Knockout of stim2a Increases Calcium Oscillations in Neurons and Induces Hyperactive-Like Phenotype in Zebrafish Larvae

Stromal interaction molecule (STIM) proteins play a crucial role in store-operated calcium entry (SOCE) as endoplasmic reticulum Ca2+ sensors. In neurons, STIM2 was shown to have distinct functions from STIM1. However, its role in brain activity and behavior was not fully elucidated. The present study analyzed behavior in zebrafish (Danio rerio) that lacked stim2a. The mutant animals had no morphological abnormalities and were fertile. RNA-sequencing revealed alterations of the expression of transcription factor genes and several members of the calcium toolkit. Neuronal Ca2+ activity was measured in vivo in neurons that expressed the GCaMP5G sensor. Optic tectum neurons in stim2a−/− fish had more frequent Ca2+ signal oscillations compared with neurons in wildtype (WT) fish. We detected an increase in activity during the visual–motor response test, an increase in thigmotaxis in the open field test, and the disruption of phototaxis in the dark/light preference test in stim2a−/− mutants compared with WT. Both groups of animals reacted to glutamate and pentylenetetrazol with an increase in activity during the visual–motor response test, with no major differences between groups. Altogether, our results suggest that the hyperactive-like phenotype of stim2a−/− mutant zebrafish is caused by the dysregulation of Ca2+ homeostasis and signaling.


Introduction
In any cell, the precise regulation of Ca 2+ flow is possible because of the existence of different branches of calcium signaling that allow Ca 2+ influx into the cytosol and then into Ca 2+ depots, including the endoplasmic reticulum (ER), mitochondria, and cytosolic Ca 2+ -binding proteins. One such pathway is store-operated Ca 2+ entry (SOCE), which is responsible for restoring Ca 2+ levels in the ER, the largest store of Ca 2+ [1]. The main proteins that are involved in SOCE are stromal interaction molecule (STIM) proteins, Orai and transient receptor potential (TRP) channels. STIMs are localized to the ER membrane and act as Ca 2+ sensors via their luminal EF-hand Ca 2+ -binding site. Upon a decrease in Ca 2+ levels in the ER lumen, the EF-hand changes its conformation, leading to the formation of clusters of STIM molecules and their translocation to the plasma membrane [2][3][4]. In the plasma membrane, STIM proteins form complexes with plasma membrane channels, including Orai (Orai1, Orai2, or Orai3) [5] and TRP (especially members of the TRPC and TRPV families) [6]. Upon these interactions, Ca 2+ enters the cell to refill ER stores via Sarco/endoplasmic reticulum Ca 2+ -adenosine triphosphatase (SERCA) [5,7,8]. Upon cellular stimulation, the opposite process of Ca 2+ release from the ER to the cytosol occurs via such channels as ryanodine receptors and inositol 1,4,5-triphosphate 3 receptors [1,9]. Under normal conditions, the level of Ca 2+ in the cytosol is kept at a very low advantages. For example, partial knockout may prevent lethality. The majority of components of SOCE are present in the zebrafish genome, including stim1a, stim1b, stim2a, stim2b, orai1a, orai1b, and orai2 [46], the expression of which in the zebrafish brain and heads of larvae was confirmed by our research group [10]. Store-operated calcium entry was studied in zebrafish but only at the early stages of embryo development [47]. Tse et al. reported that SOCE-like activity in neurons derived from dissociated neurospheres of zebrafish [48]. Zebrafish Stim1 was shown to be crucial for the axon guidance of motor neurons and the regulation of calcium signaling [49]. We recently showed that the knockout of one orthologue of Stim2, stim2b, caused hyperactivity and increased the susceptibility to seizures in zebrafish larvae [50]. Considering that Stim2 isoforms might have diverse effects on brain activity, the role of Stim2a should also be investigated. Thus, the present study sought to complement our previous work with stim2b knockout zebrafish and form a complete picture of the role of Stim2a in the brain. We combined several behavioral tests with in vivo calcium imaging to reveal that changes in neuronal activity affect behavior in stim2a −/− animals. We also identified some candidate genes that could be responsible for such changes.

Genotyping of stim2a Knockout Zebrafish Line and Phenotypic Analysis
We generated the stim2a −/− zebrafish line using CRISP/Cas9 technology. The deletion of 5 bp in front of the SOAR domain sequence produced an early stop codon. This resulted in the truncation of the target protein at 318 amino acid residues ( Figure 1A,B). Animals with the mutation were identified by genotyping DNA that was isolated from fin tissue using a restriction length fragment polymorphism (RLFP). The 5 bp fragment that is recognized by the MBOII restriction enzyme was absent in stim2a −/− zebrafish ( Figure 1B). Larvae from mutant animals did not exhibit developmental abnormalities and exhibited no changes in viability until 120 hpf (Figure 2A). Adult fish were viable and fertile. No visible anatomical changes were detected. Thus, the lack of stim2a did not induce a significant morphological phenotype ( Figure 2B). advantages. For example, partial knockout may prevent lethality. The majority of components of SOCE are present in the zebrafish genome, including stim1a, stim1b, stim2a, stim2b, orai1a, orai1b, and orai2 [46], the expression of which in the zebrafish brain and heads of larvae was confirmed by our research group [10]. Store-operated calcium entry was studied in zebrafish but only at the early stages of embryo development [47]. Tse et al. reported that SOCE-like activity in neurons derived from dissociated neurospheres of zebrafish [48]. Zebrafish Stim1 was shown to be crucial for the axon guidance of motor neurons and the regulation of calcium signaling [49]. We recently showed that the knockout of one orthologue of Stim2, stim2b, caused hyperactivity and increased the susceptibility to seizures in zebrafish larvae [50]. Considering that Stim2 isoforms might have diverse effects on brain activity, the role of Stim2a should also be investigated. Thus, the present study sought to complement our previous work with stim2b knockout zebrafish and form a complete picture of the role of Stim2a in the brain. We combined several behavioral tests with in vivo calcium imaging to reveal that changes in neuronal activity affect behavior in stim2a −/− animals. We also identified some candidate genes that could be responsible for such changes.

