Introduction

RNA Binding Fox-1 Homolog 1 (RBFOX1, also referred to as A2BP1 or FOX1) encodes an RNA splicing factor that is specifically expressed in brain, heart and muscle in human adults (GTEX, https://gtexportal.org/home/gene/RBFOX1). This gene regulates the expression and splicing of large gene networks and plays an important role in neurodevelopment [1, 2]. Rare genetic variations, including point mutations and copy number variants, have been reported in RBFOX1 in patients with neurodevelopmental disorders such as autism spectrum disorder (ASD) [3,4,5,6], and RBFOX1 haploinsufficiency results in a syndrome characterised by impaired neurodevelopment [7, 8]. In addition, transcriptomic analysis of brains from autistic individuals revealed decreased levels of RBFOX1 and dysregulation of RBFOX1-dependent alternative splicing [6, 9]. RBFOX1 has not only been related to neurodevelopmental conditions, but increasing evidence points to both rare and common variants in this gene as contributors to several psychiatric and neurological disorders [5, 6, 10,11,12]. Interestingly, common variants in RBFOX1 were found significantly associated with the cross-trait phenotype of the most recent genome-wide association studies (GWAS) meta-analysis of psychiatric disorders [13] and RBFOX1 was identified as the second most pleiotropic locus in a previous cross-disorder GWAS meta-analysis, showing association of common variants with attention-deficit/hyperactivity disorder (ADHD), ASD, bipolar disorder (BIP), major depression (MD), obsessive-compulsive disorder, schizophrenia (SCZ) and Tourette’s syndrome (TS) [14]. Finally, Rbfox1-/- mutant mice present a heightened susceptibility to seizures and neuronal hyperexcitability [15], and Rbfox1 neuron-specific knockout mice show pronounced hyperactivity, stereotyped behaviour, impairments in fear acquisition and extinction, reduced social interest and lack of aggression [6], behaviours that are related to different psychiatric disorders. All these data suggest a major role for RBFOX1 in psychopathology, although the mechanisms underlying its pleiotropic effects are not well understood.

In the last years, zebrafish have become a powerful model to study psychiatric disorders thanks to their high genetic similarity to human and their well-defined behavioural phenotypes, which can be easily assessed in the laboratory and compared to human psychiatric phenotypes [16,17,18]. RBFOX1 has two orthologous genes in zebrafish, rbfox1 (a2bp1, NCBI gene ID: 449554) and rbfox1l (a2bp1l, NCBI gene ID: 407613). While the human gene RBFOX1 is expressed in both the brain and skeletal and cardiac muscle (GTEX, https://gtexportal.org/home/gene/RBFOX1), rbfox1 is mainly expressed in brain – but is also transcribed in heart –, and rbfox1l is exclusively expressed in skeletal and cardiac muscles at early developmental stages [19, 20] and shows a low and restricted expression in only some neuronal populations of the adult zebrafish brain [21]. In this study we focused on rbfox1, a gene that encodes a major protein isoform with 84% identity to the human protein [19], given its strong brain expression during development. To date, the expression of rbfox1 at later stages has not been investigated nor its role in zebrafish neurodevelopment and behaviour.

Genetic studies in humans have pointed to a pleiotropic contribution of RBFOX1 to several psychiatric conditions. Here, we have characterised the effect of altered rbfox1 function on zebrafish behaviour, and our data help describe mechanisms underlying its pleiotropic effects on the onset of neurodevelopmental and psychiatric disorders.

Material and methods

Zebrafish strains, care and maintenance

Adult zebrafish and larvae (Danio rerio) were maintained at 28.5 °C on a 14:10 light-dark cycle following standard protocols. All experimental procedures were approved by the Animal Welfare and Ethical Review board of the Generalitat de Catalunya. Behavioural experiments were performed using two different rbfox1 mutant strains with different genetic backgrounds. rbfox1sa15940, on the Tübingen Long-fin (TL) background, is a transgenic line obtained from the European Zebrafish Resource Center of the Karlsruhe Institute of Technology (KIT). This line contains an intronic point mutation at the -2 position of a 3’ acceptor splicing site of rbfox1 before the second/third exon of rbfox1 annotated isoforms (A > T, Chr3:28068329, GRCz11) (Supplementary Fig. 1). The second line, rbfox1del19, on the Tübingen (TU) background, was created using CRISPR/Cas9 genetic engineering and causes a frameshift deletion of 19 bp within exon 2 or 3 of rbfox1 annotated isoforms (Chr3:28068264-28068282, GRCz11) (Supplementary Fig. 2). Homozygous mutant fish (HOM, rbfox1sa15940/sa15940 and rbfox1del19/del19), heterozygous (HZ, rbfox1sa15940/+ and rbfox1del19/+) and wild-type (WT, TL rbfox1+/+ and TU rbfox1+/+) fish were used for all behavioural experiments. For both rbfox1sa15940 and rbfox1del19 lines, homozygous, heterozygous and wild-type fish were obtained from heterozygous crosses to ensure a common genetic background.

