RNA editing of the IQ domain in Ca_v1.3 channels modulates their Ca²⁺-dependent inactivation

Summary

Adenosine-to-inosine RNA editing is crucial for generating molecular diversity, and serves to regulate protein function through recoding of genomic information. Here, we discover editing within Ca_V1.3 Ca²⁺ channels, renown for low-voltage Ca²⁺-influx and neuronal pacemaking. Significantly, editing occurs within the channel’s IQ domain, a calmodulin-binding site mediating inhibitory Ca²⁺-feedback (CDI) on channels. The editing turns out to require RNA adenosine deaminase ADAR2, whose variable activity could underlie a spatially diverse pattern of Ca_V1.3 editing seen across the brain. Edited Ca_V1.3 protein is detected both in brain tissue and within the surface membrane of primary neurons. Functionally, edited Ca_V1.3 channels exhibit strong reduction of CDI; in particular, neurons within the suprachiasmatic nucleus show diminished CDI, with higher frequencies of repetitive action-potential and calcium-spike activity, in wildtype versus ADAR2 knockout mice. Our study reveals a mechanism for fine-tuning Ca_V1.3 channel properties in CNS, which likely impacts a broad spectrum of neurobiological functions.

Introduction

Adenosine-to-inosine (A-to-I) RNA editing is a versatile post-transcriptional mechanism that allows pinpoint recoding of transcripts at the resolution of single nucleotides. This mechanism can drastically impact both the expression levels and functional properties of resulting proteins, thereby expanding the repertoire of protein customization (Keegan et al., 2001). The underlying chemistry involves ADAR enzymes (adenosine deaminases acting on RNA) that catalyze the deamination of adenosine (A) to generate inosine (I) at certain nucleotide positions within RNA. Because inosine is decoded as guanosine (G) during translation, resulting protein products feature exquisitely customized amino-acid composition. In the mammalian brain, characterization of such RNA editing has mainly focused on a restricted subset of ion channels and neurotransmitter receptors—editing of AMPA/kainate-type glutamate receptors (GluRs) yields changes in Ca²⁺ permeability or agonist desensitization, while that of serotonin 2C receptors leads to altered G-protein signalling (Burns et al., 1997; Seeburg et al., 1998). Functionally, the adenosine deaminase enzyme ADAR2 is responsible for RNA editing that recodes a glutamine to arginine in the selectivity filter of GluR-B subunits (‘Q/R editing’ of GluR-B); consequently, ADAR2 knockout mice exhibit increased Ca²⁺ permeability, with concomitant epilepsy and death (Higuchi et al., 2000). More broadly, a generalized dysregulation of brain RNA editing in humans may contribute to epilepsy, depression, and suicidal tendencies (Gurevich et al., 2002; Schmauss, 2003; Sergeeva et al., 2007). Indeed, it is likely that numerous other editing substrates remain to be identified in the mammalian brain, given the high inosine content of mRNA in neural tissue (Paul and Bass, 1998).

In particular, we wondered whether RNA editing might fine-tune the calmodulin (CaM) regulation of voltage-gated calcium channels (VGCCs). This Ca²⁺ feedback regulation would be an especially attractive target for editing, because structure-function analysis reveals that even single amino-acid substitutions at critical channel hotspots can markedly alter modulatory properties (Dick et al., 2008; Tadross et al., 2008; Zuhlke et al., 2000), and such regulation impacts functions as diverse as neurotransmitter release, neuronal pacemaking, neurite outgrowth, and gene expression (Dunlap, 2007). Figure 1A cartoons such regulatory hotspots, which are located on the amino- and carboxyl-termini of the pore-forming α₁ subunits of VGCCs. The best-studied locus is a CaM-binding domain approximating a consensus IQ element satisfying the amino-acid pattern IQxxxRGxxxR (Jurado et al., 1999), with × denoting any residue. CaM binding at this IQ domain is critical for CaM/channel regulation (Liu et al., 2010), and mutations in the central isoleucine strongly attenuate Ca²⁺ regulation (Shen et al., 2006; Yang et al., 2006).

Figure 1.

Detecting RNA editing in the Ca_V1.3 IQ domain

(A) Cartoon of Ca_V1.3 channel pore-forming α₁ subunit (α_1D), with hotspots for CaM/channel regulation shown on the amino terminus (NSCaTE (Dick et al., 2008)), and in the CI region of the carboxyl terminus (EF, EF-hand (Peterson et al., 2000); preIQ and IQ domains (Erickson et al., 2003; Pitt et al., 2001)).

(B) DNA sequencing chromatogram from direct analysis of PCR of genomic DNA, yielding unique coding for IQ domain at this level.

(C) Top, exemplar PCR products from RT-PCR analysis of rat thalamus. Bottom, direct sequencing results of RT-PCR from thalamus, showing distinct doublets of adenosine (A) and guanosine (G) (arrows) to either generate MQDY or IQDC from unedited IQDY. Colony screening reveals additional IQ domain editing from IQDY to IRDY.

(D) Direct sequencing results for RT-PCR of cochlea, heart, DRG and pancreatic β-islets, showing identical patterns to the genomic analysis in panel B, indicating no editing.

Here, we reveal the existence of ADAR2-mediated RNA editing of the IQ domain of Ca_V1.3 channels. This editing appears specific to the central nervous system, and proteomic analyses indeed confirm the presence of edited Ca_V1.3 channel proteins within native brain tissues. Adding to the theme of specificity, no RNA editing was found for Ca_V1.3 coding regions outside of the IQ domain, nor was IQ-domain editing present in any other members of the Ca_V1–2 channel family. All these features suggest that Ca_V1.3 editing may entail distinctive sequelae for the CaM-dependent inactivation (CDI) of these channels, particularly in relation to the availability of these low-voltage activated channels to support neurotransmission at ribbon synapses (Evans and Zamponi, 2006; Yang et al., 2006) and repetitive firing within neurons throughout the brain (Chan et al., 2007). Accordingly, we demonstrate that RNA editing of the Ca_V1.3 IQ domain can strongly diminish the CDI of these channels. Furthermore, as a first step towards evaluating the neurobiological impact of Ca_V1.3 IQ-domain editing, we characterized repetitive firing properties of neurons in the suprachiasmatic nucleus (SCN), an oscillatory brain region that contributes a central biological clock for circadian rhythms in mammals. Of particular relevance, SCN oscillations appear to be substantially driven by L-type Ca²⁺ currents, most of which are carried by Ca_V1.3 channels (Marcantoni et al.,; Pennartz et al., 2002; Xu and Lipscombe, 2001). Importantly, we now find that RNA editing alters both CDI and the frequency of repetitive electrical activity, as judged by comparison of CDI and SCN rhythmicity in wild-type and transgenic mice wherein RNA editing was eliminated. Additionally, since the chemical compound Bay K 8644 selectively diminishes CDI and augments overall current amplitude in L-type Ca²⁺ channels (Tadross et al., 2010), we utilized this agent as a selective pharmacological mimic of altered Ca_V1.3 IQ domain editing. Indeed, the effects of Bay K 8644 on SCN rhythmicity were strikingly reminiscent of those produced upon transitioning from wild-type to transgenic mice lacking RNA editing. Accordingly, our experiments demonstrate that regulation of mammalian circadian rhythmicity constitutes one of potentially many important consequences of Ca_V1.3 RNA editing.

