G protein βγ subunits directly interact with and activate phospholipase C

Phospholipase C (PLC) enzymes hydrolyze membrane phosphatidylinositol 4,5 bisphosphate (PIP 2 ) and regulate Ca 2+ and protein kinase signaling in virtually all mammalian cell types. Chronic activation of the isoform downstream of protein–coupled receptors to the development of cardiac We previously shown that PLC ε -catalyzed hydrolysis of Golgi-associated phosphatidylinositol 4-phosphate (PI4P) in cardiac myocytes depends on G protein βγ subunits released upon stimulation with endothelin-1. PLC ε binds and is directly activated by Ras family small GTPases, but whether they directly interact with G βγ has not been demonstrated. To identify PLC ε domains that interact with G βγ , here we designed various single substitutions and truncations of wild-type PLC ε and tested them for activation by G βγ in transfected COS-7 cells. Deletion of only a single domain in PLC ε was not sufficient to completely block its activation by G βγ , but blocked activation by domains completely abrogated PLC ε activation by G βγ , but activation by the GTPase Rho was retained. In vitro reconstitution experiments further revealed that purified G βγ directly interacts with a purified fragment of PLC ε (PLC ε –PH–RA2) and increases PIP 2 hydrolysis. Deletion of the RA2 domain decreased G βγ binding and eliminated G βγ stimulation of PIP 2 hydrolysis. These results provide first evidence that G βγ directly interacts with PLC ε and yield insights into the mechanism by which βγ subunits activate PLC ε .


Introduction
G protein coupled receptors (GPCRs) regulate a wide variety of cellular functions through stimulation of a few canonical signal transduction cascades. One major GPCRstimulated pathway is the agonist-dependent hydrolysis of phosphatidylinositol 4,5 bisphosphate (PIP 2 ) by phospholipase C (PLC) to regulate intracellular Ca 2+ and protein kinase activation (3). PLC signaling is stimulated by both receptor tyrosine kinases (RTKs) and GPCRs and regulates multiple cellular functions in normal and disease states. Thirteen PLC isoforms have been identified with some common structural features (4)(5)(6)(7). All PLC isoforms contain a pleckstrin homology (PH) domain, an EF-hand domain, a highly conserved catalytic core domain (composed of the X and Y domains separated by a linker region) and a phospholipid-binding C2 domain. The diversity in regulation of the PLCs arises in large part from regions outside of the conserved core domains.
The PLCβ family of PLCs are directly activated by Gα q or Gβγ subunits with varying efficacies and potencies (8)(9)(10)(11)(12)(13). PLCε is a more recently discovered PLC isoform and is also downstream of GPCRs, as well as RTKs (14)(15)(16)(17). PLCε binds directly to members of the Ras family of small GTPases including Ras, Rap, and Rho. Ras family members can be activated by both GPCRs and RTKs and subsequently activate PLCε (7). G protein βγ subunits have also been shown to regulate PLCε in a COS-7 cell co-transfection system, but direct activation by Gβγ has not been demonstrated (18).
PLCε has a unique domain architecture where apart from conserved core domains, it has an amino-terminal Cysteine-rich (Cys) and CDC25 guanine nucleotide exchange factor (GEF) domains that activates Rap, and two carboxy-terminal Ras-association domains (RA1 and RA2) (Fig. 1A). A putative PH domain was identified in silico, but its properties have not been studied biochemically (18). The RA2 domain binds directly to Ras and is required for activation by Ras (14). A domain required for activation by Rho is in the Y catalytic domain but a direct binding site for Rho has not been identified (19). While Gβγ has been shown to activate PLCε, the domains on PLCε required for this activation have not been delineated (19).
We recently found that Gβγ-dependent regulation of PLCε at the Golgi apparatus in cardiac myocytes regulates local PI4P hydrolysis, protein kinase D (PKD) activation, gene expression and hypertrophy (2). Thus understanding how PLCε is regulated by Gβγ is important to fully understand this process. Here we identified two independent domains of PLCε that are involved in its regulation by Gβγ and demonstrate direct interactions between Gβγ and PLCε.
