Cacna1b alternative splicing impacts excitatory neurotransmission and is linked to behavioral responses to aversive stimuli

Presynaptic CaV2.2 channels control calcium entry that triggers neurotransmitter release at both central and peripheral synapses. The Cacna1b gene encodes the α1-pore forming subunit of CaV2.2 channels. Distinct subsets of splice variants of CaV2.2 derived from cell-specific alternative splicing of the Cacna1b pre-mRNA are expressed in specific subpopulations of neurons. Four cell-specific sites of alternative splicing in Cacna1b that alter CaV2.2 channel function have been described in detail: three cassette exons (e18a, e24a, and e31a) and a pair of mutually exclusive exons (e37a/e37b). Cacna1b mRNAs containing e37a are highly enriched in a subpopulation of nociceptors where they influence nociception and morphine analgesia. E37a-Cacna1b mRNAs are also expressed in brain, but their cell-specific expression in this part of the nervous system, their functional consequences in central synapses and their role on complex behavior have not been studied. In this report, we show that e37a-Cacna1b mRNAs are expressed in excitatory projection neurons where CaV2.2 channels are known to influence transmitter release at excitatory inputs from entorhinal cortex (EC) to dentate gyrus (DG). By comparing behaviors of WT mice to those that only express e37b-CaV2.2 channels, we found evidence that e37a-CaV2.2 enhances behavioral responses to aversive stimuli. Our results suggest that alternative splicing of Cacna1b e37a influences excitatory transmitter release and couples to complex behaviors.


Introduction
Presynaptic Ca V 2.2 channels control neurotransmitter release throughout the nervous system where their activity impacts a wide range of neuronal functions. Ca V 2.2 channels dominate in supporting transmission of sensory information from nociceptors to spinal cord dorsal horn neurons [1][2][3][4][5][6]. Mice that lack Ca V 2.2 channels have impaired nociception [7][8][9] and intrathecal Ca V 2.2 calcium channel blockers are analgesic [10]. Ca V 2.2 channel activity also contributes to other behaviors in mice including locomotion, exploration, startle [9,11], ethanol intake [12], and aggression [13]. In humans, inhibition of Ca V 2.2 by intrathecal ziconotide can trigger psychotic episodes and anxiety, further emphasizing the importance of Ca V 2.2 channels in higher level, brain function and complex behaviors [14][15][16].
The Cacna1b gene encodes the Ca V α 1 pore-forming subunit of all Ca V 2.2 channels, and alternatively spliced exons in Cacna1b influence channel function and sensitivity to G-protein coupled receptors (GPCRs) [17][18][19][20]. The expression of alternatively spliced exons in Cacna1b is regulated by factors that depend on tissue type, cell type, development, and disease state [21][22][23][24]. Of special interest are mutually exclusive exons, e37a and e37b, which encode sequences in the C-terminus of Ca V 2.2 [25]. Previous work has shown that e37a-Cacna1b mRNAs are enriched in a subset of transient receptor potential vanilloid 1 (Trpv1) expressing nociceptors of dorsal root ganglia (DRG) [26]. Compared to e37b-Ca V 2.2, e37a-Ca V 2.2 channels are trafficked more efficiently to the cell surface [27,28], they are inhibited more strongly by G i/o -protein coupled receptors including the μ-opioid receptor [19,20,29], and they enhance actions of intrathecal morphine analgesia in vivo [19,20,29]. E37a-Ca V 2.2 and e37b-Ca V 2.2 channels contribute to basal thermal and mechanical nociception but e37a-Ca V 2.2 channels have a preferred role in the maintenance of thermal and mechanical hyperalgesia induced by inflammation [30,31]. The unique function of e37a-Ca V 2.2 channels in nociception and sensitivity to GPCR inhibition motivated us to explore the expression pattern and potential function of e37a-Ca V 2.2 channels in brain.
Here we show that e37a-Cacna1b mRNAs are expressed at low level throughout the brain but they are enriched in subsets of excitatory projection neurons. E37a-Cacna1b mRNAs are more abundant in Ca 2+ /calmodulin-dependent protein kinase II excitatory projection neurons (CaMKIIα + PNs) compared to cholecystokinin-expressing interneurons (CCK + INs). We show that e37a-Ca V 2.2 channels contribute to transmitter release at cortico-hippocampal excitatory synapses and they inhibit exploratory and novelty-induced anxiety-like behaviors. Our results suggest that e37a-Ca V 2.2 channels influence presynaptic transmitter release at specific synapses in the brain and are linked to behavioral responses to aversive stimuli.

