Endocrine Reviews 24 (6): 765-781
Copyright © 2003 by The Endocrine Society
Insights into G Protein Structure, Function, and Regulation
Theresa M. Cabrera-Vera,
Jurgen Vanhauwe,
Tarita O. Thomas,
Martina Medkova,
Anita Preininger,
Maria R. Mazzoni and
Heidi E. Hamm
Department of Molecular Pharmacology and Biological Chemistry (T.M.C.-V., J.V., T.O.T., M.M., A.P., H.E.M.), Institute for Neuroscience, Northwestern University, Chicago, Illinois 60611; Department of Molecular Pharmacology (M.R.M., H.E.M.), Vanderbilt University Medical Center, Nashville, Tennessee 37232; and Department of Psychiatry, Neurobiology, Pharmacology, and Biotechnology (M.R.M.), University of Pisa, Pisa, 56057, Italy
Correspondence: Address all correspondence and requests for reprints to: Heidi E. Hamm, Ph.D., Department of Pharmacology, Vanderbilt University Medical Center, 442 Robinson Research Building, 23rd and Pierce Drive, Nashville, Tennessee 37232. E-mail: heidi.hamm{at}mcmail.vanderbilt.edu
 |
Abstract
|
|---|
In multicellular organisms from Caenorhabditis elegans to Homo sapiens, the maintenance of homeostasis is dependent on the continual flow and processing of information through a complex network of cells. Moreover, in order for the organism to respond to an ever-changing environment, intercellular signals must be transduced, amplified, and ultimately converted to the appropriate physiological response. The resolution of the molecular events underlying signal response and integration forms the basis of the signal transduction field of research. An evolutionarily highly conserved group of molecules known as heterotrimeric guanine nucleotide-binding proteins (G proteins) are key determinants of the specificity and temporal characteristics of many signaling processes and are the topic of this review. Numerous hormones, neurotransmitters, chemokines, local mediators, and sensory stimuli exert their effects on cells by binding to heptahelical membrane receptors coupled to heterotrimeric G proteins. These highly specialized transducers can modulate the activity of multiple signaling pathways leading to diverse biological responses. In vivo, specific combinations of G
- and Gß
-subunits are likely required for connecting individual receptors to signaling pathways. The structural determinants of receptor-G protein-effector specificity are not completely understood and, in addition to involving interaction domains of these primary acting proteins, also require the participation of scaffolding and regulatory proteins.
- I. Introduction
- II. G Protein Structure
- III. Molecular Basis for G Protein Activation
- IV. Structural Determinants of Receptor-G Protein Specificity
- V. Receptor-Independent Activators of G Protein Signaling (AGS Proteins)
- VI. The Receptor-G Protein Interface as a Therapeutic Target
- VII. G Interaction with Effectors
- VIII. Gß Interaction with Effectors
- A. Gß-Dimer composition directs effector and receptor coupling
- B. Structural determinants of effector specificity
- C. Novel Gß-effectors
- D. Additional role for Gß5
- IX. Molecular Basis for G Protein Inactivation
- A. Intrinsic GTPase activity
- B. G Interaction with GTPase-activating proteins (GAPs)
- X. Regulation of G Protein Function by Covalent Modification
- A. G protein lipidation
- B. The role of lipid modifications in G protein membrane association and consequent signaling functions
- XI. Advances for the Future: Investigating the Dynamic Nature of G Protein Signaling
|
I. Introduction
|
|---|
WHEN A LIGAND such as a hormone, neurotransmitter, or glycoprotein interacts with a heptahelical receptor on the surface of the cell, the ligand either stabilizes or induces a conformation in the receptor that activates a heterotrimeric G protein (composed of -, ß-, and -subunits) on the inner membrane surface of the cell (1). In the inactive heterotrimeric state, GDP is bound to the G-subunit. Upon activation, GDP is released, GTP binds to G, and subsequently G-GTP dissociates from Gß and from the receptor (Fig. 1
). Both G-GTP and Gß are then free to activate downstream effectors. The duration of the signal is determined by the intrinsic GTP hydrolysis rate of the G-subunit and the subsequent reassociation of G-GDP with Gß (1, 2). This article will review current knowledge and recent progress in defining the molecular mechanisms that regulate the activity and specificity of G protein signaling cascades. In addition, we will briefly discuss the use of dynamic experimental approaches that are likely to provide new insights into G protein regulation in the future.
|
II. G Protein Structure
|
|---|
The solution of crystal structures for inactive (GDP-bound), active (GTP-bound), and transition state (GDP-ALF4-) Gt (3, 4, 5) or Gi (6), as well as structures for the inactive heterotrimeric complexes (7, 8), has provided the framework for understanding the biomechanics of G proteins as molecular switches. For a detailed discussion of the specific intramolecular contacts within heterotrimeric G proteins, the reader is referred to reviews in Refs. 2 and 9 . Herein, we will briefly discuss key structural and functional features common to all heterotrimeric G proteins to understand the framework and interpretation of recent work in this field.
