Animals

Male C57BL/6J mice (weighing 22.74 ± 0.63 g, 6 weeks old) were purchased from Jiangsu Jicui Yaokang Biotechnology Co., Ltd. (Nanjing, China) and reared in the Laboratory Animal Center of Shanghai University of Sport (SPF grade, license No.: SYXK 2020-0007). The mice were fed separately in individually ventilated cages, with 5 individuals per cage, at an appropriate temperature (24 ℃ ± 1 ℃), humidity, and a 12/12 h light/dark cycle. Our preliminary experiments involved one-week adaptive feeding of 6-week-old female (n = 20) and male (n = 20) C57BL6/J mice before their random allotment to the sham or SNI model groups (n = 10/group). Pain behavioral tests, including mechanical withdrawal threshold (MWT) and cold allodynia response time, were conducted pre-surgery and on postoperative days 3, 7, 14, 21, 28, 35, and 42. Male and female mice exhibited similar pain behaviors (mechanical and cold allodynia) in the SNI model, with no statistically significant differences (Supplementary Figures S1b–S1c). In consideration of the variations in the estrous cycle of female mice and the convenience of implementation in cage feeding and swimming interventions, male C57BL/6 mice were uniformly used to streamline the experimental design and reduce potential variability.

MiR-183 knockout (KO) mice (provided by Professor Peng Chang-geng from Tongji University) were designed by Kunshan Pengji Kaifeng Biotechnology Co., Ltd., and prepared by Shanghai Nanfang Model Biotechnology Co., Ltd. The Cas9/gRNA gene-editing strategy was applied, and miR-183 mutant mice were constructed and bred to obtain miR-183 KO homozygous mice. Our research group was gifted with two 12-week-old male miR-183 KO homozygotes. Given the long breeding cycle required for miR-183 KO mice and the need to meet the sample size of the experimental design, male and female miR-183 KO mice (weighing 22.18 ± 1.33 g, 6 weeks old) were selected for the experiment. The Scientific Research Ethics Committee of Shanghai University of Sport approved all animal experimental protocols (NO. (027)2023DW010).

Animal models

Based on a previous description, a mouse model of SNI-induced NP was established for the NP models [22]. First, the mice were anesthetized with a gas mixture of 5% isoflurane (R510-22-10, RWD, Shenzhen, China) and oxygen in a box. Subsequently, the animals were covered with an anesthetic mask, and their anesthetic pathways were switched. The dose was reduced to 1.5% isoflurane mixed with oxygen. Then, a longitudinal incision was made 2 mm into the middle and lower femurs of the left hind limb, and the sural, tibial, and common peroneal nerves (branches of the sciatic nerve) were visualized. The two thick ends of the common peroneal and tibial nerves were tightly ligated with 6 − 0 silk sutures. Afterward, the nerve tissues were cut 1–2 mm distal to the ligation, and the muscles and skin were subsequently aligned and sutured. Care was exercised during the operation to avoid damaging the surrounding blood vessels. The sham group underwent a surgical procedure similar to that of the SNI group. Only the sural, tibial, and common peroneal nerves were exposed without nerve ligation and subsequently sutured.

Swimming exercise protocol

Mice were scheduled for swimming intervention after unilateral SNI modeling, and an exercise protocol based on a previous description was implemented [23, 24]. Five mice were placed simultaneously in a clear plastic container measuring 45 cm in length, 35 cm in width, and 21 cm in height, and they were allowed to swim. The container had at least 17 cm of warm water (34–36 °C). The mice of the swimming group were trained for swimming adaptability 1 week before SNI modeling. During the adaptive phase, the mice swam for 10 min on day 1, and the duration was increased to 10 min per day until 60 min on day 6. Formal swimming commenced on postoperative day 3 and continued for 6 weeks, comprising 30 sessions. For the first 6 sessions, the swim duration was gradually increased from 30 min to 50 min, and the remaining 24 sessions lasted 60 min (Supplemental Figure S1a). During training, the mice were gently touched on the back of the neck, or the water was gently stirred to prevent floating behavior [24]. After the training, the mice’s hair was dried with a warm air blower before they were placed back in the cages[25].

