Autism spectrum disorder-like behaviors in developing zebrafish exposed to particulate matter

Abstract

Autism spectrum disorder (ASD) is a group of neurodevelopmental disorders that can impact communication and social behaviors. Evidence suggests that the causes of ASD are likely a combination of genetic and environmental factors, such as air pollution. Particulate matter (PM) is the solid and liquid portion of air pollution that can vary in size and has been associated with many health impacts, including cardiorespiratory impacts, and has more recently been found to be associated with the prevalence of ASD. However, little is known about the phenotypic presentations of this association between PM and ASD, therefore, the zebrafish (Danio rerio) model was employed to study behaviors often associated with ASD as a result of PM exposure. Zebrafish larvae were exposed for a total of 5 days to PM standard reference material (SRM1649b) and a commonly used home remedy, melatonin, beginning at 6 h post-fertilization and various behavioral assays were performed on subsequent days for a total of 13 days. Observed and quantified behaviors were compared to a positive control, valproic acid (VPA). Generally, PM exposure did not elicit behavior resembling that of VPA exposure and the interactions between PM and VPA did not induce additive or synergistic behavioral patterns, as expected. Melatonin supplementation did not ameliorate most of the observed behavioral impacts of PM or VPA exposure. These results have prompted additional questions about the phenotypic presentations of ASD as a result of PM exposure and contribute to growing knowledge about disease-environment interactions.

1. Introduction

According to the CDC, 1 in 31 children in the United States is diagnosed with autism spectrum disorder (ASD) (CDC, 2025). ASD is a group of neurodevelopmental disorders that are characterized by changes to communication and social behavior (NIMH, 2025) and impacts people from all socioeconomic, racial, and ethnic backgrounds (CDC, 2025). ASD is considered to be a product of a combination of genetic and environmental factors (NIMH, 2025), including environmental contaminants such as air pollution.

Fine particulate matter (PM_2.5) is the solid and liquid portion of air pollution that is 2.5 μm or smaller in aerodynamic diameter (US EPA, 2025). Due to its small size, PM_2.5 has the ability to travel deep into the lung and alveolar spaces, even entering systemic circulation, thus enabling the induction of system-wide impacts. PM_2.5 is associated with various human health impacts, including cardiorespiratory conditions and chronic diseases, likely as a result of the induction of oxidative stress and downstream compensatory effects (Feng et al., 2016; He et al., 2022). PM_2.5 is also a complex mixture, containing a variety of chemical components that have been associated with human health impacts, including metals, polycyclic aromatic hydrocarbons, polychlorinated biphenyls, and more (He et al., 2022; Jia et al., 2017; Lin et al., 2023).

Although much research focuses on the cardiorespiratory impacts of PM_2.5 exposure, recent evidence has highlighted other health impacts, including the association between PM_2.5 exposure and the development of ASD (Fujiwara et al., 2016; Perera et al., 2019). Epidemiological studies have reported ASD prevalence to be associated with both preand postnatal exposure to PM_2.5 (Talbott et al., 2015; Flores-Pajot et al., 2016), while mice exposed to PM_2.5 in early development exhibited behavioral deficits in communication and social interaction, accompanied by the prevalence of inflammatory markers (Li et al., 2018). Although some work has been done to elucidate linkages between PM_2.5 and ASD (Chang et al., 2018; Rodulfo-Cárdenas et al., 2023), little is known about how PM_2.5 exposure can influence ASD-like behavioral modifications in other model organisms.

The most commonly referenced potential mechanism of action for the human health impacts of PM exposure is through the generation of reactive oxygen species as a stress response to chemical constituents of PM and the resulting oxidative stress, (Feng et al., 2016; He et al., 2022) which can impact many physiological systems. The nervous system, including neurodevelopmental processes, can be particularly sensitive to oxidative stress, experiencing altered neuronal growth and development, synapse function, and neuronal network formation (Sies and Jones, 2020; Qin et al., 2024; Liu et al., 2023). More specifically, airborne particles have been found to induce neurotoxicity as a result of neuroinflammation, oxidative stress, apoptosis, and other changes to neuronal structures (Qin et al., 2024; Liu et al., 2023). Therefore, the working hypothesis was that PM induces- or contributes to-the development of ASD through reactive oxygen species generation, resulting systemic oxidative stress and impacts on the developing nervous system, and induction of the ASD phenotype.

Zebrafish (Danio rerio) are biomedical model for human health and are frequently used to determine behavioral changes related to toxicant exposure, including PM_2.5 toxicity (Roper et al., 2018; Roper et al., 2020). Their small size, predictable behavioral markers of toxicity, external development, and fast generation times are just some of the characteristics that make zebrafish the idea model organism for the study of various toxicants and disease states (Bauer et al., 2021; Bambino and Chu, 2017). In fact, zebrafish have been validated as a suitable model organism for the study of both PM_2.5 toxicity (Roper et al., 2018) and of ASD (Chen et al., 2018; Lee et al., 2018; Zimmermann et al., 2015). Induction of ASD symptoms in zebrafish by valproic acid (VPA) exposure has been successful, which presented as ASD-related behavioral modifications and representative gene expression patterns (Lee et al., 2018; Zimmermann et al., 2015).

