Glucocorticoid receptor interaction with TrkB promotes BDNF-triggered PLC-γ signaling for glutamate release via a glutamate transporter

Abstract

An increase in glucocorticoid levels and down-regulation of BDNF (brain-derived neurotrophic factor) are supposed to be involved in the pathophysiology of depressive disorders. However, possible crosstalk between glucocorticoid- and BDNF-mediated neuronal functions in the CNS has not been elucidated. Here, we examined whether chronic glucocorticoid exposure influences BDNF-triggered intracellular signaling for glutamate release via a glutamate transporter. We found that chronic exposure to dexamethasone (DEX, a synthetic glucocorticoid) suppressed BDNF-induced glutamate release via weakening the activation of the PLC-γ (phospholipase C-γ)/Ca²⁺ system in cultured cortical neurons. We demonstrated that the GR (glucocorticoid receptor) interacts with receptor tyrosine kinase for BDNF (TrkB). Following DEX treatment, TrkB-GR interaction was reduced due to the decline in GR expression. Corticosterone, a natural glucocorticoid, also reduced TrkB-GR interaction, BDNF-stimulated PLC-γ, and BDNF-triggered glutamate release. Interestingly, BDNF-dependent binding of PLC-γ to TrkB was diminished by DEX. SiRNA transfection to induce a decrease in endogenous GR mimicked the inhibitory action of DEX. Conversely, DEX-inhibited BDNF-activated PLC-γ signaling for glutamate release was recovered by GR overexpression. We propose that TrkB-GR interaction plays a critical role in the BDNF-stimulated PLC-γ pathway, which is required for glutamate release, and the decrease in TrkB-GR interaction caused by chronic exposure to glucocorticoids results in the suppression of BDNF-mediated neurotransmitter release via a glutamate transporter.

Most patients with depression exhibit prolonged elevation of a glucocorticoid stress hormone, cortisol (1, 2). The blood level of glucocorticoids (cortisol in humans and corticosterone in rodents) is regulated by the hypothalamic-pituitary-adrenal (HPA) axis (2). When excessive stress is prolonged, abnormally increased amounts of glucocorticoids may damage the CNS and cause depressive symptoms, which can be decreased with antidepressants (3–5).

Glucocorticoids function via the glucocorticoid receptor (GR), which regulates gene transcription. Glucocorticoids contribute to glucose homeostasis, cell differentiation, and inflammation (6). Additionally, glucocorticoids and the GR influence neuronal functions such as hippocampal long-term potentiation/depression (7–9) and cognitive function governed by the prefrontal cortex (10). The GR potentiates the response to NMDA in dopamine-sensitive neurons in the ventral tegmental area (11) and modulates the NMDA receptor function in the spinal cord following peripheral nerve injury (12), suggesting that the GR is involved in synaptic plasticity.

Beyond the promotion of cell differentiation, nerve growth, and neuronal survival, brain-derived neurotrophic factor (BDNF) plays a crucial role in synaptic function (13–15). For instance, BDNF increases neurotransmitter release (16, 17). We reported that BDNF rapidly induces glutamate transporter-mediated glutamate release via phospholipase C-γ (PLC-γ)/Ca²⁺ signaling and that antidepressants enhance PLC-γ/Ca²⁺ signaling (18, 19). Growing evidence has suggested a close relationship between BDNF and the pathophysiology of depression (20, 21). The BDNF level was low in the brains of suicide victims, most of whom had depressive disorders (22). BDNF plays a critical role in cognition, learning, and memory, and patients with depression exhibit deficits in these brain functions (23, 24). Both BDNF and glucocorticoids/GR are involved in synaptic function and the pathophysiology of depression. However, the possible influence of glucocorticoids on the acute action of BDNF is poorly understood.

Here we report that chronic treatment with glucocorticoids suppressed BDNF-triggered PLC-γ signaling for glutamate release via a glutamate transporter. We found that the GR interacted with receptor tyrosine kinase for BDNF (TrkB), playing an important role in BDNF action.

