Fyn-mediated Phosphorylation of NR2B Tyr-1336 Controls Calpain-mediated NR2B Cleavage in Neurons and Heterologous Systems*

Cleavage of the intracellular carboxyl terminus of the N-methyl-d-aspartate (NMDA) receptor 2 subunit (NR2) by calpain regulates NMDA receptor function and localization. Here, we show that Fyn-mediated phosphorylation of NR2B controls calpain-mediated NR2B cleavage. In cultured neurons, calpain-mediated NR2B cleavage is significantly attenuated by blocking NR2B phosphorylation of Tyr-1336, but not Tyr-1472, via inhibition of Src family kinase activity or decreasing Fyn levels by small interfering RNA. In HEK cells, mutation of Tyr-1336 eliminates the potentiating effect of Fyn on calpain-mediated NR2B cleavage. The potentiation of NR2B cleavage by Fyn is limited to cell surface receptors and is associated with calpain translocation to plasma membranes during NMDA receptor activation. Finally, reducing full-length NR2B by calpain does not decrease extrasynaptic NMDA receptor function, and truncated NR1/2B receptors similar to those generated by calpain have electrophysiological properties matching those of wild-type receptors. Thus, the Fyn-controlled regulation of NMDA receptor cleavage by calpain may play critical roles in controlling NMDA receptor properties during synaptic plasticity and excitotoxicity.

The N-methyl-D-aspartate (NMDA) 2 receptor is a calciumpermeable ionotropic glutamate receptor requiring glutamate, glycine, and membrane depolarization for activation. These receptors play crucial roles in brain development, excitatory neurotransmission, and synaptic plasticity but also may contribute to the pathogenesis of neurological disorders, including ischemia, epilepsy, and Huntington's disease (1)(2)(3)(4). NMDA receptors are made from heteromeric assemblies of different subunits called NR1 and NR2 (NR2A to -D) (5)(6)(7)(8). The NR2B subunit is essential for both neonatal and mature NMDA receptors and is highly expressed in the entire embryonic brain and the adult forebrain (9). In the hippocampus, the major neonatal receptor is believed to contain two NR1 and two NR2B subunits, whereas the major receptor in more mature synapses probably contains two NR1 subunits, an NR2A subunit, and an NR2B subunit. Analysis of NR2B-null mice has suggested the physiological importance of NR2B, since the absence of NR2Bcontaining NMDA receptors in NR2B Ϫ/Ϫ mice leads to perinatal death, whereas mice lacking NR2A, NR2C, and NR2D subtypes are viable (10 -12).
NR2B subunits have long C-terminal tails that link NR2B to intracellular pathways. Like NR2B-deficient mice, mice expressing NR2B subunits with truncated C-terminal domains die shortly after birth (13), showing the physiological importance of the intracellular region of NR2B. Therefore, processes that modify the C-terminal region of NR2B may alter the physiological events mediated by NR2B-containing receptors. Calpain and tyrosine kinases regulate NMDA receptors through the C-terminal domain. Calpain, a Ca 2ϩ -dependent neutral cysteine protease, cleaves the C-terminal domain of NR2B in HEK cells, neuronal cultures, and animal models of ischemia and seizures (14 -17). In acutely isolated and cultured cortical neurons, prolonged stimulation of NMDA receptors reduces NMDA receptor-mediated currents in a manner requiring calpain activation (18). In cultures of hippocampal neurons, calpain-mediated cleavage of NR2B produces truncated receptors that remain on the cell surface but potentially have novel properties (16,19). These studies indicate that calpain-mediated truncation of NR2B may regulate NMDA receptor function and localization.
The NR2B subunit is tyrosine-phosphorylated in neurons (20 -22). Fyn, a Src family tyrosine kinase, phosphorylates Tyr-1252, Tyr-1336, and Tyr-1472 in the C-terminal domain of NR2B subunit in vitro, with Tyr-1336 and Tyr-1472 being the most readily phosphorylated sites in cell culture systems (23). A previous study limited to in vitro experiments suggested that Fyn and Src might regulate cleavage of NR2 subunits in opposing manners. Here, using in situ approaches in transfected cells and neurons, we have used molecular biological, physiological, and receptor trafficking approaches to show that phosphorylation of tyrosine residue Tyr-1336 by Fyn controls calpain-me-gent (Upstate Biotechnology) according to the manufacturer's instruction. We transfected 10 or 11 DIV neurons in 12-well plates with 500 ng of corresponding siRNA plasmid (pKD-Fyn-v6 or pKD-NegCon-v1), and Fyn protein expression was assessed 96 h after transfection (15)(16) by immunoblotting. The species of the target for pKD-Fyn-V6 is human, but pKD-Fyn-V6 has a ϳ95% sequence homology to Rat Fyn and has no homology to other known rodent genes (based on BLAST searches).
In Vitro Phosphorylation and Proteolysis of NR2B by Purified Calpain I-Twenty-four hours after transfection, HEK293 cells were rinsed with phosphate-buffered saline and then scraped into phosphorylation buffer (200 mM Tris-HCl, 10 mM MgCl 2 , 5 mM MnCl 2 , 0.4 M Na 2 VO 4 , 500 M ATP, 0.8 mM EGTA). Samples were briefly sonicated on ice, incubated at 56°C for 30 min to inactivate endogenous kinases and phosphatases, and then incubated with recombinant active Fyn protein (100 ng/ml) for 40 min at 30°C. After phosphorylation, samples were incubated with 2 units/ml calpain I (Calbiochem) for 2 min at 30°C in a buffer containing final concentrations of 1 mM CaCl 2 , 5 mM dithiothreitol, and 40 mM HEPES, pH 7.2. The reaction was stopped by an addition of SDS sample buffer, and the reaction mixtures were then loaded on an 8% SDS-polyacrylamide gel.