Genotyping
Genomic DNA was extracted from P7-P9 animal tissue using Phire Animal Tissue Direct kit II (ThermoFisher Scientific, F140WH) according to manufacturer instructions. PCR was performed with AmpliTaq Gold® 360 mastermix (Thermo Fisher Scientific, 4,398,881) using the following conditions: a hot start of 95 0 C for 10 min, followed by 35 cycles (95 0 C, 30 s; 60 0 C, 30 s; 72 0 C, 1 min), and a final step of 72 0 C for 7 min. Primers and expected products are shown in Table 1. Primers were added to the same mixture for genotyping.

In-situ hybridization (BaseScope™)
To detect e37a-Cacna1b splice variants in brain tissue, we utilized a variation of in-situ hybridization, BaseScope™. Briefly, 'Z' probes contained a complementary sequence that binds Cacna1b mRNA between e36 and e37a, and a signal amplification complex containing the Fast Red dye. The punctate pattern of the Z probe signals is due to the ability of this technique to detect single RNA molecules [34]. Hematoxylin counterstaining was used to label nuclei. In brief, deeply anesthetized adult C57BL/6 mice either males or females (Euthasol, Virbac, 200-071) were transcardially perfused with 1x PBS and subsequently with 10% neutral buffered formalin (NBF) fixative solution (Sigma, HT501128). Brains were post-fixated in 10% NBF at 4°C for 24 h, washed 1x PBS, and dehydrated in a PBS-15% sucrose solution for at least 18 h at 4°C, or until tissue sank to the bottom of the tube. Tissue was transferred into PBS-30% sucrose solution for a second dehydration step for about 18 h at 4°C or until tissue sank to the bottom of the tube. Next, tissue was cryopreserved in optimal cutting temperature formulation (Fisher, 4585), with isopentane pre-chilled in dry ice. 12 μm sections from brain were generated in a cryotome (Shandon, 77,200,222) and individual sections were placed in 15 mm Netwell™ inserts (Corning, 3478). Sections were allowed to free-float in 1x PBS, and mounted on positively charged microscope slides (VWR, 48311-703). Sections were air-dried 10-20 min before in-situ hybridization (ISH). To aid tissue adhesion, air-dried sections were incubated at 60°C in a drying oven for 30 min, then post-fixed in 10% NBF at 4°C for 15 min and immediately transferred to 50, 70%, and two rounds of 100% ethanol for 5 min each. Slides were then allowed to air dry for additional 5 min before incubating with RNAscope® hydrogen peroxide solution for 10 min (ACD, 322381). Sections were then washed in Milli-Q® water and transferred to RNAscope® Target Retrieval solution preheated to 99°C for 15 min (ACD, 322000). Sections were briefly washed in Milli-Q® water, transferred to 100% ethanol for 3 min, then placed in drying oven at 60°C for 30 min. Sections were isolated with a hydrophobic barrier pen (ACD, 310018) and allowed to air-dry overnight at RT. Next sections were incubated with RNAscope® Protease III for 30 min (ACD, 322381) at 40°C in the ACD HybEZ™ Hybridization System (ACD, 310010). BaseScope™ probe spanning exon junction 36:37a of Cacna1b mRNA (ACD, 701151) was applied to each section for 2 h at 40°C in the ACD HybEZ™ Hybridization System. Base-Scope™ Detection Reagents AMP 0 -AMP 6 and FastRed (ACD, 322910) were applied according to manufacturer protocol and washed using RNAscope® Wash Buffer (ACD, 310091). To visualize nuclei, sections were counterstained in Gil's Hematoxylin I for 2 min at RT (American Master Tech, HXGHE) following signal amplification. Sections were washed in tap water 3 times, briefly transferred to 0.02%ammonia water, and washed again in tap water. Finally, sections were dried at 60°C for 15 min and mounted using VectaMount™ mounting medium (Vector Laboratories, H-5000). Cell counting was performed using Fiji cell counter plug in [35,36].

