixing protocol underestimates

the size of this pool for kinetic reasons, and the pool may also be limited by inaccessibility of some of the vesiclebound GUI,-GTP to the added PLC-pl. The extent of the burst may be increased by mixing G,,,,-GTP and PLC-51 in detergent solution. Fast kinetic measurements will be required to determine the actual extent of GTP hydrolysis by the G~I1-PLC-B1 complex, and such measurements should also provide an accurate measurement of the enhanced hydrolytic rate. Toexaminetheeffectof PLC-51 ontherateof nucleotide exchange by Gu,,, the binding of [YS]GTPyS to receptorG,,,, vesicles was first measured in the absence or presence of 25 nM PLC-51 (Figure 5). Binding was monophasic. The rate of binding, which gives an estimate of the rate of GDP-GTP exchange, was essentially the same in the absence and presence of PLC-pl (kobs = 0.35 min’). The absolute amount of [35S]GTPyS that bound to the vesicles was somewhat higher (1 .&fold f O.Bfold; n = 12) in the presence of the enzyme. We do not know the reason for this difference, which might result from stabilization of G,,,, by PLC-91. The extent of [?S]GTPyS binding indicates the concentration of active GwyrI that can be regulated by the Ml receptor (see Ross, 1991) and the ratio of the steady-state GTPase rate to this value yields a molar turnover number for the GTPase. For the reconstituted Ml muscarinic receptor-Gq/,, system, PLC-pl increased the GTPase turnover number from 0.14 -c 0.04 min’ to 1.3 +

0.3

min-’

(n =

12)-roughly

a lo-fold

stimulation.

The molar turnover number for steady-state GTP hydrolysis must be slower than the rate constants for both GTPGDP exchange and hydrolysis of bound GTP. This is true in the absence of PLC-91. In the presence of PLC-pl, however, the observed steady-state turnover number was at least 50% faster than the rate constants for GTP binding and GDP release that we had previously determined in the absence of PLC-91. We therefore compared the rates and extents of carbachol-stimulated binding of [yJ2P]GTP and [a-32P]GTP in the presence and absence of PLC-51. [yJ2P]GTP binding measuresonly bound GTP; [a-32P]GTP binding measures bound GTP plus bound GDP (see Berstein et al., 1992). The addition of PLC-91 had no effect on the rates of binding of either nucleotide. However, PLC-51 reproducibly decreased the steady-state binding of both nucleotides by 30% to 50% (data not shown). These observations, taken together, suggest that PLC-91 creates a pool of GUI1 that both hydrolyzes bound GTP rapidly and releases bound GDP too fast to be measured manually.

The

are discussed

mechanistic

implications

of such

an effect

below.

Discussion The data presented here indicate that PLC-f31 stimulates the hydrolysis of GTP that is bound to G9/,,, its physiological regulator. Because GTP binding activates G proteins such that they can stimulate their effector proteins, the GAP activity of PLC-91 implies that it accelerates the termination of its own activation. Such an effect at first appears to be paradoxical, but the kinetics of G protein-mediated signaling in cells strongly predicts the existence of some

