gtp -activating proteins for heterotrimeric … 2000 annurev.pdf · gtp hydrolysis rates prompted...

P1: FPP/FPW P2: FPP

July 29, 2000 15:11 AR102 CHAP26

Annu. Rev. Biochem. 2000. 69:795–827Copyright c© 2000 by Annual Reviews. All rights reserved

GTPASE-ACTIVATING PROTEINS FOR

HETEROTRIMERIC G PROTEINS: Regulatorsof G Protein Signaling (RGS)and RGS-Like Proteins

Elliott M. Ross and Thomas M. WilkieDepartment of Pharmacology, University of Texas Southwestern Medical Center, Dallas,Texas 75390-9041; e-mail: [email protected], [email protected]

Key Words signal transduction, GAP, GTP-binding protein, scaffold

This review is dedicated to the memory of Eva Neer, whose research andcollegial behavior were an inspiration to the G protein community.

■ Abstract GTPase-activating proteins (GAPs) regulate heterotrimeric G proteinsby increasing the rates at which theirα subunits hydrolyze bound GTP and thus returnto the inactive state. G protein GAPs act allosterically on Gα subunits, in contrastto GAPs for the Ras-like monomeric GTP-binding proteins. Although they do notcontribute directly to the chemistry of GTP hydrolysis, G protein GAPs can acceleratehydrolysis>2000-fold. G protein GAPs include both effector proteins (phospholipaseC-β, p115RhoGEF) and a growing family of regulators of G protein signaling (RGSproteins) that are found throughout the animal and fungal kingdoms. GAP activity cansharpen the termination of a signal upon removal of stimulus, attenuate a signal eitheras a feedback inhibitor or in response to a second input, promote regulatory associationof other proteins, or redirect signaling within a G protein signaling network. GAPs areregulated by various controls of their cellular concentrations, by complex interactionswith Gβγ or with Gβ5 through an endogenous Gγ -like domain, and by interactionwith multiple other proteins.

CONTENTS

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 796STRUCTURE AND PHYLOGENETIC RELATIONSHIPS

OF RGS PROTEINS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 799The RGS Domain. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 799Conserved Domains in RGS-Flanking Regions. . . . . . . . . . . . . . . . . . . . . . . . . . 803

MECHANISM AND CONSEQUENCES OF G PROTEIN GAP ACTIVITY. . . . . . 803

0066-4154/00/0707-0795/$14.00 795

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

796 ROSS ¥ WILKIE

Mechanism of G-Protein GAP Activity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 804Regulatory Consequences of GAP Action. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 806

G PROTEIN EFFECTORS AS GAPS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 809PLC-β . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 809Rho GEFs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 810Cyclic GMP Phosphodiesterase. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 811Why Should Effector Proteins Be GAPs?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 811

GAPS AND SIGNAL PATHWAY SPECIFICITY. . . . . . . . . . . . . . . . . . . . . . . . . . 812Selectivity In Vitro . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 812RGS-Gα Selectivity in Cells. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 813Receptor-Selective Activities of RGS Proteins. . . . . . . . . . . . . . . . . . . . . . . . . . 814

REGULATION OF RGS PROTEINS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 814Regulation of RGS Protein Synthesis and Accumulation. . . . . . . . . . . . . . . . . . . 814Covalent Modification of RGS Proteins and Their Gα Targets. . . . . . . . . . . . . . . 815Cellular Binding Partners of RGS Proteins. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 816

REGULATION BY Gβγ SUBUNITS AND A GGL DOMAIN . . . . . . . . . . . . . . . 817Gβγ Inhibits GAP Activity in Single-Turnover Assays. . . . . . . . . . . . . . . . . . . . 818The GGL Domain: Selective Association of RGS Proteins with Gβ5 . . . . . . . . . . 818A Model for Gβ-RGS Interactions: Evidence from Gβγ -Gated K+ Channels. . . . 819

CELLULAR REGULATION BY RGS PROTEINS: AC. ElegansModel . . . . . . . . 821

INTRODUCTION

G protein–mediated signaling is intrinsically kinetic. Signal amplitude is deter-mined by the balance of the rates of GDP/GTP exchange (activation) and of therates of GTP hydrolysis (deactivation); this balance determines the amount of Gprotein in the activated GTP-bound state. The activation limb of the GTPase cycleis accelerated by G protein–coupled receptors (GPCRs); the deactivation limb isaccelerated by GTPase-activating proteins (GAPs). GAPs for heterotrimeric G pro-teins include two G protein–regulated effectors, the Gq-stimulated phospholipaseC-β (PLC-β) and the G13-stimulated p115RhoGEF. The recently identified regula-tors of G protein signaling (RGS) proteins, which are found throughout the highereukaryotes, are also G protein GAPs. This review focuses on their mechanisms ofaction and roles in signal modulation.

GTP hydrolysis was initially thought to be an unregulated function of Gα

subunits that provides an intrinsic control over the activation lifetime of a G protein.This idea was consistent with the first report of the rate of hydrolysis of GTP boundto Gs(1), which corresponded well to the rate of deactivation of adenylyl cyclase ina plasma membrane (2). Both events displayed half-lives of∼15 s. However, otherG proteins that mediate responses with rapid recovery rates (t1/2< 1 s) displayedsimilarly slow hydrolysis rates in vitro (3). Vision presented a particularly obviousinconsistency between rates of hydrolysis and deactivation (3, 4); the half-time forhydrolysis of GTP bound to isolated Gt is∼15 s, but light responses terminate muchmore quickly (∼100 ms). Several studies indicated that the rate of Gαt-catalyzedGTP hydrolysis in intact photoreceptor cells is as fast as physiologic deactivation,

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

GTPASE-ACTIVATING PROTEINS 797

which suggests that a GTPase-accelerating factor was lost upon purification of Gt(5–8).

The discrepancy between fast physiologic deactivation rates and slow in vitroGTP hydrolysis rates prompted the first searches for proteins that would acceler-ate hydrolysis of Gα-bound GTP. Effort was further stimulated by the discoveryof a GAP for the monomeric GTP-binding protein p21ras (9, 10). Two G pro-tein GAP activities were finally identified in 1992. Berstein et al (11) found thatPLC-β, the principal effector protein regulated by Gq, also has selective Gq GAPactivity in that it could accelerate GTP hydrolysis by Gq≥ 50-fold. Arshavsky &Bownds (12) found that the addition of the regulatoryγ subunit of cyclic GMPphosphodiesterase (PDE) to washed photoreceptor membranes accelerated thehydrolysis of GTP by Gt (transducin) 4–5-fold. Gt is the proximal activator ofthe PDE in photoreceptor cells, and PDE thus appeared to be another effectorwith intrinsic GAP activity. PDEγ was subsequently shown not to be a Gt GAPitself, but it does potentiate the GAP activity of RGS9 (13). More recently, aRhoGEF was also shown to be both an effector for G13 and a G13/12-selective GAP(14, 15).

RGS proteins were independently identified by three groups. Siderovski andcoworkers (16) first identified this gene family, but they were initially misled aboutits function because of an apparent basic helix-loop-helix motif in the prototype,GOS8 (renamed RGS2). However, they did notice similarity to Sst2p, a feedbackinhibitor of a Gα subunit that mediates the mating response in yeast (see 17 forreview). De Vries et al (18) then showed that GAIP (Gα-interacting protein), arelated protein, binds to Gα subunits. The reality of RGS proteins as G proteinregulators gained further support when Koelle & Horvitz (19) reported that Egl-10is an inhibitor of Go in Caenorhabditis elegansand is related to Sst2p, GOS8,GAIP, and several mammalian expressed sequence tags (ESTs). Shortly thereafter,Berman et al (20) and others (21–23) showed that RGS proteins display GAPactivity toward members of the Gαi and Gαq families.

At this time,>20 mammalian RGS proteins have been identified, as have severalmore distantly related protein families (Figure 1; see below). Most RGS proteinsand several of their relatives display G protein GAP activity. The physiologicalfunctions of G protein GAP activity is known in only a few cases: feedback inhibi-tion of the mating pheromone response in yeast, acceleration of termination of thephotoresponse in the eye, and inhibitory signaling in neurons ofC. elegans. Thediscovery of new and different functions can be expected. GAPs have the potentialto inhibit signaling by depleting the GTP-activated form of Gα, to change signalingkinetics, or to alter signal specificity. RGS proteins also interact with a variety ofother proteins and lipids, and may thus exert both positive and negative regulatoryfunctions distinct from, or in addition to, their GAP activity. This review aims toconvey what is known about the action of RGS proteins and other G protein GAPs,as well as to point out possible new functions of this recently recognized class ofproteins. Because G protein signaling is found throughout the eukaryotes, interestin RGS proteins extends from basic biochemistry through whole-animal biology.Other reviews with multiple viewpoints are also available (24–32).

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

798 ROSS ¥ WILKIE

Figure 1 Mammalian RGS proteins.Left: A cladogram constructed from amino acidsequence identities within the RGS domain defines five subfamilies RZ, R4, R7, R12, andRA (see Figure 2). The scale shows approximate amino acid identity calculated as 100%minus the sum of the horizontal distance to and from the common branch point (e.g. axinand conductin are 64% identical). All sequences are human except for RGS8 (rat), RGSZ2(mouse), and RET-RGS1 (bovine). Ortholog/paralog relationships remain uncertain forRET-RGS1 and RGSZ1; the only cloned RET-RGS1 cDNA is bovine and the onlyRGSZ1 cDNAs are murine and human. Accession numbers: GAIP (AAC62919); RGSZ1(AAC62013); RGSZ2 (AF191555); RET-RGS1 (AAC48721); RGS1 (Q08116); RGS2(P41220); RGS3 (P49796); RGS4 (P49798); RGS5 (O15539); RGS6 (P49758); RGS7(P49802); RGS8 (P49804); RGS9 (AAC64040); RGS10 (O43665); RGS11 (AAC69175);RGS12 (AAC39835); RGS13 (O14921); RGS14 (O43566); RGS16 (AAC39642); axin(AAC51624); conductin (AAD20976).Right: Most proteins within each subfamily are alsohomologous in regions flanking the RGS domain; homologies include definable functionaldomains shown as labeled blocks on the diagrams of each protein’s structure. Functions ofthese domains, where known, are discussed in the text. Abbreviations: APC, adenomatouspolyposis coli; GGL, Gγ -like; DEP, PDZ, and PTB, protein interaction domains; PP2A,protein phosphatase 2A; GSK.

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


STRUCTURE AND PHYLOGENETIC RELATIONSHIPSOF RGS PROTEINS

The RGS proteins are related by a conserved RGS domain that is∼130 amino acidresidues in length (Figure 1). The RGS domain alone is capable of binding Gα

subunits and accelerating GTP hydrolysis (33, 34), although other domains maycontribute to affinity and/or selectivity for G protein targets (33, 35). RGS domainshave been found in fungi,Dictyostelium discoideum,and animals. They are yetto be identified in higher plants, although plants express both heterotrimeric Gproteins and GPCRs.

The mammalian RGS proteins, of which>20 are now known, can be groupedinto five subfamilies based on sequence similarity within the RGS domain (Fig-ure 2; see 29 for an alternative grouping). Regions outside the RGS domain alsoshow similarity within each subfamily. RGS proteins in the RZ, R4, R7, andR12 subfamilies all display GAP activity. We group the RA subfamily (axin andconductin) with the RGS proteins because of sequence similarity, although theseproteins have not been shown to have GAP activity or to bind Gα subunits. Werefer to proteins with domains more distantly related to the RGS domain as RGS-like (or RGL) proteins. These include the GPCR kinases (GRKs), some of whichdisplay GAP activity (36, 37); D-AKAP2, neither of whose RGL domains haveyet been shown to bind to a Gα subunit 38 (RK Sunahara & SS Taylor, manuscriptin preparation); and the p115RhoGEFs, which are both stimulated by Gα13 andare G13/12GAPs (14, 15). The C-terminal GAP domain of PLC-β (40) is the mostdistantly related to the RGS domain, if at all, but we include PLC-β in the RGLproteins because of its GAP activity. Sequence similarity among the RGL domainsis listed in the legend to Figure 2; other sequence alignments have been proposedelsewhere (15, 36, 38). It seems most likely that the RGL proteins evolved from anancestor common to the RGS proteins, although convergent evolution from distinctgene lineages remains plausible for the RhoGEFs and likely for the PLC-βs.