Behavioral tests

Mechanical hypersensitivity and cold allodynia assessments were conducted as described at baseline (preoperative) and on postoperative days 3, 7, 14, 21, 28, 35, and 42 (Fig. 1a). A pain behavior adaptation test was also performed during the adaptive feeding period. The mice were placed in a bright, quiet behavioral chamber at about 25 °C and acclimated for 30 min. Then, they were placed on a perforated metal grid chassis (the grid size was 0.5 × 0.5 cm²) with a cage covered by 8 × 8 × 5 cm³ of breathable plastic boxes to separate the mice. The MWT [26, 27] was identified to evaluate the mechanical allodynia by treating the plantar surface of the hind paw using the Von Frey fiber measurement kit (Aesthesio, Danmic Global, USA). Appropriate fiber ends were obtained to stimulate the lateral plantar surface of the mouse’s hind paws. During the test, the fiber filaments were gently placed vertically on the sole of the mouse’s foot, and the hand was slowly lifted upward until the fibers bent. The MWTs of both hind paws were recorded separately when the mice withdrew their hind paws three times or licked the soles of their feet by raising their legs. The stimulation force started from small to large, generally beginning at 0.16 g, and was repeated 5 times. Each stimulation had a 5-minute interval. Cold allodynia was assessed using an acetone test [27, 28] on the hind paw. Approximately 30 µL of acetone was added to a 1 mL syringe, and 15 µL of acetone was dropped onto the surface of the hind paw’s plantar surface each time to avoid hitting the mouse’s legs or body. The total response time of licking, withdrawing/guarding, or flicking the hind paw during the 1 min was recorded. A significant response time (seconds) indicated high cold allodynia. To ensure the validity and consistency of the results and limit confounding factors, we evaluated each mouse twice on the same hind paw, with a 5-minute interval between evaluations. The average of two tests was calculated to quantify the results. Mice were randomly assigned to the swimming or control groups after surgery. The randomization sequence was generated using a computer-based random number table. Investigators conducting behavioral tests were blinded to the post-surgery intervention assignment (whether mice underwent exercise). However, the distinction between the SNI and sham groups could be inferred from the morphological changes in the operated hindpaw, making complete blinding infeasible. We acknowledge this as a limitation of our study.

Fig. 1

Swimming exercise alleviated allodynia induced by SNI and reduced FTO mRNA and protein levels in the damaged L4-L6 DRGs. a Experimental schedules. Mice were scheduled for 6 weeks of swimming intervention after unilateral spared nerve injury (SNI) modeling. L4-L6 DRGs on the ipsilateral side were collected at the end of 6-week swimming. b Comparison of mechanical withdrawal threshold (MWT) and (c) cold allodynia response time on the ipsilateral side in each group. Behavioral tests were conducted pre-surgery and on postoperative days 3, 7, 14, 21, 28, 35, and 42. N= 15/group, data presented as mean ± SEM, analyzed by two-way repeated measures ANOVA with Bonferroni correction for multiple comparisons. ##: p < 0.001, indicating a comparison between the SNI and Sham groups. **: p < 0.001, indicating comparison between the SNI + Swim and SNI groups. d Immunofluorescence staining found the co-localization of FTO and DRG neurons. Green represents FTO, red represents NeuN (neuronal cell body marker), and blue represents DAPI. N = 3/group, scale bar = 20 µm, magnification = 200×. Green represents FTO, red represents CACNA2D2, and blue represents DAPI. Meanwhile, (e) the co-localization of FTO and CACNA2D2 was also found. N = 3/group. f Average fluorescence intensity of FTO and CACNA2D2 in each group. Unilateral SNI increased the fluorescence intensity of FTO and CACNA2D2 and decreased after swimming. N = 3/group, data presented as mean ± SD, analyzed by one-way ANOVA with Tukey’s test for multiple comparisons. **:p < 0.001. g Changes in RNA expression levels of Fto, miR-183, Cacna2d2, Bdnf, and Trkb; and (h) changes in mRNA expression levels of m6A regulatory factors (Mettl3, Mettl14, Wtap, Alkbh5, Ythdf2) in each group. N = 5/group, data presented as mean ± SD, analyzed by one-way ANOVA with Tukey’s test for multiple comparisons. **: p < 0.001. i Changes in m6A methylation in L4-L6 DRGs of SNI mice in each group. The vertical axis represents the percentage of m6A in total RNA. N = 5/group, data presented as mean ± SD, analyzed by one-way ANOVA with Tukey’s test for multiple comparisons. **: p < 0.001. j Changes in FTO, CACNA2D2, BDNF, and TrkB protein expression levels in each group. Swimming exercise decreased the mRNA and protein levels of FTO, Cacna2d2, BDNF, and TrkB; increased the expression of miR-183; increased the total m6A levels of SNI mice. N = 5/group, data presented as mean ± SD, analyzed by one-way ANOVA with Tukey’s test for multiple comparisons. **: p < 0.001. Antibodies: FTO (RRID: AB_3713456), CACNA2D2 (RRID: AB_3713457), BDNF (RRID: AB_10862052), TrkB (RRID: AB_444716), and β-actin (RRID: AB_2242334)

