Transcriptional Regulation of T-type Calcium Channel CaV3.2

Background: Expression of the T-type Ca2+-channel CaV3.2 has to be tightly regulated for proper calcium homeostasis. Results: Overexpression of the transcription factor Egr1 strongly activates the CaV3.2 promoter and can be counteracted by the repressor REST. Conclusion: Egr1 and REST “bi-directionally” regulate the CaV3.2 promoter. Significance: Our results have important implications for calcium homeostasis and dynamics in health and disease. The pore-forming Ca2+ channel subunit CaV3.2 mediates a low voltage-activated (T-type) Ca2+ current (ICaT) that contributes pivotally to neuronal and cardiac pacemaker activity. Despite the importance of tightly regulated CaV3.2 levels, the mechanisms regulating its transcriptional dynamics are not well understood. Here, we have identified two key factors that up- and down-regulate the expression of the gene encoding CaV3.2 (Cacna1h). First, we determined the promoter region and observed several stimulatory and inhibitory clusters. Furthermore, we found binding sites for the transcription factor early growth response 1 (Egr1/Zif268/Krox-24) to be highly overrepresented within the CaV3.2 promoter region. mRNA expression analyses and dual-luciferase promoter assays revealed that the CaV3.2 promoter was strongly activated by Egr1 overexpression in vitro and in vivo. Subsequent chromatin immunoprecipitation assays in NG108-15 cells and mouse hippocampi confirmed specific Egr1 binding to the CaV3.2 promoter. Congruently, whole-cell ICaT values were significantly larger after Egr1 overexpression. Intriguingly, Egr1-induced activation of the CaV3.2 promoter was effectively counteracted by the repressor element 1-silencing transcription factor (REST). Thus, Egr1 and REST can bi-directionally regulate CaV3.2 promoter activity and mRNA expression and, hence, the size of ICaT. This mechanism has critical implications for the regulation of neuronal and cardiac Ca2+ homeostasis under physiological conditions and in episodic disorders such as arrhythmias and epilepsy.

Low voltage-activated (T-type) Ca 2ϩ channels are expressed in multiple organs, including the CNS and heart (1)(2)(3)(4), where they play a key role in many cellular processes, such as shaping of neuronal discharge patterns, secretion of hormones and neurotransmitters, amplification of dendritic excitatory postsynaptic potentials, maintenance of circadian rhythms, and pacing of the heart (4 -6). T-type Ca 2ϩ channels comprise a subfamily of three Ca V 3 pore-forming channel subunits (Ca V 3.1, Ca V 3.2, and Ca V 3.3), encoded by members of the Cacna1 gene family (Cacna1g, Cacna1h, and Cacna1i). The common biophysical characteristics of these three Ca V 3 channel subtypes are activation at subthreshold voltages, comparatively slow activation and deactivation kinetics, and complete inactivation during a sustained depolarization (7). In addition to their common characteristics, the Ca V 3 channels also exhibit diverging properties. Ca V 3.3 channels display particularly slow inactivation kinetics, and Ca V 3.2 channels are significantly more sensitive to nickel than Ca V 3.1 and Ca V 3.3 (6, 8 -10).
The importance of T-type Ca 2ϩ channels for normal cellular function is underscored by the pathophysiological alterations associated with genetic and acquired Ca V 3.2 "channelopathies." Gain-of-function mutations in the Ca V 3.2 gene are associated with idiopathic generalized/absence epilepsy (11)(12)(13). Likewise, acquired increases in thalamic and hippocampal Ca V 3.2 expression contribute to the development of chronic epilepsy (14,15). In addition, overexpression of Ca V 3.2 channels in myocytes may result in the development of several cardiac dysfunctions, including ventricular arrhythmias (16,17). Epileptic seizures as well as cardiac arrhythmias share the episodic onset of symptoms. Because no overlapping mutations for both conditions have been identified in Ca V 3.2, transcriptionally mediated changes in Ca V 3.2 levels might constitute an attractive mechanism explaining the common episodic onset. However, despite the importance and potency of transcriptional regulation, only little is known about the key mechanisms controlling expression of the Ca V 3.2 gene. Recently, an intriguing transcriptional mechanism of T-type Ca 2ϩ channel regulation has been described. Repressor element-1 (RE-1) 2 -silencing transcription factor (REST, also known as NRSF (neuron-restrictive silencer factor)) was found to function as a transcriptional regulator of Ca V 3.2 in the heart of mice (16,18). REST was originally described as a repressor of neuronal gene expression and can bind to a neuron-restrictive silencer element in the genome, also known as RE-1. Although its levels are generally low, neuronal REST expression is up-regulated after extended periods of neuronal hyperactivity, as demonstrated after seizures, neuropathic pain, and ischemia (19 -22).
Here, we have used bioinformatic and molecular approaches to characterize the Ca V 3.2 promoter in detail and to identify potential mechanisms regulating Ca V 3.2 transcription. Our analyses show for the first time that the transcription factor early growth response 1 (Egr1/Zif268/Krox-24) mediates Ca V 3.2 promoter activation. Moreover, this effect of Egr1 is potently antagonized by the transcriptional repressor REST. The functional interactions described here may have important implications for Ca V 3.2 regulation under physio-and pathological conditions.