Fluorescence-activated cell sorting (FACS)
Adult CCK;Dlx5/6;tdT and CaMKIIα;tdT mice were deeply anesthetized with isofluorane, brains removed and dissected (in less than 45 s) in Earl's Balanced Salt Solution (EBSS) (Sigma, E3024) containing 21 U/mL of papain. Cerebral cortex or hippocampal tissue was dissociated using a modified version of Worthington Papain Dissociating System® (Worthington Biochemical Corporation, LK003150). After incubating with papain for 45 min at 37°C on a rocking platform, tissue was triturated with three sequential diameter fire-polished glass pipettes. Next, cell suspensions were centrifuged at 300 x g for 5 min. After discarding supernatants, pellets were resuspended in 3 mL of EBSS containing 0.1% of  All primers are reported in 5′ to 3′ direction ovomucoid protease inhibitor and 0.1% bovine serum albumin (Worthington, LK003182) to quench papain. Cell suspension was centrifuged at 270 x g for 6 min and resuspended in EBSS (3 mL). To isolate tdT-expressing cells, we performed FACS in a Sony SH800 flow cytometer using a 561-nm laser to excite and a 570-630 nm filter for event selection. At least 300,000 events were collected directly into TRIzol™ LS Reagent (Ther-moFisher Scientific, 10,296,028). Collection was performed keeping 1:3 (v/v) sorted cell suspension: TRIzol™ LS ratio. Cell suspension were kept on ice throughout the sorting session.

RT-qPCR
Total RNA from sorted cells was extracted using TRI-zol™ LS and isopropanol precipitation with the addition of 30 μg of GlycoBlue® Coprecipitant (ThermoFisher Scientific, AM9516) to facilitate visualization of RNA pellet. 300 ng of total RNA from sorted cells was primed with oligo-dT and reverse transcribed with Superscript IV First-Strand Synthesis System (ThermoFisher Scientific, 18,091,050) according to manufacturer instructions. To quantify the amounts of 37a relative to total Ca V 2.2 mRNA, we used a set of primers that amplify e37a. A forward primer was designed to target the splice junction between e36 and e37a (e36-37a: CTGCGTG. TTGCCGGATT) and a reverse primer to target a sequence within e37a (e36-37aR: 5'ACCTACGAGGG-CAGTTCTT). The e36-37a amplification was normalized to a PCR reaction using a set of primers that amplify between constitutive exons 35 and 36 (e35-36F: 5′ GGAAACATTGCCCTTGATGATG, e35-36R 5′ CAGTGGCACTCCTGAACAATA. Fig. 6c). The amplification efficiencies of both sets of primers were tested using serial dilutions of cDNA obtained from DRG samples (Fig. 6d). First-strand cDNA was diluted 1:5 and 4 μL of this dilution were used in a 20 μL qPCR reaction containing 10 μL of EvaGreen® 2X mastermix (Biotum, 31,003) and 0.8 μL of forward and reverse primers (10 μM). RT-qPCR reactions were run on an ABI 7500 Fast Real-Time PCR system (Applied Biosystems) with the following conditions: 1 cycle 95°C for 2 min, 45 cycles (95°C for 15 s and 60°C for 1 min). Each sample from at least five different mice per genotype (biological replicates) was run in triplicate (technical replicates). Ct values were determined by 7500 Software v2.3 (Applied Biosystems). Relative quantification of gene expression was performed with the 2 -ΔΔCt method [37]. To confirm band identity, all bands were cloned and sequenced using NEB® PCR Cloning Kit (NEB, E1202S). End-point PCR was performed to validate the specificity of primers directed to 37a (Fig. 6c). We used 37a-Cacna1b cDNA clone (Addgene, plasmid #26569) and 37b-Cacna1b cDNA clone (Addgene, plasmid #26571). Glutamate decarboxylase 2 (Gad2) mRNA was quantified using TaqMan® real-time PCR assays (ThermoFisher Scientific) with probe Mm00484623_m1. Levels of mRNA were normalized to glyceraldehyde 3-phosphate dehydrogenase (Gapdh) using the probe Mm99999915_g1. Cycling conditions were similar to those described above.

