Article Text
Abstract
Objectives The mechanism of transformation to intestinal metaplasia in Barrett's oesophagus has not been clarified. We previously reported that bile acids activate the Cdx2 promoter via nuclear factor kappa B (NF-κB) and stimulate production of Cdx2 protein in oesophageal keratinocytes, resulting in production of intestinal-type mucin. Krüppel-like factor 4 (KLF4) is an important transcription factor in the development of intestinal mucosa and has similar functions as Cdx2. In the present study, we investigated the direct effects of bile acids on KLF4 expression as well as the precise mechanisms of expression in cultured oesophageal squamous epithelial cells.
Methods We investigated the expression of KLF4 in rat and human Barrett's epithelium specimens, while the response of that expression to bile acids was studied using a KLF4 promoter luciferase assay. In addition, oesophageal squamous epithelial cells were transfected with a KLF4 expression vector, after which their possible transformation into intestinal-type epithelial cells was investigated.
Results In both rat and human tissues, Barrett's epithelium strongly expressed KLF4. Furthermore, a bile acids mixture increased KLF4 promoter activity, and mRNA and protein expression in oesophageal epithelial cells. Results from mutation analysis of the KLF4 promoter suggested that the NF-κB binding site is responsible for bile acid-induced activation of the KLF4 promoter. In addition, KLF4 and Cdx2 stimulated each other by directly binding to the promoter of the other, while transfection of the KLF4 expression vector in oesophageal epithelial cells induced production of MUC2 protein.
Conclusion Bile acid-induced sequential expression of KLF4 followed by MUC2 production may have an important role in the development of Barrett's epithelium.
- Bile acid
- Barrett's oesophagus
- Cdx2
- KLF4
- Barrett's metaplasia
- oesophageal reflux
Statistics from Altmetric.com
Significance of this study
What is already known about this subject?
Cdx2 is a key mediator in the development of Barrett's oesophagus.
Bile acids directly augment Cdx2 via NF-κB.
Cdx1 is also an important molecular mediator of Barrett's oesophagus.
What are the new findings?
KLF4 is expressed in Barrett's epithelium.
The expression of KLF4 in oesophageal keratinocytes in response to bile acids induces metaplastic changes during Barrett's epithelium development.
The transcriptional network related to KLF4 and Cdx2 has important roles in development of this disease.
How might it impact on clinical practice in the foreseeable future?
Molecular targets for treatment of Barrett's oesophagus will be defined.
Introduction
Barrett's oesophagus is an acquired condition, in which stratified squamous epithelium is replaced by metaplastic columnar epithelium in the distal oesophagus.1 The condition is associated with chronic gastro-oesophageal reflux disease (GORD)2 and reflux of duodenal contents with bile acids is generally considered to be one of the most important risk factors in its development.3 In association with tissue damage and regeneration, one or a few stem cells may attempt to adapt to this new environment by altering the patterns of some gene expressions, and thus undergo profound phenotypic changes that lead to a different type of epithelium that is more resistant to such a novel environment. The causal link between bile acid reflux and alterations of some transcription factors has been studied; however, the precise mechanism of promotion of Barrett's epithelium formation by bile acid reflux remains to be characterised.
A number of studies have found that Cdx2 is a key mediator in the development of Barrett's oesophagus.4 5 We previously reported a two-step mechanism involved in the development of Barrett's epithelium, in which bile acids activate the Cdx2 promoter via nuclear factor kappa B (NF-κB) and stimulate the production of Cdx2 protein in oesophageal immature keratinocytes, with a resulting production of intestinal type mucin.6 In addition to Cdx2, we also recently showed that bile acids induce the expression of Cdx1 in oesophageal immature keratinocytes, and demonstrated an interplay mechanism between Cdx1 and Cdx2 that causes upregulation of each other by directly binding to the promoter of the other, stimulating the development of Barrett's epithelium.7
Krüppel-like factors (KLFs) are zinc finger-containing transcription factors that exhibit homology to the Drosophila melanogaster segmentation gene product Krüppel. KLFs comprise a family of evolutionarily conserved zinc finger transcription factors that regulate numerous biological processes, including proliferation, differentiation, development and apoptosis.8 Among them, KLF4 (gut-enriched Krüppel-like factor) is highly expressed in epithelial cells of the small and large intestines.9 The expression pattern of KLF4 is similar to that of Cdx2 in the intestine, as it is expressed mainly in non-proliferating differentiating and differentiated cells of the upper crypt and villus/surface mucosa, where Cdx2 is also expressed.8 During embryogenesis, the expression of KLF4 begins to rise, which correlates with a critical period of gut epithelium morphogenesis, similar to that of Cdx2.10 Furthermore, colonic goblet cells in KLF4−/− mice do not show a normal goblet cell morphology and the goblet cell marker MUC2 exhibits patchy expression throughout the colonic epithelium.11 These findings suggest that KLF4 plays a fundamental role in the development of intestinal mucosa.
