Profilin II Regulates the Exocytosis of Kainate Glutamate Receptors*

The trafficking of ionotropic glutamate receptors to and from synaptic sites is regulated by proteins that interact with their cytoplasmic C-terminal domain. Profilin IIa (PfnIIa), an actin-binding protein expressed in the brain and recruited to synapses in an activity-dependent manner, was shown previously to interact with the C-terminal domain of the GluK2b subunit splice variant of kainate receptors (KARs). Here, we characterize this interaction and examine the role of PfnIIa in the regulation of KAR trafficking. PfnIIa directly and specifically binds to the C-terminal domain of GluK2b through a diproline motif. Expression of PfnIIa in transfected COS-7 cells and in cultured hippocampal neurons from PfnII-deficient mice decreases the level of extracellular of homomeric GluK2b as well as heteromeric GluK2a/GluK2b KARs. Our data suggest a novel mechanism by which PfnIIa exerts a dual role on the trafficking of KARs, by a generic inhibition of clathrin-mediated endocytosis through its interaction with dynamin-1, and by controlling KARs exocytosis through a direct and specific interaction with GluK2b.

Kainate receptors (KARs) 2 compose a family of ionotropic glutamate receptors, which play an important role in the regulation of synaptic transmission and neuronal excitability (1,2). At variance with the closely related family of AMPA receptors, the main postsynaptic receptors involved in fast synaptic transmission, KARs exert their function by acting at either pre-or postsynaptic sites. Because their physiological functions critically depend on their specific localization as well as their density in these different neuronal compartments, it is important to better understand the mechanisms by which their trafficking is regulated in neurons. KARs are heterotetrameric receptor channels composed of various combinations of five subunits GluK1, GluK2, GluK3, GluK4, and GluK5 (formerly referred as GluR5, GluR6, GluR7, KA1, and KA2, respectively). The diversity of KARs is increased by the existence of splice variants for GluK1, GluK2, and GluK3 subunits. KAR subunits isoforms display different levels of expression at the plasma membrane depending on the alternative splicing of their C terminus (3)(4)(5). When expressed as homomers in heterologous cells or in cultured neurons, GluK2a is highly addressed to the plasma membrane, whereas GluK2b is present at low levels (3,4,6). The low level of expression of homomeric GluR6b can be explained by restricted export of the subunit from the endoplasmic reticulum (3). Native KARs likely exist as heteromers of different GluK subunits, as well as of different splice variants of the same GluK subunit, as for instance, in the case of GluK2a/GluK2b heteromers, which form native KARs (7). Oligomerization of KAR subunits plays a major role in their surface expression (4,8). The regulation of ionotropic glutamate receptors trafficking to the cell surface is probably controlled by a combination of endoplasmic reticulum retention and export signals, as well as by domain-specific protein interactors. KAR subunits and splice variants diverge in their cytoplasmic C-terminal region, which opens the possibility that their trafficking and function are differentially regulated by proteins that interact specifically with subunit specific domains (7). GluK2a and GluK2b interact with two different sets of cytosolic proteins partners, some of which have been identified by proteomic analysis (7). Most reports have concentrated on the regulation of trafficking of homomeric GluK2a receptors by PDZ domain binding proteins (9), BTB-Kelch protein, and KRIP6 (10) and have not taken into consideration the fact that both GluK2a and GluK2b splice variants likely co-assemble in the brain.
Among proteins found to interact with GluK2b in a proteomic screen, profilin IIa (PfnIIa) appears as an interesting candidate for activity-dependent regulation of KARs (7). Mammalian profilins are actin-binding proteins, highly conserved with respect to their affinities for G-actin, poly-L-proline, and phosphoinositides. Profilins act as regulators of various cellular processes such as cytoskeletal dynamics and membrane transport (11). In mammals, several profilin isoforms have been identified. Whereas profilin I (PfnI) is ubiquitously expressed, PfnIIa is specifically expressed in the brain (12) and is a necessary element in key steps of neuronal differentiation and synaptic plasticity (13). PfnIIa interacts with proteins implicated in membrane trafficking, i.e. by interacting with and controlling dynamin-1 activity. Overexpression of PfnIIa inhibits endocytosis, whereas the lack of PfnIIa in neurons results in an increase in endocytosis and membrane recycling (14). Interestingly, profilins are targeted to dendritic spines after robust synaptic activation suggesting a potential role in synaptic plasticity (15).
Here, we have studied the potential role of the interaction between GluK2b and PfnIIa in the trafficking of KARs composed of GluK2a and GluK2b. We describe the specific interaction of PfnIIa to a diproline motif in GluK2b, and we show that PfnIIa controls membrane trafficking of heteromeric KARs in heterologous cells and in hippocampal neurons. Our results indicate that PfnIIa acts at two different levels, as a general regulator of clathrin-mediated endocytosis through its interaction with dynamin-1 and by controlling exocytosis of KARs through a direct interaction with the diproline motif of GluK2b.

