Research ArticlePain

Brain-derived neurotrophic factor stimulation of T-type Ca2+ channels in sensory neurons contributes to increased peripheral pain sensitivity

See allHide authors and affiliations

Science Signaling  24 Sep 2019:
Vol. 12, Issue 600, eaaw2300
DOI: 10.1126/scisignal.aaw2300

Treating neuralgias

The trigeminal ganglion (TG) is a cluster of neurons near the temporal bone that provides nociceptive sensation in the head and is associated with various pain disorders, including migraine and trigeminal neuralgia. Increased expression of the neurotrophic factor BDNF and its receptor TrkB is implicated in the heightened sensitivity of the TG. Wang et al. uncovered a cross-pathway kinase cascade through which BDNF-TrkB sensitizes the TG to mechanical and inflammatory stimuli. The pathway triggered the opening of a subtype of Ca2+ channels, and the consequent pain-associated behavior in rats was diminished by intra-TG injection with respective channel blockers or kinase inhibitors. These findings may be relevant to patients that suffer from TG-associated neuralgias and may lead to the development of effective therapies.

Abstract

Although brain-derived neurotrophic factor (BDNF) is implicated in the nociceptive signaling of peripheral sensory neurons, the underlying mechanisms remain largely unknown. Here, we elucidated the effects of BDNF on the neuronal excitability of trigeminal ganglion (TG) neurons and the pain sensitivity of rats mediated by T-type Ca2+ channels. BDNF reversibly and dose-dependently enhanced T-type channel currents through the activation of tropomyosin receptor kinase B (TrkB). Antagonism of phosphatidylinositol 3-kinase (PI3K) but not of its downstream target, the kinase AKT, abolished the BDNF-induced T-type channel response. BDNF application activated p38 mitogen-activated protein kinase (MAPK), and this effect was prevented by inhibition of PI3K but not of protein kinase A (PKA). Antagonism of either PI3K or p38 MAPK prevented the BDNF-induced stimulation of PKA activity, whereas PKA inhibition blocked the BDNF-mediated increase in T-type currents. BDNF increased the rate of action potential firing in TG neurons and enhanced the pain sensitivity of rats to mechanical stimuli. Moreover, inhibition of TrkB signaling abolished the increased mechanical sensitivity in a rat model of chronic inflammatory pain, and this effect was attenuated by either T-type channel blockade or knockdown of the channel Cav3.2. Together, our findings indicate that BDNF enhances T-type currents through the stimulation of TrkB coupled to PI3K-p38-PKA signaling, thereby inducing neuronal hyperexcitability of TG neurons and pain hypersensitivity in rats.

INTRODUCTION

Brain-derived neurotrophic factor (BDNF) is a highly conserved and nonglycosylated polypeptide primarily isolated from pig brain that serves as a neurotrophic factor via autocrine and/or paracrine mechanisms (1). Tropomyosin receptor kinase B (TrkB), which is located predominantly (but not exclusively) in peripheral tissues and the central nervous system, has been identified as an endogenous receptor for BDNF (2). Acting through this receptor, BDNF plays regulatory roles in a variety of physiological functions, including survival, differentiation, and synaptic plasticity of neurons (3, 4). In addition, in vitro studies have suggested that BDNF might participate in the nociceptive (pain) signaling of primary sensory neurons (5), in which TrkB receptors are abundantly expressed (6). Moreover, compelling evidence has demonstrated that increased BDNF production is essential for the development of neuropathic pain and inflammatory pain (7, 8), with the application of BDNF inducing pronociceptive effects (8). Nevertheless, direct proof and the detailed molecular mechanism of BDNF’s hyperalgesic action remain elusive.

Mammalian voltage-gated Ca2+ channels are categorized into three families: the Cav1 family, which encodes L-type channels; the Cav2 family, which encodes N-, P/Q-, and R-type channels; and the Cav3 family, which encodes T-type channels (911). Among them, the low voltage–activated (LVA) T-type channels (Cav3 family) have the unique ability to modulate neuronal excitability (12) with small depolarization and play regulatory roles in low-threshold exocytosis (13). The cloning of the α1 subunits of T-type channels revealed the existence of three different subtypes, termed Cav3.1, Cav3.2, and Cav3.3, with specific expression patterns and distinct pharmacological profiles in the brain and peripheral organs (1416). Aberrant function and/or expression of T-type channels is associated with pathological conditions such as slow-wave sleep, seizure susceptibility, absence of epilepsy, and pain perception (17, 18). In the primary afferent pain pathway, stimulation of T-type channels not only shapes the firing patterns and changes the action potential (AP) frequency of pain-sensing neurons but also plays a regulatory role in quantal neurotransmitter release at dorsal horn synapses (19, 20). Accumulating evidence from genetic (21) and pharmacological analyses (22, 23) has indicated huge therapeutic potential of targeting T-type channels for pain management (17, 24, 25).

Here, we examined the effect of BDNF on T-type channels in small-sized trigeminal ganglion (TG) neurons (<30 μm) and uncovered underlying molecular mechanisms associated with the nociceptive effects of BDNF. Our results indicate that, through TrkB, BDNF stimulates T-type channels through phosphatidylinositol 3-kinase (PI3K) and p38-dependent protein kinase A (PKA) signaling. This cascade contributes to TG neuronal hyperexcitability and pain hypersensitivity in rats.

RESULTS

BDNF enhances T-type channel currents in a dose-dependent manner

Whole-cell recordings from dissociated TG neurons of adult rats are usually used to investigate peripheral nociceptive mechanisms in vitro (26, 27). In this study, we limited our patch-clamp experiments to small-diameter neurons (<30 μm) because most of these cells are primarily involved in nociceptive processing/signaling (27, 28). To isolate T-type channel currents (hereafter, T-currents), recordings were carried out in an extracellular bath solution containing 5 μM nifedipine (a blocker of L-type channels) and 0.2 μM ω-conotoxin-MVIIC (isolated from Conus magus, a blocker of N-type and P/Q-type channels). Whole-cell currents were obtained by holding the neuron at −110 mV and stepping to −40 mV. The addition of 100 μM NiCl2 inhibited the peak amplitude of the remaining current by ~82.3% (Fig. 1A), indicating effective T-current separation. Application of BDNF [50 ng/ml (1.9 nM)] to TG neurons markedly enhanced the T-currents (Fig. 1, B and C), and the current amplitude partially recovered within 3 min after BDNF was washed out (Fig. 1, C and D). Further determinations of the response indicated that BDNF enhanced T-currents in a dose-dependent manner (Fig. 1E). The relationship between the dose of BDNF and the degree of increase was described by a sigmoidal Hill equation, with a median effective dose (ED50) of 29.4 ng/ml (1.1 nM) and a Hill coefficient of 0.89 (Fig. 1E). Moreover, the examination of the biophysical properties of the T-type channels indicated that BDNF, at 50 ng/ml (used hereafter unless specified otherwise), had no notable effect on the voltage dependence of the activation properties (V50,act) but altered the potential of steady-state inactivation, and induced a significant depolarizing shift by ~9.3 mV in the inactivation potentials of T-type channels (V50,inact) (Fig. 1, F to H). Although the biophysical mechanisms require further investigation in detail, the positive change of V50,inact implies that the decreased proportion of T-type channels in the steady-state inactivation might be one factor responsible for the BDNF-induced increase in T-currents.

