Modulation of NMDA Receptor- Dependent Calcium Influx and Gene Expression Through EphB Receptors

See allHide authors and affiliations

Science  18 Jan 2002:
Vol. 295, Issue 5554, pp. 491-495
DOI: 10.1126/science.1065983


Protein-protein interactions and calcium entry through theN-methyl-d-aspartate (NMDA)–type glutamate receptor regulate synaptic development and plasticity in the central nervous system. The EphB receptor tyrosine kinases are localized at excitatory synapses where they cluster and associate with NMDA receptors. We identified a mechanism whereby EphBs modulate NMDA receptor function. EphrinB2 activation of EphB in primary cortical neurons potentiates NMDA receptor–dependent influx of calcium. Treatment of cells with ephrinB2 led to NMDA receptor tyrosine phosphorylation through activation of the Src family of tyrosine kinases. These ephrinB2-dependent events result in enhanced NMDA receptor–dependent gene expression. Our findings indicate that ephrinB2 stimulation of EphB modulates the functional consequences of NMDA receptor activation and suggest a mechanism whereby activity-independent and activity-dependent signals converge to regulate the development and remodeling of synaptic connections.

During the development of the central nervous system, patterned neuronal activity drives the specification and maturation of synaptic connections (1, 2). The NMDA-type excitatory glutamate receptor regulates these activity-dependent processes in part by controlling the entry of Ca2+ into neurons, which then activates signaling pathways that orchestrate neuronal development (3). Prior to the development of synapses, young neurons differentiate and begin the process of axon growth and guidance by a mechanism that is largely independent of neuronal activity (4). The molecular mechanisms that allow activity-dependent processes to build upon activity-independent cues are not clear.

The ephrins and their receptors, the Eph tyrosine kinases, are cell-surface proteins that play a role in mediating the initial interaction between axons and dendrites as well as other activity-independent processes during neural development (5, 6). The Eph receptor proteins are classified into the EphA and EphB families on the basis of their preference for binding of glycosyl-phosphatidylinositol–linked ephrinA ligands or transmembrane ephrinB ligands, respectively. At the time of synaptogenesis, EphBs are localized to the postsynaptic region of excitatory synapses, suggesting a role for EphBs in synaptic function (7, 8). In cultured neurons, ephrinB binding of EphB promotes the interaction of EphB with NMDA-type glutamate receptors (9), suggesting that EphB receptor tyrosine kinases might modulate NMDA receptor function.

We investigated whether ephrinB2 activation of EphB could modulate Ca2+ conductance through the NMDA receptor. To assay for changes in the concentration of intracellular Ca2+ ([Ca2+]i), we incubated cultured cortical neurons from embryonic day 18 (E18) rats with the Ca2+ indicator dye fura-2-AM (10). Although these neurons expressed NR1 and NR2B receptor subunits and responded to membrane depolarization, treatment of these neurons with glutamate (20 to 40 μM) to activate the NMDA receptor in the presence of inhibitors of the L-type Ca2+ channel (nimodipine) and the AMPA-type glutamate channel (6-cyano-7-nitroquinoxaline-2,3-dione; CNQX) resulted in only a small increase in [Ca2+]i (Fig. 1A) (11). This suggests that in cultured embryonic neurons, NMDA receptor subunits are expressed before synapse formation, but are not yet fully functional. We considered the possibility that ephrinB/EphBs might enhance the ability of the NMDA receptor to regulate the influx of Ca2+. Cultured cortical neurons were exposed to activated ephrinB2-Fc, a fusion protein consisting of the extracellular domain of ephrinB2 fused to the Fc portion of human immunoglobulin G1 that is multimerized with antibodies to cluster and activate EphB. Neurons were assessed for changes in [Ca2+]i in response to glutamate stimulation. EphrinB2 treatment followed by glutamate stimulation resulted in a large increase in [Ca2+]i (Fig. 1, A and B). The increase in glutamate-stimulated [Ca2+]i in ephrinB2-treated cultures required the NMDA receptor, because the NMDA receptor antagonist 2-amino-5-phosphonovalerate (D-APV) blocked this increase in [Ca2+]i (Fig. 1C).

