METTL3 protein is upregulated in HPH animals

To reveal the role of RNA methylation in the progression of PH, we prepared hypoxic mouse models of PH (Fig. 1A, B), and assessed the expression of key enzymes associated with RNA methylation, including METTL3, METTL14, Wtap, FTO and Alkbh5. Among the three 'writer' proteins, METTL3 was the most significantly upregulated one in the lungs of hypoxic pulmonary hypertension (HPH) mice, while METTL14 and Wtap exhibited smaller increases in expression (Fig. 1C). Oppositely, the m6A 'eraser' protein, FTO, was significantly decreased in the HPH mouse lung, whereas Alkbh5 levels remain unchanged. Elevated levels of METTL3 protein were also observed in lung tissues and pulmonary arteries (PAs) of hypoxic PH rats (Fig. 1D, E), indicating the potential regulatory significance of METTL3 in PH progression. Consequently, we focused on METTL3 in following investigation.

Fig. 1

METTL3 protein is upregulated during PH development. A, B Male C57BL/6 mice (6-week-old) were subjected to either normoxia or hypoxia for 3 weeks. The right ventricular systolic pressure (RVSP) (A) and right ventricular hypertrophy index (RVHI) (B) for both PH and control animals were assessed (n = 5). C The protein levels of METTL3, METTL14, Wtap, FTO and Alkbh5 in the lungs of PH and control animals were detected by western blotting (n = 3). D, E The protein levels of METTL3 in rat lung (rLung, D) and pulmonary artery (rPA, E) were also examined by western blotting (n = 3). β-actin was used as a loading control for western blotting. Nor: normoxia; Hyp: hypoxia. Data were analyzed by a two-tailed unpaired t test. Statistical significance is denoted by *P < 0.05, **P < 0.01 and ***P < 0.001. ns non-significance

Smooth muscle-specific knockout of Mettl3 aggravates hypoxia-induced PH in mouse

To elucidate the role of METTL3 in PH progression in vivo, we developed smooth muscle-specific Mettl3 knockout mice (Fig. 2A). The knockout mice of SMMHC-CreERT2;Mettl3fl/fl (Mettl3SMCKO) and the control mice of Mettl3fl/fl were subjected to either hypoxia (10% O2) or normoxia (21% O2) exposure for 3 weeks. Hemodynamic analysis revealed that right ventricular systolic pressure (RVSP) and right ventricular hypertrophy index (RVHI) in Mettl3fl/fl mice were considerably increased under hypoxia compared to those under normoxia (Fig. 2B, C). Moreover, hypoxic Mettl3SMCKO mice exhibited an even greater elevation in RVSP and RVHI than Mettl3fl/fl mice under hypoxia (Fig. 2B, C). This suggests that smooth muscle-specific knockout of Mettl3 exaggerates hypoxia-induced PH hemodynamic changes. In addition, histological analysis revealed a significant augmentation of pulmonary arterial wall thickness and remodeling in Mettl3SMCKO mice compared to Mettl3fl/fl mice (Fig. 2D–F). Western blotting analysis confirmed a significant reduction of METTL3 in the PAs of Mettl3SMCKO mice (Fig. S1), thereby validating the effective knockout of Mettl3. Immunostaining also confirmed the lack of METTL3 in the PAs of Mettl3SMCKO mice, associated with an increase in pulmonary arterial wall thickness, as indicated by anti-α-SMA staining (Figs. 2G, S2). Enhanced pulmonary vascular remodeling in Mettl3SMCKO mice under both normoxia and hypoxia was further confirmed by anti-Calponin immunostaining, which demonstrated pronounced co-localization with α-SMA (Fig. S2).

