Gγ identity dictates efficacy of Gβγ signaling and macrophage migration

G protein βγ subunit (Gβγ) is a major signal transducer and controls processes ranging from cell migration to gene transcription. Despite having significant subtype heterogeneity and exhibiting diverse cell- and tissue-specific expression levels, Gβγ is often considered a unified signaling entity with a defined functionality. However, the molecular and mechanistic basis of Gβγ's signaling specificity is unknown. Here, we demonstrate that Gγ subunits, bearing the sole plasma membrane (PM)–anchoring motif, control the PM affinity of Gβγ and thereby differentially modulate Gβγ effector signaling in a Gγ-specific manner. Both Gβγ signaling activity and the migration rate of macrophages are strongly dependent on the PM affinity of Gγ. We also found that the type of C-terminal prenylation and five to six pre-CaaX motif residues at the PM-interacting region of Gγ control the PM affinity of Gβγ. We further show that the overall PM affinity of the Gβγ pool of a cell type is a strong predictor of its Gβγ signaling–activation efficacy. A kinetic model encompassing multiple Gγ types and parameterized for empirical Gβγ behaviors not only recapitulated experimentally observed signaling of Gβγ, but also suggested a Gγ-dependent, active–inactive conformational switch for the PM-bound Gβγ, regulating effector signaling. Overall, our results unveil crucial aspects of signaling and cell migration regulation by Gγ type–specific PM affinities of Gβγ.

Among G␤␥-controlled activities, chemokine GPCR activation-governed cell migration plays a key role in many physiological functions, including embryonic development and immune responses. Altered cell motilities are implicated in pathological processes such as immune deficiencies, lack of wound healing, tissue repair, and cancer metastasis (14 -17). We have recently shown that G␤␥ is a key regulator of inhibitory G protein (G i )-coupled GPCR activation-induced macrophage migration (18). In addition to PI3K-PIP3 signaling at the leading edge, we demonstrated that G␤␥-mediated activation of PLC␤ pathway is essential for macrophage migration.
Mammalian cells express 12 G␥ and 5 G␤ subunits, and form stable G␤␥ dimers with the exception of G␤5, giving rise to 48 possible combinations of G␤␥ (19,20). It has been shown that most G␥ subtypes comparably interact with the two most predominant G␤ types in cells, G␤1 and G␤2, with the exception of G␥11 for G␤2 (21). Similar affinities of G␣ i 1 for G␤␥ types have also been demonstrated (20). Studies have suggested the possibility of specific G␤␥ subtypes possessing higher affinities toward certain GPCRs or effectors. Using in vitro reconstituted heterotrimers and activated GPCRs, heterotrimers with certain G␥ subtypes exhibited higher affinities for specific GPCRs (20). In addition, specific structural motifs in GPCRs, preferring interactions with certain G␤␥ isoforms, also have been reported for adenosine family receptors (22,23). Assigned cellular functions to the availability of specific G␤ or G␥ subtypes have also been shown (24,25). For instance, modulation of Golgi vesiculation and cellular secretions by G␥11 and differential ion channel control by G␥9 and G␥3 subunits have been demonstrated (24,25). G␥3 and G␥5 were shown to control predisposition of mice to seizures (26).
Although these investigations have primarily assigned subunit identity of either G␤ or G␥ subtype to specific signaling activities and cellular functions, molecular and mechanistic basis of such a signaling specificity has not been provided. G␤ subunits have a conserved structure with a Ͼ80% identity among their isoforms. However, G␥ isoforms show a significant sequence diversity ranging from ϳ20 -80% (19,27). Therefore, if the G␤␥ diversity is a crucial modulator of its signaling and associated cell behaviors, the G␥ identity in these dimers is likely to be a primary regulator of G␤␥ signaling. Although G␤␥ is classically considered plasma membrane (PM) bound, recent work has shown that, upon GPCR activation, G␤␥ translocates from the PM to internal membranes (IMs) until an equilibrium is reached (25). Interestingly, translocation half-time (Tt1 ⁄2 ) and the extent ͉T͉ are governed by the type of accompanying G␥ subunit (25,29). These results further suggest that the PM affinity of a G␤␥ is G␥ subtype-dependent. Because, G␥ provides the only PM anchor for G␤␥, the accompanying G␥ subtype-dependent translocation ability of G␤␥ suggests that the G␥ subunit controls the PM affinity of the accompanying G␤␥. Considering that the majority of G␤␥-effector interactions take place at the PM, G␥-governed PM affinity is likely to be crucial for G␤␥ signaling. Thus, this study is focused on examining how cells employ a selected group of G␥ subunits to tune signal propagation from activated GPCRs to G␤␥ effectors, controlling signaling and macrophage migration.

G␥ subtype identity-specific control of PI3K␥ activation by G␤␥
PIP3 is a major regulator of lamellipodia formation in the leading edge of migratory cells (30). Because G␤␥-PI3K interaction leads to the PIP3 generation (Fig. 1A), we examined whether PIP3 production is controlled in a G␥ subtypedependent manner. To interact with G␤␥ and catalyze PIP3 production, PI3K subunit p110 should translocate to the PM upon activation (31,32). The signaling circuit that drives PIP3 production is composed of GPCRs, G␤␥, and PI3K␥ subunits (Fig. 1A). PIP3 generation was measured using the translocation of a fluorescently tagged PIP3 sensor (Akt-PH-mCherry) from cytosol to the PM. We have previously shown that localized blue opsin activation results in a robust PIP3 production at the leading edge and directional motility of RAW264.7 cells (33). Because both blue opsin and chemokine receptor 4 (CXCR4) activate G proteins with nearly similar efficiencies, blue opsin was employed to induce macrophage migration (Fig. S1, A-C). Activation of G i -coupled GPCR, blue opsin, induced a robust PIP3 production in RAW cells (Fig. 1B, top). This suggests that the type of G␤␥ in RAW cells supports PI3K activation. However, the same receptor activation in HeLa cells failed to produce an observable PIP3 response (Fig. 1B, lower). Interestingly, both cell types produced similar G protein activation upon blue opsin activation, when measured using translocation of YFP-G␥9 and YFP-G␥3, respectively, indicating equivalent G protein activations (Fig. S1, D-F). Therefore, either absence of proper type of G␤␥ or low expression of PI3K␥ or both are the source of this lack of PIP3 production. Similar to HeLa cells, PC12 cells also failed to show PIP3 production upon GPCR activation. However, unlike HeLa cells, PC12 cells exhibited augmented PIP3 generation upon expression of PI3K␥, directing the study toward the type of G␤␥ in HeLa cells (Fig. 1C). Real-time PCR data from RAW and HeLa cells revealed that they express substantially different G␥ subunit profiles (Fig. 1D). Compared with RAW cells, HeLa cells show an ϳ6-fold lower expression of G␥3, and the expression of G␥4 is also significantly lower in HeLa cells. Expression of G␥3 in HeLa cells resulted in an elevated basal PIP3, even without GPCR activation (Fig. 1F). Blue opsin activation resulted in a robust PIP3 production in these cells. Although G␥4 expression did not promote an elevation of basal PIP3, opsin activation exhibited a minor increase in PIP3 in HeLa cells (Fig. 1F, yellow arrows). Nevertheless, G␥9-expressing HeLa cells failed to induce PIP3 production either at the basal state or upon opsin activation (Fig. 1F). Real-time PCR data indicated that overexpression of a G␥ subunit results in the reduction of the fractional contribution of endogenous G␥ subunits to the pool, making the introduced G␥ subunit dominant, creating nearly a mono-G␥ system (Fig. 1E).

