Innate Immune Dysfunction and Neuroinflammation in Autism Spectrum Disorder (ASD)

Cytokines are key chemical mediators of the immune response produced by cells when activated by exposure to an antigen or immunogen. They are fast-acting cell-signaling molecules that require appropriate regulation to maintain a balance between pro- and anti-inflammatory responses. This balance is especially critical during neurodevelopment as cytokines and their receptors are also expressed by neurons, and they are intricately involved in the proper development of the central nervous system (CNS). For example, these chemical messengers play essential roles in regulating synapse formation, maturation and elimination, cell renewal processes, and survival and apoptosis of cells of the CNS. Disrupted cytokine and chemokine expression during critical neurodevelopmental periods show the substantial influence of these signaling molecules on cells within the CNS, including radial glial cells and neural progenitors, neurons, microglia, astrocytes, and oligodendrocytes (reviewed in (Deverman and Patterson, 2009; Zengeler and Lukens, 2021).

Increases in these inflammatory cytokines have been identified repeatedly in ASD, providing evidence of peripheral innate activation and dysregulation compared to typically developing children (Table 1). Early studies in ASD cytokine research identified increases in inflammatory cytokines in whole blood cultures and TLR4-stimulated peripheral blood mononuclear cells (PBMC) (Croonenberghs et al., 2002; Jyonouchi et al., 2001). Activation of TLR4 on innate immune cells, including monocytes, initiates a proinflammatory cascade and production of cytokines including TNF-α, IL-1β, and IL-6. In adults with severe autism, serum IL-1β and IL-6 were significantly higher than healthy controls (Emanuele et al., 2010). In a large cohort of ASD children ages 2–5, significant elevations of IL-1β, IL-6, and IL-12p40 were seen compared to typically developing children. Moreover, these elevated inflammatory cytokines were associated with worsening behaviors (Ashwood et al., 2011a). Similar increases in IL-1β and IL-12 (p70) were also seen in a cohort of high-functioning ASD males, along with increases in IL-1RA and T helper (T_H) cell-associated cytokines IL-5, IL-13, and IL-17 (Suzuki, 2011). In a recent study, individuals with ASD had increased inflammatory cytokines in serum, including IL-1β, IL-6, IL-12, IL-23, and TNF-α, providing additional evidence of peripheral innate immune activation (Ricci, 2013). We recently characterized a cohort of ASD children based on PBMC responses to lipopolysaccharide (LPS) stimulation and found those with higher innate inflammatory responses or increased T cell activation had more behavioral impairments (Careaga, 2017). A subset (endophenotype) of children with ASD that exhibited increased innate immune response and severe behavioral and cognitive problems emerged, accounting for at least a third of the subjects tested. When considering the balance of pro- and anti-inflammatory mediators, the regulatory cytokine TGF-β is often decreased in ASD and associated with worsening behaviors (Ashwood et al., 2008; Okada, 2007; Hashim et al., 2013). This cytokine is critical for dampening the inflammatory responses of monocytes and macrophages during immune response resolution, and deficiency may contribute to prolonged or dysfunctional activation of these cells.

Chemokines are cytokines that mediate chemotaxis and recruitment of immune cells to sites of inflammation. Several chemokines associated with innate immune activation are elevated in ASD. Increases in monocyte chemoattractant protein (MCP)-1, regulated on activation, normal T cell expressed and secreted (RANTES), eotaxin, IL-8, and C-X-C motif chemokine 1 (CXCL1, previously known as Gro-α) have been identified in children with ASD (Ashwood et al., 2011a; Suzuki, 2011). Supporting these findings, Shen et al. recently found elevations of RANTES, macrophage inflammatory protein (MIP)-1α, and MIP-1β and decreases in interferon gamma-induced protein (IP)-10 (also known as CXCL10) and monokine induced by gamma interferon (MIG, also known as CXCL9) in children with ASD. Furthermore, MIP-1α, MIP-1β, and IP-10 concentrations were associated with impairments in social behaviors (Shen, 2016). Increases in CXCL5 have also been seen in ASD (Mostafa and Al-Ayadhi, 2015). Additionally, neonatal blood spots from ASD children were found to have increased MCP-1 (Zerbo et al., 2014). Elevated MCP-1 in amniotic fluid was also associated with an increased risk of having a child with ASD (Abdallah et al., 2012b). These data suggest that chemotaxis and migration of monocytes/macrophages may be impacted in autism after diagnosis, at birth, and potentially in the womb during pregnancy.

