G protein βγ translocation to the Golgi apparatus activates MAPK via p110γ-p101 heterodimers

The Golgi apparatus (GA) is a cellular organelle that plays a critical role in the processing of proteins for secretion. Activation of G protein–coupled receptors at the plasma membrane (PM) induces the translocation of G protein βγ dimers to the GA. However, the functional significance of this translocation is largely unknown. Here, we study PM-GA translocation of all 12 Gγ subunits in response to chemokine receptor CXCR4 activation and demonstrate that Gγ9 is a unique Golgi-translocating Gγ subunit. CRISPR-Cas9–mediated knockout of Gγ9 abolishes activation of extracellular signal-regulated kinase 1 and 2 (ERK1/2), two members of the mitogen-activated protein kinase family, by CXCR4. We show that chemically induced recruitment to the GA of Gβγ dimers containing different Gγ subunits activates ERK1/2, whereas recruitment to the PM is ineffective. We also demonstrate that pharmacological inhibition of phosphoinositide 3-kinase γ (PI3Kγ) and depletion of its subunits p110γ and p101 abrogate ERK1/2 activation by CXCR4 and Gβγ recruitment to the GA. Knockout of either Gγ9 or PI3Kγ significantly suppresses prostate cancer PC3 cell migration, invasion, and metastasis. Collectively, our data demonstrate a novel function for Gβγ translocation to the GA, via activating PI3Kγ heterodimers p110γ-p101, to spatiotemporally regulate mitogen-activated protein kinase activation by G protein–coupled receptors and ultimately control tumor progression.

G protein-coupled receptors (GPCRs) modulate a wide variety of cell functions through activating cognate heterotrimeric G proteins, arrestins, and other signaling molecules (1). In the classical GPCR signaling system, GPCRs at the plasma membrane (PM), once activated by hormones or neurotransmitters, function as guanine nucleotide exchange factors to enhance the exchange of GDP for GTP from Gα subunits, leading to the dissociation of active GTP-bound Gα and free Gβγ dimers, which can separately activate downstream effectors, such as adenylyl cyclases, phospholipases, mitogen-activated protein kinases (MAPKs), phosphoinositide 3-kinases (PI3Ks), and ion channels (2)(3)(4). In recent years, several studies have demonstrated that, after activation by GPCRs at the PM, some Gβγ dimers can translocate from the PM to intracellular organelles, including the Golgi apparatus (GA), likely via passive diffusion, and that the efficiency of translocation is determined by Gγ anchoring to the PM, as well as Gγ interaction with the receptors (5)(6)(7)(8)(9)(10). Although the Golgi-localized Gβγ complex can activate phospholipase C (11,12) and protein kinase D (13,14) and regulate post-Golgi trafficking (14)(15)(16), Golgi structure (13,16,17), insulin secretion (17), and cardiomyocyte hypertrophic growth (11), the physiological and pathophysiological functions of Gβγ translocation from the PM to the GA are still largely undefined.
The MAPKs extracellular signal-regulated kinases 1 and 2 (ERK1/2) and PI3Ks are crucially involved in many fundamental cellular processes and are directly associated with the pathogenesis of human diseases, particularly cancer. It is known that almost all GPCRs can activate the MAPK Raf-MEK-ERK1/2 pathway. However, ERK1/2 activation by different GPCRs or the same GPCR in different cell types may be mediated through different biochemical pathways, involving distinct signaling molecules, such as Gβγ subunits, arrestins, small GTPases, receptor tyrosine kinases (RTKs), and protein kinases (18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29). Although these complex activation mechanisms have been studied extensively and arrestins have been shown to function as scaffolds in ERK1/2 activation on the endosomal compartment, the spatial aspects of ERK1/2 activation by GPCRs remain poorly explored and signal initiation by G proteins is generally considered to occur at the PM (23,27,28,30).
PI3Ks are a family of lipid kinases that specifically phosphorylate the inositol moiety of phospholipids at the 3' position and can be divided into three classes based on their sequence homology and substrate specificity. Class I PI3Ks include PI3Kα, β, γ, and δ isoforms, which are heterodimers consisting of a catalytic subunit and a regulatory subunit, and the PI3Kγ catalytic subunit p110γ can form a complex with the regulatory subunit p101 or p87. It has been known that the cell surface GPCRs can activate PI3Kβ, γ, and δ isoforms and Gβγ dimers activate PI3Kβ and γ isoforms. PI3Kγ activation by Gβγ has been extensively studied, and these studies have shown that Gβγ can directly interact with both p110γ and p101 (31)(32)(33).
Over the past decades, multiple signaling cascades have been defined to contribute to prostate tumorigenesis. Among these cascades, the ERK1/2 and PI3K pathways are hyperactivated in prostate cancer and enhanced activation of ERK1/ 2 and PI3Ks is correlated with disease progression, androgen independence, and poor prognosis (33)(34)(35)(36). However, the molecular mechanisms responsible for the hyperactivation of the ERK1/2 and PI3K pathways in prostate cancer remain poorly understood. Here we study the function of Gβγ translocation to the GA in ERK1/2 activation by the chemokine receptor CXCR4 in human androgen-insensitive prostate cancer (DU145 and PC3) cells and human embryonic kidney 293 (HEK293) cells. CXCR4 is a crucial factor involved in bone metastasis of prostate cancer and CXCR4 antagonists inhibit prostate cancer growth and metastasis (37,38). We demonstrate that Gβγ translocation to the GA activates the ERK1/2 pathway through the PI3Kγ heterodimer p110γ-p101. We also show that knockout of the most translocatable Gγ9 subunit and p110γ markedly inhibits prostate cancer cell migration, invasion, and metastasis. These data reveal a novel function of Gβγ translocation to the GA to spatiotemporally regulate MAPK activation by GPCRs and control tumor progression.