Whole-cell Recording-Experiments were performed with standard whole-cell voltage clamp techniques using transfected HEK293 cells 12-24 h after transfection and rat embryonic hippocampal neurons maintained in culture for 17-21 days DIV. NMDA-evoked currents were recorded at room temperature using NMDA (100 M) along with glycine (10 M) in a Mg 2ϩfree extracellular recording medium containing 500 nM tetrodotoxin, 155 mM NaCl, 3 mM KCl, 3 mM CaCl 2 , and 10 mM HEPES, pH 7.35. To examine NMDA receptor function, cells were voltage-clamped at Ϫ50 mV using an intrapipette solution containing 100 mM Trizma phosphate (dibasic), 28 mM Trizma base, 11 mM EGTA, 2 mM MgCl 2 , 0.5 mM CaCl 2 , and 10 mM Mg-ATP, pH 7.35, 290 mosM/kg H 2 O. Each application was 2 s, and the interval between applications was more than 1 min to prevent current desensitization. Solution switching was carried out using an SF-77B fast step solution delivery device (Warner Instrument Co., Hamden, CT). Recording signals were amplified using an Axopatch-1D amplifier, filtered at 5 kHz, and then saved using an IBM PC running pCLAMP 8.01 software (Axon Instruments, Inc. Foster City, CA) for off-line analysis. NMDA, glutamate, and CalI3 were dissolved in extracellular solution. Some aliquots of CalI3 were solubilized in Me 2 SO; final Me 2 SO concentration was 0.001% (v/v). This concentration had no effect on whole-cell NMDA currents (data not shown). In order to observe the extrasynaptic NMDA receptor-mediated effects, synaptic NMDA receptors in hippocampal culture neurons were blocked with 50 M bicuculline and 10 M MK801 for 10 min. In addition, ifenprodil (10 M) was used to confirm that NMDA receptor-mediated current in bicuculline/MK801-pretreated neurons was primarily from extrasynaptic NMDA receptor (since they contain primarily NR1/2B). To examine the pharmacological effect of ifenprodil on NMDA receptors, ifenprodil was preadministered to the recorded neurons alone for more than 30 s before neurons were co-applied with NMDA/glycine and ifenprodil. Since the effect of ifenprodil could not be washed away within 30 min (data not shown), only one neuron per culture dish was recorded during these experiments.
Neuronal Cultures-Primary cultures of hippocampal and cortical neurons from embryonic day 17-19 Sprague-Dawley rats (Charles River Laboratories, Wilmington, MA) were prepared as described previously (16,24). Briefly, the hippocampi were dissected, gently minced, trypsinized (0.027%, 37°C; 7% CO 2 for 20 min), and then washed with 1ϫ HBSS. Neurons were seeded to a density of 6 ϫ 10 5 viable cells/35-mm culture dish or similar relative density onto 12-well plates or 60-mm culture dishes. All plates and culture dishes were coated with poly-D-lysine (100 g/ml) prior to seeding neurons. Cultures were maintained at 37°C with 5% CO 2 , fed with neurobasal medium with B27 supplement, and the cultures were used at 17-21 DIV.
Western Blotting Analysis-Western blotting was performed as described previously (16,24). Protein content was estimated using BCA Protein Assay (Pierce). Equal amounts of total protein (15-20 g/lane) were resolved on denaturing 8 -10% SDSpolyacrylamide gels, and transferred to nitrocellulose membranes. Membranes were blocked with 3% nonfat milk and incubated with primary antibody for 2 h at room temperature or overnight at 4°C. Blots were then incubated with appropriate horseradish peroxidase-conjugated secondary antibodies (Sigma) for 2 h at room temperature and then washed; bands were visualized using enhanced chemiluminescence (Pierce). Reaction product levels were quantified by scanning densitometry using NIH Image 1.62 (available on the World Wide Web).
Transfection of HEK293 Cells-HEK293 cells were grown on tissue culture dishes in minimum Eagle's medium containing 5% horse serum, 5% fetal bovine serum supplemented with 2 mM glutamine, 100 units/ml penicillin/streptomycin and placed in a 5% CO 2 incubator at 37°C. Transfection was performed using calcium phosphate as previously described (25). Treatments were performed 18 -24 h following transfection. Ketamine (500 M) was included in the media during transfection to prevent NMDA receptor activation as previously described (25).
Calcium Imaging-HEK293 cells were cotransfected with NR1, SAP-102, and GFP with NR2B wild type, NR2BY1336F, NR2BY1472F, or NR2Bdel1036 (NR2B truncated at amino acid 1036) each with or without FynY531F. Eighteen to twenty hours later, transfected cells were identified via green fluorescence protein fluorescence and selected for measurement of intracellular calcium responses. After loading with 5 M fura-2-acetoxymethyl ester (Molecular Probes) for 30 min, cells were washed with Hanks' balanced salt solution twice. Intracellular calcium was measured by a dual excitation wavelength method (340/380 nm) using a Nikon microscope in conjunction with Metafluor imaging software (Universal Imaging, West Chester, PA) (26).
