EZH2‐mediated inhibition of KLF14 expression promotes HSCs activation and liver fibrosis by downregulating PPARγ

Abstract Objectives Induction of deactivation and apoptosis of hepatic stellate cells (HSCs) are principal therapeutic strategies for liver fibrosis. Krüppel‐like factor 14 (KLF14) regulates various biological processes, however, roles, mechanisms and implications of KLF14 in liver fibrosis are unknown. Materials and Methods KLF14 expression was detected in human, rat and mouse fibrotic models, and its effects on HSCs were assessed. Chromatin immunoprecipitation assays were utilized to investigate the binding of KLF14 to peroxisome proliferator‐activated receptor γ (PPARγ) promoter, and the binding of enhancer of zeste homolog 2 (EZH2) to KLF14 promoter. In vivo, KLF14‐overexpressing adenovirus was injected via tail vein to thioacetamide (TAA)‐treated rats to investigate the role of KLF14 in liver fibrosis progression. EZH2 inhibitor EPZ‐6438 was utilized to treat TAA‐induced rat liver fibrosis. Results KLF14 expression was remarkably decreased in human, rat and mouse fibrotic liver tissues. Overexpression of KLF14 increased LD accumulation, inhibited HSCs activation, proliferation, migration and induced G2/M arrest and apoptosis. Mechanistically, KLF14 transactivated PPARγ promoter activity. Inhibition of PPARγ blocked the suppressive role of KLF14 overexpression in HSCs. Downregulation of KLF14 in activated HSCs was mediated by EZH2‐regulated histone H3 lysine 27 trimethylation. Adenovirus‐mediated KLF14 overexpression ameliorated TAA‐induced rat liver fibrosis in PPARγ‐dependent manner. Furthermore, EPZ‐6438 dramatically alleviated TAA‐induced rat liver fibrosis. Importantly, KLF14 expression was decreased in human with liver fibrosis, which was significantly correlated with EZH2 upregulation and PPARγ downregulation. Conclusions KLF14 exerts a critical anti‐fibrotic role in liver fibrosis, and targeting the EZH2/KLF14/PPARγ axis might be a novel therapeutic strategy for liver fibrosis.


| INTRODUC TI ON
Various chronic liver diseases cause liver fibrosis, and advanced liver fibrosis leads to liver cirrhosis, which results in approximately 1.16 million deaths per year worldwide. Liver fibrosis also generates a permissive niche for the development of hepatocellular carcinoma. 1,2 Therefore, it is urgent to uncover the underlying mechanism of liver fibrogenesis and develop effective targets for better anti-fibrotic therapies. Liver fibrosis is featured by excessive accumulation of extracellular matrix (ECM), and the central event for this process is hepatic stellate cells (HSCs) activation. 3,4 Under physiological condition, HSCs are quiescent and the most distinctive feature is abundant vitamin A stored on cytoplasmic lipid droplets (LD), 5 and this phenotype is mainly maintained by several adipogenic transcription factors, such as peroxisome proliferator-activated receptor γ (PPARγ), CCAAT/enhancer-binding proteins (C/EBPs), liver X receptor α (LXRα) and sterol regulatory element-binding protein 1c (SREBP-1c). 6,7 Upon activation, the quiescent HSCs transdifferentiate to myofibroblasts, which are fibrogenic, proliferative, contractile and chemotactic, accompanied by rapid loss of LD. 8 Recent studies reported that recovery of the LD content in activated HSCs could convert the activated HSCs to the quiescent phenotype, and induction of HSCs deactivation and apoptosis are principal therapeutic strategies for liver fibrosis. [9][10][11] The Krüppel-like factors (KLFs) are a family of transcription factors containing 17 members, which are implicated in embryogenesis, proliferation, apoptosis, differentiation and development. [12][13][14] KLFs consist of three evolutionarily conserved cysteine and histamine zinc fingers (C-terminal C 2 H 2 DNA binding domain), which recognize and bind to CACCC motifs or GC-rich sequences in the promoter, and mediate transactivation or transrepression of the target genes. 15 In recent years, KLF14 has evoked significant attention. KLF14 inhibited KRAS-mediated proliferation of pancreatic cells and induced apoptosis. 16 Overexpression of KLF14 led to G2/M arrest, and KLF14 acted as a tumour suppressor in breast ductal carcinoma and colon cancer. 17 In hepatocellular carcinoma cells, lncRNA DGCR5 sponged miR-346 and upregulated KLF14, and therefore inhibiting proliferation and migration. 18 Similarly, lncRNA HAND2-AS1 sponged miR-1275, upregulated KLF14 expression, and therefore inhibiting proliferation and invasion of colorectal cancer cells. 19 In summary, these studies suggested that KLF14 mainly exerted inhibitory roles in proliferation, migration and survival. Furthermore, KLF14 has recently emerged as a master regulator of multiple metabolic phenotypes in adipose tissue, 20 and it is closely correlated with type 2 diabetes and obesity. 21,22 In addition, KLF14 not only activates the generation of lipid-mediated signalling molecules, 23 but also mediates lipid metabolism. 14 Considering the vital role of KLF14 in proliferation, survival, migration and lipid metabolism, we surmised that KLF14 might regulate biological processes of HSCs, and therefore regulating liver fibrosis.