EXPERIMENTAL PROCEDURES
cDNA Constructs-Myc-GluK2a and Myc-GluK2b were described in Ref. 16. Myc or superecliptic pHluorin tobacco etch virus (SEP-TEV) sequences were introduced after the signal sequence in the GluK2 cDNA by PCR. Site-directed mutagenesis was performed using the QuikChange XL kit (Stratagene). SEP-TEV (SGGSGGDYDIPTTENLYFQGELK-TVDAD) was amplified by PCR introducing appropriate restriction sites for subcloning (17). The Myc epitope from Myc-GluK2a was exchanged by pHluorin-TEV by subcloning. The C terminus region of SEP-TEV-GluK2a was replaced by the C terminus of either GluK2b or GluK2b/AA. CDNAs were sequenced and expressed in COS-7 cells to verify molecular weight by Western blot analysis with a C-terminal antibody when available.
Immunocytochemistry-COS-7 cells were transfected with cDNAs using the FuGENE kit (Roche Applied Science, Meylan, France). Cultured mouse hippocampal neurons were obtained as described previously (3) from mouse pups derived from PfnIIa knock-out mice (13) and transfected with Lipofectamine 2000 (Roche Applied Science). For surface labeling, cells were incubated for 30 min at 4°C with primary antibodies (1/500 dilution) in culture media and immediately fixed with 4% paraformaldehyde, 4% sucrose for 15 min at 37°C. For intracellular labeling, after fixation, cells were permeabilized with 0.3% Triton X-100 for 2 min and rinsed in PBS/ 0.3% BSA. Primary antibodies were then incubated for 30 min at 20°C, washed with PBS/BSA, and incubated with the secondary fluorescent antibodies (anti-mouse antibody, Alexa Fluor 568 and anti-rabbit antibody, Alexa Fluor 488) for 30 min at 20°C and extensively washed with PBS/BSA. Coverslips were then mounted with VECTASHIELD (Vector Laboratories).
Endocytosis Experiments-After 24 h of expression, COS-7 cells transfected with the appropriate cDNAs were incubated for 30 min at 37°C with the anti-Myc antibody and with transferrin Alexa Fluor 488. Cells were acid washed for 2 min at 20°C with cultured media adjusted at pH 2.2. Cells were then fixed, permeabilized, and incubated with secondary fluorescent antibodies (anti-rabbit antibody, Alexa Fluor 568) for 30 min at 20°C.
Exocytosis Experiments-COS-7 cells or hippocampal neurons from profilin knock-out mice were transfected with SEP-TEV-GluK2b and GluK2a or SEP-TEV-GluK2b/AA and GluK2a with the different PfnIIa cDNAs. Neurons were transfected at 6 days in vitro on 5 coverslips. Experiments were performed after 24 h of expression. Coverslips were incubated with culture medium without serum containing 300 units/ml of TEV enzyme (Invitrogen) for 10 min at 37°C followed by 10 min at 20°C. Cells were then either directly fixed (corresponding to time 0 in the figures) or further incubated in culture medium at 37°C for the time indicated in the figures. After fixation at different times, cells were incubated with an anti-GFP monoclonal antibody for 30 min at 20°C (labeling of extracellular receptors). Cells were permeabilized and incubated with a polyclonal anti-GFP antibody for 30 min at 20°C. Cells were then incubated with anti-mouse coupled to Alexa Fluor 568 and anti-rabbit coupled to Alexa Fluor 488. For image acquisitions, the exposure settings and gain were kept the same to compare the different experimental conditions. Quantity of exocytosed receptors was measured as the ratio between extracellular (EC) receptors and intracellular (IC) receptors.
Image Analysis-For endocytosis experiments, cells were imaged with confocal microscope Leica DMR TCS SP2 AOBS on an upright stand (Leica Microsystems, Mannheim, Germany), using an objective HCX Plan Apo CS 40ϫ oil. The lasers used were argon (458 nm), green helium-neon (543 nm), and red helium-neon (633 nm). In the other experiments, cells were visualized with an upright epi-fluorescence microscope (Leica DMR, Leica Microsystems, Wetzlar, Germany). Images were acquired with a CoolSnap camera (Roper Scientific) and Metamorph software and analyzed with a macro written in MATLAB. Statistical analyses were conducted with GraphPad PRISM software as indicated in the figure legends.
Immunoprecipitation-Immunoprecipitation was performed as described previously (16). Immunoprecipitated proteins were directly eluted with SDS gel load buffer, run on polyacrylamide gels, and transferred on membrane for immunoblotting. Gels were quantified with a Syngene apparatus (catalog no. CG2XE/D2-E from Ozyme).
Peptide Affinity Chromatography-A synthetic peptide corresponding to the C-terminal domain of GluK2b was coupled via its N-terminal extremity to an activated CH-Sepharose 4B (Amersham Biosciences), according to the manufacturer's instructions. PfnI or PfnIIa were expressed in bacteria, purified, and incubated overnight with the corresponding resins. After extensive washes, samples were directly loaded on SDS gels and Western blotted with antibodies directed against PfnI or PfnIIa.
