EphB controls NMDA receptor function and synaptic targeting in a subunit-specific manner

Abstract

Dynamic regulation of the localization and function of NMDA receptors (NMDARs) is critical for synaptic development and function. The composition and localization of NMDAR subunits at synapses are tightly regulated and can influence the ability of individual synapses to undergo long-lasting changes in response to stimuli. Here, we examine mechanisms by which EphB2, a receptor tyrosine kinase that binds and phosphorylates NMDARs, controls NMDAR subunit localization and function at synapses. We find that, in mature neurons, EphB2 expression levels regulate the amount of NMDARs at synapses, and EphB activation decreases Ca(2+)-dependent desensitization of NR2B-containing NMDARs. EphBs are required for enhanced localization of NR2B-containing NMDARs at synapses of mature neurons; triple EphB knock-out mice lacking EphB1-3 exhibit homeostatic upregulation of NMDAR surface expression and loss of proper targeting to synaptic sites. These findings demonstrate that, in the mature nervous system, EphBs are key regulators of the synaptic localization of NMDARs.

Figure 1.

EphB2 regulates localization of NMDAR receptors to synapses in mature cortical neurons. A–C, Confocal microscopy maximum projection images of cultured cortical neurons at 21 DIV expressing eGFP and shRNA vector (control), EphB2.1 shRNA, or EphB2.1 shRNA plus “rescue” EphB2 (functional EphB2 OE), immunostained for GFP (green), NR1 (red), and the presynaptic marker vGlut (blue). The magnified sections (top) of high-contrast image with arrows show spine (yellow arrows) and shaft (white arrows) synapses, defined as the locations where NR1, GFP, and vGlut immunostaining colocalize. The bottom panels show same region with anti-NR1 staining in red. D–F, Cumulative probability histograms of synaptic NR1, NR1 at spine synapses, and NR1 at shaft synapses. Functional EphB2 OE using a rescue construct in the context of endogenous EphB2 knockdown caused a significant increase in the amount of NR1 colocalizing with vGlut, whereas knockdown of EphB2 resulted in a decrease (Kolmogorov–Smirnov test, p < 0.001). Amount of NR1 for each condition is normalized to the maximum intensity observed from all three conditions. Control, n = 27 cells, 432 synapses; EphB2.1 shRNA, n = 27, 359; functional EphB2 OE, n = 21, 319.

Figure 2.

EphB2 expression regulates mEPSC amplitude but not frequency in mature cortical neurons. Whole-cell patch-clamp recordings were made from 21–23 DIV cultured rat cortical neurons expressing eGFP and shRNAi vector (control; blue), EphB2 shRNA (EphB2.1 shRNA; red), or EphB2.1 shRNA plus “rescue” EphB2 (functional EphB2 OE; green). Recordings were at −65 mV in Mg²⁺-free solution. A, Example whole-cell patch-clamp recordings from cortical neurons in each condition; functional EphB2 OE neurons (row 3) show occasional miniature synaptic events at higher amplitude (left) that are blocked by the NMDAR antagonist APV (right). B, Mean traces of mEPSCs after NMDAR blockade with APV. Control: n = 5 cells without APV, 2298 mEPSCs; n = 5 cells with APV, 2490 mEPSCs; EphB2 shRNA: n = 5 cells without APV, 2528 mEPSCs; n = 5 cells with APV, 2040 mEPSCs. Functional EphB2 OE: n = 3 cells without APV, 1037 mEPSCs; n = 3 cells with APV, 1052 mEPSCs. C, No change in mEPSC frequency was observed for any condition (ANOVA, p > 0.05). D, Quantification of mean mEPSC amplitude before and after application of 50–100 μm APV. An increase in mEPSC amplitude was observed with functional overexpression of EphB2, whereas NMDAR blockade with APV significantly reduced mEPSC amplitude for all conditions (ANOVA, *p < 0.01, **p < 0.001). E, Cumulative probability histograms of mEPSC amplitude for each condition. Error bars indicate SEM.

Figure 3.

EphB2 regulates synaptic localization of functional NMDARs in mature cortical neurons. Whole-cell patch-clamp recordings were made from 21–23 DIV cultured rat cortical neurons expressing eGFP and shRNA vector (control; blue), EphB2 shRNA (EphB2.1 shRNA; red), or EphB2 shRNA plus “rescue” EphB2 (functional EphB2 OE; green) (A–E). A, Mean traces of EPSCs for each condition. B, Sample trace of whole-cell patch-clamp recording illustrating decay time. Decay time was calculated as the time from peak amplitude of the current to 30% of the peak amplitude, indicated by the arrows. C, Quantification of average decay time for each condition. D, E, Cumulative probability histograms of mEPSC decay times for each condition, plotted together (D) and individually for clarity (E). Control/EphB2.1 shRNA, p < 0.05; EphB2.1 shRNA/functional EphB2 OE, p < 0.0001; without APV/with APV, p < 0.0001 for all conditions; control with APV/functional EphB2 OE with APV, p < 0.0001; EphB2.1 shRNA with APV/functional EphB2 OE with APV, p < 0.0001; K-S tests; N as in Figure 1. These findings indicate that the slow NMDAR component of mEPSCs is reduced by EphB2 knockdown (EphB2.1 shRNA) and increased when EphB2 is functionally overexpressed (EphB2.1 shRNA plus “rescue” EphB2). F, Whole-cell patch-clamp recordings were made from 21–23 DIV cultured rat cortical neurons expressing eGFP and vector (control; blue), EphB2.1 shRNA (red), or EphB2 (EphB2 OE; green). Normalized amplitude plot of the mean mEPSCs recorded at +50 mV in control, EphB2 shRNA, and EphB2 OE neurons. G, Quantification of NMDAR component of the mEPSCs recorded at +50 mV in the presence of Mg²⁺ (control, n = 368 events/6 cells; EphB2 shRNAi, n = 744/11; EphB2 OE, n = 1247/9). *p < 0.05, **p < 0.001, ANOVA. Error bars indicate SEM.

