Protein kinase C-theta (PKCθ) phosphorylates and inhibits the guanine exchange factor, GIV/Girdin

Abstract

Gα-interacting, vesicle-associated protein (GIV/Girdin) is a multidomain signal transducer that enhances PI3K-Akt signals downstream of both G-protein–coupled receptors and growth factor receptor tyrosine kinases during diverse biological processes and cancer metastasis. Mechanistically, GIV serves as a non-receptor guanine nucleotide exchange factor (GEF) that enhances PI3K signals by activating trimeric G proteins, Gαi1/2/3. Site-directed mutations in GIV’s GEF motif disrupt its ability to bind or activate Gi and abrogate PI3K-Akt signals; however, nothing is known about how GIV’s GEF function is regulated. Here we report that PKCθ, a novel protein kinase C, down-regulates GIV’s GEF function by phosphorylating Ser(S)1689 located within GIV’s GEF motif. We demonstrate that PKCθ specifically binds and phosphorylates GIV at S1689, and this phosphoevent abolishes GIV’s ability to bind and activate Gαi. HeLa cells stably expressing the phosphomimetic mutant of GIV, GIV-S1689→D, are phenotypically identical to those expressing the GEF-deficient F1685A mutant: Actin stress fibers are decreased and cell migration is inhibited whereas cell proliferation is triggered, and Akt (a.k.a. protein kinase B, PKB) activation is impaired downstream of both the lysophosphatidic acid receptor, a G-protein–coupled receptor, and the insulin receptor, a receptor tyrosine kinase. These findings indicate that phosphorylation of GIV by PKCθ inhibits GIV's GEF function and generates a unique negative feedback loop for downregulating the GIV-Gi axis of prometastatic signaling downstream of multiple ligand-activated receptors. This phosphoevent constitutes the only regulatory pathway described for terminating signaling by any of the growing family of nonreceptor GEFs that modulate G-protein activity.

Gα-interacting vesicle-associated protein (GIV/Girdin) is a large, multidomain protein that is required for growth factors [EGF (1, 2), insulin-like growth factor (IGF) (3), VEGF (4), and insulin (5–7)] to enhance PI3K-Akt signals and trigger cell migration (1). GIV also enhances Akt signaling downstream of G-protein–coupled receptors (GPCRs) [e.g., lysophosphatidic acid receptor (LPAR)] (6–8). GIV plays critical roles during diverse biological processes, including epithelial wound healing, macrophage chemotaxis, autophagy, neuronal development, tumor cell migration, vascular repair, maintenance of pluripotency of cancer stem cells, and tumor angiogenesis (reviewed in ref. 9), by enhancing Akt (a.k.a. protein kinase B, PKB) signals downstream of both growth factor receptor tyrosine kinases (RTKs) and GPCRs. GIV performs these functions via multiple mechanisms—it activates trimeric G proteins Gαi1/2/3 via its guanine nucleotide exchange factor (GEF) motif (amino acids 1676–1696) (8), binds and reorganizes the actin cytoskeleton via its actin-binding domain (1), modulates multiple growth factor signaling pathways by directly binding to RTKs (2, 5) and enhances the Akt pathway via synergistic activation of classes 1A and 1B PI3-kinases (11). The C-terminally located GEF motif has emerged as a key modulator of GIV’s functions. Cells expressing a GEF-deficient F1685A (FA) mutant display a phenotype strikingly different from those that express the wild-type (WT) protein in that they fail to activate Gi (6) or couple Gi to ligand-activated RTKs (2), maximally enhance PI3K-Akt signaling in response to growth factors or ligands for GPCRs (2, 6, 8, 12), generate actin stress fibers (6), trigger cell migration (2), or arrest autophagy (5). Instead, they enhance the MAPK/ERK pathway (2), display increased cortical actin (6), and trigger mitosis (2).

