Enhancing m⁶A modification in the motor cortex facilitates corticospinal tract remodeling after spinal cord injury

Abstract

Spinal cord injury typically causes corticospinal tract disruption. Although the disrupted corticospinal tract can self-regenerate to a certain degree, the underlying mechanism of this process is still unclear. N⁶-methyladenosine (m⁶A) modifications are the most common form of epigenetic regulation at the RNA level and play an essential role in biological processes. However, whether m⁶A modifications participate in corticospinal tract regeneration after spinal cord injury remains unknown. We found that expression of methyltransferase 14 protein (METTL14) in the locomotor cortex was high after spinal cord injury and accompanied by elevated m⁶A levels. Knockdown of Mettl14 in the locomotor cortex was not favorable for corticospinal tract regeneration and neurological recovery after spinal cord injury. Through bioinformatics analysis and methylated RNA immunoprecipitation-quantitative polymerase chain reaction, we found that METTL14 regulated Trib2 expression in an m⁶A-regulated manner, thereby activating the mitogen-activated protein kinase pathway and promoting corticospinal tract regeneration. Finally, we administered syringin, a stabilizer of METTL14, using molecular docking. Results confirmed that syringin can promote corticospinal tract regeneration and facilitate neurological recovery by stabilizing METTL14. Findings from this study reveal that m⁶A modification is involved in the regulation of corticospinal tract regeneration after spinal cord injury.

Introduction

Spinal cord injury (SCI) has a high global prevalence and can be neurologically devasting; no effective treatment is available (GBD 2016 Traumatic Brain Injury and Spinal Cord Injury Collaborators, 2019). Disruption of the corticospinal tract (CST), which is critical for fine motor movements, is common in SCI (Guo et al., 2021; Davleeva et al., 2023; Mu et al., 2024). Some degree of CST regeneration can occur after SCI; however, the ability of the CST to form effective connections with downstream motor units across the site of injury is limited (Savio and Schwab, 1990; Zheng et al., 2005). Investigating the mechanisms underlying CST regeneration after SCI has potential to provide clinical benefit.

N⁶-methyladenosine (m⁶A) modification is the most common type of epigenetic RNA modification and plays a crucial role in central nervous system (CNS) development and disease (Wang et al., 2020; Zhang et al., 2022; Liu et al., 2023; Zeng et al., 2023). Methyltransferase 14 (METTL14) and methyltransferase 3 (METTL3) form the m⁶A methyltransferase complex, which has been extensively investigated (Liu et al., 2014a, 2022a; Xiong et al., 2023). m⁶A modifications are recognized by specific binding proteins called readers (Liu et al., 2023; Knight et al., 2024). Notable readers include members of the YTH domain–containing protein (YTHDF1–3 and YTHDC1–2) and insulin-like growth factor 2 mRNA–binding protein families, which mediate downstream effects (Huang et al., 2018; Sikorski et al., 2023). Conversely, erasers like alpha-ketoglutarate dependent dioxygenase (FTO) and alkB homolog 5 (ALKBH5) can eliminate m⁶A (Mazloomian and Meyer, 2015). The multifaceted functions of m⁶A-tagging include mRNA splicing, stability, nuclear export, localization, translational efficiency activation, and the decay of target mRNA stability (Wang et al., 2014; Deng et al., 2018). In CNS development, knocking down Mettl14 reduces neural stem cell self-renewal by regulating histone modifications (Wang et al., 2018b). After peripheral nervous system injury, axonal regeneration depends on the pro-protein synthesis effect of METTL14. Knockdown of Mettl14 diminishes the robust axon regeneration induced by phosphatase and tensin homolog (Pten) deficiency (Weng et al., 2018). Our earlier study demonstrated that Mettl14 expression in neurons is significantly upregulated after chronic compressive SCI, suggesting that METTL14 may serve as a potential neuroprotective factor in acute trauma (Li et al., 2023). However, further investigation is needed to determine whether CST regeneration after SCI is regulated by METTL14. In this study, we systematically investigated the role of METTL14 in CST regeneration after SCI and explored potential METTL14-related drug treatment strategies.

Methods

Animals

Eight-week-old female C57BL/6J mice weighing between 20 and 22 g were purchased from Hunan SJA Laboratory Animal Co., Ltd. (Changsha, China, license No. SCXK (Xiang) 2019-0004). To exclude the influence of severe urinary retention on the observation and evaluation of the repair effect of the SCI mice, female mice were selected owing to their shorter and more accessible urethra, which facilitates urination (in contrast, the narrower and longer urethra in male mice causes urination difficulties and higher mortality). All mice were housed in the Zoology Department of Central South University with a 12/12 hour light/dark cycle. Animals had free access to food and water. All procedures were approved by the Animal Experiment Ethics Committee of Central South University (approval No. CSU-2022-0328; July 5, 2022). All experiments were designed and reported according to the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines (Percie du Sert et al., 2020).

Spinal cord injury model establishment and treatment

A total of 85 mice were used. Thirteen were randomly assigned to the sham group without being subjected to SCI. The remaining 72 mice were randomly assigned to the following groups: SCI only, SCI + NC-Mettl14, SCI + sh-Mettl14, SCI + vehicle, and SCI + syringin group. Among the 72 SCI mice, 61 survived injury. The SCI only group consisted of 25 mice which were divided into 3 days post-injury (dpi), 7 dpi, 14 dpi, and 28 dpi (n = 4/time point) subgroups as well as a control group (n = 9) for behavior and histologic analysis. The NC-Mettl14 group, Sh-Mettl14 group, vehicle group, and syringin groups consisted of 36 mice (n = 9/group), including 3 mice per group for CST tracing.

The mice were anesthetized with an intraperitoneal injection of pentobarbital sodium 75 mg/kg (Merck, Darmstadt, Germany). Briefly, the fur was removed, the skin disinfected, and an incision made in the middle of the back. The fascia and muscles were separated layer-by-layer and the T10 lamina removed to expose the spinal cord. A microknife was used to completely transect the dorsal spinal cord to a depth of approximately 1 mm (except for the sham group). Then, the muscle and fascia were sutured in layers and the mice returned to their cages. The following week, we assisted the mice to urinate twice a day. Syringin was dissolved in dimethyl sulfoxide (DMSO) and then prepared with physiological saline to a concentration of 20 mg/mL. Syringin (MCE, Monmouth, NJ, USA, 118-34-3) solution and normal saline (vehicle) were administered by gavage at a dose of 100 mg/kg per day for 4 weeks (Li et al., 2017).

The lentivirus for knocking down METTL14 was constructed by OBiO Technology (Shanghai) Corp., Ltd. (Shanghai, China) and administrated 2 weeks before establishment of the SCI model. Briefly, after anesthetization, the mice were fixed with a stereotactic instrument (Stoelting, Chicago, IL, USA). The head fur was removed, and an incision made on the head to expose the skull. A dental drill was used to remove the skull above the sensorimotor cortex bilaterally. The bregma was set as the location point and a Hamilton microneedle was used to inject 1 μL sh-Mettl14-lentivirus or corresponding negative control lentivirus (NC-Mettl14) at each site (six sites/hemisphere; from bregma, ±1.0/−0.1/0.55 mm, ±1.0/−0.6/0.55 mm, ±1.0/−1.1/0.55 mm, ±1.4/−0.1/0.55 mm, ±1.4/−0.6/0.55 mm, and ±1.4/−1.1/0.55 mm, anteroposterior/mediolateral/dorsoventral (AP/ML/DV), 0.4 μL/site at a rate of 0.05 μL/min) (Dias et al., 2018). After injection at each site, the syringe was left in place for 2 minutes. After all sites were injected, the skin was sutured and the mice return to their cages. A flow chart of the experimental procedures is shown in Figure 1.

Figure 1.

Experimental flow chart.

BDA: Biotin dextran amine; BMS: Basso Mouse Scale; dpi: days post injury.

Anterograde tracing of corticospinal tracts

Anterograde CST tracing was performed using 10% biotinylated dextran amine (BDA; Invitrogen, Carlsbad, CA, USA) as previously described (Guo et al., 2021). Two weeks after SCI, the mice were anesthetized and then fixed with a stereotactic instrument. After exposing the skull, the bone above the sensorimotor cortex bilaterally was removed using a dental drill. BDA was injected into both sensorimotor cortexes (6 sites/hemisphere; from bregma, ±1.0/−0.1/0.55 mm, ±1.0/−0.6/0.55 mm, ±1.0/−1.1/0.55 mm, ±1.4/−0.1/0.55 mm, ±1.4/−0.6/0.55 mm, and ±1.4/−1.1/0.55 mm, AP/ML/DV, 0.4 μL/site at a rate of 0.05 μL/min) (Dias et al., 2018). The needle was kept in place for 2 minutes after each injection to allow for BDA infiltration.

