Therapeutic Potential of Heat Shock Protein 90 Inhibitor 17-DMAG in Regulating METTL3 for Kidney Fibrosis Treatment

Visual Abstract

Abstract

Key Points

• Targeting METTL3 with the heat shock protein 90 inhibitor 17-DMAG mitigated kidney fibrosis in CKD.

• The drug repositioning through differentially expressed gene and enrichment analyses identified 17-dimethylaminoethylamino-17-demethoxygeldanamycin as a potential agent to alleviate kidney fibrosis.

• The N-terminal heat shock protein 90 inhibitor 17-dimethylaminoethylamino-17-demethoxygeldanamycin suppressed c-Jun–METTL3 signaling, attenuating N6-methyladenosine methylation and kidney fibrosis.

Background

Kidney fibrosis is a major pathological feature of CKD, characterized by excessive deposition of extracellular matrix proteins, leading to progressive loss of kidney function. N6-methyladenosine (m6A) RNA methylation has emerged as a crucial epigenetic modification implicated in various diseases, including kidney fibrosis. METTL3, an m6A writer, plays a key role in promoting fibrosis by stabilizing profibrotic gene expression. Therefore, targeting METTL3 represents a promising therapeutic strategy for CKD treatment. In this study, we explored the therapeutic potential of 17-dimethylaminoethylamino-17-demethoxygeldanamycin (17-DMAG) in regulating METTL3 to mitigate kidney fibrosis.

Methods

Through transcriptome-based drug repositioning, we identified 17-DMAG as a potential inhibitor of METTL3. Differentially expressed gene analysis was performed to assess the enrichment of 17-DMAG in CKD-related gene expression profiles. The antifibrotic effects of 17-DMAG were evaluated in in vitro and in vivo models. The mechanism by which 17-DMAG downregulates METTL3 was also investigated.

Results

17-DMAG significantly reduced METTL3 expression in renal epithelial cells in a dose-dependent and time-dependent manner. In in vivo mouse models of kidney fibrosis, 17-DMAG treatment attenuated METTL3 levels, reduced total m6A modification, and effectively mitigated fibrosis, as evidenced by decreased collagen deposition and profibrotic marker expression. Mechanistically, 17-DMAG, a heat shock protein 90 (HSP90) N-terminal inhibitor, induced a heat shock response that sequentially upregulated HSP70 expression. The elevated HSP70 levels inhibited c-Jun N-terminal kinase activity, thereby suppressing the c-Jun transcription factor and ultimately leading to the downregulation of METTL3 expression. MeRIP-Seq analysis revealed that 17-DMAG reversed unilateral ischemia-reperfusion injury–induced m6A epitranscriptomic changes in fibrosis-related genes, including GSK3B, which is involved in fibrotic pathways.

Conclusions

N-terminal HSP90 inhibition, along with subsequent c-Jun suppression, contributed to the mechanism underlying 17-DMAG–induced METTL3 downregulation. Through this regulatory pathway, 17-DMAG effectively suppressed METTL3 expression and attenuated kidney fibrosis in both in vitro and in vivo models.

Introduction

Kidney fibrosis, which manifests as excessive deposition of extracellular matrix proteins in the kidney interstitium, resulting in structural and functional damage to the kidney, is an essential pathology of CKD.^1,2 Consequently, kidney fibrosis is a major target of CKD treatment. Among various epigenetic modification mechanisms, RNA methylation, particularly N6-methyladenosine (m6A) modification—methylation at the sixth nitrogen position of adenosine—has been identified as the most prevalent internal mRNA modification and is associated with various diseases, including kidney diseases.^3,4

m6A modifications are catalyzed by m6A methyltransferases (METTL3/14, Wilms tumor 1-associating protein, RBM15/15B, vir-like methyltransferase complex catalytic subunit, and ZC3H13, termed writers), removed by demethylases (fat mass and obesity-associated protein, ALKBH5, and ALKBH3, termed erasers), and recognized by m6A-binding proteins (YTHDC1/2, YTHDF1/2/3, IGF2BP1/2/3, heterogeneous nuclear ribonucleoprotein, and eIF3, termed readers). Among these proteins, only METTL3 has a binding pocket for S-adenosyl methionine within the methyltransferase domain, allowing it to transfer a methyl group to its target adenosine, whereas other proteins act structurally as scaffolds or facilitate the engagement of the methyltransferase complex on pre-mRNA.⁵ As a result, many researchers have studied METTL3 inhibition as a potential therapeutic strategy for reducing m6A levels.

Recent studies have shown that m6A RNA methylation is associated with fibrosis in various organs, such as the heart, lung, liver, and kidney.^6,7 Therefore, METTL3 represents a promising target for CKD therapies, with studies showing that METTL3 inhibition can reduce the development and progression of kidney fibrosis both in vitro and in vivo.⁷

In our previous study,⁸ we showed that targeting METTL3 attenuates kidney fibrosis by inhibiting METTL3 with siRNA and the METTL3-specific inhibitor STM2457, and we observed elevated METTL3 protein expression in the tubular cell nuclear area of kidney tissue from the patients with CKD. Furthermore, we identified neutrophil extracellular trap 1 (NET1) as a key gene for m6A modification because its mRNA expression was regulated by m6A methylation and its expression modulated epithelial-to-mesenchymal transition (EMT) marker gene expression. In this study, through transcriptome-based drug repositioning, we discovered that the heat shock protein 90 (HSP90) inhibitor 17-dimethylaminoethylamino-17-demethoxygeldanamycin (17-DMAG) downregulates METTL3 expression. Using a dataset of disease samples, we processed data for analysis, identified differentially expressed genes (DEGs) from selected disease and drug RNA profiles, and evaluated drug candidates on the basis of reverse DEG analysis, selecting 17-DMAG as the most effective candidate for preventing kidney fibrosis in in vitro and in vivo models (Supplemental Figure 1). We also explored the mechanism by which 17-DMAG inhibits METTL3 expression.

Methods

This study was approved by the Institutional Animal Care and Use Committee of our institute (2024-0026). For details, see Supplemental Methods.

DEG Analysis and Enrichment Analysis

DEG analysis was performed using the GEO2R tool (National Center for Biotechnology Information) on the GSE66494 dataset from the Gene Expression Omnibus after log2 transformation, background correction, and quantile normalization. The resulting gene signatures were then matched against the Library of Integrated Network-Based Cellular Signatures (LINCS) drug-induced gene expression profiles using rank-based Kolmogorov‒Smirnov statistics,⁹ in which the direction of gene regulation was reversed to identify compounds that could counteract CKD-associated gene expression.

