Introduction

Heterotrimeric G proteins regulate a wide variety of effectors downstream of G protein-coupled receptors (GPCRs). One family of effector enzymes are the phospholipase Cβ (PLCβ) enzymes. The four PLCβ isoforms (PLCβ1-4) cleave phosphatidylinositol-4,5-bisphosphate (PIP2) to inositol-1,4,5-trisphosphate (IP3) and diacylglycerol (DAG). These second messengers in turn increase intracellular Ca²⁺ and activate protein kinase C (PKC). All PLCβ isoforms are activated by direct binding of Gα_q, released by G_q-coupled receptors. PLCβ2, PLCβ3, and to a lesser extent PLCβ1, are also stimulated by binding of Gβγ heterodimers, released by G_i-coupled receptors (1, 2).

PLCβs share four core domains with other PLC enzymes, including a pleckstrin homology (PH) domain, four EF hands, a catalytic triose phosphate isomerase (TIM) barrel split by a regulatory linker (X–Y linker) into X and Y subdomains, and a C2 domain. The PLCβ subfamily is defined by its unique proximal and distal C-terminal domains (CTDs) that follow the C2 domain. The proximal CTD (pCTD) includes the autoinhibitory the Hα2′ helix, which is displaced when Gα_q binds (3), and the distal CTD (dCTD), which contributes to membrane binding and contains a second functionally critical Gα_q binding site (2). The mechanism by which Gα_q binds to and activates PLCβ has been well characterized through functional and structural studies (3–6), but the mechanism by which Gβγ activates PLCβ is much less clear.

The Gβγ heterodimer has no intrinsic enzymatic activity yet regulates a wide variety of effector enzymes via membrane recruitment and/or allostery (7, 8). Prior studies of Gβγ regulation of PLCβ identified two potential, non-overlapping binding sites for Gβγ. The first was the PH domain, which serves primarily as a protein-protein interaction site in the PLCβ subfamily (9, 10). The PH domain is required for Gβγ stimulation of PLCβ and chimeras of PLCδ that contained the PH domain of PLCβ gained sensitivity to the G protein subunit (11). The second Gβγ binding site was mapped to a helix in the Y subdomain of the TIM barrel. Peptides corresponding to this region blocked Gβγ activation of PLCβ and crosslinked to Gβγ (12, 13). However, Gβγ binding to this site appeared to preclude the PLCβ active site from engaging the membrane for PIP2 hydrolysis. The mechanism of Gβγ-dependent activation is further complicated by reports that, in cells, the process requires prior or coincident activation by Gα_q (14–16).

Recent structural work has shed light onto how Gβγ interacts with PLCβ. Cryo-electron microscopy (cryo-EM) reconstructions of Gβγ and PLCβ on liposomes and nanodiscs showed two Gβγ molecules bound to one PLCβ3, one to the PH domain and one to the EF hand domain (17). The latter domain had not previously been implicated in Gβγ-dependent activation. Gβγ binding did not induce conformational changes within the lipase, and despite its proximity to a lipid bilayer, PLCβ3 remained in its autoinhibited conformation; the active site remained occluded by the X–Y linker and inhibitory interactions between the Hα2′ helix and the catalytic core persisted. Because no allosteric changes in the lipase were observed, a model was proposed in which Gβγ activates PLCβ3 by recruiting it to and orienting it at the plasma membrane (17). Consistent with this idea, the same study demonstrated Gβγ-dependent partitioning of PLCβ3 to a lipid bilayer (17), although several prior studies using purified components did not show Gβγ-dependent recruitment of the lipase to membranes (18–20). Additionally, the Gβγ–PLCβ3 interface(s) responsible for activation in the cellular environment are unknown. Finally, it is also unclear if G protein activation liberates sufficient free Gβγ to recruit PLCβ3 in cells.

Here we investigate the mechanism by which Gβγ binding to PLCβ3 increases lipase activity using crosslinking, cryo-EM single particle analysis (SPA), and functional assays in living cells. We report cryo-EM reconstructions of Gβγ–PLCβ3 that reveal a third binding site for Gβγ involving the PH, EF hand, and C2 domains. We show that all three structurally determined interfaces contribute to Gβγ-dependent activation in cells. Notably, we find that free Gβγ does not promote recruitment of PLCβ3 to the plasma membrane, and that membrane-tethered PLCβ3 can still be activated by Gβγ. Thus, Gβγ binds to multiple sites on the lipase to further stimulate PIP₂ hydrolysis concomitant with activation of PLCβ3 by Gα_q, most likely by orienting the enzyme at the plasma membrane. We propose that Gβγ is best understood as a critical positive allosteric modulator of PLCβ3, as opposed to a bona fide activator in cells.

Results

Cryo-EM reconstructions of Gβγ–PLCβ3 complexes in solution

We first attempted to determine the solution structure of a soluble Gβγ–PLCβ3 complex, in which the prenylated Gγ C68 is mutated to serine (Gβγ C68S) (21), eliminating the need for lipids and/or detergents. Complexes of Gβγ C68S–PLCβ3 could be isolated by size exclusion chromatography (SEC) but were too unstable for structural determination, in agreement with previous studies (17). We turned to crosslinking to isolate a stable complex (22). Solvent-exposed cysteines in the lipase (human PLCβ3 residues 193, 221, 358, 516, 824 and 834) were mutated to serines in the background of PLCβ3 Δ892, a C-terminal truncation which lacks the distal CTD yet retains robust activation by Gβγ (23–25). Given the evidence that Gβγ binds to the PH domain, an E60C mutation was installed in the PH domain (PLCβ3 Δ892 PH_cys) to facilitate crosslinking with Gβγ C68S (Gβ1 contains fourteen cysteines, with C204 and C271 solvent-exposed, while Gγ2 contains two cysteines with only C68 solvent-exposed). As a control, C516 was retained in the X–Y linker (PLCβ3 Δ892 XY_cys). Neither variant underwent self-crosslinking, in contrast to PLCβ3 Δ892 which retains the endogenous cysteines, but only PLCβ3 Δ892 PH_cys crosslinked efficiently to Gβγ C68S (Figure S1). Bismaleimidoethane (BMOE), an 8 Å crosslinker had a crosslinking efficiency of >50%, and a 1:1 complex was observed with PLCβ3 Δ892 PH_cys, consistent with a persistent and specific interaction (Figure S1B). To confirm the BMOE-crosslinked Gβγ–PLCβ3 Δ892 PH_cys complex was functional, crosslinking was repeated using wild-type Gβγ and PLCβ3 Δ892 PH_cys, resulting in ∼3-fold greater activity than the reaction without the crosslinker (Figure S1C). Similar results were also obtained with the 14.7 Å BM(PEG)2 crosslinker. The crosslinked Gβγ–PLCβ3 Δ892 PH_cys complexes were purified using SEC and subjected to cryo-electron microscopy (cryo-EM) single particle analysis (SPA). Two reconstructions were independently refined to 4 Å and 7 Å resolution in the BMOE data set (Figures 1, S2, S3, Table 1), and one 4.4 Å reconstruction in the BM(PEG)2 data set (Figures S2, S4, Table 1).