Figure 1

EphrinB2 activation of EphB enhances glutamate-stimulated Ca2+ influx through the NMDA receptor. (A) Trace of average [Ca2+]iduring Ca2+ imaging for E18 + 1 DIV cultured cortical neurons in the presence of nimodipine and CNQX (n = 30 cells). Neurons were stimulated with glutamate (40 μM; Gibco-BRL), washed with artificial cerebrospinal fluid, treated with ephrinB2-Fc for 40 min, and again stimulated with glutamate. (B) [Ca2+]i in response to glutamate stimulation before (−) and after (+) treatment with ephrinB2-Fc (eB2) or BDNF was quantified [*P < 0.01 versus (−), analysis of variance (ANOVA); n = 30 cells per condition]. (C) Neurons were treated as in (A) and then stimulated with glutamate in the presence (+) or absence (−) of the NMDA receptor antagonist D-APV (10 μM; RBI), and [Ca2+]i was compared [*P < 0.01 versus (−), ANOVA; n = 13 cells per condition]. (D) Average [Ca2+]i in E18 + 1 DIV neurons overexpressing pCDNA3 (vector), EphB2TR, or EphB2WT in response to glutamate stimulation before (−) and after (+) ephrinB2 treatment was compared [*P < 0.01 versus (−), ANOVA; n = 10 cells,pCDNA3;n = 11 cells, EphB2WT; n = 32 cells, EphB2TR]. (E) Representative examples of neurons transfected with either EphB2WT or EphB2TR and GFP and visualized by fluorescence microscopy (top panels) and by fura-2-AM labeling after glutamate stimulation (30 μM) before (−) and after (+) treatment with ephrinB2-Fc.

To determine whether the ephrinB2 effect is selective for the NMDA receptor, we depolarized neurons with 55 mM KCl to trigger Ca2+ influx through voltage-sensitive Ca2+channels before and after ephrinB2 treatment. EphrinB2 treatment had no effect on Ca2+ influx caused by membrane depolarization, even when the depolarization stimulus was submaximal (11). Unlike ephrinB2, Fc alone or brain-derived neurotrophic factor (BDNF), which activates the trkB receptor tyrosine kinase, did not result in an increase in glutamate-stimulated [Ca2+]i(Fig. 1B) (11).

EphrinB binding activates the EphB receptor tyrosine kinase and promotes its association with cytoplasmic proteins that mediate the effects of receptor activation (12). We thus investigated whether the cytoplasmic domain of EphB2, including the tyrosine kinase domain, is required for the ephrinB2 potentiation of NMDA receptor–dependent Ca2+ influx (10). In neurons transfected with either vector control (pCDNA3) or an EphB2 wild-type construct (EphB2WT), glutamate-stimulated NMDA receptor–mediated Ca2+ influx was increased after ephrinB2 treatment (Fig. 1, D and E). However, in neurons expressing an EphB2 construct containing a deletion of the EphB2 cytoplasmic domain (EphB2TR) (9), ephrinB2 treatment did not cause an increase in glutamate stimulation of NMDA receptor–dependent Ca2+influx (Fig. 1, D and E). Thus, the cytoplasmic region of EphB2 is required for the increase in glutamate-stimulated Ca2+influx through the NMDA receptor.

