Collapsin Response Mediator Protein 2 (CRMP2) Interacts with N-Methyl-d-aspartate (NMDA) Receptor and Na+/Ca2+ Exchanger and Regulates Their Functional Activity*

Background: NMDA receptor and Na+/Ca2+ exchanger are involved in glutamate-induced calcium dysregulation in neurons. Results: CRMP2 interacts with and modulates activity of the NMDA receptor and Na+/Ca2+ exchanger. Conclusion: CRMP2 is involved in regulation of Ca2+ homeostasis in neurons exposed to glutamate. Significance: CRMP2 interaction with NMDA receptor and Na+/Ca2+ exchanger affects their activity and is important for glutamate-induced Ca2+ dysregulation. Collapsin response mediator protein 2 (CRMP2) is traditionally viewed as an axonal growth protein involved in axon/dendrite specification. Here, we describe novel functions of CRMP2. A 15-amino acid peptide from CRMP2, fused to the TAT cell-penetrating motif of the HIV-1 protein, TAT-CBD3, but not CBD3 without TAT, attenuated N-methyl-d-aspartate receptor (NMDAR) activity and protected neurons against glutamate-induced Ca2+ dysregulation, suggesting the key contribution of CRMP2 in these processes. In addition, TAT-CBD3, but not CBD3 without TAT or TAT-scramble peptide, inhibited increases in cytosolic Ca2+ mediated by the plasmalemmal Na+/Ca2+ exchanger (NCX) operating in the reverse mode. Co-immunoprecipitation experiments revealed an interaction between CRMP2 and NMDAR as well as NCX3 but not NCX1. TAT-CBD3 disrupted CRMP2-NMDAR interaction without change in NMDAR localization. In contrast, TAT-CBD3 augmented the CRMP2-NCX3 co-immunoprecipitation, indicating increased interaction or stabilization of a complex between these proteins. Immunostaining with an anti-NCX3 antibody revealed that TAT-CBD3 induced NCX3 internalization, suggesting that both reverse and forward modes of NCX might be affected. Indeed, the forward mode of NCX, evaluated in experiments with ionomycin-induced Ca2+ influx into neurons, was strongly suppressed by TAT-CBD3. Knockdown of CRMP2 with short interfering RNA (siRNA) prevented NCX3 internalization in response to TAT-CBD3 exposure. Moreover, CRMP2 down-regulation strongly attenuated TAT-CBD3-induced inhibition of reverse NCX. Overall, our results demonstrate that CRMP2 interacts with NCX and NMDAR and that TAT-CBD3 protects against glutamate-induced Ca2+ dysregulation most likely via suppression of both NMDAR and NCX activities. Our results further clarify the mechanism of action of TAT-CBD3 and identify a novel regulatory checkpoint for NMDAR and NCX function based on CRMP2 interaction with these proteins.

HIV-1 protein (TAT-CBD3), decreased surface expression of NMDAR in dendritic spines, antagonized glutamate-induced Ca 2ϩ dysregulation, and protected brain tissue against ischemic damage (16). We hypothesized that TAT-CBD3 alters CRMP2 interaction with NMDAR, attenuates NMDAR activity, and thus exerts its neuroprotective action. However, whether TAT-CBD3 alters CRMP2-NMDAR interaction has not yet been experimentally proven. In addition, suppression of NMDAR activity as a sole mechanism of TAT-CBD3-evoked neuroprotection appears to be inconsistent with our finding that both NMDAR and NCX rev are important for glutamateinduced Ca 2ϩ dysregulation, and inhibition of both mechanisms is necessary to maintain Ca 2ϩ homeostasis (12). To resolve this issue and clarify the mechanism of the neuroprotective effect of TAT-CBD3, in the present study, we tested whether CRMP2 interacts with NMDAR and NCX and whether TAT-CBD3 alters these interactions and inhibits both NMDAR and NCX rev .
In this paper, using live cell fluorescence microscopy applied to rat hippocampal neurons in culture, we show that TAT-CBD3, in addition to NMDAR, inhibits NCX rev . Using co-immunoprecipitation, we demonstrated that CRMP2 physically interacts with the NR2B subunit of NMDAR as well as with NCX3. TAT-CBD3 disrupted CRMP2-NMDAR interaction without changing NMDAR localization but augmented CRMP2-NCX3 interaction and stimulated NCX3 internalization. Knockdown of CRMP2 did not affect NCX3 localization but eliminated TAT-CBD3-induced NCX3 internalization and prevented NCX rev inhibition. However, TAT-CBD3 inhibited not only NCX rev but also NCX operating in the forward mode, suppressing cytosolic Ca 2ϩ extrusion in exchange for external Na ϩ . Based on our results, we propose a model that explains the role of CRMP2 in regulation of NMDAR and NCX and elucidates the mechanism of the TAT-CBD3 neuroprotective effect.
