Activity-dependent Transcriptional Regulation of M-type (Kv7) K⁺ Channels by AKAP79/150-mediated NFAT Actions

Summary

M-type K⁺ channels, encoded by the KCNQ2-5 (Kv7) gene family, play key roles in regulation of neuronal excitability; however, less is known about the mechanisms controlling their transcriptional expression. Here, we discovered a novel mechanism regulating KCNQ2/3 transcriptional expression by neuronal activity in rodent neurons, involving activation of calcineurin and Nuclear Factor of Activated T-cells (NFAT) transcription factors, orchestrated by A-kinase-anchoring protein (AKAP)79/150. The signal requires Ca²⁺ influx through L-type Ca²⁺ channels and both local and global Ca²⁺ elevations. We postulate increased M-channel expression to act as a negative-feedback to suppress hyper-excitability of neurons, demonstrated by profoundly up-regulated KCNQ2/3 transcription in hippocampi from wild-type mice after drug-induced seizures, an effect nearly eliminated in AKAP150−/− mice. Thus, we suggest a distinct role of AKAP79/150 and the complex it organizes in activity-dependent M-channel transcription, which may potentially serve throughout the nervous system to limit over-excitability associated with disease states such as epilepsy.

Introduction

The family of A-kinase anchoring proteins (AKAPs) has emerged as a convergent point of diverse signals to achieve spatiotemporal specificity. Besides the extensive studies on its regulation of ion channel activity and trafficking, AKAP79/150 (human AKAP79/rodent AKAP150) has been shown to be intimately involved in synaptic plasticity, and learning and memory (Horne and Dell’Acqua, 2007; Lu et al., 2007; Tavalin et al., 2002; Tunquist et al., 2008; Weisenhaus et al., 2010). A direct role of AKAP79/150 in gene transcription has been implicated, highlighting nuclear or plasma membrane complexes it organizes with signaling components of cAMP/CREB or calcineurin (CaN)/nuclear factor of activated T-cells (NFAT) signaling pathways (Oliveria et al., 2007; Sample et al., 2012). NFAT transcription factors are activated by intracellular Ca²⁺ (Ca²⁺_i) signals in concert with CaN, and play critical roles in neural development, axon growth, and amyloid beta neurotoxicity (Graef et al., 1999; Graef et al., 2003; Hudry et al., 2012; Wu et al., 2012). AKAP79/150 has been suggested to play a role in the regulation of neuronal excitability and susceptibility to seizures (Tunquist et al., 2008). Recent structural and biochemical studies further reveal the stoichiometry of the core AKAP79-dimer/PKA/CaN complex, and suggest the mechanism of CaN activation by Ca²⁺/CaM binding to AKAP79, and NFAT activation by dissociation of CaN from the AKAP79 complex (Gold et al., 2011; Li et al., 2012). However, few genes in the nervous system presumed to be regulated by NFAT have actually been identified.

Voltage gated M-type (KCNQ, Kv7) K⁺ channels, expressed in a wide variety of neurons, play critical roles in modulation of neuronal excitability and action potential firing (Delmas and Brown, 2005). KCNQ2 and KCNQ3 underlie most neuronal M-currents which are partly regulated by AKAP79/150-mediated PKC phosphorylation (Hoshi et al., 2005; Hoshi et al., 2003; Zhang et al., 2011). Yet, despite the importance of M channels in control over neuronal excitability, very little is known about their transcriptional regulation, which would have profound implications for nervous function. Mechanisms of transcriptional up-regulation have not been described, but rather down-regulation by the transcriptional repressor, REST, in sensory neurons (Mucha et al., 2010; Rose et al., 2011).

In this paper, we discover a distinct role of AKAP79/150 in modulation of M-currents, by mediating activity-dependent regulation of KCNQ2-3 channel gene transcription. We examined the novel hypothesis that neuronal activity, which is regulated by M current, induces NFAT-mediated transcriptional up-regulation of the very KCNQ channels that can dampen excitability. Increased expression of KCNQ2/3 channels operates in a negative-feedback manner to suppress hyper-excitability of neurons. We show that AKAP79/150 orchestrates a signaling complex that includes CaN and L-type (Ca_V1.3) Ca²⁺ channels, the activity “reporter” of the neurons. This signaling pathway may potentially serve throughout the nervous system to limit over-excitability, which underlies myriad disorders such as chronic pains, epilepsies and cardiovascular dysfunction.

Results

Neuronal stimulation induces increased M-channel expression

We first examined whether M-channel transcription and expression in sympathetic neurons of rodent superior cervical ganglion (SCG) are regulated by neuronal stimulation, using both quantitative real-time PCR (qPCR) and patch-clamp electrophysiology. As previously reported (Hadley et al., 2003), strong KCNQ2 and KCNQ3, but little KCNQ1, transcripts express in juvenile rat SCG neurons (Fig. 1A). The relative expression levels for KCNQ1-3 transcripts, normalized by expression level of the housekeeping β-actin RNA, were 0.002 ± 0.001 *10⁻³, 1.11 ± 0.03 *10⁻³, and 0.74 ± 0.18 *10⁻³ (n=3), respectively. We then compared the effect of neuronal stimulation on the levels of KCNQ2 and KCNQ3 transcripts. Neuronal activity was mimicked by two methods, 1) depolarization by perfusion (10–15 min) of a high-K⁺ solution (50 mM), or 2) application of acetylcholine (ACh, 1 mM), which is a more physiological way to depolarize and excite the neurons via stimulation of nicotinic ACh receptors. Perfusion by only Ringer’s solution served as a negative control. Total RNA was extracted after 7 hours and qPCR performed (Fig. 1B). We detected significant increases in both KCNQ2 and KCNQ3 mRNA in neurons stimulated by 50-K⁺ or ACh (Fig. 1C). For KCNQ2, the relative expression levels in neurons treated with high-K⁺ or ACh were 2.05 ± 0.44 (n=4, p<0.05), and 1.80 ± 0.19 (n=4, p<0.05), respectively. For KCNQ3 transcripts, they were 1.76 ± 0.51 (n=4, p<0.05), and 1.56 ± 0.22 (n=4, p<0.05), respectively.

Fig. 1. Neuronal stimulation induces increased KCNQ2 and KCNQ3 mRNA and M-current amplitudes.

A, quantitative real-time (qRT)-PCR analysis of KCNQ1-KCNQ3 mRNA levels in cultured rat SCG neurons. Data are expressed as the relative quantification normalized to the expression of β-actin mRNA. B, the protocol used is shown: cultured SCG neurons were exposed to either 50 mM K⁺ or ACh (1 mM) for 15 min. After returning to normal medium for 7 h, or 2–3 d, qRT-PCR or patch-clamp analysis were performed, respectively. C, bars summarize the fold-increase (mean ± SEM) in KCNQ2 and KCNQ3 mRNA from SCG neurons after stimulation by 50 mM K⁺ or ACh. Data are normalized to the expression of β-actin RNA. D, representative M-current traces from rat SCG neurons 2–3 d after stimulation by 50-K⁺, ACh or Ringer only, using the indicated voltage protocol are shown. The grey traces are in the presence of the M-channel blocker, XE991. E, Bars summarize the current density (mean ± SEM) for the groups of cells as in D. *p<0.05, **p<0.01.

