TENS setting range as follows: the voltage peak value from 0.75–47.20 Vpp; the voltage values from 0.375–23.600 V; the current value 0.3675–23.1280 mA; the electric current density 0.1470–9.2512 A/m²; the frequency 25.00 Hz; stimulating the inner and outer sides of the knee joint in rats.

All animal experiments were approved by the Animal Study Committee of Kunming Medical University (No. KMMU2023MEC203) and were conducted according to the requirements of NIH Guidelines for care and use of laboratory animals. A total of 60 rats (Purchased from Kunming Medical University, Department of Animal, No. SYXK K2020-006), three-month-old SD female rats (200 ± 10 g) were maintained in standard conditions with a controlled temperature (21–23 °C) and a strict 12 h light/dark cycle. All the rats were fed with standard rat chow and allowed free access to distilled water ad libitum at all times during acclimatization and experimental treatment periods.

Before the operation, fasting deprivation for 12 h, water deprivation for 6 h and the rats were anesthetized with intraperitoneal injection of sodium pentobarbital (0.1%, 35 mg/kg; Merck Ltd., Germany). ATFL model was constructed by partial resection of the anterior talo-fibular ligament of the right ankle. After the operation, ampicillin (6,000 IU/kg/day, HYZS Ltd., China) was injected intramuscularly for 3 days.

A total of 36 ATFL rats were randomly divided into the model control (C) group, the low/medium/high (L/M/H) intensity of TENS group (low intensity: 3.16 Vpp, 1.58 V, 1.55 mA; medium intensity: 6.16 Vpp, 3.08 V, 3.02 mA; high intensity: 12.90 Vpp, 6.45 V, 6.32 mA) for 1/2/3 weeks.

Then a total of 24 ATFL rats were divided into eight groups, including the model control (C) group and the low/medium/high (L/M/H) intensity of TENS group treated with TENS (manufactured FMT which induced by TENS) or treated with FMT for 3 weeks.

When the animal model (the weights were 226.00–238.38 g, 249.25–255.00 g, 240.75–245.88 g after 1, 2, 3 weeks of induction) was euthanized by an overdose of anesthetic (150.0 mg/kg sodium pentobarbital, i.m.), the judgement standard of death was pupils dilated, absence of light reflex, breathing and heartbeat cease according to the Guideline for ethical review of animal welfare in Chinese. The intestinal samples were collected from the colon bags, including ATFL.1/2/3W.C.1/2/3, ATFL.1/2/3W.L/M/H.1/2/3. All samples are placed in sterile PBS and analyzed with 16S rDNA sequencing (Novogene Ltd., China).

The intestinal samples were extracted by using the magnetic bead method of the Soil and Fecal Genomic cDNA Extraction Kit (TianGen, China) according to the manufacturer’s instructions. The V4–V5 region of the 16S rDNA gene was amplified by polymerase chain reaction (PCR) with a universal F′ and a unique bar-coded fusion R′ (341 F: ACTCCTACGGGAGGCAGCAG; 806 R: GGACTACHVGGGTWTCTAAT). Then, 15 mL of Phusion^® High-Fidelity PCR Master Mix (New England Biolabs), 0.2 mM primers and 10 ng of genomic DNA template was added to all PCR mixtures. The PCR conditions were as follows: denaturation at 98 °C for 1 min, followed by denaturation at 98 °C for 10 s, annealing at 50 °C for 30 s and extension at 72 °C for 30 s for a total of 30 cycles, and finally maintenance at 72 °C for 5 min. The amplicons were purified using AMPure beads (Axygen Co., United States). Barcoded libraries were generated by emulsion PCR and sequenced in the V4 to V5 reverse direction on a 318 chip using the 400 bp sequencing kit of the Ion Torrent Personal Genome Machine (PGM Co., United States) system according to the manufacturer’s instructions. The output sequences of each sample were no less than 50,000 pairs corresponding to 25,000 clean targets, and informatics methods (strategies: PE101/PE150/PE250/PE300, R language packages: QIIM2, ggplots) were applied.

Briefly, 12 ATFL rats were randomly selected and divided into the four groups as previously and treated with TENS for 3 weeks as the donor rats. During the experimental cycle, the rats were induced to excrete fresh feces (3.0 g) by anal stimulation method at 9 a.m. of a day. Further sterile PBS was added, mixed and evenly dissolved (1:10), and vortex for about 0.5 min until there were no visible fecal particles. The collected samples were centrifuged at 2,000 rpm and 4 °C for 10 min and the fecal residue was discarded. Then the supernatant was centrifuged at 8,000 rpm, 4 °C for 5 min to obtain total bacteria. 0.1 mL of the diluted bacterial solution was cultured and calculated according to the formula viable count = average colony count × dilution ratio/dose volume. Finally, the 1 mL bacterial suspension (10⁹ CFU/mL) was transplanted to recipient rats one time per day for 3 weeks and the model control group was given 1 mL PBS. After the 3 weeks treatment with FMT, the rats were conducted with fasting deprivation for 12 h and the fecal samples were collected. The clearance of intestinal microbiota ahead of FMT with antibiotics metronidazole, neomycin C, ampicillin, vancomycin (50 mg: 50 mg: 50 mg: 25 mg diluted in 200 mL PBS, 200 μL/day per rat for 5 days).