Cell Culturing-Primary cultures of hippocampal neurons were prepared from postnatal day 1 rat pups according to institutional animal care and use committee-approved protocols and procedures previously published for hippocampus (17). For fluorescence measurements, neurons were plated on glass bottom Petri dishes without preplated glia as described previously (17). Twenty-four hours after plating, 35 g/ml uridine plus 15 g/ml 5-fluoro-2Ј-deoxyuridine were added to inhibit proliferation of non-neuronal cells. Cultures were maintained in a 5% CO 2 atmosphere at 37°C in minimum essential medium supplemented with 10% NuSerum (BD Biosciences) and 27 mM glucose.
Calcium and Sodium Imaging-We used rat hippocampal neurons cultured for 12-14 days in vitro (12-14 days in vitro). The bath solution contained 139 mM NaCl, 3 mM KCl, 0.8 mM MgCl 2 , 1.8 mM CaCl 2 , 10 mM HEPES, pH 7.4, 5 mM glucose, and 65 mM sucrose. Sucrose was used to maintain osmolarity similar to that in the growth medium (340 mosM) (18,19). Fluorescence imaging was performed with a Nikon Eclipse TE2000-U inverted microscope using Nikon objective Super Fluor ϫ20, 0.75 numerical aperture and a Photometrics Cool SNAP HQ camera (Roper Scientific, Tucson, AZ) controlled by MetaFluor software version 6.3 (Molecular Devices, Downingtown, PA). The excitation light was delivered by a Lambda-LS system (Sutter Instruments, Novato, CA). The images were acquired every 15 s during the time course of the experiment.
For cytosolic Ca 2ϩ measurements, neurons were loaded with either 2.6 M Fura-2AM or 2.6 M Fura-2FF-AM for 60 min at 37°C in the presence of 0.015% Pluronic F-127. The excitation filters (340 Ϯ 5 and 380 Ϯ 7 nm) were controlled by a Lambda 10-2 optical filter changer (Sutter Instruments, Novato, CA). Fluorescence was measured from individual neuronal somata through a 505-nm dichroic mirror at 535 Ϯ 25 nm. The changes in [Ca 2ϩ ] c were followed by recording Fura-2 or Fura-2FF F 340 / F 380 ratio. In some experiments, the changes in [Ca 2ϩ ] c were followed simultaneously with changes in [Na ϩ ] c using a Ca 2ϩsensitive fluorescent dye, Fluo-4FF-AM, and a Na ϩ -sensitive dye, SBFI-AM. Neurons were co-loaded with 2.5 M Fluo-4FF-AM for 30 min and 5.3 M SBFI-AM for 1 h at 37°C. The excitation wavelengths were 340 Ϯ 5 and 380 Ϯ 7 nm for SBFI and 480 Ϯ 20 nm for Fluo-4FF. Fluorescence was recorded from individual neuronal somata through a 505-nm dichroic mirror at 535 Ϯ 25 nm. The changes in [Na ϩ ] c were followed by recording the SBFI F 340 /F 380 ratio. The changes in [Ca 2ϩ ] c were followed by recording Fluo-4FF F 480 and normalized as F/F 0 . Quantification of Fura-2, Fura-2FF, and Fluo-4FF signals was carried out per the manufacturer's instructions. [Ca 2ϩ ] c and [Na ϩ ] c were calculated using the Grynkiewicz method (20), assuming a K d for Fura-2 of 0.225 m, a K d for Fura-2FF of 5.5 M, a K d for Fluo-4FF of 9.7 M, and a K d for SBFI of 11.3 mM. In all experiments, the fluorescence background was subtracted from the signals. Because Ca 2ϩ binding and spectroscopic properties of fluorescent dyes can differ significantly in intracellular milieu, the cytosolic Ca 2ϩ concentrations presented in this paper should be deemed estimates as stated previously by other investigators (21,22).
Co-immunoprecipitation-Co-immunoprecipitation experiments were performed on freshly prepared cell lysates from rat hippocampal neuronal cultures at 12-14 days in vitro. Lysates were clarified to remove any additional precipitate by incubating with Protein A-agarose beads for experiments with rabbit primary antibodies or Protein G-agarose beads for experiments with mouse primary antibodies (Santa Cruz Biotechnology, Inc.) for 2 h at 4°C. Then the lysates were incubated overnight with primary anti-NCX1, anti-NCX3, or anti-CRMP2 antibodies under gentle agitation at 4°C followed by incubation with Protein A-agarose beads or Protein G-agarose beads for 2 h at 4°C. The immunocaptured complexes were then washed three times with lysis buffer before being heated at 70°C in equal volumes of SDS loading dye (Invitrogen). Samples were then processed by immunoblotting as described previously (5,16). Blots were probed with anti-NCX1 (NCX1 mAb R3F1), anti-NCX3 (NCX3 polyclonal antibody 95209), anti-NR2B (BD Biosciences), or anti-CRMP-2 (Sigma-Aldrich) (all 1:1000). All blots are representative of at least three experiments.
