All the zebrafish larvae and adults used in this study were reared at the University of Miami Zebrafish facility as per IACUC protocol #18–129. Water temperatures were maintained at 28°C and both adult and larval zebrafish were exposed to a circadian cycle of 14 h light/10 h dark. Water housing adult zebrafish was continuously monitored for pH and conductivity to maintain conditions within an optimal range (pH 7–8.1; conductivity 350–800 mOsm). Upon collection, embryos were rinsed briefly in reverse osmosis water and reared in 10 cm petri dishes containing ‘system water’ (taken from water housing adults). Dishes were cleaned daily to remove unfertilized eggs and prevent fungal growth that could limit oxygen and stunt early embryonic growth. Larvae used for behavioral assays were raised ~50 larvae per petri dish to minimize competition and developmental delays.

Syngap1ab mutants were generated using CRISPR/Cas9 genome editing technology (Varshney et al., 2016). One cell stage WT embryos were injected with small guide RNA (sgRNA; Integrated DNA Technologies-IDT Coralville IA) designed using CHOPCHOP online software (Montague et al., 2014) to target exon 4 of syngap1a gene (400 pg) and exon 5 of syngap1b gene (400 pg), along with Cas9 protein (PNA Bio Thousand Oaks CA; 100 pg). Embryos were either injected with syngap1a or syngap1b sgRNA using a foot-pedal-controlled Milli-Pulse Pressure Injector (MPPI-3 from Applied Scientific Instrumentation ASI Eugene OR). Resulting mosaic F0 larvae were reared to adulthood and crossed to wild-type animals to generate an F1 generation for Sanger sequencing to identify syngap1a and syngap1b mutant alleles with indels resulting from CRISPR editing (see below). Upon identification of the mutant syngap1a and syngap1b alleles, adults were in-crossed to obtain syngap1ab double mutants. Adult syngap1ab mutant fish, used to spawn larvae for experiments, span F2-F5 generations. To obtain syngap1ab+/− larvae, adult male syngap1ab−/− were outcrossed to WT (AB/TL, https://zfin.org/action/genotype/view/ZDB-GENO-031202-1) females. To obtain syngap1ab−/− larvae, syngap1ab−/− adults were in-crossed. For simplicity, zebrafish that are heterozygous in both syngap1a and syngap1b genes are denoted as syngap1ab+/−, and those homozygous in both syngap1a and syngap1b genes are denoted as syngap1ab−/−. All the wild-type (WT) larvae used for this study were AB/TL unless otherwise stated and are denoted as syngap1ab+/+.

The molecular identity of CRISPR alleles was determined by obtaining a small caudal fin sample from each anesthetized (200 mg/L of Tricaine (MS-222)) F1 adult fish. Genomic DNA was extracted by digesting each fin sample in 50 μL of 50 mM NaOH at 95°C for 1 h, the HotSHOT method (Samarut et al., 2016). To determine larval genotypes, each larva was anesthetized by either placing them on ice for 30 min or by using 200 mg/L Tricaine (MS-222). Upon complete anesthetization, larval genomic DNA (gDNA) samples were isolated using the HotSHOT method, as detailed above but using only 20 μL of 50 mM NaOH.

Gene-specific primers (Table 1) for both genes were designed using Primer3 software. Primer3 input sequences for each gene was selected based on their Cas9 target regions. For PCR, each reaction mixture contained 5 μL of 10x GOTaq Polymerase (Promega Madison WI), 0.5 μL from each 10 μM forward primer and reverse primer, 3 μL of nuclease-free water, and 1 μL of gDNA (from either larvae or adult-fin-clip digestions). Resulting PCR products were sent out for sequencing (Eurofins Genomics, LLC Louisville KY) and the results were read and analyzed using the ApE—A plasmid Editor v2.0.61 (Davis and Jorgensen, 2022) and SnapGene viewer software to determine the syngap1a and syngap1b mutant alleles (Supplementary Figure S1).

To better characterize zebrafish syngap1ab splice variants, mRNA sequences (both published and predicted), were obtained from the NCBI protein database. To test for evidence of isoform expression, these sequences were searched against Expressed Sequence Tags (EST) and Transcriptome Shotgun Assembly (TSA) databases. Expressed isoforms were then BLASTed against the UCSC zebrafish genome browser to identify unique and common exons (Figure 1).