Primary cortical neuron culture and treatment

Primary cortical neurons were acquired from C57BL/6J mice embryos (days 14–16 of gestation) from 12 pregnant mice obtained from Hunan SJA Laboratory Animal Co., Ltd as previously described (Li et al., 2021). Briefly, the pregnant mice were anesthetized and sacrificed using the decapitation method. Then, the embryos were harvested to remove their cerebral cortexes, which were washed with cold phosphate-buffered solution. After the dura and blood vessels on the surface were carefully removed, the cortical tissue was placed in precooled Dulbecco’s modified Eagle medium solution (Gibco, New York, NY, USA). The brains were then cut into tissue blocks measuring 1 mm³ and homogenized using a 1 mL pipette tip to obtain a cell suspension. After the cell suspension was filtered with a 70 μm-diameter filter and centrifuged at 300 × g for 5 minutes, the cells were resuspended in the neurobasal medium (Cyagen, Chicago, IL, USA) containing 2% B27 (Gibco), 1% glutamine (Gibco), and 1% penicillin-streptomycin (Merck). The cell suspension was seeded in a 24-well plate or a microfluidics culture plate (SND450, Xona, Chicago, IL, USA) and the medium was changed 4 hours later. Primary cortical neurons were divided into two parts, cell bodies and processes, and were specifically labeled with TUJ1.

To investigate the effects of METTL14 and tribbles homolog 2 (TRIB2) on neurite outgrowth, we transfected neurons with METTL14-knockdown lentiviruses, lentiviruses overexpressing TRIB2 (OE-Trib2), and their negative control lentiviruses at a multiplicity of infection of 1:30 using manufacturer protocol. All lentiviruses were synthesized by OBiO Technology Corp., Ltd. After 12 hours of lentivirus treatment, the culture medium was replaced. The neurons were then cultured for an additional 3 days for subsequent testing.

Immunofluorescence staining

After anesthetization, the mice were successively perfused with normal saline and 4% paraformaldehyde via the left ventricle. The cerebral cortex and T6–T12 spinal cord segment were harvested. After dehydration with 20% and 30% sucrose solutions, the tissues were embedded in optimum cutting temperature compound (Sakura, Torrance, CA, USA) for frozen sectioning. Both the brain and spinal cord tissues were sectioned into slices with 20 μm thickness. For immunofluorescence staining, the slices were washed with a phosphate-buffered solution after rewarming. The slices were penetrated and blocked with 0.1% phosphate-buffered solution-Triton-X100 solution and 3% bovine fetal albumin, respectively. Subsequently, the slices were incubated overnight at 4°C with primary antibodies. The next day, after washing with 0.1% phosphate-buffered solution-Tween 20 solution, the slices were incubated with corresponding secondary antibodies at 25°C for 1 hour. For BDA staining, after penetration and blocking, the spinal slices were incubated with avidin-fluorescein isothiocyanate (Merck, 1:1000) at 25°C for 1 hour. Then, after washing with 0.1% phosphate-buffered solution-Tween 20 solution twice, the slices were stained with 4’,6-diamidino2-phenylindole (GeneTex, Alton Pkwy Irvine, CA, USA). Images were captured using a fluorescence microscope (Zeiss, Oberkochen, Germany). All antibodies used in the study are listed in Additional Table 1.

Additional Table 1.

The antibodies used in the study

Antibody	Species	Dilution	Supplier	Cat#	RRID
METTL14	Rabbit	WB: 1:750 IF: 1:100	Zenbio, Chengdu, China	508530	AB 3076146
METTL3	Rabbit	WB: 1:750 IF: 1:100	Zenbio, Chengdu, China	382974	AB 3076147
YTHDF1	Rabbit	WB: 1:750	Huabio, Hangzhou, China	HA500350	AB 3071435
YTHDF3	Rabbit	WB: 1:2000	Proteintech, Wuhan, China	25537-1-AP	AB 2847817
TUJ1	Rabbit	IF: 1:400	Abcam, Cambridge, UK	ab78078	AB 2256751
SYN	Rabbit	IF: 1:400	Abcam, Cambridge, UK	ab32127	AB 2286949
NeuN	Mouse	IF:1:400	Abcam, Cambridge, UK	ab104224	AB_10711040
TRIB2	Rabbit	WB: 1:1000 IF: 1:200	Affinity, Munroefalls, OH,USA	DF2692	AB 2839898
GFAP	Goat	IF: 1:500	Abcam, Cambridge, UK	ab53554	AB 880202
F4/80	Rabbit	IF: 1:200	Santa Cruz, Dallas, TX, USA	sc-377009	AB 2927461
p-JNK	Rabbit	WB: 1:1000	CST, Danvers, MA, USA	4668T	AB 3076148
JNK	Rabbit	WB: 1:750	Wanleibio, Shenyang, China	WL01295	AB 3064853
P-P38	Rabbit	WB: 1:750	Wanleibio, Shenyang, China	WLP1576	AB 2922420
ACTIN	Rabbit	WB: 1:20000	Proteintech, Wuhan, China	81115-1-RR	AB_2923704
Anti-Rabbit (HRP)	Goat	WB: 1:5000	Abcam, Cambridge, UK	ab6721	AB_955447
Anti-Rabbit AF488	Donkey	IF: 1:400	Abcam, Cambridge, UK	ab150073	AB_2636877
Anti-Rabbit AF594	Donkey	IF: 1:400	Abcam, Cambridge, UK	ab150076	AB_2782993
Anti-Goat AF488	Donkey	IF: 1:400	Abcam, Cambridge, UK	ab150129	AB_2687506
Anti-Mouse AF488	Donkey	IF: 1:400	Abcam, Cambridge, UK	ab150105	AB_2732856

GFAP: glial fibrillary acidic protein; HRP: horseradish peroxidase; IF: immunofluorescence; JNK: Jun N terminal kinase; METTL14: methyltransferase-like protein 14; Mettl3: methyltransferase-like protein 3; p-JNK: phospho-JNK; p-P38: phospho-P38; SYN: synaptophysin; Trib2: tribbles pseudokinase 2; WB: Western blot; Ythdf1: YTH N6-methyladenosine RNA binding protein F1; Ythdf3: YTH N6-methyladenosine RNA binding protein F3.

Hematoxylin-eosin staining

The mice treated with syringin or normal saline were anesthetized and then successively perfused with normal saline and 4% paraformaldehyde via the left ventricle. The heart, spleen, liver, lung, and kidney were harvested. After gradient alcohol dehydration, the tissues were embedded in paraffin, sequentially sectioned (5 μm thickness), and subjected to hematoxylin-eosin (H&E) staining as per manufacturer instructions (Solarbio, Beijing, China).

Quantitative polymerase chain reaction

Total RNA was extracted from the sensorimotor cortex or neurons using a TRIzol reagent (Invitrogen). After the RNA concentration was measured using a microplate reader (Thermofisher, Waltham, MA, USA), RNA was reversed and transcribed into complementary DNA using a PrimeScript^TM RT Reagent kit (Promega, Madison, MI, USA). Subsequently, a GoTaq qPCR Master Mix kit (Promega) was applied to detect the cycle threshold (Ct). The differential gene expression was calculated using the 2^–ΔΔCt method (Harshitha and Arunraj, 2021). The primers for qPCR are listed in Additional Table 2.

Additional Table 2.

Primers for qPCR

Primers	Forward (5’-3’)	Reverse (5’-3’)
Mettl14	CTGAGAGTGCGGATAGCATTG	GAGCAGATGTATCATAGGAAGCC
Mettl3	CTGGGCACTTGGATTTAAGGAA	TGAGAGGTGGTGTAGCAACTT
Ythdfl	CACAGTGACTCCCTCAACAAG	AGGTGGTAACATCCCCAATCTT
Ythdβ	GATCAGCCTATGCCATATCTGAC	CCCCTGGTTGACTAAAAACACC
Fto	TTCATGCTGGATGACCTCAATG	GCCAACTGACAGCGTTCTAAG
Alkhb5	GCGCGGTCATCAACGACTA	ATCAGCAGCATACCCACTGAG
Wtap	TGCACGCAGGGAGAACATTC	TGAACTTGCTTGAGGTACTGGAT
Ntrk3	GCCAAGTGTAGTTTCTGGCG	CAGACACAATTTGCAGGGCA
Cntf	TCTGTAGCCGCTCTATCTGG	GGTACACCATCCACTGAGTCAA
Igflr	GTGGGGGCTCGTGTTTCTC	GATCACCGTGCAGTTTTCCA
NeuroDl	ATGACCAAATCATACAGCGAGAG	TCTGCCTCGTGTTCCTCGT
Nrg1	ATGGAGATTTATCCCCCAGACA	GTTGAGGCACCCTCTGAGAC
Pten	TGGATTCGACTTAGACTTGACCT	GCGGTGTCATAATGTCTCTCAG
Cers2	ATGCTCCAGACCTTGTATGACT	CTGAGGCTTTGGCATAGACAC
Ptprs	GGTGAACAACATACCCCCGAC	TCCCACCTCTGTGTAAGCCA
Wtip	CACGGCGCGAGACTACTTT	TTGCTGGAACCCGGAGTACA
Egr2	GCCAAGGCCGTAGACAAAATC	CCACTCCGTTCATCTGGTCA
Taf13	TTCTCGAAAGAATTGCGCTGT	GCCTTGTGAGTCATTTCAGTGA
Gstp-ps	GCCCAGGTGGATATGGTGAATGATG	AGCAGGTCCAGCAAGTTGTAATCG
Mta3	ACCTGCCCCTGTCCTACTG	CCTTGGCACACTTCTCACATC
Ppill	CCTGGAGACTAGCATGGGAGT	AGCCCCAGTGAACTTCAGGT
Oxsrl	CATTGTGGCAAAGGGGGAAC	TGACGCCGAAATCTGCAATCT
Cap2	CAGGGTTAGACGGACCTCC	CTCGGCCACCATACTGTTTATC
Mt3	TGGGTATTTAGAGATCCATCCCG	GCTCAGAGATGTGTGATTCGGAG
Trib2	ATACACAGGTCTACCCCTATCAC	ATGCGACAAGTTCGGAGTCTC
GAPDH	AATGGATTTGGACGCATTGGT	TTTGCACTGGTACGTGTTGAT

