Brain-derived Neurotrophic Factor (BDNF)-induced Synthesis of Early Growth Response Factor 3 (Egr3) Controls the Levels of Type A GABA Receptorα4 Subunits in Hippocampal Neurons*

Altered function of γ-aminobutyric acid type A receptors (GABAARs) in dentate granule cells of the hippocampus has been associated with temporal lobe epilepsy (TLE) in humans and in animal models of TLE. Such altered receptor function (including increased inhibition by zinc and lack of modulation by benzodiazepines) is related, in part, to changes in the mRNA levels of certain GABAAR subunits, including α4, and may play a role in epileptogenesis. The majority of GABAARs that contain α4 subunits are extra-synaptic due to lack of the γ2 subunit and presence of δ. However, it has been hypothesized that seizure activity may result in expression of synaptic receptors with altered properties driven by an increased pool of α4 subunits. Results of our previous work suggests that signaling via protein kinase C (PKC) and early growth response factor 3 (Egr3) is the plasticity trigger for aberrant α4 subunit gene (GABRA4) expression after status epilepticus. We now report that brain derived neurotrophic factor (BDNF) is the endogenous signal that induces Egr3 expression via a PKC/MAPK-dependent pathway. Taken together with the fact that blockade of tyrosine kinase (Trk) neurotrophin receptors reduces basal GABRA4 promoter activity by 50%, our findings support a role for BDNF as the mediator of Egr3-induced GABRA4 regulation in developing neurons and epilepsy and, moreover, suggest that BDNF may alter inhibitory processing in the brain by regulating the balance between phasic and tonic inhibition.


Altered function of ␥-aminobutyric acid type A receptors (GABA A Rs) in dentate granule cells of the hippocampus has been associated with temporal lobe epilepsy (TLE) in humans and in animal models of TLE. Such altered receptor function (including increased inhibition by zinc and lack of modulation by benzodiazepines) is related, in part, to changes in the mRNA levels of certain GABA A R subunits, including ␣4, and may play a role in epileptogenesis. The majority of GABA A Rs that contain
␣4 subunits are extra-synaptic due to lack of the ␥2 subunit and presence of ␦. However, it has been hypothesized that seizure activity may result in expression of synaptic receptors with altered properties driven by an increased pool of ␣4 subunits. Results of our previous work suggests that signaling via protein kinase C (PKC) and early growth response factor 3 (Egr3) is the plasticity trigger for aberrant ␣4 subunit gene (GABRA4) expression after status epilepticus. We now report that brain derived neurotrophic factor (BDNF) is the endogenous signal that induces Egr3 expression via a PKC/MAPK-dependent pathway. Taken together with the fact that blockade of tyrosine kinase (Trk) neurotrophin receptors reduces basal GABRA4 promoter activity by 50%, our findings support a role for BDNF as the mediator of Egr3-induced GABRA4 regulation in developing neurons and epilepsy and, moreover, suggest that BDNF may alter inhibitory processing in the brain by regulating the balance between phasic and tonic inhibition.

receptor (GABA A R)
is an integral ligand gated ion channel that mediates the majority of inhibition in the central nervous system. Being a hetero-oligomeric complex, it is composed of five membrane spanning subunits that are chosen from the products of 19 different genes (␣ 1-6 , ␤ 1-3 , ␥ 1-3 , ␦, ⑀, , 1-3 , and ). These genes are differentially transcribed during development and in various regions of the adult brain and spinal cord (1)(2)(3)(4)(5). Alteration in the function of GABA A Rs has been associated with a variety of diseases whose etiology leads to an imbalance between inhibition and excitation in specific populations of neurons (6 -8).