The RGS Domain

The structure of the RGS domain was defined by X-ray crystallographic analysis ofa complex of RGS4 and Gαi1 by Tesmer et al (41). Although the crystals containedboth intact proteins, only the 128-amino-acid RGS domain was resolved withinRGS4; the N- and C-terminal regions (50 and 27 residues) were disordered. Thecorrespondence of the genetically conserved sequence to the ordered structuresupports the identity of the RGS domain as a discrete unit of folding and function.The RGS4 structure was complemented by an NMR structure of a fragment ofGAIP that included only the RGS domain (42). The overall structures of the twoRGS domains are strikingly similar, which suggests that the RGS domain does notundergo major changes when it binds a Gα subunit. The only significant differencebetween the structures lies in an interhelix loop that is a point of contact with Gα.This difference probably reflects changes that occur upon Gα binding, rather than

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

800 ROSS ¥ WILKIE

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


a difference between RGS4 and GAIP, because the sequence of this region isconserved among RGS proteins (Figures 2 and 3).

The RGS domain (Figure 3) is a globular and mostly helical structure basedon a four-helix bundle (helices 4–7) and a second subdomain composed of the N-and C-terminal helices 1–3 and 8–9 (Figure 3). Further truncation of this minimaldomain has not produced active protein (33, 34), which suggests the importanceof both subdomains for overall stability and activity. Some of the most conservedresidues among the RGS and RGL proteins are those that face into the helicalbundles and are presumably important for structural stability. Other conservedresidues contribute to Gα binding (41, 42a). The RGS domain makes principalcontact with Gα over a discontinuous interface that includes three nearby interhelixloops: 3/4, 5/6, and 7/8. They make contact with all three of the switch regionsin the GTP-binding ras-like domain of Gα. The switch regions demonstrate thegreatest changes in structure upon activation and deactivation. The RGS domaindoes not make direct contact with bound GTP, which supports the idea that it doesnot directly take part in hydrolysis. This mechanism contrasts with that of GAPsfor the monomeric GTP-binding proteins (ras, etc.), which insert a catalyticallyimportant arginine residue into the GTP-binding site (43). The cognate arginine inGα subunits is provided by the helical domain of the Gα subunit (44). The structureof the RGS-Gα interface, particularly the pattern of stabilization of the Switch IIregion on Gα, is discussed below in terms of the mechanism of GAP activity.

Analysis of the five RGS subfamilies points to other regions of functional impor-tance. Many positions within the RGS domain contain residues found primarilyin one subfamily (Figure 2 and 3; see also 42a). Among the RGS GAPs, theseresidues are most frequent on the front face of helices 2 and 3, on the back faceof helices 6 and 7, and on loops 4/5 and 5/6. They include residues necessary foroverall structure and for the Gα contact sites, but they also appear to perform otherfunctions. Helix 7 is a particularly interesting example because it lies immediatelyadjacent to the most conserved portion of the RGS domain (helices 3, 4, and 8).It contains six residues conserved in nearly all RGS GAPs, interspersed with severalsubfamily-specific residues. This pattern suggests that helix 7 provides important

←−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−Figure 2 Amino acid sequence alignment of the RGS domains from a representativemember of each mammalian RGS subfamily (for complete alignment, see www.AnnualReviews.org). The features noted in the mammalian RGS domain include the positionof the α-helices in RGS4 and GAIP (rods) (41, 42), Gα contact residues (diamonds)(41), highly conserved amino acid identities (black), amino acid similarities (gray), andsubfamily-specific residues in R4 (blue), RZ (orange), R7 (green), R12 (red), and RA(purple). Gi and/or Gq GAP activity has been observed for all RGS proteins except theRA subfamily. Similarity/identity of amino acid residues in representative RGS and RGLproteins to the overall consensus sequence in the 127-amino-acid core domain are as fol-lows: RGSZ1, 48/39%; RGS4, 50/40%; RGS7, 43/37%; RGS12, 43/37%; axin, 37/28%;D-AKAPb, 24/18%; GRK2, 24/18%; GRK1, 23/19%; p115RhoGEF, 14/11%.

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

802 ROSS ¥ WILKIE

Figure 3 Relative conservation of amino acid sequence in the RGS family mapped ontothe structure of the RGS domain of RGS4. Amino acid sequence alignments of all membersof the RZ, R4, R7, and R12 subfamilies were generated (see www.AnnualReviews.org,Electronic Materials), and percent identity at each position was calculated. Relative con-servation at each position was mapped onto the structure of RGS4 (41).Top: RGS domainwith the Gα contact surface below.Bottom: As above but rotated 180◦ about the verticalaxis.α-Helices are numbered according to Tesmer et al (41; see also Figure 2).

contacts for binding to subfamily-specific regulatory proteins and that it may bean excellent target for mutational studies to test the function(s) of such proteins.

Several RGS and RGL proteins contain sizable inserts in the 4/5 and 6/7 loopsof the RGS domain, the side opposite that of the Gα binding site. The functionsof these inserts are mostly unknown, but they may mediate interactions with otherregulatory molecules (e.g. reference 45).1

1The RGS domain of axin has a tertiary structure similar to that of RGS4 and GAIP. It bindsan APC-derived peptide on helices 4 and 5, with a minor contact on the helix 2/3 loop. Thisis the side of the molecule opposite the G binding site (45a).

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


Conserved Domains in RGS-Flanking Regions

Even the shortest RGS and RGL proteins have recognizable functional domainsand motifs within their flanking sequences. These modules, most of which mediateprotein-protein interactions, are unique to individual subfamilies (Figure 1). Suchgrouping suggests that these flanking domains contribute to subfamily-specific ac-tivities, subcellular localization, or regulation. Specific functions of these domains,where known, are referred to elsewhere in this review. Flanking sequences in theRGL proteins also have clearly defined and independent regulatory functions: pro-tein kinase activity in GRKs, GEF activity in the p115s, and phospholipase activityin the PLC-βs.

MECHANISM AND CONSEQUENCES OF G PROTEINGAP ACTIVITY

GAPs bind the activated form of Gα and increase the rate at which bound GTP ishydrolyzed (Scheme I). Isolated Gα subunits hydrolyze bound GTP slowly, withrate constants (khyd) between 0.1 min−1 (Gz) and 2–4 min−1 (Gs, Gi, Go, Gt) at 30◦C(46). These rates correspond to half-lives for the GTP-activated state of 10 s to 7 minand are much too slow to account for the rates of deactivation in most G proteinsignaling pathways. GAPs accelerate GTP hydrolysis by as much as 2000-fold,such that GAP-stimulated hydrolysis rates (kgap) measured in vitro correspond toor exceed cellular deactivation rates. For example, PLC-β1 accelerates hydrolysisof its upstream activator Gαq-GTP> 1000-fold, from khyd = 0.012 s−1 to kgap =15 s−1 at 30◦C. The RGS4-Gαq complex displays an even greater kgap, 25 s−1,>2000-fold above khyd(47). RGS4 also accelerates hydrolysis of Gαo-GTP> 1000-fold, and RGSZ1, RGS-GAIP, and RET-RGS1 stimulate hydrolysis of Gαz-GTP>400-fold (48, 49). Although some GAPs may be less efficient, many reports ofsmaller rate enhancements (<100-fold) probably reflect either suboptimal assayconditions or an inappropriate Gα-GTP substrate (47, 50).

(Scheme I)

GAP activity is usually assayed in vitro by measuring the rate of hydrolysis of[γ -32P]GTP prebound to a purified Gα, referred to as a “single turnover” assay be-cause each Gα hydrolyzes only one molecule of labeled GTP. GAP activity is thenquantitated as if the GAP were an enzyme that converts a Gα-GTP substrate to theproducts Gα-GDP and Pi (Scheme I). This convention allows definition ofVmax=[GAP] · kgapandKm = (k−1+ kgap)/k1, the most convenient and reproducible pa-rameters for describing GAP activity in vitro. Together, these parameters providemeasures of maximal relative rate enhancement and apparent potency. Values ofVmaxandKm vary markedly for different GAPs and Gα-GTP substrates. Reported

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

804 ROSS ¥ WILKIE

values ofKm range from∼2 nM for the RZ subfamily and Gαz to 0.6µM forRGS4 and Gαo-GTP (48, 49). Maximal values ofkcat/Km exceed 108 M–1 · s–1,near the diffusion-limited maximum for an enzymatic reaction (48, 49). Althoughsubcellular localization of RGS proteins and their assembly into multiprotein sig-naling complexes may significantly influence RGS-Gα selectivities in vivo, valuesof Km should be a good estimate of the relative stability of RGS-Gα complexes incells and may suggest which GAPs act on which Gα subunits. Note that an RGSmolecule can sequentially bind and drive GTP hydrolysis on multiple moleculesof Gα-GTP during a “single-turnover assay.” RGS proteins can thus act catalyti-cally, consistent with their formally enzyme-like behavior. Such catalytic behaviormay be infrequent in cells, however, where RGS proteins may be relatively stablyassociated with G proteins and receptors during signal transduction.

GAP activity can also be detected more sensitively, and probably in a more phys-iological context, by monitoring the acceleration of agonist-stimulated steady-stateGTPase activity in a system composed of trimeric G protein, receptor, and GAPreconstituted in phospholipid vesicles (48, 50, 55). In such assays, the receptorpromotes GDP/GTP exchange such that hydrolysis becomes largely rate-limiting,and GAP activity is monitored as an increase in steady-state GTPase rate. The sen-sitivity of this assay probably reflects both orientational effects of the phospholipidbilayer on the interaction of receptor, G protein, and GAP and important mech-anistic interactions among the proteins (47, 56). Absolute quantitation is moredifficult in this format because rate acceleration depends both on the GAP andon the efficiency of receptor–G protein coupling. Quantitation of GAP activity invitro and the relative utilities of different GAP assays have been discussed in detailelsewhere (50).

Mechanism of G-Protein GAP Activity

Unlike GAPs for most of the monomeric GTP-binding proteins, GAPs for Gα

subunits do not directly take part in the chemistry of GTP hydrolysis but ratheralter the conformation of the Gα-GTP complex such that it becomes a better hy-drolase (41, 57). The mechanism of GAP action is therefore based on the multistepreaction scheme by which Gα subunits bind and hydrolyze GTP (Scheme II). Af-ter Gα binds GTP (reaction 1), the initial low-affinity Gα-GTP complex convertsto the active form Gα*-GTP (reaction 2), which can stimulate effector proteinsand which has reduced affinity for Gβγ (46). Formation of Gα*-GTP requiresMg2+ and can be monitored by enhanced fluorescence of a conserved tryptophanresidue near the GTP binding site (58, 59). Hydrolysis of bound GTP proceedsthrough a bipyramidal transition state intermediate, Gα‡-GTP (reaction 3). It wasfortuitously possible to determine the probable conformation of this intermediatebecause AlF−4 binds to Gα-GDP at the site usually occupied by the gamma phos-phoryl group of GTP and approximates the bipyramidal hydrolytic intermediate(44, 60–62). Gα-GDP-AlF4, which is activated insofar as it regulates effectors andbinds Gβγ weakly (46), has a conformation distinct from those of Gα*-GTPγSor Gα*-Gpp(NH)p (60, 61, 63–66). Hydrolysis and dissociation of Pi (reaction 4)

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


are essentially concomitant; no Gα-GDP-Pi species has been observed duringsteady-state hydrolysis (although bound orthophosphate has been observed insome crystals (67). The detailed chemical mechanism of these reactions has beenreviewed by Sprang (44).

