|Home | About | Journals | Submit | Contact Us | Français|
Transmembrane signals initiated by a broad range of extracellular stimuli converge on nodes that regulate phospholipase C (PLC)–dependent inositol lipid hydrolysis for signal propagation. We describe how heterotrimeric guanine nucleotide–binding proteins (G proteins) activate PLC-βs and in turn are deactivated by these downstream effectors. The 2.7-angstrom structure of PLC-β3 bound to activated Gαq reveals a conserved module found within PLC-βs and other effectors optimized for rapid engagement of activated G proteins. The active site of PLC-β3 in the complex is occluded by an intramolecular plug that is likely removed upon G protein–dependent anchoring and orientation of the lipase at membrane surfaces. A second domain of PLC-β3 subsequently accelerates guanosine triphosphate hydrolysis by Gαq, causing the complex to dissociate and terminate signal propagation. Mutations within this domain dramatically delay signal termination in vitro and in vivo. Consequently, this work suggests a dynamic catch-and-release mechanism used to sharpen spatiotemporal signals mediated by diverse sensory inputs.
Phospholipase C (PLC) catalyzes the hydrolysis of phosphatidylinositol 4,5-bisphosphate [PtdIns(4,5)P2] to the second messengers inositol 1,4,5-trisphosphate [Ins(1,4,5)P3] and diacylglycerol in an essential step for the physiological action of many hormones, neurotransmitters, growth factors, and other extracellular stimuli (1–3). These cascades use signaling complexes consisting of Gα subunits of the Gq family (Gαq, 11, 14, and 16) of heterotrimeric guanine nucleotide–binding proteins (G proteins) and PLC-β isozymes (β1-4) (4–6). Agonist-stimulated receptors increase exchange of guanosine triphosphate (GTP) for guanosine diphosphate (GDP) on Gαq, GTP-bound Gαq engages and activates PLC-β, and PLC-β increases up to three orders of magnitude the rate of hydrolysis of GTP by its activating G protein (7–9). Coordination from upstream and downstream inputs sharpens time frame, amplitude, and on-off cycling of these signaling nodes. Although kinetic analyses revealed much about the dynamics of Gαq/PLC-β signaling complexes (10–12), how PLC-βs simultaneously act as effectors and GTPase activating proteins (GAPs) has remained unknown. Here, we describe the structure of PLC-β3 in an activated complex with Gαq, which together with supporting biochemical and physiological analyses reveals its mechanism of transmembrane signaling.
The three-dimensional structure of an AlF4−-dependent complex of Gαq bound to PLC-β3 was solved by molecular replacement using PLC-β2 [Protein Data Bank (PDB) code 2FJU] and Gαq (PDB 2BCJ) as search models and refined at 2.7-Å resolution (table S1). PLC-β3 engages Gαq through three distinct regions (Fig. 1, A and B). First, an extended loop between the third and fourth EF hands of PLC-β3 directly buttresses switch residues critical for GTP hydrolysis by Gαq. Second, the region of PLC-β3 that connects the catalytic TIM barrel and the C2 domain interacts with both switches 1 and 2 of Gαq. Third, a segment composed of a helix-turn-helix at the C terminus of the C2 domain resides primarily within a shallow declivity on the surface of Gαq formed by switch 2 and α3. Other effectors are known to engage this region within Gα subunits (Fig. 1C). GTP hydrolysis by Gα subunits is independently accelerated by a large family of regulator of G protein signaling (RGS) proteins (8, 13, 14), and PLC-β3 interacts with a surface on Gαq that overlaps almost completely with portions of Gα subunits needed for engagement of RGS proteins (Fig. 1C). Consistent with a biologically relevant interface (15), the complex of PLC-β3 and Gαq buries ~3200 Å2 of solvent-exposed surface area.
