RUNX1 Induces Central Neuropathic Pain by Activating Microglia and Triggering the Inflammatory Response in Spinal Cord Injury

Abstract

Runt-related transcription factor 1 (RUNX1) is a highly conserved transcription factor involved in immune regulation, inflammation, and nociceptive neuron differentiation. However, its role in central neuropathic pain (CNP) induced by SCI remains unclear. Here, we investigated RUNX1 expression after SCI and its impact on CNP. Using scAAV-mediated RUNX1 knockdown in the spinal cord of SCI rats, we assessed nociceptive behaviors of mechanical allodynia and thermal hyperalgesia by hind paw withdrawal threshold (PWT) and paw withdrawal latency (PWL) before injury and at 21-, 35-, and 49-days post-injury. RUNX1 mRNA and protein levels were analyzed via RT-qPCR and Western blotting, while immunohistochemistry was used to examine its distribution. Our findings revealed that RUNX1 expression was significantly elevated in microglia following SCI. RUNX1 knockdown notably increased PWT and PWL as well as a significant decrease in the aberrant sprouting of nociceptive fibers in SCI rats. Furthermore, in lipopolysaccharide (LPS)-treated BV-2 microglia, RUNX1 knockdown markedly reduced microglial activation and inflammation. A similar reduction in microglial activation and neuroinflammation was observed in SCI rats following RUNX1 knockdown. These findings suggest that RUNX1 contributes to CNP after SCI by promoting microglial activation and neuroinflammation, identifying it as a potential therapeutic target for SCI-induced neuropathic pain.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10753-025-02336-y.

Introduction

With the continuous annual advancements in medicine, the survival rate subsequent to spinal cord injury (SCI) has experienced a remarkable increase. Nevertheless, a series of complications, such as pain, spasticity, intestinal disorders, pressure ulcers, and emotional problems, persistently exert a substantial influence on the quality of life of patients [1]. Central neuropathic pain (CNP), frequently considered the most severe form of pain ensuing from SCI, is defined as “pain resulting from a lesion or disease of the somatosensory nervous system” [2]. It encompasses spontaneous pain as well as evoked pain, which is characterized by allodynia (nociceptive responses to non-noxious stimuli) and hyperalgesia (augmented responses to noxious stimuli).

Patients with CNP report higher pain intensity, longer pain duration, and lower quality of life compared to those with non-neuropathic pain [3]. They consume more medical resources, such as undergoing surgeries and using painkillers. Moreover, they are accompanied by a greater variety and more severe complications, including depression and sleep disorders, which impose a burden on families and society [4–6]. However, the existing treatment modalities, such as antiepileptic and antidepressant medications, surgical interventions, and behavioral therapies, yield only a very limited reduction in neuropathic pain. Additionally, they often lead to the development of tolerance or give rise to side effects.

Microglia, the resident macrophages within the central nervous system (CNS), contribute to the maintenance of homeostasis and are implicated in the pathogenesis of numerous CNS disorders, among which persistent pain is included [7, 8]. In the superficial dorsal horn after SCI, the quantity of both resident microglia and macrophages originating from peripheral blood monocytes increases, and these cells swiftly transform from a quiescent state to an activated one, with such activation enduring for a period spanning from weeks to months. SCI not only causes the destruction of the anatomical structure and changes in the microenvironment in the injured area but may also affect the structure and function of spinal cord segments caudal to the injury level. The continuous activation of microglia and macrophages in the lumbar spinal dorsal horn can induce and maintain below-level SCI-induced CNP [9, 10]. Some interventions, for instance, the intrathecal administration of glial inhibitors such as minocycline and propentofylline, can alleviate below-level CNP induced by SCI, inducing the reversion of microglia to a resting morphological phenotype and attenuating the electrophysiological and behavioral manifestations associated with pain [11, 12]. Once activated, microglia and macrophages play a significant role in neuroinflammation by inducing a series of cellular responses [10, 13]. The increased secretion of pro-inflammatory mediators, such as inflammatory cytokines and chemokines, establishes and maintains SCI-related CNP. For example, the increase in TNF-α, IL-6 and IL-1β is associated with pain behaviors [14, 15]. Targeting inflammation after SCI may improve regeneration and functional recovery as well as alleviate pain [16–19].

Runt-related transcription factor 1 (RUNX1), which is also referred to as core-binding factor subunit α2 (CBFA2) or acute myeloid leukemia 1 (AML1), is a Runt domain transcription factor [20]. As a key regulator of myeloid progenitor cell proliferation and differentiation, RUNX1 is a highly conserved transcription factor that modulates immune response, inflammatory response, embryological development, angiogenesis, hematopoiesis and tumorigenesis [21–24]. RUNX1 is essential for the differentiation of nociceptive sensory neurons and expressed in most nociceptors during embryonic development but in adult mice, becomes restricted to nociceptors marked by expression of the neurotrophin receptor Ret [25]. In addition, RUNX1 plays a regulatory role in the expression of numerous nociceptive transduction ion channels and receptors, including thermal receptors of the transient receptor potential (TRP) class, sodium-gated, adenosine triphosphate (ATP)-gated, and hydrogen-gated channels, the opioid receptor (MOR), as well as members of the Mas-related G protein-coupled receptor (Mrgpr) class of G protein-coupled receptors [25–27].

The relationship between RUNX1, inflammation, and pain has been investigated in several models [25, 28, 29], but its expression pattern following SCI and its role in SCI-induced CNP remains unclear. To address this, we characterized the temporal expression and cellular localization of RUNX1 in the spinal cord following spinal cord contusion. We then assessed its impact on pain behaviors and C-fiber alterations. Further in vivo and in vitro experiments examined how RUNX1 expression influences microglial activation and proinflammatory cytokine production. Our findings suggest that RUNX1 contributes to neuropathic pain after SCI by promoting microglial activation and inflammation. This study provides new insights into RUNX1’s role in nociception and offers a foundation for developing targeted therapies for SCI-induced CNP.

Materials and methods

Subjects

Pathogen-free adult male Sprague Dawley rats, which were 6 weeks old upon arrival and sourced from Peking University Health Science Center, were utilized in all experiments. These rats were housed in a pathogen-free and precisely temperature-controlled (23 ± 3℃) environment maintained on a 12-hour light/dark cycle, with ad libitum access to food and water. All procedures and protocols were approved by the Institutional Animal Care and Use Committee of Peking University (SA2020180). This study also adhered to the guidelines of the Committee for Research and Ethical Issues of IASP published in PAIN^®, 16 (1983) 109–110. The rats were acclimatized to the environment for 7 days prior to being introduced to the behavioral apparatus and undergoing a series of experimental procedures.