GST Pulldown-GST pulldown experiments were performed using either the GluK2a C terminus domain (70 amino acids), the GluK2b C terminus domain (27 amino acids), or the GluK2b C terminus domain with mutations of two prolines into alanines. Cytosolic protein extracts from mouse brain or COS-7 cells were incubated with beads coupled to GST, GST-Cterm-GluK2a, GST-Cterm-GluK2b, or GST-Cterm-GluK2b/AA overnight at 4°C. Beads were washed, suspended in loading buffer, run on SDS gels, and transferred onto membranes for immunoblotting.
Surface Biotinylation-One day after transfection, COS-7 cells were washed with PBS (pH 8.0) and incubated with 0.5 mg/ml EZ-Link TM sulfo-NHS-S-LC-biotin (Pierce) in PBS for 30 min at 4°C. After three washes, cells were scrapped in a lysis buffer containing 25 mM Hepes, 150 mM NaCl, 1% Triton X-100, and a mix of protease inhibitors. After centrifugation, the supernatant was incubated with immobilized streptavidin bead agarose overnight at 4°C and washed extensively. Samples were analyzed by Western blot with antibodies corresponding to the proteins of interest.

RESULTS
Direct Interaction between PfnIIa and GluK2b-We have previously shown that PfnIIa interacts with GluK2b in the brain by peptide pulldown and immunoprecipitation experiments (7). To further characterize the interaction between PfnIIa and GluK2b, we first determined whether this interaction was direct or whether it required an intermediate protein partner (Fig. 1). For this, we incubated either PfnI or PfnIIa proteins produced in Escherichia coli with GluK2a and GluK2b C-terminal peptides linked to Sepharose resins (Fig.  1A). This experiment indicated that the last 15 amino acids of GluK2b representing its C-terminal domain directly bind Pf-nIIa. This interaction was specific for PfnIIa because PfnI did not bind to GluK2b (as well as to GluK2a, data not shown) and was specific for the C-terminal domain of GluK2b as Pf-nIIa did not bind to the C-terminal domain of GluK2a. In control experiments, we found that GluK2a bound to the first two PDZ domains of PSD-95 (data not shown) (16).
Profilins interact with a wide variety of proteins via their poly-L-proline binding site. GluK2b C-terminal domain contains two consecutive prolines that could correspond to the consensus binding sequence to PfnIIa (Fig. 1B). To determine whether these amino acids are responsible for the binding of GluK2b to PfnIIa, we replaced the two prolines by alanines in GluK2b (GluK2b/AA). We expressed Myc-GluK2a, Myc-GluK2b, and Myc-GluK2b/AA with PfnIIa in COS-7 cells and tested interaction of PfnIIa by a co-immunoprecipitation assay (Fig. 1C). PfnIIa was co-immunoprecipitated with myc-GluK2b, but it was not detected in the Western blots after immunoprecipitation of Myc-GluK2a or of Myc-GluK2b/AA. These results confirm the specific interaction of PfnIIa with the C-terminal domain of GluK2b, but not of GluK2a, and show that the diproline motif in the C-terminal domain of GluK2b is essential for its direct interaction with PfnIIa.
We confirmed these results using immunocytochemistry experiments in COS-7 cells expressing the different combinations of subunits with or without PfnIIa (Fig. 2D, n ϭ 4). Surface labeling of receptors was measured and normalized to the level of expression for each cell. This level was measured as the GFP signal corresponding either to pEGFP or to PfnIIa-GFP plasmids. As expected, the labeling corresponding to extracellular Myc-GluK2b expressed alone was very low (Ͻ 1%) and was undetectable in cells co-expressing PfnIIa. The relative amount of Myc-GluK2a at the surface was not affected by co-expression of PfnIIa (32 Ϯ 11% without PfnIIa, 25 Ϯ 10% with PfnIIa, n ϭ 40 cells). However, PfnIIa decreased the relative amount of Myc-GluK2a when expressed in combination with GluK2b (34 Ϯ 4% without PfnIIa, 21 Ϯ 2% with PfnIIa, n ϭ 40 cells). The decrease of surface expressed KARs was even more pronounced when monitoring the Myc-GluK2b subunit in the heteromer (15 Ϯ 2% without PfnIIa, 6 Ϯ 1% with PfnIIa, n ϭ 70 cells) (Fig. 2D, right panel). PfnIIa had no effect on the plasma membrane expression of receptors composed of GluK2a/Myc-GluK2b/AA subunits (15 Ϯ 1% without PfnIIa, 18 Ϯ 2% with PfnIIa, n ϭ 50 cells). These results indicate that PfnIIa decreases the amount of homo-and heteromeric KARs containing GluK2b at the plasma membrane through a direct interaction between GluK2b and PfnIIa in COS-7 cells.
