H-Ras Modulates N-Methyl-D-aspartate Receptor Function via Inhibition of Src Tyrosine Kinase Activity*

Tyrosine phosphorylation of the NR2A and NR2B subunits of the N-methyl-d-aspartate (NMDA) receptor by Src protein-tyrosine kinases modulates receptor channel activity and is necessary for the induction of long term potentiation (LTP). Deletion of H-Ras increases both NR2 tyrosine phosphorylation and NMDA receptor-mediated hippocampal LTP. Here we investigated whether H-Ras regulates phosphorylation and function of the NMDA receptor via Src family protein-tyrosine kinases. We identified Src as a novel H-Ras binding partner. H-Ras bound to Src but not Fyn both in vitro and in brain via the Src kinase domain. Cotransfection of H-Ras and Src inhibited Src activity and decreased NR2A tyrosine phosphorylation. Treatment of rat brain slices with Tat-H-Ras depleted NR2A from the synaptic membrane, decreased endogenous Src activity and NR2A phosphorylation, and decreased the magnitude of hippocampal LTP. No change was observed for NR2B. We suggest that H-Ras negatively regulates Src phosphorylation of NR2A and retention of NR2A into the synaptic membrane leading to inhibition of NMDA receptor function. This mechanism is specific for Src and NR2A and has implications for studies in which regulation of NMDA receptor-mediated LTP is important, such as synaptic plasticity, learning, and memory and addiction.

The N-methyl-D-aspartate (NMDA) 1 receptor is a ligandgated calcium channel that plays an essential role in neuronal development, addiction, and learning and memory (1,2). Tyrosine phosphorylation modulates NMDA receptor function; for example, inhibition of tyrosine kinase activity decreases NMDA receptor-mediated currents, whereas treatment with tyrosine phosphatase inhibitors increases these currents (3). Tyrosine phosphorylation of NR2A and NR2B occurs via Fyn and Src protein-tyrosine kinases (4,5), resulting in potentiation of NMDA channel activity (6,7). NMDA receptor phosphorylation by Src and Fyn is modulated by many mechanisms including protein kinase C (8) and the scaffolding protein RACK1 (9). Thus, the modulation of NMDA receptor function by Src protein-tyrosine kinases likely involves the convergence of diverse signaling pathways generating a complex mechanism of receptor regulation.
Glutamate activation of the NMDA receptor and the subsequent increase in intracellular calcium are essential for the induction of long term potentiation (LTP), a candidate mechanism underlying synaptic plasticity (10) thought to mediate learning and memory (11,12). NMDA receptor-mediated LTP is associated with an increase in tyrosine phosphorylation of NR2 subunits via a mechanism that requires Src family protein-tyrosine kinases (5). In the CA1 region of the hippocampus, activation of Src occurs within 5 min of LTP induction, and LTP can be prevented by application of Src-specific inhibitors (13). In addition, Fyn knockout mice show impaired LTP, which can be rescued by introduction of a Fyn transgene (14,15). Thus, both Src and Fyn are necessary for the induction of LTP (13)(14)(15)(16).
Recently, increased tyrosine phosphorylation of NR2A and NR2B and subsequent enhanced LTP was observed in the hippocampus of H-Ras null mice (17). H-Ras functions as a molecular switch, existing in an active GTP-bound or inactive GDP-bound form (18). It is an upstream initiator of the mitogen-activated protein kinase pathway, which in neurons can be activated by the influx of calcium resulting from activation of the NMDA receptor (19 -21). Many components of the H-Rasmitogen-activated protein kinase pathway are associated with the postsynaptic density (PSD (22)), a region of the synapse packed with signal transduction complexes, including Src family protein-tyrosine kinases, which colocalize with the NMDA receptor (23)(24)(25). A number of Ras family regulatory proteins have also been identified in the PSD, suggesting that the H-Ras-mitogen-activated protein kinase pathway may have varied synapse-specific functions (26 -30). Because NMDA receptor channel activity is positively regulated by Src tyrosine phosphorylation that is essential for LTP, we hypothesized that H-Ras negatively regulates NMDA phosphorylation and function via inhibition of a Src protein-tyrosine kinase. Here we present data showing that Src and H-Ras interact in vitro and in the brain. The interaction is specific for Src because Fyn, another member of the Src family expressed in the PSD, was unable to bind H-Ras. H-Ras binds via the Src kinase domain and inhibits Src kinase activity, decreasing phosphorylation and, subsequently, the membrane level of NR2A. Furthermore, overexpression of H-Ras in hippocampal slices results in a decrease in NMDA-mediated LTP. Taken together our results imply that H-Ras negatively regulates NMDA receptor channel activity by decreasing the number of NR2A-containing NMDA receptors in the synaptic membrane.