Phospholipase 415

C-81 Is a GAP for G,

mechanism to accelerate GTP hydrolysis on trimeric G proteins. Although the rate constant for hydrolysis of bound GTP by purified G proteins usually varies from 0.8 min-’ to4 min-I in vitro (Gilman, 1987), termination of cellular signaling is almost always much faster. For example, GI hydrolyzes GTP at only 2 min-’ (Higashijima et al., 1987). However, activation of the M-type K’ channel, which is regulated by G, (Yatani et al., 1988) terminates with a rate constant of 135 min-l upon removal of agonist (Breitwieser and Szabo, 1988). The photoresponse in retinal rod outer segments is also terminated on the 100 ms time scale upon removal of light, but transducin (G,) hydrolyzes GTP as slowly as does G, (see Vuong and Chabre, 1991). Thus, while there are many mechanisms for desensitizing G protein-mediated signaling at the levels of receptor and effector, rapid termination of the activated state of the G protein is clearly required. Because there are no previous reports that a G proteinregulated effector protein can act as a GAP, it is important to demonstrate clearly that the source of the GAP activity is the PLC-81 itself. Both PLC and GAP activities are expressed in the same ratio in highly purified preparations of both recombinant PLC-81 expressed in HeLa cells and PLC-81 from bovine brain. Monoclonal antibodies raised against the bovine brain PLC-81 block both the PLC-81 stimulating activity of Gq/,, (Figure 3) and the Gwl, GAP activity of PLC-81 with the same rank order of selectivity. They do not influence the receptor-stimulated GTPase activity of Gqtr in the absence of PLC-81. Last, PLC-81 does not display GAP activity for either G, or G,, G proteins that do not activate PLC-81 (Table 1). We speculate that many G protein-regulated effecters (channels, adenylylcyclases, cyclic GMP phosphodiesterase) may also be GAPS for their respective G proteins. No GAP activity was noted for adenylylcyclase in early reconstitution experiments (May et al., 1985) but retrospective review of those data indicate that low efficiency reconstitution and suboptimal cyclase:G, stoichiometry may have prevented us from observing GTPase stimulation by the cyclase. These experiments will be repeated shortly. Cyclic GMP phosphodiesterase was recently proposed to act as a GAP, but the supporting data were indirect (Arshavsky et al., 1991). The components of the rhodopsin-Gt system are abundant and easy to purify, so that this question should be addressed quickly. It is an attractive mechanism to use an effector rather than a separate protein as a G protein GAP, both because it is parsimonious and because it is inherently specific to a defined signaling pathway. While many G proteinmediated pathways require the existence of a GAP to provide physiologically appropriate speed of deactivation, others might not. Variation among effecters in their ability to stimulate GTP hydrolysis could allow different effecters under the control of a single G protein to display distinct temporal patterns of regulation. An effector that is a strong GAP will transmit a signal that ends abruptly when hormone is removed, but an effector with little or no GAP activity will generate a more prolonged signal. For example, adenylylcyclase deactivates relatively slowly (Cassel et al., 1977) and might thus be a weak GAP for G., its

regulator. The G.-stimulated Ca*+ channel, which displays rapid regulatory kinetics (Yatani and Brown, 1989) would be predicted to be a highly active G. GAP. Using effecters as GAPS can also allow effector-specific modulation of signaling. Feedback regulation of GAP activity could be exerted by either covalent or allosteric regulation. Such regulation is implicit in the observation by Arshavsky et al. (1991) that the second messenger cyclic GMP inhibits a brief period of rapid GTP hydrolysis by transducin in rod outer segments. It will be important to test the ability of either phosphatidylinositol-4,5-diphosphate or inositol-1,4,5-triphosphate to inhibit the GAP activity of PLC-81. An effect of inositol phospholipids on the GAP activity of PLC-81 would be particularly provocative because of the reports that anionic lipids inhibit the regulatory activity of rasGAP (Tsai et al., 1989a, 1989b). It is tempting to draw an analogy between the GAP activity of PLC-81 on its regulator, GqIl, and the still uncertain functions of the GAPS for low molecular weight GTPbinding proteins. When rasGAP was first identified, McCormick and coworkers noted that mutations in a region of the ras sequence that destroyed its signaling capacity (the so-called effector-binding domain) also destroyed its response to rasGAP (Adari et al., 1988; Sollag and McCormick, 1991 for review). This observation prompted the suggestion that rasGAP is the downstream effector molecule that is regulated by ras. However, other experiments that indicated that overexpression of GAP antagonizes the ras signal were interpreted to mean that GAP is an inhibitory regulator that simply promotes ras deactivation. Further experiments with rasGAP and other GAPS remain inconclusive, with the primary difficulty arising from the fact that no ras-regulated effector has been identified and no alternative activity for rasGAP has been demonstrated. Because it is clear that PLC-81 is both effector and GAP in the mammalian Gq/,, pathway, the idea that GAPS for small GTP-binding proteins are also their effecters deserve more direct biochemical investigation. The structural heterogeneity of G protein-coupled effectors is remarkable. PLC-81 is a soluble monomer, channels and adenylylcyclases are multispan integral membrane proteins of two distinctly different topologies, and phosphodiesterase is aweakly membrane-bound a8Y2tetramer. The effecters show no detectable sequence similarity; thus, no assignment a priori of a distinct GAP domain is feasible. The GAPS for small GTP-binding proteins are alsostrikingly heterogeneous (Hall, 1992) suggesting that GAP activity may have evolved after the effector function of these proteins was established. However, it should be possible to use monoclonal antibodies (Figure 3) and peptide fragments of PLC-81 to lead us to the regions in the C-terminal domain of PLC-81 that mediate its GAP activity. Klnetlc Consequences of the GAP Actlvlty of PLCgl Signaling via G proteins depends upon regulating the steady-state concentration of the activated G protein-GTP complex. Activation reflects receptor-catalyzed release of GDP and binding of GTP and deactivation reflects hydrolysis of bound GTP to GDP. Receptor amplifies the hormonal signal by sequentially promoting the activation of multiple

