We enrolled 20 POTS children and 20 healthy controls from the pediatric outpatient division of Zhejiang University Medical College's Affiliated Children's Hospital between September 2018 and December 2020. Complete assessment of medical history, physical check-up, and laboratory testing were performed for all 20 children with POTS, including ECG, EEG, blood glucose and biochemical examinations, and cranial CT or MRI, so as to eliminate the cardiac, metabolic, neurologic, and psychogenic etiologies. Meanwhile, the recruited 20 healthy controls were matched as far as possible in terms of age, gender, height, weight, and residence location, who were screened by evaluation of medical history, physical check-up, and laboratory testing, such as ECG, Holter monitoring, and standing test. These 40 children were used as sequencing groups. Another 40 POTS patients and 20 healthy children from Shenmu County Hospital and Shaanxi Provincial People's Hospital were used as validation subjects. The enrolled children were all informed about the research objectives beforehand and agreed to participate. Their parents/guardians have all signed a written form of informed consent. For this research, approval was obtained from the ethics committees of Xi'an Jiaotong University, School of Medicine (No. 2018-84).

Prior to the head-up test, any possible autonomic function-affecting drug was discontinued by the children. The test was conducted in a tranquil room with adequate temperature. A Multi-Lead Physiological Monitor (Dash 2000, General Electric, New York) was utilized for persistent surveillance of the cardiac rate and blood pressure during the test. The protocol constituted 10-min (at least) lying down and subsequent 10-min standing. In the event of a positive response within the 10-min standing, the test was terminated. Children with normal variation ranges of cardiac rate and blood pressure during the test underwent a head-up tilt test on the next day.

Prior to the head-up tilt test, all children were instructed to fast for 4 h (at least) and discontinue any possible autonomic function-affecting drug. A Multi-Lead Physiological Monitor (Dash 2000, General Electric) was utilized for persistent surveillance of cardiac rates and blood pressures of these children, who were laid on a HUT-821 tilt table (Juchi, Beijing, China). Tilting of the table was performed at a 60° angle upon the cardiac rate stabilization. The surveillance of the cardiac rate and blood pressure continued until the emergence of a positive response during the first 10 min. A response would be considered positive if the cardiac rate was elevated by ≥40 beats·min⁻¹, or if the maximum heart rate reaches the diagnostic criteria (≥130 beats/min for 6- to 12-year-olds, ≥125 beats/min for 12- to 18-year-olds); at the same time, blood pressure changes a little (systolic blood pressure drops <20 mmHg, diastolic blood pressure decreases <10 mmHg) and any two of the following symptoms present during tilting: dizziness/vertigo, thoracic tightness, headache, cardiopalmus, pallor, blurring of vision, weariness, or syncope. Any child with a positive response was diagnosed with POTS.

For POTS, its diagnostic criteria were (1) a normal cardiac rate in supine position; (2) over two clinical symptoms on standing, including dizziness/vertigo, aching head, lightheadedness, weariness, pallor, blurring of vision, thoracic tightness, cardiopalmus, trembling hands, and syncope; (3) cardiac rate elevation by ≥40 times/min and (or) the maximum heart rate reaches the diagnostic criteria (≥130 beats/min for 6- to 12-year olds, ≥125 beats/min for 12- to 18-year-olds), and at the same time, blood pressure changes a little (systolic blood pressure drops <20 mmHg, diastolic blood pressure decreases <10 mmHg); (4) relief or mitigation of symptoms by recumbence and ≥1-month persistence of symptoms; and (5) exclusion of other cardiovascular, metabolic, or neurologic lesions (Lin et al., 2017; Wang, 2018).

Prior to blood collection, both POTS and healthy children were instructed to fast. Blood samples (5 ml) were drawn at 8:00 am from subjects in supine position into PAXgene Blood RNA Tubes (BD). Immediately following blood sampling, the tubes were inverted 8–10 times mildly. In every tube, whole blood (2.5 ml) and additives (7 ml) were added. The blood samples were allowed to stand for 4 h at ambient temperature and subsequently preserved at −80°C.