Biotinylation Assays-Surface biotinylation was used to assess receptor expression on the cell surface. HEK293 cells, grown in 60-mm dishes, were transfected with various cDNA combinations. Twenty-four hours after transfection, cell surface proteins were labeled with biotin and isolated (16). Cells were washed twice with cold rinsing solution (phosphate-buff-ered saline with Ca/Mg (1 mM MgCl 2 and 0.1 mM CaCl 2 , pH 7.35)) (27) and incubated in rinsing solution containing 1 mg/ml EZ-Link Sulfo-NHS-Biotin (Pierce) for 30 min at 4°C. Cells were washed twice with quenching solution (rinsing solution with 100 mM glycine added), incubated in quenching solution for 20 min, and solubilized at 4°C in radioimmune precipitation buffer ((150 mM NaCl, 1 mM EDTA, and 100 mM Tris-HCl, pH 7.4), 1% (v/v) Triton X-100, 1% (w/v) deoxycholate, and 0.1% (w/v) SDS containing protease inhibitors (1 g/ml leupeptin, 250 M phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 mg/ml trypsin inhibitor, and 1 mM iodoacetamide). Samples were cleared by centrifugation at 16,300 ϫ g at 4°C for 20 min. A small sample of cleared lysates was saved for analysis (designated the lysate fraction); the remaining portion of lysates was incubated with avidin beads (Pierce) at 4°C overnight. After overnight incubation, samples were centrifuged at 16,300 ϫ g for 15 min, and the supernatant was saved as the intracellular fraction. The beads were washed once in radioimmune precipitation buffer, twice in cold high salt buffer (0.1% Triton X-100, 500 mM NaCl, 5 mM EDTA, 50 mM Tris, pH 7.5), and once in low salt buffer (50 mM NaCl, pH 7.5), and biotinylated proteins were eluted with SDS sample buffer (62.5 mM Tris-Cl, pH 6.8, 2% SDS, and 100 mM mercaptoethanol) at 37°C for 30 min. The remainder of the procedure followed standard Western blotting protocols. To confirm that biotin only labeled surface proteins, the integrity of the cell membrane during biotinylation was tested by immunoblotting with an anti-actin antibody. Actin immunoreactivity in the biotinylated fraction was 9 Ϯ 2% (n ϭ 24) of that in the whole cell lysate.
In selected experiments, biotinylations were performed under nondenaturing conditions by solubilizing cell membranes in radioimmune precipitation buffer (lysis buffer) lacking SDS. Without SDS in the radioimmune precipitation buffer, integral membrane proteins exposed to the extracellular space and thus labeled by biotin are not dissociated from membraneassociated proteins bound to such integral membrane proteins by noncovalent interactions. This allows us to possibly detect the changes in association of nonintegral membrane proteins (such as calpain) with biotinylated, cell surface-exposed proteins in different experimental conditions. Statistical Analysis-Data are shown as the mean Ϯ S.E. Experiments were analyzed using Student's t test (paired) to compare two conditions or ANOVA followed by planned comparisons of multiple conditions (28). For most comparisons, nonnormalized OD values were used. Significance was set at p Ͻ 0.05. All electrophysiology data, such as current amplitude, current desensitization, and time constant (fitted with a single exponential component function using the Levenberg-Marquardt nonlinear least squares algorithm), were calculated using Clampfit software (pCLAMP 8.01; Axon Instruments, Inc., Foster City, CA). We have designated "n" to refer to independent experiments in biochemical studies and independent single cells in electrophysiological studies.

Reduction of Calpain-mediated Proteolysis of NR2B by Inhibition of Src Family Kinase (SFK) Activity in
Neurons-SFKmediated tyrosine phosphorylation and calpain-mediated pro-teolysis occur in the NR2B C terminus (23, 29 -31). To test the possibility that phosphorylation of NR2B by SFK changes its sensitivity to calpain, we investigated the effect of PP2, a specific inhibitor of SFK (32,33), on calpain-mediated cleavage of NR2B. In primary hippocampal cultures (Fig. 1A), neurons pretreated with or without 10 M PP2 for 30 min had similar levels of full-length NR2B (180 kDa, detected by an N-terminal antibody), showing that inhibition of SFK activity did not acutely change NR2B protein expression. Exposure of cultured neurons to 100 M glutamate and 100 M glycine for 30 min in the absence of PP2 preapplication decreased full-length NR2B levels and generated a 115-120-kDa N-terminal breakdown product (previously shown to be generated by calpain) (16) (Fig. 1, A  and B). In parallel cultures, the application of 10 M PP2 30 min before glutamate and glycine stimulation decreased the loss of full-length NR2B and production of the breakdown product ( Fig. 1, A-C; full-length NR2B, without PP2, 48 Ϯ 10%; with PP2, 82 Ϯ 4%; p ϭ 0.0069; NR2B breakdown product, without PP2, 53 Ϯ 6%; with PP2, 30 Ϯ 4%; p ϭ 0.0068; n ϭ 6). To control for the possibility that PP2 application globally decreased calpain activity, we examined calpain-mediated spectrin break-down. Glutamate treatment produced similar levels of a 150-kDa calpain-generated spectrin breakdown product in neurons with or without PP2, showing that PP2 application did not decrease calpain activity toward all substrates (Fig. 1, A and D). Taken together, these results show that blocking SFK activity slows calpain-mediated cleavage of NR2B specifically and suggest that SFK activity controls NR2B subunit levels in hippocampal neurons in response to glutamatergic stimulation.