To study whether PfnIIa also affected surface trafficking of KARs in hippocampal neurons in culture, we took advantage of cultures derived from PfnIIa knock-out mice (13). As previously described in wild-type hippocampal neurons, endogenously expressed GluK2a was able to heteromerize with GluK2b and promoted forward trafficking of GluK2b to the neuronal plasma membrane (3). We thus expressed Myc-GluK2b (or Myc-GluK2b/AA) with or without PfnIIa in hippocampal neurons and performed live labeling of surface Myc-GluK2b (Fig. 3A). We quantified the level of extracellular KARs and normalized to the GFP signal as for COS-7 cells (Fig. 3B). Overexpression of PfnIIa markedly decreased the plasma membrane expression of GluK2b (Myc-GluK2b alone, 50 Ϯ 3%; with PfnIIa, 8 Ϯ 1%). PfnIIa did not affect the surface expression of the mutant form of Myc-GluK2b, which does not bind PfnIIa (Myc-GluK2b/AA alone: 17 Ϯ 3% with PfnIIa, 21 Ϯ 3%). However, the plasma membrane expression of Myc-GluK2b/AA was significantly lower than that of the wild-type Myc-GluK2b (Myc-GluK2b/AA, 18 Ϯ 3%, n ϭ 10 cells; Myc-GluK2b, 50 Ϯ 3%, n ϭ 50 cells). When Myc-GluK2b was expressed in cultured neurons from wild-type mice, the co-expression of PfnIIa did not change its surface expression (Myc-GluK2b alone, 16 Ϯ 3%; with PfnIIa, 14 Ϯ 2%) (data not shown). Overall, these results indicate that PfnIIa decreases surface expression of GluK2b containing KARs in neurons.
Impact of PfnIIa on Endocytosis of KARs-The decreased surface expression of GluK2b-containing KARs in the presence of PfnIIa could be assigned either to an inhibition of forward trafficking mechanisms or to facilitation of endocytosis. Contrary to this hypothesis, PfnIIa is known to inhibit endocytosis via its interaction with dynamin-1 (14). PfnIIa associates with dynamin-1 via its C-terminal proline-rich domain and thereby interferes with the assembly of the endocytic machinery (14). To understand how this generic inhibition of endocytosis by PfnIIa affects GluK2b trafficking, we used two mutants of PfnIIa (S134Y (PfnIIaY) and S138D (PfnIIaD) mutants), which were previously characterized as lacking interaction with dynamin-1 and fail to inhibit the clathrin-mediated endocytosis of transferrin (14).
We checked whether these mutations affected the binding of PfnIIa with GluK2b subunit (Fig. 4). We examined the binding of dynamin-1 and GluK2b to PfnIIa mutants, by GST pulldown experiments (Fig. 4A, n ϭ 3). As expected, we did not observe any pulldown of dynamin-1 with the PfnIIa mutants, after incubation of the GST beads with a brain extract and quantification of anti-dynamin-1 labeling in Western blots. We next incubated extracts from COS-7 cells expressing the Myc-GluK2b subunit with GST beads. We found that PfnIIaY or PfnIIaD bind GluK2b, as well as wild-type PfnIIa (n ϭ 5). Conversely, we performed a peptide pulldown assay with the C terminus peptide of GluK2b incubated with extracts of COS-7 cells expressing either PfnIIa, PfnIIaY, or Pf-nIIaD (Fig. 4B, n ϭ 2). We observed that the same amount of PfnIIa and PfnIIa mutants were pulled down with the GluK2b C terminus peptide. These results were confirmed by immunoprecipitation experiments on extracts from COS-7 cells expressing Myc-GluK2b and PfnIIa (Fig. 4C, n ϭ 3). Myc-GluK2b was immunoprecipitated with an anti-Myc antibody, and Western blots were probed with an antibody against PfnIIa. Similar amounts of PfnIIa, PfnIIaY, or PfnIIaD were co-immunoprecipitated with the Myc-GluK2b subunit. Thus, although PfnIIaY and PfnIIaD fail to interact with dynamin-1, these two mutations do not appear to affect the binding of PfnIIa to the GluK2b subunit. PfnIIa associates with dynamin-1 via its C-terminal proline-rich domain and thus probably associates with GluK2b via another domain.
Using these mutants, we could then differentiate between a specific function for the PfnIIa-GluK2b interaction and a more general effect of PfnIIa on endocytosis of membrane proteins. We first performed biotinylation experiments in COS-7 cells expressing heteromeric GluK2a/Myc-GluK2b receptors with or without PfnIIaY or PfnIIaD (Fig. 5A, n ϭ 3). Extracellular and intracellular pools of receptors were purified on streptavidin beads, and samples were run on SDS gels, blotted with anti-Myc antibody, and quantified. Expression of PfnIIaY or PfnIIaD decreased the relative surface levels of GluK2b containing KARs to similar levels as with PfnIIa (Myc-GluK2b in GluK2a/GluK2b receptor expressed with PfnIIa, 18 Ϯ 4%; with PfnIIaY, 18 Ϯ 6%; with PfnIIaD, 16 Ϯ 5%). Quantification of the relative amount of surface receptors by immunocytochemistry in COS-7 cells expressing PfnI-IaY or PfnIIaD yielded similar results (Fig. 5B). By both biochemical and immunocytochemical methods, we found that mutated forms of PfnIIa did not affect membrane trafficking  Dyn1) and Myc-GluK2b (mK2b) on GST-PfnIIa beads and its mutated forms (WT; Y, S134Y; D, S138D). sn, starting material. Beads were incubated with extracts of COS-7 cells expressing either dynamin-1 or Myc-GluK2b. Western blots were incubated with anti-GST, anti-dynamin-1, or anti-Myc antibodies. The interaction was quantified as the ration of bound protein reported to the quantity of bait. This value is then normalized using a value of 1 for the WT (n ϭ 3). PfnIIa WT interacts with dynamin-1 and GluK2b, whereas both PfnIIa mutants only pull down the Myc-GluK2b. B, peptide pulldown