A short C-terminal peptide in Gγ regulates Gβγ signaling efficacy

Abstract

G protein beta-gamma (Gβγ) subunits anchor to the plasma membrane (PM) through the carboxy-terminal (CT) prenyl group in Gγ. This interaction is crucial for the PM localization and functioning of Gβγ, allowing GPCR-G protein signaling to proceed. The diverse Gγ family has 12 members, and we have recently shown that the signaling efficacies of major Gβγ effectors are Gγ-type dependent. This dependency is due to the distinct series of membrane-interacting abilities of Gγ. However, the molecular process allowing for Gβγ subunits to exhibit a discrete and diverse range of Gγ-type–dependent membrane affinities is unclear and cannot be explained using only the type of prenylation. The present work explores the unique designs of membrane-interacting CT residues in Gγ as a major source for this Gγ-type–dependent Gβγ signaling. Despite the type of prenylation, the results show signaling efficacy at the PM, and associated cell behaviors of Gβγ are governed by crucially located specific amino acids in the five to six residue preprenylation region of Gγ. The provided molecular picture of Gγ–membrane interactions may explain how cells gain Gγ-type–dependent G protein-GPCR signaling as well as how Gβγ elicits selective signaling at various subcellular compartments.

INTRODUCTION

G protein heterotrimers (Gαβγ) interact with the inner leaflet of the plasma membrane (PM) primarily through their covalent lipid modifications (fatty acylations). These modifications provide an additional layer of G protein activity regulation (Wedegaertner et al., 1995; Resh, 2013). G protein α (Gα) subunits are N-terminally (NT) modified with a 14-carbon (14-C) myristate and/or a 16-C palmitate group, while G protein γ (Gγ) subunits are prenylated at their carboxy-terminal (CT) (Wedegaertner et al., 1995; Pedone and Hepler, 2007). Prenylation of the 12 Gγ types, with either a 20-C isoprenoid geranylgeranyl or a 15-C farnesyl group, involves stable thioether bond formation at the CaaX motif cys in the Gγ CT (Wedegaertner et al., 1995). Despite being anchored to the PM by just one of the two possible prenyl attachments, Gβγ shows multiple Gγ-type–dependent PM affinities and signaling abilities (O’Neill et al., 2012; Senarath et al., 2018). For instance, the expression of Gγ3, which exhibited the highest PM affinity, allowed cells to achieve the highest Gβγ-effector activation at the PM (O’Neill et al., 2012; Senarath et al., 2018). On the contrary, cells expressing Gγ9, the Gγ with the lowest PM affinity, showed almost no Gβγ-effector activation. In these studies, we measured the PM affinity by taking the inverse half-time of Gβγ translocation from the PM to internal membranes (IMs) induced by GPCR activation.

Besides prenyl anchor–PM interactions, interactions between residues in the CT of Gγ with the PM have also been shown (Matsuda et al., 1994; Higgins and Casey, 1996). The PM possesses a phospholipid bilayer structure with embedded proteins and forms a fluidic mosaic (Singer and Nicolson, 1972; Engelman, 2005). The major phospholipid constituents of the PM are phosphatidylethanolamine, phosphatidylserine, phosphatidylinositol, phosphatidylcholine, and sphingomyelin (Hessel et al., 2003; Jastrzebska et al., 2011; Khan et al., 2013). These lipids contain two acyl lipid anchors linked to a polar phosphate head group via a glycerol molecule. Their negatively charged hydrophilic head groups face the cytosol, while the acyl groups form the core of the PM bilayer. Because the polar head groups of both phosphatidylserine and phosphatidylinositol are negatively charged, their predominance provides a net negative charge to the inner leaflet of the PM (Dowhan, 1997; Knight and Falke, 2009). Changes in the PM lipid composition have been shown to alter its association with G proteins (Escriba et al., 2003; Vogler et al., 2004).

We have demonstrated that Gγ3 and Gγ2 possess the highest PM affinity and their expression provides cells with the highest Gβγ-effector activities at the PM (Senarath et al., 2018). Both Gγ2 and Gγ3 possess hydrophobic phe and positively charged lys and arg residues at their preprenylation (pre-CaaX) region. However, the pre-CaaX regions of Gγ with low PM affinity and low effector activation ability (i.e., Gγ9, Gγ1) contain neutral gly and negatively charged glu residues (O’Neill et al., 2012; Senarath et al., 2018). Therefore, we hypothesize that the Gγ-dependent differential effector activation abilities of Gβγ at the PM are governed by the CT residues of Gγ subtypes as well as the chemical properties of the PM. Additionally, the three-dimensional structure of a protein is crucial for its biological functions and is determined by both its primary sequence and the surrounding chemical environment (Anfinsen, 1973; Das et al., 2015). PM-interacting proteins are either embedded in or interacting with the PM, and their structure-function attributes are heavily influenced by the properties of the PM. The fluid mosaic membrane structure permits several modes of movement to the proteins interacting with the bilayer. Rotational movements allow proteins to sample their immediate neighborhood and interact with effectors (Simons and Vaz, 2004; Engelman, 2005). Lateral diffusion enables proteins to migrate from one PM microdomain to another and reach distant effector molecules (Hepler, 2014; Czysz et al., 2015). Lateral mobilities of proteins in biological membranes are affected by the degree of crowding, membrane domains (i.e., lipid rafts), and interactions with cytoskeletal components (Frick et al., 2007; Owen et al., 2009; Ramadurai et al., 2010). The crystal structure of GRK2-Gβγ shows a rotation of Gβγ by ∼90°, while moving more than 100 Å from the origin, which 1) allows for Gβγ-GRK2 interaction to take place and 2) exposes both the receptor and GαqGTP to GRK2 (Tesmer et al., 2005; Nishimura et al., 2010; Samaradivakara et al., 2018). Therefore, lateral and rotational movements of Gβγ on the PM are likely to affect the activation of other Gβγ effectors as well. Because several CT residues of Gγ interact with the PM, we propose that differential sequence properties of CT domains in Gγ subtypes result in distinct lateral as well as rotational movements of Gβγ at the PM, regulating the efficacy of the associated Gβγ signaling.

Like Gγ, many other G proteins, including Ras and Ras-like proteins, comprise a pre-CaaX and a prenylated-cys, encoded by the CaaX motifs at their CT (Seabra, 1998; Maurer-Stroh et al., 2007). These proteins play critical roles in many cellular functions, including regulation of protein trafficking, cell proliferation, differentiation, and survival (Seabra, 1998; Wennerberg et al., 2005; Vogler et al., 2008). However, compared with the severalfold-longer and primarily polybasic pre-CaaX regions in Ras proteins, the five to six residue short pre-CaaX in Gγ is unique. Further, among Gγ subtypes, only Gγ2, γ3, and γ4 possess a distinctive hydrophobic character in the pre-CaaX. Therefore, we ask what functional roles pre-CaaX residues in Gγ play in tuning Gβγ signaling.

RESULTS

Selection criteria of residues in Gγ-CT to tune Gβγ–PM interactions

Besides prenylation, Ras superfamily proteins use a 15–20 residue polybasic pre-CaaX region for PM anchoring, establishing electrostatic interactions with the PM. To understand how Gγ accomplishes sufficient PM anchoring by using only a five to six residue pre-CaaX region, we examined the chemistry of its individual residues. All Gγ types (except Gγ13) possess a conserved phe residue, that is, phe65 in Gγ3 and phe60 in Gγ9, which functions as the last Gβ contact point (Figure 1A, blue box). The pre-CaaX region spans from this conserved phe to prenylated-cys. Owing to its proximity, this region is likely to interact with the PM (Figure 1B). Among the 12 Gγ subtypes, Gγ9 imparts the highest GPCR activation-induced translocation rate (from the PM to IMs) for Gβγ (O’Neill et al., 2012; Senarath et al., 2018). Therefore, Gγ9 expression allows cells to maintain the lowest Gβγ concentration, as well as the lowermost Gβγ-effector signaling at the PM (O’Neill et al., 2012; Senarath et al., 2018). Contrarily, Gγ3 exerts the lowest translocation rate for Gβγ, and thus cells could maintain the highest Gβγ-effector activity at the PM. In addition to the type of prenylation, we have previously shown evidence for pre-CaaX region-regulated control of the PM affinity and effector activity of Gβγ at the PM (Ajith Karunarathne et al., 2012; Senarath et al., 2018). Pre-CaaX sequences such as those in Gγ3 contain primarily positively charged and hydrophobic amino acids (Figure 1A, underlined). Thus, we postulate that the positively charged amino acids establish electrostatic interactions with the negatively charged phospholipid head groups. The hydrophobic residues interact with the hydrophobic core of the PM (Figure 1C, top) (Senarath et al., 2018).