EXPERIMENTAL PROCEDURES
Bioinformatic Analysis-The genomic sequence of the rat Ca V 3.2 gene was obtained from the UCSC genome browser. Potential transcription start sites were identified using the Eponine software (threshold value of 0.99) (23). Comparative analysis of the nucleotides of the Ca V 3.2 gene of different species was performed with PhyloP (PHAST package) and Vector NTI (9.0) using default parameters. Potential transcription factor (TF) binding sites were identified using the MathInspector RegionMiner software tool (Genomatix).
Cloning and Plasmids-The mammalian expression vectors pCMV-Egr1, pCMV-myc-REST and pCMV-FLAG-NLS-REST DBD were kindly provided by Prof. Gerald Thiel (University of Saarland Medical Center, Homburg, Germany).
Cell Culture, Transfections, and Luciferase Assays-NG108-15 cells were maintained at 37°C and 5% CO 2 in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) heat-inactivated fetal calf serum (Hyclone), 100 units/ml penicillin/streptomycin, 2 mM glutamine, and 1ϫ HAT (sodium hypoxanthine, aminopterin, and thymidine; Invitrogen). Transfection was performed in 48-well tissue culture plates (80% confluency) using Lipofectamine (Invitrogen) following the manufacturer's protocol. Briefly, 0.05 g of Ca V 3.2 luciferase reporter plasmid with firefly luciferase and 0.0125 g of control pRL-TK vector with the Renilla luciferase gene (Promega) together with the amount of overexpression plasmids as indicated were mixed with 25 l of Opti-MEM medium (Invitrogen). The mixture was incubated for 20 min at room temperature and then added to the appropriate wells. Cells were grown in serum-free culture medium at 37°C and 5% CO 2 . After 12 h, the serum-free medium was replaced by serum containing medium. The cells were collected 48 h after transfection. The luciferase assay was performed using the Dual Luciferase Reporter Assay System (Promega) according to the manufacturer's specifications. Renilla and firefly luciferase activities were determined using the Glomax Luminometer (Promega). The results are given as firefly/Renilla relative light units.
Quantitative Real Time RT-PCR-Transcript quantification was performed by quantitative real time RT-PCR analysis according to the ⌬⌬C t method. ␤-Actin was amplified from all samples to normalize expression. Quantitative RT-PCR was performed in a 6.25-l reaction volume containing 3.125 l of Maxima SYBR Green/Rox qPCR Master Mix (Fermentas), 1.5 l of diethyl pyrocarbonate H 2 O, 1.25 l of cDNA, and 0.1875 l of each primer (10 pmol/ml; Ca V 3.2 and Egr1, same primers as described above; ␤-actin forward, 5Ј-CGT GAA AAG ATG ACC CAG ATC A-3Ј; ␤-actin reverse, 5Ј-GGA CAG CAC AGC CTG GAT G-3Ј). Reactions were performed in triplicate. After preincubation for 10 min at 94°C, 40 PCR cycles (20 s at 94°C, 30 s at 59°C, and 40 s at 72°C) were performed on an ABI Prism 9700HT system (PE Applied Biosystems, Foster City, CA).
Chromatin Immunoprecipitation (ChIP) Assays-ChIP on Cultured Cells-NG108-15 cells (6 wells; 80% confluency) were transiently transfected with pCMV-Egr1 or the empty pCMV vector (0.4 g/well) using Lipofectamine as described above. 48 h after transfection the cells were cross-linked in DMEM with 1% formaldehyde for 10 min at 37°C. Cells were washed twice in cold PBS containing protease inhibitors (PIs; Complete Protease Inhibitor Mixture Tablets; Roche Applied Science) and collected into conical tubes. Cells were spun down and lysed in 200 l of SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris, pH 8.1 with PIs) and incubated on ice for 10 min.
ChIP on Brain Tissue-Mice were decapitated under deep isoflurane anesthesia (Forene). Hippocampi were removed quickly, snap-frozen, and stored at Ϫ80°C until further processing. 1% Formaldehyde was added to the tissue (200 l/hippocampus), and the tissues were incubated for 10 min at 37°C. Next, hippocampi were washed twice in cold PBS with PIs, suspended in 200 l of SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris, pH 8.1, with PIs), and incubated on ice for 10 min.
