Main

Autism spectrum disorder (ASD) encompasses a range of neurodevelopmental conditions characterized by difficulties in social interaction and communication, along with restricted and repetitive behaviors1. Despite extensive research, ASD’s etiology remains elusive, involving a complex interplay of genetic and environmental factors. Animal models have been critical in exploring the pathophysiology of ASD, offering insights into the underlying mechanisms and potential therapeutic strategies2,3.

Rodent models have traditionally been at the forefront of ASD research. They have the advantages of well-characterized genetics, rapid reproduction rates and the availability of sophisticated tools for genetic manipulation and behavioral analysis. However, the limited neuroanatomical, physiological and behavioral similarity between rodents and humans has prompted the exploration of alternative models2. Nonhuman primates are advantageous as animal models for neurological disorders owing to their close genetic and physiological similarities to humans, which allow for more accurate modeling of complex brain functions and behaviors. However, their use is also associated with important ethical concerns, high costs and longer breeding cycles2,4. In this regard, pigs are increasingly considered as particularly relevant for modeling human diseases, including neurological disorders5,6. With their closer anatomical and physiological resemblance to humans, particularly with regard to brain structure and function, pigs provide a promising alternative for studying human neurological conditions7.

Valproic acid (VPA), a common anticonvulsant and mood stabilizer, has been linked to an increased risk of ASD when used during pregnancy8,9,10. Although the exact underlying mechanisms are not entirely understood, VPA exposure is known to alter gene expression through epigenetic mechanisms such as histone deacetylase inhibition11,12,13. Although previous VPA animal models of ASD, primarily in rodents and nonhuman primates, have provided important insights into ASD, these models show limitations due to species differences, ethical concerns or high costs14,15,16,17.

The present study explores the feasibility of a pig model of ASD using embryonic VPA exposure. This model allows for an in-depth investigation of the neurodevelopmental and molecular underpinnings of ASD in an organism that can better simulate human condition than rodents and that is more cost-effective than primate models. We varied VPA dosages and timing to explore neurodevelopmental impacts, using behavioral tests and neurobiological analyses to confirm the model’s relevance for ASD research.

Results

Optimizing VPA dosage

The total gestation period of minipigs is around 114 days. In developing an ASD pig model, we initially administered 200 mg/kg/day of VPA intraperitoneally to sows during mid-gestation (days 50–54, VPAM200; Fig. 1a), mirroring a method used in marmoset studies15. This developmental stage is characterized by rapid brain growth in pig embryos18, marked by the active generation of deep-layer neurons in the cerebral cortex19. Unfortunately, this high dosage proved to be lethal, resulting in the death of all seven piglets in the VPA group shortly after birth, exhibiting severe developmental anomalies including reduced body size, skeletal deformities and cranial defects (Fig. 1b). X-ray analysis showed pronounced abnormalities such as underdeveloped mandibles and spinal deformities (Fig. 1c). In contrast, the control group saw 9 of 12 piglets survive with normal developmental metrics (Fig. 1d). The VPA-treated group also had notably smaller litters and significantly lower brain weights and body measurements compared with controls (Fig. 1e–h), confirming the dosage’s teratogenic effects20,21 and its unsuitability for a viable pig ASD model.

Fig. 1: Impact of a high-dose VPA treatment during mid-gestation on neonatal outcomes.
figure 1

a, A diagram showing intraperitoneal administration of VPA at 200 mg/kg/day (VPAM200) to pregnant pigs from gestational day (GD) 50–54. b, Photos comparing neonatal mortality between control (con) and VPA-treated pigs. c, X-rays illustrating skeletal abnormalities in VPA-exposed neonates, with arrows indicating obvious deformities in the lower jaw, limbs and spine. d, A pie chart showing survival rates of neonates. e–h, Graphs comparing neurodevelopmental measures such as brain weight (e), brain length (f), head circumference (g) and overall body length (h) in dead neonates from control (n = 3) and VPA-treated (n = 7) groups. The data are presented as mean ± s.e.m. **P < 0.01, ****P < 0.0001; two-tailed Student’s t-test.

Source data

Full size image

We subsequently reduced the VPA dosage to 50 mg/kg during mid-gestation (days 50–54, VPAM50), leading to the successful birth of 13 healthy piglets without visible abnormalities, suitable for behavioral experiments and tissue analysis (Table 1). A further experiment administered a 75 mg/kg dose of VPA during early gestation (days 16–20, VPAE75), a period for neural tube closure22. This developmental period has been shown to be vulnerable to VPA exposure, which can cause ASD in both human and animal models14,23,24. This treatment resulted in 6 out of 13 piglets surviving. Despite initial losses, these piglets were healthy and underwent behavioral tests, brain histology, Golgi staining and transcriptome analysis (Table 1).