Genotyping of stim2a Knockout Zebrafish Line and Phenotypic Analysis
We generated the stim2a −/− zebrafish line using CRISP/Cas9 technology. The deletion of 5 bp in front of the SOAR domain sequence produced an early stop codon. This resulted in the truncation of the target protein at 318 amino acid residues ( Figure 1A,B). Animals with the mutation were identified by genotyping DNA that was isolated from fin tissue using a restriction length fragment polymorphism (RLFP). The 5 bp fragment that is recognized by the MBOII restriction enzyme was absent in stim2a −/− zebrafish ( Figure 1B). Larvae from mutant animals did not exhibit developmental abnormalities and exhibited no changes in viability until 120 hpf (Figure 2A). Adult fish were viable and fertile. No visible anatomical changes were detected. Thus, the lack of stim2a did not induce a significant morphological phenotype ( Figure 2B).

Higher Thigmotaxis and Hyperactivity in the Open Field Test in stim2a −/− Larvae
To investigate behavior in stim2a −/− fish, we evaluated thigmotaxis in the open field test using 4 dpf old larvae ( Figure 3A). Thigmotaxis reflects the preference of animals for the edge of the well compared with the central area [41]. The experiment was performed for 15 min. The data analysis was divided into three phases, 5 min each. During the first phase, the animals could be assumed to be stressed because they were transferred from the dish to a novel place. Both WT and mutant animals spent more time at the edge of the well compared with the center, which can be interpreted as anxiety-like behavior ( Figure 4B). Mutant fish exhibited higher mobility than WT fish both at the edges and in the center, indicating hyperactivity. During the second phase (5-10 min interval), WT animals began to explore the whole well, manifested by a similar amount of movement in both the center and the edge of the well ( Figure 4A). In contrast, mutant fish still preferred swimming at the edge of the well and avoided the center (Figure 4). In the third phase (10-15 min interval), both groups behaved similarly, spending a similar amount of time in the center and at the edge ( Figure 4B). Mutant fish reached a higher maximum acceleration ( Figure 4C) and spent more time moving at high speed (Supplementary Figure S1A) than the WT during the first 10 min of the experiment. During all phases of the experiment, stim2a −/− mutant larvae covered a significantly bigger total distance than WT fish ( Figure 4D). stim2a −/− mutant larvae moved with a higher velocity (Supplementary Figure  S1B) and spent more time moving (Supplementary Figure S1C). These results indicate that stim2a −/− zebrafish are hyperactive and required more time to habituate to the new environment, reflecting an increase in thigmotaxis.

Higher Thigmotaxis and Hyperactivity in the Open Field Test in stim2a −/− Larvae
To investigate behavior in stim2a −/− fish, we evaluated thigmotaxis in the open field test using 4 dpf old larvae ( Figure 3A). Thigmotaxis reflects the preference of animals for the edge of the well compared with the central area [41]. The experiment was performed for 15 min. The data analysis was divided into three phases, 5 min each. During the first phase, the animals could be assumed to be stressed because they were transferred from the dish to a novel place. Both WT and mutant animals spent more time at the edge of the well compared with the center, which can be interpreted as anxiety-like behavior ( Figure 4B). Mutant fish exhibited higher mobility than WT fish both at the edges and in the center, indicating hyperactivity. During the second phase (5-10 min interval), WT animals began to explore the whole well, manifested by a similar amount of movement in both the center and the edge of the well ( Figure 4A). In contrast, mutant fish still preferred swimming at the edge of the well and avoided the center (Figure 4). In the third phase (10-15 min interval), both groups behaved similarly, spending a similar amount of time in the center and at the edge ( Figure 4B). Mutant fish reached a higher maximum acceleration ( Figure 4C) and spent more time moving at high speed (Supplementary Figure S1A) than the WT during the first 10 min of the experiment. During all phases of the experiment, stim2a −/− mutant larvae covered a significantly bigger total distance than WT fish ( Figure 4D). stim2a −/− mutant larvae moved with a higher velocity (Supplementary Figure S1B

Lower Phototaxis and Hyperactivity in stim2a −/− Larvae in the Light/Dark Preference Test
We then investigated whether stim2a −/− zebrafish larvae had different light/dark preference from WT zebrafish, known as phototaxis [51]. Specifically, we analyzed the amount of time the fish spent in the light half and dark half of the Petri dish ( Figure 3B). Wildtype fish traveled a longer distance ( Figure 5A) and spent more time in the light zone ( Figure 5B). In contrast, the stim2a −/− mutant zebrafish did not exhibit significant phototaxis and swam a comparable distance as the WT fish ( Figure 5A) and spent as much time in either zone of the Petri dish ( Figure 5B). Another difference between the WT and mutant fish was the overall mobility. The total distance traveled in both zones of the Petri dish was approximately four times higher in the mutant fish than in the WT fish ( Figure 5C). These data suggest that stim2a deletion led to hyperactivity and lower phototaxis. movement) while swimming either at the edge of the well or in the center. (C) Box plots show the maximum acceleration (in millimeters per squared second) of WT and stim2a −/− fish in total throughout the well. (D) Heat maps show the total distance covered by the larvae in each phase of the experiment. * p < 0.05, *** p < 0.001 (paired Wilcoxon rank-sum test for comparisons between the edge and center; unpaired Wilcoxon rank-sum test for comparisons between WT and stim2a −/− mutants). n = 32 for WT larvae. n = 33 stim2a −/− larvae. Number of experiments: 3.

Lower Phototaxis and Hyperactivity in stim2a −/− Larvae in the Light/Dark Preference Test
We then investigated whether stim2a −/− zebrafish larvae had different light/dark preference from WT zebrafish, known as phototaxis [51]. Specifically, we analyzed the amount of time the fish spent in the light half and dark half of the Petri dish ( Figure 3B). Wildtype fish traveled a longer distance ( Figure 5A) and spent more time in the light zone ( Figure 5B). In contrast, the stim2a −/− mutant zebrafish did not exhibit significant phototaxis and swam a comparable distance as the WT fish ( Figure 5A) and spent as much time in either zone of the Petri dish ( Figure 5B). Another difference between the WT and mutant fish was the overall mobility. The total distance traveled in both zones of the Petri dish was approximately four times higher in the mutant fish than in the WT fish ( Figure  5C). These data suggest that stim2a deletion led to hyperactivity and lower phototaxis.