Characterization of Gβγ translocation to the GA in response to CXCR4 activation
Individual YFP-tagged Gγ subunits were expressed together with Gβ1, a commonly expressed Gβ subunit, Gαi1 subunit, and the Golgi marker pmTurquoise2-Golgi in DU145, PC3, and HEK293 cells. Gβγ translocation to the GA in response to CXCR4 activation by stimulation with stromal cell-derived factor 1α (SDF1α) was measured by the increase in total YFP signal at the GA by confocal microcopy in live cells (Fig. S1A). All 12 Gγ subunits were mainly expressed at the PM in unstimulated cells, and SDF1α stimulation induced Gγ translocation from the PM to intracellular compartments at different magnitudes and rates. However, translocation of Gγ9 to the GA was dramatically more efficient than that of any other Gγ subunit ( Fig. 1A and Fig. S1B). The total Gγ9 signal at the GA increased by approximately 80% (Fig. 1B). The quantification of relative expression of Gγ9 at the PM, nucleus, and GA showed that SDF1α stimulation markedly reduced Gγ9 expression at the PM and increased Gγ9 expression at the GA (Fig. 1C). Translocation of Gγ9 from the PM to the GA occurred with a half time (t 1/2 ) of about 5 s (Fig. 1D). The closest Gγ subunit to Gγ9 was Gγ11, which GA translocation increased by only 30%. In contrast, Gγ3 showed the least translocation (<10%) (Fig. 1B). Across Gγ subunits, the kinetics and extent of translocation were similar in DU145, PC3, and HEK293 cells (Fig. 1, B and D). These data indicate that Gγ9 is a unique Golgi-translocating Gγ subunit.
In order to discriminate the roles of translocating and nontranslocating Gγ subunits in MAPK activation, we used the CRISPR-Cas9 genome editing system to selectively knock out Gγ9 or Gγ3 in DU145, PC3, and HEK293 cells. Gγ9 and Gγ3 knockout cells were verified by Western blotting using Gγ-specific antibodies (Fig. 2E). Gγ9 knockout did not affect Gγ3 expression, and vice versa (Fig. 2E). Gγ9 and Gγ3 knockout cells were also confirmed by the respective depletion of transiently expressed YFP-Gγ9 and YFP-Gγ3, without affecting YFP-Gγ2 expression (Fig. S3A). Gγ9 knockout in all three cells almost completely inhibited ERK1/2 activation by SDF1α, whereas Gγ3 knockout did not have a clear effect (Fig. 2F). Gγ9 knockout abolished ERK1/2 activation after SDF1α stimulation at all time points tested in PC3 cells (Fig. S3, B and C). Furthermore, transient expression of single guide RNA (sgRNA)-resistant Gγ9 successfully rescued ERK1/ 2 activation by SDF1α in Gγ9 knockout cells ( Fig. 2G and Fig. S3D). These data strongly indicate that the normal expression of endogenous Gγ9 is essential for ERK1/2 activation by CXCR4.
Constitutive targeting of Gβγ to the GA directly activates ERK1/2 To further define the role of Gβγ translocation to the GA in ERK1/2 activation, a rapamycin-inducible translocation system was utilized to specifically recruit Gβγ dimers to either the GA or the PM. In this system, specific targeting peptides (i.e., the mutant KDELr-D193N for GA targeting and amino acids 1-11 of Lyn for PM targeting) were fused to FK506-binding protein (FKBP), while individual cytosolic Gγ subunits were fused to the FKBP-rapamycin binding (FRB) domain. This system has been used previously to recruit Gβγ to the GA and the PM in HeLa cells and cardiomyocytes (11,14). The system was confirmed by using venus-Gβ1 and Gγ9 (Fig. S4). Of interest, rapamycin-induced recruitment of Gγ2, Gγ3, or Gγ9 to the GA (Golgi-Gγ), each in complex with Gβ1, was able to activate ERK1/2 in DU145, PC3, and HEK293 cells, whereas recruitment to the PM (PM-Gγ) had no obvious effect (Fig. 3A). Rapamycin incubation to induce Gγ9 translocation to the GA activated ERK1/2 in a time-dependent fashion (Fig. 3, B and C). These data suggest that different Gβγ dimers can specifically activate ERK1/2 if they are present at the GA.
To verify if Golgi-Gβγ-mediated activation of ERK1/2 indeed occurred at the GA, we measured the effects of wellknown Golgi disruptors, including brefeldin A (BFA), ilimaquinone, monensin, nigericin, nocodazole, and swainsonine, which induce Golgi fragmentation via distinct mechanisms (40). Incubation with each of these Golgi disruptors caused Golgi fragmentation as indicated by GalT, a trans-Golgi marker (Fig. S5), and abolished ERK1/2 activation induced by Golgi-Gγ9 (Fig. 3D). BFA dose dependently inhibited ERK1/2 activation by Golgi-Gγ9 expression, and the IC 50 values were 0.37 ± 0.03, 0.19 ± 0.05, and 0.18 ± 0.03 μM (n = 3) in DU145, PC3, and HEK293 cells, respectively (Fig. 3, E and F). These data demonstrate that the integrity of the Golgi structure is required for ERK1/2 activation by Golgi-Gγ. As BFA treatment to disrupt the GA inhibits CXCR4 expression at the PM, likely by interfering its anterograde transport from the endoplasmic reticulum to the cell surface (41), we did not measure the effects of these Golgi disruptors on ERK1/2 activation by SDF1α.
To further investigate the role of PI3Kγ in ERK1/2 activation, we generated p110γ knockout DU145, PC3, and HEK293 cells using the CRISPR-Cas9 system (Fig. 5A). Similar   to the results obtained from Gγ9 knockout cells, SDF1α was unable to activate ERK1/2 in p110γ knockout cells (Fig. 5B). Furthermore, inducible translocation of Gγ9 to the GA failed to activate ERK1/2 in p110γ knockout cells (Fig. 5C). These data demonstrate that ERK1/2 activation by SDF1α and Gβγ translocation to the GA depends on PI3Kγ.
To define the role of regulatory subunits of PI3Kγ in ERK1/ 2 activation by Gβγ, we determined the effect of siRNAmediated knockdown of p101 and p87. Similar to p110γ knockout by CRISPR-Cas9, p101 knockdown by siRNA markedly inhibited ERK1/2 activation by SDF1α and Golgi-Gγ9 in PC3 cells, whereas p87 knockdown had no effect (Fig. 5, D and E). To study if p110γ and p101 were expressed at the GA, p110γ and p101 tagged with GFP or DsRed were transiently expressed together with the Golgi marker pmTurquoise2-Golgi in PC3 cells. As expected, p110γ and p101 were mainly expressed in the cytoplasm and extensively colocalized. Both p110γ and p101 were partially colocalized with the Golgi marker (Fig. S6). These data suggest that p110γ-p101 heterodimers, but not p110γ-p87 heterodimers, mediate ERK1/2 activation by Gβγ on the GA.