Alkhb5: AlkB homolog 5; Cap2: cyclase-associated protein 2; Cers2: ceramide synthase 2; Cntf. ciliary neurotrophic factor; Egr2: early growth response 2; Fto: alpha-ketoglutarate dependent dioxygenase; Gapdh: glyceraldehyde-3-phosphate dehydrogenase; Gstp: glutathione S-transferase P; Igflr. insulin like growth factor 1 receptor; Mettl14: methyltransferase-like protein 14; Mettl3: methyltransferase-like protein 3; Mta3: metastasis associated 1 family member 3; NeuroDl: neurogenic differentiation 1; Nrgl: neuregulin 1; Ntrk3: neurotrophic receptor tyrosine kinase 3; Oxsr1 : oxidative stress responsive kinase 1; Pten: phosphatase and tensin homolog; Ptprs: protein tyrosine phosphatase receptor type S; qPCR: quantitative polymerase chain reaction; Taf13: TATA-box binding protein associated factor 13; Trib2: tribbles homolog 2; Wtap: Wilms tumor 1-associated protein; Wtip: Wilms tumor protein 1-interacting protein; Ythdf1 : YTH N6-methyladenosine RNA binding protein F1; Ythdf3: YTH N6-methyladenosine RNA binding protein F3.

Western blot assay

Total proteins were extracted from the sensorimotor cortex or neurons using radioimmunoprecipitation assay lysis buffer (Beyotime, Shanghai, China). The protein concentration was detected using a bicinchoninic acid assay kit (Beyotime). The denatured proteins were separated in 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis according to their molecular weight by electrophoresis. Then, the proteins were electrotransferred to a polyvinylidene fluoride membrane under 300 mA for 90 minutes. After blocking with 5% non-fat milk for 2 hours, the membranes were incubated with primary antibodies at 4°C overnight. The next day, the membranes were washed with Tris buffered saline with Tween three times, and incubated with corresponding secondary antibodies at room temperature for 2 hours. The bands were detected using enhanced chemiluminescence (Share-Bio, Shanghai, China) and the ChemiDoc system (Bio-Rad, Hercules, CA, USA). The antibodies used are listed in Additional Table 1.

Measurement of total N⁶-methyladenosine

The total m⁶A level was measured using an m⁶A RNA Methylation Quantification Kit (Epigen Tek, New York, NY, USA) according to manufacturer instructions. Briefly, total RNA from the mouse cortex and cultured neurons were extracted using a Trizol agent. 2 μL negative control, 2 μL positive control (0.5 ng/μL), and 200 ng sample RNA were added to the designated wells. After incubating for 60 minutes at room temperature, the m⁶A RNA was captured by adding the capture antibody and detection antibody in order. Then, the samples were washed with diluted washing buffers five times. The developer solution was added to each well and incubated at room temperature for 8 minutes in the dark. Then, the same amount of stop solution was added to stop the enzyme reaction. Absorbance was measured on a microplate reader at 450 nm.

Neuroelectrophysiology analysis

Neuroelectrophysiology analysis was applied to evaluate neurological connectivity 28 days after SCI as previously described (Li et al., 2021). In brief, after anesthetization, a stereotactic device was used to determine the fixed position of the stimulation electrodes (Blackrock, Salt Lake City, UT, USA). Stimulating electrodes were placed on the skull surface (1.0/0.5 mm, –4.0/0.6 mm, caudal/lateral to the bregma, respectively). The recording electrodes were inserted into the contralateral hind limb muscles and the reference electrode was inserted into the dorsal subcutaneous tissue above the SCI lesion. Stimulation was performed (3 V, 333 Hz) and recorded at 5-second intervals. The sum of wave peak and trough voltages was recorded as the motor evoked potential. Latency time was recorded from the beginning of stimulation to the occurrence of dramatic changes in waves.

Behavioral assessment

The Basso mouse scale (BMS) was used to evaluate locomotor function (Basso et al., 2006). In the 9-point BMS, 0 indicates complete paralysis and 9 indicates normal locomotor function. Two professionally trained investigators who were blinded to grouping assessed hindlimb locomotor abilities independently. The final score was obtained by averaging the BMS scores measured by the two investigators. BMS scores were evaluated before the injury as well as 1, 3, and 7 days and 2, 3, and 4 weeks after SCI.

The inclined grid walking task was performed before injury and 1, 2, 3, and 4 weeks after injury. This task requires animals to climb a 45° slope in order to reach the top of a 30 cm long, 2 cm square metal grid (Steward et al., 2008). All animals underwent a 10-day pretraining period before the surgery. Two well-trained observers who were blinded to grouping independently recorded the number of times the hind paws slipped under the grid (missed steps) and calculated the error rate by counting the missed steps over the total number of steps. The final error rate was obtained by averaging the recordings measured by the two observers.

The forced swim test was used to evaluate swimming ability as previously described (Smith et al., 2006). Before the surgery, the mice were trained in three sessions to swim continuously from one side to the other in a water tank (5 cm × 15 cm × 10 cm). At 28 days post-injury, the mice were placed in the water tank for 30 seconds. A high-definition camera was used to record the swimming process. The Louisville Swim Scale was used to assess swimming ability, which included scores for hind limb movement, hind limb alternation, forelimb dependence, trunk stability, and body angle (Smith et al., 2006). The final error rate was obtained by averaging the recordings measured by two independent observers who were blinded to group assignment.

m⁶A-immunoprecipitation-quantitative polymerase chain reaction

m⁶A-immunoprecipitation-qPCR (meRIP-qPCR) was performed to detect methylation of Trib2 mRNA using RiboMeRIP^TM m⁶A Transcriptome Profiling Kit (RibBio, Guangzhou, China). Neuronal RNA was collected and the concentration was adjusted to 1 μg/μL with nuclease-free water. RNA fragmentation buffer was added to the RNA and reacted at 70°C for 7 minutes to break the RNA into 200-nt fragments. For the preparation of anti-m⁶A magnetic beads, magnetic beads A/G and IP buffer were mixed and the mixture placed on a magnetic column for 1 minute. The supernatant was removed and the adsorption step repeated. Next, 5 μg anti-m⁶A antibody was added and the mixture was placed on a shaker for 30 minutes to obtain anti-m⁶A magnetic beads. Before immunoprecipitation, the fragmented RNA was split by 1/10 as an Input group. The configured MeRIP reaction solution (including fragmented DNA) was added to the anti-m⁶A magnetic beads. The supernatant was removed after 1 minute of magnetic column adsorption and the precipitate was placed on ice. The elution buffer was used to separate magnetic beads and RNA fragments. The complementary DNA synthesis and PCR were performed as described above. The specific primers for the m⁶A site of Trib2 were predicted using SRAMP (http://www.cuilab.cn/sramp; Zhou et al., 2016) and are listed in Additional Table 3.

Additional Table 3.

Primers for m6A-qPCR

Primers	Forward (5’-3’)	Reverse (5’-3’)
Site 1	GCTGCGAACAGAGACCCACT	GACAACTCGGGTCCTTTTGTTA
Site 2	GAAAGGAGTTCTGGGACACAGG	CCTTCTCTGAACCAAGAAATGTG
Site 3	TGTTTTTGGGTAGGTGACACG	TGAAGCATCCAAGAGTCCGC

m6A: N6-methyladenosine; qPCR: quantitative polymerase chain reaction.