The clinical significance of GIV’s GEF motif is emphasized by the observation that its expression is dysregulated during cancer progression (2). Early during oncogenesis GIV’s C-terminal GEF motif is alternatively spliced such that its exclusion triggers mitosis and favors early tumor initiation and growth, whereas its inclusion during oncogenic progression triggers cell migration and favors late tumor invasion. On the basis of these observations, GIV’s GEF function has been likened to a “rheostat” whose abundance determines the rapidity of G-protein cycling and the resultant signal enhancement that drives prometastatic phenotypes (9). Although the GEF motif of GIV has emerged as a major platform for signal enhancement, the mechanisms by which cells may selectively down-regulate this platform have remained elusive. Given the importance of GIV’s GEF function in so many biological processes and its clinical significance (9), it is essential to identify the pathways that specifically inhibit this nonreceptor GEF. Here we describe how GIV’s GEF activity is modulated by phosphorylation within the GEF motif and how that affects PI3K-Akt signaling in cells.

Results

Identification of Ser¹⁶⁸⁹ as a Key Residue That Regulates GIV’s GEF Activity.

We recently reported that Ser(S)¹⁶⁸⁹ and Asn¹⁶⁹⁰ within the GEF motif of GIV are required to bind and activate Gαi3 efficiently (6). By phylogenetic analysis we noted that GIV’s GEF motif is conserved in vertebrates from fish to mammals, but Ser¹⁶⁸⁹ evolved later and emerged for the first time in birds (Fig. 1A and Fig. S1). On the basis of the predicted position and orientation of Ser¹⁶⁸⁹ within the GIV-Gαi3 interface (Fig. 1B) we hypothesized that phosphorylation at Ser¹⁶⁸⁹ would affect GIV’s ability to bind and activate Gαi. We first generated nonphosphorylatable Ser¹⁶⁸⁹ to Ala [S1689→A (SA)] and phosphomimetic Ser¹⁶⁸⁹ to Asp [S1689→D (SD)] GIV mutants and tested their ability to bind Gαi3 in coimmunoprecipitation and GST-pulldown assays. When Gαi3 was immunoprecipitated from lysates of Cos7 cells (SV-40 transformed African green monkey kidney fibroblasts) coexpressing GIV proteins FLAG-tagged Gαi3, GIV-WT but not GIV-SD coimmunoprecipitated with Gαi3, and GIV-FA interacted weakly as seen previously (6) (Fig. 1C). In vitro binding assays using purified GST-tagged Gαi3 and His-tagged GIV-C-terminus (CT) WT or mutant proteins (amino acids 1660–1870) containing the GEF motif showed that the phosphomimetic SD mutant failed to bind GST-Gαi3, whereas binding of the nonphosphorylatable SA mutant remained unchanged (Fig. 1D). Taken together, these results indicate that phosphorylation of GIV at Ser¹⁶⁸⁹ disrupts GIV-Gαi3 interaction both in vitro and in cells.

Fig. 1.

GIV cannot bind or activate Gαi3 when Ser¹⁶⁸⁹ in its GEF motif is mutated to Asp. (A) Diagram showing the phylogenetically conserved sequence of the GEF motif of GIV. Sequences obtained from the accession numbers (in brackets) were aligned using Clustal W. The conserved Phe¹⁶⁸⁵ (F1685, green), previously identified as the key residue that establishes a hydrophobic interaction with Trp²¹¹ and Phe²¹⁵ of Gαi3 (6), and the predicted phosphoregulated Ser¹⁶⁸⁹ (yellow) are shown. Conserved residues are shaded in black and similar residues in gray. (B) Homology model of Gαi3 in complex with GIV 1678–1689 generated (6) using the structure of the synthetic peptide KB-752 bound to Gαi1 [Protein Data Bank ID 1Y3A] as a template (32). Yellow, Gαi3 subunit; red, GIV’s GEF motif; white spheres, hydrogen; red spheres, oxygen; blue sphere, nitrogen. Ser¹⁶⁸⁹ lies in close proximity to the Gαi-GIV interface. (C) GIV fails to coimmunoprecipitate with Gαi3 when Ser¹⁶⁸⁹ in GIV is mutated to Asp. Immunoprecipitation was carried out with anti-FLAG mAb on equal aliquots of Cos7 lysates coexpressing FLAG-Gαi3 and GIV-WT, SD, or FA followed by incubation with protein-G beads. Lysates (Lower) and bound immune complexes (Upper) were analyzed for FLAG (Gαi3-FLAG), GIV, and Gβ by immunoblotting (IB). Binding of GIV-SD (lane 3) or FA (lane 4) mutants to Gαi3-FLAG is dramatically reduced compared with that of GIV-WT (lane 2), whereas Gβ binds equally under all conditions. (D) Mutation of Ser¹⁶⁸⁹ in GIV to Asp [S1689→D (SD)] virtually abolishes binding of His-GIV-CT to GST-Gαi3. Equal aliquots (1.2 μg) of WT (lanes 1 and 2), SD (lane 3), or SA (lane 4) His-GIV-CT proteins were incubated with 5 μg GST or GST-Gαi3 preloaded with GDP and immobilized on glutathione beads. Bound proteins (Top) were analyzed by IB for His (GIV-CT). Equal loading of GST and His-GIV-CT proteins was confirmed by Ponceau S staining (Middle) and by immunoblotting for His (inputs, Lower). (E) Activation of His-Gαi3 by His-GIV-CT is virtually abolished in the presence of the GIV-SD (open circles) mutant compared with GIV-WT (solid circles). The steady-state GTPase activity of His-Gαi3 (50 nM) was determined in the presence of purified His-GIV-CT-WT and SD. Gαi3 activation is expressed as percent of the steady-state GTPase activity of Gαi3 alone. Results are shown as mean ± SEM of three experiments.