Melatonin (MEL) is a naturally-occurring hormone that aids in sleep and circadian rhythm regulation (Mi et al., 2019). Due to its natural properties, synthetic melatonin supplementation has been previously used to treat symptoms associated with ASD, specifically sleep disorders, but its effectiveness in mitigating PM_2.5-induced ASD symptoms has not yet been validated (Schroder et al., 2021).

Given the combination of epidemiological and animal evidence suggesting that PM_2.5 can contribute to ASD pathogenesis, we aimed to i) validate zebrafish as an appropriate neurobehavioral model for ASD symptoms and ii) identify ASD-like behavioral modifications that occur as a result of PM_2.5 exposure along with the effects of melatonin intervention between treatment groups.

2. Methods

2.1. Chemicals and reagents

NIST SRM1649b Urban Dust (mean particle diameter = 24.3 μm, percent particles under 7.07 μm = 10 %) was obtained (Sigma Aldrich). SRM1649b was resuspended in DMSO, sonicated for 30 min, centrifuged at 13,000 xg for 5 min, and the supernatant was isolated to obtain the soluble fraction. The supernatant was then diluted with zebrafish embryo water to reach a concentration of 200 μg/mL in 1 % DMSO. The actual exposure concentration was 100 μg/mL due to volume dilution with co-exposures. Melatonin was obtained from (ThermoFisher) and diluted in DMSO and embryo water to a concentration of 2 μM Melatonin (MEL) in 1 % DMSO. Actual exposure concentration was 1 μM due to volume dilution with co-exposures. Valproic Acid (VPA) was obtained from Sigma Aldrich and diluted in DMSO and embryo water to a concentration of 100 μM in 1 % DMSO. The actual exposure concentration was 50 μM due to volume dilution with co-exposures. Concentrations used for exposures were determined based on previous developmental zebrafish exposures, inducing intended biological effects, but not significant mortality or morphological changes. The final PM concentration (100 μg/mL) has been shown to induce behavioral alterations (Roper et al., 2018). The Melatonin concentration used (1 μM) has verified biological activity but not adverse or lethal effects (Mi et al., 2019; Zeng et al., 2024). The VPA concentration (50 μM) was used because it has been found to induce ASD-like behaviors in developing zebrafish at the target age (Chen et al., 2018). Treatment composition and naming conventions are provided (Table 1).

Table 1.

Treatment composition and names.

Treatment	PM (100 μg/mL)	VPA (50 μM)	Melatonin (1 μM)
Control
MEL			√
VPA		√
VPA + MEL		√	√
PM	√
PM + MEL	√		√
PM + VPA	√	√
PM + VPA + MEL	√	√	√

^*Concentrations are listed as final exposure concentrations within each treatment.

2.2. Zebrafish husbandry

5D wildtype zebrafish were maintained in the zebrafish facility at the University of Mississippi (Protocol #24–010) under standard husbandry conditions, including temperatures at 28 ± 1 °C and 14:10 h light-dark cycle. Feeding occurred twice daily for adults (Gemma Micro 150) and once daily in larvae, beginning at 5 dpf (GEMMA Micro 75). Healthy breeding groups (10–12 adults per tank) were used for proximity spawning procedures.

2.3. Exposure conditions

Adult zebrafish were spawned to obtain embryos. At 6 hpf of development, shield stage was verified with visual inspection and embryos were pooled and randomly sorted into a 24-well plate for a static exposure setup (11 embryos per well). A total n = 33/treatment divided across 3 replicates was achieved. The number of required animals and observations per treatment was determined with a priori power analysis to allow for power ≥ 80 %, where α = 0.05, based on previous means and variances (G*Power Software Version 3.1). One of the following treatments were randomly assigned to each well: control, MEL, VPA, VPA + MEL, PM, PM + MEL, PM + VPA, or PM + VPA + MEL (Table 1). Chorions shed during exposure were removed and discarded daily. Conclusion of the exposure at 5 dpf was followed by a triple rinse with clean culture water, behavioral analysis, and transfer into larval culture tanks (no drip) containinĝ1 in. of fish water. Feeding with larval food (gemma 75) began at 5 dpf and continued once daily for the duration of the experiment. ~1 mL fresh adult culture water to all tanks, as needed due to evaporation and food contamination to all exposure tanks after 5 dpf. Handling between and during behavioral tests was limited to reduce stress. Exposure and testing timelines are provided (Fig. 1).

Fig. 1.

General experimental timeline schematic. Images generated using Biorender.com

2.4. Developmental analysis

Survival, general morphology, and hatching were observed and recorded for the duration of treatment and testing periods.

2.5. Behavioral analysis

Behavioral endpoints designed to detect general developmental toxicity or alterations in behaviors that are often associated with ASD phenotypes were assessed. The larval photomotor response (LPR) is frequently used to detect general developmental toxicity in 5 dpf zebrafish (Roper et al., 2018). SD in humans can impact sociality and communication. Zebrafish are social organisms (even at early life stages) and the model is well-established for behavioral testing. Therefore, the behavioral endpoints utilized, including light/dark response, shoaling, mirror response, and paired social interactions can be sensitive to change and offer insight about potential behavioral phenotypes of PM-induced ASD (Chen et al., 2018).