M-current (I_M) amplitudes in SCG neurons were then quantified to assay expression of functional M channels. As in the previous qPCR experiments, neurons were perfused by 50-K⁺, ACh, or regular Ringer’s solution, for 15 min, and after 1, 48, 60 or 72 hours studied under perforated-patch voltage-clamp. We did not observe a significant difference of I_M amplitudes between neurons treated with regular Ringer’s and 50-K⁺ solutions 1 hour after stimulation, but we observed significant up-regulation of I_M amplitudes in neurons treated with 50-K⁺ solution after 48, 60 or 72 hours, indicating altered expression of M-channels is involved. We thus decided to measure I_M amplitudes 48–60 hours after stimulation in this paper, and examples of I_M traces recorded from such neurons before and after application of the M-channel specific blocker, XE991 (Zaczek et al., 1998) are shown in Fig. 1D. I_M amplitudes were normalized to membrane capacitance and the current density used to indicate expression of functional M-channels. In neurons treated with 50-K⁺ or ACh-containing solutions, I_M amplitudes were significantly augmented (Fig. 1E). For neurons treated with regular Ringer’s or 50-K⁺ containing solutions, the current densities were 0.78 ± 0.10 pA/pF (n=14), and 1.24 ± 0.12 pA/pF (n=14, p<0.01), respectively. For neurons treated with only Ringer’s or ACh, the current densities were 0.81 ± 0.06 pA/pF (n=12), and 1.25 ± 0.15 pA/pF (n=14, p<0.01), respectively.

The activity-dependent regulation of M channel expression is mediated by NFAT

NFAT signaling is critical to neural development and axon growth (Graef et al., 2003), as well as transcriptional regulation of several voltage-dependent K⁺ channels, e.g., up-regulation of K_V4.2 mRNA in cardiomyocytes (Gong et al., 2006) and down-regulation of K_V2.1 mRNA in arterial smooth muscle (Amberg et al., 2004). In the rat SCG neurons that we study here, NFATc1-c4 has been shown to be expressed and, when activated, to translocate from cytoplasm to nucleus by electrical stimulation and kinase inhibitors (Hernandez-Ochoa et al., 2007). We performed qPCR on SCG neurons and detected transcripts for NFATc1-c4 isoforms (data not shown). We then asked which transcription factors mediate the up-regulation of M channel expression seen here, hypothesizing activity-dependent production of Ca²⁺/CaN and NFAT activation to be crucial.

To obtain evidence for elevated M channel expression by NFAT, we monitored I_M amplitudes after direct activation or suppression of CaN/NFAT signaling. First, we tested the effect of exogenous expression of a constitutively-active NFAT mutant (CA-NFAT) that does not require CaN to be activated. SCG neurons, transfected with either EGFP-tagged CA-NFAT or only EGFP, were treated with only Ringer’s or 50-K⁺ solutions, and I_M recorded after 48–60 hours. Successfully transfected neurons were identified by EGFP fluorescence (Fig. 2A). In CA-NFAT expressing neurons treated with only Ringer’s solution, the tonic amplitude of I_M was much larger (1.44 ± 0.22 pA/pF, n=10) than in control neurons (0.85 ± 0.09 pA/pF, n=12), with current augmentation similar to that normally seen after high-K⁺ stimulation (1.55 ± 0.18 pA/pF, n=14, p<0.01). Furthermore, high-K⁺ treatment did not further increase I_M amplitudes in CA-NFAT expressing neurons (1.63 ± 0.17 pA/pF, n=13) (Figs. 2A–B), suggesting high-K⁺ treatment induces M-channel expression through activation of NFATs. Our second test was to knock down endogenous NFAT activity using shRNA for NFATc1 or NFATc2. We transfected EGFP-tagged shRNA for NFATc1, NFATc2, or scrambled shRNA as a control, into SCG neurons from wild-type mice, which also showed increased I_M when treated with high-K⁺ solution (1.28 ± 0.08 pA/pF, n=13), compared with only Ringer’s (0.87 ± 0.05 pA/pF, n=18) (Fig. 2C, D). In NFATc1-shRNA or NFATc2-shRNA expressing neurons, the increase in I_M amplitude by 50-K⁺ was largely blunted (0.93 ± 0.10 pA/pF, n=13, and 1.03 ± 0.06 pA/pF, n=13, respectively) compared to neurons transfected with scrambled shRNA (1.39 ± 0.11 pA/pF, n=18, p<0.01) (Fig. 2C, D). These data suggest that activation of NFAT transcription factors underlies the increased expression of M current by neuronal stimulation. The shRNA data also suggest that both NFATc1 and NFATc2 activity are required.

Fig. 2. NFAT mediates the up-regulation of M-channel expression.

A, shown are representative M-current traces from rat SCG neurons stimulated by 50 mM K⁺, or Ringer only, transfected with the EGFP-tagged constitutively-active (CA)-NFAT mutant, or EGFP only. Insets are wide-field (W.F.) or fluorescent micrographs of a neuron transfected with CA-NFAT, which is tonically localized to the nucleus. B, Bars summarize the current density (mean ± SEM) for the groups of cells as in A. C, shown are representative M-current traces from mouse SCG neurons stimulated by 50 mM K⁺, or Ringer only, transfected with EGFP-tagged shRNA for NFATc1, NFATc2 or scrambled shRNA as a control. D, Bars summarize the current density (mean ± SEM) for the groups of cells as in C. *p<0.05, **p<0.01.

Neuronal stimulation elicits nuclear translocation of NFATc1/2 from the cytoplasm

Having implicated NFATc1-c2 in transcriptional regulation of M channels, we wanted to probe the relationship between Ca²⁺_i signals and NFAT translocation in real time. Thus, rat SCG neurons transfected with plasmids encoding EGFP fused to the N-terminus of NFATc1 (EGFP-NFATc1) were loaded with fura-2 AM, and EGFP localization monitored simultaneously with [Ca²⁺]_i. SCG neurons were stimulated by the 50-K⁺ solution for 10–15 min and then switched back to regular Ringer’s solution. EGFP-NFATc1 translocation was quantified by measuring mean EGFP fluorescence from regions in the cytoplasm and nucleus, and the nuclear-to-cytosolic ratio calculated, while [Ca²⁺]_i was simultaneously monitored by calculating the ratio of emitted fluorescence collected from 340 and 380 nm excitation. Images of fura-2 and EGFP emission from an example neuron, before, during and after 50-K⁺ treatment, are shown in Fig. 3A. At rest, EGFP-NFATc1 localized mostly to the cytoplasm. A rapid increase of [Ca²⁺]_i throughout the soma was observed after neurons were stimulated by depolarization with 50-K⁺. The nuclear translocation of EGFP-NFATc1 from the cytoplasm commenced much more slowly, was essentially complete with ~20 min, and lasted for at least 30 min (Fig. 3A, n=11) (Supplemental Movie 1). We performed similar simultaneous imaging of NFAT and [Ca²⁺]_i on neurons transfected with EGFP-tagged NFATc2-c4. We observed similar, rapid [Ca²⁺]_i elevations in neurons transfected with EGFP-NFATc2-c4, but only observed NFAT nuclear translocation for EGFP-NFATc2 (Figs. 3C–E, n=20, 16, 22).

Fig. 3. Neuronal stimulation elicits translocation of NFATc1/c2 from the cytoplasm to nucleus.