Fig. 1 BDNF enhances T-currents in TG neurons.

(A) Representative current traces (left) and analysis (right) of the effects of NiCl2 (Ni2+, 100 μM) on T-currents. Insets, stimulation protocol. Data are means ± SEM; n = 7 cells from three rats. **P < 0.01, paired t test. (B to D) Time course of T-current changes (B), with representative traces [(C), from points a, b, and c in (B)], and quantified T-current density (D) under control conditions, during exposure to BDNF [50 ng/ml (1.9 nM)], and during washout. Data are means ± SEM; n = 8 cells from three rats. **P < 0.01, paired t test. (E) Dose-response curve of T-currents in TG neurons cultured with BDNF. The curved line represents the best-fit line based on nonlinear regression modeling using the Hill equation. Data are means ± SEM; n indicated above each. (F to H) Effect of BDNF (50 ng/ml) on features of T-type channel activation (F) and steady-state inactivation (G), quantified for respective effects on half-maximum potentials for activation/inactivation (V50) (H). Insets, stimulation protocol. Traces are representative, and graph (H) shows means ± SEM; n = 12 cells from four rats. *P < 0.05, paired t test.

TrkB mediates the BDNF-induced T-type channel response

TrkB has been identified as the functional receptor for BDNF in vivo (2, 3). We thus assessed the participation of TrkB in T-current responses to BDNF. Immunoblot analysis of TG protein lysates revealed that TrkB was endogenously expressed in rats (Fig. 2A and fig. S1). This native expression profile of the TrkB protein was further confirmed by immunohistochemical staining of TG tissue sections (Fig. 2B). Small-sized unmyelinated sensory neurons are classified into peptidergic neurons expressing the calcitonin gene-related peptide (CGRP) and nonpeptidergic, isolectin B4-binding (IB4+) nociceptive neurons. We performed double staining for TrkB in TG sections and found that TrkB was colocalized with CGRP and IB4 but exhibited little coimmunolabeling with the 200-kDa neurofilament protein (NF200), a typical marker of myelinated afferents (Fig. 2B). Next, we examined the relative involvement of TrkB in the BDNF-mediated T-current increase. ANA-12, a potent TrkB antagonist, alone had no effect on T-currents (Fig. 2C), whereas pretreating cells with ANA-12 nearly abrogated the BDNF-induced T-current response (Fig. 2C). To further test the BDNF-induced T-type channel response through TrkB-dependent signaling, we examined whether another activator of TrkB might mimic the BDNF response. Application of 7,8-dihydroxyflavone, a small-molecule specific TrkB agonist, significantly enhanced T-currents in rat TG neurons (Fig. 2D).

Fig. 2 TrkB mediates the BDNF-induced increase in T-currents.

(A) Western blotting analysis of TrkB abundance in rat TG cells. GAPDH, glyceraldehyde-3-phosphate dehydrogenase (loading control). Blots are representative of ≥3 experiments. (B) Double-label immunofluorescent staining of TrkB (red) with isolectin B4 (IB4), calcitonin gene-related peptide (CGRP), or neurofilament 200 (NF200), all green, in rat TG sections. Colocalization quantified from 872 cells from three experiments (right). Scale bar, 35 μm. (C and D) Time course of T-current changes (represented, left; quantified, right) in TG neurons after either (C) application of BDNF (50 ng/ml) alone or pretreated with ANA-12 (10 nM), or (D) application of 7,8-dihydroxyflavone (7,8-DHF, 0.5 μM). a and b indicate the points used for the representative current traces, inset. Data are means ± SEM; n = 8 (C) or 9 (D) cells from three rats. *P < 0.05 and **P < 0.01, paired t test.

The TrkB-induced T-current increase is mediated by PI3K but not AKT

We further investigated the underlying mechanism of the TrkB-induced T-current increase. Evidence has suggested a pivotal role of PI3K-AKT signaling cascades in TrkB-mediated biological activities (29). Thus, we first examined whether the BDNF-induced response in T-type channels was dependent on PI3K. Pretreating TG neurons with LY294002, a potent PI3K inhibitor, abrogated the BDNF-induced T-current increase; by contrast, LY303511, a structurally inactive analog of LY294002, elicited no such effects (Fig. 3, A to C), indicating PI3K signaling involvement. AKT is a major downstream effector of PI3K signaling. We therefore determined whether the BDNF response also required AKT (fig. S2). The level of phosphorylated AKT (p-AKT) was significantly increased after treatment with BDNF, whereas the level of total AKT (t-AKT) remained unchanged (Fig. 3D). Pretreatment of TG cells with the AKT inhibitor III attenuated the BDNF-induced stimulatory effects (Fig. 3D). The involvement of AKT in the BDNF-induced T-type channel response was further determined. Unexpectedly, in neurons pretreated with the AKT inhibitor III, BDNF still caused a robust increase in the peak amplitude of T-currents (Fig. 3E), demonstrating that the BDNF-induced T-current response was independent of AKT. To support this hypothesis, we used an adenoviral-based short hairpin RNA (shRNA) knockdown approach to examine the effect of BDNF on T-currents in AKT-silenced TG neurons. Compared with the robust expression of AKT protein in cells transduced with a control shRNA, immunoblot analysis indicated that the level of AKT was markedly decreased in AKT-knockdown cells (Fig. 3F and fig. S3); however, AKT knockdown had no effect on the BDNF-induced increase in T-currents (Fig. 3G). Together, these findings demonstrated that the BDNF-induced T-current increase in rat TG neurons required PI3K but was independent of AKT.

Fig. 3 The BDNF-induced T-current increase requires PI3K.