According to current knowledge, 16 genes encode for G-subunits, five genes encode for Gß-, and 12 genes encode for G-subunits (10). Classically, G proteins are divided into four families based on similarity of their -subunits: Gi/o, Gs, Gq/11, and G12/13 (Table 1). G-subunits contain two domains: a GTPase domain that is involved in the binding and hydrolysis of GTP and a helical domain that buries the GTP within the core of the protein (Fig. 2A). The helical domain is the most divergent domain among G families and may play a role in directing specificity of receptor- and effector-G protein coupling. Comparison of Gt-GDP and Gt-GTPS crystal structures has revealed the presence of three flexible regions, designated switches I, II, and III, which become more rigid and well ordered in the GTP-bound active conformation (3, 4). Little is known about the structure of the extreme amino (N-) and carboxy (C-) terminal domains of G-subunits because in the isolated G protein crystal structures solved thus far, the N and C termini of G were either removed from the protein or disordered (3, 4, 5, 6). However, in two separate crystal structures of heterotrimeric complex, the N-terminal helix is ordered by its interaction with the ß-propeller domain of Gß (Refs. 7 and 8 and Fig. 2A). Biochemical studies suggest that these terminal regions play a key role in the activation process and in directing specific protein-protein interactions, as is discussed in the following section.
View larger version (52K):
[in this window]
[in a new window]
|
FIG. 2. Schematic diagrams of Gt, the RGS4-Gi1 complex, and effector contact sites on Gß. A, Ribbon diagrams depicting the probable membrane orientation of heterotrimeric Gt. The refined rhodopsin structure is from Ref. 240 [Protein databank (PDB) file 1F88]. Gt (purple); GDP molecule (red); Gß (green); G (yellow); rhodopsin helices [color gradient from red (N terminus) to navy blue (C terminus]; the retinal molecule within rhodopsin (magenta). Diagrams were generated using coordinates from PDB files (1GOT and 1BOK) and visualized with WebLab ViewerPro. B, Ribbon diagram depicting the RGS4/Gi1 complex. RGS 4 (green); Gi1. (red); Gi1 ß-sheets (cyan); GTP molecule (magenta). Diagrams were generated using coordinates from PDB file 1AGR and visualized with WebLab ViewerPro. C, Solvent-accessible surface model of Gß11 highlighting residues identified as important mediators of effector interaction (26 ). The crystal coordinates of Gß11 [PDB entry 1TBG] were used to generate a surface model of the dimer in Graphical Representation and Analysis of Structural Properties. Gß (gray); G (pink). The area on Gß that is covered by G in the G protein heterotrimer crystal structure is highlighted in light green. The effector-interacting residues on Gß are circled with colored dashed lines as follows: ß-adrenergic receptor kinase (orange); PLCß2 (red); AC II (green); K+ channel (blue); Ca2+ channel (yellow). G-GDP, when bound to Gß, covers all these distinct yet partially overlapping effector interaction regions on Gß and, thus, blocks Gß regulation of all the effectors. [Figure 2C reprinted with permission from Ford et al.: Science 280:1271–1274, 1998 (26 ). © 1988 American Association for the Advancement of Science.]
|
|
The Gß-subunit of heterotrimeric G proteins has a ß-propeller structure containing seven WD-40 repeats (Ref. 7 and Fig. 2A). The G-subunit interacts with the Gß-subunit through an N-terminal coiled coil and makes extensive contacts along the base of the Gß-subunit (Ref. 7 and Fig. 2A). The Gß-dimer binds to a hydrophobic pocket present in G-GDP. GTP binding to G removes the hydrophobic pocket and reduces the affinity of G for Gß (4).
|
III. Molecular Basis for G Protein Activation
|
|---|
The rate-limiting step in G protein activation is the release of GDP from the nucleotide-binding pocket. GDP is spontaneously released from the heterotrimeric G protein at a rate that varies depending on the G-subunit. For example, the Go GDP release rate (koff) is 0.19 min-1 whereas the Gi2 release rate is 0.072 min-1 (11). However, the inactive state of the G-subunits is controlled by Gß binding. Higashijima et al. (12) showed that in the absence of Mg2+, Gß increases the affinity of Go for GDP about 300-fold. GDP release is greatly facilitated by receptor activation of the G protein (13). Mutations (14, 15, 16) of residues in the critical TCAT guanine nucleotide-binding motif present in the ß6-5 loop of the GTPase domain (4) enhance receptor-independent spontaneous GDP release. Iiri et al. (14) identified such an activating mutation in Gs (A366S) in male patients with pseudohypoparathyroidism and gonadotropin-independent precocious puberty. Enhanced GDP release was also observed when similar mutations were generated in Gi [A326S; (15)] and Go [C325S; (16)] suggesting that this region serves as a common mediator of GDP release. Posner et al. (15) also demonstrated that GDP release can occur without inducing a large conformational change in G.