NCX3, NR2B-NMDAR, and CRMP2 Immunocytochemistry-Cultured neurons were treated with vehicle (0.1% DMSO), TAT-CBD3, or TAT-scramble peptide for 10 min at room temperature. Then neurons were fixed with 4% paraformaldehyde for 15 min and then washed with PBS. Cells were incubated with a blocking solution containing 2.5% IgG-and proteasefree BSA (Jackson ImmunoResearch Laboratories, West Grove, PA), 2.5% goat serum, and 0.1% Triton X-100 in PBS for 1 h at room temperature. Cells were incubated overnight at 4°C with the primary rabbit polyclonal anti-NCX3 (NCX3 polyclonal antibody 95209, 1:500), mouse monoclonal anti-NR2B (1:500; BD Biosciences), or rabbit polyclonal anti-CRMP2 (1:500; Sigma). Then cells were incubated with a secondary donkey anti-rabbit antibody or goat anti-mouse antibody conjugated with AlexaFluor 488 for anti-NCX3 and anti-NR2B or Alexa-Fluor 568 for anti-CRMP2 (1:1000) (Invitrogen). Cells were incubated for 1 h at room temperature. The images were taken using a laser, spinning-disk confocal microscope based on a Nikon Eclipse TE2000-U equipped with a Yokogawa spinning disk confocal unit CSU-10, a back-thinned EM-CCD camera Andor iXon EM ϩ DU-897 (Andor Technology, South Windsor, CT), and a motorized flat top stage Prior H-117 (Prior Scien-tific, Rockland, MA). This setup was controlled by Andor iQ version 1.4 software (Andor Technology, South Windsor, CT). To visualize immunostaining, neurons were illuminated at 488 nm using an air-cooled krypton/argon laser T643-RYB-A02 (Melles Griot, Carlsbad, CA). Fluorescence was collected through a 505-nm dichroic mirror and a 535 Ϯ 25-nm emission filter using an objective Nikon CFI Plan Apo ϫ100, 1.4 numerical aperture. The fluorescence intensity profiles were generated through randomly chosen cross-sections using Auto-Quant software (MediaCybernetics, Silver Spring, MD).
Statistics-Statistical analysis of the experimental results consisted of one-way analysis of variance followed by Bonferroni's post hoc test (GraphPad Prism version 4.0, GraphPad Software Inc., San Diego, CA). The data represent mean Ϯ S.E. of at least three separate, independent experiments.

TAT-CBD3 Inhibits NMDAR and Attenuates
Glutamate-induced Ca 2ϩ Dysregulation-In experiments with Fura-2FFloaded rat hippocampal neurons (12-14 days in vitro), compared with vehicle control (Fig. 1A), TAT-CBD3 (10 M) significantly attenuated glutamate-induced Ca 2ϩ dysregulation (Fig. 1B), whereas CBD3 without TAT (10 M) that cannot cross the membrane (Fig. 1C) and TAT alone were without effect (Fig. 1D). In this and subsequent figures, thin, gray traces represent fluorescence signals from somata of individual cells, whereas thick, red traces represent average signals Ϯ S.E. The change in cytosolic Ca 2ϩ in neurons exposed to TAT-CBD3, expressed as average area under the curve, showed a decline by 70% compared with vehicle-, TAT-, or CBD3-treated neurons (Fig. 1E). Similarly, TAT-CBD3 strongly attenuated NMDAevoked increases in cytosolic Ca 2ϩ , whereas CBD3 without TAT was ineffective (Fig. 2, A, C, and D): from a peak of 1.03 Ϯ 0.05 M to 0.17 Ϯ 0.05 M cytosolic Ca 2ϩ with TAT-CBD3 (p Ͻ 0.01, n ϭ 5 individual experiments with a total of 107 analyzed neurons) and from a peak of 1.15 Ϯ 0.05 M to 1.10 Ϯ 0.07 M cytosolic Ca 2ϩ with CBD3 without TAT (p Ͼ 0.05, n ϭ 5 individual experiments with a total of 98 analyzed neurons). Previously, we established an IC 50 for the neuroprotective action of TAT-CBD3 to be about 2 M and found that 10 M TAT-CBD3 provided maximal protection against glutamate excitotoxicity (16). Therefore, in the present study, we used 10 M TAT-CBD3. As a positive control, we used AP-5 (20 M), a potent and efficacious NMDAR antagonist (25). AP-5 completely blocked NMDA-induced increase in cytosolic Ca 2ϩ (Fig. 2B). Collectively, the results revealed that TAT-CBD3-mediated protection against glutamate-induced Ca 2ϩ dysregulation and NMDA-evoked increases in cytosolic Ca 2ϩ required CBD3 translocation into the cell and most likely occurred due to its interaction with intracellular molecular targets.