Results

Detecting RNA editing of the IQ domain in Ca_V1.3 channels

A schematic of the pore-forming α₁ subunit of VGCCs, together with the main elements supporting CaM-mediated CDI, furnishes the structural context of our search for RNA editing (Figure 1A). The presence of an NSCaTE Ca²⁺/CaM binding site tunes the dynamic Ca²⁺ sensitivity of CDI (Dick et al., 2008; Tadross et al., 2008), whereas PreIQ and IQ domains harbor functionally important binding sites for both apoCaM (Ca²⁺-free CaM) and Ca²⁺/CaM (Erickson et al., 2003; Liu et al., 2010; Pitt et al., 2001). Ca²⁺-driven movements of CaM among these various sites trigger CDI, with the collaboration of an EF-hand-like module that transduces CaM movements into altered channel gating (de Leon et al., 1995; Kim et al., 2004; Peterson et al., 2000). Though the collective action of several modules produces CDI, even single mutations in some of these structural hotspots can severely modify CDI (Dick et al., 2008; Peterson et al., 2000; Tadross et al., 2008; Zuhlke et al., 2000). Nowhere is this single-residue alteration of CDI better known than in the IQ domain (Dunlap, 2007), which thus serves as the focus of our screen.

At the genomic level, the predicted amino-acid sequence at the core of the IQ domain is IQDY. These are coded by the nucleotides ATACAGGACTAC, as explicitly confirmed by PCR amplification and sequencing of the rat genomic DNA (Figure 1B). Also as expected, amplification of the corresponding Ca_V1.3 IQ domain in rat thalamic mRNA yielded a seemingly homogeneous PCR product of 300 bp (Figure 1C, upper panel). However, direct sequencing of this RT-PCR product showed obvious evidence of RNA sequence variability at two adenosine positions, characterized by conspicuous A/G double peaks in sequencing chromatograms (Figure 1C, bottom half, arrows on green and gray bands). These double peaks indicated that in addition to the canonical IQDY sequence, alternative sequences like MQDY, IQDC, and MQDC could also be manifest at the protein level, as summarized in Figure 1C. To detect even rare occurrences of RNA sequence variability, we employed colony screening, where RT-PCR products were cloned into bacterial colonies, and sequencing performed on amplified DNA from individual colonies. This approach not only confirmed the two sites of variability above, but also revealed a rarer locus where CAG (Q) was modified to CGG (R), which encodes an IRDY sequence (Figure 1C, bottom row). These instances of RNA sequence variability were consistent with RNA editing, and could produce the amino-acid variations shown in Figure 1C. Yet further potential combinatorial variation of the IQ domain is detailed in supplementary Figure S1A.

In contrast to the ready detection of RNA sequence variability within the Ca_V1.3 IQ domain, further regions of editing were not observed. Transcript-scanning of the complete α_1D subunit from total rat brain RNA, using direct sequencing of RT-PCR products, gave no indication of sequence variability outside of the IQ module. Furthermore, analysis of total brain RNA for the paralogous IQ domains of other Ca_V channels (Ca_V1.2, Ca_V1.4, Ca_V2.1, Ca_V2.2 and Ca_V2.3) also failed to reveal such variation (supplementary Figure S1B).

Outside of the central nervous system (CNS), Ca_V1.3 is functionally important in cochlea, heart (Platzer et al., 2000; Shen et al., 2006), pancreas (Liu et al., 2004; Safa et al., 2001; Taylor et al., 2005), and other tissues. Yet, no RNA sequence variability at the Ca_V1.3 IQ domain was observed in rat cochlea, heart, pancreatic β-islet, and dorsal root ganglion cells (Figure 1D), despite ADAR2 expression in these contexts (Gan et al., 2006; Melcher et al., 1996). Overall, CNS modulation of RNA sequence within the Ca_V1.3 IQ region appeared rather special.

Before turning to the mechanisms underlying this RNA sequence variability, we tested whether such variability produces veritable diversity at the protein level, using state-of-the-art mass spectrometry. Ca_V1.3 complexes isolated from whole mouse brain were trypsinized, labelled with mTRAQ, and analyzed via HPLC-MS/MS multiple reaction monitoring (MRM, see supplementary Figure S2 for details). Signals for peptides containing FYATFLMR, FYATFLMRDYFR, KFYATFLIQDCFR, and KFYATFLIR isoforms of the IQ domain were detected, as well as that of the unedited IQ domain (FYATFLIQDYFR). BLAST analysis confirmed that the variant sequences are unique within the mouse genome. Hence, I-to-M, Q-to-R and Y-to-C recoding of amino acids are present within the actual Ca_V1.3 protein.

ADAR2 mediates editing of the Ca_V1.3 IQ domain

With assurance of genuine IQ-domain alterations, we considered the mechanism supporting this variability, noting that RNA editing of adenosine to guanosine is believed to require enzymes belonging to a class known as Adenosine Deaminases Acting on RNA (Keegan et al., 2001). These ADARs bind to duplex stem-loop structures within pre-mRNA, and then catalyze deamination of adenosines to inosine (I) (Figure 2A). This action effectively alters the codon within the mature edited mRNA, because inosine is decoded as guanosine by the translation machinery. To test whether the specific deaminase isoform ADAR2 is responsible for the Ca_V1.3 IQ domain variability, we compared results from wildtype GluR-B^R/R mice to those of ADAR2^−/−/GluR-B^R/R knockout animals (Higuchi et al., 2000), focusing in particular upon the lumbar and whole-brain regions. Direct DNA sequencing of RT-PCR products from these regions gave strong qualitative indications of sequence variability (Figure 2B, left) at each of the colored locations identified earlier in thalamus. For quantification, we measured the relative heights of chromatogram peaks for adenosine and guanosine at these loci, enabling specification of a percent-recoding metric shown as light-colored bar graphs (Figure 2B, right). Reassuringly, measurement of chromatogram areas yielded identical estimates of percent recoding (Figure 2E). Additionally, as an independent measure of percent recording, a colony screening method produced a closely similar quantitative profile of sequence variability (Figure 2B, right, darker-colored bars). The quantitative analyses revealed an overall rank order of RNA sequence variability (most frequent to rarest) of: ATA (I) recoding to ATG (M), followed at a slightly lower frequency by TAC (Y) recoding to TGC (C), followed much more rarely by CAG (Q) recoding to CGG (R). Another perspective came with extensive colony analysis of mouse whole brain, yielding an overall frequency distribution of IQ-domain sequence combinations (Figure 2F). Given this rich assortment of variants in wild-type mice, we undertook the key genetic experiment regarding the origin of this variability. Indeed, the ADAR2 knockout was devoid of sequence variability (Figure 2C), thus arguing strongly that ADAR2 is necessary for Ca_V1.3 IQ domain editing.