Figure 1.

Table 1.

In all crosslinked Gβγ–PLCβ3 Δ892 PH_cys complexes, Gβγ engages PLCβ3 via a surface formed primarily by the PH and EF1/2 domains that differs from the two interfaces observed previously (17). The relative orientation and local resolution of Gβγ with respect to PLCβ3 varies in each case, suggesting the interface is conformationally heterogeneous, most likely due to the absence of a membrane (Figure S2). Even though all the cysteines in Gβ are retained, and both Gβ C271 and C204 are in the “hotspot” interaction surface of Gβγ, only a single crosslink between PLCβ3 E60C and Gβ Cys271 is observed. In all Gβγ–PLCβ3 reconstructions reported, Gβ C204 is ∼25 Å from PLCβ3 E60, well beyond the range of either BMOE or BM(PEG)2. There are also no solvent-exposed cysteines in Gβγ at the EF hand binding site within ∼45 Å of PLCβ3 E60. This is consistent with a specific, persistent interaction that results in the crosslink between Gβ 271 and PLCβ3 E60C. Indeed, density for the crosslinker is observed in the 4 Å BMOE reconstruction, and the orientation of Gβγ and PLCβ3 in the 7 Å BMOE and BM(PEG)2 reconstructions are also consistent with crosslinking via this site.

In the crosslinked complexes, as in the prior structures, PLCβ3 Δ892 PH_cys is autoinhibited by its X–Y linker and pCTD (Figure 1) (3, 4, 9, 17, 26). This is consistent with previous reports demonstrating the membrane is essential for regulation by Gβγ and that its activation mechanism is independent of the PLCβ CTDs (2). The crosslinked Gβγ is situated such that it allows simultaneous binding of Gα_q to PLCβ with the C-terminal helix of Gγ nearly in the same plane as the phospholipase active site. Thus, the solution reconstruction may represent a membrane-localized complex, but not a catalytically active state.

In the 4 Å reconstruction (Figures 1, 2A), the Gβγ–PLCβ3 Δ892 PH_cys interface buries ∼1,400 Ų surface area. This is more extensive than the Gβγ-PH domain or Gβγ-EF hand interfaces, which bury ∼800 Ų and ∼1,100 Ų respectively (Figures 2B, 2C) (17). The primary Gβγ interface is formed by residues on the side of the WD40 toroid, rather than its face, which is the typical effector interface. Nevertheless, the crosslinked Gβγ–PLCβ3 Δ892-PH_cys interface includes several residues known to be critical for enzyme activation, and the specific interactions differ from the other reported Gβγ–PLCβ3 interfaces. In the BMOE complex, Gβ D228 interacts with R199 and K183 in PLCβ3, whereas in the Gβγ–PLCβ3–Gβγ reconstruction Gβ D228 interacts with R24 or K238 in the PH domain and EF hand interfaces, respectively (Figure 2A-C). In our Gβγ–PLCβ3 Δ892 PH_cys structure, Gβ K301 and R304 on blade 6 make electrostatic interactions with PLCβ3 D655 and E34, respectively (Figure 2A). These interactions were not reported in the liposome-bound Gβγ–PLCβ–Gβγ complex (17), and provide a structural explanation for observations reported over thirty years ago by Neer and coworkers on the importance of Gβ blades 6 and 7 in lipase activation (27, 28). On the other hand, Gβ W99, a critical residue for effector activation including PLCβ2 (29), does not interact with the lipase in the crosslinked complexes, but does interact in the Gβγ–PLCβ3–Gβγ complex. Taken together with previous structural findings, our results suggest that Gβγ can bind to PLCβ3 at several sites with modest affinity.

Figure 2.

Functional analysis of Gβγ–PLCβ3 interfaces: IP accumulation

To assess the functional relevance of the Gβγ–PLCβ3 interfaces observed in cryo-EM reconstructions (Figure 2) we carried out cell-based inositol phosphate (IP) accumulation assays in COS7 cells, where PLCβ3 is increased by the overexpression of either Gβγ or Gα_q. Mutations of residues in PLCβ3 or Gβ₁ at any one of the three observed interfaces decreased Gβγ-dependent activation of the lipase (Figures 2D-G). PLCβ3 mutants with decreased responsiveness to Gβγ are most likely impaired in binding the Gβγ heterodimer, as their expression and activation by Gα_q were minimally altered (Figures 2D-G; Figure S5). PLCβ3 K183E, which eliminates an electrostatic interaction with Gβ E226 and D228 in the crosslinked complexes, has 3-fold lower Gβγ-stimulated activity (Figures 2D,E), as do PLCβ3 R24E, L40G, and R185L, which disrupt interactions with Gβγ observed in the liposome-tethered reconstructions (Figure 2E) (17, 29). Gβ D228R, which interacts with PLCβ3 in all reconstructions (Figures 2F, G) (17, 30), decreased activation of PLCβ3 by ∼20-fold, which cannot be fully attributed to reduced protein expression (Figure S5). Mutations in Gβ blades 6 and 7, Gβ K301E and R304D, also decreased PLCβ3 activation, in agreement with previous reports (23) (Figure 2F,G).