In this study, we reported that KLF14 was decreased in liver fibrosis and during HSCs activation. KLF14 overexpression increased the LD accumulation, inhibited HSCs activation, proliferation, G2/M transition, survival and migration by transactivating PPARγ.
Mechanistically, KLF14 downregulation was mediated by enhancer of zeste homolog 2 (EZH2)-regulated histone H3 lysine 27 trimethylation (H3K27me3). In animal studies, adenovirus-mediated KLF14 overexpression ameliorated thioacetamide (TAA)-established rat liver fibrosis through activating PPARγ signalling. Furthermore, in vivo administration of EPZ-6438, a specific inhibitor for EZH2, dramatically alleviated TAA-established liver fibrosis in rats. Importantly, KLF14 expression was dramatically decreased in human with fibrosis, which was significantly correlated with EZH2 upregulation and PPARγ downregulation. Collectively, this study uncovers a novel mechanism underlying liver fibrogenesis, which might contribute to better anti-fibrotic therapies.

| Statistical analyses
The Student's t test or one-way ANOVA were performed to analyse data (presented as mean ± SD) using the Prism 5.0 GraphPad Software (La Jolla, CA, USA). If P < .05, the difference was considered statistically significant.
Detailed information of other materials and methods are provided in the Supplementary Materials. F I G U R E 1 KLF14 is inversely correlated with liver fibrosis and HSCs activation. A, H&E, Masson's trichrome, Sirius red and α-SMA staining of normal and TAA-treated rat liver tissues (n = 7, Scale bars: 100 μm). B, H&E, Masson's trichrome, Sirius red and α-SMA staining of normal and CCl4-treated mouse liver tissues (n = 6, Scale bars: 100 μm). C, The normal and TAA-treated rat liver tissues were subjected to RT-qPCR (n = 7) and Western blotting (n = 2) analyses for detection of α-SMA and KLF14 expression. D, The normal and CCl4-treated mouse liver tissues were subjected to RT-qPCR (n = 6) and Western blotting (n = 2) analyses for detection of α-SMA and KLF14 expression. E, The quiescent and activated R-HSCs were subjected to RT-qPCR and Western blotting analyses for detection of α-SMA and KLF14 expression (n = 3). F, Immunohistochemistry staining of KLF14 in human liver tissues (normal and cirrhotic, n = 3, Scale bars: 100 and 20 μm). G, Immunofluorescence staining of KLF14 and α-SMA in human liver tissues (normal and cirrhotic, n = 3, Scale bars: 20 μm). Yellow arrows indicated HSCs. **P < .05, **P < .01, ***P < .001 versus the control group

| KLF14 is inversely correlated with liver fibrosis and HSCs activation
To study the function of KLF14 in liver fibrosis, we examined KLF14 expression in TAA-induced rat liver fibrotic model and CCl4-induced mouse liver fibrotic model. Firstly, we assessed the degree of hepatic fibrosis in two models by haematoxylin and eosin (H&E), Masson's trichrome, Sirius red, and α-smooth muscle actin (α-SMA) staining.