on the GluK2b C terminus peptide. Western blots were incubated with an anti-PfnIIa antibody (n ϭ 3). The GluK2b C terminus peptide equally pulled down the three forms of PfnIIa. C, immunoprecipitation of PfnIIa wild-type and its mutants in COS-7 transfected cells. GluK2b and each PfnIIa were expressed in COS-7 cells. Extracts were immunoprecipitated with an anti-Myc antibody, and Western blots were probed with anti-Myc and anti-PfnIIa antibodies. Quantification of the Western blots from three different experiments. The same quantity of PfnIIa or PfnIIa mutants was co-immunoprecipitated with GluK2b subunit. DECEMBER 17, 2010 • VOLUME 285 • NUMBER 51 of receptors containing GluK2b/AA, which does not bind Pf-nIIa. We repeated these experiments in hippocampal neurons cultured from PfnIIa knock-out mice expressing the Myc-GluK2b or Myc-GluK2b/AA with or without the mutated forms of PfnIIa (Fig. 5C). Again, we found that expression of the mutated forms of PfnIIa, that still bind the GluK2b subunit but not dynamin-1, decreased the surface expression of KARs containing GluK2b. If inhibition of endocytosis by Pf-nIIa was involved in the regulated membrane expression of KARs, we may have expected a greater decrease of surface receptors in the presence of PfnIIaY or PfnIIaD. Our results suggest that PfnIIa does not inhibit clathrin-dependent endocytosis of GluK2b.

Profilin II Controls Trafficking of KARs
Endocytosis Pathways of KARs Co-expressed with PfnIIa-We directly tested this hypothesis by examining the mechanisms and the rate of endocytosis of GluK2b-containing receptors expressed in COS-7 cells and the influence of PfnIIa (Fig. 6). Transferrin is internalized through the classical receptor-mediated endocytosis pathway involving clathrincoated pits and dynamin (19). We analyzed the pathway for endocytosis of KARs by incubating the cells for 30 min at 37°C with transferrin labeled with Alexa Fluor 488 (Fig. 6, A  and B). When expressed alone, GluK2b was detected in intracellular compartments as big aggregates corresponding to endocytic events, and it was in great part colocalized with transferrin (76 Ϯ 1%, n ϭ 12 cells). Intracellular aggregates containing GluK2b also contain transferrin, indicating that this subunit is recycled via clathrin-coated pits. However, quantification of the rate of endocytosis was difficult because of the very low level of GluK2b expressed at the plasma mem-brane. We thus evaluated the colocalization of transferrin and GluK2b that compose heteromeric GluK2a/GluK2b receptors. When co-expressed with GluK2a in COS-7 cells, MycGluK2b was colocalized with transferrin in 50 Ϯ 4% (n ϭ 15) of intracellular aggregates. Co-expression of either PfnIIa or PfnIIaY did not change the level of colocalization (48 Ϯ 4 and 49 Ϯ 5%, respectively, n ϭ 15 cells). When expressed with GluK2a, Myc-GluK2b/AA also colocalized to the same extent with transferrin (48 Ϯ 2% without PfnIIa; 55 Ϯ 3% with PfnIIa). Overall, these results suggest that GluK2b (both as a homomer or as part of a heteromeric receptor) is endocytosed through the same pathway as transferrin, hence clathrin-mediated endocytosis. Expression of PfnIIa does not qualitatively change the internalization pathway.
Because PfnIIa has been shown to be a negative regulator of clathrin-mediated endocytosis (14), we next measured the rate of endocytosis of receptors composed of GluK2a/Myc-GluK2b in COS-7 cells and its regulation by PfnIIa (Fig. 6C). To do this, we incubated COS-7 cells expressing Myc-tagged receptors (Myc-GluK2b in heteromeric GluK2a/Myc-GluK2b, Myc-GluK2b/AA in heteromeric GluK2a/Myc-GluK2b/AA, or Myc-GluK2a in Myc-GluK2a as a control) for 30 min at 37°C with an anti-Myc antibody. Receptors remaining at the cell surface were then labeled with a secondary antibody conjugated with Alexa Fluor 488. Cells were then permeabilized, and a secondary Alexa Fluor 568 antibody was used to label internalized receptors. The relative amount of internalized versus surface receptors was decreased in cells co-expressing PfnIIa (without PfnIIa 47 Ϯ 3%, n ϭ 50 cells; with PfnIIa 24 Ϯ 2%, n ϭ 50 cells) but not PfnIIaY in which dynamin-1 binding is abrogated (43 Ϯ 2%, n ϭ 50 cells). Because this mutant binds GluK2b as efficiently as wild-type PfnIIa, the direct interaction between PfnIIa and GluK2b is not likely to be involved in decreased endocytosis of GluK2b. In addition, the internalization of GluK2b/AA, which does not bind PfnIIa, is similarly decreased in the presence of PfnIIa (without PfnIIa 33 Ϯ 3%, n ϭ 50 cells; with PfnIIa 17 Ϯ 2%, n ϭ 50 cells; with PfnIIaY 30 Ϯ 2%, n ϭ 50 cells). Finally, GluK2a expressed as a homomer is not endocytosed through a clathrin-mediated pathway (no colocalization with transferrin) (data not shown), and its level of internalization (18 Ϯ 3%, n ϭ 35) is not affected in cells co-expressing either PfnIIa (16 Ϯ 2%, n ϭ 40) or PfnIIaY (21 Ϯ 2%, n ϭ 20). Interestingly, GluR2b/AA, which lacks binding to PfnIIa, displays decreased endocytosis but not increased surface expression in presence of PfnIIa, as would be expected. This may be in some way related to the lower surface expression of GluK2b/AA as compared with GluK2b, for reasons that are not clear. Overall, these results strongly suggest that the decreased endocytosis of GluK2bcontaining KARs by PfnIIa probably reflects a general inhibition of clathrin-mediated endocytosis by PfnII, which is not specific for GluK2b.