Functional analysis of Gβγ–PLCβ3 interfaces: PLCβ3 interaction with Gβγ

Under physiological conditions, PLCβ3 is activated in response to GPCR stimulation that activates G_q heterotrimers, rather than overexpression of Gα_q or Gβγ, as in the previous experimental setting. To assess the functional significance of the three Gβγ–PLCβ3 interfaces downstream of receptor activation we developed a BRET assay to monitor Gβγ binding to PLCβ3 in live cells in real time. This effort was complicated by the fact that a large fraction of PLCβ3 is cytosolic, whereas Gβγ is anchored to the plasma membrane. Because of this, recruitment of PLCβ3 to the plasma membrane by any means (e.g. binding to Gα_q·GTP) would cause an increase in bystander BRET between PLCβ3 and Gβγ that would sum with BRET due to direct interactions between the two. To eliminate this confound, we anchored HiBit-labeled PLCβ3 to the plasma membrane with a C-terminal CAAX motif, which prevents increases in bystander BRET due to translocation and thus isolates signals due to interactions between Venus-Gβγ and the lipase at the plasma membrane. Activation of angiotensin AT₁ receptors induced a rapid increase in BRET between HiBit-PLCβ3-CAAX (coexpressed with LgBit and Gα_q) and Venus-Gβγ (Figure 3A). This signal was completely blocked by membrane-localized GRK3ct, which binds and sequesters free Gβγ, and enhanced by membrane-localized GRK2RH, which binds and sequesters free Gα_q·GTP (Figure 3A). The latter observation is complementary to our previous finding that sequestering Gβγ enhances Gα_q binding to PLCβ3; both findings suggest that PLCβ3 competes with G protein subunits for binding to the complementary G protein subunits (6).

Figure 3.

We next used this interaction assay to test a series of fifteen HiBit-PLCβ3-CAAX mutants designed to disrupt the three structurally determined Gβγ–PLCβ3 interfaces. Mutations in the two interfaces observed in the liposome-tethered reconstructions significantly reduced agonist-induced BRET between Venus-Gβγ and HiBit-PLCβ3-CAAX (Figure 3B). For example, L40E in the PH domain interface and R185E in the EF hand interface almost completely abolished receptor-mediated signals (Figure 3B). In contrast, K183E in the crosslinked complex interface did not significantly reduce agonist-induced BRET signals, even though this mutation significantly impaired activation by Gβγ in our IP accumulation assays. This is likely due to the position of the Venus BRET acceptor in the crosslinked complex, which is much less conducive to energy transfer from the donor (HiBit + LgBit) to the PLCβ3 N-terminus than the Venus BRET acceptors in the liposome-tethered complexes. Therefore, this assay may be unable to detect Gβγ–PLCβ3 interactions at the crosslinked site. Surprisingly, mutations at residues D167 in the PH domain interface and R215 in the EF hand interface significantly increased receptor-mediated interaction between Venus-Gβγ and HiBit-PLCβ3-CAAX (Figure 3B). Overall, these results support the conclusion that the functional defects observed in our IP accumulation assays reflect loss of Gβγ binding.

Notably, the almost complete loss of agonist-induced BRET signals after disruption of either of the PH or EF hand interfaces suggests that, when attached to a membrane, Gβγ dimers might cooperate to bind at both sites. This is consistent with the relatively weak binding of Gβγ in vitro and difficulties isolating stable Gβγ–PLCβ3 complexes in solution. We then asked if cooperative binding of G protein subunits might extend to Gα_q, i.e. if binding of Gα_q promotes binding of Gβγ. Using the same Gβγ–PLCβ3 interaction assay, we first compared signals downstream of AT₁, which activates both G_q and G_i/o heterotrimers, to those downstream of the dopamine D2 receptor (D2R), which activates only G_i/o heterotrimers. D2R-mediated signals were detectable, but significantly smaller than AT₁-mediated signals (Figure S6A). Under the same conditions, D2R liberated more free Venus-Gβγ than AT₁ when detected by the memGRK3ct-Nluc sensor (Figure S6B), suggesting that the presence of Gα_q·GTP promotes Gβγ binding to HiBit-PLCβ3-CAAX. To further test this idea we constructed HiBit-PLCβ3-CAAX mutants with disrupted Gα_q binding to the proximal CTD (HiBit-PLCβ3-CAAX-LE) or the distal CTD (HiBit-PLCβ3-CAAX-EEE) (6). Interaction of both mutants with Venus-Gβγ was significantly impaired compared to wild-type HiBit-PLCβ3-CAAX (Figure S6C). These results suggest that Gβγ binding to the PH and EF interfaces is facilitated by binding of Gα_q·GTP.

Functional analysis of Gβγ–PLCβ3 interfaces: receptor-evoked PIP2 hydrolysis

To better understand the role of Gβγ binding to different interfaces in receptor-mediated PLCβ3 activation, we carried out PIP2 hydrolysis assays based on binding of the PH domain of PLCδ to PIP2 at the plasma membrane. PIP2 hydrolysis releases Nluc-PH-PLCδ (Nluc-PH) into the cytosol, which is detected as a loss of bystander BRET to a marker on the plasma membrane (mem-Venus; Figure 3C). We first established that endogenous Gβγ contributes to activation of endogenous PLCβ3 (the only PLCβ isoform expressed at significant levels in HEK 293 cells) downstream of AT₁ receptors. Accordingly, PIP2 hydrolysis after activation of AT₁ was inhibited by GRK3ct (Figure S7A, B). Because AT₁ receptors activate both G_q and G_i/o heterotrimers, we were interested to know the source of the Gβγ contributing to PLCβ3 activation. We found that pertussis toxin (PTX) partially blocked AT₁-mediated PIP2 hydrolysis, consistent with Gβγ from G_i/o heterotrimers stimulating lipase activity (Figure S7A, B). However, GRK3ct still inhibited PIP2 hydrolysis when PTX was present, indicating Gβγ from G_q heterotrimers contributes as well (Figure S7A, B).

