Hepatocyte KLF6 expression affects FXR signalling and the clinical course of primary sclerosing cholangitis

Primary sclerosing cholangitis (PSC) is characterized by chronic cholestasis and inflammation, which promotes cirrhosis and an increased risk of cholangiocellular carcinoma (CCA). The transcription factor Krueppel‐like‐factor‐6 (KLF6) is a mediator of liver regeneration, steatosis, and hepatocellular carcinoma (HCC), but no data are yet available on its potential role in cholestasis. Here, we aimed to identify the impact of hepatic KLF6 expression on cholestatic liver injury and PSC and identify potential effects on farnesoid‐X‐receptor (FXR) signalling.


| INTRODUC TI ON
The ubiquitously expressed transcription factor Krüppel-likefactor-6 (KLF6) has been established as a tumour suppressor gene and regulator of cell growth, differentiation, tumorigenesis, and signal transduction in numerous studies. 1 Members of the KLF family are zinc-finger-proteins and have a conserved zinc finger DNA-binding domain, which is able to bind to GC-rich motifs on diverse target genes. 2 KLF6 can act both as activator and repressor of transcription. 3,4 Several studies indicated a general role for KLF6 in suppressing tumorigenesis by controlling key mechanisms involved in tumour formation, such as cell growth, differentiation, adhesion and endothelial motility. [5][6][7][8][9] KLF6 expression or its functional loss have been associated with the progression and outcome of various tumours in patients, including hepatocellular carcinoma (HCC). [10][11][12] Beside playing a role in HCC, KLF6 has been established as an important transcription factor in a variety of liver diseases, including hepatic fibrogenesis, acute liver injury, ischaemia reperfusion damage and non-alcoholic fatty liver diseases (NAFLD). [13][14][15] In the previous studies, we characterized KLF6 as a regulator of autophagy in acute liver injury and as a mediator of hepatic glucose and lipid metabolism in NAFLD by binding directly to the promoter regions of autophagy-regulating factors and hepatic glucokinase, as well as a post-transcriptional regulation of peroxisome proliferator activated receptor alpha (PPARα) expression. [16][17][18] Interestingly, no data are yet available on a potential role of KLF6 in cholestatic liver diseases.
Primary sclerosing cholangitis (PSC) is a cholestatic liver disease characterized by chronic inflammation and consecutive constriction of the intra-and extra hepatic bile ducts resulting in progressive biliary fibrosis and cirrhosis. 19 The underlying mechanisms are still insufficiently understood. The prevalence of PSC is notably increased in patients with chronic inflammatory bowel diseases, especially ulcerative colitis, predominantly in males. 20 Chronic cholestasis such as in PSC drives hepatocellular injury followed by inflammation promoting cell proliferation and tumorigenesis. Consecutively, PSC is associated with an increased risk of cholangiocellular carcinoma (CCA) and HCC. 21 To reduce cholestasis in patients with so-called dominant strictures, dilatation therapy via endoscopic retrograde cholangiopancreatography may be performed as a therapeutic option. While there is no established Medicine 6, Nürnberg, Germany as KLF6, have been identified to play crucial roles in the response to liver injury, far beyond their contribution to tumour formation. In this study, we investigate the effects of KLF6 expression on cholestasis and the development of billiary inflammation in patients with chronic cholestasis as well as in an animal model. We further aimed to identify the underlying mechanisms in a cell culture model. Our data indicate a novel mechanistic interaction of KLF6 with target genes of the FXR regulated bile acid metabolism. medical therapy, antibiotics are considered for treatment of bacterial cholangitis. As high-dose therapy appears harmful, some patients might benefit from low-dose ursodeoxycholic acid (UDCA) and early stage clinical studies suggest a beneficial effect of Farnesoid X Receptor (FXR) agonists in PSC. 22 With established cirrhosis, liver transplantation is the only therapeutic option in progressive disease. [23][24][25] Nuclear receptors, such as FXR, are ligand-activated transcription factors controlling crucial biological functions including cell growth, glucose and lipid metabolism, as well as bile acid synthesis and systemic bile acid accumulation and therefore represent promising therapeutic targets for various diseases. 26 Several FXR agonists are currently being tested in clinical trials as treatment options for cholestatic liver diseases. Among them, obeticholic acid (OCA) is already an established treatment option for primary biliary cholangitis (PBC), another autoimmune cholestatic liver disease. [27][28][29] While we have established KLF6 as a regulator of the nuclear receptor PPARα, others found KLF6 to repress the nuclear receptor PPARγ and interact with its cofactor Retinoid-X-Receptor alpha (RXRα). 18,30,31 However, no data are available yet on a potential interaction between KLF6 and the nuclear receptor FXR.
As FXR signalling is established as an important factor in cholestasis, we aimed to investigate the interaction of KLF6, FXR and its target genes in cholestatic liver diseases and potential effects of KLF6 expression on clinical features of PSC in a well-defined cohort.

| Patient material and ethical considerations
For this study we analysed retrospective data and stored sam-

| Statistical analysis
Statistical significance was determined using an unpaired (or paired, when applicable), two-tailed t test or by one-way ANOVA (with Tukey's post-hoc test for individual experimental conditions) as well as Kaplan-Meier survival analysis were performed with GraphPad Prism 8 (GraphPad Software Inc). Significance was assumed at P ≤ .05. If not stated otherwise all data are presented as mean ± SEM.
All additional information on the individual experimental methods and materials are described in the Data S1 section.  Table 1. Based on hepatocyte expression levels of KLF6 in patients' liver biopsies, as assessed by immunohistochemistry, we identified two clusters of either high (>80%) or low (<80%) KLF6 expression in hepatocytes. This pattern was even more pronounced when exluding patients, TA B L E 1 Comparison of demographic, clinical, histological and laboratory data of PSC patients with KLF6 low (n = 9) and KLF6 high (n = 10) hepatocyte expression levels; a comparison of the groups showed no significant differences for the different parameters. T = total number  Figure 1A).