Optogenetic determination of PM affinities of 12 G␥ subunits using Tt1 ⁄2 of G␥
Measurement of G protein activation upon ligand addition is prone to experimental artifacts because of inconsistencies associated with the agonist injection and variations in its diffusion through the culture media. This hinders calculation of precise Tt1 ⁄2 as well as extent of translocation G␤␥. Thus, to measure the dependence of G␥ type on translocation properties of G␤␥, optically controlled activation of GPCR-G protein signaling was used as follows. HeLa cells expressing blue opsin together with each of YFP-tagged 1-13 G␥ types were examined for translocation upon activation of blue opsin (Fig. 2, A-C). Cells were supplemented with 11-cis retinal for 5 min before opsin activation. G␤␥ translocation was measured using YFP fluorescence dynamics in IMs (F IM versus time curves), and the data were fitted to the logistic function because GPCR activation results in an approximately sigmoidal increase in G␤␥ in IMs, which reaches saturation over time.
Using the fitted curves, Tt1 ⁄ 2 and the extent of translocation ͉T͉ ϭ (T max Ϫ T base ) of individual G␥ subtypes were calculated (Table S1). The plot of ͉T͉ versus Tt1 ⁄ 2 (Fig. 2D) exhibited a strong exponential decay correlation (adjusted R 2 ϭ 0.94). This suggests that the G␥ types with moderate to slow translocation rates are translocation-deficient (small ͉T͉). These data also indicate that ͉T͉ and Tt1 ⁄2 of G␥ are linked and likely to be controlled by the ability of G␤␥ to interact with the PM G protein ␥ subtype dictates G␤␥ signaling (Fig. 2E). Considering the link between the Tt1 ⁄2 of G␤␥ dissociation from the PM and free energy of the associated transition state (⌬G) the Tt1 ⁄2 of G␤␥ was considered as an index of the PM residence time and the PM affinity of G␤␥ because the Tt1 ⁄2 includes the effects of G␤␥ shuttling between IMs and the PM.

G protein ␥ subtype dictates G␤␥ signaling
Tt1 ⁄ 2 values of G␥9 translocation were identical in HeLa, RAW, and HEK cells (Fig. S1, A-F). This demonstrates that translocation properties of G␥ types are conserved among cell types, suggesting conserved PM affinities of G␥ types. Although G␥ types only possess two types of lipid anchors (geranylgeranyl and farnesyl) at their carboxyl terminal cysteine (in the CaaX motif), they exhibit a discrete series of Tt1 ⁄2 values (Table S1). Therefore, distinct regions of PM-interacting pre-CaaX motifs of G␥ subunits appear to provide further control over their PM affinities, resulting in a discrete series of Tt1 ⁄2 values. Here, G␥ types are grouped, based on their PM affinities. D, plot of Tt1 ⁄2 versus ͉T͉ shows an exponential decay relationship. E, schematics of GPCR activation-induced G protein heterotrimer activation and dissociation. LoAf-G␤␥ translocates from the PM to IMs faster compared with HiAf-G␤␥, whereas HiAf-G␤␥ interacts with effectors to initiate signaling pathways leading to cellular responses efficiently compared with LoAf-G␤␥.

G␥-dependent control of chemokine pathway-mediated RAW cell migration
Because the subtype identity of G␥ controls PIP3 formation, we examined whether cell migration is also controlled in a G␥ type-dependent manner. Real-time PCR data showed that ϳ30% of G␥ in WT RAW 264.7 cells is G␥3, a high PM affinity (HiAf) G␥ type (Tt1 ⁄ 2 ϭ ϳ270 s) (Fig. 1D). In response to localized optical activation of blue opsin, RAW cells migrate efficiently with a leading edge velocity (LE) of 0.82 m/min and trailing edge velocity (TE) of 0.51 m/min (Fig. 3, A and B). Knockdown of endogenous G␥3 using the most effective shRNA identified by screening five constructs (Fig. S2) resulted in a complete cessation of cell migration (Fig. 3, A and B). Nonspecific shRNA did not affect WT RAW cell migration. Expression of HiAf-G␥2 (Tt1 ⁄ 2 ϭ ϳ181 s) in G␥3 knockdown cells resulted in rescuing the lost migration ability with LE: 0.61 m/min and TE: 0.28 m/min (Fig. 3, A and B). Expression of low PM affinity (LoAf) G␥ subtypes showed a marked reduction of migration, i.e. G␥9 (Tt1 ⁄ 2 ϭ ϳ5 s) 3 LE: 0.20 m/min, TE: 0.03 m/min, and G␥1 (Tt1 ⁄ 2 ϭ ϳ13 s) 3 LE: 0.24 m/min, TE: 0.04 m/min (Fig. 3, A and B). Further, expression of moderate PM affinity (MoAf) G␥4 (Tt1 ⁄ 2 ϭ ϳ116 s) also reduced the migration ability of WT RAW cells substantially (LE: 0.38 m/min, TE: 0.18 m/min). Although, LoAf G␥ expressing cells occasionally showed lamellipodia formation at the leading edge, trailing edge retraction was not observed. These data collectively suggest that the higher the PM affinity of G␥, the greater the migration ability of RAW cell. To examine the universal nature of HiAf G␥ subunit requirement in chemokine pathways, we examined whether the introduction of HiAf G␥3 helps nonmigratory HeLa cells to migrate. Localized opsin activation in HeLa cells expressing G␥3 showed a distinct trailing edge retraction with lamellipodia formation at the leading edge, resulting in a net movement of the cell. No such responses were observed in WT or G␥9 expressing HeLa cells for similar signaling activation (Fig. S3).