Next, we reconstituted the same PIP2 hydrolysis assay in genome-edited HEK 293 cells lacking endogenous PLCβ proteins (ΔPLC cells) by expressing wild-type (wt) or mutant HiBit-PLCβ3. PIP2 hydrolysis mediated by expressed HiBit-PLCβ3 was inhibited by GRK3ct to roughly the same extent as that mediated by endogenous PLCβ3 in parental cells (Figure 3C). Notably, PIP2 hydrolysis under these conditions was completely blocked by GRK2RH, consistent with the idea that activation of PLCβ3 by Gβγ in cells requires Gα_q·GTP (Figure 3C). PIP2 hydrolysis signals mediated by HiBit-PLCβ3 variants with impaired Gβγ binding were significantly smaller than signals mediated by wt HiBit-PLCβ3 (Figure 3D). Among the most defective variants were the L40E PH domain and R185E EF hand mutants, the latter supporting very modest PIP2 hydrolysis comparable to that remaining when Gβγ is sequestered by GRK3ct (Figure 3D). Consistent with our Gβγ sequestration results, both of these mutants were still significantly impaired in the presence of PTX (Figure S7C, D). In agreement with our IP accumulation assays, K183E in the crosslinked complex interface significantly reduced agonist-induced PIP2 hydrolysis (Figure 3D). Conversely, in agreement with our Gβγ interaction results, the D167A mutant in the PH domain interface significantly increased receptor-mediated PIP2 hydrolysis (Figure 3D). Overall, there was an excellent correlation between Gβγ binding and PIP2 hydrolysis across the fifteen mutants we tested, and good agreement with our IP accumulation assays. There were no significant differences between wt HiBit-PLCβ3 and any mutant with respect to association with Gα_q-Venus (Figure 3E, F) or expression level (SI Appendix, Table S1), ruling out indirect effects on Gα_q binding or protein stability. These findings indicate that all three of the structurally resolved Gβγ–PLCβ3 interfaces contribute to activation of PLCβ3 downstream of AT₁ receptors.

Gβγ facilitates activation of membrane-anchored PLCβ3

PLCβ3 is not tightly anchored to the plasma membrane, and Gβγ has been proposed to increase lipase activity by recruiting the holoenzyme to the membrane. We previously found that AT₁ activation results in recruitment of PLCβ3 to the plasma membrane, but concluded that this was primarily mediated by binding to Gα_q·GTP (6). To test the possibility that Gβγ binding contributes to membrane recruitment of PLCβ3, we used our previously established bystander BRET recruitment assay (Figure 4A). We found that none of the mutations that inhibited Venus-Gβγ binding and PIP2 hydrolysis significantly impaired recruitment of HiBit-PLCβ3 to the plasma membrane (Figure 4B), indicating that recruitment depends entirely on binding to Gα_q·GTP (6). As an alternative test of this hypothesis, we asked if Gβγ increased the activity of HiBit-PLCβ3-CAAX, which is tightly anchored to the plasma membrane and therefore not subject to translocation from the cytosol. We found that GRK3ct was effective at inhibiting PIP2 hydrolysis mediated by HiBit-PLCβ3-CAAX (Figure 4C), similar to its effect on PIP2 hydrolysis mediated by endogenous PLCβ or HiBit-PLCβ3. Likewise, mutations in HiBit-PLCβ3-CAAX that impaired Gβγ binding also impaired PIP2 hydrolysis (Figure 4C, D). Taken together, these results suggest that Gβγ does not facilitate PLCβ3 activation by recruiting the lipase to the plasma membrane, but rather enhances the catalytic activity of Gα_q-bound PLCβ3 through a distinct allosteric mechanism at the membrane.

Figure 4.

Discussion

PLCβ enzymes have critical roles in numerous processes, from cardiovascular and vascular smooth muscle function to opioid sensitivity (31–35). PLCβ basal activity is tightly controlled, and direct binding of Gα_q and Gβγ subunits, released downstream of G_q- and G_i/o-coupled receptors, is essential for robust PIP2 hydrolysis (2). The mechanisms by which Gα_q interacts with PLCβ to increase lipase activity have been well-characterized through structural and functional studies (2, 3, 5, 6, 36). Much less is known about how Gβγ binds to and activates PLCβ1-3. Prior studies established roles for the PH domain and, to a lesser extent, the TIM barrel in Gβγ-dependent activation (11–13, 18, 22, 37). The cryo-EM structure of a liposome-tethered Gβγ– PLCβ3 complex confirmed that Gβγ bound directly to the PH domain, and surprisingly, that a second Gβγ molecule bound to the EF hands (17). However, the functional relevance of both binding sites remained unclear. Moreover, binding of Gβγ also failed to induce conformational changes in the lipase necessary for activation, despite the proximity to a membrane surface. A model was proposed in which Gβγ activates PLCβ3 via membrane recruitment and orientation (17). However, this model contradicted previous studies showing Gβγ did not recruit PLCβ to the plasma membrane (12, 18, 22, 25, 38–40). Thus, the question of mechanism remained unsettled.

In this study, we used cryo-EM, activity and signaling assays, and BRET to validate Gβγ–PLCβ3 interactions, assess their contribution to PIP2 hydrolysis, and establish whether Gβγ activates PLCβ3 via membrane recruitment. Using cryo-EM, we identified a third Gβγ binding site on the PLCβ3 PH domain and confirmed all structurally determined Gβγ binding sites are necessary for maximum G protein-stimulated activity (Figures 1-3). Multiple Gβγ subunits binding to a single effector is not without precedent, as two Gβγ molecules must bind to non-overlapping sites on phosphatidylinositol 3-kinase γ (PI3Kγ) for maximum membrane recruitment and activation (41, 42). The structures of Gβγ–PLCβ3 complexes all tether the lipase to the membrane (Figure 5), but it is unlikely that all three sites could be occupied simultaneously during catalysis. In the crosslinked Gβγ–PLCβ3 complex, the orientation of Gβγ is unlikely to be compatible with the binding of the lipase active site to the membrane (Figures 1, 5A). We propose this reflects an encounter complex, or pre-activation state, given the specificity and efficiency of the crosslinking reaction (Figure S1). In contrast, both Gβγ molecules in the liposome-tethered Gβγ–PLCβ3 complex and the lipase active site can simultaneously interact with the membrane (Figure 5B). Our observation that disruption of either of these sites can severely compromise Gβγ binding suggests a degree of cooperativity between these sites.