With extension of exposure to TAA or carbon tetrachloride (CCl4), the degree of the fibrosis gradually aggravated ( Figure 1A To detect the expression of KLF14 during HSCs activation, we isolated and cultured the rat primary HSCs (R-HSCs). The R-HSCs cultured for 1 day were considered as quiescent HSCs, F I G U R E 2 KLF14 overexpression promotes lipid droplets accumulation in HSCs and inhibits HSCs activation. A, KLF14 expression in the indicated HSCs was measured by RT-qPCR and Western blotting analyses. B, Lipid droplets accumulation was measured by Oil Red O staining (Scale bars: 20 μm). C, Levels of α-SMA and COL1A1 were measured by RT-qPCR and Western blotting analyses. ***P < .001 compared with the control group. n = 3 F I G U R E 3 KLF14 overexpression inhibits HSCs proliferation and migration, and induces G2/M arrest and apoptosis. (A,B) Cell proliferation was assessed by CCK-8 and EdU incorporation assays (Scale bars: 100 μm). (C,D) Cell cycle and cell apoptosis were detected by flow cytometry. E, Cell migration ability was measured by Transwell migration assay (Scale bars: 100 μm). **P < .01, ***P < .001 versus the control group. n = 3 and the R-HSCs cultured for 10 days were deemed as activated HSCs. 24 Firstly, we measured the purity of R-HSCs by flow cytometry analyses, which showed the purity was 97.63% ( Figure   S1A). The quiescent HSCs had obvious blue-green autofluorescence under ultraviolet excitation, and they were positive for Desmin immunofluorescence staining ( Figure S1B). Furthermore, the activated R-HSCs were positive for α-SMA immunofluorescence staining ( Figure S1C). These results indicated the purity of R-HSCs was appropriate for cellular and molecular studies.
Then, we found that α-SMA expression increased dramatically in activated HSCs, while KLF14 expression decreased significantly ( Figure 1E). F I G U R E 4 KLF14 transactivates the adipogenic gene PPARγ expression in HSCs. A, The mRNA levels of PPARγ, C/EBPα, C/EBPβ, C/ EBPδ, LXRα and SREBP-1c were measured by RT-qPCR analysis. B, The protein level of PPARγ was assessed by Western blotting analysis. C, Luciferase reporter assays of indicated LX-2 cells co-transfected with pCMV-KLF14 and PPARγ promoter luciferase construct. D, Luciferase reporter assays of indicated LX-2 cells co-transfected with pCMV-KLF14 and serially truncated or mutant PPARγ promoter luciferase constructs. E, ChIP-qPCR assay was performed to assess the direct binding of KLF14 to the PPARγ promoter in LX-2 cells. *P < .05, **P < .01, ***P < .001, # P > .05 versus the control group. n = 3 F I G U R E 5 PPARγ is essential for the inhibitory role of KLF14 overexpression in HSCs. A, For PPARγ inhibition, the LX-2-control and LX-2-KLF14 cells were treated with GW9662 (1 μmol/L) for 24 h or stably transfected with LV-shPPARγ. The expression of KLF14 and PPARγ was assessed by Western blotting analysis. B, Lipid droplets accumulation was measured by Oil Red O staining (Scale bars: 20 μm). C, Expression of α-SMA and COL1A1 was measured by RT-qPCR and Western blotting analyses. D, Cell proliferation was assessed by CCK-8 assay. (E,F) Cell cycle and cell apoptosis were detected by flow cytometry. G, Cell migration ability was measured by Transwell migration assay (Scale bars: 100 μm). **P < .01, ***P < .001, ns, P > .05 versus the control group. ## P < .01, ### P < .001 versus the KLF14 group. n = 3 Immunohistochemistry staining showed that KLF14 expression was lower in human cirrhotic liver tissues than normal liver tissues, and KLF14 was localized in the nucleus ( Figure 1F, Figure S2A).
