ALKBH3‐Mediated M1A Demethylation of METTL3 Endows Pathological Fibrosis:Interplay Between M1A and M6A RNA Methylation

Abstract

Epigenetic modifications serve as crucial molecular switches for pathological fibrosis; howbeit the role of m¹A in this condition remains enigmatic. Herein, it is found that ALKBH3 exerts a pro‐fibrotic effect in pathological skin fibrosis by reshaping N6‐methyladenosine (m⁶A) RNA modification pattern. First, ALKBH3 exhibited specific upregulation within hypertrophic scars (HTS), accompanied by N1‐methyladenosine (m¹A) hypomethylation. Moreover, multiomics analyses identified METTL3, a critical writer enzyme involved in m⁶A modification, as a downstream candidate target of ALKBH3. Therapeutically, ablation of ALKBH3 inhibited the progression of HTS both in vitro and in vivo, while exogenous replenishment of METTL3 counteracted this antifibrotic effect. Mechanistically, ALKBH3 recognizes the m¹A methylation sites and prevents YTHDF2‐dependent mRNA decay of METTL3 transcript. Subsequently, METTL3 stabilizes collagen type I alpha 1 chain (COL1A1) and fibronectin1 (FN1) mRNAs, two major components of extracellular matrix, and therefore eliciting the pathological transformation of HTS. This observation bridges the understanding of the link between m¹A and m⁶A methylation, the two fundamental RNA modifications, underscoring the participation of “RNA methylation crosstalk” in pathological events.

Introduction

A series of cutaneous fibroproliferative diseases, including hypertrophic scar (HTS), keloid, and scleroderma, are clinically characterized by skin fibrosis.^[1,2] Skin fibrosis is a process characterized by excessive proliferation of fibroblasts and abnormal deposition of extracellular matrix (ECM) and is often accompanied by intolerable itching, contracture deformities, and functional impairments.^[3] Considering its effects on physical, mental, and social health, skin fibrosis is increasingly becoming recognized as one of today's major health-care challenges.^[4] However, therapeutic results for skin fibrosis frequently fall short of expectations.^[5] The Transforming Growth Factor-β/Smad (TGF-β/Smad) signaling pathway, the canonical pathway mediating skin fibrosis,^[5,6] participates in diverse biological processes across multiple organs and systems.^[7] The broad involvement of this pathway makes the identification of specific therapeutic targets challenging. Hence, exploring novel pathogenic mechanisms and identifying therapeutic targets for skin fibrosis are highly important.

N1-methyladenosine (m¹A) is an important posttranscriptional RNA modification identified in transfer RNAs (tRNAs), ribosomal RNAs (rRNAs), and messenger RNAs (mRNAs).^[8,9] In mRNAs, m¹A has been detected in every segment and functions as a unique type of base methylation to block Watson–Crick base pairing and alter mRNA structural stability^[10,11] The dynamics of m¹A methylation are mediated by methyltransferases (“writers”: TRMT6, TRMT61A, TRMT61B, and TRMT10C), demethylases (“erasers”: ALKBH1, ALKBH3, and FTO) and RNA-binding proteins (“readers”: YTHDFs).^[12] m¹A modification has recently been recognized to play a crucial role in RNA regulation, thus participating in various biological processes, such as gene expression, RNA stability regulation, posttranscriptional regulation, and disease occurrence.^[13] Notably, multiple RNA modifications are involved in the pathogenesis of skin fibrosis. For instance, N6-methyladenosine (m⁶A) hypermethylation may contribute to keloid pathogenesis by regulating the Wnt/β-catenin pathway.^[14] However, the regulatory roles of m¹A, another important form of RNA modification, in skin fibrosis remain to be fully addressed.

Thus, we aimed to identify the molecular mechanisms and clinical potential of m¹A modification in skin fibrosis. Our study reveals for the first time that the decreased global m¹A level, driven by elevated ALKBH3 expression, are a key feature of skin fibrosis. Additionally, by using RNA sequencing (RNA-seq) and methylated RNA immunoprecipitation–sequencing (MeRIP-seq), we found that ALKBH3 upregulates methyltransferase-like 3 (METTL3) expression via YTHDF2. The increased METTL3 stabilized COL1A1 and FN1 mRNAs through YTHDF1-mediated m⁶A modification, thereby advancing pathological skin fibrosis. Therapeutically, silencing ALKBH3 exhibited promising effects on skin fibrosis both in vitro and in vivo, which were diminished when METTL3 expression was restored, suggesting a novel strategy for treating skin fibrosis. Importantly, this study establishes a link between m¹A and m⁶A methylation, the two fundamental RNA modifications, underscoring the participation of a “RNA methylation crosstalk” in pathological events.

Results

Reduced RNA m¹A Modification and Elevated ALKBH3 Expression are Observed in Hypertrophic Scars

Hypertrophic scars (HTS), which are typical manifestations of skin fibrosis, were chosen as a model to elucidate the role of m¹A in pathological skin fibrosis. We first compared the global m¹A modification level between HTS and normal skin (NS). Notably, HTS samples exhibited significantly lower m¹A levels, as demonstrated by the m¹A dot blot assay (Figure 1A). Furthermore, we analyzed the expression of m¹A writers and erasers, and we observed significant upregulation of ALKBH3 mRNA expression in HTSs, while the expression of other regulators remained unchanged (Figure 1B). This finding was further supported by the results of a previous single-cell analysis (GSE156326),^[15] which also revealed that the increase in ALKBH3 expression was predominantly localized to fibroblasts (Figure 1C–E). Although ALKBH3 expression was also detected in endothelial cells (ECs) and smooth muscle cells (SMCs) (Figure 1D), no significant difference in expression was observed between HTS and NS in these cell types (Figure S1A, Supporting Information). Moreover, ALKBH3 had no significant impact on the phenotypes of ECs (Figure S1B–G, Supporting Information) or SMCs (Figure S1H–L, Supporting Information). Therefore, the increased expression of ALKBH3 in fibroblasts remains a critical driving factor in pathological skin fibrosis. Immunofluorescence (IF) staining and western blot (WB) analyses revealed elevated protein expression of the demethylase ALKBH3 in HTSs (Figure 1F,G), which corresponded with a reduction in the m¹A levels in the HTS. The trend was further confirmed in fibroblasts isolated from HTS tissues (Figure 1H,I). Moreover, high ALKBH3 expression was significantly correlated (R = 0.9236, p value < 0.0001) with more advanced HTS stages, as assessed by the modified Vancouver Scar Scale (mVSS) (Table S1, Supporting Information), underscoring the functional importance of ALKBH3 in HTS (Figure 1J).