(Scheme II)

By definition, GAPs act by decreasing the activation energy for hydrolysis1G‡,either by energetically stabilizing the Gα‡-GTP transition state complex or by rel-atively destabilizing the preceding lower-energy state (presumably Gα*-GTP).The importance of stabilizing the Gα‡-GTP state was supported by the findingof Berman et al (68) that RGS4 binds more tightly to the transition state analogGαi1-GDP/AlF4than to Gα*-GTPγS (60, 61). A crystal structure of an RGS4-Gαi1-GDP/AlF4 complex also shows the Gαi1 to be in the transition-state-like configu-ration (41). Several other RGS proteins also bind preferentially to the GDP/AlF4-bound form of their Gα targets (20, 22, 24, 69, 70). The selectivity of GAPs fortransition state–like structure extends to the monomeric GTP-binding proteins,which do not bind AlF−4 under normal conditions but will do so when bound to theirrespective GAPs (44, 71). Preferential binding to Gα-GDP/AlF4 relative to Gα-GTPγS is not universal. RGSZ1 binds Gαz-GTPγS and Gαz-GDP/AlF4 with sim-ilar affinity (72); the same is true for PLC-β1 and Gαq. In these cases, the GAP mayaccelerate conversion of Gα-GTP to Gα*-GTP. Such an effect would be predictedfor effectors and may explain how PDEγ potentiates the GAP activity of RGS9.

The structure of the RGS4 complex with Gαi1-GDP/AlF4 (41), combined withmutational and kinetic data, has helped elucidate the mechanism of the GAP activ-ity of RGS proteins. RGS proteins make a broad and discontinuous interface withGα subunits. The interface includes each of the three switch regions of Gα, regionsnear the GTP binding site that undergo large conformational changes upon activa-tion. Each switch is in direct contact with one or more of three interhelix turns ofthe RGS domain (Figure 3). Tesmer et al (41) noted that the major change in Gαi1upon binding RGS4 is decreased mobility of Switch II, as indicated by its decreasedcrystallographic temperature factor. This result led them to suggest that GAPs actby stabilizing Switch II in a catalytically active conformation through multipleinterprotein contacts rather than by altering the position or orientation of one ormore specific catalytic residues. This mechanism is probably general because keyresidues in the contact site of the RGS domain are conserved and the structure offree GAIP is nearly identical to that of RGS4 bound to Gαi1-GDP/AlF4 (42).

The functional importance of residues on both sides of the RGS-Gα interface hasbeen tested by mutational analysis, with the conclusion that RGS interface residuesdo not participate directly in catalyzing GTP hydrolysis, but rather stabilize thecatalytic conformation of the Gα subunit. Mutation of single residues of RGS4at or near the Gα interface can inhibit GAP activity profoundly (73), including

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

806 ROSS ¥ WILKIE

several residues that are not well conserved among RGS domains; however, eachposition can tolerate some substitutions with slight inhibition of GAP activity. Evena highly conserved asparagine residue (Asn128in RGS4) that lies near the Gα activesite can be multiply substituted with minor effects, primarily on affinity of RGS-Gα binding (48, 73, 74). On the Gα side of the interface, mutation of residues ineach of the three switch regions can diminish responsiveness to RGS proteins oralter selectivity among them (75–78), which again suggests that a relatively globalconformational effect on Gα leads to accelerated hydrolysis.

The N-terminal helix of Gα is also important for its response to GAPs. N-terminal proteolysis (25–30 residues), phosphorylation (tested only for Gαz),or palmitoylation (at Cys3) decreases affinity for RGS proteins by∼100-fold(49, 79, 80); myristoylation of the N-terminal glycine substantially increases affin-ity (79). Because mutation of Ser16 or Ser27 in Gαz also reduces the basal GTPhydrolytic rate by 20–30% in addition to inhibiting response to GAPs (81), theN-terminal region may be important for maintaining the structure of the nearbyactive site rather than for mediating GAP effects. Alternatively, it may contributeto the interface with the RGS protein, even though such contact was not detectedin the crystal structure of the Gαi1-RGS4 complex. In those crystals, the Gαi1 Nterminus made a presumably artifactual crystal contact with the RGS4 moleculein the adjacent unit cell (41). Additional regions of Gα and RGS protein outsidethe known molecular interface also contribute to GAP activity. These include bothflanking regions of RGS4, which are important for interaction with Gq but notGi (35), and theα-helical domain of Gαt, which contributes to recognition of Gt(82). Both interactions probably enhance RGS-Gα affinity rather than altering themechanism of GAP action.

Although mutagenesis efforts have mainly reinforced the original hypothesisof Tesmer et al (41), three potentially useful mutants have also been produced.First, selection for pheromone-sensitive mutations in the yeast Gα subunit Gpa1pdemonstrated that substituting a conserved glycine residue in Switch II makesGα subunits essentially insensitive to GAP activity (77, 83). Such GAP-resistantmutants in the Gi and Gq classes may facilitate cellular analysis of the physiologicaleffects of GAPs. Second, mutation of Asp229 in Gαs to Ser conferred limited GAPsensitivity (84), which demonstrates that the Gs class of Gα subunits is capable ofresponding to GAPs and encourages the search for endogenous Gs GAPs. Third,studies of RGS mutants with diminished GAP activity that retain the ability to bindtheir Gα targets (48, 85) may lead to the design of efficacious dominant negativeRGS proteins that may be used to disrupt GAP regulation in cells.

Regulatory Consequences of GAP Action

GAPs can promote rapid signal termination upon removal of an agonist, inhibitsignaling during continued stimulation, or both. Controlling which effect predomi-nates will determine the physiological function of a GAP, but using a GAP for onlyone of these functions demands coordinated regulation of the entire GTPase cycle.

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


Many RGS proteins act as inhibitors; Sst2p in yeast and EGL-10 and EAT-16 inC. elegansare good examples (17, 19, 86). A GAP could also selectively suppressbasal signaling by quenching spontaneously activated G protein without signifi-cantly inhibiting signaling in response to stimuli. In contrast to these inhibitoryeffects, GAPs can also accelerate signal termination without attenuating signalamplitude. In mice that lack RGS9, the principal RGS protein in photoreceptorcells, visual recovery after a light flash takes longer than in wild-type mice, butthe amplitudes of signals elicited by short or long flashes are altered only slightlyif at all (87). Similarly, coexpression of RGS4 or RGS8 increases the speed withwhich a G protein-gated K+ channel is regulated without significantly inhibitingthe agonist-initiated signal (88–92). Such control of signal kinetics is vital in cellssuch as neurons, which must turn on and off rapidly.

Kinetic studies of the Gq GAP activity of PLC-β1 during steady-state, agonist-stimulated signaling have begun to explain how a GAP can accelerate deactivationwithout attenuating signal amplitude. Both agonist-stimulated phospholipase ac-tivity and each of the intermediary steps in the GTPase cycle for this system canbe measured with the use of reconstituted phospholipid vesicles that contain het-erotrimeric Gq, ml muscarinic cholinergic receptors, and the PLC substrate PIP2(47, 56). In this system, PLC-β1 accelerates the rate of hydrolysis of Gαq-GTPfrom ∼0.01 s−1 to ∼15 s−1, but stimulation of phospholipase activity by ago-nist in the presence of GTP remains substantial, 50–100-fold above basal. Signalamplitude is maintained because addition of PLC-β1 (or RGS4, also a Gq GAP)increases the rate of receptor-catalyzed GDP/GTP exchange from∼0.007 s−1 to∼1.8 s−1 (47). Experiments using m2 receptors, Gi, and RGS4 produced similarbehavior at steady-state (48). This work led to the conclusion that, in the presenceof agonist and GAP, rapid steady-state GTP hydrolysis is catalyzed principally bya complex of receptor, G protein, and GAP (47, 56). This complex both exchangesGDP for GTP and hydrolyzes bound GTP quickly because all three componentsremain associated through many hydrolytic cycles (inner cycle in Figure 4). In theabsence of a GAP, the receptor probably dissociates during the relatively long life-time of the activated Gα-GTP complex because receptor affinity for GTP-boundGα is relatively low. In this case, receptor reassociation with G protein becomesthe rate-limiting step for the overall cycle (93). In the presence of an active GAP,however, the receptor is still associated with the G protein after GTP hydrolysisand can then rapidly reactivate the G protein. By allowing the receptor to remainbound, a GAP indirectly facilitates rapid activation in addition to directly causingrapid deactivation. Signal strength is maintained while an agonist is present, buttermination is fast when the agonist is removed.

Persistent association of receptor, G protein, and GAP during a prolonged stim-ulus may result simply from a combination of mutual affinity of the proteins andfavorable kinetics (35, 47, 56). This is probably the case for the proteoliposomesdescribed above. Rapid cycling and maximal phospholipase activation are delayedfor ≤90 s after addition of an agonist, suggesting that binding of receptor to Gprotein and subsequent recruitment of GAP are required to initiate formation of the

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

808 ROSS ¥ WILKIE

Figure 4 GAPs kinetically regulate association of signaling proteins during the GTPasecycle. A GAP (G) can accelerate the overall GTPase cycle while maintaining a significantpool of activated Gα*-GTP by kinetically favoring continued association of agonist-ligandedreceptor (R) and Gα. Without a GAP, hydrolysis of Gα-bound GTP is slow (t1/2∼ 10 s)(outer cycle). Receptor dissociates after Gα*-GTP is formed, such that the rate-limitingstep both in reactivation and in the overall cycle is the reassociation of R with Gα-GDP topromote the next round of GDP/GTP exchange (as originally suggested by Levitzki (see181). In the presence of an active GAP (inner cycle), Gα*-GTP is hydrolyzed before thereceptor dissociates. It can, therefore, promote rapid GDP/GTP exchange (t1/2< 500 ms)before the GAP and/or effector dissociates. Although steady-state GTPase activity is high,GDP/GTP exchange is fast enough to maintain significant Gα in the GTP-bound, activatedstate. Upon dissociation of agonist, however, signaling terminates promptly. If the proteinsare not associated prior to addition of agonist, the transition from outer to inner cyclepresumably occurs by the binding of a GAP to the transient R-Gα*-GTP complex (bluearrow, upper left) or by its activation. A GAP would be most likely to dissociate followingGTP hydrolysis (blue arrow, lower right). Gβγ has been omitted for simplicity. Figuremodified from Biddlecome et al (56).

complex (shaded arrow in Figure 4, left). Although signal termination is fast uponremoval of an agonist (t1/2< 3 s; perhaps as fast as GTP hydrolysis), the complexappears to remain associated for at least 10–20 s after the agonist dissociates (56).In cells, however, RGS proteins or associated proteins may also act as scaffoldsto organize receptor, G protein, and GAP (and, perhaps, other proteins) in a stablesignaling complex even in the absence of stimulus (35, 88, 94, 95; see 96 for reviewof scaffolds). This would account for the finding that RGS4 and RGS8 actually

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


accelerated the onset of signaling as well as its termination when signal outputwas measured using a Gβγ -gated K+ channel as a monitor (88–90, 92). Note thatthis general scheme also allows for the modulation of GAP activity (see below).Conversion from a slow-response regimen to a faster pattern of responses can thusbe controlled by the concerted regulation of receptor, GAP, and their associationwith G protein.

G PROTEIN EFFECTORS AS GAPS

The first recognized G protein GAP was the Gα-regulated effector protein PLC-β1 (11). Only a few other G protein effectors are known to have GAP activity:p115RhoGEF, a G13-regulated effector and G13GAP (15); possibly some isoformsof adenylyl cyclase (97); and, conceptually at least, theγ subunit of retinal cyclicGMP PDE, which potentiates the Gt GAP activity of RGS9 (12, 13, 98, 99). Thereason that GAP activity evolved in effector proteins remains speculative. RGSproteins are generally assumed not to have effector functions, but regulation oftheir GAP activity certainly confers a signaling function either in feedback orcrosstalk circuits. For example, genetic evidence suggests that EAT-16, a Gq GAPfound in neurons ofC. elegans, is regulated by Gαo (86). In addition, the signal-ing modules outside the RGS domain may generate their own signals (e.g. thePP2A-like domain of axin) or regulate other proteins to which they bind. Activ-ities of these flanking domains and their potential control by G proteins largelyremain to be elucidated. The finding that the RGS domains of GRK2 and 3 bindGαq, display modest Gq GAP activity (37), and are themselves regulated by Gβγ

(100, 101) is a good example of how such signaling via RGS proteins may takeplace.