Because binding of GIV’s GEF motif to Gαi is required for GIV to activate Gαi (12), we predicted that the S→D mutation would inhibit GIV’s ability to activate Gαi3. When we compared the ability of GIV-CT-WT and the SD mutant to enhance the steady-state GTPase activity of Gαi3, we found that GIV-dependent activation of Gαi3 with the SD mutant protein was completely abolished (Fig. 1E). These findings identify Ser¹⁶⁸⁹ as a key residue within GIV’s GEF motif whose phosphorylation uncouples GIV from Gαi and thereby could potentially reversibly modulate GIV’s GEF function and G-protein activity.

Protein Kinase C-Theta Phosphorylates GIV at Ser¹⁶⁸⁹.

Because the evolutionarily conserved sequence flanking Ser¹⁶⁸⁹ (Fig. 1A and Fig. S1) resembles a consensus motif ([pS/T]-X-R/K) known to be targeted by second-messenger-dependent protein kinase Cs (PKCs) (13), we carried out in vitro kinase assays using purified rat brain PKC and recombinant His-GIV-CT-WT and the SA mutant. Two other second-messenger-dependent AGC kinases, PKA and PKB (Akt), were included as additional controls. We found that all three kinases could phosphorylate GIV-CT in vitro. PKA and PKB could phosphorylate GIV-CT-WT and SA to an equal extent (Fig. 2A), indicating that they phosphorylate GIV-CT at a site or sites other than Ser¹⁶⁸⁹. In the case of PKC, the degree of phosphorylation dropped dramatically in the SA mutant (Fig. 2A and Fig. S2A), suggesting that PKC phosphorylates GIV-CT at more than one site and that Ser¹⁶⁸⁹ is one of the major sites. These findings demonstrate that among the AGC kinases tested, PKC is the specific kinase that targets Ser¹⁶⁸⁹. To further pinpoint the PKC isoform(s) responsible for phosphorylation of GIV at Ser¹⁶⁸⁹ we carried out similar in vitro kinase assays, using recombinant forms of all three classes (conventional, novel, and atypical) of PKC isoforms. Under the conditions tested, there was little or no detectable phosphorylation with some conventional (βI and γ), novel (ε, δ, and η), and atypical (λ) isoforms of PKC (Fig. 2B and Fig. S2 B and C). By contrast, other conventional (α, βII) and atypical (ζ) isoforms efficiently phosphorylated both GIV-CT-WT and SA to an equal extent, indicating that each of these isoforms phosphorylates GIV-CT at residues other than Ser¹⁶⁸⁹. Only the novel θ isoform could phosphorylate GIV-CT-WT but not SA (Fig. 2B), demonstrating that PKCθ is the specific PKC isoform responsible for phosphorylating GIV at Ser¹⁶⁸⁹.

Fig. 2.