Results

Chronic Dexamethasone (DEX) Treatment Suppressed BDNF-Induced Glutamate Release by Inhibiting PLC-γ/Ca²⁺ Signaling.

We examined BDNF-induced glutamate release in cultured cortical neurons after exposure to DEX (a synthetic GR-selective agonist). DEX pretreatment (48 h) suppressed BDNF-induced glutamate release in a dose-dependent manner (Fig. 1Ai). After various durations of DEX (1 μM) exposure, we found that DEX exposure for 24 or 48 h inhibited BDNF-induced glutamate release, whereas shorter exposure times (10 min to 12 h) did not (Fig. 1Aii). When the dose-dependency of BDNF on glutamate release was examined, DEX inhibited BDNF-induced release at any dose of BDNF (Fig. 1Aiii). In this study, the following experiments were performed with 100 ng/ml of BDNF. Cell viability [supporting information (SI) Fig. S1 A and B] and the number of glutamatergic and GABAergic synapses (Fig. S1 C and D) were unchanged by DEX exposure for 48 h. The endogenous GR expression in vitro and in vivo during neuronal maturation is shown in Fig. S2 A and B. In this study, we applied DEX at days in vitro (DIV) 4–5 because the expressions of the GR and synaptic proteins markedly increase (Fig. S2A), and BDNF-induced glutamate release begins at approximately DIV 5 (18). We confirmed the inhibitory effects of TTX (a Na⁺ channel blocker) and TBOA (a glutamate transporter inhibitor) on BDNF-induced glutamate release (Fig. S3A). Tetanus toxin, an exocytosis inhibitor, had no effect on BDNF-induced glutamate release (Fig. S3B). These results suggest that BDNF-induced glutamate release occurs via a glutamate transporter in a Na⁺-dependent manner as we previously reported (18).

Fig. 1.

Chronic DEX treatment suppressed BDNF-activated PLC-γ/Ca²⁺ signaling for glutamate release in cultured cortical neurons. (Ai) Dose-dependent inhibitory effect of DEX on BDNF-induced glutamate release. DEX (0.01–10 μM) was applied at DIV 4. After 48 h, BDNF (100 ng/ml, 1 min) was added. Con means no DEX treatment. Data represent mean ± SD. (n = 4). (ii) Time-course analysis of the DEX effect on BDNF-induced glutamate release. DEX (1 μM) was added at DIV 5 for the indicated durations. Data represent mean ± SD. (n = 3). (iii) Dose-dependency of BDNF on glutamate release after DEX treatment. DEX (1 μM, at DIV 4) was applied for 48 h. Then, BDNF (3–100 ng/ml) was added. Data represent mean ± SD. (n = 5). ***P < 0.001, **P < 0.01, *P < 0.05 versus basal, ^###P < 0.001, ^##P < 0.01, ^#P < 0.05 versus BDNF-induced release in Con. (t test). (B) DEX prevented BDNF-increased intracellular Ca²⁺. Data from 61 randomly selected cells for each experimental condition. The fluorescence ratio (F/F₀; BDNF-induced/basal) was calculated. DEX (1 μM) was applied at DIV 4. After 24 h later, Ca²⁺ imaging was performed. ***P < 0.001 versus vehicle (BSA) in the control. ^###P < 0.001 versus BDNF-induced increase in the control. (t test). (Ci) Immunoprecipitation with anti-PLC-γ (Upper) or anti-TrkB (Lower) antibodies was carried out. Blotting was performed with anti-pTyr, anti-PLC-γ, or anti-TrkB antibodies. DEX (1 μM) was applied at DIV 4. After 48 h, BDNF was applied for the indicated duration. (ii) Quantification of pPLC-γ or pTrkB. Data represent mean ± SD. (n = 4). Data were normalized to the level of BDNF (0.5 min) in Con. N.D.: not detected. **P < 0.01, *P < 0.05 versus BDNF-induced in Con. (t test). (iii) DEX did not affect pAkt or pERK1/2. BDNF was applied for the indicated duration.