Figure 2.

RNA editing of Ca_V1.3 regulated by ADAR2 and during development

(A) Schematic of ADAR2 mechanism. Transcription of pre-mRNA with formation of putative ECS duplex (top); recruitment of ADAR2 to duplex (second from top); conversion of adenosine to inosine (third from top); edited mature transcript ready for exportation to cytoplasm and translation (bottom).

(B) Profile of editing in mouse lumbar and whole brain. Left column, direct DNA sequencing of mouse RT-PCR products. Right column, percent editing at three locations (I-to-M, Q-to-R and Y-to-C), as calculated by measuring electropherograms heights for adenosine versus guanosine (translucent bars), or by colony counting from colony screening analysis (filled bars).

(C) No editing in ADAR2^−/−/GluR-B^R/R knockout mice. Format as in panel B.

(D) Developmental profile of RNA editing in mouse and rat brains. Left panel, direct DNA sequencing of RT-PCR products from brains of different ages (n = 3 animals per age). Right panel, percent editing at three locations (I to M, Q to R and Y to C), as calculated by measuring electropherograms heights for adenosine versus guanosine (embryonic day 14, unfilled bars at zero level; postnatal day 4, translucent bars; and postnatal day 7, filled bars).

(E) Equivalence of peak-height and area metrics for RNA editing, based on sequencing chromatography. Symbols represent editing percentages calculated via peak heights of sequencing chromatograms, plotted as a function of editing percentages calculated via area under sequencing chromatograms, performed for data sets in panel B. Open circles, lumbar; filled circles; whole brain; solid line, line of identity.

(F) Overall frequency distribution of Ca_V1.3 IQ-domain variants, taken from mouse whole-brain via colony counting method. IRDC combination was never observed.

Spatio-developmental RNA editing in rat and mouse

Given the nuanced distribution of ADAR2 throughout the brain, we next explored the spatio-temporal occurrence of Ca_V1.3 RNA editing across the CNS. Accordingly, the editing analysis introduced in Figure 2B was applied to individual brain regions, such as frontal cortex, hippocampus, medulla oblongata and cerebellum of rat brain. The analysis revealed that editing was spatially regulated across the rat brain, with frontal cortex and hippocampus showing the most editing (supplementary Figures S3A and S3B). These general trends from rat were recapitulated in the mouse brain (Figure S3C), with subtle intra-species differences present at the quantitative level. As well, we explicitly confirmed the presence of Ca_V1.3 IQ domain editing in human brain (supplementary Figure S4A).

Beyond spatial regulation, we also observed marked developmental modulation of IQ domain editing (Figure 2D). At embryonic day 14, the Ca_V1.3 IQ domain lacks detectable editing. By contrast, editing was observed as early as postnatal day 4, and reached adult levels by postnatal day 7. Similar trends were observed in both rats and mice, with bar-graph population summaries shown at the far right.

Selective modulation of CDI by RNA editing

With the existence and molecular basis of IQ-domain editing in hand, we investigated the critical question of functional impact on CaM-mediated CDI of Ca_V1.3 channels. Prior structure-function work on Ca_V1.3 would suggest that RNA editing of the critical isoleucine-glutamine (IQ) di-peptide residues might well modulate this important Ca²⁺ feedback system (Yang et al., 2006). To test for such an outcome, we performed electrophysiological analysis of recombinant Ca_V1.3 channels bearing the key IQ domain variants supported by RNA editing. As baseline, Figure 3A displays the wildtype Ca_V1.3 profile. Exemplar Ba²⁺ currents (top panel, black trace), as evoked by maintained depolarization to near the peak of I–V relations, showed little decay, indicative of minimal voltage-dependent inactivation (VDI). By contrast, exemplar Ca²⁺ currents evoked at the same potential showed a rapid decay (top panel, red trace), as produced by robust CaM-mediated CDI (Yang et al., 2006). Inactivation profiles, averaged over many cells, are displayed in the next two panels below. The fraction of peak Ba²⁺ current remaining after 50-ms depolarization to various potentials (r₅₀) hovers near unity, consistent with little VDI. By contrast, strong CDI is apparent in the sharp decline of the Ca²⁺ r₅₀ relation, which exhibits a U-shaped voltage dependence characteristic of a genuine Ca²⁺-driven process. Pure CDI was quantified by the f-value, calculated as the difference in r₅₀ measured in Ba²⁺ and Ca²⁺ at −10 mV. The difference between Ca²⁺ and Ba²⁺ relations then specifies CDI measured in isolation. The multi-second recovery from CDI is reported in the third panel by the fraction of peak current recovered after increasing durations at the holding potential (F_recovery). Finally, as for activation, the bottom two panels display the Ba²⁺ tail-activation relation (G_norm), and the normalized peak Ba²⁺ current versus voltage curve (I_norm).

Figure 3.

Modulation of Ca²⁺-dependent inactivation by IQ domain editing

(A) Wildtype (IQDY) channel gating properties. Top panel, exemplar traces of currents evoked from holding potential of −90 mV to test potential of −10 mV, with Ba²⁺ as charge carrier (black), Ca²⁺ as charge carrier (red). Throughout, vertical scale bar pertains to Ca²⁺ current, and Ba²⁺ currents are scaled down ~3× to facilitate visual assessment of kinetic decay. Second panel, averaged inactivation profiles shown by r₅₀, the fraction of peak currents remaining after 50-ms depolarization to indicated voltages (V). VDI characterized by Ba²⁺ data; CDI was quantified by a f-value that is the difference of r₅₀-values between Ca²⁺ and Ba²⁺ profiles at −10 mV. Symbols represent the average of n = 10 cells. Third panel, semi-log plot of fractional recovery from Ca²⁺-dependent inactivation. Symbols are averages of 9–11 cells. Fourth panel, Ba²⁺ tail-activation curves averaged from ~15 cells. Bottom panel, peak Ba²⁺ current versus voltage relation, averaged from 10–13 cells. Currents were normalized to maximum values before averaging. For second through bottom panels, s.e.m. bars are shown when larger than symbol size.

