Social Deficits and Repetitive Behaviors Are Improved by Early Postnatal Low-Dose VPA Intervention in a Novel shank3-Deficient Zebrafish Model

Mutations of the SHANK3 gene are found in some autism spectrum disorder (ASD) patients, and animal models harboring SHANK3 mutations exhibit a variety of ASD-like behaviors, presenting a unique opportunity to explore the underlying neuropathological mechanisms and potential pharmacological treatments. The histone deacetylase (HDAC) valproic acid (VPA) has demonstrated neuroprotective and neuroregenerative properties, suggesting possible therapeutic utility for ASD. Therefore, SHANK3-associated ASD-like symptoms present a convenient model to evaluate the potential benefits, therapeutic window, and optimal dose of VPA. We constructed a novel shank3-deficient (shank3ab–/–) zebrafish model through CRISPR/Cas9 editing and conducted comprehensive morphological and neurobehavioral evaluations, including of core ASD-like behaviors, as well as molecular analyses of synaptic proteins expression levels. Furthermore, different VPA doses and treatment durations were examined for effects on ASD-like phenotypes. Compared to wild types (WTs), shank3ab–/– zebrafish exhibited greater developmental mortality, more frequent abnormal tail bending, pervasive developmental delay, impaired social preference, repetitive swimming behaviors, and generally reduced locomotor activity. The expression levels of synaptic proteins were also dramatically reduced in shank3ab–/– zebrafish. These ASD-like behaviors were attenuated by low-dose (5 μM) VPA administered from 4 to 8 days post-fertilization (dpf), and the effects persisted to adulthood. In addition, the observed underexpression of grm5, encoding glutamate metabotropic receptor 5, was significantly improved in VPA-treated shank3ab–/– zebrafish. We report for the first time that low-dose VPA administered after neural tube closure has lasting beneficial effects on the social deficits and repetitive behavioral patterns in shank3-deficient ASD model zebrafish. These findings provide a promising strategy for ASD clinical drug development.