Behavioral assays Elevated plus maze
After 30 min of habituation to the testing room, mice were placed on the elevated plus maze. Mice were recorded for 10 min with an infrared-sensitive digital camera in dimmed light. After each trial, the maze was cleaned with clorhexidine gluconate to reduce odor cues by previous subjects. The time spent in open areas and the frequency of boundary crosses were recorded and analyzed with Ethovision XT 8.0 (Noldus, Leesburg, VA). Exploratory behavior was measured by percent time spent in and percent of entries into the open arms.

Novelty induced Hypophagia
Prior to testing, mice were housed 2-3 per cage. Mice received 3 consecutive days of training (days 1-3) in a dark room in their home cage to find a reward (sweet milk). Training sessions consisted of presenting mice with a standard dual bearing sipper tube (Quick Quench 5 oz. bottle) inserted between the wire bars of the cage roof and containing 1:3 sweetened condensed milk with water. Mice were trained to find the sweet milk in their home cage (HC) in the dark for 3 days. On the fourth day mice were tested in their HC in the dark room. For testing, all but one of the mice were removed from their HC and placed into a holding cage containing shavings taken from its own HC. The mouse left in the HC was exposed to the sweet milk solution. This indicates the beginning of the trial. The latency to drink was recorded over a 5-min period with an infrared camera. Following completion of the 5-min trial, the second and third (if applicable) animals were rotated into the HC and tested in the same manner. On fifth day, novel cage (NC) testing was conducted by placing a single mouse into a clean cage of the same dimensions as their HC, but with no shavings and under bright lighting conditions. Mice were again presented with a sipper tube bottle of diluted sweetened milk, and the latency to drink was determined off-line using a digital timer. Novelty induced anxiety-like behavior was measured by the relative difference between HC and NC in the latency to approach reward.

Brain slice electrophysiology
Briefly, adult mice were deeply anesthetized with isofluorane. The brain was immediately removed and rapidly placed in ice-cold artificial cerebrospinal fluid ACSF (mM): NaCl, 119; NaHCO 3 , 26; KCl, 2.5; NaH 2 PO 4 , 1; CaCl 2 , 2.5; MgSO 4 , 1.3; Dextrose; 11.0. 400 μm coronal slices from vHPC were prepared using a Leica 1000 vibrating slicer. After cutting, slices were held 1 h at room temperature in ACSF with constant oxygenation, then transferred to the recording chamber with continuous perfusion (1-2 ml/min). Field excitatory postsynaptic potentials (fEPSPs) were recorded by placing a microelectrode filled with ACSF into the superior blade of DG. A bipolar tungsten stimulation electrode (FHC Inc) was placed in nearby in this layer of DG and the medial perforant path (mPP) was identified by pre-pulse inhibition. Stimuli were applied using a current stimulus isolator (A365, WPI Inc). To ensure consistency in current application the output current from the stimulus unit was measured before the experiment was performed. Recordings were filtered at 2 kHz and digitalized at 20 kHz using a multiclamp 700A amplifier and acquired with Clampex 10.2 (Molecular Devices). The initial fEPSP slope was measured by fitting a straight line using Clampfit 10.2 (Molecular Devices). At least 2 slices were recorded per mouse per genotype.

Statistical analysis
Student's t-test and ANOVA repeated measures were performed using SPSS Statistics (IBM). The genotype was unknown to the experimenter during data acquisition and during analysis for all experiments.