Activated Gαq does not impinge on the active site of PLC-β3 and indeed is at least 40 Å from the calcium ion cofactor needed for PtdIns(4,5)P2 catalysis. Comparison of the structure of PLC-β3 in complex with Gαq to previous structures (16, 17) of PLC-β2 indicates that lipase activation does not involve Gαq-dependent propagation of a conformational change to the active site of the lipase (fig. S1A). Indeed, the active site of PLC-β3 is occluded by a portion of its X/Y linker (Fig. 1B): a poorly conserved loop that separates the two halves of the catalytic TIM barrel in all PLC isozymes. Our previous structural analyses illustrated that this region similarly occludes the active site of PLC-β2, and deletion of the negatively charged X/Y linker in PLC-β2, as well as in PLC-ε and -δ1, resulted in marked activation (17). Similarly, deletion of the highly negatively charged X/Y linker of PLC-β3 caused a large increase in lipase activity (fig. S1B), indicating that PLC-β3 also is robustly autoinhibited by its X/Y linker. Presumably, this region of PLC-β3 is forced out of the active site by steric and electrostatic repulsion mediated by the surface of the plasma membrane coupled to the engagement of Gαq (Fig. 1D). A similar mechanism was proposed previously for activation of PLC-β2 by Rac1, which binds entirely through the PH domain of PLC-β2 at substantial distance from the active site of the lipase (16). Consequently, Gαq, Rac1, and likely other activators such as Gβγ activate PLC isozymes by anchoring and orienting them at substrate membranes to release autoinhibition by the X/Y linker and promote access of PtdIns(4,5)P2 to the lipase active site.
PLC-β isozymes are effectors of Gαq as well as GAPs that enhance the intrinsic GTPase activity of the engaging Gα subunits. The structure of activated Gαq bound to PLC-β3 explains the integration of these reciprocal functions.
The catalytic core of the 13 mammalian PLC isozymes includes a pleckstrin homology (PH) domain, a set of four EF hands, a catalytic TIM barrel, and a C2 domain (18) (Fig. 1A). The canonical Gα effector-binding region of Gαq, located between α3 and switch 2, is occupied by a helix-turn-helix (Hα1/Hα2) that immediately follows the C2 domain of PLC-β3 (Fig. 2A). PLC-δ isozymes terminate immediately after their C2 domain, which is the last common domain found in all PLC isozymes (18). Thus, grafting of Hα1/Hα2 onto the C terminus of the C2 domain of PLC-β3 provides a large binding surface that makes numerous contacts with Gαq (~1750 Å2 of solvent accessible surface area buried). Only PLC-β isozymes are activated by Gαq, and the highly conserved Hα1/Hα2 motif is found in all PLC-βs (Fig. 2B) but not in other PLC isozymes. PLC-βs also contain a long C-terminal region that extends about 300 residues past the Hα1/Hα2 module. This long C-terminal extension previously was thought to be important for interaction with Gαq. However, absence of this region did not affect high affinity binding of PLC-β3 to Gαq [dissociation constant (Kd) ~200 nM; fig. S2], and it is not present in the PLC-β3 construct used for structure determination. The C-terminal domain is important for membrane association, but whether it has additional function(s) in the signaling complex remains unclear.
Pro862 of PLC-β3 lies within the turn between Hα1 and Hα2, makes extensive contacts with multiple residues of Gαq, and forms the center of a Gαq-binding interface (Fig. 2A). The side chain of the preceding Asn861 supports this turn by forming a hydrogen bond with the backbone amide of Lys864. This Asn-Pro couplet is preserved in three of the four PLC-β isozymes (it is Asp-Pro in PLC-β4) and presumably defines the turn because of helix capping and breaking propensities of Asn/Asp and Pro, respectively. The turn is bracketed by Leu859, which inserts into a hydrophobic pocket formed by residues in α3 and switch 2, and by Ile863, which interacts with tandem glutamates in α3. Tyr855 in Hα1 and Asp870 in Hα2 also support the binding interface at the periphery. The binding of Gαq to Hα1/Hα2 of PLC-β3 recapitulates almost entirely the interaction of Gαq with several guanine nucleotide exchange factors (GEFs) for Rho, including p63RhoGEF, Trio, and Kalirin, which use a helix-turn-helix grafted onto the end of a DH/PH cassette to bind the α3/Sw2 declivity of Gαq (19, 20) (Fig. 2C). PLC-β3 and p63RhoGEF use identical residues in their primary interfaces with Gαq, and other effectors also engage this region of Gα subunits in similar fashion (Fig. 2C).
The role of Hα1/Hα2 residues in Gαq-mediated activation was examined by mutational analyses. Whereas expression of Gαq or PLC-β3 alone in COS-7 cells had no effect, their coexpression resulted in a large increase in inositol lipid hydrolysis (fig. S3A). In contrast, coexpression of PLC-β3 with mutation L859→E859 (21) [PLC-β3(L859E)] with Gαq had no effect over a broad range of conditions (fig. S3B). Gβγ independently activates PLC-β3, and coexpression of either PLC-β3 or PLC-β3(L859E) with Gβ1γ2 resulted in similar levels of activation (fig. S3C). Mutation of the analogous residue (Leu810) in PLC-β1 also completely abrogated Gαq-dependent stimulation (fig. S3D).
