The PDZ Protein GIPC Regulates Trafficking of the LPA1 Receptor from APPL Signaling Endosomes and Attenuates the Cell’s Response to LPA

Lysophosphatidic acid (LPA) mediates diverse cellular responses through the activation of at least six LPA receptors – LPA1–6, but the interacting proteins and signaling pathways that mediate the specificity of these receptors are largely unknown. We noticed that LPA1 contains a PDZ binding motif (SVV) identical to that present in two other proteins that interact with the PDZ protein GIPC. GIPC is involved in endocytic trafficking of several receptors including TrkA, VEGFR2, lutropin and dopamine D2 receptors. Here we show that GIPC binds directly to the PDZ binding motif of LPA1 but not that of other LPA receptors. LPA1 colocalizes and coimmunoprecipitates with GIPC and its binding partner APPL, an activator of Akt signaling found on APPL signaling endosomes. GIPC depletion by siRNA disturbed trafficking of LPA1 to EEA1 early endosomes and promoted LPA1 mediated Akt signaling, cell proliferation, and cell motility. We propose that GIPC binds LPA1 and promotes its trafficking from APPL-containing signaling endosomes to EEA1 early endosomes and thus attenuates LPA-mediated Akt signaling from APPL endosomes.


Introduction
Lysophosphatidic acid (LPA) mediates diverse biological effects including cell migration, differentiation, proliferation and survival [1,2]. LPA induces these effects by binding to, and activating at least six different G protein coupled receptors (GPCRs), termed LPA 1 through LPA 6 [1][2][3], which are differentially expressed in different tissues and have distinct effects in animal models [1,2]. These receptors are coupled to three classes of heterotrimeric G proteins, G q/11 , G i/o and G 12/13 , which mediate cellular responses to LPA [1,2].
LPA receptors 1-3 are the most studied and share high sequence homology (,55% overall sequence identity) except for their carboxy-terminus (CT) [3,4]. LPA 1 and LPA 2 but not LPA 3 contain the Class I PDZ binding motif sequence X-(S/T)-X-(V/I/ L)-COOH (where X is any amino acid) at the extreme CT [3]. LPA 2 CT, but not LPA 1 or LPA 3 , interacts with the PDZ domain proteins NHERF2 and MAGI-3 which couple LPA 2 to PLC-b3, RhoA and Erk signaling [3], demonstrating that the CT can couple LPA receptors to specific signaling pathways and thereby confer the specificity of the responses to each receptor [3,4].
We noticed that LPA 1 has a PDZ binding motif (SVV) identical to that present in two other proteins, semaphorin family member SemF and the melanosomal membrane protein GP75 [5,6], which interact with the PDZ protein GIPC [7]. Like LPA 1 , GIPC plays a key role in cell motility as GIPC (a.k.a. Synectin) knock out mice have defects in endothelial cell migration and angiogenesis [8,9].
We therefore wondered if GIPC might interact with the PDZ binding motif of LPA 1 to regulate its activity.
GIPC (GAIP-interacting protein, C terminus) was originally identified based on its ability to bind to the RGS (regulator of G protein signaling) protein GAIP (RGS19), a GTPase activating protein (GAP) for heterotrimeric G proteins [7]. We subsequently found that GIPC binds to the TrkA nerve growth factor receptor [10][11] and is required for efficient endocytosis and trafficking of TrkA from peripheral (APPL) signaling endosomes to juxtanuclear (EEA1) endosomes [11]. GIPC accomplishes this in part by binding to the actin based molecular motor myosin VI (Myo6) [12] and in part by binding to APPL [11,13], a Rab5 effector protein found on a subpopulation of peripheral endosomes. APPL is required for recruitment of GIPC to endosomes, and regulates key events in signal transduction from endosomes [14][15][16]. Additional studies demonstrated that GIPC also binds to the receptor tyrosine kinase VEGFR2 [17] as well as to G protein coupled receptors (GPCRs) such as the lutropin (hLHR) [18] and dopamine D2 (D2R) receptors [19] and promotes their endocytic trafficking. Previous studies of LPA 1 trafficking indicate that LPA 1 is taken up by endocytosis in clathrin coated pits, traffics through Rab5 endosomes, and recycles back to the cell surface [20][21][22]. Thus, we reasoned that interaction between GIPC and LPA 1 might also affect trafficking of LPA 1 .
Here we show that GIPC directly binds to the PDZ binding motif of LPA 1, forms a complex with LPA 1 and APPL, and promotes LPA 1 trafficking from APPL signaling endosomes to early endosomes, resulting in downregulation of LPA 1 induced Akt signaling and cell proliferation.

Immunoblotting
Proteins separated by SDS-PAGE were transferred to PVDF membranes (Millipore, Billerica, MA). After blocking with PBS containing 5% nonfat milk, membranes were incubated with primary antibodies at room temperature (1 h) or at 4uC (overnight), followed by incubation (1 h) at room temperature with goat anti-rabbit Alexa Fluor 680 F(ab') 2 (Molecular Probes) and goat anti-mouse IRDye 800 F(ab') 2 (Rockland). Infrared imaging with two-color detection and quantification of Western blots was performed according to the manufacturer's protocols using the Odyssey Infrared Imaging System (LiCor Biosciences, Lincoln, NE).