Furthermore, the results of double immunofluorescence staining confirmed that KLF14 was obviously decreased in human cirrhotic liver tissues, and KLF14 was expressed both in hepatocytes and F I G U R E 6 Downregulation of KLF14 in activated HSCs is mediated by EZH2. A, The indicated HSCs were subjected to DNMTs inhibitor (5-azadC, 2 μmol/L), G9a inhibitor (UNC0642, 2 μmol/L) and pan-HDAC inhibitor (ITF-2357, 100 nmol/L) for 48 h, or EZH2 inhibitor (EPZ-6438, 10 μmol/L) for 72h. KLF14 expression was measured by RT-qPCR and Western blotting analyses. B, Expression of EZH2 and KLF14 was assessed by RT-qPCR and Western blotting analyses. C, ChIP analysis for EZH2, H3K27me3 and IgG, and subsequent qPCR in KLF14 promoter using 2 independent primer sets in quiescent and activated R-HSCs; IgG as a negative control. *P < .05, **P < .01, ***P < .001, # P > .05 versus the control group. n = 3 HSCs (α-SMA positive cells) ( Figure 1G, Figure S2B). Collectively, KLF14 expression is inversely correlated with liver fibrosis and HSCs activation, which might play a role in liver fibrogenesis.

| KLF14 overexpression promotes lipid droplets accumulation in HSCs and inhibits HSCs activation
Increasing evidences show that KLF14 is a master regulator of multiple metabolic phenotypes, especially in the lipid metabolism, 21 and KLF14 acts as a transcriptional activator and promotes lipid generation. 23 Thus, we are determined to explore whether KLF14 could regulate the LD content in HSCs and therefore influencing HSCs activation. To this end, we ectopically upregulated KLF14 expression by lentivirus transfection in LX-2 and HSC-T6 cells, and stable cell lines were established by puromycin selection. As expected, lentivirus transfection resulted in robust KLF14 overexpression ( Figure 2A).
Oil Red O staining results showed that KLF14 overexpression dramatically promoted the LD accumulation in both cells ( Figure 2B).
Furthermore, KLF14 overexpression decreased the α-SMA and collagen A1 (COL1A1) levels ( Figure 2C). In addition, we utilized the freshly isolated R-HSCs, which were quiescent and KLF14 expression was at the highest level. By lentivirus transfection, KLF14 expression was silenced in quiescent R-HSCs ( Figure 2A). Oil Red O staining results showed that KLF14 silencing facilitated the process of LDs disappearance in R-HSCs, and the changes in cell shape of R-HSCs also indicated that KLF14 silencing promoted transdifferentiation of R-HSCs ( Figure 2B). Furthermore, KLF14 silencing promoted α-SMA and COL1A1 expression ( Figure 2C). Thus, KLF14 overexpression promotes LD accumulation and inhibits HSCs activation.

| KLF14 overexpression inhibits HSCs proliferation and migration, and induces G2/M arrest and apoptosis
Previous studies reported that KLF14 could regulate cell proliferation, cell cycle transition, survival and chemotaxis of several cells. 16,17,19 Thus, we aimed to investigate whether KLF14 regulated these processes of HSCs. By performing Cell Counting Kit-8 (CCK-8) analysis, we found that KLF14 overexpression decreased the pro-

| KLF14 transactivates the adipogenic gene PPARγ expression in HSCs
Maintenance of the quiescent phenotype of HSCs is mediated by adipogenic transcription factors, including PPARγ, C/EBPs, LXRα and SREBP-1c. 6,7 As KLF14 overexpression promoted the LD accumulation in the activated HSCs and converted activated HSCs to the quiescent phenotype, we aimed to investigate whether these effects were achieved by modulating the expression of adipogenic genes.
Our data showed that KLF14 overexpression significantly promoted PPARγ mRNA expression, however, mRNA levels of C/EBPα, C/EBPβ, C/EBPδ, LXRα and SREBP-1c were not significantly changed ( Figure 4A, Figure S4A), which was validated by Western blotting analysis ( Figure 4B). By luciferase reporter assay, we found that Collectively, these data indicate that KLF14 transactivates the adipogenic gene PPARγ expression in HSCs.

| PPARγ is essential for the inhibitory role of KLF14 overexpression in HSCs
Previous studies reported that PPARγ was essential for LD synthesis, and it inhibited HSCs activation, proliferation, migration and induced apoptosis and senescence. [25][26][27][28][29][30] Considering PPARγ was transactivated by KLF14 in HSCs, we aimed to investigate whether KLF14 exerted  Figure S4B). Thus, PPARγ is essential for the inhibitory role of KLF14 overexpression in HSCs.