NR2B Is Phosphorylated at Tyrosine 1336 during NMDA Receptor Activation in Neurons-NR2B can be phosphorylated at Tyr-1336 and Tyr-1472 by Fyn, an SFK directly implicated in tyrosine phosphorylation of NMDA receptors in the brain (23,34). Since Fyn may change NR2 cleavage by calpain in vitro (35), we first tested whether Fyn mediated phosphorylation of NR2B at Tyr-1336, Tyr-1472, or both sites during NMDA receptor activation. Cultures were treated with glutamate and glycine, and the time course of Tyr-1336 and Tyr-1472 phosphorylation was determined using NR2B phosphospecific anti-Tyr-1336 or anti-Tyr-1472 antibodies (Fig. 2, A and B). Levels of phosphorylation of Tyr-1472 were similar over the period of 30 min of glutamate stimulation ( Fig. 2A). However, increased phosphorylation of Tyr-1336 was noted rapidly (by 7.5 min after NMDA receptor activation) and increased 2.2-fold by 30 min (0 min, 100 Ϯ 2%; 7.5 min, 151 Ϯ 22%; 30 min, 219 Ϯ 14%, p ϭ 0.0095 at 7.5 min and p ϭ 0.0004 at 30 min both versus 0 min, n ϭ 5 each) (Fig. 2B). The progressive phosphorylation of NR2B at Tyr-1336 was associated with a gradual reduction of full-length NR2B and generation of the NR2B breakdown product (Fig. 2, C and D). Thus, whereas the level of full-length NR2B declined with NMDA receptor activation, the immunoreactivity of phospho-Tyr-1336 NR2B increased. In addition, phosphorylation of Tyr-1336 in this condition is mediated by SFK, since PP2 (10 M) eliminated phospho-Tyr-1336 immunoreactivity at all of the indicated time points (Fig. 2, A and B). Furthermore, PP2 slowed both the reduction of full-length NR2B and the generation of NR2B-derived breakdown products over a time course that generally resembled that for phosphorylation of NR2B-Tyr-1336 ( Fig. 2, A, C, and D). As in our previous work (16,19), inclusion of a calpain inhibitor decreased NR2B breakdown (data not shown) but increased the amount of Tyr-1336 phosphorylation (Glu, 190 Ϯ 7%; CalI3 ϩ Glu, 330 Ϯ 18%; n ϭ 6; p ϭ 0.0016 versus Glu) (Fig. 2E), suggesting that NR2B-Tyr-1336 is a substrate for calpain. MK-801 also blocked Tyr-1336 phosphorylation (control, 100 Ϯ 6%; MK801 ϩ Glu, 94 Ϯ 8%; n ϭ 6, p ϭ 0.57 versus control; Fig. 2E). These results indicate that SFK-mediated phosphorylation and calpain-mediated proteolysis, which both regulate NR2B through the C terminus, are facilitated by NMDA receptor activation in neurons. Since SFK inhibition decreases calpain-mediated NR2B cleavage, this also suggests that SFK-mediated phosphorylation of NR2B directly accelerates the calpain-generated NR2B breakdown.
Fyn Accelerates Calpain-mediated NR2B Cleavage in Neurons-To assess whether calpain-mediated NR2B cleavage during NMDA receptor activation in neurons is controlled by Fyn, we decreased endogenous Fyn protein expression using an siRNA against Fyn coded in a DNA-based vector (36). The Fyn siRNA, which has no homology to other known rodent genes (based on BLAST searches of the sequences), decreased endog- . Neurons were harvested 30 min after the addition of agonists, and samples were subjected to SDS-PAGE and Western blot analysis with an N-terminal NR2B antibody (top) and AB38 (recognizing a calpaingenerated spectrin breakdown product) (bottom). The lower band in the top is a calpain-mediated NR2B breakdown product (16). B and C, bar graphs demonstrate the levels of full-length NR2B and NR2B-derived breakdown after the indicated treatments. The expression level of each control was normalized to 100% (based on full-length NR2B), and the levels of each band after various treatments were calculated as the percentages of corresponding control values. D, bar graph shows elevated levels of the calpain-generated spectrin breakdown after the indicated treatments. Data are means Ϯ S.E. from six experiments. *, p Ͻ 0.05; **, p Ͻ 0.001, ANOVA.