Tyrosine phosphorylation of the modulatory NR2 subunits of the NMDA receptor occurs both during development as well as in paradigms of synaptic plasticity, and can regulate NMDA receptor channel properties (13–17). In our cortical culture system, NR2B is the predominant NR2 subunit expressed (18, 19). To examine NR2B tyrosine phosphorylation, we stimulated cultured cortical neurons with aggregated ephrinB2-Fc or Fc, immunoprecipitated NR2B from neuronal lysates with antibodies to NR2B (anti-NR2B), and monitored tyrosine phosphorylation of the immunoprecipitated protein using anti-phosphotyrosine, PY99 (20). EphrinB2 treatment of neurons increased tyrosine phosphorylation of NR2B (Fig. 2A). When we expressed an expression construct for NR2B with either a wild-type EphB2 construct (WT) or an EphB2 mutant that lacks kinase activity (VM and KR) in human embryonic kidney (HEK) 293T cells, we found that NR2B became tyrosine phosphorylated only when coexpressed with EphB2WT (Fig. 2B), suggesting that EphB2 kinase activity or a kinase that associates with activated EphB is important for the ephrinB2-stimulated increase in NR2B tyrosine phosphorylation.

Figure 2

Increased tyrosine phosphorylation of the NMDA receptor after ephrinB activation of EphB. (A) Effect of ephrinB2-Fc treatment on NR2B tyrosine phosphorylation. Cortical neurons at E18 + 1 DIV were stimulated with ephrinB2-Fc or Fc and then lysed, and NR2B was immunoprecipitated with anti-NR2B. Western analysis was performed with anti-phosphotyrosine (PY99; Santa Cruz), anti-NR2B (NR2B; Affinity Bioreagents), or anti–phospho-EphB2 (n = 6). (B) Role of EphB2 protein kinase activity in NR2B tyrosine phosphorylation. HEK293T cells were transfected with NR2B alone, EphB2WT alone, NR2B and EphB2WT, or NR2B and catalytically inactive EphB2 constructs (EphB2VM or EphB2KR). Lysates were analyzed as in (A) and immunoblotted with anti-phosphotyrosine, anti-NR2B, or anti-EphB2 (n = 3). (C) Sites of NR2B tyrosine phosphorylation by EphB2. HEK293T cells were cotransfected with EphB2 and various NR2B constructs: NR2BWT, NR2B mutated at single tyrosines to phenylalanine (NR2BY1252F, Y1336F, Y1472F), or the triple mutant [(TM)Y1252F/Y1336F/Y1472F]. Western analysis was performed with antibodies that specifically recognize phosphorylated tyrosines 1252, 1336, or 1472 on NR2B (n = 3). The antibody to phosphotyrosine (PY99) does not detect tyrosine phosphorylation of NR2BTM or NR1 when these constructs are coexpressed with EphB2WT. (D) Role of NR2B tyrosine phosphorylation for glutamate-dependent Ca2+ influx. Average [Ca2+]i in HEK293T cells expressing NR1 together with NR2BWT, or NR2BTM, and either EphB2WT, EphB2VM, or EphB2EE stimulated with 0.04 μM glutamate [*P < 0.01; untreated (UT), n= 30; NR1/NR2B, n = 16; B2WT/NR1/NR2B,n = 73; B2WT/NR1/NR2BTM, n = 55; B2VM/NR1/NR2B, n = 59; B2EE/NR1/NR2B, n= 20). The expression levels of EphB2, NR1, and NR2B were verified by Western blot. Untransfected (non-GFP–labeled) cells did not respond to glutamate.

We also transfected into HEK293T cells an expression vector encoding EphB2 along with a vector encoding wild-type NR2B, or NR2B mutants in which tyrosines in the terminal 400 amino acids of the cytoplasmic tail of NR2B were converted to phenylalanines (20). This mutational analysis and the subsequent generation of phospho-specific antibodies to putative sites of tyrosine phosphorylation revealed that when EphB2 and NR2B are expressed together, NR2B becomes tyrosine phosphorylated on tyrosines 1252, 1336, and 1472 (Fig. 2C). To determine whether phosphorylation of NR2B at 1252, 1336, and 1472 by EphB2 affects Ca2+ flux through the NMDA receptor, we expressed NR1 together with NR2BWT or NR2B mutated at tyrosines 1252, 1336, and 1472 (NR2BTM) and EphB2WT or kinase-dead EphB2 (EphB2VM) in HEK293T cells. Stimulation of cells expressing NR1, NR2BWT, and EphB2WT with 0.04 μM glutamate resulted in a robust increase in Ca2+influx, whereas in the absence of EphB2, with expression of EphB2VM, or expression of NR2BTM, the Ca2+ response was significantly lower (Fig. 2D). These results suggest that phosphorylation of NR2B at tyrosines 1252, 1336, and 1472 is required for the EphB2-mediated enhancement of Ca2+ influx through the NMDA receptor.