GIV is phosphorylated on Ser¹⁶⁸⁹ by PKCθ. (A) In vitro kinase assays were carried out with recombinant PKA, PKB, and rat brain PKC and equal aliquots (2 μg) of either WT (lane 1) or a nonphosphorylatable SA His-GIV-CT mutant (lane 2). Phosphorylated proteins were detected by autoradiography (Upper three panels), and equal loading of His-GIV-CT substrates was confirmed by Coomassie Blue staining (Lower panel). Although PKA, PKB, and PKC phosphorylate GIV-CT in vitro, Ser¹⁶⁸⁹ is phosphorylated exclusively by PKC. (B) In vitro kinase assays were carried out with 100 ng of novel PKC isoforms and analyzed by autoradiography. The theta (θ) isoform of PKC specifically phosphorylates GIV at Ser¹⁶⁸⁹. (C) PKCθ-mediated phosphorylation of GIV at Ser¹⁶⁸⁹ abolishes the ability of His-GIV-CT to bind GST-Gαi3. Equal aliquots (2 μg) of WT or SA His-GIV-CT were either mock treated (lanes 1, 2, and 4) or phosphorylated with 100 ng PKCθ (lanes 3 and 5) before their use in GST pulldown assays with 10 µg control GST (lane 1) or 5 µg GST-Gαi3 (lanes 2–5) preloaded with GDP and immobilized on glutathione beads. Bound proteins (Upper) and inputs (Lower) were analyzed by immunoblotting (IB) for His (His-GIV-CT).

To determine whether phosphorylation of GIV-CT at Ser¹⁶⁸⁹ disrupts Gαi-GIV interaction we carried out pulldown assays with GST-Gαi3 and in vitro phosphorylated His-GIV-CT. Binding of GIV-CT-WT to Gαi3 was virtually abolished after phosphorylation with either rat brain PKC (Fig. S2D) or recombinant PKCθ (Fig. 2C). This disruption is solely attributable to phosphorylation at Ser¹⁶⁸⁹ by PKCθ because the nonphosphorylatable SA mutant bound equally before and after phosphorylation (Fig. 2C). Furthermore, binding of full-length, endogenous GIV to GST-Gαi3 was reduced by ∼50% when HeLa cells were stimulated with 12-myristate-13-acetate (PMA), a synthetic phorbol ester that activates PKCθ, as well as other PKCs (14) (Fig. S2E). These results implicate PKCθ as the kinase responsible for phosphorylating GIV’s GEF motif at Ser¹⁶⁸⁹ and inhibiting its ability to bind Gαi.

PKCθ Phosphorylates GIV at Ser¹⁶⁸⁹ Downstream of RTKs and GPCRs.

To investigate whether phosphorylation of GIV at Ser¹⁶⁸⁹ occurs in cells and whether this phosphoevent requires PKCθ, we generated a phosphospecific-antibody to selectively detect GIV only when it is phosphorylated at Ser¹⁶⁸⁹ (Fig. S3). We found that phosphorylation of GIV at Ser¹⁶⁸⁹ was undetectable in serum-starved HeLa cells, but readily detectable when cells were stimulated with 10% FBS (vol/vol), growth factors (EGF and insulin), LPA (ligand for LPAR), or PMA (Fig. 3). When cells were depleted of endogenous PKCθ by siRNA, phosphorylation of GIV at Ser¹⁶⁸⁹ was virtually abolished in cells grown in 10% (vol/vol) FBS and significantly reduced in cells responding to a variety of ligands (Fig. 3). Reduced GIV phosphorylation observed in cells responding to acute ligand stimulation could be due to residual endogenous PKCθ after incomplete depletion, activation of an alternative kinase(s) that can also phosphorylate GIV at Ser¹⁶⁸⁹, or acute inhibition of the phosphatase(s) that dephosphorylates this phosphoSer. Whichever is the case and regardless of the receptor stimulated, these results demonstrate that a variety of stimuli can trigger phosphorylation of GIV at Ser¹⁶⁸⁹ and that PKCθ is a major kinase that drives this phosphoevent in response to a variety of ligands.

Fig. 3.