2.5.1. Larval photomotor response

The larval photomotor response (LPR) was recorded at 5 dpf in all surviving larvae, as described (Roper et al., 2018; Chen et al., 2018). In short, surviving zebrafish (n = 31–33 larvae/group) were individually placed in wells of a 96-well plate (each well filled with culture water), placed in the Zebrabox chamber, and subjected to 5 min of acclimation, then 4 alternating 3-min light/dark periods. Standard measurements of movement were recorded during each light and dark phase using the automatic tracking feature in the ViewPoint software (Version 3.52.3.51, ViewPoint Behavior Technology, Civrieux, France). Total movement (mm) was the total distance travelled during each phase of the test.

2.5.2. Light/dark preference

The light/dark assay was performed at 10 dpf to examine exploratory behavior, as described (Chen et al., 2018). Custom 24-well plates were created so that each well contained a clear/light half and a dark half. All surviving larvae were placed individually in wells with 2 mL of clean culture water, allowed to acclimate for 5 min and then recorded for 6 min in the Zebrabox. Videos were analyzed using the EthoVisionXT (Version 13, Noldus, Leesburg, VA, USA) software to gather the total duration (sec), total distance moved (mm), and frequency of appearance in the light areas. During analysis, larvae without movement (consistent outputs of 0 or NA values) for the duration of the test were excluded.

2.5.3. Shoaling

The shoaling assay was performed at 11 dpf, as described (Chen et al., 2018). Surviving larvae were placed in 9 cm petri dishes with 25 mL water in groups of 11 larvae (less if survival was <11). A 5-min acclimation period was followed by 6 min of video recording. Videos were analyzed using EthoVisionXT (Version 17.5, Noldus, Leesburg, VA, USA) for shoaling behavior. Measured endpoints included nearest neighbor distance, mean distance between individuals (inter-individual distance), time spent in- and out of-proximity to other individuals, body contact time, relative movement (time moving towards, away from, and no interaction), mean direction of movement, and acceleration state and time (SI Fig. 1). Larvae without movement (consistent outputs of 0 or NA) for the duration of the test were excluded.

2.5.4. Mirror attack

Custom 6-well plates were created to contain a curved mirror region that covered 25 % of the outer diameter, modified according to previously-reported specifications (Chen et al., 2018). All surviving larvae (n = 19–22/group) were individually placed in the wells containing 9 mL water, allowed to acclimate for 5 min, and then recorded for 6 min. EthoVisionXT (Version 17.5) was used to identify the mirror and non-mirror zones and to analyze the following endpoints for each larva: total distance moved, total velocity, distance to mirror, and frequency, duration, and velocity in mirror zone. Measurements of each endpoint for individual larva in the mirror zone were normalized to the value of the measured endpoint within the marked non-mirror zone (SI Fig. 2). Larvae that didn’t move for the duration of the test (consistent outputs of 0 or NA) were excluded from analysis.

2.5.5. Social contact

At 13 dpf, surviving larvae (n = 9–22 larvae/group) were placed 2/well in 6-well plates containing 9 mL of culture water. The social contact was performed, as described (Chen et al., 2018) They were allowed to acclimate for 5 min and then recorded for 6 min. EthoVisionXT was used to analyze total distance moved, mean velocity, distance between subjects, proximity of subjects, and body contact time. Unmoving larvae (consistent outputs of 0 or NA) were excluded from analysis.

2.6. Statistical analysis

Statistical analyses and data representation were conducted using R Version 4.2.2 in RStudio Version 2023.09.0 + 463 (R Core Team, 2022). The significance level (α) for all tests was set at 0.05. Kaplan-Meier survival analysis with log-rank multiple comparisons was conducted with survival data and the Kruskal-Wallis non-parametric test with Dunn’s post-hoc (Bonferroni adjusted), when applicable, was used for hatching analysis. Behavioral data, including all endpoints from the LPR, light/dark, mirror, and social assays were analyzed by analysis of variance (Three-way ANOVA with type II error). Assumptions of normality and homoscedasticity were verified visually and then transformed, as needed. If transformation still resulted in violated assumptions, Kruskal-Wallis non-parametric test was run. Model fitting was based on significant main effects or interactions identified. Tukey’s post-hoc for parametric tests and Dunn’s for non-parametric tests were conducted, when applicable. Statistical analyses were conducted using n = 3 (technical replicates), with each sample containing 11 larvae (biological replicates) to account for biological variation. Graphical interpretations were generated using R Version 4.2.2 in RStudio Version 2023.09.0 + 463 (R Core Team, 2022) and the ggplot2 package.

3. Results and discussion

3.1. Survival and hatching

Survival was significantly impacted by certain treatments (Fig. 2a, P = 0.002). Larvae exposed to VPA and PM + VPA exhibited over 40 % reduced survival compared to the control by the conclusion of the 13-day period (P ≤ 0.05). Survival in the VPA + MEL and PM + VPA + MEL groups was not significantly different from the control, despite the appearance of a significant reduction (P ≥ 0.2, Kruskal-Wallis). Survival was significantly reduced in the PM + VPA group compared to the MEL group (P = 0.012).

Fig. 2.

a) Survival and b) hatching across treatment groups expressed as percentages. Asterisks denote groups that differed significantly from the control (P ≤ 0.05). Combinations of two or more treatments are denoted by “+” in the legend.