(A and B) Time course of changes in T-current amplitude induced by BDNF (50 ng/ml) in TG neurons pretreated with either LY294002 (20 μM A) or LY303511 (20 μM B). a and b indicate points used for representative current traces, inset. Data (C) are means ± SEM; n = 10 cells from four rats and eight cells from three rats, respectively. **P < 0.01, paired t test. (D) Western blotting assessment of AKT activation in TG cells after BDNF (50 ng/ml) alone and with pretreatment with AKT inhibitor III (AKT III) (10 μM). Blots were representative of at least three experiments. *P < 0.05, unpaired t test. (E) As described in (A) to (C) with AKT inhibitor III (10 μM; n = 9 cells from three rats). (F) Western blotting analysis of AKT abundance in negative control shRNA (NC-shRNA) or AKT-shRNA transfected TG cells. Blots (left) are representative of the means ± SEM (right) from at least three experiments. **P < 0.01, unpaired t test. (G) Representative current traces (left) and analysis (right) of the effect of AKT knockdown on BDNF (50 ng/ml)–induced T-currents. Data are means ± SEM, n = 11 to 15 cells from three to four rats. *P < 0.05, one-way ANOVA.

The TrkB-induced T-current increase is mediated by p38-dependent PKA signaling

PKC acts downstream of TrkB (30) and plays critical roles in T-type channel regulation (31). We thus determined the participation of PKC in the BDNF-mediated T-current response. Pretreating TG neurons with 50 nM calphostin C, which inhibits the “classical” (or “conventional”) and “novel” PKC isoenzymes, did not affect the BDNF-mediated T-current increase (Fig. 4A). The calphostin C used in the assay was effective, given that it abrogated the ability of phorbol 12-myristate 13-acetate (PMA) to increase T-currents (Fig. 4B). Accumulating evidence has also suggested a regulatory contribution of PKA (3, 31, 32). Further determination of PKA activity in TG cells showed that BDNF significantly enhanced PKA activity, and the effect was abrogated by pretreating TG cells with either ANA-12 or LY294002 but not by pretreating cells with AKT inhibitor III (Fig. 4C). Dialyzing small-sized neurons with the PKA peptide inhibitor PKI 6–22 prevented the BDNF-induced increase in T-currents (Fig. 4, D and E). Similar results were observed in cells pretreated with another PKA inhibitor, KT-5720 (Fig. 4E), supporting the participation of PKA in the BDNF-induced T-current response. Mitogen-activated protein kinase (MAPK) signaling has been shown to mediate cross-talk between PI3K and PKA signaling (33). Consequently, the involvement of the MAPK family in the BDNF-induced T-type channel response was examined (fig. S4). Immunoblot analysis indicated that exposure of TG cells to BDNF significantly increased the level of phosphorylated p38 (p-p38), whereas total p38, phosphorylated extracellular signal–regulated kinase, and phosphorylated c-Jun N-terminal kinase were not affected (Fig. 4F). Pretreating cells with LY294002, but not KT-5720, eliminated this BDNF-induced p38 activation (Fig. 4G and fig. S5). Further determinations of PKA activity suggested that PKA might be downstream of p38 signaling. BDNF significantly enhanced PKA activity, and preincubation of cells with the p38 MAPK inhibitor SB203580 abolished this effect nearly completely (Fig. 4H). Immunoblot analysis demonstrated that exposure of TG cells to BDNF significantly increased the phosphorylated PKA (p-PKA, PKA activation), which was abrogated by pretreatment with SB203580 (Fig. 4I and fig. S6). SB203580 also abrogated the T-current response induced by BDNF; by contrast, the negative control compound SB202474, which is structurally related to SB203580 but does not inhibit p38 MAPK, elicited no such effects (Fig. 4, J to L). Collectively, these results indicate that p38 MAPK mediates signaling between PI3K and PKA in the BDNF-induced T-type channel response.

Fig. 4 TrkB-mediated stimulation of T-type channels requires the p38-dependent PKA signaling.

(A) Time course of T-current changes (left) and analysis (right) indicating the effects of BDNF (50 ng/ml) on T-currents in cells pretreated with calphostin C (50 nM). n = 9 cells from three rats. a and b on the plot indicate which points were used for sample traces, inset. (B) Representative current traces (left) and analysis (right) of PMA (20 μM)–induced T-current response in TG cells pretreated with control or calphostin C (50 nM). n = 10 cells from three rats. **P < 0.01, paired t test. (C) Effect of pretreating TG cells with ANA-12 (10 nM), LY294002 (20 μM), or AKT inhibitor III (10 μM) on the BDNF (50 ng/ml)–induced increase in PKA activity, measured by the fluorescence-based PepTag Non-Radioactive Protein Kinase Assay system, in which the phosphorylation of a PKA substrate was detected. Data are means ± SEM from triplicate replicates from each of three experiments. *P < 0.05 and **P < 0.01, one-way ANOVA. (D) As in (A) but pretreated with PKI 6–22 (1 μM). n = 8 cells from three rats. (E) Effects of pretreatment of TG neurons with PKI 6–22 or KT-5720 (both 1 μM; n = 7 cells from three rats) on the BDNF (50 ng/ml)–induced T-type channel response. **P < 0.01, paired t test. (F and G) Western blot analysis of phosphorylated p38 (p-p38), phosphorylated extracellular signal–regulated kinase (p-ERK), and phosphorylated c-Jun N-terminal kinase (p-JNK) in TG cells upon BDNF application (50 ng/ml) alone (F) or with LY294002 (20 μM) or KT-5720 (1 μM) pretreatment (G). Blots are representative of at least three separate experiments. *P < 0.05 and **P < 0.01, unpaired t test. (H and I) Graph of PKA activity (H), representative blots (I; left, n = 3), and analysis (I, right) of PKA activation (p-PKA) in TG cells upon application of BDNF (50 ng/ml) with or without pretreatment with SB203580 (10 μM). *P < 0.05 and **P < 0.01, one-way ANOVA. (J to L) Representative traces (J and K) and analysis (L) of T-current changes induced by BDNF (50 ng/ml) in TG cells pretreated with SB203580 or SB202474 (10 μM; n = 8 and 9 cells, respectively, from four rats). a and b as in (A). All data are means ± SEM. **P < 0.01, paired t test.