In addition to the TCAT motif, most recent work has identified residues within the helical domain as well as within the N- and C-terminal domains of G-subunits that are also integral mediators of spontaneous GDP release. For example, in Gt mutation of three residues located in the inward-facing surface of the 5-helix causes a dramatic increase of basal nucleotide exchange rate in addition to enhanced receptor-catalyzed nucleotide exhange rate (17). Mutation of five residues within the switch IV helical domain in Gs decreases the rate of GDP release, GTPS binding, and GTP hydrolysis (18) and disruption of contacts between the helical and GTPase domains also influences basal GDP dissociation rates (19, 20). By fluorescently labeling the C-terminal residue Cys347 of a Gt/Gi chimera, Yang et al. (21) determined that the C terminus moves into a more hydrophobic environment upon AlF4- activation. The authors suggest that this movement may reflect an interaction between the C terminus and the 2-ß4 loop of Gt/Gi. In addition, these divergent terminal domains have been implicated as the source of variation in the intrinsic GDP release rates among G-subunits. Substitution of 31 N-terminal residues of a Gt/i chimera (low intrinsic exchange rate) with corresponding 42 residues of Gs (high intrinsic exchange rate) significantly enhanced the nucleotide exchange rate (22). This same group also reported that disruption of a specific contact between Val30 (N terminus) and Ile339 (C terminus) alters the rate of GTPS binding, which was inferred as an indirect index of GDP release. Hence, structural interactions between N and C termini of Gt are important to the maintenance of a slow GDP release rate for Gt.
Receptor-mediated GDP release is dependent on the ability of the receptor to interact with the G protein and trigger conformational changes in G that cause release of GDP. Comparing the crystal structure with biochemical data, we can deduce that the receptor contacts G at a site that is more than 20 Å away from the guanine nucleotide binding site (1, 21), thus working at a distance to release GDP. Current theory is that receptor contact with the C terminus of the G-subunit leads to conformational changes that are propagated through G to the GDP binding site (1, 21). However, the requirement for Gß in receptor-G protein interaction and G protein activation suggests that Gß may actively participate in GDP release by opening an exit route for the guanine nucleotide to leave the complex (23). The heterotrimer contains a prominent cavity between G and Gß that is believed to be oriented toward the plasma membrane (7, 24). Activated loops of the receptor might use this cavity to tilt Gß away from G causing the contacts between G and Gß to be disrupted, including contacts near switch I and the ß3-2 loop in G, the potential exit route for the nucleotide. In this way the receptor could use Gß as a lever to release GDP (25). Ala substitutions in Gß at the G-Gß interface near the GDP exit route inhibit receptor-induced GDP/GTP exchange without affecting G-Gß binding (26). Thus, the Gß-dimer is not merely a passive binding partner with the sole purpose of stabilizing G but, rather, Gß actively participates in receptor-mediated G protein activation.
|
IV. Structural Determinants of Receptor-G Protein Specificity
|
|---|
For the purposes of this review, we will limit ourselves to a discussion of regions within G and Gß that have been determined to mediate receptor-G protein specificity. For a thorough review of specific sites on heptahelical receptors, which direct receptor-G protein coupling specificity, the reader is referred to Refs. 27 and 28 for reviews. The extreme C terminus of G (in particular the last five residues) has been established as an important mediator of receptor-G protein interaction (23, 29, 30, 31). For example, ADP ribosylation of residue -4 by pertussis toxin uncouples Gi/Go proteins from receptors (32). Phosphorylation of a tyrosine residue at -4 in Gq/11 was shown to be required for coupling to metabotropic glutamate receptors (33) although this has not been demonstrated in intact cells. In addition, the requirement of phosphorylation at the tyrosine residue of Gq/11 cannot be generalized as M1 muscarinic receptors, and thrombin receptors were shown to couple readily to Gq/11 proteins in reconstitution experiments (34, 35). Many examples of mutations in this region that alter receptor-G protein specificity have been also reported (36, 37, 38). In addition, several investigators have generated sequence-specific C-terminal peptides or antibodies targeting the C-terminal domain to study receptor-G protein interaction. Antibodies recognizing G C-terminal domains block receptor-G protein signaling (39). Instead, sequence-specific C-terminal synthetic peptides either stabilize the active agonist-bound form of the receptor mimicking the G protein (40, 41, 42) or serve as competitive inhibitors of receptor-G protein interface (43). Although blocking peptides are commonly interpreted as evidence of a direct receptor-G protein contact site, peptides may also stabilize or disrupt regions of the protein that transmit conformational changes to the guanine nucleotide binding motif and thereby indirectly affect receptor-mediated G protein activation.
The C terminus is not the only region directing receptor-G protein interactions. Several G-subunits possess identical or nearly identical residues within the extreme C-terminal domain yet exhibit differential coupling to receptors. For example, within the last 11 amino acids of Gi1 and Gt, only a single residue is divergent, yet the serotoninIB receptor fails to couple to Gt and readily couples to Gi. Investigation into the molecular determinants of this specificity indicated that two residues within the 4-helix of Gi1 are critical mediators of this receptor-G protein coupling profile (44, 45). Key residues for coupling specificity have been also identified within the N terminus (36, 39, 46), the 2-helix, and 2-ß4 loop regions (47, 48) as well as within the 4-helix and 4-ß6 loop domain (44, 48, 49). Segments of Gß- and G-subunits may also contribute to the receptor interacting surface of heterotrimers (46, 50, 51, 52, 53). Using a peptide specific for -helical residues in Gs, Krieger-Brauer et al. (54) blocked ß-adrenergic receptor-mediated activation of both Gs- and Gß-effectors. In contrast, a C-terminal sequence-specific peptide for Gs only prevented Gs-mediated effector activation, suggesting that the extreme C terminus of Gs is required for G-mediated signaling but is not critical for ß-adrenergic receptor recognition and dissociation of G from Gß (54). Together, these studies suggest that the relative importance of the C terminus for directing receptor-G protein interactions may be dependent on G and receptor subtypes. Receptor-G protein specificity is clearly not mediated solely by one structural feature of G-subunits but appears to result from a network of specific contacts between the receptor and G protein which differs for each G-subunit and for each receptor and results in a large number of possible combinations that can bring remarkable specificity into a system with only a few central players. As suggested by Blahos et al. (36), one of the difficulties in isolating the specific determinants of receptor-G protein coupling has been that G protein coupling may still occur even when interactions at certain contact points are weak, absent, or negative if these frailties can be overcome by a stronger interaction at other contact points or when regions that may weaken coupling are removed from either the receptor or the G protein.
