To begin examining whether BDNF regulates Tyr¹⁴⁷²-NR2B phosphorylation, we exposed cerebrocortical PSDs to varying concentrations of BDNF for 10 min. BDNF (2–25 ng/ml) elicited an increase in Tyr¹⁴⁷²-NR2B phosphorylation (Fig. 1A), using equivalent concentrations of NR2B protein (Fig. 1B). BDNF evoked a maximal, 1.8-fold enhancement of phosphorylation per unit protein at 2 ng/ml of BDNF within 10 min (Fig. 1C). Higher doses of BDNF had lesser effects: 5 ng/ml of BDNF increased phosphorylation 1.5-fold, while 10 ng/ml and 25 ng/ml of BDNF only increased phosphorylation 1.3-fold.#

To define the acute effects of BDNF, we studied the time course of BDNF action on Tyr¹⁴⁷² phosphorylation (Figs. 1D–F). Incubation with 2 ng/ml BDNF increased phosphorylation 1.4-fold after 5 min. Incubation for 10 min elicited a maximal, 1.8-fold enhancement of phosphorylation. Incubation with BDNF for 30 min, however, had a much lesser effect. Decrease in phosphorylation with longer incubation was apparently the result of dephosphorylation since there was no change in NR2B protein level during the incubation (Fig. 1D).#

To determine whether other trophins also regulate Tyr¹⁴⁷²-NR2B phosphorylation, we examined NGF and NT-3 (Figs. 2A–C). BDNF increased phosphorylation 1.8-fold, while neither NGF nor NT-3 had effects, indicating that phosphorylation was specifically and selectively modulated by BDNF.#

To determine whether BDNF regulates Tyr¹⁴⁷² phosphorylation through activation of trkB, we examined K252a, a trk tyrosine kinase inhibitor (Berg et al. 1992; Knusel et al. 1992; Knusel and Hefti 1992; Nye et al. 1992; Tapley et al. 1992). BDNF increased phosphorylation 1.8-fold (Figs. 2D–F). Pretreatment with K252a alone had no effects on phosphorylation, but dramatically blocked BDNF-activated phosphorylation (Figs. 2D–F).#

To define signaling pathways involved in BDNF-mediated Tyr¹⁴⁷² phosphorylation, we initially focused on ERK1 and PI-3K, which reportedly mediate BDNF synaptic actions (Gottschalk et al., 1999). Two pharmacological inhibitors of each effector were used (PD98059 and U0126 for ERK1, and wortmannin and LY294002 for PI-3K). PD98059 (PD, 100 μM) alone had no effect on phosphorylation, but prevented the BDNF enhancement of phosphorylation (Figs. 3A–C). Two concentrations (25 μM or 40 μM) were used for U0126, another inhibitor to ERK1. At either concentration, U0126 itself did not change phosphorylation; however, the BDNF effect on phosphorylation was blocked after preincubation with U0126.#

The PI-3K inhibitor wortmannin was also examined at different concentrations. Wortmannin [0.1 μM (W1) or 1 μM (W2)] alone did not alter phosphorylation (Figs. 3B and C), but blocked BDNF-mediated enhancement. LY294002 (LY, 50 μM), another PI-3K inhibitor, also failed to alter phosphorylation when applied alone, but prevented the BDNF effect.#

To investigate the specific enzyme regulating BDNF-induced phosphorylation, we began with Fyn, a tyrosine kinase that regulates Tyr¹⁴⁷² phosphorylation and LTP (Nakazawa et al., 2001). Because tyrosine phosphorylation of Fyn increases its catalytic activity, we initially determined whether BDNF regulates phosphorylation of the kinase itself. BDNF (2 ng/ml), after 10 min of exposure, elicited a 2.8-fold increase (Figs. 4A–C). PP2 (1 μM), a specific inhibitor of Fyn, decreased phosphorylation (66.3±3.5% of baseline) and also abolished the BDNF-mediated enhancement.#

To determine whether Fyn is required for BDNF-mediated phosphorylation of NR2B, we selectively inhibited the enzyme. The results are presented in Figs. 4D–F. PP2 (1 μM) alone had no effects on phosphorylation. However, PP2 preincubation prevented BDNF-mediated enhancement. PP3 (1 μM), an inactivated form of PP2, had no effect either on baseline phosphorylation or on BDNF-mediated enhancement, suggesting the specificity of PP2 actions.#