Knockout of Gγ9 and PI3Kγ reduces prostate cancer cell migration, invasion, and metastasis
We next used PC3 cells to determine the effect of Gγ9 and PI3Kγ knockout on cancer cell migration and invasion in vitro.
In the transwell migration assay, migration of PC3 cells lacking Gγ9 and p110γ in response to SDF1α stimulation was markedly inhibited as compared with control cells. In contrast, Gγ3   knockout PC3 cells migrated normally (Fig. 6A). PC3 cell migration was significantly enhanced after rapamycin induction in cells expressing Golgi-Gγ9. The Golgi-Gγ9-induced motility was completely blocked by the ERK1/2 pathway inhibitors U0126, PD98059, and GW5074, as well as the PI3Kγ inhibitor AS-604850 (Fig. 6B). Similar to migration, PC3 cell invasion in response to SDF1α stimulation was attenuated in Gγ9 and p110γ knockout cells, but not in Gγ3 knockout cells, as compared with control cells (Fig. 6C). Inducible recruitment of Gγ9 to the GA also enhanced PC3 cell invasion, which was blocked by MAPK and PI3Kγ inhibitors (Fig. 6D). Finally, we determined the effect of Gγ9 and PI3Kγ knockout on PC3 metastasis in vivo. Gγ9 and p110γ knockout PC3 cells expressing luciferase were injected into the left ventricle of athymic nude mice to allow the cells to disseminate to multiple organs, predominantly lumbar and vertebral bones. Whole-body bioluminescence imaging for luciferase activity and subsequent bioluminescence showed that the overall tumor sizes were dramatically smaller in the group of mice injected with Gγ9 and p110γ knockout cells after 3 weeks as compared with those in the mice injected with control cells (Fig. 6, E and F). These data demonstrate a crucial role played by Gγ9 and PI3Kγ in prostate tumor metastasis.