FIGURE 1:

Molecular rationale for Gβγ PM-affinity control by preprenylation (pre-CaaX) residues of Gγ and the importance of phe-duo next to the prenylated-cys in Gγ3 to enhance Gβγ PM-affinity and PI3K activation. (A) Sequence alignment of the CT regions of the 12 Gγ subtypes. Conserved phe: green box, pre-CaaX region: blue box, and prenylated cys: brown, which undergoes prenylation and carboxymethylation. (B) Gβγ crystal structure (PDB ID: 2BCJ) modified to show its interaction with the PM using PyMOL software. The blue box: pre-CaaX, green residue: the conserved phe (the last GγG–β contact). The prenyl anchor–PM interaction is shown in black. (C) Hypothesized interactions between the CT of Gγ9 and Gγ3 with the PM. Polar-charged and hydrophobic groups in pre-CaaX residues interact with polar head groups and the hydrophobic core of the PM, respectively. Negatively charged glu (red) residues likely to modulate Gγ–PM interactions. This model provides molecular reasoning for PM-affinity differences among Gγ members including Gγ9 and Gγ3. Chemical structures were drawn using ChemDraw software. (D) Time-lapse images of HeLa cells expressing GFP-Gγ9, Gγ3, or their mutants with the Gi/o-coupled light-sensing GPCR, blue opsin before and after GPCR activation. Cells incubated with 10 µM 11-cis-retinal upon exposure to blue light show Gγ-type–dependent Gβγ translocation from the PM to IMs. Yellow arrows indicate the IMs. The plot shows Gγ translocation measured using F_IM (IM fluorescence). Scale bar: 5 µm. Average curves were plotted using n ≥ 10 cells from ≥3 independent experiments. Error bars: SEM. The bar graph and the whisker box plot show the half-time (t_1/2) and the extents of translocation, respectively. Note the Gγ3-like properties in the Gγ9-mutant and Gγ9-like behavior in the Gγ-3 mutant. Error bars: SD. (E) Time-lapse images of HeLa cells expressing GFP-Gγ9, Gγ3, or their mutants, with blue opsin and the PIP3 sensor (Akt-PH-mCh). Cells show Gγ-type–dependent PIP3 generation at the PM upon blue opsin activation. Yellow arrows indicate the Akt-PH-mCh accumulation at the PM. The corresponding plot shows the dynamics of PIP3 generation in cells with different Gγ types, measured using the mCh fluorescence at the PM (F_PM). Scale bar: 5 µm. Average curves plotted using n ≥ 10 cells from ≥3 independent experiments. Error bars: SEM. Bar graph shows distinct rates of PIP3 generation, and the whisker box plot shows the variation in the extent of PIP3 generation exhibited by WT Gγs and their mutants. Error bars: SD; *p < 0.05.

Therefore, to examine how Gγ types yield a discrete series of PM affinities for Gβγ, regulating its signaling at the PM using one of the two possible lipid anchors, we systematically mutated pre-CaaX residues (Figure 1A, green box) in both Gγ3 and Gγ9. Because these Gγ types provide the two extreme PM affinity characteristics for Gβγ, our strategy was to generate Gγ3-like mutants from farnesylated Gγ9 and Gγ9-like mutants from geranylgeranylated Gγ3, by altering only residues in their pre-CaaX regions. We have previously demonstrated that the transfected Gγ type becomes the most prominent and the dominant Gγ over the endogenous Gγs in a cell line (Senarath et al., 2018). This observation agrees with Gγ-specific distinct signaling changes observed in cells upon Gγ transfection. While the expression-level differences of Gγ among cells can influence the extent of their effects, we mitigate this by considering cells with only a defined green fluorescent protein (GFP)-Gγ–expression range for translocation and signaling measurements. For instance, we use a constant excitation intensity and consider only cells with an approximately ±30% emission range. This range is selected because these cells show a predominantly PM-bound Gγ with a minor presence at the IMs (Thul et al., 2017). As shown in sample images in Figure 1D, the selection of cells with a defined range of fluorescence intensities allows us to have cells with near-similar Gγ WT and mutant expressions for signaling quantification.

Two phe residues (phe-duo) adjacent to the prenylated-cys in Gγ3 are essential for the enhanced PM affinity and PI3K activation of Gβγ.

Gγ sequences display two distinct groups of residues in their pre-CaaX region: nonhydrophobic as in Gγ1, γ5, γ7, γ9, γ10, γ11, γ12, and γ13 and hydrophobic as in Gγ2, γ3, γ4, and γ8. Gγ3, γ2, and γ4 provide the highest Gβγ-governed PI3K (phosphatidylinositol 3-kinases) activation (Senarath et al., 2018). All of them possess a conserved phe-duo next to the prenylated and carboxymethylated cys. Therefore, we predicted that benzyl groups in phe residues interact with the PM, strengthening Gγ anchoring to the PM (Figure 1C, top, black circle). In contrast, Gγ9 possesses two gly residues next to the prenylated-cys and is expected to interact with the PM relatively loosely (Figure 1C, bottom). To explore these hypotheses, we systematically mutated pre-CaaX residues in Gγ3 and Gγ9. To gain the precise temporal control of GPCR-G protein activation using optogenetic signaling control, we expressed the light-sensitive Gi/o-coupled GPCR, blue opsin, and examined the behaviors of mutant Gγs compared with their WTs upon GPCR activation. Cells were incubated with 11-cis-retinal to make blue opsin light-activatable. GPCR activation-induced translocation of Gγ (and their mutants) was imaged using amino terminally (NT) GFP-tagged Gγ. Blue opsin was activated by exposing cells to 445 nm blue light at 1 Hz.

When the phe-duo in Gγ3 was replaced with two gly residues (Gγ3-phe-phe→gly-gly: NPFREKKGGCALL) compared with the WT Gγ3 (t_1/2 = 200 ± 20 s), the mutant-expressing cells exhibited more than twofold faster translocation (t_1/2 = 93 ± 5 s) (Figure 1D, images, plot, bar graph, and Table 1) (one-way analysis of variance [ANOVA]: F_{1, 44} = 14.98, p = 3.55 × 10^–4; Supplemental Table S1, A and B). In addition, this Gγ3-phe-phe→gly-gly mutant showed a significantly higher extent of Gβγ translocation (∼76%) over that of WT Gγ3 (Figure 1D, plot, box plot, and Table 1) (one-way ANOVA: F_{1, 44} = 14.98, p = 3.54 × 10^–4; Supplemental Table S2, A and B). This Gγ3 mutant is still expected to be geranylgeranylated because it carries the residues CALL as the CaaX motif, similar to WT Gγ3 (Wedegaertner et al., 1995). The ability of translocation as Gβγ upon GPCR activation, and its predominant PM-bound cellular distribution, indicate that the Gγ3-phe-phe→gly-gly mutant is functional in the Gβγ heterotrimer. Therefore, the observed significant changes in Gβγ translocation in these mutant-expressing cells indicate that, in addition to the type of prenylation, the phe-duo in the pre-CaaX is also crucial for Gβγ–PM interactions. Because WT Gγ3 cells exhibit a robust PIP3 production upon Gi/o-coupled GPCR activation (Senarath et al., 2018), we examined PIP3 production in Gγ3-phe-phe→gly-gly (NPFREKKGGCALL) mutant-expressing HeLa cells. Compared with WT Gγ3 cells, the extent of PIP3 generation in mutant Gγ3 cells was significantly reduced (by ∼38%) (one-way ANOVA: F_{1, 65} = 5.73, p = 0.02). A Tukey post hoc test showed that the mean extent of PIP3 generation in mutant Gγ3 cells (47.1 ± 7.3 arbitrary fluorescence units [AFU]) is significantly lower than that in WT Gγ3 cells (76.6 ± 7.4 AFU) (Figure 1E, plot and box plot; Supplemental Table S2, A and B). Additionally, mutant Gγ3 cells showed a significantly attenuated rate of PIP3 generation (3.9 × 10^–3 s^–1) compared with WT Gγ3 cells (6.7 × 10^–3 s^–1) (Figure 1E, images, plot, bar graph, and Table 2) (one-way ANOVA: F_{1, 43} = 7.67, p = 0.0083; Supplemental Table S3, A and B).

TABLE 1:

Translocation properties of Gγ mutants.