Impact of GluK2b-PfnII Interaction on Exocytosis of KARs-These previous results raise the question of how the interaction between GluK2b and PfnIIa affects KAR trafficking. The generic inhibitory effect of PfnIIa on clathrin-mediated endocytosis should lead to an increased amount of surface GluK2b. This contradicts our observation that PfnIIa decreases amount of surface-expressed GluK2b, when expressed either in a homomeric or heteromeric form. We tested the hypothesis that PfnIIa inhibited the exocytosis of GluK2b-containing receptors. To do this, we tagged GluK2b with SEP, and we inserted a tobacco etch virus (TEV) protease cleavage site after the signal sequence of either GluK2b (SEP-TEV-GluK2b) or GluK2b/AA (SEP-TEV-GluK2b/AA) proteins (Fig. 7). Superecliptic pHluorin is a pH-sensitive derivative of green fluorescent protein that does not emit any fluorescent signal under acidic conditions, preventing fluorescence emission from acidic intracellular compartments (20). The protease TEV cuts proteins exposed at the cell surface that display a specific proteolysis site composed of seven amino acids flanked by spacers as described (17) and does not cut endogenous extracellular proteins. After 24 h of expression of the different receptors and PfnIIa, the TEV protease was directly applied on live cells to remove the SEP from surface expressed GluK2b (time 0). Cells were then washed in culture medium, returned at 37°C, and fixed after different times of recovery. Only proteins newly exported at the cell surface during these time periods were labeled with a monoclonal anti-GFP (and a secondary anti-mouse antibody coupled to Alexa Fluor 568) incubated before permeabilization. Intracellular SEP-TEV subunits are labeled with a polyclonal anti-GFP (secondary anti-rabbit coupled to Alexa Fluor 488). We then quantified the amount of extracellularly expressed receptors relative to the amount of intracellular receptors. Results are expressed as percentage of recovery from a value of 100% corresponding to uncut receptors. To evaluate the efficiency of the cutting by TEV protease, we labeled the cells before and after incubation by the enzyme.
In COS-7 cells, ϳ80 -90% of SEP-TEV subunits was cut after incubation with the TEV protease. In the absence of Pf-nIIa, the recovery SEP-TEV-GluK2b/GluK2a KARs at the plasma membrane of transfected COS-7 cells was completed within 1 to 2 h (recovery at 1 h, 86 Ϯ 7%, n ϭ 45 cells; at 2.5 h, 120 Ϯ 8%, n ϭ 40 cells) (Fig. 7A). When PfnIIa was co-expressed with SEP-TEV-GluK2b/GluK2a KARs, only 30% of KARs was recovered at the plasma membrane after 2.5 h (recovery at 1 h, 31 Ϯ 4%, n ϭ 45 cells; at 2.5 h, 32 Ϯ 5%, n ϭ 40 cells). The level of recovery was 60% after 3.5 h (recovery at 3.5 h, 63 Ϯ 4%, n ϭ 20 cells). We next verified that the decreased rate of exocytosis was directly linked to the specific interaction between PfnIIa and GluK2b and not to a more general effect of PfnIIa through unknown interactions with cytosolic proteins. We first observed that PfnIIa did not affect the rate of recovery of fluorescence SEP-TEV-GluK2b/AA, which does not bind PfnIIa (recovery at 1 h without PfnIIa, 74 Ϯ 5%, n ϭ 45 cells; with PfnIIa, 86 Ϯ 5%, n ϭ 40 cells) (Fig.  7B). Interaction of PfnIIa with the endocytic machinery did not influence the rate of recovery of fluorescence at the cell surface. Indeed, PfnIIaY in which dynamin-1 binding is abrogated but still interacts with GluK2b, affected exocytosis of SEP-TEV-GluK2b/GluK2a at the same rate as PfnIIa (recovery at 1 h, 23 Ϯ 4%, n ϭ 40 cells; at 2.5 h, 38 Ϯ 5%, n ϭ 40 cells) (Fig. 7A). The fact that the rate of recovery reaches similar values after 2.5 h with PfnIIa and PfnIIaY may reflect the fact that exocytosis of GluK2a/GluK2b receptors is fully blocked by PfnIIa.
