Expanded characterization of the valproic acid zebrafish larval model of autism spectrum disorder: Insights into dysregulation in the gut Authors: Lesha Pretorius, Johanna Adriana Coetzee, Alexio Pedro dos Santos, Carine Smith Published: 2026 ABSTRACT AbstractBackgroundRelative lack of data on mechanisms underpinning poor gastrointestinal (GI) health in autism spectrum disorder (ASD) limits patient management strategies. Although the valproic acid (VPA) model of ASD is widely used, a comprehensive profile of the model in zebrafish is still lacking, and almost no data are available on ASD‐like GI symptomology or underlying mechanisms.MethodsASD‐like craniofacial morphology, anxiety‐like behavior, GI anatomical and functional dysregulation, and dysregulation of the monoaminergic system in both the brain and the gut were simultaneously assessed in zebrafish larvae at 7 days postfertilization.ResultsVPA exposure was associated with increased intraocular distance, neuroligin 3 (NLGN3) dysregulation, decreased muscle tone, and an anxious phenotype (increased thigmotaxis and dark zone preference [light–dark preference test]). In the brain, protein expression levels of tryptophan hydroxylase 2 (TPH2) and serotonin (5‐HT), but not serotonin transporter (SERT), were downregulated, as well as tyrosine hydroxylase (TH) and dopamine transporter (DAT), whereas dopamine (DA) was upregulated. In the gut, upregulated serotonergic (increased TPH1 and 5‐HT) and downregulated dopaminergic (decreased TH, increased DAT) protein expression was linked to thinner‐walled, longer intestinal tracts, and decreased spontaneous peristaltic activity.ConclusionsDopaminergic dysregulation aligns with the ASD literature. Novel findings include an association of hyposerotonergic tone with anxiety‐like outcomes as core symptomology in ASD. Furthermore, we demonstrate compartment‐specific dopaminergic and serotonergic dysregulation in the gut, which provides support for relatively decreased gut motility in ASD. INTRODUCTION Autism spectrum disorder (ASD) is a widely studied neurodevelopmental disorder (NDD). Most research in this niche focuses primarily on neurodevelopmental aspects of ASD, rather than the broader range of associated symptomology, which severely affects the quality of life in individuals with ASD. The role of dysregulated monoaminergic neurotransmitter signaling in ASD has been linked to comorbid conditions such as anxiety, but its effects on gut function have received relatively little attention to date. 1 This limits our understanding of primary ASD‐associated dysregulation at the level of the gastrointestinal tract (gut). An accurate preclinical model of ASD can overcome these obstacles, as well as the ethical obstacles to conducting research in this vulnerable population. The valproic acid (VPA) model of ASD is a recognized simulation of the NDD and is based on the known increased risk of ASD in offspring of mothers administered VPA during gestation. 2 , 3 , 4 VPA is prescribed to treat epilepsy, migraines, and bipolar disorder. 5 It modulates γ‐aminobutyric acid (GABA) levels in the central nervous system, blocks voltage‐gated ion channels, and inhibits histone deacetylase activity. 5 The effects of in utero VPA exposure appear to be consistent across species (zebrafish, rodents, and nonhuman primates), resulting in similar patterns of abnormal neurodevelopment. 6 , 7 , 8 , 9 , 10 , 11 Notably, behavioral symptoms observed in VPA models often exhibit symptoms common to human ASD, such as altered social interactions and heightened anxiety, 2 , 12 as well as central serotonergic and dopaminergic dysregulation. 1 However, the role of dysregulation outside the brain in ASD symptomology remains to be elucidated. Furthermore, the only two (repurposed) pharmaceuticals—risperidone and aripiprazole—which are currently approved by the Food and Drug Administration (FDA) for use in ASD (for management of irritability and aggression), are not highly effective in treating the core symptoms of ASD 13 , 14 , 15 and are also linked to adverse effects on gut health. 16 , 17 , 18 , 19 This highlights the priority to elucidate peripheral manifestations of ASD, so that ASD‐specific therapeutics with a more acceptable benefit–risk profile may be developed. The aim of this study was, therefore, first, to confirm the VPA‐induced ASD‐like phenotype in terms of central neurological signaling dysregulation and anxiety outcome and, second, to simultaneously expand the characterization of this model in terms of gut health. A particular strength of the study is the thorough consideration of VPA induction dose to ensure the physiological relevance of the phenotype achieved. METHODS Induction dose determination To mimic the ASD risk associated with prenatal VPA exposure in a manner that is physiologically relevant—and thus translatable to the human condition—a zebrafish equivalent dose was calculated from an estimated human in utero exposure dose, based on some key assumptions. First, VPA freely crosses the placental barrier 20 , 21 and reaches fetal circulation at concentrations equivalent to, or higher than, concentrations in maternal circulation. 22 Second, most VPA formulations are highly bioavailable (90%–100%). 23 , 24 Third, VPA is highly plasma protein bound (~90%); that is, only approximately 10% occurs as the bioactive free fraction. 25 Lastly, VPA is largely distributed to the extracellular fluid compartment, which expands during pregnancy (40%–50%) (estimated to be 21 L, a 50% increase from the theoretical 14 L in nonpregnant individuals). 