Spinal Cord Injury (SCI) Surgery and Intraspinal Microinjection

CNP was induced by means of the spinal cord contusion model [11, 30–33]. Rats were anesthetized intraperitoneally with 2% sodium pentobarbital (0.3 ml per 100 g body weight). The surgical procedures for inducing SCI involved a dorsal midline incision in the skin from T4-L2, followed by exposing the spinous processes and vertebral laminae through dissection of the bilateral vertebral muscles. A T8-9 laminectomy was conducted with rongeurs to expose the dorsal surface of the T10 spinal cord while keeping the dura intact. The rostral and caudal vertebral bodies were attached with spinal adapter (RWD, Shenzhen, China) in order to stabilize the spinal cord, and then the laminectomized section was positioned under the impactor tip of a PSI-IH precision striking device (IH impactor; Precision Systems and Instrumentation, Lexington, KY, USA), which could manipulate the injury severity by precise control of strike velocity, contusion depth and contusion time. Surgery was performed at the T10 level of the spinal cord with a force of 200 kilodynes, 0 s dwell time to induce moderate contusion. The incision was cleaned with saline, and the muscles and skin were sutured layer by layer. Rats were placed on a heating pad to maintain the body temperature and promote awakening after surgery. For postoperative animal care, penicillin sodium (once a day for 7 days) was intraperitoneally injected and lactate Ringer’s solution (5 ml/100 g, once a day for 3 days) was administered subcutaneously. The bladder was manually expressed until spontaneous urination recovered.

Immediately after SCI surgery, a glass capillary filled with scAAV-scrambled or scAAV-shRUNX1 viral solution purchased from OBiO Co., Ltd. (Shanghai, China) was injected into the spinal cord dorsal horn in 4 sites along the spinal cord (0.5 µl/site, flow rate of 0.2 µl/min). The needle was positioned on the entry zone of the dorsal root, and it was deepened into the parenchyma to 0.3 mm below the pia mater to reach the dorsal horn. To prevent reflux, the glass capillary was slowly withdrawn after remaining in place for at least 2 min after the injection.

Behavioral Testing

All behavioral tests were performed in awake, unrestrained rats by the same experienced examiners blinded to group assignment. Before assessment, all rats were habituated to the experimental sets and test environment for several hours a day over a total of 3 days and at least 30 min immediately prior to each test. The testing sequence was arranged such that the von Frey test was conducted first, followed by the thermal pain test. Additionally, the rat must exhibit one or more signs of supraspinal awareness such as vocalizing, licking, looking at, or guarding the paw, or moving away from the stimulus after the application of the stimulation that elicits a paw withdrawal response. Baseline and post-interventional measurements were performed at the same time point and the latencies and thresholds were averaged for both hind paws.

Mechanical Allodynia

Mechanical allodynia was tested using the von Frey test. Rats were placed in independent plexiglass chambers on a shelf with elevated wire mesh. The hind paw withdrawal threshold (PWT) to punctate mechanical stimulation was recorded as the lowest force (g) that elicited a hindpaw withdrawal response accompanied by supraspinal behaviors in 50% of the applications. The left and right glabrous plantar surface of the hind paws were stimulated with graded von Frey filaments (Touch Test Sensory Evaluators, North Coast Medical Inc, Gilroy, CA) with forces, ranging from the manufacturer-designated 3.61 (0.4 g) to 5.18 (15.0 g). Filaments were applied with pressure to bend and maintained for a duration of 3 s. Alternatively, the application was terminated when the animal retracted its hindpaw without ambulation, which was considered a positive withdrawal response. An interval of at least 30 s was needed between every two stimuli. The von Frey test [34] was performed as previously described with the up-down paradigm. Experimental parameters were measured a maximum of 9 times for each rat, or stopped when four consecutive positive or negative responses occurred. The PWT was tested before and after surgery on days 21, 35 and 49 and the values were averaged to quantify alterations in mechanical sensitivity.

Thermal Hyperalgesia

Thermal hyperalgesia was evaluated according to Hargreaves’ method as previously described. Briefly, the rats were placed in independent chambers on a glass plate, and a radiant heat stimulus (intensity of 40) (Zhongshi, Beijing, China) was applied onto the glabrous surface of the hind paw through the glass plate 4 times each side. When the tested hind paw moved or flinched, the light beam was turned off and the paw withdrawal latency (PWL) in seconds was recorded. Thermal PWL was defined as the time interval from the moment the stimulus started to the moment the paw stopped. The intensity of the light beam was adjusted to achieve baseline latency between 10 and 12 s and a cut-off time of 20 s was imposed to prevent tissue damage. Between trials for the same paw, at least 3 min intervals were needed to limit sensitization and the paw testing order was chosen randomly to prevent an order effect. The left and right paw measurements of each rat were averaged to determine the mean PWL and the testing time was the same as von Frey test.

Lentivirus Construction

Lentivirus was generated by transfecting 70–80% confluent HEK293T cells. Briefly, HEK293T cells were co-transfected with lentiviral packaging plasmids and the shRNA plasmid targeting the RUNX1 sequence for 48 h, which was purchased from Shanghai Genomeditech (Shanghai, China) and verified by DNA sequencing. zKirus (Zhongke, Guangzhou, China) was used to assist the transfection. The cell supernatants were collected, centrifuged, and concentrated to obtain the lentiviral working solution. The sequences of shRNAs are listed in the supplementary Table S1.

Cell Culture, Stable Cell Lines Establishment and LPS Treatment

Mouse microglia BV-2 cells were incubated at 37 ℃ with 5% CO₂ and grown in DMEM medium supplemented with 10% FBS and 1% penicillin-streptomycin. The stable BV-2 cell lines with RUNX1 gene knockdown were constructed through lentivirus infection. In brief, after the cells seeded in 24-well culture plates reached about 60% confluence, LV-Scrambled or LV-shRUNX1 lentivirus and 8 µg/ml polybrene were used to replace the previous culture medium for transfection. After 6 to 8 h of transfection, the medium containing virus was removed and the cells were cultured in DMEM medium supplemented with 10% FBS for an additional 24 h. The lentivirus positive cells were then selected using DMEM medium supplemented with 10% FBS and 2.5 µg/ml puromycin. Positive colonies were picked up and further cultured. The stable BV-2 cell lines were treated with 1 µg/ml LPS in DMEM medium for 6 h to stimulate an inflammatory response. Cells were collected by centrifugation for RT-qPCR and Western blotting, whereas supernatants were collected for nitrite and cytokine concentration analysis.

