All animal experiments were performed following ratification by the Animal Ethics Committee of the First Affiliated Hospital of Harbin Medical University (approval no. 2019010) conforming to Principles and Procedures of the National Academy of Animal Health Care Guidelines. All efforts were made to minimize animal suffering and minimize the number of animals used.

A total of 100 adult specified-pathogens free-grade Sprague-Dawley male rats weighing 200–220 g and 12–14 weeks old (Beijing Vital River Laboratory Animal Technology Co., Ltd.) were used in the present study. These rats were housed at 25±2°C and humidity of 50±10% with a 12-h light/dark cycle and given free access to water and food. Of them, 10 out of 100 rats were chosen as sham-operated rats and the remaining rats were adopted for the generation of rat models of SCI. Rats were anesthetized with 3% pentobarbital sodium via intraperitoneal injection [cat. no. P3761; Sigma-Aldrich; Merck KGaA; 30 mg/kg (1 ml/kg according to body weight)]. Following dissection of the paravertebral muscles, laminectomy was performed on the spinal segment (T9-T11). Subsequently, rats with SCI were subjected to an aneurysm clip at the T10 level for 60 sec. Subsequent to trauma, SCI rats immediately underwent intrathecal injection with lentiviral vectors containing miR-128 and ULK1 at the dose of 10 µl and 2 nmol (12). Finally, the incision was sutured layer by layer with silk sutures. Sham-operated rats only underwent laminectomy. All rats were injected with penicillin and analgesics for 3 days post-operation and then received manual bladder expression thrice a day. The success rate of modeling was 89% and no sham-operated rats succumbed (the remaining mice were used for experimental reserve). Following successful modeling, rats experienced spinal cord hemorrhage, delayed hindlimb extension and tail swing. Rats received manual bladder expression once a day until bladder function was restored. The Basso, Beattie and Bresnahan (BBB) scale of rats was performed on miR-128 treated on days 1, 7, 14, 21 and 28 following SCI. At the 28th day post-operation with the completion of behavioral tests, 0.2 ml whole blood was obtained from each rat and the rats were euthanized by asphyxia with 25% capacity volume/min of CO₂ gas displacement rate. The spinal cord tissues were harvested for the related tests. Lentiviral vectors (LVs) including LV-control, LV-miR-128, LV-sponge miR-128 (lentiviral vector-mediated inhibition of miR-128), short hairpin RNA against negative control (sh-NC) and shRNA against ULK1 (sh-ULK1) were constructed by Genomeditech.

The hindlimb motor function of rats was evaluated on days 1, 7, 14, 21 and 28 post-operation using the BBB scoring system (13,14). In an open field (125 cm × 125 cm), after rats adapted to the environment, the locomotor function was observed and scored (5 min). The observation was performed in a double-blinded manner with two trained and non-experimental staff. A total of three average recorded values were taken as the BBB score. BBB was a system that followed the recovery of hindlimb function from a score of 0 (no observed hindlimb movements), to a score of 21 (a normal ambulating rodent).

Bioinformatic websites RAID version 2.0 (http://www.rna-society.org/raid2/), miRDB version 5.0 (http://www.mirdb.org/miRDB/download.html), TargetScanHuman version 7.2 (http://www.targetscan.org/vert_72/) and miRWalk version 2021 (http://mirwalk.umm.uni-heidelberg.de/) were employed to predict the targeted relationship among miR-128, ULK1 and FasL. Venn map online website (http://bioinformatics.psb.ugent.be/webtools/Venn) was adopted for the validation of predicted results. The dual-luciferase reporter gene assay was employed to verify the targeting relationship between miR-128 and ULK1, as well as between ULK1 and FasL. Briefly, pmirGLO firefly luciferase-expressing vectors (Promega Corporation) was used to construct the wild type (WT) and complementary sequence mutated (MUT) site with miR-128 (PGLO-ULK1 WT and PGLO-ULK1 MUT). FasL WT and MUT (PGLO-FasL WT and PGLO-FasL MUT) and ULK1 overexpression (oe) vector and the corresponding NC vector were constructed. The reporter plasmids were co-transfected with the plasmids containing oe and NC of miR-128 and ULK1 into 293T cells respectively using the Lipofectamine^® 2000 transfection reagent (Invitrogen; Thermo Fisher Scientific, Inc.) based on the manufacturer's protocol. At 48 h post-transfection, Cells were lysed and centrifuged at 6,500 × g for 1 min at 4°C, followed by harvesting of the supernatant. The dual-luciferase reporter assay system (E1910, Promega Corporation) was employed to detect the luciferase activity. Relative luciferase activity=firefly luciferase/Renilla luciferase.