To check for nonsense-mediated mRNA decay in syngap1ab mutants, qPCR was used to quantify relative syngap1ab expression levels in syngap1ab mutant larvae compared to WT larvae, both groups at 7 days post-fertilization (dpf). To extract RNA, larvae were anesthetized by placing them on ice for 30 min before using TRIzol (Life Technologies, Carlsbad, CA) following manufacturer’s protocol. For each genotype (WT, syngap1ab+/−, and syngap1ab−/−), we conducted at least eight experimental replicates with 20 larvae pooled together per replicate. 1 μg of RNA was used as input for RT-PCR. To enhance syngap1 cDNA in each sample, syngap1a and syngap1b gene-specific primers were used with Eukaryotic translation elongation factor 1 like 1 eef1a1l1 (ZFIN) as the internal control. cDNA was made using SuperScript III (Invitrogen™/ ThermoFisher Scientific) and incubating at 50°C for 1 h followed by 15 min at 70°C. qPCR was carried out using GoTaq qPCR Probe Kit (Promega) in a QuantStudio3 RT-PCR system (Applied Biosystems™, Waltham MA) following manufacturer’s protocol. Cycling conditions were as follows: Activation step of 95°C for 10 min, followed by PCR with 40 cycles of 95°C for 15 s and 60°C for 1 min, followed by a melt curve of 95°C for 15 s and 60°C for 1 min. Relative levels of gene expression were calculated using the ΔΔCt method. For this method cycle threshold Ct values for syngap1a and syngap1b genes were first normalized to Ct values of the internal control eef1a1l1 by calculating ΔCt: Ct_syngap1a-Ct_eef1a1l1 and Ct_syngap1b-Ct_eef1a1l1. Fold-changes in syngap1 gene expression were then compared in WT, syngap1ab+/−, and syngap1ab−/− larvae by calculating 2^-ΔΔCt with ΔΔCt:ΔCt_{syngap1ab mutant}-ΔCt_WT. Fold changes in syngap1ab mutants were calculated by dividing WT values. Group comparisons were made using 2-way ANOVA (gene and genotype) followed by Tukey’s multiple comparison test.

Brain regions (mouse) or whole brains (zebrafish) were excised from C57BL6 mice or adult zebrafish. Tissues were lysed in 10 volumes (for zebrafish, each brain was considered ~15 mg) of lysis buffer (50 mM Tris pH 8.0, 100 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% TritonX-100, 0.2% SDS, 0.5% Sodium deoxycholate, with complete Protease inhibitor EDTA-free mix (Roche/MilliporeSIGMA Burlington MA)) using a Dounce homogenizer to obtain homogenized tissue samples. Each sample was then diluted by 1:10 using lysis buffer and ~ 20 μL from each sample was loaded into each gel lane. Based on these experimental settings, each sample, i.e., for both zebrafish and mouse tissue samples, contained about 29 μg of proteins per 20 μL. Samples were first probed with SYNGAP1 antibodies (Abcam ab3344 or NOVUS nbp2-27541) and were followed by probing with secondary antibodies (anti-rabbit IgG IRDye680; LICOR 926-68071 or anti-goat IgG IRDye680; LICOR 926-68074). Resulting signals were measured and imaged using the fluorescence-based Odyssey CLx Imaging System (LICORbio Lincoln NE).

To test for the potential effects of syngap1a and syngap1b mutant alleles on larval survival, three batches of syngap1ab+/− in-crosses were genotyped at six dpf. Data were analyzed using Prism GraphPad software (v9.1) to determine whether observed allele representation differed significantly from predicted using a Chi-square test.

To test whether the larvae show habituation in response to acoustic stimuli, a short-term habituation assay was used as described in Wolman et al. (2015). After 30 min of light adaptation, 6 dpf WT and syngap1ab+/− larvae were presented with 5 phases of acoustic stimulation using a DanioVision™ observation chamber (Noldus Leesburg VA). Phase 1 consisted of 10 tap stimuli (intensity level 3) delivered at a 20 s inter stimulus interval (ISI). Phase 2 consisted of 10 tap stimuli (intensity level 5) delivered at also at a 20 s ISI. Phase 3, the habituation test, consisted of 30 tap stimuli (intensity level 5) delivered at 1 s ISI. Phase 4 was a 3 min rest period. Lastly phase 5 consisted of 10 tap stimuli (intensity level 5), delivered at a 20 s ISI. Total distance moved by each larva per second was analyzed to measure short-term habituation. Data were analyzed using Prism GraphPad software (v9.1) to compare WT and syngap1ab+/− groups using a 2-way ANOVA (Phase and Genotype) followed by a Tukey’s multiple comparison test.