Gγ type	Sequence	t1/2 (s)	Extent of translocation (AFU)
Gγ3 WT	NPFREKKFFCALL	200 ± 20	21.3 ± 2.2
Gγ3-FF→GG	NPFREKKGGCALL	93 ± 5	37.1 ± 3.6
Gγ3-FF shifted	NPFFFREKKCALL	56 ± 5	40.76 ± 3.17
Gγ3-KK→GG in FF shifted	NPFFFREGGCALL	29 ± 3	51.7 ± 14.7
Gγ3-F65→G	NPGREKKFFCALL	115 ± 18	29.7 ± 3.7
Gγ3-RE→GGGG	NPFGGGGKKFFCALL	134 ± 6	22.3 ± 11.9
Gγ9 WT	NPFKEKGGCLIS	12 ± 1	48.4 ± 3.7
Gγ9-GG→FF	NPFKEKFFCLIS	171 ± 12	24.7 ± 2.6
Gγ9-KEK→GGG	NPFGGGGGCLIS	13 ± 2	30.1 ± 8.5
Gγ9 with Gγ3 pre-CaaX	NPFREKKFFCLIS	110 ± 5	13.6 ± 4.0

AFU: Arbitrary fluorescence units.

TABLE 2:

PIP3 generation properties of Gγ mutants.

Gγ type	Sequence	Rate of PIP3 generation (10–3 s–1)	Extent of PIP3 generation (AFU)
Gγ3 WT	NPFREKKFFCALL	6.7 ± 0.3	76.6 ± 7.4
Gγ3-FF→GG	NPFREKKGGCALL	4.0 ± 0.1	47.1 ± 7.3
Gγ3-FF shifted	NPFFFREKKCALL	4.4 ± 0.6	56.6 ± 8.5
Gγ3-KK→GG in FF shifted	NPFFFREGGCALL	4.2 ± 0.6	68.4 ± 6.4
Gγ3-F65→G	NPGREKKFFCALL	4.0 ± 0.4	54.9 ± 6.2
Gγ3-RE→GGGG	NPFGGGGKKFFCALL	5.1 ± 0.2	51.3 ± 7.1
Gγ9 WT	NPFKEKGGCLIS	2.1 ± 0.3	18.3 ± 2.5
Gγ9-GG→FF	NPFKEKFFCLIS	4.5 ± 0.5	46.2 ± 4.9
Gγ9-KEK→GGG	NPFGGGGGCLIS	2.7 ± 0.4	33.9 ± 5.5
Gγ9 with Gγ3 pre-CaaX	NPFREKKFFCLIS	4.0 ± 0.7	49.1 ± 3.8

To further validate the crucial role of this phe-duo in regulating the PM affinity, and the efficacy of Gβγ-mediated PI3K activation, we replaced the two adjacent gly residues next to prenylated-cys in Gγ9 with two phe residues (Gγ9-gly-gly→phe-phe: NPFKEKFFCLIS). Compared with WT Gγ9 cells (12 ± 1 s), these mutant Gγ9 cells (171 ± 12 s) exhibited an ∼14-fold higher t_1/2, indicating a slower Gβγ translocation (Figure 1D, images, plot, bar graph, and Table 1) (one-way ANOVA: F_{1, 60} = 431.24, p = 4.41 × 10^–29; Supplemental Table S1, A and B). Compared with WT Gγ9 cells, this mutant also exhibited a significant reduction (∼50%) in the translocation extent (Figure 1D, images, plot, box plot, and Table 1) (one-way ANOVA: F_{1, 59} = 13.62, p = 4.90 × 10^–4; Supplemental Table S2, A and B). Although Gγ9 expression suppresses Gβγ-induced PIP3 production in HeLa cells (rate = 2.1 × 10^–3 s^–1), Gγ9-gly-gly→phe-phe mutant cells exhibited a significantly enhanced rate of PIP3 production (4.5 × 10^–3 s^–1) (one-way ANOVA: F_{1, 66} = 28.58, p = 1.20 × 10^–6). A Tukey post hoc test showed that the mean extent of PIP3 generation in Gγ9 mutant-expressing cells (46.2 ± 4.9 AFU) is more than twofold higher than that in WT-Gγ9–expressing cells (18.2 ± 2.5 AFU) (Figure 1E, images, plot, box plot, and Table 1) (Supplemental Table S4, A and B). Overall, these data suggest that phe-duo next to the prenylated-cys in Gγ significantly enhances the PM affinity and PI3K signaling of Gβγ. Additionally, compared with Gγ3 WT, Gγ3-phe-phe→gly-gly mutant heterotrimers showed a clear IM presence (Figure 1D, bottom images, yellow arrows). Further, Gγ9-gly-gly→phe-phe mutant heterotrimers exhibited a more prominent PM-bound distribution than Gγ9 WT (Figure 1D, bottom images). Therefore, these data indicate that the relative changes to Gγ PM affinity due to the presence of the phe-duo or lack thereof also controls heterotrimer-PM interactions. We next examined the förster resonance energy transfer (FRET) between GFP-Gγ and mCh-Gβ1 to examine intact Gβγ dimer formation. Similar to their WT Gγs, both Gγ3 and Gγ9 mutants exhibited comparable FRET changes (donor/FRET ratios) upon photobleaching the acceptor (mCh) (Supplemental Figure S1). This indicates that the above pre-CaaX mutations in Gγ do not disrupt Gγ-Gβ interactions.

Because we explore the unique designs of membrane-interacting Gγ pre-CaaX residues as a major source for Gγ-type–dependent Gβγ signaling, we specifically selected Gγ3 and Gγ9 because they respectively represent a geranylgeranylated and a farnesylated Gγ, as well as demarcate the two extreme ends of the Gγ-PM–affinity range. The aim was to generate mutants with moderate-PM affinities. When we examined tissue-specific expression, Gγ types such as Gγ5, 10, and 12 show dominant expressions in most tissues (Tennakoon et al., 2021). Though these Gγs are geranylgeranylated, they occupy the moderate-PM–affinity range. The primary difference between the above Gγs and Gγ3 is that they do not possess the phe-duo in the pre-CaaX. When we mutated Gγ3-WT to Gγ3-phe-phe→gly-gly, the mutant exhibited translocation properties more similar to those of Gγ5, 10, and 12 than to those of Gγ3. The translocation t_1/2 in Gγ3-WT (200 s) was changed to 93 s in the Gγ3-phe-phe→gly-gly mutant, similar to 70–100 s t_1/2 in moderate-PM-affinity Gγs. On the contrary, the gly-gly→phe-phe mutation in Gγ9 changed the t_1/2 from 12 s to 171 s. Our newly added data show that this Gγ9-gly-gly→phe-phe mutant remains farnesylated (Supplemental Figure S2), while the Gγ3-phe-phe→gly-gly mutant is geranylgeranylated. The use of Gγ3, Gγ9, and their mutants, therefore, allowed us to demonstrate the significant role of pre-CaaX in modulating the PM affinity of Gβγ using a limited number of mutations.

Location of the phe-duo in the Gγ3 pre-CaaX motif is crucial for the PM affinity of Gβγ.

The above Gγ3 and Gγ9 mutants (Figure 1, D and E, and Table 1) established that the phe-duo significantly enhances the PM affinity and signaling activation ability of Gβγ at the PM. Among all prenylated G proteins, including heterotrimeric and Ras family members, only Gγ3, γ2, and γ4 possess this unique phe-duo next to the prenylated-cys (Supplemental Figure S3). Interestingly, cells expressing these Gγ types showed robust PI3K and PLCβ activations upon Gi pathway activation (Senarath et al., 2018). Therefore, to examine whether the location of phe-duo in the pre-CaaX region of Gγ tunes the PM-anchoring strength of Gβγ, a Gγ3 mutant was generated by shifting phe-duo to the beginning of the pre-CaaX region (Gγ3-phe-phe shifted: NPFFFREKKCALL). Compared with WT-Gγ3 HeLa cells (t_1/2 = 200 ± 20 s), mutant Gγ3 cells (t_1/2 = 56 ± 5 s) exhibited a nearly fourfold decrease in the translocation t_1/2 upon blue opsin activation (Figure 2A and Table 1). This indicates faster Gβγ translocation in mutant Gγ3-expressing cells. Additionally, a one-way ANOVA (F_{1, 38} = 27.01, p = 7.16 × 10^–6) and Tukey post hoc tests showed an approximately twofold increase in the extent of translocation for mutant Gγ3 compared with the WT-Gγ3 (40.8 ± 3.2 vs. 21.3 ± 2.2 AFU). Cells expressing this mutant also exhibited a significant reduction in the rate of PIP3 generation (4.4 × 10^–3 s^–1) compared with WT-Gγ3 cells (6.7 × 10^–3 s^–1) (Figure 2B and Table 2). A one-way ANOVA (F_{1, 53} = 8.54, p = 0.005) determined that there is a significant reduction in PIP3 generation in Gγ3-phe-phe–shifted mutant cells compared with that of WT-Gγ3 cells (56.6 ± 8.5 vs. 76.6 ± 7.4 AFU).