Although hatching did not differ significantly between treatment groups, some trends were observed (Fig. 2b, P > 0.05). Larvae in the VPA group reached 100 % hatching by 2 dpf, whereas all other groups (including the control) reached 100 % by 3 dpf. Additionally, treatment with PM in any combination appeared reduce the hatch rate by up 40 % compared to their non-PM-exposed counterparts at 2 dpf. For example, VPA + PM-exposed groups had 59 % hatched, while VPA-exposed groups had 100 % hatched by 2 dpf. However, by 3 dpf, all surviving larvae had hatched in all treatment groups.

A challenge in the present study was the incidence of mortality, since even control groups often suffered less than anticipated (~80 %) survival, likely due to the subsequent handling required for behavioral testing across multiple days during development. However, the survival of some treatment groups was still found to differ from the control, indicating that survival in the control larvae wasn’t decreased so much as to not distinguish differences between groups. Although previous evidence in VPA-treated larvae points to successful subsequent handling and behavioral testing for up to 13 dpf with low mortality, even at high VPA concentrations (≤1500 μM) (Chen et al., 2018), it is possible that selecting subsets of each group to be tested each day may be more effective and reduce stress throughout the battery of behavioral assays across multiple days. Zebrafish larvae have also been shown to experience increased mortality when exposed to 100–200 μg/mL PM (Roper et al., 2018), which is consistent with the observed results. The reduced survival is a potential limiting factor for the behavioral assays conducted at the later time points, as this reduced the number of tested larvae per group, potentially skewing analysis results based on reduced replication.

Hatching was not significantly impacted by any treatment group overall, but a trend of divergence between treatment groups was observed at 2 dpf. Delayed (but not overtly reduced) hatching can be indicative of a number of factors, including reduced hatch signaling (Trikíc et al., 2011; Gajbhiye et al., 2024), general reduced motor activity (Gajbhiye et al., 2024), or presence of stressors (Gajbhiye et al., 2024; Wisenden et al., 2022) at specific time points. Hatch timing in fish is flexible by design to allow for adjustment based on factors that could influence survival, so indicators of predatory threats or chemical stressors have the potential to delay or alter hatching, depending on the context (Gajbhiye et al., 2024; Wisenden et al., 2022).

3.2. Larval photomotor response

Across all phases of the larval photomotor response (light and dark), treatment with PM, MEL, or VPA resulted in altered mean swim distances (Fig. 3, P < 0.001). Groups exposed to PM or VPA exhibited average decreases of 53 % and 30 % in swim distance compared to their unexposed counterparts (e.g., PM + MEL’s unexposed counterpart being MEL treatment), respectively. Melatonin supplementation, when combined with other exposures, elicited a 43 % decrease in swim distance, on average, compared to their unexposed counterparts. However, larvae exposed to only MEL appeared to maintain a similar activity pattern to the control. Larvae exposed to VPA (alone or in combination with other exposures) exhibited activity patterns that were more sporadic and irregular compared to the control. Groups exposed to PM alone and PM + MEL demonstrated the expected activity patterns, but the activity patterns were less defined than in the control groups (hyperactivity in the dark, hypoactivity in light). Significant interaction effects were also observed between the treatments, including interactions between PM and MEL (P = 0.01) and between VPA and MEL (P = 0.01).

Fig. 3.

Mean swim distance (mm) during the larval photomotor response (LPR). Asterisks identify groups that are significantly different from control (P ≤ 0.05, Three-way ANOVA, n = 3/treatment). Shaded areas denote dark phases.

The larval photomotor response, which subjected 5 dpf zebrafish to changing light conditions, examined the movement of larvae in each condition. Healthy larvae typically exhibit hyperactivity during dark phases and hypoactivity during light phases. Increased anxiety or decreased comfort could present as hypoactivity (overall and/or in certain phases) (Chen et al., 2018). In the present study, zebrafish exposed to PM experienced overall hypoactivity during tests, with varying, but diminished light-driven behaviors, indicating overt behavioral alterations with intact sensory processing abilities. VPA exposures elicited more erratic behavior, as evidence by the decreased light-driven behaviors and overall swim distance. PM and VPA treatments appeared to have an additive effect on swim distance, and contrary to our hypothesis that melatonin would ameliorate behavioral alterations due to its antioxidant properties, melatonin also appeared to exacerbate the impacts of both PM and VPA on the LPR. The reason for this effect of MEL + PM on the LPR is not known, but other research on melatonin has highlighted a potential for decreasing sensory neuron excitability (Zhang et al., 2018), which could have contributed to the observed decreased sensation and response to the changing light conditions.

3.3. Light/dark preference

In the light dark preference assay, various measured endpoints, including total distance moved in the light zone and light zone transition frequency, differed based on whether or not VPA was present and various individual treatments also differed from one another (SI Fig. 3ab, P ≤ 0.05). All treatment groups that included VPA in the treatment (VPA, VPA + PM, VPA + MEL, and PM + VPA + MEL) had distance moved in the light zone (mm) and light zone transition frequency values that were consistently 70 % less than those exposed to treatments without VPA (Fig. 4, P < 0.001). VPA treatment inversely impacted the time to first light zone entry (latency), where larvae exposed to treatments containing VPA experienced a 17 % increase in latency compared to those not containing VPA (SI Fig. 3 cd, P = 0.008). Total distance moved in the light zone and transition frequency were similarly impacted by individual treatment groups. Larvae exposed to VPA alone and VPA + MEL transitioned to the light zone 84 % less frequently compared to the those in the control group (Fig. 4, P ≤ 0.05), but no other groups exhibited altered transition frequency compared to the control. Larvae exposed to VPA in any combination exhibited at least a 75 % reduction in transition frequency compared to larvae in the PM + MEL group (Fig. 4, P ≤ 0.03,). No effects of PM nor MEL treatment were observed (Fig. 4, P > 0.05).

