C57BL/6 mice were purchased from SLAC Laboratory Animal Co., Ltd. Atg7^flox/flox mice were gifts from Dr. Jianrong Wang, Medical College of Soochow University (20). SQSTM1^flox/flox mice were gifts from Dr. Hongting Zheng, Xinqiao Hospital, Army Medical University (21). All mice were housed with a standard 12-h light/dark cycle, climate-controlled, and pathogen-free facility. All experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals, and the study protocols were approved by the Laboratory Animal Welfare and Ethics Committee of Chongqing University. To generate tubular-specific Atg7-deficient and SQSTM1-deficient mice, Atg7^flox/flox and SQSTM1^flox/flox mice were crossed with γGTcre mice (Cyagen Biosciences). Gender-matched wild-type and specific knockout mice (6–8 weeks old) from the same litter were selected randomly to indicated groups. Two animal studies were performed as follows: a) mice were divided into four groups (n = 8 per group): (i) Ctrl with sham operation, (ii) canagliflozin (Cana) (Selleck S2760; 20 mg/kg in PBS, daily intragastric administration till the mice were harvested), (iii) UUO (sacrificed 7 days after the performance), and (iv) Cana + UUO, and b) mice were divided into four groups (n = 8 per group): (i) Ctrl with sham operation, (ii) canagliflozin (Cana) (Selleck S2760; 20 mg/kg in PBS, daily intragastric administration for consecutive 21 days), (iii) I/R (sacrificed 21 days after the performance), and (iv) Cana + I/R. I/R and UUO operation was performed according to our previous study (5).

Human proximal renal tubular cell line HK2 was purchased from the Cell Bank of the Chinese Academy of Sciences. HK2 cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM, Corning) containing 10% FBS (HyClone) in a 5% CO₂ incubator at 37°C. TGF-β1 (5 ng/ml) was added to the medium to induce the fibrotic protein expression and lipid accumulation.

The kidney of each mouse was removed and fixed with 4% paraformaldehyde before embedding. H&E staining was performed for pathological analysis. For IHC staining, kidney sections were incubated with primary antibody LC3 or FTO overnight and then a horseradish peroxidase (HRP) polymer secondary antibody was used. Oil Red O staining and Sirius Red staining were performed using a corresponding kit based on the manufacturer’s instructions (Solarbio, Chondrex). Images were collected and analyzed with a fluorescence microscope (Leica DM 2500).

Levels of blood urea nitrogen (BUN) and total cholesterol (TC), low-density lipoprotein cholesterol (LDL), and high-density lipoprotein cholesterol (HDL) in serum were assayed by a corresponding kit (BioAssay Systems, USA, DIUR-500, Nanjing Jiancheng, China, TC: A111-1-1, LDL: A113-1-1, A112-1-1) according to the manufacturer’s construction.

Levels of TG in kidney tissue were measured by using a commercial kit (Nanjing Jiancheng, A:110-1-1). Tissues were weighted and added with PBS (1:9) for homogenization, and the supernatant was obtained. Then, the supernatant and the working solution of the kit were mixed and subjected to measurement of the absorbance at 546 nm. Cellular TG levels were measured using a commercial TG Quantification Colorimetric Kit (BioVision; USA, K622) according to the manufacturer’s protocol. In brief, cells were homogenized in solution containing 5% NP-40 in water and then heated to 100°C. Then, 20 μl cell lysate was adjusted to the volume of 50 μl/well with a triglyceride assay buffer. Finally, the absorbance at 570 nm was measured.

The nucleus and cytoplasm protein of HK2 cells were isolated by using a Nucleus/Cytosolic Extraction Kit (Fude Biological Technology, China, FD0199) based on the manufacturer’s instructions. Briefly, cells were incubated with Nc-Buffer A for 10 min on ice and then added with Nc-Buffer B. After centrifugation, the supernatant was collected to obtain the cytoplasm protein. Nc-Buffer C was added to the deposit to collect the nuclear protein. All the Nc-Buffer should be added with 1% protease inhibitor before use. Nuclear and cytoplasmic proteins were subjected to immunoblot analyses.