To further define the role of Fyn, we examined cortical PSDs isolated from Fyn knockout mice. BDNF increased phosphorylation in PSDs from wild-type mice (WT) 1.9-fold (Fig. 5C). In Fyn knockout (Fyn^−/−) PSDs, basal phosphorylation was decreased to only 20% of wild-type control (Fig. 5D). Moreover, in the presence of BDNF, no enhancement of phosphorylation (Fig. 5C) was observed in Fyn knockout mice, which remained at 20% (Fig. 5D) of wild-type controls, suggesting that Fyn regulates the phosphorylation.#

To confirm that Fyn mediates the phosphorylation, we examined exogenously applied, catalytically active Fyn. In PSDs from Fyn knockout mice, basal Tyr¹⁴⁷² phosphorylation was decreased and BDNF had no effect on phosphorylation (Figs. 5C and D). However, addition of active Fyn (20 ng/ml) dramatically enhanced phosphorylation by approximately 40-fold (Figs. 5G). In addition, BDNF did not further alter phosphorylation in the presence of active Fyn in Fyn knockout PSDs.#

To determine whether BDNF regulates phosphorylation of Tyr¹⁴⁷² in intact cells as in the PSD preparations, we investigated effects of the trophin in dissociated hippocampal neurons. Cells were dissected from day 16 wild-type mouse embryos and cultured in vivo for 10–14 days. Incubation with 20 ng/ml of BDNF for 10 min increased phosphorylation 2.5-fold (Figs. 6A–C), reproducing effects on cerebrocortical PSD preparations.#

To begin defining the role of BDNF-mediated Tyr¹⁴⁷² phosphorylation in synaptic plasticity, we examined NMDA receptor activity in Fyn knockout animals in which phosphorylation was blocked. Using cell-attached single-channel recordings made from hippocampal neurons obtained from wild-type mice, we found that application of BDNF rapidly increased the charge through NMDA receptor channels 2-fold after 10–15 min of exposure (Figs. 7A and C). In contrast, NMDA receptors in neurons obtained from knockout mice did not display any increase in channel charge in response to BDNF (Figs. 7B and D). Moreover, NMDA receptor charges were significantly different for wild-type cells and Fyn knockout cells (Fig. 7E).#

It is known that BDNF increases NMDA single-channel open probability via postsynaptic trkB receptors and that this effect is dependent on the presence of NR2B-containing receptors (Crozier et al. 1999; Levine and Kolb 2000); our results indicate that phosphorylation of Tyr¹⁴⁷² may play a critical role. This residue in the cortical NR2B is phosphorylated within 10 min of BDNF exposure. The rapid response correlates well with electrophysiological studies indicating that BDNF acutely enhances synaptic transmission in hippocampal neurons (Levine et al., 1995). The effects of BDNF on phosphorylation were rapid and transient, suggesting that both protein kinase(s) and phosphatase(s) participate. The enhancement of phosphorylation was maximal at 2 ng/ml of BDNF, a low concentration that implies the involvement of trkB, the high-affinity receptor for BDNF. K252a, an inhibitor of trk receptor tyrosine kinase, blocked BDNF enhancement of phosphorylation, confirming trkB mediation. In contrast to BDNF, NGF or NT-3 failed to regulate phosphorylation, in agreement with the findings that neither NGF nor NT-3 modulates synaptic transmission (Levine et al., 1996).#

We have also begun to identify the precise molecular signaling cascades leading to BDNF-mediated Tyr¹⁴⁷² phosphorylation. Trk autophosphorylation triggers activation of several downstream signaling molecules, including PI-3K, ERK1 and PLC-γ1. We focused on PI-3K and ERK1 as these signaling molecules are rapidly activated after application of BDNF to neonatal hippocampal slices (Gottschalk et al., 1999), and activation of these signaling pathways is involved in BDNF-dependent learning and memory (Yamada and Nabeshima, 2003). BDNF greatly attenuated synaptic fatigue at CA1 synapses induced by a train of high-frequency, tetanic stimulation (HFS), while the activation of MAPK and PI-3K is required for this BDNF action at presynaptic sites (Gottschalk et al., 1999). Our results suggest involvement of both PI-3K and ERK1 in BDNF-induced Tyr¹⁴⁷² phosphorylation in the PSD. Consequently, these signaling molecules apparently mediate BDNF fast actions at both pre- and postsynaptic sites, indicating multiple mechanisms of BDNF-induced synaptic transmission.#