Fig. 2

Smooth muscle-specific knockout of Mettl3 promotes the progression of PH. A Schematic illustrating conditional smooth muscle-specific Mettl3 loss-of-function mouse model. B, C The RVSP B was measured in mmHg by right heart catheterization, and the RVHI C was determined as the ratio of the weight of RV to the sum of LV plus ventricular septum (RV/(LV + S)) for knockout mice SMMHC-CreERT2;Mettl3fl/fl (Mettl3SMCKO) and the control mice Mettl3fl/fl (n = 6). D Representative images of hematoxylin and eosin (HE)-stained lung sections in both Mettl3SMCKO and Mettl3fl/fl mice. Scale bars, 100 μm. E, F The wall thickness of the PAs was calculated for 6 mice, with n = 120 per group. The relative wall thickness was given by (outer perimeter—inside perimeter)/outer perimeter (E), and the relative wall area by (outer area—inside area)/outer area (F). G Representative double-labeled immunostaining with antibodies against METTL3 and α-SMA in the PAs of SMMHC-CreERT2;Mettl3fl/fl (Mettl3SMCKO) and the control mice Mettl3fl/fl. Scale bars, 50 μm (merged) and 20 μm (magnified). H The mRNA levels of METTL3 and PCNA in mouse PAs were evaluated by qRT-PCR (n = 5). β-actin was used as an internal reference for qRT-PCR. I The m6A levels in total RNA from mouse PAs were analyzed using methylation quantification kit (P-9005, EpiGentek) (n = 6). Nor: normoxia; Hyp: hypoxia. Data were analyzed by using a one-way ANOVA followed by Tukey's multiple comparisons test. Statistical significance is denoted by * P < 0.05, ** P < 0.01 and *** P < 0.001

The qRT-PCR assay and m6A methylation quantification revealed a significant upregulation in METTL3 expression (Fig. 2H) and m6A level (Fig. 2I) in the PAs of Mettl3fl/fl mice under hypoxia compared to normoxia. Conversely, a significant reduction in METTL3 expression and m6A level was found in the PAs of Mettl3SMCKO mice relative to Mettl3fl/fl mice under both normoxia and hypoxia. Trace amounts of METTL3 were detected in the PAs of knockout mice, potentially originating from tissues such as pulmonary artery endothelial cells (PAECs) and pulmonary artery fibroblasts (PAFs) where METTL3 was not knocked out. Another possibility is the detection of truncated transcripts resulting from exon deletion (including exons 2 and 3, and potentially 4 through alternative splicing) (Figs. S3, S4; Table S3). Sequence analysis revealed that frameshift mutations in these truncated transcripts introduce premature termination codons, precluding the synthesis of functional METTL3 protein. Additionally, the qRT-PCR assay highlighted an elevated level of PCNA in the PAs of Mettl3SMCKO mice compared to Mettl3fl/fl mice under hypoxia (Fig. 2H). Furthermore, Ki67 expression significantly increased in both normoxic and hypoxic conditions following Mettl3 knockout, suggesting that METTL3 plays a broad regulatory role in cell proliferation mechanisms, extending beyond those induced by hypoxia alone (Fig. S5). Overall, our findings suggest that Mettl3 deletion might aggravate the development of PH in mice, possibly through altering cellular proliferative phenotype.

METTL3 deficiency drives phenotypic switching in PASMCs

To clarify the role of METTL3 in influencing PASMCs phenotype, lentivirus-mediated METTL3-specific shRNA was employed to silence METTL3 in rat PASMCs (rPASMCs). Three days after infection, METTL3 expression was dramatically repressed at both mRNA and protein levels in rPASMCs treated with shMETTL3 compared to the shNC group (Fig. 3A, B). Subsequent RNA sequencing (RNA-seq) was performed to analyze the transcriptional shifts in rPASMCs exposed to shMETTL3 versus the control group. Quantitative analysis revealed 1506 differentially expressed transcripts in METTL3-silenced rPASMCs [padj-value < 0.001; fold change (FC) ≥ 2], consisting of 656 up- and 850 down-regulated genes (Fig. S6A). Kyoto encyclopedia of genes and genomes (KEGG) analysis of these differentially expressed genes (DEGs) highlighted pathways impacted by METTL3 inhibition, including vascular smooth muscle contraction, cell adhesion, and calcium signaling (Fig. S6B).