Proteins of cells and tissues were isolated using a RIPA buffer containing protease and phosphatase inhibitor cocktail. Total protein was quantified using the BCA method (Fude Biological Technology, China, FD2001). Equal amounts of protein from each group were separated through an SDS-polyacrylamide gel. The following antibodies were used: STAT6 (sc-374021), p-STAT6 (sc-136019), Arg-1 (sc-166920), TGF-β1 (sc-146), α-SMA (sc-53142), FN (sc-18827), LC3I/II (sc-398822), SQSTM1 (sc-28359), CPT-1α (sc-393070), PPARα (sc-398394), and GAPDH (sc-32233) from Santa Cruz Biotechnology. FTO (AF6936) and N6-Methyladenosine (m6A) Rabbit Polyclonal Antibody (AF7407) were from Beyotime. Protein half-life assays were performed according to our previous protocol (22). Briefly, cell lysates from the control or Cana group were collected at different time points after cycloheximide (CHX) administration and subjected to immunoblot analyses with the anti-STAT6 and anti-GAPDH antibodies. The prestained protein marker (Vazyme Biotech Co., Ltd., MP102-01) was used to identify the specific bands.

For indirect immunofluorescence, cells were fixed on round glass coverslips (Fisher Scientific) with chilled methanol. After incubation with the primary antibodies and the respective secondary antibodies for 50 min each, coverslips were washed with PBS and mounted with antifade mounting solution (Invitrogen). A fluorescence microscope (Leica DM 2500) was used to obtain representative images.

Cell viability assay was performed according to the previously described method. Briefly, the potential cytotoxicity of Cana in HK2 cells was assessed by the functional impairment of the mitochondria with 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT, Sigma). Approximately 1 × 10⁴ cells per well were plated in a 96-well plate and incubated overnight. The cells were treated with different doses of Cana for 48 h. Then, MTT was added into the cells followed by incubation at 37°C for 2 h. The medium was removed, and 100 μl isopropanol/HCl was added into each well. Absorbance at 570 nm was measured by a synergy 2 multimode microplate reader (BioTek, Seattle, USA).

HK2 cells were treated with 5 μg/ml actinomycin D (GC16866, GLPBIO, USA) to inhibit mRNA transcription. Cells were collected at 0, 3, and 6 h. RNA was extracted as below described and used for quantitative polymerase chain reaction (qPCR).

Total RNA was isolated from cells and kidney tissues using TRIzol reagent (CWBiotech, China, CW0580S). Equal amounts of mRNA were synthesized using a iScript cDNA Synthesis Kit based on the manufacturer’s instructions (Vazyme Biotech, China, R333-01) with a 0.2-ml PCR tube (Nest, China, 401001). Real-time PCR was performed using a qPCR SYBR Green Master Mix (Yeasen, 11184). Primer sequences were listed in the supplementary.

HK2 cells were harvested and lysed in an IP lysis buffer (Beyotime Biotechnology, P0013) containing a protease inhibitor cocktail and RNase inhibitor (Beyotime Biotechnology, R0102-2KU). After centrifugation, the supernatant was collected and 10% of the lysate was taken out for input. The remaining lysate was added to anti-FTO antibody incubated with protein A/G agarose beads at 4°C for 24 h. IgG was used for negative control RNA immunoprecipitation (RIP) reaction. 10% of the complex was collected to test the efficiency of immunoprecipitation after resuspending with a lysis buffer. The remaining lysate was subjected to centrifugation, and RNA was extracted as described above. RNA enrichment was normalized to the input control.

Total RNA was extracted as described above. An EpiQuik™ CUT&RUN m6A Enrichment kit (EpiGentek, USA, P-9018) was used to monitor the status of m6A in RNA. 200 ng of RNA was collected and stored at -80°C after cleavage. Anti-m6A antibody or IgG was incubated with affinity beads for 90 min and then washed with a wash buffer. A protein digestion buffer containing proteinase K was added and incubated with an antibody–RNA complex for 15 min at 55°C. Next, the RNA binding beads were resuspended and added to the complex and then the RNA was eluted with an elution buffer. The input of RNA and eluted RNA was subjected to qPCR as described above.