Tissue Processing

The segments of spinal cord, which were 1 cm in length and centered on the injury site, and the lumbar enlargement (L4-L5) spinal cord tissues of rats were quickly removed on an ice-cold plate and stored at −80 ℃ until extraction. The tissues were carefully cut into the smallest pieces possible and homogenized for subsequent RNA and protein extraction.

For subsequent histopathological staining, the rats were euthanized and intracardially perfused with about 100 ml saline, followed by a rapid perfusion of 50 ml and then a slow perfusion of 150 ml of 4% paraformaldehyde (Biosharp, Beijing, China). The injury epicenter and lumbar (L4-L5) spinal cords were removed and post-fixed with the same fixative, and placed in 10%, 20% and 30% sucrose solutions for gradient dehydration at 4 ℃.

Real-time PCR

Spinal cord and BV-2 cells were used for the gene expression analysis over time by RT-qPCR. Total RNA from rats’ spinal cord and BV-2 cells was extracted using standard method of TRIzol Reagent (Tiangen, Beijing, China) according to the manufacturer’s instructions. RNA concentrations of the obtained tissues and cells were determined using NanoDrop Microvolume (Thermo Scientific, Massachusetts, USA). 2 µg total RNA from each sample was reverse transcribed to cDNA using FastKing gDNA Dispelling RT SuperMix kit (Tiangen, Beijing, China). Reaction conditions were 42 ℃ for 15 min and 95 ℃ for 3 min. The RT-qPCR was performed in QuanStudio 5 real-time PCR system (Thermo Scientific, Massachusetts, USA) using SuperReal PreMix Plus (SYBR Green) (Tiangen, Beijing, China). The PCR conditions were an initial incubation of 95 ℃ for 15 min, followed by 40 cycles of 95 ℃ for 10 s and 60 ℃ for 32 s. The primer sequences used were custom synthesized (Xianghong, Beijing, China) based on the published sequences or GenBank accession numbers (supplementary Table S1). The gene expression level of the target mRNA was quantified relative to the internal control (GAPDH for rats and β-actin for mice) and differences in gene expression were presented as fold-changes relative to the control group. Cycle threshold (CT) was quantified as the relative amounts of amplified transcripts. Relative quantification was calculated as .

Western Blotting

Total proteins from BV-2 cells and thoracic spinal cord segments were extracted by homogenization in ice-cold RIPA lysis buffer (Applygen, Beijing, China) containing protease inhibitor (Applygen, Beijing, China). Protein concentration was measured using the bicinchoninic acid (BCA) protein assay kit (Applygen, Beijing, China). Samples were heated for 5–10 min at 100 ℃ with 5× loading buffer. Equal amounts of proteins (20–50 µg for spinal cord samples, 40 µg for cell samples) were separated by suitable SDS-PAGE separation gels according to different molecular weights of target proteins and then transferred onto PVDF membranes (Millipore, Sigma, Burlington, USA). The membranes were saturated in blocking solution (5% BSA in 1×TBST) for 1.5 h at room temperature prior to incubation with primary antibodies: rabbit anti-RUNX1 (Abcam, Cambridge, UK), rabbit anti-IL-1β (Proteintech, Wuhan, China), rabbit anti-TNF-α (Proteintech, Wuhan, China), rabbit anti-IL-6 (Abcam, Cambridge, UK), rabbit anti-actin (Abcam, Cambridge, UK), mouse anti-GAPDH (CST, Boston, USA) in the antibody dilution buffer overnight at 4 ℃. After rinsing in TBST three times, the membranes were incubated with appropriate horseradish peroxidase-conjugated secondary antibodies (Beyotime, Shanghai, China) for 1 h at room temperature. The chemiluminescent signal was visualized using ECL or ECL Plus reagents (Yeasen, Shanghai, China) and detected with a chemiluminescence detection system (Tanon, Shanghai, China). The intensity of the immunoreactive bands was assessed via ImageJ software (https://imagej.net/ij/index.html).

Immunohistochemistry

The injured spinal cord segments were embedded in paraffin and sectioned into serial Sect. (5 µm), deparaffinized in 100% xylene, rehydrated in an ethanol-wash series, and washed in PBS. The slides were boiled in 10 mM sodium citrate buffer (pH 6.0) for 5 min for heat-induced antigen retrieval, and subsequently cooled at room temperature for about 30 min. Slides were washed with PBS and incubated with the primary antibody rabbit anti-RUNX1 (Abcam, Cambridge, UK) at 37 ℃ for 2 h. After being washed three times with PBS, the slides were incubated with the enzyme-labeled goat anti-rabbit IgG secondary antibody (ZSGB-BIO, Beijing, China) at room temperature for 30 min. An appropriate amount of freshly prepared 3,3’-Diaminobenzidine (DAB) chromogenic solution (ZSGB-BIO, Beijing, China) was added. After rinsing with tap water, the cell nuclei were counterstained with hematoxylin. Subsequently, the sections were dehydrated and mounted with neutral resin. The completely dried sections were scanned by a NanoZoomer S210 (Hamamatsu, Shizuoka, Japan) to obtain images.

Immunofluorescence

The spinal cords were embedded in OCT compound and transversely cut into cross Sect. 10 μm thick using a cryostat for immunofluorescence staining. The L4-5 spinal slices were incubated in a blocking solution for 30 min at room temperature and then incubated overnight at 4 ℃ with the following primary antibodies: rabbit anti-IBA1 (CST, Boston, USA) for microglia, rabbit anti-Substance P (SP, Immunostar, Wisconsin, USA) and rabbit anti-calcitonin gene-regulated peptide (CGRP, CST, Boston, USA) for the subset of peptidergic nociceptive afferents, and biotin-conjugated isolectin-B4 (IB4, Sigma, St. Louis, USA) for a subset of non-peptidergic nociceptive afferents. Meanwhile, double-labeling procedures were performed on thoracic spinal cord slides for immunofluorescence, using mouse anti-IBA1 (Invitrogen, California, USA), mouse anti-GFAP (CST, Boston, USA), and mouse anti-NeuN (CST, Boston, USA) in combination with rabbit anti-RUNX1 (Abcam, Cambridge, UK). The slices were allowed to equilibrate to room temperature and incubated with the respective secondary antibodies for 1 h at room temperature. For double immunofluorescence staining, all sections were appropriately incubated with a mixture of primary and secondary antibodies. The slices were analyzed using Zeiss LSM 800 confocal microscopy (Zeiss, Oberkochen, Germany) or Pannoramic MIDI automatic digital slide scanner (3D HISTECH, Budapest, Hungary).