With regard to carcinogenesis, it was reported that KLF4 expression is downregulated in human adenomatous polyps and cancer of the colon.12 Notably, in colorectal adenomas and adenocarcinomas, the level of Cdx2 protein is markedly reduced.13 Also, Cdx2 was shown to activate a KLF4 promoter construct in Chinese hamster ovary (CHO) cells and colon cancer related RKO cells,14 15 while an inter-regulation mechanism between KLF4 and Cdx2 has been speculated.
In the present study, we investigated whether alterations of KLF4 expression in response to bile acids in oesophageal keratinocytes induce metaplastic changes during Barrett's epithelium development. Furthermore, we investigated the transcriptional network connecting KLF4 and Cdx2 in development of the disease.
Materials and methods
Rat model of Barrett's oesophagus
To induce Barrett's oesophagus, we employed Levrat's model with minor modifications, as previously described.6 7 16 In brief, the gastro-oesophageal junction was cut and the oesophageal end separated. The distal end of the oesophagus was then reimplanted 2 cm beyond the ligament of Treitz in an end-to-side fashion into a loop of the jejunum and the proximal end of the stomach was ligated. Six months after formation of oesophageal–jejunal anastomoses, the rats were killed and their oesophagi removed.
Patients and tissues
Human oesophageal tissues were collected after obtaining informed, written consent from all subjects. During endoscopy procedures, biopsy specimens of normal squamous mucosa from the distal oesophagus (n=6) and Barrett's oesophagus without dysplasia (n=6) were taken, then snap-frozen in liquid nitrogen. Barrett's oesophagus was histologically defined as the presence of columnar epithelium containing goblet cell metaplasia.
Cell culture and bile acid treatment
Five cell lines, including Het-1A (a human normal oesophageal cell line immortalised by viral SV40 transfection; American Type Culture Collection, ATCC, Manassas, Virginia, USA), OE33 (a human oesophageal adenocarcinoma cell line; European Collection of Cell Cultures, ECACC, Salisbury, Wiltshire, UK), OE19 (a human cell line established from an adenocarcinoma obtained from the gastric cardia/oesophageal gastric junction; ECACC), SW480 (a human colorectal adenocarcinoma cell line, ATCC) and HeLa (a human cervical adenocarcinoma cell line, ATCC), were used in this study. Primary cultures of oesophageal keratinocytes from normal rat oesophagi were established, as previously described.6
A mixture of bile acids (Sigma Chemicals, St. Louis, Missouri, USA), which included cholic acid, glycocholic acid and taurocholic acid, was used as a stimulant, as previously described.7
Vector construction and luciferase assay
We amplified 1700 bp of the KLF4 promoter (accession No. AF117109) by PCR, then cloned that into the MluI and BglII sites of a pGL3-basic luciferase vector (Promega, Madison, Wisconsin, USA) to generate pKLF4/1700-Luc (−1735 to −36), which generated pKLF4/1080-Luc (−1115 to −36), pKLF4/425-Luc (−460 to −36), pKLF4/233-Luc (−268 to −36), and pKLF4/35-Luc (−70 to −36). The position +1 refers to the major transcription start site identified in the KLF4 gene.9 We also amplified 1541 bp of the Cdx2 promoter (accession No. NC_000071) by PCR, then cloned that into the KpnI and BglII sites of a pGL3-basic luciferase vector to generate pCdx2/1541-Luc (−1415 to +125), which generated pCdx2/1014-Luc (−888 to +125), pCdx2/631-Luc (−506 to +125), pCdx2/438-Luc (−313 to +125), pCdx2/319-Luc (−194 to +125), pCdx2/219-Luc (−94 to +125), and pCdx2/74-Luc (+52 to +125), as previously described.6 Furthermore, 1750 bp of the MUC2 promoter (accession No. AF221746) was amplified by PCR, then cloned into the MluI and BglII sites of a pGL3-basic luciferase vector to generate pMUC2/1750-Luc (−1804 to −55), which generated pMUC2/823-Luc (−877 to −55), pMUC2/463-Luc (−517 to −55), pMUC2/214-Luc (−268 to −55), pMUC2/80-Luc (−134 to −55), and pMUC2/39-Luc (−93 to −55). The position +1 refers to the major transcription start site identified in the MUC2 gene.17 As an internal control for the dual luciferase assay, pRL-TATA-Renilla-Luc was used.6 To produce mutated KLF4 promoter constructs for pM/KLF4-Luc, 5′-ggcggccgccagtacttcaccggccgagagagcgagcgcggctcc-3′ was used (nucleotide substitutions indicated in bold), to produce mutated Cdx2 promoter constructs for pM/Cdx2-Luc, 5′-cggcgggtcattccaagtctctacagcttactggcaaggaggtgggaggaaa-3′ was used, and to produce mutated MUC2 promoter constructs, for pM/MUC2-Luc, 5′-cttggcaaataatacgtgaatatttcgcacctccctcgtcctccgccctcg-3′ was used.
cDNA encoding full-length mouse KLF4 (NCBI NM-010637) was amplified by PCR and cloned into a pcDNA5/FRT/V5-His-TOPO Vector (Invitrogen, Carlsbad, California, USA). Vector DNA without KLF4 sequences was used as a negative control. A Cdx2 expression vector was also constructed, as previously reported.6
Het-1A, OE33, and OE19 cells were separately cultured and transfected with 0.5 μg of each promoter vector and 0.02 μg of pRL-TATA-Renilla-Luc in each well, with Lipofectamine 2000 (Invitrogen). At 24 h after transfection of the luciferase vectors, the cells were stimulated with various concentrations of the bile acids mixture or the vehicle alone for 3 h, then cell lysates were used to determine luciferase activity. Also, Het-1A cells were cultured and transfected with 0.2 μg of each indicated promoter vector and a total of 0.2 μg of each indicated expression vector or an empty vector, along with 0.02 μg of pRL-TATA-Renilla-Luc in each well for 24 h, then the cell lysates were used for measurement of luciferase activity.
Immunohistochemistry
Immunohistochemistry was performed as previously described.6 To identify KLF4-expressing cells, tissue sections were incubated with the anti-KLF4 antibody (1:100; Medical & Biological Laboratories, Nagoya, Japan), followed by incubation with secondary biotinylated anti-rabbit immunoglobulin (DAKO, Carpinteria, California, USA). Bound antibodies were detected using a 3-amino-9-ethylcarbazole substrate–chromogen system (DAKO). The sections were counter-stained with haematoxylin.
RNA extraction and real-time PCR
Extraction of total RNA was performed as previously described.6 DNase I-treated RNA was reverse transcribed into cDNA using a ReverTra Ace α kit (Stratagene Toyobo, Tokyo, Japan). A real-time fluorescence PCR assay based on SYBR Green (Applied Biosystems, Foster City, California, USA) was then performed using the primers described in table 1.
Primary cultured cells were transfected with control non-specific siRNA (Qiagen, Hilden, Germany), p50 siRNA (Santa Cruz Biotechnology, Santa Cruz, California, USA), or p65 siRNA (Santa Cruz Biotechnology) using Lipofectamine 2000. The reduced levels of p50 or p65 mRNA expression induced by transfection of each siRNA were determined using siRNA specific primers (Santa Cruz Biotechnology).