The stim2a −/− Zebrafish Larvae Reacted to Light Changes but Exhibited an Increase in Mobility during the Low-Activity Phase
To further characterize the behavioral patterns and check vision in stim2a −/− mutants, we performed the visual-motor response test (see Figure 3C). This test is based on the ability of zebrafish larvae to react to modulations of light intensity by emitting a startle response, similar to masking

The stim2a −/− Zebrafish Larvae Reacted to Light Changes but Exhibited an Increase in Mobility during the Low-Activity Phase
To further characterize the behavioral patterns and check vision in stim2a −/− mutants, we performed the visual-motor response test (see Figure 3C). This test is based on the ability of zebrafish larvae to react to modulations of light intensity by emitting a startle response, similar to masking phenomena in higher vertebrates [52]. In this test, we measured the long-term changes in their activity that resulted from switching the light and dark phases. We analyzed changes in the total distance traveled during three phases of the experiment: baseline, low activity, and high activity ( Figure 6). A decrease in mobility occurred upon the reaction of larvae to the sudden change in light when they were switched to the low activity phase. The zebrafish larvae regained their mobility during the subsequent high activity phase in the dark. activity that resulted from switching the light and dark phases. We analyzed changes in the total distance traveled during three phases of the experiment: baseline, low activity, and high activity ( Figure 6). A decrease in mobility occurred upon the reaction of larvae to the sudden change in light when they were switched to the low activity phase. The zebrafish larvae regained their mobility during the subsequent high activity phase in the dark. Figure 6. The stim2a −/− larvae reacted to changes in the light and exhibited hyperactivity during the low activity phase in the visual-motor response test. Activity was recorded during a 30 min period that consisted of a baseline phase (basal, 0% light illumination), low activity phase (light; 70% light illumination), and high activity phase (dark; 0% light illumination). Box plots show the distance traveled (in millimeters) during the respective phase. *** p < 0.001 (Wilcoxon rank-sum test). n = 53 WT larvae. n = 52 stim2a −/− larvae. Number of experiments: 6.
During the baseline phase, we did not detect any difference in the mobility between WT and mutant larvae. Upon the subsequent increase in light intensity, both genotypes exhibited a decrease in mobility. Interestingly, the stim2a −/− larvae moved a significantly longer distance when the light was turned on. Finally, both WT and mutant zebrafish exhibited an increase in activity during the subsequent high activity phase after turning the light off. However, no significant difference in the total distance traveled was observed between the groups during this phase, similar to the baseline phase ( Figure 6). The results of the visual-motor response test further supported the hyperactivity phenotype of stim2a −/− larvae that was observed in the open field test ( Figure 4) and light preference test ( Figure 5), in which the mutant fish exhibited hyperactivity upon the abrupt switching of the light. Importantly, the mutants were able to distinguish between the light and dark phases as efficiently as the WT animals, indicating that they were able to react to light.

Calcium Imaging Revealed an Increase in Neuronal Activity in the Brain in stim2a −/− Larvae
We performed in vivo Ca 2+ imaging to explore the possible changes in neuronal activity in stim2a −/− zebrafish larvae. We can clearly see that the brain consists of tightly packed somata (almost no extracellular space) in WT and stim2a −/− fish (Supplementary Videos S1 and S2 and Supplementary Figure S3A-B). Spontaneous Ca 2+ oscillations in neurons in WT and stim2a −/− larvae were monitored using the GCaMP5G Ca 2+ sensor. In all the analyzed zebrafish lines, we chose the neuronal cells from the sub-region in the periventricular gray zone of the optic tectum to analyze the spontaneous Ca 2+ activity as it was showing intensive oscillations and was easily identified in each experiment to allow comparisons between them. The average oscillation frequency was significantly higher in stim2a −/− fish compared with WT ( Figure 7A), which indicates that neurons were firing with higher frequency in stim2a −/− larvae because of an increase in neuronal activity in this brain region. However, the average Ca 2+ amplitude was significantly low in stim2a −/− mutants as compared with WT ( Figure 7B). After that, we checked the changes in the Ca 2+ oscillation frequency ( Figure 7C) and average Ca 2+ amplitude ( Figure 7D) after 600 µ M glutamate treatment WT and stim2a −/− larvae (Supplementary Figure 6. The stim2a −/− larvae reacted to changes in the light and exhibited hyperactivity during the low activity phase in the visual-motor response test. Activity was recorded during a 30 min period that consisted of a baseline phase (basal, 0% light illumination), low activity phase (light; 70% light illumination), and high activity phase (dark; 0% light illumination). Box plots show the distance traveled (in millimeters) during the respective phase. *** p < 0.001 (Wilcoxon rank-sum test). n = 53 WT larvae. n = 52 stim2a −/− larvae. Number of experiments: 6.
During the baseline phase, we did not detect any difference in the mobility between WT and mutant larvae. Upon the subsequent increase in light intensity, both genotypes exhibited a decrease in mobility. Interestingly, the stim2a −/− larvae moved a significantly longer distance when the light was turned on. Finally, both WT and mutant zebrafish exhibited an increase in activity during the subsequent high activity phase after turning the light off. However, no significant difference in the total distance traveled was observed between the groups during this phase, similar to the baseline phase ( Figure 6). The results of the visual-motor response test further supported the hyperactivity phenotype of stim2a −/− larvae that was observed in the open field test ( Figure 4) and light preference test ( Figure 5), in which the mutant fish exhibited hyperactivity upon the abrupt switching of the light. Importantly, the mutants were able to distinguish between the light and dark phases as efficiently as the WT animals, indicating that they were able to react to light.
2.5. Calcium Imaging Revealed an Increase in Neuronal Activity in the Brain in stim2a −/− Larvae We performed in vivo Ca 2+ imaging to explore the possible changes in neuronal activity in stim2a −/− zebrafish larvae. We can clearly see that the brain consists of tightly packed somata (almost no extracellular space) in WT and stim2a −/− fish (Supplementary Videos S1 and S2 and Supplementary Figure S3A-B). Spontaneous Ca 2+ oscillations in neurons in WT and stim2a −/− larvae were monitored using the GCaMP5G Ca 2+ sensor. In all the analyzed zebrafish lines, we chose the neuronal cells from the sub-region in the periventricular gray zone of the optic tectum to analyze the spontaneous Ca 2+ activity as it was showing intensive oscillations and was easily identified in each experiment to allow comparisons between them. The average oscillation frequency was significantly higher in stim2a −/− fish compared with WT ( Figure 7A), which indicates that neurons were firing with higher frequency in stim2a −/− larvae because of an increase in neuronal activity in this brain region. However, the average Ca 2+ amplitude was significantly low in stim2a −/− mutants as compared with WT ( Figure 7B). After that, we checked the changes in the Ca 2+ oscillation frequency ( Figure 7C) and average Ca 2+ amplitude ( Figure 7D) after 600 µM glutamate treatment WT and stim2a −/− larvae (Supplementary Video S3 and Supplementary Figure S3C-D). We observed a significant increase in the average Ca 2+ oscillation frequency in the case of the stim2a −/− mutant. However, the difference in average Ca 2+ oscillation amplitude between WT and stim2a −/− mutant after glutamate treatment was not significant. Video S3 and Supplementary Figure S3C-D). We observed a significant increase in the average Ca 2+ oscillation frequency in the case of the stim2a −/− mutant. However, the difference in average Ca 2+ oscillation amplitude between WT and stim2a −/− mutant after glutamate treatment was not significant.