| Downregulation of KLF14 in activated HSCs is mediated by EZH2
Epigenetic modifications are responsible for gene downregulation,  Figure 6A). In addition, knockdown of endogenous EZH2 with specific LV-shRNA led to robust reactivation of KLF14 expression in LX-2 cells ( Figure 6B).
These data suggested that targeting EZH2 reactivated KLF14 expression. Furthermore, to determine whether KLF14 gene silencing in activated HSCs was directly regulated by EZH2-mediated H3K27me3, we performed ChIP assay with antibodies against EZH2, H3K27me3 or control IgG in the quiescent and activated R-HSCs. To assess the EZH2 occupancy and H3K27me3 modification in KLF14 promoter, the immunoprecipitated DNA samples were quantified using 2 independent primers specific for KLF14 promoter region (−567/−409 bp; −315/−153 bp). The results of ChIP-qPCR assay showed that EZH2 was recruited to the KLF14 promoter and mediated H3K27me3 modification during HSCs activation ( Figure 6C). Hence, we reveal EZH2-regulated H3K27me3 modification is responsible for KLF14 downregulation in activated HSCs.

TAA-induced rat liver fibrosis
Recently, EPZ-6438 has been approved by FDA as the first EZH2 inhibitor for the treatment of epithelioid sarcoma, 35 and EPZ-6438 treatment inhibited the progression of several cancers. [36][37][38] Considering KLF14 overexpression inhibited liver fibrosis in vitro and in vivo, and KLF14 expression was robustly reactivated upon EPZ-6438 in vitro treatment, we aimed to investigate whether in vivo administration of EPZ-6438 affected the progression of liver fibrosis. The rat fibrotic model was established by TAA injection, and EPZ-6438 was orally administered accordingly in group Ⅲ ( Figure 8A). Our data demonstrated that EPZ-6438 administration mitigated the severity of TAA-induced hepatic fibrosis, as represented by H&E, Masson's trichrome, Sirius red staining, as well as α-SMA immunostaining ( Figure 8B). Additionally, the hydroxyproline content, serum ALT and AST levels decreased significantly in fibrotic rats following EPZ-6438 treatment ( Figure 8C,D). Western blotting showed that EPZ-6438 treatment decreased H3K27me3 level which was enhanced by TAA, as expected. Furthermore, EPZ-6438 treatment reduced TAA-induced α-SMA and COL1A1 expression, while EPZ-6438 treatment increased TAA-decreased KLF14 and PPARγ expression ( Figure 8E,F). Taken together, these results demonstrated that EZH2 inhibitor EPZ-6438 upregulated KLF14 expression and ameliorated TAA-induced rat liver fibrosis ( Figure 8G).
Subsequently, we found that the expression of KLF14 and PPARγ was dramatically decreased in fibrotic livers, and was further downregulated with the progression of fibrosis. On the contrary, the expression of α-SMA and EZH2 was increased with the progression of fibrosis ( Figure 9B). Notably, PPARγ expression was positively correlated with KLF14 expression, and KLF14 expression was negatively correlated with EZH2 and α-SMA expression ( Figure 9C), and these results were consistent with the in vitro data and animal data.
Collectively, these findings suggest that KLF14 suppression due to EZH2 elevation in fibrotic liver contributes to liver fibrosis progression by downregulating PPARγ.