RNA Interference
Knockdown in HEK cells was achieved using a duplex siRNA targeting human GIPC1 (sense sequence 5-AGAGGUGGAA-GUAUUCA-AGdT-dT) purchased from Dharmacon Inc., (Chicago, IL). A negative control siRNA (Silencer #1) was purchased from Ambion (Austin, TX). Transfection of HEK-293 cells was performed using Oligofectamine according to the manufacturer's protocol (Invitrogen) with 50 nM siRNA, 0.8 mg/ml siRNA to lipid ratio, and a cell density of , 100 cells/mm 2 surface area.

Protein Purification and In Vitro Binding Assays
GST, GST-GIPC, GST-mouse LPA 1 tail (aa 311-364), GSTmouse LPA 2 tail (aa 305-348) and mutants were expressed in E. coli and purified on glutathione Sepharose 4B (Amersham). For the in vitro binding assay 10 mg GST or GST fusion protein prebound to glutathione Sepharose beads were incubated with [ 35 S]Met (GE Healthcare)-labeled GIPC-PDZ domain prepared using the TnT Quick Coupled Transcription/Translation System (Promega, Madison WI) in 300 ml binding buffer (50 mM Tris HCl, pH 7.4, 100 mM NaCl, 0.5%NP-40) overnight at 4uC. For experiments involving cell lysates, 3 mg GST or GST-GIPC were incubated with 500 ml cell lysate. Beads were sedimented and washed extensively in binding buffer and boiled in Laemmli sample buffer. Bead-bound proteins were separated by SDS-PAGE.

Endocytosis Assay for LPA 1
This assay was performed essentially as described previously [11]. HEK cells stably expressing LPA 1 were grown on cover slips pre-coated with fibronectin (BD Biosciences, Bedford, MA). Cells were serum starved in DMEM at 37uC for 4 h, incubated on ice with anti-FLAG IgG (1:1,000) for 0.5 h, washed with ice-cold PBS (3X), and shifted to fresh medium containing 1-10 mM LPA at 37uC for various times prior to fixation and processing for immunofluorescence.

Immunofluorescence
HEK cells were fixed with 3% paraformaldehyde in 100 mM phosphate buffer, pH 7.4, for 30 min, permeabilized with 0.1% Triton X-100 in 1% BSA for 10 min, and incubated with primary antibodies for 1 h followed by goat anti-rabbit Alexa-594 and/or anti-mouse Alexa-488 F(ab') 2 (Molecular Probes) for 1 h. Fluorescence images were taken with either an AxioImager M1 (Carl Zeiss, Thornwood, NY) equipped with a digital ORCA-ER camera (Hamamatsu), a PerkinElmer UltraView Vox Spinning Disk Confocal unit connected to an Olympus IX81 inverted microscope and a EMCCD camera (Hamamatsu), or an inverted Olympus FluoView 1000 confocal microscope equipped with a CH350 CCD camera (Hamamatsu). Images were processed with Adobe Photoshop 5.0 (Adobe Systems, Mountain View, CA). Fluorescence images of double-labeled samples were evaluated using the colocalization analysis features of the Volocity software (PerkinElmer, Waltham, MA).

Deglycosylation Assay
Glycosylation assays (PNGase F treatments) were performed using the N-Glycanase-PLUS kit (ProZyme, San Leandro, CA) according to the manufacturer's protocol. Briefly, HEK cells stably expressing FLAG-tagged LPA 1 or empty vector were lysed in 0.1% SDS, 50 mM Tris HCl, pH 7.5, and 50 mM b-Mercaptoethanol supplemented with protease inhibitors, and protein concentration was determined by the Bradford assay. Proteins (40 mg) were diluted in 45 ml of the above lysis buffer, and NP40 was added to a final concentration of 0.75%. 1 ml N-Glycanase-PLUS (Activity $10 U/ml) was added to half the samples, and the mixtures were incubated at 37uC for 3.5 h. Laemmli SDS sample buffer was added, proteins were resolved by SDS-PAGE, transferred to PVDF membranes, and analyzed by Western blotting using rabbit anti-FLAG IgG.

Statistical Methods
Data in graphs are presented as the mean 6 standard error of the mean (S.E.M) for n trials. Statistical analysis was carried out by Student's t-test, as appropriate, using 95% confidence limits. Specifics are detailed in the figure legends.

Cell Migration Assay
Migration assays were performed as described by Klemke et. al. [25]. Briefly, Boyden chambers containing polycarbonate membranes (tissue culture-treated, 6.5 mm diameter, 10 mm thickness, 8 mm pores, TranswellH; Costar Corp., Cambridge, MA) were coated on both sides with human fibronectin for 2 h at 37uC. Cells were transfected with control or GIPC siRNA, and after 24 h they were incubated in serum free DMEM for an additional 24 h. 1610 5 cells in 100 ml serum free DMEM containing 1 mM sodium pyruvate and 0.25% fatty acid free BSA were added to the top of each well; the bottom of each well contained the same medium with or without 1 mM LPA. Cells were allowed to migrate for 3 h at 37uC and subsequently stained with crystal violet (Sigma). Cells that migrated to the bottom of the filter in each well were counted under the microscope to assess cell migration.