Cell 416

G protein molecules, and the extent of amplification is a measure of the number of G protein molecules that can be activated by one receptor during the lifetime of the activated G protein-GTP complex. Our data indicate that PLC-31 stimulates the rate of hydrolysis of Go,,-bound GTP by at least 50-fold but has no effect on the rate of GTP-GDP exchange. This GAP activity allows appropriately rapid deactivation upon removal of hormone and may, as discussed below, enhance the specific control of G protein-mediatedsignaling through individual effectors. At first glance, there is an inconsistency between the observed rate of steady-state GTP hydrolysis and the rates of receptor-catalyzed GTP-GDP exchange. Under maximal stimulation by receptor and PLC-31, the turnover number for Gq/,, at steady state is greater than the nucleotide exchange rate constant (1.1 mol of GTP hydrolyzed per min per mol of Gqlfl, determined by receptor-stimulated GTPyS binding, compared with 0.35 per min; compare Figures 1 and 4). The overall rate of a reaction cannot be greater than that of any partial reaction. Consequently, if both values are correct, PLC-31 must elicit formation of a pool of Gq/,,, presumably a Gql,-PLC-31 complex, that turns over GTP very quickly but that is not detected in conventional GTP binding assays. We propose the following mechanism to explain these observations. When receptor catalyzes GDP release and GTP binding to GwrI, GTP-liganded G@,, binds to PLC-31. If PLC-pi-promoted GTP hydrolysis is fast, receptor will not dissociate from the Gqr,-PLC-31 complex during the lifetime of the bound GTP and will rapidly displace bound GDP, thereby reinitiating the cycle. It is the retention of bound receptor that allows rapid steady-state turnover at a rate that would be impossibly rapid if a new receptor had to bind Gq/,, via lateral diffusion prior to each nucleotide exchange event. The proposed PLC-91-G~Ir complex is likely to be only moderately stable. It should survive on the time scale of the GTPase, though, because the negatively cooperative interaction between GTP and receptor is much less than that between receptor and GDP (perhaps by 20-fold based on GDP-GTP competition; Florio and Sternweis, 1999; Berstein et al., 1992). Note that by maintaining association of receptor with Gqlll and PLC-91, this mechanism permits rapid deactivation of effector upon removal of agonist but does not decrease the steady-state signaling activity as much as would be predicted if the only effect of PLC-31 were to decrease the lifetime of activated molecules of Gq/,,. The existence of a receptor-GM,,-PLC-31 complex is consistent with two observations of this study. First, a distinct and relatively small pool of GM,,-bound GTP (200%~ 40%) displayed the PLC-f31-induced burst of rapid hydrolysis (Figure 3). Second, the addition of PLC-31 to the receptor-Gg/,, vesicles decreased the maximal binding of GTP or GDP, but the rate of the remaining GTP and GDP binding was unaltered. PLC-31 also did not change the rate of GTPyS binding and increased maximal GTPyS binding only slightly. These two populations of GM,, thus have characteristics expected for the putative receptor-Gu,,-PLC01 complex. In addition, the magnitude of the burst is roughly equal to the difference between maximal GTPyS