Table 1 Summary of animal information
Full size table

Behavioral assessment of VPA-treated piglets

Inspired by rodent behavioral research methods, we developed and utilized custom-designed apparatus to conduct a series of behavioral tests on VPA-treated piglets. These tests included gait analysis, the open field test, novel object exploration, the resident–intruder test, the social preference test and the water maze test. Considering that pigs are typically weaned at about one month of age, we conducted behavioral tests on piglets starting at one and a half months. At this age, piglets exhibit robust independent growth and active behavior. In minipigs, this developmental stage is roughly equivalent to a human toddler aged around 2–3 years25, which is the period when most children with ASD are typically diagnosed.

Imbalanced gait pattern of VPA-treated piglets

Children with ASD frequently encounter motor issues such as poor control and unstable posture26,27,28, often displaying asymmetrical gait patterns linked to abnormalities in the motor cortex and cerebellum29. In our study, piglets treated with 75 mg/kg VPA during early gestation (VPAE75) showed significant stride length discrepancies; the right front foot’s stride was longer than the left, and the left hind foot’s stride was longer than the right, unlike in control piglets (Fig. 2a,b and Supplementary. Video 1). A similar, though less pronounced, pattern was observed in piglets treated with 50 mg/kg VPA during mid-gestation (VPAM50; Supplementary Fig. 1a). These findings highlight motor coordination issues akin to those seen in human ASD.

Fig. 2: Assessment of motor coordination, anxiety, learning and memory in control and VPA-treated piglets.
figure 2

a, A diagram of gait pattern and stride length measurement of the rear left foot (RL), rear right foot (RR), front let foot (FL) and front right foot (FR). b, A comparison of stride lengths for the left and right front and hind feet between control (n = 8) and VPA-treated piglets (n = 6). c, A diagram of the open field test setup. d, Time spent in the central zone of the open field arena (n = 8 for control and n = 6 for VPA). e, A diagram of the novel object exploration test, featuring toys including a yellow dog, a blue dinosaur, a green ball and a red ball. f, Latency to first sniff after introducing toys, total sniffing instances and time spent sniffing each toy relative to total sniffing duration (n = 7 for control and n = 6 for VPA). g, A diagram of the water maze test, with the red dot marking the entry point and the green square indicating the spatial cue. h, Time to locate the platform and time spent in the correct quadrant of the water maze (n = 15 for control and n = 19 for VPA). Data from the VPAE75 group piglets were used for a–f and pooled data from both VPAE75 and VPAM50 groups for g and h. The data are presented as mean ± s.e.m. *P < 0.05, **P < 0.01; two-tailed Student’s t-test for b, d and f; two-way ANOVA for the left panel in h; and one-tailed Student’s t-test for the right panel in h.

Source data

Full size image

Reduced center exploration of VPA-treated piglets in the open field test

We assessed the anxiety levels of piglets using an open field test adapted from rodent models using a 2.1 m × 2.1 m arena (Fig. 2c and Supplementary Video 2). After positioning the piglets at the arena’s center, we recorded the time spent in the central area for over 15 min. Piglets from the VPAE75 group exhibited significantly less time in the central area compared with controls (Fig. 2d), indicating heightened anxiety in a novel environment, which mirrors increased anxiety levels seen in some human patients with ASD.

Increased latency for novel object exploration of VPA-treated piglets

The novel object exploration experiment assessed the anxiety and perceptual preferences of piglets when exposed to unfamiliar objects in an open field arena (Fig. 2e and Supplementary Video 3). Notably, the VPAE75 group showed abnormal behavior upon first encountering the toys; they paused and looked around for about 50 s before exploring, indicating heightened anxiety (Fig. 2f). Although initial hesitance was evident, subsequent analysis showed no significant differences in the frequency or duration of toy sniffing between the VPA-treated and control groups, suggesting unaffected intrinsic motivation to explore (Fig. 2f). Detailed assessments of sniffing frequency and the time spent on each toy revealed similar preferences for toy colors and shapes across both groups. The VPAM50 group exhibited a similar delay in exploration initiation but no significant changes in sniffing behavior or toy preferences were noted (Supplementary Fig. 1b). Despite consistent exploration drive and perceptual preferences, VPA treatment increased anxiety and fear responses to new stimuli, aligning with the behavior seen in the open field test.

Impaired spatial learning and memory in VPA-treated piglets

We next assessed the spatial learning and memory of piglets using a Morris water maze-inspired paradigm, training them to locate a hidden underwater platform with a spatial cue (Fig. 2g and Supplementary Video 4). Combining data from the VPAM50 and VPAE75 groups, we found that VPA-treated piglets took significantly longer to locate the underwater platform at the first day of training compared to controls. Over the next three days of training, VPA-treated piglets continued to be slower than controls in locating the platform, although these differences were not statistically significant (Fig. 2h). On the fifth day, when the platform was removed, VPA-treated piglets spent less time in the cue quadrant (Fig. 2h, P = 0.0397, one-tailed Student’s t-test). This suggests a potential VPA-induced reduction in spatial learning and memory abilities in piglets.