Fig. 3

Elimination of METTL3 induces a phenotypic switch in PASMCs from contractile to synthetic. A, B METTL3 expression levels were assessed in rPASMCs infected with shNC or shMETTL3 lentiviruses by qRT-PCR (A) and western blotting (B), respectively. The bar chart depicts the relative METTL3 protein level (n = 3). C A heatmap displays the expression of VSMCs markers in transcriptome sequencing (n = 3). D–H The RNA levels of α-SMA, SM22, Smoothelin, Calponin, PCNA and MMP2 in rPASMCs (D) and hPASMCs (G) infected with shNC or shMETTL3 were determined by qRT-PCR (n = 3). The protein levels of α-SMA, SM22, Smoothelin and Calponin in rPASMCs (E) and hPASMCs (H) infected with shNC or shMETTL3 were assessed by western blotting (n = 3). The protein levels of METTL3 in shNC or shMETTL3 hPASMCs were assessed (F) (n = 3). I The mRNA levels of SM22, α-SMA, Smoothelin and Calponin in mouse PAs were determined by qRT-PCR (n = 5). J Representative images of EdU labeling depicts the proliferation of rPASMCs upon inhibition and overexpression of METTL3. EdU-positive cells was quantified across 10 random fields, with DAPI staining highlighting all cells (n = 3). Scale bar represents 200 μm. The bar chart illustrates the proportion of EdU-positive cells. K Representative images from wound healing assay display the migration of rPASMCs following inhibition and overexpression of METTL3 (n = 3). Scale bar represents 1000 μm. Bar chart elucidates the changes in wound width at 72 h. β-actin was used as an internal reference for qRT-PCR and as a loading control for western blotting. Data were analyzed by a two-tailed unpaired t test, except the expression levels in PAs were analyzed by a one-way ANOVA followed by Tukey's multiple comparisons test. Statistical significance is denoted by * P < 0.05, ** P < 0.01 and *** P < 0.001

Among the DEGs, contractile marker genes such as α-SMA (Acta2), SM22 (Tagln), Smoothelin (Smtn) and Calponin (Cnn1) were significantly downregulated upon METTL3 inhibition (Fig. 3C). This downregulation was further confirmed by qRT-PCR and western blotting in rPASMCs (Fig. 3D, E) and human PASMCs (hPASMCs) (Fig. 3F–H). Conversely, overexpression of METTL3 (Fig. S7A) led to upregulation of α-SMA, SM22, Smoothelin and Calponin in rPASMCs (Fig. S7B) and hPASMCs (Fig. S7C). Similarly, a downregulation of SM22, α-SMA, Smoothelin and Calponin was observed in the PAs of Mettl3SMCKO mice relative to Mettl3fl/fl mice under hypoxia (Fig. 3I). The proliferative marker, PCNA, exhibited heightened levels after METTL3 silencing (Fig. 3D). Simultaneously, the migration marker MMP2 also upregulated following METTL3 inhibition (Fig. 3D) but downregulated with METTL3 overexpression (Fig. S7B-C). In addition, EdU incorporation and wound healing assay further demonstrated that METTL3 knockdown facilitated, whereas its overexpression reduced, the proliferation and migration of rPASMCs (Fig. 3J, K).

Collectively, these findings demonstrate that METTL3 deficiency led to a significant shift in PASMCs from a contractile to a synthetic phenotype, thereby potentiating PH progression.

Knockdown of METTL3 supresses miR-143/145 expression via m6A-dependent impairment of miRNA maturation

To explore the mechanism by which METTL3 mediates phenotypic transition of PASMCs, small RNA sequencing was conducted in rPASMCs treated with shNC or shMETTL3 lentiviruses. Inhibition of METTL3 resulted in a disordered miRNA expression profile, with a marked decrease of multiple miRNAs including miR-204-5p, miR-129-2-3p, miR-149-5p, miR-28-5p, miR-184, miR-425-5p, miR-145-5p, miR-23a-3p, miR-129-5p, miR-143-3p and miR-328a-3p (Fig. 4A, B). Among them, miR-143-3p and miR-145-5p have been identified as pivotal regulators of VSMCs phenotype [23]. Similarly, we verified that knockout of Mettl3 markedly reduced the levels of miR-143-3p and miR-145-5p in the PAs of Mettl3SMCKO mice compared with Mettl3fl/fl mice under both normoxia and hypoxia conditions (Fig. 4C). Furthermore, METTL3 overexpression resulted in elevated miR-143-3p and miR-145-5p levels in rPASMCs (Fig. S8A) and hPASMCs (Fig. S8B), suggesting that METTL3 plays a regulatory role in the expression of miR-143-3p and miR-145-5p.

