Reelin is an extracellular matrix protein with serine protease activity which highly secreted in several regions of cerebral and cerebellar cortex (Fatemi, 2002). It is encoded by the RELN gene (OMIM *600514) on chromosome 7q22, whose expression is significantly regulated by complex epigenetic modifications (D’Arcangelo et al., 1995; DeSilve et al., 1997; Folsom and Fatemi, 2013). The initial protein precursor is processed and subsequently secreted as a modular glycoprotein able to bind several receptors, including very low density lipoprotein receptors (VLDLRs) and apolipoprotein E receptor 2 (ApoER2) (Fatemi, 2005). The interactions with these receptors initiate a signalling cascade that involves the recruitment of Disabled-1 (Dab1) and the activation of Src family kinases (Fatemi, 2005). This leads to the reciprocal activation of Dab1 domain, which starts itself a crucial signalling cascade for correct neuronal migration and cortical layer formation (Lammert and Howell, 2016; Armstrong et al., 2019). Of note, this cascade also results in the phosphorylation of the N-methyl-d-aspartate receptor (NMDAr) subunits NR2A and NR2B (Chen et al., 2005; Folsom and Fatemi, 2013).

In addition to a cardinal regulatory role in neuronal migration and cortical stratification during the embryonal period, Reelin is involved in synaptic plasticity, memory processing, and cognition in the adult brain (Fatemi, 2005; Armstrong et al., 2019). It is particularly abundant in the Cajal-Retzius neurons of the marginal zone in the cerebral cortex, but high levels of Reelin can be also observed in several regions of the cerebellum, including the surface of the developing cortex, the deep nuclei, and the internal granular layers, where it promotes the migration of Purkinje cells (Armstrong et al., 2019). An impaired Reelin production in the cerebral cortex may lead to cortical abnormalities (e.g., heterotopia, polymicrogyria, and lissencephaly) (Armstrong et al., 2019). In the adult brain, the protein is mainly secreted by a subpopulation of GABAergic interneurons and its distribution pattern is different in comparison to the developing brain, in line with the different functional roles (Ishii et al., 2016).

A growing interest on the pathogenic contribution of Reelin deficiency in neuropsychiatric conditions, including autism spectrum disorder (ASD), has emerged in the last two decades (Fatemi, 2005). ASD is an early-onset developmental condition of variable severity characterized by impaired communication and social interactions, and restrictive or repetitive patterns of thoughts and behaviours (Hutcheson et al., 2003). Although an underlying genetic cause is strongly suspected, the lack of evident classic mendelian inheritance mechanisms suggests a complex genetic aetiolgy (Hutcheson et al., 2003). The observation of brain volume abnormalities in patients with ASD and the reports on a possible autism susceptibility gene on chromosome 7 further stimulated the interest on Reelin as a possible disease biomarker (Folsom and Fatemi, 2013). The aim of this review was to explore the complex and controversial results achieved in the numerous studies investigating the possible pathophysiological link between Reelin and ASD.

We analyzed the relevant studies published between 2001 and 2020 regarding the role of RELN in ASD. Electronic databases (MEDLINE, EMBASE, and the Clinical Trial Database) and papers from the scientific literature were systematically searched using the following terms: “reelin,” “RELN gene,” “reeler mouse,” “reeler mouse and autism,” “reelin and autism,” “reelin and autism spectrum disorder,” and “reelin and autism mouse models”. The abstracts of retrieved references were reviewed and prioritized by relevant content and quality of the reported evidence. The reference lists of the selected articles were used to search for further relevant papers. Only articles in English were reviewed.

A growing interest raised on the study of RELN after the cloning of the mouse reeler mutation (D’Arcangelo et al., 1995). In this model, there is an abnormal migration of normal appearing neurons which eventually leads to several structural defects (e.g., cerebellar hypoplasia, abnormal cell bodies orientation, deficient neuronal positioning, and inverted cortical lamination) (Falconer, 1951; Goffinet, 1979; D’Arcangelo et al., 1995; Fatemi et al., 2001; Magdaleno et al., 2002; Ohshima et al., 2002). The phenotype associated with biallelic variants in the mouse homolog (reln) is characterized by motor impairment (tremors and ataxia) and structural abnormalities involving the cerebellum, cerebral cortex, and hippocampus (DeSilva et al., 1997; Lossi et al., 2019). The discovery of a link between RELN pathogenic variants and an autosomal recessive form of lissencephaly with cerebellar hypoplasia (Lissencephaly type 2, LIS2, OMIM #257320) questioned the use of the reeler mouse as a model to study neuropsychiatric conditions in humans, including autism (Hong et al., 2000). Furthermore, Reelin deficiency-associated behavioral features may be difficult to measure in animal models and the data published so far mainly rely on subtle functional or structural changes (Fatemi, 2002). Therefore, the identification of a measurable biomarker for a better understanding of the pathological processes and the clinical correlations involved in RELN-related human conditions remains crucial (Tueting et al., 1999; Liu et al., 2001; Salinger et al., 2003; Laviola et al., 2006; Ognibene et al., 2007; Laviola et al., 2009; Lossi et al., 2019).