Protein extraction and western blot analysis
Protein extraction and western blot analysis were performed as previously previously.6 The membranes were incubated with anti-KLF4 (1:200; Abnova, Taipei, Taiwan), anti-p50 (1:200; Santa Cruz Biotechnology), anti-p65 (1:200; Santa Cruz Biotechnology), or anti-β-actin (1:3000; Sigma Chemicals) antibodies, followed by horseradish-peroxidase-conjugated anti-mouse or anti-rabbit immunoglobulin (DAKO).
Immunofluorescence cytochemistry
Immunofluorescence cytochemistry was performed as previously described.6 The cells were labelled with anti-KLF4 (1:100), anti-Cdx2 (1:50; BioGenex, San Ramon, California, USA), anti-MUC2 (1:100; Santa Cruz Biotechnology), anti-p50 (1:100), anti-p65 (1:100), and anti-cytokeratin (CK) 20 (1:200; Santa Cruz Biotechnology) antibodies. Binding of the primary antibodies was detected using FITC-conjugated anti-mouse, anti-rabbit, or anti-goat immunoglobulin, or rhodamine-conjugated anti-mouse or anti-rabbit immunoglobulin (DAKO). The cells were nuclear counter-stained with 4′,6-diamidino-2′-phenylindole dihydrochloride (DAPI) (Pierce Biotechnology, Rockford, Illinois, USA).
Chromatin immunoprecipitation
Chromatin immunoprecipitation (ChIP) analysis was performed using an EpiQuik Chromatin Immunoprecipitation Kit (Epigentek Group, Brooklyn, New York, USA), according to the method reported by D'Amico et al.18 Het-1A cells were transiently transfected with a KLF4 promoter vector and stimulated with a bile acids mixture for 3 h, after which ChIP analysis was performed. Also, Het-1A cells were transiently transfected with a KLF4 promoter vector, Cdx2 promoter vector, MUC2 promoter vector and KLF4 expression vector, Cdx2 expression vector, or empty vector, after which ChiP analysis was performed. Total DNA prior to immunoprecipitation was used as the input value. Chromatin was immunoprecipitated with anti-KLF4, anti-Cdx2, anti-p50, and anti-p65 antibodies, or IgG as a negative control. Immunoprecipitated DNA–protein complexes were isolated and a real-time PCR assay was then performed using the primers described in table 1.
Statistical analysis
All data are expressed as the mean ± SEM. Multiple comparisons were performed with ANOVA, followed by a Dunnett test. Statistical comparisons between two groups were done with a Mann–Whitney U test. p values less than 0.05 were considered to be statistically significant.
Results
Expression of KLF4 mRNA in adult rat tissues
KLF4 mRNA was found to be expressed throughout the gastrointestinal tract of adult rats, with high levels of expression observed in the jejunum, ileum, proximal, and distal colon, while a lower level of expression was found in the oesophagus (figure 1A).
Immunohistochemistry examinations of rat Barrett's epithelium
Six months after the procedure, columnar-lined epithelia consisting of absorptive cells and goblet cells were observed above the oesophageal–jejunostomy in the rats. KLF4-positive cells with nuclear staining were observed in the columnar epithelia above the oesophageal–jejunostomy, mainly in the surface villi, whereas there was a small number of cells in the crypts (figure 1B).
Expression of KLF4 mRNA in human normal oesophagus and Barrett's oesophagus
We also determined KLF4 mRNA expression levels in endoscopic biopsy specimens of normal oesophagus and Barrett's oesophagus obtained from the human subjects. KLF4 mRNA expression levels of Barrett's oesophagus were significantly higher than those of normal squamous epithelium (figure 1C).
