ASD are a behaviorally defined syndrome that consists of a heterogeneous group of neurodevelopmental disorders [1]. Various gene mutations or microdeletion syndromes are known to cause ASD behavioral symptoms; however, single gene mutations are likely to account for only a small fraction of ASD cases [2]. In the majority of ASD subjects, no causative gene mutations have been identified. In these cases, it is more likely that complex interactions among multiple genetic and environmental factors contribute to the onset and progress of ASD, as seen in other polygenic diseases.

Currently, first-line treatment measures for ASD are behavioral and pharmacological interventions. Unfortunately, favorable responses to these measures are not universally achieved; some ASD patients respond well to these measures, while others do not. This is partly attributed to both the heterogeneity of the ASD population [1] and a high frequency of comorbid medical conditions that may affect behavioral symptoms through pain and discomfort [3]. At this time, there are no objective biomarkers for identifying the children who will not be responsive to the first-line therapeutic measures.

Although there are alternative treatments, including those proposed by practitioners of complementary and alternative medicines (CAM) [4],[5], these are not universally effective either, and research has not shown which are beneficial to what types (subgroups) of ASD subjects. There is also a lack of both reliable objective biomarkers and understanding of the mechanisms of actions [6]. Parents of ASD children may turn to such CAM treatments, hoping for the best. There is an urgent need for identifying objective biomarkers that may guide optimal effective measures for ASD children who fail to respond to the first-line intervention measures.

However, the etiology of poor responses is also multifactorial, and it is extremely difficult to identify biomarkers by focusing on `poor responders’ to conventional first line treatments. Previous genetic workups indicate that mutational analysis may not be practical for segregating ASD subjects who have an apparent involvement of multiple genes [1]. Instead, it may be more feasible to focus on ASD children who are not only poor responders but also have objective clinical and laboratory characteristics that are distinguishable from other ASD subjects. The ASD-IS children described in our previous studies, as detailed in the next paragraph, may be ideal for such a purpose

In our pediatric allergy/immunology (AI) clinic, we have evaluated a substantial number of ASD children for possible food allergy (FA) and underlying immune condition. Evaluations of these ASD children led us to recognize that some of them manifest clinical features indicating immunodysregulation and/or chronic inflammation. Namely, these children exhibit chronic and/or recurrent sino-pulmonary infection; adverse reactions to multiple medications; persistent GI symptoms with high frequency of NFA; and prolonged courses of illness. Moreover, these children do not respond well to pharmacological and behavioral intervention measures per parental report [7],[8]. In this study, we categorized such ASD subjects as ASD-IS. Further studies of these children, using peripheral blood mononuclear cells (PBMCs), revealed abnormalities of innate immune responses that were distinguishable from both ASD and normal controls [9]. Changes in transcriptional profiles of isolated PB Mo, which did not overlap with those found in either the ASD and normal controls, were also found [9]. Based on these findings, we hypothesized that ASD-IS children have distinguishable innate immune abnormalities, which can be associated with behavioral changes following immune insults, typically microbial infection.

It should also be noted that our previous results revealed discrepancies between transcriptional profiles of PB Mo and protein levels of cytokine produced by PBMCs, a mixture of lymphocytes, PB Mo, and other lineage cells. Despite the upregulation of mRNA of various cytokines by PB Mo, the production of such cytokines by PBMCs was lower in ASD-IS subjects; stimulants used were targeted to PB Mo [9]. This could be associated with the influence of cells other than PB Mo, such as regulatory T cells (Treg cells) or suppressive (tolerizing) dendritic cells (DCs) contained in PBMCs. Alternatively, post-transcriptional regulation in PB Mo, such as those exerted by microRNA (miRNA) [10],[11], may also affect the protein levels of cytokines depending on the stage of PB Mo activation. We hypothesized that if intrinsic post-translational regulation has a role in the discrepancy between protein and mRNA expression of cytokines, isolated PB Mo may reveal distinct patterns of cytokine production depending on the activated state of PB Mo. That is, post-transcriptional regulation exerted by miRNA and other factors are expected to change rapidly depending on the activated stage of PB Mo [10]. It should also be noted that such changes of post-transcriptional regulation may affect the neuroimmune network [12]. Interestingly, in our observation, ASD-IS children are characterized by rapid changes in behavioral symptoms following immune insults, as observed in PANS children [13].

To test our hypothesis, we examined cytokine production by purified PB Mo from ASD-IS and control ASD and non-ASD subjects. We have reported previously reported a high prevalence of NFA in young ASD children with chronic GI symptoms, including ASD-IS children [7]. Since the presence of NFA may indicate innate immune abnormalities that predispose to NFA at least in the gut mucosa [14], ASD and non-ASD controls were further separated into two subgroups, children with or without NFA. In addition, non-ASD/PANS subjects were added to the non-ASD control groups, since the ASD-IS children we recruited are often diagnosed with PANS by CAM practitioners prior to entrance into this study.