Rat SCG neurons were transfected with EGFP-tagged NFATc1-c4, and simultaneous live-cell fura-2 and EGFP imaging performed during and after stimulation by 50-K⁺, or ACh. Neurons were bath-loaded with fura-2 as the AM-ester. A–B, shown are the representative images of cells transfected with EGFP-NFATc1 with (A) and without (B) L-type Ca²⁺ channel agonist FPL-64716 added to the 50-K⁺ solution, with the 340/380 nm ratios and EGFP images acquired every 15 s from the experiments plotted versus time. C, shown are representative fura-2 and EGFP images of cells transfected with EGFP-tagged NFATc2-c4. D–E, bars summarize data (mean ± SEM) from groups of cells in A–C, and cells treated with ACh, for fura-2 and EGFP imaging, respectively. *p<0.05. See also Supplemental Movie 1.

In hippocampal neurons, L-type Ca²⁺ channels have been suggested as pivotal for CaN/NFAT signaling (Graef et al., 1999; Oliveria et al., 2007); however, the L-type current is only <5% of total I_Ca in rat SCG neurons (Plummer et al., 1989). Thus, we tested whether including the L-channel agonist, FPL-64716 (Baxter et al., 1993), in the 50-K⁺ solution would induce greater nuclear translocation of NFATc1. However, the absence of FPL-64716 allowed similar [Ca²⁺]_i elevations and robust, but only slightly smaller NFATc1 nuclear translocation (p<0.05) by 50-K⁺ (n=19) (Figs. 3B–E). Later in this paper, we systematically explore the subtypes of I_Ca involved in the CaN/NFAT signaling cascade. We also observed rapid [Ca²⁺]_i elevations and EGFP-NFATc1 nuclear translocation when neurons were excited using ACh (n=10, Figs. 3D–E, for the statistics, see Supplemental Material). Thus, in sympathetic neurons, neuronal activity induces nuclear translocation of NFATc1 and NFATc2 that is coupled with strong increases in [Ca²⁺]_i.

Since the responses of exogenously-expressed signaling proteins may differ from endogenous ones, we also performed experiments to test the nuclear translocation of endogenous NFAT by immunostaining/confocal microscopy. We again chose to examine the nuclear translocation of NFATc1. Cultured rat SCG neurons were treated with 50 K⁺ or ACh for 15 min, fixed and immunostained by antibodies against NFATc1 before stimulation (NS) or at 15–120 min after stimulation. Tyrosine hydroxylase (TH) was used as a sympathetic neuronal marker and DAPI was used to stain nuclei. The subcellular distribution of endogenous NFAT was visualized by confocal microscopy, and nuclear staining levels calculated as the ratio of nuclear to cytoplasmic staining (Figs. 4A–B). NFATc1, TH or DAPI images are displayed in red, green or blue, respectively, so in the merged DAPI+NFATc1 images, purple regions indicate greater NFATc1 localization to the nuclei. Consistent with the transfected EGFP-NFAT data, both types of stimulation increased endogenous NFATc1 nuclear staining within 15 min, and the augmented level of nuclear NFATc1 persisted for at least 120 min (Figs. 4C–D).

Fig. 4. Neuronal stimulation elicits nuclear translocation of endogenous NFATc1 from the cytoplasm.

A–B, shown are confocal images of rat SCG neurons immunostained with the nuclear stain DAPI, with antibodies against tyrosine hydroxylase (TH) as a sympathetic neuronal marker, or against NFATc1. Images were taken either of non-stimulated (NS) neurons, or over a range of times after 50-K⁺ or ACh stimulation. C–D, summary of these experiments. Translocation of NFATc1 was calculated as the nucleus/cytoplasm ratio.

In sum, we demonstrate that stimulation of rat sympathetic neurons by either 50-K⁺ or ACh elicits translocation of both exogenous EGFP-tagged NFATc1-c2 and endogenous NFATc1 from the cytoplasm to the nucleus, concurrent with increases in [Ca²⁺]_i.

Inhibition of CaN/NFAT signaling pathways abolishes activity-dependent up-regulation of M channel expression

To further confirm the involvement of CaN-activated NFAT in the regulation of M channel expression, endogenous NFAT signaling was inhibited pharmacologically by either the CaN inhibitor, cyclosporine A (CsA), or a stearated (St), membrane-permeable peptide, composed of MAGPHPVIVITGPHEE (St-VIVIT), that inhibits the CaN/NFAT signaling pathway by competitively blocking the binding of CaN to NFAT, preventing NFAT dephosphorylation (Aramburu et al., 1999). Cultured rat SCG neurons were pre-treated with CsA or the St-VIVIT peptide for 1 hour before stimulation. The neurons were then 1) fixed and immunostained by antibodies against NFATc1 before stimulation by 50-K⁺, or at 15–120 min after stimulation, and imaged under confocal microscopy (Fig. 5A), or 2) loaded with fura-2, stimulated by 50 K⁺ for 10–15 min and simultaneous [Ca²⁺]_i and EGFP-NFATc1 imaging performed as before (Fig. 5B). In both experiments, we found translocation of NFATc1 induced by 50-K⁺ stimulation was blocked by CsA or the St-VIVIT peptide.

Fig. 5. Inhibition of CaN/NFAT signaling pathways abolishes activity-dependent up-regulation of M channel expression.

A, confocal images of SCG neurons immunostained with the nuclear stain DAPI, and with antibodies against tyrosine hydroxylase (TH) and NFATc1. Images were taken either of non-stimulated (NS) neurons or over a range of times after 50-K⁺ stimulation in the presence of a stearated (St), membrane-permeable peptide (St-VIVIT) that competes with CaN binding to NFAT. Summary of the experiment is shown below. Control data (no St-VIVIT) are superimposed, reproduced from Fig. 4C. B, Live-cell simultaneous imaging of [Ca²⁺]_i and EGFP on rat SCG neurons transfected with EGFP-NFATc1, with cyclosporine A (CsA)-added 50 K⁺ solution. C, shown are representative M-current traces from rat SCG neurons stimulated by high-K⁺, or Ringer only, either alone, or in the presence of the St-VIVIT peptide or CsA, applied for 1 h before and at the time of stimulation. D, Bars summarize the current density (mean ± SEM) for the groups of cells as in C. ***p<0.001.

Perforated-patch experiments then tested the effect of blocking CaN/NFAT signaling on I_M amplitudes. Neurons were pretreated with CsA, or the St-VIVIT peptide, 1 hour before and throughout the 50-K⁺ stimulation, and studied after 48–60 hours. Consistent with previous results, for the control neurons, we observed significantly augmented I_M amplitudes after 50-K⁺ stimulation (1.90 ± 0.18 pA/pF, n=18, p<0.001), compared with neurons treated with regular Ringer’s solution (0.95 ± 0.09 pA/pF, n=10) (Fig. 5C). Pre-treatment with CsA or the St-VIVIT peptide did not affect I_M amplitudes in neurons treated with regular Ringer’s solution (0.93 ± 0.16 pA/pF, n=11, and 0.93 ± 0.12 pA/pF, n=10, respectively), but both abolished the effect of augmented I_M amplitudes induced by 50-K⁺ stimulation (0.97 ± 0.08 pA/pF. n=17, p<0.001, and 0.87 ± 0.11 pA/pF, n=10, p<0.001, respectively) (Fig. 5C, D).