Fig. 4

Loss of METTL3 impairs miR-143/145 cluster processing in an m6A-dependent manner. A A heatmap displays differentially expressed miRNAs in rPASMCs following infection with shNC or shMETTL3 lentiviruses based on small RNA sequencing (n = 3). B Differentially expressed miRNAs in rPASMCs were validated by qRT-PCR (n = 3). C The expression levels of miR-143-3p and miR-145-5p were detected in the PAs of either Mettl3SMCKO or Mettl3fl/fl mice by qRT-PCR (n = 6). snoRNA202 was used as an internal reference in qRT-PCR for miRNA. D–I HEK293T cells infected with shNC or shMETTL3 lentiviruses were further transfected with pri-miR-143 or pri-miR-145 overexpression plasmids. The inhibition of METTL3 was verified by qRT-PCR (D, G). The pri-miR-143, miR-143-3p, miR-143-5p (E), pri-miR-145, miR-145-3p, and miR-145-5p (H) were detected by qRT-PCR (n = 3). The m6A enrichment of pri-miR-143 (F) or pri-miR-145 (I) was ascertained by MeRIP-qPCR (n = 3). J–L The shNC and shMETTL3 HEK293T cells were transfected with plasmids overexpressing either the wild-type or m6A-mutant versions (mut, A-to-T mutation) of human pri-miR-143 and pri-miR-145 (J), and the pri-miR-143, miR-143-3p (K), pri-miR-145, and miR-145-5p (L) were detected by qRT-PCR (n = 3). β-actin or snoRNA202 was used as an internal reference in qRT-PCR for pri-RNA or miRNA (B, C), respectively. Green fluorescent proteins (GFP) encoded by the pri-miRNA overexpression vector was used as an internal reference in qRT-PCR to evaluate pri-miRNA processing. A two-tailed unpaired t test (B–I), or one-way ANOVA followed by Tukey's multiple comparisons test (K, L), were used to estimate the significance. Statistical significance is denoted by * P < 0.05, ** P < 0.01 and *** P < 0.001. ns: non-significance

We subsequently explore how METTL3 regulates the expression of miR-143-3p and miR-145-5p. Previous studies have reported that METTL3 can modulate pri-miRNA processing through m6A modification [24, 25]. To elucidate the role of METTL3 in regulating miR-143/145 maturation, we transfected HEK293T cells, that have been infected with shNC or shMETTL3 lentiviruses, with plasmids overexpressing human pri-miR-143 or pri-miR-145. Our findings demonstrated that METTL3 knockdown increased pri-miR-143 but decreased miR-143-3p and miR-143-5p levels compared with the shNC groups (Fig. 4D, E), suggesting that METTL3 suppression impedes pri-miR-143 processing. Similar results were observed for pri-miR-145 processing upon METTL3 silencing (Fig. 4G, H). MeRIP-qPCR analyses verified that both pri-miR-145 and pri-miR-143 undergo m6A methylation (Fig. 4F, I). Moreover, silencing METTL3 led to reduced m6A levels in these pri-miRNAs (Fig. 4F, I), suggesting that METTL3-mediated m6A reduction impairs the processing of pri-miR-143 and pri-miR-145. Intriguingly, an increase in miR-143-3p and miR-145-5p levels under hypoxic conditions paralleled significant elevations in METTL3 expression in the pulmonary arteries (PAs) of Mettl3SMCKO mice compared with Mettl3fl/fl mice (Figs. 4C, 2H). Additionally, depletion of Mettl3 led to high levels of pri-miR-143 and pri-miR-145 in the PAs of Mettl3SMCKO mice compared to Mettl3fl/fl mice under both normoxic and hypoxic conditions (Fig. S9). These results underscore the role of METTL3 in regulating the production of miR-143-3p and miR-145-5p.