Cell Proliferation Assay
Cell proliferation was assessed using a previously described crystal violet staining method [26]. Briefly, HEK cells stably expressing LPA 1 or empty vector were transfected with control siRNA or GIPC siRNA in 12 well plates using lipofectamine 2000. 24 h after transfection cells were trypsinized, and 2610 4 cells were transferred to each well of a 96 well plate and cultured at 37uC. At specific time points (0-72 h) cells were fixed with 3.7% paraformaldehyde for 5 min, and stained with 0.05% crystal violet for 30 min. To determine cell numbers, the crystal violet in the wells was solubilized in methanol and absorbance (OD 540 nm) determined directly using a plate reader.

GIPC Specifically Interacts with the PDZ Binding Motif of LPA 1
To determine if GIPC can interact with LPA 1 we transiently coexpressed GIPC-GFP and N-terminally tagged FLAG-LPA 1 in HEK293 cells and immunoprecipitated LPA1 with anti-FLAG IgG. We found that GIPC-GFP co-immunoprecipitated with FLAG-LPA 1 (Fig. 1A), suggesting that GIPC and LPA 1 are present in the same protein complexes.
To determine if GIPC interacts with the PDZ binding motif of LPA 1 we carried out GST pull-down assays with GST-LPA 1 (aa 311-364) on cell lysates from HEK293 cells transiently transfected with FLAG-GIPC. We found that GIPC bound to GST-LPA 1 (Fig. 1B, lane 3), but did not bind to GST-LPA 1 DC, lacking the PDZ binding motif (-SVV) (Fig. 1B, lane 4). To find out if the interaction between GIPC and LPA 1 is direct and whether the PDZ domain of GIPC is sufficient for the interaction we performed pull down assays using GST-fusion proteins and [ 35 S] Met-labeled, in vitro translated, GIPC-PDZ (aa 125-225). GIPC-PDZ bound to GST-LPA 1 (Fig. 1C, lane 3) but not to GST-LPA 1 -AAA, a mutated version of GST-LPA 1 in which the last three amino acids were mutated to alanine (Fig. 1C, lane 4). Interaction with the cytoplasmic tail of LPA 2 was much weaker (Fig. 1C, lane 5) even though it also has a class-I PDZ binding motif. To verify the specificity of GIPC's interaction with the PDZ binding motif of LPA 1 we mutated the last three amino acids of LPA 1 cytoplasmic tail (-SVV) to resemble the C-terminal sequence of LPA receptor subtypes 2 (-STL), 3 (-NGS), 4 (-STF) and 5 (-SAL). In vitro translated GIPC-PDZ bound to the PDZ binding motif of LPA 1 whereas interactions with other PDZ binding motifs were much weaker (Fig. 1D), suggesting that GIPC interacts specifically with LPA 1 and can distinguish the PDZ binding motif of LPA 1 from closely related PDZ binding motifs of other members of the LPA receptor family. Taken together these results demonstrate that GIPC directly binds to the PDZ binding motif of LPA 1 , that this interaction is specific for LPA 1 , and that it is mediated via the PDZ domain of GIPC and the C-terminal PDZ binding motif of LPA 1 .

LPA 1 and GIPC Traffic Together to APPL Endosomes
We have previously shown [11] that GIPC binds to the receptor tyrosine kinase TrkA and regulates its trafficking and signaling through interaction with APPL, a Rab5 effector that serves as a marker for APPL signaling endosomes [14][15][16]. To investigate if GIPC similarly regulates trafficking and signaling of LPA 1 we prepared HEK293 cell lines stably expressing FLAG-LPA 1 (HEK-LPA 1 ) or empty vector (HEK-pIRES) (Fig. S1). We chose HEK293 cells because they were previously shown to express LPA 1 but not LPA 2 or LPA 3 , [27], and therefore any response to LPA observed is likely to be via activation of LPA 1 and its downstream signaling network. First we followed the trafficking of LPA 1 and its association with GIPC and APPL in these cells. In serum starved HEK-LPA 1 cells stably expressing LPA 1 , LPA 1 colocalized with GIPC along the PM (Figs. 2A, and S3 upper panel). Similar results were also obtained in HeLa cells transiently expressing LPA 1 (Fig. S2). By 2-5 min after stimulation with LPA, LPA 1 had been partially internalized and accumulated on peripheral vesicles located just beneath the plasma membrane that colocalize with both GIPC and APPL ( Fig (Fig. S4). Thus our results suggest that, like TrkA [11], after agonist stimulation LPA 1 is internalized and passes first through APPL endosomes located at the cell periphery and then to EEA1 early endosomes located in the juxtanuclear region.
To determine if LPA 1 and GIPC are internalized via clathrin mediated endocytosis we performed double labeling for clathrin and GIPC or LPA 1 (Fig. S5). We found that 2-3 minutes following addition of LPA, both LPA 1 and GIPC colocalized with clathrin in punctate structures at or just beneath the plasma membrane indicating that following LPA stimulation, GIPC and LPA 1 are internalized into clathrin coated pits which pinch off the plasma membrane to form clathrin coated vesicles.
To find out if ligand stimulation affects the association between LPA 1 and GIPC we immunoprecipitated LPA 1 from HEK293 cells transiently expressing FLAG-LPA 1 before and after stimulation with LPA (5-30 min). GIPC co-immunoprecipitated with LPA 1 at all time points, but the amount of GIPC that coimmunoprecipitated with LPA 1 gradually declined after ligand stimulation (Fig. 2B). Collectively the immunofluorescence and biochemical results suggest that, as for TrkA [11], GIPC associates with LPA 1 at the plasma membrane, GIPC and LPA 1 travel together to APPL endosomes (2-5 min), and they dissociate from one another before LPA 1 reaches early (EEA1) endosomes (30 min).