FIGURE 2:

Tuning of PM affinity and PI3K activation ability of Gβγ by the relative location of phe-duo in pre-CaaX and the last Gβ-interacting phe in Gγ. (A) The plot shows the variation in the kinetics of Gβγ translocation from the PM to IMs in HeLa cells expressing Gγ3 mutants compared with WT-Gγ3 and Gγ9–expressing cells, with blue opsin activation upon exposure to blue light after incubating cells with 10 µM 11-cis-retinal (F_IM ,IM fluorescence). Average curves plotted using n ≥ 10 cells from ≥ 3 independent experiments. Error bars: SEM. Bar graph shows the differences in translocation t_1/2 values while whisker box plot shows the variation in translocation extents in WT and mutant Gγ-expressing cells. Error bars: SD. (B) The plot shows the variance in the PIP3 generation kinetics in Gγ3 mutant–expressing cells compared with WT-Gγ3 and Gγ9–expressing cells upon blue opsin activation (F_PM, PM fluorescence). Average curves plotted using n ≥ 10 cells from ≥3 independent experiments. Error bars: SEM. Bar graph shows different rates of PIP3 generation while the whisker box plot compares the magnitudes of PIP3 generation. Error bars: SD; *p < 0.05.

To examine whether the observed activity in the Gγ3-phe-phe–shifted mutant is due to the polybasicity of the pre-CaaX region, we mutated the two lys residues located just before the CaaX motif in the Gγ3-phe-phe–shifted mutant (NPFFFREKKCALL) to two gly residues (Gγ3-lys-lys→gly-gly: NPFFFREGGCALL). Gγ3-NPFFFREGGCALL mutant cells exhibited only a minor increase in Gβγ translocation as indicated by its slightly reduced t_1/2 (29 ± 3 s) with a statistically similar translocation extent (at p = 0.05) compared with the Gγ3-NPFFFREKKCALL mutant (Figure 2A) (one-way ANOVA: F_{1, 28} = 55.41, p = 0.036; Supplemental Table S2, A and B). Furthermore, the PIP3 generation was equally attenuated in cells expressing these two mutants compared with the WT-Gγ3–expressing cells (at p = 0.05) upon blue opsin activation (Figure 2B). These observations suggest that the hydrophobic character in the pre-CaaX region and its location relative to prenylated-cys regulate Gβγ signaling at the PM.

The effect of the last Gβ-interacting phe in Gγ on Gβγ signaling at the PM is minor.

Because this conserved phe residue (phe65 in Gγ3) interacts with a hydrophobic pocket in Gβ (Akgoz et al., 2002), we examined how vital its contribution is to the PM affinity and effector activation ability of Gβγ at the PM. Relative to WT-Gγ3 cells, cells expressing a mutant Gγ3 in which phe65 was replaced with a gly (Gγ3-phe65→gly: NPGREKKFFCALL) showed a nearly twofold lower t_1/2 (200 ± 20 vs. 115 ± 18 s), indicating an enhanced rate of Gβγ translocation (Figure 2A and Table 1). The enhanced ability of Gβγ translocation in these Gγ3 mutant-expressing cells is further validated by the elevated extent of translocation compared with WT-Gγ3 cells (29.7 ± 3.7 vs. 21.3 ± 2.2 AFU) (Figure 2A and Table 2) (one-way ANOVA: F_{1, 40} = 4.28, p = 0.04511; Supplemental Table S2, A and B). The increased translocation rate of the mutant can be due to the loss of proper orientation of Gγ with Gβ in the Gβγ dimer. An in vitro assay showed an ∼40% reduction in PLCβ2 activity upon a mutation of an equivalent phe in Gγ5 (phe59→A) (Akgoz et al., 2002). Nevertheless, compared with WT-Gγ3, Gγ3-phe65→gly mutant HeLa cells exhibited an ∼40% reduction in the rate of PIP3 generation (4.0 × 10^–3 s^–1) (one-way ANOVA: F_{1, 54} = 15.66, p = 2.23 × 10^–4; Supplemental Table S3, A and B). In comparison to WT-Gγ3, this mutant Gγ3-expressing cell also exhibited a significantly reduced extent of PIP3 generation (76.6 ± 7.4 vs. 55.0 ± 6.2 AFU) (Figure 2B and Table 2) (one-way ANOVA: F_{1, 77} = 4.39, p = 0.039; Supplemental Table S4, A and B).

The contribution of positively charged residues in the pre-CaaX of Gγ to the PM affinity and signaling of Gβγ at the PM is minor.

Both Gγ3 and Gγ9 have homologous regions at the beginning of pre-CaaX consisting of three residues, arg-glu-lys in Gγ3 and lys-glu-lys in Gγ9 (Figure 1A). We hypothesized that the positive charges of arg and lys side chains collectively allow Gβγ to transiently interact with the negatively charged polar head groups of PM phospholipids. In contrast, negatively charged glu may likely create a repulsive force (Figure 1C). Considering these opposite charge characteristics, “+ - +”, we anticipated a “pseudo-spring’’–like behavior allowing for transient interactions–repulsions for Gβγ with and from the PM (Figure 1C). We mutated these charged residues in the pre-CaaX regions of Gγ3 and Gγ9 to examine this hypothesis. We first generated a Gγ9 mutant of which lys-glu-lys residues in the pre-CaaX were replaced with three gly residues (Gγ9-lys-glu-lys→gly-gly-gly: NPFGGGGGCLIS). Cells expressing this Gγ9 mutant showed near-similar translocation (t_1/2 = 13 ± 2 s vs. 12 ± 1 s) (Figure 3A and Table 1) and PIP3 generation characteristics (rate = 2.6 × 10^–3 s^–1 vs. 2.1 × 10^–3 s^–1) (Figure 3B and Table 2) to WT-Gγ9 cells (Table 1). Additionally, the extents of Gβγ translocation in cells expressing this Gγ9 mutant (as well as WT-Gγ9) and the Gγ3-phe-phe→gly-gly mutant were not significantly different (one-way ANOVA; F_{1, 36} = 4.67, p = 0.04) (Figure 3A; Supplemental Table S2, A and B). This further suggests the crucial contribution of the phe-duo for Gγ-PM interactions.

FIGURE 3:

The effect of positively charged residues in the pre-CaaX on PM affinity and PI3K activation ability of Gβγ. (A) The plot shows Gi-coupled blue opsin activation-induced Gβγ translocation kinetics in HeLa cells expressing pre-CaaX region–mutated Gγ3 or Gγ9, to remove positively charged residues (F_IM, IM fluorescence). Average curves plotted using n ≥ 10 cells from ≥3 independent experiments. Error bars: SEM. Differences in translocation t_1/2 and extent of translocation are shown, respectively, in the bar graph and whisker box plot. Error bars: SD. (B) The plot shows the influence of the above Gγ3 and Gγ9 mutants on PIP3 generation compared with WT-Gγ3 and Gγ9-expressing cells, upon blue opsin activation (F_PM, PM fluorescence). Average curves plotted using n ≥ 10 cells from ≥3 independent experiments. Error bars: SEM. Bar graph compares rates of PIP3 generation while the whisker box plot shows the magnitude differences in PIP3 generation. Error bars: SD; *p < 0.05.

Because the mutant Gγ9-lys-glu-lys→gly-gly-gly (NPFGGG­GGCLIS) did not show a significant change in Gβγ activity at the PM compared with WT-Gγ9 cells, to further examine whether the “pseudo-spring” behavior exists, we generated a Gγ3 mutant by replacing arg-glu with two gly and also introducing two extra gly residues (Gγ3-arg-glu→gly-gly-gly-gly: NPFGGGGKKFFCALL). Interestingly, Gγ3-arg-glu→gly-gly-gly-gly mutant cells exhibited a minor reduction in PM affinity (Gβγ translocation t_1/2 = 134 ± 6 s) (Figure 3A and Table 1). However, the change in the rate of PIP3 generation in the mutant cells compared with WT-Gγ3 cells was not statistically significant (Figure 3B and Table 1). Overall, these data suggest that, unlike polybasic C-termini of Ras family proteins, contributions from the charged residues in the pre-CaaX of Gγ to the PM affinity and the PI3K activation ability of Gβγ are relatively minor.