The study follows the protocols approved by the Institutional Review Board at Rutgers-New Jersey Medical School. Informed consent was obtained prior to blood sample obtainment.

This study addressed cytokine production profiles by PB Mo in subgroups of ASD subjects categorized on the basis of clinical features and allergy/immunology s as detailed in the method section. To our knowledge, previous studies addressing PB Mo functions in ASD children did not subdivide ASD subjects and involved smaller numbers of ASD subjects than this study [26]. Also novel to this study, three non-ASD controls groups were included, that is, non-ASD/NFA, PANS, and typically developing normal controls without NFA or PANS. This experimental design permits us to dissect what findings are specific for the ASD-IS subjects. In addition, we analyzed the data depending on the `flare’ versus `non-flare’ states in the ASD-IS and PANS subjects. This is for testing our hypothesis that changes in cytokine production profiles in PB Mo can be associated with fluctuating behavioral symptoms following immune insults in the ASD-IS and PANS children.

In this study, parental impression of behavioral changes in ASD-IS children following immune insults were verified first, using the ABC and CSHQ. Behavioral symptoms, per parental reports, would typically `flare’ for a prolonged period following immune insults in ASD-IS children, despite resolution of acute physical symptoms such as fever or other signs of acute inflammation. When these results were compared to baseline data obtained during `non-flare’ states, we found significantly higher scores in subdomains I (irritability), II (lethargy), and IV (hyperactivity) in ABC scores (Table 3 and Figure 1) during `flares’ in ASD-IS children. These findings support parental impression of worsening behavioral symptoms in the `flare’ state following immune insults. In the ASD-IS children, ABC scores were assessed after resolution of acute symptoms of immune insults such as upper respiratory infection. Therefore, it is unlikely that behavioral changes are solely attributed to the pain and/or discomfort associated with acute illnesses.

Mounting evidence indicates that in other neuropsychiatric conditions such as schizophrenia, inflammatory cytokines such as IL-1Β and IL-6 have a role in onset and progress of neuropsychiatric symptoms [27]-[31]. We have reported previously that PBMCs obtained from ASD-IS children produced reduced amounts of pro-inflammatory cytokines (IL-1Β and IL-6) and IL-10, a counter-regulatory cytokine [1]. In these studies, most samples were obtained when ASD-IS children were in the `non-flare’ states. Thus, the altered PB Mo functions following a `flare’ may have not been captured in our previous study. Alternatively, other lineage cells, other than PB Mo, may have suppressed the functions of PB Mo. In order to address these functions of PB Mo more precisely, this study employed purified PB Mo from ASD-IS children in the `flare’ versus `non-flare’ states.

First, we assessed the differences in cytokine production profiles between PBMCs and PB Mo. We measured cytokine production by PBMCs and PB Mo prepared from the same blood samples in 21 subjects randomly selected from all the study groups. As shown in Figure 2, PB Mo spontaneously produced higher amounts of most cytokines measured except for TGF-Β. The concentration of PB Mo in the monocyte cultures was then adjusted to be roughly equivalent to that of the monocyte concentration in PBMC cultures. Thus, it is very unlikely that changes in cell concentrations are responsible for the differences in spontaneous production of cytokines by PBMCs and PB Mo. Instead, it is more likely that cytokine production by PB Mo is suppressed by other lineage cells present in PBMCs. For example, counter-regulatory cytokines such as TGF-Β and IL-10 produced by inducible regulatory T (Treg) cells are known to suppress functions of effecter cells including Mo [32]. The fact that production of TGF-Β, one of the major regulatory cytokines in the periphery, was higher than that produced by PB Mo indicates a possibility that TGF-Β produced by other lineage cells suppress spontaneous cytokine production by PB Mo. Based on these findings, we reasoned that the use of purified PB Mo will likely yield more information regarding changes in monocyte function in ASD-IS children.

Spontaneous production of cytokines by PB Mo was then analyzed in all the study groups. The data from ASD-IS and PANS children were analyzed in the `flare’ versus `non-flare’ states. PANS children are characterized by acute onset neuropsychiatric changes following immune insults (typically microbial infection) [18]. In our experience, ASD-IS children are often diagnosed with PANS on the basis of their fluctuating behavioral symptoms following immune insults. In the PANS children recruited to this study, triggers of acute onset of neuropsychiatric symptoms varied from viral infection to bacterial infection, and none of them qualified for diagnosis of PANDAS (pediatric autoimmune neuropsychiatric diseases associated with streptococci) [13].

Our results of monocyte cytokine profiles revealed higher spontaneous production of inflammatory cytokines (IL-6, and IL-1Β) in the `flare’ state than in the `non-flare’ states in ASD-IS children (Tables 4 and 5). In PANS children, such changes were not observed. When purified PB Mo were further stimulated by various stimuli of innate immunity, we found that despite increased spontaneous production of IL-1Β, `flare’ ASD-IS cells are still capable of producing higher amounts of IL-1ß in response to LPS and zymosan than `non-flare’ ASD-IS cells. IL-1ß production by `flare’ ASD-IS cells are also higher than other ASD/non-ASD control cells, except for PANS cells. In contrast, IL-6 production by `flare’ ASD-IS cells in response to stimuli was equivalent or even lower than that of `non-flare’ ASD-IS cells (Table 5 and Figure 4-C). It should also be emphasized that such changes were not observed in PANS children in `flare’ versus `non-flare’ states, although they tended to exhibit increased production of IL-1 ß by PB Mo.

