Patients and healthy controls
There were 133 patients with ASD (102 males and 31 females, 4–12 years old) who had attended the G.E. Sukhareva Scientific-Practical Centre for Mental Health of Children and Adolescents and/or the Mental Health Research Center in Moscow between February 2016 and December 2018 enrolled in the study. ASD was diagnosed using the criteria established by the DSM-5 . Additionally, the patients underwent the following tests:
Autism Mental Status Examination (AMSE)  is an 8-item observational assessment that prompts the examiner to observe and document patients’ social, communicative, and behavioral functioning in the context of a routine clinical examination. The AMSE was developed by psychiatrists with autism expertise and is intended to guide clinical judgment in the context of diagnostic decision-making .
Childhood Autism Rating Scale (CARS)  is a scale for the quantification of the severity of autism pathology. The CARS assesses the child on a scale from 1 to 4 in each of 15 dimensions or symptoms (relating to people; emotional response; imitation; body use; object use; listening response; fear or nervousness; verbal communication; non-verbal communication; activity level; level and reliability of intellectual response; adaptation to change; visual response; taste, smell and touch response; and general impressions). Total scores of or above 30 strongly suggest the presence of autism. Children who have a score from 30 to 36 have mild to moderate autism, while those with scores ranging from 37 to 60 points have severe autism [46, 47].
Communication Questionnaire (SCQ) is a parent questionnaire designed for detecting risk for ASD . The SCQ was originally designed as a screening tool for children 4 years of age or older enrolled in epidemiological research or for studies comparing individuals with ASD and other clinical groups .
Exclusion criteria included (1) neurodevelopmental disorders of known etiology (Rett syndrome, fragile X syndrome, or tuberous sclerosis, etc.); (2) clinically significant sensory or motor impairment; (3) significant medical conditions known to affect brain development (neonatal brain damage, genetic and/or metabolic syndromes involving the CNS, severe nutritional or psychological deprivation); (4) and a history of inflammatory disorders and allergies.
As in the other studies [47, 50,51,52,53], the patients were divided into two groups based on their CARS scores. Group I (n = 62) consisted of patients with mild-to-moderate disease severity (the CARS scores of 30–36). Group II (n = 71) included patients with severe ASD (CARS scores more than 36).
A control group included 27 healthy children matched to an ASD group by gender and age. For each subject of the control group, a physical examination, visual analysis of expert-level EEG and comparative EEG-mapping of the brain, and a blood analysis were performed in order to exclude any subclinical condition. None of the controls belonged to the same families as the ASD cases.
The parents of each child signed an informed consent for venous blood sampling (8 mL) and conducting research with the biomaterial sampled from the child.
The design of the study was approved by the Ethics Committee of Research Centre for Medical Genetics.
Blood collection and blood cell population assessment
Blood was collected in EDTA and Li-Heparin tubes by venipuncture. The tubes were centrifuged at 400×g for 10 min at 4 °C to pellet the cells. Plasma was collected, aliquoted, and stored at − 70 °C for up to 3 months. ASD patients and healthy controls who had plasma samples with signs of hemolysis were excluded from the study. Major blood cell populations (WBC, neutrophils, lymphocytes, monocytes, eosinophils, platelets) were assessed in the fresh blood samples using the hemocytometer Abacus 5 “Diatron”. There were no significant differences between ASD children and healthy controls.
Plasma cfDNA concentration
Cells were removed from the EDTA blood samples by centrifugation at 400×g, followed by mixing 3 mL of plasma with 0.3 mL of the solution containing 1% sodium lauryl sarcosylate, 0.02 M EDTA, and 75 μg/mL RNAse A (Sigma, USA). This was incubated for 45 min, then treated with proteinase K (200 μg/mL, Promega, USA) for 24 h at 37 °C. After two cycles of the purification with saturated phenolic solution, DNA fragments were precipitated by adding two volumes of ethanol in the presence of 2 M ammonium acetate. The precipitate was then washed with 75% ethanol twice, then dried and dissolved in water. The concentration of DNA (cfDNA index) was determined by measuring fluorescence intensity on «LS 55» («PerkinElmer», England) spectrometer after DNA staining with the PicoGreen (Molecular Probes/Invitrogen, CA, USA). Relative standard error of the index cfDNA was 10 ± 4%.
Plasma nuclease activity
Nuclease activity levels in Li-Heparin plasma samples were measured by the radial diffusion technique based on radial enzyme diffusion in an agarose gel containing the substrate (dsDNA). Plasma samples (2 μl) were added to radial wells (1-mm diameter) in 1% agarose gel containing 0.05 М Tris рН 7.6, 10 mM magnesium chloride, 1 mM calcium chloride, 25 μg/mL ethidium bromide, and 0.5 mg/mL chicken DNA and were incubated at 37 °C overnight in a wet chamber. The gel was visualized in ultraviolet light on the Gel Doc XR system (BioRad, USA). Enzymatic activity was calculated from a standard curve constructed from the studies on the hydrolysis of chicken DNA by bovine pancreatic DNase (SigmaAldrich) of known concentrations.