Increased aggression of VPA-treated piglets

Emotion dysregulation and aggression, common in children with autism30, were evaluated in piglets by observing reactions to an intruder (Fig. 3a and Supplementary Video 5). Upon introducing an intruder, resident piglets sniffed and occasionally displayed aggressive behaviors such as pushing or biting. The VPAE75 group showed a greater tendency to attack and sniff the intruder, although the time to initiate the first sniffing did not differ significantly from the control group (Fig. 3b,c). Compared with control piglets, VPA-treated piglets spent a shorter percentage of time sniffing the intruder’s head (Fig. 3c). However, possibly due to an insufficient number of animals, this difference was not statistically significant. In addition, the VPA-treated piglets in the VPAM50 group exhibited more attacks, although this difference was not statistically significant (Supplementary Fig. 1c). They showed a significant decrease in the percentage of time spent sniffing the head and tail, along with a significant increase in time spent sniffing the body of the intruder (Supplementary Fig. 1c). Altogether, results of the resident–intruder tests of the two batches of animals indicate increased aggression and altered social behaviors of VPA-treated piglets.

Fig. 3: Social behaviors of control and VPA-treated piglets.
figure 3

a, A diagram of the resident–intruder test, with R representing the resident and I the intruder. b,c, A quantitative analysis of the total attack duration (b), latency to initiate the first sniffing, total sniffing time and the proportion of time spent sniffing different body parts of the intruder (c). d, A diagram of the second phase of the sociability test featuring S1 and a toy simulation piglet. e, Latency for initial sniffing of S1 and the total sniffing time directed at S1 or the toy. f, A diagram of the third phase of the sociability test with S2. g, Latency for initial sniffing of S1 and total sniffing time directed at S1 or S2. All data are presented as mean ± s.e.m. *P < 0.05, **P < 0.01; two-tailed Student’s t-test or Wilcoxon test. n = 8 (control) and 6 (VPAE75).

Source data

Full size image

Altered social interaction pattern of VPA-treated piglets

We modified the mouse three-chamber sociability test to assess social communication and novelty preferences in the VPAE75 group (Fig. 3d–g and Supplementary Video 6). Initially, each piglet explored an open field arena for 15 min. In the second phase, a same-sex unfamiliar pig (stranger 1, S1) and a similar-sized toy pig were introduced at opposite corners. (Fig. 3d). Both control and VPA-treated piglets showed a clear preference for S1 over the toy, demonstrating innate social behavior (Fig. 3e). In the third phase, another unfamiliar piglet (stranger 2, S2) replaced the toy (Fig. 3f). Both groups quickly engaged with S2, displaying a preference for new social interactions. However, VPA-treated piglets initiated contact with S1 more rapidly than controls, as indicated by a shorter latency in their first sniffing of S1 (Fig. 3g). This altered social interaction pattern in VPA-treated piglets, characterized by quicker engagement with familiar partners, may indicate reduced social interest in the novel social partner, despite their overall preference for social novelty remaining intact.

Cortex malformation in VPA-treated piglets

Patients with ASD commonly show brain structure alterations, including disorganized cortical matter and increased neuronal density31. In our study, we evaluated the brain morphology of the VPAE75 group piglets (Fig. 4a). We found that the brain-to-body weight ratio was significantly lower in the VPA-treated group than in controls (Fig. 4b). Measurements of the cortex’s width, height and length were slightly smaller in the VPA group, although these differences were not statistically significant (Fig. 4c).

Fig. 4: Neuroanatomical abnormalities in VPA-treated piglets.
figure 4

a, Schematics of the brain measurement parameters. b, A quantitative analysis of brain weight. c, Dimensional statistics for the widest, highest and longest measures of the cerebral cortex. d, Comparative photographs showing cortical subregions of control and VPA-treated piglets in dorsal and side views, with regions of interest in different colors. The red arrows and arrowheads indicate differences in the sulci of the STG and MTG. e–g, Surface area measurements of the PFC (e), M1 (f) and temporal cortex (g). The data points in histograms represent individual piglets (n = 3 for both control and VPAE75 groups), presented as mean ± s.e.m. *P < 0.05, **P < 0.01; two-tailed Student’s t-test.

Source data

Full size image

We subsequently analyzed key cortical areas linked to ASD, such as the prefrontal cortex (PFC), temporal cortex and primary motor cortex (M1), to identify developmental abnormalities (Fig. 4d and Supplementary Fig. 2). Although the PFC surface in both control and VPA-treated piglets seemed normal, extending smoothly to connect with the olfactory bulb and central sulcus, the total surface area of the PFC was significantly smaller in the VPA-treated group (Fig. 4e). This reduction may explain the observed abnormalities in social interaction and increased anxiety and aggression in VPA-treated piglets.