PLC-ββ

The PLC-β family of PIP2-selective phospholipases are the major recognizedeffectors for the Gq class in addition to being highly Gq-selective GAPs. PLC-β1binds GTP-bound Gαq with a Kd estimated to be∼2 nM and accelerates hydrolysisof Gαq-bound GTP from 0.01 s−1 to 15 s−1 at 30◦C, ∼1000-fold (11, 47, 56).Although comparably detailed measurements are not available for other PLC-β

isoforms and Gq family members, all the PLC-βs tested are Gq GAPs. The GAPactivity of PLC-β4 is about equal to that of PLC-β1, the maximal activity ofPLC-β2 is approximately that of PLC-β1 but its EC50 is ∼40 times higher, andPLC-β3 appears to be a somewhat weaker GAP than is PLC-β1 (G Biddlecome,S Mukhopadhyay, EM Ross, unpublished data). TheDrosophila visual PLC-βNorpA appears to act as a GAP for the endogenous Gq in photoreceptor membranes(102; B Cook & Z Selinger, unpublished data). In addition, Gq, G11, and G16(all Gqclass) are about equally sensitive to the GAP activity of PLC-β1 (11; unpublisheddata). Unlike the RGS proteins, PLC-β1 is strictly selective for the Gq class. Wehave been unable to detect any GAP activity of PLC-β1 toward Gi, Go, Gz, or Gs.

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

810 ROSS ¥ WILKIE

PLC-βs bear no obvious structural relationship to the RGS proteins, but, byusing mutation and proteolysis, it has been possible to map domains importantfor interaction with Gαq. PLC-βs share a catalytic core with the PLC-γ s andPLC-δs, but they are notable for a C-terminal extension of∼40 kDa (103–105).Removal of the C-terminal region blocks the ability of the PLC-β to respond toactivated Gq (106, 107) or to act as Gq GAPs, as do several deletion mutationsin this region (O Ilkaeva & RH Paulssen, unpublished data). Isolated fragmentsof the C terminal domain as small as 18 kDa display Gq GAP activity, but theirdiminished affinity (100 nM vs 2 nM for intact PLC-β1) and complex interactionssuggest that intact PLC-β1 may have more than one contact site for Gαq (40, 108).Nevertheless, expression of the C-terminal domain in cells has been reported toinhibit Gq-mediated signaling, which suggests at least that it sequesters cellularGαq (107, 109).

Because the GAP activity of PLC-β1 is similar in many ways to that of RGSproteins, it is tempting to speculate that the PLC-βs evolved from the fusion ofgenes for an RGS protein and a PLC-δ, which lacks the C-terminal GAP domainand has no GAP activity. Both PLC-δ and RGS proteins are found in the fungi,which lack PLC-βs. Weak similarity can be found in alignments of the aminoacid sequences of PLC-β GAP domains and RGS domains, but any phylogeneticrelationship between RGS proteins and PLC-βs remains uncertain.

Rho GEFs

Genetic evidence indicates that activation of the monomeric GTP-binding pro-tein Rho by GPCRs is probably mediated by G13 (110–113), and an N-terminalregulatory domain of a family of Rho-selective GTP exchange factors (GEFs) isslightly similar in sequence to the RGS domain. This information led Hart et al(14) and Kozasa et al (15) to discover both that the Rho GEF activity of p115 isstimulated by Gα13 (but not Gα12) and that p115 is an active and selective Gα13GAP (and a much weaker Gα12 GAP). The hydrolysis by p115 of GTP bound toGα13 was accelerated at least 10-fold in a single-turnover assay. Although GAPactivity did not obviously saturate even at 100 nM p115, competitive inhibitionby GDP/AlF4-bound Gα13 suggests that its Kd for p115 is∼3 nM. The GAP p115is highly selective for the G12 class of Gα subunits, reminiscent of the selectiveinteraction of PLC-β and Gq. It was inactive as a GAP toward members of theGi, Gq, and Gs classes. A related RhoGEF displayed similar selectivity for Gα12/13in binding assays (114). Although Gα12 was not a good GAP substrate for p115and did not stimulate RhoGEF activity, GDP/AlF4-bound Gα12 potently inhibitedthe G13 GAP activity of p115 (14, 15). Gα12 may thus be a true antagonist of theGα13-p115-Rho pathway.

The N-terminal RGL domain of p115 displayed G13 GAP activity when ex-pressed separately, supporting the validity of its weak sequence similarity to theRGS domain. As mentioned for PLC-β, the p115RhoGEFs thus appear to havearisen from the fusion of genes for an otherwise regulated exchange factor and a

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


primordial RGS domain. Such a history implies not just that GAP activity couldbe conferred in this way, but that regulation by Gα subunits may have arisen byfirst using a RGL domain for Gα binding, followed by the evolution of regulatoryresponsiveness.

Cyclic GMP Phosphodiesterase

Although not a GAP itself, theγ subunit of the cyclic GMP PDE from vertebratephotoreceptor cells both forms the Gαt binding site on the PDEαβγ 2 heterote-tramer and potentiates the GAP activity of RGS9/Gβ5. This “co-GAP” activitywas discovered because PDEγ increased the rate of hydrolysis of GTP added towashed photoreceptor membranes to which a small amount of Gt had been restored(12). Because RGS9/Gβ5 remained membrane bound, it was able to combine withrhodopsin and Gt to reveal this activity (13, 98, 115–117). The co-GAP activityof PDEγ is crucial for appropriate termination of the response to light: Mice thatexpress a mutant PDEγ that interacts poorly with Gt display a prolonged recov-ery after a light flash (99), a phenotype similar to that of the loss of RGS9 (87).The structural basis of the co-GAP effect of PDEγ is unknown. PDEγ selectivelypotentiates the GAP activity of RGS9 (and perhaps other R7 subfamily members)(13), but slightly inhibits the GAP activity of RGS4, RGS16, RGS6, and GAIP(42a; 117a,b). Evolutionary relationships led Sowa et al (42a) to speculate thatPDEγ binds directly to RGS9, but McEntaffer et al (117c) found no evidence ofhigh affinity binding of the isolated proteins.

Why Should Effector Proteins Be GAPs?

Why should an effector protein promote the deactivation of the Gα that regulatesit? The energetics of the GTPase reaction does not demand that an effector, de-fined as a preferential ligand of Gα*-GTP relative to Gα-GDP (Scheme II), eitheraccelerate or inhibit GTP hydrolysis. One function of intrinsic GAP activity is toallow each effector of a single G protein to display distinctive response kinetics anddynamics. For example, different PLC-β isoforms may respond differently to thesame Gq-linked receptor based on their different GAP activities. In general, intrin-sic GAP activity allows an effector to respond rapidly to a stimulus and its removalbut maintain high signal throughput while the stimulus is present (56; see Figure 4).Viewed slightly differently, GAP activity poises an effector to respond to an in-hibitory input, perhaps via a distinct, but associated, RGS protein. Having GAPactivity reside in the effector also allows a newly activated G protein to re-main active until it encounters the effector, whereupon it would either deacti-vate quickly if no agonist were present or, in the continued presence of agonist,enter a rapid and controlled GTPase cycle and maintain its activity (Figure 4).This mechanism will suppress basal (background) signal output without sacrific-ing the maximal signaling capacity of the pathway. Last, GAP activity allows aneffector to enhance the selectivity of its upstream G protein among different re-ceptors. If a receptor is to activate an effector with intrinsic GAP activity, it must

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

812 ROSS ¥ WILKIE

bind the G protein target tightly enough to remain associated throughout multipleGTPase cycles (Figure 4). A receptor that can activate a given G protein but thatbinds with low affinity would be unable to maintain activation if the effectormolecule were a GAP. Such a receptor could use that G protein to convey its signalto an effector without GAP activity but could not signal to an effector that is a GAP.

GAPS AND SIGNAL PATHWAY SPECIFICITY

The ability of RGS proteins to select among Gα substrates is clearly necessaryfor their actions as signal regulators. Selectivity of an RGS protein for a G proteintarget may be manifest merely as appropriate modulation of signal amplitude froma receptor. If a receptor regulates multiple G proteins, however, then selectivemodulation of one Gα by an RGS protein can qualitatively change the nature ofoutput from a single receptor. RGS proteins can be highly specific for Gα targets.EGL-10 is highly selective for Goover Gqin C. elegansneurons (19, 86), and neitheryeast RGS protein (Sst2p or Rgs2p) can complement a deficiency of the other (118).Similarly, p115RhoGEF acts only on G12classα subunits, and PLC-β acts only onGq class members. It has been more difficult to delineate the selectivity of most ofthe mammalian RGS proteins among their potential Gi and Gq class targets. Someselectivity is conferred by restrictive patterns of cellular expression (e.g. RGS9and Gαt in retinal photoreceptor cells), but little is known about selective RGS-Gα

interactions in a less-specialized cell that expresses multiple RGS proteins and Gα

subunits. Investigators have tried to identify and evaluate physiologic targets ofRGS proteins by measuring selectivity in GAP assays and relative attenuation ofsignaling upon transfection or microinjection. Results from these approaches oftenagree, but observed selectivity can also be assay-dependent; neither approach iscertain to reflect the action of an RGS protein in its natural environment.

Selectivity In Vitro

The selectivity of RGS proteins among potential Gα targets can be evaluated invitro by comparing their behaviors in single-turnover GAP assays (Km, kcat), theirabilities to compete with substrate in such assays (Ki), or their binding affinitiesaccording to coprecipitation or coadsorption assays. Relative activities of differentGAPs can also be compared by monitoring elevation of agonist-stimulated GTPaseactivity (see 50 for description of GAP assays formats and analysis). RGS and RGLGAPs have displayed varied patterns of selectivity in such assays. For example,GAIP displays GAP activity toward both Gi and Gq class substrates. It is relativelynonselective among Gi class members (20, 49, 119); it is a weak GAP for Gαq [lowKm but low relative stimulation of hydrolysis (49); interacts poorly in yeast-two-hybrid assay]. In contrast, RGSZ1 and RET-RGS1, which are closely related toGAIP by sequence and which are equally active as Gαz-GAPs, are highly selectivefor Gαz relative to other Gi class members and display very low GAP activity

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


toward Gαq (49). Similarly, RGS6 and RGS7 (in complex with Gβ5; see below)are selective for Gαo relative to other Gi class members (166). The selectivity ofRGS2 for Gq over Gi class Gα subunits is more complex: Selectivity for Gq ishigh in transfected cells (120, 121) and in single-turnover assays (55), but RGS2is an active Gi GAP in the steady-state assay (55). Although the receptor-coupled,steady-state assay appears more physiologically relevant than the single-turnoverassay and allows quantitative comparison of RGS proteins and Gα substratesover a wide range of activities, its greater sensitivity may detect nonphysiologicalactivities. For many other RGS proteins, selectivity in vitro has not convincinglyindicated their cellular preferences (when such preferences are known).

Amino-acid residues on RGS proteins and Gα subunits that determine selectiv-ity have been identified in a few cases, but selectivity apparently involves multipleinteractions throughout the RGS-Gα interface. For example, the selectivity ofRGS2 for Gαq is largely conferred by three Gαq residues, Cys106, Asn184, andGlu191, which apparently reduce affinity for Gαi by steric hindrance (121). RGS16can accelerate GTP hydrolysis by the D229S mutant of Gαs (substitution by thecognate residue of Gαi), although wild type Gαs neither binds RGS16 nor is asubstrate (84). Mutations at this site can also alter the selectivity of RGS-GAIPamong Gαi isoforms (76). RGS4 N128F becomes more selective for Gαq relativeto Gαi; mutation of Lys180 in Gαi1 to proline, the homologous residue in Gαq,restores some of the interaction with the mutant RGS4 (48). Theα-helical domainof Gαt is also important for its selective interaction with RGS9 (82).

