The Problem: Autism as a Heterogeneous Condition
The Rise of Diverse Preclinical Models
Behavioral and Neuroimaging Approaches
Integrating Human Models and Multi-omics
Toward an Integrated Model Ecosystem
References and Further Reading
Autism research shows that different experimental models each capture only partial aspects of the condition, but together reveal shared biological pathways such as synaptic dysfunction and circuit-level disruption. By integrating findings across animal, cellular, and human systems, scientists are beginning to identify common mechanisms that can guide more precise, personalized interventions.
Image credit: Daisy Daisy/Shutterstock.com
Preclinical autism research has evolved into a multi-model framework of diverse experimental systems that more accurately reflect the biological complexity of this condition. Rather than representing a single disorder, ASD encompasses a collection of individually rare conditions linked by overlapping behavioral and molecular features.2 Together, these approaches allow researchers to investigate converging mechanisms across cellular, genetic, developmental, and neurobiological features of ASD.
The Problem: Autism as a Heterogeneous Condition
Autism spectrum disorder (ASD) is a highly heterogeneous condition that affects about 2 % of American children, with boys 4.3 times more likely to be diagnosed with ASD than girls. Globally, ASD is estimated to affect approximately 1–3 % of the population. Despite rising prevalence rates that reflect advancements in diagnostic capabilities, there remains a lack of sufficient tools and treatments for ASD, necessitating additional research using in vitro or in vivo models.1,2
Defining features of ASD that distinguish it from other neurodevelopmental disorders include impaired social relationships and repetitive/stereotypic behaviors. Though highly variable between individuals, social interaction deficits in ASD may manifest as poor eye contact, limited facial expression, delayed or absent speech, aggressive or disruptive responses, flat affect, difficulty understanding questions and directions, and resistance to physical touch, such as cuddling or holding.1
Other behavioral features of ASD can include hyperactivity, repetitive movements, sensitivity to light, sound, or touch, specific food preferences, and/or self-harming activities like head banging. This wide spectrum of behavioral and sensory symptoms reflects the underlying biological complexity of ASD, which arises from a diverse combination of genetic and environmental factors, rather than a single unifying cause.
As a result, no single experimental model can fully capture the breadth of ASD-related phenotypes. Indeed, each individual genetic risk factor often accounts for less than 1 % of cases, underscoring the need for multiple complementary model systems. Rather, different model systems are utilized to reproduce specific aspects of the condition, whether behavioral, molecular, or circuit-level, highlighting the need for multiple complementary approaches to study ASD.2
Molecular Diagnostic Tests in Children With Autism Spectrum Disorder
Video credit: JAMANetwork/Youtube.com
The Rise of Diverse Preclinical Models
In response to this complexity, preclinical research has transitioned from single-model approaches to a broader range of experimental systems that capture specific features of ASD.
Owing to its multifactorial origin, complex genetics, and heterogeneity in clinical phenotypes, it is difficult to faithfully model ASD.2
Diverse species are used as animal models to study ASD, among them non-human primates, domestic animals, rodents, birds, fish, and invertebrates. Non-human primates, such as macaques, share key brain regions with humans involved in social behavior, enabling more direct interspecies comparisons; however, their use raises ethical considerations and requires highly trained personnel to handle them.1
Rodents, particularly mice and rats, are widely used in preclinical research due to their anatomical, biochemical, electrophysiological, and genetic similarities. The low cost of rodents, short gestation periods, and ability to produce multiple offspring are key advantages of these animal models.
Genetic Models of ASD
To date, several point mutations and copy number variants (CNVs) have been identified as potential genetic risk factors for ASD that can explain up to 50 % of all cases. However, many of these variants are rare, including de novo mutations and inherited variants, and hundreds to over a thousand candidate risk genes have been identified across studies. Many of the genes implicated in ASD development affect biological processes ranging from synaptic function and neuronal activity to neurogenesis and chromatin remodeling.1,2
Advanced gene-editing technologies, such as homologous recombination and clustered regularly interspaced short palindromic repeats (CRISPR)/Cas9, have been used to generate knock-out or knock-in genetic animal models of ASD. For example, mutations in the SH3 and multiple ankyrin repeat domains protein 3 (SHANK3) gene contribute to both ASD and its syndromic forms, such as Phelan-McDermid syndrome.
SHANK3 knockout rodents exhibit impaired social behaviors and self-injurious repetitive grooming, linked to disruptions in synaptic transmission and corticostriatal circuitry function, which allow researchers to identify converging neurobiological mechanisms that can improve current treatment strategies. Similarly, SHANK3-knockout cynomolgus monkeys experience motor deficits, disordered sleep, repetitive tendencies, and altered social and learning capacities. In addition to SHANK3, other ASD mouse models have been developed to mimic de novo CNVs affecting chromosomal loci 1q21.1, 3q29, 7q11.23, 15q11-13, 16p11.2, and 22q11.2.3,4
Syndromic ASDs are conditions with clinically defined somatic abnormalities that are associated with psychobehavioral features of ASD. In addition to Phelan-McDermid syndrome, other syndromic ASDs with respective mouse models include Fragile X syndrome, Rett syndrome (with mutations in methyl-CpG-binding protein 2, MECP2), and tuberous sclerosis complex (TSC). Importantly, experimental data obtained from these mouse models should be interpreted carefully due to the potential impact of somatic abnormalities on ASD-related dysfunction.4
Environmental Models of ASD
External factors like maternal alcohol use and exposure to environmental toxins have been implicated in the development of ASD, in addition to the presence of gestational diabetes, maternal obesity, or infections during pregnancy that can impact neonatal brain development. These environmental factors illustrate the wide range of biological mechanisms implicated in ASD, further complicating efforts to define a single unifying model.