We employed two approaches to define the requirement of Fyn for BDNF-mediated effects. Biochemical studies using PSD preparations revealed that, within 10 min of exposure, BDNF enhanced phosphorylation of Fyn 2.8-fold, potentially leading to increased activity and Tyr¹⁴⁷² phosphorylation. PP2, a specific tyrosine kinase inhibitor of Fyn, blocked phosphorylation of both Fyn and Tyr¹⁴⁷². Although PP2 is also a potent inhibitor of p56^c-src and HcK (Hanke et al., 1996), these kinases are not present in the PSDs (Wu et al., unpublished results), ruling out their participation in Tyr¹⁴⁷² phosphorylation. In PSDs isolated from Fyn-deficient mice, BDNF effects were abolished, confirming the involvement of Fyn in BDNF-mediated phosphorylation. Of note, the basal Tyr¹⁴⁷² phosphorylation in Fyn-deficient PSDs was only 20% of that in the wild-type PSDs. The simplest explanation for this finding is that Fyn is the major PSD kinase mediating up to 80% of Tyr¹⁴⁷² phosphorylation; presumably, other kinase(s) regulate the remaining 20%. In addition, exogenously applied active Fyn alone restored Tyr¹⁴⁷² phosphorylation to 800% over controls. BDNF did not exhibit any additional enhancement of phosphorylation because the exogenous Fyn employed was already active. Of note, BDNF alone increased Tyr¹⁴⁷² phosphorylation in the wild-type PSD roughly 2-fold, instead of the 40-fold exerted by Fyn in the knockouts (equal to 8-fold of the wild-type baseline). BDNF may activate only some of the Fyn pools in the PSD. The above results, showing the loss of BDNF Tyr¹⁴⁷² phosphorylation in Fyn knockout mice and the restoration of the phosphorylation by exogenously added active Fyn, suggest that Fyn is the downstream kinase phosphorylating Tyr¹⁴⁷²-NR2B. It is well established that the primary effector molecules, such as ERK1 and PI-3K, transduce trk activation and initiate the subsequent reactions, leading to the phosphorylation of receptors by the specific kinases, e.g. Fyn in the present studies. As BDNF activates the primary proteins ERK1 and PI-3K (Wu et al., 1999), as well as the downstream kinase Fyn (present studies), activation of these primary molecules by BDNF ultimately leads to Fyn activation and the subsequent Tyr¹⁴⁷² phosphorylation. Taken together, our observations support the findings of Mizuno et al. (2003), showing that Fyn is a crucial mediator of signal transduction from trkB to NR2B.#

In the present studies, we examined the functionally critical PSDs instead of hippocampal neurons to emphasize the postsynaptic mechanisms of trkB-mediated ERK1 and PI-3K signaling cascades. We believe that our findings are real for the following reasons. First, the isolated PSDs are functionally intact structures. Isolation of PSDs does not lead to the aggregation of signaling components that normally do not exist together. For example, using immunocytochemistry, our previous findings revealed the localization of trkB and BDNF to PSDs in situ (Aoki et al., 2000). In addition, trkB activation in the isolated PSDs leads to activation of the primary effector molecules PI-3K, PLC-γ1, SHC and ERK1, but not SOS (Wu et al., 1999), potentially regulating BDNF-mediated postsynaptic receptor phosphorylation. Second, we have previously demonstrated that BDNF acutely enhanced phosphorylation of the NMDA receptor subunits NR1 in the isolated PSDs and synaptoneurosomes (intact synaptic apparatus from which PSDs were derived) (Suen et al., 1997) and NR2B (Lin et al., 1998) in the isolated PSDs. In particular, BDNF elicited a 2-fold increase in NR2B tyrosine phosphorylation, mimicking the effects of LTP (Rosenblum et al. 1996; Rostas et al. 1996). Subsequently, we discovered that BDNF acutely increased tyrosine phosphorylation of the AMPA receptor subunit GluR1 in the isolated PSDs via activation of the NMDA receptors (Wu et al., 2004). Third, our present studies demonstrated that BDNF evoked similar increase in Tyr¹⁴⁷²-NR2B phosphorylation in both the isolated PSDs and cultured hippocampal neurons. Together, our studies indicate that the isolated PSDs contain intact machinery for BDNF signaling cascades, ruling out the possibility of artifactual results.#