5 , 26 Based on these parameters, the zebrafish equivalent exposure dose was estimated at approximately 17.19–71.23 μmol/L (Figure 1). A potential limitation of this approach was that potential zebrafish clearance of the drug was not included as a factor in our calculation, and that the administration of VPA via immersion for 48 h may differ from repeated bolus administration in humans. This may have had some effect on relative VPA bioavailability. Nevertheless, the calculated dose range compared well with doses previously shown to induce ASD‐like phenotypes in zebrafish larvae (Table 1). A comprehensive dose–response spanning the entire calculated range (6.25, 12.5, 25, 37.5, 50, and 75 μmol/L VPA) was performed (refer to Figures S1–S5), and the lowest dose that illustrated successful induction of an ASD‐like phenotype (37.5 μmol/L) was used to expand characterization of the VPA‐induced zebrafish larval model of ASD. A full toxicological report was executed according to OECD guidelines to confirm that the chosen induction dose was indeed subtoxic (refer to Table S1 and Figure S6). TABLE 1VPA exposure regimens previously used to induce autism spectrum disorder (ASD)‐like phenotypes in larval zebrafish models.Dose (μmol/L)Exposure (hpf)Age at data collection (dpf)References0.1, 0.3, 1, 3, 10, 306–482[27]0.5, 5, 15, 30, 504–1206[28]1, 2.5, 55–24 and 5–4821–28, 90[9]1, 5, 15, 25, 48, 754–1207, 21, 42[29]5, 10, 20, 40, 80, 160, 320, 640 a , 1280 a and 2560 a 5–1206[30]5, 50, 500, 1000 a , 1500 a 8–1085, 10, 11, 12, 13[31]10, 20, 30 a , 40 a 8–1085, 13, 90–120[32]2510–242, 5–8, 30[7]25, 504–96120[10]25, 50, 1001–13 and 14–375[33]480–486, 30, 70, 120[11]754–1207/21[34]2506–1205[35]Abbreviations: dpf, days postfertilization; hpf, hours postfertilization. a Doses that induced major morphological defects and/or mortality. VPA exposure regimens previously used to induce autism spectrum disorder (ASD)‐like phenotypes in larval zebrafish models. Abbreviations: dpf, days postfertilization; hpf, hours postfertilization. Doses that induced major morphological defects and/or mortality. Dose conversion from prescribed human doses (ranging from 600 to 2500 mg/d) to fetal exposure‐equivalent zebrafish larval doses (17.19–71.23 μmol/L). This dose conversion considered the increased fluid volume of pregnant women, as well as the fact that valproic acid (VPA) is well absorbed, crosses the placental barrier freely, and is highly plasma protein bound (90%). Ethical considerations and zebrafish larviculture All experimental protocols were approved by the Stellenbosch University Research Ethics Committee for Animal Care and Use (Ref# REC:ACU 2025‐27020). Wild‐type zebrafish (Danio rerio) embryos were obtained from the Zebrafish Research Unit (Experimental Medicine Group, Department of Medicine, Stellenbosch University) and maintained in embryo medium (E3; 5 mmol/L NaCl (S9888, Sigma‐Aldrich, St Louis, United States), 0.17 mmol/L KCl (P9333, Sigma‐Aldrich, St Louis, United States), 0.33 mmol/L CaCl2·2H2O (C3881, Sigma‐Aldrich, St Louis, United States), 0.33 mmol/L MgSO4·7H2O (M1880, Sigma‐Aldrich, Bengaluru, India), 1.3 × 10−5% w/v methylene blue (KIMIX, CAS 122965‐43‐9, Cape Town, South Africa) in reverse osmosis water) at ambient conditions of 28°C and 40%–60% humidity, with 14:10‐h light–dark cycle. At the experimental endpoint (7 days postfertilization [dpf]), larvae were euthanized by ethyl 3‐aminobenzoate methane sulfonate salt (MS‐222; A5040, Sigma‐Aldrich, Wisconsin, United States) overdose (400 μg/mL in E3) for at least 10 min or until cessation of heartbeat, followed by fixation in 4% paraformaldehyde (PFA; 252549, Sigma‐Aldrich, St Louis, United States) for postmortem analyses. VPA exposure Zebrafish embryos were randomly assigned to either the control or ASD‐like group. Larvae in the ASD‐like group were exposed to 37.5 μmol/L VPA (P4543, Sigma‐Aldrich, Wuxi, China) in E3 via immersion from 4 h postfertilization (hpf) to 52 hpf (refreshed daily) and subsequently maintained in E3 under standard conditions until 7 dpf. Control larvae were not exposed to VPA and maintained in E3 for the duration of the protocol, with media refreshing at the same time points for both groups. Intraocular distance Intraocular distance (IOD) was determined as a quantitative indicator of craniofacial anomalies previously reported in the VPA model. Fixed larvae (7 dpf; n = 12–15) were positioned in 2% low‐melting‐point agarose—preshaped using a zebrafish mold stamp (Stampwell Larvae 1; Idylle, Paris, France)—to allow dorsal orientation. Larvae were visualized with an inverted microscope (ECLIPSE Ti2, Nikon, Tokyo, Japan), using NIS‐Elements D software for image acquisition. Micrographs were exported to Image J for analysis. IOD was measured as the distance spanning the fissure between the forebrain and midbrain as a landmark, with the eyes as lateral boundaries (Figure 2). Dorsal view of larval zebrafish head, depicting landmarks used to determine intraocular distance. Behavioral analysis To assess anxiety‐like behavior, the light–dark transition test (LDTT), thigmotaxis assay, and light–dark preference test (LDPT) were performed in 24‐well plates (1 larva in 1 mL E3/well). Behavior was recorded using an automated tracking system (DanioVision, Noldus, Emmerich am Rhein, Germany) and analyzed using EthoVision XT v15 (Noldus, Emmerich am Rhein, Germany). Videos were acquired at 25 frames/s, with a minimum movement threshold of 0.2 mm. LDTT The protocol was performed as previously described. 