Although the majority of ASD studies have identified increases in inflammatory cytokines and chemokines, findings are not always consistent. Reduced cytokines and chemokines were found in neonatal blood spots from ASD females, with no differences in ASD compared to typically developing (TD) in adjusted regression model estimates (Abdallah et al., 2012a). In one study utilizing multi-plex analysis of 29 cytokines in plasma, the concentration of eight cytokines/chemokines was consistently lower in ASD. However, this study used siblings of ASD as controls Manzardo, 2012). ASD is a highly heritable disorder, and families may broadly display immune dysfunction across members; therefore, using siblings as controls can be problematic. Other contradictory findings can be due in part to differences in patient populations and mismatched controls, variations in sample type and processing, and different immunological assays and technologies used for research. Systematic reviews and meta-analyses can help identify consistencies and eliminate some of the noise produced by differences in research. One such meta-analysis looked across 17 studies and concluded that there are consistent elevations in IL-1β, IL-6, and IFN-γ found in individuals with ASD. The chemokines eotaxin, IL-8, and MCP-1 are also frequently elevated, and concentrations of the critical regulatory cytokine active- TGF-β1 are decreased in ASD (Masi, 2015). Results from another recent meta-analysis support these findings, with elevations in IL-1β, IL-6, TNF-α, and IFN-γ seen across 38 studies (Saghazadeh, 2019).

Cytokines seen repeatedly in ASD, such as IL-1β, IL-6, and TNF-α, are produced during activation of the innate arm of the immune system. This arm is responsible for the first line of defense against foreign antigens. When physical or chemical barriers of the host are breached, cells of the innate arm called monocytes act as sentinels, which are recruited to assaulted tissues by chemokines released at sites of infection. Here, these cells join their tissue-resident counterparts to become macrophages. They are critical during the early immune response for producing inflammatory cytokines to initiate an acute phase response in an attempt to control the infection directly (Dey et al., 2015). Monocyte and macrophage mediators produced as a result of nuclear factor kappa light-chain-enhancer of activated B cells (NF-κB) binding to cytokine gene promoters and inflammasome activation influence downstream adaptive responses and later, under optimal conditions, lead to the resolution of inflammation. Conversely, dysfunction in the activity of these cells can lead to excessive and chronic inflammation, inappropriate adaptive immune responses, autoimmunity, and other immune mediated conditions—many of which have been seen in ASD (Hughes et al., 2018). A hallmark of macrophages is their plasticity. These cells skew to a spectrum of phenotypes depending on microenvironment exposures. In vitro, they polarize to a classical inflammatory “M1” phenotype through stimulation with LPS plus IFN-γ or TNF-α. This polarization leads to production of the canonical proinflammatory cytokines IL-1β, IL-6, and TNF-α and helps to drive T helper cell (T_H)-1 differentiation. Conversely, macrophages can be alternatively activated to an “M2” phenotype when exposed to IL-4 and IL-13 in vitro, leading to production of IL-10 and suppression of the “M1” phenotype (Italiani and Boraschi, 2014).

Immune research has revealed increased numbers of innate immune circulating cells, including monocytes, in children with ASD (Sweeten et al., 2003; Tural Hesapcioglu, 2019) (Fig. 1). In some studies, monocyte numbers do not differ; however, monocyte activation is dysregulated. For example, Enstrom et al. found altered cytokine production after activation of TLRs on isolated peripheral CD14⁺ monocytes from ASD children, with increases in the canonical innate inflammatory cytokines TNF-α, IL-1β, and IL- 6 following activation of TLR2. IL-1β was also increased after TLR4 activation with LPS, and concentrations of IL-1β positively correlated with ASD-associated behaviors, including impaired social interactions and non-verbal communication (Enstrom, 2010). Monocytes from ASD children also had increased expression of CD95, which may reflect increased activation (Ashwood et al., 2011b). Whole peripheral blood cultures from ASD children also showed evidence of increased activation of monocytes through elevated production of innate-associated cytokines (Croonenberghs et al., 2002). Isolated PBMC showed elevated gene expression of caspases, including caspase-1 (also known as IL-1 converting enzyme). Caspase-1 is a proteolytic cleavage enzyme with critical involvement in inflammasome activity, leading to activation of IL-1β (Siniscalco et al. 2012).