KCNQ2 and KCNQ3 genes are regulated by NFAT

To prove that NFAT-mediated transcriptional regulation is causative of stimulation-induced increases in M-channel expression and I_M up-regulation, we sought to localize the site(s) of NFATc1-c2 on KCNQ2 and KCNQ3 channel promoter/enhancer regions. Thus, we developed luciferase (firefly)-reporter assays using various promoter/enhancer domains, constructed by PCR from KCNQ2 and KCNQ3 genomic DNA. Instead of transfecting the reporter constructs in SCG neurons, which has a very low efficiency, we used the PC12 sympathetic neuron-like cells, which express M channels (Villarroel, 1996) and NFATs (Cano et al., 2005) that are activated by CaN dephosphorylation (Canellada et al., 2006). We first performed a bioinformatic analysis of the promoter and 1^st-intron regions of rat KCNQ2 and KCNQ3 genes to look for potential NFAT binding domains, using the program MatInspect (version 3.3) (http://www.genomatix.de/cgi-bin/./eldorado/main.pl), and found three such domains in KCNQ2 and one in KCNQ3, containing five total potential NFAT binding sites with the core motif GGAAA or TTTCC. Thus, we made four luciferase-reporter constructs encompassing the corresponding putative NFAT binding domains, with luciferase expression as the read-out for NFAT activation and binding to the reporter constructs (Fig. 6A).

Fig. 6. KCNQ2 and KCNQ3 genes are regulated by NFAT.

A, shown are the putative NFAT binding domains on rat KCNQ2 and KCNQ3 promoter or 1^st-intron regions and the luciferase-reporter constructs to be tested. The numbering is relative to the initial methionine, defined as +1. The domains cloned into vector pGL3-Basic are shown as the purple boxes, and the predicted NFAT binding sites are the blue boxes with the AAAGG or CCTTT consensus motif for NFAT binding capitalized. B, C, bars are the luminescence (mean ± SEM) from firefly luciferase, normalized to the Renilla luciferase as the internal control, for KCNQ2 reporter construct #1–#3 and KCNQ3 reporter construct #1 transfected into PC12 cells. Bars show the response by at 2 d after treatment with 50-K⁺, ACh (1 mM), or Ringer only, either alone (B), or in the presence cyclosporine A (C) for 1 h and at time of stimulation, applied 24 hours after transfection. ***p<0.001.

PC12 cells were transfected with the four luciferase-reporter constructs encompassing the corresponding putative NFAT binding domains, and a constitutively-active Renilla reniformis luciferase construct. One day later, the cells were stimulated as before by regular Ringer’s, high-K⁺ or ACh, for 15 min, with termination by returning the cells to the culture medium. Cells were lysed after 2 days and the resulting luciferase luminescence measured. Fig. 6B shows the results from KCNQ2 reporter constructs Q2RC1-3 and the KCNQ3 reporter construct, Q3RC1. Significant firefly luciferase luminescence, normalized to the Renilla luciferase control, was observed three days after transfection for constructs Q2RC1-3 and Q3RC1. Moreover, the luminescence increased at least 2-fold (p<0.001) for constructs Q2RC1, Q2RC2 and Q3RC1, but not for construct Q2RC3, following stimulation of the cells by high-K⁺ or by ACh (n=5). There was a negligible response from cells transfected with empty vector for any stimulation. Our luciferase data predict regions Q2RC1 and Q2RC2 of the KCNQ2 gene and Q3RC1 of the KCNQ3 gene to be critical for transcriptional up-regulation. Lastly, exposure of cells to CsA for 1 hour before stimulation by high-K⁺ or ACh did not alter the basal firefly luciferase luminescence for any of the reporter constructs; however, the increased luciferase luminescence induced by high-K⁺ or ACh was abrogated (n=5) (Fig. 6C), suggesting the reporter signals are due to CaN/NFAT.

AKAP150−/− (KO) mice lack NFATc1 nuclear translocation and augmented expression of M channels

AKAP79/150 recruits CaN to multiple targets (Wong and Scott, 2004), including the Ca_V1.2 Ca²⁺ channel that serves as the Ca²⁺- and activity-dependent reporter that drives NFATc4 activation in the hippocampus (Oliveria et al., 2007). Thus, we probed the involvement of AKAP79/150 in CaN/NFAT regulation of M-channel expression in SCG neurons isolated from AKAP150+/+ (WT) and AKAP150−/− (KO) mice. We first transfected SCG neurons isolated from both groups of mice with EGFP-NFATc1, and simultaneously monitored [Ca²⁺]_i and EGFP-NFATc1 localization as previously. We observed similar [Ca²⁺]_i elevations for neurons isolated from both WT and KO mice (n=14 and 20), but NFAT nuclear translocation only for neurons from WT mice (Figs. 7A–B). Such data are summarized in Figs. 7C–D (for statistics, see Supplemental Information). Thus, the absence of AKAP150 abolishes NFATc1 nuclear translocation induced by 50-K⁺ stimulation. We then compared I_M levels between neurons isolated from AKAP150+/+ and AKAP150−/− mice by patch-clamp. As before, mouse SCG neurons were treated with regular Ringer’s or 50-K⁺ solutions, and I_M amplitudes evaluated after 48–60 hours. There was no statistical difference in I_M amplitudes from the two groups of neurons when treated with Ringer’s solution (0.87 ± 0.05, n=18, and 0.83 ± 0.09, n=20), suggesting that the lack of AKAP150 does not affect basal expression levels of M channels. However, the effect of augmented I_M amplitudes after 50-K⁺ treatment (1.28 ± 0.08 pA/pF, n=38, p<0.01) was wholly absent in neurons from AKAP150−/− mice (0.81 ± 0.08 pA/pF, n=28) (Fig. 8A, B).

Fig. 7. AKAP150−/− (KO) mice lack NFATc1 nuclear translocation after 50-K⁺ stimulation.

A–B, shown are images of the fura-2 and EGFP from wild-type (A) or AKAP150−/− (B) SCG neurons transfected with EGFP-tagged NFATc1, with intensities of 340/380 nm ratio and EGFP plotted from images acquired every 15 s during the experiments. C–D, bars summarize data (mean ± SEM) from groups of cells as in A–B for fura-2 and EGFP imaging, respectively. ***p<0.001.

Fig. 8. AKAP150−/− (KO) mice lack augmented expression of M channels after 50-K⁺ stimulation, which is “rescued” by transfection of AKAP79.

A, shown are current traces recorded from SCG neurons isolated from AKAP150+/+ or AKAP150−/− mice 2–3 d after treatment with 50 mM K⁺, or Ringer only. B, Bars summarize the current density (mean ± SEM) for the groups of cells as in A. C, Neurons from AKAP150−/− mice were transfected with WT AKAP79, or the indicated mutants that lack the PKC (ΔA-AKAP79) or CaN (AKAP79ΔCaN) binding domains, or the CaN-binding sequence, PIAIIIT (AKAP79ΔPIX). Shown are representative current traces recorded from such neurons 2–3 d after treatment with 50-K⁺, or Ringer only. D, bars summarize the current density (mean ± SEM) for the groups of cells as in C. **p<0.01.