EXPERIMENTAL PROCEDURES
Materials-Active recombinant Src and Fyn tyrosine kinases, H-Ras-GST-agarose, pUSE-SrcWT, pUSE-H-RasWT, Raf-1 H-Ras binding Domain GST beads, anti-Src-agarose, monoclonal anti-Src, anti-Fyn, and anti-H-Ras antibodies were purchased from Upstate Biotechnologies (Lake Placid, NY). Anti-Src[pY418] antibodies were purchased from BIOSOURCE (Camarillo, CA). Phosphatase inhibitor cocktails, polyclonal anti-H-Ras, and anti-c-Src antibodies were purchased from Sigma. Anti-HA antibodies, all secondary antibodies, protease inhibitor tablets, and Expand PCR system were purchased from Roche Applied Science. Sequencing and generation of primers was carried out by the Gallo Center molecular biology core. Restriction enzymes and TNT in vitro translation kit were purchased from Promega (Madison, WI). pGBK-T7 was purchased from Clontech (Palo Alto, CA). [ 35 S]methionine (15 mCi/ml, 3000 Bq) and Amplify were from Amersham Biosciences. LipofectAMINE PLUS was purchased from Invitrogen. L(-tk) cells stably transfected with NR1ϩNR2A were a generous gift from Merck Sharp and Dohme.
Animals-Srcϩ/Ϫ and FynϪ/Ϫ mice (129vImJ/C57BL6J hybrids) were purchased from Jackson Laboratories. FynϪ/Ϫ mice were mated in house with 129 wild type mice to generate Fynϩ/Ϫ mice. Srcϩ/Ϫ and Fynϩ/Ϫ mice were mated to generate Src Ϫ/Ϫ and FynϪ/Ϫ mice. The genotyping of mice was determined by reverse transcription-PCR analysis of products derived from tail mRNA. The mean age of animals used in this study was 4 weeks. Male Sprague-Dawley rats, 3-4 weeks old, were purchased from Simonsen.
In Vitro Translations-[ 35 S]Methionine-labeled proteins were generated in rabbit reticulocyte lysates (TNT kit, Promega) using the appropriate cDNAs. The translation reactions were analyzed by SDS-PAGE and fluorography.
In Vitro Pull-down Assay-30 units of Src kinase or 75 units of Fyn kinase were incubated with 5 g of H-Ras-GST-agarose or GST-Sepharose for 2 h at 4°C with mixing. Agarose pellets were washed (1ϫ PBS, 1% Triton X-100), and proteins were resolved by SDS-PAGE and analyzed by Western blotting using anti-Src (1:500), anti-Fyn (1:500), or anti-H-Ras (1:5000) monoclonal antibodies. The H-Ras-GST-agarose pull-down and competition assays using radiolabeled in vitro translated proteins were incubated and resolved as above. Gels were fixed in 40% methanol, 10% acetic acid, incubated in Amplify, and dried down. Radiolabeled proteins were detected by overnight fluorography (Ϫ80°C).
Immunoprecipitation-Immunoprecipitation was performed with 5 g of the appropriate antibodies and 500 g of protein homogenate as described previously (9). Transfection-L(-tk) cells were cultured on 100-mm plates, and expression of NR1 and NR2A was induced as previously described (32). When 60% confluent, cells were transfected with a total of 10 g of pUSE-SrcWT and pUSE-H-RasWT cDNA using LipofectAMINE PLUS in accordance with the manufacturer's instructions.