Fig. 4.

Transition frequency in the light dark preference assay of exposed larvae, displayed as a) VPA treatment and b) individual treatments. The asterisk denotes that the two groups differed significantly from one another (P ≤ 0.05, n = 3/treatment). Groups not sharing the same letter are significantly different from one another (P ≤ 0.05, n = 3/treatment). Lines extending above and below boxes are the maximum and minimum values, respectively.

The measurements taken during the light dark assay can indicate changes in exploratory behavior into a less familiar environment (light zone). The observed induction of decreased exploratory behavior (indicated by decreased transition frequency and increased latency), was expected, given that VPA exposures have previously resulted in reduced light zone crosses in zebrafish and reduced exploratory behavior in rodents (Chen et al., 2018; Nicolini and Fahnestock, 2018). The observed impact of VPA on exploratory behavior is consistent with its established role as an ASD phenotype induction agent, as ASD can be associated with increased anxiety about- and/or reduced affinity for-novel or changing environments (NIMH, 2025; Fuld, 2018). Although these endpoints differed between some individual treatments, VPA still appeared to be the driver of the behavioral impacts. For example, PM + MEL groups differed from all other groups treated with VPA, but the VPA-treated groups did not differ from one another, indicating that addition of VPA to any treatment elicited behavioral changes. This pattern is confirmed by the observed main effects of VPA. The slight patterns observed with the combination treatments prompt considerations into the hypothesized antagonistic action of MEL with PM and VPA, and the hypothesized additive/synergistic relationship between PM and VPA. Although no significant effects of PM or MEL were observed, the transition frequency of PM and MEL groups trended lower than the control while exposure to PM + MEL together elicited an increase in transition frequency, which was not expected, given the effects of each component alone. The VPA + MEL group displayed behavior closer to the expected pattern, where its mean frequency fell between the means of VPA and MEL alone. Previous studies in mice found that the reduction in exploration and increase in anxious/depressive behaviors induced by PM exposure can be ameliorated by curcumin, a neuroprotective agent (Vastegani et al., 2023), indicating that amelioration of PM toxicity by neuroprotective compounds can be observed at the behavioral level. Finally, the expected additive or synergistic relationship between VPA and PM was not necessarily demonstrated, indicating that this relationship may not exist. Overall, PM or MEL treatment failed to elicit significant changes to exploratory behavior in larval zebrafish, while VPA exposure consistently reduced it.

3.4. Shoaling

The shoaling assay was performed at 11 dpf by observing the interactions of all surviving larvae within each treatment group in a petri dish. Several endpoints indicative of shoaling and social interaction were measured. Generally, larvae exposed to VPA or MEL (with or without PM) swam closer to other individuals compared to the non-VPA and non-MEL exposed larvae, respectively. Exposure to PM + VPA induced changes to the distances between all individuals within the shoaling group significantly (P < 0.05). Additional details are provided in the supplementary information, including discussion of the potential limitations that shoaling behavior is not as consistent or pronounced at 11 dpf as at adulthood (SI Figs. 4–5).

3.5. Mirror attack

The mirror attack assay consisted of measuring various endpoints of each larva’s behavior when placed in a well with a mirror to assess potential aggression towards- or affinity for- an unfamiliar (Gerlai et al., 2000). These endpoints included distance moved, velocity, distance to mirror, frequency and duration of mirror visits, latency to first mirror visit, and more. Endpoints for individual larva were normalized to the value of the measured endpoint within the marked non-mirror zone (SI Fig. 2).

The time spent moving in the mirror zone compared to the non-mirror zone was significantly impacted by some of the treatments (Fig. 5a, SI Fig. 6, P ≤ 0.05). Generally, larvae in the control, MEL, and PM + MEL treatment groups spent less time moving in the mirror zone than in the non-mirror zone (indicated by -Δ values), while larvae in the other treatment groups spent equal or more time moving in the mirror zone than the non-mirror zone (Δ = 0 or + Δ). Multiple comparisons highlighted that these changes in time spent moving in the mirror versus non-mirror zones significantly differed between VPA + MEL (Δ = 0.31 s) and both MEL (Δ = 0.09 s) and PM + MEL (Δs = 0.12 s) treatments (P ≤ 0.05). Additionally, a significant main effect of VPA treatment was present, as groups with VPA exposure had positive delta values, while those without VPA had delta values falling at 0 (no change) or negative (decrease) (P = 0.003). However, none of the tested groups differed significantly from the control (SI Fig. 7).

Fig. 5.

Impacts of various treatments on a) time moving (s) and b) duration spent in high acceleration state (s) in the mirror zone, normalized to behavior in the non-mirror zone. Dotted line highlights the value zero, indicating no change in value from the non-mirror zone. Groups not sharing the same letter are statistically significant from one another (P ≤ 0.05, n = 3/treatment). X indicates that no data was collected for a group due to mortality and/or lack of movement. Mean larvae per test plate in each group (left to right) = 9, 10, NA, 9, 9, 7, 5, 8. Lines extending above and below boxes are the maximum and minimum values, respectively.