36 Briefly, the test comprised a 40‐min protocol divided into four equal 10‐min phases: acclimation phase (lights off), basal activity phase (lights off), light phase (bright white light illumination at 100% of on‐board bulb capacity), and dark phase (lights off). Lighting transitions were instantaneous. Behavioral responses were quantified as the total distance moved (TDM) during each phase. The larval reflex response (light turned on) and light–dark visual motor response (light turned off) following sudden change in lighting conditions were calculated as the change in distance moved (∆DM) in the 60 s immediately before and after each transition. Thigmotaxis This assay was performed using the LDTT lighting protocol. Each well was divided into inner and outer zones with an equal surface area using software‐defined regions of interest (Figure 3). Thigmotaxis behavior was quantified in terms of the percentage of total time spent (%TTS) and the percentage of the total distance moved (%TDM) in the outer zone for the basal 10‐min phase, using the formulae: %TDM=Distance moved in outer zonepertest phaseTotal distance moved in the wellpertest phase×100 %TTS=Cumulative duration in outer zone during test phaseTotal time of the test phase×100 Visual representation of a 24‐well plate thigmotaxis assay layout, indicating inner and outer zones. LDPT The larvae were tracked over a 15‐min period consisting of a 10‐min acclimation phase (light off) and a 5‐min light phase (light turned on, 10% of the onboard bulb capacity). Each well was divided into a light and a dark half (Figure 4). The zone preference was calculated at the %TTS and %TDM in the dark zone during the light phase, as follows: %TDM=Distance moved in dark zoneTotal distance moved in the well×100 %TTS=Cumulative duration in dark zoneTotal time of the light phase×100 Visual representation of a 24‐well plate light–dark preference test (LDPT) layout, indicating light and dark zones. Pentylenetetrazol assay To assess the susceptibility of the ASD‐like phenotype to excitotoxicity via modulation of signaling via GABA receptors, a pentylenetetrazol (PTZ; P6500, Sigma‐Aldrich, St Louis, United States) assay was performed. Zebrafish larvae (7 dpf, n = 20) were transferred to a 96‐well plate (1 larva/well) in 150 μL of E3 and subsequently supplemented with 100 μL of either E3, 10 mmol/L PTZ, or 10 mmol/L PTZ and 4 μmol/L diazepam (DZP; Valium, Roche, Gauteng, South Africa) (positive control). Behavioral tracking was performed for 30 min in perceived darkness (with infrared lighting to facilitate activity tracking), starting immediately after the administration of PTZ and DZP. Gut functional assessment Epithelial lysosomal abundance As an indicator of lysosomal abundance in the mid‐intestinal region of the zebrafish larvae, neutral red (NR; 72210, Sigma‐Aldrich, Buchs, Switzerland) vital staining was performed at the experimental endpoint (7 dpf) as previously described. 37 Briefly, larvae were exposed to 2 mg/mL NR via immersion for 5 h, washed, anesthetized using 160 μg/mL MS‐222, and imaged live at 200× magnification using light microscopy. Spontaneous peristaltic activity At the experimental endpoint (7 dpf), larvae were anesthetized and orientated for imaging as described for the NR staining. Each larva was recorded for 5 min. The time for which active contractions were observed was averaged to represent seconds active/min. Gut motility assay At 6 dpf, control and ASD‐like zebrafish larvae were randomly allocated to six‐well microplates (20 larvae/10 mL) and incubated in Nile Red (72485, Sigma‐Aldrich, Buchs, Switzerland; 1 ng/mL dissolved in E3) for 16 h. Larvae were washed thrice to remove residual dye and then incubated in E3 for 24 h before endpoint analysis. Magnesium sulfate (208094, Sigma‐Aldrich, Wisconsin, United States; 33.2 mmol/L) and atropine (A0257, Sigma‐Aldrich, St Louis, United States; 5 mmol/L) were included as positive controls to mimic accelerated and delayed gut motility, respectively. Larvae were anesthetized and orientated as described in the NR protocol before live images of gut fluorescence were acquired at 100× magnification (Nikon ECLIPSE Ti2 inverted microscope, NiS‐Elements D, version 5.30.02 software). Brightfield images of the gut were overlaid with fluorescent images to accurately demarcate the region of interest before quantification of gut fluorescent intensity (using ImageJ software). Immunohistochemistry of the brain and gut A whole‐mount immunofluorescence staining protocol was used to quantify neuroligin 3 (NLGN3) expression, an ASD‐associated molecular marker, and to comprehensively characterize the dopaminergic and serotonergic neurotransmitter systems. Fixed (7 dpf) zebrafish larvae (4% PFA overnight at 4°C) were permeabilized in ice‐cold acetone: methanol (1:1) and subsequently incubated in blocking buffer (phosphate‐buffered saline (PBS, P4417, Sigma‐Aldrich, St Louis, United States) supplemented with 20% fetal bovine serum (A5670201, ThermoFisher Scientific, New York, United States) and 5% donkey serum (C06SB, Celtic Molecular Diagnostic, Cape Town, South Africa)) for 3 h at room temperature. Primary and secondary antibody incubations were performed consecutively (both overnight at 4°C) using the following antibodies: NLGN3 (1:1000, MA5‐24253, Invitrogen, Massachusetts, United States), tryptophan hydroxylase 1 (TPH1; 1:200, NBP2‐67580, Bio‐techne, Minnesota, United States), tryptophan hydroxylase 2 (TPH2; 1:250, NB100‐74555, Novus Biologicals, Colorado, United States) serotonin (5‐HT; 1:1000, S5545, Merck, Yavne, Israel), 