Immunocytochemistry

BV-2 cells were seeded into confocal dishes. When the cell confluence reached 70%, they were treated with 1 µg/ml LPS for 6 h. Cells were washed with PBS three times and fixed using the ice-cold 4% paraformaldehyde (Biosharp, Beijing, China) for 20 min. Cells were permeabilized with Triton (0.3%) and blocked with 5% goat serum. Then, cells were incubated overnight at 4℃ with primary antibodies: rabbit anti-RUNX1 (Abcam, Cambridge, UK) and rat anti-F4/80 (CST, Boston, USA). Cells were washed with PBS and then incubated with secondary antibodies: goat anti-rabbit Alexa Fluor 594 (ZSGB-BIO, Beijing, China) and goat anti-rat Alexa Fluor 647 (Invitrogen, California, USA) for 30 min. Imaging was conducted using Zeiss LSM 800 confocal microscopy.

Immunohistochemistry, Immunofluorescence and Immunocytochemistry Quantification

For immunohistochemistry, the scanned sections were observed under 20× magnification. The RUNX1 staining in the dorsal horn of gray matter (GM) and white matter (WM) was detected, and the optical density of RUNX1 relative to that in the sham group was quantitatively analyzed. There were 4 rats in each group, and 4 different sections were selected from each rat.

For immunofluorescence, the identical regions of the L4-5 spinal cord encompassing the spinal dorsal horn in each group were assessed and photographed under consistent exposure conditions to acquire the raw data. The integrated fluorescence intensity of each pixel was normalized to that of the sham group.

For immunocytochemistry, three random fields were selected for the cells in each group, and the integrated density of RUNX1 and F4/80 in each cell was quantified to obtain average fluorescence signal (OD/cells).

All images were analyzed using ImageJ.

ELISA

The expression levels of inflammatory factors TNF-α, IL-1β, IL-6 and IFN-γ were assayed using corresponding mouse ELISA kits (Elabscience, Wuhan, China) according to the manufacturer’s instructions. Absorbance at 405 nm was measured using a microplate reader (Tecan, Männedorf, Switzerland), and the background value was subtracted from the detection data. The concentrations of these cytokines were quantified via the external standard curve.

NO Concentration Measurement

Nitric oxide (NO) radicals are produced by activated microglia. The continuous activation of microglia increases the expression of inducible nitric oxide synthase (iNOS) and leads to the accumulation of excessive NO, which can give rise to neuroinflammation and diseases [35]. The production of NO was measured using the Griess Reagent (Beyotime, Shanghai, China). The kit was left to reach room temperature. After adding 50 µl of the culture supernatant to a 96-well plate, 50 µl of Griess Reagent I and 50 µl of Griess Reagent II were successively added. The absorbance was measured at 540 nm using a microplate reader. The concentration of NO in the supernatant was calculated according to the standard curve.

Statistical Analysis

All data are presented as the means ± SD. For comparisons between two groups, an independent-samples t-test was employed. Statistical comparison of multi-comparison was performed using one-way analysis of variance (ANOVA) followed by post hoc Turkey or Games-Howell test depends on the results of the homogeneity of variance test. Data from the behavior tests were analyzed via two-way repeated measures ANOVA followed by Bonferroni test. A p value < 0.05 was considered to be statistical significance. All statistical analysis were performed using IBM SPSS statistics 27.

Results

SCI Triggers CNP and Elevates Expression of RUNX1

To investigate the effects of RUNX1 in SCI-induced CNP, we first examined how spinal cord contusion in a rat model influences CNP symptoms. Male rats underwent either sham surgery or moderate spinal cord contusion, and pain symptoms, including mechanical allodynia and thermal hyperalgesia, were assessed over time.

Rats in the SCI and sham groups exhibited similar paw withdrawal threshold (PWT) to innocuous mechanical stimuli and paw withdrawal latency (PWL) to thermal beam in the hindpaws before SCI. Immediately after injury, the hindpaws of the rats were paralyzed; however, the rats showed sufficient spontaneous recovery of hindpaw movement until dpi 21 to accomplish the CNP behavior tests. Rats with SCI showed mechanical and thermal hypersensitivity symptoms in the hindpaws from dpi 21 and sustained through the end of the experiment (49 days post-injury), namely a significant decrease in the PWT and PWL compared with their baseline values and those of the sham-treated rats (n = 12, PWT: time×group, F = 262.175, p < 0.001; time, F = 262.157, p < 0.001; group, F = 1108.715, p < 0.001. PWL: time×group, F = 16.476, p < 0.001; time, F = 21.876, p < 0.001; group, F = 18.650, p < 0.001.) (Fig. 1A, B).

Fig. 1.

SCI induces CNP and enhances a sustained expression of RUNX1 at the injury over time. (A, B) The PWT (A) and PWL (B) results of rats in the sham and SCI groups. (C) Quantitative RT-qPCR analysis of RUNX1 in the sham and SCI rats at the lesion site on dpi 7, 21, 35, 49 (n = 12 rats/group). (D, E) Representative Western blotting results and quantitative analysis for the expression level of RUNX1 in the sham group and SCI group at dpi 7, 21, 35, 49. (F-H) Representative image of immunohistochemistry stain (F) and quantification analysis of RUNX1 in the sham and SCI groups in white matter (WM) (G) and gray matter (GM) (H) of the spinal cord at dpi 49 (n = 4 rats/group). Scale bar = 50 μm. Values are presented as the mean ± SD in both groups. Significant difference from the sham group, *p < 0.05, **p < 0.01, ***p < 0.001; from the time point of pre-injury, ^#p < 0.05, ^##p < 0.01, ^###p < 0.001. dpi, days post injury; SCI, spinal cord injury

Furthermore, the dynamics of RUNX1 expression were assessed in the spinal cord by RT-qPCR and Western blotting analysis at 7, 21, 35 and 49 days. The levels of RUNX1 mRNA and protein expression were considerably low in the sham group. However, the expression of RUNX1 mRNA significantly increased at dpi 7 and remained elevated until dpi 49 after SCI (Fig. 1C). Similarly, SCI induced a rapid-onset and long‐lasting expression of RUNX1 protein from dpi 7 to dpi 49, and the increase occurred at dpi 7, peaked at dpi 35, and remained elevated until dpi 49 (Fig. 1D, E).