Electrophysiology Recordings and Analysis-Patch clamp recordings were obtained from NG108-15 cells. Patch pipettes with a resistance of 3-4 megaohms were fabricated from borosilicate glass capillaries and filled with an intracellular solution containing 110 mM CsF, 20 mM tetraethylammonium, 2 mM MgCl 2 , 10 mM HEPES, 11 mM EGTA, 5 mM ATP, and 0.5 mM GTP, pH 7.2, adjusted with CsOH, 300 mosmol. Patch clamp recordings were performed in an artificial cerebrospinal fluid (ACSF) bath solution containing 125 mM sodium methanesulfonate, 3 mM KCl, 1 mM MgCl 2 , 5 mM CaCl 2 , 4 mM 4-aminopyridine, 20 mM tetraethylammonium, 10 mM HEPES, 10 mM glucose (pH 7.4, 315 mosmol). Tight-seal, whole-cell recordings were obtained at room temperature (21-24°C) according to standard techniques. Membrane currents were recorded using a patch clamp amplifier (Axopatch 200B, Axon Instruments, Union City, CA). Series resistance compensation was employed to improve the voltage-clamp control (Ͼ80%) so that the maximal residual voltage error did not exceed 1.5 mV. Voltage recordings were corrected online for a liquid junction potential of 10 mV. Whole-cell Ca 2ϩ currents were elicited with depolarizing voltage steps to Ϫ10 mV. The magnitude of I CaT was quantified as the transient component of the resulting current traces. The recorded current can be attributed to Ca V 3.2 as they were largely blocked by application of 100 M Ni 2ϩ .
Viral Vector Production-Recombinant AAV1/2 genomes were generated by large scale triple transfection of HEK293 cells. The adeno-associated virus (AAV)-Syn-Egr1-IRES-Venus plasmid, helper plasmids encoding rep and cap genes (pRV1 and pH21), and adenoviral helper pF⌬6 (Stratagene) were transfected using standard CaPO 4 transfection. Cells were harvested ϳ60 h after transfection. Cell pellets were lysed in the presence of 0.5% sodium deoxycholate (Sigma) and 50 units/ml Benzonase endonuclease (Sigma). rAAV viral particles were purified from the cell lysate by HiTrap TM heparin column purification (GE Healthcare) and then concentrated using Amicon Ultra Centrifugal Filters (Millipore) until a final stock volume of 400 l was reached. Purity of the viruses was validated by Coomassie Blue staining of SDS-polyacrylamide gels loaded with 7-15 l virus stock.
Infusion of AAV vectors-Adult Ca V 3.2 ϩ/ϩ mice (Ͼ60 days old; weight Ͼ20 g) were anesthetized with 6 mg/kg xylazine (Rompun; Bayer) plus 90 -120 mg/kg ketamine, intramuscular (Ketavet; Pfizer). Intracerebral injection of viral particles in the left CA1 hippocampal region was performed stereotactically at the coordinates Ϫ2 posterior, Ϫ2 lateral, and 1.7 ventral relative to bregma. Holes the size of the injection needle were drilled into the skull, and 1 l of viral suspension containing ϳ10 8 transducing units was injected using a 10-l Hamilton syringe at a rate of 100 nl/min using a microprocessor controlled mini-pump (World Precision Instruments). After injection, the needle was left in place for 5 min before withdrawal. 14 Days after infection, mice were decapitated under deep isoflurane anesthesia (Forene), and hippocampi were removed. All experiments were performed in accordance with the guidelines of the University of Bonn Medical Center Animal Care Committee.
Statistical Analysis-Student's t tests and one-way ANOVA followed by Bonferroni's multiple comparison tests were used to evaluate the statistical significance of the results. Values were considered significantly at p Ͻ 0.05. All results are plotted as the mean Ϯ S.E.

Bioinformatic Prediction of Ca V 3.2 Promoter Region-To
determine key molecular mechanisms underlying Ca V 3.2 expression control, we first aimed to identify the Ca V 3.2 promoter region using bioinformatics. The sequence upstream of the translation start site of the rat Ca V 3.2 gene was analyzed for the presence of promoter region characteristics including transcription start sites and a high level of conservation between species. By using the Eponine software tool, four transcription start sites were found within the upstream region of the Ca V 3.2 gene, located 183, 467, 695, and 1064 bp upstream of the start ATG (Fig. 1). Furthermore, conservation analysis of the upstream Ca V 3.2 gene showed a high degree of homology throughout the first exon (80% identity between rat, mouse, and human) and within the first 700 bp upstream of the start ATG (Ͼ65% identity). A gradual decrease in sequence homology was observed for the more upstream sequences, with a sequence homology of less than 60% 1400 bp upstream of the start ATG ( Fig. 1). Based on our bioinformatic analysis, we hypothesized that the 1400 nucleotides upstream of the start ATG contain the major regulatory promoter elements.
In Vitro Delineation of Ca V 3.2 Promoter Region-We next examined whether the bioinformatically predicted Ca V 3.2 promoter region is indeed sufficient for basal activity in neuronal cells. For this, we selected NG108-15 neural cells, which express Ca V 3.2 mRNA under naïve conditions ( Fig. 2A). We cloned the predicted full-length rat Ca V 3.2 promoter region (Ca V 3.2Ϫ1426) into a firefly luciferase reporter plasmid and measured reporter activity in transiently transfected NG108-15 cells. Luciferase activity of the Ca V 3.2 promoter was ϳ8-fold higher than the pGL3 control plasmid, which lacks a promoter (Fig. 2B), suggesting that this region of the Ca V 3.2 gene has significant promoter activity.
