Excitatory GABA induces BDNF transcription via CRTC1 and phosphorylated CREB-related pathways in immature cortical cells

Authors

  • Mamoru Fukuchi,

    1. Department of Biological Chemistry, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama, Japan
    Search for more papers by this author
  • Yuya Kirikoshi,

    1. Department of Biological Chemistry, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama, Japan
    Search for more papers by this author
  • Atsumi Mori,

    1. Department of Biological Chemistry, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama, Japan
    Search for more papers by this author
  • Reika Eda,

    1. Department of Biological Chemistry, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama, Japan
    Search for more papers by this author
  • Daisuke Ihara,

    1. Department of Biological Chemistry, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama, Japan
    Search for more papers by this author
  • Ichiro Takasaki,

    1. Division of Molecular Genetics Research, Life Science Research Center, University of Toyama, Toyama, Japan
    2. Department of Pharmacology, Graduate School of Science and Engineering, University of Toyama, Toyama, Japan
    Search for more papers by this author
  • Akiko Tabuchi,

    1. Department of Biological Chemistry, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama, Japan
    Search for more papers by this author
  • Masaaki Tsuda

    Corresponding author
    1. Department of Biological Chemistry, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama, Japan
    • Address correspondence and reprint requests to Masaaki Tsuda, Department of Biological Chemistry, Faculty of Pharmaceutical Sciences, University of Toyama, 2630 Sugitani, 930-0194, Japan.

      E-mail: tsuda@pha.u-toyama.ac.jp

    Search for more papers by this author

Abstract

Although the excitatory action of GABA has been shown to activate the expression of brain-derived neurotrophic factor (BDNF), its molecular mechanisms remain unclear. Using cultured rat cortical cells, we here demonstrated that GABA induced Bdnf mRNA expression mainly via L-type voltage-dependent Ca2+ channels (L-VDCC) at the early stage and inhibited it at the late stage of the culture, which corresponded to the excitatory and inhibitory states of cortical cells. The excitatory GABA-induced Bdnf mRNA expression was controlled by multiple Ca2+ signaling pathways including Ca2+/calmodulin-dependent protein kinase (CaMK), mitogen-activated protein kinase (MAPK) and calcineurin (CN). The Bdnf-promoter IV (Bdnf-pIV) was activated by GABA, mainly via cAMP-response element (CRE)/CREB, and this was prevented by the over-expression of a dominant negative CREB. The nuclear translocation of CREB-regulated transcriptional coactivator 1 (CRTC1) was selectively induced by the GABA-induced CN pathway to activate Bdnf-pIV. On the other hand, GABA-induced Gal4-CREB-dependent transcription, which was controlled by multiple Ca2+ signaling pathways, was prevented when the serine at position 133 of Gal4-CREB was mutated to alanine. Taken together, the excitatory action of GABA transcriptionally activated Bdnf expression through the combination of nuclear-localized CRTC1 and phosphorylated CREB in immature cortical cells, and may be the molecular mechanisms underlying Bdnf expression to control neuronal development.

image

We demonstrated that GABA induced Bdnf expression at the early stage of the culture, in which GABA exerted its excitatory action. The excitatory GABA-induced Bdnf expression was controlled by multiple Ca2+ signaling pathways evoked via L-VDCC. Both the CREB coactivator, CRTC1 and CREB phosphorylation participated in excitatory GABA-induced Bdnf transcription. Our present study indicates the mechanism underlying the excitatory GABA-induced Bdnf expression in immature neurons and provide new insights into molecular mechanisms underlying Bdnf expression to control neuronal development.

Abbreviations used
APV

DL-2-amino-5-phosphonovaleric acid

BDNF

brain-derived neurotrophic factor

CaMK

Ca2+/calmodulin-dependent protein kinase

CaMKK

CaMK kinase

CaRE3

Ca2+-responsive element 3

CBP

CREB-binding protein

CN

calcineurin

CREB

CRE-binding protein

CRE

cAMP-response element

CRTC

CREB-regulated transcriptional coactivator

DAPI

4′,6-diamidino-2-phenylindole

DMEM

Dulbecco's modified Eagle medium

ERK

extracellular-signal-regulated kinase

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

GFP

green fluorescent protein

HBS

HEPES-buffered saline

KCC2

K+–Cl cotransporter 2

L-VDCC

L-type voltage-dependent Ca2+ channel

MAP2

microtubule-associated protein 2

MAPK

mitogen-activated protein kinase

MEK1/2

MAPK/ERK kinase 1/2

NKCC1

Na+–K+–Cl cotransporter 1

NMDA-R

NMDA receptor

PBS

phosphate-buffered saline

RNAi

RNA interference

shRNA

short hairpin RNA

tPA

tissue plasminogen activator

GABA is the main inhibitory neurotransmitter in the CNS. In contrast to its inhibitory action, GABA also functions as an excitatory neurotransmitter in early development of the brain (Ben-Ari et al. 2007). The excitatory action of GABA has been shown to play a crucial role in the proliferation of neural stem cells and neural differentiation at early developmental stages. Cortical neural stem cells, such as radial glia, or migrating neurons were previously shown to express functional GABAA receptors in the developing brain (LoTurco et al. 1995; Behar et al. 1996; Owens and Kriegstein 2002), and stimulating these receptors with GABA inhibited stem cell proliferation and promoted neural differentiation (LoTurco et al. 1995). In the adult brain, neurogenesis occurs in the dentate gyrus of the hippocampus, and the excitatory action of GABA from GABAergic interneurons was shown to promote the differentiation of immature to mature neurons (Tozuka et al. 2005).

The intracellular concentration of Cl ([Cl]i) is low in mature neurons, and, therefore, activation of the ionotropic GABAA receptor causes Cl influx into neurons and membrane hyperpolarization, resulting in its inhibitory action. On the other hand, because of the higher [Cl]i in immature neurons, activation of the GABAA receptor induces Cl efflux and membrane depolarization (Ben-Ari 2002). These different responses to GABA have been attributed to the differential expression of two Cl transporters, the Na+–K+–2Cl cotransporter 1 (NKCC1) and K+–Cl cotransporter 2 (Delpire 2000) in immature and mature neurons (Plotkin et al. 1997; Rivera et al. 1999).