BDNF-induced glutamate release depends on an intracellular Ca²⁺ increase via IP₃-sensitive Ca²⁺ channels (IP₃ receptor) (18). As expected, chronic DEX treatment weakened BDNF-increased Ca²⁺ (Fig. 1B and Fig. S4 A and B). We confirmed that U73122 (a PLC-γ inhibitor) and xestospongin C (Xest C, an IP₃ receptor inhibitor) blocked BDNF-increased Ca²⁺, although BDNF still increased Ca²⁺ in the presence of APV (an NMDA receptor inhibitor), CNQX (an AMPA receptor inhibitor), or bicuculline (a GABA_A receptor inhibitor) (Fig. S4 C and D), indicating the importance of the PLC-γ/IP₃ pathway. We confirmed that both U73122 and Xest C blocked BDNF-induced glutamate release in the control and DEX-treated cultures (Fig. S4E). These results suggest that BDNF-induced glutamate release depends on the PLC-γ pathway.

Next, we focused on PLC-γ activation (phosphorylation). A significant decline in BDNF-activated PLC-γ following chronic DEX exposure was observed, although TrkB (upstream of PLC-γ) was equally activated by BDNF with or without DEX (Fig. 1 Ci and ii). In other pathways activated by TrkB, DEX did not change activation of Akt (pAkt, phosphorylated Akt) or ERK1/2 (pERK1/2) stimulated by BDNF (Fig. 1Ciii).

Recently, activation of TrkB signaling within several hours of glucocorticoid exposure was reported (25). Indeed, TrkB, PLC-γ, Akt, and ERK1/2 were activated by short-term application of glucocorticoids (DEX or corticosterone) (Fig. S5 A and B). These activations reached their maximum at 2–4 h after the application and returned to the basal level at 6 h. As expected, BDNF induced much higher activation of TrkB signaling (including PLC-γ, Akt, and ERK1/2) compared with that induced by sole acute DEX (2 h) or corticosterone (2 h) exposure (Fig. S6 A–C). In contrast to chronic exposure, such a short-term treatment with DEX (2 h) or corticosterone (2 h) did not affect the exogenous BDNF-stimulated TrkB signaling, including PLC-γ. Subsequently, we focused on the suppression of BDNF-dependent PLC-γ signaling for glutamate release after long-lasting glucocorticoid exposure.

DEX-Dependent GR Down-Regulation Was Involved in the Suppressed Responses to BDNF.

To investigate the mechanisms underlying the DEX-suppressed responses to BDNF, the possible involvement of the GR was examined. When DEX was coapplied with RU486 (a GR antagonist), BDNF-stimulated PLC-γ activation and glutamate release were not inhibited (Fig. 2 Ai and ii and Fig. 2B), suggesting that DEX acts via the GR. Thus, endogenous GR expression after DEX addition was examined. Marked down-regulation of the GR was observed following DEX application for 24 to 48 h (Fig. 2C). DEX induced GR down-regulation in a dose-dependent manner (Fig. 2 Di and ii), while the mineral corticoid receptor (MR, the other receptor for glucocorticoid) and TUJ1 (class III β-tubulin, a neuronal marker) expression was intact (Fig. 2Di). Immunocytochemistry with anti-microtubule-associated protein 2 (MAP2) and anti-GR antibodies was performed, and the quantification of the staining indicated a DEX-dependent GR reduction in neurons (Fig. 2 Ei–vii). As expected, RU486 recovered GR down-regulation by DEX (Fig. 2 Fi and ii). Corticosterone also down-regulated the GR and BDNF-induced glutamate release (Fig. S7 A–C). Moreover, the suppression of BDNF-activated PLC-γ by corticosterone was also observed (Fig. S7 D and E). We confirmed that the GR was markedly reduced in the homogenates of the cerebral cortex prepared from rats after DEX administration, although the MR level was intact (Fig. S7 F and G). These results suggested that the inhibitory action of glucocorticoid results from the down-regulation of the GR.