To pinpoint the exact region responsible for Ca V 3.2 promoter activation and to identify potential stimulatory and inhibitory regions, NG108-15 cells were transiently transfected with Ca V 3.2 deletion reporter constructs (Ca V 3.2Ϫ1188, Ϫ1020, Ϫ947, Ϫ312, Ϫ280, and Ϫ105). Each deletion fragment was tested multiple times with either three or four wells per construct and using two independent DNA isolations per construct. The basal activity of the first deletion fragment (Ca V 3.2Ϫ1188) was significantly lower (p Յ 0.001) than the activity of the full-length Ca V 3.2Ϫ1426 construct (Fig. 2C), suggesting the presence of stimulatory elements in the 238 nucleotides upstream of the Ca V 3.2Ϫ1188 fragment. The activity of the second deletion fragment (Ca V 3.2Ϫ1020) was as high as that of full-length Ca V 3.2Ϫ1426, indicating the presence of an unknown repressor(s) in the 168 nucleotides upstream of the Ca V 3.2Ϫ1020 fragment. Moreover, the two subsequent deletion fragments (Ca V 3.2Ϫ947 and Ca V 3.2Ϫ312) again showed a reduced basal activity, comparable with the basal activity of the Ca V 3.2Ϫ1188 fragment, whereas the basal activity of the two smallest deletion fragments (Ca V 3.2Ϫ280 and Ca V 3.2Ϫ105) was again at the level of the basal activity of the full-length construct (Ca V 3.2Ϫ1426). These results indicate that several stimulatory and inhibitory regulatory elements are spread throughout the entire predicted Ca V 3.2 promoter region and that in vitro all seven Ca V 3.2 deletion fragments exhibit promoter activity (Fig. 2D). Furthermore, the strong promoter activity of the smallest Ca V 3.2Ϫ105 fragment suggests that this fragment functions as the Ca V 3.2 core promoter.
Identification of Putative Transcription Factor Binding Sites in Ca V 3.2 Promoter Region-To identify potential regulatory mechanisms underlying Ca V 3.2 expression, we first aligned the bioinformatically predicted rat Ca V 3.2 promoter region with genomic Ca V 3.2 sequences of mouse and human. The homologous sequences of rat (1464 nucleotides upstream from the start ATG), mouse (1468 nucleotides), and human (1597 nucleotides) were analyzed for the presence of enriched TF binding sites using the Genomatix RegionMiner software tool. TF binding sites were ranked based on their overrepresentation value calculated against either the whole genome (Z-score genome) and all annotated promoter regions of the genome (Z-score promoter). For the predicted rat Ca V 3.2 promoter region, 255 different TF matrices were found with the highest Z-score genome for Egr1 followed by zinc finger protein 161 (ZFP161 also known as ZF5) and Sp4 transcription factor (SP4) ( Table 1). The highest Z-score promoter was also found for Egr1 followed by SP4 and zinc finger protein 219 (ZNF219). For the mouse Ca V 3.2 promoter region, 239 different TF matrices were found with again the highest Z-score genome for Egr1 followed by ZF5 and brain and reproductive organ-expressed (BRE), whereas SP4, Egr1, and SP1 were found with the highest Z-scores calculated against all promoter regions. Finally, 203 TF matrices were found in the human Ca V 3.2 promoter region with nuclear respiratory factor 1 (NRF1), ZF5, and SP1 having the highest Z-score genome and SP1 (with two different matrices) and Egr1 with the highest Z-score promoter. Combining the overrepresentation values of rat, mouse, and human, we noticed that Egr1 was represented in the top Z-score lists of all three species (Table 1), pointing to a critical role of Egr1 for Ca V 3.2 promoter regulation.
Additional bioinformatic analysis of the Ca V 3.2 chromosomal region revealed that the overrepresentation of Egr1 was restricted to the evolutionary conserved promoter region (supplemental Fig. 1). Only one additional Egr1 binding site was found more than 1000 base pairs upstream of the identified Ca V 3.2 promoter region. Therefore, we decided to first examine if the cluster of Egr1 binding sites present in the conserved Ca V 3.2 promoter region mediates stimulation by Egr1.
Egr1 Strongly Activates Ca V 3.2 Promoter-To determine whether Egr1 actively regulates Ca V 3.2 promoter activity, an expression vector for Egr1 (24) was transfected into NG108-15 cells. Although NG108-15 cells express Egr1 constitutively (Fig.  3A), quantitative RT-PCR revealed significantly higher levels of Ca V 3.2 mRNA levels in Egr1-overexpressing NG108-15 cells (1.6-fold up-regulation; p ϭ 0.0054; Fig. 3B). In addition, overexpression of Egr1 also induced a robust up-regulation of luciferase activity of the full-length Ca V 3.2Ϫ1426 reporter construct (p Յ 0.001), whereas no up-regulation was observed for the promoterless pGL3 control plasmid (Fig. 3C). These results show that the cluster of Egr1 binding sites present in the 1426-bp fragment is sufficient to mediate stimulation of Ca V 3.2 transcription by Egr1.