In Vitro Cell Models for the Efficacy of 17-DMAG in Regulating METTL3, Attenuating Epithelial‒Mesenchymal Transition and TGF-β–Induced Fibrosis

Renal cell carcinoma cell lines (ACHN human renal adenocarcinoma cell line [ACHN] and SN12C) were used to evaluate the suppressive effect of 17-DMAG on METTL3. We examined the expression of the METTL3 protein and its relative mRNA expression at different concentrations (0, 0.1, 0.5, 1, 2.5, and 5 μM) and time points (0, 2, 4, 8, 12, 16, and 24 hours).

Furthermore, we established an in vitro TGF-β–induced fibrosis model using human kidney cell line (HK-2). HK-2 cells (1×10⁵ cells/well) were seeded in six-well plates. We examined fibrosis-related EMT and RNA methylation markers at different time points (0, 8, 24, and 48 hours) in the presence of various concentrations of TGF-β (1, 10, and 100 ng/ml). In vitro experiments were performed at least twice to evaluate the statistical significance of the results.

In Vivo Mouse Models of Kidney Fibrosis

In this study, we used two different mouse models of kidney fibrosis: the unilateral ureteral obstruction (UUO) model, which represents obstructive nephropathy, and the unilateral ischemia-reperfusion injury (IRI) model, which represents the AKI-to-CKD transition. For details about sample size, animal preparation, and surgical procedure, see Supplemental Methods and Supplemental Table 1.

Dose Determination and Administration of 17-DMAG

The mice were assigned to two different models according to the procedure (UUO or unilateral IRI). Within each model, the mice were divided into four groups: Sham+Vehicle, Sham+17-DMAG, UUO/unilateral IRI+Vehicle, and UUO/unilateral IRI+17-DMAG. 17-DMAG was administered intraperitoneally at a dosage of 20 mg/kg three times per week through a 30-gauge short-needle insulin syringe.^10,11 Dimethyl sulfoxide was used as the solvent for 17-DMAG and the vehicle.

Histological Examination for Fibrosis Quantification and METTL3 Immunohistochemistry

Paraffin-embedded kidney tissue sections with a thickness of 4 μm were stained with Sirius red stain to assess the degree of tissue fibrosis and for METTL3 immunohistochemistry. The area of fibrosis, METTL3-positive area, and total tissue area were quantified through ImageJ software (version 1.53e; Wayne Rasband, National Institutes of Health, US).⁸ Quantification was performed by an individual masked to sample identity. For primary antibodies, see Supplemental Table 2.

Western Blot and Quantitative Reverse Transcription‒PCR

Protein quantification and mRNA expression were analyzed through Western blot and quantitative reverse transcription‒PCR, respectively.

Measurement of the m6A RNA Methylation Level

The levels of m6A were quantified through the EpiQuik m6A RNA Methylation Quantification Kit (Epigentek, Farmingdale, NY) according to the manufacturer's protocol.¹²

MeRIP-Seq

Methylated RNA immunoprecipitation sequencing (MeRIP-seq) and data analysis were performed as previously described with minor modifications.¹³

Chromatin Immunoprecipitation Assays

To evaluate binding of specific primer DNA and the transcriptional factors, chromatin immunoprecipitation (ChIP) assay was performed.

Ultrasound Examination of Mouse Kidneys

The parenchymal thickness and vascular index were evaluated using a clinical ultrasound equipment (Aplio i800, Canon Medical Systems), with an ultrahigh-frequency linear transducer (i33LX9, center frequency 33 MHz). The Superb Microvascular Imaging (SMI, Canon Medical Systems) was used for measuring vascular index.¹⁴

Statistics

For comparisons of multiple groups, the Kruskal–Wallis test was performed using Conover multiple-comparison post hoc test to determine statistical significance among groups. The unpaired t test was used to evaluate differences between two groups. Data were considered significant at P ≤ 0.05. All error bars represent SD.

Results

Drug Discovery Through Enrichment Analysis of DEGs for CKD

For drug discovery, we used transcriptomics-based bioinformatics to identify drugs that downregulate METTL3 in CKD. We chose the LINCS cloud database for drug analysis and the Gene Expression Omnibus database for disease analysis. We adopted the reverse DEG enrichment method to find optimal drugs for certain diseases. In this method, a high enrichment score indicates high drug potential, which can help patients recover from disease.

For disease-related DEGs, we used GSE66494 dataset and performed DEG analysis derived from two comparisons: normal versus CKD samples and METTL3-low versus METTL3-high CKD samples. The DEGs for normal versus CKD samples and METTL3-low versus METTL3-high CKD samples are shown as a heatmap (Figure 1, A and B). In addition, we stratified CKD samples from the GSE66494 dataset on the basis of the expression of key fibrosis markers, including TGFB1, ACTA2, and COL1A1 to evaluate the role of METTL3 in kidney fibrosis. For each marker, CKD samples were divided into the top 20% (high expression) and bottom 20% (low expression) groups. Notably, METTL3 expression was consistently elevated in the high fibrosis marker groups compared with the low groups (Figure 1, C–E), suggesting a potential link between METTL3 and fibrotic gene expression programs in CKD.

Figure 1.

DEG analysis and enrichment score derivation. (A) Heatmap for DEGs in CKD versus normal kidney samples (GSE66494). (B) Heatmap of DEGs associated with high versus low METTL3 expression in CKD samples (GSE66494). (C–E) Boxplots showing METTL3 expression levels in CKD samples categorized by expression of fibrosis markers: (C) TGFB1, (D) ACTA2, and (E) COL1A1. In all cases, METTL3 is significantly upregulated in the high-expression groups (top 30%) compared with the low-expression groups (bottom 30%). (F–G) Enrichment graphics showing positive connectivity scores between 17-DMAG–induced DEGs and DEGs from (F) CKD versus normal and (G) METTL3-high versus METTL3-low CKD samples. (H) Volcano plot of DEGs upon 17-DMAG treatment from GSE151514. The genes satisfying P value < 0.05 and |logFC|<1 were indicated in blue; genes associated with m6A regulation are labeled. 17-DMAG, 17-dimethylaminoethylamino-17-demethoxygeldanamycin; DEG, differentially expressed gene; FTO, fat mass and obesity-associated protein; HNRNP, heterogeneous nuclear ribonucleoprotein; VIRMA, vir-like methyltransferase complex catalytic subunit; WTAP, Wilms tumor 1-associating protein.

