Experimental animals and grouping

In this study, we used 8-week-old female C57BL/6N mice (Beijing Viton Lever Laboratory Animal Technology Co., Ltd.) which were housed in the Experimental Animal Center of China Rehabilitation Science Institute (SPF-level barrier environment). The experiments were approved by the Animal Experimentation and Laboratory Animal Welfare Committee of Capital Medical University (Reference: AEEI-2021-127).

The experimental mice were grouped as follows. First, a severe injury model was established by forceps. Mice which showed any form of movement in the lower limbs within 1 day of modeling were excluded while the remaining mice were divided into a target gene NeuroD1(ND1) and Neurogenin-2(NG2) overexpression group (SCI-ND1 + NG2 group) and a SCI non-intervention group (SCI group) by applying a random number method. Healthy mice from the same batch were selected as the normal control group (the Control group). A total of 82 mice were enrolled in the group and the specific experimental protocol is shown in Fig. 1A.

Fig. 1

Schematic diagram of the experimental procedure and detection of viral expression in mice. A The same batch of mice was randomly divided into three groups, in which the SCI and SCI-ND1 + NG2 groups were modeled for spinal cord injury and underwent continuous BrdU intraperitoneal injection for two weeks to label new neurons. For the SCI-ND1 + NG2 group, AAV spinal cord in situ injection was performed on day 7 after mapping. For each group, four mice were selected for perfusion sampling for immunofluorescence detection on postoperative days 7, 28, 56, and 70, respectively. On postoperative day 70, four animals were selected from each group for WB, six animals from each group for MEPS, and four animals from each group for EB. BrdU: 5-bromo-2´-deoxyuridine, thymidine and nucleotide analogs, i.p: intraperitoneal injection, MEPS: electroencephalographic evoked potentials, EB: Evans Blue assay, WB: protein blotting assay. B Representative immunofluorescence results of GFAP (cyan), mCherry (red) and EGFP green at the site of injury; (B1, 2) local magnification schematics; the arrows point to several representative co-localized cells. C ND1 and NG2 protein expression levels at the site of spinal cord injury.

Spinal cord injury model and AAV injection

The mice were weighed and placed on a holding pad, anesthetized with 1.5% isoflurane, and positioned at the T10 stage of the spinal column. Then, we shaved fur from the area near T10 on the back, sterilized and exposed the muscles spine layer by layer, and used biting forceps to remove the spinous process and vertebral plate of the T9 segment, thus exposing a circular area of approximately 3 mm in diameter with the T10 spinal cord segment as the center. Next, we prepared a model of transverse spinal cord injury using forceps with a tip of 0.5 mm for 5 s (F12013-10, RWD Life science Co., Ltd, Shenzhen, China). Blackening of the transverse spinal cord region and lower limb convulsions were observed after clamping. Then the tissue was provided with adequate hemostasis and sutured layer by layer with a No. 5 absorbable suture. Following suturing, we applied rehydration fluids (saline 0.5 mL) and antibiotics (penicillin 20000U/only) on a daily basis for 3 d to prevent infection. Postoperative assisted voiding care was performed every 12 h until active voiding was resumed. Thereafter, body weight changes were recorded daily. The mice were euthanized with an overdose of pentobarbital sodium overdose if they lost 20% of their body weight after surgery, had a body temperature below 35 °C, or showed self-injurious behavior.

In this study, we selected NeuroD1 and Ngn2 as target regulators and overexpressed their target genes NeuroD1 and Ngn2 by administering viral vectors (AAV2/9 serotype) containing the GFAP promoter. Supplementary Figure S1 shows the vector map for HBAAV2/9-GFAP-m-Neurod1-3xflag-EGFP used in NeuroD1 overexpression, while Supplementary Figure S2 shows HBAAV2/9-GFAP-m-Neurog2-3xflag-mcherry used for Ngn2. Seven days after SCI modeling, mice were given 1.5% isoflurane anesthesia and fixed in a stereotaxic instrument. Then, the spinal cord tissues were exposed, and an ultra-micro special positioning syringe pump(QSI, Stoelting, Wood Dale,United States) was used to inject 3 μl of AAV virus solution (1 μl per point) close to the midline at the center of the injury site by a 5 μl microinjector needle (Hamilton, Bonaduz, Switzerland). The coordinates for the injection were 5 mm at the head end and 5 mm at the tail end, at a depth of 1 mm, and a speed of 1 μl/min; the needle was held for 3 min after each point was injected. Successful injection was indicated by the lack of liquid flow-out following removal of the needle. We ensured that each mouse was fully hemostatic and then provided sutures layer by layer. Finally, the mice were placed in a warm environment until they woke up.