5‐HT transporter (SERT; 1:250, ab13030, Abcam, Cambridge, United Kingdom), tyrosine hydroxylase (TH; 1:200, MAB318, Merck, St Louis, United States), dopamine (DA; 1:250, MAA851GE21, Cloud‐Clone Corp, Wuhan, China), dopamine transporter (DAT; 1:500, PA5‐78382, Invitrogen, Taiwan, China), Alexa Fluor 488 donkey anti‐mouse (1:250, A21202, Invitrogen, Oregon, United States), Alexa Fluor 488 donkey anti‐goat (1:250, ab150129, Abcam, Cambridge, United Kingdom), Alexa Fluor 488 donkey anti‐rabbit (1:250, A21206, Invitrogen, Oregon, United States), and Alexa Fluor 594 donkey anti‐mouse (1:250, A21203, Invitrogen, Oregon, United States). The head (dorsal orientation) and gut (lateral orientation) regions of whole‐mount zebrafish larvae were imaged at 100× magnification using Nikon equipment. Image acquisition settings were optimized—as described in detail elsewhere 38 —for each parameter separately using unstained and single‐stain controls to ensure that no nonspecific or background signal was detected. For each parameter, images for all samples were acquired using (the optimized) identical settings. Images were analyzed using Image J (Fiji). Briefly, regions of interest in the brain were demarcated using the ZebrafishBrain atlas provided online (zebrafishucl.org/zebrafishbrain#about‐1) by the University College London. A standardized “demarcation shape” was applied to all images of a particular parameter to standardize the area size of the region of interest for which fluorescence was quantified across images. For the gut sections, a brightfield image was overlaid with the immunostained images to allow for standardized demarcation of the mid‐intestinal region relative to landmarks (swim bladder and cloaca). Regions of interest were standardized for size to ensure that area was not a confounder in the quantification of fluorescence. Fluorescence data were expressed as integrated density (mean gray value/area, or g/cm). Statistical analysis In terms of scientific rigor, precautions were in place. For behavioral assessments, plate‐based assessments were performed using plate layouts allowing the inclusion of all groups on each plate, and initial analyses were completed automatically by onboard activity‐tracking software to remove bias. For all other analyses, researchers were blinded to ensure an unbiased approach. Unblinding was done at the point of statistical analysis just before data interpretation. In terms of determination of the number of larvae to be used for each type of assessment, power analyses were performed using G*Power software and previous data from our group to predict the numbers required for a statistical power of 0.8. The numbers used align with those reported in the literature. Statistical analysis was conducted using GraphPad Prism version 10.5.0 (www.graphpad.com, San Diego, California, United States). Outliers were removed using the ROUT method (Q = 1%), and normality was assessed using the Shapiro–Wilk test. Behavioral data were presented as 1‐min binned data (means ± standard deviation [SD]) in line graphs for qualitative assessment. All quantitative data are presented as individual data scatter plots, using mean ± SD for normally distributed data or medians and 95% confidence intervals (CI) for nonparametric data. To avoid confusion, in datasets (per figure) where both parametric and nonparametric data are presented, all statistical analyses and data presentation followed the format for nonparametric assessment (also refer to figure legends). For parametric data, Welch's t‐test was performed, and for nonparametric data, the Mann–Whitney U‐test was performed. For the PTZ assay, a one‐way analysis of variance (ANOVA) and Fisher's least significant difference (LSD) test were performed. Significance was set at p ≤ 0.05. RESULTS Apart from the observed morphological malformations of the snout, VPA exposure resulted in significantly increased IOD and dysregulated NLGN3 expression compared to controls (Figure 5A‐F). This confirmed the successful induction of an ASD‐like phenotype at both the morphological and molecular levels. Hallmarks of the autism spectrum disorder (ASD)‐like phenotype in zebrafish larvae (7 days postfertilization [dpf]): (A–C) Intraocular distance, n = 12–15. (D–F) Brain NLGN3 expression, n = 33–37. Data are presented as means ± standard deviations (SDs). Statistical analysis: Welch's t‐test; *p < 0.05; ***p < 0.001. Micrographs were captured at 100× magnification. Scale bar = 100 μm. NLGN3, neuroligin‐3. In the PTZ assay, VPA‐exposed larvae exhibited an exaggerated sensitivity to PTZ‐induced hyperlocomotion (Figure 6A,B). This response was partially rescued in the presence of diazepam, confirming that the modulated PTZ response was GABA‐receptor dependent, which further confirms the achievement of an ASD‐like phenotype. Behavior of autism spectrum disorder (ASD)‐like versus control larvae (7 days postfertilization [dpf]) during the PTZ excitotoxicity assay. (A) Qualitative assessment of locomotion is facilitated by the presentation of movement data in 1‐min bins, with data presented as means ± standard deviations (SDs), n = 20. (B) Quantitative data are presented as total locomotion. Data are presented as medians ±95% confidence intervals (CIs). Statistical analyses: One‐way analysis of variance (ANOVA) with Fisher's least significant difference (LSD) test. *p < 0.05; ***p < 0.001; ****p < 0.0001. CTRL, control; DZP, diazepam; PTZ, pentylenetetrazol. Furthermore, behavior assessment in VPA‐exposed larvae versus controls at 7 dpf confirmed