TNF-α expression ratio of M1/M2 macrophages is a potential adjunctive tool for the diagnosis of autism spectrum disorder

Background The etiology of autism spectrum disorder (ASD) is complex. Its pathobiology is characterized by enhanced inammatory activities; however, the exact ASD pathobiology remains unclear. Some cases of ASD are dicult to diagnose using existing psychological assessments because the careful exclusion of other psychiatric disorders is challenging. To distinguish between the appropriate targets for interventions and research, the demand for identifying ecient diagnostic biomarkers is increasing. This study aimed to nd an inammatory indicator benecial for the diagnosis of ASD. Methods Cytokine mRNA expression, including tumor necrosis factor-α (TNF-α), was measured in the differentiated M1 and M2 macrophages of ASD patients (n = 29) and typically developed (TD) individuals (n = 30). TNF-α expression was also measured in the monocytes of ASD patients (n = 7) and TD individuals (n = 6). Results TNF-α expression in M1 macrophages and TNF-α expression ratio of M1/M2 macrophages were markedly higher in ASD patients than in TD subjects; however, this difference was not observed in M2 macrophages (M1: p < 0.01; ratio of M1/M2: p < 0.0001; M2: p > 0.05), suggesting that this indicator could be a useful tool for diagnosing ASD (M1: sensitivity = 34.5%, specicity = 96.7%, area under the curve (AUC) = 0.74, positive likelihood ratio (PLR) = 10.34; ratio of M1/M2: sensitivity = 55.2%, specicity = 96.7%, AUC = 0.79, PLR = 16.55). However, there was no signicant difference in the TNF-α expression in monocytes between ASD and TD individuals (p > 0.05). Conclusion These ndings suggest that TNF-α expression in differentiated macrophages represents a novel adjunctive tool for the diagnosis of ASD.


Introduction
Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by impaired social interaction, poor communication skills, and repetitive or restrictive patterns of behavior and interests (1).
The global prevalence of ASD is currently estimated to be 1 in 59 children (2). Although its etiology remains largely unknown and complex, a growing body of evidence implicates the disturbance in the immune system as a mechanism of psychiatric disorders, including ASD (3)(4)(5). Alterations in the peripheral innate and adaptive immune responses, as well as prominent in ammation in the brain, have been reported in patients with ASD (6,7). In the periphery, increased expression of pro-in ammatory cytokines, including interleukin-1 beta (IL-1β), IL-6, tumor necrosis factor alpha (TNF-α), transforming growth factor beta, IL-8, and IL-17 has been observed in the plasma, serum, and peripheral blood mononuclear cells (PBMCs) of patients with ASD (7,8). Antioxidant networks (e.g., superoxidase dismutase expression and activity) and the responses to innate immunity stimuli (e.g., lipopolysaccharide and zymosan) have been found to be dysregulated in the monocytes of ASD patients (9)(10)(11). In the central nervous system (CNS), a positron emission tomography study and a few postmortem studies have shown microglial activation in the brains of ASD patients (12)(13)(14)(15).
Macrophages are myeloid cells that are derived from circulating blood monocytes and are known to behave similarly to microglia (16). These cells play a critical role in immune function and are implicated in the pathology of medical conditions, such as autoimmune and allergic diseases (17)(18)(19).
Macrophages can be differentiated into two types: M1 (classically activated macrophages) and M2 (alternatively activated macrophages). Bacterial infection, interferon-γ produced by activated T cells, and TNF-α produced by antigen-presenting cells induce the polarization towards M1 macrophages, resulting in phagocytosis and an increase in the release of pro-in ammatory cytokines, such as IL-1β, IL-6, and TNF-α (20)(21)(22). The uncontrolled production of pro-in ammatory cytokines by M1 macrophages results in the disruption of tissue homeostasis and could increase the risk for neuropsychiatric disorders (23)(24)(25). In contrast, the polarization towards M2 macrophages is induced by IL-4, IL-10, IL-13, and glucocorticoids and is involved in anti-in ammatory reactions, tissue remodeling, angiogenesis, and tumor progression, with the secretion of growth factors and neurotrophic factors (26,27). M2 macrophages are also implicated in the pathobiology of neuropsychiatric and neurodegenerative disorders, such as bipolar disorder, Parkinson's disease, and multiple sclerosis (25,28,29). Moreover, recent ndings have indicated that the polarization of macrophages is a biologically relevant indicator for the evaluation of medical conditions, such as cancer prognosis (30, 31).
To date, the diagnosis of ASD has usually relied on a clinical interview using DSM-5, and the most accurate diagnosis is achieved using non-biological tools, namely the Autism Diagnostic Observation Schedule-2 (ADOS-2) and the Autism Diagnostic Interview-Revised (ADI-R) (32, 33). The sensitivity and speci city are 85.9% and 82.9% for ADOS-2 and 66.2% and 89.5% for ADI-R. When combined, the sensitivity and speci city become 42.2% and 100%, respectively (34). These tools are highly powerful; however, for more objective assessments, biological diagnostic tools are necessary.
In this study, we aimed to investigate cytokine expression in M1 and M2 macrophages of ASD patients and typically developed (TD) individuals to establish an e cient diagnostic biomarker to aid in the diagnosis of ASD.