These data place PLCε activation directly downstream of GPCRs.

Results
Regulation of PLCε by Gβγ subunits. G proteindependent activation of PLCε is difficult to reconstitute with purified G proteins (14,19) so to understand the mechanism by which Gβγ regulates PLCε, we confirmed previous work showing that Gβγ co-transfected with PLCε increases accumulation of inositol phosphates (IP) in COS-7 cells labeled with [ 3 H]-inositol (18). When Gβ and γ subunits are transfected in the absence of transfected PLC, increased accumulation of IP accumulation is observed due to stimulation of endogenous PLCs present in COS-7 cells (Fig 1B and C) (20). However, cells co-transfected with Gβγ and PLCε markedly increased IP accumulation in a concentration-dependent manner ( Fig. 1B and   C). To confirm that free Gβγ is necessary for activation of PLCε in COS-7 cells, we included Gα i1 in the co-transfection to bind and sequester free Gβγ. In the presence of Gα i1 , Gβγdependent accumulation of IP was inhibited (Fig. 1C). These results indicate that activation of PLCε by Gβγ requires free Gβγ and confirms previously reported Gβγ-dependent activation of PLCε (18).
Effect of carboxy-terminal domain mutation and deletion on activation of PLCε by Gβγ. Gβγ has been reported to activate Ras (21). Mutation of the lysine residues 2150 and 2152 in the PLCε RA2 domain to glutamate ( Fig. 2A) inhibits activation of PLCε by Ras (14). Western blot analysis of the mutants showed similar expression of PLCε WT and K-E (Fig. 2B). To investigate whether activation of PLCε by Gβγ in COS-7 cells is mediated indirectly through endogenous activated Ras, we tested PLCε 2150/2152 K-E for activation by Gβγ and Rho to rule out indirect activation by endogenous Ras. Co-transfection of PLCε K-E with Gβγ or constitutively active Rho (RhoG14V) in COS-7 cells resulted in marked accumulation of IP ( Fig.  2C left and middle panels) in contrast to constitutively activated Ras (RasG12V) cotransfected cells (Fig. 2C right panel). These results confirm previously published data indicating that activation of PLCε by Gβγ or Rho is independent of endogenous Ras activation in COS-7 cells (18).
To determine if the RA2 domain of PLCε is required for its activation by Gβγ, we next deleted residues 2114-2281 comprising the RA2 domain and tested for activation by Gβγ, Rho and Ras ( Fig. 2A). The unstimulated basal activity of PLCε-ΔRA2 when co-transfected with empty vector was significantly lower than PLCε WT despite similar levels of expression ( Fig. 2B and D). RasG12V was unable to activate this mutant as expected (Fig. 2D, right panel). Co-transfection of this mutant with RhoG14V caused significant accumulation of IP but not to the levels of PLCε WT indicating the enzyme is still active and can be activated ( Fig.  2D middle panel). Gβγ transfection resulted in a small but statistically significant increase in IP accumulation when PLCε-ΔRA2 is co-transfected compared control with Gβγ only transfected ( Fig. 2D left panel). The reduced levels of IP accumulation may be because of the lower basal activity of PLCε-ΔRA2. However, because the Rho activation is relatively robust it suggests that the RA2 domain is important for Gβγ stimulation. This will be explored in more detail later.

Effect of amino-terminal domain deletions on
PLCε activation by Gβγ. To investigate the involvement of the amino-terminal domains of PLCε in Gβγ-mediated activation, a series of amino-terminal truncations of PLCε were created (Fig. 3A). It has been reported that the amino-terminal domains are not essential for PLCε activation by Gβγ or Rho (19). The basal activity of PLCε-CDC25-RA2 was significantly higher than PLCε WT when co-transfected with empty vector (Fig. 3C), while PLCε-PH-RA2 and EF-RA2 had decreased basal activity compared to PLCε WT ( Fig. 3D and E). There was no correlation between expression levels of the mutant PLCε proteins (Fig. 3B) and observed basal activities (Fig. 3C, D, and E). PLCε-PH-RA2 and EF-RA2 had significantly reduced capacity for activation by free Gβγ, while activation by RhoG14V was largely retained (Fig. 3C, D, and E). The reduced accumulation of IP in the presence of the mutants (PH-RA1 and EF-RA2) and Gβγ or RhoG14V may be due to their lower basal activity, however Gβγ stimulation of the EF-RA2 PLCε was still minimal. Overall, our results indicate that deletion of the aminoterminal domains is not sufficient to completely inhibit activation of PLCε by Gβγ but significantly reduces it.