Ca2+ signals evoked via voltage-dependent Ca2+ channels (VDCCs), which are induced by the excitatory action of GABA, are known to promote the differentiation of immature neurons in adult neurogenesis (Tozuka et al. 2005) and regulate the expression of brain-derived neurotrophic factor (BDNF) (Berninger et al. 1995; Obrietan et al. 2002). BDNF, a neurotrophin, is initially translated from multiple transcripts as a precursor protein (preproBDNF). After cleavage of the signal peptide, proBDNF is converted intra- or extracellularly into mature BDNF by several proteases such as furin, proprotein convertase, and tissue plasminogen activator (tPA) (Greenberg et al. 2009). BDNF contributes to the expression of various neuronal functions in the brain, such as neuronal survival, differentiation, and synaptic plasticity (Park and Poo 2013). To elucidate the role of BDNF expression in the differentiation of immature neurons, it is important to investigate how the induction of BDNF gene (Bdnf) expression is regulated by the excitatory action of GABA in immature neurons at the transcription level, the mechanisms of which remain unknown. The transcription factor, cAMP-response element (CRE)-binding protein (CREB), is a candidate molecule that controls GABA-induced Bdnf expression. The excitatory action of GABA was shown to increase CREB phosphorylation in immature hypothalamic neurons (Obrietan et al. 2002). Although the phosphorylation of CREB at Ser133 promotes the recruitment of CREB-binding protein (CBP) and p300 to stimulate transcription (Chrivia et al. 1993), it has been suggested that the phosphorylation of CREB is not always sufficient to stimulate CREB-mediated transcription. As potent coactivators for transcription, CREB-regulated transcription coactivators (CRTCs), which interact with the basic leucine zipper domain of CREB independently of phosphorylated CREB, have been focused on (Conkright et al. 2003; Iourgenko et al. 2003). It has already been established that dephosphorylation of CRTC by calcineurin (CN) promotes its translocation from the cytoplasm to the nucleus to control CREB-mediated transcription (Bittinger et al. 2004; Screaton et al. 2004). However, how CREB phosphorylation and CRTCs functionally interact to control CREB-mediated transcription in neurons remains unclear. Therefore, we focused on the transcriptional mechanisms of the excitatory GABA-induced transcription of Bdnf, particularly in terms of the translocation of CRTC1 to the nucleus and phosphorylation of CREB.

Primary cultures of neuronal cells are a good model to investigate the reversible actions of GABA in vitro; the action of GABA can be switched from excitatory to inhibitory during the culture (Ganguly et al. 2001). Using a primary culture of rat cortical cells, we first confirmed the reversible actions of GABA and found that excitatory GABA induced the transcription activation of Bdnf through the phosphorylation of CREB at Ser133 and nuclear translocation of CRTC1, which were controlled by multiple Ca2+ signaling pathways including CN, extracellular-signal-regulated kinase (ERK)/mitogen-activated protein kinase (MAPK), and Ca2+/calmodulin-dependent protein kinase (CaMK). The results of this study provide new insights into the excitatory GABA-mediated differentiation of neurons not only in early development of the brain, but also in adult neurogenesis in the hippocampus, which may be mediated by the activation of Bdnf gene expression.

Materials and methods

Reagents

GABA, nicardipine, muscimol, baclofen, DL-2-amino-5-phosphonovaleric acid (APV), and FK506 were purchased from Sigma-Aldrich (St Louis, MO, USA) (-)-bicuculline methochloride from Tocris Bioscience (Bristol, UK), and STO609, KN93, and U0126 from Calbiochem (San Diego, CA, USA).

Cell culture

A primary culture of rat cortical cells was prepared from the cerebral cortex of Sprague–Dawley rat (Japan SLC, Hamamatsu, Japan) embryos at embryonic day 17, as described previously (Tabuchi et al. 1998). All animal care and experiments were carried out in accordance with the Guidelines for the Care and Use of Laboratory Animals of the University of Toyama and ARRIVE guidelines. After the dissected tissue was treated with trypsin solution and DNase I solution, cells were re-suspended in neurobasal medium containing 1 × B27 supplement (Invitrogen, Carlsbad, CA, USA), 2 μg/mL gentamicin, and 2 mM glutamine. The suspended cells were seeded in poly-l-lysine-coated 6-well culture plates (Thermo Scientific, Waltham, MA, USA, 2.0 × 106 cells/well). Half of the culture medium was exchanged for fresh medium every 3 days.

Ca2+imaging

Cortical cells were seeded in a poly-l-lysine-coated glass-bottom dish (Asahi Glass Co., Ltd., Tokyo, Japan). Changes in [Ca2+]i in single cells were measured using the fura-2 technique with minor modifications (Fukuchi et al. 2009).

Quantitative RT-PCR analysis

Quantitative RT-PCR was performed according to methods previously described (Fukuchi et al. 2009). Primer sequences for measuring each mRNA were as follows: Bdnf-eIV, 5′-TCGGCCACCAAAGACTCG-3′ and 5′-GCCCATTCACGCTCTCCA-3′; Nkcc1, 5′-TCTCTAGAAGTCTAGGGCCAG-3′ and 5′-CTGAGATGCCCAGAAGAACCA-3′; Kcc2, 5′-TGGAGGCCACGCTTCCGATA-3′ and 5′-AGCGAGCTGCACTGAGAGAC-3′; Gapdh, 5′-TCCATGACAACTTTGGCATCGTGG-3′ and 5′-GTTGCTGTTGAAGTCACAGGAGAC-3′.