Control of RAW cell migration by CaaX and pre-CaaX residues in the carboxyl termini (CT) of G␥
Because the CT of G␥ provides sites for G␤␥ dimers to anchor and interact with the PM, which is required for G␤␥ signaling, properties of their CT on RAW cell migration was examined. The CT sequences of G␥ exhibit a significant diversity ( Fig. 2C) (19,27). Sequence alignment and structural data show that, after the conserved Phe-59 residue in all G␥ subtypes (except G␥13), CT region loops out from a conserved hydrophobic pocket on G␤ (Figs. 2C and 4A), delineating its last contact point with G␤ ( Fig. 4A) (34). The pre-CaaX region of G␥ (between Phe-59 and the CaaX) therefore appears to interact with the PM and partially modulates the PM affinity of G␤␥. The lack of electron density for the CT of G␥ in G␤␥ crystal structures indicates that this region is unstructured and suggests dynamic interactions with the PM. We employed a group of G␥ mutants composing the body of HiAf-G␥ with a substituted CaaX and/or pre-CaaX motifs from LoAf-G␥ and vice versa (Fig. 4B). Translocation properties of these mutants resembled properties of WT G␥ in which the introduced CT motifs were originated (Table S1). For instance, G␥9 with pre-CaaX plus CaaX of G␥3 (G␥9-␥3CT) exhibited similar translocation properties to G␥3. On the contrary, G␥3 with pre-CaaX plus CaaX regions of G␥9 (G␥3-␥9CT) exhibited similar translocation properties to G␥9 (Fig. 4C). The incorporation of an extra cysteine to G␥3 CaaX moiety eliminated the translocation ability of G␤␥3 (Fig. S4A). This is likely because of the second geranylgeranyl lipid anchor attachment. Deletion of cysteine from the CaaX motif resulted in complete disruption of PM localization of G␤␥9, limiting it only to the cytosol (Fig. S4B), indicating the lipid anchor requirement for PM interaction of G␤␥. The cells expressing above-mentioned mutants were also examined for their ability to modulate RAW cell migration, for instance, G␥9-␥3CT mutant-induced cell migration. On the contrary, cells expressing G␥3-␥9CT mutant failed to migrate, recapitulating migration behavior of RAW cells expressing G␥9 G protein ␥ subtype dictates G␤␥ signaling (Fig. 4, D and E). Collectively, these data suggest that CaaX and pre-CaaX residues of the CT of G␥ control the PM affinity and the signaling efficiency of G␤␥.

Modulation of RAW cell migration potential by G␥ subtype-dependent activation of PI3K␥
Because PIP3 is a key regulator of chemokine-induced cell migration, we examined if PIP3 production is G␥-type depen-dent. RAW cells expressing the PIP3 sensor, Akt-PH-mCherry, showed a significant PIP3 accumulation at the leading edge upon localized optical activation of blue opsin ( Fig. 5B and Movie S1). Inhibition of G␤␥ with gallein and PI3K␥ with wortmannin ceased PIP3 production and migration of RAW cells (Fig. 5, A and B). A gallein-like compound, fluorescein did not show any effect for either PIP3 production or migration. Cells expressing G␥3 showed a leading edge PIP3 production and a

G protein ␥ subtype dictates G␤␥ signaling
directional migration similar to the responses exhibited by WT RAW cells (Fig. 5C and Movie S2). On the contrary, plots show that G␥9 expressing RAW cells exhibit mild or no PIP3 production. These cells further failed to migrate as well. (Fig. 5C and Movie S3). G␥3 knockdown cells showed neither PIP3 production at the leading edge nor cell migration upon opsin activation (Fig. 5, C and D and Movie S4). Additionally, RAW cells expressing G␥3 mutants composed of either pre-CaaX or CaaX motifs or both from G␥9 failed to produce PIP3 at the leading edge and subsequently migrate (Fig. 5, F and G). Interestingly, cells expressing G␥9-␥3CT mutant (both CaaX and pre-CaaX from G␥3) showed both PIP3 production and cell migration. However, G␥9 mutants with either pre-CaaX alone or CaaX alone from G␥3 failed to show PIP3 production or cell migration. This can be understood by examining PM affinities (Tt1 ⁄ 2 values) of G␥ types and their mutants listed in Table S1. The order of Tt1 ⁄ 2 is G␥3 Ͼ ␥9-␥3CT Ͼ ␥3-␥9CaaX Ͼ ␥9-␥3CaaX Ͼ ␥3-␥9CT Ͼ ␥9. PIP3 dynamics in RAW cells expressing G␥ types exhibited a reasonable fit to the logistic function with an adjusted R 2 Ͼ 0.93 (Fig. 5E). This comparative PIP3 response Figure 5. G␥ type-dependent activation of PI3K␥ during macrophage migration. A, GPCR-mediated PIP3 generation pathway and its selected inhibitory points. B, RAW 264.7 cells expressing Akt-PH-mCherry and blue opsin, supplemented with 50 M 11-cis-retinal. On localized blue opsin activation with 445 nm (white box), WT cells showed the PIP3 production at the activated leading edge. Cells treated with PI3K inhibitor wortmannin and G␤␥ inhibitor gallein inhibited both PIP3 production and cell migration, confirming that PIP3 is required for directional cell migration. C, RAW 264.7 cells expressing Akt-PH-mCherry, blue opsin-mTurquoise, a G␥ subunit (G␥3, G␥9, G␥4, G␥12) and supplemented with 50 M 11-cis-retinal. On localized blue opsin activation with 445 nm (white box), G␥3 expressing cells showed PIP3 generation at the leading edge. However, G␥4, G␥12, and G␥9 expressing cells showed minor/no PIP3 accumulation. G␥3 knockdown cells also showed no PIP3 generation. D, plots show the PIP3 generation with G␥3, G␥9, G␥4, G␥12 overexpression and G␥3 knockdown compared with the WT. E, smoothed curves fitted to logistic function show G␥ type-dependent differential PIP3 responses. F, RAW 264.7 cells expressing Akt-PH-mCherry, blue opsin-mTurquoise, each of the mutant G␥ types and supplemented with 50 M 11-cis-retinal. Upon migration induction, cells expressing the mutant G␥9-␥3CT showed both PIP3 as well as migration. Failure to exhibit migration in G␥9-␥3CaaX cells shows the significance of the pre-CaaX motif of G␥ in G␤␥ signaling. G␥3-␥9CT mutant cells exhibited neither PIP3 production nor migration. G, the plot shows PIP3 generation in RAW cells expressing CT mutants of G␥ (error bars, S.E.; n ϭ 15; scale bar, 5 m).