INTRODUCTION
Autism spectrum disorder (ASD) encompasses a group of neurodevelopmental syndromes characterized by deficits in social interaction and communication as well as repetitive behaviors and restricted interests (American Psychiatric Association, 2013). There is strong evidence for the involvement of inherited genetic factors in ASD (accounting for at least 80% of the variation in disease risk) (Jutla et al., 2019;Liu et al., 2021). Furthermore, mutations in numerous genes encoding synaptic proteins have been identified in patients with ASD and intellectual disability (Verpelli and Sala, 2012;Lai et al., 2014). According to a meta-analysis, monogenic mutations in SHANK3, which encodes the major postsynaptic density (PSD) scaffolding protein at excitatory glutamatergic synapses, are found in approximately 0.69% of ASD cases and up to 2.12% of all moderate to profound intellectual disability cases (Leblond et al., 2014). De novo mutations, interstitial deletions, and terminal deletions have been identified in ASD (Durand et al., 2007;Moessner et al., 2007;Gauthier et al., 2009;Boccuto et al., 2013;Leblond et al., 2014). Additionally, SHANK3 mutations underlie Phelan-McDermid syndrome (PMS, also known as 22q13.3 deletion syndrome), a rare autosomal dominant neurodevelopmental disorder characterized by autistic-like behaviors, absent to severely delayed speech, developmental delay, and moderate to profound intellectual disability as well as neonatal hypotonia and minor dysmorphic facial features (Phelan et al., 1993;Wilson et al., 2003;Phelan, 2008;Bonaglia et al., 2011;Phelan and McDermid, 2012). The genomic rearrangements in PMS are diverse, ranging from simple 22q13 deletions (72%), ring chromosomes (14%), and unbalanced translocations (7%) to interstitial deletions (9%), all leading to SHANK3 haploinsufficiency (Bonaglia et al., 2011). Although the severity of the developmental delay tends to vary with deletion size (Sarasua et al., 2011;Zwanenburg et al., 2016), individuals with the same size deletion may exhibit vastly different degrees of disability (Dhar et al., 2010). Thus, SHANK3 deficits appear to profoundly disrupt the neural circuitry required for social behavior, communication, and cognition.
The SHANK3 gene (also known as ProSAP2, at 22q13.33) is the best studied of the three SHANK family members, which encoding an extensive number of mRNA and protein isoforms via multiple intragenic promoters and alternative splicing (Durand et al., 2007;Wang et al., 2014b). In the brain, Shank3 mRNA is enriched in the cortex, thalamus, striatum, hippocampus, dentate gyrus, and cerebellar granule cells (Peca et al., 2011;Wang et al., 2016;Monteiro and Feng, 2017), suggesting important functions in synaptic plasticity underlying cortical organization, sensory processing, behavioral control, and cognition.
Owing to the strong genetic association between SHANK3 deficiency and ASD, many studies have focused on the neurodevelopmental functions of this particular gene. Numerous animal models of SHANK3 deficiency, including zebrafish, Drosophila, rat, mouse, and monkey models, demonstrate ASDlike behaviors, suggesting a causative role of SHANK3 deficiency in ASD (Peca et al., 2011;Wang et al., 2016;Monteiro and Feng, 2017). Wang et al. (2016) reported that homozygous Shank3 knockout mice displayed core behavioral features of ASD as well as impaired mGluR5-Homer association at the PSD, resulting in corticostriatal circuit abnormalities that may underlie learning deficits and ASD-like behaviors. In addition, monkey models also displayed core ASD features including impaired social interactions, repetitive behaviors, delayed vocalization, and reduced brain network activities (Tu et al., 2019;Zhou et al., 2019). The zebrafish genome harbors two homologs of human SHANK3, shank3a, and shank3b. In our previous studies, we generated the first shank3b loss-of-function mutation in zebrafish and reported prominent ASD-like behaviors (Liu et al., 2018). However, we did not examine the effects of shank3a and shank3b double mutant combinations, which would be more analogous to mammalian models harboring a single SHANK3 mutation, or assess potential pharmacological strategies to mitigate behavioral deficits.
Current treatment options for ASD are limited, especially pharmacotherapies (Wang et al., 2014a;Penagarikano et al., 2015). Evidence-based treatments for ASD children are restricted mainly to educational practices, and intensive behavioral interventions such as Treatment and Education of Autistic and Related Communication-Handicapped Children (TEACCH) and the Early Start Denver Model (ESDM) (Dawson et al., 2010;Hyman et al., 2020). Outcomes of these behavioral therapies vary markedly according to intervention intensity, disease severity, and a variety of other factors (Sandbank et al., 2020). Further, education and behavioral interventions do not target the underlying neurobiological mechanisms (Wang et al., 2014a;Weitlauf et al., 2014) and are costly both for educational institutions and primary caregivers (Lord et al., 2018(Lord et al., , 2020. Similarly, current pharmacologic treatments address only the associated symptoms or comorbidities, including agitation and hyperactivity, rather than the core symptoms and underlying causes (Fung et al., 2016;Lord et al., 2018Lord et al., , 2020Muhle et al., 2018). Risperidone and aripiprazole are approved by the United States Food and Drug Administration (FDA) to treat comorbidities common in ASD, including irritability, and agitation (Lord et al., 2018(Lord et al., , 2020, but similar to behavioral interventions, these evidence-based pharmacologic treatments lack sufficient biological support. Based on available evidence, behavioral interventions should be implemented as early and intensively as possible following ASD diagnosis to improve the cognitive and adaptive outcomes of preschoolers (Weitlauf et al., 2014;Muhle et al., 2018). The preschool years are critical for acquiring language and social skills, key areas of difficulty in ASD, as this period coincides with the temporal window of enhanced plasticity in relevant neural circuits (Franz and Dawson, 2019). Similar to early intervention, pharmaceutical treatments appear effective in animal models when administered early, and Muhle et al. (2018) even suggested greater emphasis on early drug treatment rather than strict adherence to the standard timeline of efficacy based on studies in adults and adolescents. For instance, early postnatal treatment improves social deficits in adult mice with mutations in the ASD risk gene cntnap2 (Franz and Dawson, 2019). Pharmacological inhibitors of histone deacetylase (HDAC) have garnered interest as possible ASD therapeutics due to demonstrated neuroprotective efficacy (Fischer et al., 2010). The class I HDAC inhibitor valproic acid (VPA) was found to reduce repetitive behaviors in a small randomized controlled trial involving 13 ASD children (Hollander et al., 2006). In addition, several studies have reported that VPA can attenuate irritability in young ASD patients (Hellings et al., 2005;DeFilippis and Wagner, 2016). Further, three daily VPA treatments transiently restored social preference deficits in adult Shank3-deficient mice, although the effect disappeared within a few days following treatment (Qin et al., 2018). Therefore, the effects of VPA on shank3 mutant models warrant further study.
Here we investigated the developmental characteristics of shank3-deficient zebrafish, neurobehavioral features relevant to ASD, and the effects of various VPA treatment regimens. We speculated that VPA administration in the early postnatal period would be more effective at reversing the core ASD-like deficits in shank3-deficient zebrafish than juvenile or adult treatment.

