Functional characterisation of single nucleotide variants of the psychiatric risk gene cacna1c in the zebrafish

Several genome-wide association studies have associated CACNA1C variants with psychiatric disorders. The molecular mechanisms involved are poorly understood. Taking advantage of the zebrafish larva as a model, we investigated how two different mutations in cacna1c – sa10930 (nonsense mutation) and sa15296 (splice site mutation), affect neuronal function. We characterized changes in cacna1c mRNA, neurotransmitter levels and behaviour, as well as whole-brain activity using single electrode local field potential recordings. Both point mutations resulted in a significant reduction in cacna1c mRNA, as well as social behaviour and prepulse inhibition deficits. Whereas sa15296 mutants displayed abnormal locomotor and open-field behaviour, we observed normal behaviour in the sa10930 mutants. Brain recordings from both mutants had lower spectral power while sa15296 displayed significant seizure-like activity. Finally, sa10930 homozygotes showed increased dopamine and serotonin levels, decreased gamma-aminobutyric acid (GABA) levels, and unchanged glutamate levels while homozygous sa15296 larvae showed increased levels of serotonin and glutamate, and unaffected levels of GABA and dopamine. Our work provides new insights into the functional role of CACNA1C in behavioural, electrophysiological and biochemical traits linked to psychiatric disorders. We show a functional role for the non-coding mutation (sa15296) in the cacna1c in vivo animal model. Consistent with existing hypotheses, our data suggest that disruption of gene expression, neurotransmission, and cortical excitability are involved in CACNA1C-related mechanisms of psychiatric disorders.


Introduction
Neuropsychiatric disorders contribute to about 14% of the global burden of diseases [1]. Although the aetiology of these disorders is complex, involving the interplay of both genetic and environmental factors [2], their heritability is high. For schizophrenia (SCZ), bipolar disorder (BD) and autism spectrum disorders (ASD), heritability is estimated at up to 80% [3]. Many putative psychiatric susceptibility genes have been identified over the last decade via genome-wide association studies (GWAS) [4,5]. One gene implicated by GWAS in mental disorders is CACNA1C, which encodes the alpha-1 subunit, CaV1.2, of the L-type voltage-dependent calcium channel (LTCC) [6]. Although GWAS have implicated CACNA1C in mental disorders, the specific underlying molecular pathogenesis mechanisms are still unknown. We, therefore, investigated in this study, how mutations in CACNA1C affect psychiatry-related phenotypes in vivo using the zebrafish model.
CACNA1C is expressed in the brain, heart, smooth muscle, and endocrine cells [6,7]. Neuronal CaV1.2 channels are involved in gene expression regulation, neurotransmitter release, and integration of dendritic information in the brain [6][7][8]. In particular, neuronal CaV1.2 are involved in several processes relevant to psychiatric disorders such as learning, memory, and brain development [7,9].
Variations in CACNA1C have been associated with BD, SCZ and ASD, as well as other psychiatric disorders including attention deficit hyperactivity disorder (ADHD), and major depressive disorder [4,5,7]. How single nucleotide polymorphisms (SNPs) relate to molecular mechanisms associated with these disorders remains unclear. Interestingly, while some of the variants are within gene coding regions, most of the variants identified thus far are within non-coding regions [7,10,11]. The SNP rs1006737 confers a significant risk for BD, SCZ and ASD [12][13][14][15]. rs1006737 lies within intron 3 (a non-coding region) of the CACNA1C gene and is thought to be associated with altered gene expression [11,16,17]. The SNP rs4765905 is reported to affect interactions with the CACNA1C promoter region, thus, leading to an alteration in its expression [11]. Furthermore, two nonsense mutations in CACNA1C predicted to result in loss-of-function (LOF), have been identified through whole-exome sequencing in a SCZ population [10]. In addition, missense mutations in exons 8 or 8a of CACNA1C cause a rare subtype of ASD called Timothy syndrome 1 via a gain-of-function (GOF) mechanism [9,18]. Thus, there is a large interest in characterising the consequences of CACNA1C aberrations, as recently illustrated [19].