PKCθ is required for phosphorylation of GIV at Ser¹⁶⁸⁹ after ligand stimulation. HeLa cells treated with scrambled (Scr) or PKCθ siRNA were serum starved (lanes 1 and 2) and stimulated with serum (lanes 3 and 4), PMA (10 min; lanes 5 and 6), insulin (5 min; lanes 7 and 8), EGF (5 min; lanes 9 and 10), or LPA (20 min; lanes 11 and 12) before lysis. (Upper) Equal aliquots of lysates were analyzed for phospho-Ser1689 GIV (pS1689 GIV), total GIV (GIV-CT Ab), PKCθ, and tubulin by quantitative immunoblotting (IB). (Lower) Bar graphs show the ratio of pS1689-GIV to total GIV, normalized to control siRNA-treated HeLa cells within each condition. Results shown are representative of five independent experiments. The concentration and duration of stimulation in each case are based on the peak activation of Akt via GIV’s GEF motif observed previously (2, 5, 6, 8, 12).

PKCθ Interacts and Colocalizes with GIV in Cells.

Next, we asked whether PKCθ binds GIV and, if so, whether this interaction is regulated by ligand stimulation. We found that endogenous GIV coimmunoprecipitated with YFP-tagged PKCθ (and vice versa) exclusively after cells were stimulated with growth factors (EGF and insulin) or with PMA (Fig. 4A and Fig. S4 A and B). These results demonstrate that GIV and PKCθ interact in cells and that ligand stimulation is required to trigger this interaction. Moreover, GIV partially colocalized with PKCθ at the plasma membrane (PM) in ligand-stimulated HeLa cells (Fig. 4B), suggesting that GIV binds PKCθ at the PM, where activated GIV can be down-regulated by PKCθ.

Fig. 4.

PKCθ binds and colocalizes with GIV at the cell periphery after ligand stimulation. (A) GIV coimmunoprecipitates with YFP-PKCθ from lysates of cells stimulated with PMA (lane 3) or insulin (lane 4), but not from serum-starved cells (lane 2). Cos7 cells expressing YFP-PKCθ were serum starved (0% FBS, 16 h) and then stimulated with either PMA (200 nM, 10 min; lane 3) or insulin (100 nM, 5 min; lane 4) before lysis. Immunoprecipitation was carried out with anti-GFP mAb (lanes 2–4) or control IgG (lane 1) from equal aliquots of lysates (Lower), followed by incubation with protein-G agarose beads. Bound immune complexes (Upper) were analyzed for GIV (CT Ab) and GFP (YFP-PKCθ) by immunoblotting (IB). (B) Endogenous PKCθ and GIV partially colocalize at the cell periphery after ligand stimulation. In starved HeLa cells endogenous PKCθ is predominantly cytosolic, whereas GIV is predominantly located at the perinuclear Golgi region as demonstrated previously (7, 33). After stimulation with EGF (50 nM) or insulin (100 nM), small but significant pools of GIV and PKCθ were seen in patches at the PM (arrowheads, black and white single panels) by confocal microscopy. Yellow pixels in the merged images at the PM denote patchy colocalization (arrowheads) of GIV (red) and PKCθ (green) in stimulated cells, but not in starved cells. (Scale bar, 10 μm.)

Phosphorylation of GIV at Ser¹⁶⁸⁹ Inhibits Akt Activation and Cell Migration, but Triggers Proliferation.