CRMP2 Directly Interacts with NR2B-NMDAR and TAT-CBD3 Disrupts This Interaction and Decreases NMDAR
Activity-The inhibition of NMDAR activity by TAT-CBD3, a membrane-permeant CRMP2 fragment, suggested a direct interaction between CRMP2 and NMDAR. In the present study, co-IP experiments revealed an interaction between CRMP2 and NR2B-containing NMDARs (Fig. 3A). Similar findings were reported previously (26). Our experiments were focused on CRMP2 interaction with NR2B-NMDAR because this type of NMDAR is predominantly involved in excitotoxicity (27,28). TAT-CBD3 disrupted the CRMP2-NMDAR complex (Fig. 3A). It is possible that CRMP2 up-regulates NMDAR activity via binding to the NR2B subunit and that TAT-CBD3mediated dissociation of CRMP2 from NMDAR leads to diminution of NMDAR activity (see Fig. 2C). It is also conceivable that CRMP2 serves as an anchor protein, retaining NMDAR in the plasma membrane. In this case, disruption of CRMP2-NMDAR interaction should lead to NMDAR internalization that could explain TAT-CBD3-evoked suppression of NMDAR activity. To test this possibility, we used immunocytochemistry in conjunction with confocal fluorescent microscopy. In these experiments, neurons were treated for 10 min with a vehicle (0.1% DMSO) or 10 M TAT-CBD3 and then fixed, permeabilized with 0.1% Triton X-100, stained, and analyzed using confocal microscopy. Fig. 3, B-E, shows representative inverted fluorescent images and fluorescence intensity profiles for the respective images. The experiments were repeated three times under each condition. The total number of analyzed neurons was 58 (vehicle-treated) and 67 (TAT-CBD3-treated). Under each condition, all analyzed cells had similar patterns of NR2B tk;2distribution, as illustrated in Fig. 3, B and D. The NMDAR appeared to be localized in both the plasma membrane and cytoplasm, the latter probably reflecting internalized NMDAR (Fig. 3, B-E). A similar NMDAR distribution pattern bas been reported previously by others (29). Application of secondary antibody alone failed to produce staining (data not shown), indicating the lack of nonspecific interactions. The monoclonal anti-NR2B antibody (BD Biosciences) used in these experiments demonstrated excellent specificity, detecting only a single band in immunoblotting experiments (Fig. 3B, inset). This ensures that immunostaining shown in Fig. 3, B and D, is exclusively due to NR2B-NMDARs. Contrary to our expectations, immunocytochemistry experiments failed to provide evidence for TAT-CBD3-induced NMDAR relocalization (Fig. 3, D and  E). This suggested that disruption of the CRMP2-NR2B complex might affect NMDAR activity without its translocation from the plasma membrane. CRMP2 Directly Interacts with NCX3 and TAT-CBD3 Strengthens This Interaction but Leads to NCX3 Internalization and Decreases Its Activity-Recently, we showed that inhibition of both NMDAR and NCX rev is necessary for robust attenua- tion of glutamate-induced Ca 2ϩ dysregulation (12). Based on these findings, we hypothesized that TAT-CBD3, in addition to inhibiting NMDAR, should also inhibit NCX rev . To test the effect of TAT-CBD3 on NCX rev activity, we induced NCX reversal by replacing external Na ϩ for equimolar N-methyl-Dglucamine (NMDG). Prior to Na ϩ /NMDG replacement, neurons were preincubated with ouabain (1 mM), an inhibitor of Na ϩ /K ϩ -ATPase, for 10 min to increase cytosolic Na ϩ and thus facilitate NCX reversal. The Na ϩ /NMDG replacement produced a robust increase in cytosolic Ca 2ϩ and a decrease in cytosolic Na ϩ (Fig. 4, A and B). Because Na ϩ /NMDG replacement could evoke release of endogenous glutamate (30), these experiments were performed in the presence of AP-5 (20 M) to prevent activation of NMDAR and exclude Ca 2ϩ influx via this mechanism. CNQX (20 M), an antagonist of AMPA/kainate receptor, did not influence Na ϩ /NMDG-induced increase in cytosolic Ca 2ϩ (data not shown), indicating that AMPA/kainate receptor is not involved. In our previous paper (12), we demonstrated that MK801, an NMDAR antagonist, also appeared to be a potent inhibitor of NCX rev . In the present study, MK801 completely inhibited Na ϩ /NMDG-induced increase in cytosolic Ca 2ϩ and slowed down a decrease in cytosolic Na ϩ (Fig. 4, C  and D), implicating NCX rev in these ion alterations. Thus, in our experiments, Na ϩ /NMDG replacement caused an increase in cytosolic Ca 2ϩ due to a reversal of NCX.