E37a-Cacna1b mRNAs are expressed in brain
We were interested in knowing if e37a-Cacna1b mRNAs were expressed in discrete regions of the brain. We used exon-specific primers for RT-PCR to test if e37a-Cac-na1b mRNAs are expressed in medial prefrontal cortex (mPFC), ventral hippocampus (vHPC), and amygdala (AMY). We RT-PCR amplified RNA from microdissected tissue using primers in constitutive exon 35 and alternative exon 37a (Fig. 1a). PCR products of the expected size (~340 bp) were amplified from all three brain regions (Fig. 1b, lanes 1, 3, 5). BsrGI digest of PCR amplified products generated two bands of the predicted size for e37a (~280 and~60 bp) for all three brain regions and 37a-Ca V 2.2 cDNA control (Fig. 1a, lanes 8, 9,  10 and 11, respectively).
To obtain more information about the cell-specific expression pattern of e37a-Cacna1b mRNA we used Base-Scope™ in situ hybridization. The specificity of e37aspecific probes was confirmed using brain sections of mice null for e37a (e37b-only mice) [19]. In WT mice, e37a-signal was observed in cell nuclei likely because transcription and mRNA splicing are coupled [38]. In the vHPC, the e37a signal was observed in nuclei of CA1 and CA3 pyramidal layers, stratum pyramidale (s.p.), as well as in interneuron populated layers, stratum radiatum (s.r.), lacunosum-moleculare (s.l.m.), stratum lucidum (s.l.), and stratum oriens (s.o.) (Fig. 2b). mPFC and neocortex contained e37a-Cacna1b mRNA puncta in nuclei, but the e37a-Cacna1b expression pattern was distributed through all cortical layers except for layer I (LI), and not restricted to any specific cell layer (Fig.  2c). E37a-containing nuclei were scattered throughout AMY as well (Fig. 2d). To assess the specificity of the e37a probe, we compared the percentage of nuclei showing signal for e37a in sections from WT and e37bonly mice. We found that~8.5% of nuclei were positive for e37a in sections from WT mice, compared to1 .6% in e37b-only mice (% e37a + nuclei ± SE. WT = 8.5 ± 0.55, n = 6; e37b-only = 1.7 ± 0.30, n = 4. Mann-Whitney test, p < 0.003, Fig. 2e). These results indicate the majority of the signal is specific for e37a-Cacna1b mRNA, although~20% of the signal is non-specific.
Similar to the vHPC, we observed an enrichment of e37a-Cacna1b in dorsal hippocampus (dHPC) in s.p., s.r., s.l.m. and s.o. layers (Fig. 3a). In the DG, e37a signal was observed in the granular cell layer (g.c.l.) and hilus Fig. 1 e37a-Cacna1b splice isoforms are expressed in mPFC, vHPC and AMY. a Schematic showing the approximate location of PCR primers to amplify e37a-containing cDNA (horizontal arrows), the approximate location of a unique BsrGI site within the e37a sequence (vertical arrow), and the expected size for BsrGI digested products. b Representative image of an agarose gel electrophoresis of undigested PCR products of RT samples from mPFC, vHPC and AMY (lanes 1, 3, and 5 respectively). Negative controls for each RNA sample without reverse transcriptase (lanes 2, 4, and 6 respectively). Positive control with a plasmid that contains cDNA for e37a-Cacna1b (lane 7). BsrGI digested PCR products derived from RT-PCRs from mPFC, vHPC, AMY, and e37a-Cacna1b plasmid (lanes 8, 9, 10, 11 respectively) (h.) but not in the molecular layer (m.l.) (Fig. 3a). Some background e37a signal was observed in similar brain areas in e37b-only mice, however this represented only2 0% relative to brain sections from WT mice (Fig. 3b). Our results show that e37a-Cacna1b mRNA is expressed in cortical and hippocampal areas including in projection neurons in s.p. Given the presence of projection neurons in s.p., we tested if e37a-Ca V 2.2 channels contribute to synaptic responses at glutamatergic synapses.