Gene expression analysis using Real-Time quantitative PCR (RT-qPCR)

Total RNA was extracted from the whole brain of 7 TL rbfox1+/+, 7 TL rbfox1sa15940/+ and 7 TL rbfox1sa15940 /sa15940 adult zebrafish and from 5 pools of 10 whole larvae of each genotype rbfox1+/+, rbfox1sa15940/+ and rbfox1sa15940 /sa15940 to perform RT-qPCR. Primers were designed to amplify cDNA from all the rbfox1, rbfox1l, rbfox2, rbfox3a and rbfox3b protein-coding isoforms described in the GRCz11 Ensembl database except for the rbfox1-203 isoform (http://www.ensembl.org/Danio_rerio/, Supplementary Table 1). In the case of rbfox1 primers, the forward primer binds to the exon situated after the intronic point mutation in the rbfox1sa15940 line and the reverse primer binds to the subsequent exon. Results were normalised to the expression levels of reference housekeeping genes: the ribosomal protein L13a (rpl13a) and the eukaryotic translation elongation factor 1 alpha 1a (eef1a1a) housekeeping genes in the case of adult brains, and the ubiquitin-conjugating enzyme E2A (ube2a) in the case of larval samples, as the expression level of this gene is more stable through developmental stages (Supplementary Table 1). The relative expression of the genes and the fold change were calculated using the 2-ΔΔCT comparative method [22, 23].

In situ hybridisation (ISH)

A specific mRNA probe targeting rbfox1 (NCBI Reference Sequence: NM_001005596) was prepared and ISH experiments were performed in larvae (28 h post fertilisation (hpf), 2-, 3-, 4-, and 5-days post fertilisation (dpf)) and dissected adult brains of WT fish from TL and TU lines. Further details are described in the Supplementary Methods.

Generation of a rbfox1 zebrafish loss-of-function line using CRISPR/Cas9

We used the CRISPR/Cas9 technology to generate stable rbfox1 loss-of-function mutants (Supplementary Fig. 2). Briefly, we designed 20 bp sequences (crRNA) targeting rbfox1 next to a PAM sequence (Supplementary Table 2). scRNA, tracrRNA and Cas9 were purchased, and 1 nL of injection solution was injected into the cell of single-cell stage zebrafish embryos. After 24 hpf, the injection efficiency and crRNA efficacy were assessed and injected embryos (called F0 thereafter) with high injection efficiency were raised to adulthood. F0 were then crossed with WT zebrafish, generating F1 animals heterozygous for different mutations. DNA extraction and PCR followed by DNA Sanger sequencing analysis of F1 at 24 hpf identified the batches of F1 siblings that were more likely to contain a high proportion of frameshift mutations and the selected batches were raised to adulthood. F1 was screened to select a frameshift mutation and two siblings (one male and one female) that carried the same 19 bp frameshift mutation were in-crossed to generate F2 offspring that were 25% wild type, 50% heterozygous and 25% homozygous for the 19 bp mutation. The genotype of each F2 zebrafish was assessed to grow the animals and establish the mutant line. Further details of the method are described in the Supplementary Methods.

Behavioural tests

A battery of behavioural tests was performed on adult zebrafish (3-6 months-old) using mixed groups of both sexes: open field test, shoaling test, visually-mediated social preference (VMSP) test, black and white test, and aggression test (Supplementary Fig. 3). All the experiments were performed with homozygous mutant fish (rbfox1sa15940/sa15940 and rbfox1del19/del19), heterozygous (rbfox1sa15940/+ and rbfox1del19/+) and wild-type (TL rbfox1+/+ and TU rbfox1+/+) fish. In this first batch of experiments, the proportion of females in the tanks for both lines ranged from 50 to 70%. A second batch of experiments was performed with the rbfox1sa15940 fish segregating them by genotype and sex. Sex was phenotypically determined as it cannot be genotyped in zebrafish [24, 25]. In all instances, all fish were genotyped, sized-matched and 13 individuals were randomly selected per condition. These fish were then housed together in a single tank per condition for one week prior to and during the behavioural experiments. For each experiment, fish were placed in a different tank once tested to avoid retesting them and when the experiment was finished, fish were housed together in the original tank. All tanks were kept in close proximity within the facility and were subjected to identical environmental and housing conditions. To prevent any potential housing bias, caretakers responsible for the fish were blind to both the genotype and experimental details.