Material And Methods
This study was approved by the ethics committee of Nara Medical University School of Medicine and was carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki) for experiments involving humans. All participants were given a complete description of the study and have provided written informed consent before enrollment.

Participants And Clinical Assessments
The participants included 29 ASD patients (age: 28.0 ± 6.7 years, 5 females) and 30 TD individuals (age: 27.2 ± 5.6 years, 5 females) of Japanese ethnicity. All participants have been born and have been living in Japan. ASD patients were recruited from the outpatient service of the Department of Psychiatry, Nara Medical University Hospital and its a liated psychiatric clinic. ASD diagnosis was based on the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) criteria. At least two experienced psychiatrists separately examined the patients, and a diagnostic consensus was reached. Further evaluation was performed using ADOS-2 by psychiatrists and trained staff (35). Autism symptom severity was assessed via self-report using the Autism Quotient-Japanese version (AQ-J)(36, 37). The full intelligence quotient (FIQ) of each participant was estimated using the Similarities and the Symbol search subtests of the Wechsler Adult Intelligence Scale, Third Edition (38). The following exclusion criteria were used: participants under 17 years old; those with low intelligence (FIQ < 70); those diagnosed with other neurological disorders and mental illnesses as evaluated by the Mini-International Neuropsychiatric Interview; and those who used steroids. We did not evaluate Asperger's syndrome.

Monocyte Isolation And Macrophage Differentiation
We used a magnetic-activated cell sorting (MACS) system (Miltenyi Biotec, San Diego, CA, USA) for monocyte isolation, and subsequently we used Human M1 or M2 Macrophage Differentiation Kit (R & D Systems, Minneapolis, MN, USA) according to the manufacturer's protocol. Brie y, whole human blood samples were collected through venipuncture from all participants during daytime (09:00-16:00) and were stored on ice. Immediately, PBMCs were separated from whole blood using density-gradient centrifugation with the separation medium Lymphoprep (Axis Shield, Oslo, Norway), separation tubes, and Leucosep (Greiner Bio-One, North Carolina, USA). CD14 + monocytes were isolated using the MACS system with CD14 microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany) from PBMCs. These cells were used as monocytes.
For macrophage differentiation, CD14 + monocytes were resuspended in phosphate-buffered saline solution (Wako, Osaka, JAPAN), containing 0.5% bovine serum albumin (Sigma-Aldrich, St. Louis, MO, USA), 2 mM ethylenediaminetetraacetic acid, and 1% Penicillin-Streptomycin mixed solution (Nacalai Tesque, Kyoto, JAPAN). They were seeded at a density of 1 × 10 6 cells/mL onto M1 or M2 differentiation medium incubated with the recombinant human granulocyte-macrophage colony-stimulating factor (GM-CSF) or the recombinant human macrophage colony-stimulating factor (M-CSF), respectively at 37 °C in a humi ed atmosphere of 5% CO 2 . On day 3, half of each culture medium was replaced with a fresh medium, and the M1 or M2 macrophages were obtained on day 6.