the achievement of a phenotype exhibiting limited capacity to mount appropriate behavioral responses to stimuli, as well as anxiety‐like behavior. In the LDTT (Figure 7A–F), basal (unstimulated) activity levels did not differ between groups, but ASD‐like larvae exhibited a significantly exaggerated startle reflex response to acute bright‐light exposure, a failure to freeze when exposed to bright white light, and a blunted visual‐motor and hyperlocomotion response after light‐to‐dark transition. Behavior of autism spectrum disorder (ASD)‐like versus control larvae (7 days postfertilization [dpf]) during the light–dark transition test (LDTT). Qualitative assessment of locomotion is facilitated by presentation of movement data in 1‐min bins, with data presented as mean ± standard deviations (SDs) (A). Quantitative data from the LDTT are presented as the total distance moved (TDM) during the basal phase (B), the light phase (D), and the dark phase (E). In addition, magnitudes of the startle response following sudden bright white light exposure and the light‐to‐dark visual‐motor response (L‐D VMR) are presented in (C) and (F), respectively. Data are presented as medians ±95% confidence intervals (CIs), n = 53–60 per group. Statistical analysis: Mann–Whitney U‐test; *p < 0.05; ****p < 0.0001. In the thigmotaxis assay (Figure 8A,B), under basal conditions, the ASD‐like group revealed a significant increase in outer‐zone preference, as indicated by the increase in %TDM and %TTS in the outer zone. Outer‐zone preference of autism spectrum disorder (ASD)‐like larvae (7 days postfertilization [dpf]) in the thigmotaxis assay. In (A), the TDM (%) and (B), the TTS (%) in the outer zone (OZ). Data are presented as medians ±95% confidence intervals (CIs), n = 52–60 per group. Statistical analysis: Mann–Whitney U‐test; *p < 0.05; ****p < 0.0001. TDM, total distance moved; TTS, total time spent. In the LDPT, ASD‐like larvae exhibited a significant decrease in light‐zone preference (significantly decreased %TTS and %TDM) (Figure 9A,B). Together, the observations in the thigmotaxis assay and LDPT confirm an anxiety‐like phenotype in the ASD‐like larvae. Light‐zone preference of autism spectrum disorder (ASD)‐like larvae (7 days postfertilization [dpf]) in the light–dark preference test (LDPT). In (A), the TDM (%) and (B), the TTS (%) in the light zone. Data are presented as medians ±95% confidence intervals (CIs), n = 34–36 per group. Statistical analysis: Mann–Whitney U‐test; ***p < 0.001. TDM, total distance moved; TTS, total time spent. In terms of signaling outcome (as assessed by protein expression levels) in the brain, both 5‐HT and DA synthesis enzymes (TPH2 and TH) appeared to be downregulated in ASD‐like larvae. However, DA expression was elevated, perhaps due to decreased reuptake via DAT, which exhibited decreased expression (Figure 10A‐I). In contrast, total 5‐HT expression was also decreased, likely because SERT remained unaffected by VPA exposure (Figure 11A‐I). Functional assessment using appropriate imaging techniques could confirm this preliminary interpretation. Expression of key components of the dopaminergic pathway—TH (A–C), DA (D–F), and DAT (G–I)—in the brain of autism spectrum disorder (ASD)‐like larvae (7 days postfertilization [dpf]). Data are presented as medians ±95% confidence intervals (CIs). Statistical analysis: Mann–Whitney U‐test; *p < 0.05; ****p < 0.0001. Scale bar = 100 μm. DA, dopamine; DAT, dopamine transporter; TH, tyrosine hydroxylase. Expression of key components of the serotonergic pathway—TPH2 (A–C), 5‐HT (D–F), and SERT (G–I)—in the brain of autism spectrum disorder (ASD)‐like larvae (7 days postfertilization [dpf]). Data are presented as medians ±95% confidence intervals (CIs). Statistical analysis: Mann–Whitney U‐test; *p < 0.05. Scale bar = 100 μm. 5‐HT, serotonin; SERT, serotonin transporter; TPH2, tryptophan hydroxylase 2. Turning attention to the relatively neglected topic of gut dysregulation in ASD, VPA exposure induced intestinal maladaptation such as thinner‐walled but longer intestinal tracts (Figure 12A–C). Furthermore, the significant increase in the length of NR‐positive‐stained mid‐intestinal epithelium suggests dysregulated lysosomal abundance in this group (Figure 12D,E). In terms of functional assessment of the gut, spontaneous peristaltic activity was significantly reduced in the ASD‐like group (Figure 12F), whereas the Nile Red assay also illustrated relatively lower gut motility in the ASD‐like group compared to controls (Figure 12G,H). Intestinal maladaptation in the autism spectrum disorder (ASD)‐like phenotype (7 days postfertilization [dpf]). (A–C) Epithelial lysosomal abundance, (D, E) spontaneous peristaltic activity (F), and gut motility (G–H) are presented. Micrographs were captured at 40×, 100×, and 200× magnification. Data are presented as (B, C, E, F) means ± standard deviations (SDs) and (H) median ± 95% confidence interval (CI), n = 14–16. Statistical analysis: Welch's t‐test (B, C, E, F) and Kruskal–Wallis test (G), *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. NR+, neutral red‐positive stained. In terms of neurotransmission profile in the gut, the expression levels of TH were significantly decreased in the ASD‐like larvae as observed in the brain compartment, but in contrast, DAT expression was increased, which could align with a net unchanged DA abundance in the ASD‐like gut (Figure 13A‐I). Abundance of 5‐HT‐associated parameters in the ASD‐like gut was very different from the picture in the brain: TPH1 and 5‐HT