37a-Ca V 2.2 channels influence release probability at medial perforant path-dentate gyrus synapses
Glutamate release at two key synapses of hippocampus, mPP-DG and Schaffer Collaterals-CA1 (SC-CA1) depend, in part, on Ca V 2.2 channel activity [39]. Therefore, we compared synaptic transmission at mPP-DG and SC-CA1 synapses in vHPC of WT and e37b-only mice (Fig. 4a). We recorded fEPSPs by stimulating mPP and SC and recording in DG and CA1, respectively (Fig.  4a). fEPSPs at mPP-DG and SC-CA1 synapses were not obviously different between WT and e37b-only mice based on fEPSP input/output relationships (I/O). The relationship between fEPSP slope and electrical stimulation intensity were similar between WT and e37b-only mice for both synapses (F 1, 13 = 2.4, p = 0.941, repeatedmeasures ANOVA, WT: n = 8; e37b-only: n = 7. Fig. 4b) and (F 1, 13 = 4.9, p = 0.85, repeated-measures ANOVA, WT: n = 8; e37b-only: n = 7. Fig. 4c). We have previously shown that Ca V 2.2 protein levels are similar between WT and 37b-only mice in the brain ( [19], supplementary Figure 3). Those results combined with our synaptic physiology suggest that the absence of 37a-Cacna1b mRNA does not affect basic synaptic function of mPP-DG and SC-CA1 pathways. To determine if e37a-Ca V 2.2 channels have a presynaptic role, we performed pairedpulse ratio (PPR) at these synapses as described below.
Presynaptic Ca V 2.2 channels influence short-term plasticity [40,41], which is a property linked to presynaptic calcium entry. Given that e37a influences both the number of Ca V 2.2 channels trafficked to the cell surface as well as GPCR inhibition of Cav2.2 channel function [19,28], we analyzed short-term plasticity using PPR at mPP-DG and SC-CA1 synapses of WT and e37b-only mice. We used a stimulus intensity adjusted to 30-40% of maximum for each individual recording (Fig. 4b, c). At mPP-DG synapses, we observed a consistent difference in the size of paired-pulse inhibition at short intervals (50-200 ms) between WT and e37b-only recordings (F 1, 13 = 0.75, p = 0.01, repeated-measures ANOVA, WT: n = 8, e37b-only: n = 7). The average PPR at wild-type synapses was 0.9 compared to 0.7 at synapses of e37bonly mice. Thus, our data suggest that at short intervals, transmitter release probability is higher at mPP-DG synapses of e37b-only mice compared to WT (e37b plus e37a). At longer stimulus intervals, (0.5-1 s), there was no consistent difference in the size of paired pulse inhibition between WT and e37b-only recordings (F 1, 13 = 1.5, p = 0.9, repeated-measures ANOVA, WT: n = 8, e37bonly: n = 7). SC-CA1 synapses facilitate in response to paired stimuli applied at intervals shorter than 200 ms. There were small, but inconsistent differences in paired pulse facilitation of fEPSP at SC-CA1 synapses from WT and e37b-only mice (F 1, 13 = 1.98, p = 0.84, repeatedmeasures ANOVA, WT: n = 8, e37b-only: n = 7) and at longer intervals where there was no facilitation, thus PPRs were not consistently different in WT and e37bonly recordings (F 1, 13 = 1.03, p = 0.92, repeatedmeasures ANOVA, WT: n = 8, e37b-only: n = 7).
Our findings suggest that mPP-DG synapses containing e37a-Ca V 2.2 channels as compared to all e37b-Ca V 2.2 containing synapses, have reduced release probability. This influences the size of the synaptic response when stimuli occur in rapid succession (intervals 200 ms or less). This result might be consistent with enhanced inhibition of e37a-Ca V 2.2 channels by GPCRs compared to e37b-Ca V 2.2 channels at mPP termini, as observed in studies of e37a and e37b clones expressed in cell lines [19,28,29]. By contrast, either e37a-Ca V 2.2 channels are not expressed at SC termini or their unique properties do not influence synaptic transmission at SC-CA1 synapses based of the experiments performed in our studies. We next analyzed the subcellular distribution of e37a-Cacna1b mRNAs in EC which contains cell bodies of mPP axons that project to DG [42][43][44].

e37a-Cacna1b mRNAs are enriched in CaMKIIα expressing projection neurons
First, we probed for e37a-Cacna1b mRNAs using Base-Scope™ in sections of EC. We found low levels of e37a-Cacna1b expression across the multiple layers of EC with the exception of LI (Fig. 5). To determine which cells express e37a-Cacna1b mRNA, we used a combination of genetic labeling and FACS coupled with RT-qPCR (Fig. 6a). All three Ca V 2 channels (Ca V 2.1, Ca V 2.2, and Ca V 2.3) contribute to synaptic transmission at excitatory SC-CA1 and mPP-DG synapses [45][46][47]. By contrast, synaptic transmission at CCK + INs termini which synapse onto granular cells of DG depend exclusively on the activity of Ca V 2.2 channels [48][49][50]. We compared e37a-Cacna1b mRNA expression in PNs and CCK + INs using mice expressing the red fluorescent protein, tdT, in CaMKIIα-expressing cells to mark PNs (CaM-KIIα + PNs) [51] and in CCK-expressing cells to mark CCK + INs (Fig. 6a). Figure 6a shows a schematic of the genetic labeling for CaMKIIα + PNs and intersectional labeling of CCK + INs (and see methods). We used the interneuron marker, Gad2 to demonstrate enrichment of this mRNA in CCK + INs relative to CaMKIIα + PNs (Fold change, mean ± SE. CCK + IN = 6.96 ± 1.56, n = 7; CaMKIIα + PN = 1 ± 0.41, n = 5. Mann-Whitney U test, p = 0.012, Fig. 6b).