Statistical Analysis
Differences between the ASD patients and TD individuals in terms of demographic characteristics, i.e., age, educational level, and FIQ, were examined using Welch's t-test, and Fisher's exact test was used to examine the differences in sex and allergic diseases. qRT-PCR data are presented as median values and interquartile ranges. Comparisons were performed using the Mann-Whitney U test. RNA concentration data were presented as the mean and standard error of the mean. The comparison was performed using one-way analysis of variance. Positive likelihood ratios (PLRs) were used to evaluate the ability of cytokine expression in M1 and M2 macrophages in predicting ASD diagnosis. The PLR is one of the best methods to evaluate diagnostic accuracy and gives the change in the odds of having a diagnosis in patients with a positive test (39). A PLR of 10 indicates a 10-fold increase in the odds of a particular condition in a patient with a positive test result. Normally, PLRs of greater than 10 signi cantly increase the probability of a disease (40). The PLR is calculated as sensitivity/(1 -speci city) through the receiver operating characteristic (ROC) curves. Correlation analyses between gene expression levels and age, FIQ, and the severity of ASD symptoms were performed using Spearman's rank correlation. Logistic regression analysis was used to examine the association between TNF-α mRNA expression in macrophages and the prevalence of allergic diseases. All statistical analyses were performed using Prism 8 (GraphPad), except for logistic regression analysis, which was performed using SPSS version 26 (IBM). P-values < 0.05 were considered statistically signi cant and are indicated in the gures as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Table 1 presents the demographic data of the participants in this study. There were no signi cant differences in age (U = 403, p = 0.63) or sex (χ 2 [1] = 0.003, p = 1.00) between the ASD and TD groups. ASD patients showed signi cantly lower educational level (U = 213.5, p = 0.0004) and lower FIQ (U = 233.5, p = 0.0019) compared with TD individuals. The prevalence of allergic diseases was higher in ASD patients than in TD participants (χ 2 [1] = 5.77, p = 0.02). Allergic diseases included asthma, atopic dermatitis, and food allergies.

Cytokine expression in M1 and M2 macrophages: differences between ASD patients and TD participants
Prior to the measurement of the mRNA expression levels of cytokines, we have con rmed that the total RNA concentrations of harvested M1 and M2 macrophages were not different between ASD and TD groups (Suppl. Figure 1). This suggests no difference in the number of differentiated M1 and M2 macrophage cells between the two groups. Then, we measured the mRNA expression levels of CCR7 and IL-10 as markers for M1 and M2 macrophages, respectively (41,42). The mRNA expression level of CCR7 was higher in M1 than in M2 macrophages (U = 27, p = 0.0079), whereas that of IL-10 was higher in M2 than in M1 macrophages (U = 0, p < 0.0001), thus con rming the differentiation into M1 and M2 macrophages (Suppl. Figure 2).
Because a previous study has revealed the abnormal expression of IL-17A and IL-17RA in ASD patients (43,44), we also examined the mRNA expression level of IL-17RA, but found no signi cant difference in M1 or M2 macrophages between the two groups (M1: IL-17RA, p = 0.42; M2: IL-17RA, p = 0.45) (Fig. 1E). From the calculated mRNA expression ratios of M1/M2 macrophages, which may indicate the pattern of in ammatory responses of these macrophages, we found that only that of TNF-α was markedly higher in ASD patients than in TD individuals (M1/M2: TNF-α, p < 0.0001, r = 0.50) (Fig. 1A). Importantly, logistic regression analysis revealed no associations between the TNF-α expression in M1, M2, and M1/M2 macrophages and prevalence of allergic diseases (Suppl. Table 1).