We generated statistically significant DEGs from LINCS drugs and calculated ranked enrichment scores to determine the top 25 candidate drugs with potential to reverse disease-associated gene expression (Table 1). Drug ranking was based on the average enrichment score calculated from two DEG signatures: one derived from normal versus CKD samples and the other from METTL3 low versus high expression groups. To consider adverse effects, we excluded drugs with unknown mechanism of action, conventional chemotherapy drugs, antibiotics, and some cell cycle–related drugs from among the top 25 hit drugs. Our intent was to exclude compounds whose primary mechanism of action involves general cell killing because these are less likely to be suitable for chronic conditions, such as CKD. After validation of some remaining candidate drugs, 17-DMAG was finally determined to be the most potent drug for CKD treatment (Supplemental Figures 2 and 3).

Table 1.

Top 25 candidate chemical drugs with high enrichment score

Rank	Chemical	Mode of Action	Enrichment Score Normal versus CKD	Enrichment Score METTL3 Low versus High	Enrichment Score Average
1	TG-101348	JAK inhibitor|FLT3 inhibitor	1.69	1.53	1.61
2	LY-2603618	CHK inhibitor	1.57	1.64	1.605
3	BRD-K31342827	CDK inhibitor|PKC inhibitor	1.54	1.67	1.605
4	KU-55933	ATM kinase inhibitor	1.67	1.52	1.595
5	Sunitinib	FLT3 inhibitor|KIT inhibitor|PDGFR inhibitor|RET inhibitor|VEGFR inhibitor	1.74	1.44	1.59
6	17-DMAG	HSP inhibitor	1.52	1.66	1.59
7	BRD-K24346336	—	1.53	1.61	1.57
8	Amsacrine	—	1.7	1.43	1.565
9	BRD-K52321331	—	1.43	1.69	1.56
10	Etoposide	Topoisomerase inhibitor	1.83	1.25	1.54
11	L-690488	Inositol monophosphatase inhibitor	1.64	1.44	1.54
12	NSC-95397	CDC inhibitor	1.49	1.54	1.515
13	Palbociclib	CDK inhibitor	1.74	1.28	1.51
14	PAC-1	Caspase activator	1.58	1.43	1.505
15	BRD-K23018071	—	1.71	1.3	1.505
16	Auranofin	NFKB inhibitor	1.41	1.59	1.5
17	BRD-K68552125	PKC activator	1.44	1.54	1.49
18	AT-CSC-18	—	1.35	1.63	1.49
19	BU-239	Imidazoline receptor agonist	1.47	1.49	1.48
20	EX-527	—	1.41	1.55	1.48
21	Dexamethasone	—	1.52	1.44	1.48
22	Camptothecin	—	1.65	1.3	1.475
23	Clofarabine	—	1.61	1.34	1.475
24	Dabrafenib	RAF inhibitor	1.63	1.31	1.47
25	Nutlin-3	MDM inhibitor	1.59	1.34	1.465

17-DMAG, 17-dimethylaminoethylamino-17-demethoxygeldanamycin; ATM, ATM serine/threonine kinase; CDC, cell division cycle; CDK, cyclin-dependent kinase; CHK, checkpoint kinase; FLT3, fms-related receptor tyrosine kinase 3; HSP, heat shock protein; JAK, Janus kinase; KIT, KIT proto-oncogene, receptor tyrosine kinase; MDM, mouse double minute; NFKB, nuclear factor kappa B; PDGFR, platelet-derived growth factor receptor; PKC, protein kinase C; RAF, Raf-1 proto-oncogene, serine/threonine kinase; RET, ret proto-oncogene; VEGFR, vascular endothelial growth factor receptor.

Two sets of DEG signatures showed positive enrichment scores (1.52 and 1.66) when compared against the transcriptional profile induced by the HSP90 inhibitor 17-DMAG, indicating that 17-DMAG may reverse disease-associated gene signatures (Figure 1, F and G). To identify m6A-associated genes regulated by 17-DMAG, we analyzed GSE151514, a dataset profiling transcriptomic changes upon 17-DMAG treatment. DEG analysis revealed only METTL3, IGFBP1, and IGFBP3 seemed to be regulated by 17-DMAG (Figure 1H).

Mechanism by Which 17-DMAG Downregulates METTL3 in Renal Cell Carcinoma

To validate 17-DMAG as a regulator of METTL3 expression, we treated renal cell carcinoma cell lines (ACHN and SN12C) with increasing doses of 17-DMAG and observed a dose-dependent decrease in METTL3 protein levels (Figure 2A). HSP70 is known to be upregulated when HSP90 is inhibited.¹⁵

Figure 2.

17-DMAG decreases METTL3 expression in renal carcinoma cells. (A) Western blot analysis of METTL3 and HSP70 expression in ACHN and SN12C cells treated with different concentrations of 17-DMAG. (B) qRT‒PCR results showing METTL3 mRNA levels in ACHN cells treated with various concentrations of 17-DMAG. (C) qRT‒PCR results showing time-dependent METTL3 mRNA expression in ACHN cells treated with 17-DMAG. (D) Venn diagram showing the number of transcription factors predicted from the transcription factor prediction analysis and references. For each transcription factor, the putative binding site count and reference source are shown in the table. (E) qRT‒PCR results showing JUN, ATF2, SMAD2, SMAD3, TBP, and YY1 mRNA levels after 17-DMAG treatment. (F) ChIP assay confirming that c-JUN binds to the METTL3 promoter. P < 0.05. Mann‒Whitney U test with post hoc Conover test. (G) Western blot for METTL3, HSP70, and c-JUN expression in ACHN cells treated with various HSP90 inhibitors. N.C means normal control. (H) Western blot analysis of METTL3, HSP70, c-JUN, p-JNK, and JNK expression in ACHN cells treated with 10 μM of 17-DMAG, and KNK437 for 24 hours. (I) Illustration of the inhibition of METTL3 expression by 17-DMAG occurring through a series of molecular interactions involving HSP90, HSF1, HSP70, JNK, and c-Jun. Red circles mean phosphorylation. ACHN, ACHN human renal adenocarcinoma cell line; ChIP, chromatin immunoprecipitation; DMSO, dimethyl sulfoxide; HSP90, heat shock protein 90; JNK, Jun N-terminal kinase; JUN, Jun proto-oncogene; qRT‒PCR, quantitative reverse transcription‒PCR; TBP, TATA-box binding protein; TF, transcription factors.

In addition, 17-DMAG also suppressed METTL3 expression transcriptionally in a dose-dependent manner and in a time-dependent manner, exhibiting fluctuation at 12 and 16 hours (Figure 2, B and C). To exclude the possibility of proteasomal degradation, we co-treated cells with the proteasome inhibitor MG132. The lack of METTL3 restoration under MG132 treatment indicated that 17-DMAG likely acts through transcriptional suppression rather than degradation (Supplemental Figure 4).