(B) Gating profile for MQDY channel variant. Format as in panel A. Slowed CDI onset and enhanced CDI recovery apparent in top three panels, with wildtype profiles reproduced as dashed gray curves, and shaded region emphasizing important differences. Activation gating unchanged, as shown in bottom two panels.

(C) Gating profile for IRDY variant, showing lesser effects on CDI. Format as in panel B.

(D) Gating profile for MRDY variant, showing strongest effects on CDI. Format as in panel B.

Compared to this reference behavior, Y-to-C editing of the IQ domain (e.g., IQDY recoding to IQDC) had little functional effect (Figure S5A). Beyond this, however, all other edited forms of Ca_V1.3 exhibited substantial alterations of CDI, with little change of either VDI or activation characteristics. Channels bearing the IQ-to-MQ variant of the IQ domain demonstrated a clearly weaker CDI (Figure 3B, top two panels, wildtype: f = 0.72 ± 0.01; MQ: f = 0.45 ± 0.03), and perhaps a hint of faster recovery from inactivation. Wildtype profiles are reproduced as dashed curves, and the red shading emphasizes the effects of editing. The flat exemplar Ba²⁺ current and r₅₀ relation (Figure 3B, top two panels) demonstrates unchanged VDI, and steady-state inactivation (VDI) was also unperturbed in all IQ variants (supplementary Figure S5B). Likewise, activation properties were essentially unchanged (Figure 3B, bottom two panels). Channels with the IQ-to-IR variant of the IQ domain exhibited similar though weaker alteration of CDI (Figure 3C, f = 0.60 ± 0.01), and closely similar recovery from inactivation as control. Most strikingly, the IQ-to-MR variant demonstrated pronounced effects—approximately 50% reduction in the onset of CDI (Figure 3D, f = 0.33 ± 0.01), and sharply accelerated recovery from inactivation—both actions highly significant.

To assess whether editing affects the ability of Ca_V1.3 channels to target to the neuronal surface membrane, we generated cDNAs encoding both unedited (IQDY) and various edited forms of Ca_V1.3 channels (MQDY, IRDY, MRDY or IQDC). These channels were also endowed with an extracellular HA tag to facilitate subsequent immunocytochemical assays of surface-membrane expression. As a preliminary check, electrophysiological characterization of heterologously expressed channels confirmed the absence of appreciable functional effects of the HA epitope itself (supplementary Figure S5C). We then transiently expressed the suite of HA-tagged Ca_V1.3 clones in primary hippocampal neurons. Immunocytochemistry revealed similar surface expression patterns between the unedited and edited forms of Ca_V1.3 variants (supplementary Figure S5D), arguing that transport of channels to the neuronal surface membrane was largely unaffected by editing. In addition, expression patterns of transfected Ca_V1.3 were similar to those of endogenous channels (supplementary Figure S5D).

Modification of SCN rhythmicity in ADAR2 knockout mice

As a first step towards explicitly resolving the biological significance of RNA editing of the Ca_V1.3 IQ domain, we turned to neurons in the suprachiasmatic nucleus (SCN), where Ca_V1.3 currents figure prominently in triggering the spontaneous action potentials that underlie circadian rhythms (Pennartz et al., 2002). Molecular analysis clearly confirmed RNA editing of the IQ domain in SCN (Figure 4A₁). Furthermore, whole-cell voltage-clamp recordings from individual SCN neurons in acute brain slices detected robust CDI, seen by comparison of mean current waveforms obtained with 10 mM Ba²⁺ versus Ca²⁺ as the charge carrier (Figure 4A₂). As baseline, Ba²⁺ currents (measuring VDI) decayed with a similarly slow timecourse in neurons from either wildtype (GluR-B^R/R) or ADAR2 knockout mice (ADAR2^−/− /GluR-B^R/R) (Higuchi et al., 2000); this feature is illustrated by the close similarity of blue- (wild-type) and cyan-colored (knock-out) Ba²⁺ current waveforms (Figure 4A₂), respectively averaged from wildtype (n = 7) and knockout (n = 6) neurons. By contrast, Ca²⁺ currents from wildtype neurons decayed more rapidly (black trace, n = 7)), indicative of substantial CDI in the wildtype SCN. Importantly in knockout neurons, Ca²⁺ currents decayed with still greater rapidity (red trace, n = 6), just as expected without RNA editing of the Ca_V1.3 IQ domain. These trends were entirely corroborated by population analysis of multiple neurons (Figure 4A₃), particularly over the 0–10 mV range where Ca_V1.3 channel CDI would likely predominate (supplementary Figure S6A and S6B). In this regard, despite the contribution of other Ca²⁺ channel subtypes to overall current (Cloues and Sather, 2003), most of the observed RNA-editing effects on CDI could be attributed to Ca_V1.3 channels, because little CDI was observed upon pharmacological blockade of Ca_V1.3 channels (supplementary Figure S6C), and a comparatively high level of intracellular Ca²⁺ buffering was used (5 mM EGTA) to preferentially suppress Ca_V2 channel CDI (Liang et al., 2003; Soong et al., 2002; Tadross et al., 2008).

Figure 4.

Comparison of SCN rhythmicity in wild-type and ADAR2 knockout mice.

(A₁) Analysis of RNA editing of Ca_V1.3 IQ domain in mouse SCN. Format as in Figure 2.

(A₂) Timecourse of normalized current decay in voltage-clamped SCN neurons in acute slice preparations. Overall format as in Figure 2A, top. Blue trace averaged from 7 wildtype neurons, with 10 mM Ba²⁺ as charge carrier. Cyan trace averaged from 6 knockout neurons, with 10 mM Ba²⁺ as charge carrier. Black trace averaged from 7 wildtype neurons, with 10 mM Ca²⁺. Red trace averaged from 6 knockout neurons, with 10 mM Ca²⁺. Ca²⁺ traces shown for 0-mV depolarizing voltage step. Ba²⁺ trace shown for −10-mV step, to account for ~10-mV surface-charge shift.

(A₃) Analysis of inactivation in SCN neurons, using r₁₀₀ metric defined as the fraction of peak current remaining after 100-ms depolarization. Ba relation averaged from 13 wildtype or knockout neurons. Ca WT relation averaged from 7 wildtype neurons. Ca KO relation averaged from 5–6 knockout neurons. Red shading highlights effects of editing on CDI, with P<0.05 difference between wildtype and knockout relations denoted by *.

(B) Na spikes in wildtype (WT, top black) and knockout mouse (KO, bottom red) SCN neurons.