GIPC Depletion Disrupts LPA 1 Trafficking
Next we investigated the effects of GIPC depletion on LPA 1 trafficking at 0 and 15 min after LPA stimulation. In serum starved cells LPA 1 was present largely at the plasma membrane in both GIPC-depleted cells and controls (not shown). At 15 min after addition of LPA, in controls LPA 1 appeared both at the PM and in vesicles inside the cell where it partially colocalized with the early endosome marker EEA1 (Fig. 3A, upper panel). By contrast in GIPC-depleted cells fewer vesicles with LPA 1 were seen in the cytoplasm, and colocalization between LPA 1 and EEA1 was markedly reduced (Fig. 3A, middle panel). Quantification of the overlap between LPA 1 and EEA1 (Fig. 3B) using Volocity software revealed a 32% decrease in the average overlap coefficient (OC) in GIPC depleted cells (OC = 0.45) compared to controls (OC = 0.66). The decreased localization of LPA 1 in EEA1 early endosomes at 15 min after LPA addition suggests that in GIPC depleted cells there is a delay in trafficking of LPA 1 from the plasma membrane or peripheral vesicles to early endosomes.
To test if following GIPC depletion, LPA 1 accumulates in peripheral (APPL) signaling endosomes we carried out double labeling for LPA 1 and APPL1 0-10 min after LPA stimulation (Fig. 3C). We found that in control cells, colocalization between APPL1 and LPA 1 in APPL endosomes peaked at 3 min and was barely detected at 10 min after LPA stimulation. In contrast, in GIPC depleted cells, colocalization between APPL1 and LPA 1 increased 3 min after LPA stimulation but remained high even after 10 min. Taken together, these results suggest that GIPC promotes trafficking of LPA 1 from peripheral APPL signaling endosomes to early endosomes after internalization of the receptor from the plasma membrane.

GIPC Depletion Enhances LPA 1 Signaling
To investigate the effects of GIPC depletion on LPA 1 signaling we stimulated HEK-LPA 1 cells with LPA and assessed activation (phosphorylation) of Erk and Akt-two signaling pathways that mediate cell survival, proliferation and motility. We found that at both 5 and 20 min after LPA stimulation GIPC depletion (,70%) enhanced Akt activation by ,2-fold ( Fig. 4A and B) but had no effect on pErk levels ( Fig. 4A and C). Similar findings were obtained using different clones of HEK-LPA 1 cells. Transfection of siRNA resistant GIPC into GIPC depleted cells reversed the effect of GIPC siRNA on Akt phosphorylation in a dose dependent manner (Fig. 5A, lanes 5-7) verifying that the effects of GIPC expression on Akt phosphorylation are not due to off target effects of the siRNA.
We showed previously that GIPC recruits GAIP (RGS19), a GAP for Gai proteins [28], and inhibits Gi signaling [10]. To determine if Gai activity is required for LPA 1 mediated Akt phosphorylation we pre-treated GIPC depleted and control cells with pertussis toxin (PTX), an inhibitor of Gai/GPCR coupling, before LPA stimulation. PTX abolished LPA induced Erk activation but did not affect activation of Akt (Fig. 5B). Notably, PTX did not inhibit the increased Akt phosphorylation seen in GIPC depleted cells (Fig. 5B), indicating that the effect of GIPC on Akt phosphorylation is most likely not mediated through Gai subunits.

APPL is Present in LPA 1 Complexes and APPL Depletion Inhibits Akt Activation
APPL directly binds GIPC as well as the TrkA receptor [11,13] and promotes Akt signaling and cell survival [14]. To determine whether LPA 1 forms a complex with APPL and GIPC we immunoprecipitated FLAG-LPA 1 from HEK-LPA 1 cells at steadystate (10% FBS) and immunoblotted for APPL and GIPC. We found that APPL and GIPC co-immunoprecipitated with LPA 1 (Fig. 6A, lane 3), indicating that LPA 1 is present in the same protein complexes as GIPC and APPL.
To determine if APPL is required for enhancing Akt phosphorylation following GIPC depletion we treated HEK-LPA 1 cells with control siRNA (Fig. 6B, lanes 1-3), GIPC siRNA alone (Fig. 6B, lanes 4-6), APPL siRNA alone (Fig. 6B, lanes 7-9) or both GIPC and APPL siRNA (Fig. 6B, lanes 10-12) and stimulated the cells with LPA. Depletion of GIPC led to enhanced Akt signaling as before, whereas depletion of APPL or double knockdown of GIPC and APPL reduced Akt signaling. The reversal of Akt enhancement in the double knockdown suggests that APPL is required for the enhancement of Akt signaling. Taken together, these results suggest that following ligand stimulation APPL associates with LPA 1 protein complexes and mediates Akt activation downstream of LPA 1 .