binding and the amount of GTP plus GDP bound at steady state, also consistent with the model. Additional experiments are needed to explain how the relative activation of PLC-91 and the GTPase activity of G9/,, are maintained at steady state. Quenched flow or low temperature assays will be needed to evaluate this or any other kinetic model. It is particularly important to measure both the PLC-plstimulated k, for hydrolysis and to reevaluate the stimulation of PLC activity in the presence of GTP, which was not measurable in our first study of this system (Serstein et al., 1992). We report here that PLC-91 serves as a GAP for its G protein regulator, Gull. This is a new mode of regulation in G protein pathways that provides a novel mechanism for enhanced temporal control of individual effecters without introducing new protein components. These findings also suggest the presence of undiscovered feedback controls to allow cellular adaptation to incoming signals. Experimental

Procedures

Protein Purltlcation Human Ml and M2 muscarinic cholinergic receptors were produced in Sf9 cells and purified as described by Parker et al. (1991) and Haga and Haga (1985). Q-Adrenergic receptor from turkey erythrocytes were purified as described by Brandt and Ross (1986). An approximately equimolar mixture of bovine hepatic trimeric G, and G,,, referred to as G,,,, was purified as described by Blank et al. (1991). Both G, and GI1 are regulated by Ml muscarinic receptors with about equal efficiency (Berstein et al., 1992). G. and Go were prepared as described by Sternweis et al. (1981) and Higashijima et al. (1990). Bovine cerebral PLC-f31 was prepared as described by Lee et al. (1987) and Ryu et al. (1987a, 1987b). Recombinant PLC-(31 was produced in HeLa cells and purified as described by Park et al. (1992). Monoclonal antibodies against PLC-51 were raised and purified from ascites fluid asdescribed by Suh et al. (1988). Molar concentrations of proteins were estimated as follows: muscarinic receptors by the binding of PHlquinuclidinyl benrilate (Berstein et al., 1992); 5-adrenergic receptors by the binding of PHldihydroalprenolol (Parker et al., 1991); G. and G, bythe Mg%timulated binding of [%]GTPyS (Berstein et al., 1992); GWI1 according to protein-bound GDP (Berstein et al., 1992); and PLC-(+l according to the amount of total protein, which was determined according to Ryu et al. (1987a, 1987b) and calculated using a molecular weight of 150,000. The concentration of Gql,, accessible to receptors after reconstitution, coupled G@,,, was defined as the total amount of carbachol-stimulated PS]GTPyS binding (Berstein et al., 1992) which is altered slightly by the addition of PLC-51 (see Figure 5).

= E = P ma m a* z 6 z-

25

.

20

.

15 10 5 0 /_ 0

2

4 6 Tcme(mm)

6

10

Figure 5. PLC-Bl Does Not Significantly Alter GTPyS Binding to Ml Receptor-GrtI Vesicles P%]GTPyS binding was measured as described in the absence (open circles) or presence (closed circles) of 26 nM PLC-51. Parallel assays indicated that steady-state GTP hydrolysis was 3 and 35 fmol of GTP hydrolyzed per mln for the same volume of vesicles without and with PLC-fll, respectively. Vesicles contained 8.6 fmol of receptor.

Phospholipase 417

C-51 Is a GAP for G,

Reconetltutlon Receptor and G protein were coreconstituted into phospholipid vesicles by gel filtration of a mixture of proteins and lipids in detergent solution, as described previously (Berstein et al., 1992). Reconstituted vesicles were incubated with 5 mM dithiothreitol (DTT) at 0°C for 1 hr to activate receptors (Pedersen and Ross, 1982). PLC-81 or its buffer (20 mM Tris-HCI [pH 7.5),0.1 mM DlT, 1 mM EGTA, 0.4 M NaCI) were added prior to assay. Other details have been described previously (Berstein et al., 1992).