These three tyrosines on NR2B are phosphorylated by Fyn, a member of the Src family of tyrosine kinases (21). Members of the Src family physically associate with both EphB and NMDA receptors, and Src family members can modulate the open time of the NMDA receptor channel (13, 17, 22, 23). We thus considered the possibility that ephrinB binding to EphB activates a Src family member, which would then phosphorylate NR2B, thereby enhancing the ability of the NMDA receptor to conduct Ca2+ upon glutamate stimulation. We treated cultured cortical neurons at E18 + 1 DIV with aggregated ephrinB2-Fc, Fc, or BDNF, which is known to activate Src family members, and assessed Src activation using an antibody specific for phospho-Tyr416, because phosphorylation of Tyr416 occurs when Src is activated (20, 24). EphrinB2 treatment of neurons led to an increase in the amount of Src that was phosphorylated at Tyr416 (Fig. 3A), and to an inducible interaction between EphB2 and Src (19). Although BDNF treatment also activates Src, BDNF does not alter glutamate-stimulated Ca2+ influx and, in contrast to ephrinB2, leads to NR2B tyrosine phosphorylation at sites other than 1252, 1336, and 1472 (Fig. 1B) (19, 25,26). These findings suggest that ephrinB2 treatment leads to the selective localization and activation of Src and to NR2B tyrosine phosphorylation.

Figure 3

Requirement of Src tyrosine kinase family members to mediate the effects of ephrinB2 treatment. (A) Effects of ephrinB2 stimulation on the activity of Src family members. Cortical neurons at E18 + 1 DIV were stimulated with Fc, ephrinB2-Fc, or BDNF, and Western analysis was performed with anti-Src phospho-Tyr416 (BioSource International), anti-Src (UBI), or anti–phospho-EphB2 (n = 3). (B) The role of Src family members in NR2B tyrosine phosphorylation. Neurons were treated with PP2, PP3, or SU6656 at 0.5 μM and analyzed as in Fig. 2A (n = 6). (C) HEK293T cells were cotransfected with NR2B alone, EphB2WT alone, NR2B and EphB2WT, or NR2B and EphB2EE, and lysates were analyzed by Western blot with anti-NR2B phospho-Tyr1472, anti-NR2B, or anti-EphB2 (n = 3). (D) Role of Src family members in glutamate-stimulated Ca2+ influx. Neurons were treated with PP2 (0.5 μM) or PP3 (0.5 μM), and the effect of glutamate-stimulated [Ca2+]i before (−) and after (+) ephrinB2 treatment was quantified [*P < 0.01 versus (−) and untreated (UT), ANOVA; n = 18 cells per condition]. (E) Average [Ca2+]i in neurons overexpressing pCDNA3, FynDN, or FynCA following glutamate stimulation after (+) ephrinB2 treatment (*P < 0.01 versus before ephrinB2 treatment, ANOVA; n = 9 cells, pcDNA3;n = 30 cells, fynDN; n = 9 cells, fynCA).

