Portal de Periódicos da CAPES

PIKE-mediated PI3-kinase activity is required for AMPA receptor surface expression

2011; Springer Nature; Volume: 30; Issue: 20 Linguagem: Inglês

10.1038/emboj.2011.281

1460-2075

Chi Bun Chan, Yongjun Chen, Xia Liu, Xiaoling Tang, Chi Wai Lee, Lin Mei, Keqiang Ye,

Protein Kinase Regulation and GTPase Signaling

Article16 August 2011free access PIKE-mediated PI3-kinase activity is required for AMPA receptor surface expression Chi Bun Chan Chi Bun Chan Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, GA, USA Search for more papers by this author Yongjun Chen Yongjun Chen Institute of Molecular Medicine and Genetics and Department of Neurology, Georgia Health Sciences University, Augusta, GA, USA Search for more papers by this author Xia Liu Xia Liu Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, GA, USA Search for more papers by this author Xiaoling Tang Xiaoling Tang Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, GA, USA Search for more papers by this author Chi Wai Lee Chi Wai Lee Department of Cell Biology, Emory University School of Medicine, Atlanta, GA, USA Search for more papers by this author Lin Mei Lin Mei Institute of Molecular Medicine and Genetics and Department of Neurology, Georgia Health Sciences University, Augusta, GA, USA Search for more papers by this author Keqiang Ye Corresponding Author Keqiang Ye Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, GA, USA Search for more papers by this author Chi Bun Chan Chi Bun Chan Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, GA, USA Search for more papers by this author Yongjun Chen Yongjun Chen Institute of Molecular Medicine and Genetics and Department of Neurology, Georgia Health Sciences University, Augusta, GA, USA Search for more papers by this author Xia Liu Xia Liu Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, GA, USA Search for more papers by this author Xiaoling Tang Xiaoling Tang Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, GA, USA Search for more papers by this author Chi Wai Lee Chi Wai Lee Department of Cell Biology, Emory University School of Medicine, Atlanta, GA, USA Search for more papers by this author Lin Mei Lin Mei Institute of Molecular Medicine and Genetics and Department of Neurology, Georgia Health Sciences University, Augusta, GA, USA Search for more papers by this author Keqiang Ye Corresponding Author Keqiang Ye Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, GA, USA Search for more papers by this author Author Information Chi Bun Chan1,‡, Yongjun Chen2,‡, Xia Liu1, Xiaoling Tang1, Chi Wai Lee3, Lin Mei2 and Keqiang Ye 1 1Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, GA, USA 2Institute of Molecular Medicine and Genetics and Department of Neurology, Georgia Health Sciences University, Augusta, GA, USA 3Department of Cell Biology, Emory University School of Medicine, Atlanta, GA, USA ‡These authors contributed equally to this work *Corresponding author. Department of Pathology and Laboratory Medicine, Emory University School of Medicine, 615 Michael Street, Atlanta, GA 30322, USA. Tel.: +1 404 712 2814; Fax: +1 404 712 2979; E-mail: [email protected] The EMBO Journal (2011)30:4274-4286https://doi.org/10.1038/emboj.2011.281 PDFDownload PDF of article text and main figures. Peer ReviewDownload a summary of the editorial decision process including editorial decision letters, reviewer comments and author responses to feedback. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info AMPAR (α-amino-3-hydroxy-5-methyl-isoxazole-4-propionic acid receptor) is an ion channel involved in the formation of synaptic plasticity. However, the molecular mechanism that couples plasticity stimuli to the trafficking of postsynaptic AMPAR remains poorly understood. Here, we show that PIKE (phosphoinositide 3-kinase enhancer) GTPases regulate neuronal AMPAR activity by promoting GluA2/GRIP1 association. PIKE-L directly interacts with both GluA2 and GRIP1 and forms a tertiary complex upon glycine-induced NMDA receptor activation. PIKE-L is also essential for glycine-induced GluA2-associated PI3K activation. Genetic ablation of PIKE (PIKE−/−) in neurons suppresses GluA2-associated PI3K activation, therefore