| Hepatocyte KLF6 expression is associated with clinical outcome in PSC
Based on these results and in order to investigate a potential influence of hepatocyte KLF6 expression on clinical features of PSC, we then arbitrarily separated these patients in two groups with either high expression (>80%, n = 10) or low expression (<80%, n = 9) of KLF6 in hepatocytes for further analysis. There were no significant differences of pANCA levels, the histological presence of fibrosis or ductular proliferation comparing the groups (Table 1).
Patients with high KLF6 expression showed a trend towards lower serum levels of ALT (alanine aminotransferase), γGT (gamma glutamytransferase) and AP (alkaline phosphatase), with higher AST (aspartate aminotransferase). However, these differences were not significant (Table 1). Interestingly, all three patients that developed CCA in the course of the disease as well as most who deceased during the observation period had low hepatocyte KLF6 expression ( Figure 1B,C).
To compare the clinical outcome of the patients, we performed a Kaplan-Meier analysis to test the influence of KLF6 expression on patients' survival. Here, given the small sample size, it was remarkable that those patients who had higher hepatocyte KLF6 expression had significantly better survival rates than patients with low KLF6 expression ( Figure 1D). This effect was still present, while attenuated, when we declared the time of surgery in the four individuals who underwent LTX as negative endpoint in the Kaplan-Meier analysis ( Figure 1E).
In contrast, in our PBC cohort (see demographics in Table S2), the distribution of hepatocyte KLF6 expression was less accentuated, based on a more even distribution between 1% and 92% ( Figure S1C). When applying the 80% cutoff, in contrast to PSC, in PBC patients with high KLF6 expression we observed a trend towards higher transaminase levels and a significantly elevated γGT, indicating potentially different roles of hepatic KLF6 in two distinct cholestatic diseases, PSC and PBC (Table S2). Our data indicate that higer hepatic KLF6 expression is associated with a better outcome in PSC. WT animals ( Figure 2). Additionally, liver tissue sections were evaluated using a blinded scoring system to quantify steatosis, ballooning, lobular inflammation, portal inflammation, necrosis, ductular reaction and fibrosis by an expert hepatopathologist. Following BDL,

| Attenuated liver injury with accentuated cholestasis in a model of hepatocyte specific Klf6knockout mice following bile duct ligation
DeltaKlf6-hep animals showed slightly more cells with ballooning but significantly less necrotic cells were detected ( Figure 2D). So in summary, a conditional Klf6 knockout affects the phenotype of the liver injury following BDL.

| Klf6 knockout affects expression of FXR target genes following BDL
The mRNA levels of Klf6 showed a robust (>20-fold) increase after BDL in WT mice, but not in DeltaKlf6-hep animals indicating that the sharp increase in Klf6 in BDL-treated WT mice is hepatocytespecific ( Figure 3A). As FXR activation is involved in cholestasis and PSC, we next measured mRNA expression in the whole liver tissue KLF6 is a transcriptional repressor of the FXR target gene NR0B2.
In different rat models, treatment with FXR agonists lowers serum levels of ALT and AST, decreases necrosis, inflammation, and bile duct proliferation following BDL. 32,33 Our studies indicate that FXR signalling is over-activated in BDL-treated DeltaKlf6-hep knockout mice but not in sham-treated animals. Furthermore, the expression of Klf6 was increased in WT mice following BDL. The results suggest a close interaction of both factors, although interestingly in our PSC cohort we did not see significant differences of FXR expression in relation to KLF6 expression ( Figure S1D). However, to examine the mutual  Figure S2A), which was also described previously. 16 To analyse the transcriptional effects of KLF6 on FXR-controlled genes/ promoter elements we performed luciferase reporter assays with con- here KLF6 overexpression had no effect on the mRNA expression of FXR target genes ( Figure S2). In summary, KLF6 represses the induction of NR0B2 and attenuates the activation of ABCB11 following CDCA stimulation.

| KLF6 binds directly to the promoter region of NR0B2 and represses its expression
Members of the KLF family bind with varying affinity to DNAbinding sites of target gene promoters and therefore function as showed that KLF6 binds directly to the promoter region of NR0B2 ( Figure 5F).
Our findings indicate that KLF6 influences FXR signalling by potentially binding to FXREs and therefore KLF6 may work as an agonist or antagonist. ChIP assay confirmed a physical interaction between KLF6 and the promoter region of the FXR target gene

| D ISCUSS I ON
The clinical course of PSC is primarily characterized by cholestasis promoting fibrogenesis and the occurrence of cirrhosis. For advanced stage PSC, liver transplantation is the only curative option.
Here, we describe a novel interrelation between hepatic KLF6 expression and overall-survival as well as a potential role in the formation of CCA in a small, but well-characterized cohort of PSC-

ACK N OWLED G EM ENTS
The authors thank Lena Wingerter, Sami Jafoui, Alina Schlinkheider, Claudia Arndt and Nadine Köhler for excellent technical assistance.