Despite considerable variation in the shape of M1 among individuals, we found that the total surface area of the M1 was significantly smaller in VPA-treated piglets than in controls (Fig. 4d,f and Supplementary Fig. 2b). This reduction indicates substantial developmental impairment in the M1 cortex, probably contributing to the abnormal gait patterns observed in the VPA-treated piglets. The temporal cortex, essential for auditory and language processing and often linked to ASD, includes the superior temporal gyrus (STG), middle temporal gyrus (MTG) and inferior temporal gyrus (ITG) (Fig. 4d and Supplementary Fig. 2b). No significant differences in the overall surface area of the temporal cortex were noted between control and VPA-treated groups (Fig. 4g); however, notable morphological variations were present. Despite individual variability, VPA-treated piglets uniquely exhibited a posteriorly extending sulcus in the STG and lacked or showed altered sulci in the MTG that typically divide it into two subregions in controls (Fig. 4d). These morphological changes may mirror auditory, and language deficits observed in ASD, indicating potential developmental similarities in the pig model.

Abnormal neuronal morphology of VPA-treated piglets

Dendritic development deficits are prevalent in neurodevelopmental disorders like ASD, substantially impacting neurological and behavioral functions. In our study, we examined pyramidal neurons in the medial PFC of VPA-treated piglets using Golgi staining to assess morphological abnormalities. We observed that soma sizes were significantly larger in the VPAE75 group compared with controls (Fig. 5a,b). Further, Sholl analysis revealed fewer intersections in dendritic branches for both basal and apical dendrites in the VPA group, with a more pronounced reduction in basal dendrites. Most notable discrepancies were observed at 120–180 μm for basal and 140–280 μm for apical dendrites from the soma, indicating reduced dendritic complexity (Fig. 5c–e). In addition, the total dendritic lengths were considerably shorter, with basal and apical dendrites at 52.5% and 67.8% of control lengths, respectively (Fig. 5d,e).

Fig. 5: Altered neuronal morphology in the PFC of VPA-treated piglets.
figure 5

a, Golgi-stained neuronal images from the PFC. b, Soma size quantification post-Golgi staining, with data points for individual neurons presented as mean ± s.e.m. (n = 30 neurons from 3 piglets for the control group and n = 53 neurons from 3 piglets for the VPAE75 group). c, Traces of neuronal dendrites. d,e, Sholl analysis results for basal (d) and apical (e) dendrites including total dendritic length. The data are presented as mean ± s.e.m. (n = 16 neurons from 3 piglets for the control group and n = 17 neurons from 3 piglets for the VPAE75 group). r, radius; n, number of intersections. f, Images of dendritic spines on a secondary basal dendrite of a pyramidal neuron from control and VPA-treated piglets. g, Statistical analysis of dendritic spine density and types, with data points for individual dendrites presented as mean ± s.e.m. (n = 22 neurons from 3 piglets for the control group and n = 23 neurons from 3 piglets for the VPAE75 group). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; a linear mixed-effects model was used for b, d and e. A two-way ANOVA was used in the left panels in d and e, and a two-tailed Student’s t-test was used for the right panels in d and e.

Source data

Full size image

To further examine the impact of VPA treatment on dendritic spine development in pyramidal neurons, we analyzed the density of dendritic spines on secondary basal dendrites. While the overall spine density of VPA-treated piglets was not significantly different from controls, the VPA-treated piglets displayed a significant reduction in the density of mature mushroom-shaped and stubby spines, indicating impaired maturation of excitatory synapses (Fig. 5f,g). This decrease in mature spine density suggests potential disruptions in the functional connectivity of the PFC in VPA-treated piglets.

Gene expression alterations in the PFC of VPA-treated piglets

To assess differential gene expression in the PFC of 3-month-old piglets treated with VPA during early gestation, we conducted RNA sequencing on six samples (three VPA-treated and three controls). We identified 106 differentially expressed genes (DEGs): 63 upregulated and 43 downregulated (Supplementary Data 1, P < 0.05; |log2(fold change)| >1). Principal component analysis (PCA) and unsupervised hierarchical clustering confirmed significant differences between groups, with samples of the same group clustering together (Fig. 6a–c). Notable among the DEGs were genes involved in dopamine synaptic transmission, including dopa decarboxylase (DDC), dopamine receptors D2 (DRD2) and D3 (DRD3), validated further by quantitative reverse-transcription polymerase chain reaction (RT-PCR) (Fig. 6d). Gene Ontology (GO) enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses highlighted significant enrichment in ASD-related pathways such as dopaminergic synaptic transmission, glutamatergic transmission regulation and G protein-coupled receptor signaling (Fig. 6e,f and Supplementary Table 1). The DEGs also included high-risk ASD genes such as KATNAL232, ADORA2A33 and SCNA434, with six DEGs identified as high-risk candidate genes of ASD by the Simon Foundation (Supplementary Table 1).