The contribution of residues across Hα1/Hα2 of PLC-β3 was examined (Fig. 2D and fig. S4). In each case, the relative sensitivity of the PLC-β3 mutant to activation by Gαq versus Gβ1γ2 was compared under conditions where maximal response to each activator was observed. Whereas single substitutions throughout Hα1/Hα2 did not affect Gβγ-stimulated activity (fig. S4), certain of these mutations (Y855A, L859A, N861A, P862A, and I863A) resulted in substantial or complete loss of the capacity of Gαq to promote PLC-β3–dependent increases in inositol phosphate accumulation (Fig. 2D).
The binding and lipase activities of PLC-β3 mutants also were tested by using purified proteins (fig. S5). PLC-β3, PLC-β3(L859E), and PLC-β3(L859A) exhibited similar basal lipase activities (fig. S5A) and were similarly activated by Gβ1γ2 (fig. S5B). However, the binding affinity of PLC-β3(L859A) for Gαq in the presence of AlF4− was sevenfold lower than PLC-β3, and no AlF4−-dependent binding of PLC-β3(L859E) was observed (fig. S5C). Activities of PLC-β3 isozymes mutated in Hα1/Hα2 also differed markedly in a signaling complex reconstituted with purified P2Y1 receptor, heterotrimeric Gq, and PLC-β3. The P2Y1 receptor agonist, 2MeSADP, promoted robust activation of PLC-β3, but PLC-β3(L859E) was completely refractory to activation and intermediate activation was observed with PLC-β3(L859A) (Fig. 2E).
Alanine-scanning mutagenesis previously identified two small segments (residues 243 to 245 and 256 to 257) of Gαq necessary for elevated production of inositol phosphates (22). These regions contribute to interactions with Hα1/Hα2 (fig. S6A). Additional alanine substitutions were made in Gαq, and, of those Gαq mutants that expressed as stable trypsin-resistant proteins, most exhibited a predictable loss in capacity to activate PLC-β3 (fig. S6B).
The loop between the end of the TIM barrel and the beginning of the C2 domain comprises a second distinct segment of PLC-β3 that makes extensive contacts with active Gαq, including switches 1 and 2 (Fig. 3A). This interface includes a series of interdigitated pairs of charged residues, specifically (in PLC-β3/Gαq) Asp709/Arg202, Lys710/Glu191, and Asp721/Lys41; these in turn are supported by additional charged residues (Glu703 and Arg707) of PLC-β3. Alanine substitution of several of these residues in PLC-β3 compromised the capacity of Gαq, but not Gβ1γ2, to activate PLC in COS-7 cells (Fig. 3B).
Residues adjacent to both borders of the C2 domain (Val724 and Tyr847) converge to envelop His218 of Gαq, which is wedged between the afore-mentioned interface and the start of Hα1/Hα2 to anchor two of the three major interfaces within the Gαq•PLC-β3 complex (Fig. 3A). Mutation of His218 results in loss of capacity of Gαq to activate PLC-β3 (fig. S6B).
PLC-δ isozymes are not regulated by Gαq, presumably because of lack of both Hα1/Hα2 and the Gαq-interacting residues found in PLC-β isozymes between the TIM barrel and the C2 domain. Thus, we hypothesized that G protein–dependent regulation could be engineered into PLC-δ1 (fig. S7A). Surface plasmon resonance (SPR) analyses revealed that, whereas PLC-δ1 did not exhibit AlF4−-dependent binding to Gαq, introduction of the Hα1/Hα2 segment of PLC-β3 into PLC-δ1 conferred binding (fig. S7B). Moreover, receptor- and guanine nucleotide–stimulated lipase activity was observed with the chimeric isozymes but not PLC-δ1 (Fig. 2F), and the median effective concentration (EC50) of GTPγS for activation of PLC-δ1(Hα1/Hα2) by GTPγS was 50 nM (fig. S7C). Thus, Hα1/Hα2 is a small, linear module used for functional engagement of Gαq.
An extended loop between EF hands 3 and 4 of PLC-β3 interacts with the GTP-binding region of Gαq (Fig. 4A). This loop is highly conserved in all PLC-β isozymes, is not found in PLC-δ1 (Fig. 4B) or other PLC isozymes, and interacts with the active site of Gαq. Asn260 of the EF3/4 loop promotes GTP hydrolysis by interaction with the side chain of Gln209 of Gαq (Fig. 4C), which rearranges during GTP hydrolysis to stabilize the transition state mimicked by GDP•AlF4−•H20. Asn260 also interacts with Glu212 to stabilize switch 1 for GTP hydrolysis. The interactions of Asn260 of PLC-β3 with Gαq are recapitulated by a functionally equivalent asparagine in RGS9 (23) (Fig. 4C) and other RGS proteins (24, 25).