Cells expressing a Gγ9 mutant carrying the entire pre-CaaX region of Gγ3 (Gγ9 with Gγ3 pre-CaaX: NPFREKKFFCLIS) showed an approximately ninefold slower translocation (t_1/2 = 110 ± 5 s) compared with WT-Gγ9–expressing cells (t_1/2 = 12 ± 1 s) (Figure 3A and Table 1). A one-way ANOVA (F_{1, 59} = 30.78, p = 7.43 × 10^–7) showed that the extent of Gβγ translocation in these mutant Gγ9 cells is significantly lower than that of WT-Gγ9 cells. Compared with WT-Gγ9 cells, mutant Gγ9 cells also exhibited an approximately twofold higher rate of PIP3 generation (2.1 × 10^–3 s^–1 vs. 3.9 × 10^–3 s^–1) (Figure 3B and Table 2). These data further validate our hypothesis that the pre-CaaX amino acids of Gγ tune the PM affinity and effector activation ability of Gβγ. Moreover, our findings emphasize the critical role of phe-duo adjacent to prenylated-cys in Gγ2, γ3, and Gγ4 in enhancing their PM affinity compared with the contribution from other residues in the pre-CaaX region.

Hydrophobic pre-CaaX residues in Gγ tune Gβγ-mediated partial adaptation of PIP2 hydrolysis

Gq-coupled GPCRs such as M1 and M3 muscarinic and gastrin releasing peptide (GRP) receptors efficiently induce the hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP2) into diacylglycerol (DAG) and inositol trisphosphate (IP3) via the activation of phospholipase-C β (PLCβ). However, within ∼30–45 s after the initial Gq-GPCR activation, a universal partial adaptation of PIP2 hydrolysis begins, allowing for a partial synthesis of PIP2 at the PM (Figure 4A, up to 600 s). This adaptation should create low-intensity steady-state signaling of IP3 and DAG, where the equilibrium of PIP2 hydrolysis ⇋ synthesis is reached. Our recent work shows that this partial adaptation of PIP2 hydrolysis is governed by Gβγ (Weiland, 1978). Our data suggest that immediately upon Gq pathway activation, a highly potent Gαq_GTP-PLCβ-Gβγ sandwich complex is formed, which is responsible for the intense and near-complete PIP2 hydrolysis (Figure 4A, yellow box). Thus far, we have shown that Gβγ transiently interacts with the PM and control effectors in the vicinity, including PI3Ks and G protein–coupled inwardly rectifying potassium (GIRK) channels (Senarath et al., 2018). Therefore, PM affinity-governed gradual dissociation of Gβγ from the PLCβ sandwich complex can be expected. This can generate the less-potent lipase Gαq_GTP-PLCβ, shifting the PIP2 hydrolysis ⇋ synthesis equilibrium to the right and reaching the steady-state. We further show that the kinetics of this PIP2 hydrolysis adaptation process and the steady-state intensity are Gγ-type dependent (Weiland, 1978). Because hydrophobic pre-CaaX residues of Gγ control the PM affinity of Gβγ (Figures 1 and 2), we examined whether the regulation of Gq-induced PIP2 hydrolysis adaptation is also pre-CaaX dependent.

FIGURE 4:

Gγ tunes Gβγ-mediated partial adaptation of PIP2 hydrolysis. (A) Plots of GRPR (Gq-GPCR)-induced PIP2 hydrolysis upon its activation with 1 µM bombesin, its partial adaptation, followed by Gβγ-mediated PIP2 rehydrolysis upon blue opsin (Gi-GPCR) activation (with blue light) and the second adaptation (PIP2 resynthesis). Red curve: fast (indicated by the blue tangent line) and complete adaptation of PIP2 upon termination of blue light after 30 s. Black curve: slower (indicated by the gray tangent line) and minor adaptation of PIP2 hydrolysis under sustained blue light. Both the red and black arrows show the steady-state PIP2 hydrolysis. Bar chart and whisker box plot show significantly lower PIP2 resynthesis (adaptation) with a slower rate under sustained blue opsin activation over near-complete and faster adaptation after blue light termination at 30 s. Error bar: SD. (B) Optogenetic signaling termination shows distinct regulation of Gαq-PLCβ–mediated PIP2 hydrolysis by Gγ9 and Gγ3. Time-lapse images of HeLa cells expressing GRPR, mCh-PH (PIP2 sensor), blue opsin, and GFP-Gγ9 (top two panels) or Gγ3 (bottom two panels) showing PIP2 hydrolysis and attenuation upon blue opsin activation in the Gq-active background (with GRPR activation). Images and corresponding plots show significantly different rates of PIP2 hydrolysis attenuation after termination of blue opsin activation at 30 s (red curves), which removes Gβγ rapidly, compared with the hydrolysis partial adaptation observed under continuous blue light (black curves). Yellow arrows indicate the PIP2 sensor (mCh-PH) initial localization on the PM, its movement to the cytosol (PIP2 hydrolysis), and partial relocalization to the PM (adaptation) during and after blue opsin activation. Average curves were plotted using n ≥ 10 cells from ≥3 independent experiments. Scale bar: 5 µm. Error bars: SEM. (C) Whisker box plot shows the distinct regulation of PIP2 resynthesis rates due to its hydrolysis attenuation by Gγ3 and Gγ9 after termination of blue opsin activation at 30 s. Here, Gγ9-expressing cells showed a twofold higher attenuation rate compared with Gγ3-cells (p < 0.01). Under sustained blue opsin activation condition also Gγ9-expressing cells showed a twofold higher rate of PIP2 hydrolysis adaptation compared with Gγ3-expressing cells (p < 0.05). Error bars: SD; *p < 0.05, **p < 0.01.

Gβγ modulates the partial adaptation of PIP2 hydrolysis.

Gq-coupled GRPR, blue opsin, and the PIP2 sensor mCh-PH were expressed in HeLa cells. We employed blue opsin to gain precise temporal control of Gi/o heterotrimer activation and Gβγ generation (Figure 1D) (Senarath et al., 2018). Before imaging, cells were incubated with 10 µM 11-cis-retinal for 3 min in the dark, allowing for light-activatable blue opsin generation. Upon activation of GRPR with 1 µM bombesin, the characteristic transient PIP2 hydrolysis and its partial adaptation were observed (Figure 4A, up to 600 s). After the steady-state of PIP2 hydrolysis ⇋ synthesis is reached, we activated blue opsin by exposing cells to 445 nm blue light (1 Hz). In this Gαq-active background, Gi/o activation-induced Gβγ generation prompted rehydrolysis of PIP2 (Figure 4A, after 600 s, black trace under blue box). Here, we optogenetically controlled Gβγ availability under two blue light exposure conditions, that is, 1) blue light termination at the peak of PIP2 rehydrolysis (at ∼30 s) (Figure 4A, red curve), and 2) continuous blue light (Figure 4A, black curve) (Table 3). In condition 1, blue opsin becomes inactive after ∼50 ms of blue light termination (at 30 s) and Gαi/o_GTP→Gαi/o_GDP conversion occurs in ∼100–400 ms (Tsang et al., 2006; Sprang, 2016). Because Gi/o_GDP has greater than 100-fold higher affinity for Gβγ than that of Gαi/o_GTP (Tuteja, 2009; Mahoney and Sunahara, 2016), an abrupt loss of Gβγ from the Gαq_GTP-PLCβ-Gβγ sandwich complex is expected. As anticipated, the termination of 30 s blue light illumination resulted in a faster PIP2 resynthesis (Figure 4A, red curve, blue tangent) compared with that in continuous blue light exposure (condition 2) (Figure 4A, black curve, gray tangent). The faster rate of PIP2 resynthesis under blue light termination resulted in the process reaching the steady-state within ∼150 s (Figure 4A, red arrow). However, under continuous blue light exposure, it took more than ∼6 min to reach the steady-state (Figure 4A, black arrow). The rate of PIP2 resynthesis after blue light termination was approximately twofold higher (7.0 × 10^–3 s^–1) compared with that under sustained blue light (1.5 × 10^–2 s^–1) (one-way ANOVA: F_{1, 24} = 14.10, p = 7.26 × 10^–4) (Figure 4A, bar graph Supplemental Table S5). Furthermore, blue light termination resulted in a near-complete PIP2 resynthesis (96.2 ± 6.1%). In contrast, the continuous blue light exposure condition showed only a partial adaptation, with a relatively lower PIP2 resynthesis (49.4 ± 12.7%) (Figure 4A, box plot). We hypothesize that the distinct rates and extents of PIP2 resynthesis observed in the above two conditions result from the differences in Gβγ availability. Owing to Gαi/o_GTP hydrolysis-associated abrupt removal of Gβγ upon blue light termination, Gαq_GTP-PLCβ-Gβγ is expected to experience a faster loss of Gβγ compared with the gradual loss of Gβγ under sustained blue light.