expression were elevated, whereas SERT levels again appeared unaffected by VPA exposure (Figure 14A‐I). Expression of key proteins in the dopaminergic pathway—TH (A–C), DA (D–F), and DAT (G–I)— in the gut of autism spectrum disorder (ASD)‐like larvae (7 days postfertilization [dpf]). Data are presented as medians ±95% confidence intervals (CIs). Statistical analysis: Mann–Whitney U‐test; *p < 0.05; **p < 0.01. Scale bar = 100 μm. DA, dopamine; DAT, dopamine transporter; TH, tyrosine hydroxylase. Expression of key proteins in the serotonergic pathway—TPH1 (A–C), 5‐HT (D–F), and SERT (G–I)—in the gut of autism spectrum disorder (ASD)‐like larvae (7 days postfertilization [dpf]). Data are presented as medians ±95% confidence intervals (CIs). Statistical analysis: Mann–Whitney U‐test; *p < 0.05. Scale bar = 100 μm. 5‐HT, serotonin; SERT, serotonin transporter; TPH1, tryptophan hydroxylase 1. DISCUSSION Current data contribute significantly to our understanding of core ASD pathology in the gut, as well as illustrate how the neurotransmitter profile in the gut differs from central observations. First, our meticulous approach to induction of the condition and the accuracy of our achieved model are clear from its recapitulation of core ASD hallmarks. Second, functional tests (behavioral data and gut motility) confirm that the anxiety‐like, relatively constipated phenotype may in fact form part of primary ASD pathology, which aligns with anxiety and poor gut health reported in human ASD. Third, and importantly, we present novel data demonstrating compartment‐specific dysregulation of role players in serotonergic and dopaminergic signaling pathways. First, in terms of ASD induction, current data align with results reported in humans and VPA‐induced models across species and research groups. The observed increase in IOD supports existing evidence that VPA exposure induces craniofacial abnormalities in zebrafish larvae. 31 Notably, alterations in IOD appear to be a consistent feature of the model; however, discrepancies exist in the literature regarding the direction of change—likely due to protocol variations in terms of dose and duration of VPA exposure. 31 , 39 Craniofacial alterations observed in VPA models parallel clinical findings in ASD, where macrocephaly has been widely reported in affected children. 40 , 41 , 42 This phenotype is thought to arise from excessive synaptic growth and insufficient neural pruning during early neurodevelopment. 43 This also aligns with the observed dysregulated expression of NLGN3, which is in line with literature naming nlgn3 mutation as an ASD‐associated outcome. 44 NLGN3 is involved in the formation and stabilization of both excitatory and inhibitory synapses, where it facilitates their function by regulating the localization of neurotransmitter receptors and synaptic transmission. The current observation of dysregulated NLGN3 expression thus supports an interpretation of ASD‐like neurodevelopmental deficit. Related to this, GABAergic dysregulation is a well‐established feature of human ASD 45 and has similarly been reported in VPA‐induced animal models. 46 , 47 Consistent with this, the PTZ assay data in the present study indicate increased susceptibility to acute neuronal hyperexcitability, suggesting reduced inhibitory tone and impaired GABAergic neurotransmission in the ASD‐like phenotype. This heightened seizure susceptibility aligns with evidence of disrupted excitatory–inhibitory (E/I) balance in ASD 48 , in which diminished inhibitory control permits excessive network excitation. Such inhibitory deficits may also partially explain the exaggerated startle response or sensory sensitivity—a behavioral hallmark of ASD 49 —observed in VPA‐exposed zebrafish larvae in the current study following abrupt illumination changes. Furthermore, the model reliably recapitulated anxiety‐like behavior—that is, increased thigmotaxis and decreased light‐zone preference—aligning with literature on the model. 29 , 34 The neurochemical dysregulation observed in the brain of ASD‐like larvae aligns with behavioral observations and reports in literature. For example, in a rat VPA model, decreased TH and DAT 50 and increased DA expression 51 have been linked to autistic‐like behavior. Similarly, low medial prefrontal DA was reported in children with ASD. 52 These studies were focused on primary ASD symptomology such as aggression, altered social interaction, and cognitive function and did not specifically report on anxiety. In terms of serotonergic dysregulation, current results illustrating relatively lower levels of TPH2 and 5‐HT in the brain align with a report of TPH2 mutation in human ASD 53 and increased anxiety‐like behavior in zebrafish exposed to a TPH2 inhibitor. 54 The association between a relatively decreased serotonergic signaling capacity and the anxiety‐like profile suggested by the data in the current study is supported by a recent experimental study and systematic review reporting anxiolytic‐like effects of acute 5‐HT‐releasing agents in zebrafish models of anxiety, 55 but remains to be confirmed in an ASD‐like context. In the gut, the ASD‐like phenotype in the current study is characterized by an increased 5‐HT abundance, which likely results from the observed upregulated TPH1 protein expression (assuming activity of the enzyme was not affected). Because these larvae were not free feeding and too young to have established their adult‐like gut microbiome at 7 dpf, confounding effects of dietary nutrients or microbial stimulation on enterochromaffin‐associated activation of TPH1 would be negligible, suggesting that upregulated TPH1 may be a primary ASD pathology in this model. The most common gut issues in individuals with ASD are dysmotility (constipation and diarrhea) and abdominal pain. 56 , 57 , 58 , 59 In terms of the implication for gut motility and potential gut health problems in the ASD context, current data on peristaltic contractility and gut motility align with a potential constipation‐like phenotype, as also reported in rodent VPA‐exposure models of ASD. 60 , 61 Interestingly, neuronal rather than enterochromaffin‐derived 5‐HT appears to be essential for gastrointestinal motility. 