Flow cytometry of neuron apoptosis

An Annexin V-FITC Apoptosis Detection Kit (Beyotime) was used to detect apoptotic neurons using manufacturer protocol. Briefly, 50,000 neurons were collected and counted to allocate to each sample. Neurons were resuspended with 195 μL annexin V-fluorescein isothiocyanate. Then, another 5 μL of annexin V-fluorescein isothiocyanate were added to the neurons and mixed gently. 10 μL propidium iodide staining solution was subsequently added and mixed gently again. The cell suspension was incubated in the dark at room temperature for 15 minutes, then placed on ice. A flow cytometer (BD-LSRII, Franklin Lakes, NJ, USA) was used to detect the percentage of apoptotic neurons.

Virtual screening

The crystal structure of human Mettl3/Mettl14 protein complexes was predicted through the PDB database (https://www.rcsb.org/structure/7RX7) (Berman et al., 2000). The two-dimensional SDF structure file was processed from online databases (Selleck FDA Approved Drug Library, TargetMol Natural Compound Library, MedChemExpress Natural Product Library, Pharmacodia Natural Product Library, BioBioPha BBP Natural Product Library, APExBIO Anti-Cancer-Compound-Library Plus, Selleck Preclinical/Clinical Compound Library, and MedChemExpress Immunology/Inflammation Compound Library, https://cpt.tsinghua.edu.cn/compoundlibrary/) and three-dimensional chiral conformations were generated (approximately 35,000 molecules). The SiteMap module in Schrodinger (https://www.schrodinger.com/) was used to predict the binding site of Mettl14 protein, and then the Receiver Grid Generation module was used to set the most appropriate enclosing box to perfectly wrap the predicted binding site. On this basis, the active site of the protein was obtained. Each processed ligand from the eight compound libraries was sequentially docked with the active pocket of Mettl14 protein (using HTVS, SP, and XP respectively, with gradually improved docking accuracy). The docking fraction and MMGBSA binding free energy served as a good basis for ligand selection. The QikProp module was used to predict all ADME/T parameter scores of these ligands, resulting in a total of 51 type parameters. These ligands were evaluated based on the Lipinski Five Principles. The top 35 natural small molecule compounds selected from the database are listed in Additional Table 4. To predict Mettl14 protein structure and whether the small molecule compounds can bind to mouse-derived proteins, we employed AlphaFoldA2 (https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb). Subsequently, the binding efficiency of small molecule compounds was validated using this predicted structure.

Additional Table 4.

Top 35 natural small molecule compounds selected from database

PubChem ID	Compound name	XP G score	MM-GBSA dG bind (kcal/mol)	Rule of five
131846037	Salviaplebeiaside	-9.032	-49.56	0
5316860	Syringin	-8.532	-54.48	0
129317357	Evodosin A	-8.397	-46.66	0
5281222	Butein	-8.012	-48.04	0
92918	5-Methylcytidine	-7.942	-31.04	0
91885216	Drimiopsin D	-7.71	-34.6	0
44420939	Coccineone B	-7.508	-41.06	0
68213	Maclurin	-7.313	-39	0
5316511	Demethoxycapillarisin	-7.22	-32.34	0
6020	Puromycin Aminonucleoside	-7.197	-31.32	0
46886154	3,5,7-Trihydroxychromone	-7.153	-30.55	0
181994	Dalbergioidin	-7.068	-39.13	0
3182	Dyphylline	-7.055	-34.14	0
5281626	Desmethylbellidifolin	-7.022	-39.5	0
5281811	Tectorigenin	-6.998	-39.27	0
10735190	14-Deoxycoleon U	-6.979	-51.86	0
71307363	1,4,5,6-Tetrahydroxy-7-Prenylxanthone	-6.929	-45.6	0
107721	Taxiphyllin	-6.919	-39.11	0
9971218	(-)-5,7-Dihydroxy-3-(4-hydroxybenzyl)-4-chromanone	-6.913	-46.4	0
92030426	CID 92030426	-6.911	-45.02	0
75492722	Scillascillin	-6.753	-42.22	0
5281656	Norathyriol	-6.741	-40.5	0
9944143	Barpisoflavone A	-6.715	-34.7	0
214348	Deferasirox	-6.698	-41.35	0
5281706	Cajanin	-6.681	-34.38	0
4330531	10-Hydroxycamptothecin	-6.671	-54.55	0
5282074	2'-Hydroxygenistein	-6.658	-33.57	0
6708707	Dihydrorobinetin	-6.656	-37.42	0
361512	Citreorosein	-6.643	-42.47	0
60198001	Urolithin C	-6.614	-43.41	0
45269778	O-Demethylforbexanthone	-6.612	-40.55	0
442788	Homoferreirin	-6.52	-37.96	0
5281657	Norlichexanthone	-6.519	-39.26	0
9916414	Subelliptenone G	-6.507	-40.77	0
2817763	3,5 -Dithiocyanatopyridine-2,6-diamine	-6.5	-39.29	0

Analysis of differentially expressed genes

Differentially expressed genes (between control and SCI mice cortex) were identified with fold change ≥ 2 and P valure < 0.05. Gene ontology analysis was performed based on the upregulated genes in the SCI mice cortex using clusterProfiler (Yu et al., 2020). The RNA sequencing data in this study is referenced from GSE141583 (Cheng et al., 2022).

Statistical analysis

Statistical analyses were performed using Prism 8.0.1 (GraphPad Software Inc., Boston, MA, USA). Data are presented as means ± standard deviation. P < 0.05 was considered significant. Comparisons between two groups were performed using the unpaired t-test. Comparisons of multiple groups were performed using one-way analysis of variance followed by Tukey’s post hoc test. BMS scores and axon density analysis were compared using two-way repeated measures analysis of variance.

Results

Mettl14 is highly expressed in sensorimotor neurons after spinal cord injury

m⁶A modification is common in CNS disease (Lei and Wang, 2022). To detect the expression of m⁶A-related genes and proteins in the sensorimotor cortex after SCI, qPCR, western blotting, and immunofluorescence were conducted. As shown in Figure 2A, Mettl14 expression increased on 3 dpi, peaked on 7 dpi, and remained elevated. The western blot results indicated the same for METTL14 protein (Figure 2B and C). Other m⁶A-related genes were also detected. The expression of Mettl3, Ythdf1, Ythdf3, and Wtap increased, while that of Fto decreased (Figure 2D–J). Western blot analysis demonstrated that METTL3, YTHDF1, and YTHDF3 were highly expressed on 7 dpi (Figure 2K–N). Immunofluorescence staining showed that METTL14 was upregulated and mainly expressed in neurons (NeuN⁺ cells); it was less expressed in astrocytes (GFAP⁺ cells) and microglia/macrophages (F4/80⁺ cells) (Figure 2O, P, and Additional Figure 1 ^{(4.5MB, tif)} A and B). The m⁶A quantification assay showed that RNA extracted from the sensorimotor cortex had an elevated m⁶A level (Figure 2Q).

Figure 2.

Mettl14 is highly expressed in the sensorimotor cortex after SCI.

(A) Quantification of relative expression of Mettl14 (normalized by sham) by qPCR (n = 3 per group). (B, C) Quantification of relative expression of METTL14 protein (normalized by sham) by Western blotting (n = 4 per group). (D–J) Quantification of relative expression of Mettl3, Ythdf1, Ythdf2, Ythdf3, Fto, Alkbh5, and Wtap by qPCR (n = 3 per group). (K–N) Quantification of relative expression of METTL3, YTHDF1, and YTHDF3 proteins by Western blotting (n = 3 per group). (O) Representative images of neurons (NeuN, AF488) and METTL14 expression (red, AF594) in cerebral sections of mice in sham and 7 dpi groups, showing that SCI increased the NeuN⁺METTL14⁺ signaling in the cerebral cortex sections in the 7 dpi group. Scale bar: 50 μm. (P) Quantification of number of METTL14⁺NeuN⁺ cells in O (n = 4 per group). (Q) Quantification of total m6A level (n = 3 per group). Data are presented as the mean ± SD. *P < 0.05, **P < 0.01 (one-way analysis of variance followed by Tukey’s post hoc test (A, C) or unpaired t-test (D–J, L–N, P, Q)). Alkbh5: AlkB homolog 5, RNA demethylase; dpi: days post injury; Fto: alpha-ketoglutarate dependent dioxygenase; Mettl14: methyltransferase-like protein 14; Mettl3: methyltransferase-like protein 3; ns: not significant; SCI: spinal cord injury; Wtap: WT1 associated protein; Ythdf1: YTH N⁶-methyladenosine RNA binding protein F1; Ythdf2: YTH N⁶-methyladenosine RNA binding protein F2; Ythdf3: YTH N⁶-methyladenosine RNA binding protein F3.