G protein ␥ subtype dictates G␤␥ signaling
analysis illustrates that cells expressing only HiAf-G␥ subtypes, including G␥3, G␥2, and G␥9-␥3CT mutant elicited a significant PIP3 generation. Such a robust PIP3 production appears to be required for cell migration. Fitted curves also showed that both WT and HiAf-G␥3 expressing RAW cells possess comparable mean rates of PIP3 production, 0.0022 s Ϫ1 and 0.0030 s Ϫ1 , respectively. However, the mean rate of PIP3 generation in MoAf-G␥4 expressing cells (0.0009 s Ϫ1 ) was closer to G␥9 (0.0007 s Ϫ1 ) and G␥12 (0.0012 s Ϫ1 ) than to G␥3. Collectively, these data indicate that only G␥ types with the highest PM affinity support significant PIP3 production and RAW cell migration.

G␥ subtype-dependent control of G␤␥-mediated PLC␤ activation
Recently, we demonstrated that G i -coupled GPCR activation-induced RAW cell migration requires an increase in cytosolic calcium (Ca 2ϩ ) which is governed by G␤␥-mediated activation of PLC␤ to induce trailing edge retraction (18). Thus, we examined if PLC␤ activity in RAW cells is also controlled in a G␥ subtype-dependent manner, in the same way it controlled PI3K␥ activation. Ca 2ϩ mobilization upon endogenous G i -coupled complement component 5a receptor (c5aR) activation in RAW cells with 10 M c5a (35) was measured using a fluorescence probe for Ca 2ϩ , Fluo-4 AM. WT and HiAf-G␥3 expressing cells showed Ca 2ϩ responses to a higher degree (Fig. 6, A, B,  and D), whereas LoAf-G␥9 expressing RAW cells showed minor or no Ca 2ϩ response upon c5aR activation (Fig. 6, C and  D). Interestingly, MoAf-G␥4 and G␥12 expressing cells only exhibited a relatively weak response (Fig. 6, E and F). Replacement of the entire CT or CaaX motif alone in G␥3 with those of G␥9, respectively, resulted in loss of Ca 2ϩ mobilization ability of WT G␥3 (Fig. 6, G and H). Although expression of G␥9-␥3CaaX mutant failed to elicit Ca 2ϩ mobilization, mutant G␥9-␥3CT showed a Ca 2ϩ response, which is equivalent to responses exhibited by WT as well as G␥3 expressing RAW cells (Fig. 6, G and H). In addition, we confirmed that the resultant Ca 2ϩ responses are similar for G␥ with different fluorescent tags (Fig. 6, A-D).

Tt1 ⁄2 of G␥ is a strong predictor of G␤␥ effector activation ability
The purpose was to examine the hypothesis that the extent of G␤␥ effector responses elicited upon GPCR activation in a cell can be predicted using the averaged Tt1 ⁄ 2 of the endogenous G␥ pool. The experimental process to test this concept is given in Fig. 7A. PIP3 production in HeLa cells expressing each of the 12 G␥ subtypes upon blue opsin activation was measured and plotted against the Tt1 ⁄ 2 of G␥ types (Fig. 7A, blue box). The extent of PIP3 production in each G␥ expressing cell was considered as the effector activation, ͉EF͉ exp G␥ , and was measured using baseline-normalized increase of Akt-PH-mCherry fluorescence at the PM because of PIP3 production (Fig. 7B). Tt1 ⁄2 values of each G␥ type translocation were also similarly calculated by measuring YFP-G␥ translocation (Fig. 2, A-C). The fitted straight line on the resultant ͉EF͉ exp G␥ versus Tt1 ⁄2 (HeLa effector plot, blue box) exhibited an R 2 value of 0.94 (Fig. 7C). This indicates a linear relationship between the G␤␥ effector responses and the PM affinities of G␤␥. Next, translocation properties of endogenous G␤␥ pool in HeLa and RAW cells were measured using blue opsin activation-induced YFP-G␤1 translocation (Fig. 7D). Because G␤ translocates with endogenous G␥, Tt1 ⁄2 of G␤ was considered as an indicator of endogenous G␥ translocation and we termed it average Tt1 ⁄2 (avg-Tt1 ⁄2 ). The fast Tt1 ⁄2 of G␤ observed in G␥9 expressing cells (Tt1 ⁄2 of G␤1 ϭ 7 Ϯ 2 s and G␤2 ϭ 6 Ϯ 1 s) confirms that G␤ represents translocation properties of endogenous or introduced G␥ (Fig. 7, D and E and Table S2). The avg-Tt1 ⁄2 value observed for RAW cells (221 Ϯ 5 s) was greater than avg-Tt1 ⁄2 of HeLa cells (93 Ϯ 2 s) ( Fig. 7D and Table S2). These results suggest that, compared with HeLa cells, RAW cells express more HiAf-G␥ types. These data are also in agreement with real-time PCR data of G␥ mRNA (Fig. 1D). To ensure that the type of G␤ does not influence endogenous G␥ translocation, similar experiments were performed in both HeLa and RAW cells, however expressing YFP-G␤2 ( Fig. 7E and Table S2). The observed Tt1 ⁄2 of G␤1 and G␤2 were comparable, suggesting that the type of G␤ does not alter the Tt1 ⁄2 of G␥.
Next, blue opsin activation induced PIP3 production in WT HeLa, and WT RAW cells were measured to obtain effector activity induced by endogenous G␤␥ (͉EF͉ exp or ⌬[PIP3]). Data show that RAW cells possess a 2-fold higher effector activation ability than HeLa cells (Fig. 7F). The avg-Tt1 ⁄2 obtained above for both HeLa and RAW cells (Fig. 7E) were then extrapolated on the HeLa effector plot (Fig. 7C) to obtain predicted effector activities (͉EF͉ calc ). The ratio of experimental and calculated effector activities (͉EF͉ exp :͉EF͉ calc ) for HeLa and RAW cells were found to be 0.82 and 0.99, respectively. This shows that ͉EF͉ exp :͉EF͉ calc ratios for both HeLa and RAW cells are closer to 1. Therefore, the avg-Tt1 ⁄2 of endogenous G␥ pool is a strong predictor of a cell's G␤␥ effector-activation ability.