(C) Overlays of time-aligned Na spikes from preparation in panel B, confirming diminished depolarization prior to Na spikes in KO mice. Solid lines represent graphically the mean depolarization rate averaged over multiple spikes. Symbols plot mean depolarization rate (±SEM) preceding Na spikes in exemplar wild-type versus KO mouse presented in panel B.

(D) Average frequency of action potential (Na spikes) in wild-type (n = 6) versus KO (n = 10) mouse SCN neurons. Upper scale tick, 1 Hz. Asterisk denotes significance at P < 0.05 level.

(E) Ca spike activity (in 1 µM TTX) recorded in wildtype (top, black) versus knockout mouse SCN neurons (bottom, red). In the knockout, Ca spike frequency is decreased, and troughs are depolarized. Red trace at extreme bottom shows abolition of Ca spikes by 10 µM nimodipine.

(F) Time-aligned, averaged Ca spikes confirm that KO mouse neurons manifest a depolarization of minimal troughs between spikes. Data averaged from n = 6 WT and n = 7 KO mouse SCN slices. Bars, standard error of mean.

(G) Decreased average frequency of Ca spikes in wild-type versus KO mouse SCN slices analyzed in panel F (p < 0.01). Format as in panel D.

(H) Exemplar Ca spikes from wild-type mouse SCN slice, illustrating effect of Bay K 8644 to increase spike frequency and deepen troughs between spikes.

(I) Time-aligned, averaged Ca spikes confirm that Bay K 8644 deepens troughs between Ca spikes. Averaged from n = 11 mouse SCN slices. Format as in panel F.

(J) Increased average frequency of Ca spikes in wild-type SCN slices exposed to 1–10 µM Bay K 8644 (P < 0.05), averaged from same slices as analyzed in panel I. Format as in panel D.

Having explicitly established effects of RNA editing on CDI within SCN neurons, we tested for potential corresponding consequences on SCN rhythmicity. Under current clamp of SCN neurons in acute slices of wildtype mice (GluR-B^R/R), we observed spontaneous discharges of sodium action potentials (‘Na spikes’) characteristic of this preparation (Figure 4B, top black). By contrast, SCN neurons of ADAR2 knockout mice (ADAR2^−/− /GluR-B^R/R) (Higuchi et al., 2000) exhibited Na spikes that fired at clearly lower frequencies (Figure 4B, bottom red; and Figure 4D), with a decreased depolarization rate preceding Na action potentials (Figure 4C). This suite of effects in the ADAR2-deficient setting is consistent with a loss of RNA editing leading to increased Ca_V1.3 CDI, with corollary diminution of Ca_V1.3 pacemaking current. Two important controls warrant mention. Firstly, the ‘wildtype’ GluR-B^R/R mice used as baseline were engineered for constitutive expression of the R-containing form of GluR-B subunits at the Q/R-editing site (Higuchi et al., 2000); hence, the alteration of Na spike activity seen upon transitioning to ADAR2 knockout animals (Figures 4B–4D) could not have arisen trivially from a loss of Q/R editing of GluR-B subunits. Secondly, we determined that Q/R editing of GluR-B subunits in the SCN of non-engineered wild-type mice was complete (supplementary Figure S4B), thus excluding the possibility that engineering wildtype mice for constitutive expression of the R-form of GluR-B would, in itself, alter baseline excitability.

Nonetheless, altering RNA editing of targets other than those considered thus far could still account for the rhythmicity effects up to this point. Accordingly, we analyzed the actions of ADAR2 elimination upon a persistent pattern of membrane potential oscillations that persists after application of a saturating concentration of TTX, as illustrated by the exemplar trace from a wildtype mouse (Figure 4E, upper black trace). These repetitive ‘Ca spikes’ feature a more readily defined role for L-type channels, because these spikes arise from a simpler system wherein oscillations are generated by recurring depolarization via L-type Ca²⁺ current that in turn begets an ensuing phase of repolarization via Ca²⁺-activated K currents (Belle et al., 2009; Pennartz et al., 2002). Importantly, while representing a simpler phenomenon, Ca spikes may closely reflect more integrated Na spike behavior (Pennartz et al., 2002). Significantly, then, the specimen trace from an ADAR2-deficient mouse (Figure 4E, middle red record) exhibits a reduced frequency of Ca spikes, concurrent with depolarization of troughs between spikes. Population averages from several SCN slices confirmed the attenuated Ca spike frequency upon ADAR2 elimination (Figure 4G); and corresponding averages of time-aligned Ca spikes confirmed depolarization of troughs between Ca spikes (Figure 4F). Both effects of ADAR2 elimination accord well with heightened Ca_V1.3 CDI and resultant attenuation of Ca_V1.3 current. In particular, diminished low-threshold depolarizing current explains the decrement in Ca spike frequency, while reduced Ca²⁺ entry during spikes would moderate Ca²⁺-activated K current and thereby repolarization between Ca spikes. Indeed, the role of Ca_V1.3 in driving Ca spikes was explicitly confirmed by abolishing spontaneous fluctuations with the L-type channel inhibitor nimodipine (Figure 4E, bottom red trace). In all, this spectrum of effects on the simpler system of Ca spikes hinted more strongly that RNA editing of Ca_V1.3 channels contributes to the altered SCN rhythmicity upon loss of ADAR2.

Still, ADAR2-mediated editing of several other membrane currents involved in repetitive Ca spiking could explain even these results (Figures 4E–4G). Accordingly, we investigated the actions of Bay K 8644, a highly-selective, L-type-channel-specific agonist. Though this compound has been available for some time, particularly relevant aspects of its actions have only recently become clear. Importantly, beyond its well-known ability to augment overall current, this compound also diminishes Ca²⁺-dependent inactivation (CDI), as demonstrated in our recent detailed biophysical analysis of Bay K 8644 actions on Ca_V1.3 (Tadross et al., 2010). Given this functional profile, Bay K 8644 should act much like a selective pharmacological mimic of altered Ca_V1.3 IQ-domain editing. In particular, this compound should mirror the transition from an ADAR2 knockout context (more CDI and less current) to a wild-type context (less CDI with more current) —so long as RNA-editing-induced alteration of Ca spiking does arise from modified Ca_V1.3 CDI. Indeed, we observed a striking analogy between the effects of Bay K 8644 (Figure 4H–4J) and those produced upon transitioning from knockout to wild-type mice (Figure 4E–4G). Specifically, Bay K 8644 produced both an increase in overall Ca spike rate, and hyperpolarization of troughs between Ca spikes. More precisely, Bay K 8644 simulated an exaggerated wild-type phenotype, wherein reduction of CDI by RNA editing was enhanced beyond the normal wild-type level. It is possible that the effects of Bay K 8644 would be still stronger if applied to the knockout condition, since the initial level of CDI would be stronger. In all, these effects of Bay K 8644 on SCN Ca spikes, highly analogous to those in our transgenic experiments, argue well that RNA editing of Ca_V1.3 channels contributes to SCN rhythmicity.