More recently, Jyonouchi et al. grouped ASD based on monocyte production ratios of IL-1β and IL-10 after stimulation of several TLRs (high, moderate, and low ratios). The IL-1β/IL-10 ratio groups differed significantly in monocyte expression of miRNAs and mitochondrial respiration. When probed with β-glucan, monocytes from the high-ratio group responded with increased pro-inflammatory responses (IL-1β and TNF-α), while the low-ratio group responded with increased anti-inflammatory responses (Jyonouchi and Geng, 2019; Jyonouchi et al., 2017; Jyonouchi et al., 2019). Overexpression of the IL-17A receptor on ASD monocytes has recently been identified, along with increased expression of NF-κB and inducible nitric oxide synthase (iNOS), markers of in flammatory activity (Nadeem, 2018). These markers further increased expression after treatment with IL-17A, the ligand for the IL-17A receptor. IL-17 is associated with autoimmune disorders, including rheumatoid arthritis and multiple sclerosis, and IL-17A was recently identified to have substantial involvement in the MIA rodent model of ASD (Choi et al., 2016). The transcription factor NF-κB plays a significant role in mediating innate inflammatory immune responses (Liu, 2017). Evidence supports increased NF-κB transcriptional activity, as elevated NF κB mRNA and protein are observed in ASD peripheral blood and ASD lymphoblastic cell lines. However, some inconsistencies exist across studies, potentially due to differing methodologies (reviewed in (Liao and Li, 2020). We recently investigated gene expression in TLR activated monocytes and found increased gene expression of NFKB1, a subunit of NF-κB, in TLR-activated monocytes from autistic children. We also found dysregulated gene expression of translational machinery in monocytes from ASD children compared to typically developing controls, which may be contributing to increased inflammatory responses in a subset of these children as a failure to dampen a prolonged immune response through regulation of translation (Hughes, 2022a). We further identified increased IL-6 production in LPS-stimulated monocytes from ASD children. This increase was associated with restricted and repetitive behaviors, supporting the role of excess inflammation associated with innate immune responses in aberrant behavior. Differences in gene expression and cytokine production were only seen after cells were activated; there were no differences at baseline (Hughes, 2022a; Hughes, 2022b). Similarly, Yamauchi et al. found increased gene expression of TNF-α in monocyte-derived macrophages from ASD in dividuals; however, this was only after differentiation to M1- macrophages. Neither baseline monocytes nor differentiated M2- macrophages showed this elevation (Yamauchi et al., 2021).

Aberrant numbers or activation of other innate immune cells are also present in ASD, including significantly increased numbers of peripheral myeloid dendritic cells (mDCs) (Breece et al., 2013) (Fig. 1). Changes in the mDC population may be due to increased differentiation of monocytes into dendritic cells. Further work has shown that peripheral mDCs from ASD children express more surface CD80/CD86 co-stimulatory molecules than TD children (Saad et al., 2017). Differences in DC correlated with amygdala volume, gastrointestinal issues, and impaired behaviors (Breece et al., 2013). In further studies, ASD children with gastrointestinal issues had significantly increased monocyte, eosinophil, and neutrophil infiltration in duodenum, colon, and terminal ileum biopsies (Ashwood et al., 2003; Furlano, 2001). In addition, significantly increased numbers of natural killer (NK) cells are seen in ASD children compared to controls (Ashwood et al., 2011b). Transcriptional profiling of whole blood from children with ASD also revealed increased expression of genes belonging to the NK cell KEGG pathway (Gregg et al., 2008). Later examination confirmed abnormalities in gene expression associated with NK cells, including increased expression of inhibitory killer cell immunoglobulin receptors (KIR) and cytolytic genes. Furthermore, ASD children had increased CD56⁺ NK cell frequency and dysfunctional activity in isolated NK cells under both resting and stimulative conditions (Enstrom et al., 2009). Function of NK cells was increased under basal conditions, but there was a failure to respond to stimulus in the form of tumor cells.