When in an inflammatory state or when subjects are recovering from inflammation, proinflammatory immune responses are generally suppressed by intrinsic suppressive mechanisms of cells or by other lineages cells with suppressive functions. This will help self-limit the inflammatory processes. However, our results indicates that ASD-IS PB Mo are still capable of producing significant amounts of IL-1ß in response to stimuli even in the `flare state, despite spontaneous production of high amounts of IL-1ß. Since our assay used the purified PB Mo, our results indicate that ASD-IS monocytes have impaired suppressive mechanisms for IL-1ß production following activation. IL-6 production did not increase in response to stimuli in `flare’ ASD-IS cells. Production of functional IL-1ß requires additional enzymatic steps as opposed to other cytokines [33]. Therefore, it is possible that ASD-IS children have impaired regulation in the specific steps of IL-1ß production.

Dysregulated production of IL-1Β and IL-6 has been known to lead to chronic inflammatory conditions of the CNS, joints, and GI tracts. For example, in cryopyrin associated periodic fever syndrome (CAPS), autosomal dominant mutation of cryopyrin leads to excessive production of IL-1Β, leading to chronic systemic inflammatory condition affecting the CNS, joints, and other organs [34]. In juvenile idiopathic arthritis (JIA), excessive production of IL-6 and IL-1Β (for systemic JIA) has been implicated in their pathogenesis [35]. In these conditions, blockers of these inflammatory cytokines have been effective for treating these conditions [35],[36]. Interestingly, in a rodent maternal immune activation (MIA) model for autism, a single injection of IL-6 to pregnant mice at gestational age of 12.5 days was reported to result in deficits in certain behaviors in the adult offspring [37]. Likewise, increased serum levels of IL-1Β in offspring have also been reported in MIA models [38]. In patients with primary autoinflammatory conditions causing excessive action of IL-1 ß, CNS symptoms that partly resemble ASD behavioral symptoms are well controlled by exogenous IL-1 blockers [34]. These reports along with our results indicate a possibility that changes in IL-1 ß responses may be associated with changes in behavioral symptoms observed in ASD-IS children following immune insults. These findings further support the need for mechanistic studies to assess the feasibility for IL-1 ß and/or IL-6 blockage strategy as a strategy for treating ASD-IS children.

In chronic inflammation, IL-10, a counterregulatory cytokine, is generally upregulated as part of the immunoregulatory actions [39]. However, in the ASD-IS children, spontaneous IL-10 production did not differ between the `flare’ and `non-flare’ states. In addition, we found a significant decrease in IL-10 production by `flare’ ASD-IS cells in response to stimuli of innate immunity as compared to `non-flare’ ASD-IS cells (Figure 5). This was not found in PANS children. Production of IL-10, one of the key regulators of inflammatory responses, by PB Mo is an important intrinsic regulatory mechanism for controlling excessive inflammation [39]. In the state of microbial infection, IL-10 production is often downregulated to mount an immune defense effectively. However, excessive, prolonged downregulation of IL-10 could cause dysregulated, prolonged inflammation. Thus, our results indicate a possibility that monocytes from ASD-IS children also have intrinsic defects in regulatory mechanisms of IL-10 production.

Lastly, it should be noted that the cytokine production profiles in PB Mo differed between ASD-IS and PANS subjects in this study. This finding indicates that although fluctuating behavioral symptoms are somewhat overlapping in ASD-IS and PANS subjects, underlying mechanisms may not be identical in these two groups.

In summary, studies of purified PB Mo from ASD-IS children indicate that they have dysregulated innate immune responses. Specifically, these relate to the production of IL-1 ß and IL-10, revealing significant changes in `flare’ versus `non-flare’ states. These responses, for the most part, were not observed in ASD/non-ASD controls. These results indicate that further analysis of the regulatory mechanisms of PB Mo in ASD-IS subjects may lead to the identification of biomarkers and even treatment options for ASD-IS subjects who typically do not respond well to the first-line therapeutic measures.

HJ is the PI of this study and all the experimental design was done by her along with obtainment of the IRB approval, recruitment of study subjects in the clinic, sample obtainment, supervision of bioassays, data analysis, and manuscript preparation. GL conducted most part of bioassay conducted in this study under the supervision of HJ and helped data analysis and manuscript preparation by HJ. AD is a statistician and mainly contributed to statistical analysis, discussing data analysis and data presentation with the PI. All authors read and approved the final manuscript.

HJ has been conducting studies in ASD subjects with apparent inflammatory components. This study is an extension of her previous studies.