Preparation of Cell Homogenates-L(-tk) cells were washed once with cold PBS, harvested, and resuspended in lysis buffer (20 mM Tris-HCl, pH 7.5, 10 mM EGTA, 2 mM EDTA, 0.25 M sucrose, 1% deoxycholate, and protease and phosphatase inhibitors as per the manufacturer's instructions). Samples were sonicated briefly, and lysis was allowed to proceed for 30 min on ice. Determination of the protein concentration was made using a BCA kit (Pierce). In addition, samples were also normalized with respect to SrcWT by Western blot and densitometry (NIH Image Version 1.62).
Generation of Phospho-NR2A-specific Polyclonal Antibody-The NR2A-derived peptide RLLEGNFY(PO 3 )GSLFSV corresponding to amino acids 1318 -1331 was generated by SynPep and used to immunize rabbits. Test bleeds were made over the course of three months, after which terminal bleeds were taken. The antigenicity of the bleeds was analyzed by slot blot using the immunizing peptide and showed increasing recognition of the antigen over three months. The terminal bleed was tested in the same way and then subsequently purified using the phospho-peptide coupled to an inert support. Specificity for the phosphorylated NR2A epitope was analyzed by slot blotting the immunizing phospho-peptide and a non-phospho form of the same epitope. No cross-reaction with the non-phospho-peptide was detected.
Raf-1 Binding Assay-Raf-1 H-Ras binding domain GST-agarose and rat total brain homogenate (250 g) were diluted in PBS and incubated for 30 min at 4°C as per the manufacturer's instructions. Raf-1-H-Ras complexes were resolved by SDS-PAGE and analyzed by Western blotting using anti-H-Ras antibodies.
Preparation of Tat-H-Ras Fusion Protein-pTAT-H-RasWT was expressed in and purified from Escherichia coli as previously described (33). Tat fusion protein was detected using anti-HA antibodies.
Immunohistochemistry-After incubation with 1 M Tat-H-Ras, L(-tk) cells were washed in cold wash buffer (PBS, 0.1% Triton X-100), fixed in ice-cold methanol for 3 min, and blocked in wash buffer containing 0.3% normal goat serum for 4 h at room temperature. Immunofluorescence was performed using rat monoclonal anti-HA antibodies (1:100) incubated overnight at 4°C. Staining was detected with secondary antibodies conjugated to Texas Red (1:500) incubated for 2 h at room temperature in the dark. Slides were mounted using Vectashield and viewed with a Zeiss 510 meta laser-scanning confocal microscope. Images shown are individual middle sections of projected Z series and were processed using Adobe Photoshop (Adobe Systems Inc).
Electrophysiology-Transverse hippocampal slices (350 m) were prepared from 3-5-week-old male Sprague-Dawley rats. Slices were maintained for at least 2 h in aCSF that contained 126 mM NaCl, 1.2 mM KCl, 1.2 mM NaH 2 PO 4 , 0.01 mM MgCl 2 , 2.4 mM CaCl 2 , 18 mM NaHCO 3 , and 11 mM glucose saturated with 95% O 2 , 5% CO 2 at 25°C. After recovery, slices were submerged and continuously superfused with aCSF at 25°C. Field excitatory post-synaptic potentials (fEPSPs) were recorded from stratum-radiatum of CA1 region with glass microelectrodes filled with 2 M NaCl. Picrotoxin (100 M) was added to the bath solution to block GABA A receptor-mediated inhibitory postsynaptic potentials. To evoke fEPSPs, Schaffer collateral/commissural afferents were stimulated with 0.1-Hz pulses using steel bipolar microelectrodes at intensities adjusted to produce an evoked response that was 40 -50% that of the maximum-recorded fEPSP for each recording. LTP was induced by high frequency stimulation (100 Hz, 1-s duration, 2 trains at 10-s intervals) at the same intensity as the test stimulus, and synaptic responses were monitored for 60 min after LTP induction. Data were collected using an Axopatch-1D amplifier (Axon instruments), filtered at 2 kHz, and digitized at 5-10 kHz. Compiled data were analyzed and expressed as the mean percent of fEPSP slope ϮS.E. over the base-line levels.