Effect of simultaneous amino-terminal and carboxy-terminal domain deletions on PLCε activation by Gβγ.
The results of the foregoing experiments could indicate that some PLCε activation by Gβγ requires interaction with only the catalytic core. An alternate hypothesis is that PLCε activation by Gβγ may involve multiple interaction sites at the carboxy-terminus and the amino-terminus. To test these hypotheses, we made a series of combined truncations of PLCε that lack both the amino and carboxy-terminal domains as shown in Figure 4A. COS-7 cells were transfected with either Gβγ or RhoG14V and one of the mutant PLCε constructs shown in Figure 4A. A western blot showing expression levels of these constructs are shown in Figure 4B. Although the basal activity of CDC25-RA1 is significantly lower than PLCε WT as shown in Figure 4C, cotransfection with Gβγ resulted in a small but statistically significant accumulation of IP relative to control without PLCε co-expression. RhoG14V was also able to activate this mutant. Deletion of the CDC25 domain completely eliminated Gβγ-dependent accumulation of IP above levels of cells transfected with Gβγ alone (Fig. 4D, left panel). In contrast, co-transfection of this PLCε mutant with RhoG14V resulted in some Rho-stimulated accumulation of IP confirming that this construct can be activated to some level ( Fig. 4D right panel). EF-RA1 is also not activated upon co-transfection with Gβγ. In this case co-transfection with RhoG14V also did not result in accumulation of IP above levels of cells transfected with Gβγ or Rho alone (Fig. 4E). These data together suggest that PLCε activation by Gβγ requires both the amino (Cysteine-rich and CDC25 domains) and carboxy termini (RA2 domain) for full activation of PLCε.
Constitutive membrane localization of PLCε-EF-RA1 rescues Rho-dependent activation, but not Gβγ-dependent activation. Confounding interpretation of these experiments is the reduction in basal activity associated with many of the domain deletion mutations. In particular, the EF-RA1 construct shows little basal activity and no activation by either Rho or Gβγ. We hypothesized that this could be due to a loss in membrane association of the enzyme. To test this, we designed a EF-RA1 variant with a Cterminal CAAX box to allow for carboxyterminal prenylation and constitutive membrane localization as shown in previous studies (22) (Fig. 5A). Western blot analysis of the mutants showed similar levels of expression of EF-RA1 and EF-RA1 CAAX, which were slightly increased relative to PLCε WT expression (Fig.  5B). In contrast to EF-RA1 transfected cells, EF-RA1 CAAX transfected cells showed elevated Regulation of PLCε by Gβγ basal activity above empty vector transfected cells indicating that this construct was indeed active when localized to the plasma membrane (Fig. 5C). Interestingly, the EF-RA1 CAAX mutant rescued activation by Rho but not Gβγ (Fig. 5C). The data indicate that a PLCε mutant lacking the Cysteine-rich, CDC25, and RA2 domains can be activated by Rho when targeted to the membrane, but cannot be stimulated by Gβγ. This data strongly supports these domains as being required for Gβγ activation.

PLCε directly interacts with Gβγ and increases PIP 2 hydrolysis.