Treatment with Pentylenetetrazol (PTZ) and Glutamate Resulted in Minimal Differences between stim2a −/− Mutants and WT in the Visual-Motor Response Test
We further explored the alterations of neuronal activity, detected by in vivo calcium imaging. We administered two drugs that stimulate neuronal activity, PTZ and glutamate, and performed the visual-motor response test (see Figure 3C). Pentylenetetrazol induces seizures, and zebrafish were shown to exhibit behavioral reactions in response to this treatment beginning at 3 dpf [53,54]. Glutamate increases neuronal activity and locomotor activity [50,55]. Importantly, glutamate was shown to penetrate zebrafish larvae and reach brain neurons. Thus, the main effect of this drug on animal behavior is primarily attributable to an increase in brain activity [50,55]. We performed PTZ treatment from a lower to a higher dose [53]. At a lower dose of PTZ, we could not detect any differences in the WT compared to the mutant (Supplementary Figure S2). However, both the mutant and WT fish reacted to both 15 mM PTZ and 600 μM glutamate treatments, reflected by an increase in the distance traveled during all phases of the experiment (Figure 8). Interestingly, the mutant larvae reacted to 600 μM glutamate with a significantly milder induction of locomotor behavior during the baseline phase ( Figure 8B). No difference in the distance traveled was observed between the two genotypes that were exposed to 15 mM PTZ ( Figure 8A). Both WT and stim2a −/− larvae significantly increased their locomotor activity, which was the result of the occurrence of seizure-like episodes. Consistent with the results that are presented in Figure 6, we observed the most pronounced difference between the WT and stim2a −/− fish during the low activity phase under the notreatment condition ( Figure 8A,B). Both mutant and WT larvae exhibited hyperactivity and no decrease in mobility with 15 mM PTZ treatment in the light ( Figure 8A). Finally, we did not observe any difference in the behavior between stim2a −/− and WT larvae during the high activity phase with either treatment (Figure 8A,B).

Treatment with Pentylenetetrazol (PTZ) and Glutamate Resulted in Minimal Differences between stim2a −/− Mutants and WT in the Visual-Motor Response Test
We further explored the alterations of neuronal activity, detected by in vivo calcium imaging. We administered two drugs that stimulate neuronal activity, PTZ and glutamate, and performed the visual-motor response test (see Figure 3C). Pentylenetetrazol induces seizures, and zebrafish were shown to exhibit behavioral reactions in response to this treatment beginning at 3 dpf [53,54]. Glutamate increases neuronal activity and locomotor activity [50,55]. Importantly, glutamate was shown to penetrate zebrafish larvae and reach brain neurons. Thus, the main effect of this drug on animal behavior is primarily attributable to an increase in brain activity [50,55]. We performed PTZ treatment from a lower to a higher dose [53]. At a lower dose of PTZ, we could not detect any differences in the WT compared to the mutant (Supplementary Figure S2). However, both the mutant and WT fish reacted to both 15 mM PTZ and 600 µM glutamate treatments, reflected by an increase in the distance traveled during all phases of the experiment (Figure 8). Interestingly, the mutant larvae reacted to 600 µM glutamate with a significantly milder induction of locomotor behavior during the baseline phase ( Figure 8B). No difference in the distance traveled was observed between the two genotypes that were exposed to 15 mM PTZ ( Figure 8A). Both WT and stim2a −/− larvae significantly increased their locomotor activity, which was the result of the occurrence of seizure-like episodes. Consistent with the results that are presented in Figure 6, we observed the most pronounced difference between the WT and stim2a −/− fish during the low activity phase under the no-treatment condition ( Figure 8A,B). Both mutant and WT larvae exhibited hyperactivity and no decrease in mobility with 15 mM PTZ treatment in the light ( Figure 8A). Finally, we did not observe any difference in the behavior between stim2a −/− and WT larvae during the high activity phase with either treatment (Figure 8A,B).