Fig. 6: Transcriptome analysis of DEGs in the PFC of VPA-treated piglets.
figure 6

a, PCA differentiates the VPA-treated group from controls based on 106 DEGs (n = 3 for both control and VPAE75). PC, principal component. b, Unsupervised hierarchical clustering of 106 DEGs reveals distinct separation between control (Con) and VPA-treated groups (VPAE75). c, A volcano plot displaying DEGs in the PFC between the Con and VPA-treated groups. d, Quantitative RT-PCR validates altered expression of selected genes in the PFC of VPA-treated pigs. The data are based on tissue samples from six animals for each group (Con and VPAE75+M50) and are shown as mean ± s.e.m. *P < 0.05, **P < 0.01; two-tailed Student’s t-test. e, GO biological pathway analysis for DEGs. f, KEGG enrichment analysis shows significant pathways, with circle size representing gene count and color denoting adjusted P value. ECM, extracellular matrix.

Source data

Full size image

Discussion

VPA dosage sensitivity in ASD pig model versus monkeys and rodents

In developing the ASD pig model, a lower VPA dosage was used compared with previous monkey and rodent studies, yet it resulted in more severe birth defects in pigs than in monkeys14. This finding underscores VPA’s teratogenic effects and highlights species-specific variations in drug sensitivity. These differences may be due to genetic factors affecting drug metabolism, developmental stages and physiological factors such as organ size and placental structure, influencing drug delivery. In rodent models of ASD generated by prenatal VPA treatment, animals often exhibit a high percentage of tail deformities, including short tail, simple bent, short and bent tail and a double flexure tail35. However, we did not observe clear tail deformities in VPA-treated piglets of either the VPAM50 group or the VPAE75 group, suggesting a relatively milder teratogenic effect of VPA at a lower dosage or species-specific differences in VPA toxicity affecting tail development. These interspecies discrepancies emphasize the need for precise dosage calibration and thorough evaluations in translational research to ensure the safe and accurate application of findings from animal models to human conditions.

Validation of the VPA-induced ASD pig model: behavioral, molecular and neuroanatomical perspectives

A pivotal aspect of this research involved a thorough behavioral analysis of VPA-treated piglets. We observed that control piglets demonstrated strong social preferences, validating pigs as an effective model for exploring social behaviors. VPA-treated piglets, while maintaining the basic social preferences, displayed altered interaction patterns, such as reduced head sniffing and quicker transitions to familiar piglets, alongside abnormal gait, and increased anxiety—traits paralleling ASD comorbidities1,36. These behavioral outcomes enhance the credibility of our pig model in simulating ASD-like behaviors.

At the molecular level, we noted a significant alteration in the expression of several ASD high-risk genes in the PFC of VPA-treated piglets, a finding that reflects molecular etiology in human patients with ASD. The abnormal expression of these genes in our pig model validates its relevance and paves the way for further investigations into the molecular underpinnings of ASD.

We observed neuroanatomical reductions in key brain regions such as the PFC and M1, areas critical for social, linguistic and motor functions that are often impacted in ASD. Interestingly, while the overall surface area of the temporal cortex remained unchanged, there were specific morphological alterations within its sulci and gyri, indicating mild developmental anomalies. In addition, changes in neuronal soma size and dendritic architecture in the PFC were noted, which are closely related to ASD pathology. Our findings showed enlarged neuronal soma size in the PFC of VPA-treated piglets, contrasting with previous studies that generally report smaller neuronal cell volumes in postmortem ASD brain tissues37,38. This discrepancy could be due to species-specific responses to VPA, as well as differences in the brain regions and developmental stages examined. Furthermore, we observed simplified dendritic arborization and reduced maturation of dendritic spines in VPA-treated piglets. While idiopathic autism is generally associated with increased spine density39, reductions in dendritic arborization and spine density, including dendrite shrinkage in Rett syndrome, have also been reported40,41. These discrepancies underscore the heterogeneity of ASD pathogenesis and may be influenced by differences in the brain regions and developmental stages analyzed. The altered neuronal soma size and reduced dendritic complexity in the PFC of VPA-treated piglets suggest developmental deficits in cortical neuron morphogenesis and wiring, potentially contributing to the excitation–inhibition imbalance in ASD-associated neuronal circuits.

