Clinical specimens

Fifty pairs of peripheral blood samples were collected from patients with TBI. in specific, a 1 mL sample of peripheral blood was drawn from patients within 30 min after they had been diagnosed with a TBI in our department, and another 1 mL sample was drawn from these patients after they had received 2 weeks of treatment. The patients ranged in age from 18 to 60 years.

All blood collection vessels were pre-treated with sodium citrate anticoagulant. The collected blood samples were immediately stored in an ultralow-temperature refrigerator at −80 ℃. All patients met the diagnostic criteria for TBI specified in the 11th revision of the International Classification of Diseases. The patients had no history of neurological or psychiatric disorders, substance abuse, severe liver and kidney dysfunction, or recent use of glucocorticoids. In addition, they were all male and none had any infectious diseases.

Small RNA sequencing

Five pairs of matched (i.e., pre-treatment and post-treatment) peripheral blood samples were subjected to small-RNA sequencing. Total RNA was extracted from samples using the TRIzol method and then separated by polyacrylamide gel electrophoresis (PAGE). Subsequently, small RNA was recovered by selecting bands comprising 18–30 nt. Next, each end of a small RNA was joined to a 3′ or a 5′ connector, as appropriate, and the resulting modified small RNA was reverse-transcribed and then amplified by PCR. The resulting DNA was separated by PAGE and then the band comprising approximately 140 bp was dissolved in a solution of ethidium bromide. The resulting mixture was used to complete library construction. The constructed library was tested for quality and yield using an Agilent2100 and ABI StepOnePlus Real Time PCR System (Life Technologies, Santa Clara, CA, United States), and then sequenced on an Illumina NovaSeq 6000 platform. We subjected the original disembarkation data to the following filtering process to obtain high-quality small RNA data: (a) remove the 5′ connector, filter out reads without the 3′ connector, and retain the sequence before the 3′ connector; (b) filter out low-quality reads (those with a mass < 20 or more than one base number) to obtain high-quality reads; (c) filter out reads without an insert fragment and those with an insert fragment of less than 18 nt; (d) filter out reads containing polyadenine (i.e., reads in which more than 70% of the bases were adenine). The resulting clean small RNA sequences were used in subsequent analyses.

Bioinformatics analysis

Basic Local Alignment Search Tool 2.2.25 was used to screen for small RNA sequences with an identity greater than 97%. Those that met this condition were aligned with ribosomal RNAs (rRNAs), small conditional RNAs (scRNAs), small nucleolar RNAs (snoRNAs), small nuclear RNAs (snRNAs), and transfer RNAs (tRNAs) selected from GenBank and Rfam databases. This allowed as many rRNAs, scRNAs, snoRNAs, snRNAs, and tRNAs as possible to be identified and then removed from the samples. Subsequently, due to the unique secondary structure of miRNAs, we were able to align small RNA sequences with a reference genome and then use secondary structure prediction to infer the potential existence of new miRNAs.

As this was an miRNA omics study, principal component analysis (PCA) was used to reduce information consisting of thousands of dimensions (i.e., the expression levels of thousands of miRNAs in the samples) to comprehensive indicators consisting of several dimensions (i.e., principal components). This facilitated inter-sample comparisons and ensured that the information contained in the original data was retained as much as possible. We used R (http://www.r-project.org/) for PCA, and constructed a two-dimensional coordinate plot based on the values of each sample in the first principal component (PC1) and the second principal component (PC2).

We conducted a heatmap analysis of all miRNAs, namely existing miRNAs, known miRNAs, new miRNAs, and the first type of miRNA. In specific, with reference to miRNA expression levels, we performed hierarchical clustering analyses of the relationship between samples and miRNAs, and depicted the clustering results in heatmaps. The hierarchical clustering analyses examined the expression levels of miRNAs in different samples and their corresponding genes. Each column in the heatmap represented a sample, and each row represented an miRNA. The expression levels of miRNAs in different samples were represented by different colours, with red indicating higher expression levels than normal and green indicating lower expression levels than normal. Subsequently, we conducted Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses of the differentially expressed miRNA target genes in each sample. First, we input the miRNA target genes to the GO term mapping database (http://www.geneontology.org/). Then, we calculated the number of miRNA target genes matching each term and thereby obtained a GO function list of miRNA target genes. Next, we applied hypergeometric tests to identify GO entries that were significantly enriched in miRNA target genes compared with the background. Subsequently, we conducted significant enrichment analysis using KEGG PATHWAY to identify the most important biochemical metabolic pathways and signal transduction pathways in which the miRNA target genes were involved.