Time spent moving within the mirror zone is a potential indicator of attempts to move towards- or interact with- a subject in the mirror (Gerlai et al., 2000). Although no groups differed significantly from the control in time spent moving, some treatment groups still differed from one another. Within MEL-treated groups, the resulting effect (increase or decrease in time spent moving), depended on whether PM or VPA was present. Addition of VPA induced an increase in time spent moving in the mirror zone, while PM induced a decrease. Given that PM and VPA had an opposite impact on this social behavior indicator, especially in the presence of MEL, contradicts the hypothesis that PM and VPA would have similar impacts on ASD-like behaviors. Zebrafish larvae exposed to VPA exposure have previously exhibited decreased time spent in the mirror zone (Chen et al., 2018), similar to the behaviors observed in the present study in the PM-, but not VPA-exposed groups. Treatment with PM + MEL elicited a response that trends more similarly to the control and MEL treatments than VPA + MEL did, indicating that VPA could elicit a different behavioral phenotype than PM. However, PM treatment alone elicited no change between the mirror and non-mirror zones, which falls between the responses of larvae exposed to PM + MEL and the other 2 PM treatments (PM + VPA, PM + VPA + MEL).

The duration of high acceleration events can be indicative of how long larvae spent attacking or charging within the mirror zone compared to the non-mirror zone. As in other measured endpoints of the mirror attack test, duration of high acceleration events did not differ significantly in any groups compared to the control, but some groups differed from one another (Fig. 5b). Similar to the pattern observed for time spent moving, the VPA + MEL treatment group exhibited an opposite reaction (increased high acceleration event duration) compared to groups exposed to PM + MEL and PM + VPA + MEL (no change or decreased duration). This may indicate that larvae exposed to VPA, even with MEL supplementation, display more aggressive behaviors towards the mirror. Interestingly, addition of PM to the VPA + MEL treatment elicits an apparent inversion (or rescue), as the larvae exposed to PM + VPA + MEL experience the opposite effect on high acceleration event duration compared to the VPA + MEL group. However, previous VPA exposures in larval zebrafish have resulted in decreased mirror attacks (Chen et al., 2018), which contradicts the present results. As seen in the time spent moving endpoint, PM did not appear to have a similar impact as VPA on the duration of high acceleration events in the mirror zone.

3.6. Social behavior

The behavior of paired larvae from the same treatment was analyzed for various sociality-related endpoints. Due to poor survival and general motor function of all VPA-treated groups at 13 dpf, only control, PM, and melatonin-treated groups were analyzed. The cumulative time each group spent without body contact was impacted by PM and MEL treatments (Fig. 6, P ≤ 0.05). PM exposure elicited a 21 % decrease in the amount of body contact time (Fig. 6a, P = 0.05), while MEL exposure elicited a 38 % increase (Fig. 6b, P = 0.02). Additionally, the larvae exposed to PM alone displayed less no body contact time compared to those exposed to MEL alone (P = 0.03), but did not differ from the control- or PM + MEL-exposed groups (Fig. 6c, P ≥ 0.05). All other measured endpoints, including time spent moving towards and away from one another, time in proximity, distance between individuals, body contact, distance moved, and velocity were not altered in any treatment groups (SI Table 1, P > 0.05).

Fig. 6.

Cumulative time with no body contact (s) during the social assay, displayed to compare a) PM treatments, b) MEL treatments, and c) all treatments. Significantly different groups are denoted by asterisks or by groups not sharing the same letter (P ≤ 0.05, n = 3/treatment). Lines extending above and below boxes are the maximum and minimum values, respectively, while data points indicate outliers.

The social assay involved placing two individuals from the same treatment group in a single chamber to observe their interactions with one-another at 13 dpf. ASD-related social alterations could present as either extreme in any of the measured endpoints (e.g., more or less body contact, time spent in proximity, etc.). The present study found that PM elicited a reduction in no-body contact time compared to those not exposed to PM, while the opposite effect was seen in larvae exposed to MEL. Although pro-social effects of PM exposure were not expected, the decrease in no body contact time suggests a prosocial effect that could present as an increase in body contact time (the inverse measurement). However, this increase in body contact time was not observed for any of the individual treatment groups, which may be explained by the distance thresholds used for tracking and measuring body contact, proximity, and no body contact separately. Larvae in the PM and treatment groups had decreased no body contact time compared to one another, but not compared to the control, indicating potential slight alterations in social behavior. A potential link between the measured decrease in no body contact time in zebrafish larvae and behavioral presentations in humans with ASD is the social concept of personal space. People with ASD can experience challenges with social interactions, reading social cues, and emotional perception (Asada et al., 2016). Human adolescents with ASD have been observed to exhibit reduced interpersonal distances compared to their typically developing peers (Asada et al., 2016). Although the present study did not identify changes in distance between individuals, the observed reduction in no body contact time with PM exposures points to an induction of social deficits, since allowing reasonable and context-specific personal space is important for maintaining social relationships (Asada et al., 2016). An alternative explanation for the reduction in no body contact time could be due to a change in visual or sensory processing, as melatonin has been associated reduced sensitivity to pain in some human and animal studies and is also hypothesized to reduce sensitivity to other stimuli (Gagnon and Godbout, 2018). Visual or sensory impairment could interfere with identifying-, tracking-, and leaving space for- other larvae, ultimately impacting no body contact time. Melatonin is also thought to have an antioxidant role in neurodegenerative diseases (Polimeni et al., 2014). Given that combination treatments that contained melatonin did not elicit alternative behaviors compared to the control, a potential rescue of the slight behavioral impacts of PM may have occurred.