PM-residing ability of G␤␥ produced upon GPCR activation on its effector activation potential
Similar to HiAf-G␥ types, moderate affinity G␥ (MoAf-G␥) types also maintain considerably high G␤␥ concentrations on the PM after GPCR activation (Fig. 2B). However, it was unclear why MoAf-G␥ expression does not promote robust PIP3 productions, as seen in HiAf-G␥ expressing cells (Figs. 5E and 7B). To comprehend this, a model was proposed in which G␤␥ on the PM stays in a transiently active conformation (G␤␥-PM*) which can activate effectors. We assume that the lifetime () of this transiently active conformation is dependent on the corresponding G␥ type or more specifically the PM affinity of G␥ type. Using G␥12 as a model MoAf-G␥, we first examined if the observed lack of translocation in G␥12 is controlled by factors other than its CT. We substituted CT of G␥12 with CT of G␥3 and G␥9, respectively (Fig. S5). Compared with the moderate translocation observed in G␥12 (Tt1 ⁄2 ϭ ϳ80 s) (Table S1), G␥12-␥9CT mutant showed a fast translocation with Tt1 ⁄2 ϭ ϳ8 s, resembling translocation properties of G␥9. As expected, G␥12-␥3CT mutant translocated slower that G␥12 (Tt1 ⁄2 ϭ ϳ232 s) (Fig. S5). Similar changes were observed for G␥3-␥9CT mutant (Fig. 4C). Because the CT of G protein ␥ subtype dictates G␤␥ signaling G␥ does not interact with the receptor, the fast translocating mutants of MoAf-and HiAf-G␥ suggest that their heterotrimers are equally activated by the GPCR, as seen for heterotrimers with LoAf-G␥ through their intense G␤␥ translocation. These observations indicate that, although MoAf-G␤␥ are liberated from the heterotrimer and reside on the PM, a fraction of them are not conformationally appropriate for G␤␥ effector activation. These findings are consistent with recent reports suggesting that K-Ras possesses orientation-dependent effector binding (36).

Data-guided computational modeling of G␥ subtypes-driven signal propagation
As an extension to our previous model (37), reactions for G␤␥ effector activation in a multi-G␥ containing cell were modeled, to decipher how a diverse group of G␥ types in a cell controls G␤␥ effector activities. The optimized model (Variation 3) encompasses reactions in Fig. 8A (see Equations S1 and S2, Variations). The model includes novel circuits for G␤␥ to (a) activate effectors, (b) translocate to IMs, and (c) be in fluctuating conformationally active and inactive states on the PM G␤␥ PM n* %G␤␥ PM n . Considering that many proteins fluctuate among multiple conformations, it is likely that G␤␥ fluctuates among these structures while one particular conformation has a stronger affinity toward G␤␥ effectors (Fig. 8, A and B) (36,38,39). The kinetic curves show that G␤␥ in the mono-G␥3 system is primarily in the PM-bound active state (G␤␥-PM*), and is available for effector activation (Fig. 8C). The simulations show low concentrations of G␤␥-PM* for mono-G␥4 and mono-G␥9 systems compared with mono-G␥9. Simulations also demonstrate that, in mixed G␥ systems, G␥3 is the primary contributor for the active G␤␥ on the PM (PM*) (Fig. 8D). Concentrations of activated effectors ([␤␥ PM tot EF]) were plotted as a function of time. The mono G␥-systems exhibited effector responses (Fig. 8E), and are similar to the experimental PIP3 and Ca 2ϩ responses observed in HeLa and RAW cells expressing specific G␥ subtypes, that we also defined as mono-G␥ systems. Simulations show that mono-G␥3 systems rapidly activate 90% of the effectors in 195 s, whereas G␥4 and G␥9 systems exhibit minor effector activities.
Effector activation by mixed G␥ [␤␥ PM 3ϩ4ϩ9 EF] system with equal compositions of HiAf-G␥3, MoAf-G␥4, and LoAf-G␥9 (Fig. 8D) exhibited that G␥3 interacts with effectors the most (Fig. 8E). In this system at 600 s, 99% of the effectors are activated and 82% of these effectors are bound to G␥3 (i.e. [␤␥ PM 3 EF]), 16% to G␥4 ([␤␥ PM 4 EF]), and 2% to G␥9 ([␤␥ PM 9 EF]). Here, effector activations by MoAf-G␥4 and LoAf-G␥9 are lower with respect to mono-G␥ systems. Thus, the model predicts that in mixed systems, physiological responses are primarily governed by HiAf-G␥. This is also observed for 1:1 mixture of MoAf-G␥4 and LoAf-G␥9, where the activity is determined by the available higher affinity G␥ subtype (i.e. G␥4, Fig. 8F). It is noteworthy that, if translocation rate constants are set equal for individual G␥ type (i.e., k in n ϭ k out n ), this activity dominance of HiAf-G␥ is not observed. When fluctuation between activeinactive G␤␥ conformations was not incorporated, equal effector activity from G␥3 and G␥4 is observed, contradicting experimental observations.