TAT-CBD3 (10 M) strongly inhibited the Na ϩ /NMDG-induced increase in cytosolic Ca 2ϩ (Fig. 5, A and B), suggesting that TAT-CBD3 inhibits NCX rev . TAT-CBD3 A6K, a mutated version of CBD3 that demonstrated increased conformational stability and greater anti-nociceptive activity in several rodent models of neuropathic pain compared with the parent TAT-CBD3 peptide (31), also significantly inhibited Na ϩ /NMDGinduced increase in cytosolic Ca 2ϩ (Fig. 5C), whereas TATscramble peptide and CBD3 without TAT were without effect (Fig. 5, D and E). Similar results were obtained with these peptides when NCX reversal was induced by application of 5 M gramicidin (data not shown). Fig. 5F shows a summary of the data, expressed as average area under the curve, obtained with Na ϩ /NMDG-induced increases in cytosolic Ca 2ϩ and the effects of various peptides.
Na ϩ /Ca 2ϩ exchange is electrogenic; therefore, it is possible to record ion currents mediated by NCX (32,33). Previously, this approach was used in experiments with cardiomyocytes (24,34). We successfully adapted this technique to neurons in culture (12,23) and have previously provided a detailed explanation as to why NCX rev activity, but not voltage-gated Ca 2ϩ channel activity, is the main contributor in generating these currents (30). In these experiments, we employed a voltage ramp protocol with a change in voltage from Ϫ120 mV to ϩ80 mV. Because of favorable Na ϩ and Ca 2ϩ concentrations across the plasma membrane, NCX operates in the reverse mode under positive voltages, extruding 3 Na ϩ out of the cell for every Ca 2ϩ transported in. This results in a net outward current, the peak of which when recorded at ϩ80 mV can be used as a measure of NCX rev activity (12). Consequently, in addition to Ca 2ϩ imaging, we used an electrophysiological, whole-cell, voltage ramp protocol to demonstrate NCX rev inhibition by TAT-CBD3. Ni 2ϩ , an NCX inhibitor (24,34), was used as a positive control. Ni 2ϩ (5 mM) significantly decreased (by 62 Ϯ 7%, n ϭ 5) the peak ion current generated by the voltage ramp, indicating that this current is predominantly mediated by NCX rev (Fig. 6A). TAT-CBD3 (10 M), but not TAT-scramble (10 M), applied to neurons for 5 min also significantly attenuated the peak ion current (by 53 Ϯ 6%, n ϭ 5), suggesting that indeed TAT-CBD3 decreases NCX rev activity (Fig. 6, B and C). Fig. 6D shows a summary graph, analyzing changes in the peak ion current under different conditions. Previously, we showed that CRMP2 directly interacts with CaV2.2 (35) and NMDAR (Fig. 3). Here, we hypothesized that CRMP2 could also physically interact with NCX and that TAT-CBD3 inhibition of NCX rev could possibly be due to alterations of a putative CRMP2-NCX interaction. In co-immunoprecipitation experiments, we found no evidence for interaction of NCX1 with CRMP2 (Fig. 7A). However, CRMP2 did co-immunoprecipitate with NCX3 (Fig. 7, B and C), suggesting a physical interaction between these proteins. Surprisingly, TAT-CBD3 notably strengthened the CRMP2-NCX3 interaction, whereas TAT-scramble was without effect (Fig. 7C).