E37a-Ca V 2.2 influences behavioral responses to aversive stimuli in mice
To determine if the presence of e37a-Ca V 2.2 channels are important for certain aspects of mouse behavior, we compared WT and e37b-only mice in a series of exploratory and novelty-induced anxiety-like behaviors. E37b-only mice were compared to WT mice in elevated plus maze (EPM) and novelty-induced hypophagia (NIH) assays. We first tested WT and e37b-only male mice in C57BL/6;I-129 background. In EPM, e37b-only mice entered more frequently and spent more time in open arms (OA) relative to WT (% entries in OA, mean ± SE: WT = 30.8 ± 1.7%, n = 8; e37b-only, 43.7 ± 5.7%, n = 7. Mann-Whitney U test, p = 0.004. Fig 7a, left panel. % time spent in OA, mean ± SE: WT = 7.2 ± 1.6, n = 8; e37b-only = 17.0 ± 3.9, n = 7. Mann-Whitney U test, p = 0.021. Fig. 7a, middle panel). By comparison, no difference was detected in overall distance travelled in the elevated plus maze between WT and e37b-only mice  Fig. 7a, middle panel). Interestingly, e37b-only female mice showed reduced locomotor activity compared to WT female mice (Distance traveled, mean ± SE: WT = 32.74 ± 3.09 m, n = 8; e37bonly = 26.53 ± 2.14 m, n = 7. Mann-Whitney U test, p = 0.041. Fig. 7a, right panel). Our results suggest that e37a-Ca V 2.2 channels have an inhibitory influence, relative to e37b-Ca V 2.2 channels, on exploratory behavior in elevated plus maze in both males and females. Furthermore, we also show that e37a-Ca V 2.2 channels enhance overall locomotion in female mice, but not male mice compared to e37b-Ca V 2.2 channels.
Next, we assessed the ability of mice from both genotypes to resolve a conflict between a context that induces heightened aversion and a drive to approach an appetitive stimulus in the NIH assay (novelty-induced anxiety-like behavior). In this task, over the course of 3 days, rodents learn that a sipper tube delivers sweetened condensed milk. On the fourth day, latency to approach and drink in the home cage (HC) is measured. On the fifth day, mice are placed in a novel cage with a (NC) mildly aversive environment, and latency to drink is assessed. Mice typically show increases in latency to approach and drink from the sipper tube in this novel aversive environment, a marker of novelty induced anxiety-like behavior [52]. To determine the levels of novelty-induced anxiety-like behavior, time to approach in HC was subtracted from time to approach in NC (Δ Time to approach). E37b-only male mice in C57BL/6: I129 background overcame the aversive environment faster than WT mice from similar background (Δ Time to approach, mean ± SE: WT = 90.9 ± 20.3 s, n = 7; e37b- Gad2 mRNA expression was normalized to Gapdh mRNA levels. c Upper panel. Approximate location of primers (arrows) to amplify a sequence spanning e35-36 and e36-37a. Lower panel. Representative image to show the specificity of both sets of primers. Note the lack of amplification for e36-37a set of primers in the presence of the e37b-Cacna1b clone. As expected, e35-36 primers amplified both e37a-Cacna1b and e37b-Cacna1b clones. d Upper panel. Melting curve for e36-37a and e35-36 sets of primers. The derivative of fluorescence as a function of temperature was plotted (−dF/dT) against temperature. One single peak in each plot strongly suggest the presence of one product of amplification for both e36-37a and e35-36 sets of primers. Lower panel. Standard curves to assess the PCR efficiency for e36-37a and e35-36 sets of primers. Open circles indicate individual measurements of Ct values at a given dilution. All points for each sets of primers were considered to calculate slope of standard curve and PCR efficiency. e Quantification of e37a-Cacna1b mRNA in RNA isolated from CaMKIIα + PNs and CCK + INs. Data are shown as mean (filled symbols) ± SE, and individual values for each mouse (empty symbols). * p < 0.05 only, 20.4 ± 7.7 s, n = 7. Mann-Whitney U test, p = 0.01. Fig. 8a, right panel). We confirmed this result in e37bonly and WT mice in C57BL/6 (Δ Time to approach, mean ± SE: WT = 124.3 ± 23.0 s, n = 15; e37b-only = 57.2 ± 11.9 s, n = 14. Mann-Whitney U test, p = 0.036. Fig. 8b, right panel). Similarly, latency to approach was shorter for e37b-only female mice compared to WT female mice in C57BL/6 (Time to approach, mean ± SE s: WT = 156 ± 29 s, n = 6; e37b-only = 53 ± 30 s, n = 7. Mann-Whitney U test, p = 0.016. Fig. 8b). Our results suggest that e37b-only mice are more likely to approach the reward than WT mice despite the aversive environment.