To reveal the specific mechanism of action of 17-DMAG, we examined transcription factors that bind to the METTL3 promoter region (−2000 approximately +100). The analysis was performed with PROMO v3.0.2 software, which incorporates TRANSFAC v8.3.^16,17 We also investigated several transcription factors that have been suggested to bind to the METTL3 promoter by other groups.^12,18–28 We present a Venn diagram and a table depicting the transcription factors that satisfied both the analysis and reference criteria (Figure 2D). We confirmed the transcription of several transcription factors under 17-DMAG treatment. AP-1 is a transcription factor protein composed of c-Jun and its binding partner genes. Jun proto-oncogene (JUN) was significantly downregulated by 17-DMAG in ACHN cells. The level of ATF2 was increased and SMAD3 was also slightly decreased by 17-DMAG treatment (Figure 2E). Although the binding site of c-Jun in the METTL3 promoter has already been revealed previously,¹⁹ we performed reconstruction to determine whether Jun functions as a transcription factor in renal cancer cells through a ChIP assay (Figure 2F). The binding site prediction revealed seven potential binding sites for c-Jun, but the illustration depicts only four sites (BS1, BS2, BS3, and BS4), and five sets of primer regions were designed (PR1, PR2, PR3, PR4, and PR5). The ChIP assay results revealed a significant binding signal for only PR1 and PR2, involving BS1, BS2, and BS3 as binding sites for JUN gene. In addition, only PR1 and PR2 presented a suppressed signal upon 17-DMAG treatment. The Western blot analysis confirmed that overexpression of c-Jun led to an increase in METTL3 expression (Supplemental Figure 5B). These data led us to conclude that the suppression of c-Jun expression may be a mechanism of METTL3 suppression by 17-DMAG.

To determine whether this effect was specific to 17-DMAG or common to other HSP90 inhibitors, ACHN cells and HK-2 cells were treated with N-terminal HSP90 inhibitors, such as 17-AAG, luminespib, HSP-990, and a C-terminal HSP90 inhibitor NCT-58 (Figure 2G and Supplemental Figure 6B). The N-terminal inhibitors decreased both METTL3 and c-Jun levels while increasing HSP70 expression. But C-terminal inhibitor NCT-58 did not reproduce these effects, indicating the drug effect specificity of the N-terminal HSP90 inhibition (Figure 2G). To further evaluate the relationship between induced heat shock proteins and METTL3 expression, we treated ACHN cells and HK-2 cells with 17-DMAG and pharmacological modulators of heat shock proteins (Figure 2H and Supplemental Figure 6A). Co-treatment with KNK-437, a pan-HSP inhibitor, partially rescued the expression of both METTL3 and c-Jun in 17-DMAG–treated cells, highlighting the involvement of heat shock response components in METTL3 regulation. In addition, overexpression of HSPA1B in HEK293 cell showed restored protein levels of c-JUN and METTL3 reduced by 17-DMAG (Supplemental Figure 5A). At the mRNA level, HSPA1B knockdown attenuated 17-DMAG–induced downregulation of METTL3 mRNA, suggesting an inverse relationship between HSP70 and METTL3 expression (Supplemental Figure 7). These findings collectively suggested that 17-DMAG inhibited METTL3 expression through the induction of a heat shock response through N-terminal inhibition of HSP90, which initiated a cascade of molecular interactions involving HSP90, HSP70, and c-Jun, potentially mediating METTL3 downregulation through c-Jun–dependent pathways (Figure 2I).

Effects of 17-DMAG on METTL3 Expression and m6A Modification In Vitro

We investigated the effects of 17-DMAG on METTL3 expression in human kidney tubular epithelial HK-2 cells, a model for TGF-β–induced fibrosis. Treatment with increasing concentrations of 17-DMAG resulted in a dose-dependent reduction in METTL3 mRNA and protein levels, although the total m6A remained largely unchanged after 17-DMAG treatment (Figure 3, A–C). Our previous study⁸ revealed that METTL3 inhibition attenuates EMT by decreasing the stability of NET1 mRNA; thus, we evaluated NET1 expression according to 17-DMAG treatment. Relative NET1 mRNA expression was decreased after treatment with 1 μM 17-DMAG (Figure 3D). Accordingly, the MeRIP-qPCR results revealed that the m6A modification levels of the NET1 transcript were significantly reduced under 1 μM 17-DMAG treatment (Figure 3E), suggesting that 17-DMAG–mediated downregulation of METTL3 decreased the m6A levels of NET1 mRNA. In addition, mRNA stability assays using actinomycin D showed decreased NET1 mRNA stability with 17-DMAG (Figure 3F). In TGF-β–treated HK-2 cells, 17-DMAG suppressed the upregulation of EMT markers, including N-cadherin and vimentin, and reduced NET1 mRNA expression (Figure 3G). Western blot analysis revealed that 17-DMAG reduced METTL3 protein levels in both TGF-β–treated and untreated HK-2 cells, accompanied by decreased expression of mesenchymal markers, including N-cadherin, vimentin, and fibronectin, under the TGF-β–treated condition (Figure 3H). Under these conditions, however, the effect of METTL3 knockdown on the relative magnitude of 17-DMAG–mediated reductions in TGF-β–induced mesenchymal markers varied by gene. For example, although METTL3 knockdown significantly reduced TGF-β–induced N-cadherin expression (P = 0.047), additional 17-DMAG treatment resulted in little statistical difference in N-cadherin expression between control and siMETTL3-treated cells (P = 0.11), suggesting a central contribution of METTL3 to the N-cadherin response. By contrast, METTL3 knockdown significantly reduced vimentin and alpha-smooth muscle actin (α-SMA) expression under TGF-β treatment (P = 0.039 for vimentin; P = 0.03 for α-SMA), but this difference remained evident even after 17-DMAG treatment (P = 0.013 for vimentin; P = 0.013 for α-SMA; Supplemental Figure 8). These data suggest that anti-EMT and antifibrotic effects of 17-DMAG are, at least in part, mediated by METTL3 regulation. The total m6A levels did not significantly differ regardless of 17-DMAG application (Figure 3I). For evaluating the effect of 17-DMAG to the METTL3–METTL14 complex, we performed immunoprecipitation of the METTL3–METTL14 complex using an anti-METTL14 antibody after 17-DMAG treatment in HK-2 cells. The binding of METTL3 to METTL14 was modestly reduced after 17-DMAG treatment as shown below, probably because of a decrease in total METTL3 protein levels by the treatment (Figure 3J).

Figure 3.