In our previous study, we had also reported that TAT-CBD3, in an activity-dependent manner, down-regulated surface expression of NR2B-containing NMDAR in dendritic spines (16). Consistent with this, in our present study, we found that TAT-CBD3 disrupts the CRMP2-NMDAR interaction (Fig. 3). In stark contrast, TAT-CBD3 enhanced the CRMP2-NCX3 interaction, and consequently, we expected a commensurate increase in NCX3 surface expression. In immunocytochemistry experiments to test this hypothesis, neurons were treated for 10 min with vehicle, TAT-CBD3, or TAT-scramble (both at 10 M) and then fixed, permeabilized with 0.1% Triton X-100, stained, and analyzed using confocal microscopy. Fig. 8 shows representative inverted fluorescent images and fluorescence intensity profiles for the respective images. The experiments were repeated three times under each condition. In vehicletreated neurons, NCX3 was found to be localized in both plasma membrane and, to a lesser extent, the cytoplasm (Fig. 8,  A and B).  ments (Fig. 8A, inset). This selectivity ensures that immunostaining shown in Fig. 8 (A, C, and E) is exclusively due to NCX3. In contrast to our expectations regarding an increase in NCX surface expression, after treatment with TAT-CBD3, we found that TAT-CBD3 triggered internalization of NCX3, resulting in a decreased expression in the plasma membrane and an increased pool of internalized NCX3 (Fig. 8, C and D). A similar pattern was observed in 63 of 65 analyzed neurons. TAT-scramble did not change NCX3 localization (Fig. 8, E and F; a similar pattern was observed in 48 of 51 analyzed neurons). The rapid NCX internalization induced by TAT-CBD3 probably explains the decrease in NCX rev activity following preincubation with TAT-CBD3.
To establish a link between CRMP2 and TAT-CBD3-triggered NCX3 internalization, we down-regulated CRMP2, using siRNA. We hypothesized that if TAT-CBD3-induced NCX3 internalization depends on CRMP2 association with NCX3, then CRMP2 down-regulation should prevent or attenuate this process. To test this hypothesis, neurons were concomitantly transfected during plating with a GFP construct and siRNA against CRMP2. As illustrated in Fig. 9, A-D, neurons expressing GFP had an almost complete lack of CRMP2 (Fig. 9, A-D). Notably, in non-transfected cells, CRMP2 was found in both processes and neuronal somata. In all analyzed neurons (n ϭ 21), expression of GFP coincided with significant CRMP2 down-regulation. Consequently, in the following experiments, we considered GFP fluorescence as an indicator of neurons with down-regulated CRMP2. We compared NCX3 expression in non-transfected versus transfected cells. The NCX3 expression pattern was similar in both types of cells (Fig. 9, E-H and I-L). Importantly, pretreatment with 10 M TAT-CBD3 for 10 min failed to induce internalization of NCX3 in neurons with down-regulated CRMP2 (Fig. 9, I-L). In addition, CRMP2 down-regulation robustly attenuated TAT-CBD3induced inhibition of NCX rev (Fig. 10). The results of these experiments demonstrate that CRMP2 is necessary for the effect of TAT-CBD3. Overall, our data show that the increased CRMP2 association with NCX3 correlates with TAT-CBD3-induced NCX3 internalization and subsequent inhibition of the exchanger.
If our findings about TAT-CBD3-induced internalization of NCX3 are correct, then we predict that in addition to a decrement in NCX rev activity, we should also observe a diminution of the activity of NCX operating in the forward mode (NCX for ). To test this hypothesis, we treated neurons with ionomycin (5 M), a Ca 2ϩ ionophore. Ionomycin caused delayed Ca 2ϩ dysregulation in neurons (Fig. 11A), demonstrating that neurons possess very effective mechanisms antagonizing elevations in cytosolic Ca 2ϩ . It is known that NCX for cannot function in the absence of FIGURE 7. CRMP2 co-immunoprecipitates with NCX3, but not with NCX1. TAT-CBD3 strengthens CRMP2-NCX3 interaction. A, immunoprecipitation (IP) was performed with IgG or anti-CRMP2 antibody followed by Western blotting (WB) with anti-NCX1 antibody. B, immunoprecipitation was performed with IgG or anti-NCX3 antibody followed by Western blotting with anti-CRMP2 antibody. C, prior to immunoprecipitation, where indicated, neurons were incubated either with 10 M TAT-CBD3 or with 10 M TAT-scramble peptide for 10 min. Each experiment was repeated three times.  external Na ϩ (36). Consequently, to demonstrate the role of NCX for in ionomycin-induced Ca 2ϩ increase, we inhibited NCX for activity by removing external Na ϩ and replacing it with equimolar NMDG. In these experiments, ouabain was omitted from the bath solution, so that cytosolic Na ϩ remained low, and Na ϩ /NMDG replacement did not cause an increase in cytosolic Ca 2ϩ . In addition, these experiments were performed in the presence of AP-5 (20 M), an NMDAR antagonist, to exclude any contribution by NMDARs to the ionomycin-induced Ca 2ϩ increase. The removal of external Na ϩ significantly accelerated the onset of ionomycin-induced Ca 2ϩ dysregulation, illustrating a key role for NCX for in maintaining low cytosolic Ca 2ϩ (Fig.  11B). As we expected, TAT-CBD3 significantly accelerated Ca 2ϩ dysregulation in neurons treated with ionomycin (Fig.  11C). To quantify our observations, we measured the time from the beginning of ionomycin application to the onset of Ca 2ϩ dysregulation (t dys ) (Fig. 11D). A similar approach was used recently to provide a statistical analysis of glutamate-induced Ca 2ϩ dysregulation (37,38). The statistical analysis confirmed that suppression of NCX for , by removing external Na ϩ or by application of TAT-CBD3, significantly accelerated the onset of Ca 2ϩ dysregulation in neurons (Fig. 11D). These observations suggest that TAT-CBD3 decreases the activities of NCX regardless of the mode of operation and supports a model whereby TAT-CBD3-induced NCX internalization is probably responsible for the suppression of NCX activity.