Discussion
Considering diverse and unique tissue-and cell type-specific G␥-type distribution patterns, the G␥ identity-specific regulation of G␤␥ signaling can have a broader impact on the current understanding of GPCR-G protein signal transduction. If G␤␥ were to be a unitary signaling entity, cells would have intense G␤␥ signaling on all occasions of GPCR activation, which can be deleterious. For instance, RAW cells have a G␥ profile with HiAf-G␥ that supports PI3K activation and PIP3 production. However, for a usually immobile cell type like HeLa, intense PIP3 production may not serve a purpose and thus HiAf-G␥ expression is not required. Supporting this notion, G␥3 expression allowed HeLa cells to produce PIP3 upon GPCR activation. The fundamental difference identified between the introduced G␥3 over endogenous G␥ types in HeLa cells was the ability of G␥3 to make G␤␥ more available at the PM, where PIP3 production takes place. To catalyze PIP2 to PIP3, G␤␥ recruits and activates PI3K subunits to the PM (40). Of the 12 G␥ types, only G␥3 and G␥2 promoted PIP3 production. This is likely because of the weak translocation properties of G␤␥3 that allows maintenance of a relatively higher concentration of free G␤␥ on the PM. The G␥-dependent differential PIP3 generation in HeLa cells hints at a plausible mechanism of how G␤␥ effectors are recruited to the PM and activated by PM-bound fraction of HiAf-G␤␥. We recently showed that G␤␥ controls PLC␤ activation, induces Ca 2ϩ mobilization, governing the trailing edge retraction during RAW cell migration (18). Similar to PI3K␥, PLC␤1 and PLC␤2 are also cytosolic (41,42). Our data suggest that PM targeting and/or activation of these G␤␥ effectors are likely to be governed by the PM affinity of G␤␥. The extent of effector responses suggests that the stronger the PM affinity of G␤␥, the greater its potential to control signaling. Here we employed Tt1 ⁄ 2 as an index for the residence time on the PM or the PM affinity of G␤␥. The free energy of the translocation (⌬G) is considered as the energy required to dislodge G␤␥ from the PM to the IM. Thus it is a direct measure of PM affinity of G␤␥ to the PM. ⌬G is related to a first-order reaction equilibrium constant (K eq ) by ⌬G ϭ ϪRT ln K eq . For the G␤␥ translocation process, and the half-time, t1 ⁄ 2 can be expressed as k ϭ 0.693/t1 ⁄ 2 , thus ⌬G ϭ ϪRT ln(t 1/2 out / 1/2 in . This indicates that the longer the residence time on the PM, the greater the PM affinity. Translocation t1 ⁄ 2 of G␤␥ is a complex measure which includes the shuttling of G␤␥ between the PM and IMs. However the initial reaction is dominated by G␤␥ dislodging from the PM (k in ), Figure 6. PLC␤ activation induced differential Ca 2؉ response with different G␥s. RAW 264.7 cells expressing different WT G␥s and G␥ mutants were stimulated with 10 M c5a addition to activate endogenous c5a receptors (c5aRs) after 30-min Fluo-4 incubation. Cells were imaged at 40ϫ magnification to capture the Ca 2ϩ response. A-C, control (A) (mCherry-GPI and untransfected), (B) G␥3 expressing cells showed greater Ca 2ϩ response compared with (C) G␥9 expressing cells, which showed almost no Ca 2ϩ . Scale bar, 10 m. D, plot shows the difference in Fluo-4 signal (GFP fluorescence) increase in cells, indicating differential Ca 2ϩ release to the cytoplasm depending on the G␥ subtype they overexpress. Also, it shows that the fluorescent tag of the G␥ subtype is not affecting the Ca 2ϩ response (n ϭ 8). E and F, MoAf-G␥4 and G␥12 expressing cells showed minor Ca 2ϩ response with c5AR activation (n ϭ 8). G and H, G␥9 mutants with G␥3 CaaX and G␥3 CT showed an increased Ca 2ϩ response compared with WT G␥9, whereas G␥3 with G␥9 CaaX and G␥9 CT showed a reduced Ca 2ϩ response compared with WT G␥3, confirming differential G␤␥-effector interactions with respect to the difference in the PM affinity thus different PM residence times of G␤␥ (error bars, S.E.; scale bar, 10 m). The ͉EF͉ was measured using PIP3 production on the PM in HeLa cells expressing each of the 12 G␥ types and Akt-PH-mCherry. D, HeLa cells expressing blue opsin-mTurquoise, mCherry-G␥9, either YFP-G␤1 or YFP-G␤2, respectively, supplemented with 50 M 11-cis-retinal. On blue opsin activation, both G␤1 and G␤2 exhibited Tt1 ⁄2 closer to that of G␥9, further confirming that the translocation properties of G␤ represent the prominent G␥ subtype expressed in the cell. E, HeLa and RAW cells expressing blue opsin-mTurquoise and either YFP-G␤1 or YFP-G␤2, respectively, were supplemented with 50 M 11-cis-retinal. Cell was imaged for YFP and blue opsin was activated with 445 nm light every 3 s. G␤ translocation exhibited the average translocation properties of the entire pool of endogenous G␥. G␤ type does not influence translocation properties of endogenous G␥ in HeLa cells. The plot shows that Tt1 ⁄2 of G␤1 and G␤2 translocation was closer to the Tt1 ⁄2 of the most abundant G␥ of each cell type. F, blue opsin activation-induced experimental ͉EF͉ (PIP3 response) measured in WT HeLa and RAW cells expressing blue opsin and the PIP3 sensor (error bars, S.E.; n ϭ 10; scale bar, 5 m).

G protein ␥ subtype dictates G␤␥ signaling
thus, as shown above, Tt1 ⁄ 2 is a fair approximation of the PM affinity of G␤␥.
G␥ subunits interact with the PM through the prenyl group. The type of prenylation is decided by the CaaX motif sequence of G␥. The prenylation with 20-carbon geranylgeranyl lipid provides a higher PM affinity to G␤␥ compared with the 15-carbon farnesyl lipid attachment. Except for farnesylated G␥9, G␥1, and G␥11, all other G␥ types are geranylgeranylated.