RGS-Gα Selectivity in Cells

Genetic complementation assays in yeast provided the first tests of RGS functionin cells, but also demonstrated that exogenous RGS proteins, when overexpressed,can act promiscuously on nonphysiologic substrates (70, 122). Similarly, trans-fection studies in mammalian cells have shown that almost any RGS protein canattenuate signals mediated by Gi and/or Gq. Nonetheless, evidence of regulatoryspecificity has emerged from such studies. As mentioned above, RGS2 displayscellular selectivity for Gq over Gi targets (120, 121, 123). In transfected CHO cells,RGS16 inhibited activation of p38 MAP kinase by platelet activating factor moreeffectively than platelet activating factor-initiated activation of ERK2; RGS1 dis-played the opposite pattern of selectivity (124). This behavior presumably reflectsdifferential targeting of Gq or Gi by these RGS proteins even though Gi and Gq areapparently activated by the same platelet activating factor receptor (see also 123).The best evidence for Gα-selective action of RGS proteins will emerge from stud-ies of natural cells whose endogenous signaling pathways can be evaluated withoutoverexpression of transfected genes. An early example comes from studies of Ca2+

channel regulation in dorsal root ganglion cells (125). In this study,α2-adrenergicactivation of Gαi and Gαo was enhanced by antibodies against RGS4 and GAIP,respectively, whereas antibodies against RGS2, RGS12, and RGS10 did not altertheir signaling (125). Microinjection of the RGS proteins confirmed these results.

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

814 ROSS ¥ WILKIE

RGS proteins can thus alter signal inputs from single or multiple receptors andtarget individual downstream signaling pathways.

Receptor-Selective Activities of RGS Proteins

RGS proteins can also selectively regulate different receptors that use the same Gprotein. In pancreatic acinar cells, RGS4 blocked Gq-mediated PLC-β activationand consequent Ca2+ signaling initiated by m3 muscarinic cholinergic receptors,cholecystokinin receptors, or bombesin receptors, all of which signal through Gqclass Gα subunits (126). However, inhibition of m3 muscarinic stimulation was∼30-fold more potent than inhibition of cholecystokinin action, and stimulation bybombesin was intermediately sensitive. RGS1 and RGS16 were 1000- and 100-foldselective, respectively, in inhibiting muscarinic Ca2+ signaling action relative tocholecystokinin. Receptor selectivity did not reflect differential use of Gq class Gαsubunits by each receptor because the same pattern of inhibition was observed incells from wild-type and several single and double Gq-class knockout mice (126).The receptor-selective action of RGS4 requires the concerted action of both theN- and C-terminal flanking regions and the RGS domain; the RGS domain aloneinhibited signaling from each receptor equally and displayed 10,000-fold lowerpotency (35). It is unknown whether selectivity among receptors is conferred bydirect receptor-RGS4 binding, by subcellular colocalization, or by distinct kineticproperties of the receptors. In the case of RGS12, an N-terminal PDZ domainbinds specific chemokine receptors (94), but receptor-RGS4 binding has not beenobserved.

REGULATION OF RGS PROTEINS

The ubiquity and diverse activities of G protein GAPs virtually demand theirregulation. Cellular levels of many RGS mRNAs are modulated by developmentalfactors and more acute stimuli (see list in 27). The activities of RGS proteins aremodulated by covalent modification and modification of their Gα substrates, byalteration of their subcellular localization, and by interaction with multiple, newlydiscovered binding partners.

Regulation of RGS Protein Synthesis and Accumulation

Even before the yeast RGS protein Sst2p was shown to be a GAP for the Gprotein Gpa1p, its expression was found to be induced at the transcriptional levelby the mating factor that initiates Gpa1p signaling (17). Because Sst2p helpsovercome cell cycle arrest induced by mating factor, Sst2 induction provided thefirst indication that RGS proteins can act as negative feedback regulators of thepathways that induce their expression. Similarly, the mammalian proteins RGS1and RGS2 were identified as genes that are induced by activators of B-lymphocytesand monocytes.

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


The mRNAs for RGS proteins are now known to be regulated by a varietyof stimuli; many of these experiments have been recently reviewed (27, 32). Asexpected for proteins that regulate G protein signaling, almost all RGS genes areexpressed in the brain. Several are restricted to discrete regions (27, 128, 129),and many are regulated by neurotransmitters, trophic agents, and second messen-gers (55, 130–133). It is not yet known how much of this regulation is feedbackfrom downstream signals and how much represents inputs from distinct pathways.Many RGS mRNAs are also regulated in non-neural cells and tissues. They in-clude RGS3 and RGS4 in heart (134), GAIP and RGS16 in colon adenocarcinomacells (135, 136), RGS2 inβ-TC3 insulinoma cells (137), RGS2 and RGS16 inT-lymphocytes (123), and RGS16 in ventricular cardiomyocytes (138) and liver(K Yu & T Wilkie, unpublished data). Regulatory inputs are diverse. They includefibroblast growth factor and pressure overload in the heart, interleukins and mito-gens in T cells, the tumor suppressor p53, and glucose-dependent insulinotropicpeptide.

Regulation of the cellular concentrations of RGS proteins post-transcriptionallyis also important, but the natural cellular concentrations of RGS proteins have rarelybeen measured because of the low levels at which they are usually expressed. Al-though translational regulation of RGS protein accumulation has not yet beenstudied, the cellular concentrations of RGS proteins are, in some cases, modu-lated by regulating their proteolysis. The half-life of Sst2p in yeast is increasedby phosphorylation by the MAP kinase Fus3p (45). The phosphorylation site, inthe 4/5 loop of the RGS domain, is near two PEST motifs, which mark rapidlydegraded proteins. Regulation of RGS7 protein stability is particularly striking.RGS7 is completely degraded by ubiquitin-directed proteolysis within 60 min af-ter addition of cycloheximide, but it is protected by TNFα-initiated activation ofthe stress-activated MAP kinase p38 (133). The TNFα-responsive phosphoryla-tion site is in a Ser/Thr-rich sequence unique to RGS7 that lies just N-terminalof the Gγ -like (GGL) and RGS domains. TNFα-induced stabilization of RGS7protein is apparently significant in vivo because injection of lipopolysaccharideincreases RGS7 levels in brains of wild type mice but causes a much smaller ef-fect in brains of mice that lack functional TNF receptors. Stabilization of RGS7may thus contribute to some of the neurological effects of sepsis (133). RGS7 inkidney is protected from rapid proteolysis by binding to polycystin (139), a pro-tein frequently mutated in polycystic kidney disease. Because RGS7, polycystin,and Gβ5 are expressed coincidently in developing renal tubules, polycystin maymodulate the regulatory activities of RGS7/Gβ5 in addition to inhibiting its degra-dation.

Covalent Modification of RGS Proteins and Their Gα Targets

The activities of RGS proteins are regulated by covalent modification of theRGS proteins themselves and of their Gα substrates. Regulation of the RGS pro-tein turnover (see above) and of the regulatory behavior of PLC-β isoforms by

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

816 ROSS ¥ WILKIE

phosphorylation (140, 141) has now been noted in several cell types. The Gz GAPactivity of RGS proteins is also regulated by phosphorylation near the N termi-nus of Gαz. Gαz, a member of the Gi class that is expressed at low levels in thebrain and a few other secretory tissues, is phosphorylated at two nearby sites byprotein kinase C and at one of the two sites by PAK, which potentially allows theregulation of Gz by several signaling pathways (81). Phosphorylation at either siteinhibits GAP activity. The extent of the inhibition can exceed 90%, but it variesamong RGS proteins and according to which site is phosphorylated (49, 80, 81).Gαz is naturally phosphorylated by PKC in platelets and elsewhere (142, 143), andcan be phosphorylated by PAK in 293 cells in response to recombinant activatedRac1 (81). Gα12 is also phosphorylated near its N terminus by PKC (144), andit will be of interest to find whether phosphorylation alters its interactions withp115RhoGEF.

GAP activity is modulated by the reversible palmitoylation both of RGS pro-teins and of their Gα targets. Several RGS proteins are palmitoylated near their Ntermini and at a highly conserved cysteine residue in the RGS domain. N-terminalpalmitoylation, at a pair of cysteine residues in the R4 subfamily (145, 146) andat a cysteine string in the RZ subfamily (119), is implicated in subcellular local-ization and membrane association, although details of the relationship are unclear.Palmitoylation in the RGS domain strongly inhibits GAP activity when measuredin detergent solution in a single-turnover assay, but can potentiate GAP activityin receptor-G protein proteoliposomes (147). This bidirectional control of GAPactivity depends in part on which RGS protein is studied. It is not known whetherRGS protein palmitoylation is regulated in cells, but in RGS4-expressed Sf9 cellspalmitate turns over at both sites on the 10–60 min time scale.

Gα subunits are also reversibly palmitoylated at a conserved N-terminal cys-teine residue (151–153). Gα palmitoylation inhibits responsiveness to the GAPactivity of RGS proteins, primarily by increasing Km (79). Inhibition can be nearlytotal, but the extent of inhibition in one report correlated with the activity of theRGS protein on the particular Gα-GTP substrate. Thus, the activity of RGSZ1 onGαz was inhibited>90% by Gαz palmitoylation, but other RGS-Gα pairs that in-teracted less productively or with lower affinity were inhibited less. Many studiesof palmitoylation-deficient mutants have linked Gα palmitoylation both to signal-ing activity and to the membrane localization, but the diversity of these results hasnot allowed generalization that supports or refutes the physiologic importance ofthe inhibition of GAP activity (49, 80, 81).

Cellular Binding Partners of RGS Proteins

RGS proteins can bind a wide variety of other cellular proteins in addition to Gα

subunits. These interactions may modulate GAP activity or subcellular localiza-tion, or may allow RGS proteins to regulate novel signaling pathways. Physio-logical roles for most of these interactions are not known, but they are under in-tense study. Specific binding partners and interaction domains appear to correlate

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


within RGS subfamilies; these are listed in Figure 1. In addition to the cysteinestrings, N-terminal amphipathic helices and GGL-domains (described separately),DEP, PTB, PDZ, Rap binding, and LOCO domains have been identified by se-quence homology and/or yeast two-hybrid screens (94; 154a,b). PDZ-binding mo-tifs were discovered on two RGS proteins, RGS12 and GAIP (94, 154c). ThePDZ-domain protein GIPC (for GAIP interacting protein, C terminus) may ori-ent GAIP within a multiprotein complex to regulate vesicular trafficking (154c).Lipids may also regulate RGS activity. For example, the GAP activity of most RGSproteins is inhibited by PIP3 but not other phosphoinositides, and PIP3-dependentinhibition is reversed by Ca2+-calmodulin (SG Popov, UM Krishna, JR Falck& TM Wilkie, J. Biol. Chem.275: In press). The opposing effects of PIP3 andCa2+-calmodulin may convey Ca2+-responsive feedback regulation of RGS GAPactivity.

RGS proteins themselves may provide scaffold functions to regulate signalingapart from or in addition to, their GAP activity. Axin, the prototype for the RAsubfamily, provides a perhaps extreme example of the interrelated protein bindingactivities of RGS proteins. Axin, a component of the wnt developmental signal-ing pathway, is not known to have GAP activity or to bind to a Gα. It appearsinstead to be a scaffolding protein. Its RGS domain binds adenomatous polyposiscoli protein (154); they together control wnt-dependent nuclear translocation ofthe transcription factorβ-catenin (155, 156). In the absence of wnt, axin bindsbothβ-catenin and a glycogen synthase kinase at sites outside the RGS domainand promotes glycogen synthase kinase-dependent phosphorylation ofβ-catenin.Phosphorylation targetsβ-catenin for degradation (155). If the RGS domain ofaxin is removed, the resulting mutant still bindsβ-catenin but prevents its phos-phorylation, thus mimicking wnt signaling by stabilizingβ-catenin (157). Axinhas also been reported to bind MEKK1 in yet another region (158). Each domainof axin thus provides scaffolding sites for the regulated interaction of multiplesignaling molecules in an important developmental pathway.