Next, we employed SRAMP (http://www.cuilab.cn/sramp/) for further analysis and identified potential m6A modification sites within both pri-miR-143 and pri-miR-145 in human, rat and mouse (Figs. 4J, S10, S11, and S12; Supplementary Table S5). To investigate the effects of m6A modifications on miRNA processing, we generated plasmids overexpressing human pri-miR-143 and pri-miR-145 with adenosines at these sites replaced by thymines (Fig. 4J). Our experiments revealed that mutation at these m6A sites resulted in effects akin to those observed with METTL3 inhibition: increased levels of pri-miR-143 but decreased miR-143-3p levels compared with the wild-type (WT) controls (Fig. 4K, L). Similar results were observed for pri-miR-145, indicating that METTL3-mediated m6A modification plays a crucial role in the processing of pri-miR-143 and pri-miR-145.

Previous studies have identified hnRNPA2B1 as an m6A mediator during miRNA maturation, and its inhibition impedes miRNA processing [26]. Our results showed that silencing hnRNPA2B1 significantly reduced the expression of miR-143-3p and miR-145-5p (Fig. S13A, B), whereas overexpression of hnRNPA2B1 increased their levels (Fig. S13C, D). Silencing METTL3 reversed the hnRNPA2B1-induced enhancement of miR-143-3p and miR-145-5p expression (Fig. S13D), indicating a potential role for hnRNPA2B1 in m6A-mediated processing of miR-143/145 cluster.

In summary, our findings demonstrate that inhibiting METTL3 reduces miR-143/145 levels through m6A-dependent disruption of miRNA maturation. Furthermore, hypoxia-induced upregulation of METTL3, along with the subsequent increase in miR-143/145, may protect against the development of PH.

miR-145-5p and miR-143-3p regulate phenotypic transformation of PASMCs via their specific targets KLF4 and FSCN1

To determine the METTL3-mediated role of miR-143/145 in PASMCs phenotypic modulation, rPASMCs were transfected with miR-143-3p or miR-145-5p mimic and inhibitor. The data indicated that introduction of miR-143-3p or miR-145-5p mimic upregulated contractile markers such as Smoothelin, α-SMA, Calponin and SM22 (Fig. 5A, C). In contrast, their inhibitors led to downregulation of these contractile proteins (Fig. 5B, D). Introduction of miR-143-3p or miR-145-5p mimic yielded similar effects in hPASMCs as observed in rPASMCs (Fig. S14A, B), highlighting the pivotal role of miR-143-3p and miR-145-5p in influencing the PASMC phenotype. The EdU assay demonstrated that transfection with either miR-143-3p or miR-145-5p mitigated the cellular proliferation in both rPASMCs (Figs. 5E, F, S15A) and hPASMCs (Fig. S15B, C). Furthermore, a wound healing assay indicated that the introduction of miR-143-3p or miR-145-5p mimic suppressed the migration of rPASMCs, further validating their functional importance in modulating PASMC behavior (Figs. S15D, 5G, H).