Plasmid constructs, virus production, and intrathecal injection

An adeno-associated virus (AAV) vector of shFTO and AAVFTO of the mouse was built. Plasmid construction, virus packaging, and titer determination were completed by OBiO Technology (Shanghai, China). AAV vectors typically begin to express at around 1–2 weeks, peak at approximately 3 weeks, and then maintain sustained expression. For the AAV vectors with shFTO, three sequences were designed for verification (Obio Technology), and the one with the highest knockdown efficiency was selected for the formal experiment (Supplemental Figures S6a–S6d). Supplemental Appendix 1 shows the DNA primer fragment, the framework of virus vector construction, carrier name, serotype, and titer detection. The AAV vectors with serotype AAV2/9 of AAV-FTO-shRNA (pAAV-U6-shRNA(FTO)-CMV-EGFP-WPRE, 5 µL, titer >4.68 × 1012 v.g./mL) or AAV scramble-shRNA (pAAV-U6-shRNA(NC2)-CMV-EGFP-WPRE, 5 µL, titer >8.03 × 1012 v.g./mL) or AAV-FTO(pcAAV-CMV-EGFP-P2A-FTO-3×FLAG-WPRE, 5 µL, titer >5.46 × 1012 v.g./mL) or negative control (pcAAV-CMV-EGFP-P2A-MCS-3×FLAG-WPRE, 5 µL, titer >3.57 × 1013 v.g./mL) were injected into the L5–L6 intervertebral space of the mouse spine through a previously described intrathecal injection technique [17, 29,30,31]. These findings support the notion that intrathecal injection is an effective method for predominantly targeting DRG neurons and can induce corresponding changes in expression.