Mettl14 knockdown in the sensorimotor cortex inhibits corticospinal tract regeneration and worsens functional recovery after spinal cord injury

To explore the effect of METTL14 on CST regeneration and neurological recovery after SCI, we used a Mettl14-lentivirus to knock down Mettl14 expression in sensorimotor cortex (Figure 3A). The knockdown efficiency was verified by immunofluorescence staining and western blot analysis. As Additional Figure 2 ^{(6.4MB, tif)} A and B depict, Mettl14 expression increased in sensorimotor neurons and sh-Mettl14 administration reduced this effect. The results of the western blot represented a similar tendency (Additional Figure 2 ^{(6.4MB, tif)} C and D). m⁶A quantification showed that sh-Mettl14 reduced the m⁶A level in cortical neurons (Additional Figure 2 ^{(6.4MB, tif)} E). Then, we investigated whether METTL14 could influence functional recovery after SCI. BMS scores showed that compared with the control and NC-Mettl14 groups, locomotor recovery in the sh-Mettl14 group was worse at 2 weeks post-injury until the observation endpoint (Figure 3B). The inclined grid walking task indicated that after Mettl14 knockdown, the frequency of hind paws slipping under the grid was increased at 2 weeks after injury until the observation endpoint (Figure 3C). The forced swim test also suggested that Mettl14 knockdown impaired swimming ability after SCI (Figure 3D and E). As shown in Figure 3F–H, compared with the NC-Mettl14 group, after sh-Mettl14 administration, the motor evoked potential decreased and the latent period became longer, which indicated worse neurological connectivity recovery. To explore the effect of sh-Mettl14 on CST regeneration after SCI, we conducted antegrade BDA tracing and calculated the axon density at different distances rostral to the lesion. Mice in the sh-Mettl14 group showed less axon density adjacent to the lesion than those in the SCI + NC-Mettl14 group, indicating a decreased CST regeneration capacity. However, no BDA-labeled CST axons were detected in the coronal section 5 mm caudal to the lesion core (Figure 3I–K). Interestingly, we found no significant difference in synaptophysin (SYN) density in the injured area after SCI; however, sh-Mettl14 treatment reduced the number of SYN⁺BDA⁺ dots, indicating that sh-Mettl14 decreased effective synapse formation (Figure 3L–N). These results suggest that Mettl14 knockdown in the sensorimotor cortex reduced CST regeneration and impaired locomotor function after SCI.

Figure 3.

Mettl14 knockdown in the sensorimotor cortex inhibits CST regeneration and worsens functional recovery after SCI.

(A) Schematic diagram of lentivirus injection sites. Created by Adobe Illustrator 2021 (Adobe, San Jose, USA). (B) BMS scores (n = 6 per group). (C) Error rates of grid walking over time (n = 6 per group). (D) Mouse performance in forced swim test at 28 dpi. (E) Quantification of the swim test in D using the Louisville Swim Scale (n = 6 per group). (F) Representative images in neuroelectrophysiology analysis. (G, H) Quantification of MEP and latent period in F (n = 6 per group). (I) Representative images of BDA tracing, showing that after SCI, the BDA-labeled axon could regenerate adjacent to the epicenter, while Mettl14 knockdown induced the CSTs drawback. Scale bars: 200 μm. (J) Coronal section of the BDA tracing spinal cord 5 mm caudal to the lesion core, showing that no regenerative CSTs could cross the injury site. Scale bar: 200 μm. (K) Quantification of axon densities in the spinal sections rostral to the lesion of sham, NC-Mettl14, and sh-Mettl14 groups (n = 3 per group). (L) Representative images of BDA-labeled CSTs (BDA, green) and synapse (SYN, red, AF594) in sham, NC-Mettl14, and sh-Mettl14 groups, showing that Mettl14 knockdown induced the BDA⁺SYN⁺ signaling loss. Scale bar: 50 μm. (M) Quantification of relative SYN⁺ intensity in K (n = 6 per group). (N) Quantification of number of BDA⁺SYN⁺ dots per 100 μm axons in K (n = 6 per group). The data are presented as means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 (one-way analysis of variance followed by Tukey’s post hoc test (E, G, H, M, N) and two-way analysis of variance followed by Tukey’s post hoc test (vs. sh-Mettl14; B, C, K)). BDA: Biotin dextran amine; BMS: Basso Mouse Scale; CST: corticospinal tract; dpi: days post injury; MEP: motor evoked potential; METTL14: methyltransferase 14; NC: negative control; ns: not significant; SCI: spinal cord injury; Sh: short hairpin; SYN: Synaptophysin.

Mettl14 knockdown decreases axon regeneration in vitro

To investigate the effect of METTL14 on axon regeneration, we transfected primary cortical neurons with lentivirus to knockdown Mettl14 (sh-Mettl14). The results of qPCR and western blotting verified the knockdown efficiency of Mettl14 lentivirus (Figure 4A–C). The m⁶A quantification assay also indicated a decreased m⁶A modification level after sh-Mettl14 treatment (Figure 4D). To observe the effect of sh-Mettl14 on regenerative associated genes, qPCR was performed to detect the expression of genes that promote axonal regeneration (Ntrk3, Cntf, Igf1r, NeuroD1, and Nrg1) and those that inhibit axonal regeneration (Pten, Cers2, and Ptprs). As shown in Figure 4E–L, sh-Mettl14 reduced the expression of genes that promoted axonal regeneration and increased the expression of genes that inhibited axonal regeneration. Immunofluorescence staining showed that Mettl14 knockdown reduced neurite outgrowth of cortical neurons (Figure 4M and N). Next, we used a microfluidics culture plate to further simulate the situation of axonal injury in vivo. After culturing for 7 days, the axons crossing the 450 μm-microtunel were cut off using a 200 μL-pipette (Figure 4O). The results showed that the knockdown of Mettl14 resulted in impaired neuronal axonal regeneration (Figure 4P and Q). Therefore, METTL14 appears to be indispensable for neural growth and regeneration.

Figure 4.

METTL14 knockdown decreases axon regeneration in vitro.

(A) Quantification of relative expression of Mettl14 in NC-Mettl14 and sh-Mettl14 treated neurons by qPCR (n = 3 per group). (B) Western blots of METTL14 in NC-Mettl14 and sh-Mettl14 treated neurons. (C) Quantification of relative protein expression levels in B (n = 3 per group). (D) Total m⁶A level in NC-Mettl14 and sh-Mettl14 treated neurons (n = 3 per group). (E–L) Relative expression of Ntrk3, Cntf, Igf1r, NeuroD1, Nrg1, Pten, Cers2, and Ptprs in NC-Mettl14 and sh-Mettl14 treated neurons by qPCR (n = 3 per group). (M) Representative images of neurites (TUJ1, red, AF594) in NC-Mettl14 and sh-Mettl14 treated neurons, showing Mettl14 knockdown decreased TUJ⁺ signaling. Scale bar: 100 μm. (N) Quantification of neurite length in M (n = 4 per group). (O) Pattern diagram for evaluating axonal regeneration using an axonal transection model. Created by Adobe Illustrator 2021 (Adobe). (P) Representative immunofluorescence images of the regenerated axons on the microfluidic cultures (TUJ1, green, AF488) in which cortical neurons were seeded on the other side, showing that Mettl14 knockdown decreased TUJ1⁺ signaling of the cultured neurons. Scale bar: 200 μm. (Q) Quantification of total axonal length in H (n = 4 per group). The data are presented as means ± SD. **P < 0.01 (unpaired t-test (A, C–L, N, Q)). Cers2: Ceramide synthase 2; Cntf: Ciliary neurotrophic factor; Igf1r: Insulin like growth factor 1 receptor; Mettl14: methyltransferase-like protein 14; NC: negative control; NeuroD1: neurogenic differentiation 1; Nrg1: neuregulin 1; Ntrk3: neurotrophic receptor tyrosine kinase 3; Pten: phosphatase and tensin homolog; Ptprs: protein tyrosine phosphatase receptor type S; qPCR: quantitative polymerase chain reaction; Sh: short hairpin.

METTL14 activates the TRIB2/mitogen-activated protein kinase signaling pathway in an m⁶A-dependent manner

To further explore how METTL14 affected axon regeneration, we analyzed the whole-gene expression changes in sensorimotor neurons after SCI using public datasheets (GSE141583) (Cheng et al., 2022). Datasheet analysis revealed that 74 genes were upregulated and 72 genes were downregulated, as shown in Figure 5A. We listed the top 15 upregulated gene ontology terms in sensorimotor neurons at 7 dpi according to the datasheets, which showed that the MAPK signaling pathway was significantly activated (Figure 5B). Then, we detected the top 10 upregulated genes using qPCR after knockdown of Mettl14 in cortical neurons. We found that Trib2 expression was markedly decreased after Mettl14 knockdown (Figure 5C), indicating that TRIB2 may be a downstream target of METTL14. Next, the western blot showed that after Mettl14 knockdown, TRIB2 protein expression in cortical neurons was decreased (Figure 5D and E). To explore whether METTL14 regulated Trib2 in an m⁶A-dependent manner, we performed meRIP-qPCR (Figure 5F). Three potential sites for m⁶A modifications in Trib2 mRNA were screened: after treatment with sh-Mettl14, the ratio of IP/Input was markedly decreased in sites 1 and 3, indicating that knockdown of Mettl14 decreased m⁶A methylation of Trib2 mRNA in neurons (Figure 5G). Furthermore, the TRIB2/mitogen-activated protein kinase (MAPK) pathway was activated in the sensorimotor cortex after SCI and expression of the TRIB2, p-JNK, and p-P38 proteins was elevated (Figure 5H–K). The results of immunofluorescence staining also suggested that TRIB2 was mainly expressed in neurons and was elevated in sensorimotor neurons after SCI (Figure 5L and M). These results suggest that the TRIB2/MAPK pathway is regulated by METTL14 in an m⁶A-dependent manner.