We conducted electrophysiological studies using hippocampal neurons to investigate the role of BDNF-mediated Tyr¹⁴⁷² phosphorylation in modulation of the NMDA receptor activity. Specifically, we used Fyn knockout mice because Fyn appears to be required for BDNF-mediated phosphorylation. Fyn-deficient mice are viable and fertile but display neurological defects including blunted LTP, impaired spatial learning and altered hippocampal development (Grant et al., 1992). Consequently, Fyn is apparently necessary for certain forms of synaptic plasticity. Our results revealed that basal activity of the NMDA receptor in Fyn knockout neurons was similar to that in wild-type neurons. Nevertheless, lack of Fyn impaired BDNF-mediated enhancement of channel activity, indicating that Fyn-mediated phosphorylation may be required for BDNF-mediated synaptic plasticity.#

BDNF is known to regulate LTP. Since NMDA receptors are critical for the induction of some forms of LTP, BDNF-induced modulation of NMDA receptors may influence the threshold for eliciting LTP. Furthermore, interaction between BDNF/TrkB signaling and NMDA receptors appears to be important in spatial memory (Mizuno et al., 2003). In addition, continuous intracerebroventricular infusion of PP2 in rats delayed memory acquisition, which is accompanied by diminished phosphorylation of Fyn and NR2B, but not trkB. These observations indicate the importance of Fyn-mediated modulation of NMDA receptor activity in BDNF-induced synaptic plasticity, correlating well with our findings.#

Using expression in heterologous cells, Kohr and Seeburg (1996) reported that NR2A, but not NR2B, containing receptors are affected by Fyn. Nonetheless, we believe that our results are real for the following reasons. First, the Kohr and Seeburg findings were obtained with recombinant channels expressed in heterologous system, whereas our findings were obtained with native channel in primary neurons. The signaling pathways in these two quite different systems may not be the same. Second, our primary goal was to examine elements of BDNF-induced changes in NMDA receptor activity, and we have a clear demonstration of Fyn involvement. It is apparently working through the NR2B subunit because the NR2A subunit is not phosphorylated in response to BDNF treatment (Lin et al., 1998). Third, another indicator that the effect is mediated by the NR2B subunit is that we are recording the activity of extrasynaptic NMDA receptors which are thought to be NR1-NR2B diheteromers that do not contain the NR2A subunit.#

The exact functions of BDNF-mediated Fyn-dependent Tyr¹⁴⁷²-NR2B phosphorylation remain to be determined. Nonetheless, we propose that the phosphorylation may play significant roles in various synaptic processes. For example, BDNF-mediated Tyr¹⁴⁷² phosphorylation may alter the conformation of the NR2B subunits, leading to altered NMDA receptor conductivity and channel activity. In addition, protein tyrosine phosphorylation reportedly regulates protein–protein interactions (Pawson and Scott, 1997). NR2B is associated with a variety of proteins in the NMDA receptor complex through the C-terminal tail (Sheng, 1996), a region that binds to the functionally critical PDZ domain of PSD-95 (Kornau et al., 1995). Consequently, BDNF-mediated Tyr¹⁴⁷² phosphorylation in this domain may alter protein interactions of signaling molecules (Pawson and Scott, 1997), resulting in modulation of synaptic transmission and synaptic plasticity. Moreover, it has been shown that tyrosine phosphorylation of acetylcholine receptors at neuromuscular junctions increases clustering of the receptors (Wallace et al., 1991). As the C-terminal tail of NR2B is important in synaptic localization and clustering of the receptor (Mori et al., 1998), BDNF-mediated Tyr¹⁴⁷² phosphorylation may similarly regulate synaptic NMDA receptor clustering that is critical for rapid and efficient synaptic signaling.#