Effects of bile acids on KLF4 promoter activity
The bile acids mixture had a stimulatory effect on KLF4 promoter activity in a dose-dependent manner, with an approximately twofold increase in transcriptional activation in Het-1A, OE33, and OE19 cells (figure 2A,B,C). We constructed a series of reporter plasmids containing different lengths of the KLF4 promoter. The transcriptional activity of these constructs was analysed in Het-1A cells with 200 μm of the bile acids mixture. The plasmid pKLF4/1080-Luc exhibited a similar level of activation by the bile acids mixture as that shown by pKLF4/1700-Luc. However, the plasmids pKLF4/425-Luc, pKLF4/233-Luc, and pGL3-basic without the KLF4 promoter showed no activation response to bile acid stimulation (figure 2D). These results revealed that bile acid-induced activation of the KLF4-promoter is controlled by a site located between −1155 and −460. To determine whether bile acids function as a direct transcriptional activator of KLF4, we examined the KLF4 promoter region for the putative NF-κB binding site using the computational program TESS (http://www.cbil.upenn.edu/). We identified a putative NF-κB binding site from −605 to −596 (ggcagttccc) and speculated that bile acids might bind to the KLF4 promoter in this region. Therefore, to investigate the role of the NF-κB site following bile acid-induced stimulation of KLF4 expression, the element of the putative NF-κB binding site was mutated, which completely abolished bile acid-induced activation of the KLF4 promoter (figure 2D).
To confirm whether bile acids bind to the KLF4 promoter, a ChIP assay was performed using Het-1A cells. Real-time PCR analysis was performed to amplify the promoter region of KLF4 from −650 to −424 that contains the NF-κB binding site. The amount of transcript in the bile acid-treated samples was significantly higher than that in the vehicle-treated samples of DNA immunoprecipitated with the anti-p50 and anti-p65 antibodies (figure 2E).
Direct effects of bile acids on KLF4 mRNA and protein expressions in oesophageal epithelial cells
To determine whether bile acids augment KLF4 mRNA expression, we investigated the direct effect of a bile acids mixture on KLF4 mRNA expression using Het-1A and OE33 cells, and found that the bile acids augmented KLF4 mRNA expression in a dose-dependent manner (figure 3A,B). We also investigated the direct effect of the bile acids mixture on KLF4 expression using primary cultured oesophageal keratinocytes, and found that bile acids augmented KLF4 mRNA and protein expression in a dose-dependent manner (figure 3C,D). When examined using immunofluorescence cytochemistry, KLF4 protein expression was augmented with positive nuclear staining by the bile acids mixture (figure 3E). Furthermore, we investigated the direct effects of the bile acids mixture on p50 and p65 protein expressions. The results of western blotting analysis revealed that the bile acids augmented p50 and p65 protein expressions (figure 3D). In addition, p50 and p65 nuclear translocations were shown by immunofluorescence cytochemistry following addition of the bile acids mixture (figure 3E).
To determine whether KLF4 induction by bile acids occurs via NF-κB activation, we used an siRNA approach with primary cultured cells. p50 and p65 mRNA expression was significantly decreased in p50 siRNA- and p65 siRNA-transfected samples, respectively (data not shown). Furthermore, KLF4 mRNA expression was significantly decreased in cells transfected with specific p50 and p65 siRNAs as compared to the control non-specific siRNA transfected cells following bile acid treatment (figure 3F).
Homologous auto-regulations of KLF4
The specificity of the KLF4 expression vector was confirmed by western blot analysis (supplementary figure 1). Transfection of the KLF4 expression vector into Het-1A cells increased KLF4 promoter activity in a dose-dependent manner (figure 4A). We also constructed a series of reporter plasmids containing different lengths of the KLF4 promoter. The plasmid pKLF4/1080-Luc exhibited a level of activation following stimulation with the KLF4 expression vector similar to the activation shown by pKLF4/1700-Luc. However, the plasmids pKLF4/425-Luc and pKLF4/233-Luc, as well as PGL3-basic without the KLF4 promoter showed no activation response to KLF4 stimulation (figure 4B). These results revealed that KLF4-induced activation of the KLF4-promoter is controlled by a site located between −1115 and −460.
A previous report indicated that the KLF4 promoter has three GC boxes that bind to KLF4.14 Therefore, to confirm whether KLF4 binds to the KLF4 promoter, a ChIP assay was performed using Het-1A cells. Real-time PCR analysis was performed to amplify the promoter region of KLF4 from −781 to −614, which contains the three GC boxes. The amount of transcript from the KLF4 transfected cells was significantly greater than that from the empty vector transfected cells (figure 4C).