DNA transfection and reporter assay

DNA transfection was performed using the calcium phosphate precipitation method (Tabuchi et al. 1998, 2005). Cortical neurons cultured at 3 days in culture were washed once with pre-warmed Dulbecco's modified Eagle medium (DMEM) and incubated in 2 mL DMEM. The plasmid solution [total 6 μg of plasmids (pGL4.12-Bdnf-pIV: phRL-TK(−) (Promega, Madison, WI, USA) = 10 : 1, pGL4.12-Bdnf-pIV: phRL-TK(−): CREB expression vector (WCREB, ACREB, or MCREB) = 10 : 1 : 1, pGL4.12-Bdnf-pIV: phRL-TK(−): CREB expression vector (Flag-CREB wild type or Flag-CREB R314A) = 10 : 1 : 2, pGL4.12-Bdnf-pIV: phRL-TK(−): shRNA expression vector = 10 : 1 : 10, or pFR-Luc: phRL-TK(−): pFA2-CREB (Gal4-CREB(1–280) expression vector) = 10 : 1 : 10) in 0.25 mM CaCl2 solution] was mixed with an equal volume of 2 × HEPES-buffered saline (pH 7.09) and left at 25°C for 15 min to allow calcium phosphate to precipitate with the plasmids. Cells were then exposed to the precipitate for 15 min. After cells were washed three times with DMEM, they were placed back in conditioned medium. Cells were incubated for another 48–72 h and used for the reporter assay. The transfected cells were treated with GABA or vehicle for 6 h, and the lysate was extracted with 1 × passive lysis buffer (Promega). Dual (firefly and Renilla) luciferase activity was measured using a Dual-luciferase Reporter Assay System (Promega) with TD-20/20 Luminometer (Promega). To normalize the efficacy of DNA transfection, a phRL-TK(int−) vector (Promega) was co-transfected with a firefly reporter plasmid. In Fig. 5a(iii), we investigated promoter activity by measuring firefly luciferase activity. To evaluate the efficiency of DNA transfection, a green fluorescent protein (GFP) expression vector was transfected into the cortical cells. The transfection efficiency was 3.75 ± 0.484% (We counted more than 700 cells from five independent experiments) based on GFP-positive neurons (Figure S1). Although the efficiency using the calcium phosphate precipitation method was low, it was sufficient to measure promoter activity (Tabuchi et al. 1998, 2000, 2005).

Immunostaining

Cells were seeded on an 18-mm circle coverslip coated with poly-l-lysine. At 4 days in culture, cortical cells were fixed in phosphate-buffered saline (PBS)(+) containing 4% formaldehyde and 4% sucrose for 15 min at 25°C, and were then treated with blocking-PBS(+) containing 0.3% Triton X-100, 3% bovine serum albumin, and 3% normal goat serum for 1 h at 25°C. After blocking, cells were incubated with the primary antibody against microtubule-associated protein 2 (1 : 500, Sigma-Aldrich) or GFP (1 : 1000, Medical & Biological Laboratories Co., Ltd., Nagoya, Japan), and anti-CRTC1 antiserum [1 : 1000, generously provided by Dr. Hiroshi Takemori (Laboratory of Cell Signalling and Metabolism, National Institute of Biomedical Innovation, Osaka, Japan)]. After the cells were washed in PBS(+), they were incubated with the CF488- or CF594-conjugated secondary antibody against rabbit or mouse IgG (Biotium Inc., Hayward, CA, USA). Nuclei were counter-stained with 300 nM 4′, 6-diamidino-2-phenylindole (DAPI, Molecular Probes, Eugene, OR, USA). After another wash, coverslips were mounted on the slides with Fluoromount (Diagnostic BioSystems, Pleasanton, CA, USA). Confocal fluorescent images were taken with a LSM 700 confocal microscope (Carl Zeiss, Oberkochen, Germany) and these images were analyzed using ZEN 2009 software (Carl Zeiss). Signal intensities of CRTC1 in the cytoplasm and nucleus were estimated in reference to those of microtubule-associated protein 2 (cytoplasm) and DAPI (Nucleus). The cell containing CRTC1 in its nucleus was determined by higher immunofluorescence intensity of CRTC1 in the nucleus than that in the cytoplasm. The numbers of cells containing CRTC1 in their nuclei and total number of cells were counted, and the percentage of cells with nuclear CRTC1 was calculated. We counted more than 450 neurons from 10 independent experiments.

Knockdown of endogenous CRTC1

The pSUPER RNAi System (OligoEngine, Seattle, WA, USA) was used to knock down endogenous CRTC1 expression, according to the manufacturer's instructions. The oligonucleotide sequences were as follows: shScramble: 5′-GATCCCCAATTAGACGAGTGCGTCGCTTCAAGAGAGCGACGCACTCGTCTAATTTTTTTA-3′ and 5′-AGCTTAAAAAAATTAGACGAGTGCGTCGCTCTCTTGAAGCGACGCACTCGTCTAATTGGG-3′; shCRTC1: 5′-GATCCCCCCTTCGAGGAGGTCATGAATTCAAGAGATTCATGACCTCCTCGAAGGTTTTTA-3′ and 5′-AGCTTAAAAACCTTCGAGGAGGTCATGAATCTCTTGAATTCATGACCTCCTCGAAGGGGG-3′. Underlined sequences indicate target sequences. The target sequence was designed to knock down human, mouse and rat CRTC1. To evaluate knockdown efficiency, the expression vectors of human CRTC1 (pCMV-Sport6-human CRTC1, 0.4 μg) and shRNA (0.4 μg) were co-transfected into NIH3T3 cells, which were cultured on a poly-l-lysine-coated 35 mm culture dish (Asahi Glass Co., Ltd), using Lipofectamine Reagent (Invitrogen) and Plus Reagent (Invitrogen) according to the manufacturer's instructions. A GFP expression vector (0.2 μg) was also co-transfected to confirm the efficiency of DNA transfection. Twenty-four or forty-eight hours after DNA transfection, cell lysates were prepared for immunoblot analysis. To evaluate knockdown efficiency of endogenous CRTC1 in cultured rat cortical cells, the shRNA and GFP expression vectors were co-transfected into the cells at 3 days in culture, and the expression of endogenous CRTC1 in GFP-positive cells was investigated 3 days later using immunostaining.

Immunoblot analysis

The cell lysate extracted from cultured rat cortical cells was mixed with an equal volume of 2 × Laemmli sample buffer (Bio-Rad Laboratories, Hercules, CA, USA) and the protein extract was separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel-electrophoresis on a 10% resolving gel. After transferring the proteins to a polyvinylidene difluoride membrane (Bio-Rad Laboratories), the membrane was incubated with anti-CRTC1 antiserum (1 : 1000), anti-GFP (1 : 1000), anti-α-tubulin (1 : 1000, Sigma-Aldrich), anti-phospho-CREB (Ser133) (1 : 1000, Cell Signaling Technology, Beverly, MA, USA), and anti-CREB (1 : 1000, Cell Signaling Technology) antibodies overnight at 4°C and then with horseradish peroxidase-conjugated rabbit or mouse anti-IgG (1 : 5000, GE Healthcare, Buckinghamshire, UK) for 1 h at 25°C. Proteins were detected with the enhanced chemiluminescence protocol (ECL Western blotting detection reagents, GE Healthcare).