Zebrafish and Embryo Maintenance
Wild-type (WT) zebrafish of Tu strains were acquired from Children's Hospital of Fudan University. They were raised and maintained under standard laboratory conditions at 28.5 • C in "system water" under a 14 h light/10 h dark cycle according to standard protocols (Kalueff et al., 2014;Evans and Erickson, 2019). Freshly fertilized eggs were collected from multiple breeding tanks containing 25 females and 25 males. All animal experimental procedures were in compliance with local and international regulations, and approved by the institutional animal care committee of Children's Hospital of Fudan University.

Generation of shank3a and shank3b Double Deficient Zebrafish Model
Zebrafish shank3a and shank3b genes and their exon/intron boundaries were identified by searching the NCBI database (gene ID: shank3b, NC_007115.7; shank3a, NC_007129.7). Mutations in shank3a and shank3b were generated using CRISPR/Cas9 editing as previously reported (Hwang et al., 2013;Mali et al., 2013). The CRISPR/Cas9 target of shank3a was 5 -GGACCCCAGCCCTCCTCCCGTGG-3 and that of shank3b was 5 -GGGCGTGTTGTTGCCACGGCCGG-3 (Liu et al., 2018;Supplementary Table 1). In vitro-transcribed RNA of the guide RNA (120 ng each) and Cas9 mRNA (500 pg) were microinjected into WT zebrafish embryos (F 0 ) at the one-cell stage. The progeny were propagated via a series of out-crossings with WT zebrafish and genotyping of each generation. Eventually, these animals were in-crossed to obtain the homozygous knockouts shank3a −/− and shank3b −/− . The shank3ab −/− homozygous line was obtained by mutant crossing and subsequent genotyping.

Behavior Tests for Adult Zebrafish
All behavioral experiments were conducted on 2-3.5 month old male zebrafish between 10 a.m. and 4 p.m. Behaviors were recorded for 30 min after 1-2 min habituation period using a video camera (zebrabox) suspended above the test tank. Zebrafish were returned to their home tanks immediately after completion of the test. The raw data was analyzed using Viewpoint software.

Open Field Test
The free-swimming open field test was performed in novel tanks as previously described (Liu et al., 2018). Each tank was 30 cm × 30 cm × 30 cm, and its walls consisted of opaque partitions. Swim velocity was calculated as the total distance traveled divided by the total swim time.
For the danger awareness test, the tank was virtually divided into two equal areas, peripheral and center, and greater peripheral swimming distance relative to central swimming distance was measured as a metric of danger awareness.
For analyzing repetitive and stereotyped behaviors, we used a double-blind method to score the swimming pattern within each minute, and counted the number of the four stereotyped swimming pattern episodes separately.