Syngap1ab mutants and WT 6 dpf larvae larvae were placed in a 96-well plate containing system water. Prior to starting the experiment, larvae were dark adapted for 1 h and at 28°C in the Noldus DanioVision™ behavioral observation chamber. Ethovision® XT 11 software (Noldus) was used to program the delivery of stimuli and to analyze the results. Images were captured at 40 Hz. For the visual-motor response (VMR) assay, larvae were exposed to 5 min of 12% (high-light settings) lights-on stimulus followed by 5 min of lights-off stimulus. Larval movements were recorded for four consecutive light on/off cycles and their activity/movements were recorded for a total duration of 40 min. Larvae were then genotyped using the genotyping assays previously described and analyzed for the total distance moved/time. The resulting data were further analyzed using GraphPad Prism 9.1 software to compare WT, syngap1ab+/−, and syngap1ab−/− groups using Kruskal-Wallis ANOVA followed by Dunn’s multiple comparison tests.

To test whether the observed hyperactivity is due to increased initiation of movements or increased distance moved, the displacement and dwell-time data from raw, exported data produced by Ethovision® XT 11 software were analyzed using MATLAB scripts¹ to examine behavioral probability distributions. Resulting data were plotted using Prism GraphPad (V9.1) and WT and syngap1ab groups compared using a two-sample Kolmogorov–Smirnov test.

While mammals including humans have a single SYNGAP1 gene that encodes a 1,343 amino acid, ~149 kDa protein (Komiyama et al., 2002), there are two syngap1 ohnologs present in the zebrafish genome. The zebrafish syngap1a gene is found on chromosome 16 and encodes a 1,290 amino acid, ~146 kDa protein whereas the syngap1b gene is found on chromosome 19 and encodes a 1,507 amino acid, ~160 kDa protein. Like rodents and humans, both zebrafish Syngap1a and Syngap1b have four, highly conserved protein–protein interacting domains: pleckstrin homology (PH), C2, RasGAP, and coiled-coiled (CC).

We characterized zebrafish Syngap1 isoforms based on NCBI databases and comparisons to mammalian isoforms. In mammalian Syngap1, there are four well-characterized, alternatively-spliced Syngap1 isoforms, α1, α2, β, and γ, that vary at their C-termini (McMahon et al., 2012). To assess whether zebrafish syngap1a and syngap1b genes might encode similar isoforms, we curated all published and predicted syngap1ab isoforms. Similar to the mammalian isoforms, for both zebrafish syngap1a and syngap1b genes, exons encoding the four protein interaction domains occur in all isoforms with many alternative exons at both N- and C-termini. Based on expressed sequence databases, we were able to identify five syngap1a and eleven syngap1b isoforms (Figure 1). Of the human C-term isoforms, α1 is the most studied and is localized to the post-synapse by a four amino acid (QTRV) PDZ-interacting domain. Interestingly, we were able to find a zebrafish syngap1b C-term isoform X5 with a putative mammalian-like PDZ-interacting domain that had ten of eleven of the terminal amino acids identical to those of the human SYNGAP1 α1 isoform (Figure 1B). We were unable to find mammalian-like isoforms (α2, β, and γ) due to the highly variable C-terminal ends in the zebrafish isoforms.

To better understand how pathogenic mutations in syngap1 result in altered behaviors, we used CRISPR/Cas9 to generate zebrafish loss-of-function mutations. In humans, pathogenic variants span the SYNGAP1 gene (Hamdan et al., 2011; Berryer et al., 2013; Vlaskamp et al., 2019; Gamache et al., 2020). To mutate a region of the gene that would affect the majority of isoforms, we targeted the earliest shared exons, exon 4 and exon 5 in syngap1a and syngap1b respectively, to generate loss-of-function alleles. CRISPR/Cas9 induces indels causing reading frame shifts and introducing premature stop codons. Upon sequence analyses of F1 adult crispants, we selected two mutant alleles for syngap1a (syngap1a + 7 nucleotide insertion and syngap1a − 22 nucleotide deletion) and one allele for syngap1b with a − 14 nucleotide deletion, all of which would be predicted to result in a severely truncated proteins that were less than 200aa (Figure 2B). To best recapitulate human SYNGAP1 variant haploinsufficiency, we used double-heterozygous larvae for syngap1a + 7 and syngap1b − 14, and to further assess complete loss-of-function, we used double-homozygous larvae (here onwards denoted as syngap1ab+/− and syngap1ab−/− respectively).