Egr1 expression in cells is highly dynamic and can be altered by various stimuli. To examine the relationship between Egr1 levels in the cell and the transcriptional activity of the Ca V 3.2 promoter, we transfected NG108-15 cells with increasing amounts of Egr1 (2.5-75 ng/well) and determined the respective Ca V 3.2 luciferase activity. We observed that Ca V 3.2 activity was augmented as a function of gradually increasing Egr1 concentrations (Fig. 3D), with a plateau phase reached at ϳ20 ng/well Egr1. Higher amounts of Egr1 resulted in a decrease in Ca V 3.2 promoter activity, indicating a saturation of Egr1-induced Ca V 3.2 promoter activity at ϳ20 ng/well.

Egr1 Increases Functional Expression of T-type Ca 2ϩ
Channels-To examine whether increased neuronal Egr1 augments T-type Ca 2ϩ channels on a functional level, NG108-15 cells were transfected with Egr1, and Ca 2ϩ currents were recorded in whole cell condition. Ca 2ϩ currents were isolated pharmacologically by blocking Na ϩ and K ϩ conductance (see "Experimental Procedures"). From a holding potential of Ϫ80 mV, Ca 2ϩ currents were elicited by a voltage step to Ϫ10 mV (Fig. 3E). In Egr1-transfected cells, T-type currents were strongly increased (Fig. 3E, compare left and right current  traces). On average, the magnitude of T-type currents was increased more than 2-fold in Egr1-transfected cells (Fig. 3F, 35.5 Ϯ 4.5 pA, n ϭ 13, versus 109.3 Ϯ 14.2 pA, n ϭ 13, in control and Egr1 transfected neurons respectively, p Յ 0.001), indicating a major functional role for Egr1 in regulating I CaT . The recorded currents can be attributed to Ca V 3.2, as they were largely blocked by application of 100 M Ni 2ϩ (Fig. 3F, gray traces, n ϭ 6 versus n ϭ 5 in control and Egr1-transfected neurons).
Specific Binding of Egr1 to Ca V 3.2 Promoter-To determine the region of the Ca V 3.2 promoter involved in the strong Egr1mediated up-regulation, Egr1 was cotransfected with the Ca V 3.2 reporter deletion constructs. Overexpression of Egr1 showed a gradual decrease in luciferase activity for the deletion fragments, with the largest luciferase activity for the full-length Ca V 3.2Ϫ1426 reporter fragment (10-fold up-regulation; p Յ 0.001) and the lowest luciferase activity for the smallest Ca V 3.2Ϫ105 fragment (1.4-fold up-regulation; p ϭ 0.0026) (Fig. 4A). These results imply that Egr1 effectively up-regulates promoter activity of Ca V 3.2 reporter constructs mainly via sequences located more upstream in the Ca V 3.2 promoter region.
To further support our hypothesis that the Ca V 3.2 promoter could be a target of Egr1 and to further pinpoint the region responsible for Egr1-induced Ca V 3.2 up-regulation, we performed ChIP experiments. 18 putative Egr1 responsive elements (EREs) are located within the Ca V 3.2 promoter region (supplemental Fig. 1 and 2 and Fig. 4B), of which only the first two EREs are located within the full-length Ca V 3.2Ϫ1426 reporter construct. Because the up-regulation in luciferase activity after Egr1 stimulation was significantly different between full-length Ca V 3.2Ϫ1426 and Ca V 3.2Ϫ1188 (Fig. 4A), we hypothesized that the two most upstream EREs have the highest binding efficacy. For the ChIP experiments, NG108-15 cells were transfected with Egr1 and compared with empty vector-transfected control cells. We examined binding of Egr1 to the Ca V 3.2 promoter by using an Egr1-specific antibody. A rabbit-IgG no-antibody reaction served as a negative control, and three different primer pairs were used to cover the Ca V 3.2 promoter region. Intriguingly, the most upstream ChIP fragment 3 of the Ca V 3.2 promoter (containing the two predicted upstream EREs; Fig. 4B) was strongly enriched after immunoprecipitation with anti-Egr1 (Fig. 4, C and D). No significant difference was observed for the ChIP fragments 1 and 2. These observations indicate that Egr1 binds to the Ca V 3.2 promoter in neural cells and suggest that Egr1 overexpression leads to Ca V 3.2 promoter activation by the use of the upstream EREs.
REST Binds Ca V 3.2 Gene in NG108-15 Cells-Next, we examined whether Egr1-induced up-regulation of Ca V 3.2 can be counteracted by inhibitory elements located in the Ca V 3.2 gene. One repressor element known to be involved in Ca V 3.2 transcriptional regulation is REST. The Ca V 3.2 gene contains a highly conserved binding site for REST (RE-1) in its first intron (Fig. 5A), and this binding site has been reported capable of effectively binding REST (16,18). To investigate whether Egr1induced up-regulation of Ca V 3.2 could be counteracted by REST, we first analyzed NG108-15 cells for their endogenous REST mRNA expression. RT-PCR analysis revealed a clear band for the NG108-15 sample using primers against fulllength REST and the truncated REST4 variant (Fig. 5B), indicating sufficient REST expression in the neuronal cell line.