|
V. Receptor-Independent Activators of G Protein Signaling (AGS Proteins)
|
|---|
A novel class of signaling proteins, termed AGS proteins, has been identified (55, 56). AGS proteins activate heterotrimeric G proteins independently of receptor activation. The mechanism for AGS activation differs among members of this family. AGS1 has been found experimentally to promote GTPS binding. AGS2 selectively associates with Gß, whereas AGS3 binds to G and exhibits a preference for GDP-G vs. GTP-G. AGS3 has been shown to prevent the reassociation of Gß with the G-subunit and function as a guanine dissociation inhibitor for Gi-subunits (57). AGS3 contains a G protein-regulatory motif. This G protein regulatory motif or GoLOCO repeat is an approximately 20-amino acid domain found in several proteins that interact with and/or regulate G proteins, e.g., AGS3, the Partner of Inscuteable and its mammalian homolog, LGN, Purkinje cell protein 2, and Rap1 GTPase-activating protein (GAP). The physiological role of these proteins in vivo remains to be determined, but one possible role for these proteins may be in the regulation of G proteins that do not reside near the plasma membrane and cannot be activated directly by receptors, e.g., G proteins in the Golgi that regulate vesicular trafficking (58). Little is known about the role of this pool of G proteins, and the discovery of AGS proteins may stimulate research into a new dimension of heterotrimeric G protein signaling.
|
VI. The Receptor-G Protein Interface as a Therapeutic Target
|
|---|
Traditionally, the extracellular surface and transmembrane domains of G protein-coupled receptors have served as a target for the development of drugs that can selectively activate or inactivate specific cellular pathways. However, some receptor isoforms, such as the dopamine D2L and D2S receptors, and the D4 receptor variants differ only on the intracellular surface of the protein (59, 60) and cannot be readily distinguished by targeting the ligand-binding site. Moreover, many receptors promiscuously couple to several G protein subtypes in what may be a tissue- or cell-specific phenomenon. Therefore, additional therapeutic targets will certainly be required to more specifically influence intracellular signaling events. One avenue being explored by our laboratory and others is the use of peptide inhibitors that target the receptor-G protein interface (43, 61, 62). Currently, these peptides represent either G C-terminal-specific sequences or peptides isolated from a combinatorial library based on C-terminal G-sequences and screened for high-affinity receptor binding (31). These studies are based on the idea that the C terminus of G-subunits serves as a key receptor contact site and mediator of receptor-G protein specificity. In the short term, these peptides may provide useful tools for exploring specificity of G protein-mediated signaling.
The delivery of peptide inhibitors represents a challenge to the therapeutic use of these tools. Possible delivery systems include the use of inducible retroviral minigene vectors (64), incorporation of peptides into liposomes (65), or the fusion of peptides to a viral peptide sequence that carries the C-terminal peptide into the cell (66). Alternatively, peptidomimetics may prove to be more stable and bioavailable. Selective targeting to specific organs is likely to prove beneficial, because Akhter et al. (67) have demonstrated that transgenic mice selectively expressing a Gq C-terminal minigene in the myocardium exhibit a marked inhibition of 1B-adrenergic receptor-mediated inositol phosphate production and blockade of cardiac hypertrophy. The identification of peptide inhibitors with high affinity for specific receptor subtypes and/or variants would also allow for more selective inhibition of signaling pathways. Despite the significant hurdles, targeting the receptor-G protein interface will clarify the complex coordination of players in signaling cascades and may prove therapeutically useful in the future.
|
VII. G Interaction with Effectors
|
|---|
Once G-GTP has dissociated from the Gß-dimer, G can directly interact with effector proteins to continue the signaling cascade. The specific effector proteins activated by G are dependent on the G-subtype and are summarized in Table 1. Well-defined G effectors, such as adenylyl cyclase (AC) and phospholipase C (PLC), have been the topic of several excellent reviews (68, 69).