17-DMAG attenuates METTL3 and EMT in vitro. (A) qRT‒PCR results showing METTL3 mRNA levels after 17-DMAG treatment. (B) Western blot analysis of METTL3 expression in HK-2 cells treated with various concentrations of 17-DMAG. (C) m6A RNA methylation levels in HK-2 cells after 17-DMAG treatment. (D and E) NET1 mRNA and m6A modification levels decreased after 17-DMAG treatment. (F) NET1 mRNA stability was reduced in the presence of actinomycin D and 17-DMAG. (G and H) Western blot showing the effects of METTL3 silencing and 17-DMAG treatment on HSP70, EMT markers, and fibrosis markers in the TGF-β–induced fibrosis model using HK-2 cell. (I) m6A modification levels in TGF-β–treated and untreated HK-2 cells after 17-DMAG treatment. (J) Coimmunoprecipitation of METTL3 with METTL14 in HK2 cells after 17-DMAG treatment. METTL3 binding to METTL14 was slightly reduced after treatment. Mann‒Whitney U test with post hoc Conover test was used for statistical analysis. α-SMA, alpha-smooth muscle actin; EMT, epithelial-to-mesenchymal transition; HK-2, human kidney cell line; m6A, N6-methyladenosine.

17-DMAG Reduced METTL3 Expression In Vivo

The effect of 17-DMAG on METTL3 was examined in two in vivo mouse models of kidney fibrosis as mentioned above. The periodic acid–Schiff staining, immunohistochemical staining, and immunofluorescence staining in kidney tissues of UUO mouse presented predominant distribution of METTL3 in proximal tubule cells (Supplemental Figure 9). Immunohistochemical staining of mouse kidney tissue revealed that METTL3 was significantly overexpressed in the injury group compared with the normal group, and 17-DMAG treatment significantly decreased the expression level of METTL3 compared with that in the nontreated subgroup in the injury group, regardless of the injury mechanism (Figure 4, A–D). Western blot analysis revealed that the METTL3 protein level was increased by injury, similar to the results of immunohistochemical staining, and 17-DMAG treatment after injury significantly attenuated METTL3 expression (Figure 4, E–G, and Supplemental Figure 10). The relative expression of METTL3 mRNA was diminished by 17-DMAG treatment in the UUO mouse group, although the difference was not statistically significant (Supplemental Figure 11). In addition, treatment with 17-DMAG decreased the m6A level in the injury group, in both the UUO and unilateral IRI mouse models (Figure 4, H and I). These findings suggested that 17-DMAG treatment attenuated METTL3 expression during kidney injury and consequently decreased the total m6A level.

Figure 4.

17-DMAG effectively inhibits METTL3 in vivo. (A–D) Immunohistochemical staining of METTL3 in UUO-induced and unilateral IRI–induced kidney fibrosis models with and without 17-DMAG treatment. (E–G) Western blot analysis of METTL3 and other m6A-related proteins in fibrosis models. (H and I) m6A modification levels in UUO and unilateral IRI model mice after 17-DMAG treatment. Mann‒Whitney U test with post hoc Conover test was used for statistical analysis. IHC, immunohistochemistry; IRI, ischemia-reperfusion injury; UUO, unilateral ureteral obstruction.

17-DMAG Suppressed Kidney Fibrosis In Vivo

Next, we evaluated the effect of 17-DMAG on the degree of kidney fibrosis. Sirius red staining of kidney tissues and subsequent quantification revealed significantly greater fibrosis in the untreated group than in the 17-DMAG–treated group (Figure 5, A and B). Western blot analysis showed that 17-DMAG significantly downregulated molecular fibrosis markers, including α-SMA and Col1A1 expression in the UUO model. A comparable but statistically nonsignificant reduction of fibrosis markers was detected in the unilateral IRI model in the Kruskal–Wallis test. However, there was a significant decrease of α-SMA and TGF-β according to 17-DMAG treatment in unilateral IRI mice (Figure 5, C and D).

Figure 5.

17-DMAG attenuates kidney fibrosis in vivo. (A and B) Sirius red staining and quantification of fibrosis in UUO and unilateral IRI models. (C and D) Western blot analysis of fibrosis and EMT markers in mouse kidney tissues. (E and F) Western blot and qRT‒PCR analysis of NET1 expression in UUO model mice. (G and H) B-mode and microvascular ultrasound images showing kidney parenchymal thickness and the vascular index in fibrosis models. (I–L) Quantification of kidney parenchymal thickness and the vascular index in mouse models. Mann‒Whitney U test with post hoc Conover test was used for statistical analysis.

While the levels of EMT markers such as vimentin and N-cadherin were slightly decreased after 17-DMAG treatment, their levels were less affected in the UUO mouse model (Supplemental Figures 12 and 13). Similar to the in vitro results, NET1 protein (Figure 5, E and F) and mRNA (Supplemental Figure 11) expression was decreased after 17-DMAG treatment in the UUO mouse model. In both the UUO and unilateral IRI mouse models, 17-DMAG treatment significantly preserved the kidney parenchyma and improved the vascular index compared with those in the control groups (Figure 5, G–L). These findings suggest that 17-DMAG treatment mitigates kidney fibrosis by preserving kidney structure and blood flow.

17-DMAG Modulated m6A Modification in Unilateral IRI–Induced Fibrosis Models

We investigated the effects of 17-DMAG on m6A modification in unilateral IRI–induced kidney fibrosis models through MeRIP-seq. The distribution of m6A sites was enriched near stop codons and in 3′ UTRs (Figure 6A). Motif analysis revealed enrichment of the D = A, G, or U, R = A or G, A = m6A, C = C, H = A, C, or U-like motif in differentially methylated peaks, which is consistent with previously reported m6A profiles (Figure 6B).

Figure 6.

m6A RNA methylation profile in in vivo kidney fibrosis models. (A) Distribution of m6A modification peaks in kidney samples. (B) DRACH motif enrichment in differentially methylated peaks. (C) Increased m6A peaks in the unilateral IRI group compared with the sham group, indicating fibrosis-related gene set enrichment. Conducted hallmark gene set analysis using 386 genes corresponding to the 504 peaks increased in the unilateral IRI models, we found that fibrosis-related gene sets, including “EPITHELIAL_MESENCHYMAL_TRANSITION” and “KRAS_SIGNALING_UP” (PMID: 31570767), as well as inflammation-related gene sets, such as “IL2_STAT5_SIGNALING,” “IFN_GAMMA_RESPONSE,” and “IFN_ALPHA_RESPONSE,” were enriched (FDR Q <0.05). (D) Reduced m6A peak intensity after 17-DMAG treatment. Hallmark gene set analysis using 386 genes corresponding to the 504 peaks increased in the unilateral IRI models proved that several fibrosis-related gene sets, including “EPITHELIAL_MESENCHYMAL_TRANSITION” and “KRAS_SIGNALING_UP” (PMID: 31570767), as well as inflammation-related gene sets, such as “IL2_STAT5_SIGNALING,” “IFN_GAMMA_RESPONSE,” and “IFN_ALPHA_RESPONSE,” were enriched (FDR Q <0.05). (E) Venn diagram showing the number of commonly presented genes between untreated unilateral IRI group with increased m6A peaks and 17-DMAG–treated unilateral IRI group with decreased m6A peaks. (F) m6A modification and RNA expression levels of the GSK3B gene in unilateral IRI and/or 17-DMAG–treated groups. Relative m6A enrichment was assessed by MeRIP-seq (upper panel), and RNA expression levels were quantified by RNA sequencing of input RNA, represented as fragments per kilobase of transcript per million mapped reads (FPKM; lower panel). (G) Integrative Genomics Viewer tracks of MeRIP-seq data at the GSK3B locus. The y axis indicates sequencing read counts, and the box marks the predicted m6A-modified regions. DMP, differentially methylated peaks; DRACH, D = A, G, or U, R = A or G, A = m6A, C = C, H = A, C, or U; FDR Q, false discovery rate (FDR)-adjusted P value (Q value); FPKM, fragments per kilobase of transcript per million mapped reads; KRAS, Kirsten rat sarcoma virus oncogene homolog; MeRIP-seq, methylated RNA immunoprecipitation sequencing.