Statistics

All values represent the mean ± SE for a number of separate experiments performed in duplicate, as indicated in the corresponding figures. Statistical analyses were performed using a two-way factorial anova. However, an interaction was present in almost all experiments (data not shown); therefore, each group was analyzed using a one-way anova with Sheffe's F test (Table S1). The relationship between developmental changes in the expression levels of Nkcc1 and Kcc2 was analyzed using the Spearman rank correlation (Fig. 1c).

Figure 1.

Switching from the excitatory to inhibitory action of GABA during the development of cultured rat cortical cells. (a) Changes in [Ca2+]i in the presence of GABA (100 μM, i and ii), muscimol (100 μM, iii), or baclofen (25 μM, iv) at 4 days in culture. Nicardipine (Nica, 5 μM) was added before the GABA treatment (ii). Changes in [Ca2+]i in approximately 50 cells were measured. (b) Spontaneous Ca2+ oscillation in mature cortical cells (at 14 days in culture), and the effect of DL-2-amino-5-phosphonovaleric acid (APV) (200 μM, i) or GABA (ii) on changes in [Ca2+]i. (c) Changes in Cl transporters during the culture. Total RNA was extracted at 4, 7, or 14 days in culture to measure the quantities of Nkcc1 and Kcc2 mRNA expression using quantitative RT-PCR analysis. Data represent the mean ± SE (n = 6). The relationship between Nkcc1 and Kcc2 was analyzed using the Spearman rank correlation, and an inverse correlation was observed (rs = −0.521, < 0.05).

Results

Switching from the excitatory to inhibitory action of GABA during development of rat cortical cells in culture

We first investigated the cellular responses of rat cortical cells to GABA at the early stages of the culture. Using fura-2 fluorescence ratio imaging to monitor changes in the intracellular concentration of Ca2+ ([Ca2+]i), we found that [Ca2+]i transiently increased with the addition of GABA at 4 days in culture [Fig. 1a(i)], but did not with vehicle (data not shown). This transient increase in [Ca2+]i was effectively prevented by pre-treatment with nicardipine, a blocker for L-type voltage-dependent Ca2+ channels (L-VDCC) [Fig. 1a(ii)], which indicated that L-VDCC is the main entrance for extracellular Ca2+ into cells. Furthermore, the addition of muscimol, an agonist of GABAA receptors [Fig. 1a(iii)], but not baclofen, an agonist of GABAB receptors [Fig. 1a(iv)], induced an increase, which indicated the specific involvement of GABAA receptors in the increase in [Ca2+]i induced by GABA at the early stages of culture.

On the other hand, a prolonged culture of cortical cells caused a synchronized Ca2+ oscillation at 14 days in culture (Fig. 1b), which could be spontaneously generated at the later stages of culture as a result of the maturation of synapses in culture (Murphy et al. 1992; Wang and Gruenstein 1997). At this later stage, we investigated the effect of GABA on [Ca2+]i and found that the Ca2+ oscillation was abolished upon the addition of APV or GABA (Fig. 1b). The spontaneous and synchronized Ca2+ oscillation was observed in cortical cells from 9 to 11 days in culture onward, but not at 4 or 7 days in culture (data not shown).

Developmental changes in two cation-chloride cotransporters, the chloride accumulator NKCC1 and chloride extruder K+–Cl cotransporter 2, have been shown to play a pivotal role in the developmental changes in [Cl]i and excitatory/inhibitory GABA responses (Ben-Ari et al. 2007). Therefore, we examined changes in the mRNA expression of these Cl transporters. Nkcc1 mRNA expression levels decreased at 14 days in culture (Fig. 1c), whereas those of Kcc2 mRNA gradually increased at 7 and 14 days in culture (Fig. 1c). We found that the developmental changes in the expression levels of Nkcc1 inversely correlated with those of Kcc2 (Fig. 1c, rs = −0.521, < 0.05).

The excitatory action of GABA induced Bdnf mRNA expression through Ca2+ signals evoked via L-VDCC

Aid et al. (2007) demonstrated that the rat Bdnf gene consists of eight 5′-untranslated exons and one 3′-exon encoding preproBDNF, the mRNA expression of which is regulated by alternative promoters located upstream of each exon. Among the alternative Bdnf transcripts, the expression of exon IV-containing Bdnf transcripts (Bdnf-eIV) is the most abundant in the brain (Tao et al. 1998; Aid et al. 2007). As it has also been demonstrated that the expression of Bdnf-eIV mRNA is activated by Ca2+ signals evoked via L-VDCCs (Tao et al. 1998; Tabuchi et al. 2000), we focused on changes in Bdnf-eIV mRNA expression and its responses to GABA during the culture. The basal expression of Bdnf-eIV mRNA was higher at 14 days in culture in culture than at 4 and 7 days in culture (Fig. 2a). The addition of GABA markedly induced the expression of Bdnf-eIV mRNA at 4 and 7 days, but reduced it at 14 days in culture (Fig. 2a).

Figure 2.