Experiments were completed between 9:00 and 18:00 and recorded using StreamPix 7 software (Norpix) and a digital camera. Fish were left for 30 min to habituate to the testing room before the experiment. To determine the appropriate number of individuals in each group, we utilised GPower 3.1 software [26]. The calculation was based on data from previous experiments conducted within the same experimental setup, aiming to guarantee sufficient statistical power to detect differences between the groups in the behavioural tests without using more animals than necessary. During the experiments, genotypes were alternated to prevent any bias resulting from the time of day or other possible confounders and all individuals were tested in the same setup and testing tanks. Most of the measures were performed automatically using a tracking system and analysed using software, ensuring blinding of the data. However, for a few measures that required manual quantification, a retrospective blinding process was implemented to ensure that there was no bias during the manual quantification of phenotypes. Further details of the tests are described in the Supplementary Methods.

Statistical methods

Statistical analysis of RT-qPCR and behavioural data were performed with GraphPad Prism 8 (GraphPad Software, La Jolla California USA). The data sets were assessed for normality using D’Agostino-Pearson and Shaphiro-Wilk normality test and either a one-way ANOVA test followed by a Tukey’s post-hoc test or a Kruskal-Wallis test with Dunn’s correction for multiple testing were used to compare between multiple groups. Statistical analysis of the visually-mediated social preference test (VMSP) was performed by a two-way ANOVA with Sidak’s post-hoc test or a Kruskal-Wallis test followed by a Dunn’s correction for multiple testing. Standard deviation (SD) is indicated in the figures for each group of data. In the behavioural tests, the median of the individual speed was used instead of the mean as it was more representative of non-normal data caused by a high degree of freezing behaviour.

Results

rbfox1 expression is restricted to neurons during development and is localised to specific forebrain, midbrain and hindbrain areas in adulthood

During early development (28 hpf), rbfox1 is expressed in spinal cord and hindbrain lateral neurons (Fig. 1A). At later developmental stages (2–5 dpf) rbfox1 expression is widespread in the mid- and hindbrain (Fig. 1A). These findings are in line with previous published data [27]. Furthermore, we found that during development rbfox1 is also expressed in the heart, in line with what has previously been described elsewhere [19].

Fig. 1: rbfox1 shows restricted neuronal expression during development and is localised to specific forebrain, midbrain and hindbrain areas during adulthood.
figure 1

rbfox1 in situ hybridisation on (A) zebrafish whole mount larvae and (B) adult zebrafish brains, TL background. A rbfox1 whole mount in situ hybridisation on zebrafish larvae at 28 h post fertilisation, 3-, 4- and 5-days post fertilisation. B rbfox1 in situ hybridisation on adult zebrafish brains, (a–c) forebrain and (d–e) midbrain transverse sections. A anterior thalamic nuclei; ATN, anterior tuberal nucleus; CP, central posterior thalamic nucleus; D, dorsal telencephalic area; GL, glomerular cellular layer; HaV, ventral habenular nucleus; Hv, ventral zone of periventricular hypothalamus; ICL, internal cellular layer; PGZ, periventricular grey zone; PPv, ventral part of the periventricular pretectal nucleus; TeO, optic tectum; TL, torus longitudinalis; TPp, periventricular nucleus of posterior tuberculum; Val, valvula cerebelli; Vd, dorsal nucleus of ventral telencephalic area; Vl, lateral nucleus of ventral telencephalic area; VM, ventromedial thalamic nuclei; Vv, ventral nucleus of ventral telencephalic area. Scale bars: 100 µm (a, b, c, d); 200 µm (a’, d’, d”, e).