Figure 5.

METTL14 activates Trib2/MAPK signaling pathway in an m⁶A-dependent manner.

(A) Volcano plot of upregulated and downregulated genes in the sham and 7 dpi groups. (B) The top 15 upregulated GO terms in sensorimotor neurons 7 days after SCI. (C) Relative expression of the top 10 upregulated genes in NC-Mettl14 and sh-Mettl14 treated neurons by qPCR (n = 3 per group). (D) Western blot of TRIB2 in NC-Mettl14 and sh-Mettl14 treated neurons. (E) Quantification of relative protein expression levels in D (n = 3 per group). (F) Pattern diagram of meRIP-qPCR for Trib2 m⁶A modified site detection. (G) m⁶A level of Trib2 in NC-Mettl14 and sh-Mettl14 treated neurons by meRIP-qPCR (n = 3 per group). (H) Western blots of TRIB2, p-JNK and p-P38 protein expression in the sham and SCI groups. (I) Quantification of TRIB2, p-JNK and p-P38 protein expression levels in H (n = 3 per group). (L) Representative images of neurons (NeuN, green, AF488) and TRIB2 expression (red, AF594) in the sham and SCI groups, showing that SCI increased NeuN⁺TRIB2⁺ signaling in the cerebral cortex sections. Scale bar: 50 μm. (M) Quantification of the number of TRIB2⁺NeuN⁺ cells in L (n = 4 per group). The data are presented as means ± SD. *P < 0.05, **P < 0.01 (unpaired t-test (C, E, G, I–K, M)). DAPI: 4′,6-Diamidino2-phenylindole; GO: gene ontology; JNK: Jun N terminal kinase; MeRIP: Methylated RNA immunoprecipitation; m⁶A: N⁶-methyladenosine; Mettl14: methyltransferase-like protein 14; MAPK: mitogen-activated protein kinase; meRIP-qPCR: m⁶A-immunoprecipitation-qPCR; NC: negative control; ns: not significant; p-JNK: phospho-JNK; p-P38: phospho-P38; qPCR: quantitative polymerase chain reaction; SCI: spinal cord injury; Sh: short hairpin; Trib2: tribbles pseudokinase 2.

METTL14 promotes axon regeneration by activating the TRIB2/MAPK pathway

To further investigate whether METTL14 affects axon regeneration by regulating the TRIB2/MAPK pathway, we transfected primary cultured neurons with lentivirus to knockdown Mettl14. As shown in Figure 6A–D, Mettl14 knockdown downregulated TRIB2, p-JNK, and p-P38 protein expression, while OE-Trib2 treatment promoted their expression. Mettl14 knockdown decreased the expression of Ntrk3, Cntf, Igf1r, NeuroD1, and Nrg1 (positive regulation of axon regeneration) and increased the expression of Pten, Cers2, Ptprs (negative regulation of axon regeneration), while OE-Trib2 reduced the effect of Mettl14 knockdown (Figure 6E–L). Immunofluorescence staining showed that Mettl14 knockdown inhibited neurite growth while OE-Trib2 treatment promoted it (Figure 6M and N). The microfluidic plate experiments also indicated that Mettl14 knockdown reduced axonal length, while treating with OE-Trib2 increased it (Figure 6O and P). These results demonstrated that METTL14 appear to be an indispensable factor for axon regeneration via activation of the TRIB2/MAPK signaling pathway.

Figure 6.

METTL14 promotes axon regeneration by activating the Trib2/MAPK pathway.

(A) Western blots of TRIB2, p-JNK and p-P38 proteins in NC-Mettl14, sh-Mettl14, and sh-Mettl14+OE-Trib2 treated neurons. (B–D) Quantification of TRIB2, p-JNK and p-P38 protein expression levels in A (n = 3 per group). (E–L) Relative expression of Ntrk3, Cntf, Igf1r, NeuroD1, Nrg1, Pten, Cers2, and Ptprs in NC-Mettl14, sh-Mettl14, and sh-Mettl14+OE-Trib2 treated neurons by qPCR (n = 3 per group). (M) Representative images of immunofluorescence staining of TUJ1 (red, AF594) in NC-Mettl14, sh-Mettl14, and sh-Mettl14+OE-Trib2 treated neurons, showing that overexpression of TRIB2 counteracted the impact of Mettl14 knockdown. Scale bar: 100 μm. (N) Quantification of neurite length in N (n = 4 per group). (O) Representative immunofluorescence images of the regenerative axons (TUJ1, green, AF488) on the microfluidic cultures, showing that overexpression of TRIB2 counteracted the impact of Mettl14 knockdown. Scale bar: 200 μm. (P) Quantification of total axonal length in G (n = 4 per group). The data are presented as means ± SD. *P < 0.05, **P < 0.01 (one-way analysis of variance followed by Tukey’s post hoc test (B–L, N, P)). Cers2: Ceramide synthase 2; Cntf: ciliary neurotrophic factor; DAPI: 4′,6-diamidino2-phenylindole; Igf1r: insulin like growth factor 1 receptor; JNK: Jun N terminal kinase; MAPK: mitogen-activated protein kinase; Mettl14: methyltransferase-like protein 14; NC: negative control; NeuroD1: neurogenic differentiation 1; Nrg1: neuregulin 1; Ntrk3: neurotrophic receptor tyrosine kinase 3; OE: over-expression; Pten: phosphatase and tensin homolog; Ptprs: protein tyrosine phosphatase receptor type S; Sh: short hairpin; Trib2: tribbles pseudokinase 2.

Syringin acts as a stabilizer of METTL14 and can promote axon regeneration

Based on the above findings, we reasoned that METTL14 was indispensable for neural regeneration and functional recovery after SCI. Considering that METTL14 mainly functions by forming complexes with METTL3, we searched the RCSB PDB database and discovered the three-dimensional structure of the human METTL3/METTL14 complex (PDB ID: 7RX7). Based on the active site predicted by Sitemap, we detected a pocket-like structure in this protein (Figure 7A). To screen the natural small-molecular compounds that target and stabilize METTL14, we performed molecular docking and virtual screening of natural small-molecular compounds from a library which contains more than 35,000 compounds. Three candidate compounds were selected as METTL14 regulators based on the Lipinski Five Principles and binding affinity (Additional Table 4). Among these, syringin is particularly interesting because the results of molecular docking showed that syringin had a high affinity with METTL14 and was able to enter the molecular pocket of human METTL14 (Figure 7B and C). Furthermore, AlphaFold2 (DeepMind, London, UK) was used to obtain the crystal structure corresponding to mouse-derived METTL14 protein. Among the three small-molecule compounds, syringin had the maximum affinity for mouse METTL14 (Additional Figure 3 ^{(1.8MB, tif)} A–D). To investigate whether syringin influences METTL14 protein expression, cortical neurons were treated with syringin at different concentrations. Western blotting showed that 40 μM syringin increased METTL14 expression in cortical neurons in vitro (Figure 7D and E). As shown in Figure 7F, after cycloheximide was added to inhibit protein synthesis, the expression level of METTL14 was higher in the syringin group than the DMSO group over time, indicating that syringin acted as a stabilizer of METTL14. Next, the results of qPCR showed that syringin treatment promoted the expression of Ntrk3, Cntf, Igf1r, NeuroD1, and Nrg1 (positive regulation of axon regeneration) and decreased the expression of Pten, Cers2, and Ptprs (negative regulation of axon regeneration) (Figure 7G–N). Immunofluorescence staining showed that syringin greatly promoted neurite outgrowth (Additional Figure 4 ^{(1.9MB, tif)} A and B). In the axonal regenerative assay, syringin also facilitated robust axonal regeneration after axonal rupture (Figure 7O and P). Interestingly, we found that syringin did not affect neural apoptosis, as detected by flow cytometry (Additional Figure 4 ^{(1.9MB, tif)} C and D). These results provided solid evidence that syringin can promote axon regeneration in vitro by stabilizing METTL14.

Figure 7.

Syringin acts as a stabilizer of METTL14 and could promote axon regeneration.