GIPC Depletion Promotes LPA 1 Mediated Cell Proliferation and Cell Motility
Next we investigated if GIPC depletion can affect cell growth in the presence of LPA. GIPC was depleted from HEK-LPA 1 cells, and the growth of GIPC depleted vs control HEK-LPA 1 cells was followed for 96 h after siRNA transfection. GIPC depletion resulted in a 72% increase in the number of cells per well (43,000+/26,000 vs 25,000+/28,000 cells/well) (Fig. 7A). GIPC depletion did not significantly affect growth of HEK-pIres controls that do not express FLAG-LPA 1 (Fig. 7A), indicating that enhancement of cell growth is mediated through LPA 1 . In addition, the number of HEK-LPA 1 cells that incorporated BrdU was increased from 18% in controls to 23% in GIPC depleted cells (data not shown), suggesting a slightly faster cell cycle. These results are consistent with a role for GIPC in down-regulating LPA 1 mediated cell growth or cell survival.
Because LPA 1 is also known to trigger cell motility we next examined the effect of GIPC depletion on cell migration by analyzing movement of HEK-LPA 1 cells across a porous membrane in a Boyden chamber in the presence of concentration gradient of LPA [25]. GIPC depletion enhanced motility of HEK-LPA 1 cells in that increased numbers of cells migrated across the membrane both in the presence and absence of a concentration gradient (Fig. 7B). These results demonstrate that GIPC inhibits cell motility in cells expressing LPA 1 . Previously, LPA 1 was shown to possess intrinsic basal activity even in the absence of ligand binding [29]. Thus the inhibitory effect of GIPC on cell motility in the absence of ligand is most likely due to inhibition of the basal activity of LPA 1 .
Based on our results we propose a working model (Fig. 8) in which GIPC associates with LPA 1 at the PM in a ligand independent manner, and following ligand stimulation the receptor and GIPC are internalized in clathrin-coated vesicles and associate with APPL which activates Akt signaling. Subsequently, GIPC promotes LPA 1 trafficking to early (EEA1) endosomes and thus terminates APPL/Akt signaling. Depletion of GIPC delays Figure 2. Trafficking of LPA 1 and GIPC to Endosomes. A, Upper panel: In serum starved cells stably expressing LPA 1 (asterisks), GIPC is concentrated at the plasma membrane (arrowheads) where it colocalizes with LPA 1 . Middle panels: 2 min following stimulation with LPA, LPA 1 colocalizes with APPL and GIPC (arrowheads) in endocytic vesicles at the cell periphery. Lower panel: 30 min following stimulation, LPA 1 colocalizes with EEA1 in early endosomes concentrated in the juxtanuclear region (arrowheads). Boxed regions are enlarged (2.26) in the insets. Bar = 10 mm. HEK-LPA 1 cells were serum starved for 4 h, incubated on ice with mouse anti-FLAG IgG to label FLAG-LPA 1 at the cell surface, washed in PBS and shifted to fresh medium containing LPA (10 mM) 2 or 5 min before fixation. Cells were processed for immunofluorescence using affinity purified rabbit anti-GIPC, anti-APPL1 or anti-EEA1 IgG, followed by goat anti-rabbit Alexa-488 and goat-anti-mouse Alexa-594 F(ab') 2 (the latter to detect FLAG-LPA 1 ). Images were taken with a PerkinElmer UltraView Vox Spinning Disk Confocal unit connected to an Olympus IX81 inverted microscope and a EMCCD camera (Hamamatsu) using a 60X oil immersion lens (1.42 NA). B, GIPC co-immunoprecipitates with FLAG-LPA 1 from serum-starved cells (lane 7) at all time points after LPA stimulation, but the interaction gradually decreases after LPA stimulation (lanes 8-10). The relative abundance of GIPC that coprecipitated with FLAG-LPA 1 is indicated beneath each band. As expected, both LPA 1 and GIPC are absent from immunoprecipitates of cells transiently transfected with empty vector instead of FLAG-LPA 1 (lane 6, vector control). HEK293 cells were transiently cotransfected with full-length GIPC and FLAG-LPA 1 (lanes 2-5 and 7-10) or GIPC alone (lanes 1 and 6). Cells were serum starved overnight (lanes 1, 2, 6 and 7) or starved and stimulated with 10 mM LPA for 5 (lanes 3 and 8), 15 (lanes 4 and 9) or 30 min (lanes 5 and 10) before lysis. IP was carried out on cell lysates using mouse anti-FLAG IgG and immunoblotted as is Fig. 1A. The abundance of GIPC and LPA 1 in each immunoprecipitation reaction was quantified using the LICOR imaging system, and the GIPC abundance relative to LPA 1 was calculated for each reaction. Similar results were obtained in 2 additional experiments. Input (lanes 1-5): Lysates (2%) are shown to verify comparable expression levels. doi:10.1371/journal.pone.0049227.g002  (2.26) in the insets. Images were acquired with a Zeiss AxioImager M1 microscope, and overlap in staining between LPA 1 and EEA1 was evaluated using Volocity software. Statistical significance (p value) was determined by t-test. B, Trafficking of LPA 1 is delayed in APPL1 endosomes after depletion of GIPC. Left panel: In both GIPC-depleted (GIPC siRNA) and controls (Ctrl siRNA), LPA 1 is localized along the plasma membrane after serum starvation (0 min) whereas APPL1 is found in peripheral cytoplasmic vesicles. Middle Panel: In both GIPC depleted and control cells stimulated with LPA for 3 min, LPA 1 colocalizes with APPL1 in cytoplasmic vesicles (arrowheads). Right Panel: In controls stimulated with LPA for 10 min, very few LPA 1 receptors remain in APPL endosomes (yellow, arrowhead) whereas in GIPC-depleted cells the majority of the receptors are retained in APPL endosomes (yellow, arrowhead). Boxed regions are enlarged (36) in the insets. HEK-LPA 1 cells grown on coverslips were transfected with GIPC or control siRNA. 72 h after transfection cells were serum starved for 4-6 h and subsequently incubated on ice with rabbit (A) or mouse (C) anti-FLAG IgG, shifted to fresh medium containing LPA for the indicated times, then fixed and processed for immunofluorescence using mouse anti-EEA1 IgG (A) or rabbit-anti-APPL1 IgG (C) as in Fig. 2A. Images in ''A'' were acquired with a Zeiss AxioImage M1 microscope, and those in ''C'' were acquired with an Ultra View Vox Spinning Disk Confocal. Bar = 10 mm. doi:10.1371/journal.pone.0049227.g003 trafficking of LPA 1 , prolongs its stay in APPL signaling endosomes, and enhances Akt signaling leading to increased cell motility and cell proliferation.