[IMAGE OMITTED. SEE PDF]

Inhibition of ALKBH3 Attenuates HTS Derived Fibroblasts Activation In Vitro

To explore the biological function of ALKBH3 in fibroblasts, we utilized two individual small interfering RNAs (siRNAs) to silence ALKBH3 in HTS derived fibroblasts (HDF) (Figure 2A–C; Figure S2, Supporting Information). Concordantly, ALKBH3 knockdown led to a marked increase in m¹A levels in HDFs (Figure 2D). Furthermore, a reduction in collagen deposition capacity was observed in ALKBH3-silenced HDFs, as demonstrated by quantitative reverse transcription–PCR (qRT‒PCR) (Figure 2A) and WB (Figure 2C; Figure S2,Supporting Information) analyses. ALKBH3 knockdown also resulted in decreased cell proliferation, as evidenced by both Cell Counting Kit-8 (CCK-8) and 5-ethynyl-2′-deoxyuridine (EdU) incorporation assays (Figure 2E,F). Moreover, flow cytometric analysis revealed an increase in apoptosis and alterations in the cell cycle distribution after ALKBH3 inhibition (Figure 2G,H). Notably, the knockdown of ALKBH3 resulted in negligible changes in the migration of HDFs (Figure S3, Supporting Information). Additionally, silencing ALKBH3 in normal dermal fibroblasts (NDFs) did not significantly affect collagen deposition, proliferation, or migration (Figure S4, Supporting Information). Taken together, these data suggest that ALKBH3 promotes pathological skin fibrosis primarily by driving the abnormally activated fibroblasts.

[IMAGE OMITTED. SEE PDF]

ALKBH3 Promotes the Progression of Skin Fibrosis In Vivo

Based on the profibrotic role of ALKBH3 observed in vitro, we next established mice with global Alkbh3 knockout (Alkbh3^−/− mice) (Figure 3A). The gross skin appearance and histological characteristics of these Alkbh3^−/− mice were similar to those of their wild-type (WT) littermates. The decrease in Alkbh3 expression and increase in global m¹A modification throughout the skin layers of Alkbh3^−/− mice were verified through genotyping, WB analysis, IF staining, and a dot blot assay (Figure 3B,C; Figure S5, Supporting Information). We utilized mechanical stretch-induced HTS and bleomycin-induced skin fibrosis models to mimic pathological skin fibrosis in vivo (Figure 3D,E). In the mechanical stretch-induced HTS model, the gross scar area was significantly attenuated in the Alkbh3^−/− group (Figure 3F; Figure S6A, Supporting Information). In addition, collagen deposition in the dermis was also reduced in the Alkbh3^−/− group, as revealed by qRT‒PCR and WB analyses (Figure 3G; Figure S6B,C, Supporting Information). Similarly, reductions in dermal thickness and collagen deposition were observed in Alkbh3^−/− mice of bleomycin-induced skin fibrosis model (Figure 3H,I; Figure S6D–F, Supporting Information).

[IMAGE OMITTED. SEE PDF]

To further elucidate the clinical implications of ALKBH3 inhibition in pathological skin fibrosis, we administered anti-Alkbh3 antisense oligonucleotides (ASOs) to WT mice after successful modeling (Figure 3J,K). ASOs are chemically modifiable single-stranded nucleic acid sequences that function by binding to RNA sequences through Watson–Crick base pairing.^[16] ASOs designed to selectively downregulate, upregulate, or modify the expression of key genes in patients with terminal illnesses have become well-established clinical therapeutic agents. Several ASO-based drugs have received approval from the United States Food and Drug Administration (FDA), with numerous others currently undergoing clinical trials.^[17] Treatment with anti-Alkbh3 ASOs led to a significant reduction in the gross scar area, concomitant with reduced dermal thickness and collagen deposition in mice with pathological skin fibrosis (Figure 3L–O; Figure S7, Supporting Information). Taken together, these findings underscore the role of ALKBH3 in exacerbating pathological skin fibrosis progression in vivo, thus emphasizing its potential as a crucial therapeutic target.

Multiomics Screening Identified METTL3 as the Downstream Candidate of ALKBH3

We then investigated the mechanism underlying the inhibitory effect of ALKBH3 silencing on skin fibrosis. After silencing ALKBH3 in HDFs, we performed multiomics analyses, including m¹A-MeRIP-seq and RNA-seq. As a result, silencing ALKBH3 led to notable alterations in the transcriptome, with 1707 genes downregulated and 589 upregulated (Figure 4A). Since most of the differentially expressed genes (74.3% of the 2296 genes) were downregulated in ALKBH3-deficient cells, our results are consistent with previous observations suggesting that ALKBH3-induced m¹A demethylation increases mRNA stability and consequently promotes gene expression.^[18] Moreover, the differentially expressed genes were associated with processes related to skin development and homeostasis, including process terms such as cell cycle, biosynthetic process, and signal transduction (Figure 4B). Additionally, the average numbers of m¹A peaks detected in the m¹A-MeRIP-seq libraries generated from control and ALKBH3-deficient cells were 14 301 and 13 394, respectively (Figure 4C,D). These m¹A peaks exhibited significant enrichment in 5′ UTRs, particularly near start codons (Figure S8A, Supporting Information). Notably, the differentially expressed m¹A-modified genes were associated with multiple fibrosis-related pathways, including the MAPK, Hippo, and Wnt signaling pathways (Figure S8B,C, Supporting Information). These findings underscore the regulatory role of ALKBH3 in skin fibrosis pathogenesis.

[IMAGE OMITTED. SEE PDF]

By integrated analysis of these omics data, we identified 14 candidate genes potentially linked to ALKBH3-regulated m¹A modification (Figure 4E). Subsequent Ingenuity Pathway Analysis (IPA) of these 14 genes revealed that METTL3 was located in the core of the candidate gene matrix (Figure 4F). We further confirmed that ALKBH3 silencing reduced METTL3 expression at both the mRNA and protein levels in fibroblasts, with a corresponding increase in m¹A levels (Figure 4G–J; Figure S9A,B, Supporting Information). Similar findings were observed in Alkbh3^−/− mice (Figure S9C,D, Supporting Information). Additionally, a significant positive correlation between ALKBH3 and METTL3 expression was found in skin samples, both in the GEPIA2 public database (R = 0.26, p value = 6.1e-10) (Figure 4K) and in a clinical cohort via qRT‒PCR (R = 0.8714, p < 0.0001) (Figure 4L). This correlation was further validated by HTS tissue microarray staining, which revealed either a high/high or low/low ALKBH3/METTL3 expression pattern (R = 0.7428, p < 0.0001) (Figure 4M; Figure S9E, Supporting Information).