Fig. 5

miR-145-5p and miR-143-3p modulate PASMCs phenotypic switching via their specific targets. A–D Western blotting analysis was conducted to determine the expression of SM22, α-SMA, Smoothelin, and Calponin in rPASMCs transfected with miR-143-3p mimic (A), miR-143-3p inhibitor (B), miR-145-5p mimic (C) and miR-145-5p inhibitor (D) (n = 3). Bar charts represent the relative protein levels. E–H Transfection of miR-143-3p and miR-145-5p mimics was performed in rPASMCs, followed by EdU incorporation and wound healing assay. Bar chart illustrating the proportion of EdU-positive cells upon transfection of miR-143-3p (E) and miR-145-5p (F) (n = 3). Bar chart also elucidates the changes of wound width at 72 h in wound healing assay upon transfection of the mimics of miR-143-3p (G) and miR-145-5p (H). I The expression levels of KLF4 and FSCN1 in METTL3-silenced rPASMCs were assessed by qRT-PCR (n = 3). J The mRNA levels of KLF4 and FSCN1 in mouse PAs were determined by qRT-PCR (n = 6). K–N The KLF4 mRNA and protein levels in rPASMCs transfected with miR-145-5p mimic (K–L) and miR-145-5p inhibitor (M–N) were detected by qRT-PCR and western blotting, respectively (n = 3). O–R Similarly, the FSCN1 mRNA and protein levels in rPASMCs transfected with miR-143-3p mimic (O–P) and miR-143-3p inhibitor (Q–R) were also assessed using the same techniques (n = 3). S The expression levels of KLF4 and FSCN1 in hnRNPA2B1-silenced rPASMCs were assayed by qRT-PCR (n = 3). β-actin was used as an internal reference for qRT-PCR and as a loading control for western blotting. A two-tailed unpaired t test (A–I, K–S), or one-way ANOVA followed by Tukey's multiple comparisons test (J), were used to estimate the significance. Statistical significance is denoted by * P < 0.05, ** P < 0.01 and *** P < 0.001. ns: non-significance

To discern the molecular mechanisms through which miR-143/145 influence PASMCs phenotype, we aimed to identify their targets. Transcriptome analysis, followed by qRT-PCR assays, revealed a significant upregulation of Krüppel-like Factor 4 (KLF4) and fascin actin-bundling protein 1 (FSCN1) in METTL3-silenced rPASMCs (Figs. 3C, 5I). Remarkably, depletion of Mettl3 markedly resulted in a pronounced elevation of KLF4 and FSCN1 expression levels in the PAs of Mettl3SMCKO mice compared to Mettl3fl/fl mice under hypoxic conditions (Fig. 5J). Previous studies identified KLF4 as a target of miR-145-5p in human embryonic stem cells [27] and FSCN1 as a target of miR-143-3p in esophageal squamous cell carcinoma [28] (Fig. S16). In rPASMCs, we observed that the miR-145-5p mimic reduced the protein levels of KLF4 without affecting its mRNA levels (Fig. 5K, L), whereas its inhibitor increased the protein levels of KLF4 without altering its mRNA levels (Fig. 5M, N). Similarly, the miR-143-3p mimic decreased both mRNA and protein levels of FSCN1 (Fig. 5O, P), whereas its inhibitor enhanced them (Fig. 5Q, R). This consistent negative correlation confirmed the direct targeting of KLF4 by miR-145-5p and FSCN1 by miR-143-3p, highlighting their roles in regulating the PASMC phenotype.

The significant reduction in miR-143/145 expression following hnRNPA2B1 silencing led us to propose a regulatory role for hnRNPA2B1 on KLF4 and FSCN1 expression. Confirming our hypothesis, hnRNPA2B1 knockdown resulted in elevated levels of KLF4 and FSCN1 (Figs. S13A, 5S), underscoring the impact of hnRNPA2B1 on the biogenesis of miR-143/145.

Taken together, the results suggest that RNA methylation-mediated modulation of miR-143/145 cluster plays an essential role in PASMCs phenotypic transformation via specific targets.

A miR-143/145-KLF4 positive feedback loop facilitates PASMCs phenotypic transition

Previous studies have highlighted the key role of KLF4 in regulating phenotypic switching in smooth muscle cells [29, 30]. Our findings reinforced this by showing that KLF4 overexpression decreased the mRNA and protein levels of contractile markers, including SM22, α-SMA, Calponin and Smoothelin in rPASMCs (Fig. 6A, B).