Heterologous inter-regulation mechanism of Cdx2 stimulated by KLF4
Transfection of the KLF4 expression vector into Het-1A cells increased Cdx2 promoter activity in a dose-dependent manner (figure 5A). We constructed a series of reporter plasmids containing different lengths of the Cdx2 promoter. The plasmids pCdx2/1014-Luc, pCdx2/631-Luc, pCdx2/438-Luc, pCdx2/319-Luc, and pCdx2/219-Luc exhibited a level of activation following stimulation with a KLF4 expression vector that was similar to the activation shown by pCdx2/1541-Luc. However, the plasmids pCdx2/74-Luc and PGL3-basic without the Cdx2 promoter showed no activation responses to KLF4 stimulation (figure 5B). These results revealed that KLF4-induced activation of the KLF4-promoter is controlled by a site located between −94 and +52. We identified a putative Sp-1 binding site from −91 to −82 (tccccgcctct) and speculated that KLF4 might bind to a Cdx2 promoter in this region. Therefore, to investigate the role of the Sp-1 site in KLF4-induced stimulation of Cdx2 expression, that element of the putative Sp-1-binding site was mutated, which completely abolished the KLF4-induced activation of the Cdx2 promoter (figure 5B).
To confirm whether KLF4 binds to the Cdx2 promoter, a ChIP assay was performed using Het-1A cells. Real-time PCR analysis was performed to amplify the region of the Cdx2 promoter from −194 to −48 that contains the Sp-1 binding site. The amount of transcript from the cells transfected with the KLF4 expression vector was significantly higher than that from cells transfected with an empty vector (figure 5C).
Finally, Cdx2 protein expression following stimulation with a KLF4 expression vector was evaluated in Het-1A cells using immunofluorescence cytochemistry. Cells transfected with the KLF4 construct were found to express the Cdx2 transcript (figure 5D).
Heterologous inter-regulation mechanism of KLF4 stimulated by Cdx2
Transfection of the Cdx2 expression vector into Het-1A cells increased KLF4 promoter activity in a dose-dependent manner (figure 6A). We also constructed a series of reporter plasmids containing different lengths of the KLF4 promoter. The plasmids pKLF4/1080-Luc, pKLF4/425-Luc, and pKLF4/233-Luc exhibited a level of activation following stimulation with the Cdx2 expression vector similar to the activation shown by pKLF4/1541-Luc. However, the plasmids pKLF4/35-Luc and PGL3-basic without the KLF4 promoter showed no activation response to Cdx2 stimulation (figure 6B). These results revealed that Cdx2-induced activation of the KLF4-promoter is controlled by a site located between −268 and −70. We identified multiple putative Cdx2 binding sites and concluded that Cdx2 might bind to the KLF4 promoter in these regions. To confirm whether Cdx2 binds to the KLF4 promoter, a ChIP assay was performed with Het-1A cells. Real-time PCR analysis was performed to amplify the region of the KLF4 promoter from −259 to −56, which contains multiple Cdx2 binding sites. The amount of transcript from the cells transfected with the Cdx2 expression vector was significantly higher than that from cells transfected with an empty vector (figure 6C).
Next, KLF4 protein expression following stimulation with a Cdx2 expression vector was evaluated in Het-1A cells using immunofluorescence cytochemistry and cells transfected with the Cdx2 construct were shown to express the KLF4 transcript (figure 6D).