DISCUSSION
Sustained elevations in cytosolic Ca 2ϩ play a causative role in glutamate-induced neuronal death (6 -8). Elevated cytosolic Ca 2ϩ activates Ca 2ϩ -dependent degradation enzymes, such as proteases and phospholipases, and thus hastens cell death (3)(4)(5). Therefore, it is vitally important for neurons to fight against rampant Ca 2ϩ elevations and maintain low cytosolic Ca 2ϩ . In neurons, Ca 2ϩ extrusion by NCX for and Ca 2ϩ uptake by mitochondria are major mechanisms that antagonize Ca 2ϩ increases in the cytosol. Mitochondria may accumulate substantial amounts of Ca 2ϩ before they undergo permeability transition and become damaged (39). Following Ca 2ϩ -dependent activation of the permeability transition pore, mitochon- dria lose the ability to accumulate Ca 2ϩ and, subsequently, release previously accumulated Ca 2ϩ (40). Thus, although mitochondria can accumulate large amounts of Ca 2ϩ , their Ca 2ϩ uptake capacity is finite. On the other hand, the amount of Ca 2ϩ in the extracellular milieu is practically infinite, and consequently, attenuation of Ca 2ϩ influx into neurons could be a much more effective strategy in neuronal defense against glutamate-induced Ca 2ϩ dysregulation.
Ca 2ϩ influx via NMDAR and NCX rev are two major mechanisms of Ca 2ϩ entry into neurons exposed to glutamate (9,41). Recently, we demonstrated that both mechanisms have to be suppressed to prevent glutamate-induced Ca 2ϩ dysregulation (12). TAT-CBD3 strongly attenuates glutamate-induced Ca 2ϩ dysregulation. Based on these observations, we hypothesized that TAT-CBD3 should inhibit both NMDAR and NCX rev . The results presented in this paper strongly support this hypothesis and link NMDAR and NCX activity to CRMP2 signaling.
CRMP2 is a member of the CRMP protein family originally known as an axon guidance and outgrowth regulator (13)(14)(15). CRMP2 is cleaved by calpain following focal ischemia, thereby implicating it in neurotoxicity associated with ischemia (42). It appears that cleaved CRMP2 is neuroprotective, whereas cleavage of CRMP3 and CRMP4 is detrimental for neurons (42)(43)(44). CRMP2 physically interacts with numerous proteins in the cell, including NMDAR and CaV2.2 (26,35,(45)(46)(47)(48)(49). It has been shown that TAT-CBD3 disrupts interaction of CRMP2 with CaV2.2, leading to its internalization and a decrease in ion currents mediated by this channel (35). In our previous paper, using a pH-sensitive fluorescent NR2B sensor, we showed that TAT-CBD3 decreases surface expression of NMDAR in dendritic spines (16), but it remained unclear whether TAT-CBD3 disrupts a CRMP2-NMDAR interaction. It was also unknown whether CRMP2 interacts with NCX and whether TAT-CBD3 affects NCX rev activity. Here, we demonstrate for the first time that CRMP2 physically interacts with NCX3 and that TAT-CBD3 strengthens this interaction, triggers NCX3 internalization, and consequently suppresses NCX activity in both forward and reverse modes. In contrast to strengthening the CRMP2-NCX3 complex, TAT-CBD3 disrupted CRMP2-NMDAR interaction, suggesting different mechanisms by which CRMP2 interacts with NMDAR and NCX3.
Although NCX3 internalization may account for the suppression of NCX activity, the mechanism of NCX3 internalization and the role of TAT-CBD3-augmented CRMP2-NCX3 interaction in this process are not clear. It is conceivable that TAT-CBD3 competes with CRMP2 for binding with other proteins (e.g. CaV2.2 (35,45) and NMDAR (Fig. 3)) and releases CRMP2 from complexes with these proteins (Fig. 12). If so, then the liberated CRMP2 may bind to other proteins (e.g. NCX3) possibly via a different binding site, with which TAT-CBD3 does not interfere. Overall, this may lead to CRMP2 redistribution and increased binding to NCX3. That NCX3 internalization is no longer induced by TAT-CBD3 in CRMP2ablated neurons suggests a causative link between increased CRMP2 association with NCX3 and NCX3 internalization. However, additional experiments are necessary to completely prove this hypothesis.