Notably, METTL3-mediated m⁶A methylation has been identified as a critical factor in fibrotic processes such as cardiovascular and pulmonary fibrosis.^[19,20] By analyzing clinical samples, we found that METTL3 and m⁶A methylation play a crucial role in the occurrence of pathological skin fibrosis (Figure S10, Supporting Information). We utilized two distinct siRNAs to silence METTL3 expression in HDFs, which resulted in a significant reduction in the m⁶A level (Figure S11, Supporting Information). Furthermore, the inhibition of METTL3 led to a decrease in the collagen deposition function and the growth capacity of fibroblasts, indicating a profibrotic role of METTL3 in pathological skin fibrosis (Figures S11 and S12, Supporting Information).

These findings corroborate the profibrotic actions of ALKBH3 in pathological skin fibrosis. Taken together, this suggests that METTL3, an m¹A-modified gene, is potentially regulated by ALKBH3.

The ALKBH3-METTL3 Axis is Critical in Skin Fibrosis Pathogenesis

To verify the relationship between ALKBH3 and METTL3, we altered METTL3 expression after ALKBH3 inhibition in HDFs by exogenously overexpressing METTL3. As expected, the expression of METTL3 was significantly increased at both the mRNA (Figure 5A) and protein levels (Figure 5B; Figure S13A, Supporting Information). Furthermore, the reintroduction of METTL3 effectively restored the endogenous METTL3 expression that had been reduced by ALKBH3 knockdown (Figure 5A,B; Figure S13A, Supporting Information). Notably, exogenous METTL3 overexpression also mitigated the inhibitory effects of ALKBH3 depletion on collagen deposition (Figure 5A,B; Figure S13A, Supporting Information) and significantly rescued the proliferative capacity of fibroblasts with ALKBH3 knockdown (Figure 5C–F; Figure S13B–D, Supporting Information). To validate these findings in vivo, we utilized Alkbh3^−/− mice and overexpressed METTL3 with a lentiviral vector (Lv-METTL3) following a previously established lentiviral transfection strategy (Figure 5G).^[21] In the mechanical stretch-induced HTS model, METTL3 overexpression (Figure 5H–J) reversed the effects of ALKBH3 knockout, including reduced scar area and collagen deposition (Figure 5H–L). A similar reversal of dermal thickness and collagen deposition was observed in the bleomycin-induced fibrosis model (Figure 5M–Q).

[IMAGE OMITTED. SEE PDF]

Additionally, to further confirm the dependency of ALKBH3's pro-fibrotic effects on METTL3, we conducted experiments in which ALKBH3 was overexpressed while METTL3 was simultaneously knocked down in HDFs. As expected, the simultaneous ALKBH3-overexpression and METTL3-silenced group showed comparable METTL3 expression, compared to NC group (Figure 6A–C; lanes 1 and 4). Consistently, the profibrotic effects of ALKBH3-overexpression, including collagen deposition (Figure 6A–C) and proliferative capacity (Figure 6D–G), were diminished upon METTL3 deprivation, supporting the hypothesis that ALKBH3 exerts its profibrotic effects through METTL3-dependent pathways.

[IMAGE OMITTED. SEE PDF]

ALKBH3 Regulates the Expression of METTL3 in a YTHDF2-Dependent Manner

We then explored the epigenetic mechanisms underlying the ALKBH3-mediated m¹A modificationl of METTL3. Since previous studies have demonstrated that YTHDF proteins are responsible for the recognition of m¹A, we first tested the binding status of YTHDF1, YTHDF2, and YTHDF3 to METTL3 mRNA. RNA immunoprecipitation (RIP) demonstrated that YTHDF2 specifically recognized METTL3 mRNA. However, YTHDF1 and YTHDF3 exhibited only limited interactions with METTL3 mRNA (Figure 7A). Moreover, YTHDF2 expression exhibited a strong positive correlation (R = 0.12, p value = 0.0041) with METTL3 expression in skin samples represented in the GEPIA2 public database (Figure S14A, Supporting Information), completely consistent with the hypothesis that YTHDF2 is necessary for the recognition of METTL3.

[IMAGE OMITTED. SEE PDF]

Since YTHDF2 is responsible for RNA degradation,^[22] we examined its role in METTL3 mRNA stability. Intriguingly, knockdown of ALKBH3 decreased the RNA stability of METTL3; however, this effect was reversed by subsequent depletion of YTHDF2 (Figure 7B–E: Figure S14B, Supporting Information). Furthermore, we have performed overexpression of ALKBH3 combined with YTHDF2 knockdown and compared the results to the group with ALKBH3 overexpression alone (Figure 7F; Figure S14C,D, Supporting Information). Our findings indicate that YTHDF2, as an m¹A reader protein, promotes METTL3 degradation and therefore inhibits its expression. On the contrary, upon ALKBH3 overexpression, which removes the m¹A methylation modification, YTHDF2 knockdown did not impact the mRNA stability of METTL3 (Figure 7G). These loss-of-function and gain-of-functions results aggregate the fact that YTHDF2 promotes METTL3 degradation by recognition of m¹A methylation of METTL3.

We then determined the specific m¹A modification site of METTL3 mRNA. Previous literature has highlighted that m¹A methylation sites are often enriched in the 5′ UTR and near the start codon region of mRNAs.^[11] In the 5′ UTR of METTL3, we identified six potential m¹A sites based on m¹A-meRIP-seq analysis [c.35A (AGAAT), c.52A (GGAGC), c.65A (GGAGG), c.94A (AGAGG), c.110A (GGAGA), and c.118A (GGAAT), identified in the m¹A-seq peak, and mutated each A to a T. These wild-type and mutant 5′ UTR sequences were then cloned into the pmirGLO vector. The luciferase reporter assay revealed that the c.A52T mutation resulted in a significant decrease in luciferase activity, while mutations at other sites did not remarkably alter the luciferase signal (Figure 7H). Similarly, following previous protocol,^[23] we observed m¹A-probe2 (carrying c.52-m¹A methylation site) interacts with YTHDF2 in the RNA pulldown assay, which is identical with the m¹A-probe 7 (positive control group). However, other m¹A probes and control probe showed limited signals (Figure 7I). Taken together, these results showed c.52-m¹A methylation site serves as the recognition site of YTHDF2, which is responsible for the RNA degradation of METTL3.

Notably, METTL3 is a well-known m⁶A writer and a crucial component of the m⁶A methyltransferase complex.^[24] We found that the reduced expression of METTL3 in ALKBH3-deficient cells and Alkbh3^−/− mice resulted in decreased overall m⁶A levels (Figure 7J,K). Collectively, these data shedding light on a novel process of “RNA methylation crosstalk” between m¹A and m⁶A methylation.