Although SCZ, BP and ASD are three distinct disorders, several studies have suggested that they have shared genetic risks, common pathologies and symptoms [20,21]. The association of CACNA1C SNPs with multiple disorders suggests shared genetic convergence across disorders [5] and thus highlights the need to investigate the functional roles of CACNA1C using relevant disease endophenotypes rather than focusing on rigid disease classification.
Most of the functional roles of CaV1.2 in psychiatric disorders have been obtained from genetic rodent models [9,19] and to a lesser extent using zebrafish genetic models [22,23]. Except for a recent aforementioned study [23], the role of cacna1c within the context of normal brain function, behaviour, and disease in zebrafish, has been underexplored. Furthermore, although most of the SNPs identified in humans are located in non-coding regions [11], there is no genetic animal model reported with mutations in non-coding regions of Cacna1c. The goal of the present study was to investigate how different mutations in cacna1c affect brain function in zebrafish larvae. Two mutant zebrafish lines generated via N-ethyl-N-nitrosourea (ENU) mutagenesis were obtained from the zebrafish international resource centre. The first line, sa10930, results in a premature stop codon in exon 6, whereas the second line, sa15296, carries an essential splice site mutation in intron 35 [24]. To understand the relationship between the gene variants and central nervous system (CNS) function, we examined the morphological, molecular, behavioural, electrophysiological, and biochemical characteristics of mutant larvae versus their wild type (WT) counterparts. Together, these comprehensive experiments provide new knowledge on how variants of cacna1c may contribute to psychiatric disease.

Zebrafish mutant lines and husbandry
Two mutant lines cacna1c sa10930 (hereafter sa10930 for homozygotes and sa10930/WT for heterozygotes) and cacna1c sa15296 (hereafter sa15296 for homozygotes and sa15296/WT for heterozygotes) that were generated by ENU mutagenesis from the Zebrafish Mutation Project (Sanger, UK) were obtained as fertilised embryos from the Zebrafish International Resource Center (Eugene, Oregon, USA) and raised to adulthood.
Adult zebrafish stocks were maintained under standard conditions [25] in an approved fish facility with a 14 hours' light and 10 hours' dark cycle. Fertilised eggs from the natural spawning of adult fish lines were collected, transferred to petri dishes (n = 60), and raised in an incubator at 28°C in E3 medium. All the experiments were approved by the Norwegian Food Safety Authority experimental animal administration's supervisory and application system (FOTS-18/106800-1; ID 15469 and 23935).

Morphological analysis
Live anaesthetised larvae were mounted using 2% methylcellulose. Images of larvae were taken with the Leica M205 FA microscope at the same resolution for direct comparison.
Mannitol exposure Larvae were exposed to 250 mM mannitol (#M4125, Sigma) as described in [26] from 3 days postfertilization (dpf) and replenished daily until the last day larvae were used for experiments. The plates housing larvae were changed after every two days to avoid solute crystals that form due to evaporation of the solution. 3 dpf was chosen as the treatment start day because oedema was visible from this day onwards.
Whole-mount in-situ hybridization (WISH) WISH was used to investigate the spatial gene expression patterns of cacna1c using digoxigenin labelled riboprobes as earlier described [27]. Briefly, 5 and 7 dpf larvae were fixed in 4% (Thermofischer) and the agarose bleach gel method as described previously [28]. cDNA was synthesized from 0.5 µg RNA templates using the SuperScript First-Strand Synthesis System for rt-PCR (Invitrogen) as described by the manufacturer. The cDNA was diluted by a factor of 60.
The rt-qPCR was performed in triplicates with 9 µl of the diluted cDNA and 2× PowerUp Sybr

Locomotor activity
For measuring locomotor activity, larvae (6 dpf) were acclimated 15 min in the ZebraBox (testing chamber) in a 48 well plate under the same illumination conditions (light or dark) as the tracking conditions followed by a 10 minutes recording session in a 1 min time bin. The parameters measured were the average total distance travelled in millimetres (mm) and the duration spent in inactivity in seconds (s).

Light-dark transition test
For the light-dark transition test, larvae (6 dpf) were acclimated in the test chamber for 15 minutes followed by 10 min each of tracking in (1) darkness (0 % light) 2) 100% light, and (3) darkness.
