Autism spectrum disorder (ASD) is a term that encompasses a group of neurodevelopmental conditions. The common dominator of these conditions is a defect in two core domains, social/communication skills and restricted or repetitive behaviors (American Psychiatric Association, 2013). There is no approved biomarker for ASD diagnosis, and the current diagnostic method is based on clinical evaluation, which tends to vary greatly between the affected individuals. The advancement in high-throughput proteomics techniques, which made measuring multiple proteins simultaneously a reality, motivated scientists to undertake proteomics discovery studies using body fluids of ASD subjects in an effort to identify biomarkers for diagnosis and understanding of ASD pathogenesis (Voineagu et al., 2011; Saghazadeh et al., 2019; Hewitson et al., 2021). Many of these studies have identified dysregulation in various components of the immune system as they observed upregulation in pro-inflammatory markers, such as IFN-γ, TNF-α, IL-6, and IL-8, in the brain of ASD individuals at both the protein and mRNA levels (Voineagu et al., 2011; Saghazadeh et al., 2019), thus suggesting that pro-inflammation may play a pivotal role in ASD pathogenesis.

In addition, autoimmune abnormalities have been observed in ASD (reviewed by Edmiston et al., 2017). The notion that maternal autoantibodies may play a role in children's neurodevelopment arose when maternal IgG antibodies were found in the cerebrospinal fluid of fetuses and newborns, suggesting that maternal autoantibodies can pass through the blood–brain barrier (Adinolfi et al., 1976). Several studies further observed that maternal autoantibodies of ASD subjects are reactive toward fetal brain proteins (Braunschweig et al., 2008, 2013). Moreover, studies on the blood of ASD children found a significant increase of autoantibodies in ASD compared to healthy controls; these autoantibodies were specific for myelin basic protein (MBP), ribosomal P protein, and nuclear antigens (Singh et al., 1993; Mostafa et al., 2014, 2016). The majority of published studies have tested a limited number of autoantibodies in ASD, and, there is yet to be a study that utilizes high-throughput technology to test a large number of autoantibodies in ASD cases. Therefore, the overall landscape of ASD's autoantibody reactivity across a large spectrum of self-antigens is still unknown.

Autoantibodies originate from autoreactive B-cells. In a normal scenario, B-cells producing antibodies that bind with a high affinity to self-antigens are eliminated, while B-cells producing antibodies with medium-/low-binding affinity are more likely to escape the elimination processes (Ludwig et al., 2017). Many low-affinity autoantibodies have anti-inflammatory features depending on the antibody subclass (IgM) and the post-translational modifications it has undergone (glycosylation/sialylation) (Karsten et al., 2012; Mannoor et al., 2013). However, in many cases, autoantibodies against self-antigens can interfere with physiological processes and induce devastating life-long illnesses. For instance, autoantibodies may disrupt the physiological function of the cells by different mechanisms such as mimicking receptor function, inducing cell signaling and inflammatory processes, blocking neurotransmission, and inducing cell lysis (Ludwig et al., 2017).

Collectively, the studies mentioned above have presented emerging evidence linking immune system dysregulation to ASD. Nevertheless, the comprehensive autoantibody profile of individuals with ASD and the potential mechanisms through which these autoantibodies might influence ASD remain to be fully elucidated. Therefore, investigating the autoantibody profile of ASD individuals might provide insights into the autoimmune status of ASD. Herein, we screen for 1,623 autoantibodies using the serum of 93 ASD individuals and 28 healthy controls between the ages of 6 and 15 years. To the best of our knowledge, this is the first study that screens serum autoantibodies of ASD using high-throughput KoRectly Expressed (KREX) technology. The study aims to explore the autoreactivity of IgG antibodies against self-antigens in ASD individuals to identify potential autoantibodies that could underly ASD pathology.

Several studies have identified the presence of autoantibodies against fetal brain and neuronal proteins in ASD children and their mothers (Piras et al., 2014; Abou-Donia et al., 2019; Mazón-Cabrera et al., 2019; Ramirez-Celis et al., 2021). The implication of those autoantibodies and whether or not they interfere with normal developmental and biological processes are still unknown. In our study, we performed quantitative autoantibody profiling using 1,623 antigens for ASD subjects and healthy controls in an effort to identify differentially expressed autoantibodies and to elucidate the biological relevance of those autoantibodies in ASD. An aggregate of 534 autoantibodies were identified across the screened samples, 29 of which were differentially expressed in ASD (Figure 3; Table 2). The highest FC differences between ASD cases and HCs were observed in two proteins, PSIP1 (FC ~ 1.7) and NAP1L3 (FC ~ 1.6), which were upregulated in ASD. PSIP1 is a nuclear protein and a transcription coactivator that is involved in neurogenesis processes such as neuroepithelial cell differentiation (Chylack et al., 2004). Furthermore, NAP1L3 is also a nucleosome protein and a histone chaperone that plays a crucial role in epigenetic transcription regulation (Heshmati et al., 2018). Interestingly, both proteins have chromatin-binding abilities and were enriched in the chromatin-binding term in GO analysis. Moreover, other corresponding proteins also enriched in chromatin binding GO term are HDAC4, CTNNB1, and ZKSCAN3; however, unlike PSIP1 and NAP1L3, these proteins exhibited reduced expression in ASD compared to HCs. HDAC4 encodes for histone deacetylase 4, a protein involved in the deacetylation of the core histones, thus regulating transcription (Wang et al., 2014). Deletion in HDAC4 has been reported in ASD (Williams et al., 2010). CTNNB1 is a beta-catenin protein involved in the Wnt pathway in the neurons. Interestingly, according to the Simons Foundation Autism Research Initiative (SFARI) database, CTNNB1 nonsense and missense mutations have been reported in ASD subjects (O'Roak et al., 2012). No evidence has been reported on the relationship between ASD pathogenesis and ZKSCAN3.