METTL3, Relying on YTHDF1, Controls the m⁶A Modification and Stabilization of COL1A1 and FN1 mRNAs

To further investigate the downstream targets of METTL3, we performed m⁶A-MeRIP-seq analysis on HDFs. The MeRIP-seq identified ≈9792 m⁶A peaks per sample. These m⁶A peaks were predominantly enriched in the 3′UTRs, particularly near the stop codon. Notably, significant m⁶A modification peaks were observed in the COL1A1 and FN1 genes, suggesting that these genes might be regulated by METTL3 (Figure 8A,B). m⁶A RIP experiments further confirmed that the m⁶A modification levels of COL1A1 and FN1 were markedly reduced upon METTL3 knockdown (Figure 8C). Additionally, m¹A RIP analysis of ALKBH3-silenced HDFs revealed no significant changes in m¹A modification levels, indicating that COL1A1 and FN1 expression is regulated through m⁶A rather than m¹A (Figure S15, Supporting Information). These findings provide further support for the hypothesis that COL1A1 and FN1 are targets of METTL3.

[IMAGE OMITTED. SEE PDF]

We then explored the detailed mechanism underlying the m⁶A regulation of COL1A1 and FN1 by METTL3. Since m⁶A modifications have been revealed to play vital roles in the maintenance of RNA stability, we first tested whether METTL3 regulates the RNA stability of COL1A1 and FN1. As a result, we found that the RNA stability was dramatically decreased in METTL3-deficient cells (Figure 8D), consistent with previous observations that METTL3 enhances the expression of COL1A1 and FN1. Importantly, YTHDF family members are also the major components responsible for recognizing m⁶A-modified transcripts,^[25] we attempted to identify the potential reader protein. First, RIP analysis revealed that both COL1A1 and FN1 showed a strong interaction with YTHDF1; while very limited signals were captured in the anti-YTHDF2/YTHDF3 groups (Figure 8E). To confirm the role of YTHDF1 in regulating COL1A1 and FN1, we silenced YTHDF1 with two individual siRNAs and measured the expression levels of COL1A1 and FN1. Notably, YTHDF1 silencing resulted in a significant reduction in both mRNA expression levels and protein abundance (Figure 8F; Figure S16, Supporting Information), accompanied by decreased RNA stability (Figure 8G), while protein stability remained unaltered (Figure S17, Supporting Information). These findings indicated that YTHDF1 acts as a nucleus m⁶A reader, stabilizing m⁶A-modified mRNA, thereby highlighting the crucial role of m⁶A in collagen regulation.

However, m⁶A -MeRIP-seq analysis revealed that another critical collagen gene, COL3A1, did not exhibit significant m⁶A modification peaks, suggesting that COL3A1 expression is not influenced by METTL3-mediated m⁶A methylation (Figure S18, Supporting Information). It is well known that the imbalance in the COL1A1/COL3A1 ratio is a key factor in the formation of pathological scars, typically characterized by excessive deposition of COL1A1 and a relative reduction in COL3A1. Therefore, the high expression of METTL3 in pathological scars, through YTHDF1-dependent m⁶A methylation, enhances the stability of COL1A1 and FN1 mRNAs, exacerbating the imbalance in the COL1A1/COL3A1 ratio, ultimately affecting the structure and function of pathological skin fibrosis.

Discussion

In this study, we showed the crucial role of m¹A-m⁶A crosstalk in pathological skin fibrosis (Figure 8H). RNA modification is a crucial aspect of epigenetics and plays an important role in the mechanisms governing gene expression and cell fate determination. Recently, attention has been focused on the epigenetic regulatory network. For instance, diverse histone deacetylase (HDAC) inhibitors exert anticancer effects by modulating and orchestrating m⁶A modifications, revealing a “histone-RNA crosstalk” process in ocular melanoma.^[26] Several studies have reported interactions between m⁶A and DNA methylation,^[27,28] such as the interaction between YTHDC2 and TET1, which is crucial for transposon activity and the fate of human embryonic stem cells (hESCs).^[27] Here, we propose the existence of an RNA modification network for the first time, uncovering the synergistic role of m¹A and m⁶A modification-related proteins in promoting pathological skin fibrosis, thereby broadening our current understanding of RNA methylation.

Previously, research on RNA methylation modifications has predominantly focused on their oncogenic role in cancer, including impacts on cell proliferation and apoptosis, invasion and metastasis, and metabolic reprogramming. For instance, ALKBH3-mediated SP100 demethylation has been implicated in tumorigenesis in ocular melanoma, as determined by both in vitro and in vivo experiments.^[13] Moreover, few studies have mentioned that METTL3, a canonical writer protein for m⁶A methylation, plays a pathogenic role in other fibrotic diseases. Notably, METTL3-mediated m⁶A methylation promotes liver fibrosis by enhancing the secretion of TGF-β.^[29] Additionally, the highly specific small-molecule inhibitor of METTL3, STM2457 (Chemicals, DC53045, Shanghai, China), has been shown to attenuate kidney fibrosis in vivo.^[30] However, the relationship between RNA modifications and skin fibrosis remains unclear. In this study, we found that ALKBH3-mediated m¹A demethylation of METTL3 transcription affects the m⁶A methylation levels of collagen, leading to the development of skin fibrosis. This finding is the first to elucidate the critical roles of two types of methylation modifications in skin fibrotic diseases, thereby expanding our current understanding of methylation modifications.

We anticipate that targeting the ALKBH3-METTL3 axis for the treatment of pathological skin fibrosis could be applied clinically in the near future. Although both ALKBH3 and METTL3 contribute to fibrosis, it is likely that ALKBH3 represents a more suitable clinical target for intervention. First, ALKBH3 functions as an upstream regulator of METTL3 and plays a central role in the regulation of skin fibrosis. Second, literature reports that METTL3 serves as an important regulator for immune homeostasis,^[31,32] pro-angiogenic functions,^[33,34] wound healing process,^[35,36] indicating that targeting METTL3 may result in various side effects.

To further advance clinical translation, we explored the clinical relevance of our findings and showed that treatment with ASO-Alkbh3 effectively reduced fibrosis severity in a skin fibrosis model. ASOs can precisely modulate the transcript levels of both precursor mRNAs (pre-mRNAs) and mature mRNAs, demonstrating notable target specificity. This capability could broaden the therapeutic landscape for a spectrum of diseases previously considered untreatable. For example, eteplirsen, which targets exon 51 of dystrophin pre-mRNA, restores the translational reading frame and is approved for Duchenne muscular dystrophy (DMD) treatment.^[37] Additionally, nusinersen, an ASO drug targeting Survival motor neuron-2 (SMN-2) mRNA has been approved for the treatment of spinal muscular atrophy (SMA).^[38] Notably, the current treatments for skin fibrosis are limited and often ineffective, making ALKBH3-targeted therapy a valuable therapeutic approach. However, the clinical application of ASOs has certain limitations: ASOs possess a high negative charge, are susceptible to enzymatic degradation and are rapidly cleared from the circulatory system.^[39] To address these challenges, researchers have explored various ASO delivery systems, including lipid nanoparticles, liposomes, polymer nanoparticles, and bioconjugates,^[40,41] which hold promise for revolutionizing the treatment landscape of numerous diseases in the near future.