To test whether Gβγ directly interacts with PLCε, PLCε truncations lacking the Cysteine-rich and CDC25 domains (PH-RA2), or the Cysteine-rich, CDC25, and RA2 domains (PH-RA1) were expressed and purified (Figs. 6A and 7A). To test for direct binding of Gβγ to PLCε, purified biotinylated Gβγ (bGβγ) was mixed with purified PLCε-PH-RA2 and precipitated with neutravidin beads. PLCε-PH-RA2 was readily detected associated with beads bound to bGβγ, but not beads alone (Fig. 6B). The two bands observed are likely the result of proteolysis that occurs during the incubation. To test the specificity of Gβγ interaction with PLCε, purified PLCε-PH-RA2 and bGβγ were incubated with or without Gα i1 -GDP to sequester free Gβγ. Gα i1 -GDP strongly inhibited the pulldown of PLCε-PH-RA2 by bGβγ (Fig. 6B, C), demonstrating a direct and specific interaction between PLCε and Gβγ.
Unlike the PLCβ isoforms, reconstitution of PLCε activation with purified components has been difficult to demonstrate, and generally the amount of PLCε activation observed upon incubation with purified upstream activators such as Rho and Ras is minimal (19,23). Nevertheless, to test the specificity of Gβγ activation of PLCε, purified PLCε-PH-RA2 and Gβγ were incubated with or without Gα i1 -GDP in reconstituted lipid vesicles containing [ 3 H]inositol PIP 2 , and the amount of [ 3 H]-IP 3 produced was measured. Figure 6D shows a small but statistically significant increase in PIP 2 hydrolysis by PLCε-PH-RA2 when Gβγ is added. This activation was inhibited by Gα i1 -GDP confirming that this increase is through direct interactions of PLCε with Gβγ. This is also the first direct demonstration of Gβγdependent activation of PLCε using reconstituted proteins.
Our data from the COS-7 cell cotransfection experiments indicate that PLCε activation by Gβγ requires both the amino (Cysteine-rich and CDC25 domains) and carboxy-terminus (RA2 domain) that are potentially compensatory in the absence of one or the other. Therefore, we hypothesized that if PLCε is directly regulated by Gβγ binding to the RA2 domain, Gβγ will bind to PLCε-PH-RA2 but not PH-RA1. To test this, purified bGβγ and amino-and carboxy-terminally truncated PLCε mutants (PH-RA2 or PH-RA1) were incubated and then precipitated with neutravidin beads. As shown in Figure 7B and C, PLCε-PH-RA2 was strongly pulled down with bGβγ bound to beads. In contrast PLCε-PH-RA1 was only weakly pulled down with bGβγ.
We next tested the ability of purified Gβγ to activate purified PLCε-PH-RA2 or PH-RA1 in reconstituted lipid vesicles containing [ 3 H]-PIP 2 . PH-RA1 had significantly less basal activity than PH-RA2 (Fig. 7D). Gβγ also activated PLCε-PH-RA2, but not PH-RA1, in a concentration-dependent manner (Fig. 7D) consistent with the decreased binding of Gβγ to PH-RA1 (Fig. 7B) and the lack of activation of this construct in COS-7 cells (Fig. 4D). Overall, these results suggest that the activation of PLCε involves a direct interaction with Gβγ that requires the carboxy-terminus RA2 domain and the amino terminal Cysteine-rich and CDC25 domains.

Discussion
In our previous work we showed that inhibiting Gβγ with the carboxy terminus of G protein coupled receptor kinase 2 (GRK2ct), or with gallein, inhibited PLCε-dependent PI4P hydrolysis in cardiac myocytes (2). Activation of PLCε can be mediated by small GTPases and/or Gβγ (14,15,18,19).
To date, direct regulation of PLCε by Gβγ has not been reported. If indeed free Gβγ subunits activate PLCε, Gα i1 -GDP should form an inactive heterotrimer with Gβγ, preventing PLCε activation. We confirmed that co-transfection of cells with cDNAs encoding Gβγ, PLCε and Gα i1 completely prevented activation of PLCε Regulation of PLCε by Gβγ mediated by free Gβγ. Our results also demonstrate that Gβγ directly interacts with PLCε to increase PIP 2 hydrolysis. In addition, the results from domain deletion analysis of PLCε show an unexpected involvement of the RA2, and CDC25 domains in the activation of PLCε by Gβγ.