Fig. 2.

RU486 blocked DEX-decreased PLC-γ activation, glutamate release, and GR expression. (Ai) RU486 (RU, a GR antagonist) blocked DEX-decreased PLC-γ activation. DEX (1 μM) and RU486 (1 μM) were coapplied at DIV 4. Subsequently, 48 h later, BDNF was added for 1 min. (ii) Quantification of pPLC-γ. Data represent mean ± SD. (n = 4). Normalization to the level in BDNF-stimulated PLC-γ with no pretreatment. ***P < 0.001 (t test). (B) RU486 blocked the inhibitory effect of DEX on BDNF-induced glutamate release. Data represent the mean ± SD. (n = 4). ***P < 0.001 (t test). (C) Time-course of DEX-decreased GR expression. DEX (1 μM, at DIV 4) was applied for 10 min or 1–48 h. (Di) Dose-dependency of DEX on GR down-regulation. After DEX (0.01–100 μM, at DIV 4) exposure for 48 h, GR, MR, and TUJ1 were detected. (ii) Quantification of the GR. Data represent mean ± SD. (n = 5). Normalization to the level in 0 μM. ***P < 0.001 (t test). (Ei and iv) Immunostaining with anti-MAP2 and (ii and v) anti-GR antibodies. (iii and vi) Merged images. Upper: control. Lower: DEX-treated. DEX (1 μM, at DIV 4) was applied for 48 h. (Scale bar, 50 μm.) (vii) GR immunoreactivities of randomly selected regions from cell bodies or neurites. Normalization to the level in Con. N indicates the number of selected regions. At least 20 neurons from 6 coverslips were examined. (Fi) RU486 inhibited GR down-regulation by DEX. TUJ1 is the control. (ii) Quantification of the GR. Data represent mean ± SD. (n = 5). Normalization to the level in no treatment. ***P < 0.001, **P < 0.01 (t test).

Then, the effects of GR overexpression were examined. After viral infection, about 85% of MAP2-positive cells in either the control (GFP) or GR-overexpressing (GR and GFP) cultures were GFP-positive, indicating that the majority of neurons were infected (Fig. 3Ai). Blotting with anti-GR and anti-GFP antibodies showed GR overexpression (Fig. 3Aii), which enhanced BDNF-induced glutamate release (Fig. 3B). DEX failed to reduce BDNF-activated PLC-γ in GR-infected cultures (Fig. 3 Ci and ii). Next, siRNA was used to examine the function of endogenous GR. Approximately 60% of endogenous GR was depleted by GR-siRNA (Fig. 3D), and BDNF-induced glutamate release was decreased in GR-siRNA-transfected cultures (Fig. 3E). GR-siRNA reduced BDNF-activated PLC-γ (Fig. 3 Fi and ii) but not pAkt or pERK1/2 (Fig. S8 A–C). These results suggest that the amount of GR expression is critical for BDNF-stimulated PLC-γ signaling and glutamate release.

Fig. 3.

BDNF-induced glutamate release was enhanced by GR overexpression and reduced by GR down-regulation. (Aia and d) MAP2- and (b and e) GFP-positive images. (c) Overlay of a and b. (f) Overlay of d and e. (a–c) GFP-infected control. (d–f) Overexpression of both GFP and GR. (Scale bar, 50 μm.) Cells were infected at DIV 4, and fixed at DIV 6. (ii) GR overexpression was checked by anti-GR and anti-GFP antibodies. (B) GR overexpression enhanced BDNF-induced glutamate release. None: no infection. Con: sole GFP. GR: both GFP and GR. Infection was performed at DIV 4, and glutamate was measured at DIV 6. Data represent mean ± SD. (n = 4). ***P < 0.001 (t test). (Ci) DEX-decreased BDNF-activated PLC-γ was recovered by GR overexpression. DEX (1 μM) was applied at DIV 5 (24 h after infection). After 24 h, BDNF (1 min) was applied. (ii) Quantification of pPLC-γ. Data represent mean ± SD. (n = 7). Normalization to the level in BDNF-activated without DEX in GFP-infected. ***P < 0.001 (t test). (D) Endogenous GR was decreased after GR-siRNA transfection. Scramble siRNA had no effect. MR and TUJ1 are the controls. SiRNA was transfected 48 h before lysates were collected. (E) BDNF-induced glutamate release in GR-siRNA-transfected cultures was reduced. Data represent mean ± SD. (n = 4). (t test). ^##P < 0.01 versus BDNF-induced in the scramble. (Fi) BDNF-activated PLC-γ was decreased by the GR-siRNA. (ii) pPLC-γ was quantified. Data represent mean ± SD. (n = 4). Normalization to the level in BDNF-activated PLC-γ in None. ***P < 0.001 (t test).