Despite the significant findings of this study, several limitations should be acknowledged, as they may affect the interpretation and generalizability of our results. First, due to the limited availability of clinical samples, HTSs were the only means by which we could validate the role of m¹A in skin fibrosis. Nevertheless, in our study, we employed two animal models, namely the mechanical stretch-induced HTS and bleomycin-induced skin fibrosis models, both of which are classic representations of skin fibrosis diseases. Furthermore, while our study utilized ALKBH3 global knockout mice, we acknowledge the potential involvement of multiple cell types in the pathogenesis of skin fibrosis. Our comprehensive single-cell analysis and functional validation studies consistently demonstrated that fibroblasts serve as the primary effector cells mediating ALKBH3-dependent fibrotic processes. However, to further substantiate these findings and elucidate potential contributions from other cellular components, future studies employing conditional knockout models will be essential. This will provide more robust and specific evidence regarding the cell-type-specific role of ALKBH3 in skin fibrosis. Additionally, while ALKBH3 is known to exhibit diverse functions across various RNA species, including tRNAs, rRNAs, mRNAs, and mitochondrial RNAs,^[8,9] its role in the m¹A demethylation of tRNAs/rRNAs in the context of skin fibrosis remains unclear and warrants further investigation.

In summary, this study is the first to reveal the existence of “RNA methylation crosstalk” mediated by m¹A and m⁶A, providing new insights into RNA modifications. Additionally, our findings reveal a novel fibrosis mechanism where ALKBH3-mediated m¹A demethylation enhances METTL3 expression, which in turn promotes fibrosis by elevating m⁶A levels of COL1A1 and FN1 transcripts. This insight offers a new avenue for targeted therapeutic intervention.

Experimental Section

Patients and Tissue Specimens

Hypertrophic scar and normal skin samples were obtained from patients who underwent surgery at the Department of Plastic and Reconstructive Surgery at Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine. Written informed consent was obtained before sample collection in accordance with the Declaration of Helsinki and with approval from the Human Research Ethics Committee of Shanghai Jiao Tong University School of Medicine (patient information is summarized in Table S2, Supporting Information). The ethics permit number for the use of clinical samples collected during surgery was 2018-129-T107.

Cell Culture and Treatment

The isolation and culture of fibroblasts from human HTS and NS tissue were conducted as previously described.^[42] Briefly, surgical specimens were dissected into 5 mm×5 mm pieces and incubated in 0.3% dispase II (0.3 g mL⁻¹; Gibco, 17 105 041) at 4 °C for 12 h. Then, the epidermis was removed, and the dermis was minced and incubated in collagenase NB4 (3 mg mL⁻¹; Nordmark, S1745401) at 37 °C for 4 h to isolate dermal fibroblasts. The EC and SMC cell line were purchased from the American Type Culture Collection (Manassas, VA, USA). Fibroblasts and ECs were cultured in DMEM (Gibco, USA) supplemented with 10% foetal bovine serum (FBS; Gibco, USA) and 1% penicillin‒streptomycin (Gibco, USA) at 37 °C in a humidified atmosphere with 5% CO_2. Notably, the primary fibroblasts could be passaged up to 6 passages. SMCs were cultured in Smooth Muscle Cell Medium (SMCGS, Cat #1152, ScienCell) supplemented with 10 mL of fetal bovine serum (FBS, Cat. No. 0010, ScienCell), 5 mL of smooth muscle cell growth supplement (SMCGS, Cat. No. 1152, ScienCell), and 5 mL of penicillin/streptomycin solution (P/S, Cat. No. 0503, ScienCell) at 37 °C in a humidified atmosphere with 5% CO₂.

Dot Blot Assay

Total RNA was extracted using TRIzol Reagent (Invitrogen, USA), and 2 or 3 µg of RNA was spotted onto two separate nitrocellulose membranes (Millipore, INYC00010), one for total RNA detection and the other for m¹A or m⁶A methylation analysis. For total RNA detection, the membrane was stained with 0.02% methylene blue for 30 min, followed by a brief wash with nuclease-free water briefly before capture. For methylation analysis, the membrane was cross-linked under ultraviolet (UV) light, blocked with 5% milk for 1 h at room temperature, and then incubated overnight at 4 °C with anti-m¹A antibody (ab208196, Abcam, USA) and anti-m6A antibody (A19841, abclonal) at 4 °C overnight. Afterward, the membrane was incubated with horseradish peroxidase (HRP)-conjugated anti-rabbit IgG (#4412, CST, USA) for 1 h at room temperature (RT) and visualized using enhanced chemiluminescence (ECL) (Millipore, WBKLS0100).

RNA Isolation and qRT–PCR

Total RNA was extracted from cultured cells using TRIzol Reagent (Invitrogen, USA), and cDNA was synthesized using PrimeScript RT Reagent Mix (Takara Bio, RR036A) according to the manufacturer's instructions. qRT‒PCR was performed on an ABI QuantStudio 6 Flex system using SYBR Premix (Takara, RR066A) according to the manufacturer's instructions. The sequences of the primers used are summarized in Table S3 (Supporting Information).

Western Blot (WB) Analysis and IF Staining

WB analysis and IF staining were performed according to the protocol described in our previous study.^[43] The antibodies used for WB analysis and IF staining are listed in Table S4 (Supporting Information). The original WB images are available in Figures S19 and S20 (Supporting Information). Image-Pro Plus 6.0 software was used for quantitative analysis.

Plasmid Construction and RNA Interference

Knockdown of RNA expression in fibroblasts was achieved by transfection with siRNA sequences synthesized by Zorin Biotechnology Co., Ltd. (Shanghai, China). The sequences of the siRNAs used are listed in Table S5 (Supporting Information). The METTL3 and ALKBH3 overexpression cassettes were generated via PCR, inserted into MSCV and CMV vectors respectively, and confirmed by DNA sequencing. (the sequences used to construct the overexpression plasmid are summarized in Table S6, Supporting Information). Transfection of siRNAs and overexpression plasmids was performed using Lipofectamine 3000 transfection reagent (Invitrogen, L3000008) according to the manufacturer's instructions.

Cell Proliferation Assays

CCK-8 colorimetric assays were employed to assess cell proliferation. A total of 2000–3000 cells were seeded into 96-well plates (Corning, USA) in 100 µL of medium. 3 h prior to detection, 10 µL of CCK-8 solution (Dojindo, Japan) was added, and the samples were incubated at 37 °C. To measure the absorbance of the samples at 450 nm, a microplate reader (ELX800, BioTek, USA) was used.