Regulation of PLCε by Ras is through direct binding to the RA2 domain (14). The regulation of PLCε by Gβγ is not understood. It was previously reported that PLCε lacking the amino-terminal domains retained activation by Gβγ subunits, Ras, and Rho, suggesting activation was not dependent upon the PLCε amino-terminal domains (19). In addition, it was also reported that PLCε-CDC25-C2 can be stimulated by Gβγ subunits, but not by Ras, suggesting that Gβγ activation was independent of the RA2 domain (19). Our results generally support these findings.
The most compelling data indicating that Gβγ may be interacting with both the CDC25 domain and the RA2 domain is that simultaneous deletion of the amino-terminal Cysteine-rich and CDC25 domains and the RA2 carboxy-terminal domain resulted in complete loss of activation by free Gβγ, but not Rho (Fig.  4D). Further evidence that the CDC25 and RA2 domains interact with Gβγ is that simultaneous deletion of the amino-terminal Cysteine-rich, CDC25, and PH domains and the RA2 carboxyterminal domain led to loss of PLCε activation. Addition of a CAAX sequence to target this mutant to the membrane rescued Rho activation but not Gβγ activation (Figs. 4 and 5). Since deletion of the cysteine rich domain alone has no effect on Gβγ regulation, these data support a role for both the amino-terminal CDC25 domain and the carboxy-terminal RA2 domain in Gβγdependent activation of PLCε.
The inability of Gβγ to activate the PLCε-PH-RA1 mutant in COS-7 cells implies that this mutant lacks a binding site for Gβγ. The purified PLCε-PH-RA1 protein had enzymatic activity but was not activatable by Gβγ and only weakly bound to Gβγ in pull down assays for protein binding supporting the COS-7 cell data. Purified PLCε-PH-RA2, which retains one of the two Gβγ binding sites (RA2) implicated in the transfection experiments, bound Gβγ more strongly than PLCε-PH-RA1, and was activated by Gβγ, and implicating the RA2 domain as a binding site for Gβγ. These results demonstrate for the first time that PLCε is directly stimulated by G protein βγ subunits and a requirement for the RA2 domain. The activation of PLCε-PH-RA2 by Gβγ in vitro may be weak because it is missing the important CDC25 domain binding site, or because as has been raised previously, PLCε activation is notoriously difficult to activate in vitro. We have been unable to purify full length PLCε to discriminate between these possibilities.
In summary we have demonstrated that PLCε is directly regulated by Gβγ through binding to the RA2 domain with additional contributions from the CDC25 domain. Potential mechanisms for regulation of PLCε by Gβγ are shown in Figure 8. In the first scenario, free Gβγ interacts with and activates PLCε by binding to the amino-terminal CDC25 domain and to the carboxy-terminal RA2 domain through two independent sites, and interaction with both is required for full activation. An alternative model is that, in the threedimensional structure of PLCε, the RA2 and CDC25 domains are in close proximity and form a single Gβγ binding site.
Our previous observation using a neonatal rat ventricular myocyte (NRVM) model of cardiac hypertrophy showed that PLCεmediated PI4P hydrolysis at the Golgi is Gβγdependent (2). Furthermore, PLCε is also implicated in the development of cardiac hypertrophy and has been shown to localize to a subcellular compartment in NRVMs (1). Our studies indicate that Gβγ activation of PLCε is direct and likely occurs at the Golgi apparatus in cardiac myocytes. These studies extend our knowledge of the regulation of PLCε by Gβγ and could result in novel strategies for targeting PLCε regulation in hypertrophy, heart failure and other diseases linked to PLCε activation.

Construction of PLCε domain deletions and mutations.
Variations of the wild-type rat PLCε pCMV-scriptPLCε-FLAG were generated using QuickChange Lightning site-directed mutagenesis kit (Agilent Genomics) and manufacturers protocol followed. The following constructs were verified by sequencing; pCMV- The cDNAs PLCε-PH-RA2 (837-2281) and PH-RA1 (837-2098) variants used for in vitro assays were subcloned into pFastBac HTA to yield amino-terminally His-tagged protein.