To verify whether exocytosis, in general, is affected by the expression of PfnIIa, we expressed SEP-TEV-GluK2a alone (Fig. 7C). GluK2a is readily expressed at the surface due to efficient targeting of a forward trafficking motif. Without Pf-nIIa, the recovery of GluK2a at the plasma membrane was fast (recovery at 0.5 h, 94 Ϯ 7%, n ϭ 30 cells). Addition of PfnIIa did not change the recovery time of SEP-TEV-GluK2a (recovery at 0.5 h, 103 Ϯ 10%, n ϭ 30 cells).
We then repeated these experiments in cultured neurons derived from PfnII knock-out mice (Fig. 7, D and E). We cotransfected GluK2a with the different GluK2b constructs as indicated, and we quantified exocytosis of KARs by evaluating recovery from TEV cleavage after 0.5 to 2.5 h. The recovery of SEP-TEV-GluK2b/GluK2a KARs at the plasma membrane was not complete after 2.5 h (recovery at 1 h, 47 Ϯ 10%, n ϭ 14 cells; at 2.5 h, 69 Ϯ 10%, n ϭ 18 cells) (Fig. 7D). The extent of recovery was markedly reduced when PfnIIa was co-expressed with SEP-TEV-GluK2b/GluK2a (recovery at 1 h, 26 Ϯ 8%, n ϭ 13 cells; at 2.5 h, 20 Ϯ 12%, n ϭ 11 cells). Co-expression of PfnIIaY, which does not interact with the endocytic machinery, also affected recovery of SEP-TEV-GluK2b/ GluK2a to the same extent as PfnIIa (recovery at 1 h, 23 Ϯ 6%, n ϭ 7 cells; at 2.5 h, 24 Ϯ 5%, n ϭ 7 cells) (Fig. 7D). The recovery rate of fluorescence of SEP-TEV-GluK2b/AA, which does not bind to PfnIIa, was not affected by co-expression with PfnIIa (recovery at 2.5 h without PfnIIa, 61 Ϯ 7%, n ϭ 12 cells; with PfnIIa, 53 Ϯ 9%, n ϭ 7 cells) (Fig. 7E), indicating  (K2b) with GluK2a, SEP-TEV-GluK2b/AA, with GluK2a (K2a) or SEP-TEV-GluK2a, without or with PfnIIa or the S134Y mutant), were incubated with the TEV enzyme (10 min at 37°C followed by 10 min at 20°C) and then returned at 37°C for different recovery times as indicated. Cells were fixed and labeled for extracellular receptors with a monoclonal anti-GFP antibody at saturating concentration (1/300). After permeabilization, intracellular KARs were labeled with an anti-GFP polyclonal antibody. Secondary antibodies were anti-mouse Alexa Fluor 568 (red, EC) and antirabbit Alexa Fluor 488 (green, IC). The level of exocytosed KARs was performed as the ratio between exocytosed receptors on total intracellular receptors (n ϭ 50 cells for each conditions, three experiments for COS-7 cells, n ϭ 7 to 25 cells for hippocampal neurons, three experiments). Results are expressed as percentage of recovery; 100% corresponds to the ratio EC/IC before adding the TEV protease, representing the uncut KARs. that the inhibition of exocytosis was due to the specific interaction between GluK2b and PfnIIa.

DISCUSSION
In the present study, we demonstrate that the KAR subunit GluK2b directly and specifically interacts with PfnIIa through a diproline motif localized at the C-terminal domain of the KAR subunit. This interaction leads to decreased surface expression of GluK2b, either expressed as a homomer or as part of heteromeric GluK2a/GluK2b KARs, both in transfected COS-7 cells and in cultured hippocampal neurons derived from PfnIIa knock-out mice. We further provide mechanistic explanations for the regulation of KAR trafficking by PfnIIa. The general effect of PfnIIa on endocytosis, which was described previously (14), is also found for KARs composed of the GluK2b subunit that are endocytosed by clathrin-coated pits. Direct interaction of GluK2b with PfnIIa leads to a KARspecific inhibition of exocytosis. Taken together, our results provide evidence that PfnIIa, an actin-binding protein that is driven to dendritic spines in an activity-dependent manner, controls exocytosis of a major population of KARs composed of the GluK2b splice variant.
The interaction between PfnIIa and the GluK2b subunit was identified through a proteomic screen based on peptide and GST pulldown experiments and on indirect immunoprecipitation of KARs from brain of Myc-GluK2a transgenic mice (7). As for other KAR-interacting proteins identified in the same screen, it is important to ascertain whether there is a direct binding between the C-terminal domain of GluK2b (and GluK2a) or whether this is an indirect interaction. Here, we demonstrate with a peptide pulldown approach and purified PfnIIa produced in E. coli that this interaction is direct, and we identify a diproline motif in the C-terminal domain of GluK2b necessary for this interaction. This is the first direct interactor of the GluK2b C-terminal domain to be firmly identified. Conversely, Y2H screens have provided evidence for a direct interaction between GluK2a and PDZ domaincontaining proteins (PSD-95, syntenin, GRIP, and PICK1) (21,22), through the last four amino acids of GluK2a. In addition, two proteins of the BTB-Kelch family, actinfilin and KRIP6, bind to the C-terminal domain of GluK2a (10, 23), although the exact sequence of interaction has not been identified. Finally, a recent study has indicated that Neto2, identified by a Y2H screen acts as an auxilliary protein for KARs (24). Given the variety of proteins directly or indirectly interacting with KARs, a number of studies have attempted to understand how they control either the functional properties or trafficking of KARs (reviewed in Ref. 9).