Comparison between the Sham+Vehicle and unilateral IRI+Vehicle groups revealed a substantial increase in m6A peaks in the unilateral IRI model (504 versus 183 peaks, log2FC ≥1 or ≤ −1, P < 0.05; Figure 6C), indicating elevated m6A methylation after injury. Hallmark gene set analysis of the 386 genes linked to increased peaks showed enrichment in fibrosis-related and inflammation-related pathways, including epithelial–mesenchymal transition, Kirsten rat sarcoma virus oncogene homolog signaling, and IFN responses (false discovery rate [FDR]-adjusted P value [Q value] <0.05; Figure 6C).

By contrast, 17-DMAG treatment in the unilateral IRI model led to more decreased m6A peaks (296 versus 138 peaks; Figure 6D), a reduction in m6A levels. Analysis of 233 genes associated with decreased peaks in the 17-DMAG group showed significant depletion in several fibrosis-related and inflammation-related gene sets previously enriched in the unilateral IRI model (FDR Q value <0.05; Figure 6D).

Among the 386 genes with increased m6A peaks in the unilateral IRI group, 136 genes (35.2%) presented decreased m6A levels in the 17-DMAG–treated unilateral IRI group (Figure 6E). The full list of these 136 genes, which were differentially methylated in response to unilateral IRI kidney fibrosis and subsequently modulated by 17-DMAG, is provided in Supplemental Table 3. One such gene, GSK3B, which is known to be involved in kidney fibrosis,^29,30 exhibited increased m6A and mRNA expression in response to unilateral IRI, which was reversed by 17-DMAG (Figure 6, F and G). These findings indicate that 17-DMAG modulates m6A modification profiles and mitigates unilateral IRI–induced fibrosis.

Discussion

In a previous study, we demonstrated the association between RNA methylation and kidney fibrosis, confirming that inhibition of the m6A RNA methyltransferase METTL3 attenuates kidney fibrosis both in vitro and in vivo.⁸ This was achieved through the use of STM2457, a selective catalytic inhibitor of METTL3 that binds to its S-adenosyl methionine–binding domain.²⁸ Building on these findings, this study aimed to identify novel METTL3 regulators through drug repositioning, elucidate their mechanisms of action, and assess their efficacy in inhibiting kidney fibrosis both in vitro and in vivo.

Our results identify 17-DMAG as a novel METTL3 inhibitor using transcriptomics-based bioinformatics through DEG analysis. The average enrichment score of candidate drugs between drug-related DEGs and disease-related DEGs (CKD and METTL3 high expression conditions) was the criterion for selecting the final candidates. Another candidate drug, TG-101348, had a greater average enrichment score than did 17-DMAG, but the RT‒PCR results revealed that the METTL3 mRNA inhibition effect of 17-DMAG was greater than that of the other candidate drugs (Supplemental Figures 2 and 3). Previous research has shown that the robustness of DEG analysis can be compromised when additional random drugs are included in the drug set,³¹ but validation experiments in this study confirmed 17-DMAG as the optimal candidate (Supplemental Figures 2 and 3).

We hypothesized that the HSP90 inhibitor 17-DMAG would inhibit METTL3 expression, which was confirmed in renal cell carcinoma cell lines (SN12C,ACHN) as well as the UUO and unilateral IRI models. Our data indicate that only N-terminal HSP90 inhibition can cause c-Jun and METTL3 downregulation. The difference between N-terminal HSP90 inhibition and C-terminal HSP90 inhibition is whether heat shock response is induced.¹⁵ N-terminal HSP90 inhibition activates HSF1, which undergoes trimerization and phosphorylation, then translocates to the nucleus and induces the expression of heat shock proteins such as HSP70, whereas C-terminal inhibition does not elicit this response. Furthermore, upregulation of HSP70 suppresses c-Jun N-terminal kinase activity, leading to reduced phosphorylation of c-Jun, which in turn, results in decreased c-Jun expression.³² Taken together, these findings suggest that HSP90 inhibition leads to HSP70 activation, which reduces c-Jun expression at the transcriptional level, and that decreased c-Jun expression sequentially causes METTL3 downregulation. This consecutive signaling pathway may be a mechanism for METTL3 attenuation by 17-DMAG (Figure 2I).

While c-Jun and HSP90 are both well-established regulators of gene expression, our study specifically focused on the interplay between HSP90 inhibition and the epitranscriptomic regulation of kidney fibrosis through METTL3. We acknowledge that 17-DMAG, a HSP90 inhibitor, may exert broader transcriptional effects, potentially altering the expression of other c-Jun downstream targets. However, our data support a mechanistic model in which 17-DMAG attenuated c-Jun–mediated transcriptional activation of METTL3 by inhibiting HSP90. This selective effect on METTL3 appears to be a critical node linking transcriptional change with m6A RNA methylation in the context of fibrotic signaling.

Fibroblasts are widely recognized as the principal effectors of fibrosis. Considering that upregulation of METTL3 and global m6A methylation are observed in fibrotic kidney tissue and epithelial cells represent the predominant population in the kidney tissue, it is plausible that the upregulation of METTL3 and m6A within epithelial cells contributes to the fibrotic process. Moreover, it is well established that epithelial cells can influence fibroblasts through paracrine signaling.^33,34 Therefore, we hypothesized that targeting METTL3 in epithelial cells would modulate the expression of paracrine-related genes, which may play a critical role in regulating fibroblast function during fibrosis.