Larval locomotor activity was measured as the total distance moved (mm) over either a 10 min period in time bins of 1 min or a complete recording period.

Thigmotaxis
Thigmotaxis was measured using a 24 well plate (diameter = 16.2 mm) as the open field. Each well of the plate was divided into an outer and inner zone (diameter of inner zone = 8 mm, inner zone distance in relation to the outer zone = 4 mm) and thigmotaxis was calculated as the total distance moved (TDM) in the outer zone as described by [29,30]. The preference of each fish to remain at the periphery or explore the centre of the arena was monitored for 20 minutes of spontaneous behaviour in either light or dark conditions. Larvae aged 6 dpf were acclimated 15 min in the test chamber before the start of recording.

Shoaling
The shoaling assay was performed using a round dish of diameter 60 mm, height 15 mm, and volume 30 ml (82.1194.500, Sarstedt). Larvae (7 dpf) were gently placed in the center of the testing arena (5 larvae/arena, n = 10-14/group) and acclimated for 15 min followed by a 20 minute recording session at 30s-time bin to obtain a time-series data using the ZebraLab software. The following social dynamics: nearest neighbor distance (NND) and inter-individual distance (IID) of the shoal were measured as previously described [31,32].

Acoustic startle response (ASR) and PPI
The acoustic startle response and PPI of 6 dpf larvae were evaluated using the ZebraBox Revo (ViewPoint, France) and EthoVision (Noldus, Netherlands) as we previously showed [33]. Single larvae were placed in individual wells of a custom-made plexiglass plate (33-wells in a 96 format).
Larvae were acclimated in a 100 Lx illuminated ZebraBox 5 min before the onset of the experiment. A 660 Hz startle stimulus of duration 100 ms and 440 Hz prepulse stimulus of duration 5 ms were used. The inter-stimulus interval (ISI) for all PPI experiments was 100 ms.

Local field potential (LFP) analysis
The LFP recordings of 7 dpf larvae were performed as previously described [34]. A glass electrode was filled with artificial cerebrospinal fluid and inserted into the optic tectum of individual larvae aged 7 dpf, immobilised using a thin layer of 2% low melting point agarose. Whole-brain activity was measured for 20 minutes using a MultiClamp 700B amplifier and Digidata 1550 digitizer (Axon Instruments, USA). The Clampfit version 10.6.2 software (Molecular Devices Corporation, USA) was used for processing the LFP recordings. The data were analysed manually by two independent trained observers, blind to the genotype of the larvae.
For spectral analysis, the LFP time series were divided into time windows of 10 seconds with a 50% overlap, and the spectral power was determined for each segment using Welch's method as implemented in the python package scipy.stats. The spectrograms were visually inspected, and the fish that displayed a prominent peak at 2.5 Hz, which we interpreted as a heart-beat component, were excluded from the analysis: this resulted in discarding 4 recordings from WT larvae and 2 recordings from each of the mutant larvae. Moreover, within each sample, the time windows where the total spectral power was more than twice the median spectral power across the time windows were discarded, along with two neighbouring time windows before and after the high-power window. This was done to remove transient artefacts. The remaining time windows were pooled across the fish of the same genotype. This resulted in 1838 spectral samples for the WT larvae, sample was then normalized by the total spectral power within the range 0.5-200 Hz after filtering out the power-grid component at 50 Hz and its duplicates (100, 150 Hz) with a margin of ±0.5 Hz.
To compare the mutant fish spectra with the WT spectra, Wilcoxon rank-sum test (U-test) was used with a threshold p-value of 0.01, which was Bonferroni-corrected by the number of frequency windows (980) to 1.02 × 1 0-5.