inhibiting the subsequent surface expression of GluA2 and the formation of long-term potentiation. Our findings suggest that PIKE-L is a critical factor in controlling synaptic AMPAR insertion. Introduction PIKE (phosphoinositide 3-kinase enhancer) is a family of GTPases that directly interact and enhance the activity of phosphoinositide 3-kinase (PI3K) and Akt (Ye et al, 2000; Rong et al, 2003; Ahn et al, 2004). Through different promoters and alternative splicing of CENTG1 gene, three PIKE isoforms (PIKE-L, PIKE-S and PIKE-A) are generated (Chan and Ye, 2007). Our previous studies showed that PIKE promote cell survival, which functionally links extracellular stimuli to the PI3K/Akt pathway. For instance, PIKE-L interacts with Homer-1 and metabotropic glutamate receptor I (mGlu1) during mGlu1 activation in hippocampal neurons, leading to the activation of PI3K and prevention of neuronal apoptosis (Rong et al, 2003). It also associates with netrin-1 receptor Unc5B to activate PI3K upon netrin stimulation, thus protecting neurons from apoptosis (Tang et al, 2008). Recently, we reported that PIKE-L partnered with the DNase inhibitor SET, and prevented it from cleavage by endopeptidase AEP during excitotoxicity and stroke (Liu et al, 2008). Thus, PIKE-L is critical in maintaining neuronal survival against apoptotic stimuli. Long-term potentiation (LTP) is a cellular model of memory formation and consolidation, resulting from long-lasting alternations in the efficacy of synaptic transmission (Bliss and Collingridge, 1993). Accumulating evidences suggest that trafficking of postsynaptic α-amino-3-hydroxy-5-methyl-isoxazole-4-propionic acid receptor (AMPAR) is critical for synaptic plasticity (Genoux and Montgomery, 2007). AMPAR is an ionotropic glutamate receptor that contributes the majority of fast excitatory synaptic transmission (Palmer et al, 2005). It is a tetrameric ion channel composed of various combinations of four subunits (GluA1 to GluA4) and gates the passage of Na+, K+ and Ca2+ ions (Rosenmund et al, 1998). While incorporation of GluA2 to AMPAR renders the passage of Ca2+, AMPAR with Q/R unedited GluA2 is calcium permeable (Burnashev et al, 1992). Although many mechanisms like structural remodelling of synapse and alternation in gene transcription could lead to the change of synaptic plasticity, it is in consensus that addition of AMPAR to postsynaptic surface mediates the formation of LTP (Muller et al, 2000; Costa-Mattioli et al, 2009; Santos et al, 2009). Trafficking of AMPAR to the synaptic membrane is a dynamic process that occurs constitutively between the cytoplasm and cell surface, and is facilitated by several proteins. For example, export of AMPAR from the ER to cell surface is controlled by molecular chaperones TARP (transmembrane AMPAR regulatory proteins) (Ziff, 2007). It has also been reported that PICK1 (protein interacting with C-kinase 1) was required for AMPAR's endocytosis (Xia et al, 1999; Lu and Ziff, 2005). In addition, it is suggested that GRIP1 (glutamate receptor-interacting protein 1), a seven PDZ-containing protein that interacts with the C-terminus of GluA2, is essential for AMPAR trafficking (Dong et al, 1997; Srivastava et al, 1998; Matsuda et al, 1999; Wyszynski et al, 1999; Osten et al, 2000; Zhang et al, 2001). Functioning as an adaptor protein, GRIP1 links AMPAR with microtubule motor proteins kinesins to facilitate the receptor transportation along the cell body (Setou et al, 2002; Hoogenraad et al, 2005). Synaptic targeting of AMPAR could also be regulated during synaptic plasticity (Song and Huganir, 2002). Pioneer studies by electrophysiological method in the CA1 region of hippocampus suggest that the synaptic level of AMPAR is rapidly enhanced during tetanic stimulation (Lissin et al, 1998). This activity-induced delivery of AMPAR requires synaptic N-methyl-D-aspartic acid receptor (NMDAR) activation (Shi et al, 1999). Indeed, activation of NMDAR by either tetanic stimulation or partial agonist glycine is sufficient to induce postsynaptic AMPAR insertion (Lu et al, 2001). Studies by Man et al (2003) further reveal that the glycine-induced AMPAR trafficking depends on the GluA2-associated PI3K activity. In this model, Ca2+ influx through the activated NMDAR results in an activation of GluA2-associated PI3K and accumulation of membrane binding phosphoinositide-3,4,5-trisphosphate (PtdIns-3P), which in turn, facilitates membrane targeting of AMPAR. However, how PtdIns-3P triggers the delivery of AMPAR from its intracellular pool to cell surface remains unknown. Here, we report that PIKE-L acts as a novel interacting partner to both GRIP1 and GluA2, which enhances the formation of GluA2/GRIP1 complex during glycine-induced LTP. Moreover, PIKE-L is critical for glycine to trigger GluA2-associated PI3K activation. Depletion of PIKE in PIKE knockout (PIKE−/−) mice changes the AMPAR activity and abolishes the glycine-induced AMPAR-associated PI3K activation, thus inhibiting the subsequent GluA2 trafficking and LTP formation. Results PIKE interacts with GRIP1 To look for the binding targets of PIKE-L, we performed a yeast two-hybrid (Y2H) analysis using the GTPase domain as the bait. Four of the 12 independent positive clones were identified as overlapping fragments of GRIP1 protein. We observed the interaction between the GTPase domain of PIKE-L and PDZ domain 4–6 (amino acids 483–708) in GRIP1 regardless of which protein was used as bait or prey (Figure 1A). To verify the interaction between these two proteins in a cellular context, we conducted the binding assay in HEK293 cells. Both wild-type (WT) and GTPase dominant-negative (KS) PIKE-L interacted strongly with GRIP1 (Figure 1B). Similarly, both PIKE-A WT and PIKE-A KS associated with GRIP1 as well, indicating that the N-terminal domain (NTD) and GTPase activity of PIKE are dispensable for their associations with GRIP1. The interaction of PIKE-L and GRIP1 was readily detectable in cultured neurons or brain tissues (Figure 1C, fourth lane, and Figure 1D, second lane). The assay specificity was confirmed by the negative signal in the control IgG (Figure 1C, first lane, and Figure 1D, first lane). Immunofluorescence staining on primary hippocampal neurons revealed the co-localization (Rr=0.58±0.03, n=3) of PIKE-L and GRIP1 in dendrites (Figure 1E), which further supported the physiological interaction of these two proteins. This staining was specific as the signal was blocked by antigen used in generating PIKE-L antibody but not by GST control protein (data not shown). Figure 1.PIKE associates with GRIP1. (A) Y2H screening using PIKE-L GTPase domain as the bait. (B) Both PIKE-A and PIKE-L interact with GRIP1. HEK293 cells were co-transfected with myc–GRIP1 and various GFP-tagged PIKE-L or PIKE-A constructs. PIKE proteins were immunoprecipitated and the associated GRIP1 was detected (first panel). The expression of GFP proteins (second and third panels) and myc–GRIP1 was also examined (fourth panel). (C) PIKE-L associates with GRIP1 in neurons. Immunoprecipitation using various antibodies was performed as indicated in cultured cortical neurons (21 DIV) and the associated proteins were detected using specific antibody (top panel). Total GRIP1 (middle panel) was detected using antibody against GRIP1. Antibody against the C-terminus of PIKE-L (PIKE (C)) was used to determine the expression of PIKE-L (bottom panel). (D) PIKE-L associates with GRIP1 in rat brain. Immunoprecipitation using indicated antibody was performed in whole brain lysates and the associated PIKE-L was detected by antibody against the N-terminus of PIKE-L (PIKE (N)). (E) Co-localization of PIKE-L and GRIP1 in neurons. Immunofluorescent staining was performed on hippocampal neurons (21 DIV) using anti-PIKE-L C-terminus antibody (PIKE (C)) (green) and anti-GRIP1 antibody (red). Scale bar represents 20 μm. (F) Schematic representation of various GRIP1 deletion truncates used in the in vitro binding assay. (G) Mapping of PIKE-L interaction domain in GRIP1. Various deletion truncates of GRIP1 tagged with bacterial GST were purified and incubated with cell lysates from HEK293 cells expressing HA–PIKE-L. The GST proteins were pulled down and the associated PIKE-L was detected (upper panel). The expression of GST-tagged GRIP1 truncates (asterisked) was also examined (lower panel). (H) Schematic representation of various PIKE-L deletion truncates used in the in vitro binding assay. (I) Mapping