Src and H-Ras Interact in Vitro and in Brain-
We set out to determine whether H-Ras negatively regulates the NMDA receptor through a Src-dependent mechanism. Previously Stancato et al. (34) observed that H-Ras coimmunoprecipitates with Src in Sf9 insect cells. Because both Src and H-Ras are present in the PSD (24), we hypothesized that H-Ras may regulate Src activity via a direct interaction. To confirm that Src and H-Ras interact directly, purified Src was incubated with H-Ras-GSTagarose or GST-Sepharose alone. H-Ras-GST-agarose was capable of "pulling down" Src (Fig. 1a, lane 3), whereas no inter-action was observed between Src and GST-Sepharose (Fig. 1a,  lane 4), suggesting a direct interaction between Src and H-Ras in vitro. To determine the specificity of this interaction, the pull-down experiment was repeated using another member of the Src family, Fyn, which shares 84% sequence homology with Src and is also expressed in the PSD (35). There was no significant interaction detected between Fyn and H-Ras (Fig. 1b,  lane 3), suggesting that the interaction between H-Ras and Src is specific and not a general property of the Src family of protein-tyrosine kinases.
Next, to confirm that the Src-H-Ras interaction occurred in brain, we performed co-immunoprecipitation studies in rat brain homogenate (Fig. 1c). We found that anti-Src antibodies were capable of co-immunoprecipitating H-Ras (lane 1) and, conversely, that anti-H-Ras antibodies formed an immune complex with Src (lane 2). In summary, we show that Src and H-Ras can interact in vitro and in rat brain.
H-Ras Interacts with Src via Its Kinase Domain-Src is comprised of a unique N-terminal region, an SH3 domain, an SH2 domain, and a C-terminal catalytic domain (36). To further characterize the SrcϪH-Ras interaction, we made constructs of various domains and expressed them as [ 35 S]methionine-labeled proteins by in vitro translation (Fig. 2a). Radiolabeled Src-SH3, Src-SH2, and Src-KD were incubated with H-Ras-GST-agarose and analyzed by fluorography. Src-KD, but not SrcSH2 or SrcSH3, bound to H-Ras-GST-agarose, indicating that the kinase domain of Src contains a binding site for H-Ras (Fig. 2a, lane 6). To confirm the identification of the binding site, a competition assay was carried out in which H-Ras-GST-agarose was preincubated with either control (in vitro translated empty plasmid) or non-radiolabeled in vitro translated kinase domain (KD; Fig. 2b). Radiolabeled kinase domain ( 35 S-labeled KD) was then added to the reaction, and the incubation was allowed to proceed. Assuming that the kinase domain is indeed the binding site for H-Ras, fewer binding sites would be available for 35 S-labeled KD binding after preincubation with unlabeled KD. Indeed, there was a decrease in 35 (Fig. 2, a and b), we tested whether the binding of H-Ras to Src affected Src kinase activity. L(-tk) mouse fibroblast cells were transiently transfected with combinations of SrcWT and H-RasWT cDNA. Immunoprecipitation using anti- Src-agarose confirmed that in these cells, Src co-immunoprecipitated with H-Ras (Fig. 3a, lane 2). Therefore Src activity was measured in the presence of H-Ras using anti-Src[pY418], an antibody that recognizes active autophosphorylated Src. In cells transfected with SrcWT alone, anti-Src[pY418] detected a robust signal corresponding to Src at 60 kDa (Fig. 3b, top panel,  lane 1), suggesting that transfected Src was active. Cotransfection of H-RasWT with SrcWT resulted in a sharp decrease in active Src (Fig. 3b, top panel, lane 2). Our results show that H-Ras significantly inhibits the ability of Src to autophosphorylate Tyr-418, implying that Src kinase activity is abrogated by the binding of H-Ras to the Src kinase domain.
Because NR2A is a substrate for tyrosine phosphorylation by Src (6), we analyzed the effect of H-Ras overexpression and the consequent inhibition of Src activity on the tyrosine phosphorylation state of NR2A. To do so we used L(-tk) cells stably transfected with NR1 and NR2A and transiently transfected them with combinations of SrcWT and H-RasWT cDNA. In addition, we generated a polyclonal rabbit antibody against tyrosine 1325-phosphorylated NR2A (41) and tested for the effects of H-Ras overexpression on basal level phosphorylation of NR2A. As predicted, there was a decrease in phosphorylated NR2A in SrcWT/H-RasWT-transfected cells compared with those transfected with Src alone (Fig. 3c). Thus, co-transfection with H-RasWT decreased Src kinase activity, consequently inhibiting Src-mediated NR2A phosphorylation.