RNA extraction was accomplished from the Blood RNA tubes (PAXgene), followed by quality control. Central analysis was performed on all the quality control outcomes at the end of 2020. After ambient temperature thawing and agitation, the samples were all observed for identifying whether the tubes contained any blood clot. For the automatic RNA extraction from the samples, the Blood RNA kit (PAXgene) was utilized in combination with a QIA cube. Following elution with buffer, the RNA was preserved at −80°C.

Regarding the miRNA library input material, 2 μg of RNA was used per sample. A Small RNA Sample Prep Kit (TruSeq, Illumina, San Diego, USA) was utilized for the creation of miRNA sequencing libraries, and the sequences were assigned to every sample through the addition of index codes. Initially, the total RNA was ligated to a 3′ adaptor, and the resulting mixture was subjected to a 2-min incubation at 70°C for the secondary RNA structure breakage, followed by instant placement on ice. The next step was a 1-h specific ligation of the 3′ adaptor to the miRNA 3′ end by the T4 Rnl2 (T4 RNA Ligase 2) truncated at 28°C. An additional 1-h ligation of the ligated fragment to a 5′ adapter proceeded at 28°C by T4RNA ligase. Utilizing SS IV (SuperScript IV), the synthesis of the first-strand cDNA was accomplished as per the manufacturer's instructions. For the polymerase chain reaction (PCR) amplification, 5-fold Phusion HF buffer, RP1 (RNA PCR primer), RPLX (RNA PCR primer index) primer, dNTP, and Phusion DNA polymerase were utilized. Subsequently, a polyacrylamide gel (6%) was used to purify the PCR products. Thereafter, the 140- to 160-bp-long RNA fragments (corresponding to the length of small non-coding RNA plus 3′/5′ adaptors) were recovered and then dissolved using elution buffer (8 μL). Lastly, a Qubit spectrophotometer was utilized to assess the library concentration, while a Bioanalyzer system (Agilent 2100) and a High-Sensitivity DNA Kit were employed to examine the library quality. The index-coded samples were clustered on a cBot station with a TruSeq PE Cluster Kit v3-cBot-HS (Illumina) as per the manufacturer's instructions. Following cluster generation, sequencing of the prepared libraries proceeded on a HiSeq 2000 platform (Illumina), and single-end 50-bp reads were produced to assess transcriptomics. The sample quality control, experimentation, and data analysis were all accomplished by Genesky Biotechnologies Inc. in Shanghai, China.

Exploiting the miRDeep2's quantifier module, the expression profile was generated for the known mature miRNA, which offers read counts for known miRNAs. The generated expression profiles of raw reads for all the sample replicates were normalized by the trimmed mean of the M-value approach via the Bioconductor's edgeR. The resulting normalized profiles for all the sample replicates were subsequently analyzed by hierarchical clustering and principal component analysis (PCA) to control quality and evaluate inter-sample similarities.

Aided by HTSeq (v0.6.1), the quantity of reads mapped to every miRNA was subjected to normalization. Then, the expected fragments per kilobase million (FPKM) number was computed for every miRNA according to the gene length and the read quantity mapped to miRNA. It is common practice to predict the levels of miRNA expression in FPKM, a value reflecting the effects of both gene length and sequencing depth for the read quantity. DESeq2's R package was exploited to analyze the DEmiRNAs. Utilizing the digital miRNA expression data, this package enables routine statistical determination of differential expression with a negative binomial distribution-based model. For the false discovery rate (FDR) control, the Benjamini–Hochberg method was employed to adjust the resulting P values. The thresholds for defining DEmiRNAs were an adjusted DESeq P of <0.05 and a log2 fold variation of ≥3.

The miRanda and RNAhybrid algorithms were used for the potential target gene recognition for the dysregulated miRNAs. We considered the miRNA target genes recognized by both algorithms.