Immunohistochemistry analysis revealed that RUNX1 expression in the white matter and the dorsal horn of gray matter in the thoracic spinal cord was minimal in the sham group. Conversely, rats with CNP at dpi 49 exhibited strong RUNX1 immunoreactivity in the injured epicenter (Fig. 1F-H). Taken together, the results demonstrate that SCI induces CNP and enhances the expression of RUNX1 at the injured site.

RUNX1 Reduction Alleviates CNP Induced by SCI

Subsequent to the observation that SCI could induce a significant up-regulation of RUNX1, an intraparenchymatous injection of scAAV-shRUNX1 (SCI-shRUNX1 group) or scAAV-scrambled (SCI-nc group) was administered immediately after SCI to investigate the analgesic effect of RUNX1 suppression. Compared with rats in the sham group, rats in the SCI-nc group showed enhanced RUNX1 protein and mRNA expression (Fig. 2A-C) while scAAV-shRUNX1 treatment decreased RUNX1 protein and mRNA levels compared with scAAV-scrambled treatment (Fig. 2A-C).

Fig. 2.

RUNX1 reduction alleviates mechanical allodynia and thermal hyperalgesia after SCI. (A) Western blotting was performed to evaluate the expression levels of RUNX1 in sham, SCI-nc and SCI-shRUNX1 groups at dpi 49. (B) Quantitative analysis of the expression levels of RUNX1 based on Western blotting results. (C) Quantitative analysis of the expression levels of RUNX1 based on RT-qPCR results. (D) Changes in the PWT in response to mechanical stimuli were measured in sham, SCI-nc, SCI-shRUNX1 rats pre-injury and 21, 35, 49 days post-injury. (E) Changes in the PWL in response to radiant heat stimuli were measured at the same time points as in (D). (F) Representative photomicrographs of SP-labeled sections of the lumbar (L4-5) spinal dorsal horn from sham, SCI-nc and SCI-shRUNX1 rats. (G) Representative photomicrographs of CGRP-labeled sections of the lumbar (L4-5) spinal dorsal horn from sham, SCI-nc and SCI-shRUNX1 rats. (H) Representative photomicrographs of IB4-labeled sections of the lumbar (L4-5) spinal dorsal horn from sham, SCI-nc and SCI-shRUNX1 rats. SP, CGRP and IB4 were shown in green fluorescence, while DAPI was shown in blue fluorescence. (I) Quantitative analysis of SP immunofluorescence. (J) Quantitative analysis of CGRP immunofluorescence. (K) Quantitative analysis of IB4 immunofluorescence. Data are presented as mean ± SD. Scale bar = 50 μm, n = 3. Significant difference from the SCI-nc group, *p < 0.05, **p < 0.01, ***p < 0.001; from the time point of pre-injury, ^#p < 0.05, ^##p < 0.01, ^###p < 0.001; from the time point of dpi 21, ^&p < 0.05. PWT, paw withdrawal threshold; PWL, paw withdrawal latency

Mechanical allodynia and thermal hyperalgesia tests on both sides of the hindlimb plantar surface were applied to study whether the expression of RUNX1 influenced behavioral hypersensitivities following SCI. Significant differences were detected among various groups (n = 12, PWT: time×group, F = 51.706, p < 0.001; time, F = 60.516, p < 0.001; group, F = 250.831, p < 0.001. PWL: time×group, F = 4.428, p = 0.001; time, F = 6.903, p < 0.001; group, F = 14.993, p < 0.001.). Intraspinal injection of scAAV-shRUNX1 significantly counteracted the SCI-induced decrease in mechanical PWT and thermal PWL from dpi 21, maintaining the improvement until dpi 49 compared with scAAV-scrambled rats (Fig. 2D, E).

Two distinct populations of primary afferent nociceptors transmit pain-related information, and this transmission actually involves two parallel and potentially independent ascending pathways. One population of C-fibers contains the neuropeptides substance P (SP) and calcitonin gene-related peptide (CGRP), while the non-peptidergic C-fibers are represented by isolectin B4 (IB4) [36]. To evaluate SCI-induced and RUNX1 expression changes in primary nociceptive afferents terminating in the dorsal horn, lumbar L4-5 spinal cord sections were labeled for SP, CGRP and IB4 to identify peptidergic and non-peptidergic C-fibers with immunofluorescence, respectively (Fig. 2F-H). At dpi 49, the SCI-nc rats showed a marked increase in density in the dorsal horn of both C-fiber types as the bands of SP+, CGRP + and IB4 + nociceptive afferents widened compared with sham rats. Importantly, shRUNX1 treatment reduced the density of SP-labeling, CGRP-labeling and, IB4-labeling, partially reversing the injury-induced increases (Fig. 2I-K). In summary, the findings indicate that reducing RUNX1 alleviates CNP following SCI.

RUNX1 Is Localized in Microglia Cells at the Injured Site of the Spinal Cord

To detect the distribution of RUNX1 in spinal cord tissue, double immunostaining was performed with antibodies against RUNX1 and three major cell-specific markers such as IBA1, GFAP and NeuN to determine the cellular localization of RUNX1 expression at the injured and homologous sites of the spinal cord in sham and SCI rats 49 days after injury. As illustrated in Fig. 3, RUNX1 was expressed exclusively in microglia (Fig. 3A), but not in astrocytes (Fig. 3C) or neurons (Fig. 3E). The results were the same as those of the sham group and the SCI-induced increase in RUNX1 expression occurred in IBA1-positive microglia (Fig. 3B), whereas there was no colocalization of RUNX1 in GFAP-positive astrocytes (Fig. 3D) or in NeuN-positive neurons (Fig. 3F). Since RUNX1 is expressed in the nucleus, while the other markers are primarily localized in the cytoplasm, quantitative co-expression analysis was not feasible. Therefore, the Plot Profile function in ImageJ was employed to analyze the correlation between different fluorescent signals by measuring the gray value variations along a defined linear region, enabling spatial relationship assessment based on fluorescence distribution patterns. As shown in Fig. 3G and H, the fluorescence trends of RUNX1 and IBA1 exhibited a strong correlation in both the sham and SCI groups, whereas no such association was observed between RUNX1 and GFAP or NeuN (Fig. 3I-L). These findings indicate an upregulation of RUNX1 expression in rats with CNP following SCI, particularly in microglia located at the injured site.