Next, we cloned the Ca V 3.2 RE-1 sequence downstream of the full-length Ca V 3.2Ϫ1426 luciferase reporter construct (Fig.  5C) and compared basal activity of the Ca V 3.2Ϫ1426-REST reporter gene with the basal activity of Ca V 3.2Ϫ1426. We found no difference in basal activity for the two reporter constructs (Fig. 5D). In addition, overexpression of REST (25) or a dominant-negative variant of REST (RESTdN (26)) did not show an effect on the basal activity of any of the two reporter genes (Fig. 5E), suggesting no significant effect of REST on the Ca V 3.2 promoter fragment under study in naïve NG108-15 cells.
To investigate whether REST indeed binds the RE-1 site of the Ca V 3.2 gene, ChIP experiments on NG108-15 lysates were performed using a REST antibody. In both basal and RESToverexpressing NG108-15 cells, PCR amplicons were obtained, indicating efficient binding of REST to the Ca V 3.2 gene (Fig.  5F). Nevertheless, no difference in binding efficiency was observed between immunoprecipitates generated from basal NG108-15 lysates and lysates from REST-overexpressing cells (Fig. 5G). Thus, under unstimulated conditions, up-regulation of REST has no effect on Ca V 3.2 promoter binding and Ca V 3.2 expression.
REST Potently Counteracts Egr1-induced Ca V 3.2 Activation-Next, we examined the effect of increasing REST concentrations on the Ca V 3.2 promoter activity of Egr1-stimulated NG108-15 cells. Cotransfection of Egr1 (25 ng/well) and REST resulted in a significant repression of the Ca V 3.2Ϫ1426-REST reporter gene when cells were treated with more than 100 ng/well REST (Fig.  6A). No down-regulation of Egr1-induced Ca V 3.2 promoter activity was observed in transfected cells harboring Ca V 3.2Ϫ1426 lacking a REST binding site (Fig. 6B). In addition, RESTdN did not have an effect on any of the two promoter constructs. Intriguingly, these results indicate that recruitment of REST to the RE-1 site of the Ca V 3.2 promoter effectively represses Egr1-induced Ca V 3.2 expression.
Egr1 and REST Bind Ca V 3.2 Promoter in Vivo-To analyze whether the above described effects in NG108-15 cells have any physiological relevance in vivo, ChIP analyses using anti-Egr1 and anti-REST antibodies were carried out on brain tissues. Because expression levels of the two target genes Egr1 and REST are relatively high in hippocampal tissue (Figs. 3A and 5B), we selected mouse hippocampi for our experiments. Anti-Egr1 and anti-REST hippocampal immunoprecipitates were analyzed for their binding to the Ca V 3.2 gene. Primers specific to the Ca V 3.2 promoter region (Fig. 4B) yielded PCR amplicons from the anti-Egr1 chromatin (Fig. 7A). In addition, PCR amplicons were also obtained from anti-REST immunoprecipitates when using primers specific for the RE-1 sequence of the Ca V 3.2 gene (Fig. 7A). Hence, consistent with the finding that FIGURE 4. Egr1 binds the upstream Ca V 3.2 promoter. A, luciferase activity of the Ca V 3.2 promoter deletion fragments after overexpression with Egr1 is shown. Deletion of the upstream nucleotides of the Ca V 3.2 promoter region gradually decreases the luciferase activity after Egr1 stimulation (one-way ANOVA: *, p Յ 0.05; **, p Յ 0.01; ***, p Յ 0.001; n Ն 3). B, shown is a schematic representation of the rat Ca V 3.2 promoter and the amplicons used in the ChIP assay. Vertical black bars represent predicted Egr1 binding sites; vertical dashed lines represent the borders of the Ca V 3.2 promoter deletion fragments. Three ChIP PCR assays were designed spanning the Ca V 3.2 promoter region (ChIP1, -2, and -3). C, PCR analysis of the three ChIP products is shown. Binding efficiency of Egr1 to the Ca V 3.2 promoter region was determined in empty vector (basal) and Egr1-overexpressing (Egr1) cells. D, quantification of the ChIP experiment under the two conditions indicated substantial binding of Egr1 at the Ca V 3.2 promoter region covered by ChIP3 (t test: ***, p Յ 0.001; n ϭ 4).
Egr1 and REST can bind the Ca V 3.2 gene in NG108-15 cells, ChIP analysis of mouse hippocampi revealed that Egr1 and REST also bind the Ca V 3.2 gene in vivo.