Figure 5.

Comparisons of the predicted Gβγ binding sites in PLCβ1-4 provide some insights into the isoform-specific differences in Gβγ-dependent activation. Of the nine residues we identified as functionally relevant across the three PLCβ3 Gβγ binding sites (Figures 2, 3), only two are conserved in PLCβ. The other residues are incompatible with Gβγ binding and explain why Gβγ does not activate PLCβ4 (43, 44). In PLCβ1 and PLCβ2, seven of the residues identified are conserved, the exceptions being PLCβ3 R204 and R215. In PLCβ1, these residues are replaced by proline and valine, respectively, and in PLCβ2 they are asparagine and serine. Whether these differences are sufficient to explain why PLCβ1 is weakly activated by Gβγ while PLCβ2 is robustly activated remains to be established (1).

Previous work has shown that PLCβ3 must be preactivated, or simultaneously activated, by Gα_q·GTP in cells before Gβγ binds to further increase PIP2 hydrolysis (14). This is consistent with the fact that all Gβγ–PLCβ3 reconstructions are fully compatible with simultaneous binding Gα_q·GTP at the membrane (Figure 5C, D). Gα_q binding is essential for translocation of the lipase to the plasma membrane (6), but mutation of any of the three structurally characterized Gβγ binding sites has no impact on lipase translocation (Figure 3). Moreover, free Gβγ released by activation of G_i/o heterotrimers does not promote translocation (6). While Gβγ clearly does not increase lipase activity via membrane recruitment, its binding to the lipase is essential for maximum PIP2 hydrolysis (Figure 3), even when PLCβ3 is tethered at the membrane (Figure 4). While Gβγ recruitment and reorientation of PLCβ3 at the membrane were proposed as activation mechanisms, they were not directly tested. Here, we have elucidated this mechanism, and our results demonstrate that Gβγ is a positive allosteric modulator of PLCβ3, following direct activation of the phospholipase by Gα_q, as opposed to a bona fide activator. In this paradigm, Gβγ binding to the PH domain and/or EF hands optimizes the orientation of Gα_q–PLCβ3 at the membrane to facilitate interfacial activation and maximize PIP2 hydrolysis (Figure 5E).

Finally, our results show that Gβγ is surprisingly important for PLCβ3 activation by G_q heterotrimers in cells. While synergistic activation of PLCβ3 by Gα_q and Gβγ is well known, this synergy is typically discussed as a mechanism of crosstalk between G_q and G_i signaling, the latter being the source of Gβγ dimers (45, 46). Gβγ affinity for PLCβ3 is relatively low and only G_i/o heterotrimers are thought to be expressed at high enough levels to release sufficient free Gβγ. Here we show that PIP2 hydrolysis downstream of G_q alone is sensitive to sequestration of Gβγ dimers, consistent with a previous study in HeLa cells (47), as well as mutations that disrupt Gβγ binding. Endogenous G_q heterotrimers are evidently capable of releasing sufficient free Gβγ to occupy binding sites on PLCβ3. This is most likely enabled by cooperative binding of G protein subunits to PLCβ3 at the plasma membrane.

Materials and Methods

Cell culture and transfection

Human embryonic kidney HEK 293 cells were obtained from ATTC (CRL-1573) and used as supplied or after gene editing as described previously (15, 58); ΔPLCβ1-4 (ΔPLC) cells were generated using CRISPR/Cas9 and validated as described previously (15). Cells were transfected in 6-well plates in growth medium using linear polyethyleneimine MAX (Polysciences) at a nitrogen/phosphate ratio of 20 and were used for experiments 24-48 hours later. Up to 3.0 μg of plasmid DNA was transfected in each well of a 6-well plate.COS-7 cells were a gift from A.V. Smrcka and used for [³H]-IP_x accumulation assays.

Plasmids

SNAPf-AGTR1, HiBit-PLCβ3, HiBit-PLCβ3-CAAX, mem-link-Venus, mem-GRK3ct, mem-link-amber-GRK2RH, Nluc-PH, Gαq-Venus, CMV-LgBit, Venus-1-155-Gγ2 and Venus-156-239-Gβ1 were described previously (6). Mutations in HiBit-PLCβ3, HiBit-PLCβ3-CAAX were made by amplifying three fragments from wild-type plasmids with the desired mutation incorporated in a primer, assembling using Gibson assembly and verifying by full plasmid sequencing. Mutations in human PLCβ3 in pFastbac1, pCMV, or pCDNA3.1 were generated using the Takara infusion site-directed mutagenesis kit (Takara) (Supplemental Table 1) and verified by full plasmid sequencing. The same strategy was used to subclone Gβ₁ and Gγ₂ in pCI-Neo(59) and Gα_q in pCDNA3.1+.

Protein expression, purification, and complex formation

Protein expression

PLCβ3, PLCβ3 Δ892 and variants, Gβγ, and Gβγ C68S were expressed and purified from baculovirus-infected insect cells. Baculoviruses were generated using the FastBac recombinant baculovirus system (Invitrogen/Thermo Fisher Scientific, Inc.) in ESF 921 Insect Cell Culture Medium (Expression Systems)-adapted Sf9 (Spodoptera frugiperda) cells.

Purification of Gβγ and Gβγ C68S

High Five cells were infected with baculoviruses encoding His6-Gα_i1, Gβ₁, and Gγ₂ (49, 60). Cells were harvested 60 h post-infection by centrifugation at 2,500 x g, frozen in liquid N₂ and stored at −80 °C.