For determining the slope of the spectral power decay, we transformed the data (both frequency Students t-test or its equivalent non-parametric test, Mann Whitney was used in comparing two groups. Where more than two groups were compared, one-way analysis of variance (ANOVA) followed by Tukey's post-hoc test was performed. We used two-way ANOVA followed by

Molecular and morphological effects induced by cacna1c mutations
The sa10930 mutant line carries a specific thymidine (T) to adenosine (A) point mutation in exon 6 (NM_131900.1: c.876T>A: p.Tyr292Stop), which results in a premature stop codon (TAA) at amino acid 292 (Fig. 1a). In silico analysis predicts exon 6 to form part of the first transmembrane domain that regulates voltage sensitivity of the CaV1.2 channel (supplementary Fig. S1). Thus, premature termination may result in channel loss of function. On the other hand, the sa15296 mutant line carries a thymidine (T) to cytosine (C) point mutation in the donor splice site between intron 35-36 (NC_007115.7: g.155870T >C), resulting in an essential splice variant (Fig 1a). We predict that the sa15296 mutation may result in exon skipping and/or nonsense-mediated mRNA decay. Either of the aforementioned situations could alter the sensitivity of the channel -whether it leads to overall hypo/hyperactive channel activity is yet to be determined.
To determine the genotype of our experimental animals, restriction enzyme digestion of the PCR products was performed. The sa10930 mutation introduces an MseI restriction site into the cacna1c gene. Digestion of the 161 base pair (bp) long PCR product spanning the mutation with MseI results in two smaller products (109 and 52 bp) in fish carrying the mutation -i.e., two bands for homozygous, three bands for heterozygous, and one band for WT after digestion (supplementary Fig. S2a). On the contrary, the sa15296 mutation did not introduce a novel restriction site. However, the WT sequence already harboured an HphI restriction site. Digestion of the 181 bp long PCR product spanning the area of the mutation with HphI results in two smaller products (112 and 69 bp) in non-mutant fish i.e. one band for homozygous, three bands for heterozygous, and two bands for WT after digestion (supplementary Fig. S2b).
When heterozygous mutants of either sa15296 or sa10930 were in-crossed, the genotypes of the resulting offspring were within the expected Mendelian proportion of 25% WT, 50% heterozygous mutant, and 25% homozygous mutant (data not shown). Heterozygous mutants were morphologically indistinguishable from their WT siblings, were viable, fertile, and survived to adulthood. After 48 hpf, both sa15296 and sa10930 homozygous mutants showed several developmental abnormalities such as oedema of the pericardial region, yolk sac and gut that worsened with age, as well as craniofacial abnormalities and curved body axis (Fig. 1b). We observed no deficits in touch response of both sa10930 and sa15296 mutants in either the heterozygous or homozygous state. Since the mutations were larval lethal, homozygous mutants of the two lines did not survive past 10 dpf.
Our interest in performing a repertoire of behavioural tests across genotypes motivated us to identify ways to reduce oedema severity in homozygous mutants, as this impaired locomotor ability. We achieved this by increasing the osmolarity of the surrounding environment (Fig. 1b).
Maintaining larvae in 250 mM mannitol from 3 dpf significantly decreased the severity but did not prevent the occurrence of oedema or prolong the life span of the mutants. Importantly, mannitol did not affect larval locomotor activity, as this could potentially serve as a confounding factor in interpreting behavioural responses (supplementary. Fig. S3).
We examined the spatiotemporal expression of cacna1c transcript(s) in homozygous sa10930 and sa15296 as well as WT larvae at 5 and 7 dpf, using WISH (Fig. 1d). In WT larvae, cacna1c was expressed in the heart, brain and to a lesser extent in the musculature, which is consistent with previous reports [36]. Overall, there was a low expression of cacna1c in the brain of homozygous sa10930 and sa15296 mutant larvae at 5 dpf with reduced expression sustained up to 7 dpf (the latest developmental stage tested in our experimental setup). There was no visible expression of cacna1c in larvae stained with the sense probes.
Next, we examined whether the decreased expression of cacna1c mRNA in the two homozygous mutant lines was reflected in a change in protein levels. We performed immunostaining using 7 dpf zebrafish whole-lysate with anti-CaV1.2 antibody (Fig. 1e) S4a). Next, we performed immunostaining using anti-CaV1.3 antibody shown to have significantly less sequence homology with Cacna1c ( supplementary. Fig S4b). The results revealed a significant increase in anti-CaV1.3 signal in both homozygous sa10930 and sa15296 mutants relative AB WT (Fig. 1f).