Finally, to assess the overall quantitative sufficiency of editing-induced modifications of Ca_V1.3 CDI to modulate SCN rhythmicity, we undertook computational simulations of SCN pacemaking, utilizing refined versions of previously established models (Belle et al., 2009; Sim and Forger, 2007). Here, we incorporated Ca_V1.3 profiles appropriate for our various experimental conditions (wild-type, ADAR2-deficient, and Bay K 8644 scenarios), and then observed the consequences for spontaneous activity (see Supplementary Information, section 6). Figure 5A displays the state-diagram for the Ca_V1.3 channel utilized in the refined models, along with corresponding CDI profiles for the differing conditions. Simulated Na spikes demonstrated a marked decrement in frequency upon transitioning from wild-type to ADAR2-deficient CDI configurations (Figures 5B and 5D). This decrement in frequency was accompanied by a decreased depolarization rate prior to Na spikes (Figure 5C), similar to effects observed experimentally (Figure 4C). Moreover, simulated Ca spikes demonstrated both decreased frequency and depolarization of troughs between spikes (Figures 5E–5G), qualitatively recapitulating experimental effects (Figures 4E–4G). Finally, Bay K 8644 increased simulated Ca spike frequency and hyperpolarized troughs between Ca spikes (Figures 5H–5J), also as observed experimentally (Figures 4H–4J) Thus, projected alterations in Ca_V1.3 channel CDI by RNA editing were sufficient to explain a wide array of experimentally observed effects.

Figure 5.

In silico simulations of experimental changes in SCN activity via altered Ca_V1.3 CDI

(A) Ca_V1.3 gating mechanism and operational profiles utilized in simulations of spontaneous SCN activity. Left, gating mechanism, with first-order activation (rate constants a_r and b_r have customized voltage dependence), and with slow-CaM CDI formulation appropriate for C-lobe-dominant inactivation of these channels. Middle, simulated CDI profiles during voltage-step protocols, illustrating appropriate behavior for wild-type condition (WT, denoting mixture of mainly MQ and IQ channels) and ADAR2 knockout condition (KO, denoting pure IQ population). Compare to Figure 3 for appropriateness of simulated CDI behavior. Right, simulated WT (with mixture of MQ and IQ channels) and WT plus Bay K 8644 (slowed CDI and augmented peak current of mixture of MQ and IQ channels). See Supplementary Information, section 6, for detailed computational methods and setup.

(B) Na spikes seen with WT and KO Ca_V1.3 profiles in panel A (middle). Note the lower frequency in the ADAR2 KO setting, quantified in panel D below.

(C) Decreased depolarization rate prior to Na spikes in KO. Format as in Figure 4C.

(D) Decreased frequency of simulated Na spikes in KO configuration. Upper scale tick, 1 Hz.

(E) Ca spikes seen with WT and KO Ca_V1.3 profiles in panel A (middle). Na currents eliminated to produce Ca spikes. Note the lower frequency and depolarization of troughs between spikes in the ADAR2 KO setting; these trends are quantified further in panels F and G.

(F) Expanded view of Ca spikes, confirming trough depolarization in KO configuration.

(G) Bar graph summary of drop in Ca spike frequency in KO configuration. Format as in D.

(H) Bay K 8644 increases Ca spike frequency, and hyperpolarizes troughs between spikes. Ca_V1.3 profile as in panel A (right column). Trends shown at higher resolution in panels I and J.

(I) Hyperpolarization of troughs between Ca spikes, upon addition of Bay K 8644.

(J) Increased Ca spike frequency with Bay K 8644. Format as in panel D.

Taken together, the results in Figures 4 and 5 suggest that RNA editing of the Ca_V1.3 IQ-domain modulates SCN firing rates and thereby the central biological clock underlying circadian rhythms. Beyond the SCN, we suspect that RNA editing of Ca_V1.3 channels will orchestrate further neurobiological effects, wherever these channels act to promote pacemaking and near-threshold activity. For example, robust RNA editing of Ca_V1.3 was also detected in rat substantia nigra (supplementary Figure S4C), where these channels contribute to pacemaking and heighten the onset of Parkinson’s disease under pathological conditions (Chan et al., 2007). Overall, RNA editing of the Ca_V1.3 IQ domain could offer precise and potent tuning of neuronal activity in diverse brain regions.

Discussion

Adenosine-to-inosine RNA editing post-translationally recodes genomic information to generate molecular diversity. Many of the identified editing targets are found in the mammalian nervous system, with a historical focus on the family of GluR ion channels and serotonin 2C receptors (Schmauss, 2003; Seeburg and Hartner, 2003). Beyond this focus, the list of editing targets is expanding. For example, outside of Ca_V1.3 channels, there are instances of Ca_V channel editing exclusive of the IQ domain, though with uncertain biophysical consequences (Hoopengardner et al., 2003; Kawasaki et al., 2002; Keegan et al., 2005; Smith et al., 1998; Tsunemi et al., 2002). In other voltage-gated channels, editing of K_V1.1/K_Vβ1.1 channels speeds inactivation recovery (Bhalla et al., 2004), and editing of insect Na⁺ channels alters channel gating properties (Dong, 2007; Song et al., 2004). Here, our discovery of editing within the Ca_V1.3 IQ domain represents a significant expansion to this group, given the robust functional modulation of Ca²⁺-dependent feedback control at this particular locus, and the broad range of biological roles served by these channels (Day et al., 2006; Sinnegger-Brauns et al., 2004; Striessnig et al., 2006). Figure 6 schematically summarizes the general scope of RNA editing effects on Ca_V1.3 CDI, along with potential consequences for neuronal Ca²⁺ load in neurons..

Figure 6.

Schematic overview of RNA editing effects on Ca_V1.3 Ca²⁺-dependent inactivation (CDI) and calcium load in neurons. Channel schematic as defined in Figure 1A. Increased RNA editing (right scenario) favors channels with MQ and other edited versions of the IQ domain, decreasing overall CDI and presumably increasing cellular Ca²⁺ load (intense yellow-green coloration). Decreased editing (left scenario) favors default channels with IQ version of IQ domain, increasing overall CDI and potentially decreasing cellular Ca²⁺ load (weak yellow-green coloration). Actual neurons reside on a continuum between these two extremes, as represented by ramp schematics at bottom.