Interestingly, differential levels of autoantibodies against proteins with neuronal-related function were also detected in our cohort. For instance, EPHA3, MAPK1, PAK2, and SSNA1 are important for the axonal guidance process during neurogenesis (Brown et al., 2000; Hammarlund et al., 2009; Ding et al., 2015; Lawrence et al., 2021). Moreover, MAPK1 is also involved in cholinergic and serotonergic synapse (as shown in our GO analysis). MAPK1 is highly expressed in the human brain and is involved in various processes such as synaptic plasticity, learning, and memory (Thomas and Huganir, 2004). Although we were not able to rule out the cross-reactivity between MAPK1 and MAPK3 due to their high correlation and homology, MAPK1 and MAPK3 are both as important in the MAPK/ERK cascade and participate in the same biological processes; thus, they might both correlate due to their biological relevance. Moreover, recent studies showed that MAPK1 and MAPK3 can generally compensate for each other in most tissues and have a very similar expression pattern (Upadhya et al., 2013; Frémin et al., 2015). MAPK1 has a critical role in multiple signaling pathways such as MAPK, ErbB, Ras, and Rap1 pathways, as shown in GO enrichment analysis (Figure 5A; Upadhya et al., 2013).

Another pathway of autoantibodies that was found to be differentially expressed in nine of its proteins is the metabolic pathway, which includes amino acids and fatty acid metabolism, gluconeogenesis, and glycolysis. Amongst these autoantibodies is one that corresponds to the LDHB protein. Autoantibodies against this protein have been reported to be upregulated in ASD children and their mothers (Braunschweig et al., 2013; Ramirez-Celis et al., 2021) contrary to our cohort, where it was found downregulated. This discrepancy could potentially stem from technical differences as the aforementioned studies utilized western blotting for detection, while our study employed iOme microarray. In addition, both western blotting and iOme microarray entail different normalization approaches. LDHB is an enzyme that catalyzes the conversion of lactate to pyruvate, a critical energy substrate. A recent study using an LDHB knock-out mouse model showed that LDHB deficiency triggers gliosis, neuroinflammation, and neurodegeneration in the brain (Park et al., 2022). Only one autoantibody, corresponding to GOT1 protein, exhibited a significant but weak correlation with the ADOS-2 scores. Our GO analysis revealed that GOT1 is involved in multiple metabolic processes, such as cysteine and methionine metabolism pathways. Paradoxically, our findings revealed a negative correlation between age and the majority of the identified autoantibodies. This finding suggests that there is an overall reduction in autoantibody signal as ASD children age increases. Moreover, our study could not confirm the source of the identified autoantibodies; therefore, more longitudinal studies are needed to validate and properly curate the origins and the dynamic changes of these autoantibodies.

Considering the existing reports on ASD, autoantibodies against glyceraldehyde-3phosphate dehydrogenase (GAPDH) have previously been identified (Mazón-Cabrera et al., 2023). Unfortunately, this specific autoantigen is not included in the Sengenics autoantigen panel. On the other hand, autoantibodies against stress-induced phosphoprotein 1 (STIP1) have been previously identified in ASD (Braunschweig et al., 2013), and the autoantigen is present in the Sengenics panel; however, our results revealed no statistically significant difference in its expression. It is crucial to highlight that autoantibodies targeting these proteins were primarily identified in mothers of ASD children rather than the children themselves. Another study highlighted the significance of GFAP and SNCA autoantibodies in ASD subjects (Abou-Donia et al., 2019). However, our study indicated that GFAP did not yield statistically significant results, and SNCA did not pass the NCF threshold, leading to its exclusion from the statistical analysis. These previous studies predominantly employed Western blotting for autoantibody detection, and this methodological disparity could potentially explain the challenges in reproducing these findings within our cohort.