| D ISCUSS I ON
Reversion of activated HSCs to the vitamin-A-storing phenotype and induction of apoptosis are the main therapeutic strategies for liver fibrosis. 10,11 Previously, Chen et al of our laboratory briefly reported that KLF14 was downregulated in multiple chronic liver diseases, and overexpression of KLF14 exerted a protective role in immunemediated liver damage by inducing differentiation of regulatory T cells. 39 Furthermore, KLF14 was involved in proliferation, survival and migration in several tumour cells, and KLF14 mainly exerted inhibitory roles in these cellular processes, and functioned as a protective factor. 16,17,19 What's more, recent studies identified that KLF14 was a master regulator in lipid metabolism. 14,23 However, roles, mechanisms and implications of KLF14 in HSCs and liver fibrogenesis have never been reported. In the present study, we reported that KLF14 was dramatically decreased in human, rat and mouse fibrotic liver tissues and during R-HSCs activation. In addition, KLF14 overexpression promoted the LD accumulation in HSCs, inhibited HSCs activation, proliferation and migration, and induced G2/M arrest and apoptosis, which were consistent with the inhibitory roles of KLF14 in the other cells. Furthermore, adenovirus-mediated KLF14 overexpression ameliorated TAA-induced rat liver fibrosis. In this study, we found that, apart from the downregulation in HSCs, KLF14 expression was also downregulated in hepatocytes in the fibrotic liver tissues ( Figure 1F,G), and injection of adenovirus led to KLF14 overexpression both in hepatocytes and HSCs ( Figure S5A), these results drive us to think whether KLF14 could also exert hepatoprotective activities, besides the directly regulatory roles in HSCs. The FITCconjugated TUNEL assays showed that Ad-KLF14 ameliorated TAAinduced apoptosis of hepatocytes ( Figure S6A). Furthermore, KLF14 overexpression inhibited apoptosis of two hepatocyte cell lines (L02 and Chang liver) ( Figure S6B,C). These in vivo and in vitro results suggested that KLF14 could also exert hepatoprotective activities.
Collectively, these clinical and experimental evidences strongly indicated that KLF14 exerted profound protective roles in liver fibrosis, and these effects could be achieved at least through direct regulation of HSCs and hepatoprotective activities. β-actin and total H3 were used as loading control. G, A schematic diagram of the role EZH2-KLF14-PPARγ signalling in liver fibrosis. Upon HSCs activation, the elevated EZH2 mediates suppression of KLF14 expression, which promotes HSCs activation and liver fibrosis by downregulating PPARγ. The EZH2 inhibitor EPZ-6438 rescues EZH2-mediated KLF14 downregulation, which transactivates PPARγ expression, converts the activated HSCs to the quiescent phenotype and induces apoptosis, and therefore alleviating liver fibrosis. ***P < .001 versus the control group. ### P < .001 versus the TAA group F I G U R E 9 KLF14 expression is significantly correlated with PPARγ and EZH2 expression in patients with liver fibrosis. A, H&E, Masson's trichrome, and α-SMA staining of human liver tissues from normal (n = 9), mild fibrosis (n = 12) and advanced fibrosis (n = 9). Scale bars: 100 μm. B, RT-qPCR analyses of α-SMA, KLF14, PPARγ and EZH2 in human liver tissues (n = 30). C, The correlations of α-SMA, KLF14, PPARγ and EZH2 were analysed by Pearson correlation analysis (n = 30). *P < .05, **P < .01, ***P < .001 versus the control group The adipogenic genes play crucial roles in maintaining HSCs quiescence and inhibiting HSCs activation, such as PPARγ, C/EBPs, LXRα and SREBP-1c. 6,7 Considering KLF14 was a master regulator of lipid metabolism and was confirmed to induce HSCs deactivation, we aimed to investigate whether KLF14 exerted these effects by regulating the expression of adipogenic genes. Interestingly, we found that KLF14 dramatically upregulated PPARγ expression both in LX-2 and HSC-T6 cells. Mechanistically, as transcription factors, KLFs exert the transactivation or transrepression function by binding to the CACCC motifs or GC-rich sequences in the promoter of target genes, 15 and the role varies in different cells and diseases. KLF14 was identified as a transrepressor by the Raul Urrutia's team for the first time, which coupled to mSin3A and HDAC2, and repressed the TGFβ receptor II promoter in pancreatic epithelial cancer cell lines. 40 Subsequently, they found that KLF14 transrepressed Cyclin A2 promoter in pancreatic cancer cell lines. 16 Figure S7A). Taken together, KLF14 overexpression transactivated PPARγ, and therefore converting the activated HSCs to the quiescent phenotype, thus alleviating liver fibrosis. However, the mechanism underlying the hepatoprotective activities of KLF14 needs further exploration.