Aeuy

of Steady-St&ate

GTP Hydrolysis

Reconstituted vesicles mixed either with PLC-81 or PLC buffer (10 ul) were added to an assay cocktail (20 ul) to give the following final concentrations: 20 mM HEPES (pH 8.0) 100 mM NaCI, 4 mM MgCII, 1 mM EDTA, 0.1 mM EGTA, 1 mg/ml bovine serum albumin, 150 nM [T-~P]GTP (150-500 cpm/fmol), 500 nM GDP, 1 mM ATPTS, and muscarinic ligands as indicated in the legends. Reaction mixtures were incubated at 30°C for 15 min unless otherwise indicated. GTP hydroly sis was measured by charcoal precipitation as described previously (Brandt and Ross, 1985). ATPyS was included in the reaction mixture to inhibit a nonspecific nucleoside triphosphatase activity present in the purified recombinant PLC-51 preparation (4.4 pmol of GTP hy drolyzed per min per mg of protein). GDP was included to inhibit GTP binding activity present in the preparation of recombinant PLC-51 (0.02 mol of GTP bound per mol of PLC-81).

Assay

of Hydrolysis

of GTP Sound

to Cg,,

Reconstituted vesicles (30 ul) were incubated at 30% in a 33 ul total volume in the presence of 20 mM HEPES (pH 8.0). 100 mM NaCI, 4 mM MgClt, 1 mM EDTA, 1 mM EGTA, 0.85 mg/ml bovine serum albumin, 50 nM [y-SP]GTP (900 cpm/fmol), and 100 mM carbachol. After 1 min, 2 pl of atropine (to inhibit receptor-mediated GTP release) plus GTP (to limit hydrolysis to a single turnover) were added to give final concentrations of 100 PM and 50 uM. respectively. Four seconds later, defined as experimental zero time, purified recombinant PLC-51 (3.3 or 6.6 pmol in 0.5 or 1 .O ul) or 0.5 ul of PLC buffer was added. Subsequent GTP hydrolysis was determined as described above. Data shown have been corrected for zero-time background. The sources of reagents were reported by Berstein et al. (1992).

Bollag, G., and McCormick, F. (1991). Regulators proteins. Annu. Rev. Cell Biol. 7, 601-832. Bourne, H. R., Senders, D.A., end McCormick, superfamily. I. A conserved switch for diverse 348, 125-I 32.

and effecters

of ms

F. (1990). TheGTPase cell functions. Nature

Bourne, H. R., Sanders, D.A., and McCormick, F. (1991). TheGTPase superfamily. II. Conserved structure and molecular mechanism. Nature 349, 117-127. Brandt, D. R., and Ross, E. M. (1985). GTPase activity of the stimulatory GTP-binding regulatory protein of adenylate cyclase, G.. Accumulation and turnover of enzyme-nucleotide intermedietes. J. Biol. Chem. 280.268-272. Brandt, D. R., and Ross, E. M. (1986). Catecholamine-stimulated GTPase cycle: multiple sites of regulation by 8.adrenergic receptor and Mg” studied in reconstituted receptor-G. vesicles. J. Biol. Chem. 281, 1656-I 864. Breitwieser, G. E., and Szabo, G. (1988). Mechanism of muscarinic receptor-induced K+ channel activation as revealed by hydrolysisresistant GTP analogues. J. Gen. Physiol. 91, 469-493. Gales, C., Hancock, J. F., Marshall, C. J., and Hall, A. (1988). The cytoplasmic protein GAP is implicated as the terget for regulation by the fas gene product. Nature 332, 548-551. Casey, P. J., Fong, H. K. W., Simon, M. I., and Gilman, A. G. (1990). G,, a guenine nucleotide-binding protein with unique biochemical properties. J. Biol. Chem. 285, 2383-2390. Cassel, D., Levkovitz, GTPase cycle of turkey otide Res. 3, 393-406.

H., and Selinger, 2. (1977). erythrocyte adenylate cyclase.

The regulatory J. Cyclic Nucle-

Dratz, E. A., Lewis, J. W., Schaechter, L. E., Parker, K. R., end Kliger, D. S. (1987). Retinal rod GTPase turnover rate increases with concentration: a key to the control of visuel excitation? Biochem. Biophys. Res. Commun. 748, 379-386. Florio, V. A., and Sternweis, P. C. (1989). Mechanisms of muscarinic receptor action on G. in reconstituted phospholipid vesicles. J. Biol. Chem. 284,3909-3915. Gilman, signals.