Discussion
We demonstrate here that GIPC binds LPA 1 and that binding is direct and is mediated through the PDZ domain of GIPC and the C-terminal PDZ binding motif of LPA 1 . siRNA depletion of GIPC delayed trafficking of LPA 1 to early endosomes and resulted in enhanced LPA 1 -mediated Akt signaling and enhanced cell proliferation and cell motility. APPL, a marker APPL/GIPC signaling endosomes, was present in LPA 1 complexes and necessary for LPA 1 mediated Akt signaling. Taken together, these results support a model in which GIPC promotes trafficking of LPA 1 from APPL signaling endosomes to early (EEA1) endosomes thus attenuating LPA 1 mediated signaling and cellular responses (see Figure 8).
Both LPA 1 and GIPC have been implicated in cell migration [8,17,[30][31][32], neuronal cell activity [33,34] and cell proliferation [9,35,36]. GIPC has been shown to inhibit endothelial cell migration through interaction with Endoglin [37] or syndecan-4 [38], but it promotes migration of primary arterial endothelial cells [8]. LPA 1 has been shown to promote migration and proliferation of many cell types [30][31][32]35]. Our results showing that GIPC binds LPA 1 and regulates its activity suggest a novel mechanism by which GIPC affects cell migration. We also observed an apparent increase in cell proliferation following GIPC depletion in cells expressing LPA 1 . The effects of LPA 1 on cell proliferation are most likely indirect and are believed to reflect a combination of the secondary release of growth factors following initial LPA stimulation combined with anti-apoptotic actions [39][40][41]. The increase in cell number following GIPC depletion coincides with enhanced LPA 1 activity and presumably stems from primary effects on cell survival coupled with secondary effects on cell proliferation.
GIPC was previously shown to define the signaling specificity of b-adrenergic receptor subtypes [42]. Our finding that GIPC interacts with LPA 1 but shows much weaker or no interaction with other LPA receptor subtypes may similarly explain the differential effect of LPA 1 and LPA 2 on cell migration and proliferation [35]. In the case of LPA receptors, binding to PDZ domain proteins has recently been shown to influence the signaling outcomes of different LPA receptors [43][44][45]. The PDZ proteins NHERF2 and  1-4), possibly due to enhanced basal activity of LPA 1 . Erk phosphorylation (pERK) was not affected by GIPC depletion. HEK-LPA 1 cells were transfected with control or GIPC siRNA, serum starved overnight, stimulated with 1 mM LPA or incubated with BSA alone for 5 or 20 min, lysed in RIPA buffer and analyzed by immunoblotting using phospho-Erk (pErk), total Erk (tErk), phospho-Akt (pAkt) and a-tubulin IgG. Each treatment was done in duplicate. a-tubulin was used as a loading control. In cells transfected with GIPC siRNA (lanes 3-4, 7-8, 11-12), GIPC expression is reduced 70-80% in cells transfected with control siRNA (Ctrl, lanes 1-2, 5-6, 9-10). B-C, Densitometric analysis of the immunoblots in A demonstrating that GIPC depletion (siRNA) leads to a two-fold increase in Akt phosphorylation (B) at both 5 and 20 min after LPA stimulation (B, P,0.05) but does not significantly affect Erk phosphorylation (C) compared to controls (Ctrl siRNA). doi:10.1371/journal.pone.0049227.g004 MAGI-3 were shown to couple LPA 2 to PLC-b3, RhoA and Erk signaling [43,44], and two additional PDZ proteins, PDZ-RhoGEF and LARG, have been shown to interact with both LPA 1 and LPA 2 [45]. Because the latter proteins bind to both LPA 1 and LPA 2 , these interactions can't explain the different effects of LPA 1 and LPA 2 on cell behavior [30,35].
Shano et. al. [46] recently reported that a point mutation in the LPA 1 PDZ binding motif led to increased Akt signaling and cell proliferation. Our findings that GIPC binds to the PDZ binding motif of LPA 1 and depletion of GIPC has similar effects suggests that the findings of Shano et al can be explained by loss of interaction of LPA 1 with GIPC. In contrast, loss of interaction between LPA 1 and the PDZ proteins PDZ-RhoGEF and LARG [45] had different consequences suggesting that these proteins do not mediate the effects on Akt and cell proliferation. Previously it was shown that deletion of the LPA 1 PDZ binding motif enhances Akt signaling [46] but did not affect inositol phosphate production [21]. These observations suggest that Akt enhancement is mediated by PLC-and inositol phosphate-independent mechanisms [13].