All purification steps were performed at 4 °C unless otherwise indicated. Cell pellets were thawed in 15 mL of lysis buffer (50 mM HEPES pH 8.0, 3 mM MgCl₂, 10 mM β-mercaptoethanol, 0.1 mM EDTA, 100 mM NaCl, 10 μM GDP, and protease inhibitors (133 μM PMSF, 21 μg/mL TLCK, and 0.5 μg/mL TPCK) and lysed by four freeze-thaw cycles in liquid nitrogen. Lysate was diluted to 100 mL with lysis buffer and centrifuged at 100,000 x g for 30 min to isolate the membrane fraction. The membrane pellets were resuspended by dounce in 5 mL extraction buffer (50 mM HEPES pH 8.0, 3 mM MgCl₂, 50 mM NaCl, 10 mM β-mercaptoethanol, 10 μM GDP, and protease inhibitors), combined and diluted to 60 mL. Cholate was added to a final concentration of 1% and the mixture stirred for 1 h at 4 °C to extract membrane proteins. Detergent extracts were clarified by centrifugation at 100,000 x g for 45 min. The supernatant was diluted five-fold with buffer A (50 mM HEPES pH 8.0, 3 mM MgCl₂, 10 mM β-mercaptoethanol, 100 mM NaCl, 10 μM GDP, 0.5% polyoxyethylene(10) lauryl ether (C12E10), and protease inhibitors) and applied to a cOmplete™ His-Tag Purification Resin (Roche) column pre-equilibrated with buffer A. The column was washed with 100 mL of buffer A supplemented with 300 mM NaCl and 5 mM imidazole, then transferred to room temperature and washed with 12 mL buffer A. Gβ₁γ₂ subunits were released from His6-Gα_i1 using six 4 mL fractions of RT buffer (buffer A supplemented with 150 mM NaCl, 5 mM imidazole, 50 mM MgCl₂, 10 mM NaF, 10 μM AlCl₃, and 1% cholate). Fractions were analyzed by SDS-PAGE and Coomassie staining to assess purity. Fractions containing Gβγ were pooled and applied to a MonoQ column pre-equilibrated with 20 mM HEPES, pH 8, 1 mM DTT, 50 mM NaCl and 0.5% CHAPS, and eluted with a 50-500 mM NaCl gradient. Fractions containing purified protein were identified by SDS-PAGE, concentrated to 20-40 μM, flash frozen in liquid N₂, and stored at −80 °C.

Gβγ C68S was purified as described with some modifications. Briefly, after cell lysis, the supernatant was diluted five-fold with buffer A lacking polyoxyethylene(10) lauryl ether, glass fiber-filtered, and applied to cOmplete™ His-Tag Purification Resin (Roche) as described. The rest of the steps were carried out as described, with the omission of cholate or CHAPS from the buffers.

Purification of PLCβ3 and variants

His6-PLCβ3, PLCβ3 Δ892, PLCβ3 Δ892-PH_cys and PLCβ3 Δ892-XY_cys were expressed in Sf9 insect cells grown in S6900 II serum-free media and infected with baculovirus at an MOI of 1(52). After 48 h, cells were harvested by centrifugation, frozen in liquid N₂ and stored at −80°C. Cells were rapidly thawed and lysed by four freeze-thaw cycles in liquid nitrogen in lysis buffer (20 mM HEPES, pH 8, 50 mM NaCl, 10 mM β-mercaptoethanol, 0.1 mM EDTA, 0.1 M EGTA, 133 μM PMSF, 21 μg/ml TLCK and TPCK, 0.5 μg/ml aprotinin, 0.2 μg/ml Leupeptin, 1 μg/ml Pepstatin A, 42 μg/ml Tosyl-L-Arginine Methyl Ester (TAME), 10 μg/ml Soy Bean Trypsin Inhibitor (SBTI)) Lysed cells were collected and diluted with lysis buffer and NaCl to a final concentration of 800 mM NaCl, and centrifuged at 100,000 × g for 1 h. The supernatant was diluted 5 fold with lysis buffer containing 0.5% polyoxyethylene(10) lauryl ether (C12E10) and centrifuged again at 100,000 × g for 1 h The supernatant was loaded onto a cOmplete™ His-Tag Purification Resin (Roche) column pre-equilibrated with buffer A (20 mM HEPES, pH 8, 100 mM NaCl, 10 mM β-mercaptoethanol, 0.1 mM EDTA, and 0.1 M EGTA). The column was washed with 3 column volumes (CVs) of buffer A, followed by 3 CVs of buffer A supplemented with 300 mM NaCl and 10 mM imidazole. The protein was eluted with 3-10 CVs of buffer A supplemented with 200 mM imidazole. Proteins were concentrated and loaded onto tandem Superdex 200 columns (10/300 GL; GE Healthcare) equilibrated with SEC buffer (20 mM HEPES pH 8, 200 mM NaCl, 2 mM DTT, 0.1 mM EDTA, and 0.1 M EGTA). Fractions containing purified protein were identified by SDS-PAGE and were pooled, concentrated, and flash frozen in liquid N₂.

Crosslinking and complex isolation

Purified Gβγ-C68S, Gβγ, PLCβ3 Δ892, PLCβ3 Δ892-PH_cys, and PLCβ3 Δ892-XY_cys were buffer exchanged to remove DTT by concentrating the proteins in an Amicon Ultra 0.5 ml 30 K concentrator (Millipore-Sigma) and washing them twice with 20 mM HEPES pH 7.4, 100 mM NaCl, 0.1 mM EDTA and 0.1 mM EGTA (and 0.5% CHAPS for Gβγ). 25 μM of the buffer-exchanged Gβγ and 25 μM PLCβ3 Δ892, PLCβ3 Δ892-PH_cys or PLCβ3 Δ892-XY_cys were mixed and crosslinking initiated by addition of 200 μM BMOE or BMPEG2. Reactions were incubated for 45 min at room temperature and quenched by addition of 20 mM DTT. Crosslinking of Gβγ and PLCβ3 Δ892-PH_cys were completed as above but contained 0.5% CHAPS. Crosslinking was confirmed by SDS-PAGE by the presence of a band at ∼135 kDa, consistent with a 1:1 stoichiometric complex (Gβ MW: 35 kDa, PLCβ3 D892 MW: 100.89 kDa)