EdU Incorporation Assay

A total of 5000–6000 cells were seeded in 24-well plates. After 24 h of incubation, cell proliferation was evaluated via the incorporation of EdU with an EdU Cell Proliferation Assay Kit (Invitrogen, Click-iT EdU Imaging Kit, C10337). The steps were conducted according to the manufacturer's protocol. In brief, cells were incubated with 50 µm EdU for 3.5 h before fixation, permeabilization, and EdU staining. Then, the cell nuclei were stained with 4′,6-diamidino-2-phenylindole (1 µg mL⁻¹, DAPI; Sigma–Aldrich, D9542) for 20 min. The proportion of cells with EdU incorporation was determined with a Zeiss 710 laser scanning microscope (Thornwood, NY, USA).

Apoptosis and Cell Cycle Assays

Apoptosis was assessed as in a previous study using a fluorescein isothiocyanate (FITC)-Annexin V apoptosis kit (BD Biosciences, San Diego, CA) was described following the manufacturer's instructions.^[43] In brief, cells were washed twice with cold PBS and stained with FITC-Annexin V and propidium iodide (PI) on ice for 5 min. For cell cycle analysis, cells were collected and fixed with 75% cold ethanol at 4 °C for 2 h. The cells were subsequently rinsed with PBS 3 times and stained sequentially with RNase A and PI (Cell Cycle Assay Kit, Dojindo, Japan) according to the manufacturer's instructions. The samples were subjected to flow cytometric analysis (BD LSRFortessa analyzer, BD Biosciences).

Animals

Six- to eight-week-old C57BL/6 mice were purchased from Cyagen Biosciences, Inc. (Cyagen, Suzhou, China). Alkbh3^−/− C57BL/6 mice were generated via the CRISPR/Cas9 system. Single-guide RNAs (sgRNAs) were designed to target exons 3 to 5 of Alkbh3 and were coinjected with Cas9 into the zygotes. The pups obtained were genotyped by PCR. After genotyping, the F0 mice were subjected to serial mating to generate homozygous mutant offspring.

Animal Models

All the procedures for establishing the model were conducted in accordance with the Guide for the Care and Use of Laboratory Animals and were approved by the Committee of Animal Care and Use for Research and Education (CACURE) of Shanghai Jiao Tong University School of Medicine. The ethics permit number for the animal HTS model study was SH9H-2021-A215-1.

The stretch-induced HTS model was established according to the model developed by Aarabi et al.^[44] For the bleomycin-induced skin fibrosis model, 100 µL of bleomycin solution (B8416, diluted to 1 unit mL⁻¹; Sigma–Aldrich, St. Louis, MO) was injected intradermally into the dorsal skin at four symmetrically distributed injection sites every other day for 3 weeks.

Lentiviral Packaging

A mixture of 3 µg of the indicated plasmid, 3 µg of the pMD2.D plasmid, and 6 µg of the PsPax plasmid was transfected into HEK239T cells with Lipofectamine 3000 (Invitrogen, L3000008) in Opti-MEM I Reduced Serum Medium (Gibco, USA). The medium was replaced with fresh complete medium 6 h after transfection. Then, 48 and 72 h after transfection, the virus-containing supernatant was collected, filtered through 0.45-mm cellulose acetate filters, and concentrated with a Lenti-X Concentrator (Takara Bio, USA).

RNA-seq and Data Analysis

Total RNA was extracted from fibroblasts using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). Poly(T) oligo-attached magnetic beads were used to enrich eukaryotic mRNA. After fragmentation, the mRNA was used to construct individual cDNA libraries. After cluster generation, the prepared libraries were sequenced on the Illumina NovaSeq 6000 platform. Gene expression levels were quantified as fragments per kilobase of exon model per million mapped reads (FPKM) values. The DESeq2 algorithm was used to identify differentially expressed genes, with a false discovery rate (FDR) < 0.05 and | log2(fold change) |≥ 1.5 as the thresholds.

Gene Ontology (GO) analysis of the designated genes was performed using the Database for Annotation, Visualization, and Integrated Discovery (DAVID) tool (). Fisher's exact test was used to identify the significant GO terms, and the FDR was used to correct the P values. GO terms with P < 0.05 were considered to be significantly enriched. Enrichment maps were created using Cytoscape 3.7.0, and bubble plots were constructed using GraphPad Prism 9.0 (GraphPad Software, Inc.). The correlation networks of METTL3 with other candidate genes was constructed using IPA software (Ingenuity Systems).

MeRIP-Seq and Data Analysis

MeRIP-Seq referred to here as m¹A-MeRIP-seq, was performed in accordance with published protocols with minor modifications.^[11] In brief, RNA was randomly fragmented into lengths of ≈200 nt with RNA fragmentation reagents, and protein A/G beads were coated with an anti-m¹A antibody (202 003, Synaptic Systems, Germany) by rotating at RT for 1 h. The RNA fragments were incubated with the beads at 4 °C for 4 h. Then, the captured RNA was eluted from the beads and isolated with TRIzol Reagent (Invitrogen, USA). Both the input sample and the m¹A immunoprecipitated sample were used for library preparation with the NEBNext Ultra RNA Library Prep Kit (New England Biolabs, UK). The libraries were qualified with an Agilent 2100 bioanalyzer (Agilent, USA) and sequenced on the NovaSeq 6000 platform (Illumina, USA). The harvested paired-end reads were subjected to quality control using Q30 and 3′ adaptor trimming by cutadapt software (v1.9.3) to remove low-quality reads. Clean reads obtained from all libraries were aligned to the reference genome (hg19) using HISAT2 software (v2.0.4). Methylated sites on RNAs (peaks) were identified with MACS software, and differentially methylated sites were identified with diffReps. The peaks identified as overlapping with exons of mRNAs were chosen. In addition, GO and pathway enrichment analyses were performed on the differentially methylated protein-coding genes. The MeRIP‒seq data were deposited in the Sequence Read Archive (SRA) database (PRJNA1114374). Sequencing was conducted by Kangcheng Biotech, Inc. (Shanghai, China).

RNA-Binding Protein Immunoprecipitation (RIP)-qPCR

The m¹A and RNA-binding proteins were assessed by RIP experiments using an RNA immunoprecipitation kit (P0101, Geneseed, Shanghai, China) following the manufacturer's instructions. In brief, 1.0 × 10⁷ cells were treated with 1 mL of RIP lysis buffer. The resulting supernatants were divided into two fractions: 100 µL was kept as input, and 900 µL was incubated with specific antibody- or rabbit IgG-conjugated protein A/G magnetic beads in IP buffer supplemented with RNase inhibitors at 4 °C overnight. The immunoprecipitated RNA was digested, purified, and further analyzed by qPCR. The primers and antibodies used for the RIP-qPCR experiments are listed in Tables S3 and S4 (Supporting Information), respectively.