TNF-α mRNA expression ratio of M1/M2 macrophages as a potential biomarker for ASD
The ROC curves of cytokine expression in M1 and M2 macrophages, as well as the expression ratios of M1/M2 macrophages, were delineated to calculate the sensitivity, speci city, and PLR ( Table 2). The AUCs of TNF-α mRNA expression in M1 macrophages and its expression ratio of M1/M2 macrophages were signi cantly higher than the chance level ( Fig. 2A-C) (Suppl. Figure 3A-D). PLR was calculated to validate the usefulness of cytokine expression for ASD diagnosis. The PLR of mRNA expression ratio of TNF-α of M1/M2 macrophages was calculated as 16.55 (sensitivity: 55.2%, speci city: 96.7%) (Fig. 2C). These ndings overall suggest that the diagnostic accuracy of ASD can be remarkably improved with the reduction of false positives by the addition of TNF-α mRNA expression ratio of M1/M2 macrophages to psychiatrists' examination. 3.3. No correlations between TNF-α mRNA expression ratio of M1/M2 macrophages and age, FIQ, and the severity of ASD symptoms.
TNF-α mRNA expression in either M1 or M2 macrophages did not correlate with age, FIQ, or the severity of ASD symptoms (AQ-J score) in ASD and TD except for the correlation of TNF-α mRNA expression in M2 macrophages with AQ-J score. TNF-α mRNA expression ratios in M1/M2 macrophages did not correlate with either the age, FIQ, or AQ-J score in ASD and TD (Fig. 2D), suggesting that this biomarker can be used to evaluate a broad range of ASD patients.

TNF-α mRNA expression in monocytes between ASD patients and TD participants
Moreover, the monocytes in participants with further informed consent were collected, and their TNF-α mRNA expression level was measured to validate the necessity of macrophage differentiation. We found no signi cant difference in TNF-α mRNA expression between ASD patients and TD individuals (U = 15, p = 0.45) (Fig. 3). In contrast, TNF-α mRNA expression in M1 macrophages and TNF-α mRNA expression ratios in M1/M2 macrophages were signi cantly different between ASD patients and TD individuals (M1: U = 4, p = 0.01; M1/M2: U = 0, p = 0.001), suggesting the value of macrophages for the determination of ASD endophenotypes.

