Autism spectrum disorder (ASD) is a group of common, complex, and highly heterogeneous neurodevelopmental disorders characterized by the core symptoms including deficits in social interaction and communication and repetitive and stereotyped behaviors, interests, or activities (1). The worldwide prevalence of ASD is approximately 1%, while this estimate is much higher in high-income countries, such as the United States (2). Together with these core symptoms, individuals with ASD are usually accompanied by other conditions, including epilepsy, intellectual disability (ID), sleeping disorders, anxiety, motor-control difficulties, attention deficit hyperactivity disorders, and gastrointestinal disorders (3, 4).

Currently, the exact etiologies of ASD are still largely unknown, and no hypothesis can completely explain the etiology and pathology of ASD. However, it is generally accepted that ASD is caused by genetic and environmental factors or their interaction (5). An increasing number of studies from patients and animal models of ASD suggested that the abnormal synaptic structure and function changes are closely related to the onset of ASD (6–8). Therefore, ASD is often conceptualized as a synaptic disorder or called “synaptopathy” (9, 10), indicating that the etiology of ASD, at least in part, is attributed to synaptic abnormalities, and targeting ameliorating the dysregulation of synaptic structure and function may be a promising method to promote ASD clinical rehabilitation.

However, despite the high prevalence of ASD, no effective pharmacological treatments to cure it have been proposed (11). Clinically, the drugs are used mainly to improve co-occurring behaviors or problems but not ASD (1). Over the decades, numerous evidence has been demonstrated that different exercise interventions positively influence physical functions for people with ASD, such as improved motor skill performance (12) and reduced body mass index (13). Apart from the physical benefits, exercise intervention has been shown to improve social skills (14) and cognitive functions (15), as well as reduce maladaptive behaviors (16) and stereotypic behaviors (17). Meanwhile, improvements in academic engagement (18), sleep quality (19), and emotional problems (20) among children with ASD have also been reported. These findings suggested that exercise intervention, as a non-pharmacological treatment, is effective for ASD symptoms, particularly if these interventions are introduced early in life. However, research investigating the underlying possible molecular and cellular biological mechanisms is still insufficient, and further research is necessary.

Hence, in this review, we summarized the characteristics of synaptic structural alternations in the context of ASD and discussed their possible roles in the pathogenesis of ASD. Then, we explored the molecular and cellular mechanisms behind these abnormal changes from the perspective of ASD-associated genetic mutation, signal pathway, and environmental factors. In addition, we reviewed the effects of an exercise intervention on individuals with ASD. We listed the possible neurobiological mechanisms behind it to gain a better understanding of the synaptic impairments underlying ASD and the mechanisms of improvement using exercise intervention.

Much remains to be understood about the characteristics of synaptic structural alterations in ASD brains. The pathologies of several neuropsychiatric disorders, such as ASD, ID, Alzheimer's disease (AD), and schizophrenia, present with alterations in synapses and the dendritic spine (spine) (8). In ASD, the first report concerning this topic was proposed by Williams et al. (21). They found an apparent reduction in the spine density on the dendrites of some pyramidal neurons in adolescents and one of the adult patients. However, the morphology was normal (21). Subsequently, other studies of human brains indicated that the hippocampus and prefrontal cortex presented with decreased dendritic branching and spine density (22, 23). However, different results exist, an increased mean spine density was found in the cortical pyramidal neurons of ASD subjects compared to those age-matched subjects (24). Tang et al. (25) found that layer V pyramidal neurons in the temporal lobe of ASD individuals showed a higher dendritic spine density than the brain samples in control subjects. However, reduced spine elimination from the age of children to adolescence indicates synaptic pruning deficits. Notably, a study reported that the high density of spines was most commonly found in ASD individuals with more severe cognitive deficits (8). Thus, future studies of human neuropathologies should strive to understand the degree of correlation between the severity of cognitive deficits and spine dysmorphogenesis. The variations in findings across studies may be attributed to some confounding factors, such as the postmortem interval, symptomatic heterogeneity, and patient medical history (8). Together, these findings from human brains revealed that the abnormal changes in spine density are closely related to the pathologies of ASD. Thus, it is possible that aberrant alterations in the spine, the increased or decreased density, contribute to the disruptions of specific neural circuits, which, in turn, may underlie the socio-cognitive impairments of this disorder. The brain samples are typically collected from deceased adults, an endpoint of the disease process. Therefore, when abnormal synapse and spine changes occurring in ASD remain unknown, thereby emphasizing the need to develop better diagnostics and provide timely interventions.

To explore the specific disease-causing factors on neuropathology, synaptic function, and behavioral outcomes more thoroughly, ASD animal models provide valuable tools. Further evidence of spine pathology from ASD animal models underscores the importance of these structural perturbations in the pathogenesis of ASD and comorbid disorders. The detailed data have been described by Martínez-Cerdeño (26) and Forrest et al. (27). Overall, structural alterations of the spine are a common feature of ASD. Specifically, too much or too little spine density in ASD affects synapse and circuit-level connectivity functional changes.

The phenotypes of decreased and increased density of the spine in ASD individuals and different animal models have been widely described, but the underlying molecular mechanisms of these changes in spine density in ASD remain unclear and need to be explored further. Given the complexity and heterogeneity of ASD etiology, currently, a united molecular biology theory illustrates that all ASD cases have been lacking. In the present review, we will discuss the possible molecular mechanisms in certain ASD types on spine density changes from the perspective of brain-derived neurotrophic factor/tropomyosin receptor kinase B (BDNF/TrkB), mechanistic target of the rapamycin (mTOR)-dependent protein synthesis signaling pathway, and microglia-dependent synaptic pruning (Figure 1).