Shoaling Test
The shoaling test assesses social cohesion in homogeneous groups of zebrafish ( Figure 3A; Liu et al., 2018). A shoal refers to a loose aggregation of individuals who swim close to one another, whereas a school describes a group of fish exhibiting polarized, synchronized motion (Perathoner et al., 2016). The distance between each fish can reveal the degree of shoaling behavior (i.e., social cohesion). Six zebrafish were placed in a novel 30 cm × 30 cm × 30 cm tank with walls consisting of opaque partitions, and mean inter-individual distance was measured (Kim et al., 2017).

Social Preference Test
Sociability was evaluated as the difference score between the time spent in proximity to a conspecific sector and an empty sector (Busnelli et al., 2016). Briefly, social testing was conducted over 30 min in a standard mating tank of inner dimensions 21 cm × 10 cm × 7.5 cm. A transparent plexiglas divider was placed in the middle of the tank, which allowed sufficient visual presentation for forming a social preference, and a single shank3 mutant or WT zebrafish was placed on the left side while a group of six conspecific zebrafish (conspecific sector) was placed in the right side ( Figure 3C). Social preference behavior was quantified as a distance distribution or as presence in a zone adjacent to the group of conspecifics. The distance ratio was calculated as the distance swam in the conspecific sector divided by the total distance.

Kin Preference Test
Another test was performed to assess preference for kinship using various colored variants. The duration and frequency of contact was compared between conspecifics and a phenotypically distinct strain ( Figure 3E) in mating tanks with dimensions and configuration the same as those used in the "social preference test." Briefly, two transparent separators divided the tank into three compartments, with a single test fish placed in the middle and Kin zebrafish placed on the right and non-Kin (red color) zebrafish placed on the left. Kin preference was represented by the ratio of the time spent in the Kin-sector to the total time.

VPA Treatment and Phenotypic Assessments
To assess the extent to which VPA exposure affects morphology, 4 dpf WT or shank3ab −/− larvae were reared in Petri dishes containing blue egg water alone or blue egg water containing 5, 10, 20, or 50 µM sodium valproate (Cat No. 4543-10G, Sigma-Aldrich). The egg water with or without VPA was changed daily. At 8 dpf, larvae were observed under a microscope for mortality and any morphologic abnormalities, including distended abdominal and thoracic regions, lordosis, yolk sac edema, and pericardial edema. Adverse effects including mortality and malformation rates were calculated to determine the optimal VPA concentration for subsequent experiments (Supplementary Figure 1).
To examine the effects of early postnatal low-dose VPA exposure on autism-like behaviors, WT or shank3ab −/− larvae were exposed to blue egg water with or without 5 µM VPA from 4 to 8 dpf. At 8 dpf, each larva was pipetted into fresh paramecium liquid, and raised to 2.5 months old (juvenile) or 3.5 months old (adult). The juveniles were then examined for 30 min using the 1 versus 6 social preference assay, while adults were subjected to social preference, repetitive behavior, locomotor activity, and thigmotaxis tests to comprehensively evaluate the effects of VPA on autism-like behaviors.

Real-Time Quantitative PCR
Total RNA was extracted from 15 WT, shank3a −/− , shank3b −/− , and shank3ab −/− larvae each at 3.5-4.5 months post-fertilization (mpf) using the RNA Extraction Kit from Takara, and reverse transcribed to cDNA using the PrimeScript TM reagent Kit with gDNA Eraser (Takara) according to the manufacturer's recommendations. The Cas9 target region of shank3a and shank3b were amplified in duplicate samples from shank3ab −/− zebrafish by real-time quantitative PCR (RT-qPCR) to confirm genotype (Figures 1C,D and Supplementary Table 1).
To assess the effect of VPA exposure on synaptic genes and class I hdac genes (as VPA belongs to class I HDAC inhibitor), groups of ∼15 WT and shank3ab −/− zebrafish larvae were exposed to vehicle or VPA from 4 to 8 dpf, reared under normal conditions, then sacrificed for whole-brain total RNA isolation. The expression levels of the following genes analyzed by RT-qPCR: NMDAR subunits (grin1a, grin1b, grin2bb, grin2ca, grin2da, grin2aa), AMPAR subunits (gria1a, gria1b, gria2b), mGluR subunits (grm1a, grm1b, grm5a), and class I hdacs (hdac1, hdac3, hdac8). We selected β-actin or Rpl13α as internal controls because both are expressed in the brain throughout development. Primer sequence are shown in Supplementary Table 1.