We previously demonstrated that activation of Gαi3 via GIV’s GEF motif downstream of RTKs and GPCRs is required for enhancement of Akt phosphorylation, rearrangement of the cytoskeleton and generation of actin stress fibers, and efficient cell migration (7). We reasoned that these responses might be inhibited or reduced in cells expressing the SD mutant of GIV. To test this, we generated HeLa cell lines stably expressing siRNA-resistant GIV-WT, GIV-FA, or GIV-SD as described previously (6) and analyzed their phenotype after depletion of endogenous GIV by siRNA. We found that HeLa cell lines stably expressing GIV-SD (HeLa-GIV-SD) (Fig. S5A) failed to enhance Akt phosphorylation in response to either insulin (Fig. 5A and Fig. S5B) or LPA (Fig. 5B and Fig. S5C), much like HeLa-GIV-FA cells (6). Similarly, the arrangement of the actin cytoskeleton was altered (Fig. 5C), and cell migration was impaired (Fig. 5D) in HeLa-GIV-SD and in HeLa-GIV-FA cells versus HeLa-GIV-WT cells. From these results we conclude that Akt activation, actin cytoskeleton remodeling, and cell migration are highly sensitive to phosphorylation of Ser1689 in GIV’s GEF motif by PKCθ. Because the poorly motile HeLa-GIV-FA cells proliferate faster than HeLa-GIV-WT cells in the absence of an intact GEF motif (2), we predicted that the HeLa-GIV-SD cells might also have this phenotype. We found that the mitotic index, determined by levels of phospho-histone H3 (15), was consistently higher in HeLa-GIV-SD and HeLa-GIV-FA cells than in HeLa-GIV-WT cells (Fig. 5E and Fig. S5D). An identical trend was seen when proliferation was assessed by incorporation of bromodeoxyuridine (BrdU) into cellular DNA (16) (Fig. 5F and Fig. S5E). These results demonstrate that phosphorylation of GIV at Ser¹⁶⁸⁹ within the GEF motif can inhibit all previously defined functions of GIV’s GEF activity. On the basis of these findings we propose that targeted phosphorylation of GIV’s GEF motif at Ser¹⁶⁸⁹ by PKCθ serves as a major negative feedback loop that terminates the GIV-Gi axis of signal enhancement downstream of multiple ligand-activated receptors (Fig. 6).

Fig. 5.

Phosphorylation of GIV at Ser¹⁶⁸⁹ inhibits Akt activation, actin stress fiber formation, and cell migration and enhances cell proliferation. (A and B) HeLa cell lines stably expressing siRNA-resistant GIV-WT-FLAG (HeLa GIV-WT), GIV-FA-FLAG (HeLa GIV-FA), or GIV-SD-FLAG (HeLa GIV-SD) were treated with control (Scr) or GIV siRNA for 36 h, serum starved [0.2% (vol/vol) FBS, 16 h], and then stimulated with insulin for 5 min (A) or LPA for 20 min (B) before lysis. Whole-cell lysates were analyzed for phospho Akt (pAkt) (S473), total Akt (tAkt), and tubulin by quantitative immunoblotting (immunoblots shown in Fig. S5 B and C). Bar graphs display the ratio of pAkt/tAkt (y axis), normalized to starved control HeLa cells. Results are shown as mean ± SD of three independent experiments. (C) All three HeLa-GIV stable cell lines were depleted of endogenous GIV, fixed, and costained with phalloidin-Texas Red (F-actin, red) and DAPI/DNA (blue) and visualized by fluorescence. (Scale bar, 10 μm.) (D) HeLa cell lines were treated with scrambled (Scr) or GIV siRNA as indicated, grown to 100% confluency as monolayers, scratch wounded, and assessed for wound closure by serial imaging of the wound for 12 h (see SI Experimental Procedures). Bar graphs show quantification of the percentage of wound area closed by 12 h, which is expressed as migration index (y axis). (E and F) HeLa cells grown on coverslips were fixed and costained with DAPI/DNA (blue) and either phospho-Histone H3 (E) or anti-BrdU mAb (F). Bar graphs display the proliferation rate, i.e., expressed as mitotic index (y axis) derived by quantifying percentage of phospho-Histone- or BrdU-positive cells (see SI Experimental Procedures). Images of representative fields are shown in Fig. S5 D and E. Results are shown as mean ± SD of 8–12 fields. P values: *P < 0.05; **P < 0.01; ***P < 0.001.

Fig. 6.

Proposed model for how GIV’s GEF function is regulated by PKCθ. Upon ligand stimulation of RTKs and GPCRs GIV is recruited to the PM (2, 7), where GIV’s GEF motif activates Gi and releases free Gβγ subunits, which in turn activate Akt via class 1 PI3Ks (6). Gβγ is also known to activate PLC (18), as can multiple other pathways downstream of RTKs and GPCRs (dashed arrows). Once activated, PLC cleaves phosphatidylinositol 4,5-bisphosphate (PIP2) into inositol triphosphate (IP3) and diacylglycerol (DAG) (17). The latter is a potent stimulus for recruitment and activation of PKCθ. Activated PKCθ binds and phosphorylates GIV at Ser¹⁶⁸⁹. This phosphoevent inhibits GIV’s ability to bind and activate Gi, thereby down-regulating GIV-Gi-dependent cellular processes. Thus, PKCθ-dependent phosphorylation of GIV serves as a negative feedback loop (dashed line) that terminates the GIV-Gi pathway of signaling.