Transfection of AKAP79 in AKAP150−/− neurons “rescues” the activity-dependent transcriptional up-regulation of M channels

To test whether the above result in AKAP150 KO neurons is truly due to the lack of AKAP150, such neurons were transfected with either WT AKAP79, AKAP79 with a PKC-binding domain deletion (ΔA-AKAP79) (Klauck et al., 1996; Zhang et al., 2011), or AKAP79 with a CaN-binding domain deletion (AKAP79ΔCaN) (Klauck et al., 1996; Oliveria et al., 2007), to “rescue” the responses of augmented expression of M channels after high-K⁺ stimulation. Transfection of EGFP-tagged AKAP79 (n=9) or ΔA-AKAP79 (n=16), but not AKAP79ΔCaN (n=9), in the AKAP150−/− neurons restored the augmented I_M amplitudes after 50-K⁺ stimulation, suggesting the role of AKAP79/150 and its recruited CaN, but not PKC, in NFAT-mediated regulation of M channel expression (Fig. 8C). Within the CaN binding site on AKAP79/150 is the sequence, PIAIIIT, satisfying the consensus CaN-binding sequence PIXIXIT, and its deletion prevents the CaN-AKAP79/150 interaction (Oliveria et al., 2007). This deletion mutant (AKAP79ΔPIX) also failed to rescue the responses of augmented I_M amplitudes after 50-K⁺ stimulation in AKAP150 KO neurons (n=9) (Figs. 8C–D, for statistics, see Supplemental Information). We noticed that the “rescued” up-regulation of I_M in AKAP150−/− neurons transfected with AKAP79 or ΔA-AKAP79 was significantly larger than those in neurons from AKAP150+/+ mice (compare Fig. 8B, D, p<0.01), probably because AKAP79 was over-expressed in these neurons. However, such AKAP79 over-expression did not up-regulate I_M amplitudes in neurons treated with regular Ringer’s solution, further confirming the role of AKAP79/150 and CaN mediated NFAT signaling in activity-dependent, but not tonic, transcriptional regulation of M channels.

NFAT-mediated activation of M-current transcription requires local Ca²⁺ influx through L-type Ca²⁺ channels and global Ca²⁺ rises

To determine the source of Ca²⁺_i signal that activates the CaN/NFAT, we tested the effect of bradykinin (BK) receptor stimulation, nominally Ca²⁺-free external solution, or various subtype-specific Ca²⁺ channel blockers on both NFAT nuclear translocation and up-regulation of I_M amplitude.

We first explored whether NFAT activation requires influx of Ca²⁺ through the plasma membrane, using imaging on SCG neurons from WT mice transfected with EGFP-NFATc1. Neurons loaded with fura-2 were stimulated by either BK (250 nM), which stimulates G_q/11-coupled B₂ receptors to induce Ca²⁺ release from IP₃-sensitive intracellular Ca²⁺ stores (Cruzblanca et al., 1998; Gamper and Shapiro, 2003; Zaika et al., 2007), or nominally Ca²⁺-free external 50-K⁺ solution, for 10–15 min and then switched back to nominally Ca²⁺-free Ringer’s. BK induced an obvious [Ca²⁺]_i elevation, but EGFP-NFATc1 nuclear translocation was not observed (n=22, Supplemental Material, Fig. S1A). For neurons stimulated in Ca²⁺-free external solution, we observed neither a [Ca²⁺]_i elevation nor EGFP-NFATc1 translocation (n=12, Fig. 9A). We next used 50-K⁺ (or ACh) solution added with either 1) the L-type Ca²⁺ channel (L-channel) blocker nifedipine (10 μM), 2) the N-type Ca²⁺ channel (N-channel) blocker, ω-conotoxin GVIA (Boland et al., 1994) (ω-CgTX, 1 μM), or 3) the P/Q-type Ca²⁺ channel blocker, ω-agatoxin-TK (Adams et al., 1993) (ω-Aga-TK, 400 nM) on WT neurons to study which Ca²⁺ channels are critical for CaN/NFAT signaling. We found ω-Aga-TK to affect neither Ca²⁺ responses nor EGFP-NFATc1 nuclear translocation (n=14, Fig. 9D). With nifedipine added to the 50-K⁺ or ACh solution, the [Ca²⁺]_i elevation was undiminished, but we did not observe EGFP-NFATc1 nuclear translocation (Fig. 9B, n=19, and Fig. S1B, n=8). When ω-CgTX was added to the 50-K⁺ solution, both the [Ca²⁺]_i elevations and the EGFP-NFATc1 nuclear translocation were also diminished (n=19, Fig. 9C). Such data are summarized in Figs. 9G, H (for statistics, see Supplemental Material). Thus, influx of external Ca²⁺ ions through both L- and N-channels is required for NFAT nuclear translocation in sympathetic neurons.

Fig. 9. NFATc1 nuclear translocation requires local Ca²⁺ influx through L-type Ca²⁺ channels and global Ca²⁺ rises.

A–E, shown are images of the Fura-2 340/380 nm ratio and EGFP from wild-type SCG neurons transfected with EGFP-NFATc1, with zero external Ca²⁺ (A), or 10 μM nifedipine (B), 1 μM ω-conotoxin GVIA (ω-CgTX, C), 400 nM ω-agatoxin-TK (ω-Aga-TK, D) or ω-CgTX+ L-type Ca²⁺ channel agonist FPL-64716 (E) added to the 50 mM K⁺ solution, with intensities of the 340/380 nm ratio and EGFP plotted from images acquired every 15 s during the experiments. F, shown are images of EGFP from wild-type SCG neurons transfected with EGFP-NFATc1, with BAPTA-AM, or EGTA-AM loaded for 30 min before 50-K⁺ stimulation, with intensities of EGFP plotted from images acquired every 15 s during the experiments. EGTA data were grouped into NS (non-significant) and S (significant) [Ca²⁺]_i rises. G–H, bars summarize data (mean ± SEM) from groups of cells as in A–F for fura-2 and EGFP imaging, respectively. ***p<0.001. See also Supplemental Fig. S1.

We suspected that 1) NFAT activation requires AKAP79/150 to target CaN to L-channels, and CaN activated by localized high [Ca²⁺]_i elevations close to the inner mouth of open L-channels, and 2) NFAT translocation requires global [Ca²⁺]_i elevations, most easily through N-channels. We did two experiments to test these hypotheses. First, nuclear translocation of EGFP-NFATc1 was tested on WT neurons loaded with either the slow Ca²⁺ chelator, EGTA, or the fast Ca²⁺ chelator, BAPTA, both loaded in the cell as the cell-permeant AM-ester (Fig. 9F). EGFP-NFATc1 nuclear translocation induced by high-K⁺ stimulation was dramatically suppressed by BAPTA (n=23), consistent with our hypothesis that the initiation of NFAT signals depends on local [Ca²⁺]_i rises. However, EGTA yielded highly divergent results among cells, which we suspected was due to variable loading of EGTA-AM. Fura-2 imaging from these cells confirmed this (Fig. S1F), and these cells were then further analyzed into two groups. The “NS” (non-significant) group of cells had no statistical increase of [Ca²⁺]_i (Δ_340/380<0.05, n=9) and no NFATc1 nuclear translocation, whereas the “S” group were those with significant [Ca²⁺]_i rises (Δ_340/380>0.05, n=12, p<0.001) and displayed noticeable, although slower and smaller, NFATc1 nuclear translocations (Fig. 9F), consistent with a requirement for global [Ca²⁺]_i elevations in addition to local ones. Second, solutions that failed to induce NFATc1 translocation in previous experiments were altered to fulfill the requirement of “local” + “global” [Ca²⁺]_i rises: the 50 mM K⁺ + ω-CgTX solution was combined with FPL-64716, or with BK. Both cocktail solutions restored the elevated global [Ca²⁺]_i responses, and restored the NFATc1 nuclear translocation induced by high-K⁺ stimulation (Fig. 9E, n=12, and Fig. S1C, n=6). Such data are summarized in Figs. 9G, H (for statistics, see Supplemental Information). Taken together, our data are best explained by such local Ca²⁺_i signals occurring in microdomains containing AKAP79/150-orchestrated Ca²⁺-binding molecules, such as CaN and CaM, and L-channels, which function as the activity reporter that links neuronal activity with NFAT-mediated transcriptional regulation (see Discussion).