Statistical Analysis
Values are presented as mean ± SEM. All data were analyzed using SPSS 20.0. In all experiments, WT and shank3-deficient zebrafish were compared by two-sided unpaired Student's t-tests, while three or more groups were compared by analysis of variance (ANOVA). Genotypes within treatment groups were compared by one-way ANOVA. All experiments were conducted in triplicate using independently treated animals. A P < 0.05 was considered statistically significant for all tests.
Adult WT zebrafish typically avoid open areas near the water surface for protection against predation. To examine whether shank3 deficiency modulates these avoidance behaviors, the relative proportions of swim time and distance in the pool periphery (thigmotaxis) versus the center (dotted line in Figure 2B) were calculated in a novel square tank with opaque walls. All shank3-deficient genotypes spent a significantly greater proportion of total swim time and traveled longer distances in the center of the tank compared to WT zebrafish, and shank3ab −/− zebrafish exhibited the greatest peripheral to center distance ratio of the three mutant genotypes (Figures 2D,E). This behavior can be interpreted as reduced alertness or reduced danger awareness (Mathur and Guo, 2010).

Core ASD-Like Behaviors of shank3-Deficient Adult Zebrafish
Since ASD diagnosis is based on behavioral criteria, a valid zebrafish model should exhibit core behavioral symptoms, including impaired social interactions and repetitive and stereotyped behaviors. The shoaling test showed that adult WT zebrafish spent the majority of swimming time in compact schools, while all three shank3-deficient genotypes swam in looser and larger schools with more frequent deviation (leaving the group), resulting in a greater average inter-fish distance compared to WT zebrafish ( Figure 3B). This social deficit was particularly strong among the shank3ab −/− zebrafish group.
Social preference was further assessed by measuring conspecific proximity. WT zebrafish (3.5 mpf) maintain closer proximity with a conspecific group members on the right side ( Figure 3C). In contrast, all shank3-deficient genotypes showed reduced frequency and duration of conspecific proximity, again with the shank3ab −/− genotype demonstrating the greatest average inter-conspecific distance ( Figure 3D).
Wild type zebrafish also typically spend more time with a Kin group (conspecific and same color) than a non-Kin group in mixed populations. However, this Kin recognition and preference was markedly reduced in all shank3-deficient genotypes as measured by the proportion of time spent in close proximity with non-Kin (red-skinned) zebrafish among a mixed population ( Figure 3F). Consistent with other social behavior tests, shank3ab −/− zebrafish spent the least amount of time in proximity to other conspecifics.
Repetitive and stereotyped behavior is another core symptom of ASD. Compared to adult WTs, shank3ab −/− zebrafish demonstrated greater behavioral perseveration, including repetitive stereotypic "figure 8" swimming, cycling behavior (swimming in circles), and other locomotor changes and patterns such as stereotyped "corner" or "wall" swimming ( Figures 3G,H).