Additional mouse models harboring heterozygous or homozygous alterations in the Reln gene have been developed to investigate the potential link between Reelin and ASD (Sawahata et al., 2020). The endophenotype of Reln mutant mice is complex and include behavioral abnormalities that can be related to core clinical features observed in human patients (Sawahata et al., 2020). In general, rodent models of ASD have been reported to show excessive anxiety, aggressive behavior, and increased seizures susceptibility (Argyropoulos et al., 2013). More complex alterations involve sensory function and motor coordination (decreased in most models), as well as sleep and gastrointestinal functions (Argyropoulos et al., 2013). Some of these abnormalities are recapitulated in Reln mutant mice, supporting the association with human ASD. Increased locomotor activity in an open field test, typically reflecting increased anxiety in mouse models, was reported in Orleans hetero (BALB/C), Orleans homo (BALB/C), and ⊿C-KI (C57BL/6) mice (Lalonde et al., 2004; Sakai et al., 2016; Sobue et al., 2018; Sawahata et al., 2020). Similarly, abnormal responses in the elevated plus maze test, evaluating innate anxiety in the rodents, were described in the Orleans homo (BALB/C), and ⊿C-KI (C57BL/6) models (Lalonde et al., 2004; Sakai et al., 2016). Impaired social interaction was noticed in Orleans hetero (BALB/C), ⊿C-KI (C57BL/6), and in the recently developed Reln-del (C57BL/6) model (Sakai et al., 2016; Sobue et al., 2018; Sawahata et al., 2020). Of note, the Jackson hetero (B6C3Fe) mice exhibited instead novel object recognition deficiency, decreased fear conditioning, and reduced prepulse inhibition (Qiu et al., 2006). Abnormalities in learning and memory were further observed in the ⊿C-KI (C57BL/6) model (impaired working memory in the T-maze test) and Orleans homo (BALB/C) mice (impaired water maze test solution) (Salinger et al., 2003; Sakai et al., 2016). Impaired coordination causing abnormal responses in the stationary beams and Rotarod tests were reported in the Orleans homo (BALB/C) and in the Orleans hetero and homo (BALB/C) mice, respectively (Lalonde et al., 2004; Sobue et al., 2018). Taken together, the endophenotypes observed in Reln mutant mice argue in favor of the potential pathophysiological link between Reelin and ASD, helping to unravel the complex and still elusive underlying mechanisms.

The huge impact of ASD on patients’ quality of life reinforces the need for a better understanding of the underlying pathophysiological aspects. Despite several studies have been focusing on the hundreds of genes involved in brain development, either through gene-gene or gene-environment interactions, a definite understanding of the reasonably complex molecular machinery involved in autism pathogenesis is still lacking. Among the investigated genes, RELN certainly represents an intriguing object of study, especially considering its relevant and proteiform biological functions in human brain. The complexity of these functions likely explains the controversial results achieved by the numerous groups investigating the role of RELN variants in the susceptibility to ASD.

Several lines of evidence demonstrated that the number of GCCs repeats in the 5′UTR of RELN may play a relevant role in autism predisposition, most likely through a negative impact on protein expression and brain development. More recently, a growing evidence also emerged in favor of the association between SNPs in RELN and autism. For example, the rs736707 polymorphism (NM_173054.2:c.9606-57T > C) was detected in several heterogeneous cohorts of affected individuals from different countries, suggesting that this variant may contribute to ASD predisposition. Although this is a deep intronic variant, its possible relevance in a complex disorder such as ASD is not surprising (Wu et al., 2005; Li et al., 2008). However, the exact underlying mechanism remains elusive. It has been suggested that this SNP might affect Reelin expression through the interaction with other SNPs in RELN or distinct nearby loci (Li et al., 2008). However, further studies are needed to confirm this hypothesis and possibly clarify this intriguing association.

In this study, we critically reviewed RELN expression and haplotypes transmission in children with ASD, recapitulating the current knowledge on the contributing role of RELN variants in autism pathogenesis and predisposition. The conflicting evidence regarding the association Reelin-autism that emerged from our review study certainly reflects the relevant complexity of autism pathophysiology, involving variants in single genes as well as environmental factors, such as gastrointestinal abnormalities and immune imbalance (Cheroni et al., 2020). Furthermore, despite the efforts of the scientific community, the studies investigating this pathophysiological link still have undeniable limitations. For example, the behavioral abnormalities observed in animal models might be difficult to interpret and subtle ASD-related features in human subjects can be only approximated in rodent models (e.g., deficits in social interaction and communication) (Patterson, 2011). Taken together, these factors contribute to the conflicting results observed in the reported studies. However, the statistically significant link between autism and GCCs triplets in the 5′UTR or certain SNPs in RELN opens the route to new studies in uncharted territories (e.g., protein dosage in brain samples from affected individuals) towards an improved understanding of the pathophysiological aspects linking Reelin to ASD.