Due to the lack of survival of the VPA groups to 13 dpf, it is important to note that behaviors from the positive control groups were not recorded. However, previous rodent studies reported that VPA alters responses to external stimuli, sensory function, attention, and information processing (Nicolini and Fahnestock, 2018). Therefore, the results observed in the present study may be due to a wide range of effects involving neurological, social, and/or sensory alterations. It is recommended that VPA exposure concentrations are decreased in future approaches to allow for better sublethal behavioral analyses.

3.7. Broader discussion

3.7.1. PM chemical components

Particulate matter is a complex mixture that often contains components that individually have well-known health impacts, including carcinogenicity and development toxicity. However, evidence suggests that several components could play a role in neurodevelopmental alterations that contribute to ASD, potentially explaining some of the present results. Heavy metals, such as lead, cadmium, and mercury, which are present in SRM1649b at mass fractions of 12,864, 26.10, and 1.80 mg/kg, respectively, are well-studied and are reported as likely contributors to ASD development. In general, heavy metal exposure can lead to oxidative stress and mitochondrial injury of neurons and glia, which are predicted mechanisms of environmental influences on neurodevelopmental disorders (Ijomone et al., 2020). Epidemiological studies reported that exposure to heavy metals is associated with greater ASD prevalence and that children with ASD have elevated lead and mercury levels. Lead alone is an identified cause of ASD, given that exposure is linked to neurodevelopmental, cognitive, and communication deficits (Ijomone et al., 2020). Evidence for mercury (Hg) is less clear due to overlapping symptoms of Hg poisoning and ASD, but several studies have reported associations between blood Hg levels and ASD. Additionally, evidence for associations of Pb, Cd, and Hg with ASD are reiterated by various human clinical studies (Błażewicz and Grabrucker, 2022).

Polychlorinated Biphenyls (PCBs) are present in SRM1649b at mass fractions of 9.1–79.4 mg/kg, depending on the specific congeners. PCBs impact molecular and cellular pathways that are involved in neurodevelopment and are implicated in ASD (Panesar et al., 2020). PCBs can cause behavioral deficits and interact with genetic predispositions that increase ASD likelihood (Panesar et al., 2020). PCB exposure has been shown to interfere with synapse formation and dendritic growth (Ijomone et al., 2020), while serum PCB levels have been associated with placental DNA methylation patterns involved in child neurodevelopment (Mouat et al., 2023).

Polycyclic Aromatic Hydrocarbons (PAHs), including pyrene, chrysene, and more are in SRM1649b at mass fractions of 4.98 and 3.045 mg/kg, respectively. Exposure to PAHs has been linked to adverse neurodevelopmental outcomes, including attention deficits, altered self-regulation capacity and social competence (Perera et al., 2019). Evidence for associations PAH exposure and ASD is mixed, as some epidemiological studies have reported a positive association (Perera et al., 2019; Amen et al., 2022), while others reported no associations (Perera et al., 2019).

3.7.2. ASD-like behaviors

ASD is a disorder characterized by many different symptoms with varying degrees of severity. ASD often impacts social interactions, communication, learning, and other behaviors (NIMH, 2025). People with ASD can struggle with peer relationships and exhibit externalizing behaviors, including hyperactivity and aggression (Schroder et al., 2021). Previous studies using animal models of ASD have identified behaviors that can be characterized as ASD-like. Male rat pups with induced ASD exhibit altered communication in the form of vocalizations, altered social odor-induced behaviors, decreased social interaction time, and novel object avoidance (Li et al., 2018). Mice exposed to diesel exhaust exhibited ASD-like behaviors that presented as decreased social interactions in reciprocal interaction and social preference tests, altered communicatory vocalizations, and repetitive behaviors (Chang et al., 2018). Rats with VPA-induced ASD display decreased social interactions, exploration, and attention and information processing, but display increased repetitive/stereotypical behaviors and latency to social interactions (Nicolini and Fahnestock, 2018).

The zebrafish ASD model is induced by VPA exposure. Although the mechanism by which VPA induces ASD is not completely known, several hypotheses have withstood experimentation. These include 1) brain hyperacetylation and 2) axonal remodeling with subsequent neuronal differentiation and cerebral cortex patterning (Nicolini and Fahnestock, 2018). Other researchers reported that zebrafish with the VPA-induced ASD phenotype exhibit increased spontaneous embryonic movements, touch reactivity, and speed during the larval photomotor test (Chen et al., 2018). They also displayed social deficits, presenting as decreased crosses from one light condition to the other (light/dark assay), increased time spent in light environments, and increased nearest neighbor and interindividual distances during shoaling. Other social deficits were also present in the mirror assay, where the ASD phenotype included decreased mirror attacks and time spent in the mirror zone compared to control groups (Chen et al., 2018).