Dysregulation of Synapse-Related Protein Expression in shank3-Deficient Zebrafish
These behavioral abnormalities in shank3-deficient zebrafish suggest possible disruption of normal synaptic function, so we compared the expression levels of several important synaptic proteins among genotypes. Expression of the neuronal marker NeuN was reduced by 50% in the brains of adult shank3ab −/− zebrafish compared to age-matched WTs (Figures 4A,B). As SHANK3 is a major synaptic scaffolding protein enriched at the PSD of excitatory synapses (Jiang and Ehlers, 2013;Monteiro and Feng, 2017), we also examined expression of the postsynaptic marker homer1 and found an approximately 90% reduction in shank3ab −/− zebrafish relative to WTs (Figures 4C,D). It was reported that Shank3 deficiency in mice disrupts the presynaptic neurexin-neuroligin-mediated signaling pathway required for synapse targeting and development (Arons et al., 2012), so we further compared the expression levels of presynaptic proteins among genotypes, including the ubiquitous synaptic vesicle protein synaptophysin (Kwon and Chapman, 2011). Indeed, synaptophysin expression level was reduced by ∼64% in shank3ab −/− zebrafish compared to WTs (Figures 4E,F). Thus, shank3 deficiency reduced the expressions of several preand postsynaptic proteins which are likely important to protein transmitter signaling.

Improved ASD Core Symptoms and grm5 Receptor Expression by Early VPA Treatment
We then examined the efficacy of VPA to mitigate autismlike symptoms in these zebrafish models. In WT zebrafish, both shank3a and shank3b expression levels increased gradually from 3 to 7 dpf, a period of intense synaptogenesis . Exposure regimens of 3-7 and 4-8 dpf were thus judged as potentially suitable for optimal therapeutic effect. Based on preliminary observations, 4-8 dpf was chosen as the optimal exposure regimen (Supplementary Figure 1), and exposure concentrations (5, 20, and 50 µM) were then evaluated to identify the safety. We found that 5 µM completely eliminated mortality of shank3ab −/− larvae at 8 dpf and almost completely eliminated morphological dysgenesis (1.04%, 1/96) with no adverse effects on WT larvae (Supplementary Figure 1).