Asn260 is positioned at the active site of Gαq as part of a tight turn (residue 260 to 264) of PLC-β3 that is stabilized by Glu261 and underpinned by an extensive series of hydrogen bonds principally mediated by Asp256 and Arg255 and Arg258 (Fig. 4A). These residues are highly conserved in all PLC-βs, as are Asn251 and Leu267, which appear crucial in stabilizing the ends of the loop (Fig. 4B). The EF3/4 loop as well as other portions of EF hands 3 and 4 are disordered in the crystal structure of PLC-β2 (Fig. 4D). A likely scenario is that Gαq initially engages the EF3/4 loop of PLC-β3 to nucleate the underlying hydrogen bonding network and promote cooperative ordering of EF hands 3 and 4.
To directly examine the role of the EF3/4 region of PLC-β3 in mediating inactivation of its activating G protein, we quantified GTP hydrolysis by Gαq in the presence of purified PLC-β3 mutants (fig. S8A) by using phospholipid vesicles reconstituted with the P2Y1 receptor and heterotrimeric Gq. In the presence of receptor agonist, PLC-β3 promoted up to 100-fold stimulation of GTP hydrolysis (Fig. 5A and fig. S8B), and activation occurred with an EC50 ~ 3 nM (table S2). A chimeric PLC-β3 replacing the EF3/4 loop with the analogous region of PLC-δ1 was severely crippled in its capacity to accelerate GTP hydrolysis by Gαq (Fig. 5A). Similarly, substitution of Asn260 dramatically reduced the capacity of PLC-β3 to promote GTP hydrolysis, whereas substitution of Val262 had negligible effect. Importantly, basal and Gβγ-stimulated lipase activities of these purified mutants were unaffected (fig. S8, A and C). The EC50 values of mutant and wild-type PLC-β3 for stimulation of GTP hydrolysis also were similar (table S2). In contrast, substitution of Leu859 to Ala859 within Hα1/Hα2, which reduced the binding affinity of the complex by ~sevenfold (fig S5C), also increased the EC50 for stimulation of GTP hydrolysis by ~10-fold (table S2). Thus, the EF3/4 loop of PLC-β3 is crucial for stimulation of GTP hydrolysis by Gαq but contributes minimally to forming the signaling complex.
Loss of capacity of PLC-β3 to promote GTP hydrolysis by Gαq should decrease its capacity to turn off subsequent to Gαq-mediated activation. This idea was first tested in vitro with use of purified proteins. Addition of a P2Y1 receptor antagonist (fig. S8, B and D) to an agonist pre-activated signaling complex of the P2Y1 receptor, heterotrimeric Gαq, and wild-type PLC-β3 resulted in a rapid decline of lipase activity to levels similar to those observed in the absence of agonist (Fig. 5B). In contrast, little reversal of lipase activity occurred upon addition of P2Y1 receptor antagonist to a similarly preactivated complex containing PLC-β3(δEF) (Fig. 5B).
Rhodopsin-initiated phototransduction in Drosophila melanogaster is mediated by Gαq-dependent activation of PLC-β (26). To examine the role of the EF3/4 loop of PLC-β in a physiological system, we replaced wild-type PLC-β (NORPA) in Drosophila with a version mutated to alanine in the conserved Asn (N262) demonstrated above to be required for PLC-β-promoted GTP hydrolysis by Gαq. Flies expressing wild-type NORPA or NORPAN262A exhibited similar amplitudes of the light-induced photoresponse (Fig. 5C). In contrast, whereas termination of light resulted in rapid termination of photoresponse in wild-type flies, we observed a marked defect in recovery with norpAN262A.
PLC-β3 is a tumor suppressor, and its disruption in humans contributes to lymphomas and other myeloid malignancies (27, 28). Similarly, Gαq is an oncogene, and its constitutive activation drives ~50% of all uveal melanomas (29). Signaling through the Gαq/PLC-β axis is important for regulation of cell proliferation, and other disruptions in this node can be expected to contribute to cancer. In this regard, homozygous substitution of Arg254 within the EF3/4 loops of PLC-β4 was found in a pancreatic tumor during genome-wide profiling (30). The equivalent substitution in PLC-β3 resulted in a decrease in capacity to accelerate GTP hydrolysis by Gαq (Fig. 5D and fig. S8E).