Previous studies have used pigs as animal model to investigate the pathophysiological and histological underpinnings of neurodevelopmental disabilities caused by maternal immune activation and fetal growth restriction42,43. In the present study, the convergence of behavioral, molecular and anatomical dimensions of ASD within a single animal model is particularly noteworthy. This study establishes a comprehensive model system for investigating the etiology and potential interventions for this complex neurodevelopmental disorder. Together with previous studies, our findings highlight the substantial potential of pigs as model organisms for studying neurological disorders, extending beyond traditional laboratory species. The behavioral paradigms developed in this study can also be utilized to evaluate behavioral phenotypes in future studies using pig models.

Limitations of the VPA-induced ASD pig model

The embryonic application of VPA in pigs provides a unique approach to modeling ASD but has several inherent limitations. First, VPA treatment primarily represents an environmental risk factor and may not capture the complex genetic underpinnings of human ASD. Although VPA induces epigenetic changes associated with ASD-like symptoms, it is uncertain how these changes precisely reflect the etiology of ASD in humans, where both genetic and environmental factors have roles.

Another limitation is the dosage and timing of VPA administration. Although we have attempted to model the variability in human gestational exposure to VPA, the exact correlation between the doses used in pigs and the exposure levels in humans is challenging to ascertain. This issue is compounded by interspecies differences in metabolism, VPA absorption and placental transfer, which may lead to different outcomes in pigs compared with humans.

Furthermore, while the observed behavioral and neurobiological changes in VPA-treated piglets offer valuable insights, they may not comprehensively represent the full spectrum of ASD symptoms. The behavioral tests available for pigs, although informative, may not fully capture the more subtle social and communication deficits that are commonly seen in humans with ASD.

Neurodevelopmental disorders, including ASD, often exhibit sex-specific differences, with certain impairments being more prevalent in males1,31. However, our study did not specifically investigate the impact of biological sex on the behavioral phenotypes in our pig model, which represents a limitation of the present research. Future studies would benefit from a more comprehensive investigation of this important aspect in ASD pig models.

Potential of transgenic pig models in future ASD research

Considering these limitations, transgenic pig models have great potential for advancing ASD research. Unlike the VPA-induced model, transgenic models can directly mimic specific genetic mutations associated with ASD in humans. These models enable a focused study of gene-specific pathologies and their impacts on neurodevelopment and behavior, thus offering a more detailed understanding of ASD’s genetic mechanisms. Future research employing transgenic pig models, especially those carrying mutations in high-risk genes of ASD, such as SHANK3, FMR1 or NLGN3, could shed light on the disorder’s molecular pathways and neural circuitry alterations. As transgenic techniques in pigs continue to advance44, these models will become more feasible and invaluable for testing the effectiveness of therapies aimed at these specific genetic abnormalities.

Impaired dopamine signaling in ASD

Our transcriptome analysis of the PFC in the ASD pig model has identified a notable increase in the expression of genes in the dopamine signaling pathway, supporting the emerging link between dopamine dysregulation and ASD. Dopamine, a major neurotransmitter in the brain, is critical in various functions, including reward processing, motivation, memory, attention and motor control. The disruption of dopamine pathways is increasingly being recognized as a factor contributing to the behavioral and cognitive symptoms characteristic of ASD45,46. In human ASD research, several pieces of evidence underscore the link between abnormal dopamine signaling and ASD. Neuroimaging studies have highlighted variations in dopamine transporter and receptor activities in the brains of individuals with ASD45,46,47. Moreover, genetic studies have identified variants in genes related to dopamine metabolism and transport in patients with ASD48,49,50, strengthening the connection between dopamine dysregulation and the disorder. Our findings echo these human studies and align with previous animal research, where various ASD models have shown alterations in dopamine-related genes and behaviors. Rodent models, for instance, have demonstrated that ASD-associated gene mutations can lead to changes in dopamine levels, which in turn affect behavior51.

Several lines of research suggest an overfunctioning dopaminergic system in ASD. Positron emission tomography studies have shown higher dopamine transporter binding in the orbitofrontal cortex of individuals with high-functioning autism47. Increased dopamine D2 receptor binding in the caudate and putamen of children with autism has also been reported52. Dopamine antagonists have been shown to significantly decrease behavioral symptoms, such as hyperactivity, aggression and self-injury, and improve discrimination learning in children with autism45,53,54. However, hypoactivity of the dopaminergic system has also been suggested to be a feature of ASD by findings of reduced dopamine release in the PFC and striatum, and diminished dopamine responsiveness of the striatum in ASD55,56. Moreover, a 2022 positron emission tomography study showed lower D2/3 receptor availability throughout the D2/3 receptor-rich extrastriatal regions of the dopaminergic pathways in a group of participants with ASD57. These discrepancies in findings regarding functional alterations in the dopamine pathway in ASD reflect the complexity and heterogeneity of the neurobiological etiology of the disorder. Variations in research methodologies, study populations, experimental conditions and developmental stages of the participants could contribute to the disparate findings. Furthermore, it is likely that both excessive and insufficient dopamine signaling could be detrimental to cognitive functions associated with ASD45,55. It is also possible that different subtypes of ASD may be associated with distinct patterns of dopaminergic abnormalities.