Overall, several patterns emerge upon examination of the G-effectors. First, each G-family activates a distinct profile of effectors. The molecular basis for this divergence has not been completely elucidated. Cocrystallization studies of Gs and the catalytic domains of AC have identified specific contacts within Gs at the 2-helix (SII) and the 3-ß5 loop (70). In addition, the 4-ß6 loop of Gs also plays a role in AC activation (71). Sunahara et al. (72) demonstrated that GDP-bound Gs can also stimulate AC, albeit with a lower potency than the GTP-bound -subunit. These data are intriguing because they suggest that reassociation of G with Gß is required for the complete termination of Gs signaling, and inhibition of reassociation could prolong both G- and Gß-mediated signaling. In addition, Gß could serve to prolong signaling because it can block the PLCß-mediated acceleration of GTPase activity at Gq-proteins (73). Second, within a family, each G-subunit exhibits a differential profile of effector activation. For example, Gi2 is required to inhibit forskolin-stimulated AC activity whereas Gi3 serves to inhibit Gs-activated AC (74). In addition, 1-adrenergic receptors elevate intracellular Ca2+ by two distinct mechanisms that are dependent on the G-subunit coupled to the receptor: Gq releases Ca2+ from the endoplasmic reticulum whereas G11 activates a nonselective cation channel (75). Third, some G-subunits have only one identified effector, such as cGMP phosphodiesterase for Gt whereas others more promiscuously couple to several effector proteins. Lastly, effectors for some G-subunits have yet to be definitively identified, and the search for novel G-effectors is a rapidly growing area of research. A number of proteins that directly interact with G-subunits have been identified, yet further evidence awaits as to whether guanine nucleotide binding to G regulates the activity of these proteins in vivo in response to receptor activation. Nonetheless, some of the most recent studies identifying novel putative G-effectors are discussed below.
Using a yeast-two-hybrid screen, Jordan et al. (76) identified direct interactions between Go and Rap1 GAP, Gz GAP, and RGS17. This group also determined that Rap1 GAP interacts with Gi-proteins but not with Gq or Gs. However, receptor-mediated activation of these proteins was not demonstrated. Interestingly, Rap1 GAP interacted preferentially with GDP-bound Go, suggesting that Go-GDP may sequester Rap1 GAP away from Rap1, resulting in a sustained activation of MAPK. These findings reveal a novel mechanism of G protein function that is dependent on GDP-liganded G proteins. Gß-subunits might then be considered as inhibitors of G-GDP proteins (and vice versa).
In search of Gz-effectors, Chen et al. (77) screened a cDNA expression library using phosphorylated Gz-GTPS as a probe. This group identified two proteins that interact with Gz and named them GRIN1 and GRIN2 for G protein-regulated inducer of neurite outgrowth 1 and 2. Both GRIN1 and GRIN2 bound to activated G proteins (Go, Gi, and Gz) and were identified in neural tissue, but the regulatory mechanism for neurite growth is unknown.
The Ca2+ binding protein calnuc (nucleobindin) is a potential effector for Gi3 and Gs (78, 79). The binding of calnuc to Gi3 has been shown to be Ca2+ and Mg2+ dependent (80). This ion dependence has not been shown explicitly for Gs, probably because calnuc undergoes a conformational change after Ca2+ binding (81) that could be necessary for G protein interaction.
Bruton’s tyrosine kinase (Btk) has been identified as a novel effector for Gq proteins because Gq activates Btk both in vitro and in vivo, and this activation is required for receptor-mediated stimulation of p38 MAPK (82). However, the generalization of these results to other Gq family members remains to be determined.
Although a role for G12/13-proteins had been established in several physiological events such as stress fiber formation, cellular transformation, regulation of Na+/H+ exchange, modulation of inducible nitric oxide synthase expression, and regulation of Erk and c-jun kinase activity (83), direct interaction of G12/13 with effector proteins has been established recently, when Hart et al. (84) identified p115RhoGEF as a direct effector for G13. A RGS protein, p115RhoGEF, is also shown to serve as a GAP for both G12 and G13. However, only activated G13 is able to stimulate p115RhoGEF to trigger GDP/GTP exchange on the small molecular weight G protein Rho. In addition, the cytoskeletal-associated protein radixin has been found to interact with G13 (85) whereas an interaction between G12 and heatshock protein 90 is required for G12-induced serum response element activation, cytoskeletal changes, and mitogenic response (86).
|
VIII. Gß Interaction with Effectors
|
|---|
Initially, Gß was thought to facilitate the completion of intracellular information transfer passively by binding to G and hastening the return of the heterotrimer to the plasma membrane, thereby preventing noise or spontaneous G activation in the absence of receptor stimulation (87). This belief changed when Gß was shown to activate a K+-selective ion channel (IKACh) in cardiac atrial cells (88). Today, Gß is known to interact with and activate several effectors, including PLCß2 and ß3 (89, 90), ACs (91), ß-adrenergic receptor kinase (92), phosphoinositide 3-kinase (PI3 kinase) (93, 94), components of the MAPK cascade (95), and K+ and Ca2+ channels (88, 96, 97, 98) (Table 2). As the list of Gß-effectors continues to grow, recent attention has turned toward examining the mechanisms responsible for Gß-specific signaling.