RNA isolation and qRT-PCR

Total RNA from the unilateral L4–L6 DRGs was purified using the SteadyPure Virus DNA/RNA Extraction Kit (AG21021, Accurate Biotechnology, Hunan, China) and then reverse-transcribed using the miRNA first-strand cDNA synthesis kit (AG11717, Accurate Biotechnology, Hunan, China). RNA quality was assessed using CLARIOstar (BMG LABTECH, Germany). The reverse transcription kit mainly included miRNA RT enzyme mix, 2X miRNA RT reaction solution, and miRNA qPCR 3’ primer (10 µM). The miScript SYBR Green PCR Kit (Qiagen, 218073, Germany) was used for quantitative reverse transcription-PCR (qRT-PCR) of miRNA. The template (1 µL) was amplified via real-time PCR using the primers listed in Supplemental Appendix 2. Each sample was run in triplicate in a 10 µL reaction, consisting of 1 µL forward and reverse primers, 5 µL 2× QuantiTect SYBR Green PCR Master Mix, 1 µL 10× miScript Universal Primer, 2 µL RNase-free water, and 2 ng cDNA (2 ng/µL). SYBR Green Promix Pro Taq HS qPCR Kit (AG11702, Accurate Biotechnology, Hunan, China) was used for mRNA qRT-PCR. The template (1 µL) was amplified through real-time PCR using the primers listed in Supplemental Appendix 2. Each sample was run in triplicate in a 10 µL reaction with 0.4 µl forward and 0.4 µl reverse primers, 5 µL 2×SYBR® Green Pro Taq HS Premix II, 0.2 µl ROX Reference Dye (4 µM), 3 µl RNase-free water, and 2 ng cDNA (2 ng/µL). Reactions were conducted in a QuantStudio 6 Flex real-time PCR system (ABI, USA). U6 was used as an internal control for miR-183 normalization, and 18 S was used as an internal control for mRNA normalization. All data were normalized to U6 (for miR-183) or 18 S (for mRNA), and the ratios of miR-183 and mRNA levels from the treated groups were calculated using the ΔCt method (2 − ΔΔCt).

Western blot

Unilateral L4–L6 DRGs tissue proteins were treated with 100 mM phenylmethylsulfonyl fluoride (Beyotime, Shanghai, China) and radioimmunoprecipitation assay lysate (Beyotime) and homogenized, and the supernatants were collected. The BCA Protein Assay Kit (Beyotime) was used to determine protein concentrations, followed by heating at 100 °C for 10 min. A7.5% or 15% sodium dodecyl sulfate-polyacrylamide gel electrophoresis was performed based on the molecular weight of the target protein, which was then transferred onto a polyvinylidene fluoride membrane (0.45–0.22 μm, Millipore). The sample was 20 µg per loading lane (2 µg/µl), and the amount of marker (WJ102, Epizyme Biotech, Shanghai, China) was 5 µg per loading lane. After being blocked with 5% bovine serum albumin (BSA) for 2 h, the membranes were subjected to an overnight incubation at 4 °C with the primary antibodies for the following (Supplemental Appendix 3): FTO (Abcam, ab280081, RRID: AB_3713456, 1:1000), CACNA2D2 (Abcam, ab173293, RRID: AB_3713457, 1:500), BDNF (Abcam, ab108319, RRID: AB_10862052, 1:1000), TrkB (Abcam, ab18987, RRID: AB_444716, 1:1000), and β-actin (CST, 3700, RRID: AB_2242334, 1:1500). The proteins were detected using horse radish peroxidase (HRP)-labeled secondary antibodies goat antimouse IgG (Beyotime, A0216, 1:1000, Shanghai, China) or goat antirabbit IgG (Beyotime, A0208, 1:1000, Shanghai, China) for 60 min, visualized by Immobilon Western Chemiluminescent HRP Substrate (Millipore, WBKLS0100, Burlington, MA), and exposed using a chemiluminescent imaging system (Tanon, 5200Multi, China). The manufacturer has validated the above antibody for its specificity. The detected band was consistent with the predicted molecular weight of the target protein in this study. Based on the validation of the antibody and consistent changes in protein expression and pain-related behaviors, the specificity of the above antibody was reasonably supported. Original and uncropped immunoblots are provided in Supplementary Materials. The contrast and brightness of images were adjusted linearly across their entirety when necessary for clarity, without altering any data.

Enzyme-Linked immunosorbent assay (ELISA)

The levels of interleukin (IL)−1β, tumor necrosis factor (TNF)-α, IL-6, and IL-10 in unilateral L4–L6 DRGs tissues were determined using a mouse IL-1β ELISA kit (WEIAOBIO, Shanghai, China, EM30300S), mouse TNF-α ELISA kit (WEIAOBIO, EM30536S), mouse IL-6 ELISA kit (WEIAOBIO, EM30325S), and mouse IL-10 ELISA kit (WEIAOBIO, EM30274S). Briefly, protein extraction was performed, and protein concentration was measured. Samples for incubation were then added. The following solutions were added separately: 1x biotinylated detection antibody working solution, SABC complex working solution, colorimetric solution, and stop solution. The absorbance at 450 nm was measured, and the concentration was calculated by plotting a standard curve.