RNA Stability

HDF cells were treated with actinomycin-D (5 µg mL⁻¹, HY-17 559, MCE) for 0, 2, 4, 6, and 8 h, and the half-life of the RNA was calculated by Prism GraphPad 9.0 (GraphPad Software, Inc.).

Luciferase Reporter Assay

The DNA fragments of the METTL3-5′UTR containing the wild-type m¹A motifs and mutant motifs (potential m¹A was replaced by T) were synthesized and inserted upstream of the firefly luciferase of the pmirGLO vector. Cultured 293T cells were transfected with pmirGLO-METTL3-WT-luc or pmirGLO-METTL3-MUT-luc. Relative luciferase activity was evaluated 48 h after transfection by the Dual-Luciferase Reporter Assay System (Promega, USA). The sequences are listed in Table S7 (Supporting Information).

RNA Pull-Down

RNA‒protein pull‒down assays were performed using a PureBinding RNA‒protein pull‒down kit (P0201, Geneseed, Shanghai, China) according to the manufacturer's instructions. Biotin-labeled ssRNA probes were synthesized in vitro by Generay Biotechnology (Shanghai) Co., Ltd. (the sequences of the ssRNA probes are listed in Table S8, Supporting Information). Briefly, the cell pellets were resuspended and homogenized using 1 mL of standard lysis buffer. Five percent of each sample was used as input. Then, 100 pmol of RNA probes and 50 µL of magnetic beads were incubated with each sample at 4 °C for 1 h with rotation. The eluted protein and input samples were diluted using SDS‒PAGE loading buffer and analysed by WB.

Protein Stability

Cycloheximide (100 µg mL⁻¹, CHX; HY-12320, MCE, USA) was added to HDF cells at predetermined intervals. Cells were harvested, and the protein stability of COL1A1 and FN1 was determined using western blotting.

Statistical Analysis

GraphPad Prism 9.0 was used for statistical analysis. Quantitative data are presented as the mean ± SD values, and comparisons between two groups were performed by unpaired Student's t test. Correlations between two sets of data were assessed using simple linear regression analysis. If the variance among three or more groups was minimal, ANOVA followed by Dunnett's post hoc test or Tukey's post hoc test was used for multigroup comparisons. A p value <0.05 was considered to indicate statistical significance, and the number of asterisks denotes the level of statistical significance (^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001).

Acknowledgements

The authors thank Kangcheng Biotech, Inc. (Shanghai, China), for the assistance with RNA-seq and m¹A MeRIP-seq, and Jiayin Biotechnology Ltd. (Shanghai, China), for assistance with m⁶A MeRIP-seq. This work was supported, in part, by National Natural Science Foundation of China (82302805, 82072177, 82272264, 82472557), Shanghai Clinical Research Center of Plastic and Reconstructive Surgery Technology Commission of Shanghai Municipality (22MC1940300), “Two Hundred Talent” program, “Hengjie” Program of Shanghai Health Youth Talent Reward Foundation, China Postdoctoral Science Foundation (2022M722132).

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

L.T., S.G., R.X., and E.Y. contributed equally to this work. L.T. and S.G. designed and performed the experiments and drafted and revised the manuscript. R.X. and E.Y. designed and performed some of the experiments. S.L., X.H., X.L., and H.L. were responsible for the discussion and interpretation of the data. Y.Z. and T.Z. contributed to the revision and approval of the manuscript. All the authors reviewed and approved the final version of the manuscript.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

P. S. Tsou, J. Varga, S. O'reilly, Nat. Rev. Rheumatol. 2021, 17, 596.

S. Gu, X. Huang, S. Luo, Y. Liu, Y. Khoong, H. Liang, L. Tu, R. Xu, E. Yang, Y. Zhao, M. Yao, T. Zan, Mol. Ther. 2024, 32, p1984.

R. Ogawa, Plast. Reconstr. Surg. 2022, 149, 79e.

J. H. W. Distler, A. H. Gyorfi, M. Ramanujam, M. L. Whitfield, M. Konigshoff, R. Lafyatis, Nat. Rev. Rheumatol. 2019, 15, 705.

T. Zhang, X. F. Wang, Z. C. Wang, D. Lou, Q. Q. Fang, Y. Y. Hu, W. Y. Zhao, L. Y. Zhang, L. H. Wu, W. Q. Tan, Biomed. Pharmacother. 2020, 129, 110287.

D. Abraham, A. Lescoat, R. Stratton, Mol. Aspects. Med. 2024, 96, 101252.

X. M. Meng, D. J. Nikolic‐Paterson, H. Y. Lan, Nat. Rev. Nephrol. 2016, 12, 325.

P. A. Boriack‐Sjodin, S. Ribich, R. A. Copeland, Nat. Rev. Drug. Discovery. 2018, 17, 435.

W. Kuang, H. Jin, F. Yang, X. Chen, J. Liu, T. Li, Y. Chang, M. Liu, Z. Xu, C. Huo, X. Yan, Y. Yang, W. Liu, Q. Shu, S. Xie, T. Zhou, Cell Discovery. 2022, 8, 25.

C. Zhang, G. Jia, Genomics Proteomics Bioinformatics. 2018, 16, 155.

X. Li, X. Xiong, K. Wang, L. Wang, X. Shu, S. Ma, C. Yi, Nat. Chem. Biol. 2016, 12, 311.

Y. Liu, S. Zhang, X. Gao, Y. Ru, X. Gu, X. Hu, Cell Commun Signal. 2024, 22, 79.

X. Gu, A. Zhuang, J. Yu, L. Yang, S. Ge, J. Ruan, R. Jia, X. Fan, P. Chai, Nucleic Acids Res. 2024, 52, 2273.

C. X. Lin, Z. J. Chen, Q. L. Peng, K. R. Xiang, D. Q. Xiao, R. X. Chen, T. Cui, Y. S. Huang, H. W. Liu, Front. Cell. Dev. Biol. 2022, 10, 947337.

V. Vorstandlechner, M. Laggner, D. Copic, K. Klas, M. Direder, Y. Chen, B. Golabi, W. Haslik, C. Radtke, E. Tschachler, K. Hotzenecker, H. J. Ankersmit, M. Mildner, Nat. Commun. 2021, 12, 6242.

A. L. Panagiotopoulos, N. Karguth, M. Pavlou, S. Bohm, G. Gasparoni, J. Walter, A. Graf, H. Blum, M. Biel, L. M. Riedmayr, E. Becirovic, Mol. Ther. Nucleic. Acids. 2020, 21, 1050.