The maternal use of certain medications like selective serotonin reuptake inhibitors (SSRIs) and some antibiotics may also contribute to ASD, particularly if prescribed during highly sensitive periods of brain development. Subsequently, chemical agents like valproic acid, an antiepileptogenic drug historically associated with teratogenic effects linked to ASD, have been used to induce ASD-like features in rodents to study this aspect of its pathophysiology.2
Behavioral and Neuroimaging Approaches
Assessing ASD treatment efficacy and extrapolating its effects in humans using animal models primarily involves behavioral studies. In rodents, for example, sociability can be evaluated using the Three-Chamber Social Interaction (SIT) test, which compares how much time a mouse spends interacting with a social partner versus an inanimate object. Other ASD-like behaviors that can be monitored in rodents include ultrasonic vocalizations for communication, marble burying for stereotypy, and the T-maze or Morris water maze for assessing spatial memory and cognitive flexibility.1,4
Advanced imaging techniques are often performed alongside in vivo behavioral assays in ASD studies to elucidate associations between physiological, behavioral, and brain structural abnormalities. Resting-state functional magnetic resonance imaging (rs-fMRI), for example, can identify defects in functional connectivity and distinguish between different rodent models of ASD.2
Integrating an in vivo imaging setup with behavioral tasks and optogenetic and chemogenetic manipulations will extend our ability to identify causal relationships between circuit activity and behavioral outcomes.4
In vivo two-photon imaging is another powerful tool that captures large-scale patterns of neuronal activity, as well as the biochemical states of neurons and neural circuits. This approach enables longitudinal observation of synaptic structure, dendritic spine dynamics, and calcium signaling in living animals, providing direct evidence of circuit-level dysfunction in ASD models. Using this platform, researchers monitoring postnatal brain development in mice have observed circuit-level dysfunction and fluctuations in calcium dynamics associated with ASD symptoms.4
Integrating Human Models and Multi-omics
Human-based model systems have emerged as a critical complement to traditional animal models in preclinical autism research. Induced pluripotent stem cell (iPSC)-derived neurons isolated from patient samples retain the genetic profile of individuals with ASD, enabling studies of neuronal development, synaptic function, and gene expression in a human-specific context.
Brain organoids extend this approach by modeling early neurodevelopment in three-dimensional (3D) structures that recapitulate certain aspects of cortical organization. These systems can also capture patient-specific developmental trajectories and allow testing of therapeutic interventions in a personalized manner. Although these systems do not fully reproduce the complexity of the human brain, they provide unique insights into developmental processes that are difficult to study in vivo.2
Recent advances in systems biology and multi-omics technologies, including genomics, transcriptomics, and proteomics, have enabled large-scale analyses of molecular changes across these and other model systems. Rather than focusing on individual genes, these strategies identify converging biological pathways and network-level disruptions shared across genetically diverse forms of ASD. Multi-omics approaches are increasingly applied to iPSC-derived neurons and organoids, enabling researchers to link patient-specific genetic variants to downstream molecular and cellular phenotypes.
Toward an Integrated Model Ecosystem
The wide range of preclinical models available to study different aspects of ASD function serves as a complementary tool for comprehensively understanding this highly heterogeneous condition. Accordingly, current research emphasizes cross-validation of findings across multiple model systems rather than reliance on any single model.
By shifting away from a single unifying model to a multi-model framework, modern ASD research can better translate this complex biology into more targeted and effective interventions.2
References and Further Reading
- Li, Z., Zhu, Y., Gu, L., & Cheng, Y. (2021). Understanding autism spectrum disorders with animal models: applications, insights, and perspectives. Zoological Research 42(6); 800-824. DOI: 10.24272/j.issn.2095-8137-2021.251. https://pmc.ncbi.nlm.nih.gov/articles/PMC8645879/.
- Ranjan, J., & Bhattacharya, A. (2025). The Evolving Landscape of Functional Models of Autism Spectrum Disorder. Cells 14(12); 908. DOI: 10.3390/cells14120908. https://pmc.ncbi.nlm.nih.gov/articles/PMC12190894/.
- Amal, H., Barak, B., Bhat, V., et al. (2020). Shank3 mutation in a mouse model of autism leads to changes in the S-nitroso-proteome and affects key proteins involved in vesicle release and synaptic function. Molecular Psychiatry 25(8); 1835-1848. DOI: 10.1038/s41380-018-0113-6. https://pmc.ncbi.nlm.nih.gov/articles/PMC6614015/.
- Terashima, H., Minatohara, K., Maruoka, H., & Okabe, S. (2022). Imaging neural circuit pathology of autism spectrum disorders: autism-associated genes, animal models and the application of in vivo two-photon imaging. Microscopy 71(1). DOI: 10.1093/jmicro/dfab039. https://academic.oup.com/jmicro/article/71/Supplement_1/i81/6530483.
Last Updated: Apr 6, 2026