Effects of cacna1c mutations on larval behaviour
To assess the impact of cacna1c mutations on larval locomotor behaviour, larvae (6 dpf . 2a) and average duration of inactivity (H = 6.227, p < 0.05)] (supplementary Fig. S5a) and [sa15296: average total distance (H = 58.34, p < 0.0001) (Fig. 2b)  We evaluated if mutants displayed anxiety-like behaviour by subjecting them to the light-dark test which utilises abrupt changes in illumination from light to dark to elicit startle and/or stress responses, as measured by an exaggerated locomotor activity [37][38][39] and the open field test where we measured thigmotaxis. The normal response for larvae in a light-dark and dark-light switch is hyperlocomotion and freezing respectively [38,39]. Whereas heterozygous mutants of both lines showed a normal response in the dark-light and light-dark switch, homozygous mutants of both lines remained unresponsive to either switch. However, the extent of the response of heterozygous sa10930 in the light-dark switch was lesser than WT [(p < 0.0001), Fig. 2c conditions, although there was a tendency of mutants to stay in the outer zone when compared with their WT counterparts (Fig. 2e). On the contrary, heterozygous sa15296 mutants spent significantly more time in the inner zone in both dark (U = 211, p < 0.001) and light (U = 1291, p < 0.01) conditions than their WT siblings (Fig. 2f) indicating boldness or risk-taking behaviour [40].
Furthermore, because persons with disorders such as SCZ and ASD present social deficits [21], we tested if the cacna1c mutants also displayed social deficits by subjecting them to a shoaling 3 test. We measured two shoaling characteristics -nearest neighbour distance (NND) 4  we observed a significant increase in IID [t (16) = 2.547, p < 0.05] (Fig. 2h) in relation to their WT siblings, which is also suggestive of impaired social behaviour.
Finally, we examined whether our mutants harboured any sensorimotor deficits by exposing them to acoustic startle stimuli and measuring the extent of PPI. The zebrafish, like many animals, has an innate startle response that is attenuated when preceded by a weak, non-startling stimulus [41].
Similar to humans and rodents, the pharmacological agents apomorphine, haloperidol and ketamine modulate the zebrafish PPI thus supporting its translational value [33,41]. Presentation of a prepulse did not attenuate the startle response of both homozygous sa10930 (Fig. 2i) and sa15296 (Fig. 2j) mutants as evidenced by their decreased % PPI when compared to their WT counterparts.
Effects of cacna1c mutations on the whole-brain activity of zebrafish larvae We measured the spectrograms of homozygous mutants of the two cacna1c lines and WT fish and pooled the time-windowed spectra across the fish of the same genotype (see Methods). We band. The spectra of all fish showed large intrinsic variability (Fig. 3a-c). When we averaged the time-windowed data for each fish instead of the pooling procedure, the differences were nonsignificant both for the individual spectral power components (U-test, p > 0.05, Bonferroni corrected) and for the spectral slopes (U-test, p > 0.05). However, when pooled across time windows, homozygous fish from both cacna1c mutant genotypes showed a significantly lower spectral power in the low delta-frequency (0.5-2 Hz) range than WT (Fig. 3d, U-test, p<0.01, Bonferroni-corrected). In other frequency ranges, the mutant fish only showed sporadic differences in the beta and gamma bands (Fig. 3d). The spectral slopes, determined for the frequency range 50 -200, were significantly flatter in the mutants than in the WT fish (Fig. 3e), suggesting increased excitability in the neural circuits of the mutant fish [42]. Taken together, these results show that the cacna1c mutants, despite a large intrinsic variability, display systematic LFP-power deficits in the low delta range and the spectral slopes.
Further examination of the LFP recordings revealed seizure-like discharges in mutant larvae. Oneway ANOVA analysis indicated statistically significant differences concerning the number of seizure-like discharges [F (2, 61) = 10.92, p < 0.001]. The sa15296 mutants discharged more seizure-like activity p < 0.001 when compared to WT larvae whereas there was no statistically significant difference between the sa10930 mutants and WT larvae (Fig. 3f). Sample traces of seizure-like activity of WT and mutants are shown in (supplementary Fig. S6).