A. G. (1987). G proteins: transducers Annu. Rev. Biochem. 58, 615-649.

of receptor-generated

Acknowledgments

Haga, K., and Haga, T. (1986). Purification ofthe muscarinicecetylcholine receptor from porcine brain. J. Biol. Chem. 280, 7927-7935.

This work was supported by Fogarty International Postdoctoral Fellowship TWO4475 to G. B. and by research grants from the National Institutes of Health (GM30355) and the R. A. Welch Foundation (I-092) to E M. R. We appreciate the suggestions of Tsutomu Higashijima and the unfettered criticism of this manuscript by Melanie Cobb. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “edvertisement” in eccordance with 18 USC Section 1734 solely to indicate this fact.

Hall, A. (1990). 921-923.

Received

June

8, 1992

Refersnces Adari, H., Lowy, D. R., Willumsen, B. M., Der, C. J., and McCormick, F. (1988). Guanosine triphosphatase activating protein (GAP) interacts with the p21 ras effector binding domain. Science 240, 518-521. Arshavsky, V., Gray-Keller, M., and Bownds. M. (1991). cGMP suppresses GTPase activity of a portion of transducin equimolar to phosphodiesterase in frog rod outer segments, J. Biol. Chem. 288,1853018537. Berstein, G., Blank, J. L., Smrcka, A. V., Higashijima, T., Sternweis, P. C., Exton, J. H., and Ross, E. M. (1992). Reconstitution of agoniststimulated phosphatidylinositol4,5-bisphosphate hydrolysis using purified ml muscarinic receptor, G qlr and phospholipase C-81. J. Biol. Chem. 267.8081~8068. Blank, J. L., Ross, A. H., and Exton, J. H. (1991). Purification and characterization of two G-proteins that activate the 81 isozyme of phosphoinositide-specific phospholipase C. J. Biol. Chem. 286, 18206-18216.

ras and

GAP-who’s

Hall, A. (1992). Signal transduction Iwo GAPS. Cell 89,389-391.

controlling through

whom?

small GTPases-a

Cell

81,

tale of

Higashijima, T., Ferguson, K. M., Smigel, M. D., and Gilman, A. G. (1987). The effect of GTP and Mg” on the GTPase activity and the fluorescent properties of G.. J. &of Chem. 282, 757-761, Higashijima, T., Burnier, J., and Ross, E. M. (1990). Regulation of G, and G, by mastoparen, related amphiphilic peptides and hydrophobic amines: mechanism and structural determinants of activity. J. Biol. Chem. 285, 14176-14186. Kaziro, Y., Itoh, H., Kozasa, T., Nakafuku, Structure and function of signal-transducing Annu. Rev. Biochem. 80, 349-400.

M., and Satoh, GTP-binding

T. (1991). proteins,

Lee, K.-Y., Ryu, S. H., Suh, P.-G., Choi, W. C., and Rhee, S. G. (1987). Phospholipase C associated with particulate fractions of bovine brain, Proc. Natl. Acad. Sci. USA 84, 5540-5544. May, D. C., Ross, E. M., Gilman, A. G., and Smigel, M. D. (1985). Reconstitution of catecholaminestimulated adenylate cyclase activity using three purified proteins. J. Biol. Chem. 280, 1582915833. McCormick, F. (1989). ras GTPase activating ter and signal terminator. Cell 58, 5-8.

protein:

signal transmit-

Navon, S. E., and Fung, B. K.-K. (1984). Characterization of trensducin from bovine retinal rod outer segments: mechanism and effects of cholera toxin-catalyzed ADP-ribosylation. J. Biol. Chem. 259, 68866693. Park, D., Jhon, D.-Y., Kriz, R., Knopf, J., and Rhee, Cloning, sequencing, expression, and G,independent phospholipase C-82. J. Biol. Chem., in press.