T. Ramasamy, H. B. Ruttala, S. Munusamy, N. Chakraborty, J. O. Kim, J. Control. Release. 2022, 352, 861.

H. H. Woo, S. K. Chambers, Biochim. Biophys. Acta. Gene. Regul. Mech. 2019, 1862, 35.

B. Tu, K. Song, Y. Zhou, H. Sun, Z. Y. Liu, L. C. Lin, J. F. Ding, J. M. Sha, Y. Shi, J. J. Yang, R. Li, Y. Zhang, J. Y. Zhao, H. Tao, Pharmacol. Res. 2023, 194, 106840.

J. X. Zhang, P. J. Huang, D. P. Wang, W. Y. Yang, J. Lu, Y. Zhu, X. X. Meng, X. Wu, Q. H. Lin, H. Lv, H. Xie, R. L. Wang, Mol. Ther. 2021, 29, 3436.

D. T. Woodley, D. R. Keene, T. Atha, Y. Huang, R. Ram, N. Kasahara, M. Chen, Mol. Ther. 2004, 10, 318.

S. Zaccara, R. J. Ries, S. R. Jaffrey, Nat. Rev. Mol. Cell Biol. 2019, 20, 608.

W. Dai, N. J. Yu, R. E. Kleiner, Acc. Chem. Res. 2023, 56, 2726.

X. Wang, J. Feng, Y. Xue, Z. Guan, D. Zhang, Z. Liu, Z. Gong, Q. Wang, J. Huang, C. Tang, T. Zou, P. Yin, Nature 2016, 534, 575.

X. Jiang, B. Liu, Z. Nie, L. Duan, Q. Xiong, Z. Jin, C. Yang, Y. Chen, Signal. Transduct. Target. Ther. 2021, 6, 74.

A. Zhuang, X. Gu, T. Ge, S. Wang, S. Ge, P. Chai, R. Jia, X. Fan, Cancer Commun. 2023, 43, 1185.

T. Sun, Y. Xu, Y. Xiang, J. Ou, E. J. Soderblom, Y. Diao, Nat. Genet. 2023, 55, 1324.

S. Deng, J. Zhang, J. Su, Z. Zuo, L. Zeng, K. Liu, Y. Zheng, X. Huang, R. Bai, L. Zhuang, Y. Ye, M. Li, L. Pan, J. Deng, G. Wu, R. Li, S. Zhang, C. Wu, D. Lin, J. Chen, J. Zheng, Nat. Genet. 2022, 54, 1427.

Y. Feng, H. Dong, B. Sun, Y. Hu, Y. Yang, Y. Jia, L. Jia, X. Zhong, R. Zhao, Cell Mol. Gastroenterol. Hepatol. 2021, 12, 839.

H. R. Jung, J. Lee, S. P. Hong, N. Shin, A. Cho, D. J. Shin, J. W. Choi, J. I. Kim, J. P. Lee, S. Y. Cho, Exp. Mol. Med. 2024, 56, 355.

H. B. Li, J. Tong, S. Zhu, P. J. Batista, E. E. Duffy, J. Zhao, W. Bailis, G. Cao, L. Kroehling, Y. Chen, G. Wang, J. P. Broughton, Y. G. Chen, Y. Kluger, M. D. Simon, H. Y. Chang, Z. Yin, R. A. Flavell, Nature 2017, 548, 338.

Y. Zhang, W. Zhang, J. Zhao, T. Ito, J. Jin, A. O. Aparicio, J. Zhou, V. Guichard, Y. Fang, J. Que, J. F. Urban, Jr., J. H. Hanna, S. Ghosh, X. Wu, L. Ding, U. Basu, Y. Huang, Nat. Immunol. 2023, 24, 1256.

Q. Wang, C. Chen, Q. Ding, Y. Zhao, Z. Wang, J. Chen, Z. Jiang, Y. Zhang, G. Xu, J. Zhang, J. Zhou, B. Sun, X. Zou, S. Wang, Gut 2020, 69, 1193.

J. Ma, L. Zhang, X. Zhang, L. Zhang, H. Zhang, Y. Zhu, X. Huang, T. Zhang, X. Tang, Y. Wang, L. Chen, Q. Pu, L. Yang, Z. Cao, B. S. Ding, Sci. Transl. Med. 2024, 16, eado5266.

T. Wang, X. Li, Y. Tao, X. Wang, L. Li, J. Liu, J. Transl. Med. 2024, 22, 643.

J. Zhou, T. Wei, Z. He, Mol. Med. 2021, 27, 146.

J. S. Charleston, F. J. Schnell, J. Dworzak, C. Donoghue, S. Lewis, L. Chen, G. D. Young, A. J. Milici, J. Voss, U. Dealwis, B. Wentworth, L. R. Rodino‐Klapac, Z. Sahenk, D. Frank, J. R. Mendell, Neurology 2018, 91, 637.

R. S. Finkel, E. Mercuri, B. T. Darras, A. M. Connolly, N. L. Kuntz, J. Kirschner, C. A. Chiriboga, K. Saito, L. Servais, E. Tizzano, H. Topaloglu, M. Tulinius, J. Montes, A. M. Glanzman, K. Bishop, Z. J. Zhong, S. Gheuens, C. F. Bennett, E. Schneider, W. Farwell, D. C. De Vivo, E. S. Group, N. Engl. J. Med. 2017, 377, 1723.

M. A. Younis, H. M. Tawfeek, A. a. H. Abdellatif, J. A. Abdel‐Aleem, H. Harashima, Adv. Drug. Delivery. Rev. 2022, 181, 114083.

M. J. Mitchell, M. M. Billingsley, R. M. Haley, M. E. Wechsler, N. A. Peppas, R. Langer, Nat. Rev. Drug. Discovery. 2021, 20, 101.

S. Nummelin, J. Kommeri, M. A. Kostiainen, V. Linko, Adv. Mater. 2018, 30, 1703721.

S. Gu, X. Huang, X. Xu, Y. Liu, Y. Khoong, Z. Zhang, H. Li, Y. Gao, T. Zan, BMC Genomics 2021, 22, 613.

X. Huang, S. Gu, C. Liu, L. Zhang, Z. Zhang, Y. Zhao, Y. Khoong, H. Li, Y. Gao, Y. Liu, Z. Wang, D. Zhao, Q. Li, T. Zan, J. Invest. Dermatol. 2022, 142, 1065.

S. Aarabi, K. A. Bhatt, Y. Shi, J. Paterno, E. I. Chang, S. A. Loh, J. W. Holmes, M. T. Longaker, H. Yee, G. C. Gurtner, FASEB J. 2007, 21, 3250.