Overexpression of Egr1 Increases Ca V 3.2 Expression in Vivo-Finally, we investigated whether the Egr1-induced up-regulation of Ca V 3.2 also occurs in vivo. Egr1 overexpression was accomplished by stereotaxical delivery of an AAV encoding the Egr1 protein in the hippocampus of adult mice. Two weeks after injection, hippocampal Ca V 3.2 and Egr1 mRNA expression levels were measured. We observed a significant up-regulation of Egr1 mRNA expression after rAAV-Egr1 transduction, indicating efficient rAAV infection in the hip-pocampus (Fig. 7B, left panel). Intriguingly, a significant upregulation of Ca V 3.2 expression was also observed after infection with rAAV-Egr1 (Fig. 7B, right panel). Collectively, these data indicate that Egr1 can increase Ca V 3.2 expression not only in cultured cells but also in brain tissue.

DISCUSSION
Here, we have defined a regulatory element in the upstream Ca V 3.2 promoter that mediates activation of Ca V 3.2 transcription by Egr1. Stimulation of the Ca V 3.2 promoter by Egr1 thereby leads to an increase of I CaT . Furthermore, we observed that Egr1-mediated promoter activation  can be effectively counteracted by binding of the transcriptional repressor REST to the Ca V 3.2 gene (Fig. 8). In contrast to short term modulatory effects on I CaT (for review, see Ref. 27), the mechanisms, we observed here, are well suited for prolonged dynamic regulation of neuronal and cardiac Ca 2ϩ -homeostasis and discharge behavior.
Our bioinformatic analyses revealed a striking accumulation of adjacent binding sites for Egr1 in the upstream Ca V 3.2 promoter region. Such "homotypic" TF binding clusters are a widespread genomic feature of higher eukaryotes (28) and may be utilized to control gene expression via sophisticated regulatory mechanisms such as high affinity cooperative binding of the corresponding TF (29). In general, cooperative TF binding can be translated into on/off transcriptional responses, regulating the functional state of the corresponding gene, namely, active or inactive. In contrast, non-cooperative TF binding does not switch between a digital (on/off) transcriptional response but results in a more gradual transcriptional activation (30). Our ChIP data reveal Egr1 binding to all Egr1 binding sites of the Ca V 3.2 promoter under basal conditions, indicating a cooperative mechanism of Ca V 3.2 transcriptional regulation. However, in the presence of increased Egr1 levels, augmented Egr1 binding occurred only at the two most upstream Egr1 binding sites, suggesting these binding sites to be of importance for strong stimulus-induced Ca V 3.2 up-regulation.
Egr1 is a zinc finger transcription factor and belongs to a larger family of early response genes that also include Egr2, Egr3, Egr4, and Wilms tumor 1 (Wt1). Egr1 is rapidly and transiently induced by a variety of stimuli, including serum, growth factors, mechanical injury, stress, and ischemia and has important roles in the regulation of cell growth, differentiation, development, and apoptosis (31,32). Upon induction, Egr1 can bind ERE consensus sequences to regulate expression of downstream target genes, such as fibronectin (Fn1), fibroblast growth factor 2 (Fgf2), synapsin I (Syn1), transforming growth factor, ␤1 (Tgf␤1), phosphatase and tensin homolog (Pten), and p53 (for review, see Ref. 33). Furthermore, Egr1 has been reported to activate NGFI-A-binding protein 1 and 2 (Nab1 and Nab2). Interestingly, both proteins also appear to be important regulators of Egr1-mediated transactivation (34 -36). By binding to Egr1, Nab1 and Nab2 can strongly inhibit Egr1 activity. Nab1 and Nab2 are also expressed in NG108-15 cells, but only Nab2 is significantly up-regulated after Egr1 stimulation. 3 Congruently, Egr1 can regulate the induction of its own repressor Nab2, and can thus control its own transactivation efficiency in NG108-15 cells. This negative feedback loop provides a potential explanation for the saturation of Egr1induced Ca V 3.2 up-regulation when increasing levels of Egr1 are transfected (Fig. 3D).
Previous studies have shown that Egr1 and Sp1, for which there are also several potential binding sites within the Ca V 3.2 promoter (Table 1), can generally compete for overlapping binding motifs (37,38). Sp1 usually activates target promoter sequences but gives complex responses when in the presence of Egr-1 (39). Sp1 is also expressed in NG108-15 cells. 3 Therefore, Sp1 and Egr1 could recognize and competitively bind overlapping Sp1/Egr1 consensus sequences located in the Ca V 3.2 promoter. Intriguingly, of the predicted Egr1 binding sites in the Ca V 3.2 promoter (Fig. 4B), the two most upstream Egr1 binding sites, with the highest Egr1 binding efficiency in our ChIP experiments, do not overlap with Sp1 consensus sequences (supplemental Fig. 1). This bioinformatic finding together with our data suggest that the two upstream Egr1 binding sites that are critical for a dose-dependent Ca V 3.2 promoter control by Egr1 are exclusively controlled by Egr1 and not by Sp1.