DEX Decreased TrkB-GR Interaction and Binding of PLC-γ to TrkB.

How does chronic DEX interrupt PLC-γ signaling? Initially, endogenous PLC-γ, TrkB, or BDNF expression after DEX exposure was examined; however, the levels of these proteins were intact (Fig. S9 A and B). The level of TrkB on the cell surface was also unchanged (Fig. S9 C and D). Subsequently, the possible interaction between GR and TrkB was investigated. Following immunoprecipitation with anti-GR antibody, coprecipitated TrkB was found (Fig. 4Ai). The degree of coprecipitated TrkB was not changed by BDNF and/or DEX (Fig. 4 Aii and iii). Significant coprecipitated GR after immunoprecipitation of TrkB was also observed (Fig. 4Bi). Remarkably, DEX reduced the coprecipitated GR with or without BDNF application (Fig. 4 Bii and iii). In the control, a BDNF-dependent slight increase in the coprecipitated GR was observed. To inspect the specificity of the interaction, we used a competitive peptide to block TrkB immunoprecipitation. The peptide, containing a sequence for the epitope of the antibody, blocked the immunoprecipitation of TrkB and the coprecipitation of the GR (Fig. 4C). Moreover, GR overexpression increased TrkB-GR interaction, and GR-siRNA transfection decreased TrkB-GR interaction (Fig. S10 A–C). Corticosterone also reduced TrkB-GR interaction (Fig. S10 D and E). Immunocytochemical analysis showed that the merged level of the GR- and TrkB-positive signal was diminished because chronic DEX exposure decreased the GR levels (Fig. S11 A and B). In contrast to chronic exposure, acute DEX or corticosterone had no effect on TrkB-GR interaction with or without BDNF (Fig. S12). Interestingly, BDNF-induced binding of PLC-γ to TrkB decreased due to chronic DEX (Fig. 4 Di and ii). Marked reduction in TrkB-GR interaction in the homogenates of the cerebral cortex prepared from rats treated with DEX injection was confirmed (Fig. 4 Ei and ii).

Fig. 4.