We previously discovered that in PC12 cells, GIPC binds to APPL on peripheral endosomes and that depletion of GIPC slows down endocytosis and trafficking of TrkA and the Rab5-effector APPL to early EEA1 endosomes [11]. Here we show that GIPC depletion similarly delays trafficking of LPA 1 to early EEA1 endosomes and prolongs the residence of LPA 1 receptor on APPL1 signaling endosomes. Despite the fact that GIPC depletion has similar effects on the trafficking of TrkA and LPA1, their signaling outcomes differ: GIPC depletion reduced TrkA mediated Akt and Erk signaling but enhanced LPA1 mediated Akt signaling [11]. This illustrates that signaling outcomes can be widely divergent among different receptors. Signaling depends on protein-protein interaction networks, and each receptor has a distinctive set of binding partners. TrkA and LPA1 are representatives of two diverse families, the receptor tyrosine kinases (RTKs) and G protein coupled receptors (GPCR), respectively, which have very different modes of signaling. As discussed earlier, even closely related receptors, such as LPA1, LPA2 and LPA3, form distinct protein-protein interactions with distinct signaling outcomes. Thus the molecular mechanisms underlying the different effects of GIPC depletion on TrkA and LPA1 signaling will be fully understood only when their specific binding partners and protein interaction networks are established.
Urs et al reported that deletion of the PDZ binding motif of LPA 1 did not affect inositol phosphate signaling or the amount of LPA 1 receptor that accumulated at the surface of HeLa cells 30 min after ligand stimulation [22]. The lack of effect on receptor accumulation suggests that the PDZ binding motif is not required for internalization of receptor from the surface. Indeed, we and others have previously shown that binding of GIPC to the PDZ motif does not promote internalization of receptors from the surface but rather promotes trafficking of receptors from peripheral signaling endosomes to early endosomes [11,12,17]. The association between LPA 1 and APPL and the effects of GIPC on LPA 1 trafficking further expand the role of GIPC and APPL to regulation of the activity of G-protein coupled receptors.
We found here that following ligand stimulation, LPA 1 internalizes and traffics through APPL peripheral endosomes on its way to EEA1 early endosomes. Our results are in keeping with previous findings showing that ligand induced endocytosis of LPA 1 is dependent on dynamin2 and Rab5 and that internalized LPA 1 traverses the same endocytic pathway as the transferrin receptor in that it passes through sorting endosomes, early (EEA1) endosomes and juxtanuclear recycling endosomes [33]. GIPC is believed to affect receptor trafficking in part by binding to the Rab5 effector APPL [16,47] and in part by binding to the actin based motor protein myosin VI [12,48,49]. We demonstrated that APPL associates with LPA 1 complexes and colocalizes with LPA 1 in peripheral endosomes. We also found that APPL depletion inhibits Akt signaling in cells expressing LPA 1 . This is in keeping with previous reports that APPL is required for activation of Akt on endosomes and for cell survival [11,[14][15][16]50]. It appears that GIPC depletion prolongs LPA 1 association with APPL signaling endosomes by delaying LPA 1 trafficking to early (EEA1) endosomes, leading to increased Akt signaling and promoting cell proliferation and motility.
Our finding that interaction between GIPC and LPA 1 leads to downregulation of Akt signaling has important pathophysiological implications, given 1) that LPA 1 has been shown to promote the development of various carcinomas, 2) that mutations in the PDZ binding motif of LPA 1 induces oncogenic transformation [1,[42][43][44]46,51,52], and 3) that GIPC plays a tumor suppressor role in breast cancer progression [51,53]. Whether and how the interaction of these two proteins is abrogated during cancer progression remains unknown.
In summary, the identification of signaling pathways involving GIPC and APPL downstream of LPA 1 extend the role of these Figure 5. Enhancement of Akt activation following GIPC depletion is reversed by GIPC expression and is independent of Gai signaling. A, GIPC depleted HEK-LPA 1 cells show reduced Akt phosphorylation after transfection of siRNA resistant GIPC DNA (lanes 6-7, middle panel) verifying that GIPC is responsible for the enhanced Akt phosphorylation seen after GIPC depletion. HEK-LPA 1 cells were transfected with GIPC or control siRNA, and 12 h later they were transfected with siRNA-resistant GIPC DNA (0, 10, or 33 ng). After 24 h cells were serum starved overnight, stimulated with 1 mM LPA for 5 min, and cell lysates were immunoblotted for GIPC, pAkt and tubulin (used as loading control). B, Activation of Gai is not required for the enhanced Akt phosphorylation seen after GIPC depletion. In GIPCdepleted cells PTX treatment (lanes 7-10) prevented LPA induced Erk phosphorylation (pErk) (which is Gai dependent) but did not affect Akt phosphorylation (pAkt) compared with controls (lanes 3-6). 