At present, five different Gß-subunits and 12 different G-subunits have been identified (10, 99, 100, 101, 102), meaning that if Gß-dimers formed randomly, there would be 60 possible combinations. Although, in general, most Gß-subunits can dimerize with most G-subtypes, biochemical studies have demonstrated exceptions to the rule. For example, Gß2 dimerizes with G2 in vitro but not with G1, and Gß3 does not dimerize with either G1 or G2 (103, 104, 105). Likewise, Gß5 dimerizes poorly with G2 in a yeast-two-hybrid assay (106), but other studies suggest that these dimers can form in vitro (107) and can activate PLCß2 (102). Conversely, Gß and G combinations that were excluded in in vitro assays (104) display functional effects when transfected into cells (108). Although Gß-dimers of varying composition may form in vivo as well, Gß-dimer combinations may also exhibit cell type or tissue specificity. For example, in the retina the primary Gß-dimer is Gß11 whereas Gß12 is the most common dimer formed in the brain (109, 110).
A. Gß-Dimer composition directs effector and receptor coupling
What is the physiological significance of the formation of different Gß-dimers? Although it was previously thought that Gß-dimers were for the most part interchangeable, current research indicates that Gß-dimer composition determines the quality and efficiency of effector activation and may mediate receptor-G protein coupling specificity similar to G-subunits. For example, when nine unique dimers of Gß1 or Gß2 with G(1, 2, 3, 5 or 7) were tested for the ability to activate various PLCß isoforms, all dimers could activate the various PLCß isoforms except retinal-specific Gß11 (111, 112). Likewise, Gß11 was markedly less effective at stimulation of ACII and inhibition of ACI than other Gß dimer combinations (111, 112). A comparison of Gß12 with Gß52 demonstrated that Gß52 is a much weaker inhibitor of ACI, ACV, and ACVI. In addition, Gß12 stimulated ACII activity, whereas Gß52 inhibited the activity of this enzyme (113). In contrast, both Gß12 and Gß52 activated PLCß2 with similar potency and efficacy (114). Finally, the rank order for Gß-subtype inhibition of voltage-dependent N-type Ca2+ currents differs from enzyme activation [Gß1 = Gß2 > Gß5>>Gß3 = Gß4; (115)], and this potency difference may be related to the ability of the various Gß-subunits to physically interact with the LI-II loop of the Ca2+ channel (115). Together, these data demonstrate that the primary sequence of the Gß-subunit is a major determinant of effector coupling efficiency and specificity. Isolation of the structural features responsible for effector variation remains to be completely determined. Recently, Mirshahi et al. (116) have shown that Ser67 in Gß1 is part of a functional domain that regulates several different effectors whereas other residues of the ß-propeller seem to direct the effector specificity.
With respect to receptor-G protein coupling specificity, both Gß12 and Gß52 can couple Gq-proteins to endothelin B and M1 muscarinic receptors. However, Gß12 but not Gß52 promotes endothelin B receptor-Gi-protein interaction (107). Thus, the Gß52-dimer specifically couples Gq-proteins to receptors (117). With the exception of Gß5, the identity of the Gß-subunit does not currently appear to be a critical determinant of receptor-G protein specificity. For example, A1 adenosine receptors couple equally well to Gi-proteins containing Gß12-, Gß13-, Gß22-, or Gß23-dimers as measured by reconstitution of high-affinity agonist binding (118). In contrast, G proteins containing a farnesylated -subunit coupled less efficiently to the A1 receptor, suggesting that lipid modification of the Gß-dimer can influence receptor-G protein coupling efficiency (52).
B. Structural determinants of effector specificity
Unlike G-subunits, the conformation of Gß-dimers does not significantly change whether Gß is in the inactive heterotrimeric complex or in the free active state. One notable exception to this idea is that phosducin binding to Gß induces a conformational change primarily in blades 1 and 7, thus preventing Gß association with additional effectors (119). Once dissociated from G, Gß can interact with a number of effectors. Using alanine scanning mutagenesis, our laboratory (26) and others (120, 121) previously identified residues on Gß that contact G and that mediate a number of effector interactions including ion channels, PLCß2, and ACII (Fig. 2C). Regions important for ACII interaction map roughly to blades 2, 3, and 5, whereas the N-terminal interface of Gß interacts with G protein-activated, inwardly rectifying potassium channels, 1 and 4 (26). In addition, point mutations either on the G interacting face of blades 1–4 or mutations in the outer loops of blades 2, 6, and 7 inhibit PLCß2 activity (26, 120); whereas, PLCß3 is inhibited by point mutations within blades 2 and 5 (121). Therefore, each effector contacts a unique but overlapping set of residues on Gß, and some of these sites also represent G interacting sites. These studies are consistent with the idea that interaction with precludes Gß binding to effector proteins. Mutational studies continue to reveal the molecular basis for effector interaction as well as the structural basis for variations between Gß-subunits in effector coupling efficiency. However, one key question yet to be resolved is how Gß activates a particular effector once freed from G, in a cytoplasmic milieu full of potential partners. Signaling specificity could be brought about by factors such as discrete subcellular localization of effectors, compartmentalization of scaffolding components, and cell type-specific expression of signaling molecules (122). The formation of signaling networks that bring together specific receptors, G proteins, regulatory proteins, enzymes, and substrates is a hot area of research and will likely reveal key factors regulating signaling specificity.