RNA m6A quantification

RNA from unilateral L4–L6 DRGs was extracted using a SteadyPure Virus DNA/RNA Extraction Kit (AG21021, Accurate Biotechnology, Hunan, China), and RNA quality was assessed using CLARIOstar (BMG LABTECH, Germany). In accordance with the instructions, the m6A modification level of the total RNA was examined using an EpiQuik m6A RNA methylation quantification kit (P-9005, Epigentek Group Inc., Farmingdale, NY, USA). Briefly, 200 ng RNA (each sample) was added, and the m6A standard was placed in the corresponding hole separately. Then, the capturing m6A and detecting antibody solutions were added. The absorbance (OD450) of each hole was read at a wavelength of 450 nm and calculated based on the standard curve to quantify the total m6A level of each sample.

Methylated RNA Immunoprecipitation sequencing (MeRIP-Seq) and data analysis

The principle of MeRIP-Seq is to enrich m6A-methylated RNA fragments using the m6A antibody, identify and bind m6A-modified RNA fragments with m6A-specific antibody for immunoprecipitation (IP), and conduct high-throughput sequencing of IP RNA fragments. Regions with high levels of m6A methylation across the transcriptome were identified. Briefly, 2 µg RNA was extracted from the unilateral L4–L6 DRGs. An Agilent 2100 bioanalyzer and simpliNano spectrophotometer (GE Healthcare) were used to detect RNA integrity and concentration. In the IP experiment, the fragmented RNA (~ 100 nt) and anti-m6A polyclonal antibody were incubated at 4 °C for 2 h. Ovation SoLo RNA-SEQ System Kit (NuGEN) was then used to construct the immunoprecipitated RNA or input library. The library preparations were sequenced on an Illumina NovaSeq 6000 platform, which produced 150-bp paired-end reads. The constructed library was sequenced on the Illumina platform with a sequencing strategy of PE150.

Fastp (version 0.19.11) was used to process raw reads in fastq format. Differential gene expression analysis was performed using the edgeR R software package (3.16.4). Benjamini and Hochberg’s methods were used to correct the p-values. In general, |log2(FoldChange)| >1&padj < = 0.05 was the screening criterion for differential genes. The exome Peak R package (version 2.16.0) was used to perform differential peak calling, with parameters set to a P value ≤ 0.05 and a fold change ≥ 1. The genes associated with differential m6A peaks were classified and annotated via the same method, and Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed. Based on the p-values, dot size, color scales, and correspondence comparisons and aligned with the focus of this study, only the most relevant GO function terms and KEGG pathways were highlighted using red boxes.

MeRIP-qPCR

MeRIP-qPCR is performed to verify the modification level of a specific modification site on a target gene. The sequence fragments of IP samples and input samples were compared, and the degree of m6A methylation was calculated. The operation steps of RNA fragmentation, immunoprecipitation, and RNA purification of MeRIP-qPCR were identical to the first few steps of MeRIP-seq. Additionally, they followed the experimental steps of cDNA synthesis and quantitative PCR (qPCR). The cDNA synthesis and qPCR materials mainly included RNA enzyme inhibitors (Epicentre), SuperScriptTM III reverse transcriptase (ThermoFisher), 5x (Invitrogen), 2.5 mM dNTP RT buffer mixture (dATP, dGTP, DCTP, and dTTP, each 2.5 mM) (HyTest Ltd), primer (Yingjun Biological Technology Co., Ltd.), and qPCR SYBR Green master mix (GenSeq). In this study, MeRIP-seq was employed to identify transcriptome-wide alterations in m6A modification and to screen candidates, while specific validation of miR-183 was conducted using MeRIP-qPCR.