We acknowledge that a comprehensive understanding of kidney fibrosis requires elucidation of how epithelial cell changes influence fibroblast activation and matrix deposition. We highlighted the potential genes for epithelial-derived secretion proteins and transmembrane proteins of fibroblast based on DEGs affected by 17-DMAG treatment (Supplemental Figure 14 and Supplemental Tables 4 and 5). These data offer potential targets for exploring epithelial–fibroblast cross-talk and may serve as a resource for further mechanistic studies. Furthermore, 17-DMAG treatment resulted in a significant reduction in METTL3, EMT marker (NCad), and fibrosis marker (Col1A1) (Supplemental Figure 15). In our previous study,⁸ we demonstrated that treatment with STM2457, a S-adenosyl methionine–binding site inhibitor of METTL3, reduced cellular m6A levels and suppressed EMT marker expression in a TGF-β–induced fibrosis model using NRK49F rat fibroblasts. Therefore, 17-DMAG could attenuate fibrosis by inhibiting METTL3 in fibroblasts either indirectly through epithelial cells or through a direct effect to fibroblast m6A methylation.

Previous studies have investigated the role of HSP90 in the regulation of kidney fibrosis. HSP90 reportedly exacerbates kidney fibrosis progression by stabilizing TGF-β receptor 2 (TGF-β R2) to promote TGF-β–mediated EMT.^35,36 HSP70, which is upregulated under conditions of HSP90 inhibition, as mentioned above, has been reported to attenuate kidney fibrosis by inhibiting the TGF-β signaling pathway and eventually blocking EMT and the production of N-cadherin, vimentin, and α-SMA.^37–39 Consequently, previous studies have shown that the HSP90 inhibitor 17-AAG mitigates kidney fibrosis.^35,40 On the other hand, we investigated whether 17-DMAG regulates the NET1 gene and inhibits kidney fibrosis through METTL3 inhibition, as shown in our previous study.⁸ Our results revealed that METTL3 levels and total m6A levels were significantly attenuated both in an in vitro TGF-β–treated HK-2 cell model and in an in vivo mouse model. NET1 mRNA and protein expression were also reduced in the 17-DMAG–treated groups, both in vivo and in vitro, although there were fluctuations in EMT marker levels in vivo (UUO mouse group). Histological fibrosis was effectively mitigated in the 17-DMAG–treated groups regardless of the mechanism of injury. These results highlight that the HSP90 inhibitor 17-DMAG may attenuate kidney fibrosis by inhibiting METTL3 to regulate NET1, in addition to its previously known effect of blocking TGF-β–mediated EMT by directly inhibiting HSP90.

Our results further highlight the epitranscriptomic changes induced by 17-DMAG, particularly in the GSK3B gene, which was identified as a key target through MeRIP-seq in unilateral IRI–induced kidney fibrosis models. GSK3B, a serine/threonine protein kinase involved in multiple fibrosis-related pathways,^29,41 exhibited increased m6A modification in unilateral IRI models, which was mitigated by 17-DMAG treatment. These findings suggest that the antifibrotic effects of 17-DMAG are partially mediated by the modulation of m6A modification on key fibrosis-related genes, such as GSK3B.

Previous studies have reported that m6A modification of c-Myc mRNA is associated with polycystic kidney disease,⁴² and that genes involved in the cyclic GMP-AMP synthase–stimulator of IFN genes (cGAS–STING) pathway undergo m6A modification in kidney fibrosis.⁴³ However, in our MeRIP-seq data from unilateral IRI–induced kidney fibrosis, we did not detect m6A peaks in these genes. This discrepancy is probably attributed to differences in model systems, sample types, and the sensitivity of the MeRIP-seq technique, which can be substantially influenced by the quality of the m6A antibody and the efficiency of the immunoprecipitation. Therefore, to comprehensively assess global m6A modifications, more sensitive, quantitative, and base-resolution approaches, such as gold-standard m6A detection with raw-read information⁴⁴ or site-specific cleavage and radioactive-labeling followed by sequencing,⁴⁵ need to be considered.

In addition to these molecular findings, our ultrasound results revealed that 17-DMAG treatment preserved intrarenal blood flow in both UUO and unilateral IRI model mice, suggesting that 17-DMAG not only mitigates fibrosis but also maintains kidney function under injury conditions. Future studies should further explore the relationships among HSP90 inhibition, METTL3 downregulation, and the preservation of kidney function in CKD.

Although our data support the notion that METTL3 downregulation represents one plausible mechanistic axis by which 17-DMAG attenuates fibrogenic responses, the heterogeneous responses observed among individual EMT-related genes (Supplemental Figure 8) suggest that METTL3 suppression alone cannot fully explain the antifibrotic effect of 17-DMAG. This may be further influenced by the use of a knockdown, rather than a complete knockout, system in which residual METTL3 function remains. Therefore, while METTL3 downregulation likely constitutes an important component of 17-DMAG–mediated inhibition of fibrosis, we acknowledge that 17-DMAG may exert broader regulatory effects beyond the METTL3 axis. Future studies using METTL3-null models and unbiased pathway-level analyses will be required to delineate the relative contributions of METTL3-dependent and METTL3-independent mechanisms to the antifibrotic actions of 17-DMAG.

In conclusion, our study demonstrated that transcriptomic bioinformatics, particularly DEG-based enrichment analysis, can be used to successfully identify novel METTL3 inhibitors. 17-DMAG, a drug originally recognized for its ability to inhibit HSP90, effectively downregulated METTL3 expression at the transcriptional level in human renal carcinoma cells and mouse models of kidney injury. This mechanism involves the suppression of transcription factors, such as c-Jun, as well as the regulation of NET1 expression. Furthermore, 17-DMAG effectively attenuated fibrosis markers, reduced m6A modification, and preserved intrarenal blood flow. These findings suggest that 17-DMAG and other METTL3 inhibitors identified through drug repositioning could be promising therapeutic agents for the treatment of kidney fibrosis. To date, no HSP90 inhibitors have been approved for clinical use, largely because of safety concerns from off-target effects observed during clinical trials. Careful evaluation of reported and potential adverse effects, such as ocular, hepatic, and marrow toxicities,^46–48 are essential when considering their clinical application for kidney fibrosis.

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/JSN/F615.

Author Contributions

Conceptualization: Sung-Yup Cho, Jin Woo Choi, Jung Pyo Lee, Myoung Seok Lee, Soo Min Lee.

Data curation: Bogyeong Cho, Sung-Yup Cho, Da Som Choi, Jin Woo Choi, Seung Hyun Han, Wencheng Jin, Hae Rim Jung, Jeonghwan Lee, Jung Pyo Lee, Myoung Seok Lee, Soo Min Lee, Nayeon Shin.