Total RNA was isolated as described above and spotted onto nylon membranes. Then, the membranes were crosslinked by ultraviolet (UV) and subjected to methylene blue and then blocked for 1 h with a blocking buffer (5% milk in phosphate-buffered saline with 1% Tween 20). Corresponding secondary antibodies were added and incubated for 1 h.

The levels of global m6A in mRNA were measured by using an EpiQuik m6A RNA Methylation Quantification kit (Colorimetric) (EpiGentek, Germany) based on the manufacturer’s instruction.

HK2 cells were transfected with mRFP-GFP-LC3 for 24 h and then either left untreated or separately treated with rapamycin, CQ, or Cana. Finally, cells were gently washed with PBS and images were captured with a fluorescence microscope (Leica DM2500, USA).

The expression profile and corresponding annotation platform of GSE118337 were downloaded from the National Center for Biotechnology Information (https://www.ncbi.nlm.nih.gov/) (23). The profile was then normalized by function normalizeBetweenArrays(). The normalized microarray series matrix was processed by the limma R package (24) to calculate the log fold change, p value, and adjusted p value of gene expression for further analysis, and genes that meet |log2FC| >1 and p < 0.05 were considered differentially expressed. A volcano plot and heatmap created by R were graphed to show the overall condition of the data.

Genes in GSE118337 regulated by TGF-β1 and restored by Cana (p < 0.05, |log2FC| >1) were used to create a protein–protein interaction (PPI) network on String and visualized by Cytoscape (25, 26). Retrieved from functional enrichment results, the genes related to fibrosis and fatty acid oxidation were highlighted.

The upregulated and downregulated genes in GSE118337 were respectively used for biological process functional enrichment on Metascape (27), the results of which were used to create bar plots showing functions related to the present investigation by R. Gene sets used for functional enrichment analysis and gene set enrichment analysis (GSEA) were downloaded from GSEA (http://www.gsea-msigdb.org/gsea/index.jsp) (28, 29). Then, software GSEA 4.2.2 and R package “ClusterProfiler” were employed for GSEA and visualization (30).

Correlation between the expression level of genes was calculated by Pearson correlation analysis. The p-value of the correlation coefficient was analyzed with package “hmisc,” and p-value <0.05 was considered statistically significant. An absolute value of Pearson correlation coefficients between 0.6 and 1 indicated a strong correlation, and that between 0.4 and 0.6 indicated a moderate one. The correlation was finally visualized by software Cytoscape and R package “corrplot” (31).

We applied a methodology to quantify the autophagy level (autophagy score) of samples. First, we chose 19 genes with frequent existence in pathways related to autophagy or reported in previous studies (32). Next, principal component analysis (PCA) was performed to establish the autophagy gene signature of genes selected above. Finally, we selected principal components 1 as signature scores.

The investigators were blinded to group allocation. Data were presented as mean ± SD. GraphPad Prism 8.0 and R statistical software were performed for graphics and statistical analyses. Unpaired Student’s t tests were applied for comparison between two groups. For multiple comparison analyses, one-way ANOVA with Bonferroni’s correction were performed. The differences with *p < 0.05 was considered statistically significant.

To verify the effects of Cana on renal fibrosis and lipid metabolism, the microarray expression dataset GSE118337 retrieved from the GEO database was analyzed. As shown in Figure 1A , there were 541 genes regulated by TGF-β1 and 589 genes regulated by Cana (p < 0.05, |log2FC| >1), with 332 genes at their intersection. 61.3% of the genes changed by TGF-β1 were further restored by Cana, revealing the potential therapeutic effect of Cana on renal fibrosis ( Figure 1B ). Among them, 193 genes and 139 genes were separately upregulated and downregulated by TGF-β1 and restored by Cana. Enrichment functions of all genes regulated by both TGF-β1 and Cana (p < 0.05) were conducted by Metascape (https://metascape.org/). As shown in Figures 1C–E , gene ontology terms associated with fibrosis and fatty acid metabolism were significantly mediated and the PPI network was graphed, suggesting that fatty acid metabolism and fibrosis-related genes are closely interacted with each other. These data showed that Cana exhibited the significant modulation both on renal fibrosis and fatty acid metabolism.