The enrichment of FasL by ULK1 was evaluated by a ChIP kit (EMD Millipore). Briefly, 1% formaldehyde was added to fix the 70–80% confluent cells at room temperature for 10 min to allow the formation of DNA-protein cross-links in the cells. Subsequently, cells were lysed in 1 ml lysate buffer. Next, sonication (10 sec per time) was performed at intervals of 10 sec (15 cycles) to fractionate large fragments into appropriate size, followed by centrifugation at 6,500 × g for 1 min at 4°C. The supernatant was then collected, divided into three tubes and then added with the positive control antibody RNA polymerase II (cat. no. ab36010; 1:1,000; Abcam), the NC antibody of normal mouse immunoglobulin G (IgG; cat. no. ab6789; 1:1,000; Abcam), or the target protein-specific rabbit anti-ULK1 antibody (cat. no. ab2851; Abcam), followed by overnight incubation at 4°C. Protein Agarose/Sepharose was supplemented to precipitate the endogenous DNA-protein complex. Subsequent to transient centrifugation at 6,500 × g for 1 min at 4°C, the supernatant was discarded and the non-specific complexes were washed. The complex was de-crosslinked at 65°C and the DNA fragment was recovered by phenol/chloroform extraction and purification. The binding of ULK1 to the FasL promoter region was assessed by the specific primers in FasL promoter region using western blotting as detailed in the following section.

Spinal cord tissue located at 0.5 cm above and below the injury was attained 28 days post-surgery. The spinal cord tissues were fixed in 4% neutral paraformaldehyde (PFA) for 24 h at room temperature, routinely dehydrated with graded ethanol series (100% for 5 min, 95% for 3 min, 90% for 3 min), cleared using xylene and, embedded in paraffin, and sectioned into 4 µm-thick sections. The sections were then dewaxed twice with xylene (each time for 5 min) and rehydrated, followed by 5-min hematoxylin staining and 20-sec differentiation with hydrochloric acid alcohol. Subsequent to 5-min l% eosin staining, morphological changes in spinal cord tissue structure were observed under a light microscope (magnification, ×200; Olympus BX51; Olympus Corporation).

For Nissl staining the sections were stained with 1% toluidine blue staining solution for 10 min, followed by conventional dehydration using gradient ethanol and xylene clearing and neutral resin mounting. Quantitative analysis of the number of surviving neurons was implemented under a light microscope (magnification, ×200; Olympus BX51; Olympus Corporation). In all ~3–5 rats were randomly attained in each group for statistical analysis.

The frozen sections were air-dried at room temperature, fixed with −20°C pre-chilled acetone for 6 min and water-bathed with 50% formamide/2X citrate hybridization solution at 65°C for 2 h. Subsequently, the sections were washed using 2X saline sodium citrate solution for 3 times, followed by 30-min immersion in 0.3% Triton at ambient temperature. The sections were washed in 2 M HCl and finally washed twice in 0.1 mol/l boric acid buffer (pH 8.0; 5 min/time). Finally, the sections were blocked with 10% sheep serum at 37°C for 1 h. Each section was then added dropwise with neuronal nuclear antigen (cat. no. ab128886), neurofilament-200 (cat. no. ab82259), glial fibrillary acidic protein (1:100; cat. no. ab7260; all from Abcam) for overnight incubation at 4°C. The 0.01 M phosphate-buffered saline instead of the primary antibody was adopted as a NC. IgG was then added dropwise to the sections for 45-min incubation at room temperature. The sections were mounted with anti-fluorescent quencher and observed under Leica SP5 confocal microscope (magnification, × 200; Leica Microsystems GmbH).