RNA Sequencing Revealed Differential Gene Expression in stim2a −/− Zebrafish Larvae
To gain insights into molecular changes that were caused by stim2a deletion, we analyzed differential gene expression in the stim2a mutant and WT fish using RNA sequencing ( Figure 9A, Supplementary Table S1). A total of 392 genes showed at least two-fold change (upregulation: 336 genes, ~86%; downregulation: 56, ~14%) in expression ( Figure 9B). Using qPCR, we checked the level of expression of SOCE components, but no differential gene expression was found between genotypes, except stim2a downregulation ( Figure 9C). We then performed the GO annotation analyses of 392 genes using the PANTHER classification system [56] and zebrafish genome as a reference gene list ( Figure 10A,B, Supplementary Figure S4). The majority of the identified transcripts roughly overlapped with the plausible annotated function: binding, transporter activity, and catalytic activity. Approximately 8% of the genes are related to neurodegenerative disease pathways associated with Alzheimer's disease, Huntington's disease, and Parkinson's disease. Among the most enriched pathways were signaling via serotonin, glutamate, acetylcholine, and histamine neurotransmitters and signaling via hormones that are released from hypothalamic neurons (oxytocin, thyrotropin-releasing hormone, and gonadotropin-releasing hormone). Among the differentially expressed genes, egr4 and fosl1a (both upregulated) and fosab and npas4l (both downregulated) are isoforms of immediate-early genes that are known to be regulated by calcium signaling [13,24]. We also detected a high level of overexpression of smc1a, which encodes a cohesin complex protein that is involved in regulating nervous system development [57]. Some other wellknown genes, such as kcnc1a, slc20a2, slc25a17, slc25a25b, and slc31a1, which encode proteins that are

RNA Sequencing Revealed Differential Gene Expression in stim2a −/− Zebrafish Larvae
To gain insights into molecular changes that were caused by stim2a deletion, we analyzed differential gene expression in the stim2a mutant and WT fish using RNA sequencing ( Figure 9A, Supplementary Table S1). A total of 392 genes showed at least two-fold change (upregulation: 336 genes,~86%; downregulation: 56,~14%) in expression ( Figure 9B). Using qPCR, we checked the level of expression of SOCE components, but no differential gene expression was found between genotypes, except stim2a downregulation ( Figure 9C). We then performed the GO annotation analyses of 392 genes using the PANTHER classification system [56] and zebrafish genome as a reference gene list ( Figure 10A,B, Supplementary Figure S4). The majority of the identified transcripts roughly overlapped with the plausible annotated function: binding, transporter activity, and catalytic activity. Approximately 8% of the genes are related to neurodegenerative disease pathways associated with Alzheimer's disease, Huntington's disease, and Parkinson's disease. Among the most enriched pathways were signaling via serotonin, glutamate, acetylcholine, and histamine neurotransmitters and signaling via hormones that are released from hypothalamic neurons (oxytocin, thyrotropin-releasing hormone, and gonadotropin-releasing hormone). Among the differentially expressed genes, egr4 and fosl1a (both upregulated) and fosab and npas4l (both downregulated) are isoforms of immediate-early genes that are known to be regulated by calcium signaling [13,24]. We also detected a high level of overexpression of smc1a, which encodes a cohesin complex protein that is involved in regulating nervous system development [57]. Some other well-known genes, such as kcnc1a, slc20a2, slc25a17, slc25a25b, and slc31a1, which encode proteins that are involved in transporter activity, showed significant changes in the expression of stim2a −/− larvae ( Figure 10B). All of these data show that at the level of transcription, stim2a deletion leads to an imbalance of processes that are important for brain development and function.
involved in transporter activity, showed significant changes in the expression of stim2a −/− larvae ( Figure 10B). All of these data show that at the level of transcription, stim2a deletion leads to an imbalance of processes that are important for brain development and function.
We examined whether the differentially expressed genes belonged to the zebrafish CaTK that was published previously by our group [10]. Surprisingly, the CaTK and our current dataset shared only nine genes (anxa3a, grinab, hp, hpca, mast2, pkn3, pvalb7, slc25a25b, and stim2a). However, in our experimental dataset, we identified the cdh13, scin, dgkaa, mmp13a, LOC101882496, LOC103908715, and vwa2 genes, which encode proteins that might function in calcium signaling [46], but they were not identified in our earlier work on the zebrafish CaTK [10]. All of the differentially expressed genes in the CaTK are important for neuronal differentiation and function and could be involved in the behavioral phenotype of stim2a −/− fish.

Discussion
An important role has been established for STIM proteins in neuronal Ca 2+ homeostasis and signaling using in vitro mammalian primary neuronal cell cultures (reviewed in [1,9]). However, these molecular pathways are not precisely linked to in vivo data because Stim2 knockout in mice leads to early lethality before their nervous system is well developed (4-8 weeks of postnatal development) [31,32]. Thus, complex behavior cannot be analyzed in Stim2 knockout mice [31,32]. In zebrafish, the nervous system develops at an early stage and allows larvae to exhibit complex behavior [42,50,59]. Additionally, because of larval transparency, in vivo Ca 2+ measurements can be performed in neurons [45,60]. In the present study, we took advantage of these unique features of zebrafish to investigate behavior together with molecular analysis. The zebrafish genome has many duplicated genes, including stim2a and stim2b [10,45]. Thus, in contrast to early lethality in Stim2 knockout mouse lines, the knockout of one zebrafish isoform yields a viable animal, which we demonstrated for stim2a knockout fish in the present study and stim2b knockout fish in our recent study [50].
We showed that stim2a knockout caused distinct behavioral changes in zebrafish larvae. We observed hyperactivity in stim2a −/− zebrafish larvae, which was characterized by a significant increase in mobility. In the open field test, the acceleration and velocity of stim2a −/− larvae was higher; they also spend more time moving. We also observed significantly higher thigmotaxis (i.e., a preference for remaining close to the edge of the well) compared with WT zebrafish. Moreover, in stim2a −/− mutants, we detected the disruption of phototaxis compared with WT zebrafish. The stim2a −/− larvae reacted to changes in light in the visual-motor response test, thus excluding the possibility of visual deficits in the mutant animals. Lower light preference has been linked to a decrease in the anxiety in zebrafish larvae [50,59,61]. Such unpredictability between phototaxis and thigmotaxis indicates the involvement of different brain regions that control these two processes [51,62]. In addition, it is possible that the brain connectivity network that controls behavior may exhibit differential susceptibility to the loss of stim2 isoforms. We observed higher activity in stim2a −/− larvae during the low activity phase in the visual-motor response test, indicating that stim2a −/− mutants may be hyperactive [50,63].
An increase in locomotor activity is often associated with alterations of neurotransmission in the brain, including alterations of glutamate [64] and γ-aminobutyric acid (GABA) [53,65]. Thus, we treated the WT and stim2a −/− larvae with glutamate and PTZ. The stim2a −/− zebrafish reacted similarly to both treatments with an increase in activity during all phases of the visual-motor response test compared with the no-treatment condition. Only during the baseline phase, stim2a −/− mutants exhibited lower locomotor activity with glutamate treatment compared with WT fish. To link behavior with cellular-signaling events, we next examined Ca 2+ activity in neurons in vivo. We observed an increase in oscillation frequency but a decrease in the average amplitude of these oscillations in neurons in the optic tectum in stim2a −/− zebrafish compared with WT fish. It was shown We examined whether the differentially expressed genes belonged to the zebrafish CaTK that was published previously by our group [10]. Surprisingly, the CaTK and our current dataset shared only nine genes (anxa3a, grinab, hp, hpca, mast2, pkn3, pvalb7, slc25a25b, and stim2a). However, in our experimental dataset, we identified the cdh13, scin, dgkaa, mmp13a, LOC101882496, LOC103908715, and vwa2 genes, which encode proteins that might function in calcium signaling [46], but they were not identified in our earlier work on the zebrafish CaTK [10]. All of the differentially expressed genes in the CaTK are important for neuronal differentiation and function and could be involved in the behavioral phenotype of stim2a −/− fish.