62 In our model, increased 5‐HT expression in the gut did not result in an increase in peristaltic activity. This may suggest that the 5‐HT detected with our methodology mainly reflects enterochromaffin‐derived 5‐HT (in response to the upregulated TPH1 levels) rather than neuronal 5‐HT. However, more specific analysis is required to confirm this interpretation. Furthermore, without the assessment of postsynaptic 5‐HT receptors in the gut (e.g., 5‐HT3 or 5‐HT4), a firm conclusion on the role of 5‐HT in peristaltic activity cannot be made. Nevertheless, our observation of longer but thinner guts, together with the relatively poorer peristaltic contractility of the gut and even the observed anomalous (drooping) snout in VPA‐exposed larvae, aligns with poorer basal muscle tone, which has been reported in human ASD 63 and other preclinical ASD models. 61 , 64 We acknowledge that poor gut health in ASD is a highly complex condition. However, in our model of ASD, in the absence of dietary aversions and pharmaceutical treatments such as risperidone, the ASD phenotype seems to primarily be one of decreased gut motility. This model may thus add value in this context to tease out the compounding effects of dietary choices, pharmaceutical management, and others on gut health outcomes in ASD. Interestingly, the fact that SERT levels were not affected argues against a severe inflammatory state in the gut 65 as a major contributor to 5‐HT dysregulation and aligns with the finding of increased gut NR staining, as opposed to the decreased NR staining previously reported in a larval zebrafish model of intestinal inflammation. 66 Rather, the greater abundance of lysosomes illustrated with the NR stain suggests that the VPA model also recapitulates the mitochondrial dysregulation (and resultant upregulated lysosome‐dependent mitophagy) known to form part of the ASD symptomology, 67 which was also recently described in more detail by our group using high‐resolution imaging techniques. 68 However, because increased lysosomal accumulation would also be observed in a scenario of phagocyte accumulation, such as in an inflammatory state, a purpose‐designed study is required to inform on the extent and exact nature of inflammation and/or mitochondrial dysregulation in the model. Such studies may also provide insights into the extent to which oxidative stress‐associated low‐grade inflammation—which may have gone undetected in our assessments—may potentially contribute to longer‐term health complications in ASD. In terms of dopaminergic signaling capacity, central DA dysregulation is well established as a major contributor to ASD pathology in humans. 69 However, to our knowledge, the current study is the first to report dopaminergic dysregulation in the peripheral (gut) compartment in the context of ASD. Similar to our observations in the brain of ASD‐like larvae, we also observed a downregulation of TH protein expression levels in the gut. However, unlike in the brain, DA abundance may have been stabilized by a parallel upregulation of DA recycling via increased DAT. However, the cross‐sectional design of the current study precludes firm interpretations in terms of primary pathology and potential countermechanisms at play; for example, to what extent the increased DAT expression may have been a response to the increased 5‐HT abundance, as DAT is known to also function as a 5‐HT reuptake receptor. 70 Nevertheless, in this larval model, DA does not seem to be a major factor in primary ASD gut health outcome. Considering the nature of immunofluorescent quantification employed in the current study, dynamic changes to neurotransmitter signaling or neuronal functional outcome were not directly assessed. In this context, a specific limitation of the current dataset is the lack of quantification of postsynaptic monoaminergic receptor levels or activity, which limits the interpretation of downstream functional outcomes. Assessment of differential signaling via different 5‐HT receptor subtypes may add significantly to therapeutic target identification in the ASD context, as several 5‐HT receptor subtypes (e.g. 5‐HT3 and 5‐HT4 receptors) have been therapeutically targeted to alter gut functional outcome. 