A one-step TUNEL apoptosis assay kit (Roche Diagnostics GmbH) was employed to determine the morphological changes of cell apoptosis 28 days post-operation. Apoptotic cells were TUNEL positive in a Nikon ECLIPSE Ti inverted microscope (Nikon Corporation) and the nuclei of apoptotic cells were stained brownish-yellow. A total of 10 high-power fields were randomly observed, followed by calculation of the mean number of positive cells using BI-2000 image-analysis system (BI-2000; Visitech Systems).

Total RNA extraction from spinal cord tissues of rats from each group was implemented as per the instructions of TRIzol^® (Thermo Fisher Scientific, Inc.). All primers were synthesized by Augct Biotechnology Co., Ltd. (Table I) according to the manufacturer's protocols. Fluorescence quantitative PCR was performed in accordance with the protocols of SYBR Premix Ex TaqTM II kit (cat. no. RR820A; Xingzhi Biotech, Co. Ltd.). The reaction system was 50 µl: SYBR^® Premix Ex Taq II (2X) 25 µl, PCR upstream primer 2 µl, PCR downstream primer 2 µl, ROX reference dye (50X) 1 µl, DNA template 4 µl, and ddH₂O 16 µl. PCR was performed on ABI PRISM 7300 (Applied Biosystems; Thermo Fisher Scientific, Inc.). The reaction conditions were as follows: 10 min of pre-denaturation at 94°C, 15 sec of denaturation at 94°C and 30 sec of annealing at 55°C. After 40 cycles, amplification was performed for 1 min at 72°C. The relative expression was calculated using the 2^−ΔΔCq method (15) and standardized by U6 (for miR-128) and GAPDH (for ULK1; cat. no. abs830032; Absin Bioscience, Inc.). The experiment was performed three times.

The total protein content was extracted from the tissue according to the manufacturer's protocols provided by high-efficiency radioimmunoprecipitation assay-like buffer (cat. no. R0010, Beijing Solarbio Science & Technology Co., Ltd.). The estimation of protein concentration was implemented using a bicinchoninic acid kit (cat. no. 20201ES76; Shanghai Yeasen Biotechnology Co., Ltd.). The protein was uploaded (30 µg per well), separated by 10% polyacrylamide gel electrophoresis, and electro-blotted onto a polyvinylidene fluoride membrane. Subsequent to 1-h membrane blocking with 5% bovine serum albumin (Sigma-Aldrich; Merck KGaA) at ambient temperature, overnight membrane incubation was performed at 4°C with diluted primary antibodies to Cleaved caspase-3 (cat. no. ab2302; 1:1,000), FAS cell surface death receptor (Fas) (cat. no. ab82419; 1:1,000), FasL (cat. no. ab15285; 1:1,000) and GAPDH (cat. no. ab181602; 1:1,000). The membrane was washed thrice with Tris-buffered saline with 0.05% Tween 20, each time for 5 min and subsequently re-probed with horseradish peroxidase-tagged goat anti-rabbit IgG (cat. no. ab150077; 1:10,000 Abcam). An electrogenerated chemiluminescence (ECL) developer (Shanghai Lianshui Biotechnology Co., Ltd.) was adopted to develop the membrane. ImageJ 1.48u software (National Institutes of Health) was applied for the quantitative protein analysis. Relative protein expression was analyzed by the ratio of the gray value of the target band to that of GAPDH.

The 0.2 ml orbital whole blood was attained from each rats and centrifuged at 1,500 × g at room temperature for 10 min to collect the serum of rats. ELISA was performed in accordance with the manufacturer's protocols (cat. no. ER2101; Wuhan Msk Biotechnology Co., Ltd.) and a standard curve was drawn to determine the levels of inflammatory factors IL-1β, IL-6 and TNF-α in the serum of rats.