In summary, results of our studies raise the possibility that Fyn-dependent phosphorylation of Tyr¹⁴⁷²-NR2B may constitute a key mechanism underlying BDNF synaptic plasticity. This is the first demonstration that BDNF signaling may be regulated at a single amino acid level.#

Protein concentration was determined using the BCA protein assay kit (Pierce, Rockford, IL) with bovine serum albumin as a standard.#

Low density cultures of dissociated embryonic day 16 mouse hippocampi were prepared according to the procedures of Levine et al. (1998). Time-mated pregnant mice were sacrificed by exposure to CO2 vapor. Fetal hippocampi were dissected, and meninges were removed. Pooled tissue from each litter was triturated through fire-polished Pasteur pipettes in 2 ml of Eagle's minimum essential medium (MEM) with added glucose and 7.5% FBS. Cells were plated on poly-d-lysine-coated culture dishes at a final density of 350,000 cells/35 mm dish. Cultures were maintained in serum-free medium (SFM) at 37 °C in a 95% air/5% CO2 humidified incubator for 10–14 days. SFM consisted of a 1:1 (vol/vol) mixture of Ham's F-12 and Eagle's minimum essential medium and supplemented with insulin (25 μg/ml), transferring (100 μg/ml), putrescine (60 μM), progesterone (20 nM), selenium (30 nM), glucose (6 mg/ml), penicillin (0.5 U/ml) and streptomycin (0.5 mg/ml). These cultures contained virtually pure neurons, as judged by neuron-specific enolase immunocytochemistry.#

PSDs (80 μg protein in a final volume of 100 μl) were suspended in a HEPES buffer consisting of 50 mM HEPES (pH 7.5), 0.2 mM EGTA, 2 mM CaCl2, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 20 μg/ml soybean trypsin inhibitor and 1 mM PMSF. Vehicle or BDNF (2, 5, 10, 25 ng/ml) was added to each experimental group. In parallel, NGF (2 ng/ml) and NT-3 (2 ng/ml) were used to determine the specificity of the neurotrophin response. The reaction mixtures were incubated for 10 min in the presence of 20 μM ATP at 37 °C to define dose response relationships. To examine the time course of BDNF effects, BDNF (2 ng/ml) was incubated for 5, 10 or 30 min. In other experiments, PSDs were preincubated in the presence or absence of pharmacological inhibitors for 20 min and then incubated with vehicle or 2 ng/ml BDNF for 10 min. The reactions were stopped by adding a sample buffer (62.5 mM Tris–HCl, pH 6.8, 2% SDS, 5% β-mercaptoethanol, 10% glycerol and 0.002% bromophenol blue) and then subjected to SDS-PAGE for Western blot analyses.#

All procedures were as described previously (Suen et al., 1997) and performed at room temperature unless otherwise stated. Samples containing equal amounts of protein were denatured in the sample buffer at 90 °C for 5 min and subjected to vertical slab 7–11.5% gradient SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Electrophoresis was carried out at 70 V for 16 h. Proteins were electrophoretically transferred to Immobilon-P membrane (Millipore, Bedford, MA) at 200 mA for 1 h. The transfer buffer consisted of 2 mM Tris–HCl, 39 mM glycine, 0.0375% SDS and 20% (vol/vol) methanol. The resultant membrane was blocked with 5% fat-free milk in 0.1% TBS-T (20 mM Tris–HCl, pH 7.4, 0.1% Tween 20) for 1 h and then incubated for another 1 h with primary antibodies, anti-NR2B or anti-Fyn antibodies (1 μg/ml) followed by incubation with secondary antibodies, anti-rabbit IgG (1:5000), for 45 min. Immunopositive bands were visualized by chemiluminescence using the ESL detection kit (NEN, Boston, MA). To detect levels of phosphorylation, the same membrane was stripped in a buffer consisting of 62.5 mM Tris–HCl, pH 6.8, 2% SDS and 100 mM β-mercaptoethanol. Antibodies against phospho-Tyr¹⁴⁷²-NR2B or PY20 (1 μg/ml) were then used as primary antibodies.#