Because another SFK, Src, changes calpain-mediated cleavage of NR2A in vitro (35), we examined whether Src kinase can alter calpain-mediated cleavage of NR2B and phosphorylation of NR2B-Tyr-1336. Constitutively active Src (SrcCA) (SrcY527F) or a vector control was expressed in HEK cells with NR1/2A/NR2B/SAP-102 (Fig. 5D). Compared with the corresponding 0 min control, 30-min application of glutamate and glycine reduced the NR2B immunoreactivity to 70% of control levels (n ϭ 5; p ϭ 0.011) in cells transfected with SrcCA and to 72% (n ϭ 5; p ϭ 0.019) in cells transfected with a vector control. Calpain-generated spectrin breakdown increased to the same extent at 30 min after agonist application in both transfected conditions (Fig. 5D). Both the reduction of NR2B levels and the increase in spectrin breakdown were completely blocked by pretreatment with calpain inhibitor III (data not shown). The immunoreactivity of actin did not change before or after agonist application in any transfected condition (Fig. 5D). In addition, immunoreactivity of phosphorylated NR2B Tyr-1336 was undetectable before or after agonist treatment in cells either with or without SrcCA coexpression, showing that NR2B Tyr-1336 was not a substrate for Src kinase in this paradigm (data not shown). Taken together, these results indicate that in HEK cells, SrcCA does not alter calpain-mediated cleavage of NR2B in this transfected cell paradigm; nor does it readily mediate phosphorylation of Tyr-1336 after NMDA receptor activation induced by glutamate and glycine. Thus in this system, Fyn (but not all SFKs) accelerates calpain-mediated NR2B cleavage.
Cellular Location of Fyn Facilitation of Calpain-mediated NR2B Cleavage in HEK Cells-Since activation of NMDA receptors leads to calpain-mediated NR2B cleavage in HEK cells in both the plasma membrane and intracellular compartments (16,24), we next investigated whether Fyn-mediated facilitation of NR2B cleavage occurred in a selective cellular compartment in HEK cells. Cells expressing NR1/NR2A/NR2B with or without FynY531F were treated with glutamate and glycine. Surface receptors were biotinylated, separated using avidin beads, and detected by Western blotting using anti-N-terminal NR2B antibody. In each transfected condition (Fig. 7, B  and C), before agonist application, NR2B was found in both the cell surface fraction and the intracellular fraction. Actin was detected almost exclusively in the intracellular fraction, confirming that biotinylation occurred only at the cell surface (Fig. 7, B and C). Before agonist treatment, in both the surface and the intracellular fractions, the levels of immunoreactivity of NR2B in cells transfected with NR1/2B with or without FynY531F were similar (Fig. 7, B and C). This suggests that basal Fyn-mediated phosphorylation of NR2B Tyr-1336 does not change the NR2B levels in each compartment in this system, including the subunit delivery to the cell surface. Following 30-min agonist treatment, the immunoreactivity of NR2B in the overall lysate declined compared with the corresponding control (0 min). (Fig. 7A, 35% decrease without FynY531F; 63% decrease in wild type with FynY531F; n ϭ 6 each). In the cell surface fraction, agonist application decreased NR2B immunoreactivity by 33% in cells transfected with NR2B wild type without FynY531F and by 62% in cells transfected with NR2B wild type with FynY531F, suggesting that NR2B cleavage is enhanced by FynY531F in the surface fraction ( Fig. 7C; NR1/2B without FynY531F, 67 Ϯ 2%; NR1/2B with FynY531F, 38 Ϯ 3%; p ϭ 0.0063; n ϭ 6 each). In the intracellular fraction, the decrease in levels of NR2B by agonist application was similar among wild-type NR2B with or without FynY531F ( Fig. 7B; NR1/2B without FynY531F, control, 100 Ϯ 3%; Glu, 41 Ϯ 4%; NR1/2B with FynY531F, control, 100 Ϯ 7%; Glu, 40 Ϯ 2%; n ϭ 6 each). Overall, the level of NR2B cleavage in the intracellular fraction in the absence of FynY531F was similar to the level cleavage with FynY531F in the surface fraction. This suggests that Fyn may relieve a tonic inhibition of cleavage of NR2B in the surface fraction. Overall, these results show that FynY531F mediated facilitation of NR2B cleavage in the cell surface fraction without significant effects on cleavage of intracellular receptors.

The Mechanism of Fyn-mediated Facilitation of NR2B
Cleavage-Since cleavage of intracellular receptors was not enhanced by Fyn, we sought to better define the mechanism by which Fyn enhances calpain-generated cleavage of NR2B. In vitro, Fyn enhanced the speed of NR2B cleavage (Fig. 8, A and  B). This increase in the rate of cleavage was not noted in receptors formed from NR2BY1336F, demonstrating that the mechanism of the effects of Fyn in vitro is consistent with that observed in cell culture. Consequently, the effect of Fyn probably does not require processes observed only in situ (such as trafficking of NMDA receptors between cellular compartments). Additionally, FynY531F expression did not increase levels of NR2B on the cell surface either before or 30 min after agonist application in which calpain activity was blocked by its inhibitor, indicating that Fyn does not facilitate NR2B delivery to the cell membrane in this system ( Fig. 8C; wt/vector, control, 100 Ϯ 0.2%; Glu, 94% Ϯ 9; p ϭ 0.27 versus control; wt/FynY531F, control, 93 Ϯ 11%; p ϭ 0.44 versus control of wt/vector; Glu, 93 Ϯ 9%; p ϭ 0.24 versus control of wt/vector; n ϭ 6 each). In addition, NR2B Tyr-1336 was phosphorylated to a similar degree in both cell surface and intracellular fractions ( Fig. 8D; surface, control, 100 Ϯ 6%; Glu, 248 Ϯ 24%; intracellular, control, 100 Ϯ 8%; Glu, 210 Ϯ 18%; n ϭ 6 each). Thus, the selective enhancement of surface receptor cleavage by Fyn cannot simply reflect the distribution of phosphorylation. However, since calcium entry facilitates association of calpain with cell