of GRIP1 interaction domain in PIKE-L. Various deletion mutants of PIKE-L were expressed and purified from bacteria and incubated with cell lysates from HEK293 cells expressing myc–GRIP1. The GST proteins were pulled down and the associated PIKE-L was detected (upper panel). The expression of GST-tagged PIKE-L truncates (asterisked) was also examined (lower panel). Download figure Download PowerPoint To delineate the PIKE-L-binding site in GRIP1, we performed an in vitro mapping assay. PIKE-L strongly bound to PDZ4 but not other PDZ domains (Figure 1G; Supplementary Figure S1B). To identify the domain in PIKE-L responsible for the PIKE/GRIP1 interaction, we conducted co-immunoprecipitation assay after co-transfecting various deletion mutants of GFP-tagged PIKE-L and myc–GRIP1 into HEK293 cells (Figure 1H). While the GTPase, PH and the ArfGAP domains associated with GRIP1, the proline-rich NTD failed (Figure 1I), suggests that the NTD in PIKE-L is dispensable for the PIKE-L/GRIP interaction. PIKE-L associates with AMPAR Given that GRIP1 is a binding partner of AMPAR (Dong et al, 1997), and PIKE-L interacts with other glutamate receptor family member (Rong et al, 2003), we speculated that PIKE-L, GRIP1 and GluA2 might exist as a functional complex. To test this hypothesis, we performed co-immunoprecipitation in rat brain tissue. A PIKE-L/GRIP1 complex was readily detected no matter anti-PIKE-L or anti-GRIP1 was employed in the immunoprecipitation (Figure 2A, upper and lower panels, second lane). PIKE-L was also co-immunoprecipitated with GluA2 (Figure 2A, lower panel, fourth lane). Since the majority of AMPARs in brain are GluA1/GluA2 heterodimers (Sans et al, 2003), it is reasonable to find that both PIKE-L and GRIP1 were pulled down when anti-GluA1 was utilized (Figure 2A, upper and lower panels, third lane). This was supported by the fact that PIKE-L associated specifically with GluA2 but not GluA1 in HEK293 cells (Supplementary Figure S2A). Similarly, we did not observe any significant binding between PIKE-L and GluA3 in mouse brain lysate (data not shown). These results strongly suggest that PIKE-L, GRIP1 and AMPAR form a complex in brain. Immunofluorescent staining further demonstrated the co-localization (Rr=0.50±0.05, n=3) of PIKE-L and GluA2 in dendrites of cultured hippocampal neurons (Figure 2B). Figure 2.PIKE associates with GluA2. (A) Formation of PIKE-L/GRIP1/GluA2 complex in brain. Immunoprecipitation from whole rat brain lysates using various antibodies was performed and detected with either anti-GRIP1 antibody or antibody against the N-terminus of PIKE-L (PIKE (N)). (B) Co-localization of PIKE-L and GluA2 in neurons. Immunofluorescent staining was performed in hippocampal neurons (21 DIV) using anti-PIKE-L C-terminus antibody (PIKE (C)) (red) and anti-GluA2 antibody (green). Scale bar represents 20 μm. (C) Schematic representation of various PIKE-L deletion truncates used in the in vitro binding assay. (D) The PH domain of PIKE-L is essential for GluA2 interaction. HEK293 cells were transfected with HA–GluA2 and different deletion mutant of myc–PIKE-L. The PIKE-L truncates were immunoprecipitated and the associated GluA2 was detected (top panel). The expression of the GluA2 (middle panel) and PIKE-L (asterisked) (bottom panel) was also examined. (E) Inhibition of PIKE-L/GluA2 interaction by exogenous PIKE-L PH domain. Mouse brain lysates were first incubated with 50 μg recombinant GST protein or GST–PIKE-L PH domain (GST–PH) and the PIKE-L were then immunoprecipitated. The associated GluA2 was then detected (first panel). The expressions of GluA2 (second panel), PIKE-L (third panel) and the GST proteins (fourth and fifth panels) were also examined. (F) PIKE-L enhances GRIP1/GluA2 interaction. HEK293 cells were transfected with HA–GluA2, myc–GRIP1 and various amount of GFP–PIKE-L. Interaction of GRIP1 and GluA2 was detected using immunoprecipitation (first panel). The expression of GluA2 (second panel), GRIP1 (third panel) and PIKE-L (fourth panel) was also examined (bottom panel). Download figure Download PowerPoint To determine the GluA2-interacting motif in PIKE-L, we performed an in vitro mapping assay. The GluA2/PIKE-L interaction in 293 cells was