Effects of cacna1c mutations on the neurotransmitter levels Imbalances in neurotransmitter levels have long been hypothesised to be associated with the pathoaetiology of SCZ and BD [43,44]. We used HPLC analysis to assess if the neurotransmitter levels of glutamate, GABA, dopamine, and serotonin in the mutants were altered relative to their WT siblings at 7 dpf (supplementary Table T1 & supplementary Fig. 7).
Increased levels of both dopamine (p < 0.05) and serotonin (p < 0.01) and decreased GABA levels (p < 0.0001) were observed, while the levels of glutamate were unaffected (p > 0.05) in the homozygous sa10930 mutants (Fig. 4a). However, in the homozygous sa15296 mutants we observed an increased amount of both glutamate and serotonin at p < 0.05, while the levels of both GABA and dopamine remained unaffected at p > 0.05 (Fig. 4b). Further analysis of the glutamate/GABA ratio, a measure of excitation -inhibition, revealed that both mutant lines had significantly increased glutamate/GABA [(WT vs sa10930/sa10930; p < 0.01) (Fig. 4c) and (WT vs sa15296/sa15296; p < 0.05)] (Fig. 4d).

Effects of cacna1c mutations on downstream molecular targets
Calcium influx through CaV1.2 calcium channels regulates downstream genetic transcription pathways such as brain-derived neurotrophic factor (BNDF) and c-Fos [45,46]. We investigated the influence of low levels cacna1c on bndf and c-fos to find possible molecular pathways that may be responsible for the observed phenotypes so far described. We found a statistically significant reduction in the mRNA levels of bdnf (p < 0.05) but not c-fos (p > 0.05) in homozygous sa10930 mutants relative to WT siblings (Fig. 5a). However, in the homozygous sa15296 mutants, the mRNA levels of both bdnf (p < 0.05) and c-fos (p < 0.05) were statistically significant when compared to their WT siblings (Fig. 5b).

Discussion
Here, we investigated the molecular mechanisms affecting brain function of the psychiatric risk gene CACNA1C in genetically modified zebrafish larvae. We characterized behavioural, electrophysiological and biochemical traits linked to psychiatric disorders in two mutant lines harbouring point mutations in distinct regions (i.e., coding (sa10930) and non-coding (sa15296) regions) of the zebrafish cacna1c gene to gain further information on the functional role of CACNA1C gene variants in brain disorders.
The sa10930 mutation is a nonsense mutation that results in a premature stop codon in exon 6 and reduced gene expression in homozygotes. We found no evidence of exon skipping as a result of the sa15296 mutation although it occurs in the essential splice site. Rather, we observed reduced expression of cacna1c mRNA. Similarly, previous studies identified a non-coding SNP that resulted in reduced CACNA1C expression [11,16,17].  [23]. The aforementioned phenotypes i.e., thigmotaxis and abnormal response to darkness are reminiscent of anxiety-like behaviours in zebrafish [29,38]. Notably, several studies have also reported increased anxiety in mouse models of Cacna1c dysfunction [50][51][52][53]. We excluded homozygous mutants from thigmotaxis measurements because the quantification of thigmotaxis relies on adequate locomotor capabilities [29]. Taken together, these findings suggest an effect on behaviour that may have implications for mental traits and disorders.
Zebrafish are social animals that are attracted to conspecifics and tend to move in a group [31,32].
Although the attraction of larvae to conspecifics starts as early as 6 dpf, shoaling has been shown to develop with age but is evidenced as early as 7 dpf, albeit less robust than at older stages [54].
In zebrafish, the shoaling assay is commonly used in validating models for ASD and to a lesser extent, models for SCZ and intellectual disability [55]. By analysing two shoaling parameters, deficits in NND and IID were found in sa10930 mixed genotype larvae while for sa15296 mixed genotype larvae, deficits were only observed in the IID. It is important to note that NND is independent of shoal size while IID is dependent on shoal size. Hence, some researchers consider IID reflective of shoal cohesion as opposed to NND [31,56]. However, measuring both provide further information about the shoal dynamics. We did not use homozygous mutants in the shoaling test because of their high immobility, which could bias the results. Furthermore, it was impossible to segregate heterozygous mutants from their WT counterparts since they are morphologically indistinguishable. Thus, by taking advantage of the "heterogeneous shoaling paradigm" previously described [57], mixed populations of fish (i.e., a non-homogeneous genotype group made up of heterozygous mutants and WT siblings) were used for the shoaling assay.