REGULATION BY Gββγγ SUBUNITS AND A GGL DOMAIN

Three recent observations point to novel and important effects of Gβγ subunitsand/or Gβ5 on the action of G protein GAPs; only some relate to their previouslyknown activities (see 159 for review of Gβγ regulatory functions). Gβγ bindsto Gα negatively cooperatively with respect to GTP. Activation by GTP thereforedisplaces Gβγ from Gα (or at least alters its mode of binding) such that Gβγ

can regulate multiple effector proteins, most notably some isoforms of adenylylcyclase, K+channels, PLC-β, and protein and lipid kinases. Conversely, Gβγ bindsGα positively cooperatively with respect to GDP, thereby stabilizing GDP bindingand suppressing spontaneous, receptor-independent activation. In addition, Gβγ

helps anchor Gα to the membrane and (partly thereby) is essentially required forreceptor-catalyzed GDP/GTP exchange.

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

818 ROSS ¥ WILKIE

Gββγγ Inhibits GAP Activity in Single-Turnover Assays

Gβγ inhibits the GAP activities of both RGS proteins and PLC-β in solution-based,single-turnover assays (49, 72, 160). Inhibition has been observed for PLC-β1 andseveral RGS proteins (RGS4, RGSZ1, GAIP, and RGS10), and for several differentGα substrates (Gαi, Gαq, Gαz, Gαo). Inhibition, which appears primarily as anincrease inKm of >10-fold, can exceed 80%. IC50 values vary from 50–400 nM.Crystal structures indicate that the binding sites on Gαi for RGS4 and Gβγ overlap(41, 161), but inhibition of GAP activity by Gβγ varies among different GAPs inways inconsistent with simple competition (49, 81). It is likely that inhibition ofGAP activity by Gβγ reflects its direct interaction with both the Gα substrate andthe GAP itself.

The inhibition of GAP activity by Gβγ observed in single-turnover assaysseems inconsistent with the high GAP activities observed in steady-state GAPassays in which Gβγ is present at high local concentrations (47, 48, 56). Theseassays use receptor, heterotrimeric G protein, and GAP coreconstituted into uni-lamellar phospholipid vesicles, such that the calculated concentration of Gβγ atthe vesicle surface is≥20µM (50), well above the IC50determined in solution. Theapparent disagreement between these results may be reconciled by hypothesizingsome special orientational or permissive role of the lipid bilayer in the steady-state assay, but the problem is probably more fundamental, as discussed below.This paradox is reminiscent of the observation that Gβγ suppresses GDP/GTPexchange in isolation but is essentially required for receptor-catalyzed exchange(159); both observations suggest potentially biphasic regulation by Gβγ in cells.

The GGL Domain: Selective Association of RGSProteins with Gββ5

A major clue to the role of Gβγ in RGS action came from the finding by Snowet al (162) that several RGS proteins (RGS6, 7, 9, 11) contain a domain similar insequence to Gγ subunits, referred to as a GGL domain. They found that RGS11selectively binds Gβ5, a Gβ isoform that is found primarily in neural tissue andwhose sequence diverges notably from those of Gβ1–4 (163). RGS11 bound Gβ5when the proteins were either translated together in vitro or when the two cDNAswere cotransfected into Cos cells. Gβ5 also bound to an independently expressedGGL domain from RGS11; deletion of the GGL domain from RGS11 eliminatedbinding. Gβ5 mRNA is alternatively spliced to yield a partially soluble short form(Gβ5S) and a mostly membrane-associated long form (Gβ5L); both bound toRGS11 and its GGL domain (162). The physiologic validity of GGL-Gβ5 bind-ing was further substantiated by the observations that endogenous Gβ5 copurifieswith RGS7 from retina (164) and with RGS6 and RGS7 from brain (164a), thatrecombinant RGS11 and Gβ5 copurify from Sf9 cells (162), and that a mutantmouse whose RGS9 genes are disrupted does not accumulate Gβ5L in the retina(87).

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


The complex of RGS11 and Gβ5 displays GAP activity in vitro, with preferencefor Gαo as substrate (162). A complex of RGS9 and Gβ5L is evidently the principalGt GAP in photoreceptor cells (13, 87, 165). RGS6 and RGS7 are also selectiveGαo GAPs when combined with Gβ5 (166). The GAP activity displayed by thecomplex of RGS9 and Gβ5 can support agonist-stimulated steady-state GTPaseactivity when assayed in phospholipid vesicles coreconstituted with m2 muscariniccholinergic receptor and Gαo (but without added Gβγ ). This finding suggeststhat Gβ5 and the GGL domain can functionally replace a conventional Gβγ inpromoting receptor-Gα coupling (TK Harden & AG Gilman, unpublished).

The GGL domain is not required for G protein GAP activity in vitro (13, 162),however, and there is one report that Gβ5 inhibits the binding of RGS7 to Gαo.This result, which was obtained using tracer amounts of in vitro translated proteinsand which was not accompanied by data on RGS function, seems to be in con-flict with others (87, 162, 165). Although it may be incorrect, it is consistent withthe inhibition of GAP activity of other RGS proteins by Gβγ dimers discussedabove and may reflect a potential bidirectional action of Gβ that depends on assayconditions.

The detailed function of the GGL domain and binding to Gβ5 have been hardto ascertain because it has not been possible to isolate useful amounts either ofa GGL-containing RGS protein without Gβ5 or of Gβ5 alone. Both proteinsprobably associate shortly after translation, and each is crucial to the stability orinitial folding of the other. For example, retinas of RGS9−/− mice do not containdetectable Gβ5L despite the presence of a normal amount of its mRNA (87). Al-though Gβ5 does bind to other Gγ subunits upon overexpression of both, affinity islow and accumulation of Gβ5γ dimers is inefficient (168). Two mutations in GGLdomains have been reported to alter their selectivity among Gγ subunits (169, 170);interaction appears to be determined by multiple contacts between the extendedstructure of GGL domain (or Gγ ) and the face of the Gβ disc (161, 171). Althoughthe complex of Gβ5 and either RGS6 or RGS7 mediates receptor-Gα interactions,neither complex stimulates type II adenylyl cyclase, inhibits type I adenylyl cy-clase, or stimulates PLC-β2 (166). Conventional Gβγ dimers (Gβ1–4) displayeach of these activities. The inability of the RGS-Gβ5 complexes to regulate theseeffector proteins may reflect intrinsic differences between the Gβ5-GGL complexand other Gβγ dimers, as well as an inhibitory effect of the attached RGS domain.

A Model for Gββ-RGS Interactions: Evidence fromGββγγ -Gated K++ Channels

The implicit contradiction between the inhibition of GAP activity by Gβγ insolution and the high receptor-promoted GTPase activity observed in membranesthat contain Gβγ can potentially be explained if attachment of the proteins to aphospholipid bilayer blocks inhibition by Gβγ . A more interesting proposal isthat the association of Gβγ with Gα, receptor, or GAP is altered during receptor-stimulated GTPase turnover to make Gβγ an active regulatory participant in the

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

820 ROSS ¥ WILKIE

GTPase cycle. Examples of such behavior include the participation of Gβ5 andthe GGL domain in receptor-stimulated steady-state GTP hydrolysis (TK Harden& AG Gilman, unpublished) and the Gβγ -mediated activation of K+ channels.

Effects of RGS proteins on Gβγ -mediated regulation of inwardly rectifyingK+ channels (Gβγ -gated K+ channels or GIRKs) have provided insight into RGS-Gβγ signaling interactions by showing that GAPs can enhance the temporal acuityof signaling without diminishing signal amplitude. GIRKs are activated (opened)by Gβγ subunits (usually from heterotrimeric Gi; see 159, 172), but their kineticsof activation and deactivation are often sluggish when they are overexpressed inheterologous cells. Most markedly, channel closing is quite slow upon removal ofagonist, similar in rate to the hydrolysis of GTP by isolated Gα subunits. Doupniket al (88) and Saitoh et al (89) found that several RGS proteins increased thespeed at which GIRK channels close when an agonist is removed, consistentwith continuously accelerated hydrolysis of Gαi-bound GTP and consequentlyrapid sequestration of Gβγ . Surprisingly, RGS proteins attenuated the agonist-stimulated, steady-state K+ current only slightly, even though the deactivation ratewas increased>10-fold (see also 90). Such accelerated deactivation should haveinhibited the K+ current by 90% unless receptor-catalyzed binding was acceler-ated commensurately. The rate of channel activation upon addition of agonist wasincreased, although only about twofold (88), which suggests that GDP/GTP ex-change was much faster during continued stimulation than at its outset. BecauseRGS7/Gβ5 and RGS9/Gβ5 can mediate rapid activation and deactivation of GIRKchannels with little attenuation of amplitude (172a), it is likely that all the relevantproteins—receptor, Gα, RGS/Gβ5, and channel—remain associated in a signaltransducing complex over an extended time scale.

Accelerated deactivation upon removal of agonist without significant loss ofsignal amplitude is consistent with the idea that GAP activity allows formation ofa signaling complex of receptor, Gα, and GAP that rapidly binds and hydrolyzesGTP (26, 56, 88) (Figure 4). However, this model does not explain the increase inthe initial activation rate that is observed in the presence of RGS proteins (90). Inaddition to their GAP activities, the RGS proteins may act as scaffolds to maintainassociation of receptor and G protein prior to receptor activation. They may alsodecrease the affinity of Gα subunits for Gβγ and thus increase the availability ofGβγ upon activation (90, 91, 173). In support of this idea, B¨unemann & Hosey(173) reported that RGS3 and RGS4 by themselves produce modest Gβγ -mediatedactivation of GIRKs. Facilitated release of Gβγ by RGS protein and the inhibitionof RGS GAP activity by Gβγ (see above) may both be explained as negativelycooperative binding of RGS protein and Gβγ to Gα, such that each decreases theaffinity of Gα for the other. This conjecture clearly remains to be tested.

The behavior of Gβγ -regulated GIRK channels does not explain how Gβγ caninhibit a single round of GTP hydrolysis in solution without substantially inhibitingsteady-state GTPase activity in membranes. The answer to the latter question maybe that RGS proteins and Gβγ interact positively rather than negatively when bothproteins and Gα are bound to receptor in the presence of agonist. Both the positive

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


and negative interaction could operate using either a distinct Gβγ dimer or thecombination of a GGL domain and Gβ5, as suggested by the ability of a GGL/Gβ5complex to replace conventional Gβγ in both control of GIRK channels and inmediating rapid GTP hydrolysis (172a; TK Harden & AG Gilman, unpublished).Because GGL/Gβ5 performs these functions while covalently bound to the RGSdomain, we suggest that an RGS protein may not drive actual dissociation from Gα

but rather the reorientation of the various domains such that they, with receptor,could mediate rapid GDP/GTP exchange. In the case of the GIRKs, it is likelythat Gβγ does not completely dissociate from activated Gαi in order to promotechannel opening, but that the Gα-Gβγ interface changes during Gα activation toallow bound Gβγ to stimulate channel opening (174–177).

CELLULAR REGULATION BY RGS PROTEINS:A C. Elegans Model

What is known about the roles of endogenous RGS proteins in cellular signalingcomes primarily from genetic studies in mice, worms, and fungi. RGS9 acceleratestermination of the mammalian photoresponse following stimulation by light, whichsuggests that its major effect is to regulate signaling kinetics rather than to act as aninhibitor (87). In contrast, RGS proteins inSaccharomycesandNeurospora(Sst2pand RGS2; FlbA) apparently act as inhibitors, with Sst2p mediating a feedbackloop. Perhaps the most provocative studies of RGS physiology come from geneticanalysis of RGS regulation of Go and Gq pathways inC. elegans. Pioneering studiesidentified roles for RGS proteins in motility and egg laying, which are regulatedby the opposing activities of Gαo and Gαq and the RGS proteins, EGL-10 andEAT-16. Subsequent genetic analysis of worms with either gain-of-function orloss-of-function alleles in these Gα and RGS genes indicates an unexpected rolefor EAT-16 as an effector protein in the serotonin-stimulated Go signaling pathway.