C. Novel Gß-effectors
At the current discovery rate of Gß-effectors, the final tally of proteins that interact with the Gß-dimer is likely to exceed that for G-subunits (68). As shown in Tables 1 and 2, Gß- and G-subunits interact with a number of common effectors, such as PLCß, Bruton’s tyrosine kinase, and certain types of ACs. These effector interactions can be independent, synergistic, or antagonistic. For example, Gß-subunits potentiate ACII activation by Gs, but inhibit Gs-stimulated ACI activity. In addition, Gß-dimers interact with a number of novel effectors that are not regulated by G-subunits. These novel effector interactions expand the role of G proteins in the regulation of various cellular processes and are briefly discussed below.
Putative Gß-effectors recently identified include protein kinase D (123), PI3 kinase (93, 94), tubulin (124), KSR-1 (125), dynamin I (126), Raf-1 protein kinase (127), Tsk protein kinase (128), and calmodulin (129) (see Table 2 and references therein). Although previous data suggest that Gß-effectors bind to an overlapping domain on Gß-subunits, additional studies also indicate that Gß binding to one particular effector does not necessarily preclude Gß interaction with a second effector protein. For instance, Gß binding to calmodulin does not prevent Gß-mediated stimulation of PLCß (129). The ability of Gß to simultaneously regulate different effectors suggests that the Gß conformation is not disturbed upon effector binding. One notable exception to this idea is that phosducin binding to Gß induces a conformational change primarily in blades 1 and 7 preventing Gß association with additional effectors (119). Likewise, Gß interaction with the protein kinase KSR-1 prevents Gß-mediated stimulation of MAPK (125). However, the mechanism responsible for this exclusivity remains to be elucidated. As mentioned before, Chidiac and Ross (73) showed that Gß could prevent the acceleration of the GTPase activity of Gq by PLCß, which implies a dual role for Gß because it can stimulate PLCß activity directly and indirectly (through prolonged activation of Gq).
Although most Gß-effectors are believed to directly interact with the Gß-subunit, a role for the G-subunit has also been suggested. Using a yeast-two-hybrid screen with the protein kinase KSR-1, Bell et al. (125) identified G2, G3, and G10 as interacting proteins. The C terminus of G-subunit seems to play a direct role in modulating PLCß functions (130). To date, no specific Gß or G binding domain has been identified, although an intriguing number of Gß interacting proteins contain pleckstrin homology domains. Future research is likely to identify an increasing number of Gß-effector proteins. Recently, our laboratory found that the receptor for activated C kinase 1 and the dynein intermediate chain interact with the Gß11-dimer (131). Gß can inhibit neurotransmitter release independently of second messenger formation and ion channel modulation, perhaps by direct interaction with the exocytotic fusion machinery, because both syntaxin 1B and SNAP25B are Gß binding partners (132).
D. Additional role for Gß5
Gß1–4 share 80%–90% sequence homology and are ubiquitously expressed (133). In contrast, Gß5 shares only about 50% identity with the others and is preferentially expressed in the central nervous system (134). Gß1–4 are entirely particulate proteins, whereas Gß5 can exist both in the soluble and membrane fractions (134), and the N-terminal domain of Gß5 is significantly longer than the other Gß-subunits. Although this region is important for G interaction (135), Gß5 can dimerize with G, form functional heterotrimers with G, and interact with a number of effectors in response to receptor activation (102, 113, 114, 117, 134). However, unlike other Gß-subunits, Gß5 can readily dissociate from G under low-stringency conditions and is stable in solution without being complexed to G (136, 137). Free Gß5 has been shown to interact with certain GAPs known as regulators of G protein signaling (RGS proteins) through a G protein G-subunit-like domain (138, 139). The G protein G-subunit-like domain is a 64-amino acid region (34% identical to G5) that is present in RGS6, RGS7, RGS9, RGS11, and the Caenorhabditis elegans RGS protein EGL-10 (140). Gß5 binding to RGS proteins enhances the ability of the proteins to accelerate the GTPase activity of G-subunits (141). In addition, Gß5 binding to RGS6, -7, and -11 allows for the selective inactivation of Go (140) and may localize these RGS proteins within the cytosolic compartment (142). Is Gß5 always associated with an RGS protein in vivo, or does it shuttle between RGS proteins and G-subunits? In native preparations, RGS9 exists in a tight complex with the long splice variant of Gß5 (Gß5Long) in vertebrate photoreceptors (138). The Gß5L-variant was absent from the retinal tissue of RGS9-deficient mice despite the presence of normal levels of Gß5 mRNA (143). In contrast, Gß5Short protein levels were normal in knockout mice. Therefore, RGS9 may be required for the translation or stability of Gß5L in photoreceptor cells whereas Gß5S may be free to interact with G-subunits (143). Other questions yet to be resolved include: does Gß5 interact with other proteins outside the RGS family? Do free G-subunits have a signaling role on their own? What is the brain-specific role for Gß5? The discovery of Gß5 independent of G has clearly disproved the previous dogma that Gß-subunits associate only with G, and that only through this association do they elicit a physiological response.