If a weaker auditory stimulus (prepulse) precedes a startle stimulus within a period of 30-500 ms, the magnitude of response to the startle stimulus is suppressed in a dose-dependent manner with respect to the prepulse strength. This phenomenon, referred to as PPI, is a quantitative measure of sensorimotor gating and happens already at the first trial, hence, no learning is required.
Sensorimotor gating is representative of the brain´s ability to differentiate between relevant and irrelevant sensory input to make a proper behavioural response [58,59]. Reduced PPI is considered a useful behavioural endophenotype for SCZ in human and animal models [41,60,61]. In SCZ and other psychiatric disorders and their equivalent animal models, deficits in PPI have been reported [41,62]. Thus, defective PPI is indicative of information processing dysfunction [58]. We observed reduced PPI responses in homozygous larvae carrying both sa10930 and sa15296 mutations. This observation is similar to what was earlier reported for cacna1c LOF zebrafish larvae [23] and for humans carrying mutations in other neuropsychiatric related genes [59]. The brain circuitry and neurotransmitter systems mediating PPI in mammals are conserved in the zebrafish [63,64].
We observed a significant reduction in the expression of total bdnf in mutants versus WT using whole larval isolate. There is evidence of reduced Bdnf expression in conditional Cacna1c conditional knockout mice [65], and region-specific (↓ Bdnf in the prefrontal cortex, ↑ Bdnf in the dentate gyrus of the hippocampus) regulation of Bdnf in the Cacna1c heterozygous rats [66]. In humans, risk variants in CACNA1C were associated with decreased expression in CACNA1C and as well as region-specific BDNF regulation (↓ BDNF in the prefrontal cortex and substantia nigra, ↑ BDNF in the dentate gyrus of the hippocampus) [66]. Bdnf deficient mice showed enhanced acoustic startle response (ASR), reduced PPI of ASR and deficits in nesting behaviour (a social behaviour) [58].
Analysis of LFP spectrograms from mutant and WT larval brains was performed to test whether cacna1c mutant larvae display electrophysiological phenotypes that correspond to mental disorder-associated LFP-based phenotypes. Delta-range spectral power and spectral slope, have been shown to be altered in SCZ and ADHD, respectively [67,68]. Both of the cacna1c mutants showed a significantly lower spectral power in the low delta-frequency (0.5-2 Hz) range than WT.
In other frequency ranges, mutant larvae only showed sporadic differences in the beta and gamma band. The spectral slopes, determined for the frequency range 40 -200 Hz, were significantly flatter in the mutant than in WT, suggesting increased excitability in the neural circuits of the mutants [42]. Hyper-excitability of neural circuitry underlies a number of brain disorders, particularly those associated with seizure-like activity [42,69]. Indeed, seizure-like discharges were observed in sa15296 but not sa10930 mutants. The discharges observed in sa15296 mutants generally have a lower amplitude and a shorter duration compared to the EEG patterns of the pentylenetetrazole-induced seizure model or the genetic Dravet model [34,69]. Recent studies have revealed pathogenic variants in CACNA1C associated with epilepsy, some of which involve splice site mutations [70]. The clinical spectrum of CACNA1C variants is highly variable, likely due to a variety of reasons, such as the nature of the mutation -i.e., its location within the protein and structural domain of CaV1.2 and the varying transcript expression in different tissues [70].