abolished when the PIKE-L PH domain was deleted (Figure 2D), indicating that the PH domain is essential for the interaction. To further demonstrate that the PIKE-L associates with GluR2 through its PH domain, we introduced recombinant GST-tagged PH domain (GST–PH) or GST proteins into the mouse brain lysates. Preincubation with GST–PH but not GST protein alone diminished the association between PIKE-L and GluA2, suggesting that PIKE-L PH domain is necessary for mediating GluA2 binding (Figure 2E). We have also examined the PIKE-L-binding motif in GluA2. WT, PDZ-binding site mutated (myc–R2 SVKE) or C-terminal truncated (myc–R2 1–853) GluA2 (Supplementary Figure S2B) were co-transfected with GFP–PIKE-L into HEK293 cells, followed by co-immunoprecipitation. WT GluA2 and all of the mutants strongly associated with PIKE-L (Supplementary Figure S2C), suggesting that PIKE-L bound to the juxtapose region in the GluA2 C-terminal tail. Next, we tested if the association of PIKE-L and GluA2 depends on GRIP1. In HEK293 cells, PIKE-L interacted with GluA2 in the absence of GRIP1. The association of PIKE-L and GluA2 was not increased and even the expression of GRIP1 was quadruplicated (Supplementary Figure S2D). Therefore, GRIP1 is dispensable for PIKE-L/GluA2 association. However, the association of GluA2 and GIRP1 was augmented in the presence of PIKE-L in a dose-dependent manner (Figure 2F, first panel), indicating that PIKE-L enhanced the binding of GRIP1 to GluA2. PIKE-L is essential for glycine-induced GluA2-associated PI3K activation and GluA2 surface expression Glycine is an obligatory co-agonist of glutamate to fully activate NMDAR (Johnson and Ascher, 1987). It has been reported that glycine application to cultured hippocampal neurons elicited LTP by increasing postsynaptic AMPAR number (Lu et al, 2001). Since GRIP1 is a critical protein in facilitating GluA2 surface expression (Osten et al, 2000), we hypothesized that glycine might induce GluA2/GRIP1 association. Primary neurons were treated with glycine (200 μM) for 3 min followed by a recover period of 10 or 20 min (Man et al, 2003). In these neurons, the association of GRIP1 with GluA2 increased strongly after glycine induction in a time-dependent manner (Figure 3A, first panel). Concomitantly, PIKE-L/GluA2 interaction also increased (Figure 3A, second panel), suggesting that the formation of PIKE-L/GluA2 complex was enhanced upon glycine stimulation. On the other hand, PIKE-L constitutively associated with GRIP1, and their interaction increased slightly after 20 min stimulation of glycine (Figure 3A, third panel). The association of GluA2/GRIP1 or GluA2/PIKE-L was PI3K dependent as inhibition of PI3K by LY294002 abrogated the glycine-induced interactions (Figure 3B, first and third panels). Since GluA2 constitutively associates with PI3K that could be activated by glycine (Man et al, 2003), presumably, activation of GluA2-associated PI3K (PI3KGluA2) may serve as the molecular switch to recruit PIKE-L and GRIP1 to GluA2. Figure 3.PIKE is essential for glycine-induced GluA2-associated PI3K activation and GluA2 surface expression in neurons. (A) Glycine stimulation enhances GluA2/GRIP1 association. Cortical neurons (10 DIV) were treated with glycine (200 μM) for indicated time intervals and the interactions between GRIP1, GluA2 and GRIP1 were determined by immunoprecipitation (first to third panels). The expressions of GluA2 (fourth panel), GRIP1 (fifth panel) and PIKE-L (anti-PIKE-L N-terminus antibody) (sixth panel) under glycine stimulation were also verified. (B) PI3K is critical for GluA2/GRIP1 interaction. Cortical neurons (10 DIV) were pretreated with LY294002 (10 μM) for 1 h followed by 20 min glycine (200 μM) stimulation. The interaction between GRIP1, GluA2 and PIKE-L was determined by immunoprecipitation (first and third panels). PIKE-L level was examined using anti-PIKE-L N-terminus antibody (PIKE (N)) (fifth panel). The expressions of GRIP1 (third panel) and GluA2 (sixth panel) was also verified. (C) PIKE-L facilitates glycine-induced GluA2-associated PI3K activation. Cortical neurons (10 DIV) were infected with control adenovirus (Ctr Ad), adenovirus carrying PIKE-L (Ad-PIKE-L) or adenovirus carrying shRNA