Effect of GABA on Bdnf mRNA expression in cultured rat cortical cells. (a) Changes in the GABA response during the culture. Cortical cells were treated with 100 μM GABA for 6 h at 4, 7, or 14 days in culture, and total RNA was extracted to measure the quantity of Bdnf-eIV mRNA expression using quantitative RT-PCR analysis. Data represent the mean ± SE (n = 6). *< 0.05 and **< 0.01 versus the control at each stage. ##< 0.01 versus the control at 4 days in culture. (b) Changes in Bdnf-eIV mRNA expression with the GABA treatment in immature cortical cells. The primary culture of cortical cells was treated with GABA for 1, 3, and 6 h (i), or 6 h (ii, iii, and iv) at 4 days in culture, and total RNA was extracted for quantitative RT-PCR analysis. Bicuculline (Bic, 200 μM, ii), nicardipine (iii and iv), or DL-2-amino-5-phosphonovaleric acid (APV) (d) was added 10 min before the GABA treatment. Data represent the mean ± SE (n = 4–5). **< 0.01 versus the control. ## < 0.01 versus the same sample without bicuculline, nicardipine, or APV. ††< 0.01 versus the same sample with APV alone. (c) Changes in Bdnf-eIV mRNA expression with the APV or GABA treatment at 14 days in culture. Six hours after the treatment of cortical cells with APV or GABA, total RNA was extracted for quantitative RT-PCR analysis. Data represent the mean ± SE (n = 4–5). **< 0.01 versus the control.

The expression of Bdnf-eIV mRNA time dependently increased upon the addition of GABA [Fig. 2b(i)]. A similar increase was also observed by the addition of muscimol (data not shown). GABA-induced mRNA expression was completely inhibited by the pre-treatment with bicuculline, an antagonist of GABAA receptors, or nicardipine, a blocker of L-VDCC [Fig. 2b(ii), (iii)]. The expression of Bdnf-eIV mRNA was not changed by the bicuculline treatment in the absence of exogenously added GABA [Fig. 2b(ii)]. Induction was partially prevented by APV, an antagonist of NMDA receptor (NMDA-R), and was further prevented in combination with nicardipine [Fig. 2b(iv)]. On the other hand, the basal expression of Bdnf-eIV mRNA at 14 days in culture decreased upon the addition of GABA or APV (Fig. 2c).

Intracellular signaling pathways involved in GABA-induced Bdnf transcription

We investigated the intracellular signaling pathways involved in excitatory GABA-induced Bdnf mRNA expression. GABA-induced Bdnf-eIV mRNA expression was significantly inhibited by STO-609, KN-93, U0126, or FK506, inhibitors of CaMK kinase, CaMKs, MAPK/ERK kinase 1/2, and CN, respectively (Fig. 3a). Therefore, multiple Ca2+ signaling pathways, including CaMKs, ERK/MAPK and CN, are involved in GABA-induced Bdnf-eIV mRNA expression. Among these inhibitors, either KN93 or FK506 almost completely inhibited GABA-induced Bdnf-eIV mRNA expression.

Figure 3.

Intracellular signaling pathways involved in GABA-induced Bdnf gene expression. (a) Effect of STO609, KN93, U0126, or FK506 on GABA-induced Bdnf-eIV mRNA expression. Cortical cells were treated with 100 μM GABA for 6 h at 4 days in culture, and total RNA was extracted for quantitative RT-PCR analysis. STO609 (5 μM, i), KN93 (10 μM, i), U0126 (20 μM, ii), or FK506 (5 μM, ii) was added 10 min before the GABA treatment. Data represent the mean ± SE (n = 3–4). **< 0.01 versus the control. ##< 0.01 versus the same sample without an inhibitor. (b) Effect of nicardipine, KN93, U0126, or FK506 on GABA-induced Bdnf-pIV activation. Reporter plasmids were transfected into the cells at 3 days in culture. Cells were treated with GABA 40 h after DNA transfection, and cell lysates were prepared to measure luciferase activity. Nicardipine (Nica), KN93, U0126, or FK506 was added 10 min before the GABA treatment. Data represent the mean ± SE (n = 3–5). *< 0.05 and **< 0.01 versus the control. #< 0.05 and ##< 0.01 versus the same sample without an inhibitor. (c) Effect of the CRE mutation (i) or over-expression of wild-type or dominant negative CREB (ii) on GABA-induced Bdnf-pIV activation. Transfected cells were treated with 100 μM GABA for 6 h, and cell lysates were prepared to measure luciferase activity. Data represent the mean ± SE (n = 3–4). **< 0.01 versus the control. ##< 0.01 versus the wild-type (i).

We next examined whether Bdnf transcription was activated by the excitatory action of GABA. Using a luciferase reporter assay, we found that the activity of Bdnf promoter IV (Bdnf-pIV), which controls the expression of Bdnf-eIV mRNA, was activated by the addition of GABA, whereas its activation was completely inhibited by nicardipine (Fig. 3b), which indicated the dependency of Bdnf-pIV activation on Ca2+ signals evoked via L-VDCCs. Corresponding to changes in Bdnf-eIV mRNA expression, the GABA-induced activation of Bdnf-pIV was also prevented by KN93, U0126, or FK506 (Fig. 3b).

GABA-induced Bdnf-pIV activation was effectively prevented by the mutation of CRE [Fig. 3c(i)], which is also referred to as Ca2+-responsive element 3 (Tao et al. 2002). We next examined how CREB is involved in the GABA-induced Bdnf-pIV activation using a dominant negative CREB. GABA-induced promoter activation was slightly enhanced when wild-type CREB (WCREB) was over-expressed [Fig. 3c(ii)]. On the other hand, the over-expression of ACREB, which lacks a transcription activation domain and disrupts DNA binding of CREB, or MCREB, which is a phosphorylation mutant of CREB (Ser133Ala), prevented promoter activation [Fig. 3c(ii)]. These results indicate that CREB-CRE-dependent transcription is required for GABA-induced Bdnf-pIV activation.

Involvement of CRTC1 in the GABA-induced transcription activation of Bdnf-pIV

As the GABA-induced activation of Bdnf-pIV was controlled by CN [Fig. 3a(ii) and b], we investigated the involvement of CRTC1, one of the CRTC isoforms (Conkright et al. 2003; Iourgenko et al. 2003), which is widely and highly expressed in the brain (Watts et al. 2011) and dephosphorylated by CN, resulting in its translocation from the cytoplasm to the nucleus to activate CREB-mediated transcription (Bittinger et al. 2004; Li et al. 2009). Moreover, CRTC1 was previously shown to be required for long-term synaptic plasticity in the hippocampus (Zhou et al. 2006; Kovács et al. 2007). We first investigated the translocation of CRTC1 using immunostaining with anti-CRTC1 antiserum, and calculated the number of cells with nuclear CRTC1. CRTC1 was mainly localized in the cytoplasm of cortical cells at 4 days in culture, whereas the number of cells in which CRTC1 translocated to the nucleus increased with the GABA treatment (Fig. 4). We did not detect the nuclear localization of CRTC1 upon the FK506 treatment in either the absence or presence of GABA (Fig. 4). This nuclear translocation in response to GABA was canceled by nicardipine, but not by KN93 or U0126 (Fig. 4).