The high-resolution structure of Gαq•PLC-β highlights a dynamic interplay between regions of the complex needed to coordinate rapid activation and inactivation of this signaling node required for highly responsive, low-noise signal transduction. We propose that a conformationally flexible Hα1/Hα2 samples a relatively large volume to maximize probability of encountering Gαq, and transient interactions with Gαq guide the final folding of Hα1/Hα2. The process of coupling folding with binding to increase the rate of formation of the final encounter complex has been described for other protein complexes (31, 32) and is referred to as fly-casting. A subset of Dbl-family RhoGEFs typified by p63RhoGEF also apparently use fly-casting to engage Gαq (19, 20). In particular, p63RhoGEF uses a helix-turn-helix immediately adjacent to a conserved PH domain to engage Gq in a fashion that is recapitulated almost identically in Gαq•PLC-β. Thus, an independent module is grafted onto PLC-βs and RhoGEFs to confer binding to Gαq. The Hα1/Hα2 module in PLC-βs is encoded by a single exon, which suggests that these signaling proteins acquired capacity to bind Gαq through intergenic exon shuffling.
Engagement of PLC-β by Gαq is intimately coupled to inactivation of the complex. A primordial PLC-δ acquired an extended loop between EF hands 3 and 4 (Fig. 4D) that directly engages the switch regions of Gαq to stabilize the transition state for GTP hydrolysis. This EF3/4 loop and Hα1/Hα2 are linked evolutionarily, because both motifs are found in the two PLC-β isozymes of Caenorhabditis elegans. Indeed, they are not found separately in any PLC-β and therefore are the defining motifs of members of the PLC-β family. Taken together, we propose that Gαq is “caught” by a flycast from Hα1/Hα2 and is “released” by EF3/4 loop-mediated stimulation of GTP hydrolysis, which results in a conformational change in Sw2 and abrogation of the binding sites for both the EF3/4 loop and the Hα1/Hα2 segment. We also note that p115RhoGEF binds to Gα13 and promotes GTP hydrolysis through two different domains (33).
Rapid cycling of effector engagement and GTP hydrolysis favors the maintenance of heterotrimeric Gq/effector complexes necessary for signal acuity in a process generally referred to as kinetic scaffolding (9, 10, 34, 35). Phototransduction requires high signal amplification in rapid cycles of activation/deactivation in a signaling system organized for suppression of noise and therefore provides an excellent model for comparison of signaling mediated by PLC-βs and other effectors. Although Gαq-promoted activation of PLC-β mediates phototransduction in some metazoans such as fruit flies, mammalian rod and cone phototransduction involves Gαt-mediated activation of cyclic guanosine monophosphate (GMP) phosphodiesterase (PDE) (36). This PDE is not a GAP, and acceleration of GTP hydrolysis evolved in a separate protein, RGS9, which nonetheless stabilizes the switch regions of Gαt in much the way the EF3/4 loop of PLC-β stabilizes Gαq (23) (Fig. 4C). The binding of PDE to the effector pocket of Gαt allosterically increases binding of RGS9 (23, 37). G protein–coupled receptor kinase 2 (GRK2) and, to a lesser extent, p63RhoGEF enhance the GAP activity of RGS4 in a complex with Gαq (38). This allostery is inherent in the catch-and-release mechanism used by PLC-βs. Ablation of GAP function of PLC-β markedly prolongs deactivation of phototransduction in Drosophila (Fig. 5C), and disruption of RGS9 in mice (39) and mutation of RGS9 in human disease (40) produce analogous phenotypes.
This research was supported by NIH grants GM38213 and GM57391 (J.S. and T.K.H.), GM61454 and GM074001 (T.K.), and EY010852 (C.M.). T.K.R. was supported by a Ruth L. Kirschstein National Service Award F31 Predoctoral Fellowship and a United Negro College Fund Merck Graduate Science Research Dissertation Fellowship. K.T. was supported by a Postdoctoral Fellowship for Research Abroad from the Japan Society for the Promotion of Science. We acknowledge the outstanding help with structural analyses by B. Temple, L. Betts, J. Vanhooke, T. Charpentier, V. Arshavsky, and M. Kosloff; with analysis of linker-deleted PLC-β3 by N. Vincent Jordan; with SPR analyses by A. Kimple; with [γ32P]GTP purification by E. Lazarowski and the insightful comments on the manuscript by H. Dohlman, R. Nicholas, E. Ross, and K. Slep. Coordinates and structure factors for Gαq•PLC-β3 have been deposited under the PDB accession code 3OHM.
Materials and Methods