In adult fish, rbfox1 is expressed along the entire rostro-caudal brain axis. In the pallial region of the forebrain, rbfox1 is expressed in the glomerular (GL), external (ECL) and internal (ICL) cellular layers of the olfactory bulbs (Fig. 1B – a, a’). More caudally, rbfox1 is expressed in the dorsal telencephalic area (D) and in the dorsal (Vd), lateral (Vl) and ventral (Vv) nuclei of ventral telencephalic area (Fig. 1B – a, b). In the diencephalon, rbfox1-expressing cells have been detected in the ventral habenular nucleus (HaV), and in the anterior (A) and ventromedial (VM) thalamic nuclei (Fig. 1B – c). rbfox1 is also expressed in the periventricular layer of the thalamic and hypothalamic areas including the ventral part of the periventricular pretectal nucleus (PPv), the central posterior thalamic nucleus (CP), the periventricular nucleus of posterior tuberculum (TPp), the anterior tuberal nucleus (ATN), and the ventral zone of the periventricular hypothalamus (Hv) (Fig. 1B – d, d”). In the midbrain, rbfox1 has been detected in the periventricular grey zone (PGZ) and in the torus longitudinalis (TL) (Fig. 1B – d, d’, e). Finally, in the hindbrain rbfox1 expression is observed in the lateral division of the valvula cerebelli (Val) (Fig. 1B – e).

No differences were observed in rbfox1 expression between TU and TL backgrounds (Supplementary Figure 4) at either larval or adult stages.

rbfox1 sa15940/sa15940 zebrafish do not express the complete rbfox1 mRNA sequence and do not show alterations in the expression of the other rbfox genes

The first mutant line that we characterised, rbfox1sa15940 (A > T, Chr3:28068329, GRCz11), has an intronic point mutation at the -2 position of the 3’ acceptor splicing site before the second/third exon of all but one of the annotated rbfox1 zebrafish isoforms (Supplementary Fig. 1), which would cause that this exon is skipped during the splicing process and a shorter and aberrant rbfox1 mRNA sequence produced. This mutation would cause the deletion of 81 aminoacids, which represent 22 to 54% of the original Rbfox1 protein isoforms, leading to a significant change in conformation that may affect the functionality of the mutant protein. By RT-qPCR, we observed a strongly decreased level of the expression of the rbfox1 exon situated after the intronic point mutation in both homozygous and heterozygous rbfox1sa15940 mutant adult brains (93% and 43% respectively) compared to WT (mean HZ = 0.47; mean HOM = 0.07, WT vs. HOM: p = 0.0002, Fig. 2A). We also observed a decreased level of the expression of this rbfox1 exon in both homozygous and heterozygous rbfox1sa15940 mutant 3dpf larvae (80% and 38% respectively) compared to WT (mean HZ = 0.62; mean HOM = 0.20, WT vs. HOM: p = 0.0036, Fig. 2D). These results suggest that this mutant line can be used to examine the effect of rbfox1 function in zebrafish.

Fig. 2: Relative expression of rbfox family mRNAs in adult brains and 3dpf larvae from the rbfox1sa15940 line.
figure 2

Adult brains: Relative brain expression of (A) rbfox1, (B) rbfox1l, rbfox2, rbfox3a and rbfox3b mRNA in adult brains. mRNA expression is normalised to the average expression of the mRNA in wild-type fish and to the reference housekeeping genes ribosomal protein L13a (rpl13) (A, B) or the eukaryotic translation elongation factor 1 alpha 1a (eef1a1a) (B). Kruskal-Wallis test followed by Dunn’s multiple comparison test. n = 5-7 WT, 5-7 HZ, 5-7 HOM. Mean ± SD. C Brain expression of the rbfox family mRNAs in adult wild-type zebrafish brains represented as the quantification cycle value for each gene. n = 7 WT, 7 HZ, 7 HOM. Mean ± SD. 3dpf larvae: Relative brain expression of (D) rbfox1, (E) rbfox1l, rbfox2, rbfox3a and rbfox3b mRNA in 3 dpf whole larvae. mRNA expression is normalised to the average expression of the mRNA in wild-type fish and to the reference housekeeping gene ubiquitin-conjugating enzyme E2A (ube2a). Kruskal-Wallis test followed by Dunn’s multiple comparison test. n = 5 WT, 5 HZ, 5 HOM. Mean ± SD. F Brain expression of the rbfox family mRNAs in 3 dpf wild-type larvae represented as the quantification cycle value for each gene. n = 5 WT, 5 HZ, 5 HOM. Mean ± SD. Each point represents the results from a pool of 10 larvae.