Cryo-EM

Sample preparation and data collection

For the BMOE-crosslinked and BM(PEG)2-crosslinked Gβγ C68S–PLCβ3 Δ892-PH_cys complexes, 3.5 μL of purified complex at 1 mg/mL supplemented with 0.2 % CHAPS_(f) ∼5 min before blotting was applied onto glow-discharged Quantifoil R1.2/1.3 300-mesh grids and prepared and imaged as described for PLCβ3. For the BMOE sample, Micrographs were collected on a Titan Krios G1 electron microscope (FEI) equipped with a post-GIF K3 direct electron detector (Gatan) in the Purdue Life Science Cryo-EM facility. A dataset containing ∼4,515 images was collected in super-resolution mode with a pixel size of 0.539 Å, at a defocus range of 1-3 μm using Leginon. For each movie stack, 40 frames were recorded at a frame rate of 78 ms per frame and a total dose of 53.69 electrons/Å2. Micrographs were collected on a Titan Krios G4 electron microscope (FEI) equipped with a Post-GIF K3 direct electron detector (Gatan) in the Purdue Life Science Cryo-EM facility. A dataset containing ∼4,587 images was collected in super-resolution mode with a pixel size of 0.539 Å, at a defocus range of 1-3 μm using EPU. For each movie stack, 40 frames were recorded at a frame rate of 78 ms per frame and a total dose of 53.43 electrons/Ų.

Data processing

Micrographs were motion aligned and motion corrected using motioncor2 (61) implemented within CryoSPARC(53). CTF estimations were completed using CTFfind4 (62). Particle picking, 2D classifications, initial model generation, 3D classification and refinement were all performed using CryoSPARC. Workflows for each data set are shown in Figures S4, S5, and S6. The nominal resolution was determined based on a Fourier shell correlation cutoff of 0.143.

Model building and refinement

Crystal structures of Gβγ and PLCβ3 (PDB IDs 1GP2 and 4GNK, (5, 63)) were rigid-body fit into the cryo-EM map using Chimera (64). The model was then refined using molecular dynamic flexible fitting (MDFF)(54). MDFF configuration files were generated using VMD. During MDFF simulation, Gβγ was set as rigid with domain restraints. The MDFF simulation was conducted with a grid scaling value of 0.5 for 100 ps, followed by 3,000 steps of energy minimization until convergence of the protein RMSD. The MDFF generated model was inspected and manually adjusted in COOT(55), guided through the use of deep-learning-based amino-acid-wise model quality (DAQ) scoring (65, 66) and refined in PHENIX(56). Resulting models were assessed in PHENIX for stereochemical correctness. Maps, half maps, and coordinate files were deposited in the PDB as 9Y7H, 9YAO, and 9YAP and in the EMDB as EMD-72655, EMD-72732, and EMD-72733.

Activity assays

Inositol phosphate accumulation

COS-7 cells were seeded in 12-well culture dishes at a density of 100,000 cells per well and maintained in Dulbecco’s modified Eagle’s medium containing 10% fetal bovine serum (Atlanta Bio), 1X Glutamax (Gibco), 100 units/mL penicillin, and 100 μg/mL streptomycin (Corning) at 37 °C and 5% CO₂. Cells were transfected with 400 ng of PLCβ3 variant and 200 ng G protein subunit using Fugene 6 (Promega) at a 3:1 ratio per manufacturer’s protocol. Total DNA varied from 700-900 ng per well, with pCMV used as an empty vector. 18-24 h after transfection, the media was changed to low-inositol Ham’s F-10 medium (Gibco) containing 1.5 μCi/well myo-[2-³H(N)] inositol (Perkin Elmer) for 16-18 h, then treated with 10 mM LiCl for 1 h to inhibit inositol phosphatases. Media was aspirated, cells were washed twice with PBS, then lysed by addition of ice-cold 50 mM formic acid. Lysates containing [³H]-inositol phosphate were applied to Dowex AGX8 columns to isolate the IP species. Columns were washed twice with 10 CVs of 50 mM formic acid, then 100 mM formic acid, and eluted with 3 CVs of 1.2 M ammonium formate into scintillation vials containing scintillation fluid and counted.

Western blotting

Cells were lysed in SDS sample buffer (100 mM Tris pH 6.8, 6% sucrose, 2% SDS, 715 mM β-mercaptoethanol, and 0.02% bromophenol blue), boiled, and run on a 10% (w/v) SDS–polyacrylamide gel. Proteins were transferred to PDVF for 16 hours at 25 V, followed by incubation with an antibody against PLCβ3 (Cell Signaling Cat: D9D6S) (1:1000), Gβ1 (Thermo: Cat: PA530046) or actin (Cell Signaling: 8H10D10) (1:2000). Goat anti-rabbit HRP or goat anti-rabbit AlexaFlour 800 antibodies (1:10,000) were added before visualizing with ECL reagent (Pierce) for HRP linked antibodies. Western blots were imaged with a GeneGnome imaging system or Azure Sapphire FL respectively.

Liposome-based activity assay

Hen egg white phosphatidylethanolamine (PE) at 100 μM and soy phosphatidylinositol (PI) at 250 μM (Avanti Polar Lipids) were resuspended in CHCl₃, mixed, and dried in 312 μL aliquots in borosilicate glass tubes under N₂, sealed and stored at −20 °C until use. To prepare liposomes, the lipids were resuspended in 312 μL of sonication buffer (50 mM HEPES pH 7, 80 mM KCl, 2 mM EGTA, and 1 mM DTT) and incubated at room temperature for 5 min, then sonicated to clarity in 30 second duty cycles using a bath sonicator (Avanti Polar Lipids). Each reaction mixture contained 10 μL of liposome solution,10 μL of PLC solution (50 mM HEPES pH 7, 3 mM EGTA, 80 mM KCl, 3 mM DTT, and 3 mg/mL BSA), 5 μL of Gβγ solution or buffer (50 mM HEPES pH 7, 100 mM NaCl, 5 mM MgCl, 1 mM DTT, and 3 mM EGTA), and 5 μL of CaCl₂ solution (50 mM HEPES pH 7, 3 mM EGTA, 80 mM KCl, 1 mM DTT, and 18 mM CaCl₂). CaCl₂ solution was added last to initiate PI hydrolysis upon transfer to 30°C and incubated for 15 minutes. Control reactions contained all components except CaCl₂. Reactions were terminated by the addition of an ice-cold quench solution (50 mM HEPES pH 7, 80 mM KCl, 210 mM EGTA, and 1 mM DTT), and incubated on ice.