(A) The optimized homology modeling structure of the METTL3/METTL14 complex. The blue and red structure is the predicted small molecule (presented as white balls) binding pocket that consists of 25 amino acids. (B) The molecular structure of syringin. (C) The optimized binding mode with the lowest binding energy between syringin and the METTL3/METTL14 complex. The images were created by Maestro 13.5 (Schrödinger, New York, USA). (D) Western blots of METTL14 in syringin-treated neurons at different concentrations. (E) Quantification of relative protein expression levels in D (n = 3 per group). (F) The protein levels of METTL14 in DMSO and syringin in pre-treated neurons treated with cycloheximide (CHX). (G–N) Relative expression of Ntrk3, Cntf, Igf1r, NeuroD1, Nrg1, Pten, Cers2, and Ptprs in DMSO and syringin treated neurons by qPCR (n = 3 per group). (O) Representative immunofluorescence images of the regenerative axons (TUJ1, green, AF488) on the microfluidic cultures, showing that syringin treatment improved the regenerative axonal signaling. Scale bar: 200 μm. (P) Quantification of total axonal length in O (n = 4 per group). The data are presented as means ± SD. **P < 0.01, ***P < 0.001 (one-way analysis of variance followed by Tukey’s post hoc test (E) or unpaired t-test (G–N, P)). Cers2: Ceramide synthase 2; Cntf: ciliary neurotrophic factor; DMSO: dimethyl sulfoxide; Igf1r: insulin like growth factor 1 receptor; METTL14: methyltransferase 14; METTL3: methyltransferase 3; NeuroD1: neurogenic differentiation 1; Nrg1: neuregulin 1; Ntrk3: neurotrophic receptor tyrosine kinase 3; ns: not significant; Pten: phosphatase and tensin homolog; Ptprs: protein tyrosine phosphatase receptor type S; qPCR: quantitative polymerase chain reaction.

Syringin promotes corticospinal tract regeneration and functional recovery after spinal cord injury

To investigate the potential effect of syringin on promoting CST regeneration and neurological recovery after SCI, syringin or vehicle was administered to mice (100 mg/kg per day) by gavage for 1 month (Figure 8A; Li et al., 2017). First, we assessed whether syringin administration had any obvious adverse effects. As shown in Additional Figure 5 ^{(2.3MB, tif)} A and B, syringin administration did not affect body weight or fur color, nor did it result in appreciable cardiotoxicity, hepatotoxicity, spleen cell toxicity, pulmonary toxicity, or nephrotoxicity based on H&E staining (Additional Figure 5 ^{(2.3MB, tif)} C). Immunofluorescence staining showed that syringin upregulated METTL14 expression in sensorimotor neurons (Additional Figure 6 ^{(1.3MB, tif)} ). BMS score analysis showed that syringin improved functional recovery from the 2^nd week to the observation endpoint (Figure 8B). Compared with the vehicle group, the syringin group showed a lower frequency of hind paw slipping under the grid at 3 weeks and thereafter (Figure 8C). The swimming test suggested that syringin administration improved swimming ability as well (Figure 8D and E). As shown in Figure 8F–H, the mice in the syringin-treated group showed higher motor evoked potentials with no significant change in the latent period compared to the vehicle group. Next, we investigated the effect of syringin on CST regeneration and effective synaptic formation. Compared with the vehicle group, the axon density near the lesion was higher in the syringin-treated group; some axons even crossed the lesion, indicating that the regeneration ability of CST was enhanced (Figure 8I–K). Immunofluorescence staining showed that syringin did not affect synaptophysin density in the injured area; however, it increased the number of BDA⁺SYN⁺ dots, indicating that syringin promoted effective synaptic formation (Figure 8L–N). These results suggest that syringin is an ideal small molecule that promoted CST regeneration and improved neurological recovery after SCI.

Figure 8.

Syringin promotes CST regeneration and functional recovery after SCI.

(A) Schematic of the time in which mice were orally administrated syringin or vehicle. Created by Adobe Illustrator 2021 (Adobe). (B) BMS scores in vehicle and syringin-treated groups (n = 6 per group). (C) Error rates of grid walking over time post-injury in the vehicle and syringin-treated groups (n = 6 per group). (D) Mouse performance in the forced swim test at 28 dpi in the vehicle and syringin-treated groups. (E) Quantification of the forced swim test using the Louisville Swim Scale (n = 6 per group). (F) Representative images in neuroelectrophysiology analysis. (G, H) Quantification of MEP and latent period in F (n = 6 per group). (I) Representative images of BDA tracing in the vehicle and syringin-treated groups, showing that more BDA-positive signaling was detected adjacent to the epicenter in the syringin-treated groups. The white arrow indicates the regenerative CSTs. Scale bar: 200 μm. (J) Coronal section of the BDA tracing spinal cord 5 mm caudal to the lesion core in the vehicle and syringin-treated groups; no BDA-positive signaling was detected caudal to the lesion core in either group. The white arrow indicates the regenerative CSTs. Scale bars: 200 μm, 100 μm (J1 and J2). (K) Quantification of axon densities at different distances rostral to the lesion of the vehicle and syringin-treated groups (n = 3 per group). (L) Representative images of BDA-labeled axons (BDA, green, FITC) and synapse (SYN, red, AF594) in the vehicle and syringin-treated groups, indicating more BDA⁺SYN⁺ axons was detected in the syringin-treated group. Scale bars: 50 μm. (M) Quantification of relative SYN⁺ intensity in K (n = 6 per group). (N) Quantification of the number of BDA⁺SYN⁺ dots per 100 μm axons in K (n = 6 per group). The data are presented as means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001 (two-way analysis of variance followed by Tukey’s post hoc test (B, C, K) or unpaired t-test (E, G, H, M, N)). BDA: Biotin dextran amine; BMS: Basso Mouse Scale; CST: corticospinal tract; Dpi: days post injury; FITC: fluorescein isothiocyanate; MEP: motor evoked potential; ns: not significant; SCI: spinal cord injury; ST: corticospinal tract; SYN: synaptophysin.

Discussion

After SCI, mice experience a certain degree of functional recovery, indicating spontaneous neural regeneration (Danilov and Steward, 2015). However, the internal regeneration mechanisms deserve further elucidation. In this research, we found that spontaneous regeneration of the CST was accompanied by upregulation of METTL14 expression. METTL14 promoted CST regeneration by activating the TRIB2/MAPK pathway in an m⁶A-dependent manner. Using a METTL14 stabilizer, syringin could enhance CST regeneration, thereby facilitating neurological recovery after SCI. Our findings revealed a new regulation model of CST regeneration after SCI.

The corticospinal motor circuitry, composed of cortical motor neurons, CSTs, and spinal cord neurons, mainly regulates fine and coordinated motor movements (Kadoya et al., 2016; Song and Martin, 2023). Damage or loss of any component of this circuit will impair fine motor activity (Serradj et al., 2017; Kazim et al., 2021). Previous researchers have proposed that rupture of the CST leads to sensorimotor neuron apoptosis, resulting in massive neuronal loss (Hains et al., 2003; Liu et al., 2014b). Recent studies have refuted this. Although loss of limb function can lead to atrophy of corresponding brain regions, retrotracking of the CST neurons did not decrease after SCI (Nielson et al., 2010, 2011). Thus, considering that cortical motor neurons could be preserved in cases of distal axonal rupture and functional loss, stimulating CST regeneration has become a focus of SCI research. Synaphilin knockout in neurons can improve axonal mitochondrial transport, thereby promoting CST regeneration after SCI (Han et al., 2020). In another study, co-deletion of Pten and RhoA was conducive to remodeling the corticospinal circuits through regulating intrinsic and extrinsic factors (Nakamura et al., 2021). We also found that the METTL14 stabilizer syringin can promote CST regeneration.

m⁶A is the most frequent post-transcriptional modification in eukaryotic cells at the RNA level. m⁶A is a dynamic epigenetic modification which is mainly regulated by methyltransferases (writers, including METTL3, METTL14, and WTAP), demethylases (erasers, including FTO and ALKBH5), and RNA binding proteins (readers, including various YTH domain–containing proteins) (Jiang et al., 2021; Murakami and Jaffrey, 2022). METTL3 regulates the expression of EZH2 in an m⁶A-dependent manner, thereby affecting the proliferation and differentiation of neural stem cells (Chen et al., 2019). Mettl3 knockout induces apoptosis of newborn cerebellar granule cells, leading to cerebellar hypoplasia (Wang et al., 2018a). m⁶A modification plays an important role in CNS development as well as CNS disease (Livneh et al., 2020; Liu et al., 2022a; Tian et al., 2022). In peripheral nervous system injury, the knockdown of Mettl14 or Ythdf1 significantly reduces axonal regeneration in dorsal root ganglion cells by reducing protein synthesis (Weng et al., 2018). m⁶A modifications are also widely involved in acute and chronic SCI (Ni et al., 2022; Li et al., 2023). After SCI, Mettl14 knockdown can reduce neuronal apoptosis and improve neurological function recovery by overexpressing EEF1A2 or preventing miR-375 from maturing (Wang et al., 2021; Gao et al., 2022). In our study, we found for the first time that m⁶A modification alterations occurred in the sensorimotor cortex after SCI; in addition, Mettl14 knockdown significantly decreased CST regeneration. These findings indicate that METTL14 might have opposite effects in cortical neurons and spinal cord neurons.