BDNF induces excitability in TG neurons

We further determined the role of the BDNF-induced T-type channel response in regulating TG neuronal excitability. Initial analysis showed that the application of BDNF did not affect the whole-cell currents through voltage-gated Na+ channels (Nav) (Fig. 5A). It has been shown that the application of BDNF facilitated L-type channels in nucleus tractus solitarius neurons (34). In addition, treatment with BDNF enhanced N-type channel current amplitude in human striatal precursor cells isolated from ganglionic eminence (35). Therefore, we applied 5 μM nifedipine to block L-type channels and 0.2 μM ω-conotoxin-MVIIC to block N-type and P/Q-type channels in the external solution, and found that the application of BDNF did not affect the remaining high voltage–activated Ca2+ channel currents (R-type) in TG neurons (Fig. 5B). Furthermore, application of BDNF to TG neurons decreased the peak amplitude of the sustained delayed rectifier K+ channel currents (IDR; Fig. 5, C and D), whereas the transient outward K+ channel currents (A-type currents or IA) were not significantly affected (Fig. 5, C and D). In a bath solution (extracellular) containing 0.2 μM ω-conotoxin-MVIIC, 5 μM nifedipine, and 5 mM tetraethylammonium (TEA) (the IDR blocker), BDNF significantly increased the AP firing rate (Fig. 5E), whereas other properties of membrane excitability, including the AP threshold and input resistance, remained unaffected. Pretreating TG neurons with TrkB antagonist (ANA-12) nearly abrogated the BDNF-induced TG neuronal hyperexcitability (Fig. 5F). Similar results were observed in TG cells pretreated with PI3K inhibitor (LY294002, Fig. 5F) or PKA inhibitor (KT-5720, Fig. 5F). Furthermore, pretreating TG neurons with p38 MAPK inhibitor (SB203580) abrogated the neuronal hyperexcitability induced by BDNF (Fig. 5F). By contrast, the negative control compound (SB202474) elicited no such effects (Fig. 5F). Next, we determined the role of the BDNF-induced T-type channel response in regulating neuronal excitability under physiological recording conditions. TG neurons were incubated with a bath solution containing IDR blocker at 37°C for 10 min before taking the electrophysiological recordings. Under these conditions, application of BDNF still induced a significant increase in the AP firing rate in TG neurons (Fig. 5G). Moreover, we examined whether the BDNF-induced TG neuronal hyperexcitability was dependent on its T-type channel stimulation and found that pretreating TG neurons with NiCl2 (100 μM) prevented the BDNF-induced neuronal hyperexcitability (Fig. 5H). To further determine the involvement of Cav3.2 in the response mediated by BDNF/TrkB, we used small interfering RNA (siRNA)–mediated knockdown approach to silence Cav3.2 expression in rat TGs (Fig. 5I and fig. S7). Our results indicated that intra-TG injection of the 5′-cholesteryl–modified and 6-FAM–modified Cav3.2-siRNA resulted in a significant down-regulation of Cav3.2 protein expression in rat TGs (Fig. 5I). Knockdown of the Cav3.2 almost completely abolished the BDNF-induced increase in the rate of AP firing in small-sized TG neurons (Fig. 5J).

Fig. 5 BDNF induces TG neuronal hyperexcitability.

(A and B) Representative current traces (left) and analysis (right) showing the effects of BDNF (50 ng/ml) on voltage-gated Na+ channel (Nav) currents (A) and high voltage–activated (HVA) Ca2+ channel currents (B) in TG neurons. Neurons were held at −60 mV and stepped to 0 mV to elicit the currents. Graphs show means ± SEM; n = 8 (A) or 9 (B) cells from three rats. (C and D) As in (A) and (B) on transient outward A-type K+ currents (IA; C, left) and the sustained delayed rectifier K+ current (IDR; C, right). Data (D) are means ± SEM; n = 12 cells from four rats. *P < 0.05, paired t test. (E and F) Effect of BDNF (50 ng/ml), either alone (E; n = 17 cells from four rats) or after pretreatment with ANA-12 (10 nM, n = 9 cells from three rats), LY294002 (20 μM, n = 8 cells from three rats), SB203580 (10 μM, n = 9 cells from four rats), KT-5720 (1 μM, n = 9 cells from three rats), or SB202474 (10 μM, n = 11 cells from three rats) (F), on the action potential (AP) firing rate (spike frequency) of TG neurons. Insets (E, top) indicate the current injection protocol; resting membrane potential of the neurons was maintained at −70 mV. Graphs are means ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001, paired t test. (G and H) As described for (E) when incubated for 10 min at 37°C (G, n = 15 cells from four rats) or pretreated with NiCl2 (Ni2+, 100 μM; H, n = 12 cells from four rats). Graphs are means ± SEM. **P < 0.01, paired t test. (I and J) AP firing rate in TG cells after Cav3.2 knockdown (J), confirmed by Western blotting of TG tissue homogenates after intra-TG injection (I). Data are means ± SEM; n ≥ 3 experiments (I); n = 12 cells from five rats (J). **P < 0.01 versus NC-siRNA, unpaired t test.