membranes, including the plasma membrane (37, 38), we FIGURE 7. Cellular location of the effect of Fyn on calpain-mediated NR2B cleavage in HEK cells. A-C, representative blots and histogram showing NR2B levels in different cellular compartments in cells transfected with NR1/2B with or without FynY531F before (control) and 30 min after agonist application. Quantification of the NR2B levels by biotinylation showed that NR2B levels decreased to a greater degree in the lysate (A) and the surface fraction (C) in cells coexpressing FynY531F than in those expressing a vector control. In the intracellular fraction (B), quantification showed that the levels of NR2B with or without FynY531F decreased to a similar extent after agonist application. Biotinylation data were obtained from six independent experiments performed in triplicate. Error bars, S.E. wt, wild type. investigated whether NMDA receptor activation translocated calpain I from the cytosol to the membrane area using biotinylation assays under nondenaturing conditions before and after agonist stimulation. This protocol should leave associations of calpain with membrane components intact and lead to retention of cell surface-associated calpain in the biotinylated fraction. In transfected HEK cells, 100 M glutamate treatment increased the calpain I immunoreactivity in the cell surface fraction 1.9-fold relative to levels before agonist stimulation, with no significant changes of calpain I levels in the intracellular fraction or the lysate ( Fig. 8E; surface, CalI3, 3300 Ϯ 100; CalI3 ϩ Glu, 6220 Ϯ 380; intracellular, CalI3, 11,600 Ϯ 4600; CalI3 ϩ Glu, 12,400 Ϯ 140; lysate, CalI3, 19,300 Ϯ 360; CalI3 ϩ Glu, 20,800 Ϯ 400). Actin levels in the surface fraction were not detectably altered by 100 M glutamate application for 30 min.
Thus, overall, the facilitation of calpain-mediated cleavage of NR2B appears to reflect an increased sensitivity of NR2B to calpain with phosphorylation at Tyr-1336 and the translocation of activated calpain to the cell membrane area, creating a selective enhancement of cleavage in the cell surface fraction.
Cleavage of Full-length NR2B Does Not Decrease Extrasynaptic NMDA Receptor Function-Previously, we demonstrated that calpain does not control the overall size of electrophysiological responses of NMDA receptors in DIV17 hippocampal neurons (19). However, in hippocampal neurons, extrasynaptic NMDA receptors preferentially contain NR2B (39) linked to SAP-102, not PSD-95. This makes extrasynaptic receptors the most likely selective physiologic target of neuronal calpain. We therefore investigated whether calpain-mediated cleavage of the NR2B subunit alters extrasynaptic NMDA receptor function. To monitor pure extrasynaptic NMDA currents in cultures of hippocampal neurons, we selectively blocked synaptic NMDA receptors with an MK801trapping technique that eliminates the contribution of synaptic receptors from the whole-cell response (39 -41). In this paradigm, however, we also wished to examine whether activation of calpain would alter extrasynaptic currents, a goal requiring a control ensuring that the conditions needed to activate calpain did not unblock synaptic receptors. Since extrasynaptic receptors contain largely NR1/2B in hippocampal neurons, one can ensure that response reflects extrasynaptic receptors by their sensitivity to ifenprodil. Consistent with previous reports, blocking synaptic NMDA receptors with 50 M bicuculline in conjunction with 10 M MK-801 for 10 min reduced NMDA receptor current density by 69% from the control level of 107 Ϯ 10 pA/pF (n ϭ 12) to 33 Ϯ 5 pA/pF (n ϭ 18) (Fig. 9A). The residual currents were almost completely abolished (5 Ϯ 1 pA/pF, n ϭ 5) by ifenprodil (10 M), a selective antagonist of NR1/2B receptors, indicating that the whole-cell NMDA-evoked responses after bicuculline/MK801 pre-blocking reflected extrasynaptic currents consisting mainly of NR1/ NR2B heteromers (Fig. 9A). With MK801/bicuculline pretreatment, neurons showed no apparent increase in calpain activity based on spectrin cleavage and cleavage of NR2B (Fig. 9B). However, if 100 M glutamate was then applied to these neurons for 30 min, there was loss of full length NR2B, generation of a 115 kDa NR2B truncation and cleavage of spectrin, showing that isolated NR1/2B extrasynaptic receptors can activate calpain (Fig. 9B). We then compared extrasynaptic NMDA receptor responses before and after calpain-mediated NR2B cleavage in bicuculline/MK801 pre-blocked neurons. As shown in Fig.  9A, after application of 100 M glutamate for 30 min in bicuculline/MK801 pre-blocked neurons, the NMDA receptor current density was 34 ؎ 4 pA/pF (n ϭ 21), similar to the corresponding value obtained from bicuculine/MK801 pre-blocked neurons without glutamate treatment (33 Ϯ 5 pA/pF; n ϭ 18, p Ͼ 0.8). Inhibition of calpain activation with calpain inhibitor III significantly prevented the glutamate-induced decrease of full-length NR2B (Fig. 9B) but produced similar levels of extrasynaptic NMDA-evoked response (40 Ϯ 5 pA/pF, n ϭ 19, p Ͼ 0.3), compared with either glutamate-treated or untreated neurons (Fig. 9A). Current evoked by NMDA after glutamate application for 30 min with or without calpain inhibitor III in bicuculine/MK801 preblocked neurons was completely blocked by ifenprodil (Fig. 9A), indicating no recovery of synaptic NMDA receptors following removal of bicuculline/MK801 during glutamate and calpain inhibitor III treatment (since these receptors typically are ifenprodil-insensitive). This is consistent with previous studies showing that synaptic receptors remain blocked for this period after glutamate application (41). Thus, reducing levels of full-length NR2B without a corresponding change in the current amplitude of response to NMDA indicates that calpain-mediated loss of full-length NR2B does not cause a functional decrease in extrasynaptic NMDA receptor.