Through bioinformatics analysis, we found that TRIB2 expression was increased and the MAPK signaling pathway was activated in sensorimotor neurons after SCI. MeRIP-qPCR confirmed that METTL14 could promote TRIB2 expression by increasing m⁶A methylation of Trib2 mRNA. TRIB2 is a pseudo serine/threonine kinase that plays an important role in a variety of physiological and pathological processes (Liang et al., 2016; Mayoral-Varo et al., 2021). A recent study confirmed that TRIB2 also has kinase activity, which can regulate glucose metabolism through phosphorylation of the M2 isoform of pyruvate kinase M2, thus promoting the proliferation of cancer cells (Liu et al., 2022b). Elevated expression of TRIB2 promotes hepatic stellate cell activation by regulating the hippo signaling pathway to promote the development of liver fibrosis (Xiang et al., 2021). In the CNS, the role of TRIB2 in narcolepsy has been widely described (Viorritto et al., 2012; Katzav et al., 2013; Tanaka et al., 2017). However, whether it is involved in SCI and subsequent secondary pathological processes has never been studied. Our study confirmed that Trib2 is regulated by METTL14 in an m⁶A-dependent manner after SCI. METTL14 regulated axonal growth by promoting the expression of TRIB2, which in turn activated the MAPK pathway. MAPK activation is required for CST axon regenerative growth and mediates long-distance axon outgrowth in developing peripheral neurons (Markus et al., 2002; Zhong et al., 2007; Hollis et al., 2009). Our research shows that after SCI, METTL14 activates the TRIB2/MAPK pathway to promote CST regeneration.

Our findings suggest that METTL14 might be a potential molecular target to modulate CST regeneration after SCI. Based on this molecular pathway, we performed molecular docking and virtual screening to identify naturally occurring small-molecule compounds that can bind METTL14 and further promote neural regeneration. We found that syringin had great potential in axon regeneration. Syringin is a major bioactive phenolic glycoside in Acanthopanax senticosus with anti-osteoporosis activity (Wang et al., 2022) and prevents cardiac hypertrophy induced by pressure overload through the attenuation of autophagy (Li et al., 2017). Our results showed that syringin treatment improved the stability of METTL14 and promoted axonal extension after SCI. Although large animal safety data is lacking, syringin appears to be an important candidate compound in SCI drug design.

There are still limitations in this study. We investigated the role of Mettl14 in corticospinal tract regeneration using virus transfection. Perhaps in the future, constructing transgenic mice with corticospinal tract specific knockout may enhance the persuasiveness of the mechanism explored in this study. And oral administration has limitations such as poor compliance and first-pass elimination. Therefore, exploring more efficient drug delivery methods is our future research goal.

Collectively, this study demonstrated that METTL14 regulates CST regeneration through the TRIB2/MAPK pathway after SCI and administration of the METTL14 stabilizer syringin improves neural functional recovery, which may provide a potential therapeutic option for SCI.

Additional files:

Additional Figure 1 ^{(4.5MB, tif)} : METTL14 expression after SCI.

Additional Figure 1

METTL14 expression after SCI.

(A) Representative immunofluorescent images of astrocytes (GFAP, green, AF488) and METTL14 expression (red, AF594) in sham and 7 dpi group, showing that METTL14 rarely expressed in astrocytes in both groups. (B) Representative images of macrophage (F4/80, green, AF488) and METTL14 expression (red, AF594) in sham and 7 dpi group, showing that METTL14 rarely expressed in macrophage in both groups. Scale bars: 50 μm. DAPI: 4’,6-diamidino2-phenylindole; dpi: days post-injury; GFAP: glial fibrillary acidic protein; METTL14: methyltransferase 14; SCI: spinal cord injury.

Additional Figure 2 ^{(6.4MB, tif)} : Knockdown efficiency of METTL14 lentivirus in vivo.

Additional Figure 2

Knockdown efficiency of METTL14 lentivirus in vivo.

(A) Representative images of neurons (NeuN, green, AF488) and METTL14 expression (red, AF594) in the sham, SCI + NC-Mettl14, and SCI + sh-Mettl14 groups, showing that knockdown of METTL14 decreased the NeuN⁺METTL14⁺ signaling in the cerebral cortex sections. Scale bar: 50 μm. (B) Quantitation of number of METTL14⁺NeuN⁺ cells per field in A (n = 4 per group). (C) Western blots of METTL14 in the sham, SCI + NC-Mettl14, and SCI + sh-Mettl14 group. (D) Quantification of relative protein expression levels in C (n = 3 per group). (E) The total m⁶A level in the sham, SCI + NC-Mettl14, and SCI + sh-Mettl14 groups. The data are presented as the mean ± SD. ^**P < 0.01 (one-way analysis of variance followed by Tukey’s post hoc tests). DAPI: 4’,6-Diamidino2-phenylindole; METTL14: methyltransferase 14; NC: negative control; SCI: Spinal cord injury; Sh: short hairpin.

Additional Figure 3 ^{(1.8MB, tif)} : The optimized binding modes of the small molecular with METTL14.

Additional Figure 3

The optimized binding modes of the small molecular with METTL14.

(A) The optimized binding mode with the lowest binding energy between Evodosin A and METTL3/METTL14 complex. (B) The optimized binding mode with the lowest binding energy between Salviaplebeiaside and METTL3/METTL14 complex. (C) The binding score of the mouse-METTL14 with Syringin, Evodosin A and Salviaplebeiaside. (D) The optimized binding mode with the lowest binding energy between Syringin and mouse-METTL14. METTL3: Methyltransferase 3; METTL14: methyltransferase 14.

Additional Figure 4 ^{(1.9MB, tif)} : Syringin promotes neurite outgrowth, but does not affect neuronal apoptosis.

Additional Figure 4

Syringin promotes neurite outgrowth, but does not affect neuronal apoptosis.

(A) Representative images of the immunofluorescent stain of neurites (TUJ1, red, AF594) in the vehicle and Syringin groups, showing that the Syringin treatment improved the neurite outgrowth. Scale bar: 100 μm. (B) Quantification of neurite length in A (n = 4 per group). (C) Flow cytometry analysis of the neurons treated with vehicle and Syringin. (D) Quantification of the percentage of apoptotic neurons in C (n = 4 per group). The data are presented as the mean ± SD. ^*P < 0.05 (unpaired Student’s t-test). DAPI: 4’,6-Diamidino2-phenylindole; ns: not significant; PI: propidium iodide.

Additional Figure 5 ^{(2.3MB, tif)} : Physiology of mice treated with Syringin.

Additional Figure 5

Physiology of mice treated with Syringin.

(A) The body weight of mice treated with Syringin and vehicle controls. The data are presented as the means SD, and were analyzed by unpaired Student’s t-test. (B) The gross pathology of mice treated with Syringin and vehicle controls, showing no fur color difference were detected between the two groups. (C) Representative images of H&E-staining of heart, liver, spleen, lung, and kidney tissue showed no differences between vehicle treatment and Syringin treated group, showing that Syringin treatment did not result in appreciable cardiotoxicity, hepatotoxicity, spleen cell toxicity, pulmonary toxicity, and nephrotoxicity by H&E staining. Scale bar: 50 μm. H&E: Hematoxylin-eosin; ns: not significant.

Additional Figure 6 ^{(1.3MB, tif)} : Syringin promoted the expression of METTL14 in cortical neurons.

Additional Figure 6

Syringin promoted the expression of METTL14 in cortical neurons.

(A) Representative images of neurons (NeuN, green, AF488) and METTL14 expression (red, AF594) in vehicle and Syringin treated groups, showing more NeuN⁺/METTL14⁺ signaling in Syringin treated groups. Scale bar: 50 μm. (B) Quantitation of METTL14⁺NeuN⁺ cells in A (n = 4 per group). The data are presented as the mean ± SD. ^**P < 0.01 (unpaired Student’s t-test). DAPI: 4’,6-Diamidino2-phenylindole; METTL14: methyltransferase 14.

Additional Table 1: The antibodies used in the study.

Additional Table 2: Primers for qPCR.

Additional Table 3: Primers for m⁶A-qPCR.

Additional Table 4: Top 35 natural small molecule compounds selected from database.

Funding Statement

Funding: This study was supported by the National Natural Science Foundation of China, Nos. 82030071 (to JH), 82272495 (to YC); Science and Technology Major Project of Changsha, No. kh2103008 (to JH); and Graduate Students’ Independent Innovative Projects of Hunan Province, No. CX20230311 (to YJ).

Data availability statement:

All relevant data are within the paper and its Additional files.