Active Tat-H-Ras Transduces into Cultured Cells and Brain and Inhibits Endogenous Src Kinase Activity-To determine whether H-Ras modulates endogenous Src kinase activity and NR2A-mediated channel function, we used the Tat fusion protein transduction system (42) to elevate H-Ras protein levels in brain slices. We generated a Tat-H-Ras construct and expressed and purified the fusion protein from E. coli as described previously (33). First, we used immunofluorescence and

FIG. 3. Binding of Src and H-Ras causes inhibition of Src kinase activity and decreases NR2A phosphorylation in L(-tk) cells. a, L(-tk) cells transfected with
SrcWT Ϯ H-RasWT were lysed, and 500 g of total protein was incubated overnight with anti-Srcagarose (5 g). Immune complexes (lanes 1 and 2) were washed, resolved by SDS-PAGE, and analyzed by Western blotting with anti-Src or anti-H-Ras monoclonal antibodies. Total protein (50 g) was also included as a control for expression of transfected protein (lanes 3 and 4). n ϭ 3. IP, immunoprecipitates. b, L(-tk) cells transfected with SrcWT and H-RasWT were lysed and normalized with respect to Src by Western blotting. Lysates (ϳ50 g) from SrcWT and SrcWT/H-RasWT transfections were resolved by SDS-PAGE then analyzed by Western blot using anti-Src[pY418], an antibody that recognizes active Src (top panel). A loading control was run simultaneously (bottom panel). n ϭ 3. c, cell lysates (50 g) were resolved as before and analyzed using antiphospho NR2A (top panel) and anti-NR2A antibodies (middle panel). The bar histogram represents percentage NR2A phosphorylation normalized to total NR2A Ϯ S.D. n ϭ 3.
confocal microscopy to confirm that the fusion protein was capable of entering cells. Control and Tat-H-Ras (1 M)-transduced L(-tk) cells were fixed and analyzed using an antibody against an HA tag engineered into the Tat fusion protein (Fig.  4a). There was a clear antibody signal detected in cells treated with Tat-H-Ras (right panel) but not in control (left panel). Because the images shown represent the middle sections of a projected Z series, Tat-H-Ras has been successfully transduced through the cell membrane. Next, we tested Tat-H-Ras transduction in brain slices. Coronal rat whole brain slices were incubated in the absence or presence of Tat-H-Ras, homogenized, and fractionated to P2 (crude synaptosomes) and S2 (cell cytosol, light membranes) fractions. The integrity of the fractions was confirmed using an antibody against PSD-95 (data not shown), a protein enriched in the PSD (43), and Tat-H-Ras was detected in the P2 pellet fraction (Fig. 4b, lane 3) correlating with previous data showing that H-Ras is localized to the membrane (44,45). To determine whether transduced Tat-H-Ras was functional, we utilized the Raf-1 binding assay (Fig.  4b). Initiation of the mitogen-activated protein kinase pathway requires H-Ras to be active (i.e. GTP-bound) before it can bind to and activate Raf-1, a kinase downstream of H-Ras (46). Thus, the activation state of Tat-H-Ras can be determined by its binding to the H-Ras binding domain on Raf-1. We found that transduced Tat-H-Ras did indeed bind to Raf-1 H-Ras binding domain (Fig. 4b, lane 3), suggesting that Tat-H-Ras was successfully transduced into brain, activated, and correctly compartmentalized. Because transfection of H-Ras caused a decrease in Src kinase activity in L(-tk) cells (Fig. 3a), we examined Src kinase activity in Tat-H-Ras-treated brain slices using anti-Src[pY418] antibodies. The pY418 signal was reduced in slices treated with Tat-H-Ras (Fig. 4c, top panel, lanes 1 and 3), implying that transduction of Tat-H-Ras decreases Src kinase activity through in vivo interaction with the Src kinase domain.