Discussion
In this study, we demonstrate that Gβγ translocation from the PM to the GA is essential for ERK1/2 activation by CXCR4 in three different cells. We have shown that Gβγ dimers that contain Gγ9 are uniquely efficient with respect to translocation to the GA, both in terms of translocation rate and translocation magnitude in DU145, PC3, and HEK293 cells. These results are highly complementary to previous results in other cell types (5,7,9,10,17). The function of Gβγ translocation to the GA in activating ERK1/2 is strongly supported by three series of experiments that demonstrate that knockout of Gγ9 abolishes ERK1/2 activation by SDF1α, chemically   induced translocation of Gγ9 to the GA constitutively activates ERK1/2, and ERK1/2 activation by inducible translocation of Golgi-Gγ9 is completely blocked by Golgi-localized GRK2ct and Golgi disruptors. It is worth noting that, as with Gγ9, recruitment of Gγ2 and Gγ3 on the GA also causes ERK1/2 activation, suggesting that different Gβγ combinations are able to activate ERK1/2 if they are expressed at the GA. However, future studies using Gγ9 mutants defective in GA translocation and defining the ERK1/2 activation signal at the GA may fully clarify the importance of Gβγ translocation in ERK1/ 2 activation. Since several GPCRs have been shown to promote Gβγ translocation to the GA in different cells (5)(6)(7)(8)(9)(10)42) and our data have shown that, in addition to CXCR4, α 2 -ARmediated ERK1/2 activation is also inhibited by Gγ9 knockout, the mechanism we describe here may be commonly used by many GPCRs to activate the ERK1/2 pathway (Fig. 6G).
Gβγ is capable of activating many downstream signaling molecules (2,3). We have demonstrated here that ERK1/2 activation by Golgi-localized Gβγ is mediated through PI3Kγ, specifically its heterodimer p110γ-p101. This became evident as specific pharmacological inhibition of PI3Kγ, CRISPR-Cas9-mediated knockout of catalytic subunit p110γ, and siRNA-mediated knockdown of regulatory subunit p101 abolished ERK1/2 activation by GPCR agonists and translocation of Golgi-Gγ9. This is also supported by our data showing that both p110γ and p101 are partially expressed at the GA and a previous study showing that PI3Ks may be expressed at the GA (43). PI3Kδ may also play a role in Gβγmediated ERK1/2 activation, because its inhibition attenuated ERK1/2 activation. Indeed, PI3Kγ and δ can be activated by GPCRs and PI3Kγ can be activated by Gβγ (31)(32)(33). Although a number of previous studies have established the function of Gβγ and PI3Kγ in GPCR-mediated ERK1/2 activation, these studies suggest that the crucial interaction may occur at the PM (18)(19)(20). For example, PI3Kγ was shown to be required for M2-muscarinic receptor-and Gβγ-mediated activation of ERK1/2, likely via a typical PM RTK cascade (20). In addition, Gγ3 regulates macrophage migration via activating PI3Kγ at the PM, whereas Gγ9 has minimal effect (10). These data suggest that different Gβγ dimers may activate PI3Kγ in distinct subcellular compartments to control different cellular processes, adding to the complexity of Gβγ-mediated signaling and functional regulation. Nevertheless, our data demonstrate a signal transduction pathway in which GPCR activation at the Gβγ activates MAPK via PI3Kγ at Golgi PM induces Gβγ translocation to the GA where it activates the PI3Kγ heterodimer p110γ-p101, leading to the activation of the MAPK pathway (Fig. 6G). These data also suggest a novel function of the GA as a signaling organizing compartment in MAPK activation by GPCRs, in which the GA provides a spatial station to compartmentalize the translocation of Gβγ, activation of PI3Kγ, and activation of ERK1/2 pathway (Fig. 6G).
Another important finding of the present study is the possible pathophysiological function of Gβγ translocation to the GA and subsequent ERK1/2 activation in prostate tumor progression. Over the past decade, many studies have demonstrated that, similar to RTKs, GPCRs at the PM are involved in the initiation and progression of many different cancer types and significant efforts are currently underway to develop GPCR-and G protein-based drugs for cancer (44,45). Given the importance of the ERK1/2 pathway in the progression of prostate cancer, the molecules involved in regulation of this pathway are thought to be appealing targets for prostate cancer therapeutics (34,35). Although multiple genetic mutations in RTKs, Ras, Raf, and MEK cause constitutive activation of the ERK1/2 pathway and drive many types of malignancies (46,47), patients with prostate cancer frequently do not have these oncogenic mutations. Therefore, extensive efforts are focused on the identification of regulators that control ERK1/2 activation in prostate cancer cells (48). We have shown that Gγ9 is a very strong activator of ERK1/2 at the GA. In addition, both Gγ9 (49) and CXCR4 (38) are highly   expressed in prostate cancer cells. As such, enhanced expression and exaggerated activation of GPCRs and G proteins (e.g., CXCR4 and Gγ9) may represent crucial mechanisms responsible for the enhanced activation of the oncogenic ERK1/2 pathway in prostate cancer. As we have demonstrated that knockout of Gγ9 and PI3Kγ markedly suppresses prostate cancer cell migration, invasion, and metastasis, these data, together with previous studies showing the roles of Gβγ in prostate cancer progression (50,51), imply that Golgi-localized Gβγ, as well as PI3Kγ, may be important targets for prostate cancer therapy.
Cell culture and transfection DU145, PC3, and HEK293 cells were purchased from American Type Culture Collection. DU145 and PC3 cells were cultured in complete Roswell Park Memorial Institute (RPMI) 1640 medium (Lonza) supplemented with 2 mM L-glutamine and 10% fetal bovine serum (FBS) (Atlanta Biologicals). HEK293 cells were cultured in Dulbecco's Modified Eagle's medium with 10% FBS. The transfection was carried out using Lipofectamine 3000 (Thermo Fisher Scientific).
siRNA-mediated depletion of p101 and p87 siRNA-mediated knockdown of p101 and p87 was carried out as described (57). The cells cultured on 6-well plates were transfected with siRNA at a concentration of 30 nM using Lipofectamine 3000 for 12 h. To study the effect of p101 and p87 knockdown on ERK1/2 activation by Golgi-Gγ9, cells were transfected with siRNA, together with Gβ1, FRB-Gγ9, and Golgi-FKBP (1 μg each). The cells were then split at a ratio of 1:2 and grown for additional 24 h before starvation and stimulation with SDF1α or rapamycin.