Inositol phosphate (IP) was quantified using a modified version of the CisBio IP-One Gq assay kit. Following termination, 14 μL of each reaction mixture, 3 μL of d2-labeled IP1, and 3 μL of the cryptate-labeled anti-IP1 antibody (CisBio) were added to a 384-well low-volume white microplate at room temperature (Corning, Corning, NY). Positive controls contained assay buffer, d2-labeled IP1, and cryptate-labeled anti-IP1, and negative controls contained assay buffer, lysis and detection buffers, and cryptate-labeled anti-IP1. The plate was centrifuged at 1000 x g for 1 min, incubated at room temperature for 1 h, and fluorescence read on a SynergyNeo2 plate reader (BioTek). The concentration of IP1 was calculated from a standard curve and normalized following the manufacturer’s protocol (CisBio).

BRET

For Venus-Gβγ interaction experiments HEK 293 cells were transfected with 0.01 μg HiBit-PLCβ, 0.1 μg CMV-LgBit, 0.4 μg Gα_q, 0.3 μg Venus-1-155-Gγ2, 0.3 μg Venus-155-239-Gβ1 and 0.3 μg SNAPf-AGTR1. For Gαq-Venus interaction experiments Gα_q-Venus replaced Gα_q, and unlabeled Gβ1 and Gγ2 replaced their Venus-labeled counterparts. For PIP2 hydrolysis assays HEK cells were transfected with 0.01 μg Nluc-PH, 0.1 μg SNAPf-AGTR1 and 0.8 μg of mem-link-Venus. For PIP2 hydrolysis reconstitution assays ΔPLC cells were transfected with the same components plus 0.01 μg of HiBit-PLCβ, HiBit-PLCβ-CAAX or variants thereof. To sequester either active Gα_q or Gβγ GRK2RH or GRK3ct were added at 0.2 μg per well, respectively. For translocation bystander BRET experiments HEK293 cells were transfected with 1 μg mem-link-Venus, 0.1 μg CMV-LgBiT, 0.3 μg SNAPf-AGTR1, 0.4 Gα_q and 0.01 μg HiBit-PLCβ or variant. Cells were washed and resuspended in Dulbecco’s phosphate buffered saline (DPBS) and distributed to 96-well plates in suspension immediately before taking BRET measurements. All BRET measurements were made in the presence of furimazine (Promega or ChemShuttle; 1:1,000 as supplied or from a 5 mM stock dissolved in 90% ethanol/10% glycerol). BRET and luminescence measurements were made using a Polarstar Optima or Lumistar Omega plate reader (BMG Labtech); angiotensin II was injected from a 10-fold concentrated solution. Raw BRET signals were calculated as the emission intensity at 520–545 nm divided by the emission intensity at 475–495 nm. Net BRET signals were calculated as the raw BRET signal minus the raw BRET signal measured from cells expressing only the donor.

Statistical analysis

All analysis was carried out using GraphPad Prism Version 10.5.0.

Materials availability

All unique materials produced for this study are available from the corresponding author upon request.

Data availability

All assay data, including western blots, CPMB counts, and luminescence will be made available in the Purdue University Research Repository under a single DOI. Maps, half maps, and coordinate files were deposited in the PDB as 9Y7H, 9YAO, and 9YAP and in the EMDB as EMD-72655, EMD-72732, and EMD-72733.

Acknowledgements

We thank Steve Wilson and Drs. John. J. G. Tesmer, Val J. Watts, Thomas Klose, and Frank Vago for technical and conceptual assistance.

Additional information

Funding

A.I. was funded by Japan Society for the Promotion of Science (JP24K21281 and JP25H01016); Japan Science and Technology Agency (JPMJFR215T and JPMJMS2023); Japan Agency for Medical Research and Development (JP22ama121038 and JP22zf0127007); and The Uehara Memorial Foundation. E.K. was supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) with the grant 214362475/GRK1873/3. This work is supported by F32GM145110-01 to I.J.F, R35GM145284 to N.A.L., 1R01HL141076 and 1R01GM152701 to A.M.L.

Funding

Japan Society for the Promotion of Science (JP24K21281)

• Asuka Inoue

MEXT | Japan Society for the Promotion of Science (JSPS) (JP25H01016)

• Asuka Inoue

MEXT | Japan Science and Technology Agency (JST)

https://doi.org/10.52926/jpmjfr215t

• Asuka Inoue

MEXT | Japan Science and Technology Agency (JST)

https://doi.org/10.52926/jpmjms2023

• Asuka Inoue

Japan Agency for Medical Research and Development (AMED) (JP22ama121038)

• Asuka Inoue

Japan Agency for Medical Research and Development (AMED) (JP22zf0127007)

• Asuka Inoue

Uehara Memorial Foundation (UMF)

• Asuka Inoue

Deutsche Forschungsgemeinschaft (DFG) (214362475/GRK1873/3)

• Evi Kostenis

HHS | NIH | National Heart, Lung, and Blood Institute (NHLBI) (1R01HL141076)

• Angeline Lyon

HHS | NIH | National Institute of General Medical Sciences (NIGMS) (F32GM145110-01)

• Isaac J Fisher

HHS | NIH | National Institute of General Medical Sciences (NIGMS) (R35GM145284)

• Nevin A Lambert

HHS | NIH | National Institute of General Medical Sciences (NIGMS) (1R01GM152701)

• Angeline Lyon