TrkB-GR interaction and BDNF-dependent binding of PLC-γ to TrkB were decreased after DEX exposure. (Ai) After immunoprecipitation with the anti-GR antibody, blotting with the anti-TrkB antibody was performed. Lysates from DIV 6 cultures were used. Preimmune: preimmune serum control. noAb: no anti-GR antibody. Total: 10% input (total lysates). (ii) Immunoprecipitation of the GR was performed after BDNF stimulation with or without DEX pretreatment. DEX (1 μM, at DIV 4) was applied for 48 h. Subsequently, BDNF was added (1 min). Anti-TrkB and anti-GR antibodies were used for blotting. (iii) Quantification of the coprecipitated TrkB. Normalization to the level without DEX and BDNF. Data represent mean ± SD. (n = 7). (Bi) Immunoprecipitation with anti-TrkB antibody. Blotting with anti-GR antibody was performed. noAb: no anti-TrkB antibody. (ii) Immunoprecipitation of TrkB after BDNF stimulation with or without DEX. Blottings were performed with anti-GR and anti-TrkB antibodies. (iii) The coprecipitated GR was quantified. Data represent mean ± SD. (n = 9). Normalization to the level without DEX and BDNF. ***P < 0.001, **P < 0.01 (t test). (C) Inhibition of the coprecipitation of the GR by a competitive peptide to block the immunoprecipitation of TrkB. Lysates from GR-overexpressed cortical cultures were used. (Di) TrkB-PLC-γ interaction after DEX treatment. After immunoprecipitation of PLC-γ, coprecipitated TrkB was detected. DEX (1 μM, at DIV 4) was applied for 48 h before BDNF addition (1 min). (ii) Coprecipitated TrkB was quantified. Data represent mean ± SD. (n = 5). **P < 0.01 (t test). Data were normalized to no treatment. (Ei) DEX exposure reduced TrkB-GR interaction in vivo. P7 rats received i.p. injections (0.1–10 mg/kg i.p.) of DEX or vehicle. Samples were obtained 48 h after the injections. (ii) The coprecipitated GR was quantified. Data were obtained from the control and 0.1 mg/kg DEX. Data represent mean ± SD. (n = 4). Normalization to the level in con. ***P < 0.001 (t test).

To further assess TrkB-GR interaction, 3 types of GR plasmids containing His tags were constructed (Fig. 5A). GR-FL (full length), GR-N (including DNA binding site), GR-C (including ligand binding site), and GFP (con) were transfected into SH-SY5Y cells. As the anti-GR antibody recognized GR-FL and GR-N but not GR-C (Fig. 5Bi, the epitope for the antibody exists in the N-terminal region of GR), blotting with anti-His antibody was also performed with total lysates (Fig. 5Bii). After immunoprecipitation of TrkB, blotting with anti-GR (Fig. 5Biii), anti-TrkB (Fig. 5 Biii and iv), and anti-His antibodies (Fig. 5Biv) was conducted. GR-FL and GR-N were coprecipitated; however, GR-C failed to interact with TrkB, indicating the importance of the N-terminal region of the GR. BDNF-activated PLC-γ in GR-FL-transfected SH-SY5Y cells was enhanced compared with the control, whereas such enhancement was not detected in GR-N- or GR-C-transfected cells (Fig. 5 Ci and ii). Finally, the responses to BDNF in cortical neurons transfected with these GR plasmids were examined. PLC-γ activation and glutamate release in GR-FL-transfected neurons were reinforced; in contrast, neither PLC-γ activation nor glutamate release was enhanced by GR-N and GR-C transfection (Fig. 5 Di and ii). These results suggest that the N-terminal region of the GR interacts with TrkB; however, the C-terminal region is also required to boost BDNF-activated PLC-γ.

Fig. 5.

The N-terminal region of the GR was required for interaction with TrkB. (A) Three types of GR plasmids, GR-FL (nt1- nt2331), GR-N (nt1- nt1458), and GR-C (nt1459- nt2331), were constructed. (B) GR plasmids and GFP plasmid (con) were transfected into SH-SY5Y cells. The exogenous GR in total lysates was detected by (i) anti-GR and (ii) anti-His antibodies. As the anti-GR antibody recognized GR-FL and GR-N but not GR-C, blotting with the anti-His antibody was also performed. (iii) After TrkB immunoprecipitation, blotting was performed with anti-GR and anti-TrkB antibodies. GR-FL and GR-N were coprecipitated. (iv) GR-C failed to interact with TrkB. Blotting with the anti-His antibody was conducted. (Ci) GR-FL overexpression potentiated BDNF-activated PLC-γ in SH-SY5Y cells. BDNF was applied for 15 min. (ii) pPLC-γ was quantified. Data represent mean ± SD. (n = 7). The ratio (BDNF-stimulated/basal) of pPLC-γ was calculated. **P < 0.01 versus BDNF-stimulated in con. (t test). (D) Three types of GR plasmids and a GFP plasmid (con) were transfected into cortical neurons. BDNF-dependent (i) pPLC-γ and (ii) glutamate release were enhanced by GR-FL transfection. Data represent mean ± SD. (n = 4). ***P < 0.001 (t test).