Behavioral assessment

Behavioral assessments were performed as described previously [20]. To prevent subjective interference by the researchers involved, behavior was assessed in a single-blinded manner, which were conducted by a researcher who was not aware of the animal groupings. The mice were placed in the assessment environment in advance to fully adapt before each assessment. Assessments were carried out during the same time period to minimize the influence of diurnal habit differences on behavioral assessments; the main behavioral assessments are described below.

The Basso mouse scale (BMS)

The BMS is a widely used scale for assessing motor function in a mouse model of spinal cord injury, and classifies mice into 10 levels of motor function (0 referring to complete paralysis and 9 referring to completely normal motor function). For the BMS, mice were placed in an open field for 15 min prior to assessment; this was followed by acclimatization and 4 min of observation.

The open field test

The voluntary motor function of each experimental mouse was evaluated with a TopScanlite device (Clever Sys,Inc. Virginia, United States). Prior to evaluation, mice were placed in the evaluation environment for at least 30 min in advance. During evaluation, mice were placed in a 40 × 40 cm2 open field area and their movement trajectories were recorded for 5 min using a high-speed camera. Motor function and emotional state were determined by analyzing the speed of movement, the length of movement trajectories, and the number of times each mouse crossed the central area.

The DigitGait test

The motor function of experimental mice was evaluated with a rodent-specific digital footprint analysis system (DigiGaitTM, Mouse Specifics Inc., MA. USA). The speed of the running platform was slowly increased to 5 cm/s and at least four consecutive gait cycles were recorded using a high-speed camera (Basler A602 camera 150 fps).

Mechanical pain threshold

The up-down method was used to determine the sensory threshold of experimental mice [21, 22]. This assessment was performed by applying eight different thicknesses (0.02, 0.04, 0.07, 0.16, 1.4, 0.6, 1.0, and 1.4) of Von Frey filaments (Aesthesio, Danmic Global, San Jose, CA). Mice were placed in a plastic cage with a wire mesh bottom that had full access to the foot before evaluation. The mice were acclimated for approximately 30 min until they were allowed to explore the test environment and grooming behavior ceased. A positive response was recorded if the foot retracted, or if the animal licked or jumped up. The force applied at the time of the response was then recorded.

Thermal nociception threshold

Plantar thermal nociception was evaluated using an infrared thermal nociception tester (Bioseb, Florida, USA). During evaluation, the mice were placed in a special cage with a glass bottom. Once mice had finished exploratory behavior and grooming, an infrared light source was focused on the soles, and the system would automatically stop timing and display the retraction time threshold when the hind limbs retracted. Three evaluations were performed for each mouse and the mean value was taken at 10 min intervals. To prevent scalding, the maximum test time was 30 s.

Intraperitoneal injection of BrdU

In this study, BrdU (B5002, sigma) was injected intraperitoneally for two consecutive weeks after SCI to label the newborn cells. First, we prepared a stock solution of 10 mg/mL by fully dissolving the appropriate amount of BrdU with 1X PBS; this was then filtered, de-sterilized, and stored in aliquots at −20 °C to avoid repeated freeze-thawing. BrdU was injected into the mice intraperitoneally at a dose of 10 mg/kg twice daily.

Blood-spinal cord barrier (BSCB) permeability assay

Mice were injected intraperitoneally with 2% Evans Blue (EB, E2129, sigma) dye (0.4 mL per mouse). After 3 h of deep anesthesia with sodium pentobarbital sodium, the mice were perfused and sampled. Spinal cord tissue at the injury site was harvested, and the fluorescence intensity of spinal EB was observed under a fluorescence microscope [23].