against PIKE-L (Ad-shPIKE). After 48 h infections, the neurons were treated with glycine (200 μM) and GluA2 was then immunoprecipitated. The activity of PI3K associated with GluA2 was examined by in vitro PI3K assay. Quantitation of 32P-PIP3 (middle panel) as a representation of PI3K activity was shown (top panel) and the expression of PIKE-L under various adenovirus infections was examined (bottom panel). Results were expressed as mean±s.e.m. (**P<0.01, Student's t-test, n=3). (D) Total PI3K activity could not be enhanced by glycine stimulation. Cortical neurons (10 DIV) were infected with either control adenovirus (Ctr Ad) or adenovirus carrying shRNA against PIKE-L (Ad-shPIKE). After 48 h infection, the neurons were treated with glycine (200 μM) and the total PI3K was then immunoprecipitated and analysed directly by in vitro PI3K assay (top panel). Phosphorylation of Akt was also examined using immunoblotting (middle panel). Total Akt was checked to ensure equal loading (bottom panel) (*P<0.05; **P<0.01; ***P<0.001, Student's t-test, n=3). (E) Overexpression of PIKE-L facilitates glycine-induced cell-surface expression of GluA2 in neurons. Hippocampal neurons (21 DIV) were infected with control adenovirus (Ctr Ad, left panels), adenovirus carrying PIKE-L (Ad-PIKE-L, middle panels) or adenovirus carrying shRNA against PIKE-L (Ad-shPIKE, right panels). After 48 h infection, the neurons were treated with glycine (200 μM), stained with anti-GluA2 antibody under non-permeabilized condition and the cell surface GluA2 were visualized by confocal microscope. (F) Quantitation of cell surface GluA2 on neurons shown in (D) (**P<0.01, ***P<0.001, Student's t-test, n=8). Arbitrary units (A.U.) of GluA2 expression were shown. (G) Biotinylation assay on cell-surface expression of GluA2. Cortical neurons (DIV 10) were infected with control adenovirus (Ctr Ad), adenovirus carrying PIKE-L (Ad-PIKE-L) or adenovirus carrying shRNA against PIKE-L (Ad-shPIKE) for 48 h before glycine (200 μM) treatment. Cell-surface proteins were biotinylated and pulled down by NeutrAvidin beads and the amount of GluA2 was detected using immunoblotting analysis (top panel). The expression of PIKE-L (middle panel) was determined using anti-PIKE-L N-terminus antibody (PIKE (N)). GluA2 level (bottom panel) after adenovirus infection was also examined. Download figure Download PowerPoint PIKE-L coupled mGlu1 to PI3K pathway through binding its adaptor protein Homer (Rong et al, 2003), we thus hypothesized that PIKE-L might be necessary for glycine-induced PI3KGluA2 activation. In control hippocampal neurons, glycine treatment activated PI3KGluA2 (Figure 3C, lanes 1 and 2). Overexpression of WT PIKE-L further enhanced the activity of PI3KGluA2 after glycine stimulation (Figure 3C, lanes 3 and 4). On the other hand, depletion of PIKE-L by shRNA abolished the glycine-provoked PI3KGluA2 activity (Figure 3C, lanes 5 and 6), suggesting that PIKE-L was critical for the glycine-induced PI3KGluA2 activation. As expected, glycine did not elicit total cellular PI3K (PI3Ktotal) activation as compared with control (Figure 3D, lanes 1 and 9). This is in agreement with the previous finding that glycine-induced PI3K activation is unique to PI3KGluA2 but not the PI3Ktotal in cultured neurons (Man et al, 2003). Interestingly, PI3Ktotal activity was enhanced by AMPA stimulation and depletion of PIKE-L abolished this effect (Figure 3D, lanes 5 and 6), suggesting that PIKE-L was also important for the non-receptor-associated PI3K activation induced by AMPA. The AMPA-stimulated PI3K activation was abolished if the neurons were pretreated with AMPAR antagonist CNQX (Figure 3D, lanes 7 and 8). In contrast, NMDA stimulation did not increase the PI3Ktotal activity (Figure 3D, lanes 1 and 3). These results support that PIKE-L is also an important factor to regulate glutamate receptors-mediated PI3Ktotal activation. Next, we sought to determine the effect of PIKE-L overexpression or depletion on GluA2 surface expression by synaptotagmin/GluA2 co-staining (Perestenko and Henley, 2003). Confocal microscopy analysis on hippocampal neurons indicated that the surface GluA2 expression induced by glycine treatment was greatly enhanced if PIKE-L