Figure 4.

EphB2 attenuates Ca²⁺-dependent desensitization of NR2B- but not NR2A-containing NMDARs. A, Top, NMDA-elicited currents recorded from HEK-293 cells expressing NR1-1a/NR2B receptors in the absence (left) or presence of EphB2 (center), or the kinase-dead mutant EphB2-KD (right). Recordings at −60 mV in Mg²⁺-free solution. Bottom, Summary data showing peak current, steady-state current, and Ca²⁺-dependent desensitization of NMDA current (quantified as 1 − steady-state current/peak current × 100) (n = 8, 9, and 7 for cells in the absence of cotransfected EphB2, presence of EphB2, or EphB2-KD, respectively). B, Top, NMDA-elicited currents recorded from HEK-293 cells expressing NR1-1a/NR2A receptors in the absence (left) or presence (right) of EphB2. Bottom, Summary data showing peak current, steady-state current, and Ca²⁺-dependent desensitization of NMDA current [n = 10 and 8 cells (3 independent experiments) in the absence or presence of EphB2, respectively]. *p < 0.05, **p < 0.01. Error bars indicate SEM.

Figure 5.

Ephrin-B2 activation of EphB2 increases NR2B surface localization. A–C, Cortical neurons at 7 DIV (A), 14 DIV (B), or 21 DIV (C) were treated for 45 min with control Fc (C) or activated ephrin-B2-Fc (eB2). Biotinylated (surface) and total NR2B protein was visualized by immunoblotting with specific antibodies (top gels). β-Actin was used as a loading control for total protein (bottom gels). Absence of actin in surface (biotinylated) gels indicates validity of surface labeling. Representative immunoblots show no actin immunolabeling in the biotinylated surface fraction. The bottom bar graphs show the ratio of amount of surface NR2B to total NR2B at 7 DIV (n = 5 experiments), 14 DIV (n = 6 experiments), or 21 DIV (n = 6 experiments). Ephrin-B2-Fc versus Fc (control) conditions were analyzed by an unpaired t test. *p < 0.05. Error bars indicate SEM.

Figure 6.

Surface and total NR2B expression levels are increased in the cortex of EphB1^−/−, 2^−/−, 3^−/− TKO mice. A, Representative Western blots depicting NR2A and NR2B surface expression (left) and total expression (right) in WT, EphB1^−/−, 3^−/− DKO, and TKO mice. B–D, Quantification of NR2A surface, total, and surface/total expression. E–G, Quantification of NR2B surface, total, and surface/total expression. Values were normalized to DKO (n = 6, 9, and 7 animals for WT, DKO, and TKO, respectively). *p < 0.05, **p < 0.01. Error bars indicate SEM.

Figure 7.

NR2B surface expression is increased and total NR2A levels are decreased in the hippocampus of TKO mice. A, Left, Representative Western blots depicting NR2A and NR2B surface expression in WT, EphB DKO, and EphB TKO mice. Right, Western blots showing total expression of NR2A and NR2B in WT, EphB DKO, and EphB TKO mice. Actin was used as a loading control in total protein fraction and as a control for surface staining in surface fraction. B, Quantification of NR2A and NR2B surface, total, and surface/total expression. Values are normalized to DKO (n = 6, 9, and 7 animals for WT, DKO, and TKO, respectively). *p < 0.05, **p < 0.01, ANOVA. Error bars indicate SEM.

Figure 8.

NR1 surface expression is decreased in hippocampus of TKO mice. Top left, Western blot illustrating surface and total NR1 expression in cortex of WT, EphB DKO, and EphB TKO mice. Top right, Western blot depicting surface and total NR1 levels in hippocampus of WT, EphB DKO, and EphB TKO mice. Actin is shown as a loading control in total protein fraction and as a control for surface staining in surface fraction. Below, Quantification of NR1 surface, total, and surface/total levels in cortex and hippocampus of WT, EphB DKO, and EphB TKO mice. Values are normalized to WT (n = 2 animals each for WT, DKO, and TKO, samples were then divided and labeled with two independent reactions). *p < 0.05, ANOVA. Error bars indicate SEM.

Figure 9.

EphB TKO brains exhibit reduced synaptic expression of NR2A and NR2B. A, Lysates from brains of WT, EphB DKO, and EphB TKO animals were fractionated to isolate supernatant (S1), crude membrane (P1), and pure synaptosome (Syn) fractions. Western blots were probed with indicated antibodies and show enrichment of glutamate receptor subunits in the synaptosome fraction of all animals. PSD-95 and EphB2 are enriched in the same fraction. B–E, Syn/P1 ratio was used to compare synaptic versus nonsynaptic expression of NMDA and GluR2 glutamate receptors subunits. Compared with the WT and EphB DKO brains, EphB TKO animals exhibit reduced synaptic expression of NR2A (C) and NR2B (D) subunits of the NMDAR. There is no change in synaptic expression the GluR2 subunit (E) of AMPA receptors between different genotypes. ANOVA, *p < 0.005; n = 3 animals for each condition. Error bars indicate SEM.