Fig. 3.

RUNX1 is expressed in microglia. Photomicrographs illustrate the expression and cellular localization of RUNX1 in the injured spinal cords and homologous spinal cord sites in sham and SCI rats 49 days after surgery. (A, B) Representative images depict colocalization of RUNX1 (red) with IBA1 (green, a marker of microglial cells) in sham (A) and SCI (B) rats. (C, D) Representative images show RUNX1 (red) is not co-expressed with GFAP (green, a marker of astrocytes) in sham (C) and SCI (D) rats. (E, F) Representative images show RUNX1 (red) is not co-expressed with NeuN (green, a marker of neurons) in sham (E) and SCI (F) rats. Scale bars = 20 µm. (A’-F’) The images in the last column are the high-magnification versions of those in the preceding column. Scale bars = 10 μm. (G-L) Colocalization analysis of images (A’-F’) is performed using the Plot Profile function in ImageJ, examining linear regions with a fixed width of 10 pixels

RUNX1 Inhibition Attenuates LPS-induced Microglia Activation in BV-2 Cells

After confirming that RUNX1 is exclusively expressed in microglia subsequent to SCI, we aimed to explore its function in microglial activation. To assess the role of RUNX1 in microglial activation, we used lentivirus-mediated knockdown of RUNX1 in BV-2 mouse microglial cell lines, followed by stimulation with lipopolysaccharide (LPS), a bacterial endotoxin known to elicit a strong immune response and induce microglial activation in animal models [37].

The knockdown efficiency of shRUNX1 lentivirus was validated by RT-qPCR, confirming that RUNX1 expression was reduced by at least 50% (Fig. 4A), which met the threshold for subsequent experiments. BV-2 stable cell lines were transduced with either scrambled or shRUNX1 lentivirus and then treated with LPS or PBS for 6 h. To evaluate the impact of RUNX1 on microglial activation, we measured the fluorescence intensity of F4/80, a commonly used microglial activation marker [38], using cellular fluorescence imaging (Fig. 4B). To ensure accurate quantification, cell numbers were counted using ImageJ software, and the integrated fluorescence intensity was normalized to the total cell count.

Fig. 4.

RUNX1 inhibition attenuates LPS-induced microglia activation in BV-2 cells. (A) Quantitative analysis of the expression levels of RUNX1 based on RT-qPCR results. (B) Immunofluorescence staining of DAPI (blue), RUNX1 (red) and F4/80 (green) in the BV-2 cells. (C) Quantitative analysis of RUNX1 immunofluorescence. (D) Quantitative analysis of F4/80 immunofluorescence. Data are presented as mean ± SD. Significant difference from the LPS-LV-scrambled group, *p < 0.05, **p < 0.01, ***p < 0.001, n = 3, Scale bar = 20 μm

Our results showed that LPS stimulation increased RUNX1 expression, which was accompanied by an elevated fluorescence signal of F4/80 (Fig. 4B). In contrast, RUNX1 knockdown led to a marked reduction in F4/80 fluorescence intensity (Fig. 4C, D), indicating that RUNX1 plays a crucial role in promoting microglial activation.

RUNX1 Inhibition Attenuates LPS-induced Inflammation in BV-2 Cells

To further explore the role of RUNX1 in regulating the neuroinflammatory response, BV-2 cells were transfected with shRUNX1 lentivirus and exposed to LPS for 6 h. Western blotting, RT-qPCR, ELISA, and NO secretion analysis were performed to assess whether RUNX1 inhibition could attenuate inflammation in activated microglia.

Following LPS stimulation, the protein levels of pro-inflammatory cytokines (Fig. 5A), including TNF-α (Fig. 5B) IL-6 (Fig. 5C) and IL-1β (Fig. 5D), were significantly elevated. However, RUNX1 knockdown effectively suppressed this LPS-induced upregulation. Consistently, RT-qPCR analysis showed that while the mRNA levels of TNF-α, IL-6, and IL-1β were low in the control group, they increased significantly in the LPS-treated group but were markedly reduced upon RUNX1 suppression (Fig. 5E-G).

Fig. 5.

RUNX1 inhibition attenuates LPS-induced inflammation in BV2 cells. (A) Representative western blotting results of TNF-α, IL-6 and IL-1β in the PBS + LV-scrambled, the PBS + LV-shRUNX1, the LPS + LV-scrambled and the LPS + LV-shRUNX1 groups. (B) Quantitative analysis of TNF-α expression from western blotting results. (C) Quantitative analysis of IL-6 expression from western blotting results. (D) Quantitative analysis of IL-1β expression from western blotting results. (E) Quantitative RT-qPCR analysis of TNF-α. (F) Quantitative RT-qPCR analysis of IL-6. (G) Quantitative RT-qPCR analysis of IL-1β. (H) Mouse ELISA kits were used to quantitate the concentrations of TNF-α in the cellular supernatants. (I) Mouse ELISA kits were used to quantitate the concentrations of IL-6 in the cellular supernatants. (J) Mouse ELISA kits were used to quantitate the concentrations of IL-1β in the cellular supernatants. (K) Mouse ELISA kits were used to quantitate the concentrations of IFN-γ in the cellular supernatants. (L) The production of NO was measured by determining the total nitrate concentration in the supernatants. Data are presented as mean ± SD. Significant difference from the LPS + LV-scrambled group, *p < 0.05, **p < 0.01, ***p < 0.001; from the PBS + LV-scrambled group, ^#p < 0.05, ^##p < 0.01, ^###p < 0.001

ELISA results (Fig. 5H-K) further confirmed these findings, demonstrating that LPS stimulation led to a significant increase in TNF-α, IL-6, IL-1β, and IFN-γ secretion, while RUNX1 inhibition effectively mitigated these increases, aligning with the Western blotting and RT-qPCR data. Additionally, RUNX1 suppression reduced NO production in BV-2 microglial cells, regardless of LPS stimulation (Fig. 5L). Overall, these results suggest that RUNX1 plays a crucial role in promoting the neuroinflammatory response in microglia.