Discussion

We have shown that chronic pretreatment with DEX disturbs BDNF-stimulated PLC-γ signaling for glutamate release via a glutamate transporter. Chronic DEX caused marked GR down-regulation. GR overexpression recovered the reduction in BDNF-activated PLC-γ, and siRNA transfection for endogenous GR mimicked the inhibitory effect of DEX. Corticosterone also reduced the GR level and suppressed BDNF-stimulated PLC-γ and glutamate release. Interestingly, we found that the GR interacted with TrkB and that the TrkB-GR interaction and the BDNF-dependent binding of PLC-γ to TrkB decreased following DEX exposure.

BDNF-activated PLC-γ was specifically down-regulated by DEX or corticosterone whereas the activation of TrkB, Akt (a component of the PI3K pathway), and ERK signaling was not affected. A study on an animal model of depression showed a different responsiveness at the level of PI (phosphoinositide)-PLC after single vs. repeated stress (26). Additionally, long-term administration of antidepressants (desipramine, fluoxetine, and phenelzine) decreases PI-PLC activity in the membrane and cytosol fractions of the rat cortex (27). Meanwhile, imipramine activates PLC-β1 in the rat frontal cortex membrane (28). Recently, we reported that BDNF-activated PLC-γ was increased by imipramine (19). These studies suggest that PLC activity is critical in the pathophysiology of depression and the effects of antidepressants. In isolated rat islets, DEX suppresses PLC activation and insulin secretion (29). Thus, to reveal the cell biology of stress hormones, it may be valuable to focus on the PLC/Ca²⁺ system in neuronal or nonneuronal cells.

Glucocorticoids acutely activate TrkB signaling via the genomic function of the GR (25). Activation of TrkB, PLC-γ, Akt, and ERK was increased by short-term application of DEX and corticosterone, reaching the maximum at 2–4 h after the application, and returned to the baseline in our cultures. In contrast, no change in the BDNF-stimulated activations of TrkB signaling, including PLC-γ, was observed after such a short-term pretreatment with DEX or corticosterone. In this study, decreased responses to BDNF (not response to the glucocorticoid alone) after long-term glucocorticoid exposure (24–48 h) were discovered. Collectively, these results suggest that glucocorticoids play various functions depending on exposure time and that the mechanism underlying the down-regulation of BDNF-dependent PLC-γ signaling after chronic glucocorticoid exposure differs from the activation of TrkB signaling by acute exposure.

Down-regulations of the GR and TrkB-GR interaction caused by chronic DEX or corticosterone were observed. Such down-regulation was also observed in vivo. It is possible that the GR down-regulation may simply result in the decrease of the GR/TrkB complex. GR-overexpressing neurons showed a high response to BDNF, and siRNA for the GR mimicked the action of DEX, suggesting that moderate levels of GR expression may be essential for the BDNF/TrkB/PLC-γ system. In this study, a BDNF-dependent slight increase in TrkB-GR interaction in the control cultures was observed. In contrast, fluctuation in TrkB-GR interaction by GR overexpression or down-regulation was observed without BDNF stimulation, and endogenous BDNF was not changed by DEX. Thus, both BDNF (or phosphorylation of TrkB)-dependent and -independent mechanisms might be involved in TrkB-GR interaction.