Notably, ADAR2-mediated editing of Ca_V1.3 is exquisitely selective—editing of Ca_V1.3 is restricted to the IQ domain; IQ-domain editing is absent in other Ca_V1–2 channels; and Ca_V1.3 editing is restricted to the CNS. This selectivity suggests that editing of the Ca_V1.3 IQ domain may be critical for certain biological niches, where fine tuning of Ca²⁺ feedback on channels (CDI) is especially desirable for low-voltage activated Ca²⁺ influx. As an initial delineation of neurobiological consequences, we have focused upon the suprachiasmatic nucleus (SCN), where Ca_V1.3 currents modulate spontaneous action potentials underlying mammalian circadian rhythms (Pennartz et al., 2002). We clearly demonstrate that RNA editing substantially modulates SCN rhythmicity, a significant finding in its own right. More specifically, our data suggest that editing of Ca_V1.3 appreciably mediates this modulation. This suggestion merits two lines of discussion, given the multiplicity of potential editing targets in SCN.

Firstly, the literature is rather divided regarding the role of L-type Ca²⁺ channels in modulating SCN activity. While earlier studies (Pennartz et al., 2002; Pennartz et al., 1998) favor a substantial contribution of L-type Ca²⁺ channels to SCN pacemaking, a more recent investigation emphasizes a more subsidiary influence of these channels (Jackson et al., 2004). This seeming discrepancy may relate to differences of slice (Ikeda et al., 2003; Pennartz et al., 2002; Pennartz et al., 1998) versus isolated neuron preparations of SCN (Jackson et al., 2004). Fitting with this view, a similarly diminished role of voltage-gated Ca²⁺ channels in shaping cerebellar pacemaking has been observed between slice (Womack et al., 2004) and isolated neuron experiments (Raman and Bean, 1999). Indeed, our own experiments favoring an appreciable role of Ca_V1.3 were performed in the presumably more intact acute slice configuration.

Secondly, as for differences in SCN rhythmicity observed between wild-type and ADAR2-deficient mice (Figure 4), our data argue well for the sufficiency of altered Ca_V1.3 CDI to contribute to these differences (Figures 4H–4J, 5), and also exclude Q/R editing of GluR-B subunits as the causative mechanism (supplementary Figure S4B). Nonetheless, other potential editing targets remain to be considered. Could altered Q/R editing of kainate receptors modify SCN activity upon ADAR2 elimination (Herb et al., 1996)? Countering this possibility, addition of kainate to wild-type SCN slices increased Ca spiking frequency while depolarizing troughs between spikes (supplementary Figure S3D), contradicting the outcome seen upon transitioning from ADAR2-deficient to wild-type contexts (Figure 4E–4G). Could editing of serotonin receptors explain our findings? Contrary to this view, it is the serotonin HT-7 receptor subtype that mediates serotonin effects in SCN (Aghajanian and Sanders-Bush, 2002; Lovenberg et al., 1993), and there is no indication that HT-7 is edited like the HT-2C receptor subtype (Aghajanian and Sanders-Bush, 2002). Could editing of GABA receptors contribute? GABA can certainly regulate SCN activity (Gillespie et al., 1997; Mintz et al., 2002), and GABA receptors undergo RNA editing by ADAR2 (Ohlson et al., 2007). Opposing this hypothesis, only the α₃ subunit of GABA_A receptors is known to be edited (Ohlson et al., 2007), and the α₃ subunit is only sparsely expressed in the adult mice relevant to our studies (O'Hara et al., 1995). Finally, might editing of voltage-activated K⁺ channels play a role? Against this position, only K_V1.1 channels are known to be RNA edited (Bhalla et al., 2004), while SCN neurons have been reported to express K_V3.1 (Espinosa et al., 2008; Itri et al., 2005), K_V3.2 (Itri et al., 2005), K_V4.1 and K_V4.2 (Itri et al., 2010). In fact, K_V1.1 knockout mice exhibit intact circadian rhythms, so long as overt seizure activity is controlled (Fenoglio-Simeone et al., 2009). Overall, then, while comprehensive exclusion of alternative mechanisms is difficult to achieve, our data remain highly suggestive that RNA editing of Ca_V1.3 CDI influences SCN rhythmicity.

Beyond the SCN, editing the Ca_V1.3 IQ domain is poised to modulate numerous other brain regions, wherever Ca_V1.3 contributes to low-voltage activated synaptic transmission and pacemaking (Day et al., 2006; Sinnegger-Brauns et al., 2004; Striessnig et al., 2006). More broadly, developmental regulation of RNA editing of the Ca_V1.3 IQ domain (Figure 2D) could influence neurodevelopment via Ca²⁺-dependent transcription factors (Pasca et al., 2010; Wheeler et al., 2008; Zhang et al., 2006). Furthermore, it would be interesting if Ca_V1.3 editing contributes to the epilepsy, depression, and suicide affiliated with a generalized alterations of brain RNA editing (Gurevich et al., 2002; Schmauss, 2003; Sergeeva et al., 2007). Investigating the role of edited Ca_V1.3 channels in these and other neurobiological processes now promise rich dividends for mechanistic advance and disease insight.

Experimental Procedures

Tissue preparation and total RNA extraction

Experiments were carried out on adult Sprague-Dawley rats and C57BL mice, as approved by the institutional IACUC. Various regions of the brain and spinal cord were dissected for RT-PCR experiments. Total RNA was isolated using the Trizol method (Invitrogen, Carlsbad, CA) and first strand cDNA was synthesized with Superscript II and oligo(dT)₁₈ primers (Invitrogen, Carlsbad, CA). Negative control reactions without reverse transcriptase were performed in all reverse transcription RT-PCR experiments to exclude contamination by genomic DNA. Reverse transcription to generate the first strand cDNA was performed by standard methods.

Transcripts scanning for edited sites

For rat, transcript-scanning of the Ca_V1.3 IQ domain was done by using the primer pairs —

• sense primer: 5′-GAGCTCCGCGCTGTGATAAAGAAA-3′;

  and antisense primer: 5′-GGTTTGGAGTCTTCTGGTTCGTCA-3′

  —to amplify a 300 bp Ca_V1.3 fragment.

  For mouse, the primer pairs used were —

  sense primer: 5′-CTCCGAGCTGTGATCAAGAAAATCTGG-3′;

  and antisense primer: 5′-GGTTTGGAGTCTTCTGGCTCGTCA-3′

  —for a 299 bp amplicon.