71 , 72 Furthermore, given the descriptive nature of the current study, direct assessment of gut–brain signaling or interaction was not included. The significance of bidirectional communication along this axis would be best elucidated in studies with interventional design, which would add to our understanding of how dysregulation in one compartment may influence regulation and functional outcome in another. Looking forward to the utility of this more comprehensively characterized model, future studies thoroughly interrogating the therapeutic potential of 5‐HT or DA signaling manipulation (e.g., using prebiotics, pharmaceuticals, etc.) could further elucidate links between these systems, gut symptomology, and behavioral outcome. In addition, given the alignment of this model with clinical ASD, this model seems well suited for multiomics investigation, especially of compartments less accessible in humans, for example, in the context of therapeutic target identification. Lastly, given the well‐established influence of gut dysbiosis on intestinal alterations and behavioral changes, the investigation of the role of the microbiota in the ASD context using the VPA model may provide broader biological context and insight on gut‐based therapeutic strategies to improve/manage holistic ASD symptomology. In conclusion, current data first demonstrate alignment of anxiety behavior in the VPA‐exposed larval zebrafish with known neurochemical dysregulation of ASD, suggesting the potential of hyposerotonergic activity as a primary role player in ASD‐associated anxiety. Second, novel profiling of neurochemical dysregulation in the gut contributes to our understanding of the suggested mechanisms underlying poor gut health in ASD and provides a basis for studies aimed at more specifically elucidating communication between the two compartments. AUTHOR CONTRIBUTIONS Lesha Pretorius: Data curation; supervision; writing – original draft; writing – review and editing; investigation; methodology; validation; project administration; visualization. Johanna Adriana Coetzee: Methodology; data curation; formal analysis; validation; investigation; visualization; writing – original draft. Alexio Pedro dos Santos: Data curation; formal analysis; visualization; investigation; methodology. Carine Smith: Conceptualization; supervision; validation; funding acquisition; project administration; writing – review and editing; writing – original draft; resources; methodology. FUNDING INFORMATION Running expenses were funded by the South African National Research Foundation (SANRF) (CPRR SRUG200326510617) and the Harry Crossley Foundation. Lesha Pretorius (postdoctoral fellow) and Johanna Adriana Coetzee (PhD student) were financially supported by the SANRF, whereas Alexio Pedro dos Santos (PhD student) received a bursary from Stellenbosch University. CONFLICT OF INTEREST STATEMENT All authors confirm no conflict of interest. ETHICS STATEMENT All protocols were ethically cleared by the Stellenbosch University Research Ethics Committee for Animal Care and Use (Ref# REC:ACU 2025‐27020). Supporting information Figure S1. Behavior of VPA‐exposed vs. control larvae at 7 dpf, in the light–dark transition test (LDTT). (A) Qualitative assessment of changes in activity levels over time is facilitated by presentation of movement data in 1‐min bins. Binned data are presented as mean ± SEM. Quantitative data from the LDTT is presented as the total distance moved (TDM) during (B) the basal phase as an indicator of basal activity, (D) the TDM during the light phase as an indicator of a ‘freeze response’ and (E) the TDM during the dark phase as an indicator of light–dark transition‐induced sustained hyperlocomotion. In addition, (C) the reflex response following sudden bright white light exposure and (F) magnitudes of light‐to‐dark visual motor response (L‐D VMR). Data is presented as median with 95% CI, n = 53–60 per group. Statistical analysis: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Figure S2. The outer zone preference of VPA‐induced zebrafish larvae (7 dpf) during the thigmotaxis assay. In (A) the TDM (%) and (B) the TTS (%) in the outer zone (OZ) during the basal (unstimulated) phase are represented for all VPA‐treated groups. In (C), the TDM (%) and (D) the TTS (%) in the outer zone during the light and dark phases of the thigmotaxis assay were represented for the control (0 μM) and 37.5 μM VPA group only. All data are presented as median and 95% CI, n = 52–60 per group. Statistical analysis: **p < 0.01; ***p < 0.001, ****p < 0.0001. OZ, outer zone; TDM, total distance moved; TTS, total time spent. Figure S3. Comparing the zone preference of larvae (% total time spent; %TTS) at 7 dpf during the light–dark preference test (LDPT) following exposure to different VPA concentrations. The % TTS in the dark zone of the arena during the ‘dimmed’ light phase of the LDPT. Data is presented as median with 95% CI, n = 57–60 per group. Statistical analysis: *p < 0.05, **p < 0.01. % TTS, percentage of total time spent; LDPT, light dark preference test. Figure S4. Mid‐intestinal neutral red staining following VPA‐exposure (7 dpf). (A–F) Representative micrographs of the neutral red‐stained mid‐intestinal regions of larvae treated with different concentrations of VPA (B, 0 μM (control); C, 6.25 μM; D, 12.5 μM; E, 25 μM; F, 37.5 μM; G, 50 μM). Red line indicates the quantified length. Micrographs were captured at 200× magnification, scale bar = 100 μm. (G) Quantitative data are presented as mean ± SD, n = 31–38 per group. Statistical analysis: **p < 0.01; ***p < 0.001; ****p < 0.0001. NR+, neutral red positive stained. Figure S5. Intraocular distance of zebrafish larvae following VPA‐exposure (7 dpf). (A–D) Representative micrographs illustrate craniofacial changes observed with increasing VPA concentrations. Intraocular distance is indicated by the red line. Micrographs were captured at 100× magnification (Scale bar is 100 μm). (E) Quantitative data are presented as mean ± SD, n = 12–15 per group. Statistical analysis: **p < 0.01; ****p < 0.0001. Table S1. OECD 236 apical observation of acute toxicity. Figure S6. Representative images of developing zebrafish larvae exposed to E3 or 37.5 μM VPA at different time points.