Tat-H-Ras Decreases NR2A Retention and NR2A Phosphorylation in Synaptic Membranes-Interestingly, we found that the level of NR2A in the crude synaptosomal membrane fraction (P2) was decreased after transduction of Tat-H-Ras (Fig.  5a, top panel, lanes 1 and 3). In addition, there was a concomitant decrease in phosphorylated NR2A on Tat-H-Ras treatment (Fig. 5a, middle panel, lanes 1 and 3), consistent with the observations made for NR2A in L(-tk) cells (Fig. 3c). Because Src is capable of phosphorylating NR2B as well as NR2A in vitro (47) and because inhibition of Src-mediated phosphorylation by H-Ras decreased NR2A in the synaptic membrane, we examined whether H-Ras affects retention of NR2B in the synaptic membrane. There was no change in NR2B levels in the synaptic membrane after Tat-H-Ras treatment (Fig. 5b), suggesting that Src-mediated tyrosine phosphorylation plays a major role in specifically retaining NR2A-containing NMDA receptors in the membrane.
Retention of NR2A in the Membrane Is Mediated by a Srcspecific Mechanism-To confirm that the observed membrane depletion of NR2A was mediated via a Src tyrosine kinase mechanism, we transduced Tat-H-Ras into brain slices from Src Ϫ/Ϫ and Fyn Ϫ/Ϫ mice. Tat-H-Ras-transduced slices from Fyn Ϫ/Ϫ mice showed decreased NR2A in the crude synapto- somal P2 fraction (Fig. 6a) as observed in Tat-H-Ras-treated brain slices (Fig. 5a, top panel, lanes 1 versus 3). However, no change was seen in membrane-associated NR2A in Src Ϫ/Ϫ mice (Fig. 6b), implying that the retention of NR2A in the membrane is mediated by Src and not Fyn. There was no change in levels of NR2B in the membrane on Tat-H-Ras treatment (Fig. 6c), consistent with the results observed in rat brain slices (Fig. 5b). Therefore, not only is H-Ras inhibition a Srcspecific effect, the downstream consequences of this inhibition are specific for NR2A-containing NMDA receptors. Furthermore, even though the overall amount of NR2A in the Src Ϫ/Ϫ was significantly less than in the Fyn Ϫ/Ϫ mice (data not shown), NR2A was increased in the S2 fractions of Src Ϫ/Ϫ mice (Fig. 6b, lanes 2 and 4) compared with the Fyn Ϫ/Ϫ mice (Fig. 6a, lanes 2 and 4), supporting the hypothesis that Src is required for efficient membrane retention of NR2A.
Tat-H-Ras Inhibits the Induction of LTP in Hippocampal Slices-Manabe et al. (17) report an increase in NMDA receptor-mediated LTP in H-Ras Ϫ/Ϫ mice. Src is required for the induction of LTP (13), and mice lacking NR2A show reduced responses to LTP-inducing stimuli (48). Because Tat-H-Ras inhibited Src kinase activity and reduced NR2A levels in the membrane, we predicted that elevation of intracellular Ras levels by treatment with Tat-H-Ras would result in inhibition of LTP. fEPSPs were recorded in the CA1 region of the hippocampus from control and Tat-H-Ras-treated slices. A stable base line was established (Fig. 7b, traces 1 and 3), and LTP was induced by tetanic stimulation of afferent fibers (Fig. 7a). The magnitude of LTP in Tat-H-Ras slices was 45% that measured in control untreated slices (Fig. 7, a and b, traces 2 versus 4) or slices treated with a control Tat fusion protein (Tat-KIP 27 ; data not shown). Thus, elevating levels of H-Ras in brain inhibits Src kinase activity, reduces NR2A membrane retention and phosphorylation, and consequently inhibits NMDA receptormediated long term potentiation. DISCUSSION Tyrosine phosphorylation of the NMDA receptor by Src protein-tyrosine kinases plays an important role in the modulation of receptor function and is essential for the regulation of synaptic plasticity (1). We present here data to suggest a mechanism for the regulation of Src via the small GTP-binding protein H-Ras, which can act as a novel "switch off" mechanism for Src-mediated phosphorylation of the NMDA receptor (Fig. 8).