This assay can determine a range of 1 fg to 1 pg of DNase in 1 mL serum samples within 30 min. One unit of enzyme assayed corresponds to 1 ng of purified human DNase I. The relative standard error of the index DNase 1 was 15 ± 5%.
Determining the rate of oxidation marker 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-oxodG) in cfDNA samples
Membrane (Extra C) was moistened with 20хSSC solution and dried. The cfDNA (10 ng/mL) solution in ТЕ buffer were applied to the prepared filter in an amount of 2 μL per dot. From each sample, 3–5 dots were applied. Standard samples of oxidized genomic DNA with known content of 8-oxodG were applied to the same filter in order to plot a calibration curve of signal intensity depending on the content of 8-oxodG in the sample. Standard samples of human oxidized genomic DNA were obtained via reaction of DNA with hydrogen peroxide as previously described . The content of 8-oxodG in the control samples was determined by the ESI-MS/MS method using a AB SCIEX 3200 Qtrap machine. The filter was heated at 80 °С under vacuum for 30 min. For carrying out 8-oxodG detection, special equipment was used: an oven for hybridization, which enables temperature adjustment and has a swinging platen for mixing the solution above the membrane. The membrane was blocked (30 min, 30 °С) in a solution of 0.1% fat-free milk, 0.1% gelatin, Tris-HCL buffer, pH 7.5, 0.1 M sodium chloride. The membrane filter was then treated for 30 min (25 °С) with a conjugate of biotin with antibody to 8-oxodG (1 μg/mL) in solution A (0.1 M Tris-HCL buffer, pH 7.5, 0.1 M sodium chloride). The filter was washed (3 × 10 min) with solution A. Then, the membrane filter was treated for 20 min (25 °С) with a conjugate of streptavidin with alkaline phosphatase (1 μg/mL, Sigma) in solution B (0.1 M Tris-HCL buffer, pH 7.5, 0.1 M sodium chloride, 0.005 M magnesium chloride). The filter was washed (3 × 10 min) with solution B. Then, the filter was placed in a solution of substrates for alkaline phosphatase (Tris-HCL buffer, pH 9.5, 0.1 M sodium chloride, 0.005 M magnesium chloride, 4.4 μL/mL NBT and 3.3 μL/mL BCIP). The reaction was conducted in a dark room at 37 °С under visual control of emerging stained violet dots. After the reaction finished, the filter was washed with water and dried in the dark. The dried filter was then scanned. For the quantitative analysis of the dots, a special software was used, namely, Images6 (Research Centre for Medical Genetics, Moscow). The software determines the dot location, measures the nearest background signal, and calculates the integral dot intensity. Signals from several dots for the same sample were averaged, and the mean and standard error are calculated. The 8-oxodG content in a studied sample is calculated using the calibration curve equation. Relative standard error of the index 8-oxodG was 15 ± 5%.
Peripheral blood lymphocyte isolation and culture
Peripheral blood lymphocytes (PBL) were isolated from heparinazed peripheral blood by Ficoll-Verographin density gradient centrifugation. The cells were washed twice in RPMI-1640 medium (ICN, USA) supplemented with 10% heat inactivated donor horse serum, 2 × 10− 3 M HEPES, 2 mM l-glutamine, 2.8 × 10− 6 M 2-mercaptoethanol, and 20 μg/mL gentamycin. The cells were cultivated in flat-bottomed 24-well plates (Costar, USA), which contained 106 cells per well.
Flow cytometry analyses
PBL subsets were evaluated with a flow cytofluorometer (СyFlow, Partek, Germany). Briefly, following isolation, the subsets were washed in PBS, split into tubes, and incubated for 30 min at 4 °C in the dark with fluorochrome-labeled monoclonal antibodies (phycoerythrin (PE)-conjugated anti-human CD3, fluorescein isothiocyanate (FITC)-conjugated anti-human CD4, FITC-conjugated anti-human CD8, FITC-conjugated anti-human CD14, FITC-conjugated anti-human CD19; eBioscience, San Diego, CA). For each analysis, 20,000 events were acquired and gated on CD4, CD8, or CD14 expression and side scatter properties. Samples were first run with single fluorochrome-stained preparation for color compensation.
Flow cytometry was also applied to measure levels of phosphorylated and non-phosphorylated forms of р65 protein (a subunit of NF-kB) in freshly isolated PBL and in PBL treated with oxidized DNA fragments. Primary anti-human р65 (‘Santa Cruz’, USA) and secondary (anti-mouse-FITC, SC-2010, ‘Santa Cruz’, USA) monoclonal antibodies were used according to the common protocol: the cells were washed with 1% albumin solution in PBS, fixed with 3.7% formaldehyde for 10 min at 37 °C, washed off, and permeabilized in 90% methanol at − 20 °C. Then, the cell suspension was incubated with primary antibodies (1 μg/mL) overnight at + 4 °C (1 μg/mL in PBS in the presence of 1% albumin) and, if necessary, with secondary antibodies for 1 h at room temperature in the dark and assayed with a flow cytofluorometer (СyFlow, Partek, Germany). Data on NF-kB content in the lymphocyte nuclei were obtained using a system for cell imaging ‘CyTell’ (‘General Electric Healthcare’, USA).