The samples were harvested and fixed in 10% paraformaldehyde (Gefan Biotechnology Co., Ltd., China) for 14 days, dehydrated and gradually decalcified. Five-micrometre-thick sections were prepared using a Leica RM2245 microtome and stained with Hematein and Eosin (HE), Alcian blue (AB) & Alizarin red (Servicebio Co., China).

The Mankin score quantifies the extent of cartilage degeneration by assessing the structure, cell number, matrix staining, and tideline integrity of the cartilage. A higher Mankin score indicates more severe cartilage degeneration as follows:

Micro-computed tomography (Micro-CT) analysis was performed according to recent guidelines⁵⁶ using a SkyScan 1176 micro-CT imaging system (SkyScan, Bruker Ltd., Belgium) with a spatial resolution of 17.75 mm (X-ray source 70 kV/357 mA, 90 kV/270 mA; exposure time 250 ms/360 ms; magnification ×15; and 1.0 mm aluminum/0.1 mm copper filter). Volumetric reconstructions and analyses were performed using the built-in software NRecon 1.6 and CTAn 1.8. For the analysis of bone regeneration, the volume of interest was measured by the average grayscale value at the specific bone position (minimum to maximum degree: 0–255).

The sample were equilibrated in 0.1 Mtris-buffered saline for 10 min. After 1 hour of blockage in phosphate buffered saline (PBS) with 10% normal goat serum, the samples were incubated overnight at 4 °C with anti-bodies (NOD2, DF12125, 1:500, Affinity Co., China; IL-6, GB11117-50, 1:500, Servicebio Co., China). Then the samples were incubated with HRP-conjugated secondary antibody (GB23303, 1:500; Servicebio Co., China) for 1 h at room temperature. The results were enumerated by the ImageJ software (National Institutes of Health, Bethesda, MD, United States).

The athletic ability of the ATFL rats was analyzed by calculating step number of hind leg which was recorded as the number of step number rats ran through a 25.00 cm exercise wheel. The rats were observed once a day for 6 days/week.

A sample of the ankle was taken to keep the ligament-bone junction intact. Before starting the measurement, the length, width and thickness need to be measured (Muromachi Co., Japan), and then after bathing and stretching the fitting with normal saline, the ankle are fixed on the MTS universal materials testing machine to measure compression, bending and tensile related data.

Each rat was placed in a running wheel setup, where a camera recorded its movement at a frame rate of 30.00 Hz for a duration of 1 min. To ensure accurate tracking, key body parts were manually labeled in a set of key frames, including the knee (purple), ankle (cyan), and paw (red). The labeled dataset was used to train a deep learning model in DeepLabCut. Using a convolutional neural network, the model learned the spatial characteristics of each marker, enabling it to recognize and track these points in subsequent videos. Once trained, the model was applied to new videos for pose estimation and motion tracking, outputting coordinate data for each labeled body part. For analysis, the 1-min recording was divided into three 20-s segments. The following metrics were calculated for each segment:

A scAAV2/1-hSyn-EGFP-WPRE-pA virus (S0581-1, Taitool Bioscience Co. China) was injected into the joint cavity of ATFL rats, and the working concentration was 0.44 E13 VG (virus genome)/60 mL. After 3 weeks of treatment with TENS, the colon was collected and sliced at a thickness of 8 mm. Subsequently, the sections were analyzed with a microscope (Eclipse ci, NIS_F_Ver43000_64bit_E&Digital sight DS-FI2, NIKON Co., Japan).

IL-6 (Rattus), F′ CACTTCACAAGTCGGAGGCT, R’ AGCACACTAGGTTTGCCGAG; NOD2 (Rattus), F′ GCAAGCACTTCCACTCCATC, R′ CAACTTGAGGTGCCCAACAT; BMP-2 (Rattus), F′ GAAAACAGCAGCAGTGACC, R′ GGTGGCGTTCATGTAGGAGT; TGF-β (Rattus), F′ TGGGCACTGCTAGAGCCTAT, R′ GCGGAGATCCATACAAAGGA; NF-κB (Rattus), F′ TGTGAAGAAGCGAGACCTGG, R′ TGCTCCTCTATGGGAACTTGAA. Erysipelotrichaceae, F′ GGCGTGGATATGGTAGTGGT, R′ TAGTTCGAGCTCTGGTCTGC; Lachnospira, F′ TCATGCCTCCATTAGTTGTAAGCCT, R′ ATGAAGACTAATAACTCCAAAGAAAAAGTACGACAAC; Eubacterium, F′ ATGTTCAACGTAGGCGACCTGA, R′ TCAGTCGACGGTTCGGTCG; Phascolarctobacterium, F′ AACACATGCAAGTCGAACGG, R′ TTTCTTCATCCTGCCATGCG; Alloprevotella, F′ GTGAAAGTTCGGGGCTCAAC, R′ TCAGCGTCAGTTACACTCCG; 16S rRNA, F′ GTGCCAGCMGCCGCGGTAA, R′ TACCGCGGCTGCTGGCAC.