DISCUSSION
We described a novel shank3-deficient zebrafish, shank3ab −/− , demonstrating stable autism-like behaviors from the juvenile stage through adulthood, including social deficits and stereotyped behaviors. These deficits were generally more severe than exhibited by either shank3a or shank3b mutants. All three mutants also exhibited higher postnatal mortality and rates of morphological dysgenesis than WTs, but adults were fertile. We also found there were decreases in several pre-and postsynaptic proteins in shank3ab −/− mutants. Low-dose VPA reversed some of these autism-like behaviors, consistent with the potential efficacy of this treatment strategy for ASD patients (Hellings et al., 2005;Hollander et al., 2006;DeFilippis and Wagner, 2016). Therefore, the shank3ab −/− zebrafish line is a robust model to FIGURE 6 | Increased grm5 expression level in shank3-deficient zebrafish upon early VPA treatment. (A,B) Quantitative immunoblot blot analysis showed that the expression level of neuron protein NeuN was not significantly restored in the brain of shank3ab -/zebrafish treated with VPA relative to WT zebrafish (2 mpf). Similarly, the expressions of post-synaptic homer1 protein (C,D) and presynaptic synaptophysin protein (E,F) were not significantly increased in in the brain of shank3ab -/zebrafish treated with VPA relative to WT zebrafish (2 mpf). (G) The relative mRNA expression levels of grm1a, grm1b, and grm5a at 4.5 mpf were detected. Each group n = 3. Data are shown as mean ± SEM. *P < 0.05, **P < 0.01. explore the neurological mechanisms underlying ASD as well as potential pharmacological treatments.
Among molecular alterations, these shank3-deficient zebrafish exhibited a significant reduction in the expression levels of several synaptic proteins, including pre-and postsynaptic markers, which were consistent with previous mouse or Drosophila models. As a postsynaptic protein, change in postsynaptic homer1 protein was prominent in shank3ab −/− zebrafish, which was consistent with findings from other ASD mouse models (Tu et al., 1999;Wang et al., 2016). In addition, we also demonstrate reduced expression of the presynaptic protein synaptophysin, which was not previously detected in Shank3 deficient mouse models. This finding suggests that shank3 deficiency alters presynaptic formation and neurotransmission through direct or trans-synaptic mechanisms in zebrafish. The presynaptic functions of SHANK3 are not well characterized in comparison to postsynaptic functions. Several recent studies have suggested that SHANK3 is expressed in presynaptic terminals of rodent brain and dorsal root ganglion (DRG) neurons (Han et al., 2016). In a Drosophila Shank mutant (analogous to a SHANK3 mutation in humans), Shank was found in both axons and the presynaptic sites of neuromuscular junctions (NMJs) (Wu et al., 2017). Moreover, ultrastructural analysis of synaptic boutons at Shank mutant calyces showed disorganization of both presynaptic and postsynaptic components, and lack of synaptic clefts (Wu et al., 2017). This first demonstration of reduced synaptophysin in a vertebrate ASD model supports a presynaptic function for shank3 protein. The contribution of shank3-associated presynaptic deficits to ASD warrant further investigation. Generally, the overall morphology of the brain tissues were relatively normal in KO group as compared to WT group (Supplementary Figure 3). Moreover, the expression levels of synaptic proteins were not significantly restored between WT and shank3ab −/− fish after exposed to VPA (Figure 6). Similarly, in mouse model, neuronal morphology and density were not changed between WT and Shank3-deficient mice and also not altered by romidepsin (a highly potent class I inhibitor) treatment (Qin et al., 2018). Additionally, there were no apparent morphological changes in neurons treated with VPA compared with controls (Fujiki et al., 2013).
Postnatal low-dose VPA treatment profoundly and persistently improved social preference deficits, abnormal repetitive behaviors, and impaired thigmotaxis, suggesting activation of a compensatory mechanism under shank3 deficiency. Shank3 facilitates both synaptogenesis and the synaptic plasticity processes underlying social learning and cognition (Duffney et al., 2015). In addition, the behavioral abnormalities exhibited by Shank3-deficient animal models have been attributed to altered glutamatergic signaling (Peca et al., 2011;Wang et al., 2011;Bozdagi et al., 2013;Jiang and Ehlers, 2013), and VPA has been reported to increase synaptic transmission (Rinaldi et al., 2007;Akhtar et al., 2009).
We also found that postnatal low-dose VPA significantly reduced stereotyped swimming patterns ("figure 8" and "walling"). One possible explanation for this effect is improved transcription of grm5a, as Wang et al. (2016) demonstrated that disrupted mGluR5 scaffolding and abnormal mGluR5 signaling contribute to the excessive self-grooming and other behavioral and functional abnormalities of Shank3-defcient mice. Moreover, pharmacological enhancement of mGluR5 receptors rescued behavioral deficits, including repetitive behaviors and social deficits, in Shank3 knockout mice (Vicidomini et al., 2017). Collectively, an important inference from this study is that early VPA treatment can have long-lasting benefits on repetitive behaviors in shank3-deficient zebrafish, possibly by improving grm5a expression.