We used two small molecule inhibitors of the Src family of tyrosine kinases, PP2 and SU6656, to assess the importance of Src activation for ephrinB2-induced NR2B phosphorylation. Both PP2 and SU6656, but not PP3, the inactive analog of PP2, inhibited the ephrinB2-induced increase in NR2B tyrosine phosphorylation in neurons (Fig. 3B). Also, expression of EphB2 and NR2B in HEK293T cells with a dominant inhibitory Fyn construct (FynDN), which inhibits activation of Fyn and other Src family members, inhibited NR2B tyrosine phosphorylation (19). These findings suggest that ephrinB2 binding to EphB leads to the activation of a Src family member that in turn phosphorylates NR2B. An EphB2 mutant (EphB2-EE) that is catalytically active but cannot bind Src family members (27) did not promote tyrosine phosphorylation of NR2B (Fig. 3C). Thus, EphB activation of Src family members is required for the effects of ephrinB2 on tyrosine phosphorylation of the NMDA receptor.

We used dominant inhibitory and constitutively active Fyn constructs (FynDN or FynCA, respectively), EphB2EE, or the Src inhibitor PP2 to determine whether Src family members are required for ephrinB-dependent potentiation of Ca2+ influx through the NMDA receptor. Overexpressing FynDN, EphB2EE, or treatment with PP2 prevented the ephrinB2-induced increase in NMDA receptor–dependent [Ca2+]i (Figs. 2D and 3, D and E). These findings suggest that ephrinB2 potentiates glutamate stimulation of Ca2+ influx through the NMDA receptor by inducing a Src family member, which in turn phosphorylates NR2B. The phosphorylation of NR2B then enhances the ability of the NMDA receptor to regulate the influx of Ca2+ in response to glutamate.

One function of ephrinB2 potentiation of glutamate-stimulated Ca2+ influx might be the activation of Ca2+-regulated immediate-early genes (IEGs) (3, 28, 29). NMDA receptor stimulation can induce IEG transcription by activating the Ca2+/cAMP-responsive element binding protein (CREB) through phosphorylation of CREB Ser133(3, 30). Therefore, we assessed whether exposure of neurons to ephrinB2 affected glutamate stimulation of CREB Ser133 phosphorylation and CREB-dependent reporter-gene transcription.

Treatment of cells with ephrinB2-Fc or glutamate alone had little effect on the level of CREB Ser133phosphorylation or CREB-dependent reporter-gene expression (Fig. 4, A and B) (31). Treatment of neurons with ephrinB2-Fc, followed by glutamate stimulation, substantially increased CREB Ser133phosphorylation and reporter-gene transcription (Fig. 4, A and B). The effect of ephrinB2 is mediated by the NMDA receptor because treatment of neurons with D-APV blocked the effect of ephrinB2 (Fig. 4, A and B). EphB2TR, which blocks EphB signaling, or inhibition of Src family members with PP2 eliminated ephrinB2 potentiation of glutamate-stimulated gene expression (Fig. 4, C and D). These findings demonstrate that ephrinB2 potentiates CREB phosphorylation and activation by an EphB- and Src family–dependent mechanism.

Figure 4

Potentiation of CREB-dependent transcription after EphB activation. (A) Effect of ephrinB2 treatment on CREB Ser133 phosphorylation. Lysates from neurons treated with ephrinB2-Fc or Fc before glutamate stimulation (20 μM; 10 min) were immunoblotted with anti-CREB phospho-Ser133, anti-CREB, or anti–phospho-EphB2 (n = 5). Neurons were first treated with D-APV (10 μM) to block NMDA receptors (n = 2). EphrinB2 treatment and glutamate stimulation potentiates CREB Ser133 phosphorylation until at least DIV 7, suggesting that even as neurons mature, ephrinB2 treatment is capable of stimulating CREB phosphorylation. (B) Enhancement of CREB-dependent transcription. Neurons were transfected with reporter constructs, treated with ephrinB2-Fc or Fc followed by glutamate or control stimulation, and analyzed by the dual-luciferase reporter system. Luciferase activity is presented as fold induction over Fc and mock stimulation (NLU, normalized luciferase units). Neurons were also treated with D-APV (*P < 0.01 versus Fc + glutamate treatment, ANOVA; n = 3). (C and D) Requirement of EphB cytoplasmic domain and Src family member activation. Neurons were transfected with luciferase constructs and pCDNA3, EphB2TR, or EphB2WT (C) or treated with control, PP2 (0.5 μM), or PP3 (0.5 μM) (D) and analyzed as in (B) (*P < 0.01 versus Fc + glutamate treatment in each condition, ANOVA; n = 3). Glutamate stimulation led to a statistically significant increase in CRE-luciferase induction in all cases except for D-APV treatment.