We found that H-Ras interacts with Src both in vitro and in brain. H-Ras bound to the Src kinase domain and decreased Src autophosphorylation. Transduction of Tat-H-Ras into brain slices reduced endogenous Src kinase activity, NR2A phosphorylation, and NR2A retention in the synaptic membrane. Finally, treatment with Tat-H-Ras was capable of inhibiting NMDA receptor-mediated LTP in hippocampal slices. Our results suggest that the increase observed in NMDA receptor-mediated LTP in H-Ras null mice (17) may be because of increased Src phosphorylation of NR2A and the maintenance of NR2A-containing NMDA receptors in the membrane. The absence of H-Ras prevents the inhibition of Src kinase activity and, consequently, the loss of NR2A from the membrane. Our results also correlate with LTP data from SynGAP ϩ/Ϫ mouse hippocampal slices (49). SynGAP is a GTPase-activating protein enriched in the PSD that negatively regulates the activity of synaptic Ras (26,27). In mice heterozygous for SynGAP, implying more Ras in its GTP-bound active form, the induction of LTP is inhibited (49). This supports our observations suggesting that increased Ras in the hippocampus negatively regulates LTP (Fig. 7).
The mechanism by which inhibition of Src phosphorylation leads to a change in the level of NR2A in the synaptic membrane remains unknown. Because Src family-mediated NR2 phosphorylation is required for trafficking of NR2 subunits to the postsynaptic membrane (31,50), it is possible that inhibition of tyrosine-phosphorylated NR2A prevents its initial insertion into the membrane. However, we do not see a corresponding increase of NR2A in the cytosolic S2 fraction after Tat-H-Ras treatment (Fig. 5a, top panel, lane 4), which would occur if NR2A were unable to localize to the membrane.
Recently it has been proposed that protein turnover in the PSD regulates synaptic activity and structure (51,52). A known mediator of protein degradation in the PSD is calpain, a protease that is activated in response to increases in intracellular calcium (53). Calpain degradation promotes restructuring of the PSD (54), and some of the major constituents of the PSD, e.g. PSD-95 and the intracellular C termini of both NR2A and NR2B, are substrates for calpain-mediated proteolysis (55,56). Src and Fyn tyrosine phosphorylation exert a protective effect from calpain degradation on NR2A and NR2B, respectively (57). Therefore, inhibition of Src phosphorylation of NR2A by H-Ras may promote its degradation by calpain and result in the loss of NR2A from the synaptic membrane.
Interestingly, the inhibitory effect of H-Ras is specific for Src because Fyn was not capable of binding H-Ras in vitro (Fig. 1b), and the reduction of NR2A in the membrane fraction after Tat-H-Ras treatment was still apparent in FynϪ/Ϫ mice (Fig.  6a). In addition, because no loss of NR2B from the membrane was observed in either rat (Fig. 5b) or Src Ϫ/Ϫ mouse brain slices (Fig. 6c) after Tat-H-Ras treatment, our results suggest that the action of H-Ras is specific for NR2A. Although the sequences of Src and Fyn are highly homologous, conformational differences conferred by their N-terminal unique domains may allow selective binding to other proteins. In addition, differences in compartmentalization of binding partners may account for specific binding to subsets of highly related proteins. A number of scaffolding proteins has been isolated in the PSD (23,24). Previously we showed that Fyn but not Src is capable of binding the scaffolding protein RACK1 in a trimolecular complex with NR2B (9). The formation of this complex prevents Fyn from phosphorylating NR2B and negatively regulates the function of NR2B-but not NR2A-containing NMDA receptors (9). During the current study we showed that Src but not Fyn is capable of binding to H-Ras, causing a decrease in Src kinase activity and subsequent reduction in phosphorylation and retention of NR2A but not NR2B in the membrane. In vitro, Src and Fyn are capable of phosphorylating both NR2A or NR2B (6,35,47). We, therefore, suggest that it is compartmentalization that regulates the specificity of Src and Fyn for in vivo phosphorylation of the NR2 subunits.
In summary, we show here that Src tyrosine phosphorylation of NR2A is negatively regulated by H-Ras. The H-Ras inhibition of Src and subsequent alteration in NMDA receptor composition at the membrane contribute to negative regulation of synaptic strength. LTP is understood to represent a cellular basis for learning and memory (58), and there is a positive correlation between increased synaptic strength and learning and memory paradigms. Ras has been implicated in both regulation of LTP (17) and in learning and memory (59 -61). Therefore, the negative regulation of NMDA receptors by H-Ras may have implications for modulation of synaptic plasticity and learning and memory.