We tested for loss-of-function of syngap1 at the level of protein and mRNA. Western blot analysis carried out using adult zebrafish brain lysates showed reduced Syngap1 protein levels in syngap1ab−/− brain tissues (Figure 2C). qPCR analysis of RNA harvested from 7-day-old larvae showed reduced RNA transcripts in both syngap1ab+/− (adjusted p values for syngap1a = 0.001, and syngap1b = 0.0495) and syngap1ab−/− (adjusted p values for syngap1a = 0.0699 and syngap1b = 0.0312) mutant larvae compared to WT larvae supporting non-sense mediated decay (Figure 2D). Taken together, these results show that our CRISPR/cas9 generated Syngap1ab zebrafish mutants show reduced mRNA and protein expression, supporting the use of these alleles as haploinsufficient models.

To study potential interactions between Syngap1a and Syngap1b proteins, we analyzed larval survival from each of the nine genotypes resulting from syngap1ab+/− in-crosses. When syngap1b mutant alleles outnumbered syngap1a mutant alleles, larval survival was much lower than expected (Supplementary Figures S2A,B; p < 0.0001). By contrast, zebrafish in which there were more syngap1a than syngap1b mutant alleles were over-represented and zebrafish in which there were an equal number of syngap1a and syngap1b mutant alleles were as expected. These observations suggest that each ohnolog contributes distinct functional roles so that the syngap1a ohnolog, likely due to its inability to encode the Syngap1 α1 isoform, cannot make up for loss of syngap1b ohnolog.

Given that SYNGAP1 is known to be important for sensory processing, we wanted to test the dynamic range of syngap1ab+/− zebrafish responses to vibration stimuli of medium and high intensity as well as their ability to habituate to repeated, high-intensity, high-frequency stimuli. To assess these factors, we used an assay described in Wolman et al. (2015) in which 6 days-post-fertilization (dpf) larvae were exposed to different intensity vibrations with different inter-stimulus intervals (ISIs) during phases 1–5 of the assay (Figure 3A). We allowed larvae to acclimate to the Noldus chamber for 30 min before exposing them to vibrational stimuli. During phases 1 and 2, 10 taps of medium-intensity (level 3) and high-intensity (level 5) respectively were delivered with 20 s ISI. During phase 3, habituation was tested by delivering 30 high-intensity taps with 1 s ISI. This was followed by a 3 min rest period in phase 4. Finally, phase 5 was a repeat of phase 2 with 10 high-intensity taps delivered with 20 s ISI. Overall, syngap1ab+/− larvae (n = 130 from 3 independent crosses) responded very similarly to WT larvae (n = 110 from three independent crosses) (Figure 3B), with a normal degree of habituation to high frequency stimuli (Figure 3C).

Despite these similarities, there were subtle differences. syngap1ab+/− larvae showed consistently elevated responses to high-intensity stimuli during phases 2 and 5. To determine whether this was due to larger movements and/or an increased probability of response, we calculated median movement velocity per larva (Figure 3D) and their response probability (Figure 3E) during Phases 1, 2, and 5. For median movement velocity (calculated across taps in a given phase), a mixed effects model of phase and genotype indicated a significant effect of phase (p = 0.0018). The subsequent Tukey’s multiple comparison test showed that syngap1+/− larvae moved further in response to high-intensity vibrations in phases 2 and 5 than to medium-intensity vibrations in phase 1 (p = 0.0006 and 0.0007 respectively) while WT larvae moved similar distances during phases 1, 2, and 5. For probability of response, a mixed effects model indicated effects of both phase (p < 0.0001) and genotype (p = 0.0009). The subsequent Tukey’s multiple comparison test showed that syngap1+/− larvae had a higher response probability to stronger vibrational stimuli (p < 0.0001 Phase 1 vs. 2; p = 0.0002 Phase 1 vs. 5) and that this higher response probability to high-intensity stimuli was greater in syngap1+/− than WT larvae (p = 0.0049 Phase 2; p = 0.0014 Phase 5).

Therefore, in response to stronger stimuli, syngap1+/− are more likely to move and move faster, indicating a larger dynamic range of response in syngap1+/− larvae. This higher probability of response to the same stimulus is also consistent with higher levels of arousal in syngap1+/− larvae.