T-type channels contribute to BDNF-induced pain hypersensitivity

To further determine the functional implications of BDNF at the behavioral level, we investigated whether BDNF affects pain sensitivity in rats. The escape threshold was determined by measuring the mechanical threshold in response to von Frey filaments. Intra-TG injection of BDNF induced a marked decrease in the escape threshold (Fig. 6A); this effect recovered 3 hours after BDNF application. The BDNF-mediated mechanical pain hypersensitivity was abrogated by prior injection with ANA-12 (Fig. 6B). The participation of T-type channels in the BDNF-mediated mechanical hypersensitivity was further examined using a specific T-type channel blocker, TTA-P2 (Fig. 6B). Whereas intra-TG application of TTA-P2 did not affect the escape threshold, pretreatment with TTA-P2 significantly attenuated the mechanical hypersensitivity induced by BDNF (Fig. 6B). Furthermore, the possible role of BDNF in chronic inflammatory pain was determined. Rats exhibited significant mechanical allodynia to complete Freund’s adjuvant (CFA; Fig. 6C). The lowered escape threshold started 12 hours after subcutaneous injection of CFA, peaked at day 2, and was sustained for at least 7 days (Fig. 6C). Moreover, a significant concomitant increase in TrkB protein abundance was observed in TGs at day 2, the peak time point of mechanical allodynia (Fig. 6D and fig. S8). Therefore, the effects of TrkB blockade on CFA-induced mechanical allodynia were measured. We observed a significant alleviation of mechanical allodynia by ANA-12 in rats treated with CFA, and this effect lasted for at least 2 hours (Fig. 6E). The role of the T-type channel as a pivotal target was further verified in a rat model of CFA-induced chronic inflammatory pain. Intra-TG injection of the T-type channel blocker TTA-P2 2 days after CFA treatment effectively attenuated CFA-induced chronic pain for at least 2 hours (Fig. 6F). Although intra-TG injection of ANA-12 also alleviated the mechanical hypersensitivity induced by CFA (Fig. 6F), the hourly assessment of pain sensitivity indicated that the ANA-12 application did not elicit any additive effects to that of TTA-P2 in CFA-treated rats (Fig. 6F). To further validate the Cav3.2 T-type channel as an important cellular target for pain-alleviating effects of BDNF signaling in chronic inflammatory pain, we administered intra-TG injection of siRNA specific for the Cav3.2 T-type channel. Administration of Cav3.2-siRNA resulted in a significant down-regulation of Cav3.2 protein abundance in rat TGs (Fig. 5I). The Cav3.2-siRNA–treated CFA rats exhibited a significant decrease in mechanical allodynia as compared with controls (Fig. 6G). ANA-12 was then administered and pain behavioral phenotype was assessed on an hourly basis. Treatment with ANA-12 in Cav3.2-siRNA–treated CFA rats had no further effect on mechanical sensitivity, as evidenced by escape threshold recordings that remained stable throughout the 3-hour period after ANA-12 administration (Fig. 6G), suggesting that ANA-12 and Cav3.2-siRNA likely share the same cellular signaling pathway in vivo. In addition, the control siRNA-treated CFA rats responded to ANA-12 similarly to those that did not receive siRNA (Fig. 6E), in that ANA-12 induced a significant increase in escape threshold. These findings together demonstrate that Cav3.2 T-type channels are involved in TrkB-mediated nociceptive actions in CFA-induced inflammatory pain behavior.

Fig. 6 Involvement of peripheral TrkB activation in pain hypersensitivity.

(A) Mechanical hypersensitivity, determined by the escape threshold (ET), after intra-TG injection of BDNF (5 nmol) or vehicle. *P < 0.05 versus vehicle, one-way ANOVA. (B) Effect of pretreatment with ANA-12 (2 nmol; intra-TG) or TTA-P2 (1 nmol) on BDNF-induced mechanical hypersensitivity. *P < 0.05, one-way ANOVA. (C) Effect of CFA injection (20 μl) or normal saline (NS; 20 μl) on ET. *P < 0.05 versus corresponding NS, one-way ANOVA. BL, baseline. (D) TrkB abundance in CFA-treated or control rats, age- and sex-matched. Blots are representative of at least three experiments. *P < 0.05, unpaired t test. (E) Effect of intra-TG injection of vehicle or ANA-12 (2 nmol, arrow) 2 days after CFA on the ET in rats. *P < 0.05 versus corresponding vehicle, one-way ANOVA. (F) Effect of blockade of T-type channels by intra-TG injection of TTA-P2 (1 nmol) on the ANA-12 (2 nmol)–mediated amelioration of mechanical allodynia in CFA rats. $P < 0.05 versus CFA 2 days; **P < 0.01 versus corresponding vehicle; #P < 0.05 versus vehicle 1 hour; two-way ANOVA. (G) Effect of siRNA-mediated knockdown of Cav3.2 versus control (NC-siRNA) on ANA-12 (2 nmol, intra-TG injection; arrow)–induced alleviation of mechanical allodynia (ET) in CFA rats. +P < 0.05 versus CFA 2 days; *P < 0.05 and **P < 0.01 versus corresponding NC-siRNA; #P < 0.05 versus 0 hours (“siRNA”) NC-siRNA; two-way ANOVA. For (A) to (C) and (E) to (G), n = at least seven rats for each group.

DISCUSSION

Our findings provide mechanistic insight into a critically functional role of BDNF in regulating T-type channels in TG neurons, underlying sensory signaling and pain. These findings demonstrated that this BDNF-induced T-current response is mediated by TrkB coupled to PI3K-p38-PKA signaling. This signaling cascade contributes to TG neuronal hyperexcitability and the nociceptive activities of BDNF in rats (Fig. 7).

Fig. 7 Proposed mechanisms of TrkB signaling on T-type channels.

BDNF acts through the receptor TrkB and subsequently activates PI3K. Several pathways may stimulate T-type channels, including those of PKA, PKC, and MAPK (p38, ERK, and JNK). Data from this study implicate the involvement of the p38-dependent PKA signaling pathway in the stimulation of T-type channels rather than the AKT and PKC pathway. PI3K-dependent p38 signaling may activate PKA, which then phosphorylates Cav3 channels (mainly Cav3.2) to regulate T-currents.

In contrast to the known classical PI3K-AKT signaling pathways, the BDNF-induced T-current increase is dependent on PI3K but is not mediated by AKT phosphorylation. PI3K signaling enhances PKC activity to modulate various molecules, including voltage-gated Ca2+ channels (36). Accumulated evidence suggests divergent roles of PKC in regulating T-type channels (31, 32). For example, T-currents recorded from neonatal rat cardiomyocytes increased in response to the PKC activator PMA (37). Similarly, in mouse dorsal root ganglion (DRG) neurons, PKCα inhibition abolished the increase in T-currents stimulated by insulin-like growth factor 1 (38). By contrast, T-current inhibition by the M3 muscarinic receptor in DRG was prevented by antagonists of PKC (39). Nevertheless, the possibility that the BDNF-induced increase in T-currents was due to PI3K-PKC activation can be ruled out in this study, suggesting that other non-PKC–mediated mechanisms are involved. PKA is a critical regulator of T-type channels (32), and we found that the p38-dependent PKA does participate in the TrkB-PI3K pathway in rat TG neurons. These findings are consistent with previous reports that activation of PKA by serotonin type 7 receptor increases T-currents in rat glomerulosa cells (40). However, in retinal ganglion cells, T-current inhibition induced by the cannabinoid CB1 and CB2 receptors was prevented by PKA inhibitors (41). Moreover, biphasic effects or even no effect of PKA activation on T-currents were observed in a previous study (32). For instance, PKA activation induces an increase in Cav3.2 T-currents in mammalian heterologous expression systems at 37°C but not at room temperature (32, 42). The heterogeneous effects of PKA on T-type channels suggest that different parameters are involved in the differential modulation of T-type channel activity by PKA. The tissue- or cell-specific expression and/or activation of endogenous PKA isoforms leads to heterogeneity of PKA-dependent signal transduction pathways modulating ion channels (31, 32). Moreover, the extensive alternative splicing isoforms of Cav3 subunits and/or relevant accessory subunits contribute to the diversity of PKA-mediated channel activities (10, 43, 44). The three subunits of the mammalian Cav3 family have been shown to have different properties and regulation (16, 45). The possibility that PKA phosphorylation by an intermediate protein is involved in the observed BDNF-induced response cannot be excluded. Phosphorylation is one of the major mechanisms regulating the activity of ion channels. Although further investigation is clearly necessary, phosphorylation of the Cav3.2 T-type channel might contribute to the reversibility of BDNF response in rat TG neurons. Consistently, by using a commercially available PKA catalytic subunit that is constitutively active and that allows in vitro kinase assays of immunoprecipitated hyaluronic acid–tagged Cav3.2 channels, previous studies have nicely indicated that PKA induced direct phosphorylation of Cav3.2 (42). Moreover, it has been shown that Cav3.2 channels are highly phosphorylated in the mammalian brain, which established phosphorylation as an important mechanism involved in the dynamic regulation of Cav3.2 channel gating properties (46).