Confocal microscopy
To measure Gβγ translocation, cells were cultured on 25-mm coverslips for 24 h and then transfected with individual YFP-Gγ subunits, Gβ1, Gαi1, and pmTurquoise2-Golgi. Before imaging, the cells were starved for 48 h and then stimulated with SDF1α at 1 μg/ml. The cells were imaged for the YFP and cyan fluorescent protein fluorescence signals every 5 s using a time-lapse Leica DMi8 microscope. The translocation of Gβγ to the GA in response to SDF1α stimulation was quantified by measuring the increase of total YFP signal at the GA.
To measure Golgi-GRK2ct expression, cells were transfected with Golgi-GRK2ct and its mutant for 36 h and stained with GRK2 antibodies. To verify inducible translocation of Gβγ to the PM and the GA, cells were transiently transfected with venus-Gβ1 and FRB-Gγ9, together with either Golgi-FKBP or PM-FKBP (500 ng each) for 24 h and starved for 48 h before induction with rapamycin at 1 μM for 30 min. To measure the effect of Gα subunits on Gβγ translocation to the GA, cells were transfected with venus-Gβ1, mCherry-Gαi1, FRB-Gγ9, and Golgi-FKBP (500 ng each). To study Golgi fragmentation, cells were transfected with the Golgi marker YFP-GalT and then treated with different Golgi disruptors for 40 min. To study if p110γ and p101 were expressed at the GA, cells were transfected with DsRed-or GFP-tagged p110γ and p101, together with the Golgi marker pmTurquoise2-Golgi for 24 h. In these experiments, cells were fixed with 4% paraformaldehyde for 15 min. For antibody staining, cells were permeabilized with 0.25% Triton X-100 for 5 min and blocked with normal donkey serum for 30 min. The cells were sequentially stained with primary antibodies and secondary antibodies. The images were captured using a Zeiss LSM780 confocal microscope equipped with a 63× objective.