We explored the expression levels of the four other rbfox zebrafish genes in both adult brains and 3dpf larvae. We observed no differences in the expression of rbfox1l, rbfox2, rbfox3a and rbfox3b in brain between WT and mutant rbfox1sa15940 adult fish (Fig. 2B) and in the expression of rbfox1l, rbfox2 and rbfox3a between WT and mutant rbfox1sa15940 larvae (Fig. 2E). The expression of rbfox3b was slightly reduced in homozygous mutant rbfox1sa15940 larvae compared to WT and HZ (WT vs. HOM: p = 0.049, HZ vs. HOM: p = 0.011, Fig. 2E). We also found that rbfox1l expression in the WT adult brain was very low compared to the expression of the other rbfox genes in this tissue, as its Cq (quantification cycle value) is much higher in the RT-qPCR analysis (Fig. 2C). Unlike in adult brains, the expression levels of all the rbfox genes in whole WT 3dpf larvae were similar, with Cq ranging between 22 and 26 (Fig. 2F).

Finally, the expression levels of the rbfox genes at 6dpf could not be assessed as all the reference genes considered (rpl13a, eef1a1a, ube2a and tmem50a) did not show stable expression among individuals and genotypes at this larval stage.

rbfox1 sa15940 mutation in zebrafish produces behavioural alterations

We performed a battery of five behavioural tests (open field test, shoaling test, VMSP test, black and white test and aggression test) (Supplementary Fig. 3) in TL WT rbfox1+/+, heterozygous (HZ) rbfox1sa15940/+, and homozygous (HOM) rbfox1sa15940/sa15940 adult fish, to investigate whether this rbfox1 mutation affects behaviour (Fig. 3 and Supplementary Fig. 5).

Fig. 3: Behavioural alterations observed in the rbfox1sa15940 line.
figure 3

A Open field test. Time spent in the centre of the arena, time spent freezing and total distance travelled during the open field test. One-way ANOVA followed by Tukey’s multiple comparison test. B Visually-mediated social preference test (VMSP). Social preference step. Time spent in the area close to the 1st strangers and in the opposite area, time spent freezing and total distance travelled during the social preference step of the VMSP test. Two-way ANOVA followed by Sidak’s multiple comparison test. C Visually-mediated social preference test (VSMP). Preference for social novelty step. Time spent in the areas close to the 1st or 2nd strangers, time spent freezing and total distance travelled during the preference for social novelty step of the VMSP test. Two-way ANOVA followed by Sidak’s multiple comparison test. D Shoaling test. Mean of interindividual distance, nearest neighbour distance, cluster score and total distance travelled during the shoaling test. One-way ANOVA followed by Tukey’s multiple comparison test and Kruskal-Wallis followed by Dunn’s multiple comparisons test. E Black and white test. Number of crossings between areas and time spent in the white area of the tank during the black and white test. Kruskal-Wallis followed by Dunn’s multiple comparisons test. F Mirror test. Time spent exhibiting an aggressive behaviour against the mirror. For all the experiments except for the shoaling test: HOM, rbfox1sa15940/sa15940 fish; HZ, rbfox1sa15940/+ fish; WT, wild-type TL. n = 13 WT, 13 HZ and 13 HOM for all tests except for the shoaling test. For the shoaling test: n = 2 groups of 5 individuals per genotype. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Mean ± SD.

In this mutant line, all HZ and HOM individuals spend less than 20% of the time in the centre of the open field arena and show thigmotaxis, a behaviour that could be related to anxiety or stereotypies, whereas TL WT fish do not show preference to swim close to the walls of the arena (Fig. 3A and Supplementary Fig. 5A). In addition, HZ and HOM fish spend less time freezing than TL WT fish (WT vs. HZ, p = 0.0068; WT vs. HOM, p = 0.0001; Fig. 3A) and show hyperactivity, as they swim longer distances (WT vs. HZ, p = 0.0027; WT vs. HOM, p = 0.0002; Fig. 3A). They also present a higher swimming speed than TL WT individuals (WT vs. HZ, p = 0.0026; WT vs. HOM, p = 0.0002; Supplementary Fig. 5A).

In the VMSP we did not observe differences in social preference between genotypes for this line (Fig. 3B, C, and Supplementary Fig. 5B, C). In the first step, all the genotypes prefer to stay close to the group of stranger fish rather than in the opposite corner (1st strangers vs. Opposite area: WT, p < 0.0001; HZ, p < 0.0001; HOM, p = 0.0005; Fig. 3B) and in the second step all the genotypes show an equal preference for both stimulus groups (1st strangers vs. 2nd strangers: WT, p > 0.99; HZ, p = 0.90; HOM, p = 0.61; Fig. 3C). However, in the first step of the test, mutant fish again showed hyperactivity, reflected by more distance travelled (HZ vs. HOM, p = 0.0282; WT vs. HOM, p = 0.0487; Fig. 3B) and a higher speed of HOM fish compared to TL WT individuals (WT vs. HOM, p = 0.0130; Supplementary Fig. 5B).