Discussion

The major finding of this work is the identification of PKCθ as the specific second-messenger-dependent kinase that phosphorylates and inhibits GIV’s GEF activity, thereby terminating GIV’s ability to activate Gi downstream of RTKs and GPCRs. We previously demonstrated that upon ligand stimulation of cell surface receptors GIV is translocated to the PM where it colocalizes with receptors and Gαi in multiprotein complexes, activates Gi, and enhances PI3K-Akt signals and other pathways via G-protein intermediates (2). Here we provide evidence that supports the following model (Fig. 6): Ligand stimulation of either RTKs or GPCRs leads to PM recruitment and activation of PKCθ, likely via activation of phospholipase C (PLC) (17). There are many pathways by which PLCs could be activated; because Gβγ subunits of trimeric Gi proteins are known to activate PLCβ2 (18), it is plausible that activation of Gi and release of “free” Gβγ by GIV’s GEF motif could also enhance the PLC-PKCθ pathway. Alternatively other pathways that activate PLC could also activate PKCθ, e.g., direct activation of PLCβ1 by Gq-coupled GPCRs (19) or indirect enhancement of PLCγ activity downstream of RTKs via GIV’s GEF motif (2). Regardless of the mechanism(s) involved, our results demonstrate that activated PKCθ forms complexes with GIV, presumably at the cell periphery, phosphorylates GIV at Ser¹⁶⁸⁹, inhibits GIV’s GEF function, and thereby terminates GIV’s ability to further activate Gi proteins in the vicinity of activated receptors. The striking GEF-deficient phenotype of cells expressing a constitutively phosphorylated (S¹⁶⁸⁹D) GIV mutant establishes the PKCθ-GIV pathway as a major negative feedback loop that attenuates the previously defined receptor-GIV-Gi-PI3K-Akt signaling cascade. In the case of receptor GEFs, i.e., GPCRs, their signaling is attenuated when they are uncoupled from G proteins and thereby desensitized by G-protein receptor kinases (GRKs) and second-messenger-dependent kinases, i.e. PKC (20–22) and PKA (23). Although it seems essential that like GPCRs, nonreceptor GEFs must be down-regulated, the mechanisms by which this is achieved have remained elusive. The current work describes a unique regulatory phosphorylation for GIV's GEF function, and defines a mechanism for dynamic and reversible regulation of GEF function of any member in the growing list of nonreceptor GEFs (24). Because GIV’s GEF motif is flanked by numerous Ser/Thrs, and we show here that GIV-CT is a substrate for multiple other kinases (PKA, PKB, and multiple isoforms of PKC), we speculate that other phosphomodifications might differentially modulate GIV’s GEF activity and further fine-tune GIV-dependent signaling.

Noteworthy, GIV’s GEF motif evolved in fish, whereas the conserved Ser¹⁶⁸⁹ evolved later in birds, coincided with the fusion of the two independently evolved N and C termini into one GIV molecule as seen in mammals (9). Thus, the consensus for inhibitory phosphorylation by PKCθ in GIV’s GEF motif coevolved with the addition of GEF function to N-terminal functions. Calnuc (nucleobindin 1) and NUCB2 (nucleobindin 2), two recently added members of the GIV family of nonreceptor GEFs (25), which also activate Gαi via a structurally conserved Gα-binding and activating (GBA) motif, also contain a conserved Thr at an identical position within a PKC substrate motif (Thr³²⁰ in Calnuc and Thr³²² in NUCB2). Based on the fact that GBA motifs [Ψ-(S/T)-(Φ/Ψ)-X-(D/E)-F-Ψ] (25) are immediately followed by PKC substrate sites [X-X-(pS/T)-X-R/K] in all three known members of this family we predict that their GEF activities might be similarly regulated by PKCs.