G protein ␥ subtype dictates G␤␥ signaling
However, only G␥3 and G␥2 supported RAW cell migration, suggesting factors additional to the type of prenylation control the PM affinity of G␥. Interestingly, pre-CaaX regions of G␥3 and G␥2 are composed of ϳ80% positively charged and hydrophobic residues, as opposed to ϳ50% in farnesylated G␥ subunits. Extensive mutagenesis to the pre-CaaX region of G␥ suggested that this five-or six-residue region modulates G␤␥-PM interactions, in which positively charged and hydrophobic amino acids strengthen the PM affinity. Previously reported translocation data of G␥ mutants with altered pre-CaaX residues further validate the role of this motif in controlling the PM affinity (29). The complete loss of PM localization observed in G␥9 upon cysteine removal from CaaX motif indicates that pre-CaaX region only serves as a strong modulator of PM affinity, whereas prenylation is essential for primary PM anchoring of G␤␥. By modulating properties of their pre-CaaX motifs, geranylgeranylated G␥ subunits managed to possess a discrete series of PM affinities.
Heterotrimers with specific G␥ types have been shown to possess higher affinities toward certain GPCRs (20,43). However, an exchange of the CT of slow translocating G␥3 and moderate translocating G␥12 with the CT of G␥9 resulted in fast translocating mutants, comparable to G␥9. This can suggest that either (a) heterotrimer activation process is controlled by the CT of G␥ through modulating G␣␤␥-GPCR interactions or (b) the PM affinity of generated G␤␥ is dependent on the CT of G␥ subunit. Regardless, the CT of G␥ should hold a crucial control over G␤␥ function, although our data strongly support possibility (b). We anticipate that, among the available G␤␥ pool, LoAf-and MoAf-G␤␥ types exist primarily to support G␣GTP generation, whereas HiAf-G␤␥ subunits activate G␤␥ effectors. Our data also support that the PM-bound G␤␥ composed of HiAf-G␥ types stay a longer fractional time in the active conformation, compared with their LoAf and MoAf associates. Lack of migration ability in G␥3-knockdown RAW cells strongly supports this notion, because the remaining MoAf-G␥ types in RAW cells lack effector activation ability. Nevertheless, we are aware that, in addition to G␥ diversity, there are converging and diverging pathways and signaling components, including integrins, secretory proteins (i.e. matrix metalloproteinases) can influence the migration potential of a cell (44 -46). Therefore, differences in cell migration potentials among cell types with diverse origins should be examined considering these potential inherent variables. Although we are in concert with these reports, our findings demonstrate that G␤␥-governed migration requires appropriate G␥ types with higher PM affinity. Supporting these findings, even a nonmigratory cell type like HeLa expressing HiAf-G␥3, undergo directional migration upon blue opsin activation.
Avg-Tt1 ⁄ 2 of endogenous G␥ measured using G␤ translocation accurately predicted the ability of native G␤␥ to control its effectors. Predicted effector activity using this method was similar to the PIP3 production observed in both RAW and HeLa cells. These observations suggest that Tt1 ⁄ 2 , and therefore the PM affinity, of a G␥ type is a strong indicator of the ability of G␤␥ to activate effectors. Thus, our observations collectively indicate that G␥ subunit diversity in a cell is a crucial factor in determining whether the cell has the ability to activate G␤␥ effectors sufficiently to orchestrate the intended behaviors, including migration.
In the kinetic model with multi-G␥ and embedded experimental observations that Tt1 ⁄ 2 ϰ PM affinity and Tt1 ⁄ 2 ϰ ͉EF͉, multiple mechanistic scenarios associated with G protein activation were attempted. The incorporation of an active-inactive conformation circuit to the PM-bound G␤␥ was required to simultaneously capture all the experimental responses observed. These include the lack of effector activation by G␤␥ associated with MoAf-G␥. The incorporated circuits to the model indicated that (a) HiAf-G␤␥ subtypes tend to readily activate effectors, initiating downstream signaling; (b) the majority of LoAf-G␤␥ types translocate away from the PM to down-regulate signaling; and (c) the fraction of MoAf-G␤␥ that did not translocate tends to minimize signaling by oscillating between PM-bound active-inactive conformational states. Although these conformational fluctuations are common for all types of G␤␥, the G␥ type and the PM affinity decide the lifetime of their active state. The ability of this model to recapitulate experimental responses indicates its reliability. Therefore, reactions and parameters embedded in our model (Table S3) are likely to closely reflect how PM affinities of G␥ subunits modulate information flow from the activated GPCRs to effectors. The model also allowed simulation of experimentally challenging in vivo conditions, including varying ratios of HiAf-G␥: LoAf-G␥ and total G␤␥ concentrations.
In summary, this study demonstrates that distinct translocation abilities of the 12 G␥ types provide G␤␥ a diverse range of PM interaction and effector activation abilities. Because most G␤␥-effector activities occur at the PM, data confirm that the PM affinities of G␥ types expressed in a cell are deterministic to the potency of G␤␥ effector as well as downstream signaling activation. Although we only show G␥ identity-dependent control of PI3K␥ and PLC␤, and their regulation of cell migration, it is likely that a plethora of G␤␥-mediated functions are similarly regulated. Because GPCR-G protein signaling is universally conserved and G␤␥ signaling pathways are major drug targets, mechanisms we describe here can have a wide influence not only on cell migration but also in many areas of signaling.

Reagents
The reagents gallein (TCI America), Fluo-4 AM (Molecular Probes, Eugene, Oregon), wortmannin 2APB (Cayman Chemical, Ann Arbor, MI), C5a (Eurogentec), U50488 hydrochloride (Tocris) were initially dissolved in DMSO and then diluted in Hanks' Balanced Salt Solution (HBSS) (Gibco) before adding to cells. 11-cis retinal (National Eye Institute) was initially resuspended in absolute ethanol and 2-l aliquots (50 M) were further diluted (2 l for each aliquot) with absolute ethanol before introducing (2 l) to cells in dark. SDF-1␣ (PeproTech) was reconstituted in deionized water to a concentration 100 g/ml and further diluted with a buffer containing 0.1% BSA before adding to cells.

G protein ␥ subtype dictates G␤␥ signaling DNA constructs and cell lines
Engineering of DNA constructs used for blue opsin-mCherry, blue opsin-mTurquoise, Akt-PH-mCherry, and YFP-tagged G␥1-G␥13 have been described previously (33,47,48). YFP-␤1 and -␤2, -opioid receptor, PI3K-CA-CFP, and mCherry-GPI were kind gifts from Professor N. Gautam's lab, Washington University, St. Louis, Missouri. G␥3, G␥9, and G␥12 mutants were generated using Gibson assembly (New England Biolabs (NEB)) (28). Parent constructs mCherry-G␥3, mCherry-G␥9, and YFP-G␥12 were PCR amplified with overhangs containing expected nucleotide mutations. DpnI (NEB) digestion was performed on the PCR product to remove the parent construct. DpnI-digested PCR product was then mixed with the Gibson assembly master mix (NEB) and incubated at 50°C for 45 min, which was followed by transformation of competent cells and plating on ampicillin LB agar plates. All the constructs used in this study possess the ampicillin-resistant pcDNA 3.1 vector backbone. Cell lines (HeLa, RAW 264.7, PC12, and HEK cells) were originally purchased from the American Tissue Culture Collections (ATCC) and authenticated using a commercial kit to amplify nine unique STR loci.

Cell culture and transfections
RAW 264.7 cells used in migration and PIP3 generation experiments were cultured in RPMI 1640 (10 -041-CV; Corning, Manassas, VA) with 10% dialyzed fetal bovine serum (DFBS; Atlanta Biologicals) and 1% penicillin-streptomycin (PS) in 60 mm tissue culture dishes. HeLa cells were maintained in minimum essential medium (MEM; CellGro) supplemented with 10% DFBS and 1% PS. Around 80% cell confluency, the growth medium was aspirated, 2 ml Versene (EDTA) (CellGro) was added, incubated for 3 min at 37°C, 5% CO 2 incubator, and then cells were lifted and suspended in Versene. The cell suspension was centrifuged at 1000 ϫ g for 3 min, Versene (EDTA) was aspirated, and the cell pellet was resuspended in its normal growth medium (RPMI/DFBS/PS for RAW and MEM/ DFBS/PS for HeLa) at a cell density of 1 ϫ 10 6 /ml. For imaging experiments, cells were seeded on 35-mm glass-bottomed dishes (8 ϫ 10 4 cells on each) with 15-mm inner diameter, prepared using no. 1 German cover glasses. Before cell seeding, dishes were washed with 2 N NaOH for 20 min, ethanol washed, and sterilized for 1 h using UV irradiation. A day following cell seeding, cells were transfected with appropriate DNA combinations using the transfection reagent PolyJet (SignaGen Laboratories), according to the manufacturer's protocol and then incubated in a 37°C, 5% CO 2 incubator. Cells were imaged after 16 h of the transfection.