In the shoaling test, we observed thigmotaxis in rbfox1sa15940 mutant fish (Supplementary Fig. 5D) and we found differences in the mean interindividual distance (IID), which was higher in HZ and HOM compared to TL WT fish (WT vs. HZ, p = 0.0194; WT vs. HOM, p = 0.0005; Fig. 3D). No differences were found in the time spent in the white chamber of the black and white test, but HOM fish cross more times the limit between areas, a sign of hyperactivity (WT vs. HOM, p = 0.0334; Fig. 3E). Finally, no differences were observed between mutants and WT fish in the aggression test, but HOM fish are significantly more aggressive than HZ fish, as they spend more time exhibiting aggressive behaviour against a mirror (HZ vs. HOM, p = 0.0083; Fig. 3F).

Taken together, these results show behavioural alterations in rbfox1sa15940 mutants in the TL genetic background that include hyperactivity, thigmotaxis and alterations in social behaviour.

We also investigated potential sex differences in the effects of this rbfox1sa15940 mutation. In a second batch of experiments, fish were segregated by sex before and during the behavioural tests. We did not observe any significant variations in behaviour between males and females within the mutant groups, and observed only slight differences in locomotion during the initial step of the VMSP among the WT fish, where WT females travelled shorter distances and exhibited lower swimming speeds compared to WT males (Supplementary Fig. 6). In this second batch, encompassing analyses of males, females, and combined groups, the mutant fish did not show hyperactivity or distinct thigmotaxis. However, alterations in social behaviour were evident in both VMSP and shoaling tests (Supplementary Fig. 7A, B and C). Furthermore, we observed behavioural differences between the TL WT groups from the first and second batches (Supplementary Fig. 7D), which could potentially account for the different outcomes obtained in these two batches.

rbfox1 del19 mutation in zebrafish affects behaviour similarly

We then repeated the battery of behavioural tests in a second rbfox1 mutant line with a TU genetic background, rbfox1del19, to investigate if rbfox1-deficiency affects behaviour also in this line. This line was created by using the CRISPR/Cas9 genome editing technique, causing a frameshift deletion of 19 bp in exon 2 that disrupts the rbfox1 coding sequence and produces a premature stop codon (Supplementary Fig. 2 and 8). We observed behavioural differences between rbfox1del19 mutants and TU WT fish in all the tests performed, although some of the behavioural changes differed from those obtained for the rbfox1sa19540 line (Fig. 4 and Supplementary Fig. 9).

Fig. 4: Behavioural alterations observed in the rbfox1del19 line.
figure 4

A Open field test. Time spent in the centre of the arena, time spent freezing and total distance travelled during the open field test. One-way ANOVA followed by Tukey’s multiple comparison test. B Visually-mediated social preference test (VMSP). Social preference step. Time spent in the area close to the 1st strangers and in the opposite area, time spent freezing and total distance travelled during the social preference step of the VMSP test. Two-way ANOVA followed by Sidak’s multiple comparison test. C Visually-mediated social preference test (VSMP). Preference for social novelty step. Time spent in the areas close to the 1st or 2nd strangers, time spent freezing and total distance travelled during the preference for social novelty step of the VMSP test. Two-way ANOVA followed by Sidak’s multiple comparison test. D Shoaling test. Mean of interindividual distance, nearest neighbour distance, cluster score and total distance travelled during the shoaling test. One-way ANOVA followed by Tukey’s multiple comparison test and Kruskal-Wallis followed by Dunn’s multiple comparisons test. E Black and white test. Number of crossings between areas and time spent in the white area of the tank during the black and white test. Kruskal-Wallis followed by Dunn’s multiple comparisons test. F Mirror test. Time spent exhibiting an aggressive behaviour against the mirror. For all the experiments except for the shoaling test: HOM, rbfox1del19/ del19 fish; HZ, rbfox1del19/+ fish; WT, wild-type TU. n = 13 WT, 13 HZ and 13 HOM for all tests except for the shoaling test. For the shoaling test: n = 2 groups of 5 individuals per genotype. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Mean ± SD.

Similar to findings in rbfox1 sa15940, rbfox1del19 mutants tend to spend less time in the centre than TU WT fish, being significant for HZ fish (WT vs. HZ, p = 0.0467, Fig. 4A) and present with thigmotaxis (Supplementary Fig. 9A). However, we also observed differences in behaviour in the open field test between rbfox1del19 and rbfox1sa15940 lines: we did not find differences in locomotor activity (nor in distance travelled or speed) and freezing behaviour between genotypes in the rbfox1del19 line (Fig. 4A and Supplementary Fig. 9A).