The Cav3.2 channel isoform is the main LVA T-type channel expressed in small-sized peripheral sensory neurons and implicated in nociceptive signaling (47, 48). Both pharmacological and genetic studies have firmly established a prominent role of the Cav3.2 channel in amplifying nociceptive signals in the periphery (47, 49, 50) and in contributing to central sensitization in the spinal dorsal horn (20). Recent studies have shown that modulation of peripheral Cav3.2 channels influences somatic and visceral nociceptive inputs and results in pain relief in several animal models of neuropathic pain (24, 25, 51). Here, we observed that the nociceptive effects of BDNF are mediated, at least in part, through stimulation of TrkB-mediated Cav3.2 T-type channels. Our present results are in line with previous in vivo studies that show that application of BDNF induces pain hypersensitivity (5, 7, 52). Contradictory data claiming an antinociceptive effect are also found in the literature. Lumbar subarachnoid transplantation of BDNF-secreting neurons in the spinal cord reversed cold and tactile allodynia and hyperalgesia in a unilateral chronic constriction injury (CCI) of the rat sciatic nerve (53). The central rather than peripheral actions of BDNF may count for the analgesic effect of BDNF, since increased BDNF production in peripheral tissues has been shown to sensitize primary afferent neurons and facilitate hypersensitivity to pain (7, 8). In addition, intraspecies differences (52, 54) and differences in the age of animals used in the various studies may be a factor in this variability (55, 56). Nevertheless, other potential mechanisms of BDNF other than Cav3.2 channels contributing to the chronic inflammatory pain have also been reported. It has been shown that nociceptor-derived BDNF regulates the excitability of spinal neurons and plays a crucial role in inflammatory pain through the abolition of N-methyl-d-aspartate acid (NMDA) receptor and ERK1 and ERK2 activation (57, 58). Further, the chronic restraint stress induced abnormal pain sensitivity (59), highlighting the key role of BDNF-mammalian target of rapamycin (mTOR) signaling pathway. However, nonspecific channel/receptor (e.g., NMDA or mTOR) or kinase (such as ERK1/2) inhibitors have been tested with little success in the treatment of a variety of chronic pain–related conditions due to their lack of selectivity and limitations of their pharmacokinetics. Therefore, targeting a specific isoform, such as selective inhibition of Cav3.2, could potentially reduce the risk of side effects of analgesics, although further investigation is clearly necessary.

In conclusion, we dissected the molecular components and signaling pathway underlying the effect of BDNF on T-type channels in TG neurons. We showed that activation of TrkB enhances T-currents through the PI3K-p38-PKA signaling cascade and results in pain hypersensitivity. As the expression of p38 was up-regulated in ipsilateral TGs in orofacial inflammatory pain model induced by CFA (60) and that pharmacological blockade of p38 activation reduced the CCI-induced thermal hyperalgesia (61), the knowledge of TrkB-mediated p38 pathway in TG neurons in the present study may also pave the way for this process to become a tool for developing pain therapeutics in clinical applications.

MATERIALS AND METHODS

Preparation of TG neurons

Animal care and use complied with the National Institutes of Health guidelines for the care and use of experimental animals. All experimental and surgical protocols were approved by the Institutional Animal Care and Use Committee of Soochow University, and every effort was made to minimize both the number of animals used and their suffering. Acute dissociation of rat (Sprague-Dawley, 200 to 250 g, either sex) TG neurons was performed as previously described (62, 63). Briefly, the rats were decapitated under anesthesia, and TGs were dissected out bilaterally. After removing connective tissue and trimming, the cleaned TGs were treated with collagenase II (2.25 mg/ml, Worthington Biochemical) and trypsin (1.5 mg/ml, Sigma-Aldrich) in Dulbecco’s modified Eagle’s medium saturated with CO2/O2 mixed gas at 37°C for 35 min. The ganglia were then dissociated mechanically by triturating with sterile fire-polished Pasteur pipettes (Thermo Fisher Scientific). After centrifugation, the pellet was resuspended and then transferred to glass coverslips coated with Matrigel (BD Biosciences). The isolated TG neurons were used for recording between 3 and 8 hours after plating. In the knockdown experiments, cells transduced with adenovirus-based shRNA were replated at 48 hours after infection.