For instance, changes in certain GABA A R subunit levels occur in dentate granule cells (DGCs) of both humans with temporal lobe epilepsy (TLE) and in animal models of TLE (6,9). These molecular responses have been hypothesized to underlie persistent changes in GABA A R function associated with epileptogenesis. Most notably, individual DGCs display an elevation of ␣4 subunit mRNAs and a decrease in ␣1 (6). Receptors that contain ␣4 subunits have unique pharmacological properties that include heightened blockade of receptor function by zinc (11)(12)(13) and decreased benzodiazepine modulation (14). In addition, the majority of GABA A Rs that contain ␣4 subunits (co-assembled with a ␤ and ␦) are located extrasynaptically and mediate tonic GABA currents, while those containing ␣(1, 2, 3, or 5) without ␦ and with ␥2 are targeted to the synapse (1,15). Although the majority of synaptic receptors do not contain ␣4 subunits, ␣4 containing receptors are precipitated using ␥2 subunit antibodies in thalamic ventrobasal neurons (14,16) suggesting that they are found in a select group of synaptic receptors in neuronal subtypes where ␣4 is abundant. Assembly of an ␣, ␤, and ␥2 subunit is also necessary for the pharmacological response to classical benzodiazepines but such a response is absent when ␣4 is present.
Recent data from our laboratory shows that seizure-induced transcriptional up-regulation of the ␣4 subunit gene (GABRA4) is regulated through the protein kinase C (PKC) pathway and specifically through binding of the inducible early growth response factor 3 (Egr3) (10). A possible endogenous mediator of Egr3 signaling in TLE has yet to be identified. Because brainderived neurotrophic factor (BDNF) mRNAs and protein are elevated during seizures in TLE patients (17,18) and in several animal models of TLE (19 -21), we used cultured primary hippocampal neurons to determine whether BDNF could be the seizure-induced signal that up-regulates GABRA4 transcription and down-regulates GABRA1.
Drug Treatments-Drug treatments were performed for Western blotting, promoter analysis, and real-time PCR exper-*This work was supported by National Institutes of Health/NINDS Grant NS050393.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 These authors contributed equally to the work. 2  iments. Drugs were dissolved in dimethyl sulfoxide (Me 2 SO) or water. Final vehicle concentration was 0.5% or less for all experiments. Drugs were diluted to 20 l in warm conditioned media and added to each dish. Cultures were treated with signaling inhibitors and returned to the incubator for 1 h. Cultures were then treated with signaling activators. Following activation, cultures were harvested at variable time intervals ranging from 45 min to 24 h after treatment. For treatments with the MAPK inhibitor (MEK1/MEK2), the following final concentration was utilized: U0126 (Calbiochem, catalog number 662006, 20 M). The general PKC inhibitor calphostin C (Calbiochem, catalog number 208725) was used a final concentration of 1 M. The TrK signaling inhibitor (K252a, Calbiochem, catalog number 420298) was used as a final concentration of 200 nM. Treatment with the signaling activators phorbol 12-myristate 13-acetate (PMA) (Sigma, 1 M) and BDNF (Calbiochem, catalog number 203702, 50 ng/ml) were also utilized. Sister control dishes received vehicle (Me 2 SO or H 2 0) during the pretreatment and treatment phases.
Transient Transfection Studies-Primary cultures in 6-well dishes (Nunc) were transfected using a modified calcium phosphate precipitation method (22,23). Eight micrograms of total DNA was transfected into each well of a 6-well dish for studies that accessed reporter activity (GABRA4) after drug treatment. Vectors containing 500 bases of the 5Ј-flanking sequence specific to GABRA4 were cloned upstream of the luciferase reporter gene in the pGL2 vector (Promega) (10). Promoter fragments confers full promoter activity in primary hippocampal or neocortical neurons, respectively.
Luciferase reporter activity was monitored using the cell culture lysis Reagent and luciferase substrate (Promega) and a Victor 1420 detection system (PerkinElmer Life Sciences). Luciferase counts were normalized to protein within each dish.