In the preference step of the VMSP test, TU WT and HZ rbfox1del19 fish show a preference to stay close to stranger fish, whereas HOM rbfox1del19 fish show no social preference (1st strangers vs. Opposite area: HOM, p = 0.6979; Fig. 4B) and spend significantly less time than TU WT fish near strangers and more in the opposite area (WT vs. HOM, p = 0.0057; Supplementary Fig. 9B). In the social novelty preference step, we observed similar behaviour in both rbfox1del19 and rbfox1sa15940 lines: none of the genotypes show preference for a group of strangers (Fig. 4C and Supplementary Fig. 9C). In line with the rbfox1sa15940 results, HOM rbfox1del19 fish present hyperactivity in the two steps of the VMSP test, reflected by a higher speed (WT vs. HOM, p = 0.0339; Supplementary Fig. 9B) and a further distance travelled (WT vs. HOM, p = 0.0380; Fig. 4C) than TU WT.

We found similar results in both rbfox1 HOM lines in the shoaling and black and white tests: mutant rbfox1del19 fish present impaired social behaviour (IID: WT vs. HZ, p = 0.0235; WT vs. HOM, p < 0.0001; HZ vs. HOM, p = 0.0047; NND: WT vs. HOM, p < 0.0001; Fig. 4D) and thigmotaxis (Supplementary Fig. 9D) and HZ and HOM rbfox1del19 performed a higher number of crossings between areas than WT (WT vs. HZ, p = 0.0040; WT vs. HOM, p = 0.0006; Fig. 4E). Finally, in contrast to HOM rbfox1sa15940 fish, HOM rbfox1del19 fish were not more aggressive than HZ fish (Fig. 4F).

In summary, both rbfox1sa15940 and rbfox1del19 mutants show hyperactivity, thigmotaxis and impaired social behaviour. However, each rbfox1 line presents particularities: rbfox1sa15940 mutants show alterations in freezing behaviour and trends of aggression while rbfox1del19 mutants have stronger social impairments. The behavioural differences reported between the two rbfox1 mutant lines might be explained by environmental effects and genetic background differences that modulate rbfox1 effect on behaviour. Indeed, we can see that some behavioural aspects are different between the two WT lines, as we observe strong differences in the freezing behaviour (Supplementary Fig. 10). Finally, even though discrepancies are reported, the effect of mutations in rbfox1 on behaviour in these two zebrafish models is in line with previous results found in Rbfox1-deficient mice [6], as summarised in Table 1.

Table 1 Summary of behavioural alterations in rbfox1sa15940 and rbfox1del19 mutant zebrafish and Rbfox1 depleted mice [6].

Discussion

In this study we have investigated the role of rbfox1 in neurodevelopmental and psychiatric disorders by studying the behavioural effects of rbfox1 function in zebrafish. This gene has previously been reported to be highly pleiotropic, contributing to several psychiatric disorders [13, 14, 28]. In addition, we have validated zebrafish rbfox1sa15940 and rbfox1del19 HOM lines as models of neurodevelopmental and psychiatric conditions.

First, rbfox1 shows a restricted expression in brain and heart across developmental stages that suggests an important role of this gene during brain zebrafish development, in line with previous findings. Indeed, a study in human neural progenitor cells demonstrated that RBFOX1 regulates splicing and expression of large gene networks implicated in neuronal development and maturation [29], and another study showed that Rbfox1 controls synaptic transmission in the mouse brain [15, 30]. Also, previous studies in mice have shown that specific Rbfox1 deficiency in the central nervous system leads to impairments in neuronal migration, axon extension, dendritic arborisation and synapse network formation, suggesting that loss of Rbfox1 function contributes to the pathophysiology of neurodevelopmental disorders [31,32,33]. Finally, several point mutations and copy number variations (CNVs) in RBFOX1 have been described in patients with neurodevelopmental disorders, such as ASD and ADHD [4,5,6, 10, 34]. We therefore hypothesise that altered rbfox1 function may affect brain maturation in zebrafish and therefore lead to impaired neuronal function and transmission during adulthood, with implications in the sensory response to the environment and in behaviour.

In addition, rbfox1 specific expression is found mainly in forebrain areas in adult WT zebrafish, including the dorsal and ventral telencephalon, thalamus and periventricular hypothalamus. Interestingly, these areas are involved in receiving and processing sensory information, stress, and in directing behaviour, especially social behaviour and emotion [35,36,37,38]. Given the important role of Rbfox1 in controlling splicing and expression in neurons, rbfox1 mutations may induce an impaired neuronal function in these areas with an impact on sensory processing, stress and behaviour in zebrafish.