Whole-cell patch-clamp recording

In this study, we sorted adult rat TG neurons into groups on the basis of soma diameter [small-sized (<30 μm) and medium-sized (30 to 45 μm)] and restricted the electrophysiological recordings to small-sized neurons, most of which are nociceptors (22, 23, 38). Whole-cell recordings were carried out at room temperature (22° to 24°C) as previously described (39, 62, 64). Coverslips plated with TG neurons were transferred to a Warner recording chamber (Warner Instruments) and perfused with external solution for T-type channel current recording containing 5 mM BaCl2, 140 mM TEA-Cl, 5 mM CsCl, 0.5 mM MgCl2, 5.5 mM d-glucose, 10 mM Hepes (pH 7.3) (with TEA-OH), and 305 mOsm. The electrodes were pulled from borosilicate glass (Sutter Instruments) and had 3 to 4 megohm resistance when filled with the internal solution composed of 110 mM CsCl, 4 mM Mg–adenosine 5′-triphosphate (ATP), 0.3 mM Na2–guanosine 5′-triphosphate (GTP), 25 mM Hepes, 10 mM EGTA (pH 7.4) (with CsOH), and 295 mOsm. The whole-cell recordings were conducted using a MultiClamp 700B patch-clamp amplifier (Molecular Devices), and the output was digitized with a Digidata 1322A converter (Molecular Devices). Data were low-pass–filtered at 2 kHz, digitized at 20 kHz, and stored on a computer equipped with pClamp 10.2 (Molecular Devices). The series resistance was adequately compensated by at least 80%. To enable recordings of isolated T-currents, recordings were carried out in extracellular bath solution containing 5 μM nifedipine (L-type channel blocker) and 0.2 μM ω-conotoxin MVIIC (N-type and P/Q-type channel blocker). In experiments for whole-cell voltage-gated potassium channel (Kv) current recordings, the extracellular solution contained 140 mM choline-Cl, 1 mM MgCl2, 5 mM KCl, 0.03 mM CaCl2, 10 mM glucose, 10 mM Hepes (pH 7.4) (with KOH), and 310 mOsm. The intracellular pipette solution contained 140 mM KCl, 0.5 mM CaCl2, 0.3 mM Na2-GTP, 3 mM Mg-ATP, 1 mM MgCl2, 5 mM EGTA, 10 mM Hepes (pH 7.4) (with KOH), and 295 mOsm. The two kinetically different Kv currents in small-sized TG neurons were separated by the two-step voltage protocol as previously described (62, 65). In current clamp and whole-cell Nav current recording experiments, the extracellular solution contained 125 mM NaCl, 2 mM KCl, 2 mM CaCl2, 2 mM MgCl2, 30 mM d-glucose, 25 mM Hepes (pH 7.4) (with NaOH), and 305 mOsm. The intracellular pipette solution contained 110 mM KCl, 10 mM NaCl, 25 mM Hepes, 0.3 mM Na2-GTP, 4 mM Mg-ATP, 2 mM EGTA (pH 7.3) (with KOH), and 295 mOsm. To perform current clamp recordings, we added 5 mM TEA (the delayed rectifier K+ channel blocker), 0.2 μM ω-conotoxin-MVIIC, and 5 μM nifedipine to the external solution. Using a pressure-pulsed microinjector (Picopump PV820, World Precision Instruments), BDNF was puff-applied through a glass pipette, the tip of which was placed 15 to 25 μm from the soma of TG neurons. In cells dialyzed with compounds, the resistance of the patch electrodes ranged from 2 to 3.5 megohms for intracellular delivery of agents, and recordings were started at least 5 min after breaking into the whole-cell configuration.

Western blot analysis

Aliquots containing 20 μg of protein were blotted as described previously (38, 66). In brief, protein samples were separated by 10% SDS–polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes (Bio-Rad). After blocking with 5% skimmed milk at room temperature for 1 hour, the blots were probed overnight at 4°C with the following primary antibodies: mouse anti-TrkB (1:500 dilution, BD Biosciences; catalog no. 610101), mouse anti-Cav3.2 (1:500, Novus Biologicals; catalog no. NBP1-22444), rabbit anti–p-AKT (1:1000 dilution, Abcam; catalog no. ab131443), rabbit anti-AKT (1:1000 dilution, Abcam; catalog no. ab179463), mouse anti–p-p38 MAPK (1:500 dilution, Cell Signaling Technology; catalog no. 9216), mouse anti-p38 MAPK (1:1000 dilution, Cell Signaling Technology; catalog no. 9228), rabbit anti–p-p44/42 MAPK (ERK1/2, 1:600 dilution, Cell Signaling Technology; catalog no. 4376), rabbit anti-ERK1/2 (1:500 dilution, Cell Signaling Technology; catalog no. 9102), rabbit anti–p-SAPK/JNK (1:2000 dilution, Cell Signaling Technology; catalog no. 4668), rabbit anti-SAPK/JNK (1:600 dilution, Cell Signaling Technology; catalog no. 9252), and rabbit anti–glyceraldehyde-3-phosphate dehydrogenase (1:5000 dilution, Sigma-Aldrich; catalog no. SAB2108266). Either anti-mouse immunoglobulin G (IgG) (catalog no. HAF007) or anti-rabbit IgG (catalog no. HAF008) horseradish peroxidase–conjugated secondary antibodies were applied (1:6000 dilution; R&D systems). Protein bands were visualized using SuperSignal West Pico Chemiluminescent substrate (Thermo Fisher Scientific Inc.). Images were captured using a Bio-Rad ChemiDoc XRS system, and the band intensities were quantified by densitometry using Quantity One software (Bio-Rad Laboratories).

Immunohistochemistry

Immunohistochemistry was performed as previously described (36, 38, 63). In brief, animals were deeply anesthetized with isoflurane and subsequently perfused transcardially with phosphate-buffered saline (PBS), followed by a fixative containing 4% paraformaldehyde. Tissue samples were sectioned at a thickness of 15 μm using a Leica CM1950 cryostat (Leica Microsystems Inc., Germany). Subsequently, tissue sections were washed in PBS and blocked for 1 hour in 0.25% Triton X-100 and 5% normal goat serum. After overnight incubation at 4°C with primary antibodies including rabbit anti-TrkB (1:400 dilution, Abcam; catalog no. ab18987), mouse anti-CGRP (1:500 dilution, Abcam; catalog no. ab81887), or mouse anti-NF200 (1:600 dilution, Sigma-Aldrich; catalog no. N2912), the sections were visualized with IB4-fluorescein isothiocyanate (5 μg/ml; Sigma-Aldrich; catalog no. L2895), goat anti-rabbit IgG conjugated to Cy3 (1:300 dilution, Abcam; catalog no. ab97075), or goat anti-mouse IgG conjugated to fluorescein isothiocyanate (1:300 dilution, Abcam; catalog no. ab6785) for 2 hours at room temperature. The slices were examined under an upright fluorescence microscope (104C, Nikon), and images were captured with a CoolSNAP HQ2 charge-coupled device camera (Photometrics). Negative controls were only incubated with the secondary antibody and did not show any specific immunostaining.

Adenovirus-based shRNA infection

Adenovirus-mediated gene silencing was conducted as previously described (36, 38, 66). Three candidate shRNAs targeting AKT (GenBank accession number NM_033230.2) were designed, and the optimal sequence (5′-GCACCTTTATTGGCTACAAGG-3′) was selected. The nonsense sequence used as the negative control (Ad-NC-shRNA) was as follows: 5′-TACTAGCACGAGCTCTGATTG-3′. PAdTrack–cytomegalovirus–green fluorescent protein (GFP) transgenic adenovirus vector carrying AKT shRNA (Ad-AKT-shRNA) recombined with the adenoviral backbone plasmid pAdEasy-1 was packaged by Genechem Co. Ltd. (Shanghai, China). After 48 hours of infection, the efficiency of shRNA knockdown was determined by immunoblot analysis as described above. For patch-clamp analysis, small-sized neurons expressing GFP were subjected to whole-cell recordings.