Real-time PCR for Gene Expression Analysis-PCR primers were designed using primer express software (PE Biosystems). Primer sets for Egr3 and BDNF were the following sequences: Egr3, Fwd 5Ј-GAGATCCCCAGCGCGC-3Ј (forward) and 5Ј-CATCTGAGTGTAATGGGCTACCG-3Ј (reverse); Egr3 Taqman, 5Ј-CAACCTCTTCTCCGGCAGCAGTGAC-3Ј. BDNF primers were purchased from Applied Biosystems (Rn00788315_m1). RNA extraction from cells in culture were performed from single wells of a 6-well dish. Cells were harvested using RNeasy Micro RNA extraction kit (Qiagen). For samples from the adult dentate gyrus, RNA was extracted using RNeasy Midi RNA extraction kit (Qiagen). Taqman onestep RT-PCR Master Mix (Applied Biosystems) was used to generate cDNAs and to quantify mRNAs. Control probes for relative abundance of rRNA or cyclophilin were used in multiplex assays (Applied Biosystems) using the ABI7900HT (Applied Biosystems). Thermocycling conditions were as follows: 50°C (30 min) 1 cycle, 95°C (15 min) 1 cycle; 95°C (15 s), 60°C (1 min) 50 cycles. Standard curves were generated from control embryonic E18 rat brain RNA or control RNA from a control dentate gyrus. To account for loading error, mRNA values were normalized by dividing GABRA4 levels by rRNA or cyclophin levels.
Induction of Status Epilepticus-Adult rats were exposed to pilocarpine-induced status epilepticus (SE) according to standard protocol (6). Pilocarpine injection triggered long duration (Ͼ30 min) seizures within 10 -30 min after injection. Rats that did not exhibit behavioral seizures of class 3 or higher on the scale of Racine (24) within 1 h of pilocarpine injection were injected with a second dose of pilocarpine (192.5 mg/kg intraperitoneal), as is standardly done in this model because equivalence is measured in seizure severity and not doses of pilocarpine. Diazepam (6 mg/kg intraperitoneal; Sigma) was administered 1 h after onset of SE to stop seizure activity and again every 2 h until rats stopped seizing completely. Control rats were treated identically to pilocarpine injected rats, except that a subconvulsive dose of pilocarpine (38.5 mg/kg) was administered.
Statistical Analysis-Experimental values were determined by expressing values as a percentage of control values defined as 100 or 0%. Statistical significance was evaluated using a 95 or 99% confidence interval or by a Mann-Whitney test.

RESULTS
Previous data from our laboratory show that Egr3 mRNA and protein levels are elevated in the dentate gyrus 24 h after pilocarpine-induced SE (10). Elevated levels result in enhanced binding of Egr3 to GABRA4 at a time when GABRA4 mRNAs are also elevated. To determine whether BDNF mRNA levels are also elevated at this time point, real-time PCR was conducted with RNA from either control animals or those animals who have undergone pilocarpine-induced SE. Consistent with several models of TLE (19 -21), BDNF mRNAs are elevated by greater than 7-fold (Fig. 1A) at 24 h after onset of SE. We have previously demonstrated that activation of PKC/MAPK signaling leads to elevated binding of Egr3 to the endogenous GABRA4 promoter in hippocampal neurons using chromatin immunoprecipitation (10). To determine whether these changes are due to elevated Egr3 expression and to determine whether BDNF can also alter Egr3 expression, cultures were treated with either BDNF or PMA for 45 min, 2 h, or 6 h. RNA lobe epilepsy; PKC, protein kinase C; Egr3, early growth response factor 3; BDNF, brain-derived neurotrophic factor; MAPK, mitogen-activated protein kinase; MEK, mitogen-activated protein kinase/extracellular signalregulated kinase kinase; PMA, phorbol 12-myristate 13-acetate; HRP, horseradish peroxidase; SE, status epilepticus; Trk, tyrosine kinase. was extracted and real-time PCR was performed. Stimulation with BDNF and PMA increases Egr3 mRNA levels at both 45 min and 2 h, with levels peaking at 45 min (Fig. 1B). Levels return to base line by 6 h after stimulation and remain at baseline for up to 24 h (data not shown). Maximal stimulation by either BDNF or PMA is blocked by the MEK inhibitor U0126 or the PKC inhibitor calphostin C (Fig. 1C).
Changes in Egr3 mRNA levels are paralleled by changes in Egr3 protein after BDNF treatment. Similar to Egr3 mRNA levels, Egr3 protein is elevated by BDNF (Fig. 2). Egr3 levels increase by almost 3-fold 1 h after treatment. Levels are maximally elevated after 2 h (5-fold) and remain elevated for at least 6 h.