Epigenetic modifications play important roles in HSCs activation and liver fibrosis, especially the DNA methylation, histone methylation (especially H3K27me3 and H3K9me2) and histone deacetylation, which lead to aberrant transrepression state. [31][32][33][34] Several studies reported that, among the ageing-related CpG sites, KLF14 promoter region was associated with chronological age. [45][46][47] Furthermore, with ageing and obesity, the level of DNA methylation in KLF14 promoter was increased significantly in several organs in mice, which led to downregulation of KLF14. 47 Moreover, the KLF14 gene was under a hypermethylation state in familial Alzheimer's disease, which regulated cell death signalling. 48 However, in this study, we found that inhibition of DNMTs using 5-azadC did not signifi- Polycomb repressive complex 2 consists of the catalytic subunit EZH2, which catalyses H3K27 trimethylation and leads to repression of gene expression. 49 Recent studies showed that EZH2 played critical role in initiation and development of various tumours, and blocking EZH2 signalling by EPZ-6438 inhibited tumour progression. [36][37][38] Strikingly, EPZ-6438 (Tazemetostat, Tazverik™), as a first-in-class, small molecular inhibitor of EZH2, received accelerated approval in January 2020 by FDA for the treatment of patients (age ≥ 16 years) diagnosed with locally advanced or metastatic epithelioid sarcoma, which are ineligible for complete surgical resection. 35 EPZ-6438 comes to be the first approved EZH2 inhibitor for clinical treatment. It is also undergoing several other clinical trials as anticancer agent. 50,51 In this study, our data showed that in vivo administration of EPZ-6438 dramatically alleviated TAA-induced rat liver fibrosis, indicating that EPZ-6438 might have a broader range of application, which is appealing and needs further clinical investigation. As previously reported, the stimulation of EZH2 expression and subsequent H3K27 trimethylation of PPARγ gene is one mechanism for MeCP2-mediated transcriptional silencing of PPARγ. 52 Our results confirmed EPZ-6438 treatment reactivated PPARγ expression in LX-2 and HSC-T6 cells ( Figure S7B). From our perspective, there might be at least three possible underlying mechanism for the suppressive effect of EPZ-6438. Firstly, EPZ-6438 inhibited the activity of EZH2, and rescued the expression of PPARγ, which was directly inhibited by EZH2 (direct regulation of PPARγ by EZH2). Secondly, EPZ-6438 abolished the suppression of KLF14 by EZH2, and reactivated KLF14 transcription, which subsequently transactivated PPARγ expression (indirect regulation of PPARγ by EZH2 via KLF14).
According to our results, EPZ-6438 administration upregulated the mRNA and protein expression of PPARγ and KLF14, which were both downregulated in fibrotic liver tissues ( Figure 8E,F).
We thought these results could at least confirm the existence of the indirect regulation, as we have confirmed KLF14 overexpression did alleviate liver fibrosis in vitro and in vivo. Thirdly, there might exist other genes also critical for anti-fibrosis, and these genes were transrepressed by EZH2, in this sense, EPZ-6438 administration might result in transactivation of these anti-fibrotic genes, and therefore alleviating liver fibrosis (the unknown way).
Collectively, EPZ-6438 could alleviate TAA-induced liver fibrosis by directly regulated PPARγ expression and indirectly regulated PPARγ expression via KLF14. Therefore, as an intermediate molecule, KLF14 did exert a significant role in the EZH2/PPARγ axis to some extent. However, whether there exists other mechanism for the effect of EPZ-6438 needs further exploration.
In summary, we reported a protective role for KLF14 in liver fibrosis. KLF14 reversed the activated HSCs to the quiescent phenotype and inhibited TAA-induced liver fibrosis by transactivating PPARγ expression. EZH2-mediated H3K27me3 modification is a novel mechanism responsible for KLF14 downregulation in HSCs activation. EPZ-6438, a specific EZH2 inhibitor, dramatically ameliorated TAA-induced rat hepatic fibrosis. Thus, KLF14 exerts a critical role in liver fibrogenesis, and targeting the EZH2/KLF14/PPARγ axis might provide a novel approach to liver fibrosis treatment.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflicts of interest.

AUTH O R CO NTR I B UTI O N S
ZD performed the experiments and wrote the manuscript. ML, ZW, ZL and YF helped in tissues collection and data analysis. LX, DT and ZD designed the project. LX and DT revised the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
All supporting data are included in the article and its additional files.