TABLE 3:

PIP2 resynthesis properties of Gγ mutants.

Gγ type	t1/2 (s)	Rate of PIP2 resynthesis (10–2 s–1)
		Sustained blue light	Terminated blue light
Gγ3 WT	200.42 ± 20.20	1.2 ± 0.9	2.5 ± 1.0
Gγ3-FF→GG	93.28 ± 4.90	1.8 ± 0.8	3.4 ± 0.2
Gγ9 WT	12.15 ± 1.19	2.1 ± 0.2	4.2 ± 0.1
Gγ9-GG→FF	170.99 ± 11.74	1.6 ± 0.1	2.5 ± 0.6
Gγ9 with Gγ3 pre-CaaX	109.78 ± 4.67	1.3 ± 0.7	3.0 ± 0.5

To further show Gβγ modulation of this Gi/o-governed PIP2 rehydrolysis ⇋ resynthesis response, we conducted similar experiments in cells additionally expressing either WT-Gγ9 or WT-Gγ3 (Figure 4A). After the partial adaptation of Gq-mediated PIP2 hydrolysis, we exposed cells to either 30 s or continuous blue light. Regardless of the Gγ type (whether Gγ9, Gγ3, or endogenous) compared with the continuous blue light condition, faster and near-complete PIP2 resynthesis responses were observed after blue light termination at 30 s (Figure 4, B and C). One-way ANOVA (F_{1, 38} = 21.91, p = 3.57 × 10^–5) and Tukey post hoc tests showed a significantly higher mean rate (approximately twofold) of PIP2 resynthesis upon blue light termination compared with the sustained blue light condition. Under the above light conditions, HeLa cells expressing pre-CaaX mutants of Gγ9 and Gγ3 also exhibited similar and significant differences in the rates of PIP2 resynthesis (Supplemental Figure S4). Collectively, these data indicate that Gβγ availability is crucial for the efficacy of the Gq-governed lipase activity of PLCβ.

Hydrophobic pre-CaaX residues influence Gβγ-governed partial adaptation of PIP2 hydrolysis.

These experiments were performed in cells that reached steady-state PIP2 hydrolysis after Gq-GPCR activation (like in Figure 4A, ∼600 s). To examine how the type of Gγ modulates PIP2 adaptation in cells after its hydrolysis by Gi/o-GPCR activation, we activated blue opsin only for 30 s in HeLa cells also expressing mCh-PH and either Gγ3 or Gγ9. Upon termination of blue light at 30 s, WT-Gγ9 cells (Figure 5, blue trace) showed a twofold higher rate of hydrolysis adaptation (4.2 × 10^–2 s^–1) compared with that of WT-Gγ3 cells (2.5 × 10^–2 s^–1) (Figure 5, black trace). A Tukey post hoc test associated with a one-way ANOVA (F_{1, 48} = 27.84, p = 3.12 × 10^–6) revealed that these mean rate differences are significant. Upon termination of blue opsin activation, a faster generation of Gαi/o_GDP is likely to sequester Gβγ. The relative mobility of Gβγ, governed by its PM affinity, could dictate how fast Gβγ is removed from effectors. Therefore, we propose that the removal of Gβγ from the Gαq_GTP-PLCβ-Gβγ complex in the Gi/o-active background is determined by the PM affinity of Gβγ. Therefore, the higher rate of PIP2 hydrolysis adaptation in Gγ9-expressing cells can be understood by the low PM affinity of Gβγ9 (Ajith Karunarathne et al., 2012) and the relatively transient PLCβ-Gβγ9 interactions. On the contrary, we anticipate Gβγ3 to have relatively robust interactions with PLCβ, allowing for a weaker PIP2 hydrolysis adaptation.

FIGURE 5:

Pre-CaaX residues influence the rate of PIP2 hydrolysis termination. Cells expressing GRPR, mCh-PH, blue opsin, and either GFP-Gγ9, Gγ3, or their pre-CaaX mutants were first activated with 1 µM bombesin, allowing PIP2 hydrolysis and partial adaption to occur before the experiment. (A) The plot compares the dynamics of blue opsin–induced PIP2 hydrolysis (BO activation) up to 30 s and resynthesis after blue light termination in the Gq-active background. The whisker box plot shows that replacement (in Gγ3) or introduction (in Gγ9) of phe-duo next to the prenylated cys significantly alters the rates of PIP2 resynthesis compared with the corresponding Gγ WT. Average curves were plotted using n ≥ 10 cells from ≥3 independent experiments. Error bars: SD. (B) Tissue-specific segregated mRNA expression of fast and slow translocating Gγ types. Human retina shows exclusive expression of fast translocating Gγ1, Gγ9, and Gγ11 (red). Note that these Gγs are excluded from the brain tissues. Brain tissues show predominant expression of Gγ2, Gγ3, and Gγ4 (phe-duo containing) (green) compared with other Gγs. Interestingly, fast translocating Gγs do not show a detectable expression in brain tissues. Gγs with relative expression above 10% were considered. Figures S5 and S6 shows Gγ diversity in cells within tissues. *p < 0.05.

We next examined whether pre-CaaX residues influence the Gq-mediated lipase activity of the Gαq_GTP-PLCβ-Gβγ complex. As indicated by one-way ANOVA (F_{1, 49} = 37.41, p = 2.34 × 10^–7) and Tukey post hoc tests, the rate of PIP2 hydrolysis adaptation was significantly lower in Gγ9-gly-gly→phe-phe mutant cells compared with that of WT-Gγ9 cells (Figure 5A, red trace and whisker box plot) (2.5 × 10^–2 s^–1 vs. 4.2 × 10^–2 s^–1). Similarly, compared with WT-Gγ9 cells, cells with the Gγ9 mutant carrying the pre-CaaX of Gγ3 showed a significant reduction in PIP2 hydrolysis adaptation rate (4.2 × 10^–2 s^–1 vs. 3.0 × 10^–2 s^–1; Figure 5A, orange trace and whisker box plot) (one-way ANOVA: F_{1, 32} = 10.66, p = 0.00261). These rate differences (Table 3) indicate that the introduction of a more hydrophobic character to the pre-CaaX of Gγ enhances the ability of Gβγ to stimulate Gq-PLCβ signaling. Distinct Gβγ translocation profiles (Figure 1D) and PIP3 generation data (Figure 1E) show the reduction in the PM affinity of Gβγ in Gγ3-phe-phe→gly-gly mutant cells compared with that of WT-Gγ3 cells. As anticipated, a significantly higher rate of PIP2 hydrolysis adaptation was observed in the mutant Gγ3-phe-phe→gly-gly cells compared with WT-Gγ3 cells (3.4 × 10^–2 s^–1 vs. 2.5 × 10^–2 s^–1) (Figure 5A) (one-way ANOVA: F_{1, 63} = 5.92, p = 0.02). Gβγ translocation is a direct indicator of the PM affinity of Gβγ.(Senarath et al., 2016) Therefore, it likely captures even minor changes in the PM affinity of Gβγ. Though we observed significant changes in the PLCβ-governed PIP2 dynamics upon expression of Gγ mutants (relative to Gγ-WT) compared with Gβγ translocation, the sensitivity of the PIP2 hydrolysis-adaptation process to the changes in Gβγ PM affinity is reduced. This is not surprising because PIP2 hydrolysis and its adaptation upon PLCβ activation is one of the downstream processes regulated by Gβγ. Further, PIP2 and PLCβ are also controlled by regulators, including phosphatases, kinases, and calcium (Weiland, 1978). Therefore, it is not surprising to observe larger PLCβ activity changes only upon drastic changes to PM affinity and thus the availability of Gβγ at the PM.