Measurement of PKA activity

PKA activity in TG homogenates was measured as previously described (66). After stimulation of cells with BDNF for 15 min, the activity of PKA was assayed by enzyme-linked immunosorbent assay using the PepTag Non-Radioactive Protein Kinase Assays (Promega Corporation) following the manufacturer’s instructions. The activity was expressed as RLU−1 (relative light units) per amount of protein. Each sample was tested in triplicate, and the average results are reported.

Escape threshold for mechanical stimulation

Rats were housed under standard conditions (22° ± 1°C, 50 to 70% humidity, 12-hour light/12-hour dark cycle, with ad libitum access to food and water) and were allowed to habituate to laboratory conditions for 1 week before the experiments. The escape threshold to mechanical stimulation was determined by an ascending series of von Frey filaments (Ugo Basile) as described in previous studies (67, 68). Von Frey stimuli were applied to the right side of the buccal pad region. Each von Frey stimulation was applied three times in each series of trials. The escape threshold intensity was determined when the rat strongly moved its head away from at least one of the three stimuli. BDNF, ANA-12, or TTA-P2 was injected into the TG (intra-TG injection) with a 30-gauge needle inserted through the infraorbital foramen, infraorbital canal, and foramen rotundum. The needle tip was positioned in the medial part of the ganglia, and the treatment agent was slowly delivered in a volume of 5 μl. Chronic inflammatory pain was induced by subcutaneous injection of 20 μl of CFA into the right side of the buccal pad. Rats received intra-TG injection of Cav3.2 siRNA (Cav3.2-siRNA, 5 μg) or a scrambled negative control siRNA (NC-siRNA) 2 days after the CFA injection. 5′-Cholesteryl–modified (69) and 6-FAM–modified Cav3.2-siRNA (GenBank accession: NM_153814.2, 5′-GCAGCCAUCC UCGUCAAUAdTdT-3′) and its NC-siRNA (5′-GACCUACGCUCACUCGAUAdTdT-3′) were purchased from RiboBio (Guangzhou, China). siRNA was mixed with polyethyleneimine (PEI, Fermentas Inc.) for 10 min at room temperature before being delivered. PEI was used as a delivery vehicle to prevent degradation and enhance cell membrane penetration of siRNAs (38). This treatment was repeated every 24 hours for 3 days, for a total of three injections. The protein abundance measurement was performed 6 hours after the last siRNA injection. This protocol was based on previous studies demonstrating an efficient knockdown of γ-aminobutyric acid type A receptor α6 subunit (70) and T-type channels in sensory neurons in vivo (38, 47).

Drugs and administration

BDNF was obtained from Sigma-Aldrich; PKI 6–22 was purchased from Tocris Bioscience; and AKT inhibitor III was obtained from Santa Cruz Biotechnology. All drugs were dissolved in deionized-distilled water as concentrated stock solutions. ANA-12, 7,8-dihydroxyflavone, LY294002, and SB203580 were purchased from Sigma-Aldrich; LY303511 was obtained from Tocris Bioscience; KT-5720 was purchased from Abcam; TTA-P2 was obtained from Alomone Labs; and SB202474 was obtained from Merck Millipore. All of these molecules were prepared as concentrated stock solutions in dimethylsulfoxide (DMSO). The concentration of DMSO in each medium was less than 0.05% and had no significant effects on T-currents. All other chemicals were purchased from Sigma-Aldrich unless otherwise indicated.

Statistical analysis

Data are presented as original traces or as the means ± SEM. For the purpose of data acquisition and analysis, Clampfit 10.2 (Molecular Devices, USA) and/or the GraphPad Prism 5.0 software package (GraphPad Software) was used. Sigmoidal BDNF dose-response curves were fitted to the data using the following Hill equation: I/Icontrol = 1/[1 + 10(logED50 − log[BDNF])n], where ED50 is the dose at which the half-maximum effect occurs, and n is the coefficient. The voltage-activation and voltage-inactivation relationships were fitted to the Boltzmann equations, respectively: G/Gmax = 1/{1 + exp.[(V50,actV)/k]} and I/Imax = 1 – 1/{1 + exp.[(V50,inactV)/k]}, where Gmax and Imax are the maximum conductance and current amplitude, V50,act and V50,inact are the half-maximum potentials for activation and fast inactivation, and k is the slope factor. Statistical comparisons were performed using one-way analysis of variance (ANOVA) followed by a post hoc Bonferroni correction or using Student’s t test if the comparisons were restricted to two means. The behavioral data were analyzed by two-way repeated-measures ANOVA followed by Bonferroni posttest. Values of P < 0.05 were considered statistically significant.

SUPPLEMENTARY MATERIALS

stke.sciencemag.org/cgi/content/full/12/600/eaaw2300/DC1

Fig. S1. Abundance of TrkB in rat TGs.

Fig. S2. Abundance of p-AKT.

Fig. S3. Knockdown of AKT in rat TGs.

Fig. S4. Abundance of p38, ERK, and JNK.

Fig. S5. Change in the abundance of p-p38 induced by BDNF.

Fig. S6. Abundance of p-PKA.

Fig. S7. Knockdown of Cav3.2 in rat TGs.

Fig. S8. Increased abundance of TrkB induced by CFA.

REFERENCES AND NOTES

Acknowledgments: We are grateful to S. Gong and S. Yu for technological support. Funding: This work was supported by the National Natural Science Foundation of China (nos. 81873731, 81771187, 81622014, 81671080, and 81571063), the Qing-Lan Project of Jiangsu Province, the Six Talent Peak Project of Jiangsu Province (JY-065), the Shanghai Natural Science Foundation (14ZR1434000), the Science and Technology Development Fund of Shanghai Pudong New Area (PKJ2015-Y19), the Jiangsu Key Laboratory of Neuropsychiatric Diseases (BM2013003), and a project funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions. Author contributions: H.W., Y.W., Y.P., Y.Z., and D.J. performed the experiments and contributed to the acquisition and analysis of data; D.J., X.J., and J.T. contributed to data interpretation; H.W., X.J., and J.T. designed the experiments, performed the analysis of data, and wrote (and all authors edited) the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Stay Connected to Science Signaling

Navigate This Article