Calpain-truncated NR2B-containing NMDA Receptors Are Functional in HEK Cells-We then attempted to confirm that receptors containing calpain-truncated NR2B remain active. Based on the expected molecular weight of the truncated form (roughly 1030 amino acids) (16) generated by calpain in neurons, we produced a construct leading to NR2B truncated at amino acid 1036 (NR2Bdel1036) by insertion of two consecutive stop codons and then examined its functional properties. In HEK cells expressing NR1/NR2Bdel1036, Western blots of biotinylated samples demonstrated that the truncated NR2B subunit was expressed on the cell surface (Fig. 9C). In addition, NMDA receptors incorporating NR2Bdel1036 produced intracellular calcium transients similar to wild type receptors when stimulated with agonists (data not shown). Electrophysiologically, brief (2-s) application of 100 M NMDA activated both wild type-and NR2Bdel1036-containing receptors with no significant difference in current density (current amplitude normalized to cell capacitance; wild-type, 90 Ϯ 15 pA/pF, n ϭ 24; NR1/NR2Bdel1036, 103 Ϯ 13.3 pA/pF, n ϭ 35). In addition, the percentage of current desensitization of NR1/NR2Bdel1036 was statistically indistinguishable from the values found in cells expressing wild-type receptors (Fig. 9D). This result indicates that calpain-truncated NR2B-containing NMDA receptors are functional in HEK cells. These data also demonstrate that Fynmediated phosphorylation of NR2B facilitates the generation of truncated receptors that function similarly to full-length receptors but lack the domains necessary for intracellular signal transduction, which may play critical roles in synaptic plasticity or excitotoxicity.

DISCUSSION
In the present study, we have demonstrated that the tyrosine kinase Fyn controls the ability of the NR2B subunit to be cleaved by calpain in neurons and in model expression systems. In neurons, SFK inhibitors and siRNA specific for Fyn slow calpain-mediated cleavage of NR2B but do not alter cleavage of spectrin. In HEK cells, calpain-mediated cleavage is accelerated by constitutively active Fyn but not Src. Calpain-mediated cleavage is enhanced in association with phosphorylation of NR2B Tyr-1336 but not NR2B Tyr-1472, and site-directed mutagenesis demonstrates that removal of Tyr-1336 (but not Tyr-1472) removes the ability of Fyn to potentiate NR2B cleavage by calpain. These data demonstrate that Fyn enhances calpain-mediated cleavage of NR2B in multiple experimental conditions through a mechanism involving direct phosphorylation of NR2B.
The interactions of Fyn, NR2B, and calpain are also controlled by the specific MAGUK protein involved in NR2B binding. Both SAP-102 and PSD-95 enhance phosphorylation of NR2B Tyr-1336 by Fyn. However, as shown previously, PSD-95 blocks the ability of calpain to cleave NR2 subunits (24), and phosphorylation by Fyn was unable to override this effect. This suggests that in neurons, the NR2B cleaved by calpain is most likely bound to SAP-102 and phosphorylated by Fyn. Based on the requirement for SAP-102 (compared with PSD-95) and the preference for cleavage of NR2B rather than NR2A, most calpain-cleaved NMDA receptors are likely to be extrasynaptic NR1/2B receptors. In addition, our data show that such receptors (both extrasynaptic receptors in neurons and truncated NR1/2B in heterologous cells) are predicted to have electrophysiological features similar to full-length receptors. This shows that the primary effect of calpain in our system is not simply degradative but may create receptors with novel properties based on the dissociation of channel activity from downstream messenger systems, such as p38 mitogen-activated protein kinase dephosphorylation and phosphatidylinositol 3-kinase activation. Such receptors could then play crucial roles in excitotoxicity and synaptic modification. The mechanism by which Fyn controls cleavage of NR2B may have several different components. Fyn accelerated the cleavage in vitro of recombinant receptors in a manner that depended on NR2B Tyr-1336, suggesting that Fyn directly enhances the substrate characteristics of the C terminus of NR2B. This has been noted previously for NR2A and NR2B in brain homogenates, and in fusion proteins of NR2B from 1269 -1482 (35), suggesting that the effect of Fyn is mediated entirely within these amino acids and matching the dependence on NR2B Tyr-1336 demonstrated here. These data suggest that at least some aspect of the mechanism of the effects of Fyn is a direct biochemical effect on the substrate properties of the C-terminal region of NR2B.