Discussion
Cell-specific expression of e37a-Cacna1b pre-mRNA in brain It was previously shown that e37a-Cacna1b mRNA is more abundant in DRG than in brain;~6% to~1.2% of the Cacna1b splice isoforms contain e37a in DRG and brain respectively [27]. However, alternative splicing within cell populations of the brain varies substantially [53]. Here, we show that~8.5% of cells in cortical areas contain e37a, and these cells are distributed throughout the cortex. It is important to note that approximately1 .7% of cells in sections from e37b-only mice showed  Alternative splicing in Cacna1b pre-mRNA and control of transmitter release Interestingly, the other two members for the Ca V 2 family, Cacna1a (Ca V 2.1) and Cacna1e (Ca V 2.3) also contain alternatively spliced exons that are homologous to e37a and e37b in Cacna1b [25,54]. Recently, Thalhammer et al. demonstrated that switching e37a to e37b splice variants in Cacna1a resulted in reduced probability of transmitter release in excitatory synapses. This was attributed to differential coupling of e37 Cacna1a splice variants to the neurotransmitter release machinery [55]. Given that e37b-only mice in our study show reduced PPR compared to WT mice, our results suggest that switching splicing from e37a to e37b produced an increase in probability of transmitter release. Our findings oppose to those ones for 37 Cacna1a splice variants.
The following can help to reconcile these apparently contradictory findings. e37a in Cacna1b is thought to enhance inhibition of calcium entry through Ca V 2.2 channels by G i/o PCRs [19,28]; therefore, by eliminating e37a-Ca V 2.2 channels there might be a reduction in the inhibition of presynaptic calcium entry by G i/o PCRs, and therefore increased probability of transmitter release. Nonetheless, all these findings open the door for future exciting research on the role of splice variants of Ca V 2 channels in neurotransmission.
Cacna1b pre-mRNA alternative splicing and behavior Here we show that although e37a-Cacna1b is expressed at very low levels in the brain [27], it has robust effects on the response to aversive stimuli in both males and females and on overall locomotion in female mice. The behavioral effects are not related to changes in overall protein levels because the amount of Ca V 2.2 is similar in whole brain of both 37b-only and WT mice ( [19], supplementary Figure 3). Developmental compensation by other Ca V s is unlikely to explain our behavioral observations, because the protein levels of Ca V 2.1 channels is similar in whole brain of 37b-only and WT mice ( [19], supplementary Figure 3). Furthermore, the non-N-type current recorded in DRG in both new born and adult mice is similar between 37b-only and WT mice, which provides more support to the lack of developmental compensation by other Ca V s in our genetic mouse models [19,31]. All these observations are also in line with our electrophysiological results, where we were unable to observe differences in I/O relationships in excitatory synapses of vHPC.
The mechanisms underlying the behavioral effects of disrupting e37a splicing are yet to be determined. Our studies in hippocampus suggest a link between Cacna1b alternative splicing, hippocampal function, and behavioral responses to aversive stimuli. This link is supported by previous studies where direct excitation of DG increases exploratory behavior in rodents in the elevated plus maze [56,57]. However, a limitation of our study is that we are unable to conclusively determine what areas of the nervous system are affected by disruption of e37a splicing in Cacna1b to influence exploratory behavior, given that e37a-Cacna1b is broadly expressed in CaM-KIIα + PNs and CCK + INs, and possibly other types of interneurons. Nonetheless, our study is one of the very few linking single splicing events in Cacna1b to complex behavior.