Decreased soluble adhesion molecules were also seen in ASD plasma, including soluble platelet endothelial cell adhesion molecule-1 (sPE-CAM-1), which plays an important role in mediating innate immune cell adhesion to and migration across the endothelial wall to facilitate recruitment to inflamed tissues (Onore, 2012). In mouse models, decreased sPECAM-1 is associated with decreased vascular integrity and increased blood-brain barrier permeability, and deficient mice are more prone to autoimmunity (Woodfin et al., 2007). See Table 2 for a comprehensive list of aberrant cellular findings associated with innate peripheral immune dysfunction.

Microglia are specialized macrophages that reside in the brain and CNS. They play a critical role not only in neurodevelopment but also in CNS homeostasis throughout life. During neurodevelopment, microglia are responsible for phagocytosing excess neuronal precursor cells to regulate neurogenesis by restraining cell production (Cunningham et al., 2013). They also play a role in supporting neuronal survival while also limiting axon outgrowth (Li and Barres, 2018). Production of brain-derived neurotrophic factor (BDNF) by microglia promotes survival and differentiation of neurons, regulation of synaptic formation and transmission, and synaptic plasticity critical for memory and learning (Miranda, 2019; Parkhurst et al., 2013). Similar to peripheral monocytes, they are responsible for surveilling their local environment and can become activated when exposed to a foreign antigen or under inflammatory conditions following infection or injury. Under these conditions, microglia release pro-inflammatory cytokines and can experience aberrant activation, including excessive proliferation. If occurring during developmentally-delicate periods, prolonged inflammatory activity of microglia can alter neurodevelopment (Li and Barres, 2018). It was originally proposed that these cells polarized in a similar fashion to peripheral macrophages. However, more recent evidence suggests that microglia are not fully committed to one phenotype or activation state (Ransohoff, 2016).

Investigations of postmortem brain tissue from ASD individuals have revealed evidence of neuroinflammation in the brain, including microglia activation (detailed in Table 3) (Fig. 1). One of the first studies to identify the presence of neuroinflammation examined tissues from 11 ASD subjects and found regional neuronal loss and evidence of neuro inflammation in the cerebral cortex, white matter, and cerebellum. Increased staining of GFAP and HLA-DR in multiple brain regions suggested marked activation of microglia and astroglia. Increased MCP-1 was also seen in cerebrospinal fluid (CSF) (Vargas et al., 2005). A later study found significantly elevated cytokines associated with innate activation, including TNF-α, IL-6, and GM-CSF, and increased IFN-γ and IL-8 in postmortem tissue from the frontal cortex of 8 ASD subjects (ASD subject). Microglia were then characterized using immunostaining to identify differences in activation phenotypes; ASD tissues showed evidence of increased microglial activation, including amoeboid presentation. Additionally, density and somal volume of microglia were increased in ASD. These researchers also found no evidence of acute inflammation, suggesting microglia have long-standing activation (Morgan et al., 2010). Furthermore, the same group later found increased spatial proximity of microglia to neurons in the dorsolateral prefrontal cortex of ASD brain, with microglia processes frequently encircling the somal bodies of neurons in ASD samples (Morgan et al., 2012). Increased density of microglia has also been identified in two separate regions of cortex in the ASD brain compared to controls (Tetreault et al., 2012). NF-κB is increased in the postmortem ASD brain, with localization to glial and neuronal cells (Young et al., 2011), and an enzyme complex that activates NF κB is also elevated in the ASD brain (Malik et al., 2011). More recently, it was shown that cells of the adaptive arm of the immune system are increased in postmortem ASD brain and seen accumulating near blood vessels, further supporting evidence of neuroinflammation in ASD (DiStasio et al., 2019).