Fig. 6

A positive feedback circuit between miR-143/145 and KLF4 promotes PASMCs phenotypic switching. A, B The mRNA and protein levels of SM22, α-SMA, Smoothelin and Calponin in rPASMCs transfected with OE-Con or OE-KLF4 lentiviruses were measured by qRT-PCR (A) and western blotting (B), respectively (n = 3). Bar charts show the relative protein levels. β-actin was used as an internal reference for qRT-PCR and as a loading control for western blotting. C The miR-143-3p and miR-145-5p levels were detected in OE-Con or OE-KLF4 rPASMCs by qRT-PCR (n = 3). snoRNA202 was used as an internal reference in qRT-PCR. D The potential binding site of KLF4 on miR-143/145 promoter was predicted using the JASPAR database (http://jaspar.genereg.net/). E The OE-Con or OE-KLF4 A7r5 cells were transfected with the luciferase reporter plasmids containing the wild-type miR-143/145 promoter or promoter with indicated mutation (site 1, − 317 to − 323; site 2, − 408 to − 414; site 3, − 827 to − 833; site 4, − 1246 to − 1252). The relative luciferase activity was measured at 48 h post-transfection (n = 3). F, G The expression levels of miR-145-5p (F), as well as KLF4, Smoothelin, SM22, α-SMA and Calponin (G) were detected by qRT-PCR in OE-Con or OE-KLF4 rPASMCs transfected with miR-145-5p mimic or its control (miR-Con) (n = 3). A two-tailed unpaired t test (A–D), or one-way ANOVA followed by Tukey's multiple comparisons test (F, G), were used to estimate the significance. Statistical significance is denoted by * P < 0.05, ** P < 0.01 and *** P < 0.001

Previous evidence has indicated that SRF/Myocd binds to the CArG cis-element on miR-143/145 promoter, enhancing miR-143/145 expression [15, 18, 23]. Inversely, KLF4 diminishes the binding affinity of serum response factor (SRF) to CArG-box elements [30,31,32,33]. Given these observations, we proposed that KLF4 might negatively regulate miR-143/145 expression at transcriptional level. Substantiating our hypothesis, we observed marked decreases in miR-143-3p and miR-145-5p levels upon KLF4 overexpression in rPASMCs (Fig. 6C). Using the JASPER database, four potential KLF4 binding sites on the promoter region of miR-143/145 were predicted (Fig. 6D). We then constructed the miR-143/145 promoter into a luciferase reporter and found that KLF4 overexpression significantly repressed the luciferase activity compared with the control group (Fig. 6E). Nevertheless, mutation of binding site 4, but not site 1, 2 and 3, abolished the suppressive effect of KLF4 on the reporter activity (Fig. 6E), confirming that KLF4 targets the miR-143/145 promoter at predicted site 4.

To further ascertain the bidirectional regulatory relationship between KLF4 and miR-145-5p, we conducted a rescue experiment by introducing either mimic control or miR-145-5p mimic into KLF4 overexpressed rPASMCs (Fig. 6F). Our results revealed that miR-145-5p overexpression reversed the KLF4-induced reduction in contractile genes (Figs. 6G, S17), highlighting the role of KLF4 in shaping the PASMCs phenotype through miR-145-5p. In sum, we illuminate a sophisticated positive feedback loop between KLF4 and miR-143/145, wherein METTL3-guided m6A methylation orchestrates PASMCs phenotypic switching.

Taken together, the current research reveals a novel epigenetic regulatory mechanism influencing miR-143/145 cluster expression in phenotypic switch of PASMCs. METTL3-driven and hnRNPA2B1-mediated m6A modification play a key role in regulating miR-143/145 cluster. METTL3 reduction mitigates miR-143/145 via m6A-dependent impairment of miRNA maturation. A miR-143/145-KLF4 positive feedback loop potentiates the repression of contractile markers genes, facilitating PASMCs phenotypic transition (Fig. 7). Our findings unmask a promising therapeutic approach via targeting m6A modified miR-143/145-KLF4 loop for PH treatment.

Fig. 7

Within PASMCs, METTL3 depletion reduces m6A modification, subsequently diminishing miR-143/145 expression via impeding miRNA processing in an m6A-dependent manner. The reduction in mature miR-145-5p and miR-143-3p enhances their respective targets, KLF4 and FSCN1. Augmented KLF4 in turn represses the transcription of miR-143/145 cluster, establishing a positive feedback loop with miR-143/145. This perpetuating cycle suppresses contractile markers, facilitating the phenotypic switch in PASMCs and pulmonary vascular remodeling. Hypoxia-induced upregulation of METTL3 and consequent increase of miR-143/145 may serve as a compensatory and protective agent against the progression of PH