Motility and egg laying are easily scored and are therefore well-studied behav-iors in worms. Gαo deficiency causes hyperkinesis; expression of activated Gαocauses sluggish locomotion, foraging and feeding, and delayed egg laying (19,178–180, 180a,b). EGL-10 is coexpressed with Gαo in the relevant cells and isprobably its attendant RGS protein. As expected for an inhibitor of Go signaling,deletion of the RGS geneEgl-10 caused a hypokinetic behavior similar to thatcaused by activated Gαo; EGL-10 was only effective in the presence of wild-typeGαo. Thus, the intensity of Gαo signaling was elevated by increasing the copynumber of the wild-type Gαo gene, by expressing a constitutively active allele ofgoa-1[Q205L]that is presumably refractory to GAP activity, or by reducing theactivity of EGL-10.

The Gαq gene,Egl-30, also opposes Gαo-mediated effects on motility and egglaying. The biochemical mechanism of antagonism between Go- and Gq-signalingremained enigmatic until Hajdu-Cronin et al (86) found that the phenotype of aconstitutively active Gαo mutant is suppressed by the loss of EAT-16, an RGS

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

822 ROSS ¥ WILKIE

protein of the R7 family. The behavior of EAT-16 suggested that it acts as a GqGAP, and expression of EAT-16 in Cos cells inhibited m1 muscarinic stimulation ofGq-mediated inositol phosphate production. Genetic analysis suggested that Gαoenhances the inhibitory activity of EAT-16 on Gq signaling; EAT-16 may be aneffector or downstream target of the Go pathway. Because EAT-16 contains a GGLdomain, it is tempting to speculate that the serotonin receptor may act through acomplex of Gαo, the EAT-16 GGL domain, and Gβ2 (theC. eleganshomologueof mammalian Gβ5). In this case, Go may stimulate EAT-16 directly to inhibit Gqsignaling inC. elegansneurons.

ACKNOWLEDGMENTS

We thank David Sierra for preparing Figures 1–3; Melanie Cobb, John Perkins, andmembers of our research groups for comments on the manuscript; and numerouscolleagues for sharing preprints and unpublished information.

Visit the Annual Reviews home page at www.AnnualReviews.org

LITERATURE CITED

1. Brandt DR, Ross EM. 1985.J. Biol. Chem.260:266–72

2. Cassel D, Eckstein F, Lowe M, Selinger Z.1979.J. Biol. Chem.254:9835–38

3. Breitwieser GE, Szabo G. 1988.J. Gen.Physiol.91:469–93

4. Vuong TM, Chabre M. 1991.Proc. Natl.Acad. Sci. USA88:9813–17

5. Vuong TM, Chabre M. 1990.Nature346:71–74

6. Jans DA, Peters R, Zsigo J, Fahrenholz F.1989.EMBO J.8:2481–88

7. Wagner R, Ryba N, Uhl R. 1988.FEBSLett.234:44–48

8. Arshavsky VY, Gray-Keller MP, BowndsMD. 1991.J. Biol. Chem.266:18530–37

9. Trahey M, McCormick F. 1987.Science238:542–45

10. Boguski MS, McCormick F. 1993.Nature366:643–54

11. Berstein G, Blank JL, Jhon D-Y, Exton JH,Rhee SG, et al. 1992.Cell 70:411–18

12. Arshavsky VY, Bownds MD. 1992.Nature357:416–17

13. He W, Cowan CW, Wensel TG. 1998.Neu-ron 20:95–102

14. Hart MJ, Jiang XJ, Kozasa T, RoscoeW, Singer WD, et al. 1998.Science280:2112–14

15. Kozasa T, Jiang XJ, Hart MJ, SternweisPM, Singer WD, et al. 1998.Science280:2109–11

16. Siderovski DP, Heximer SP, Forsdyke DR.1994.DNA Cell Biol.13:125–47

17. Dietzel C, Kurjan J. 1987.Mol. Cell. Biol.7:4169–77

18. De Vries L, Mousli M, Wurmser A, Far-quhar MG. 1995.Proc. Natl. Acad. Sci.USA92:11916–20

19. Koelle MR, Horvitz HR. 1996.Cell 84:115–25

20. Berman DM, Wilkie TM, Gilman AG.1996.Cell 86:445–52

21. Hunt TW, Fields TA, Casey PJ, Peralta EG.1996.Nature383:175–77

22. Watson N, Linder ME, Druey KM, KehrlJH, Blumer KJ. 1996.Nature383:172–75

23. Hepler JR, Berman DM, Gilman AG,Kozasa T. 1997.Proc. Natl. Acad. Sci. USA94:428–32

24. Berman DM, Gilman AG. 1998.J. Biol.Chem.273:1269–72

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


25. De Vries L, Farquhar MG. 1999.TrendsCell Biol. 9:138–44

26. Ross EM. 1995.Recent Prog. Horm. Res.50:207–21

27. De Vries L, Zheng B, Fischer T, ElenkoE, Farquhar MG. 2000.Annu. Rev. Phar-macol. Toxicol.40:235–71

28. Siderovski DP, Strockbine B, Behe CI.1999.CRC Crit. Rev. Biochem. Mol. Biol.34:215–51

29. Zheng B, De Vries L, Farquhar MG. 1999.Trends Biochem. Sci.24:411–14

30. Dohlman HG, Thorner J. 1997.J. Biol.Chem.272:3871–74

31. Wieland T, Chen C-K. 1999.Naunyn-Schmiedebergs Arch. Pharmacol.360:14–26

32. Hepler JR. 1999.Trends Pharmacol. Sci.20:376–82

33. Popov S, Yu K, Kozasa T, Wilkie TM.1997. Proc. Natl. Acad. Sci. USA94:7216–20

34. Faurobert E, Hurley JB. 1997.Proc. Natl.Acad. Sci. USA94:2945–50

35. Zeng W, Xu X, Popov S, MukhopadhyayS, Chidiac P, et al. 1998.J. Biol. Chem.273:34687–90

36. Siderovski DP, Hessel A, Chung S, MakTW, Tyers M. 1996.Curr. Biol. 6:211–12

37. Carman CV, Parent J-L, Day PW, ProninAN, Sternweis PM, et al. 1999.J. Biol.Chem.274:34483–92

38. Huang LJ-S, Durick K, Weiner JA, ChunJ, Taylor SS. 1997.Proc. Natl. Acad. Sci.USA94:11184–89

39. Deleted in proof40. Paulssen RH, Woodson J, Liu Z, Ross

EM. 1996. J. Biol. Chem.271:26622–29

41. Tesmer JJG, Berman DM, Gilman AG,Sprang SR. 1997.Cell 89:251–61

42. de Alba E, De Vries L, Farquhar MG,Tjandra N. 1999.J. Mol. Biol. 291:927–39

42a. Sowa ME, He W, Wensel TG, LichtargeO. 2000.Proc. Natl. Acad. Sci. USA97:1483–88

43. Scheffzek K, Ahmadian MR, Kabsch W,Wiesmuller L, Lautwein A, et al. 1998.Sci-ence277:333–38

44. Sprang SR. 1997.Annu. Rev. Biochem.66:639–78

45. Garrison TR, Zhang Y, Pausch M,Apanovitch D, Aebersold R, et al. 1999.J. Biol. Chem.274:36387–91

45a. Spink KE, Polakis P, Weis WI. 2000.EMBO J.19:2270–79

46. Gilman AG. 1987.Annu. Rev. Biochem.56:615–49

47. Mukhopadhyay S, Ross EM. 1999.Proc.Natl. Acad. Sci. USA96:9539–44

48. Posner BA, Mukhopadhyay S, Tesmer JJ,Gilman AG, Ross EM. 1999.Biochemistry38:7773–79

49. Wang J, Ducret A, Tu Y, Kozasa T, Ae-bersold R, et al. 1998.J. Biol. Chem.273:26014–25

50. Wang J, Tu Y, Mukhopadhyay S, Chidiac P,Biddlecome GH, et al. 1999. InG Proteins:Techniques of Analysis, ed. DR Manning,pp. 123–51. Boca Raton, FL: CRC Press

51. Deleted in proof52. Deleted in proof53. Deleted in proof54. Deleted in proof55. Ingi T, Krumins AM, Chidiac P, Brothers

GM, Chung S, et al. 1998.J. Neurosci.18:7178–88

56. Biddlecome GH, Berstein G, Ross EM.1996.J. Biol. Chem.271:7999–8007

57. Sprang SR. 1997.Science277:329–3058. Higashijima T, Ferguson KM, Sternweis

PC, Ross EM, Smigel MD, et al. 1987.J.Biol. Chem.262:752–56

59. Higashijima T, Ferguson KM, SternweisPC, Smigel MD, Gilman AG. 1987.J. Biol.Chem.262:762–66

60. Coleman DE, Berghuis AM, Lee E, Lin-der ME, Gilman AG, et al. 1994.Science265:1405–12

61. Sondek J, Lambright DG, Noel JP, HammHE, Sigler PB. 1994.Nature372:276–79

62. Chatterjee TK, Eapen AK, Fisher RA.1997.J. Biol. Chem.272:15481–87

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

824 ROSS ¥ WILKIE

63. Noel JP, Hamm HE, Sigler PB. 1993.Na-ture366:654–63

64. Lambright DG, Noel JP, Hamm HE, SiglerPB. 1994.Nature369:621–28

65. Mixon MB, Lee E, Coleman DE, BerghuisAM, Gilman AG, et al. 1995.Science270:954–60

66. Coleman DE, Sprang SR. 1999.J. Biol.Chem.274:16669–72

67. Raw AS, Coleman DE, Gilman AG, SprangSR. 1997.Biochemistry36:15660–69

68. Berman DM, Kozasa T, Gilman AG. 1996.J. Biol. Chem.271:27209–12

69. Chen C-K, Wieland T, Simon MI. 1996.Proc. Natl. Acad. Sci. USA93:12885–89

70. Chen CH, Zheng B, Han JH, Lin S-C. 1997.J. Biol. Chem.272:8679–85

71. Mittal R, Ahmadian MR, Goody RS, Wit-tinghofer A. 1996.Science273:115–17

72. Wang J, Tu YP, Woodson J, Song XL, RossEM. 1997.J. Biol. Chem.272:5732–40

73. Srinivasa SP, Watson N, Overton MC,Blumer KJ. 1998.J. Biol. Chem.273:1529–33

74. Natochin M, McEntaffer RL, ArtemyevNO. 1998.J. Biol. Chem.273:6731–35

75. Natochin M, Granovsky AE, ArtemyevNO. 1998.J. Biol. Chem.273:21808–15

76. Woulfe DS, Stadel JM. 1999.J. Biol. Chem.274:17718–24

77. DiBello PR, Garrison TR, Apanovitch DM,Hoffman G, Shuey DJ, et al. 1998.J. Biol.Chem.273:5780–84

78. Lan K-L, Sarvazyan NA, Taussig R,MacKenzie RG, DiBello PR, et al. 1998.J. Biol. Chem.273:12794–97

79. Tu YP, Wang J, Ross EM. 1997.Science278:1132–35

80. Glick JL, Meigs TE, Miron A, Casey PJ.1998.J. Biol. Chem.273:26008–13

81. Wang J, Frost JA, Cobb MH, Ross EM.1999.J. Biol. Chem.274:31641–47

82. Skiba NP, Yang C-S, Huang T, Bae H,Hamm HE. 1999.J. Biol. Chem.274:8770–78

83. Duncan JA, Gilman AG. 1998.J. Biol.Chem.273:15830–37

84. Natochin M, Artemyev NO. 1998.Bio-chemistry37:13776–80

85. Druey KM, Kehrl JH. 1997.Proc. Natl.Acad. Sci. USA94:12851–56

86. Hajdu-Cronin YM, Chen WJ, PatikoglouG, Koelle MR, Sternberg PW. 1999.Genes Dev.13:1780–93

87. Chen C-K, Burns ME, He W, Wensel TG,Baylor DA, et al. 2000.Nature.403:557–60

88. Doupnik CA, Davidson N, Lester HA,Kofuji P. 1997.Proc. Natl. Acad. Sci. USA94:10461–66

89. Saitoh O, Kubo Y, Miyatani Y, Asano T,Nakata H. 1997.Nature390:525–29

90. Chuang H-H, Yu M, Jan YN, JanLY. 1998. Proc. Natl. Acad. Sci. USA95:11727–32

91. Zerangue N, Jan LY. 1998.Curr. Biol. 8:313–16

92. Herlitze S, Ruppersberg JP, Mark MD.1999.J. Physiol.517:341–52

93. Levitzki A. 1981. CRC Crit. Rev.Biochem.10:81–12

94. Snow BE, Hall RA, Krumins AM, Broth-ers GM, Bouchard D, et al. 1998.J. Biol.Chem.273:17749–55

95. Wang L-H, Kalb RG, Strittmatter SM.1999.J. Biol. Chem.274:14137–46

96. Pawson T, Scott JD. 1997.Science278:2075–80

97. Scholich K, Mullenix JB, Wittpoth C,Poppleton HM, Pierre SC, et al. 1999.Sci-ence283:1328–31

98. Arshavsky VY, Pugh EN Jr. 1998.Neuron20:11–14

99. Tsang SH, Burns ME, Calvert PD, GourasP, Baylor DA, et al. 1998.Science282:117–21

100. Krupnick JG, Benovic JL. 1998.Annu.Rev. Pharmacol. Toxicol.38:289–319

101. Pitcher JA, Freedman NJ, Lefkowitz RJ.1998.Annu. Rev. Biochem.67:653–92

102. Ross EM, Berstein G, Karandikar M.1994. In Endothelium-Derived Factorsand Vascular Functions,ed. T Masaki, pp.211–18. Amsterdam: Elsevier Sci.