DEmiRNA joint effects on pathways were evaluated by KEGG analysis based on their associated target genes. The threshold parameters for pathway significance were P-value and FDR. Any KEGG pathway was regarded as significant when P < 0.01.

For placental isolation of total RNA in humans, an RNeasy Micro Kit (Qiagen, Hilden, Germany) was utilized. A Nanodrop spectrophotometer (Wilmington, USA) was used for the quantification of RNA. Initially, reverse transcription of total RNA (20 mg) per sample was accomplished with reverse transcriptase (SuperScript II, Invitrogen)-involving a master mix. Real-time PCR was carried out with a PRISM 7700 Sequence Detector (ABI, Warrington, UK) as per the instructions of the manufacturer. All the Assays-on-Demand primers and probes were procured from ABI. All PCRs were run in triplicate. The reaction system contains forward primers and reverse primers (0.3 μM, respectively) and 2 × Maxima SYBR Green qPCR Master Mix (12.5 μL), cDNA(200 ng), and nuclease-free water reaches 25 μL totally. U6 small nuclear RNA (snRNA) was used as an internal control of miR-21 levels. The expression of miR-21 was normalized to snRNA U6, and relative expression was calculated using the 2^−ΔΔCT method. The primers of miR-21 and U6 were as follows: 5'-TAGCTTATCAGAC TGATGTTGA-3' (hsa-mir21-5p, forward); and 5'-CTCAACTGGTGTCGTGGA-3'(hsa-mir21-5p, reverse); 5'-CTCGCTTCGGCAGCACA-3' (U6, forward); 5'-AACGCTTCACGAATTTGCGT-3' (U6, reverse).

Prior to the examination, the subjects were asked to fast for 8 h, avoid vasoactive agents, high-lipid foods, caffeine, and vitamin C for 24 h, and evade exercise for 4–6 h. In the case of adolescent women, the menstrual period was also avoided. A blood sample (1 ml) was collected at the cubital vein from every subject in the morning (7:00 am to 8:00 am) in a tube containing ethylenediaminetetraacetic acid and aprotinin. After mixing 0.5 mL of plasma with an equal volume of antioxidant buffer, a PXS-270 sulfide electrode (Shanghai, China) was used to determine the plasma H₂S. After washing with distilled water and drying, the electrode was soaked in the sample. Upon stabilization of readout, we documented the electrode potential. The standard curve method was employed for the computation of the H₂S concentration (Zhang et al., 2012).

An HP2500 color Doppler ultrasonographer (Philips Healthcare, Andover, USA) was utilized for the FMD test, whose transducer frequency was 7.5 MHz. The American College of Cardiology guideline was followed in this study to achieve brachial arterial assessment of the endothelium-dependent FMD (Corretti et al., 2002; Donald et al., 2008). The participant was laid down with their arm in a location suitable for the brachial arterial imaging. The transducer was arranged vertically on the upper brachial skin 5–10 cm above the antecubital fossa, while a cuff of mercurial sphygmomanometer was placed on the forearm. For the baseline vascular lumen diameter determination, the baseline brachial arterial image was acquired, while for the blood flow assessment, the pulsed wave Doppler velocity signal was analyzed (detection angle 40°). Next, arterial occlusion was established by inflating the cuff for 5 min to 40–50 mm Hg higher than the systolic pressure. Upon deflation of the cuff, there emerged a high-flow stimulus, and dilation of the brachial artery was noted due to shear stress enhancement. The brachial arterial image and blood flow assessment was attained at the identical position within 2 min of the cuff deflation. The maximum vascular lumen diameter during dilation was documented. The computational formula for FMD (%) was [(maximum diameter – baseline diameter)/baseline diameter]× 100% (Liao et al., 2013).