was overexpressed (Figure 3E and F). In contrast, such increase in surface GluA2 expression was significantly reduced in PIKE-depleted neurons (Figure 3E and F; Supplementary Figure S3 for the whole morphology of the infected neurons). The change in GluA2 expression was further confirmed by biotinylation on cell-surface proteins (Man et al, 2007). In agreement with the immunofluorescence staining, the number of surface GluA2 was greatly elevated in primary neurons overexpressing PIKE-L WT (Figure 3G, lanes 2 and 4). PIKE-L KS expressing cells also had elevated surface GluA2 expression after glycine stimulation as compared with control, though they revealed less surface GluA2 than WT PIKE-L (Figure 3G, lanes 4 and 6), suggesting that the GTPase activity might not play a major role in this process. On the other hand, no increase in the surface GluA2 was observed in PIKE-depleted neurons, indicating that PIKE is critical for glycine-induced membrane GluA2 retention (Figure 3G, lanes 7 and 8). Glycine-induced PI3K activity and GluA2 surface expression are abolished in PIKE−/− neurons Next, we examined the effect of PIKE ablation on GluA2/GRIP1 interaction in PIKE knockout (PIKE−/−) mice (Supplemental data; Supplementary Figure S4). Association between GluA2 and GRIP1 was diminished in PIKE−/− brain (Figure 4A, third panel), suggesting that PIKE-L was critical for GRIP1/GluA2 interaction. Moreover, we have determined the amount of GluA2 on the cell surface of neurons in PIKE−/− brain. In the purified plasma membrane of neurons from cortex and hippocampus, expressions of GluA2 were diminished in samples isolated from PIKE−/− mice (Figure 4B). We also determined the effect of PIKE ablation on glycine-induced PI3KGluA2 activity. Glycine stimulation activated the PI3KGluA2 in cultured WT but not the PIKE−/− neurons (Figure 4C, top panel), which is in agreement with the result of shRNA knockdown assay (Figure 3C). Moreover, cell surface GluA2 insertion induced by glycine was also abolished in PIKE−/− neurons (Figure 4D and E; Supplementary Figure S5 for the whole neuron morphology). To further demonstrate the important role of PIKE-L in modulating GluA2 surface expression. We infected PIKE−/− neurons with adenovirus overexpressing PIKE-L to see if GluA2 expression could be rescued. As shown in Figure 4F and G (also see Supplementary Figure S6 for the whole neuron morphology), surface expression of GluA2 was significantly enhanced in PIKE-L-restored PIKE−/− neurons in both basal and glycine-stimulated conditions. The reduced surface GluA2 expression in PIKE−/− neurons was further supported by the results of cell-surface protein biotinylation assay (Figure 4H). Hence, PIKE is indispensable for glycine-induced AMPAR surface insertion and PI3KGluA2 activation. Figure 4.Glycine-induced cell-surface expression of GluA2 is impaired in PIKE−/− neurons. (A) The association of GluA2 and GRIP1 is reduced in PIKE−/− brain. Immunoprecipitations from WT (+/+) and PIKE knockout (−/−) mouse brain lysates using control IgG (first panel), anti-PIKE-L C-terminal antibody (PIKE (C)) (second panel) and anti-GluA2 (third panel) were performed and the associated GRIP1 was detected using anti-GRIP1 antibody. The expression of GluA1 (fourth panel), GluA2 (fifth panel) and GRIP1 (sixth panel). The level of PIKE-L (seventh panel) was also examined using anti-PIKE-L N-terminus (PIKE (N)) antibody. (B) Surface expression of GluA2 is reduced in PIKE−/− brain. Plasma membrane were isolated from the cortex and hippocampus of age-matched (3-month-old) WT (+/+) and PIKE−/− (−/−) mice. The expression of GluA2 was determined by immunoblotting analysis (upper panel). Expression of EPO receptor was also determined as a loading control (lower panel). (C) Glycine-induced GluA2-associated PI3K activity is abolished in PIKE−/− neurons. After treated with glycine (200 μM), the GluA2 in WT (+/+) and PIKE-knockout (−/−) cortical neurons (10 DIV) was immunoprecipitated for PI3K assay (top panel). The expression of PIKE-L (middle panel) and GluA2 (bottom panel) in the cortical neurons were verified using anti-PIKE-L N-terminus (PIKE (N)) antibody and anti-GluA2, respectively. (D) Cell-surface expression of GluA2 under glyc