Discussion
ASD diagnosis is often based on psychological assessments, including ADOS-2 and ADI-R, which are time-consuming and relatively subjective (35, 45). Therefore, more objective indicators, such as biomarkers, are required for more accurate ASD diagnosis. In this study, we have provided evidence indicating the potential of the TNF-α mRNA expression ratio of M1/M2 macrophages with high sensitivity to increase the rate of true positive ASD diagnosis.
Although there have been several reports on the potential of monocytes as biomarkers for ASD diagnosis (11,43,46), to the best of our knowledge, there are no studies investigating M1 and M2 macrophages. We have induced the polarization of circulating blood monocytes into both M1 and M2 macrophages in vitro and have measured their cytokine expression levels. A series of transcription factors and posttranscriptional regulators play a critical role in M1 and M2 polarization (26), which controls immune responses by balancing in ammation-related cytokines; pro-in ammatory cytokines such as TNF-α were predominantly expressed in M1 macrophage, and anti-in ammatory cytokines such as IL-10 were predominantly expressed in M2 macrophages (47). In this study, the expression levels of IL-10 were not different in either M1 or M2 macrophages between ASD patients and TD individuals. In contrast, TNF-α expression was substantially increased in M1 but not M2 macrophages in ASD patients compared to TD individuals, indicating that, at least, the balance of TNF-α and IL-10 expression in macrophages and that of TNF-α expression in M1 and M2 macrophages tilted towards in ammation in ASD patients.
TNF-α is a pleiotropic cytokine regulating brain development and function (48,49). It has been shown that TNF-α is associated with the pathobiology of schizophrenia, bipolar disorder, major depressive disorder, and ASD (50)(51)(52). An increase in serum TNF-α expression has also been shown in ASD patients from childhood through adulthood (53,54), which is consistent with the current nding that there were no correlations between age and TNF-α mRNA expression levels (Fig. 2D). While TNF-α is expressed in multiple cell types, such as monocytes, T cells, mast cells, NK cells, keratinocytes, broblasts, and neurons, it is predominantly produced by the macrophages and microglia (55,56). Glial TNF-α regulates homeostatic activity-dependent synaptic connectivity, which plays a critical role in the developing brain (57,58). TNF-α derived from microglia is also known to modulate neuronal functions temporally (59). However, brain region-speci city of the effect of TNF-α, the chronicity, and its relevance to ASD symptoms in humans remain unclear. Furthermore, TNF-α has been generally accepted to be too large to cross the brain-blood barrier (BBB), which prevents macrophage in ltration in the brain (60). Nevertheless, TNF-α permeability varies depending on brain regions, i.e., its permeability is higher in the hypothalamus than in the parietal cortex (61). Transporters also help TNF-α to gain access into the brain (62), indicating that macrophage-derived TNF-α could affect brain functions under certain conditions. The breach of the BBB by TNF-α may lead to increased entry of blood macrophages into the brain (63). Furthermore, recent ndings indicate that perivascular macrophages-regularly renewed by monocytes-may be critical players in controlling brain functions (64). Thus, macrophage-derived TNF-α is likely to regulate brain functions and may contribute to the development of neuropsychiatric disorders, as is the case with microglia-derived TNF-α.
TNF-α has also been known to cause autoimmune diseases, such as rheumatoid arthritis (RA). TNF-α inhibitors, such as in iximab, etanercept, adalimumab, golimumab, and certolizumab pegol, have been prescribed for RA treatment (65). An increased number of macrophages are found in the synovial tissue. Macrophages produce TNF-α, and the number of macrophages signi cantly correlates with the degree of joint erosion and the change in the disease-activity score (66,67). Moreover, the M1 or M2 macrophage phenotypes in the synovial uid depend on the disease phase (68). It is, however, of note that no previous studies have shown a difference in TNF-α expression in either circulating blood monocytes or monocytederived macrophages between RA patients and controls. This difference, if found to exist, would indicate that macrophages in the synovial tissue differ from those derived from monocytes and are therefore speci c to RA pathology (69).
TNF-α has also been implicated in the pathobiology of allergic diseases such as atopic dermatitis. While it has been shown that a single-gene polymorphism of TNF-α is signi cantly associated with the diagnosis of atopic dermatitis (70), and that TNF-α expression in the serum is elevated in patients with atopic dermatitis compared with the controls (71), no previous studies have shown the alterations in TNFα expression in blood-circulating monocytes or monocyte-derived macrophages in these patients. We have found that ASD patients tend to suffer from allergic diseases, but not autoimmune diseases, including RA, compared with TD individuals (Table 1). Notably, we did not nd any correlation between the prevalence of allergic diseases and TNF-α expression levels (Suppl. Table 1), indicating that the aberrant expression of TNF-α in monocyte-derived macrophages is likely due to ASD, not allergic diseases.
Considering that TNF-α has been implicated in social stress-induced behaviors (72), the potential of our proposed tool for diagnosing other psychiatric disorders, such as depression, remains to be investigated. This diagnostic tool might be useful for the identi cation of more moderate ASD cases, which are rather di cult to diagnose because TNF-α mRNA expression levels did not correlate with either FIQ or the severity of ASD symptoms based on AQ-J scores (Fig. 2D). Another limitation in our study is the small number of participants; therefore, the effect of sex difference could not be determined. Further, ASD patients had lower full-scale IQ scores and fewer years of education than TD individuals. Lastly, because all participants were Japanese, future studies to assess the generalizability of our ndings to a more diverse population are needed.

Conclusion
Our ndings have revealed that the TNF-α expression ratio of M1/M2 macrophages improves diagnostic speci city up to 96.7%, AUC up to 0.79, and PLR up to 16.55, thus providing a valuable biological tool to aid in the diagnosis of ASD.

List Of Abbreviations
Autism spectrum disorder (ASD) The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.