3.7.3. Impacts of PM and VPA on tested behaviors

Given the epidemiological and experimental evidence that particulate matter exposure can play a role in the development of ASD, the behavioral profiles of zebrafish exposed to PM treatment alone were expected to resemble that of zebrafish exposed to VPA alone. The present study found that PM exposure did not typically induce similar behavioral effects as VPA exposure did, as the PM groups often did not differ from the control (Fig. 7, SI Fig. 7). When behavioral effects of PM treatment were observed, they were less noticeable than the impacts of VPA. Additionally, zebrafish receiving the PM + VPA combination treatment were expected to experience even greater effects than PM or VPA alone. Impacts of the PM + VPA treatment generally met the expectation for survival, hatching, and the LPR. However, in all other behavioral assays, the PM + VPA group either did not differ from the single PM or VPA treatment or the groups were difficult to compare due to decreased movement/high mortality in VPA groups. The dissimilarity of behaviors induced by PM compared to VPA is potentially due to the complex nature of PM mixtures. PM contains various chemical constituents with differing biological activity, which can also differ from the mechanism of action of VPA on the tested ASD-like behaviors. The lack of additive or synergistic effects between PM and VPA may also highlight the difference in the biological mechanisms, where behavioral changes occurring as a result of exposure to VPA may not be altered- or worsened- by the addition of PM, or vice versa.

Fig. 7.

Heatmap of measurements taken throughout the testing period. Percent difference of each measurement from the control group is indicated by color intensity. Red colors indicate a negative change and green colors indicate a positive change from control. P-values are listed in bottom right corners of groups that differed significantly from the control. Mirror attack measurements are excluded for readability.

3.7.4. Impacts of melatonin on tested behaviors

Melatonin can reduce oxidative stress (Muhammad et al., 2019) and is a commonly-used home remedy thought to reduce obstructive behaviors associated with ASD (Schroder et al., 2021; Polimeni et al., 2014). Delayed-release melatonin treatment in children with ASD is reported to aid with insomnia, decrease externalizing behaviors (hyperactivity and aggression), and improve scores of peer relationships (Schroder et al., 2021). Adult zebrafish exposed to melatonin have previously experienced an anxiolytic effect, as evidenced by the novel tank test (Genario et al., 2020). Melatonin supplementation can also mitigate neuroinflammation and neurodegeneration induced by oxidative stress (Muhammad et al., 2019). Since the predicted mechanism of action of particulate matter-induced health impacts is reactive oxygen species generation and subsequent oxidative injury, melatonin’s antioxidant effects may contribute to the amelioration of neurological and behavioral impacts of PM exposure. Therefore, melatonin was expected to ameliorate the impacts of VPA and/or PM treatment, but this was also not observed for most endpoints. Impacts of VPA or PM on survival and hatching were sometimes ameliorated by MEL supplementation, as seen in the PM + MEL and VPA + MEL groups, but the effect was not consistent. Behavior in the light dark assay was sometimes slightly rescued by melatonin supplementation, but not significantly. Behaviors in the LPR, shoaling, mirror, and social assays did not meet this expectation. The lack of amelioration of PM and VPA-induced behavioral alterations by melatonin prompts discussion of the effectiveness of the behavioral assays due to their heavy reliance on intact sensory processing. This consideration is important because melatonin has the ability to dull the senses and even reduce sensitivity to pain (Gagnon and Godbout, 2018), which may influence social behavior simply through reduction in sensory processing. The difference between dulled sensory processing and reduced ASD-like behaviors may not be distinguishable with the assays used in the present study. Future approaches for investigating how melatonin can be more effective in mitigating PM-induced behavioral changes may include 1) utilizing an oral route of exposure, as this is a common route in humans, 2) conducting delayed melatonin supplementation after initial PM exposure, and/or 3) conducting parallel exposures with other known anxiolytics (e.g., diazepam) (Genario et al., 2020) or antioxidants to determine potential target mechanisms.

4. Conclusions

Overall, larvae exposed to PM did not exhibit behavior that resembled that of larvae exposed to the positive control (VPA), and the interactions between treatments also did not induce the expected additive or synergistic behavioral patterns. Finally, melatonin supplementation did not ameliorate most of the impacts of PM or VPA exposure. However, some of the observed main effects of VPA, PM or MEL on specific behavioral presentations highlight areas for further investigation. Future investigations may consider further reducing exposure concentrations (especially of VPA) to allow for sublethal and subsequent behavioral analysis. Most of the assays addressing sociality and aggression followed many days of depuration in clean water, which may account for the loss of some treatment-driven behavioral consistency, so future investigations into 1) the immediate and long-term behavioral effects and 2) persistence of social behaviors into adulthood may offer more insight about the lasting effects of phenotypic ASD induction by PM or VPA and amelioration by melatonin. Additionally, it would be beneficial to assess the bioavailability of PM constituents in developing zebrafish to begin to understand absorption in the zebrafish model and to identify significant contributors to the observed phenotypes.

Biographies

Shayla Victoria: Environmental Toxicologist focusing on the use of fish for the study of human and environmental health impacts of toxicant exposures. Experience working with insecticides and air pollution. PhD in Pharmaceutical Sciences with an emphasis in Environmental Toxicology from the University of Mississippi. MS in Biology from the University of Wisconsin – La Crosse.

Courtney Roper: Environmental Toxicologist focusing on air quality through community engagement, environmental chemistry, and developmental toxicology research. Postdoctoral researcher in Environmental Toxicology at Oregon State University. PhD in Molecular Toxicology from the University of Pittsburgh. MS in Biology from New York University.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.ntt.2025.107548.

Data availability

Data will be made available on request.