Based on the bioinformatic analysis above, we next conducted an in vivo study via establishing UUO and I/R models to confirm the protection of Cana on renal fibrosis. H&E staining showed that the histological injury caused by UUO and I/R was attenuated by Cana, so did the lipid accumulation as shown by Oil Red O staining and triglyceride (TG) content in the kidney, without damage on renal function ( Figures 2A, B, F, G , S1A ). In addition, Sirius Red staining exhibited the alleviated collagen deposition by Cana ( Figures 2C, H ). Our previous study showed that STAT6, a key nuclear transcription factor, is activated in fibrosis-associated tissue disorder and it could negatively regulate PPARα related-lipid metabolism, which contributes to lipid accumulation in renal tubular cells and kidney fibrosis (5). To explore the potential mechanism of Cana on the regulation of FAO and renal fibrosis, the expressions of STAT6 signal-related protein (STAT6, p-STAT6, Arg-1) and fibrosis-related protein (α-SMA, FN) were detected by immunoblot analyses ( Figures 2D, I ). Interestingly, here we found that STAT6 was suppressed in the kidney of UUO or I/R mice with Cana treatment. The deactivation of the STAT6 pathway was accompanied by reduced expressions of fibrosis-related protein α-SMA and FN in UUO and I/R kidney. Consistently, Cana mainly upregulated the FAO in the kidney with UUO and I/R performance ( Figures 2E, J ). These data suggested that Cana may attenuate UUO- and I/R-induced renal fibrosis through STAT6 inhibition and FAO activation.

Next, we conducted in vitro studies using HK2 cells to further confirm Cana’s protection effects observed in animal studies above. MTT assay was employed to select the dose range (0–40 μM) in the following in vitro experiments ( Figure 3A ). Consistently, Western blot showed that Cana decreased the expression and activation of STAT6 signals (STAT6, p-STAT6, and Arg-1) in a dose-dependent manner along with the increased expressions of SQSTM1 and LC3II and less effect on LC3I ( Figure 3B ). Since STAT6 could be degraded by autophagy according to others’ and our previous studies, the stability of STAT6 was next detected. Cana shortened STAT6 half-life from 7.38 to 5.11 h ( Figure 3C ). Consistent with Figure 3B , immunofluorescence (IF) staining showed that Cana increased LC3 expression in a dose-dependent manner ( Figure 3D ). Moreover, a tandem mouse red fluorescent protein (mRFP)-GFP-LC3 construct was employed. Two positive controls were conducted to differentiate increased formation of autophagosomes (rapamycin) from blockage of autophagosome degradation (bafilomycin A1, BafA1). As shown in Figure 3E , cells treated with Cana contained red puncta but less yellow puncta, which was the same as rapamycin, indicating that Cana may act as an inducer of autophagy. Moreover, qPCR and Western blot assays indicated that Cana alleviated TGF-β1-induced abnormal lipid metabolism and fibrotic protein expression ( Figures 3F, G ). Then, the effects of downregulated STAT6 by Cana were further explored. Overexpression of STAT6 exacerbated TGF-β1-induced aberrant lipid metabolism and fibrotic protein expression and impaired the protective effect of Cana, which further confirmed that Cana improved FAO and fibrosis via autophagy degradation of STAT6 ( Figures 3G–I ). These data suggested that Cana suppressed STAT6 through activation of autophagy to alleviate TGF-β1-induced aberrant lipid metabolism and fibrotic protein expression.

Since we found that Cana decreased STAT6, HK2 cells treated with different doses of Cana were separated to detect STAT6 expression in the cytoplasm and nucleus, respectively. The results showed that the expression of STAT6 in both cytoplasm and nucleus was decreased after Cana treatment ( Figure 4A ). To further determine the method of STAT6 degradation, cells were treated with Cana along with proteasome inhibitor (MG132) or autophagy inhibitor (BafA1). Results showed that the expressions of SQSTM1 and LC3II were both increased after Cana treatment. Higher levels of LC3II and SQSTM1 were found in the Cana and BafA1 combined treatment group ( Figure 4B ), which was consistent with Figure 3E that Cana induced the autophagosome formation. Moreover, only BafA1 treatment reversed the decrease in STAT6 signaling.