The following antibodies were used: anti-GAPDH (GB15004, 1:5,000, Servicebio Co., China), anti-β-ACTIN (GB11001-100, 1:5,000, Servicebio Co., China), anti-BMP-2 (GB11252, 1:5,000, Servicebio Co., China), and anti-TGF-β1 (GB115750, 1:5,000, Servicebio Co., China), anti-NOD2 (DF12125, 1:5,000, Affinity Co., China), P65 (GB11997-100, 1:1,000, Servicebio Co., China).

Data was analyzed by GraphPad Prism 8 software (GraphPad, San Diego, CA, United States) and presented as the form of mean ± standard deviation (SD). The p-value of operational taxonomic unit (OTU) was performed by MetagenomeSeq test. The statistical differences were analyzed by one-way analysis of variance (ANOVA) with Tukey’s post-hoc test for multiple group comparisons (SPSS 27.0, United States), and p < 0.05 indicated statistical significance.

After 2 or 3 weeks of treatment with TENS, compared with those in the model control group, the weights in the low and high-intensity TENS groups decreased (p < 0.05, Figure 1A). After 2 weeks of treatment with TENS, compared with that in the model control group, the degree of swelling decreased in the medium and high-intensity TENS groups (p < 0.05, Figure 1B). After 1 week of treatment with TENS, the 25-cm step number increased in the TENS groups (p < 0.05, Figure 1C). After 3 days of treatment with TENS, the degree of varus and valgus ankle angles decreased in the TENS groups (p < 0.05, Figures 1D,E); the inclined plane test degrees increased in the medium and high-intensity TENS groups (p < 0.05, Figure 1F). Compared with those before the ATFL operation, the average step length, step frequency, and average ankle angle degree decreased postoperatively (p < 0.05, Figure 1G). After 3 weeks of treatment with TENS, compared with those in the model control group, the average step length and step frequency were greater in the TENS groups (p < 0.05, Figures 1H,I), and the average ankle angle was greater in the high-intensity TENS group (p < 0.05, Figure 1J). Compared with those in the model control group, the biomechanical distance, max power, and stiffiness were greater in the TENS groups (p < 0.05, Figures 1K–M).

Compared with that in the model control group, the bone mineral density (BMD) was greater in the medium and high-intensity TENS groups (p < 0.05, Figures 2A,B), the trabecular bone number (TB.n) value was greater in the TENS groups (p < 0.05), and the structure model index (SMI) was greater in the medium and high-intensity TENS groups (p < 0.05). Compared with those of the model control group, the Mankin scores were lower in the TENS groups (p < 0.05, Figures 2C,D), the cartilage thicknesses were greater in the TENS groups (p < 0.05, Figure 2E), and the bone mass ratios were greater in the medium and high-intensity TENS groups (p < 0.05, Figure 2F). Compared with that in the model control group, which was analyzed by immunofluorescence (IF), NOD2 expression was lower in the TENS groups (p < 0.05, Figures 2G,H), and IL-6 expression was lower in the TENS groups (p < 0.05, Figures 2I,J). After 3 weeks of treatment with TENS, compared with that in the model control group, NOD2 expression was lower in the TENS groups (p < 0.05, Figure 2K); BMP2 expression was greater in the TENS groups (p < 0.05); TGF-β expression was lower in the TENS groups (p < 0.05); and NF-κB expression was lower in the medium and high-intensity TENS groups (p < 0.05). Compared with that in the model control group, NOD2 expression was lower in the low and high-intensity TENS groups (p < 0.05, Figures 2L,M); BMP2 expression was greater in the medium and high-intensity TENS groups (p < 0.05); and NF-κB expression was lower in the high-intensity TENS group (p < 0.05).