It is widely accepted that treating patients with ASD should be aimed at improving the quality of life and reducing family suffering (46). At present, although there are no known medications that can effectively treat the core symptoms of ASD (1), especially social and communication deficits, some interventions can help ASD individuals to promote their quality of daily life (46). As a non-pharmacological intervention, exercise intervention has many beneficial roles in improving related symptoms in individuals with ASD. It has also been employed as an adjuvant therapy measure to treat ASD children within usual care treatments, and exercise intervention can be considered an evidence-based practice for school-aged children with ASD (47). Specifically, Yang et al. (14) found that a 12-week mini-basketball training program improved the social communication and the executive control network in preschool children with ASD. These effects may be related to the enhanced functional connectivity between the right cerebellum and the left inferior frontal gyrus in the experimental group. Remarkably, among various exercise intervention programs (e.g., swimming, jogging, yoga, and basketball), the team sports such as basketball intervention are considered a promising strategy for improving social deficits compared to individual sport (e.g., jogging), because it requires individuals to make decisions and communicate with others while playing such team sports (48–50). Regarding stereotypical behaviors, aerobic exercise has been shown to reduce this phenotype (51). One study found that low- to moderate-intensity exercise produces significant and large reductions in these behaviors (17). These studies establish that exercise intervention can positively impact the core symptoms of ASD.

Exercise intervention can not only positively impact the primary symptoms of ASD but also contribute to the improvement of ASD comorbidities (Figure 2). Research suggested that reduced participation in physical activity and a higher incidence of obesity among ASD children than peers may be related to their symptoms (52, 53). Conversely, reduced participation may limit the opportunities to communicate with peers, leading to further developing obesity and ASD symptoms (11). Therefore, it is essential to encourage ASD children to engage in physical activity. Moreover, sleep difficulties in children with ASD are widespread (54). Compared to ASD children with lower physical activity levels, those ASD children with higher physical activity levels display fewer sleep problems and better overall sleep quality, suggesting that engaging in physical activity contributes to improving sleep disorders (55). Motor skill impairment is a limitation to participation in physical activity, Rafie et al. (56) found that different games (e.g., ball games, fun games, and orienteering games) all improve the perceptual motor skills of ASD adolescents. Interestingly, exercise intervention (e.g., jogging) can also enhance the academic performance of ASD children, which reflects the improvement of attention (11).

Similarly, the positive effects are also observed in ASD animal models. In the VPA-induced animal model of ASD, several reports indicated that treadmill exercise is associated with improving aggressive behavior, spatial learning memory, and motor dysfunction (57–59). In another ASD animal model, early swimming intervention can lessen the core symptoms of ASD rats caused by SHANK3 knockout (60). Andoh et al. (45) found that voluntary wheel-running exercise reverses the ASD-like behaviors in offspring after MIA.

Altogether, appropriate exercise intervention can improve ASD-associated symptoms in ASD individuals and animal models and enhance the social adaptability of ASD individuals. However, given the complexity of the etiology and heterogenicity of phenotype in ASD, an acceptable standardized exercise intervention program has not yet been formulated to control all symptoms. Consequently, personalized exercise intervention programs should be designed depending on the various symptoms and severity of ASD individuals in future.

Many experimental studies have widely affirmed the beneficial effects of an exercise intervention on ASD, but few have explored the response of the underlying molecular mechanisms to exercise intervention. In recent years, the disorder of ASD has been recognized as “synaptopathy,” characterized by abnormal synaptic structure and functional changes. Specifically, an increased or reduced density of synapses and spine can be observed in the brain of ASD individuals and animal models. In this section, we discuss the underlying possible molecular mechanisms of exercise intervention that improve ASD symptoms by regulating the structural plasticity of synapses and the spine.

In this review, we summarized the structural alterations of the synapse/spine in ASD and its molecular mechanisms. Importantly, we also discussed that exercise intervention, as a cost-effective therapeutic strategy, significantly improves ASD symptoms. Finally, we mainly emphasized the possible underlying molecular mechanisms from the perspective of microglia-dependent synaptic pruning. Admittedly, these results concerning molecular mechanisms are from animal experiments. Although animal models are a powerful tool for exploring synaptic or cellular physiology, ultrastructure, and biochemistry alterations, species-specific differences might limit direct predictions in humans. Thus, results from clinical trials are required to provide reliable information on the possibility of alleviating the disorder.

Synaptic deficits underlie the pathogenesis of the disorder of ASD. Exercise intervention has been shown to play a significant role in improving ASD symptoms. However, few studies have explored the underlying synapse-related molecular mechanisms. Currently, the existing report is only from the perspective of animal experiments. The report has demonstrated that voluntary wheel-running exercise could ameliorate core symptoms of ASD mice induced by MIA. These effects may be caused by regulating microglia-dependent synaptic pruning deficits, but many questions remain. For example, the molecular mechanisms of exercise intervention improving the developmental synaptic pruning disorder are unclear. The present studies mainly converge on microglia-mediated synaptic pruning. Recently, astrocyte has also been shown to be involved in synaptic pruning. However, the molecular mechanisms for whether or how it mediates the process of exercise intervention regulating synaptic pruning have not been reported. In microglia-dependent synaptic pruning, the classical complement system is an important pathway to mediate microglia to recognize low active synapses, leading to synaptic elimination. However, how the classical complement system works during exercise intervention regulating synaptic pruning remains to be demonstrated. Further clarification of these questions may lead to the elucidation of the molecular mechanisms for improving ASD symptoms through exercise intervention.

WZ drafted the work. XL collected the information. GD, LZ, and KL substantively guided and revised it. All authors contributed to the conception, design of the work, collected the information, analyzed data used in the systematic review, read, approved the submitted version, agreed to be personally accountable for the authors' contributions, and to ensure the accuracy and integrity of the work.