EDTA plasma samples were analyzed for IL-1β, TNFα, IL-8, IL-17A, IL-10, and IFNγ (CYTOKINE; St.-Petersburg, Russia) using the enzyme-linked immunosorbent assay (ELISA) technique with commercially available kits in accordance with the manufacturer’s instructions.
Measuring gene expression levels using real-time PCR
Expression levels of the genes NFKB1, IL1В, IL8, IL6, ТВР, and GAPDH were measured using real-time PCR. RNA was extracted from the cells using YellowSolve kits (Clonogen, Russia) or Trizol reagent (Invitrogen) as per manufacturer’s instructions (http://tools.lifetechnologies.com/content/sfs/ manuals/trizol reagent.pdf) with the subsequent phenol-chloroform extraction and precipitation with chloroform and isoamyl alcohol (49:1). RNA samples were treated with DNase without RNase activity (Silex, Russia) to remove the DNA contaminations. RNA concentrations were determined by using the Quant-iT RiboGreen dye RNA reagent (‘MoBiTec’, Germany) in a plate reader (EnSpire equipment, Finland) (λexcit = 487 nm, λflu = 524 nm). The reverse transcription reaction was carried out using chemical reagents supplied by the Sileks company (Russia) according to the standard procedure. PCR was carried out using the corresponding primers (Syntol) and the intercalating dye SybrGreen at StepOnePlus instrument (‘Applied Byosystems’, USA). The used primers were as follows (written using the format (F; R)): NFKB1 (5′-CAGATGGCCCATACCTTCAAAT-3′; 5′-CGGAAACGAAATCCTCTCTGTT-3′); ТВР (5′-GCCCGAAACGCCGAATAT-3′; 5′-CCGTGGTTCGTGGCTCTCT-3′); GAPDH (5′-GAAGGTGAAGGTCGGAGTC-3′; GAAGATGGTGATGGGATTTC-3′); IL1B (5′-GGTGTTCTCCATGTCCTTTGTA-3′; 5′-GCTGTAGAGTGGGCTTATCATC-3′); IL8 (5′-ACTGAGAGTGATTGAGAGTGGAC-3′; 5′-AACCCTCTGCACCCAGTTTTC-3′); TNFA (5′-ATCAATCGGCCCGACTATCTC-3′; 5′-GCAATGATCCCAAAGTAGACCTG-3′).
The composition of the PCR reaction mix in a volume of 25 μL were the following: 2.5 μL of PCR buffer (700 mM/L Tris-HCl, pH 8.6; 166 mM/L ammonia sulfate, 35 mM/L MgCl2), 2 μL of 1.5 mM/L dNTP solution, and 1 μL of 30 picomol/L solution of each primer and cDNA. The conditions of PCR were chosen individually for each primer pair. The standard conditions for most primers were the following: after denaturation (95 °C, 4 min), 40 amplification cycles were conducted in the following mode: 94 °C for 20 s, 56 to 62 °С for 30 sec, 72 °С for 30 sec, and then 72 °С for 5 min. The PCR procedures were performed at StepOnePlus (Applied Biosystems, USA). PCR product lengths were checked by electrophoresis on a 1% agarose gel (if a better sensitivity was needed, a PAGE method was used). In addition, amplicon sequence identities for every primer pair were regularly verified by sequencing.
The expression levels of genes of interest were normalized to the expression levels of the respective standard gene (TBP, GAPDH). In experiments on cultivated lymphocytes, gene expression levels were measured in a series of independent tests on cells from different donors. Statistical analysis of the results was performed using a calibrating curve and after taking into account the PCR efficiency; the standard error was 2%.
Preparation of oxidized DNA samples in vitro
The model oxidized DNA fragments (DNAoxy) for the experiments were prepared using combined treatment of selected genomic DNA sample with 300 mM Н2O2 and UV light (wavelength λ = 312 nm, 1.5 min, 25 °C). The modified DNA was precipitated with two ethanol volumes in the presence of 2 M ammonium acetate, washed twice with 75% ethanol, then dried, and dissolved in water. According to data of the ESI-MS/MS method, the content of 8-oxodG in model fragments was 1200 8-oxodG molecules per 106 deoxynucleosides, or 1200AU (this value is equivalent to the actual level of 8-oxodG in the cfDNA samples under severe oxidative stress). It is assumed that the genomic DNA oxidation by H2O2 in vitro eliminates the effects of other possible factors influencing the cfDNA properties, such as changes in the methylation level or other different sequence contents .
The statistical data analysis was conducted using MS Excel, Statistica 6.0, and StatGraph software. The null hypotheses of the absence of the difference between the compared samples were tested with the Mann–Whitney U test. Samples were deemed to be distinct at p < 0.05.