Discussion
An important role has been established for STIM proteins in neuronal Ca 2+ homeostasis and signaling using in vitro mammalian primary neuronal cell cultures (reviewed in [1,9]). However, these molecular pathways are not precisely linked to in vivo data because Stim2 knockout in mice leads to early lethality before their nervous system is well developed (4-8 weeks of postnatal development) [31,32]. Thus, complex behavior cannot be analyzed in Stim2 knockout mice [31,32]. In zebrafish, the nervous system develops at an early stage and allows larvae to exhibit complex behavior [42,50,59]. Additionally, because of larval transparency, in vivo Ca 2+ measurements can be performed in neurons [45,60]. In the present study, we took advantage of these unique features of zebrafish to investigate behavior together with molecular analysis. The zebrafish genome has many duplicated genes, including stim2a and stim2b [10,45]. Thus, in contrast to early lethality in Stim2 knockout mouse lines, the knockout of one zebrafish isoform yields a viable animal, which we demonstrated for stim2a knockout fish in the present study and stim2b knockout fish in our recent study [50].
We showed that stim2a knockout caused distinct behavioral changes in zebrafish larvae. We observed hyperactivity in stim2a −/− zebrafish larvae, which was characterized by a significant increase in mobility. In the open field test, the acceleration and velocity of stim2a −/− larvae was higher; they also spend more time moving. We also observed significantly higher thigmotaxis (i.e., a preference for remaining close to the edge of the well) compared with WT zebrafish. Moreover, in stim2a −/− mutants, we detected the disruption of phototaxis compared with WT zebrafish. The stim2a −/− larvae reacted to changes in light in the visual-motor response test, thus excluding the possibility of visual deficits in the mutant animals. Lower light preference has been linked to a decrease in the anxiety in zebrafish larvae [50,59,61]. Such unpredictability between phototaxis and thigmotaxis indicates the involvement of different brain regions that control these two processes [51,62]. In addition, it is possible that the brain connectivity network that controls behavior may exhibit differential susceptibility to the loss of stim2 isoforms. We observed higher activity in stim2a −/− larvae during the low activity phase in the visual-motor response test, indicating that stim2a −/− mutants may be hyperactive [50,63].
An increase in locomotor activity is often associated with alterations of neurotransmission in the brain, including alterations of glutamate [64] and γ-aminobutyric acid (GABA) [53,65]. Thus, we treated the WT and stim2a −/− larvae with glutamate and PTZ. The stim2a −/− zebrafish reacted similarly to both treatments with an increase in activity during all phases of the visual-motor response test compared with the no-treatment condition. Only during the baseline phase, stim2a −/− mutants exhibited lower locomotor activity with glutamate treatment compared with WT fish. To link behavior with cellular-signaling events, we next examined Ca 2+ activity in neurons in vivo. We observed an increase in oscillation frequency but a decrease in the average amplitude of these oscillations in neurons in the optic tectum in stim2a −/− zebrafish compared with WT fish. It was shown earlier that STIM2 deletion and the subsequent SOCE disruption caused the decrease in oscillations of non-neuronal cells [66,67]. However, our in vivo data showed the increased Ca 2+ oscillation frequency in brain neurons stim2a −/− mutants. The Ca 2+ oscillation frequency might get compensated due to Stims interaction with other calcium channels, as shown previously, and by others [68,69]. Moreover, Stims 2a and Stim2b have different dissociation constants with Orai, which could affect the Ca 2+ oscillation frequency [70]. The qPCR analysis of the SOCE component genes showed the significant downregulation of stim2a; however, the expression of other SOCE component genes, including stim2b, were not affected, indicating that the compensation effect did not occur by the upregulation of stim2b. Moreover, the STIM2 interaction with voltage-gated calcium channels (VGCC), α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPA), N-methyl-D-aspartate receptors (NMDA), and plasma membrane Ca 2+ ATPases (PMCA) might affect the Ca 2+ oscillation frequency [12,68,71]. In differential genes expression analysis, we focused on CaTK genes (genes that are involved in Ca 2+ homeostasis and signaling) and transcription regulators. We found many genes that were affected in stim2a −/− zebrafish, including anxa3a, grinab, hp, hpca, mast2, pkn3, pvalb7, and slc25a25b (a member of CaTK genes and showed a significant change in expression). Among transcription factors, fosl1a, fosab, npas4l, and egr4 are isoforms of immediate-early genes that are regulated by Ca 2+ signaling [13,24]. The exact function of numerous c-fos orthologues (e.g., fosaa, fosab, fosb, fosl1a, fosl1b, fosl2, and fosl2l) has not been defined in the literature. However, active neurons are known to have higher c-fos gene expression as a result of a perturbation of calcium signaling [24,64]. Moreover, based on the published expression patterns of c-fos orthologues, we can conclude that they are expressed in the brain [46]. Thus, these genes have differential functions and might be differentially controlled by Ca 2+ signaling.
The phenotype of the stim2a mutant line that we investigated herein is not identical to the phenotypes of another Stim2 mutant (stim2b −/− ) that we reported previously [50]. Both mutants exhibited a hyperactive phenotype, a decrease in phototaxis, and an increase in thigmotaxis. Both lines also exhibited a milder decrease in activity upon turning the light on (Supplementary Table S1) [50]. However, responses to PTZ and glutamate treatment were different between the stim2a and stim2b mutants. For example, the stim2a −/− line behaved similarly to WT animals with PTZ treatment, whereas the stim2b −/− line exhibited greater susceptibility to this seizure-inducing drug. With glutamate treatment, the stim2b −/− line exhibited higher activity during the low activity phase, unlike the stim2a −/− line, which was less active during the baseline phase but more active during the following low activity phase. Among the notable genes in our RNA sequencing analysis were genes that are potentially involved in distinct features of the stim2a −/− line, including genes in the CaTK (grinab, hp, hpca, mast2, pkn3, pvalb7, and slc25a25b) [10] and some transcription regulators (egr4, fosl1a, fosab, npas4l). Notably, the anxa3a and smc1a genes may be responsible for some similarities in the phenotypes of the stim2a −/− and stim2b −/− lines. The anxa3a gene encodes the scaffolding protein annexin 3a, which was shown to bind plasma membrane phospholipids in a calcium-dependent manner and regulate calcium oscillations in the human promyeloblast cell line HL-60 [72]. Interestingly, these oscillations were blocked by 2-Aminoethoxydiphenyl borate (2-APB), a selective inhibitor of SOCE, suggesting the involvement of STIM molecules [72]. The smc1a gene is a homologue of human SMC1a that was impaired in a patient with Cornelia de Lange, a cohesinopathy that is characterized by the early manifestation of intractable epilepsy [57]. Our data suggest diverse roles of Stim2 isoforms in the zebrafish brain that might be involved in the maintenance of neuronal SOCE and the regulation of neuronal activity and genetic programs.
In summary, the present study showed that Stim2a deficiency in neurons led to changes in the levels of numerous mRNAs that encode proteins that are involved in calcium homeostasis. Thus, Stim2a can be considered an indirect regulator of their expression. Together with previous data on stim2b knockout zebrafish that were reported recently by our laboratory [50], we were able to conclude that both isoforms of stim2 affect neuronal activity and zebrafish behavior.