Effects of Cdx2 or KLF4 over-expression on oesophageal epithelial cells
Transfection of the Cdx2 expression vector into Het-1A cells increased MUC2 promoter activity in a dose-dependent manner (figure 7A). Transfection of the KLF4 expression vector into Het-1A cells increased MUC2 promoter activity in a dose-dependent manner (figure 7B). We also constructed a series of reporter plasmids containing different lengths of the MUC2 promoter. The plasmids pMUC2/823-Luc, pMUC2/463-Luc, pMUC2/214-Luc, and MUC2/80-Luc exhibited a level of activation following stimulation with the KLF4 expression vector that was similar to the activation shown by pMUC2/1750-Luc. However, the plasmids pMUC2/39-Luc and PGL3-basic without the MUC2 promoter showed no activation response to KLF4 stimulation (figure 7C). These results revealed that KLF4-induced activation of the MUC2-promoter is controlled by a site located between −134 and −93. We identified a putative CACCC/Sp-1 binding site from −113 to −101 (gccccacccaccc) and speculated that KLF4 might bind to the MUC2 promoter in this region. Therefore, to investigate the role of the CACCC/Sp-1 site in KLF4-induced stimulation of MUC2 expression, the element of the putative CACCC/Sp-1-binding site was mutated, which completely abolished KLF4-induced activation of the MUC2 promoter (figure 7C). To confirm whether KLF4 binds to the MUC2 promoter, a ChIP assay was performed with Het-1A cells. Real-time PCR analysis was performed to amplify the region of the MUC2 promoter from −304 to −84 that contains the CACCC/Sp-1 binding site. The amount of transcript from the cells transfected with the KLF4 expression vector was significantly higher than that from the cells transfected with an empty vector (figure 7D). We also transfected a KLF4 expression vector into Het-1A cells and observed the expression of intestine specific MUC2 mRNA in those cells. After 48 h, Het-1A cells transfected with the KLF4 expression construct induced MUC2 mRNA expression (figure 7E). Next, MUC2 protein expression following stimulation with a KLF4 expression vector was evaluated in Het-1A cells using immunofluorescence cytochemistry and those transfected with the KLF4 construct were found to express the MUC2 transcript (figure 8A). In addition, cells transfected with the KLF4 expression vector induced CK20 expression (figure 8B).
Discussion
The results of the present experiments suggest that KLF4 is an important molecular mediator in the development of Barrett's epithelium. This is the first known study to investigate the role of KLF4 expression induced by bile acids in development of the disease. Other studies have found that KLF4 gene expression has characteristic tissue distribution, with its expression noted in epithelial cells of the gut, skin, and tongue, as well as several other organs.8 9 However, it has not been fully revealed whether KLF4 is expressed in the oesophagus. Herein, we examined the expression level of KLF4 in the oesophagus, and found a lower level as compared to its expression in the small and large intestines, similar to Cdx2. The roles of KLF4 and Cdx2 in the development and carcinogenesis of the intestinal mucosa have been reported to be similar,10 12 thus we examined the role of KLF4 in Barrett's epithelium development in our study.
First, we determined whether bile acids can induce the expression of KLF4 in vivo using Barrett's epithelium formed in model rats with an oesophago-jejunal anastomosis. KLF4 expression was observed in rat Barrett's epithelium, thus bile acids are suggested to be inducers of KLF4. However, the expression pattern was found to be different from that of Cdx2, as KLF4-positive cells were observed mainly in the surface villi, whereas there was only a small number of those cells in the crypts. In contrast, Cdx2-positive cells were reported to be abundant in both surface villi and crypts of columnar glands.6 19 20 Although there is a possibility that many different types of cells are present in whole biopsy samples, our results from examinations of human tissues suggest that KLF4 expression is related to the development of Barrett's oesophagus.
Next, we investigated whether bile acids induce KLF4 expression in oesophageal epithelium in vitro. In patients with Barrett's oesophagus, the most common bile acids found in refluxant are cholic acid, glycocholic acid and taurocholic acid.21 Since a mixture of bile acids is considered to provide physiological stimulation,7 21 we investigated the changes in KLF4 promoter activity following stimulation with such a mixture and found that the bile acids stimulated KLF4 promoter activity in three types of oesophageal cells. Furthermore, as expected, KLF4 mRNA and protein expressions in oesophageal epithelial cells were augmented by treatment with the bile acids mixture.
Certain bile acids have been reported to be potent activators of NF-κB sites of the promoters of several important proteins, including Cdx1 and Cdx2.6 22 23 The NF-κB family is comprised of several members that interact as homodimers or heterodimers, which function as key regulators of both developmental and pathologic processes. Indeed, deoxycholic acid induced NF-κB subunit p50 nuclear translocation and binding to these sites of the Cdx2 promoter.23 It was also proposed that the Cdx2 promoter is positively regulated by p50 homodimers, whereas it is negatively regulated by p50–p65 heterodimers.24 Furthermore, we and others have suggested that Cdx1 is also an important molecular mediator of Barrett's metaplasia, and that bile acids stimulate Cdx1 expression by upregulation of p65.22 The present results indicate that the KLF4 promoter is positively regulated by p50 and p65 heterodimers following exposure to bile acids in rat primary cultured keratinocytes. Since activation of the NF-kB pathway is cell-type specific25 it would be interesting to also evaluate the effects of bile acids on KLF4 expression in cell lines derived from Barrett's oesophagus.