I_M amplitudes and the expression level of KCNQ2 and KCNQ3 transcripts were also assayed under these same conditions. Consistent with the EGFP-NFATc1 translocation results, there was no enhancement of I_M amplitudes in WT SCG neurons that had been stimulated with zero Ca²⁺-added (0.71 ± 0.09 pA/pF, n=10), or nifedipine-added 50-K⁺ solutions (0.81 ± 0.05 pA/pF, n=9), compared with neurons stimulated under regular Ringer’s solution (0.87 ± 0.05 pA/pF, n=18) (Figs. 10A–B). qPCR was also performed from WT SCG neurons 7 hours after perfusion of regular Ringer’s, 50-K⁺, 50-K⁺ with CsA, or 50-K⁺ with nifedipine, solutions for 15 min. We detected significant increases in the amount of both KCNQ2 and KCNQ3 mRNA in neurons depolarized by 50-K⁺ (3.2 ± 0.8 and 3.9 ± 0.7, n=4, p<0.01). This elevated expression of KCNQ2 and KCNQ3 mRNA was suppressed when CsA (1.05 ± 0.20 and 1.41 ± 0.15, n=4) or nifedipine (1.25 ± 0.28 and 1.61 ± 0.50, n=4) was present during the stimulation (Fig. 10C). Thus, removal of external Ca²⁺, blockade of CaN or addition of nifedipine during stimulation eliminates NFATc1 nuclear translocation and augmented KCNQ2/3 mRNA and I_M amplitudes, suggesting the critical role of CaN and L-type channels for transcriptional regulation of M channels.

Fig. 10. Seizures induce dramatic augmentation of KCNQ2-3 channel transcription in the hippocampus, requiring AKAP79/150.

A, shown are representative M-current traces recorded from WT SCG neurons 2–3 d after treatment with either zero Ca²⁺ or 10 μM nifedipine added 50 mM K⁺, or Ringer only. B, bars summarize the current density for the groups of cells as in A. C, bars summarize the fold-increase (mean ± SEM) in KCNQ2 and KCNQ3 mRNA from WT SCG neurons after stimulation by 50 mM K⁺, either alone or with cyclosporine A (CsA) or nifedipine added. Data are normalized to the expression of β-actin RNA. D–E, bars summarize the fold-increase (mean ± SEM) of KCNQ2 (D) and KCNQ3 (E) mRNA after seizures induced by pilocarpine or kainic acid (KA) in hippocampi from AKAP150+/+ or −/− mice. Data are normalized to the expression of β-actin RNA. E, Model for activity-dependent transcriptional regulation of M-channel expression. Neuronal activity regulates KCNQ2-3 transcription, acting as a negative feedback loop that limits neuronal hyper-excitability. AKAP79/150 binds to L-type (Ca_V1.3) Ca²⁺ channels and orchestrates a signaling complex that includes bound PKA, CaM and CaN (and other proteins not shown here) in a microdomain. L/Ca_V1.3 channels serve as the critical “sensor” of depolarization, and opening of L/Ca_V1.3 channels creates an elevated local Ca²⁺_i signal in sympathetic neurons. CaN binding to AKAP79/150 in the microdomain containing the local Ca²⁺_i signal is activated, NFAT is then dephosphorylated and activated by Ca²⁺-CaM/CaN, resulting in translocation of NFAT from the cytoplasm to the nucleus, where it acts on KCNQ2-3 gene regulatory elements. *p<0.05, **p<0.01. See also Supplemental Fig. S2.

Seizures induce dramatic augmentation of KCNQ2 and KCNQ3 transcription, requiring AKAP79/150

Our discovery that M-channel transcription is regulated by neuronal activity through NFAT/CaN signaling in sympathetic neurons led us to think this may generalize throughout the nervous system to limit neuronal hyper-excitability. Thus, we measured the relative expression level of KCNQ2 and KCNQ3 transcripts in a pathological animal model of neuronal hyper-excitability, chemo-convulsant-induced seizures in mice. As the part of the brain often serving as the focal point for dangerous human seizures, we focused on the hippocampus. We used the pilocarpine as well as kainic acid (KA) convulsant-seizure models of inducing status epilepticus (Leite et al., 2002), which corrects for any confound of altered M currents from muscarinic agonist (pilocarpine), rather than from hyper-activity. Visual scoring of seizure behavior was performed on each mouse (0 = no phenotype, 1 = vacant stare, 2 = tremors and ‘sticky’ feet, 3 = flag pole tail, 4 = full clonus) and only mice obtaining a score of 3 or 4 within 60 min after drug administration were used. We did not notice significantly different seizure stages in WT and AKAP150−/− mice as previously reported (Tunquist et al., 2008), except in response to the first dose of KA (Fig. S2). Hippocampi were isolated from mice 16–20 h after intrapleural administration of either pilocarpine or KA, or only vehicle as a control, total RNA extracted and qPCR performed. Indeed, a profound augmentation of both KCNQ2 and KCNQ3 mRNA was observed in mice after seizures induced by pilocarpine (27.9 ± 6.7 and 9.3 ± 2.2 fold, n=18) or KA (8.7 ± 2.3 and 3.1 ± 0.6 fold, n=25), compared with control mice injected with vehicle (1.1 ± 0.05 and 1.1 ± 0.10, n=17) (Figs. 10D–E). This profoundly-increased mRNA of both KCNQ2 and KCNQ3 in hippocampi from mice subjected to seizures are much greater than that seen from 50-K⁺ or ACh treatment in cultured sympathetic neurons, suggesting that seizures induce exaggeration of this transcriptional regulation, and could be conserved throughout the nervous system as a protective mechanism against hyper-excitability disorders such as epilepsy. Our mechanism predicts that this profound increase induced by seizures should likewise be dependent on AKAP150. Indeed, in AKAP150−/− mice, there was almost no up-regulation of KCNQ2 and KCNQ3 mRNA after pilocarpine-induced (1.5 ± 0.1 and 1.4 ± 0.1, n=14) or KA-induced (1.8 ± 0.4 and 1.6 ± 0.2, n=18) seizures (Figs. 10D–E), confirming the central role of AKAP150 in this phenomenon.