Measurement of ERK1/2 activation
Cells were cultured in 6-well dishes for 24 h and starved for 48 h before stimulation with SDF1α, UK14304, or rapamycin as indicated in the figure legends. After the medium was removed and the cells were washed twice with cold PBS, the cells were solubilized by the addition of 300 μl of 1X SDS gel-loading buffer. ERK1/2 activation was determined by measuring ERK1/2 phosphorylation by Western blotting as described (52,57).

Migration and invasion assays
Chemotactic migration of PC3 cells toward SDF1α was quantified using the Boyden migration chambers. Briefly, PC3 cells were suspended in serum-free RPMI1640 medium and 2 × 10 5 cells (200 μl) were subjected to transwell migration assays using SDF1α at 200 ng/ml for 48 h at 37 C. For invasion assays, the suspended cells (2 × 10 5 cells in 200 μl) were seeded in the top insert coated with diluted Matrigel solution. The migrated and invaded cells were measured by MTT assay and calculated as described (58).

Bioluminescent imaging of luciferase in animals
All animal studies were approved by the Institution Animal Care and Use Committee of Augusta University. CRISPR-Cas9-mediated knockout PC3 and control cells were cultured and transfected with the pQCXIB plasmid. The cells were selected with blasticidin S, and luciferase expression was confirmed by dual-luciferase reporter assays. Six-week-old male nude mice (3 groups, n = 10 for each group, Jackson Laboratories) were anaesthetized and PC3 cells expressing luciferase (1 × 10 5 cells in 100 μl) were injected via the left cardiac ventricle. The tumor size was measured by bioluminescent imaging of luciferase every week using an Ami Spectral Advanced Molecular Imager after intraperitoneal injection of D-luciferin (200 μl, 15 mg/ml). The data were analyzed by AMIView software and expressed as photon flux (photons/s/ cm 2 /sr).

Statistical analysis
Comparisons across groups were evaluated using one-way ANOVA, and p < 0.05 was considered as statistically significant. Data are expressed as the means ± SD.

Data availability
All data presented are available upon request from Guangyu Wu (guwu@augusta.edu).