It is noteworthy that the majority of the identified autoantibodies in this study were overall downregulated in ASD compared to HCs. The cause of this downregulation and how it might be involved in ASD pathogenesis requires further investigation. However, the overall downregulation of autoantibodies in ASD does not indicate an underlying autoimmune pathology. The presence of higher concentrations of autoantibodies does not necessarily signify pathology; on the contrary, they might serve a protective role (Shoenfeld and Toubi, 2005). For instance, autoantibodies against dsDNA were negatively correlated with lupus nephritis in systemic lupus erythematosus patients; in other words, the higher the antibody titer, the lower the severity of the disease (Witte, 2008). Furthermore, anti-TCR antibodies that circulate in myasthenia gravis patients' peripheral blood are reactive against a specific T-cell population that expresses the Vβ5.1 TCR gene. This gene is largely responsible for the production of anti–acetylcholine receptor antibodies in patients with myasthenia gravis, thus inhibiting the proliferation and IFN-γ production of those cells and reducing the severity of the condition (Jambou et al., 2003). Another example of the protective role of autoantibodies against self-antigens was tested in the Alzheimer's disease mouse model, where administrating antibodies against β-amyloid, a protein responsible for plaque formation in Alzheimer's disease, resulted in reduced amyloid plaque formation in the brain and attenuated other pathological features of Alzheimer's disease (Schenk et al., 1999). This hypothesis was observed in humans as naturally occurring autoantibodies against β-amyloid were reduced in the cerebrospinal fluid of Alzheimer's disease patients compared to healthy controls (Du et al., 2001).

The majority of the identified differentially expressed autoantibodies correspond to cytosol and nuclear proteins; hence, they are expressed intracellularly. It is important to mention that autoantibodies production is not limited to extracellular and cell-surface proteins, but rather, there have been some instances where autoantibodies are generated against intracellular proteins. For example, autoantibodies against double-strand DNA are known to be present in systematic lupus erythematosus disorder (Rahman and Isenberg, 2008); autoantibodies against SSA and SSB are known to be present in Sjögren's syndrome (Routsias and Tzioufas, 2010); and autoantibodies against myeloperoxidase (P-ANCA) and proteinase 3 (p-ANCA) are present in vasculitides (Chen and Kallenberg, 2010). The mechanism by which the immune system is exposed to those antigens is thought to be through apoptosis, where the intracellular content of the cell is expressed on the surface of the apoptotic body (Casciola-Rosen et al., 1999). Another possible mechanism is antigen mimicry; for example, the Ro60 antigen, recognized by anti-Ro60 in systematic lupus erythematosus, shares similarities with a ribonucleoprotein found in Mycobacterium smegmatis as well as with a specific amino acid sequence (58–72) in the Epstein–Barr nuclear antigen-I (EBNA-I), a latent viral protein expressed by Epstein–Barr virus (McClain et al., 2005). However, the exact mechanism of how this takes place in the context of ASD is unknown.

Other examples of autoantibody targets that have been reported to have a protective role are autoantibodies against chemokines and cytokines. These autoantibodies protect against disease models such as multiple sclerosis and rheumatoid arthritis, thus confirming the importance of beneficial autoimmunity in controlling harmful self-directed immune responses (Wildbaum et al., 2003). In addition, the presence of natural autoantibodies that arise with no prior exposure to antigen plays a crucial role in clearing apoptotic cell bodies by marking them for macrophage-mediated phagocytosis (Elkon and Casali, 2008). These natural autoantibodies recognize intracellular antigens such as nucleosomes and phosphatidylserine in the apoptotic cell body.

To the best of our knowledge, our study is the first to perform autoantibody profiling using a large number of antigens for ASD individuals. However, it is important to acknowledge that our study faced limitations, primarily related to sample size, as the number of the HCs group was smaller than that of the cases. Subsequent studies are needed to validate our findings through larger sample sizes and to explore the role of these autoantibodies in the pathogenesis of ASD. Although we excluded sequence similarity among the candidate antigens, the potential for conformational molecular mimicry remains. Furthermore, our study uncovered a negative correlation between the majority of the identified autoantibodies and the age of ASD individuals. As a result, future longitudinal studies should include samples from ASD children to better elucidate this correlation and to validate our findings.

In conclusion, this study aimed to explore autoantibody reactivity against self-antigens in the serum of ASD individuals using a high-throughput assay. We identified 29 autoantibodies that were differentially expressed in ASD. These differential autoantibodies were reactive against proteins involved in axonal guidance, synaptic function, amino acid metabolism, fatty acid metabolism, and chromatin binding. In addition, we detected an overall decline in autoantibody concentration as ASD children's ages increased. This study was limited by the imbalanced number of participants and the significant difference, in terms of age and sex, between the studied groups. Therefore, validating our findings using a different cohort with a larger sample size is warranted. Furthermore, additional longitudinal investigations are necessary to accurately ascertain the sources and evolving patterns of these autoantibodies.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

The studies involving humans were approved by Qatar Biomedical Research Institute (QBRI). The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participants' legal guardians/next of kin.

AM: conceptualization, data curation, formal analysis, and writing—original draft. HE: data curation and formal analysis. KL: formal analysis. HA: project administration, resources, and writing—review and editing. FA-S: data curation (ADOS-2 scores) and writing—review and editing. JD, NA, ME, and OA: conceptualization and writing—review and editing. AA: conceptualization. LS: conceptualization, project coordination, and writing—review and editing. SA: project coordination. JB: writing—review and editing. OE-A: conceptualization, study design, supervision, and writing—review and editing. All authors have read and agreed to the published version of the manuscript.