Formal analysis: Seung-Pyo Hong, Hae Rim Jung, Jong-Il Kim, Bada Lee, Myoung Seok Lee, Soo Min Lee.

Funding acquisition: Jung Pyo Lee.

Investigation: Bogyeong Cho, Sung-Yup Cho, Da Som Choi, Jin Woo Choi, Seung-Pyo Hong, Wencheng Jin, Jong-Il Kim, Bada Lee, Jeonghwan Lee, Jung Pyo Lee, Myoung Seok Lee, Soo Min Lee, Nayeon Shin.

Methodology: Sung-Yup Cho, Da Som Choi, Jin Woo Choi, Seung-Pyo Hong, Hae Rim Jung, Jong-Il Kim, Jung Pyo Lee, Myoung Seok Lee, Soo Min Lee, Chun Soo Lim, Min Hoan Moon.

Project administration: Sung-Yup Cho, Da Som Choi, Jin Woo Choi, Jung Pyo Lee, Chun Soo Lim, Min Hoan Moon, Nayeon Shin.

Resources: Bogyeong Cho, Sung-Yup Cho, Jin Woo Choi, Seung Hyun Han, Wencheng Jin, Jeonghwan Lee, Jung Pyo Lee, Myoung Seok Lee, Nayeon Shin.

Software: Jin Woo Choi, Jung Pyo Lee, Soo Min Lee.

Supervision: Sung-Yup Cho, Jin Woo Choi, Jung Pyo Lee, Chun Soo Lim, Min Hoan Moon.

Validation: Sung-Yup Cho, Da Som Choi, Jin Woo Choi, Hae Rim Jung, Jung Pyo Lee, Myoung Seok Lee, Soo Min Lee, Nayeon Shin.

Visualization: Sung-Yup Cho, Jin Woo Choi, Hae Rim Jung, Myoung Seok Lee, Nayeon Shin.

Writing – original draft: Jin Woo Choi, Myoung Seok Lee.

Writing – review & editing: Sung-Yup Cho, Hae Rim Jung, Jung Pyo Lee, Myoung Seok Lee, Soo Min Lee.

Funding

This work was supported by the Korean Fund for Regenerative Medicine (KFRM) grant funded by the Korea government (the Ministry of Science and ICT, the Ministry of Health & Welfare; 23B0104L1) and the National Research Foundation of Korea (NRF) grant (No. 2021R1A2C2011292) funded by the Korea government (Ministry of Science and ICT, MSIT).

Declarative Statements

All animal experiments were performed according to protocols approved by the Seoul National University Boramae Medical Center Institutional Animal Care and Use Committee. This study was approved by the Institutional Animal Care and Use Committee of our institute (2024-0026). Animal experiments were carried out in the animal laboratory of our institute under specific pathogen-free conditions in accordance with the National Research Council's Guidelines for the Care and Use of Laboratory Animals.

Data Availability Statements

Original data generated for the study will be made available upon reasonable request to the corresponding author. Data Type: Image Data; Raw Data/Source Data. Reason for Restricted Access: We have adopted the policy of making the original data available upon reasonable request to the corresponding author. This approach ensures that data sharing is conducted in a responsible manner, allowing us to verify the scientific purpose of each request and to provide appropriate guidance regarding the context and use of the data. Such a process helps maintain the integrity of the dataset, prevents potential misinterpretation, and ensures that the data are used solely for legitimate academic and research purposes.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/JSN/F616, http://links.lww.com/JSN/F706.

Supplemental Methods

Supplemental Figure 1. Overall study workflow.

Supplemental Figure 2. Gel electrophoresis band from RT‒PCR of METTL3 mRNA under (+)-Nutlin-3, PMA, and KU-55933 treatment in renal cell carcinoma cell ACHN.

Supplemental Figure 3. Gel electrophoresis bands from RT‒PCR and semiquantitative qPCR results of METTL3 mRNA under TG-101348 and 17-DMAG treatment at different time points (12 and 24 hours) and concentrates (1 and 10 μM) in renal cell carcinoma cell ACHN.

Supplemental Figure 4. Inhibition of HSP90 led to the ubiquitination and subsequent proteasomal degradation of several client proteins, and with a proteasome inhibitor MG132 prevents proteasomal degradation.

Supplemental Figure 5. HSPA1B and c-JUN overexpression rescued 17-DMAG–induced METTL3 downregulation.

Supplemental Figure 6. 17-DMAG showed METTL3 downregulation effect in normal renal epithelial HK-2 cell.

Supplemental Figure 7. HSP70 knockdown attenuates 17-DMAG–induced downregulation of METTL3 mRNA.

Supplemental Figure 8. Silencing of METTL3 attenuates the upregulation of TGF-β–induced mesenchymal and fibrosis markers and the inhibitory effect of 17-DMAG on mesenchymal markers.

Supplemental Figure 9. (A) In serial sections of UUO kidney tissue, periodic acid–Schiff staining highlighted proximal tubules with abundant cytoplasm and well-defined brush borders, and these regions corresponded to strong METTL3 immunoreactivity on the immunohistochemical staining of adjacent section. (B) METTL3 (green) co-staining with cell type–specific marker CD13 (red) showed co-expression of METTL3 in a kidney proximal tubule cell.

Supplemental Figure 10. Western blot results of m6A methyltransferases in UUO and uIRI mouse models not shown in the main Figure 4E.

Supplemental Figure 11. Real-time RT‒PCR result of relative expressions of vimentin, α-SMA, METTL3, and NET1 mRNA in UUO mouse model.

Supplemental Figure 12. Western blot result of M6A methyltransferases, demethylases, m6A-binding proteins, EMT markers, and fibrosis marker proteins in UUO mouse model.

Supplemental Figure 13. Semiquantification result of METTL3, EMT markers, and fibrosis markers in the UUO mouse model.

Supplemental Figure 14. (A) Dot plot of kidney fibrosis–related secreted proteins DEG by 17-DMAG treatment (GSE151514).

Supplemental Figure 15. Western blot result of METTL3, EMT markers, and fibrosis marker proteins in in vivo fibrosis model using rat fibroblast (NRK49F) treated with TGFβ 5 ng/ml with or without 17-DMAG ten approximately 50 nM/ml.

Supplemental Table 1. Initially included number of animals.

Supplemental Table 2. Primer sequences.

Supplemental Table 3. List of 136 genes with altered m6A methylation in kidney fibrosis and reversal by 17-DMAG treatment.

Supplemental Table 4. Kidney fibrosis–related secreted proteins among DEGs in 17-DMAG–treated cells (GSE151514).

Supplemental Table 5. Kidney fibrosis–related transmembrane proteins among DEGs in 17-DMAG–treated cells (GSE151514).

Supporting Data Values file. Values for all data points in graphs.