Some homeobox genes, including Cdx1 and Cdx2, have been shown to positively regulate their own expression, as the Cdx promoter has multiple Cdx responsive elements.7 Notably, mechanisms similar to those of Cdx1 and Cdx2 were revealed in regard to KLF4 expression induced by bile acids. KLF4 can be a transcriptional activator or repressor, and it binds to a similar DNA sequence that has either a CACCC homology or is rich in GC contents.14 26 Therefore, we examined whether KLF4 binds to the KLF4 promoter in esophageal epithelial cells. Transfection of the KLF4 expression vector into Het-1A cells increased KLF4 promoter activity and, using a number of deletion constructs of the KLF4 promoter, we confirmed that KLF4 is capable of transactivating the promoter of its own gene through three closely spaced GC boxes within the promoter. Once the expressions of KLF4 are positively regulated by bile acids, even if the induction level is low, the self-replication mechanism induces a higher expression of KLF4.
In the next step, we investigated the heterologous inter-regulation mechanism between KLF4 and Cdx2. Prior studies that utilised CHO and RKO cells revealed that Cdx2 induces KLF4 promoter activation via Cdx responsive elements within the KLF4 promoter.14 15 In the present study, transfection of the KLF4 expression vector into Het-1A cells increased the promoter activity of Cdx2 and induced production of Cdx2 protein. Using a number of deletion and mutation constructs, we also revealed that KLF4 protein binds to Sp-1 responsive elements of the Cdx2 promoter. Furthermore, transfection of the Cdx2 expression vector into Het-1A cells increased KLF4 promoter activity and induced KLF4 protein. Taken together, bile acids augment KLF4 expression, and contribute to induce auto- and inter-regulation mechanisms between KLF4 and Cdx2.
KLF4 is known to be a direct transcriptional activator of the intestine specific gene IALP.27 In addition, colonic goblet cells throughout colonic epithelia in KLF4−/− mice were found to have reduced expression of MUC2.11 Therefore, we examined whether over-expression of KLF4 in Het-1A cells could trigger their trans-differentiation to intestinal type columnar epithelial cells. Using a number of deletion and mutation constructs, we revealed that KLF4 protein binds to CACCC/Sp-1 responsive elements of the MUC2 promoter. As expected, Het-1A cells transfected with the KLF4 expression vector induced MUC2 mRNA and protein expressions. Furthermore, Het-1A cells transfected with the KLF4 expression vector induced columnar marker CK20. However, KLF4 is a weak inducer of MUC2 as compared to Cdx2, as activation of the MUC2 promoter by a Cdx2 expression vector caused an approximately 10-fold increase, whereas that induced by a KLF4 expression vector was approximately threefold. KLF4 also contributes to induction of an inter-regulation network with Cdx2, and directly and indirectly stimulates cellular trans-differentiation into intestinal metaplasia. Taken together with our previous findings6 7 these results indicate that over-expression of transcription factors, including KLF4, Cdx2 and Cdx1, induced by bile acids may change the phenotype of oesophageal stem cells into columnar cells (figure 9).
In conclusion, we found that induction of KLF4 expression in oesophageal keratinocytes in response to bile acids has important functions in the induction of metaplastic changes during Barrett's epithelium development. In addition, our results revealed that the transcriptional network related to KLF4 and Cdx2 has important roles in development of this disease.
References
Supplementary materials
online only appendix
Online only appendix
Footnotes
Funding supported in part by grants-in-aid for Scientific Research from the Ministry of Education, Science, Sports, and Culture of Japan.
Competing interests None.
Ethics approval All rat experimental protocols were approved by the institutional animal care and experimental committee of Shimane University. All human experimental protocols were approved by the ethics committee of Shimane University.
Provenance and peer review Not commissioned; externally peer reviewed.