Materials
The chemicals and reagents were obtained from the following suppliers: glutamate

Animal Husbandry
All of the experimental animals were maintained in the Zebrafish Core Facility according to standard protocols [73].

Generation and Genotyping of stim2a −/− Mutant Zebrafish Line
Using the clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9 system, the stim2a −/− zebrafish line was created by introducing a 5 bp deletion before the start of the SOAR domain, which promoted an early stop codon. The fish were genotyped using the RLFP protocol that was developed especially for the stim2a −/− line. Briefly, at~30 dpf, tail fin tissue was obtained by biopsy and digested in Tris-EDTA (TE) buffer that was supplemented with proteinase K (0.5 mg/mL final concentration) overnight at 50 • C. The isolated DNA was used for polymerase chain reaction (PCR) using forward (5 -AACTCAGCCGTCTGTGGTATGCG-3 ) and reverse (5 -TGACGTTGTAGTACTGAACCTCCACCTC-3 ) primers for the stim2a gene. The fragments were digested with MBO II restriction enzyme and visualized by electrophoresis in 2% agarose (30 min, 80 mV) with the FastRuler Low Range DNA Ladder. The heavy fragment was designated as belonging to animals with both alleles that carried the stim2a −/− mutation. The heavy fragment that was accompanied by two light fragments was designated as belonging to animals with only one allele that carried the mutation (i.e., heterozygous). The presence of two light fragments alone was designated as belonging to animals with both alleles that did not carry the mutation (i.e., WT). F2 fish were out-crossed with the AB zebrafish line, and their offspring were in-crossed to obtain homozygous mutants. All of the animals that were used in the experiments were offspring from these fish (or fish from subsequent generations). For the Ca 2+ imaging experiments, stim2a −/− fish were out-crossed with Tg(HuC:GCaMP5G) and selected for fluorescent homozygous mutants.

Behavioral Experiments
The behavioral experiments were performed according to previously published protocols [38,40,42,50]. The treatment solutions were prepared in E3 water. The final concentrations of the chemicals were the following: 600 µM glutamate (30 mM stock solution) and 15 mM PTZ (0.5 M stock solution). These solutions were mixed in the experimental wells at a 1:1 proportion in E3 water from 2× concentrated working solutions. On the day of the experiment, randomly selected 4 dpf larvae were habituated to the behavioral testing room for at least 30 min. Their locomotor activity was recorded using the ZebraBox high-throughput monitoring system (ViewPoint, Life Sciences, Lyon, France). The video recordings were further analyzed using EthoVision XT software (Noldus, Wageningen, The Netherlands). The data were exported to Microsoft Excel files for further analysis using R software (v.3.6.0, R Foundation for Statistical Computing, Vienna, Austria). Data from larvae that were not active during the entire recording period (total distance < 10 mm) were excluded from the analyses.

Open Field Test
This test was initially developed for rodents [74,75] and subsequently adapted to adult zebrafish and larvae [38,44,75]. The protocol was modified from [38]. Briefly, 2 min before recording in the test, 4 dpf larvae were transferred to a 12-well plate at a density of 1 embryo per well ( Figure 3A). The test duration was 15 min. The following parameters were analyzed: total distance traveled in either the border or central zones of the wells, the duration of movements in each area, the duration of not-movement in each area, the mean velocity of the fish, the duration of the movement with a high speed (>20mm/s), and the maximum acceleration. Heat maps showing the mean traces of larvae were generated using EthoVision XT software (Noldus). The 15 min recording period was divided during the data analysis into three phases, 5 min each. Thigmotaxis is presented as the percent time spent in the zones relative to the total time spent moving in the outer zone of the test apparatus [38].