S. G. (1992). activation of

Cell 418

Parker, E. M., Kameyama, K., Higashijima, T.. and Ross, E. M. (1991). Reconstitutively active G protein-coupled receptors purified from baculovirus-infected insect cells. J. Biol. Chem. 266, 519-527. Pedersen, S. E., and Ross, E. M. (1982). Functional reconstitution of f3-adrenergic receptors and the stimulatory GTP-binding protein of adenylate cyclase. Proc. Natl. Acad. Sci. USA 79, 7228-7232. Ross, E. M. (1989). Signal sorting and amplification through G proteincoupled receptors. Neuron 3, 141-152. Ross, E. M. (1991). Reconstitution of the 8-adrenergic receptor and its biochemical function. In The f3-Adrenergic Receptor, J. P. Perkins, ed. (Clifton, New Jersey: Humana Press), pp. 125-179. Ryu, S. H., Cho, K. S., Lee, K.-Y., Suh, P.-G., and Rhee, S. G. (1987a). Purification and characterization of two immunologically distinct phosphoinositide-specific phospholipase C from bovine brain. J. Biol. Chem. 262, 12511-12518. Ryu, S. H., Suh, P.-G., Cho, K. S., Lee, K.-Y., and Rhee, S. G. (1987b). Bovine brain cytosol contains three immunologically distinct forms of inositolphospholipid-specific phospholipase C. Proc. Natl. Acad. Sci. USA 84, 6849-8853. Smrcka, A. V., Hepler, J. R.. Brown, K. O., andsternweis, P. C. (1991). Regulation of polyphosphoinositide-specific phospholipase C activity by purified G,. Science 257, 804-807. Sternweis, P. C., Northup, J. K., Smigel, M. D., and Gilman, A. G. (1981). The regulatory component of adenylate cyclase: purification and properties. J. Biol. Chem. 256, 11517-I 1528. Strathmann, M., and Simon, M. I. (1990). G protein diversity: a distinct class of a subunits is present in vertebrates and invertebrates, Proc. Natl. Acad. Sci. USA 67, 9113-9117. Suh, P.-G., Ryu, S. H., Choi, W. C., Lee, K.-Y., and Rhee, S. G. (1988). Monoclonalantibodies to threephospholipasec isozymesfrom bovine brain. J. Biol. Chem. 263, 14497-14504. Tanaka, K., Nakafuku, M., Satoh, T., Marshall, M. S., Gibbs, J. B., Matsumoto, K.. Kaziro, Y., and Toh-e, A. (1990). S. cerevisiae genes IRA 1 and IRA2 encode proteins that may be functionally equivalent to mammalian ras GTPase activating protein. Cell 60, 803-807. Trahey, M., and McCormick, F. (1987). A cytoplasmic protein stimulates normal N-m p21 GTPase, but does not affect oncogenic mutants. Science 236, 542-545. Tsai, M-H., Hall, A., and Stacey, D. W. (1989a). Inhibition by phospholipids of the interaction between N-ras, rho, and their GTPase activating proteins. Mol. Cell. Biol. 9, 5280-5364. Tsai. M-H.,Yu,C.-L., Wei, F.-S.,andStacey, D. W.(1989b).Theeffect of GTPase activating protein upon ras is inhibited by mitogenically responsive lipids. Science 243, 522-525. Vuong. T. M., and Chabre, M. (1990). Subsecond deactivation of transducin by endogenous GTP hydrolysis. Nature 346, 71-74. Vuong, T. M., and Chabre, M. (1991). Deactivation kinetics of the transduction cascade of vision. Proc. Natl. Acad. Sci. USA 66, 98139817. Yatani, A., and Brown, A. M. (1989). Rapid 8-adrenergic modulation of cardiac calcium channel currents by a fast G protein pathway. Science 245, 71-74. Yatani, A., lyengar, R., gated atrial Nature 336. Note Added

Mattera, R.. Codina, J., Graf, R., Okabe, K., Padrell, E., Brown, A. M., and Birnbaumer, L. (1988). The G proteinK+ channel is stimulated by three distinct Gla subunits. 680-682. In Proof

Arshavsky and Bownds (Nature 357,418-417) have reported that purified cyclic GMP phosphodiesterase (PDE) stimulated GTPase activity in bleached toad photoreceptor membranes that had been depleted of endogenous PDE. Stimulation appeared to be about 4- to Sfold and with an ECm of about 40 pM.