Figure 4.

CREB-regulated transcriptional coactivator1 (CRTC1) translocated from the cytoplasm to the nucleus in response to excitatory GABA. (a) Subcellular localization of CRTC1 in cultured rat cortical cells. Cortical cells were treated with GABA for 15 min at 4 days in culture, and cells were fixed and immunostained with the anti-CRTC1 antiserum and anti-MAP2 antibody. Nuclei were counter-stained with 4′, 6-diamidino-2-phenylindole (DAPI). FK506, nicardipine (Nica), KN93, or U0126 was added 10 min before the GABA treatment. Scale bar = 20 μm. White arrowhead indicates a representative example of nuclear localized CRTC1. A similar result was observed in three independent experiments. (b) The number of cells containing CRTC1 in the nucleus was counted based on an immunofluorescence signal of CRTC1. The total cell number was determined by the number of nuclei (DAPI). The percentage of cells with nuclear CRTC1 was calculated. Data represent the mean ± SE (n = 10). **< 0.01 versus the control. ##< 0.01 versus the same sample without an inhibitor.

We further examined the effect of knockdown of endogenous CRTC1 on GABA-induced Bdnf-pIV activation, by the expression of shRNA for CRTC1 (shCRTC1). In NIH3T3 cells, we demonstrated that the over-expressed CRTC1 protein was knocked down by shCRTC1, but not by scrambled shRNA (Scramble) [Fig. 5a(i)]. Moreover, the immunostained signals of the endogenous CRTC1 protein in the primary culture of rat cortical cells were also decreased by the expression of shCRTC1, but not by that of Scramble [Fig. 5a(ii)]. The GABA-induced activation of Bdnf-pIV was inhibited under the expression of shCRTC1 [Fig. 5a(iii)]. We also examined the effects of mutant CREB (CREB R314A) on the GABA-induced activation of Bdnf-pIV, in which an arginine at position 314 was substituted for an alanine residue. A previous study showed that this mutant CREB was unable to interact with CRTC1 (Kovács et al. 2007). The level of either GABA-induced activation or the basal activity of Bdnf-pIV was lower with the CREB R314A mutant than with wild-type CREB (Fig. 5b). These results indicate that CRTC1 contributes to the excitatory GABA-induced activation of Bdnf-pIV in immature cortical cells.

Figure 5.

CREB-regulated transcriptional coactivator1 (CRTC1) was necessary for excitatory GABA-induced Bdnf-pIV activation. (a) Effects of shCRTC1 on the expression of CRTC1 (i, ii) or GABA-induced Bdnf-pIV activation (iii). (i) The expression vectors of human CRTC1 and shRNA (pSuper empty, Scramble, or shCRTC1) were co-transfected into NIH3T3 cells. Scrambled shRNA (Scramble) was used as a negative control. Cell lysates were prepared 24 or 48 h after DNA transfection. The expression level of CRTC1 was analyzed by immunoblotting. (ii) The expression vectors of green fluorescent protein (GFP) and shCRTC1 were co-transfected into cortical cells at 3 days in culture. Cells were fixed and immunostained 3 days after DNA transfection. Endogenous CRTC1 was detected in both the cytoplasm and nucleus under the non-treated condition in cultured cells, that is, CRTC1 was diffusely expressed in cells. On the other hand, CRTC1 accumulated in the nucleus with the activation of NMDA-R. To clearly evaluate the effects of shCRTC1 or shScramble, cells were fixed under the condition of NMDA-R activation before immunostaining analysis. White arrow indicates a GFP-positive cell, that is, transfected cell. Scale bar = 20 μm. (iii) The expression vectors of shCRTC1 and reporter plasmids were co-transfected into cortical cells at 3 days in culture. Cells were treated with GABA 3 days after DNA transfection, and cell lysates were prepared to measure firefly luciferase activity. Data represent the mean ± SE (n = 4). #< 0.05 versus Scramble. (b) The Flag-CREB wild-type (CREB wild) or R314A mutant (CREB R314A) expression vector and reporter plasmid were co-transfected into cortical cells at 3 days in culture. Cells were treated with GABA 40 h after DNA transfection, and cell lysates were prepared to measure luciferase activity. Data represent the mean ± SE (n = 3). *< 0.05 and **< 0.01 versus the control. #< 0.05 and ##< 0.01 versus CREB wild.

The phosphorylation of CREB was involved in GABA-induced CREB-dependent transcription

Although CRTC1 was detected in the nucleus with the GABA treatment in the presence of KN93 or U0126 (Fig. 4), these inhibitors significantly prevented GABA-induced Bdnf gene expression (Fig. 3a and b). Moreover, the over-expression of MCREB repressed GABA-induced Bdnf-pIV activation [Fig. 3c(ii)]. These results suggest the involvement of not only CRTC1, but also CREB phosphorylation at Ser133 in the activation of Bdnf transcription. In practice, we detected the increase in CREB phosphorylation with the GABA treatment (Fig. 6a). We found that the level of phosphorylation of CREB caused by GABA was reduced by nicardipine. This increase in CREB phosphorylation tended not to be detected in the presence of KN 93, but did slightly in the presence of U0126 (Fig. 6a). The addition of FK506 increased the phosphorylation of CREB in the absence of GABA, which resulted in the absence of a GABA-induced increase in CREB phosphorylation (Fig. 6a).

Figure 6.