Small-RNA sequencing

Total RNA from unilateral L4–L6 DRGs was used to prepare the library. The 3′ and 5′ adaptors were attached to the small RNA’s 3′ and 5′ ends, respectively. Then, the first cDNA was synthesized through hybridization with reverse primers. PCR amplification was conducted to generate double-stranded cDNA libraries. The Qubit2.0 Fluorometer was used to detect their concentration, and qRT-PCR was employed for accurate quantification (above 2 nM) to ensure library quality. The library was diluted to a uniform concentration of 1.5 ng/µl. Then, the Agilent 2100 bioanalyzer was used to determine the insert size of the library, with insert sizes ranging from 18 to 40 bp used for sequencing. After the library test was qualified, double-terminal sequencing was performed using the Illumina NovaSeq 6000 (Illumina, USA), generating 150 bp paired-end reads. Fastp (version 0.23.1) was used to filter sequences containing joint pollution, low quality, and a high N amount to control data quality. Known microRNA sRNA classification analysis, which eliminated data annotation for all small-RNA comparisons, and comments with all kinds of RNA were summarized. The enrichment of differentially expressed genes in the GO enrichment analysis and KEGG pathway was analyzed using Cluster Profiler software. Adj < 0.05 was used as the threshold of significant enrichment.

Histological assessments

The morphology of unilateral L4–L6 DRGs was observed via hematoxylin-eosin staining. After DRG tissue fixation, dehydration was performed using a 15% and 30% sucrose solution gradient, followed by OCT embedding and sectioning. The slices were removed from the − 80 ℃ refrigerator, balanced at room temperature for 20 min, and immersed in 1×phosphate-buffered saline buffer for 5 min to wash away OCT. Dyeing was performed in accordance with the following procedure: (1) immersion in hematoxylin for 3.5 min; (2) rinsing with running water for 20 min to wash away the floating color; (3) immersion in 1% hydrochloric alcohol and differentiation for 35 s; (4) rinsing with running water for 10 min; (5) immersion in eosin for 3.5 min; (6) immersion in 85%, 95%, and 100% ethanol solution for 1 min each and dehydration; (7) immersion in xylene 1 and 2 for 3 min each to make the slices transparent. This process was completed in the fume hood, which was sealed with a neutral resin under the hood. After the film was sealed, the tissue morphology of DRG was observed with an Olympus optical microscope (BX53), and the film was obtained.

Immunofluorescent staining

Considering the double fluorescent staining of FTO and the neuronal marker neuronal nuclei (NeuN) as an example, 1× sodium citrate was initially used for antigen repair, 5% goat serum blocking solution was added to the drop, and the unilateral L4–L6 DRG tissue section was added with FTO primary antibody (mouse source, Santa Cruz, sc271713, 1:50) proportionally diluted with 5% goat serum and incubated at 4 ℃ overnight. Next, AlexaFluor 488 fluorescent secondary antibody diluted with 5% goat serum was added to the tissues (goat antimouse, Invitrogen, A-11001, 1:1000) and incubated for 1 h under dark conditions. Next, 5% BSA sealing solution was added to the tissue and blocked at room temperature for 30 min. Next, 5% BSA-diluted NeuN primary antibody (rabbit source, Abcam, ab104225, 1:1000) was added to the tissue, which was then incubated overnight at 4 ℃. AlexaFluor 555-labeled fluorescent secondary antibody (goat anti-rabbit, Invitrogen, A-21428, 1:1000) diluted with 5% BSA was then added to the tissue, followed by incubation for 1 h. Finally, the tablets were sealed with an antiquenching tablet containing DAPI (Invitrogen, P36971). After the film was sealed, it was left for 15 min, and a laser confocal microscope (ZEISS, LSM700) was used for observation and filming.