Discussion

Our model for activity-dependent regulation of M channel transcription

We here show neuronal activity to closely regulate M-channel transcription, likely as a negative-feedback loop that limits neuronal hyper-excitability. AKAP79/150 associates with L-type (Ca_V1.3) Ca²⁺ channels, and orchestrates a signaling complex that includes bound PKA, CaM and CaN in this microdomain. Ca_V1.3 channels serve as the critical “sensor” of activity and depolarization, and their opening creates an elevated local Ca²⁺_i signal, which activates CaN bound to AKAP79/150 in the microdomain of elevated local [Ca²⁺]_i. Upon Ca²⁺_i/CaN signals, both NFATc1 and NFATc2 are dephosphorylated and translocate from the cytoplasm to the nucleus, where they act on KCNQ2-3 gene regulatory elements, up-regulating I_M, thus reducing excitability (Fig. 10F). In a variety of neurons, AKAP79/150 orchestrates PKA to phosphorylate and up-regulate the activity of L-type Ca²⁺ channels, amplifying the responses to depolarization induced by neuronal activity. In the hippocampus, CaN counter-balances PKA actions, since the Ca²⁺ ions that enter the cell through L-channels participate in inactivating those same channels via CaN (Hall et al., 2007; Oliveria et al., 2007). We show that in the same protein complex where PKA augments L-currents, AKAP79/150 directs CaN to activate NFAT and initiate a longer-term feedback loop that up-regulates M channel expression, thus countering increased neuronal excitability. This seeming dual competing action of AKAP79/150 is unexpected, intriguing, and novel.

Recent structural and biochemical studies have revealed the stoichiometry of the core AKAP79 complex as a dimer with two CaN heterodimers, a PKA homodimer, with PKA binding to each AKAP79 protomer (Gold et al., 2011). Thus, there lies the tempting possibility that AKAP79/150 not only brings PKA, PKC and CaN to both L-type Ca²⁺ channels and M-type K⁺ channels, but it also physically couples one channel to the other in the same macromolecular complex, perhaps via the two protomers of the AKAP79/150 dimer (Gold et al., 2011). Both channels are widespread with overlapped expression in the nervous system, with KCNQ2/3 clustered at the axon initial segments and nodes of Ranvier (Devaux et al., 2004; Klinger et al., 2011; Pan et al., 2006; Shah et al., 2008), and L-channels concentrated in the cell bodies and proximal dendrites of central neurons (Hell et al., 1993). Recent findings in ventricular myocytes might shed some light on the role of AKAP79/150 in physical coupling between ion channels. Ca_V1.2 channels in those cells physically interact with each other at their carboxyl-tails by AKAP79/150, resulting in the amplification of Ca²⁺ influx and excitation-contraction coupling (Dixon et al., 2012). Thus, the interaction between L-channels and M-channels could serve to fine tune the activity of various neural circuits in an activity-dependent manner.

What is the source of Ca²⁺ signal that activates NFAT/CaN signaling?

Why should L-channels, which underlie no more than 15% of I_Ca in rodent SCG neurons, be so critical for NFAT activation? Our hypothesis is that opening of specifically Ca_V1.3, as the dominant L-channel in SCG (Lin et al., 1996), creates an elevated local Ca²⁺_i signal that is sensed by CaM and CaN recruited by AKAP79/150 to the microdomain of Ca_V1.3 channels. Although we did not rigorously test for physical association of AKAP79/150 with Ca_V1.3 channels using FRET or co-IP as was done in the hippocampus for Ca_V1.2 (Oliveria et al., 2007), we strongly predict such intimate association must be the case also in sympathetic ganglia. Interestingly, blockade of the N-channels that dominate I_Ca in sympathetic neurons also abolished NFATc1 nuclear translocation, in addition to most of the 50-K⁺ or ACh-induced [Ca²⁺]_i rises. Another lab investigating NFAT translocation in the same SCG cells has suggested that influx through N-, not L-channels to be the driving force for NFAT activation by electrical stimulation (Hernandez-Ochoa et al., 2007), a result that might be compatible with the dual requirement found here. If L-channels play a central role in CaN/NFAT activation by clustering the Ca_V1.3/CaM/CaN complex through AKAP79/150, why then is there a requirement for N channels? CaN is thought to rapidly dissociate from the AKAP79 complex to interact with NFAT (Li et al., 2012). Dephosphorylated NFAT then translocates from cytoplasm to nucleus, which requires at least 5–10 minutes. During this period, NFAT must remain dephosphorylated and CaN activated. CaN is believed to remain bound to NFAT to keep it dephosphorylated during its import into the nucleus. What keeps CaN activated even when it translocates away from the “local” elevated [Ca²⁺]_i near the mouth of L-channels? We believe it is the globally elevated [Ca²⁺]_i, mostly mediated by the N-channels that underlie the majority of I_Ca in SCG cells. However, the globally elevated [Ca²⁺]_i need not have come specifically from N-channels. Thus, when the L-channel agonist FPL-64716 or BK, was included in the 50-K⁺+ω-CgTX solution, the elevated global [Ca²⁺]_i signal and NFATc1 translocation were restored. The free [Ca²⁺] needed to occupy the low affinity sites on apoCaN and cause modest activation is around 1 μM, with V_max increased more than 20-fold in the presence of Ca²⁺/CaM (Feng and Stemmer, 2001; Klee et al., 1998). Such a 1 μM [Ca²⁺]_i is consistent with the globally elevated [Ca²⁺]_i expected from stimulating SCG cells (Gamper and Shapiro, 2003). Clearly, our hypothesis needs to be confirmed by biochemical studies of the Ca²⁺/CaM affinity of CaN when it is bound to AKAP79/150, or to NFAT. We find translocation of NFATc1/c2 to lag well behind the induced Ca²⁺_i rises, similar to that seen in BHK cells or Jurkat lymphocytes, in which NFAT was shown to be rapidly dephosphorylated by CaN, but NFAT nuclear import to be >10-fold slower. This phenomenon has been described as providing for a “working memory of Ca²⁺_i signals” (Kar et al., 2012; Tomida et al., 2003), but the mechanism responsible for this temporal discrepancy is, as yet, unclear.

Our work in sympathetic ganglia should be compared with similar lines of inquiry in DRG sensory neurons, where CaN/NFAT signals have been shown to be triggered by multiple mechanisms. In those cells, NFAT translocation occurs downstream of [Ca²⁺]_i rises not only by influx of Ca²⁺ from depolarization that open VGCCs, such as from trains of action potentials, opening of TRPV channels, or high-K⁺ stimulation (Kim and Usachev, 2009), but also by release of Ca²⁺ from internal Ca²⁺ stores, such as by IP₃-mediated Ca²⁺ release from stimulation of G_q/11-coupled BK receptors (Jackson et al., 2007). However, in SCG such Ca²⁺_i signals from internal stores induced by BK alone are much smaller and could not activate NFAT, but were sufficient for the global [Ca²⁺]_i rise that we suggest maintains NFAT active during its transit into the nucleus. As to the induction of NFAT translocation by TRPV activation in DRG neurons, we suggest that mechanism to be akin to the NFAT translocation induced by AChR stimulation seen here in SCG cells. For the latter, our model supposes the AChRs to cause NFAT translocation not from Ca²⁺ influx through the AChRs themselves, but from robust depolarization, which opens L-channels, beginning the CaN/NFAT cascade. We suggest the TRPV mechanism on NFAT in sensory neurons could be similar, especially since Kim and Usachev (2009) did not block L-channels during their capsaicin stimulations. Finally, the NFAT isoform shown to translocate in an L channel-dependent manner in hippocampal neurons is NFATc4 (Oliveria et al., 2007), whereas we found translocation only for NFATc1-c2, but not NFATc3-c4. Probably, distinct NFAT subtypes are activated in distinct neuronal types.