The data of Gγ expression at the transcript level in different human tissues extracted from the FANTOM5 repository in the human protein atlas database show the tissue-specific expression of different Gγs in the human body (Uhlén et al., 2015, 2017; Thul et al., 2017). Here, we selected brain-associated 6 tissue types as a sample to illustrate the Gγ diversity. These transcriptomic data show that many distinct neurological centers of the brain, including the cerebral cortex, olfactory regions, and basal ganglia, express Gγ types such as Gγ2, γ3, and γ4 that contain phe-duo next to prenylated-cys (Figure 5B) (Uhlén et al., 2015). Retina is an outlier, and unlike the brain regions, it does not express high PM affinity Gγ2, γ3, or γ4. It does not express even Gγ types with moderate-PM affinities (Figure 5B). Because the expressions were determined using complex tissue homogenates, the cell-type–specific Gγ diversity is not visible. However, distinct cell types have exhibited significant variability in Gγ expression in tissues at the protein level (Supplemental Figures S5 and S6). For instance, while bipolar cells show a prominent Gγ13 expression, Gγ1 and Gγ9 are the primary Gγs in rod and cone photoreceptor cells in the retina. In addition to its elevated expression in the ovary and prostate, Gγ3 protein has exhibited prominent expression in the cerebral cortex, corroborating with the transcriptomic data in Figure 5B (Sjöstedt et al., 2018). Several other investigations also have indicated the unique expression profiles of Gγ types in different cell and tissue types (Fagerberg et al., 2014; Gremel et al., 2015; Syrovatkina et al., 2016). Collectively considering our data and unique Gγ expression profiles in functionally specialized tissues and cells, including the retina and the brain, we propose that pre-CaaX residues in Gγ likely play a broader regulatory role in GPCR-G protein signaling.

DISCUSSION

We believe that the following remaining questions about Gβγ are crucial to have a deeper understanding of GPCR-G protein signaling: 1) Why do not all Gγ types promote Gβγ to activate effectors to a similar extent? 2) Why does Gγ show cell-tissue–specific expressions? 3) Could the relative orientation of Gβγ with the PM be important for Gβγ-effector interactions? 4) What makes Gγ2, γ3, and γ4 more efficient in regulating PM-bound Gβγ effectors? 5) Is sampling the PM for effectors by Gβγ governed by the CT of Gγ? Identification of the unique sequence properties of Gγ2-4 compared with the rest of the Gγ-pool, especially their phe-duo next to the prenylated-cys as a key regulator of Gβγ signaling at the PM, now allows us to answer the above questions and more. Compared with Gγ2–4, the rest of the geranylgeranylated Gγ types such as Gγ5, γ7, γ10, and γ12 that are lacking hydrophobic residues exhibited a significantly reduced Gβγ signaling at the PM (Senarath et al., 2018). Therefore, we hypothesized that this phe-duo is an essential requirement in Gγ to display high Gβγ-effector activity at the PM. The perturbation of these residues in the current study revealed intricate molecular details of pre-CaaX residues and their relative location in the Gγ-CT in Gβγ signaling regulation.

The observed significant attenuation of Gβγ signaling after shifting phe-duo away from the prenylated-cys indicates that being located away from the prenyl anchor and thus the reduced proximity of benzyl side chains in the shifted phe-duo to the PM likely reduces the PM affinity of Gβγ. When the phe-duo is next to the prenylated-cys, the prenyl anchor is likely to pull Gγ toward the PM, facilitating benzyl groups of phe to form strong hydrophobic interactions with the PM. This reinforcement appears to be an essential criterion in the pre-CaaX of Gγ to enhance the PM affinity and the effector interaction ability of Gβγ at the PM. This crucial involvement of the phe-duo was further confirmed by the reduced Gβγ signaling observed in Gγ mutant cells with gly or Gγ types lacking phe residues next to prenylated-cys (as in Gγ9). Data also indicate a moderate enforcement of Gβγ–PM interactions by other hydrophobic residues in the pre-CaaX, likely through hydrophobic interactions with the PM. For instance, compared with Gγ9, leu in Gγ1 pre-CaaX appears to provide a slightly higher PM affinity. Additionally, our data also indicate a lesser influence from the charged residues in the pre-CaaX, including lys, arg, and glu, on Gβγ signaling at the PM. This is a significant difference from prenylated Ras family proteins that primarily use poly-lys and thereby electrostatic interactions to enforce PM interactions.

Our data indicate that the influence of the Gγ prenylation type on the PM affinity and the signaling of Gβγ at the PM is not as strong as initially predicted. For instance, cells expressing predictably a farnesylated Gγ9 mutant with Gγ3 pre-CaaX exhibited a Gβγ-induced PIP3 generation resembling that of Gγ3 cells. We show that the type of prenylation in pre-CaaX mutated Gγs is intact and in agreement with its CaaX sequence (Supplemental Figure S2). Both the Gγ-WT and its pre-CaaX region–altered Gγ mutants showed similar sensitivity to their corresponding prenyltransferase inhibitors but not to the other kind. For instance, Gγ9-WT (pre-CaaX→NPFKEKGG) is farnesylated, and this farnesylation was inhibited by the farnesyltransferase inhibitor, Tipifranib, but not by the geranylgeranyltransferase inhibitor, GGTI286. Prenylation of its pre-CaaX mutant, Gγ9-gly-gly→phe-phe (pre-CaaX→NPFKEKFF), was also similarly inhibited by Tipifarnib, indicating that the mutagenesis of the pre-CaaX region does not alter the type of prenylation (Supplemental Figure S2). Our findings thus collectively suggest that the sequence properties of the pre-CaaX region of Gγ allow cells to have a broad range of Gβγ activities at the PM from “on” to “off” paradigms. These data also suggest that Gγ types are evolved to have relatively short PM-interacting pre-CaaX regions compared with other prenylated proteins (i.e., Ras family proteins). By installing or avoiding hydrophobic residues at specific locations in the pre-CaaX, Gγ types provide a range of PM affinities and effector activation abilities for Gβγ at the PM. The capacity of pre-CaaX residues to significantly modify Gq pathway-mediated PLCβ signaling further establishes the broader physiological relevance of the pre-CaaX region of Gγ. It also rationalizes the evolutionary significance of the existence of 12 Gγ types with distinct PM affinities. Is Gβγ translocation due to deprenylation? Unlike palmitoylated Gα that shows a rapid turnover at the PM through depalmitoylation (Huang et al., 1999; Wedegaertner, 2012), Gγ prenylation is an irreversible process (Palsuledesai and Distefano, 2015; Wang et al., 2017). Prenylation requires –aaX residues of the CaaX motif to proceed. Upon prenylation, –aaX is cleaved by RCE1 (Ras-converting CaaX endopeptidase 1), followed by ICMT (isoprenyl carboxyl methyltransferase)-induced carboxymethylation of prenylated-cys (Wright and Philips, 2006). Even if Gγ were to be deprenylated, reprenylation could not occur due to the permanent modification of the prenyltransferase-recognizing region of Gγ. Therefore Gβγ must be translocating with the prenylated Gγ.

The dominant expression of Gγ1, γ11, and γ9 as well as the exclusive absence of Gγ2, γ3, and γ4 in the retina photoreceptor cells (Figure 5B) can be understood by envisioning the possible consequences of their signaling outcomes. The observed slow adaptation of PIP2 hydrolysis in Gγ3 cells compared with Gγ9 cells indicates a slow termination of Gβγ3 effector signaling after termination of GPCR activity (Figure 5A). During vision transduction, upon the termination of opsin activation, Gα_transducin and Gβγ should be reunited instantaneously, so that opsin can be ready for the next heterotrimer activation upon the photon reception. Therefore, we argue that the exclusion of Gγ2–4 and the inclusion of Gγ1, γ9, and γ11 in photoreceptor cells are essential for the acuity and brisk nature of phototransduction. In what way, then, does Gγ1, γ9, and γ11 facilitate vision signaling? Upon activation of opsin, Gβγ9 has shown a detectable translocation in <500 ms (Senarath et al., 2016). Therefore, we propose that this rapid translocation allows for a complete and fast separation of Gα_transducin from Gβγ, facilitating Gα_transducin-directed vision signaling. Further, the removal of Gβγ from the PM could eliminate unnecessary PM Gβγ signaling. The fast translocation of Gβγ may also be necessary to regenerate Gα_transducin heterotrimers upon Gα_GDP formation, allowing for signaling continuation. Though they show elevated expression in specific brain regions, the functional role/s of this exclusive expression of Gγ2, γ3, and γ4 are entirely unknown (Morishita et al., 1997). This complexity is exacerbated by the fact that the majority of organs express Gγ types with moderate-PM affinities. Though they are geranylgeranylated, these Gγ types lack the phe-duo next to the prenylated-cys. This deficiency appears to provide the Gγs moderate-PM affinities between Gγ2–4 and Gγ1, γ9, and γ11. We, therefore, propose that, by having Gγ types with moderate-PM affinities, cells in most organs, unlike the eye (and possibly the brain), utilize Gβγ to maintain moderate effector signaling at the PM and some signaling in cell-interior regions such as the Golgi and endoplasmic reticulum (ER). Further investigations are needed to confirm these assumptions.