However, in HEK cells, the subcellular localization of the enhancement of calpain-mediated cleavage suggests other aspects to the mechanistic effects of Fyn. Although calpain can cleave NR2B within intracellular compartments and Fyn can phosphorylate NR2B Tyr-1336 in intracellular compartments, Fyn did not enhance calpain-mediated cleavage in these compartments. This is consistent with the localization of NR2B in neurons, where most NR2B is located on the cell surface in hippocampal cultures (16), but suggests that factors beyond biochemical aspects contribute to the effects of Fyn on calpainmediated cleavage of NR2B. One factor could be increased trafficking of NR2B to the cell surface after activation of Fyn. However, the presence of constitutively active Fyn (with resultant base-line phosphorylation of Tyr-1336) did not increase levels of NR2B on the cell surface; nor did agonist treatment of receptors (with Fyn activation) during calpain blockade alter surface NR2B levels. Thus, increased trafficking of NR2B to the cell surface cannot explain the selective enhancement of cleavage by Fyn in this compartment. Instead, calpain, previously reported to associate with membranes after activation (37,38), increased in the cell surface fraction under nondenaturing conditions (Fig. 8E), suggesting that targeting of calpain to the plasma membrane mediates the selective enhancement of cleavage in the cell surface fraction. The regulatory subunit of calpain has been associated with synaptic membranes in proteomic experiments (42). These results suggest that the biochemical enhancement of NR2B degradation by Fyn in conjunction with calpain transport to the plasma membrane provides the mechanism for selective enhancement of cleavage by Fyn in the cell surface fraction. However, calpain also can associate with internal membranes after rises in intracellular calcium, suggesting that the translocation of calpain to the plasma membrane may only be a partial explanation for the compartmentalized effect of Fyn. The level of cleavage of NR2B in intracellular compartments was essentially equal to that observed with enhancement by Fyn in the surface fraction, suggesting that unidentified plasma membrane components tonically inhibit cleavage of NR2B. It is conceivable that such components (structural features of the attachment of NR2B to MAGUK proteins or plasma membrane lipids, association with RACK1 (receptor for activated protein kinase C1) or related proteins) selectively inhibit calpain-mediated NR2B cleavage in a manner that is relieved by phosphorylation by Fyn.
Fyn modulates NMDA receptors in several ways (33,(43)(44)(45)(46). Although Fyn has not clearly been shown to alter channel function directly, Fyn enhances transport of NR2 subunits to the cell surface and into the synapse under different conditions and blocks internalization of NR2B-containing receptors (47). However, these effects are most likely mediated by Tyr-1472 based on site-directed mutagenesis (23,45,46). The control of calpain-mediated cleavage appears to reflect events occurring with phosphorylation of Tyr-1336, a site that is phosphorylated in cellular systems to a slightly lower degree than Tyr-1472. This site also controls activation of phosphatidylinositol 3-kinase and dephosphorylation of p38 mitogen-activated protein kinase, potential protective mechanisms in neurons (22). This suggests that the events occurring with Tyr-1336 phosphorylation provide a set of responses distinct from the role of Fyn in organizing NR2B localization at the synapse through receptor trafficking and internalization.
The effect of Fyn here does not entirely match in vitro results examining calpain-mediated cleavage of NR2 subunits. Whereas Fyn accelerated cleavage of NR2B subunits in vitro, the same effect is seen on NR2A. However, the selective binding of NR2A to PSD-95 in synapses limits its ability to be cleaved in neurons (24). Src blocks calpain-mediated cleavage in vitro (through an unknown mechanism), but constitutively active Src did not lead to phosphorylation at Tyr-1336 in situ; nor did it alter the speed of cleavage of NR2B by calpain. This result agrees with the observation that among MAGUK proteins, only PSD-95 binds Src (48). Our data illustrate that the interaction of MAGUK proteins and SFK regulates the cleavage of NR2 subunits by calpain in living cells and that when calpain-mediated cleavage of NR2B is examined in situ, only Fyn alters this process.
Fyn significantly contributes to tyrosine phosphorylation of NMDA receptors, and Fyn mutant mice show defects in NMDA receptor-dependent LTP (43). However, NMDA receptor activation itself has not been shown to stimulate Fyn activity. In the present study, phosphorylation of NR2B Tyr-1336 increased with NMDA receptor stimulation in cultured hippocampal neurons or HEK cells expressing NR2B-containing NMDA receptors. These effects were blocked with MK801 and suggest that NMDA receptors could directly activate Fyn. However, NMDA receptor stimulation did not enhance phosphorylation of NR2B Tyr-1472, suggesting that NMDA receptor stimulation may instead direct the activity of Fyn to a different site on the NMDA receptor. Selective Fyn-mediated phosphorylation of Tyr-1336 is also seen in neurons treated with ethanol, in which it is controlled by dissociation of RACK1 from NR2B (33). Nevertheless, the present results indicate that the NR2B/Fyn/calpain interactions may act as part of a Fyn kinase cascade, which in turn regulates NMDA receptor features through promoting calpain-mediated cleavage of NR2B subunits. Further experiments are required to investigate the mechanisms by which NMDA receptor activation is coupled to Fyn-mediated phosphorylation of NR2B Tyr-1336 and calpainmediated cleavage of NR2B in neuronal pathophysiology.
In summary, the present study provides a model for understanding the control and selectivity of neuronal calpain for selected proteins in synaptic and extrasynaptic membranes. Since calpain-cleaved receptors dissociated from intracellular signal transduction elements remain on the cell surface and are active similarly to those of wild-type receptors, the present data demonstrate a new manner in which Fyn may contribute to enhancement of synaptic plasticity or excitotoxicity in neuronal systems.