SPSS (v 13.0; Chicago, USA) was used for analyzing data. The continuous data displayed are means ± SDs, while the categorical data displayed are quantities of cases. The inter-group differences were examined by χ² and t-tests. For the fold variation computation of miRNA expression, the levels of expression were compared between POTS and control groups. The screening criteria for DEmiRNAs adopted herein were a fold variation of ≥2 and P < 0.05. The quantile normalization was conducted by R software (LC-Bio, Hangzhou, China), which was also used for the GO (gene ontology) and KEGG analyses. Differences were regarded as significant when P < 0.05. The datasets for this study can be found in the [ArrayExpress] [https://www.ebi.ac.uk/arrayexpress/]. The accession number is E-MTAB-10961.

The 20 POTS patients consisted of 12 boys and eight girls, aged 10.8–13.9 years, with a mean of (12.7 ± 1.21) years. The 20 healthy controls also comprised 12 boys and eight girls, aged 11.2–14.1 years, with a mean of (13.1 ± 1.13) years. Insignificant inter-group differences were noted regarding age, height, blood pressure, or weight (P >0.05), while the cardiac rate varied significantly (P <0.05) (Table 1).

We identified 135 DEmiRNAs between the two groups. Among them, 49 are upregulated and 86 are downregulated (Figure 1). The 32 miRNAs for which a fold change ≥2 and P < 0.05 are shown in Table 2. Among the miRNAs, 13 miRNAs show a log2 fold change ≥3. They are hsa-miR-4707-3p, hsa-miR-21-5p, hsa-miR-151b, hsa-miR-151a-5p, hsa-miR-548b-5p, hsa-miR-32-3p, hsa-miR-15a-5p, hsa-miR-3613-5p, hsa-miR-574-5p, hsa-miR-18b-5p, hsa-let-7g-3p, hsa-miR-1278, and hsa-miR-1-3p.

The target genes were identified by RNAhybrid and miRanda, and only those identified by both algorithms were considered. In total, 3,820 genes were targeted by the 135 DEmiRNAs (data not shown). The target gene counts of the 13 miRNA for which a log2 fold change ≥3 is shown in Table 3. When the P-value is 0.01 and 0.05, 198 and 481 genes, respectively, are shown to be targeted by the 13 miRNAs. The target gene of hsa-miR-21-5p was SP1 when the P-value is <0.01 (Supplementary Table 1).

GO functional enrichment analyses were performed on the DEmiRNAs. The 10 most enriched functions of GO were chosen from the following three categories: CC (cellular component), BP (biological process), and MF (molecular function), according to the P-value (Figure 2, Supplementary Table 2). The main enriched BP entries are axon development, embryonic organ development, intercellular adhesion via plasma membrane adhesion molecules, and positive ERK1 and ERK2 regulation. Meanwhile, the primary enriched MF entries include actin filament binding and actin binding. As for the chief enriched CC entries, they include the actin cytoskeleton, apical plasma membrane, apical part of the cell, and basolateral plasma membrane.

The pathway enrichment evaluation for the DEmiRNAs in POTS patients was accomplished by utilizing the KEGG database, which revealed prominent enrichment of the DEmiRNAs in 36 pathways (P < 0.05) (Supplementary Table 3). Figure 3 shows a scatter chart that presents the top 10 pathways, of which the PI3K/Akt signaling pathway is closely related to vascular function.

On the basis of the foregoing findings, we recruited another 60 children and adolescents as validation participants to externally confirm the findings. The blood pressure and cardiac rate of POTS patients differed significantly from those of the healthy participants (Table 4). The plasma H₂S and FMD are higher in POTS patients. Furthermore, the fold increase in the expression level of whole-blood miR-21 is higher in POTS patients, while miR-151b, miR-let7g, and miR-1278 were reduced (P < 0.05) (Figure 4).

Next, the validation participants were diagnosed by our standard criteria for defining POTS sufferers as described earlier. The levels of whole-blood miR-21 were determined by qRT-PCR. The results in Table 5 show that according to the defining standard criteria, 40 of the 60 children were diagnosed with POTS. Of these 40 cases, 37 were predicted correctly using whole-blood miR-21 levels. Thus, whole-blood miR-21 was used to diagnose POTS, where the specificity was 100% and the sensitivity was 92.5%.