BDNF can stimulate MAPK/PKC pathways (25,26), and our previous work (10) has shown that GABRA4 responds to PKC stimulation. To investigate whether BDNF can produce similar changes in ␣4 subunit levels that occur after PKC stimulation and whether such changes are accompanied by down-regulation of ␣1 as seen in the pilocarpine model (6), hippocampal cultures were treated with BDNF or vehicle for 6 or 24 h. Following BDNF treatment, cultures were then harvested and FIGURE 2. BDNF and PMA increase Egr3 levels in primary hippocampal cultures. A, cultures were treated with vehicle (Veh) or BDNF (50 ng/ml) for 1, 2, 6, or 12 h. Whole cell extracts were prepared as described under "Experimental Procedures" and resolved by SDS-PAGE under reducing conditions. Proteins were visualized using ECL following incubation with an anti-rabbit HRP-conjugated antibody. A, a representative Western blot is shown. B, Egr3 and ␤-actin levels were quantified (n ϭ 5 independent experiments) by densitometry. Data are presented as mean Ϯ S.E. and expressed as percent change with respect to normalized Egr3 levels in vehicle-treated cultures (defined as 100%). Levels of ␤-actin did not vary with any drug treatment and were used as internal control (* ϭ significantly different from control as determined by 95% confidence interval). C, cultures were treated with vehicle or PMA (1 M) for 2 h. A representative Western blot is shown. extracts were subjected to Western blot analysis. As shown in Fig. 3A, ␣4 subunit levels increase by 57% after 6 h of treatment and 120% after 24 h, while ␣1 subunit levels show no change at 6 h and decrease by 42% after 24 h. A representative Western blot is shown in Fig. 3B.
To test the hypothesis that BDNF regulates ␣4 subunits through the up-regulation of GABRA4 transcription, primary hippocampal neurons were transfected with the GABRA4 promoter luciferase reporter construct. Eighteen hours after trans-fection, neurons were pretreated with a Trk receptor signaling inhibitor (K252a) for 1 h followed by a 6-or 12-h treatment with BDNF. Treatment with BDNF increases promoter activity by more than 50% after 12 h and activation is fully reversible by K252a (Fig. 3C). K252a alone reduces promoter activity to about 50% of base line.

DISCUSSION
Changes in GABA A R subunit gene expression may play an important role in the etiology of TLE (27). In both animal models of TLE or in TLE patients, DGCs have elevated levels of GABA A Rs with distinct pharmacological properties (6,28,29). These receptors are associated with a marked increase in ␣4 and decrease in ␣1 subunit gene expression (6). We previously reported that elevated levels of ␣4 mRNAs and protein are due to binding of Egr3 to GARBA4 in cultured hippocampal neurons and that there are increased levels and increased association of Egr3 and GABRA4 in dentate gyrus of animals exposed to SE (10). Additionally, our data showed that activation of PKC signaling elevates ␣4 mRNA levels most likely through transcriptional regulation of endogenous GABRA4 by Egr3. We now report that BDNF is a likely mediator of GABRA4 expression in TLE through its stimulation of Egr3 mRNA and protein synthesis.
Previous evidence has suggested a relationship between BDNF and the modulation of GABA A R receptor function. In primary hippocampal neurons, BDNF reduces GABAergic miniature inhibitory postsynaptic currents and causes a reduction in GABA A R subunit ␣2, ␤(2,3), and ␥2 immunoreactivity (31). A role for BDNF has not been established in the regulation of GABRA4. We now show that BDNF increases ␣4 while decreasing ␣1 subunit levels in hippocampal neurons suggesting that the neurotrophin has the potential to differentially regulate the expression of extrasynaptic and synaptic GABA A Rs.
Egr3 is a member of the early growth response (Egr) transcription factor family that includes four members (Egr1-4). Our data suggest that Egr3 mRNAs and protein are elevated in primary hippocampal neurons after BDNF treatment through PKC/MAPK activation. These changes are also seen after treatment with PMA, a drug known to activate both PKC and MAPK pathways. Our observations are consistent with those in nonneuronal systems where both Egr3 mRNA levels (32) and Egr3 promoter activity (33) are activated upon treatment with PMA. Egr3 mRNA and protein levels are also elevated in several animal models of TLE (34 -36). Taken together with our data, where the elevation of BDNF mRNA levels occurs before or at a time when Egr3 levels are also elevated, evidence suggests that BDNF is a likely endogenous regulator of Egr3 in response to SE.