RUNX1 Reduction Leads To a Decrease in Microglia Activation and Neuroinflammation after SCI

Previous research has indicated that intense microglial activation and neuroinflammation occurred in the superficial dorsal horn of the lumbar spinal cord following SCI [14, 32, 39, 40]. Notably, in cases of moderate SCI, the degree of microglial activation and neuroinflammation positively correlated with the severity of allodynia-greater activation was associated with more pronounced pain-like behaviors [14].

After confirming the role of RUNX1 in microglial activation and inflammation in vitro, we aimed to investigate its involvement in these processes in vivo and its contribution to CNP induced by SCI. To determine whether RUNX1 is involved in microglial activation, we assessed IBA1 expression in the L4-5 spinal cord via immunofluorescence following RUNX1 suppression. Consistent with previous findings [41], SCI induced a significant increase in IBA1 expression. However, RUNX1 suppression markedly reduced SCI-induced microglial overactivation compared to SCI-nc rats (Fig. 6A, B), suggesting that spinal RUNX1 modulates microglial activation in the SCI model. Moreover, RUNX1 suppression prevented SCI-induced M1 microglial polarization and promoted M2 polarization, as evidenced by decreased mRNA levels of M1 markers (cd86, inos) and increased M2 markers (cd206, arg1) in RT-qPCR analysis (Fig. 6C-F).

Fig. 6.

RUNX1 reduction leads to a decrease in microglia activation and neuroinflammation after SCI. (A) Representative images document IBA1 expression in the L4-L5 dorsal horns of rats in each of the experimental groups. (B) Quantitative analysis of IBA1 immunofluorescence. Quantitative RT-qPCR analysis of cd86 (C), inos (D), cd206 (E), arg1 (F), TNF-α (G), IL-6 (H) and IL-1β (I) in the sham, SCI-nc and SCI-shRUNX1 rats at dpi 49. Data are presented as mean ± SD. Significant difference from the SCI-nc group, *p < 0.05, **p < 0.01, **p < 0.001, n = 3

Furthermore, RT-qPCR analysis revealed that RUNX1 suppression significantly downregulated the expression of key pro-inflammatory cytokines, including TNF-α, IL-6, and IL-1β (Fig. 6G-I). Together, these findings indicate that RUNX1 contributes to SCI-induced CNP by promoting microglial activation and neuroinflammation, while its suppression mitigates these pathological processes, potentially alleviating pain.

Discussion

The role of RUNX1 in SCI-induced chronic neuropathic pain (CNP) remains largely unexplored. In this study, we demonstrated that RUNX1 is activated in the spinal cord during the subacute and chronic phases of SCI, playing a crucial role in the development and persistence of CNP. Our findings reveal that RUNX1 is primarily expressed in microglia, and its suppression in vitro alleviates microglial activation and the inflammatory response. These observations were further validated in vivo, where RUNX1 inhibition mitigated microglial activation and polarization as well as neuroinflammation following SCI. On the whole, our results highlight RUNX1 as a key regulator of SCI-induced CNP by promoting microglial activation and triggering neuroinflammation.

RUNX1 belongs to the Runt domain transcription factor family, which includes RUNX1/PEBP2αB/AML1, RUNX2/PEBP2/AML3, and RUNX3/PEBP2βC/AML2 [42]. These transcription factors interact with the cofactor CBFβ to regulate various biological processes, including neurogenesis, hematopoiesis [43], and cell proliferation and differentiation [44]. Published data reveal that RUNX1 expression is significantly upregulated post-neural injury [45, 46]. Our results showed that RUNX1 expression was significantly upregulated in rats after SCI in both white matter and gray matter, suggesting its broad involvement in CNS pathophysiology. In gray matter, RUNX1 may primarily regulate neuronal survival, synaptic plasticity, or inflammatory responses, as supported by prior studies linking RUNX1 to neuroprotection and microglial activation [47]. While in white matter, RUNX1 could play a distinct role in oligodendrocyte function, myelination, or axonal integrity [46, 48]. Further studies are needed to elucidate the region-specific mechanisms of RUNX1, particularly whether its upregulation reflects a unified response or context-dependent functions across different CNS compartments.

RUNX1 has been demonstrated to play a pivotal role in regulating the specialization of nociceptive neurons and their axonal projections to the spinal cord [49]. It is hypothesized that RUNX1 inhibits the expression of TrkB in nociceptors while facilitating the expression of TrkA and TrkC [50, 51]. However, other studies proposed that the persistent expression of RUNX1 served as a critical event for prospective nociceptors to undergo non-peptidergic differentiation. Specifically, it enables the transition from the TrkA + to the Ret + phenotype by suppressing the expression of TrkA and promoting the expression of Ret [25], including the expression of several sensory channels and receptors which are indispensable for thermal pain, inflammatory pain, and neuropathic pain [25, 44, 52]. Mice deficient in RUNX1 exhibit a significantly delayed response time to noxious thermal stimuli [25]. Additionally, RUNX1 is an indispensable prerequisite for the emergence and manifestation of neuropathic pain responses in the mouse SNI and CCI model [25, 44]. However, the mechanisms by which RUNX1 contributes to SCI-induced CNP remain poorly understood. Our findings revealed that SCI upregulated RUNX1 expression, concomitant with reduced PWT and PWL. Notably, partial knockdown of RUNX1 attenuated mechanical allodynia and thermal hyperalgesia, suggesting its potential role in pain hypersensitivity. Importantly, we discovered a significant relationship between RUNX1 and the expression levels of primary afferent neuron markers. Specifically, a reduction in RUNX1 led to a significant decrease in the fluorescence intensity of SP, CGRP, and IB4, suggesting that RUNX1 promotes the maladaptive sprouting of nociceptive fibers and enhances pain sensitivity in the spinal cord dorsal horn.

SCI induces persistent glial activation, which is fundamental mechanisms underlying SCI-induced CNP [13, 53–55]. Activated glial cells and recruited macrophages significantly contribute to neuroinflammation by triggering a cascade of cellular responses [56, 57]. Among glial cells, microglia and astrocytes play key regulatory roles in the nociceptive circuit and are activated in response to various nerve injuries in neuropathic pain conditions. In SCI models, including hemisection [39], contusion [58], and spinal neuropathic avulsion pain (SNAP) [59], microglia in the lumbar spinal dorsal horn undergo a profound transition from a quiescent to an activated state. This activation is strongly linked to the onset and persistence of SCI-induced neuropathic pain, particularly below-level CNP [39, 54]. In neuropathic pain development, spinal microglia activate into two distinct phenotypes: the pro-inflammatory M1 type which releases hyperalgesia-inducing cytokines, and the anti-inflammatory M2 type which produces pain-alleviating opioid peptides. The M1 response typically predominates over M2, resulting in neuronal hyperexcitability and central sensitization. Promising therapeutic approaches include both pharmacological inhibition of M1 polarization and active conversion from M1 to M2 phenotype, which have demonstrated significant pain-attenuating effects according to multiple studies [60–62]. Despite these advances, the precise molecular mechanisms underlying microglial activation and neuroinflammation in SCI-induced CNP remain incompletely understood.