Primary spinal cord neurons were isolated from healthy SPF-level SD rats (Beijing Vital River Laboratory Animal Technology Co., Ltd.) on day 14–17 after gestation. Briefly, spinal cords were dissected in medium without serum and trypsinized (0.125%) at 37°C for 15 min. The cell suspension was prepared and incubated into 24-well plates coated with 0.1 g/l polylysine at a cell density of 4×10⁵ cells/ml under a humidity-saturated environment with 5% CO₂ at 37°C. The medium was renewed every 2 days. On the 3rd day of culture, cytarabine was supplemented at a final mass concentration of 5 mg/l to inhibit the growth of mixed cells. Following 5 days of culture in medium without serum, the neuronal cells were treated with 30 mm H₂O₂ for 24 h to construct neuron injury models. Then, 24 h before modeling, cells were transfected with inhibitor NC (25 nM), miR-128 inhibitor (25 nM), oe-NC (25 nM), or oe-ULK1 (25 nM) for 48 h using the HG transgene reagent (Genomeditech), respectively The above vectors were all constructed by Genomeditech. The sequences are shown in Table II. The neuronal cells were then incubated for 48 h after transfection for subsequent detection.

Cells were prepared into 1×10⁴ cell/ml cell suspension using Dulbecco's modified Eagle's medium containing 10% fetal bovine serum (BioSer) and cultured in a 96-well plate. Then, 8 wells (100 µl) were set for each group and culture was implemented in a 37°C cell incubator with 5% CO₂. Subsequent to 24-h culture, the plate was removed and 10 µl MTT (Sigma-Aldrich; Merck KGaA) was added into each well for 2-h culture. The optical density value at 570 nm was read using an enzyme-linked immunoassay (cat. no. NYW-96M; Beijing. Noahway Instruments Co., Ltd.). The cell survival rate was calculated as follows: Cell survival rate=(experimental group OD-blank group OD)/(control group OD-blank group OD)x100%.

Cells were trypsinized in the absence of ethylene diamine tetraacetic acid and attained into a flow tube. Subsequent to cell centrifugation, the supernatant was discarded. As per the protocols of Annexin V-fluorescein isothiocyanate (FITC) apoptosis detection kit (cat. no. C1065; Beyotime Institute of Biotechnology), Annexin V-FITC, phosphatidyl inositol (PI), hydroxyethyl piperazine ethylsulfonic acid buffer was added to Annexin V-FITC/PI staining solution in the proportion of 1:2:50. Cells at a concentration of 1×10⁶ cells per 100 µl of the staining solution were resuspended, incubated at ambient temperature for 15 min in the dark and added with 1 ml of 4-(2-hydroxyethyl)-1-piperazineëthanesulfonic acid buffer. Flow cytometer (Bio-Rad ZE5; Bio-Rad Laboratories, Inc.) was employed to determine the excitation wavelength at 488 nm. The excitation wavelength at 525 or 620 nm was employed to detect FITC or PI fluorescence for cell apoptosis, respectively. The samples (n=10) were randomly selected in each group. Apoptosis index (AI)=number of apoptotic cells/(number of apoptotic cells + normal cells).

Statistical analysis was performed using SPSS 21.0 (IBM Corp.). Measurement data were summarized by mean ± standard deviation or inter-quartile range. An unpaired t-test was performed for the variance of data consistent with the normal distribution and homogeneity of variance. One-way analysis of variance (ANOVA) was conducted for comparison among multiple groups, followed by Tukey's post-hoc test. Data at different time-points were compared using repeated-measures ANOVA, followed by Bonferroni post-hoc test. P<0.05 was considered to indicate a statistically significant difference.