Here, we show that PfnIIa has a marked influence on the amount of GluK2b-containing KARs at the cell surface, in heterologous systems and in cultured hippocampal neurons. Whereas GluK2a is readily expressed at the surface due to efficient targeting of a forward trafficking motif, GluK2b is only detected at very low levels at the membrane (3,6). In the case of homomomeric GluK2b, one may question the functional significance of further decreasing the level of surface receptors. However, an important point of the present study is that PfnIIa, through its binding to the C-terminal domain of GluK2b, markedly decreases surface expression of GluK2a when it is co-assembled with GluK2b. This finding provides additional support to the notion that the heteromerization of the two splice variants of GluK2 (7), which only differ in their cytoplasmic C-terminal domain, is highly significant for the regulation of trafficking of KARs to and from the plasma membrane. This heteromerization was also shown to be important for the physiological regulation of KARs (7). This process could certainly be extended to other members of the glutamate receptor family, as, for instance, the NR1 subunits, which exist in eight splice variants derived from a single gene (25). The possibility that two distinct splice variants of NR1 co-exist in an NMDA receptor complex and whether this has important impact on receptor trafficking has not yet been addressed.
Profilins were described originally as actin-binding proteins, but are now regarded as regulators of complex cellular processes, including clathrin-mediated endocytosis (11). This regulation depends on an interaction of PfnIIa with dynamin-1 and appears relevant for the control of membrane trafficking of receptors. However, PfnIIa was shown to inhibit endocytosis in neurons, which is in apparent contradiction with our finding that PfnIIa decreases the level of surface expressed KARs. To provide insight into the mechanisms of action of PfnIIa, we took advantage of mutant forms of PfnIIa, which do not interact with the dynamin-1 pathway but still interact with GluK2b, and we examined whether PfnIIa regulated the endocytosis of GluK2b receptors. Interestingly, GluK2b homomers and GluK2b/GluK2a heteromers, but not GluK2a, co-localize with transferrin in endocytosis experiments, indicating that endocytosis of GluK2-containing receptors is mediated by a clathrin-dependent pathway. As in the case of transferrin, PfnIIa effectively decreases the rate of endocytosis of GluK2b-containing receptors, in contrast with mutants of PfnIIa that fail to interact with this endocytic machinery (14) but still bind to GluK2b. The inhibition of endocytosis of GluK2b by PfnIIa does not require direct and specific binding of PfnIIa to GluK2b but appears to rely on a generic regulation of clathrin-mediated endocytosis.
This down-regulation does not match with decreased levels of GluK2b-containing KARs at the cell surface. In fact, we show that the regulation of endocytosis appears to be masked by the prevailing reduction of exocytosis of GluK2b-containing receptors that relies on the specific interaction between GluK2b and PfnIIa. We have thus characterized PfnIIa as a double agent that first controls exocytosis in a protein-specific manner through a direct interaction with GluK2b and then acts as a more generic regulator of clathrin-mediated endocytosis. The mechanism by which PfnIIa inhibits exocytosis of GluK2b is unknown but may be linked to its role in the organization of the cytoskeleton and possibly to the actindependent trafficking of export vesicles to the plasma membrane. At inhibitory synapses, PfnIIa participates in a complex with gephyrin, an essential component of the protein network that participates in the dynamic assembly of inhibitory receptors (26). Although this has not yet been defined, PfnIIa could be involved in connecting gephyrin to two cytoskeletal systems, microtubules and microfilaments. In addition it is not yet known how PfnIIa regulates the clusterization and stabilization of inhibitory receptors at synapses.
PfnIIa is redistributed to postsynaptic sites upon stimulation of postsynaptic NMDA receptors and influx of extracellular Ca 2ϩ in spines, with both elements being involved in long term changes of synaptic function and architecture (15). Activity-induced profilin accumulation in synapses has been also demonstrated in conditions of a physiological learning process, namely fear conditioning in the amygdala (27), where it is accompanied with an enlargement of the postsynaptic density. We can postulate that massive redistribution of Pf-nIIa following activity-dependent plastic changes could play a role in the dynamic regulation of synaptic KARs. Interestingly, PfnIIa knock-out mice display a presynaptic phenotype with higher vesicle release probability, suggesting that PfnIIa may also be subject to redistribution to presynaptic compartments. Because GluK2-containing KARs are present at presynaptic sites where they regulate neurotransmitter release (28), it is tempting to speculate that the density of presynaptic KARs could be controlled by PfnIIa-mediated interactions. In conclusion, this study extends the regulatory functions of profilins within the context of synaptic molecules, playing a role both as a generic regulator of membrane trafficking and as a component of specific glutamate receptor complexes.