The increased neural excitability of mutants suggested by the less negative slopes from the spectral analysis was further supported by 1) the increased glutamate/GABA ratio for both sa10930 and sa15296 mutants relative to their WT siblings from neurotransmitter analysis and 2) increased levels c-fos mRNA. Imbalances in neurotransmitters, in particular dopamine, have long been hypothesized to be associated with the aetiology of mental disorders [43,44]. Disruption in GABA-, glutamate-, and dopaminergic neurotransmission is thought to contribute to perturbed cortical oscillations [71]. A recent paper identified risk genes associated with a wide range of neurotransmitter systems -glutamatergic, GABAergic, dopaminergic, serotonergic, cholinergic and opioid in SCZ [44]. We observed enhanced dopamine and serotonin, decreased GABA, and unaffected levels of glutamate in homozygous sa10930 larvae. In the homozygous sa15296 larvae, enhanced serotonin and glutamate, unaffected levels of GABA, and dopamine were observed.
Calcium channel activity modulates overall levels of neurotransmitters through the activation of G-protein coupled receptors which control the sensitivity of ionotropic neurotransmitter receptors and ion channels [44]. Therefore, mutations in calcium channel genes can alter neuronal excitability and firing patterns possibly through alterations in neurotransmission. Notably, however, neurotransmitter levels were analysed from whole-larvae in our experimental setup and therefore does not provide information regarding tissue-specific changes.
The observed reduced expression of bdnf in sa10930 and sa15296 larvae suggests that decreased cacna1c alters downstream signalling pathways such as bdnf that may be mediating the observed phenotypes herein reported.
Our multi-trait analyses provide interesting insights into the functional role of CACNA1C in psychiatric disorders using LOF mutants. Our earlier computational modelling studies predicted that LOF CACNA1C variants can lead to 1) decreased "single-cell" PPI in pyramidal cells [72,73], 2) decreased cardiac activity [74], and 3) increased delta oscillations (and overall activity) in networks of pyramidal neurons [72]. The present results lend support to the first prediction but are inconsistent with the third prediction. The reason for the incongruency with the predictions on delta-oscillation power in a mammalian cortex may be the large density of SK currents in the mammalian L5 pyramidal cells, which are likely to cause gain-of-function mutations of highvoltage-activated Ca2+ channels to counter-intuitively decrease the neuronal excitability in these cells, and vice versa [74]. It is not known whether the neuronal types giving rise to the LFP signal in the delta frequency range in zebrafish larvae express SK currents. Previous studies of Cacna1c mutations in rodents [75] and clinical samples [76] indicated that LOF mutations were associated with altered sleep quality. Moreover, LOF mutations of CACNA1C decreased high-frequency LFP oscillations during wake and REM sleep in patients [75], where a (non-significant) decrease of low-frequency delta-power can also be observed. Further research is needed to increase our understanding of the effects of CACNA1C mutations on delta oscillations and sleep. As for the second prediction, our finding of decreased mobility of homozygous cacna1c mutant larvae (as a consequence of severe pericardial oedema), may be related to reported cardiac deficits [77], but other causes such as rest behaviour, anxiety-induced freezing behaviour or neuromotor defects are also possible [40].
Even though we confirmed that the mutations harboured by the sa10930 and sa15296 resulted in over 90% decrease (LOF) of cacna1c at the mRNA level, we could not replicate the cacna1c LOF finding at the protein level due to the non-specificity of the anti-CaV1.2 antibodies used. The substantial amount of anti-CaV1.2 signal detected especially in the sa10930 mutant samples could not be Cacna1c because the mutation causes a truncated channel lacking the c-terminus region, which the epitope recognizes. However, considering the high homology of the anti-CaV1.2 epitope to the c-terminus of Cacna1d, it is likely that the observed protein signal is Cacna1d. Upon further analysis, we found the protein level of Cacna1d to be significantly increased when immune-blot lysates were probed with antibodies specific for Cacna1d. These results, altogether, suggest a compensatory upregulation of Cacna1d following cacna1c LOF. Although CACNA1D is the most likely LTCC to compensate for a LOF of CACNA1C [78,79], earlier studies conducted in Cacna1c mice with 1) global happloinsufficiency [51], 2) conditional knockout in neurons [80] and 3) conditional deletion in the hippocampus and neocortex [81] were reported to have unaffected Cacna1d protein levels. The disparity could stem from the fact that the presence of one healthy Cacna1c allele is enough to support essential functions hence no need for any compensatory expression by "sister" proteins while the timing and/or regions of the conditional deletion could be outside critical windows and/or areas that drive the need for any compensatory expression.