Cell culture

The cell line HMO6 is derived from human brain microglia and is commonly used to experimentally study the function and characteristics of these cells. HMO6 cells have multi-directional differentiation ability and can differentiate into various types of neural cells, such as neurons, astrocytes, and oligodendrocytes. In the current study, HMO6 cells were cultured in a T25 culture bottle in a medium consisting of a mixture of 0.5% penicillin–streptomycin, 90% high-glucose Dulbecco’s modified Eagle medium, and 10% foetal bovine serum, and the medium was changed every 2–3 days. HMO6 cells with logarithmic growth were seeded onto a six-well plate, and the wells were randomly divided into a control group and an intervention group. The latter group was treated with LPS (150 ng/mL) to establish a cellular model of the inflammatory response, as LPS is an immune stimulator that induces cellular inflammation and immune responses. All cells were maintained in a cell culture incubator in an atmosphere of moist air with 95% relative humidity containing 5% carbon dioxide at 37 ℃.

RNA extraction and qRT-PCR analysis

TRIzol Reagent (Ambion, Carlsbad, CA, USA) was used to isolate total RNA from HMO6 cells. A Multiskan GO1510 spectrophotometer (Thermo Fisher Scientific, Vantaa, Finland) was used to measure the quantity and purity of RNA, and a CFX Connect Real-Time PCR Detection System (Bio-Rad, USA) was used to perform real-time quantitative reverse-transcription polymerase chain reaction (qRT-PCR) analyses. First, the miRNA expression in HMO6 cells was evaluated using a miDETECT A Track™ miRNA qRT-PCR Starter Kit (RiboBio, Guangzhou, China), as described previously [15, 16]. Primers for hsa-miR-122-5p, hsa-miR-193b-3p, and U6 were designed and synthesised by RiboBio. All researchers were blinded to the clinical and pathological diagnoses of the patients from whom the samples had been collected. The target miRNAs and an endogenous control gene (U6) were amplified in triplicate wells. Finally, relative expression levels of miRNAs were calculated using the 2−△△Ct method [17].

Clinical severity assessment

The Glasgow Coma Scale (GCS) is widely used to objectively assess a patient’s level of consciousness on a scale from 3 (completely unresponsive) to 15 (responsive). GCS scores have also been widely used to grade TBI severity and prognosis, and were found to be negatively correlated with the positive rate of detection of TBI by CT. For example, the rate of intracranial injury and the need for neurosurgical treatment of patients with a GCS score of 14 are double that of patients with a GCS score of 15. The relationships between GCS scores and TBI severity are used to classify TBI into three grades: mild TBI (GCS score = 13–15; mortality = 0.1%), moderate TBI (GCS score = 9–12, mortality = 10%) and severe TBI (GCS score < 9, mortality = 40%). This study examined the correlation between the patients’ GCS scores and the levels of expression of TBI-associated miRNAs in their peripheral blood samples.

Receiver operating characteristic (ROC) curve analysis

An ROC curve, also known as a sensitivity curve, is an analytical tool created by drawing coordinate schemata on a two-dimensional plane [18, 19]. Recently, ROC curve-based approaches were tailored to and applied in machine learning and data-mining. In an ROC curve, the horizontal coordinate is the false positive rate (FPR) and the vertical coordinate is the true positive rate (TPR) [20]. Moreover, the area under an ROC curve, commonly known as the area under the curve (AUC), can be used to indicate the performance of the classifier of interest. For example, an ROC curve in which 0.5 < AUC < 1 indicates that the corresponding classifier performs better than a random guess, and the higher the AUC, the higher the accuracy of the classifier. In the current study, ROC curves were used to evaluate whether miRNA expression levels in peripheral blood had clinical utility for the diagnosis of TBI.

Statistical analysis

Statistical analyses were performed using Statistical Products and Service Solutions (SPSS) 17.0 software (SPSS, Chicago, IL, USA). Paired t tests were used to analyse the differences between the expression levels of miRNAs in patients with TBI before they had received treatment and those in such patients after they had received treatment. Student’s t tests were used to analyse the differences between the experimental and control groups. Single-factor analyses of variance were used to compare data between three or more groups. A P value of less than 0.05 was considered to indicate a statistically significant difference.