Electrophysiological evaluation

Motor-evoked potentials (MEPS) in the lower limbs were recorded using an eight-channel electrophysiological recorder (AD Instruments, New South Wales, Australia) in accordance with method described previously with modifications [24,25,26]. First, mice were anesthetized with 1.5% isoflurane and fixed in a brain stereotaxic apparatus. Then, the head fur was removed and disinfected with an iodophor. The head was then sterilized with an alcohol wipe and the skin of the head was cut using a scalpel. The right cerebral motor cortex was then exposed against the brain atlas using a cranial drill, and two 30-G stimulation electrodes were placed on the cortical tissue in the area of the insertion hole. Recording electrodes were placed on the left gastrocnemius muscle (0.5–1 cm apart) and the reference electrode was placed on the distal part of the ipsilateral lower limb. The stimulation wave width was 1 ms, the current was 7 mA, and the recording electrodes were sampled at 20 kHz. A schematic depicting the assessment is shown in Fig. 10A, B.

Tissue fixation

Mice were deeply anesthetized with sodium pentobarbital (40 mg/kg, i.p.) and fixed in the supine position on the operating table. Then, the skin and sternum were cut open sequentially to expose the heart. A perfusion pump head was then inserted into the aorta from the apical direction, and the left auricle was quickly cut and perfused, flushed with 0.9% saline, and then fixed with 4% paraformaldehyde solution following the removal of blood. After fixation was complete, we removed the injury site (a 1–2 cm area of the spinal cord tissue at the center of the injury site) and stored in 4% paraformaldehyde solution at 4 °C.

For western blotting (WB), spinal cord tissue was removed following deep anesthesia with sodium pentobarbital. The damaged area was centered to a region of approximately 5 mm in the spinal cord tissue which was snap frozen in liquid nitrogen and then stored at −80 °C to await protein immunoblot assays.

Immunofluorescence

Fixed tissues were treated with an antigen repair protocol and endogenous peroxidase was removed. Next, we blocked non-specific binding sites and incubated tissues with a range of primary antibodies: NEUN (Abcam, ab177487, 1:3000), MAP2 (Abcam, ab18383, 1:5000), DCX (Abcam, ab18723, 1:5000), GFAP (Abcam, ab7260, 1:500), and BrdU (Abcam, ab6326, 1:200). Next, the tissues were rewarmed and incubated with appropriate secondary antibodies. DAPI was used to stain cell nuclei and tissue autofluorescence was quenched. Finally, the slides were scanned with a TissueFAXS Q + (Tissuegnostics, Vienna, Austria). Data were analyzed with strata Quest (TissueGnostics, version 7.0.1.176) and ImageJ software.

Western blot

Total protein was extracted from each tissue by weighing an appropriate amount of spinal cord tissue from the site of injury. Protein concentrations were then determined and protein samples were denatured, and separated by SDS-PAGE (G2003, Servicebio). Following electrophoretic separation, proteins were transferred onto membranes. Membranes were then incubated for 30 min with a range of primary antibodies at room temperature: Anti-NeuroD1(Abcam, ab213725), Anti-Neurogenin 2(Abcam, ab109236), Anti-TGF beta 1 (Abcam, ab215715), Anti-SMAD1 + SMAD5 + SMAD9 (Abcam, ab76296), Anti-Smad3 (Abcam, ab40854), Anti-NeuN (Abcam, ab177487), Anti- Smad2 (Abcam, ab33875), p70 S6 kinas (Abmart, T55365), and PP2A alpha + beta (Abmart, T55564). Finally, membranes were incubated with appropriate secondary antibodies and positive signals were detected and imaged with Photoshop (Adobe) and Alapha (Alpha Innotech).

Statistical analysis

Statistical analysis was performed using SPSS (IBM version 26.0) and R software (version 3.5.3). Data are presented in the form of mean ± standard deviation. For the comparison of two groups of data, the data were first tested for normality followed by the chi-squared test. If the data conformed to a normal distribution, then we applied the two independent samples student-t test; if the data did not conform to the normal distribution, then we used a corrected p-value and performed the Mann–Whitney U test. For the comparison of three groups of data, the data were first tested for normality followed by the chi-squared test. If the data conformed to the normal distribution, then we performed analysis of variance (ANOVA). If the data were not normally distributed, then we applied the Kruskal–Wallis H test.