Whether autophagy is involved in the protective effect of Cana against tubulointerstitial fibrosis in vivo was further confirmed. There was no obvious different histological change and collagen deposition ( Figure S2A ). Interestingly, remarkable tubular atrophy and interstitial collagen deposition were observed in the WT group with UUO performance, which was significantly ameliorated by Cana administration. These protective effects of Cana were inhibited in the tubular-Atg7^-/- (Atg7^cKO) group ( Figures 4C, D ). Consistently, Western blot showed that Atg7 deficiency in tubular cells vanished Cana’s effect of suppressing STAT6 signaling, inhibiting fibrotic protein expression, and increasing FAO-related protein expression ( Figure 4E ). These data suggested that Cana improved renal tubular cells’ FAO and kidney fibrosis through degradation of STAT6 in an autophagy-dependent manner.

SQSTM1 is well known as the receptor of autophagy-related degradation. In the above experiments, we found that SQSTM1 was induced after Cana treatment ( Figures 3B , 4B ). To elucidate the functional role of SQSTM1, a siRNA-based strategy was used to interfere HK2 cells’ SQSTM1 expression and then detected the change of STAT6 and autophagy levels. The efficiency of siSQSTM1 was confirmed at the protein level. During knockdown of SQSTM1, the suppressed expression of STAT6 was alleviated while the increased expression of LC3II caused by Cana was inhibited ( Figure 5B ). In line with the results from Western blot, IF staining also showed that Cana barely induced LC3 expression with siSQSTM1 transfection ( Figure 5A ).

In order to further identify that Cana and SQSTM1 were related to autophagy, we systematically searched publicly available renal gene-expression data sets and chose GSE118337 for the next analysis. The Cana group and control group were selected as the object (33). Differentially expressed genes (DEGs) between the two groups were determined using the R package limma. The results represented by volcano plot showed that many DEGs were related to autophagy ( Figure 5C ). Moreover, functional enrichment analysis revealed that DEGs were significantly enriched in pathways related to autophagy ( Figure 5D ). To further explore the changes of autophagy, gene set enrichment analysis (GSEA) was used to observe the enrichment of gene signatures in different phenotypes. The results showed that autophagy-related genes were significantly enriched in the Cana-treated group ( Figure 5E ). It was concluded that Cana treatment was closely related to autophagy of renal proximal convoluted tubule epithelial cells and might upregulate the level of autophagy. Additionally, the correlation between SQSTM1 and 19 feature genes related to autophagy was calculated. The results indicated that autophagy-related genes were mostly positively correlated with SQSTM1 ( Figures 5F, G ). Then, 19 feature genes were extracted for further PCA to establish the autophagy score and rank. Samples were ranked based on gene expression of SQSTM1, and its relationship with PCA score was visualized with an alluvial diagram which showed that a high SQSTM1 expression got a high autophagy score rank and vice versa ( Figure 5H ). The above results indicated that there was a close relationship among autophagy, Cana treatment, and SQSTM1.