Valproic acid has anticonvulsant and mood stabilizing activities and are used to treat epilepsy and bipolar disorder. Generally, VPA is a HDAC inhibitor (Phiel et al., 2001), a GABA transaminase inhibitor, and a sodium channel blocker (Johannessen, 2000;Löscher, 2002;Owens and Nemeroff, 2003;Zanatta et al., 2019). Several studies confirmed the role of HDAC inhibitor of VPA. Fujiki et al. (2013) have reported that VPA, trichostatin A and sodium butyrate (all are HDAC inhibitors), but not valpromide, which is a structural analog of VPA having the same antiepileptic effect as VPA but lacking the HDAC inhibitor activity, have proapoptotic effects on neural progenitor cells (NPCs) of embryonic stem (ES) cell-derived glutamatergic neurons. In this study, we also have added a positive control -romidepsin, a highly potent class I HDAC inhibitor, to treat fish (Supplementary Table 2 and Figure 4). WT or shank3ab −/− larvae were exposed to blue egg water with or without 0.05 or 0.1 µM romidepsin from 4 to 8 dpf.
Valproic acid is a broad-spectrum inhibitor against class I (HDAC1, HDAC2, HDAC3, and HDAC8) HDAC (Chelladurai et al., 2020). Qin et al. (2018) reported that Shank3-deficient mice exhibited an abnormally low level of histone acetylation resulting from HDAC2 upregulation in the prefrontal cortex (PFC) and β-catenin/HDAC2 played a causal role in social deficits of Shank3-deficiency mouse model and the therapeutic effect of romidepsin. While the levels of HDAC1, HDAC3, and HDAC8 mRNA were largely unchanged. Given the absence of hdac2 gene in the zebrafish genome (Ko et al., 2019), we detected the expression levels of the rest three genes. Similarly, RT-qPCR analysis revealed no changes in mRNA expression levels of hdac1, hdac3, and hdac8 (Supplementary Figure 2C). In the further study, a VPA analog that does not have the HDAC inhibitory activity should be used as a control to detect the effects of the HDAC inhibitory activity of VPA.
This study also supports previous studies demonstrating that the early postnatal period is a critical therapeutic time window for long-lasting effects on core autistic symptoms (Dinstein et al., 2011). Hensch (2005) reported that novel interventions should be applied before irreversible neural function changes coinciding with the end of these critical periods. However, clinical studies have demonstrated that VPA exposure before the neural tube is closed (20-24 days of gestation in humans) increases the incidence of neurodevelopmental disorders including ASD (Rice and Barone, 2000;Jentink et al., 2010;Meador et al., 2013). Moreover, prenatal VPA exposure is actually used to establish rodent ASD models. Animals exposed to VPA during neural tube closure [E12.5 in rats according to Kim et al. (2011) and E10.5 in mice according to Kim et al. (2014)] showed an increased incidence of autism-like symptoms. When VPA was administered earlier than this critical time point, embryonic malformation was very likely to occur (Kim et al., 2019). In contrast, VPA administered after neural tube closure did not cause embryonic lethality or autism-like phenotypes (Kim et al., 2019). Furthermore, long-term VPA therapy had no noticeable noxious effect on cognition and learning in school children (Calandre et al., 1990). Thus, the optimal postnatal time window appears essential for VPA treatment efficacy against ASD.
Dose was the other key factor for effective VPA treatment. In animal models, fetal VPA exposure impairs cognitive outcome and increases malformation rate in a dose-dependent manner (Meador et al., 2013;Kaplan et al., 2015). Nicolini and Fahnestock (2018) reported ASD-like deficits following exposure of rat embryos to 350-600 mg/kg or of mouse embryos to 300-800 mg/kg VPA, while exposure to 25 µM from 10 to 24 hpf elicited social deficits in zebrafish (Baronio et al., 2018). VPA can also induce neuronal apoptosis (Ikonomidou et al., 1999). Conversely, low-dose VPA induced a dramatic rescue of core autistic deficits in shank3-deficient zebrafish, and this dose had no detectable effects on WT animals.
This study provides a new genetic zebrafish model of shank3 deficiency which displayed distinctly abnormal social behaviors and increased stereotyped behaviors. Importantly, the autismlike behaviors could be improved by postnatal low-dose VPA treatment. These findings may suggest a path for further research to identify medicinal development and allow for more in-depth understandings of future clinical drug research.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

ETHICS STATEMENT
The animal study was reviewed and approved by Research Ethics Board of Children's Hospital of Fudan University.

AUTHOR CONTRIBUTIONS
XX and QL conceived the study. CL and YW performed the experiments, analyzed the data, and wrote the initial draft of the manuscript. All authors contributed to the interpretation of the results, provided critical feedback, helped shape the analysis and manuscript, and approved the submitted manuscript.

FUNDING
This study was supported by grants from "Haiju" International Joint Laboratory of National Children's Medical Center of Children's Hospital of Fudan University (No. EK1125180106), Industry-University-Research High-tech Transformation Incubation Project of Fudan University (FDEKCXY08), and Shanghai Key Clinical Specialty Construction Project to XX. This study was also supported by grants from the Shanghai Sailing Program (No. 19YF1403800) and Clinical Research of Shanghai Municipal Health Commission (No. 20204Y0102) to CL.