We previously showed that activation of Gi by GIV’s GEF motif modulates insulin and EGF receptor signaling by specifically enhancing some pathways (e.g., PI3K-Akt signals) while suppressing others (e.g., MAPK/ERK signals) (2, 5). In the case of the PI3K-Akt signaling pathway, GIV enhances it by two synergistic but independent mechanisms: (i) activation of Gαi via GIV’s GEF motif releases free Gβγ subunits that bind and activate class 1B PI3-kinases (1, 3, 4, 6) and (ii) phosphorylation of GIV by multiple RTKs and non-RTKs at tyrosines 1764 and 1798 that directly bind and activate class 1A PI3-kinases (11). The latter is down-regulated by SH2 domain-containing tyrosine phoaphatase-1 (SHP-1), a protein tyrosine phosphatase that dephosphorylates GIV (26). Here we demonstrate that inhibitory phosphorylation of GIV’s GEF motif by PKCθ at Ser¹⁶⁸⁹ establishes another negative feedback loop that down-regulates the receptor-GIV-PI3K (class 1B) axis of signaling. Whether Ser¹⁶⁸⁹ phosphoregulation by PKCθ and dephosphorylation of tyrosines^1764/1798 by SHP-1 occur simultaneously or sequentially to regulate GIV signaling remains to be established. Because PKCθ is known to phosphorylate and inhibit SHP-1 (27), it is possible that the antagonistic relationship between SHP-1 and PKCθ adds other layers of complex regulation to the GIV-PI3K-Akt axis of signaling. Our finding that PKCθ inhibits GIV-dependent Akt activation is in keeping with and adds to the other known mechanisms by which PKCθ down-regulates the Akt pathway (28–30). Because GIV’s GEF function plays a critical role in selectively enhancing or suppressing multiple other receptor signaling pathways (2, 5) and in regulating multiple cellular processes (9), we speculate that the hitherto unknown functional significance of the ubiquitously expressed PKCθ gene (31) is to antagonize these established roles of GIV’s GEF function. In the case of cancers, our finding that PKCθ antagonizes GIV-GEF-dependent prometastatic signaling, together with the fact that PKCθ transcripts are undetectable in multiple human cancer cells (31), suggest that absence of this key negative regulator might account in part for the unrestricted GIV-GEF signaling we observe during cancer progression (9).

In conclusion, we have defined a molecular mechanism that regulates the GEF activity of the nonreceptor GEF, GIV. Mechanistic insights gained exemplify a paradigm for how activation of PKCs downstream of multiple receptors can negatively regulate the activation of Gi and how that affects GIV-Gi-dependent signaling in a variety of tissues/organs during physiology and in disease.

Experimental Procedures

A detailed description of all of the experimental procedures can be found in SI Experimental Procedures.

In Vitro GST-Pulldown Assays, Immunoprecipitation, and Quantitative Immunoblotting.

Protein–protein interactions were determined by pulldown and immunoprecipitation assays as described previously (2, 5, 11, 12).

Generation of Stable Cell Lines.

HeLa cell lines stably expressing p3xFLAG-CMV-14-GIV (GIV-3xFLAG) WT, FA, or SD mutants were generated as previously described (5, 11, 12). Two different clones were investigated for each mutant and assay. GIV-3xFLAG expression was approximately two times endogenous levels.

In Vitro Kinase and Cellular Phosphorylation Assays.

These assays were carried out using purified His-GIV-CT and commercially obtained purified kinases. Reactions were initiated by adding ATP and carried out at 30 °C in 30 μL kinase buffer [20 mM Tris⋅HCl (pH 7.5), 2 mM EDTA, 10 mM MgCl₂, 1 mM DTT] for 30–60 min. For in vivo phosphorylation assays on endogenous GIV, lysates of serum-starved (0% FBS, 6 h) HeLa cells stimulated with various ligands were analyzed for phospho-Ser1689 GIV by immunoblotting.

Steady-State GTPase Assays.

These assays were performed as described previously (5, 6, 8, 12). The His-Gαi3 protein used in this work was >95% functional on the basis of radiolabeled GTPγS binding and trypsin protection assays (12).

Data Analysis and Other Methods.

All experiments were repeated at least three times, and results were presented either as one representative experiment or as average ± SD or SEM. Statistical significance was assessed with Student’s t test: *P < 0.05; **P < 0.01; ***P < 0.001. Protein structure analysis and visualization were performed using ICM Browser Pro software (Molsoft).