M-current as a “anti-epileptogenic” vs. “anti-convulsant” target

Identified from patients with inherited neonatal syndrome, benign familial neonatal convulsions (BFNC), M channels formed by KCNQ2/3 heteromers have proven a promising therapeutic anti-epileptic target. Although more than 20 anti-epileptic drugs, including the M channel opener, retigabine, are available on the market, one third of patients cannot control their epilepsy satisfactorily due to various reasons, one of which is the fact of these drugs being only seizure suppressing, or “anti-convulsant,” but not seizure preventing, or “anti-epileptogenic” (Stafstrom et al., 2011). Transcription of KCNQ2/3 genes has been suggested to be developmentally regulated (Hadley et al., 2003; Tinel et al., 1998), which may underlie the remission of BFNC seen in the clinic. Here, we show that transcription of KCNQ2 and KCNQ3 subunits are up-regulated by stimulation, with massive up-regulation in hippocampi after seizures. Thus, as an important “anti-epileptic” target to suppress seizures, M-channels may also be as critical a pharmacological “anti-epileptogenic” target to prevent recurrent seizures, i.e., epilepsy. Another lab showed AKAP150−/− mice to be resistant to chemically-induced seizure onset (Tunquist et al., 2008), although our findings in that regard are much more subtle. Indeed, we find that the profound up-regulation of KCNQ2 and KCNQ3 transcription levels after such seizures is nearly abrogated in hippocampi isolated from AKAP150−/− mice, suggesting these mice would be much more vulnerable to epileptogenesis after seizures. This prediction will be very interesting to test, although other factors involved in enhanced seizure susceptibility, such as decreased GABA_A expression, changes in HCN channel expression, or activation of inflammatory responses (Rakhade and Jensen, 2009) must be controlled for. Interestingly, strong up-regulation of CaN and BDNF mRNA and protein levels has been reported after hypoxia-, pilocarpine-, or KA-induced seizures (Rakhade and Jensen, 2009), suggesting another mechanism by which seizures should increase CaN-dependent transcriptional actions.

The events co-coordinated by AKAP proteins range spatially from the membrane to the nucleus, and temporally over many orders of magnitude, from the second to the lifetime of the organism. These signaling complexes could play important roles to limit epileptic seizures, and to restrict undue long-term, highly-plastic phenomena, such as limiting unnecessary formation of dendritic connections and superfluous, or redundant, circuits in the brain. Both KCNQ2 transgenic dominant-negative mice and AKAP150 mutant or KO mice display impaired hippocampus-dependent learning and memory, with the latter exhibiting deficits in strength and motor coordination as well (Peters et al., 2005; Tunquist et al., 2008; Weisenhaus et al., 2010). Additionally, malfunction of AKAP79/150-NFAT signaling may underlie exaggerations of cerebral mood and disease syndromes of the peripheral nervous system, such as chronic pains and cardiovascular dysfunction.

Experimental Procedures

AKAP150 KO mice

AKAP150 (+/+) and (−/−) mice (C57BL/6 background) at UTHSCSA (originally supplied to us by Dr. G. Stanley McKnight, University of Washington) are housed in groups of five and maintained under a 12:12 h light-dark cycle with food and water provided ad libitum. Mice were backcrossed (+/+ × −/− to yield +/−) every 6^th generation. Detailed characterization of the AKAP150−/− mice has been described (Zhang et al., 2011).

Perforated-patch electrophysiology

M currents in SCG cells were studied by holding the membrane potential at −25 mV and applying a 500-ms hyperpolarizing pulse to −60 mV every 5 s. I_M amplitude was measured at −60 mV from the decaying time course of the deactivating current sensitive to the M-channel specific blocker XE991 (Zaczek et al., 1998). Further details are in Supplemental Methods.

Quantitative Real-Time PCR (qPCR)

SCG neurons were prepared using the protocol described in Supplemental Methods. Cytosine arabinoside (Ara-c, 5 μM) as a mitotic inhibitor was added in the medium to prevent astrocyte growth. On day 2 in vitro, neurons were stimulated by application of 50 mM K⁺ or ACh (1 mM) for 15 min. Stimulation was terminated by returning the neurons in culture medium. Experiments were repeated at least three times using RNA collected from at least three separate SCG cultures, and each culture was prepared from >15 rats or mice.

Dual-Luciferase-based gene reporter assays

PC12 cells were co-transfected with plasmids encoding a luciferase (firefly) based reporter and a constitutively active Renilla reniformis luciferase under the control of a thymidine kinase promoter (pRL-TK; Promega, Madison, WI) by the Lipofectamine 2000 reagent (Invitrogen, 11668-019). 24 hours later, the cells were stimulated by application of high-K⁺ or ACh (1 mM) for 15 min. In experiments with inhibitors, the cells were exposed to the inhibitors (CsA, VIVIT) for 1h and then treated with high-K⁺ stimulation in solutions containing the inhibitors. Two days after stimulation, cells were lysed, and the activity of each luciferase construct measured sequentially on a TD-20/20 luminometer (Turner Biosystem, Sunnyvale, CA) and data were calculated according to the instructions of the Dual-Luciferase Reporter assay System kit (Promega). Each experiment was conducted as least three times.

Simultaneous imaging of [Ca²⁺]_i and GFP-labeled NFAT isoforms

Rat or mice SCG neurons transfected with EGFP-NFATc1-c4 were bath-loaded with the Ca²⁺-sensitive indicator fura-2 (fura-2 AM, 2 μm) for 30 min at 37 °C in the presence of pluronic acid (0.01%). [Ca²⁺]_i elevation and nuclear import of NFAT isoforms were induced by high-K⁺-induced depolarization or nicotinic ACh receptor agonist ACh (1 mM). Cells were placed in a flow-through chamber mounted on the stage of an inverted Nikon Eclipse TE300 microscope, perfused with standard Ringer’s solution. Fluorescence was excited alternatively at 340, 380 nm and 470 nm (50–200 ms every 15 s) using a Polychrome IV monochromator (T.I.L.L. Photonics, Martinsreid, Germany) through a FURA2/GFP filter cube. Data were processed using TILLvisION 4.0 and Image J software and presented as mean ± SEM.

Pilocarpine and kainic acid (KA) administration

One-month-old male mice (C57BL/6) were given subcutaneous injections of 1) methyl scopolamine nitrate (1 mg/kg, in sterile saline) to reduce peripheral cholinergic agonist-induced side effects, and 30 min later pilocarpine (280 mg/kg, in saline), 2) KA (10 mg/kg, in saline) every hour for 3 times, or 3) saline as negative control. Mice were closely observed during and 1 h after pilocarpine or KA injections, and their seizure behaviors were assigned a rating for each 15-min period according to a seizure staging system adapted from established (Racine, 1972) rodent seizure scales (Winawer et al., 2007).

Highlights.

1. Neuronal stimulation induces up-regulation of M-channels through CaN/NFAT signals.

2. NFAT nuclear translocation requires both local and global [Ca²⁺]_i elevations.

3. AKAP79/150 is vital for NFAT activation and M-channel up-regulation by stimulation.

4. Subjecting mice to seizures induces more M-channels in hippocampus, using AKAP150.