It should be noted that, on the same membrane, we first examined the phosphorylation level using the Tyr¹⁴⁷² antibodies and then examined the protein level using NR2B antibodies after stripping off the former antibodies. This method has been employed by many investigators and has been accepted in our previous publications (Suen et al. 1997; Lin et al. 1998; Wu et al. 2004). Indeed, after stripping, the proteins in the membranes were incubated with buffer allowing returning proteins to the original, active state. We found that stripping did not alter the extent of subsequent antibody probing. Of note, our previous studies, using the method, revealed that BDNF enhanced tyrosine phosphorylation of NR2B 1.7-fold, mimicking the effects of hippocampal LTP (Rosenblum et al. 1996; Rostas et al. 1996), suggesting that the stripping did not affect the subsequent antibody binding. Moreover, we found similar results using anti-phospho-Tyr¹⁴⁷²-NR2B before anti-NR2B, indicating that the order of application of antibodies was not significant.#

To examine the effects of BDNF on Fyn phosphorylation, PSDs (200 μg protein in a final volume of 100 μl) were suspended in the HEPES buffer and preincubated in the presence or absence of PP2 (1 μM) for 20 min at 37 °C. Vehicle or BDNF (2 ng/ml) was then added and incubated for addition 10 min in the presence of 20 μM ATP. The reactions were stopped by putting on ice and then subjected to immunoprecipitation.#

After the reaction with BDNF, the mixture was first centrifuged at 6700×g at 4 °C for 5 min. The pellets were dissolved in 100 μl of 1% SDS followed by addition of 9 volumes of lysis buffer (50 mM Tris–HCl, pH 7.5, 150 mM NaCl, 10 mM EDTA, 2 mM EGTA, 1% Triton X-100, 1% CHAPS, 0.5% Nonidet P-40, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 20 μg/ml soybean trypsin inhibitor, 50 mM NaF, 1 mM PMSF, 1 mM o-vanadate and 0.5 mM microcystin-LR). The mixtures were shaken at 4 °C for 1 h. Then, the insoluble material was removed by centrifugation at 14,000×g for 5 min. Anti-Fyn antibodies (4 μg each) were added to the supernatant and incubated overnight at 4 °C. Subsequently, 50 μl of Protein-A–Sepharose CL-4B (Amersham Biosciences) was added and mixed for 2 h at 4 °C. The Sepharose–antigen–antibody complexes were pelleted by centrifugation at 14,000×g for 5 s and washed 3 times (20 min each) with a washing buffer (50 mM Tris–HCl, pH 8.1, 150 mM NaCl, 1% Triton X-100, 0.5% Nonidet P-40, 0.1% SDS and 1 mM PMSF). The immunoprecipitates were resuspended in 40 μl of sample buffer and subjected to SDS-PAGE for Western analysis to determine levels of Fyn protein and its tyrosine phosphorylation as described above.#

The intensity of bands was quantitated on a Bio-Rad (Hercules, CA) Gel Doc. Phosphorylation level per unit protein was obtained by dividing arbitrary densitometric units in immunoblot of phosphorylation over those in immunoblot of protein. Phosphorylation level per unit protein in the vehicle control was assigned a value of 100%. All experiments were performed 3 times for statistical analysis by Student's t-test. All results are mean±standard error of the mean (SEM). Representative Western blots are shown in figures.#

To study the effect of BDNF on Tyr¹⁴⁷² phosphorylation in hippocampal neurons, 6 dishes of cultured neurons (350,000 cells per dish) were treated with or without 20 ng/ml BDNF at 37 °C for 10 min. Neurons were washed with PBS twice and each dish was solubilized in 100 μl of a lysis buffer (50 mM Tris–HCl, pH 7.5, containing 0.5% SDS, 150 mM NaCl, 10 mM EDTA, 2 mM EGTA, 1% Triton X-100, 1% CHAPS, 0.5% Nonidet P-40, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 20 μg/ml soybean trypsin inhibitor, 50 mM NaF, 1 mM PMSF, 1 mM o-vanadate and 0.5 mM microcystin-LR). The lysates were then scrapped off and processed to immunoprecipitation using anti-NR2B antibodies (4 μg per tube). The immunoprecipitates were subjected to electrophoresis followed by Western analysis for levels of NR2B protein and its Tyr¹⁴⁷² phosphorylation.#