Rat models with SCI were induced and the BBB score was calculated from the normal rats and SCI rats on day 1, 3, 5, 7 and 14 after modeling (Fig. 1A). A markedly lower BBB score was seen in rats with SCI when compared with normal rats. The results of H&E staining (Fig. 1B) showed that there were no significant pathological changes in normal rats, whereas rats with SCI exhibited distinct white matter edema in the spinal cord, microvascular rupture or swelling in the gray matter with focal bleeding, degeneration, and necrosis. Nissl staining (Fig. 1C) and immunofluorescence (Fig. 1D) indicated that neurons and nestin-positive cells were both markedly reduced in rats with SCI compared to normal rats. ELISA (Fig. 1E) demonstrated that the levels of IL-1β, IL-6 and TNF-α in serum notably rose in rats with SCI in comparison with those in normal rats. TUNEL results (Fig. 1F) revealed that apoptotic cells in rats with SCI were significantly increased in rats with SCI in comparison with those in normal rats. Meanwhile, western blot analysis demonstrated that the expression of apoptosis-related factors, Fas, FasL and cleaved caspase-3 rose markedly in rats with SCI compared with those in normal rats (Fig. 1G). The above results implied that neurons in rats with SCI exhibited promoted apoptosis, accompanied by inflammation. RT-qPCR (Fig. 1H) demonstrated that miR-128 expression in rats with SCI markedly declined compared with that in normal rats.

To confirm whether the changes in miR-128 expression affected the apoptosis and inflammatory response of rats with SCI, vectors including LV-control, LV-miR-128, or LV-sponge miR-128 were first constructed and injected into rats with SCI. As shown in Fig. 2A, RT-qPCR revealed that SCI rats injected with LV-sponge miR-128 exhibited markedly downregulated miR-128 levels, but an opposite trend was observed in SCI rats injected with LV miR-128 compared with SCI rats infected with LV-control (P<0.05). SCI Rats expressing LV-miR-128 showed a significantly higher BBB score while rats transfected with LV-sponge miR-128 showed significantly lower BBB scores in comparison with rats with SCI expressing LV-control (P<0.05; Fig. 2B). Results from H&E staining revealed that SCI rats infected with LV-miR-128 showed alleviated pathological changes but SCI rats infected with LV-sponge miR-128 exhibited aggravated pathological changes when compared to rats infected with LV-control (P<0.05; Fig. 2C). The results of Nissl staining (Fig. 2D) and immunofluorescence (Fig. 2E) indicated that the number of normal neurons and nestin-positive cells of SCI rats injected with LV-miR-128 was significantly increased, but an opposite trend was observed in SCI rats injected with LV-sponge miR-128 in comparison with rats injected with LV-control (P<0.05). The results of ELISA (Fig. 2F) indicated that the serum levels of IL-1β, IL-6 and TNF-α in SCI rats infected with LV-miR-128 were significantly lower in comparison with those transduced with LV-control, whereas SCI rats infected with LV-sponge miR-128 showed a significant increase (P<0.05). TUNEL staining (Fig. 2G) demonstrated that SCI rats infected with LV-miR-128 showed a strikingly reduction in cell apoptosis while SCI rats infected with LV-sponge miR-128 showed a notable increase in apoptosis compared to rats injected with LV-control (P<0.05). Western blot analysis (Fig. 2H) demonstrated that Fas, FasL and cleaved caspase-3 expression in SCI rats infected with LV-miR-128 declined, while an opposite trend was observed in SCI rats infected with LV-sponge miR-128 in compared with those transduced with LV-control (P<0.05). These above results suggested that the highly expressed miR-128 could inhibit apoptosis and the inflammatory response in SCI rats.

RNAInter, miRDB, TargetScan and miRWalk were employed to predict the target genes of miR-128. The results revealed that there were three genes appearing in the intersection, which consisted of ULK1, ABCB9 and SGMS1 (Fig. 3A). To further verify the predicted results, vectors containing LV-control, LV-miR-128, or LV-sponge miR-128 were injected into SCI rats. RT-qPCR and western blot analysis demonstrated that ULK1 expression was upregulated in rats with SCI (P<0.05; Fig. 3B and C). The binding site between ULK1 and miR-128 was predicted using the bioinformatic website TargetScan Human version 7.2 (Fig. 3D). Results from dual-luciferase report assay confirmed that luciferase activity in the ULK1 WT was inhibited by miR-128 mimic, while ULK1 MUT was not affected (P<0.05; Fig. 3E). Furthermore, RT-qPCR and western blot analysis indicated that ULK1 expression was markedly repressed in SCI rats infected with LV-miR-128 but was elevated in SCI rats injected with LV-sponge miR-128 (P<0.05; Fig. 3F and G). All these results suggested that miR-128 could target and repress the expression of ULK1.