In the above results, we found that SQSTM1 is a key factor in the protection of Cana on renal fibrosis; we next sought to analyze its potential mechanism. To confirm Cana’s function from the perspective of m6A modification, we firstly identified whether m6A participates in the process of Cana administration. It was found that m6A levels were significantly increased in HK2 cells following Cana administration ( Figures 6A, B ). Consistently, the kidney of mice treated with Cana exhibited the same trend ( Figure 6C ). It was reported that the m6A level was mainly modified by methyltransferase or demethylase. Thus, we examined related gene expression to verify whether Cana regulates through mediating RNA methyltransferases or demethylases. As shown in Figure 6D , the expression of SQSTM1 was consistently upregulated and Arg-1 (a typical STAT6 downstream gene) expression was decreased. Notably, Cana significantly inhibited FTO expression, with no obvious change of other methyltransferase- or demethylase-related genes. In accordance with the results of qPCR, we validated that Cana deceased FTO at the protein level both in vitro and in vivo ( Figures 6E, F ). Next, the transcriptional downregulation manner of FTO by Cana was further confirmed in the cells with either proteasome inhibitor (MG132) or autophagy blocker (BafA1) treatment. Interestingly, the levels of FTO in the BafA1-treated group were dramatically increased compared with the Ctrl group. However, BafA1 treatment could not recover the decrease of FTO caused by Cana, as well as MG132, which indicated that the downregulation of FTO by Cana is not caused by increased protein degradation through either proteasome or autophagy ( Figure 6G ). Next, IF staining and Western blot were employed to further investigate the effect of FTO on autophagy. The results showed that the increased LC3II and SQSTM1 was inversed with FTO overexpression ( Figures 6H, I ). Also, a tandem mouse red fluorescent protein (mRFP)-GFP-LC3 construct was employed. As shown in Figure S3A , HK2 cells treated with Cana contained red puncta but less yellow puncta, indicating its induction of autophagy. However, overexpression of FTO attenuated Cana’s effect on autophagy induction. Moreover, we found that Cana treatment increased the stability of SQSTM1 mRNA, whereas overexpressing FTO resulted in decreased stability of SQSTM1 mRNA. These data suggested that Cana decreased FTO transcription, which negatively regulated autophagy through modulation of the SQSTM1 mRNA stability ( Figure 6J ).

Then, we profiled that SQSTM1 mRNA contained the potential m6A modification region (position: 1906) with an FTO-binding motif as shown in Figure S4 . Next, the RNA immunoprecipitation (RIP)-qPCR experiment was conducted and the results confirmed that FTO protein interacted with SQSTM1 mRNA ( Figure 6K ). Moreover, the methylated RNA immunoprecipitation (MeRIP) assay showed that Cana increased the m6A modification of SQSTM1 mRNA, whereas overexpression of FTO attenuated Cana’s effect of increasing SQSTM1 mRNA m6A modification ( Figure 6L ). With the vital role of SQSTM1 in the modulation of autophagy, we then downregulated SQSTM1 to observe the effect of Cana. As shown in Figure 6M , the knockdown efficiency was confirmed using immunoblot analysis. Cana significantly alleviated TGF-β1-induced lipid accumulation. Oppositely, siSQSTM1 exacerbated TGF-β1-induced lipid accumulation and impaired the protective effect of Cana. Collectively, these data suggested that Cana showed negative regulation on FTO and enhanced SQSTM1 mRNA stability.

To further provide evidence for the in vivo role of SQSTM1, renal tubular-specific SQSTM1 deficiency mice were generated using SQSTM1^flox/flox and γGT^Cre mice. There were no obvious differences in histological change and collagen deposition, while there were comparisons between WT and SQSTM1^cKO mice at 2 months after birth ( Figure S5A ). As shown in Figure 7A , obvious tubular atrophy and interstitial collagen deposition were presented in the WT group with UUO treatment, which was ameliorated with Cana administration. These protective effects were dampened in the SQSTM1^cKO group. Additionally, SQSTM1^cKO mice exhibited more aggravated renal morphological damage and fibrosis in comparison with the indicated WT mice. Also, Cana was found to decrease the TG content in WT mice, whereas there was no change in SQSTM1^cKO mice. Furthermore, TG content in SQSTM1^cKO mice was even higher than that in WT mice ( Figure 7B ). Consistently, immunoblot and qPCR analyses also showed that suppression of the STAT6 signaling pathway by Cana only in WT mice and the SQSTM1^cKO group showed higher levels of STAT6 and p-STAT6 expression compared with the WT group. Consistently, Cana eliminated the alteration of UUO-induced fibrosis-related protein and PPARα-related FAO gene expression, whereas SQSTM1^cKO mice exhibited more fibrosis-related gene expression and less FAO-related gene expression in comparison with WT mice ( Figures 7C, D ). These data suggested that SQSTM1 was essential for the function of Cana on renal fibrosis and SQSTM1 deletion eliminated the protection of Cana against renal fibrosis.