Reports suggest that Gβγ controls signaling in internal compartments of cells such as the Golgi (Saini et al., 2010; Hewavitharana and Wedegaertner, 2015), mitochondria (Hewavitharana and Wedegaertner, 2012; Ahmed and Angers, 2013), and the nucleus (Kino et al., 2005; Hewavitharana and Wedegaertner, 2012). Because current findings indicate a higher concentration of Gβγ in IMs upon GPCR activation in cells primarily expressing low PM-affinity Gγ types, we expect these cells to have dominant Gα signaling at the PM and Gβγ signaling in IMs. Our findings may also indicate that the pre-CaaX region of Gγ and thereby the PM affinity of Gβγ influence 1) the efficacy of heterotrimer-GPCR interaction and 2) the rate of Gβγ activity cessation upon termination of GPCR activity, as well as 3) the rates of GPCR phosphorylation and desensitization. However, these hypotheses need investigation. Our findings, additionally enabled by optogenetic GPCR-G protein signaling control, help in understanding the complex regulation of GPCR-G protein signaling at various cellular compartments and how cells achieve desired signaling selectivity based on Gβγ–membrane interactions. This knowledge may open avenues for pharmacological intervention.

MATERIALS AND METHODS

Request a protocol through Bio-protocol.

Reagents

The reagents bombesin (Tocris Bioscience, Bristol, UK), 11-cis retinal (National Eye Institute, Bethesda, MD), and Tipifarnib and GGTI286 (Cayman Chemical, Ann Arbor, MI) were dissolved in appropriate solvents according to manufacturer’s instructions and diluted in 1% Hanks' Balanced Salt solution (HBSS) supplemented with NaHCO₃ or regular cell culture medium before being added to cells.

DNA constructs and cell lines

For the engineering of DNA constructs used, mCh-PH has been described previously (Kankanamge et al., 2019). GRPR was a kind gift from the lab of Zhou-Feng Chen at Washington University, St. Louis, MO. Blue opsin-mTurquoise, GFP-Gγ9, GFP-Gγ3, mCh-Gβ1, and Akt-PH-mCh were kindly provided by N. Gautam’s lab, Washington University, St. Louis, MO. Gγ3 and Gγ9 mutants were generated by PCR amplifying the parent constructs in pcDNA3.1 (GFP-Gγ3 and GFP-Gγ9) with overhangs containing expected nucleotide mutations and DpnI (NEB) digestion (to remove the parent construct) followed by Gibson assembly (NEB) (Ratnayake et al., 2017). The HeLa cell line was originally purchased from the American Type Culture Collection (ATCC) and authenticated using a commercial kit to amplify nine unique STR loci.

Cell culture and transfections

HeLa cells used in Gβγ translocation, PIP3 generation, and PIP2 hydrolysis and adaptation experiments were cultured in MEM (from CellGro) supplemented with 10% heat-inactivated dialyzed fetal bovine serum (DFBS; from Atlanta Biologicals) and 1% penicillin−streptomycin (PS) in 60 mm tissue culture dishes and maintained in a 37°C, 5% CO₂ incubator. When the cells reach ∼80% confluency, they are lifted from the dish using versene-EDTA (CellGro) and resuspended in their growth medium at a cell density of 1 × 10⁶ /ml. For imaging experiments (translocation, PIP3 generation, and PIP2 hydrolysis), cells were seeded on 35 mm cell culture–grade glass-bottomed dishes (Cellvis) at a density of 8 × 10⁴ cells. The day following cell seeding, cells were transfected with appropriate DNA combinations using Lipofectamine 2000 transfection reagent (Invitrogen) according to the manufacturer’s protocol and stored in a 37°C, 5% CO₂ incubator. Cells were replenished with the growth medium after 5 h and were imaged after 16 h of transfection.

Live cell imaging to monitor Gβγ translocation, PIP3 generation, and PIP2 hydrolysis and subsequent synthesis

A spinning-disk XD confocal TIRF (total internal reflection) imaging system with a Nikon Ti-R/B inverted microscope, a Yokogawa CSU-X1 spinning disk unit (5000 rpm), an Andor FRAP-PA (fluorescence recovery after photobleaching and photoactivation) module, a laser combiner with 40−100 mW four solid-state lasers (with 445, 488, 515, and 594 nm wavelengths), and an iXon ULTRA 897BV back-illuminated deep-cooled EMCCD camera were used to capture time-lapse image series of live cells. In Gβγ translocation and PIP3 generation experiments, imaging was performed using a 60×, 1.4 NA (numerical aperture) oil objective. To examine the Gβγ translocation, GFP fluorescent tags on Gγ subunits (WTs and mutants) were imaged in every 1 s interval using 488 nm excitation−515 nm emission for 10 min. In PIP3 generation and PIP2 hydrolysis experiments, mCherry-tagged PIP3 and PIP2 sensors, Akt-PH and PH, were imaged using 594 nm excitation−630 nm emission red laser.

Quantification of Gβγ translocation, PIP3 generation, and PIP2 hydrolysis

Digital image analyses were performed using Andor iQ 3.1 software. In translocation and PIP2 experiments, the background-subtracted fluorescence intensity increase in IMs of individual cells was captured. Pre- and poststimulation images were generated by binning (×4) at the equilibrium. Initial baseline intensity values were subtracted from intensity values at regions of interest (ROIs) from multiple cells and single-cell fluorescence averages were plotted versus time to monitor the dynamics of translocation. The number of cells (usually one ROI per cell) and the number of independent experiments are provided in the figure legends. In PIP3 generation experiments, the background-subtracted PIP3 sensor (Akt-PH-mCh) fluorescence on the PM was captured and processed similar to Gγ translocation and PIP2 hydrolysis.

Statistical data analysis

Statistical analysis and data plotting were performed using OriginPro software (OriginLab Corporation). Results of all quantitative assays (Gβγ translocation, PIP3 generation, PIP2 hydrolysis) are expressed as mean ± SEM from n numbers of cells (indicated in the figure legends) from multiple independent experiments. After obtaining all of the baseline-subtracted data, PIP3 generation and PIP2 resynthesis rates were calculated using the Nonlinear Curve Fitting tool (NLFit) in OriginPro. In the NLFit tool, each plot was fitted to DoseResp (Dose-Response) function under the Pharmacology category by selecting the relevant range of data to be fitted. The mean values of hill slopes (P) obtained for each nonlinear curve fitting are presented as mean rates of PIP3 generation or PIP2 resynthesis. Similarly, employing the NLFit tool, Gβγ translocation plots were fitted to the MichaelisMenten function under the Pharmacology category to determine the t_1/2 of Gβγ translocation. The mean values of K_m obtained from nonlinear curve fitting for all cells are given as mean Gβγ translocation t_1/2. One-way ANOVA statistical tests were performed using OriginPro software to determine the statistical significance of mean signaling responses in different experiments. After inserting raw signaling response data from each cell for various experiments, the Tukey’s mean comparison test was performed at the p < 0.05 significance level for the one-way ANOVA statistical test.

Abbreviations used:

AFU

arbitrary fluorescence units

ANOVA

Analysis of Variance

ATCC

American Type Culture Collection

CT

carboxy terminus

DAG

diacylglycerol

DFBS

dialyzed fetal bovine serum

DNA

deoxyribonucleic acid

EDTA

ethylenediaminetetraacetic

EMCCD

electron multiplying charge-coupled device

ER

endoplasmic reticulum

FRAP-PA

fluorescence recovery after photobleaching and photoactivation

FRET

förster resonance energy transfer

GDP

guanosine diphosphate

GFP

green fluorescence protein

GIRK

G protein-coupled inwardly-rectifying potassium channels

GPCR

G protein-coupled receptor

GRK2

G protein-coupled receptor kinase 2

GRPR

gastrin-releasing peptide receptor

GTP

guanosine triphosphate

HBSS

Hanks’ balanced salt solution

ICMT

Isoprenylcysteine Carboxyl Methyltransferase

IM

internal membrane

IP3

inositol trisphosphate

mCh

mCherry

MEM

minimum essential medium

NEI

national eye institute

NLFit

nonlinear curve fitting

NT

amino terminus

PCR

polymerase chain reaction

PH

pleckstrin homology

PI3K

phosphoinositide 3-kinase

PIP2

phosphatidylinositol (4,5)-bisphosphate

PIP3

phosphatidylinositol (3,4,5)-trisphosphate

PLCβ

phospholipase C β

PM

plasma membrane

PS

Penicillin−Streptomycin

RCE1

Ras Converting CAAX Endopeptidase 1

ROI

region of interest

SD

standard deviation

SEM

standard error of mean

STR

short tandem repeat

TIRF

total internal reflection fluorescence

WT

wild type.