36 h after siRNA transfection, HEK-LPA 1 cells were cultured for another 12 h in serum-free media in the presence or absence of PTX and then stimulated for 5 min with 1 mM LPA (in 0.1% BSA, lanes 3-10) or incubated in BSA alone (0.1%) for 5 min (lanes 1-2). doi:10.1371/journal.pone.0049227.g005  proteins as regulators of GPCRs and opens exciting directions for investigation. The ability of GIPC to bind LPA 1 , APPL and myosin VI in a ligand dependent manner positions GIPC as a key target for regulation of LPA 1 activities. GIPC was previously shown to interact with additional GPCRs, including the dopamine D2 receptor and the lutropin receptor, but it is not known if APPL also associates with these receptors. Future studies will reveal if GIPC and APPL regulate signaling and trafficking of these and other GPCRs. Figure S1 Characterization of HEK-LPA 1 cell lines stably expressing FLAG-LPA 1 . A, Immunoblot of LPA 1 from HEK-LPA 1 cell lysates demonstrating receptor expression and glycosylation. A prominent broad band at ,60 kD is seen in HEK-LPA 1 cells (Lane 1) but not in HEK-pIRES controls stably expressing empty vector (lane 2). The electrophoretic mobility of FLAG-LPA 1 shifts to the predicted theoretical molecular mass (38 kD) following treatment with PNGase-F (Lane 3) which removes N-glycans. The broad mobility and fuzziness of the 38 kD band most likely is due to remaining O-glycans. Lysates from HEK-LPA 1 and HEK-pIRES cells were treated with PNGase (lanes 3-4) or sham treated (lanes 1-2), and proteins were immunoblotted with anti-FLAG IgG. B, LPA (0.01-1 mM) induces phosphorylation of Erk and Akt in HEK-LPA 1 cells (lanes 2, 4, 6 and 8) but not in HEK-pIRES cells (lanes 1, 3, 5 and 7). HEK-LPA 1 and HEK-pIRES cells were serum starved overnight, stimulated with the indicated amounts of LPA in 0.1% fatty acid free BSA for 5 min, lysed and analyzed by immunoblotting for LPA 1 (FLAG), pErk, tErk, and pAkt. C, Phase contrast microscopy of HEK-pIRES and HEK-LPA 1 cells showing that stable expression of LPA 1 induces morphological changes in HEK293 cells. HEK-pIRES controls exhibit elongated processes (arrowheads, left panel) and overall morphology similar to the parental HEK293 cell line whereas HEK-LPA 1 cells are flatter, more spread out and have shorter cell processes (right panel). (TIF) Figure S2 FLAG-LPA 1 and GIPC colocalize at the plasma membrane in HeLa cells. A, Endogenous GIPC (red, in merged image) is widely distributed throughout the cytoplasm and is also concentrated along the plasma membrane whereas LPA 1 -FLAG (green) is mainly localized at the plasma membrane where it partially colocalizes with GIPC as demonstrated by yellow overlapping pixels (arrowheads, right lower panel). HeLa cells were transfected with FLAG-LPA 1 and subsequently serum starved and processed for immunofluorescence using affinity purified rabbit anti-GIPC and mouse anti-FLAG IgG followed by goat anti-rabbit Alexa-593 and goat-antimouse Alexa-488 F(ab') 2 and examined with an Olympus FluoView 1000 confocal microscope using a 60X objective. (TIF)  In the absence of ligand LPA 1 is found at the plasma membrane in a complex with GIPC. When LPA is added, LPA 1 and GIPC move into clathrincoated pits (1). Clathrin-coated vesicles containing LPA 1 -GIPC complexes pinch off the cell membrane and uncoat and APPL is recruited (2). APPL binds pAkt to form peripheral signaling endosomes. GIPC, by binding to APPL and the motor protein myosin VI, facilitates movement of these endosomes to the juxtanuclear region (3). In juxtanuclear early endosomes, GIPC and APPL are released into the cytoplasm thus terminating APPL-pAkt signaling. Depletion of GIPC inhibits LPA 1 trafficking to EEA1 endosomes and prolongs LPA 1 signaling from APPL endosomes whereas depletion of APPL inhibits Akt signaling. doi:10.1371/journal.pone.0049227.g008 brane. Lower panels: GIPC (red) colocalizes (arrowheads) with clathrin (green) at the plasma membrane and on endocytic vesicles at 3 min after LPA stimulation. Boxed regions are enlarged (2.36) in the insets. HEK-LPA 1 cells were stimulated, processed for immunofluorescence and images acquired exactly as described for Fig. 2. Bar = 10 mm. (TIF)