After 3 weeks of TENS treatment, the top 10 most abundant genera in rats with ATFL injury were selected (Figures 3A,B; Supplementary Figures 1A,B). After 3 weeks of TENS treatment, compared with the C.3W.ATFL group, the top 10 genera of the TENS group with increased OTUs included Erysipelotrichaceae_UGG-003, Phascolarctobacterium, Alloprevotella, Eubacterium, and Lachnospiraceae_UCG-010; moreover, the genera with decreased OTUs included Candidatus_Soleaferrea, UCG-005, Rominococcacear, etc. (Figure 3C–E; Supplementary Figure 1D). The top 11 representative sequences of the top 100 genera were obtained via multiple sequence alignment (Supplementary Figure 1C). Compared with those in the C.3W.ATFL group, the median, dispersion, maximum, and minimum values in the TENS groups were different in rats with ATFL injury after 3 weeks of TENS treatment (p < 0.05, Figure 3F). Compared with those in the C.3W.ATFL group, the unifrac distances were 0.164, 0.238, and 0.303, and the unweighted UniFrac distances were 0.374, 0.420, and 0.623 in the TENS groups (Supplementary Figures 1E,F). The separation rates of PC1 and PC2 were 8.74 and 6.39%, respectively, as determined via principal component analysis (PCA; Figure 3G). There were obvious differences between the model groups and the TENS groups, and the separation rates of PC1 and PC2 were 37.82%/13.48 and 20.70%/9.77%, respectively, as determined via principal coordinate analysis (PCoA) (Figure 3H). According to the non-metric multidimensional scaling (NMDS) analysis results, the stresses of the samples were greater than 0.02 (Supplementary Figure 1G).

After 3 weeks of treatment with TENS, compared with those of the control model group, the top 10 functions of all of the groups increased metabolism, genetic information processing, and environmental information processing (among other actions) at level 1 (Figure 3I); moreover, the top 10 functions of all of the groups increased membrane transport, carbohydrate metabolism, replication and repair (among other actions) at level 2. The top 10 functions of all of the groups increased transporters, ABC transporters, general function prediction only (among other actions) (Shi et al., 2023). The microbiota exhibiting changed abundances after treatment with TENS were related to each other, and the predicted functions worked in a cosynergistic manner (Supplementary Figure 1H).

After 3 weeks of treatment with TENS, the positive GFP fluorescence intensity of virus tracing increased in the TENS groups (p < 0.05, Figures 4A,B); The FMT which induced by 3 weeks of treatment with TENS was manufactured for ATFL rats and the dominant intestinal microbiota was analyzed by qPCR as the result of 16S rDNA analysis (p < 0.05, Figure 4C). After 2 weeks of treatment with FMT, the weights in the medium/high intensity of TENS groups decreased compared to those of the model control group (p < 0.05, Figure 4D). After 2 weeks of treatment with FMT, compared with that in the model control group, the ankle swelling degrees decreased in the TENS groups (p < 0.05, Figure 4E). After 2 weeks of treatment with FMT, the step number of 25 cm increased in the low/medium intensity of TENS groups (p < 0.05, Figure 4F). After treatment with FMT, the angle of varus and valgus ankle degree, the degree of inclined plane test was similar (p > 0.05, Figures 4G–I). After ATFL operation, compared with the pre-op, the average step length, step frequency, degree of average ankle angle decreased in the post-op (p < 0.05, Figure 4J). After treatment with FMT, compared with that in the model control group, the average step length and step frequency was similar (p > 0.05, Figures 4K,L); after 2 weeks of treatment with FMT, the average ankle angle increased in the medium/high TENS groups (p < 0.05, Figure 4M). After 3 weeks of treatment with FMT, compared with that in the model control group, the bio-mechanical distance, max power, stiffiness increased in the TENS groups (p < 0.05, Figures 4N–P).

Compared with those in the model control group, the BMD, bone mineral content (BMC), TB.n and trabecular bone thickness (TB.th) in the high-intensity TENS group increased after 3 weeks of treatment with FMT (p < 0.05, Figures 5A,B). Compared with those in the model control group, the Mankin scores were lower in the TENS groups (p < 0.05, Figures 5C,D), and the cartilage thickness and the ratio of bone mass were greater in the high-intensity TENS group (p < 0.05, Figures 5E,F). Compared with that in the model control group, NOD2 expression was higher in the TENS groups (p < 0.05, Figures 5G,H), and IL-6 expression was lower in the TENS groups (p < 0.05, Figures 5I,J). After 3 weeks of treatment with FMT, compared with that in the model control group, NOD2 expression was lower in the low and high-intensity TENS groups (p < 0.05, Figure 5K); BMP2 expression was greater in the medium and high-intensity TENS groups (p < 0.05); TGF-β expression was lower in the low and medium-intensity TENS groups (p < 0.05); and NF-κB expression was lower in the high-intensity TENS group (p < 0.05). After 3 weeks of FMT, compared with those in the model control group, NOD2 and BMP2 expression increased in the medium and high-intensity TENS groups (p < 0.05, Figures 5L,M); TGF-β expression decreased in the low and medium-intensity TENS groups (p < 0.05); and NF-κB expression decreased in the high-intensity TENS group (p < 0.05).