Baculoviruses were generated using a recombinant baculovirus system (Invitrogen/Thermo Fisher Scientific, Inc.) Sf9 (Spodoptera frugiperda) insect cells (Invitrogen or Expression Systems) were used for virus production and protein expression. Cells were infected with baculovirus at an MOI=1, harvested at 48-72 hours post-infection, and flash-frozen in liquid nitrogen.
Transfection of COS-7 cells and Quantitation of Phospholipase C activity. All cell culture reagents were obtained from Invitrogen. COS-7 cells were obtained from ATCC. COS-7 cells were seeded in 12-well culture dishes at a density of 2x10 5  Purification of Free Biotinylated Gβγ. Purification of in vivo biotinylated Gβγ (bGβγ) was performed by co-expressing Gβγ with His 6 Gα i1 in High Five insect cells and nickel-agarose chromatography as described previously (25,26).
Purification of His 6 PLCε-PH-RA2 and PH-RA1. Baculovirus infected Sf9 cells expressing PLCε-PH-RA2 or PH-RA1 were lysed by dounce on ice in lysis buffer containing 20 mM HEPES, pH 8, 50 mM NaCl, 10 mM βmercaptoethanol (BME), 0.1 mM EDTA, 0.1 M EGTA, and two Roche EDTA-free protease inhibitor tablets at one-third strength. The lysate was centrifuged at 100,000 x g, and the supernatant was loaded onto a Ni-NTA column pre-equilibrated with buffer A (20 mM HEPES, pH 8, 100 mM NaCl, 10 mM BME, 0. Free Biotinylated Gβγ Pull-down Assay. Biotinylated Gβγ Pull-down Assay was performed as previously described (27) with minor modifications. Briefly, 50 ng purified PLCε-PH-RA2 or PH-RA1 was mixed with purified 15 nM bGβγ in the presence or absence of Gα i1 -GDP for 2 hours at 4°C in a rotator. 15 µL 50% magnetic neutravidin beads (GE) slurry was added to the mix and it was further incubated for 2 hours. Pellet was washed thrice and subjected to gel electrophoresis and western blotting.
Phospholipase C Activity Assay. The assay was performed as previously described (28) with minor modification. Briefly, lipids containing ~5000 cpm of [ 3 H-inositol] PIP 2 , 37.5 µM PIP 2 , and 150 µM phosphatidylethanolamine were mixed with 10 ng PLCε. Varying concentrations of Gβγ (with or without preincubation with G αi1 -GDP (1:2 ratio)) were added to the PLCε and the lipid mixture. The reaction was set at 30°C for 30 min and quenched using 5% BSA and 10% TCA. Released soluble [ 3 H] IP 3 was measured by liquid scintillation counting.
Statistical analysis. Data were analyzed using Prism 7 (GraphPad Prism Software). Mean values and standard errors were calculated, and the statistical significance established using either Student's t-test or oneway analysis of variance (ANOVA), as appropriate. All experiments were repeated at least three times.       Representative blot showing direct pull-down of PLCε-PH-RA2 with Gβ 1 γ 2 and inhibition of this interaction by Gα i1 -GDP. 50 ng purified PLCε-PH-RA2 was mixed with purified 15 nM biotinylated Gβ 1 γ 2 in the presence or absence of Gα i1 -GDP for 2 hours. 15 µl 50% magnetic neutravidin beads slurry was added and the mix was incubated for another 2 hours. Pellet was washed thrice and subjected to gel electrophoresis and Western blotting. Input is 1/20 of supernatant. C.
Quantitation of the PLCε band densities from three experiments as in B (PH-RA2 binding was normalized to 100%). The data shown are mean ± SE and analyzed by student's t test. ***P < 0.001 versus PH-RA2 binding in the absence of Gα i1 -GDP. D. 300 nM Gβ 1 γ 2 alone or with 600 nM Gα i1 -GDP was incubated with 10 ng PLCε PH-RA2. The effect on PLCε mutant activity is shown. Values are the mean ± SE of triplicate determinations and representative of three independent experiments and analyzed by one-way ANOVA with Dunnett posttest.