Knockdown of G␥3 in RAW 264.7 cells
Five shRNAs (TRCN0000036794 -98; Sigma-Aldrich) were screened in RAW cells by co-expressing with GFP-G␥3. Cells were screened for GFP expression, and the shRNA construct that induced the highest reduction in GFP-G␥3 expression was selected as the most effective shRNA. The identified TRCN0000036795 shRNA (sequence: CCGGGCTTAAGAT-TGAAGCCAGCTTCTCGAGAAGCTGGCTTCAATCTTA-AGCTTTTTG) was employed to knock down G␥3 in the sub-sequent experiments. A scrambled shRNA was used as the control.

Live cell imaging to monitor G␤␥ translocation, PIP3 generation, and optogenetic control of cell migration
Imaging system comprised a spinning-disk XD confocal TIRF (total internal reflection) imaging system composed of a Nikon Ti-R/B inverted microscope, a Yokogawa CSU-X1 spinning disk unit (5000 rpm), an Andor FRAP-PA (fluorescence recovery after photo-bleaching and photo-activation) module, a laser combiner with 40 -100 milliwatt 445, 488, 515, and 594 nm solid-state lasers and iXon ULTRA 897BV back-illuminated deep-cooled EMCCD camera. Live cell imaging was performed using a 60ϫ, 1.4 NA (numerical aperture) oil objective. In cell migration and PIP3 generation experiments, mCherry-tagged receptor blue opsin and the PIP3 sensor Akt-PH were imaged using 594 nm excitation-630 nm emission. To activate blue opsin, 50 M 11-cis retinal (National Eye Institute) was added and incubated 3-5 min in dark. After incubation, the fluorescent sensor in cells was imaged to capture basal signaling in cell migration and PIP3 generation experiments (i.e. Akt-PH-mCherry in PIP3 experiments and blue opsin-mCherry in cell migration experiments), and then receptor blue opsin was activated by shinning 445 nm blue light at 0.1% transmittance and imaging was continued for 20 min. To examine the G␤␥ translocation, YFP and GFP fluorescent tags on G␥ subunits were imaged for 10 min using 515 nm excitation, 527 nm emission or 488 nm excitation, 515 nm emission, respectively. Regular culture media or HBSS supplemented with 1 g/ml glucose preincubated in a 37°C, 5% CO 2 incubator for 30 min were used as the imaging medium. During imaging, cells were maintained at 37°C. To prevent focal plane drifts, Nikon Perfect Focus System (PFS) was engaged.

Cytosolic Ca 2؉ measurements
For intracellular Ca 2ϩ measurements, RAW cells seeded on glass-bottom dishes and maintained in at 37°C with 5% CO 2 were transfected with a G␥ subtype on the following day of cell seeding. After 12-16 h of transfection, cells were washed twice with Ca 2ϩ -containing HBSS (pH 7.2) and incubated for 30 min at room temperature with the fluorescent Ca 2ϩ indicator, Fluo-4 AM in the dark. After incubation, cells were washed twice with HBSS and 500 l of HBSS was then used as imaging medium. The fluorescence intensity of Fluo-4 AM was continuously imaged at 1-s intervals using 488 nm excitation, 515 nm emission to capture signal before activation for 50 s. Endogenous c5aRs in RAW cells were activated with 10 M c5a. Observed Fluo-4 AM fluorescence increase because of Ca 2ϩ release was baseline normalized.

Real-time PCR, transcriptome, and RNA seq data analysis
To obtain the G␥ profile of WT HeLa and WT RAW 264.7 cells, RNA was extracted from cells grown in 100 mm tissue culture dishes after reaching 90 -100% cell confluency. RNA extraction was performed using the GeneJet RNA purification kit following their given protocol. Extracted RNA was used as the template for cDNA synthesis with Radiant cDNA G protein ␥ subtype dictates G␤␥ signaling synthesis kit. cDNA product was quantified using the Nano-Drop and used for real-time PCR (Bio-Rad CFX96 Real-Time qPCR system) in 96-well plates to obtain the G␥ profile. Radiant Green Lo-ROX qPCR kit (Alkali Scientific) was used in real-time PCR experiments and ␤ actin gene was used as the housekeeping gene. To screen the G␥ profile alteration with G␥3 and G␥9 overexpression, HeLa cells were seeded in 100 mm tissue culture dishes and transfected with GFP-G␥3 and GFP-G␥9, respectively, at 70 -80% cell confluency, and RNA was extracted after confirming greater than 70% transfection efficiency by observing under the microscope. This was followed by cDNA preparation and real-time PCR.

Statistics and reproducibility
Results of all quantitative assays (G␤␥ translocation, cell migration, and PIP3 generation) are expressed as standard error of mean (S.E.) from n number of cells (indicated in the figure legends) from multiple independent experiments. Statistical analysis of cell migration data of WT and mutant G␥ subtypes was performed using two-tailed unpaired t test. p value Ͻ0.05 was considered as statistically significant.

Computational modeling
The dynamic nature of the GPCR signal transduction has been modeled by a series of ordinary differential equations (Equation S2) which encompasses the series of reactions in Fig.  8A. Computations were performed in a custom Python 2.7 script with odeint module to numerically integrate the ordinary differential equations. The equations are an extension of our previous model for a ligand-activated signal transduction and is extended to allow for effector activation by multiple types of G␤␥ subunits (HiAf, MoAf, and LoAf). The reaction mechanism is similar to our previous publication (37) with the classical GPCR activation cycle, and includes novel circuits for G␤␥ to 1) activate effectors, 2) translocate to IMs, and 3) be in a conformationally inactive structure on the PM. The equations describe the rates of heterotrimer dissociation, heterotrimer association ( Equation S1), G␣(GTP) hydrolysis, G␤␥ translocation to IMs, G␤␥ oscillation to an inactive configuration, and G␤␥ effector activation. The rates for heterotrimer dissociation and G␣(GTP) hydrolysis assume Michaelis-Menten kinetics; all others are assumed as first-or second-order reactions. The ordinary differential equations define the rates of formation or depletion of the important species in the signaling network (i.e. G␣(GDP)␤␥, G␣(GDP), G␣(GTP), G␤␥ PM , G␤␥ PM* , G␤␥ IM , and G␤␥ EF ). To incorporate multiple G␤␥ subunits and consequently the G␥ diversity, there is an ordinary differential equation for each G␤␥ type except the G␣(GTP) and G␣(GDP) concentrations. Numerically integrating these functions, the concentration of the species over time was determined.