We next investigated whether ephrinB2 potentiates glutamate stimulation of specific IEGs that might regulate synapse formation, maturation, or function. Neurons were treated with ephrinB2 or control reagents for 45 min, then stimulated with glutamate or by membrane depolarization with elevated concentrations of KCl to stimulate the influx of Ca2+ through L-type voltage-sensitive Ca2+ channels (31). Glutamate-stimulated expression of c-fos, a transcription factor that mediates cellular responses to extracellular stimuli, was enhanced after activation of EphB (Fig. 5, A and B). The effect of ephrinB2 treatment on c-fos transcription is specific to NMDA receptor activation because the enhancement of c-fos expression was blocked by D-APV and the induction of c-fos expression by KCl stimulation was not affected by EphB activation (11). Treatment of neurons with ephrinB2 also potentiated glutamate activation of two other genes, BDNFand cpg15, that have been implicated in synapse development and/or function (Fig. 5, C and D) (32,33).

Figure 5

EphB activation modulates Ca2+-dependent gene expression. (A) Analysis of c-fos expression in neurons (E18 + 2 to 3 DIV). Neurons were treated with ephrinB2-Fc or Fc for 45 min followed by stimulation with glutamate, 55 mM KCl, or control solution. RNA was collected and analyzed for c-fos mRNA expression (upper panel); the same blot was probed for GAPDH mRNA to control for loading (lower panel). (B) Effect of EphB activation on c-fosexpression. Quantification of c-fos expression normalized to GAPDH after various treatments (*P < 0.01 versus Fc + glutamate, ANOVA; n = 4). (Cand D) Analysis of BDNF and cpg15 mRNA in cortical neurons (E18 + 1 to 2 DIV for BDNF or 3 DIV for cpg15) that were first treated with Fc or ephrinB2-Fc, followed by glutamate stimulation (20 μM) for 4 hours. The level ofBDNF or cpg15 mRNA was determined by real-time PCR analysis (iCycler) and is presented as fold induction over Fc and mock stimulation (*P < 0.01 versus Fc treatment, ANOVA; n = 3).

Our findings suggest a model in which activity-independent cues may influence activity-dependent processes by potentiating or modulating gene expression during the specification, strengthening, and remodeling of synaptic connections. This model is supported by recent experiments with EphB2-deficient mice that show an in vivo role for EphB2 in activity-dependent plasticity (34,35). As neurons mature and form functional contacts, the effect of glutamate on Ca2+-dependent gene expression is tightly controlled. EphrinB activation of EphB promotes the clustering of NMDA receptors, and the modulation of NMDA receptor by EphB may sensitize the neurons to the effects of glutamate. Glutamate then stimulates increased Ca2+ influx to trigger programs of gene expression that may affect synapse formation, maturation, and plasticity. Taken together, these observations suggest that the ephrinB–EphB–NMDA receptor interaction may represent an early step in the initiation of synapse formation or maturation and may potentiate the ability of the NMDA receptor to respond to activity-dependent signals from the extracellular milieu.

  • * These authors contributed equally to this work.

  • Present address: Department of Neurobiology, Case Western Reserve University School of Medicine, 10900 Euclid Avenue, Cleveland, OH 44106, USA.

  • To whom correspondence should be addressed. E-mail: michael.greenberg{at}


View Abstract

Stay Connected to Science

Navigate This Article