Interestingly, both rbfox1sa15940 and rbfox1del19 HOM lines present alterations in behaviour. rbfox1sa15940 mutants present hyperactivity, thigmotaxis – a behaviour related to anxiety and stereotypies –, decreased freezing behaviour and altered social behaviour. rbfox1del19 mutants present similar thigmotaxis, but stronger alterations in social behaviour and lower levels of hyperactivity than rbfox1sa15940 fish. These results are in line with the behavioural alterations observed in a neuron-specific Rbfox1 KO mouse line that presents decreased Rbfox1 expression, as Rbfox1 KO mice show a pronounced hyperactivity, thigmotaxis and reduced social interest [6]. All these behavioural phenotypes can be assimilated to phenotypic alterations observed in patients with psychiatric or neurodevelopmental conditions. For example, social impairment is a symptom of ASD, hyperactivity of ADHD, and thigmotaxis is considered an anxiety-like behaviour in mouse and zebrafish.

We found differences in behaviour between rbfox1sa15940 and rbfox1del19 lines. On one hand, rbfox1sa15940 is a hyperactive line that presents with thigmotaxis and slight social impairments. On the other hand, rbfox1del19 fish present also with thigmotaxis, show only hyperactivity in one of the tests performed, and present stronger social impairments than rbfox1sa15940 fish. The phenotypic variations observed between the two zebrafish lines are likely attributable to genetic background disparities and/or environmental influences, although we cannot rule out differential effects of the two rbfox1 mutations (e.g., exon skipping in the rbfox1sa15940 line versus a frameshift deletion in the rbfox1del19line). Another factor that may contribute to the behavioural differences found between lines is the tank effect: each genotype was segregated in a separate tank and therefore behaviour could be influenced by the conspecifics present in the tank. In addition, behavioural differences between WT TL and TU strains have been previously reported, WT TL fish being considered more anxious and sensitive to anxiogenic stimuli than TU WT fish [39]. Our results are in line with these reported phenotypes, as we found that TL WT presents a strong freezing behaviour, especially in the open field test, that is not present in TU WT fish. In addition, Rbfox1 KO mice present behavioural alterations not described in the zebrafish lines such as, lack of aggressive behaviour, and behaviours that could not be tested in our zebrafish lines for practical reasons such as deficit in the acoustic startle response and impairments in fear acquisition [6]. Given the differences observed between the WT zebrafish lines, we hypothesised that the rbfox1 mutations alter behaviour differently depending on other environmental and genetic effects.

Moreover, when separating rbfox1sa15940 fish by sex in a second batch of experiments, the results obtained were different from the first batch, although social behaviour was shown to be altered as well. Indeed, the WT fish from the two batches behave differently in some tests, being more active in the second batch. These differences might be explained again by the influence of the environment and the genetic background. The fish used in this second batch come from a new generation of rbfox1sa15940 fish that were bred with a different TL WT strain and it is known that zebrafish strains are not completely inbred and genetically well-defined as it is the case with laboratory mice [40], which might lead to variations in the genetic background between these two batches. In addition, housing fish in sex-separated groups before and during the experiments has been described to affect behaviour [41].

These results suggest that, on one side, environmental effects might play a role when assessing behavioural effects of a genetic variation and, on the other side, that the effects of variants in other genes may contribute to the final phenotype, in agreement with a recent proposed genetic model for complex psychiatric disorders composed by ‘hub’ and ‘peripheral’ genes [42,43,44,45]. Our results show that the damaging effect of a mutation in rbfox1 may be modulated by genetic and environmental effects and therefore lead to different phenotypes, which is also in line with the different diagnosis of patients with rare CNVs or point mutations in the RBFOX1 gene as well as the contribution of common variants to different psychiatric disorders [6, 10, 11, 14, 46].

To conclude, all these results show that altered function of rbfox1 in zebrafish and mice leads to behavioural alterations that can be related to different neurodevelopmental and psychiatric disorders. Thus, our data contribute to a better understanding of the involvement of RBFOX1 in psychiatric disorders and point to a pleiotropic contribution of this gene that can be modulated by other environmental and genetic factors. In addition, we have validated two new rbfox1 HOM zebrafish lines to be used as models for psychiatric disorders, in which further experiments can be performed to unravel the molecular mechanisms that link RBFOX1 with psychiatric phenotypes.