Gene identification of miR-183 KO mice

Gene identification was performed on 3-week-old mice using a One Step Mouse Genotyping Kit (Vazyme, PD101-01, Nanjing, China). The experimental procedure was as follows: 1) 1–3 mm tail tip tissue was collected; 2) 1× lysis buffer was prepared by mixing proteinase K and 1× mouse tissue lysis buffer at a ratio of 1:50; 200 µL lysis buffer was required per sample; 3) 200 µL lysis buffer was added to the tail tissue, which was then incubated at 55 °C for 20 min, and heated at 95 °C for 5 min in a metal bath; 4) the mixture was mixed thoroughly and centrifuged at 12,000 rpm for 5 min at room temperature, and the supernatant was collected; 5) PCR amplification was performed using the primers listed in Supplemental Appendix 4; 6) agarose gel electrophoresis was prepared: 5 µL DNA ladder was loaded into flanking wells, followed by 5 µL PCR products in intermediate wells; electrophoresis was run at 120 V for 40 min; 7) the gel was visualize, and the genotypes were determined: homozygous miR-183 KO mice showed a single 702 bp band, heterozygotes displayed two bands at 855 and 702 bp, and wild-type mice exhibited a single 855 bp band.

Dual-Luciferase reporter assay

First, the potential m6A modification sites of FTO on miR-183-5P were predicted in accordance with the motif analysis based on MeRIP-seq. RNAhybrid 2.2 was used to predict the binding sites between mmu-miR-183-5p and BDNF. The results show that mmu-miR-183-5p had complete or partial complementary binding sites with the 3 3’-untranslated regions (UTR) of BDNF. Therefore, this fragment was designed and synthesized for validation. The miR-183-5P and BDNF 3’-UTR fragments were synthesized, XhoI/NotI restriction enzyme sites were added to both ends, and mutation sites were introduced. The obtained gene fragment was cloned into the PUC57 vector. Supplemental Appendix 5 shows the sequences of miR-183-5P, BDNF, and the mutation. An appropriate number of 293 T cells were inoculated on the day before transfection, with antibiotic-free medium added to each well, which ensured a cell density of 70% to 80% at the time of transfection. Next, 100 ng psiCHECK-mmu-miR-183-5P (V1), FTO, NC (V2), 100 ng psiCHECK-BDNF (V1), and mmu-miR-183-5p, NC (V2) were diluted in 25 µL serum-free medium and gently mixed. Fluorescence intensity was measured. 5 µl Renilla luciferase reaction buffer and substrate were added and mixed, and the Renilla luciferase activity was measured. Each sample was normalized based on firefly luciferase activity, and Renilla luciferase activities were compared and plotted (for the psiCHECK-2 vector, firefly luciferase was used as the internal reference).

Statistical analysis

SPSS 27.0, GraphPad Prism 8.0, and R version 4.2.3 were used for statistical analysis and plotting of the results. First, means ± standard deviations and medians (interquartile ranges) were used to describe continuous and categorical variables with normal and skewed distributions, respectively. Mean ± standard error of the mean was used for behavioral data. Two-factor repeated-measures analysis of variance (ANOVA) was used for body weight and behavioral data, with intergroup factors representing the groups and intragroup factors representing measurement time. Mauchly’s spherical test was performed to analyze the correlation between variables. If the spherical distribution hypothesis was not satisfied, the Epsilon correction coefficient was used to correct the degrees of freedom. A Bonferroni test was conducted for multiple comparisons to assess the differences in weight and behavioral data between groups at various time points. For miR-183 KO mice, three-way ANOVA (time × sex × treatment) was performed to assess potential sex-specific effects. One-way ANOVA was used for qRT-PCR data, MeRIP-qPCR data, protein gray value, the Renilla luciferase activity, and ELISA concentration, and total m6A level. Levene’s test was used to perform the homogeneity of variance test. Tukey’s test was used for multiple comparisons to compare the differences between the ΔΔCt and protein gray values among all groups. A p-value less than 0.05 was considered statistically significant. The statistical methods of MeRIP-seq and small RNA-seq were described in detail in the above section.