Glucocorticoids have a rapid influence on intracellular signaling (not via transcriptional activity), and this rapid action is presumed to be mediated via membrane-bound and nonmembrane-bound GR (classical GR) or putative G protein-coupled receptors on a plasma membrane (30, 31). We speculate that classical GR is involved in TrkB/PLC-γ signaling because RU486 blocked the inhibitory effect of DEX, and the overexpression and down-regulation of the GR influenced TrkB/PLC-γ signaling. Raf-1 and 14–3-3 interact with liganded- or nonliganded GR (32), implying that the GR directly influences signaling pathways in cytosol. Moreover, the GR affects the plasma membrane receptor in immune T-cells (30). The T cell receptor (TCR) makes a protein complex including the GR. After the glucocorticoid is bound to the GR, the GR dissociates from the complex, and TCR signaling is inhibited. A similar mechanism might be involved in TrkB-GR interaction. Interestingly, FMS-like tyrosine kinase 3 (Flt3), another member of the receptor tyrosine kinase (RTK) family, interacts with the GR in hematopoietic cells (33). Flt3 interacts with the N-terminal region of the GR in the presence and absence of glucocorticoid. In the present study, TrkB/PLC-γ signaling was potentiated by GR overexpression and declined by GR down-regulation without DEX, implying that the GR has a positive effect on PLC-γ signaling in a ligand-independent fashion. In our system, the N-terminal region of the GR interacts with TrkB; however, the C-terminal region is also required for the full activation of PLC-γ. These results, including those of our study, indicate an important role of GR in the signaling of the RTK family.

The influx of Ca²⁺ regulates BDNF expression (34, 35). In addition, glutamate can regulate neurotrophin expression (36, 37). BDNF is produced and released in an activity-dependent manner (38, 39). Collectively, it is possible that suppression in BDNF-stimulated PLC-γ/Ca²⁺ signaling for glutamate release may be followed by a reduction in BDNF protein. Previously, we found that antidepressants potentiated BDNF-stimulated PLC-γ/Ca²⁺ (19). Therefore, our system, in which a glucocorticoid exerts an inhibitory effect on BDNF-stimulated PLC-γ/Ca²⁺, may be useful for evaluating novel analogs as antidepressant candidates. Interestingly, the high-affinity interaction of pro-neurotrophin with a low-affinity receptor p75 was reported (40). It may be valuable to study, not only with respect to the expression of mature BDNF but also in terms of the neurotrophin form (pro-/mature) and the affinity of the ligand/receptor/adaptor interaction (present study) during stress hormone exposure.

Materials and Methods

Cells, Survival Assay, Ca²⁺ Imaging, Immunoprecipitation, Immunoblotting, Immunocytochemistry, and Animals.

Cortical neurons were prepared from P2 rats as previously reported (18). Cell viability was examined with an MTT assay. In Ca²⁺ imaging, in which fluo-3 dye was used, the ratio (F/F₀) of fluorescence was calculated based on the intensities of fluorescence before and after BDNF was added. Immunoprecipitation, immunoblotting, and immunocytochemistry were performed as previously described (18). For the in vivo approach, P7 rats received an i.p. injection of (0.1–10 mg/kg i.p.) DEX or vehicle (sesame oil). After 48 h, the brains were removed from the deeply anesthetized rats, and the cerebral tissues were homogenized. All animals were treated according to the institutional guidelines for the care and use of animals. Details of these experiments are available in SI Materials and Methods.

DEX Pretreatment.

After the cortical cultures were maintained for 4 or 5 days, DEX (1 μM, BIOMOL International L.P.) or corticosterone (1 μM, SIGMA) was added to the neurons by bath application. Subsequently, the cultures were incubated for 24 or 48 h in the presence of DEX or corticosterone before amino acid measurement, Ca²⁺ imaging, immunoprecipitation, immunoblotting, and immunocytochemistry were performed. DEX or corticosterone was dissolved in DMSO. Sole DMSO (vehicle) had no effects compared with no treatment (data not shown). RU486 (1 μM, LKT Laboratories) was applied 20 min before adding DEX.

Detection of Amino Acid Neurotransmitters.

HPLC was used to analyze amino acid neurotransmitters as described previously (18). Details can be found in SI Materials and Methods.

Viral GR Construct, GR Deletion Plasmids, and siRNA.

Detailed procedures for constructing the sindbis virus, producing GR plasmids, and transfecting siRNA are described in SI Materials and Methods.

Statistical Analysis.

Data are expressed as mean ± SD. Statistical significance was evaluated by student's t test, and probability values of less than 5% were considered significant.