Elevated Egr3 levels after BDNF stimulation are also the likely mediator of increased ␣4 subunit levels in both cultured neurons and in DGCs of TLE animals. In addition to the observation that Egr3 knock-out mice lack muscle spindles, display sensory ataxia, resting tremor, and scoliosis (37), these animals also have around 50% less GABRA4 mRNAs in the hippocampus (9). Interestingly, regulation of GABRA4 by Egr3 appears to be context independent because overexpression of Egr3 increases GABRA4 mRNA levels by over 7-fold in non-neuro- FIGURE 3. BDNF regulates levels of ␣ subunit and GABRA4 promoter activity in primary hippocampal cells. Rat hippocampal cultures, 7-9 days in vitro, were treated for 6 or 24 h with BDNF (50 ng/ml) or vehicle (H 2 0). Whole cell extracts were prepared as described under "Experimental Procedures" and resolved by SDS-PAGE under reducing conditions. Proteins were visualized using ECL following incubation with an anti-rabbit HRP-conjugated antibody. A, ␣4 (or ␣1 subunit) and ␤-actin levels were quantified by densitometry. Normalized data (␣ subunit/␤-actin) are presented as mean Ϯ S.E. and expressed as percent change with respect to vehicle-treated cultures (defined as 100%). Levels of ␤-actin did not vary with BDNF treatment and were used as an internal control (* ϭ significantly different from control as determined by 95% confidence interval, n ϭ 5; ** ϭ significantly different from control as determined by 99% confidence interval, n ϭ 5). B, representative Western blot shows effect of a 6 and 24 h treatment with BDNF on ␣4, ␣1, and ␤-actin protein levels in hippocampal neurons. C, BDNF up-regulates GABRA4 promoter activity, and activation is dependent upon TrkB receptor signaling. Cultures were transfected with the GABRA4 promoter luciferase reporter construct (Ϫ471 to ϩ71). Eighteen hours after tranfection, cells were pretreated with vehicle (Me 2 SO) or the Trk signaling inhibitor K252a (200 nM) followed by a 1-, 6-, or 12-h treatment with BDNF (50 ng/ml) or vehicle. Data are presented as mean Ϯ S.E. and are expressed as percent activity from control (% control) (* ϭ significantly different from control as determined by 95% confidence interval).
nal primary myotubes (38) suggesting that Egr3 may override the neuron-specific expression of GABRA4.
BDNF has received attention as a putative regulator of epileptogenesis (39). Intrahippocampal infusion of BDNF can cause spontaneous limbic seizures (40), and one controversial hypothesis is that BDNF may contribute to epileptogensis through mossy fiber sprouting, a process whereby DGCs send glutamatergic projections to one another to form hyperexcitable circuits. With data both supporting (41) and refuting this hypothesis (42,43), a theory is evolving that associates BDNF with epileptogenesis through its regulation of inhibitory and/or excitatory neurotransmitter systems. Given the fact that Egr3 levels dramatically increase in response to BDNF, and all GABA A R subunit genes whose mRNA levels increase in response to pilocarpine-induced SE (6) contain binding sites for Egr transcription factors (5), it will be important to parcel out the relationship between BDNF, Egr3, and GABAergic neurotransmission in the central nervous system. Equally important is the study of how BDNF selectively decreases the levels of ␣1 subunits, by an Egr-independent mechanism (5), while increasing those of ␣4. Future studies will be aimed at determining the balance between these two interrelated yet distinct BDNF-induced regulatory pathways that control GABA receptor function in the nervous system.
Acknowledgments-We thank the members of our respective laboratories for providing such a wonderful collegial environment to pursue our interests. A special thank you goes to David H. Farb for his initial vision that transcriptional regulation is key to understanding inhibition in the nervous system. We thank Jay Baraban for his generous contribution of Egr3 antibody and Sabita Bandyopadhyay for the ␣4 constructs. Finally, we dedicate our efforts to all the individuals afflicted with epilepsy in the hope that someday a better understanding may provide better therapeutics to improve their lives.