A standard step-down PCR protocol was used that included a 3-cycle decrement from 59 °C to 53 °C final annealing temperature. The number of cycles for the main PCR was 35, where denaturation was performed at 94 °C for 30 sec, annealing at 53 °C for 30 sec, and extension at 72 °C for 50 sec. The final extension was at 72 °C for 5 min. PCR products were separated on a 1% agarose gel, isolated and purified using the Qiagen gel extraction kit. The PCR product was sent for direct automated DNA sequencing (Applied Biosystems, Foster City, CA). Colony screening was performed by first sub-cloning PCR products into pGEM-T Easy vector (Promega, Madison, WI), transforming them into DH10B Escherichia coli cells, and then sending ~50 isolated clones for automated DNA sequencing. Three rats or mice were used for each group of animals. A total of 150 clones were screened to determine RNA editing for each brain or spinal cord region. To compare peak heights of the chromatogram bases, the peak height of guanosine was divided by the combined peak heights of adenosine and guanosine bases to estimate the percent of RNA editing.

Construction of rat full-length Ca_V1.3 with edited IQ motif

The first four amino acids of the Ca_V1.3 consensus IQ motif is IQDY corresponding to nucleotide sequence ATACAGGACTAC. We generated six edited Ca_V1.3 α_1D subunits from the reference wildtype α_1D-IQfull channels(Shen et al., 2006), now designated α_1D-IQDY. The edited subunits were named α_1D-MQDY, α_1D-IRDY, α_1D-MRDY, α_1D-IQDC, α_1D-MQDC and α_1D-MRDC. The α_1D-MQDY, α_1D-MQDC, α_1D-IQDC and α_1D-IQDC edited clones were generated by replacing a BstEII/NotI RT-PCR fragment containing the respective edited sites into the reference clone. The other edited clones were generated by in-vitro mutagenesis using the following primers set 5’-TGATGCGGGACTGCTTTAGG-3’ and 5’-CCTAAAGCAGTCCCGCATCA -3’ for α_1D-MRDC; 5’- TGATACGGGACTACTTTAGG -3’ and 5’- CCTAAAGTAGTCCCGTATCA-3’ for α_1D-IRDC; and 5'-TGATGCGGGACTACTTTAGG-3' and 5'-CCTAAAGCAGTCCCGCATCA-3' for α_1D-IRDC.

Whole-cell patch-clamp electrophysiology

HEK293 cells were transiently transfected α_1D-IQDY α_1D-MQDY, α_1D-IRDY, and α_1D-MRDY (1.25 µg) and rat β_2a (1.25 µg) and α₂δ (1.25 µg), using the standard calcium phosphate transfection methods(Tang et al., 2004). The β_2a and α₂δ clones were kindly provided by Dr. Terry Snutch (University of British Columbia). Electrophysiological recordings were performed as reported previously(Evans and Zamponi, 2006; Yang et al., 2006), and details are found in Supplementary Data (section 6).

SCN slice preparation

C57BL/6 wildtype (ADAR2^+/+/GluR-B^R/R) or knockout (ADAR2^−/−/GluR-B^R/R) mice(Higuchi et al., 2000) were maintained on a 12 hr light dark cycle using normal fluorescent room light. Coronal brain slices (250 µm thick) containing suprachiasmatic nucleus were obtained from 5 to 8 week-old mice anesthetized with isoflurane and decapitated. All experimental procedures were in accordance with the animal welfare guidelines of the Max-Planck-Society. The slicing chamber contained an oxygenated ice-cold solution composed of (in mM): NaCl, 125; KCl, 2.5; NaH₂PO₄, 1.25; NaHCO₃, 25; MgCl₂.6H₂O, 1.0; myo-Inositol, 3; Na-pyruvate, 2; vitamin C, 0.4; CaCl₂, 1; MgCl₂, 5; and glucose, 25. Slices were incubated for 30 min at 30°C before being stored at room temperature in artificial CSF (ACSF) containing (in mM): NaCl, 125; NaHCO₃, 25; KCl, 2.5; NaH₂PO4, 1.25; MgCl₂, 1; CaCl₂, 2; and glucose, 25; bubbled with 95% O2 and 5% CO2.

Whole-cell recording in SCN slices

Current-clamp recording were made using EPC-9 amplifier controlled by Patchmaster (Heka Elektronik, Lambrecht, Germany). Patch pipettes were pulled from borosilicate glass capillaries and had resistances of 4–6 MΩ when filled with (in mM): K-gluconate, 130; K-Cl, 10.00; EGTA, 5; N-(2-hydroxyethyl) piperazine- N’-ethanesulfonic acid (HEPES); Na₃GTP, 0.5; MgATP, 4.0; and Na-Phosphocreatine, 10.0. Brain slices were mounted on upright fixed stage microscope equipped with 40X water immersion lens and constantly perfused with the above mentioned oxygenated ACSF at a flow rate of 1.5 to 2 ml/min at room temperature. The SCN was identified as a bilaterally symmetrical, cell dense region superior to the optic chiasm and lateral to the inferior apex of the third ventricle (Pennartz et al., 1998). Individual SCN neurons were identified by IR-DIC camera. Only cells in the dorsal medial shell were patched where cluster I SCN neurons are dominant (Paxinos and Franklin, 2001). The cluster I neurons were identified by their steeply rising and monophasic AHP (Pennartz et al., 1998). After formation of gigaseal (>3 GΩ) formation, input resistance was monitored regularly by measuring voltage response by a −20 pA current injection. The reported membrane potential was corrected for the liquid junction potential −14.5 mV. For voltage-clamp recording of SCN neurons, the external solution used contained 10 mM HEPES, 140 mM tetraethylammonium methanesulfonate, 10 mM BaCl₂ or CaCl₂ (pH was adjusted to 7.4 with CsOH and osmolarity to 290–310 mosM with glucose). The external solution was constantly bubbled with 95% O₂ and 5% CO₂. The internal solution (pipet solution) contained 130 mM Cs-MeSO₃, 5 mM CsCl, 5 mM EGTA, 10 mM HEPES, 1 mM MgCl₂, 2 mg/ml Mg-ATP, pH 7.3 (adjusted with CsOH). The osmolarities of solutions used were adjusted to between 290 and 300 mosM with glucose. A junction potential of -11 mV was uncorrected for, and true voltage may be obtained by subtracting 11 mV from the reported values.

Drug Preparation

Stock solutions were prepared by dissolving nimodipine (RBI) in 100% ethanol and TTX (Alomone Labs) in distilled water. The solutions were subsequently diluted in ACSF to respective final concentrations. Nimodipine was protected from light during these procedures.

Highlights.

• ▪RT-PCR and MRM were used to detect RNA editing of IQ domain in Ca_V1.3 channels
• ▪CNS-specific editing of IQ domain in Ca_V1.3 channels is mediated by ADAR2
• ▪Edited Ca_V1.3 channels expressed in HEK cells showed two-fold reduction in CDI
• ▪ADAR2^−/− SCN neurons exhibit increased CDI and correspondingly decreased repetitive action potentials and calcium spikes