To further explore the effect of the changes in ULK1 expression on SCI rats, vectors containing sh-NC or sh-ULK1 was constructed and injected into SCI rats (Fig. 3H and I). Results showed that SCI rats transfected with sh-ULK1 exhibited significant higher BBB scores (P<0.05; Fig. 3J). In addition, H&E staining revealed that SCI rats transfected with sh-ULK1 exhibited less severe pathological changes (P<0.05; Fig. 3K).

The results from Nissl staining (Fig. 3L) and immunofluorescence (Fig. 3M) demonstrated that the number of normal neurons and nestin-positive cells were notably increased in SCI rats transfected with sh-ULK1 (P<0.05). ELISA demonstrated that the levels of IL-1β, IL-6 and TNF-α in SCI rats transfected with sh-ULK1 exhibited a notable decline (P<0.05; Fig. 3N). The results of TUNEL (Fig. 3O) showed that reduced apoptosis was observed in SCI rats transfected with sh-ULK1 (P<0.05). Western blot analysis and the results showed that protein expression of Fas, FasL and Cleaved Caspase-3 in SCI rats transfected with sh-ULK1 was significantly lowered (P<0.05; Fig. 3P). Taken together, these results suggested that miR-128 could inhibit the expression of ULK1, thus relieving SCI in rats.

To verify the miR-128/ULK1 regulatory mechanism in vitro, fetal rat spinal cord neuronal cells were isolated and cultured. The isolated spinal cord neuronal cells began to adhere following 3–6 h of culture, accompanied by protruded cells. Following 2–3 d of culture, the cell began to bulge forcefully and the connection between cells became reticular. Spinal cord neuronal cells were successfully isolated and validated by immunofluorescence testing for nestin antibody staining (Fig. 4A and B). The neuronal cells were transfected with inhibitor NC + sh-NC, miR-128 inhibitor + sh-NC, or miR-128 inhibitor + sh-ULK1. RT-qPCR and western blot analysis (Fig. 4C and D) demonstrated that miR-128 expression in neuronal cells transfected with miR-128 inhibitor + sh-NC was significantly reduced, while the mRNA and protein expression of ULK1 was significantly increased in comparison with neuronal cells transfected with inhibitor NC + sh-NC. In addition, miR-128 expression in neuronal cells transfected with miR-128 inhibitor + sh-ULK1 remained unchanged, whereas mRNA and protein expression of ULK1 were both downregulated in comparison with those in neuronal cells transfected with miR-128 inhibitor + sh-NC. The results of MTT assay (Fig. 4E) demonstrated that the neuronal cell viability was markedly reduced by miR-128 inhibitor, which was negated by additional treatment with sh-ULK1. Flow cytometry demonstrated that the apoptosis rate of neuronal cells was augmented by miR-128 inhibitor, which was annulled by further sh-ULK1 treatment (Fig. 4F). As reflected by western blot analysis, the protein expression of Fas, FasL and cleaved caspase-3 was significantly enhanced in neuronal cells following miR-128 inhibitor treatment, which was abrogated by additional sh-ULK1 treatment (Fig. 4G). Meanwhile, the ChIP assay revealed that there was significant enrichment of ULK1 and FasL (Fig. 4H). Results from the dual-luciferase report assay showed that the cells transfected with oe-NC or oe-ULK1 had no significant difference in fluorescence intensity after co-transfection with FasL MUT. Compared with the cells transfected with oe-NC, cells transfected with oe-ULK1- and FasL WT exhibited an increase in fluorescence intensity (Fig. 4I). Together, the elevation of miR-128 could downregulate ULK1, thereby attenuating SCI via FasL downregulation.