The elevated dopamine signaling observed in our ASD pig model supports a hyperactivity of the dopaminergic signaling in the PFC and provides a novel perspective on the neurochemical alterations associated with ASD. It underscores the importance of studying the dopamine pathway in the context of ASD. These results not only validate prior studies but also pave the way for potential new treatments focusing on dopamine signaling. Future comparative research involving human subjects and other animal models will be critical to verify and expand upon our findings, enhancing our overall understanding of ASD and its treatment.

Conclusion

The embryonic application of VPA in pigs represents a promising strategy for modeling ASD. Pig’s close resemblance to human physiology and neurodevelopment makes it particularly relevant for translational research, providing a potent platform for drug screening and developing ASD-specific therapeutic interventions. While the relatively small number of animals in this pilot study is a limitation, the sample size met minimal requirements for experimental research and enabled us to demonstrate the feasibility of using pigs as an ASD model. Future studies with larger sample sizes will be critical to further validate this model by enhancing the statistical power of the study.

Methods

Animals

All animal procedures were approved by the Institutional Animal Care and Usage Committee of East China Normal University (approval number P20191001). The study involved Bama miniature pigs (Sus scrofa domestica), maintained in a facility at 25 °C with a light/dark cycle and free access to food and water. Female pigs, weighing 100 ± 10 kg and aged 12 months, were monitored for menstrual cycles and paired with males around ovulation for optimal breeding. Pregnant pigs received daily intraperitoneal injections of VPA (120 mg/ml) at specified gestational stages (days 16–20 and 52–56).

Post-birth piglets were reared with their mothers until weaning at 4 weeks58, then segregated by sex. At 1.5–3 months, piglets underwent behavioral assessments, and at 3 months, they were anesthetized for brain tissue collection for molecular and histological analyses. Additional analyses included perfusion for brain dissection and X-ray photographic examinations of deceased pigs using a small animal X-ray machine.

Behavioral tests

Piglets aged 1.5–3 months were used for behavioral tests. For each piglet, a panel of behavioral tests was conducted in the following order: gait analysis, open field test, novel object exploration, resident–intruder test, sociability test and water maze test. To reduce the anxiety of pigs during behavioral tests, the experimenter fed and petted the piglets for 30 min per day for 7 days before behavioral experiments.

The gait analysis was conducted in a 7-m-long, 0.5-meter-wide corridor lined with white nonwoven fabric and flanked by obstacles. Two cameras, positioned at either end, captured the movement from the front and rear. Piglets’ feet were marked with different colored inks (red, blue, green and yellow) to distinguish footprints clearly recorded on the fabric. After navigating the corridor, the fabric was collected to measure the stride lengths of the front and hind legs (Supplementary Video 1).

The open field test took place in a 2.1 × 2.1 m square arena with 0.5-meter-high protective railings. Cameras were mounted above the arena and at two corners to record piglet activity. Placed at the arena’s center, piglets were observed for 15 min (Supplementary Video 2). The arena included a central zone of 1.05 × 1.05 m and a peripheral area, with time spent in the central zone recorded to evaluate anxiety levels in a new environment.

The novel object exploration test was conducted in the open field arena, with each of its corners holding a toy (yellow dog, blue dinosaur, green ball, red ball) secured by a 0.5-meter nylon rope, sized similar to a piglet. After a 30-minute acclimation period, piglets freely explored the arena and interacted with these toys (Supplementary Video 3). Their behavior, including sniffing latency, duration, and frequency toward each toy, was recorded for 15 minutes and analyzed to assess curiosity and anxiety of the piglet.

The water maze experiment, based on the Morris water maze, tested the spatial learning and memory of piglets (Supplementary Video 4). Using a 5-m-diameter, 1-m-deep circular pool with opaque water, piglets navigated using landmarks to locate an invisible platform. Over multiple training days, piglets entered from different positions, guided by a green cue facing the entry point. The platform’s position shifted daily, aligning beneath the green cue. Each training session lasted 60 s, repeated four times daily with a 20 min intersession interval for 5 days, recording the time for the piglet to locate the platform. After training, the platform was removed, and piglets’ inclination toward the former platform area was measured to evaluate spatial memory. Piglets were released from random points for 2 min of exploration, with their movements across four pool quadrants recorded, assessing their memory of the platform’s location.

The resident–intruder experiment, conducted in the open field arena, assessed territorial behavior and aggression in piglets (Supplementary Video 5). Following a 45 min adaptation period for the resident piglet to acclimate, a 10 min test phase introduced an unfamiliar piglet of similar size and sex. The resident piglet’s responses, including sniffing and aggressive behaviors such as pushing and biting, were recorded and analyzed. Observations focused on the time spent sniffing different body parts of the intruder and the extent of aggressive actions, providing insights into the piglets’ social and aggressive tendencies.