While the postmortem evaluation of ASD brain tissue has consistently shown neuroinflammation and microglial activation, ASD studies using positron emission tomography (PET) to study microglia ‐ in “real-time”- have provided mixed results. To date, PET evaluating microglia activation in ASD has relied on the radiolabeling of ligands that bind to the mitochondrial translocator protein (TSPO), a protein expressed on the outer mitochondrial membrane by activated microglia (Cosenza Nashat et al., 2009). The first study to evaluate microglial activity in this fashion used the first generation [¹¹C]-PK11195 radiotracer to measure TSPO expression in 20 high-functioning male ASD and typically developing control subjects (Suzuki et al., 2013). Using the binding potential as the output PET parameter, [¹¹C]-PC11195 binding was higher in the cerebellum, frontal, temporal parietal, limbic and subcortical regions. Higher radiotracer binding was interpreted as higher TSPO expression and activated microglia phenotypes in high functioning ASD subjects. However, recent studies do not corroborate increased TSPO expression. In one study evaluating the binding potential of the second generation radiotracer [¹¹C]-PBR28 in 15 ASD males, TSPO expression was seen to be lower in several brain regions (Zürcher et al., 2021). Additionally, TSPO expression did not differ in low or high-functioning ASD. In another study using the total distribution volume as the primary PET output parameter and the second generation [¹⁸F]-FEPPA radiotracer, TSPO expression was also lower in both males and females with ASD (Simpson et al., 2022). Mixed findings between studies may be due to several rea sons. The radiotracer used since the first-generation [¹¹C]-PK11195 has a high non-specific binding affinity. The PET output parameter (binding potential vs total distribution volume) can influence findings, as previously documented in a meta-analysis of patients with psychiatric disorders (Marques et al., 2019). Additionally, the measurement of TSPO for microglial activation may also introduce variability. While TSPO is expressed by activated microglia, endothelial cells during homeostasis and astrocytes in a disease context can also express TSPO (Cosenza Nashat et al., 2009). Lastly, Suzuki and colleagues did not identify if their research subjects had TSPO gene polymorphisms, as polymorphisms can influence radiotracer binding. Additional PET studies using microglial-specific proteins, correctly identifying subjects with high and low TSPO binding affinity, and analyzing both the binding affinity and total volume distribution will aid in clarifying the abovementioned findings.

Transcriptomic analyses of various brain regions support findings of increased immune responses in the ASD brain, with enrichment in immune-specific genes seen in several studies (Table 4). Increased expression of immune system-related transcripts was first seen in the temporal cortex of postmortem ASD brains. Pathway analysis resulted in 31 BioCarta gene sets that differed from controls, of which 19 involved immune pathways. Several of these belonged to innate inflammatory pathways, including NFKB, IL1R1, and IL6 (Garbett, 2008). Voineagu et al. found dysregulated genes in the ASD brain, with decreased expression of neuronal genes involved in synaptic function and upregulated genes associated with inflammatory responses and microglial activation. Using weighted-gene co-expression network analysis (WGCNA), they further identified two network modules associated with ASD, with the first involving under-expressed neuronal genes and overlapping with several autism susceptibility genes. The second module associated with ASD was enriched in overexpressed immune genes, and later work by this same group supported these findings (Voineagu et al., 2011; Parikshak et al., 2016). Similar results were also seen in a later study across several neuropsychiatric disorders, including ASD, schizophrenia, and bipolar disorder (Gandal, 2018). One study looked for enrichment of the immune module in a larger ASD data set and specifically identified M2 activation of microglia and type I interferon responses in the ASD brain. The authors argued that the increased expression of genes associated with M2 activation could be an effort to mediate M1 inflammatory skewing that occurs during type I IFN responses (Gupta et al., 2014). Meta-analysis of previous ASD cortex RNA-Seq studies utilizing data sets from GEO (Gene Expression Omnibus) unearthed newly identified differentially expressed genes in the ASD brain, including TNF signaling and complement cascade pathways (Rahman et al., 2020). Methylation studies have also reported hypomethylation of genes associated with the immune response, leading to increased expression of inflammatory genes (Nardone et al., 2014). Taken together, these studies provide evidence that neuroinflammation may play a role in ASD. However, sample sizes for postmortem ASD studies are limited and more research is needed in this area to elucidate potential mechanisms of neuroinflammation that contribute to ASD and other NDD.