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


103. Essen L-O, Perisic O, Cheung R, KatanM, Williams RL. 1996.Nature380:595–602

104. Rhee SG, Choi KD. 1992.Adv. SecondMessenger Phosphoprotein Res.26:35–61

105. Singer WD, Brown HA, Sternweis PC.1997.Annu. Rev. Biochem.66:475–509

106. Park D, Jhon D-Y, Lee C-W, Ryu SH,Rhee SG. 1993.J. Biol. Chem.268:3710–14

107. Wu D, Jiang H, Katz A, Simon MI. 1993.J. Biol. Chem.268:3704–9

108. Wang T, Pentyala S, Elliott J, Dowal L,Gupta E, et al. 1999.Proc. Natl. Acad.Sci. USA96:7843–46

109. Beekman A, Helfrich B, Bunn PA Jr,Heasley LE. 1998.Cancer Res.58:910–13

110. Van Aelst L, D’Souza-Schorey C. 1997.Genes Dev.11:2295

111. Zohn IM, Campbell SL, Khosravi-Far R,Rossman KL, Der CJ. 1998.Oncogene17:1415–38

112. Seasholtz TM, Majumdar M, Brown JH.1999.Mol. Pharmacol.55:949–56

113. Buhl AM, Johnson NL, DhanasekaranN, Johnson GL. 1995.J. Biol. Chem.270:24631–34

114. Fukuhara S, Murga C, Zohar M, IgishiT, Gutkind JS. 1999.J. Biol. Chem.274:5868–79

115. Angleson JK, Wensel TG. 1993.Neuron11:939–49

116. Angleson JK, Wensel TG. 1994.J. Biol.Chem.269:16290–96

117. Slepak VZ, Artemyev NO, Zhu C,Dumke CL, Sabacan L, et al. 1995.J.Biol. Chem.270:14319–24

117a. Nekrasova ER, Berman DM, RustlandiRR, Hamm HE, Gilman AG, et al. 1997.Biochemistry36:7638–43

117b. Natochin M, Granovsky AE, ArtemyevNO. 1997.J. Biol. Chem.272:17444–49

117c. McEntaffer RL, Natochin M, ArtemyevNO. 1999.Biochemistry38:4931–37

118. Versele M, de Winde JH, Thevelein JM.1999.EMBO J.18:5577–91

119. De Vries L, Elenko E, Hubler L, JonesTLZ, Farquhar MG. 1996.Proc. Natl.Acad. Sci. USA93:15203–8

120. Heximer SP, Watson N, Linder ME,Blumer KJ, Hepler JR. 1997.Proc. Natl.Acad. Sci. USA94:14389–93

121. Heximer SP, Srinivasa SP, Bernstein LS,Bernard JL, Linder ME, et al. 1999.J.Biol. Chem.274:34253–59

122. Druey KM, Blumer KJ, Kang VH, KehrlJH. 1996.Nature379:742–46

123. Beadling C, Druey KM, Richter G,Kehrl JH, Smith KA. 1999.J. Immunol.162:2677–82

124. Zhang Y, Neo SY, Han J, Yaw LP, LinS-C. 1999.J. Biol. Chem.274:2851–57

125. Diverse-Pierluissi MA, Fischer T, JordanJD, Schiff M, Ortiz DF, et al. 1999.J. Biol.Chem.274:14490–94

126. Xu X, Zeng W, Popov S, Berman DM,Davignon I, et al. 1999.J. Biol. Chem.274:3549–56

127. Deleted in proof128. Gold SJ, Ni YG, Dohlman HG, Nest-

ler EJ. 1997. J. Neurosci. 17:8024–37

129. Ni YG, Gold SJ, Iredale PA, TerwilligerRZ, Duman RS, et al. 1999.J. Neurosci.19:3674–80

130. Burchett SA, Volk ML, Bannon MJ,Granneman JG. 1998.J. Neurochem.70:2216–19

131. Burchett SA, Bannon MJ, Granneman JG.1999.J. Neurochem.72:1529–33

132. Pepperl DJ, Shah-Basu S, VanLeeuwenD, Granneman JG, MacKenzie RG.1998.Biochem. Biophys. Res. Commun.243:52–55

133. Benzing T, Brandes R, Sellin L, Scher-mer B, Lecker S, et al. 1999.Nat. Med.5:913–18

134. Zhang S, Watson N, Zahner J, RottmanJN, Blumer KJ, et al. 1998.J. Mol. CellCardiol. 30:269–76

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26

826 ROSS ¥ WILKIE

135. Ogier-Denis E, Petiot A, Bauvy C,Codogno P. 1997.J. Biol. Chem.272:24599–603

136. Buckbinder L, Velasco-Miguel S, ChenY, Xu N, Talbott R, et al. 1997.Proc.Natl. Acad. Sci. USA94:7868–72

137. Tseng C-C, Zhang X-Y. 1998.En-docrinology139:4470–75

138. Kardestuncer T, Wu H, Lim AL, NeerEJ. 1998.FEBS Lett.438:285–88

139. Kim E, Arnould T, Sellin L, Benzing T,Comella N, et al. 1999.Proc. Natl. Acad.Sci. USA96:6371–76

140. Yue C, Dodge KL, Weber G, SanbornBM. 1998. J. Biol. Chem.273:18023–27

141. Filtz TM, Cunningham ML, Stanig KJ,Paterson A, Harden TK. 1999.Biochem.J. 338:257–64

142. Carlson KE, Brass LF, Manning DR.1989.J. Biol. Chem.264:13298–305

143. Barovsky K, Brooker G. 1980.J. Cycl.Nucleotide Res.6:297–307

144. Kozasa T, Gilman AG. 1996.J. Biol.Chem.271:12562–67

145. Srinivasa SP, Bernstein LS, Blumer KJ,Linder ME. 1998.Proc. Natl. Acad. Sci.USA95:5584–89

146. Druey KM, Ugur O, Caron JM, ChenC-K, Backlund PS, et al. 1999.J. Biol.Chem.274:18836–42

147. Tu YP, Popov S, Slaughter C, Ross EM.1999.J. Biol. Chem.274:3860–67

148. Deleted in proof149. Deleted in proof150. Deleted in proof151. Dunphy JT, Linder ME. 1998.Biochim.

Biophys. Acta1436:245–61152. Wedegaertner PB, Wilson PT, Bourne

HR. 1995.J. Biol. Chem.270:503–6153. Ross EM. 1995.Curr. Biol. 5:107–9154. Kishida S, Yamamoto H, Ikeda S,

Kishida M, Sakamoto I, et al. 1998.J.Biol. Chem.273:10823–26

154a. Siderovski DP, Diverse-Pierluissi MA,De Vries L. 1999.Trends Biochem. Sci.24:340–41

154b. Ponting CP. 1999.J. Mol. Med. 77:695–98

154c. De Vries L, Lou X, Zhao G, Zheng B,Farquhar MG. 1998.Proc. Natl. Acad.Sci. USA95:12340–45

155. Fagotto F, Jho E, Zeng L, Kurth T, JoosT, et al. 1999.J. Cell Biol. 145:741–56

156. Rubinfeld B, Robbins P, El-Gamil M,Albert I, Porfiri E, et al. 1997.Science275:1790–92

157. Zeng L, Fagotto F, Zhang T, Hsu W,Vasicek TJ, et al. 1997.Cell 90:181–92

158. Zhang Y, Neo SY, Wang X, Han J, LinS-C. 1999.J. Biol. Chem.274:35247–54

159. Clapham DE, Neer EJ. 1997.Annu. Rev.Pharmacol. Toxicol.37:167–203

160. Chidiac P, Ross EM. 1999.J. Biol.Chem.274:19639–43

161. Wall MA, Coleman DE, Lee E, I˜niguez-Lluhi JA, Posner BA, et al. 1995.Cell83:1047–58

162. Snow BE, Krumins AM, Brothers GM,Lee S-F, Wall MA, et al. 1998.Proc.Natl. Acad. Sci. USA95:13307–12

163. Watson AJ, Aragay AM, Slepak VZ,Simon MI. 1996.J. Biol. Chem.271:28154–60

164. Cabrera JL, de Freitas F, Satpaev DK,Slepak VZ. 1998.Biochem. Biophys.Res. Commun.249:898–902

164a. Zhang JH, Simonds WF. 2000.J. Neu-rosci.20:RC59(1-5)

165. Makino ER, Handy JW, Li T, ArshavskyVY. 1999. Proc. Natl. Acad. Sci. USA96:1947–52

166. Posner BA, Gilman AG, Harris BA.1999.J. Biol. Chem.274:31087–93

167. Deleted in proof168. Lindorfer MA, Myung CS, Savino Y,

Yasuda H, Khazan R, et al. 1998.J. Biol.Chem.273:34429–36

169. Snow BE, Betts L, Mangion J, Sondek J,Siderovski DP. 1999.Proc. Natl. Acad.Sci. USA96:6489–94

P1: FPP/FPW P2: FPP

July 13, 2000 11:13 AR102 CHAP26


170. Levay K, Cabrera JL, Satpaev DK,Slepak VZ. 1999.Proc. Natl. Acad. Sci.USA96:2503–7

171. Lambright DG, Sondek J, Bohm A,Skiba NP, Hamm HE, et al. 1996.Na-ture379:311–19

172. Clapham DE, Neer EJ. 1993.Nature365:403–6

172a. Kovoor A, Chen CK, He W, Wensel TG,Simon MI, et al. 2000.J. Biol. Chem.275:3397–402

173. Bunemann M, Hosey MM. 1998.J. Biol.Chem.273:31186–90

174. Huang C-L, Slesinger PA, Casey PJ, JanYN, Jan LY. 1995.Neuron15:1133–43

175. Yamada M, Inanobe A, Kurachi Y. 1998.Pharmacol. Rev.50:723–57

176. Schreibmayer W, Dessauer CW, Voro-biov D, Gilman AG, Lester HA, et al.1996.Nature380:624–27

177. Dascal N. 1997.Cell. Signal.9:551–73178. Mendel JE, Korswagen HC, Liu KS,

Hajdu-Cronin YM, Simon MI, et al.1995.Science267:1652–55

179. Segalat L, Elkes DA, Kaplan JM. 1995.Science267:1648–51

180. Bargmann CI, Kaplan JM. 1998.Annu.Rev. Neurosci.21:279–308

180a. Lackner MR, Nurrish SJ, Kaplan JM.1999.Neuron24:335–46

180b. Miller KG, Emerson MD, Rand JB.1999.Neuron24:323–33

181. Tolkovsky AM, Levitzki A. 1978.Bio-chemistry17:3795–810

gtp -activating proteins for heterotrimeric … 2000 annurev.pdf · gtp hydrolysis rates prompted...

Documents