The sociability test in the open field arena involved three 15 min stages using tethered piglets and toys, each allowed a 0.5 m radius of movement. Initially, the test piglet explored an empty arena to acclimate. In the second stage, a stranger piglet (S1) and a pig-shaped toy were introduced at opposite corners, and the test piglet’s interactions were recorded (Supplementary Video 6). For the final stage, the toy was replaced with another unfamiliar piglet (S2), and the test continued. The time spent sniffing each stranger and the toy was analyzed to assess social and novelty preferences.

Pig perfusion

Perfusion in pigs was carried out as previously described (Supplementary Fig. 2a)59. After anesthetizing pigs with a 6 mg/kg intramuscular injection of ketamine, the onset of complete anesthesia was confirmed by the absence of consciousness and responsiveness to stimuli. Pigs were positioned abdominally on a surgical table, and their forelimbs were secured. A 7-cm-deep incision was made on both sides of each pig’s neck to expose and ligate a pulsating artery and nearby vein using cotton thread. Perfusion was initiated by inserting a retention needle connected to a peristaltic pump into the artery, and a vein was cut to prevent backflow. Sequential infusion of 1 liter of ice-cold phosphate-buffered saline, followed by 4% paraformaldehyde and 5% sucrose solutions, was administered at 100 ml/min directed toward the brain. Post-perfusion, the skull was carefully opened using a chisel, hammer, and bone forceps to expose and extract the brain tissue. The extracted brain was then fixed in 4% paraformaldehyde for 24 h at 4 °C, followed by dehydration in 20% and 30% sucrose solutions, and finally stored at −80 °C.

Brain morphological analysis

Pig brain was photographed using a digital single lens reflex camera. A sheet of white paper served as the background, enhanced by soft white lighting from both sides, effectively highlighting the sulci and gyri of the targeted brain regions. Magnification and lighting conditions were carefully maintained throughout the photography process to ensure accurate proportional comparisons of the brain region structures. Surface areas of designated brain regions were measured using ImageJ software. Different cortical regions were identified on the basis of the morphology of gyri and sulci, as described in previous literature60,61,62.

Golgi staining and analysis

For Golgi staining, the FD Rapid GolgiStain Kit (FD Neurotechnologies) was utilized. After euthanizing the piglet via heart puncture with a 30 cm sharp dissecting knife, the PFC region was extracted, cleansed with distilled water to remove the blood, and prepared for staining. Tissue samples were submerged in a 1:1 mixture of solutions A and B for 15 days in the dark, followed by a 3-day immersion in solution C, with daily changes. Post-soaking, tissues were frozen in liquid nitrogen and then sectioned at −20 °C into 100 µm slices using a Leica CM1950 cryostat and air-dried for 3 days.

The dried sections underwent two 4 min cold water washes, followed by a 15 min soak in a working solution composed of solutions D and E mixed with distilled water (1:1:2 ratio). After rinsing, the sections were dehydrated through a graded ethanol series (50%, 70%, 90% and 100%) and cleared in xylene. Neuronal dendritic morphology was imaged using a Tissue FAXS panoramic tissue scanning microscope (Tissue Gnostics) and a Leica DM2700P microscope with a 100× lens. Dendritic complexity, including branching and spine density, was quantified using Sholl analysis and measurements of total dendritic length and spine types (mushroom-shaped, stubby and thin) on the secondary dendrites of pyramidal neurons.

RNA extraction and analysis of DEGs

Pigs with similar body weights were euthanized using an injection of 6 mg/kg ketamine and 0.2 mg/kg diazepam. Cortex brain tissues were dissected, rinsed with diethylpyrocarbonate-treated water to remove RNA-degrading enzymes and then homogenized in RNase-free tubes containing TRIzol reagent (Life Technologies). After homogenization, samples were treated with DNase to remove DNA contaminants, followed by RNA extraction. The extracted RNA was then used for RNA sequencing (by Shanghai OE Biotech) and quantitative RT-PCR.

Statistical analysis

Behavioral and histological analyses were performed by at least two experimenters who were blinded to the experimental conditions whenever feasible. Statistical analyses were carried out using Prism 9.0.0.121 software (GraphPad Software). All measurements were taken from distinct samples. Data normality was assessed using Shapiro–Wilk test before statistical testing to determine whether parametric tests were feasible. The Mann–Whitney test or Wilcoxon test (paired samples) was applied for statistical analysis of nonnormally distributed data. For neuronal morphology analysis, a linear mixed-effects model, which accounts for both the number of animals and the number of samples, was used. In behavioral tests, at most one outlier per group was identified using the Grubbs’ test and excluded from the data analysis. The threshold for statistical significance was set at P < 0.05.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.