Dark/Light Preference Test
This test was performed according to previous studies [51,61,[76][77][78]. A 10 cm Petri dish was used. Half of the Petri dish was covered with a photographic filter, and the side of the dish was covered with black vinyl tape to avoid light penetrance ( Figure 3B). The test was recorded for 15 min after 2 min of the adaptation period. The total distance traveled (in millimeters) was analyzed.

Visual-Motor Response Test
This test was described elsewhere [52,79]. Briefly, a 24-well plate was used as the arena ( Figure 3C). The fish were acclimatized to the plate before the experiment. The experiments began with an exchange of half the volume of each well into the test solution. The fish were treated with either 600 µM glutamate or PTZ. Untreated fish were exposed to E3 water only. The recording began with a 2 min delay and was divided into the following phases: baseline phase (0% light illumination), low activity phase (70% light illumination), and high activity phase (0% light illumination; Figure 3C). The total distance traveled (in millimeters) was analyzed.

In Vivo Calcium Imaging
The in vivo calcium imaging of neuronal activity in the brain was performed using a Zeiss Lightsheet Z.1 microscope (40×/1.0 objective). Tg(HuC:GCaMP5G) [63] stim2a −/− zebrafish and their stim2a +/+ siblings (WT) were used in the experiment. Zebrafish larvae were immobilized at 4 dpf using 0.6 µg/µl pancuronium bromide and mounted in 1.5% low melting agarose. Time-lapses were recorded for 5 min, with 15 ms exposure and a frequency of 1 frame/sec at single-cell resolution. A plane that contained the habenula, optic tectum, and cerebellum was selected. Automated cell segmentation was done to draw the region of interest (ROI) as described elsewhere [80] in a sub-region in the periventricular gray zone of the optic tectum (as shown in the Supplementary Figure S5) using MATLAB script (Mathworks). After that, the GCaMP5G fluorescence changes due to spontaneous Ca 2+ activity were extracted from each cell from the sub-region ROI (Supplementary Figure S3A-D: plots the raw data of Ca 2+ fluorescence traces from individual neurons after automated cell segmentation for Supplementary Videos S1-S4). Then, the peaks of oscillations of Ca 2+ levels were selected using a function that detects peaks based on a slope change. The average oscillation frequency (in Hz) and the average amplitude (AU) of Ca 2+ signals were analyzed. The Wilcoxon rank-sum test was used for comparisons of average oscillation frequencies between WT and stim2a −/− animals. The data are presented as medians with Q1 and Q3 quartiles using boxplots, and dots represent data outliers. The following numbers of cells were analyzed: 168 cells from three WT animals and 392 cells from seven stim2a −/− animals.

Quantitative PCR Gene Expression Analysis
Thirty zebrafish larvae (5 dpf) per sample were anesthetized with MS-222, collected on ice, washed in phosphate-buffered saline (PBS), and snap-frozen in Qiazol (Qiagen, Hilden, Germany) until use for RNA isolation. RNA was isolated according to a previously published protocol [81]. The RNA template (500 ng) was used for cDNA synthesis using the iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA, USA). Quantitative PCR (qPCR) was performed in duplicate using 25 ng cDNA per reaction with Precision Melt Supermix (Bio-Rad, Hercules, CA, USA). The analysis of SOCE component gene expression was performed as described elsewhere [10] using previously characterized primers for stim1a, stim1b, stim2b, orai1a, orai1b, orai2. Fold changes were calculated using the ∆∆Cq method [10]. Eukaryotic translation elongation factor 1 α1, like 1 (eef1a1l1) gene expression, was used as the reference gene. Gene expression levels were analyzed using the CFX Connect RT-PCR Detection System (Bio-Rad, Hercules, CA, USA) and visualized using R software.

RNA Sequencing Analysis
The RNA isolation procedure for the RNA sequencing experiment was similar to RNA isolation for the qPCR gene expression analysis [10,81]. Briefly, the total RNA was digested using DNase I and purified using the RNA Clean and Concentrator Kit (ZYMO Research, Irvine, CA, USA) according to the manufacturer's instructions. Sequencing was performed using Illumina methodology. The preparation of cDNA libraries and sequencing using next-generation sequencing (NGS NextSeq 500, San Diego, CA, USA) were performed in cooperation with the Core Facility at the International Institute of Molecular and Cell Biology in Warsaw. The following sequencing parameters were applied: paired-end sequencing run type, 1 × 76 bp read length. This allowed approximately 120-150 million reads per sample with a 76 bp length. The reads were extracted into FASTQ format and used for the subsequent bioinformatic analysis. FASTQ files are deposited in the Sequence Read Archive (accession no. PRJNA635784). The reads were then aligned to the zebrafish Refseq genome assembly (GRCz11_genomic.fa). Genes were annotated using the GRCz11_genomic.gff Ensembl annotation file. The genes were evaluated according to the false discovery rate (FDR) and fold change between the WT and stim2a −/− samples. Genes with > 2-fold changes between WT and mutants and a significant FDR were taken for the analysis. The Danio rerio genome was used as a reference gene list, which allowed the identification of cellular components, molecular functions, and the related pathways from the gene ontology (GO) terms [10]. The most promising gene candidates were plotted as the mean ± standard error of the mean (SEM).

Statistical Analysis
The analysis output from EthoVision was exported to Microsoft Excel files and analyzed using R software (v.3.6.0). Results from larvae that were not active during the entire recording period (total distance < 10 mm) were excluded from the analysis. The data distribution was checked using the Shapiro-Wilk normality test. Data from the behavioral experiments were analyzed using paired or unpaired Wilcoxon rank-sum tests for comparisons between the groups. Adjusted values of p < 0.05 were considered statistically significant. The data are expressed as medians with Q1 and Q3 quartiles using boxplots, and dots represent data outliers unless stated otherwise. The number of fish that were used in each experiment is indicated in the figure legends. The RNA sequencing data were analyzed based on logarithmic (log 2 ) fold changes and are expressed as mean ± SEM if the log 2 fold change was >2 and <−2. The FDR was used to estimate the statistical significance of gene expression using RNA sequencing.