We next assessed visual-motor responses (VMR) (Burgess and Granato, 2007; Emran et al., 2008) in the syngap1ab models. For this assay, larval movements were recorded during four cycles of lights-on to lights-off transitions. Larval activity was measured as the total distance moved every 30 s. Larvae show robust increases in locomotor activity when presented with a sudden transition from light to darkness (Supplementary Figure S3; Figure 4A). Compared to WT larvae, both syngap1ab+/− and syngap1ab−/− larvae showed increased activity (p < 0.0001) during both lights-on and lights-off cycles (Figure 4B). During lights-off periods syngap1ab+/− showed the greatest movement but this trend was somewhat variable across different batches of larvae (Supplementary Figure S3B); during lights-on, hyperactivity was consistent across batches of larvae and was dependent on the number of mutant syngap1 alleles with syngap1ab−/− showing the greatest movement (Figure 4; Supplementary Figure S3).

For the preceding assays, WT, syngap1ab+/−, and syngap1ab−/− larvae resulted from independent crosses. To rule out the influence of distinct parental genetic backgrounds on the observed behavior, we also examined VMR in larvae resulting from an in-cross between syngap1ab+/− adults (Supplementary Figures S2C,D). Consistent with the previous results, during both lights-on (Supplementary Figure S2C) and lights-off cycles (Supplementary Figure S2D), syngap1ab−/− had the highest activity levels among all the resulting genotypes.

The hyperactivity we observed in the syngap1ab mutant models could be due to either increased movement frequency, increased distance traveled per movement, or both. To distinguish among these possibilities, we analyzed the larval movement at a higher temporal resolution. Data were sampled every 25 ms, a much higher temporal resolution than the distance per 30 s shown in the preceding figures. Zebrafish movement bouts last ~250 ms and so 40 Hz resolution is sufficient to capture the majority of bouts with multiple timepoints. We focused on two parameters: the time interval between two consecutive movement bouts, denoted as “dwell time,” and the distance traveled per bout, denoted as “displacement.” These high-resolution activity data were organized and analyzed using custom-written MATLAB scripts to assess the probability of different behaviors, described by displacement and dwell time, during light and dark conditions.

Given highly stochastic individual larval movements, to assess overall patterns of both high and low frequency events across the distribution, we captured a large number of data points from over 100 individuals per genotype and then divided the total bouts by the number of individuals to generate and “idealized larva” for each genotype (Figures 4B,C). When bouts were pooled across individuals of a genotype, there were > 10⁵ bouts per genotype and light condition (Lights-On: WT n = 119,135, syngap1ab+/− n = 196,594, and syngap1ab−/− n = 158,443; Lights-Off: WT n = 652,368, syngap1ab+/− n = 777,525, and syngap1ab−/− n = 507,720) coming from 173 WT, 167 syngap1ab+/−, and 119 syngap1ab−/− individual larvae. This analysis shows that the differences between genotypes are much more pronounced in the light/low arousal settings, than in the dark/higher arousal setting.

We next looked at probability distributions of dwell times (the time between movements) and distance traveled per movement (Figure 5). WT larvae in the dark had shorter dwell times and larger movements than WT in the light (Figures 5Ai,ii). These differences in WT light and dark behaviors are highlighted by plots below that show the relative probabilities in Dark versus Light for both dwell time and displacements (Figures 5Aiii,iv). Next, we compared syngap1ab+/− and WT in the light (Figure 5B) and in the dark (Figure 5C). In the dark, syngap1ab+/− and syngap1ab−/− had dwell time and displacement distributions that were very similar to WT larvae (Figures 5Ci–iv). By contrast, in the light, syngap1ab−/− and syngap1ab+/− mutants displayed both more frequent, and larger displacements compared to WT larvae (Figures 5Bi,ii). These differences between syngap1+/− and WT larvae in the light are highlighted by plots below that show the relative probabilities of syngap1+/− (purple) and syngap1−/− (pink) versus WT (Figures 5Biii,iv). These analyses showed that syngap1-WT (in light) resembles dark–light WT comparisons, indicating that syngap1 mutants behavior in the light resembles that of WT behavior in the dark. Taken together, behavioral experiments show context-dependent hyperactivity that is most subtle during aversive, acoustic stimuli, is intermediate in the dark, and is most pronounced in normally low arousal well-lit environments.