One of the goals of our study was to test the hypothesis that TAT-CBD3 inhibits deleterious NCX rev , which, instead of extruding Ca 2ϩ from the cell, causes Ca 2ϩ influx (50). Our results strongly support this hypothesis. However, our study also revealed an unexpected consequence of TAT-CBD3 treatment. Because TAT-CBD3 triggers NCX internalization, it suppresses both forward and reverse modes of NCX. However, NCX for is critical for mitigating increases in cytosolic Ca 2ϩ and, thus, maintaining cytosolic Ca 2ϩ at the low, physiological level. Consequently, suppression of NCX for should be detrimental for the cell. How then can one reconcile this, given TAT-CBD3mediated protection against glutamate-induced Ca 2ϩ dysregulation? A likely explanation is that TAT-CBD3-induced inhibi-    MARCH 14, 2014 • VOLUME 289 • NUMBER 11 tion of NMDAR plays the dominant role in this protection. The significantly lower Ca 2ϩ influx via NMDAR creates conditions in which, despite suppression of NCX for , cytosolic Ca 2ϩ remains at a low level. Mitochondrial Ca 2ϩ uptake and Ca 2ϩ extrusion by plasma membrane Ca 2ϩ -ATPase might also contribute to keeping cytosolic Ca 2ϩ low under these conditions. However, as we reported earlier, inhibition of NMDAR alone is not sufficient to prevent glutamate-induced Ca 2ϩ dysregulation. Suppression of NCX rev , in addition to NMDAR, is critical for maintaining low cytosolic Ca 2ϩ following exposure to glutamate (12). Our present study unequivocally demonstrates that TAT-CBD3 inhibits both NMDAR and NCX rev , hence filling a significant gap in our understanding of the mechanism of TAT-CBD3-mediated neuroprotection. Thus, a sum of the TAT-CBD3 inhibitory effects on NMDAR and NCX rev collectively results in a net protection against glutamate-induced Ca 2ϩ dysregulation.

CRMP2 Regulates Cytosolic Ca 2؉ in Neurons
In conclusion, in this paper, for the first time we demonstrated a link between CRMP2 signaling, NMDAR and NCX3 activities, and maintenance of Ca 2ϩ homeostasis in neurons exposed to excitotoxic levels of glutamate. Our experiments revealed that CRMP2 physically interacts with both NMDAR and NCX3. Moreover, we showed that TAT-CBD3 strengthens CRMP2-NCX3 interaction and induces NCX3 internalization, leading to suppression of NCX activity. Concomitantly, TAT-CBD3 disrupts the CRMP2-NMDAR interaction without detectable relocalization of NMDAR. Nevertheless, this also leads to suppression of NMDAR activity. Thus, our study impli-  We speculate that TAT-CBD3 binds to NMDAR and CaV2.2 and triggers CRMP2 dissociation from these proteins, leading to augmentation of CRMP2 binding to NCX. Disruption of the CRMP2-CaV2.2 complex leads to CaV2.2 internalization and a decrease in its activity (35). (Notably, voltage-gated Ca 2ϩ channels do not significantly contribute to glutamate-induced Ca 2ϩ dysregulation in our experiments (35).) Augmentation of CRMP2-NCX3 binding also leads to NCX internalization, and in both cases with CaV2.2 and NCX3, internalization leads to suppression of functional activity of these proteins. However, disruption of the CRMP2-NMDAR complex does not cause relocation of NMDAR and yet also results in suppression of NMDAR activity by an unknown mechanism. Inhibition of NMDAR in combination with suppression of NCX is responsible for a decrease in Ca 2ϩ influx into neurons and protection against glutamate-induced Ca 2ϩ -dysregulation. The scheme illustrates neither structural peculiarities nor true stoichiometry of molecular components involved in Ca 2ϩ signaling but serves the purpose of illustrating major CRMP2-associated events that alter Ca 2ϩ signaling regardless of their localization within the presynaptic or postsynaptic neuron.
cates CRMP2 signaling in regulation of cytosolic Ca 2ϩ in neurons exposed to glutamate, suggests different mechanisms of CRMP2 interaction with NMDAR and NCX, and further clarifies the mechanism of TAT-CBD3-mediated defense against glutamate-induced Ca 2ϩ dysregulation. Our experiments with TAT-CBD3 provide further support to our hypothesis that inhibition of both NMDAR and NCX rev is essential for robust attenuation of glutamate-induced Ca 2ϩ dysregulation. Overall, the new knowledge obtained in our study could be instrumental in designing new neuroprotective agents and in a better understanding of the mechanisms of their action.