|
IX. Molecular Basis for G Protein Inactivation
|
|---|
A. Intrinsic GTPase activity
As previously mentioned, the duration of G protein-mediated effector activation is dependent on the intrinsic GTPase activity of the G-subunit. Like the intrinsic GDP release rates, intrinsic GTP hydrolysis activity varies among G-subunits (83). For example, the catalytic rate constant value for GTP hydrolysis for Gz is approximately 200-fold lower than that of Gs and Go (83, 144, 145). The GTPase domain is highly homologous among G-subunits and the side chain of a conserved arginine residue (Arg174 in Gt; located within the helical domain) forms hydrogen bonds with oxygens of the - and -phosphates and the ß- phosphate bridging oxygen. This Arg residue plays a key role in GTP hydrolysis. Thus, mutations of either this Arg or residues contacting it have been reported to alter the GTPase activity of G-subunits (2, 146). Because of the conserved nature of the GTPase domain of G-subunits, the determinants of G-hydrolysis variability are likely to lie in the divergent helical domain and within the N and C termini or be the result of subtle flexibility and conformational changes among G-subunits. The mechanisms responsible for variations in GTP hydrolysis rates have not been studied in detail. Research in this area has focused instead on identifying proteins that directly interact with G-subunits to regulate their intrinsic GTPase activity. Some of these key studies are discussed below. For the interested reader, detailed descriptions of the mechanism of GTP hydrolysis can be found elsewhere (2, 9).
B. G Interaction with GTPase-activating proteins (GAPs)
Several years ago, researchers noted that the intrinsic GTPase activity of G-subunits occurs in vitro at a much slower rate than can account for the observed deactivation rates of G protein-controlled processes (147, 148). Therefore, speculation mounted that, in vivo, an additional protein was rapidly terminating signal transduction, returning the system to an agonist-responsive state. In mammals, the G-effectors PLCß and the -subunit of phosphodiesterase (P) were two of the earliest identified GAPs for Gq and Gt, respectively (149, 150, 151). Most recently, Scholich et al. (152) have determined that the effector ACV serves as a GAP for Gs. Thus, after activation by G, an effector can feed back on the activated G-subunit and significantly reduce the duration and amplitude of the signal generated.
In addition to effector-mediated feedback inhibition, RGS proteins enhance the GTPase activity of G-subunits, thereby reducing the duration and amplitude of both G- and Gß-mediated cellular responses (153, 154, 155, 156). RGS proteins share a common approximately 125-amino acid domain termed the RGS box (157, 158). To date, more than 30 mammalian RGS proteins have been identified (156, 158, 159, 160), each containing 23 conserved hydrophobic residues at the core of the RGS domain (155, 156, 159, 161). In vitro, the RGS core domain is both necessary and sufficient for GAP activity. However, in vivo this is not the case. Our laboratory and others (162, 163) have demonstrated that in native retinal preparations, RGS9 requires effector activation for the full expression of RGS GAP activity. Likewise, the core RGS domain of RGS16 can stimulate Go GTP hydrolysis in vitro but requires additional N-terminal residues for functional activity in vivo (164). These studies suggest that in vivo the noncatalytic domains regulate RGS GAP activity through interactions with cellular factors. Only two such factors have been identified to date, Gß5 and phosphodiesterase E (141, 151, 165). Noncatalytic domains of RGS proteins have also been suggested to mediate signal transduction pathway specificity and subcellular targeting of RGS proteins (154, 166).
GAPs for heterotrimeric G proteins accelerate GTP hydrolysis in a manner that differs from that observed with monomeric G protein GAPs. For example, Ras GAP inserts a catalytic Arg residue into the active site that participates in the hydrolysis step (2). However, this Arg finger is provided by the helical domain in heterotrimeric G proteins and mediates intrinsic GTP hydrolysis as discussed above (2). In contrast, RGS proteins bind to the switch regions on G and thereby stabilize the G transition state toward GTP hydrolysis (167). The mechanism for effector-mediated GAP activity has not been clearly delineated. By analogy, effector-mediated GAP activity may also occur through a similar stabilizing mechanism. However, differences in activity between effector GAPs and RGS GAPs have been observed. For example, Mukhopadhyay and Ross (168) demonstrated that RGS4 produces a 2-fold greater acceleration of the Gq-bound GTP hydrolysis rate in comparison to PLCß, but PLCß is 100 times more potent than RGS4. Although these findings might suggest different mechanisms of GAP activity for effectors and RGS GAPs, Gß can inhibit the GAP activity of both PLCß1 and RGS4 (73). This is consistent with the idea that Gß, effectors, and RGS proteins bind to the same region on G, namely the switch regions of the GTPase domain. Thus, RGS proteins can act as effector inhibitors as well as GAPs. On the other hand, we and others recently determined that the effector P enhances the GAP activity of the regulator of G protein signaling 9 (RGS9) core domain by increasing the affinity of the RGS9 domain for a Gt/i chimera (163, 170). These studies suggest that RGS proteins may be regulated through their participation in a signal transduction complex that may include receptors and effectors and may be localized near the plasma membrane. A similar suggestion was proposed by Chidiac and Ross (73). Our laboratory has also determined that the -helical domain of Gt (a Gi family member) is a key molecular determinant of the selectivity that the RGS9 core displays as a GAP for Gt vs. Gi (163). Therefore, RGS protein affinity and GAP activity for various G-subunits may be mediated, at least in part, by the primary structure of the G-subunit as well as by the sequence of the RGS box. Further in-depth discussion of RGS proteins can be found in one of several reviews on this topic (154, 155, 156, 161, 166).
|
Top
Abstract
I. Introduction