We found that the transcriptional repressor REST counteracts the stimulatory effect of Egr1 on the Ca V 3.2 promoter. The Ca V 3.2 gene contains a highly conserved RE-1 that can interact with REST. Originally, REST was identified to be important for the silencing of neuronal-specific genes in non-neuronal cells (40). Nevertheless, REST also has a functional role within the nervous system by regulating expression of several target genes, including Syn1, synaptophysin (Syp), the type II sodium channel genes (Scn2A and Scn2B), and the genes encoding the potassium channel subunits Kv7.2 and Kv7.3 (Kcnq2 and Kcnq3) (41,42). Recently a large scale chromatin immunoprecipitation assay (ChIPSeq) was performed to build a high resolution interactome map for REST (43). Here, Ca V 3.2 was identified as a REST-responsive gene as well as were other members of the voltage-dependent calcium channel subunit family (e.g. Cacna1a, Cacna1b, cacna1e and cacna2d2). In addition, many other ion channel genes were found positive for REST binding, including the sodium channels Scn3b and Scn10a, several potassium channels, and the hyperpolarization-activated cyclic nucleotide gated (Hcn) ion channel genes. In this context, REST may have a general role in coordinately regulating expression levels of ion channel proteins from different subfamilies, including the Ca V 3.2 gene.
REST did not repress the basal activity of the Ca V 3.2 promoter but strongly decreased Egr1-induced promoter activity. This observation suggests that because basal expression levels of Ca V 3.2 are low and the REST binding site in the Ca V 3.2 promoter is occupied during basal conditions (Figs. 5F and 7A), REST is involved in keeping basal Ca V 3.2 levels low. Therefore, an increase in REST levels cannot further down-regulate Ca V 3.2 expression. After Egr1 stimulation, REST binding to the RE-1 of the Ca V 3.2 gene might be relieved, resulting in augmented Ca V 3.2 expression levels. The Egr1-induced up-regulation can then only be repressed by higher REST availability. Therefore, REST may play an important role in keeping basal expression levels of Ca V 3.2 low before and after a transient activating stimulus. Transcriptional regulation by REST is even more complex due to the coexpression of REST4 (Fig. 5B). REST4, the truncated variant of REST that lacks the C-terminal zinc finger repressor domain can antagonize the action of fulllength REST (44) and might thus interfere with REST-induced changes.
Transient transcriptional alterations of Ca V 3.2 relate to disorders with episodic onset of symptoms such as epileptic seizures and cardiac arrhythmias. The latter has been demonstrated to relate to the modulation of Ca V 3.2 transcription by REST (16,18). Interestingly, highly dynamic changes of Ca V 3.2 expression have been observed in epileptogenesis after brain insults in thalamic as well as hippocampal principal neurons (14,15). In hippocampal CA1 pyramidal cells, Ca V 3.2 mRNA is transiently up-regulated early in epileptogenesis that is triggered by an episode of status epilepticus (SE), leading to an increase in the propensity for intrinsic burst-firing (15,45). Egr1 has been suggested as a critical factor for the establishment of long term neuronal plasticity in the hippocampal formation (46 -48). Intriguingly, Egr1 expression is also strongly increased in hippocampal neurons after SE (49,50). In addition, human twin studies showed a dysregulation of Egr1 mRNA expression in idiopathic absence epilepsies (51). In conjunction with our present data, these studies suggest that Egr1- Subsequent addition of REST can repress the activated Ca V 3.2 promoter (repressed state). REST overexpression alone does not influence Ca V 3.2 promoter activity (no effect). NRSE, neuron-restrictive silencer element. mediated transcriptional up-regulation of Ca V 3.2 may be a mechanism generalizable to a number of CNS disorders.
Notably, Ca V 3.2 mRNA peaks only transiently for 2 days after SE in CA1 pyramidal cells and afterward sharply returns to base-line levels (15). Neuronal REST expression is significantly up-regulated after global ischemia and epileptic insults (19 -21, 52, 53). Extrapolating our present in vitro data to this condition, we suggest that augmented REST levels may be involved in Ca V 3.2 repression 2 days after SE. A similar regulation by REST has been described for the hyperpolarization-activated cyclic nucleotide gated ion channel Hcn1 after kainic acid-induced SE. The SE-induced up-regulation of REST represses Hcn1 expression and the corresponding Hcn1-mediated currents (I h ) (53). REST, therefore, may play the role of a "central switch" in SE-induced channelopathies. The fact that strong expression of REST and its binding to the RE-1 of the Ca V 3.2 gene counteracts Egr1-induced Ca V 3.2 activation but does not interfere with the "basal" activity of the Ca V 3.2 promoter raises the intriguing possibility that REST up-regulation may sharpen the temporal profile of Ca V 3.2 up-regulation. This will depend to a great extent on the precise timing of Egr1 and subsequent REST induction. Further studies will be needed to analyze the effect of increased REST levels in vivo, e.g. after SE, on the Ca V 3.2 transcriptional activity.
Interfering with a transcriptional complex that alters transcription of multiple disease-relevant genes represents a potential therapeutic approach. A detailed understanding of the responsible transcriptional regulatory mechanisms will allow for specific interference strategy. Therefore, our data showing that the Ca V 3.2 promoter can be regulated by interplay of Egr1 and REST, whereby increases in cellular Egr1 activate the Ca V 3.2 promoter while REST can counteract the Egr1-induced up-regulation, represents an important step in this direction.