RUNX1 is a primary determinant of myeloid cell proliferation and differentiation and has been implicated in both prenatal and postnatal microglial activation as well as microglial maturation and differentiation [44, 63]. In our study, we observed a significant upregulation of both RUNX1 expression and microglial activation following SCI and LPS stimulation. Moreover, RUNX1 suppression resulted in a marked reduction in microglial fluorescence intensity both in vitro and in vivo. The animal experiments further established that RUNX1 critically regulated microglial polarization dynamics. Specifically, RUNX1 inhibition prevented the SCI-induced M1 polarization while simultaneously promoting the neuroprotective M2 phenotype. These findings collectively establish RUNX1 as a pivotal regulator that orchestrates microglial proliferation/activation and phenotypic polarization following SCI, crucially modulating microglial functions to influence post-injury cellular responses, neurogenesis, and ultimately determine the outcomes of SCI-induced CNP.

RUNX1 has been shown to regulate inflammation through multiple mechanisms, predominantly promoting inflammatory responses. Previous studies have demonstrated that RUNX1 interacts with the NF-κB subunit p50, thereby amplifying the TLR4-mediated inflammatory response [64]. RUNX1 silencing reduces LPS-induced IL-1β and IL-6 expression, whereas its overexpression enhances the production of these cytokines in macrophages [64]. Consistent with these findings, we observed that increased RUNX1 expression is associated with elevated levels of pro-inflammatory cytokines (such as TNF-α, IL-1β, and IL-6) and nitric oxide (NO) in both cultured microglia and SCI model. These results not only confirm RUNX1’s pivotal role in SCI-induced neuroinflammation but also suggest that targeted modulation of RUNX1 activity may represent a promising therapeutic strategy for mitigating inflammation-driven neuropathic pain following SCI.

Mechanistically, RUNX1 may contribute to the upregulation of these inflammatory mediators through several pathways. First, RUNX1 can directly bind to the promoter regions of target genes and activate transcription of inflammatory cytokines. It has been reported that RUNX1 interacts with other transcriptional regulators such as NF-κB and AP-1, enhancing their transcriptional activity and synergistically promoting the expression of pro-inflammatory genes [65]. This RUNX1–NF-κB crosstalk may be particularly relevant in the context of SCI, where NF-κB is rapidly activated and plays a central role in initiating inflammation [66]. Second, RUNX1 may regulate inducible nitric oxide synthase (iNOS), the enzyme responsible for NO production in activated microglia. Previous studies have shown that RUNX1 can bind to the iNOS promoter and enhance its transcription in response to inflammatory stimuli, thereby contributing to elevated NO levels in the injured spinal cord [46]. Furthermore, RUNX1 may facilitate a shift in microglial polarization toward a pro-inflammatory (M1-like) phenotype, characterized by increased production of cytokines and NO. This is supported by our in vivo data in Fig. 6, where RUNX1 suppression prevented SCI-induced Ml microglial polarization and promoted M2 polarization.

Several mechanisms may underlie the increased expression of RUNX1 following SCI. In our study using BV2 microglial cells, RUNX1 was found to be closely associated with microglial activation and inflammation induced by LPS. LPS is a well-established activator of Toll-like receptor 4 (TLR4), which in turn initiates the NF-κB signaling cascade and promotes the transcription of pro-inflammatory genes. Given that TLR4 expression and activity are markedly elevated following SCI and are central to the initiation of neuroinflammation [67], it is plausible that RUNX1 upregulation is at least partially mediated by TLR4 signaling. In addition to inflammatory signaling, biomechanical changes in the injured spinal cord may also regulate RUNX1 expression. Following SCI, the spinal tissue undergoes substantial remodeling of the extracellular matrix, leading to increased stiffness within fibrotic scar regions and a loss of normal viscoelastic properties [68]. Emerging evidence indicates that mechanical stimuli can modulate RUNX1 expression. For instance, static mechanical loading induces RUNX1 upregulation in bovine cartilage [69], while fluid shear stress activates RUNX1 in hematopoietic progenitors [70], supporting a role for mechanotransduction in RUNX1 regulation. Furthermore, hypoxia is a hallmark of the post-injury spinal cord microenvironment due to disrupted blood flow and metabolic stress. Hypoxia-inducible factors (HIFs), particularly HIF-1α, have been implicated in promoting RUNX1 expression under ischemic or hypoxic conditions [71, 72]. This suggests that hypoxia-driven transcriptional responses may represent another contributing mechanism to RUNX1 elevation after SCI.

Limitations

Several limitations should be acknowledged. First, our analysis focused on the subacute to chronic phases of injury (7–49 days post-SCI), as this window captures peak RUNX1 expression and stable neuropathic pain responses. However, earlier time points (1–3 days post-SCI) were not included due to logistical and ethical constraints, potentially omitting critical early dynamics in RUNX1 regulation and microglial activation. Second, the study was conducted exclusively in male rats, which limits the generalizability of our findings. Given emerging evidence of sex differences in pain mechanisms—including the sexually dimorphic roles of microglia and immune mediators—future investigations should incorporate female subjects to determine whether RUNX1-dependent pathways exhibit similar regulatory patterns across sexes.

Conclusion

In conclusion, our findings demonstrate that RUNX1 is activated in microglia following SCI and plays a pivotal role in neuroinflammation and neuropathic pain. This suggests that inhibiting RUNX1 could serve as a promising therapeutic strategy for mitigating SCI-induced CNP.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

X.W., M.Z. discussed and conceptualized the study. M.H., Y.Y. designed and performed most of the experiments. M.H., Y.Y., Y.T., and Y.L. analyzed the data. M.H., wrote the original manuscript. X.W., M.Z edited and revised the manuscript. All authors discussed the results and commented on the manuscript.

Data Availability

No datasets were generated or analysed during the current study.

Declarations

Competing Interests

The authors declare no competing interests.