GABA activated CREB-dependent transcription via its phosphorylation. (a) (i) After the treatment with GABA for 5 min, total cellular protein was prepared. Immunoblotting was performed using anti-phosphorylated CREB (Ser133) or anti-CREB antibody. The experiment was performed at least three times, and the same tendencies were observed. (ii) The relative intensities of the bands, as shown in a(i), were analyzed using Image J software. Data represent the mean ± SE (n = 3). (b) An expression vector of Gal4-CREB and reporter plasmids were co-transfected into cortical cells at 3 days in culture. Forty hours later, cells were treated with 100 μM GABA, and the lysates were prepared to measure luciferase activity. Nicardipine (Nica, i), KN93 (ii), U0126 (ii), or FK506 (ii) was added 10 min before the GABA treatment. Data represent the mean ± SE (n = 3). **< 0.01 versus the control. ##< 0.01 versus the same sample without an inhibitor. (b) Gal4-CREB wild-type or Ser133Ala-mutant was co-transfected with reporter plasmids into cells, which were then treated with GABA for 6 h. Data represent the mean ± SE (n = 3). **< 0.01 versus the control. ##< 0.01 versus the wild-type.

To further elucidate whether excitatory GABA induced CREB-dependent transcription via its phosphorylation, we used Gal4-CREB, in which the Gal4 DNA-binding domain was fused to the transcriptional activation domain of CREB. As Gal4-CREB does not contain the basic leucine-zipper DNA-binding domain of CREB, it is possible to examine the functional involvement of CREB phosphorylation in CREB-mediated transcription, independently of CRTC1. Using Gal4-CREB, we demonstrated that GABA activated CREB-dependent transcription, and this activation was completely blocked by nicardipine [Fig. 6b(i)]. GABA-induced Gal4-CREB-dependent transcription was inhibited by KN93, U0126, or FK506, but was detected slightly in the presence of U0126 [Fig. 6b(ii)]. The activation of Gal4-CREB-dependent transcription was partially, but significantly prevented when the Ser133 of Gal4-CREB was substituted for alanine (Fig. 6c). These results indicated that multiple Ca2+ signaling pathways were involved in GABA-induced CREB-dependent transcription through, at least in part, the phosphorylation of CREB at Ser133.

Discussion

Consistent with previous reports (Berninger et al. 1995; Obrietan et al. 2002), we detected the excitatory and inhibitory actions of GABA using primary cultures of rat cortical cells. Using [Ca2+]i imaging, we found that GABA exhibited its excitatory action at 4 days in culture, and its inhibitory action at 14 days in culture. We also detected developmental changes in the expression of mRNA for two Cl transporters, Nkcc1 and Kcc2, which are responsible for the reversible action of GABA. In correspondence with the change in [Ca2+]i and the expression of two Cl transporters, Bdnf-eIV mRNA expression was induced by GABA at 4 or 7 days in culture, but was repressed at 14 days in culture during the culture, which indicated that GABA induced Bdnf-eIV mRNA expression in immature neurons, but not in mature ones. We showed that the excitatory action of GABA was exerted through the GABAA receptor, which has also been demonstrated in previous studies (Berninger et al. 1995; Chen et al. 1996; Obrietan et al. 2002). An antagonist of GABAA receptor did not reduce [Ca2+]i (data not shown, refer to Obrietan and van den Pol 1995) and Bdnf-eIV mRNA expression [Fig. 2b(ii), and refer to Berninger et al. 1995] in the control culture that was not treated with GABA, suggesting that spontaneous release of GABA did not occur under our basal culture condition.

The excitatory action of GABA has already been shown to induce depolarization-dependent expression of immediate-early genes such as c-fos or Bdnf in immature hippocampal and hypothalamic neurons (Berninger et al. 1995; Obrietan et al. 2002). Although the GABA-induced Bdnf-eIV mRNA expression was partially prevented by APV, we found that the Bdnf-eIV mRNA induction was primarily mediated through L-VDCC because of its complete inhibition by nicardipine. As we have already demonstrated that NMDA-R activity is partially involved in the depolarization-induced Ca2+ influx through L-VDCC and Bdnf expression under high K+ condition of culture (Tabuchi et al. 2000), this effect of APV on the Bdnf-eIV mRNA expression is likely to be caused by a secondary effect on L-VDCC under depolarization.

Obrietan et al. (2002) demonstrated that the excitatory action of GABA activated the phosphorylation of CREB and Bdnf expression via the ERK/MAPK pathway. Using a promoter assay of Bdnf-pIV in a culture, we here clearly demonstrated that GABA-induced Bdnf-eIV mRNA expression was transcriptionally controlled mainly by CRE/CREB through Ca2+ signals evoked via L-VDCC, and the excitatory GABA-induced increase in Bdnf-eIV mRNA expression and Bdnf-pIV activation were similarly controlled by multiple Ca2+ signaling pathways including ERK/MAPK, CaMK, and CN. On the other hand, Li et al. (2009) already showed that Ca2+ signals evoked via L-VDCCs under depolarization induced CREB-dependent transcription activation through the translocation of CRTC1 from the cytoplasm to the nucleus. Consistent with their reports, we showed that GABA promoted the translocation of CRTC1 from the cytoplasm to the nucleus, which was completely blocked by FK506 or nicardipine, suggesting the essential involvement of CN in the translocation of CRTC1 induced by Ca2+ signals via L-VDCCs. The involvement of CRTC1 in GABA-induced Bdnf-pIV activation was supported by the observation that activation was suppressed by the knockdown of endogenous CRTC1, and partially by the over-expression of the CREB R314A mutant. Thus, not only the ERK/MAPK and CaMK pathways but also the pathway from Ca2+/CN to CREB through CRTC1 participates in the Bdnf gene expression induced by the excitatory action of GABA. As depolarization also induces Bdnf expression through ERK/MAPK, CaMK, and CN pathways (West et al. 2001; Kingsbury et al. 2007; Zheng et al. 2011), it is likely that the GABA-induced Bdnf-eIV mRNA expression could be controlled mostly by the same mechanisms as the depolarization-induced expression.

CRTC1-related CREB transcription was previously shown to be functional even in the absence of CREB phosphorylation (Conkright et al. 2003). However, it is more likely that the phosphorylation of CREB observed in this study was also responsible for the GABA-induced activation of Bdnf-pIV because the expression of MCREB significantly decreased activation. In support of these findings, we found that GABA increased the phosphorylation of CREB at Ser133, and this was consistent with a previous study (Obrietan et al. 2002). Using Gal4-CREB, we also found that the excitatory action of GABA induced CREB-dependent transcription via multiple Ca2+ signaling pathways such as CaMK, MAPK, and CN. The increase in CREB phosphorylation may have corresponded to the activation of Gal4-CREB-dependent transcription under the GABA treatment. For example, we found a slight increase in CREB phosphorylation by GABA in the presence of U0126, which may explain why the activation of Gal4-CREB transcription under the GABA treatment was slightly detected in the presence of U0126. On the other hand, the GABA-induced increase in CREB phosphorylation was not detected in the presence of FK506 as a result of an increase in the basal level (without GABA) of phosphorylation of CREB by FK506 (Fig. 6a), corresponding to the absence of activation of Gal4-CREB transcription in the presence of FK506. These results suggest that the increase in the level of CREB phosphorylation may be essential for inducing CREB-mediated transcriptional activation. Overall, these results indicate that both the translocation of CRTC1 and CREB phosphorylation are needed to control CREB-dependent gene transcription under the excitatory action of GABA in immature cortical cells. Although L-VDCC-mediated Ca2+ signaling pathways including MAPK and CaMK are shown to activate Bdnf transcription via CREB phosphorylation (West et al. 2001; Zheng et al. 2011), we here demonstrated that the excitatory action of GABA-mediated multiple Ca2+ signaling pathways participated in the activation of Bdnf-pIV transcription via not only CREB phosphorylation but also the nuclear translocation of CRTC1 in immature cortical cells.

Although CRTC1 interacts with the basic leucine zipper domain of CREB independently of phosphorylated CREB (Conkright et al. 2003), it remains unclear how CRTC1 could be involved in CREB-mediated transcription in the presence of phosphorylated CREB. It is important to note that the addition of U0126 or KN93 decreased Bdnf-pIV activation, but allowed translocation of CRTC1 to the nucleus with the GABA treatment, which indicates that the ERK/MAPK and CaMK pathways may be involved in regulating CREB phosphorylation, but not in the translocation of CRTC1. In support of these findings, it has already been demonstrated that the phosphorylation of CREB at serine 133, 142, and 143 is mediated by ERK/MAPK or CaMKs, which were induced by depolarization (Kornhauser et al. 2002). In addition, although the phosphorylation of CREB at Ser133 was shown to promote the recruitment of CBP to stimulate transcription (Chrivia et al. 1993), CBP is also phosphorylated by CaMKIV to stimulate activity-dependent transcription (Impey et al. 2002). Furthermore, we demonstrated that mutant Gal4-CREB (Ser133Ala) partially prevented GABA-induced CREB-dependent transcription, which supports, at least in part, the involvement of CREB phosphorylation at Ser133 in this activation. This result also suggests that other phosphorylation sites of CREB aside from Ser133 could be involved in the GABA-induced activation of CREB-dependent transcription.

Heinrich et al. (2013) recently reported that the KIX domain of CBP increased CRTC1 binding to phosphorylated CREB, which indicated that CBP could be recruited by phosphorylated CREB and, then, enhance the binding activity of CRTC1 to CREB to activate transcription. This finding suggests that the GABA-induced activation of Bdnf-pIV may be controlled by interactions between CREB, CBP, and CRTC1, which could, in turn, be controlled by ERK/MAPK, CaMK, and CN. It is noteworthy that CN was involved not only in the nuclear translocation of CRTC1, but also in controlling the phosphorylation of CREB. This suggests a dynamic equilibrium between the phosphorylation and dephosphorylation of CREB controlled by ERK/MAPK, CaMK and CN, respectively, which may contribute to interactions between CREB, CBP, and CRTC1 and explain why the activation of Bdnf-pIV was so effectively inhibited by either inhibitor. Precise molecular mechanisms have to be elucidated by further experiments.

Knockdown of NKCC1 in newly generated neurons in the hippocampal dentate gyrus was previously shown not only to decrease dendritic growth and survival, but also the expression of doublecortin, a marker of immature neurons, and NeuroD, a transcription factor involved in neuronal differentiation (Ge et al. 2006; Jagasia et al. 2009). Moreover, sustained CREB phosphorylation has been observed in immature newborn neurons in the dentate gyrus, and phosphorylation was also impaired by NKCC1 knockdown (Jagasia et al. 2009). BDNF has also been found to be involved in neurogenesis (e.g. Scharfman et al. 2005; Chan et al. 2008). These observations support a relationship between the reversible actions of GABA and BDNF expression in promoting the differentiation of neurons. Including our present study, it is clear that excitatory GABAergic stimulation of immature neurons induces Bdnf transcription activation to promote neuronal differentiation in early development of the brain or neurogenesis in the adult brain. Hong et al. (2008) directly demonstrated that the introduction of a subtle knock-in mutation into the CRE (or Ca2+-responsive element 3) of mouse Bdnf-pIV resulted in the sensory-dependent induction of Bdnf mRNA expression being disrupted in the cortex, with the formation of fewer inhibitory synapses, which suggested that the impairment in BDNF expression induced by excitatory GABAergic action could retard the development of inhibitory synapses in the developing brain. It has already been proposed that defects in the excitatory-inhibitory balance lead to the emergence of seizures and aberrant critical periods for cortical plasticity, the features of which are common between mental retardation and neurodevelopmental disorders such as autism spectrum disorder (Brooks-Kayal 2010). Thus, our present study provides new insights into understanding the molecular mechanisms of GABA-mediated neuronal development in the brain and possibly neuropsychiatric disorders associated with brain development in the early stages, or late-stage adult neurogenesis.

Acknowledgments and conflict of interest disclosure

This study was supported by JSPS KAKENHI [Grant-in-Aid for Scientific Research (B) Grant Number 20390023 (M.T.)] and the Mitsubishi Foundation (M.T.). We thank Dr. Hiroshi Takemori (National Institute of Biomedical Innovation, Japan) for providing us with the expression vector of human CRTC1 (pCMV-Sport6-human CRTC1) the anti-CRTC1 antiserum, and Dr. Jean-René Cardinaux (University of Lausanne, Switzerland) for providing us with the Flag-CREB wild-type and R314A expression vectors (pRc/RSV-FLAG-CREB wt and pRc/RSV-FLAG-CREB R314A).

All experiments were conducted in compliance with the ARRIVE guidelines. The authors have no conflicts of interest to declare.

Ancillary