Bacteria Regulate Intestinal Epithelial Cell Differentiation Factors Both In Vitro and In Vivo

Background The human colon harbours a plethora of bacteria known to broadly impact on mucosal metabolism and function and thought to be involved in inflammatory bowel disease pathogenesis and colon cancer development. In this report, we investigated the effect of colonic bacteria on epithelial cell differentiation factors in vitro and in vivo. As key transcription factors we focused on Hes1, known to direct towards an absorptive cell fate, Hath1 and KLF4, which govern goblet cell. Methods Expression of the transcription factors Hes1, Hath1 and KLF4, the mucins Muc1 and Muc2 and the defensin HBD2 were measured by real-time PCR in LS174T cells following incubation with several heat-inactivated E. coli strains, including the probiotic E. coli Nissle 1917+/− flagellin, Lactobacilli and Bifidobacteria. For protein detection Western blot experiments and chamber-slide immunostaining were performed. Finally, mRNA and protein expression of these factors was evaluated in the colon of germfree vs. specific pathogen free vs. conventionalized mice and colonic goblet cells were counted. Results Expression of Hes1 and Hath1, and to a minor degree also of KLF4, was reduced by E. coli K-12 and E. coli Nissle 1917. In contrast, Muc1 and HBD2 expression were significantly enhanced, independent of the Notch signalling pathway. Probiotic E. coli Nissle 1917 regulated Hes1, Hath1, Muc1 and HBD2 through flagellin. In vivo experiments confirmed the observed in vitro effects of bacteria by a diminished colonic expression of Hath1 and KLF4 in specific pathogen free and conventionalized mice as compared to germ free mice whereas the number of goblet cells was unchanged in these mice. Conclusions Intestinal bacteria influence the intestinal epithelial differentiation factors Hes1, Hath1 and KLF4, as well as Muc1 and HBD2, in vitro and in vivo. The induction of Muc1 and HBD2 seems to be triggered directly by bacteria and not by Notch.


Introduction
The colon provides the most favorable conditions for intestinal microbiota and harbors, with approximately 10 12 microorganisms per gram of intestinal content, the most densely populated and complex community of the human gastrointestinal tract [1,2]. During evolution a complex and intensive mutualistic relationship between bacteria and host has developed. The intestinal microflora influences the host in different ways by modulating the immune system, protecting against pathogen invasion and attachment, and contributing to digestion and nutritional uptake [3].
In a healthy gut the synergistic co-existence of intestinal microflora and the host is secured by an intact mucosal barrier. The barrier is provided by the intestinal epithelium, consisting of absorptive, goblet, Paneth and neuroendocrine cells, separating the intestinal wall from the luminal microbes. Goblet cells secrete mucins, e.g. Muc1 and Muc2 as structural proteins of the protective mucus layer covering the whole gastrointestinal tract [4]. Moreover, epithelial cells produce broad-spectrum antimicrobial peptides, including defensins [5][6][7][8]. Once secreted, the small cationic defensins are fixed in the negatively charged mucus [9]. This mucus barrier is the first front of gut defence shielding the intestinal wall from luminal microbiota.
The intestinal epithelium differentiates from multipotent stem cells located at the bottom of the crypt and undergoes a rapid and continuous regeneration [10]. This process is regulated by a complex network of different differentiation signals. For example, the early determination of secretory versus absorptive cells is regulated by antagonistic interplay of the Notch target gene Hes1 and the basic helix-loop-helix transcription factor Hath1. In progenitor cells expressing Hes1, Hath1 gene expression is blocked, directing the cells to the absorptive fate. In contrast, in progenitors with inactive Notch/Hes1 signaling, the Hath1 gene can be transcriptionally activated and these cells transit to the secretory lineage [11,12]. The predetermined cells of the secretory line require additional signals for differentiation to specific cell types such as, for goblet cells, the zinc-finger transcription factor KLF4 [13].
Dysregulation of this regulatory network may lead to defective epithelial differentiation and finally to altered function of the mucosal barrier as shown in both forms of inflammatory bowel diseases (IBD), Crohn's disease (CD) and ulcerative colitis (UC) [14,15]. There is mounting evidence that the commensal intestinal microbiota plays a key role in the pathogenesis of IBD [16]. Among other evidence, this is underlined by the observation that patients with IBD have more mucosa-adherent bacteria, some of which are even found intracellularly [17,18]. Moreover, recent studies linked intestinal epithelial differentiation to IBD development. For instance, ileal CD is associated with defective Wnt mediated Paneth cell differentiation and consequently with a diminished production of the defensins HD5 and HD6 resulting in decreased mucosal antibacterial activity [19,20]. In active UC, defective Hath1 expression [21,22] is associated with a decreased number of mature goblet cells in the upper part of the colonic crypt [21]. As a consequence, mucin synthesis in active UC is defective leading to a diminished mucus layer [23,24]. In both cases, these epithelial differentiation defects may lead to invasion of the luminal bacteria into the mucosa where they could trigger inflammation. Additionally, there is evidence that bacteria could contribute to colon cancer development. For instance, animal models showed carcinogenic properties in some bacterial species  [25,26]. Moreover, patients with colorectal cancer exhibit bacteria adhering to tumor tissue [27] and have indirect evidence of bacterial invasion [28,29]. The aim of the present study was to elucidate whether and how bacteria regulate intestinal epithelial cell differentiation. The effects of microbiota on the expression of epithelial differentiation factors Hath1, KLF4 and Hes1, as well as the mucins Muc1, Muc2 and the defensin HBD2 were analysed in vitro and in vivo. This way, we aimed to get more insight into the complex interplay between bacteria and differentiation with regard to the potential impact of the microbiota on intestinal inflammation and cancer development.

Cell Culture Experiments
The colon adenocarcinoma cell line LS174T (American Type Culture Collection, Manassas, USA) was cultivated in Dulbecco's modified Eagle medium (DMEM, Gibco Life Technologies, Eggenstein, Germany) in a humidified atmosphere at 37uC and 5% CO 2 . 10% fetal calf serum (FCS, PAA Laboratories, Pasching, Austria), 1% non-essential amino acids (Gibco Life Technologies), 1% penicillin/streptomycin (Gibco Life Technologies) and 1% sodium pyruvate (Gibco Life Technologies) was added. For experiments, cells were seeded in 12-well culture plates (Becton Dickinson, Franklin Lakes, New Jersey, USA) at a density of 0.65610 6 per well and grown to about 70% confluence. Then cells were washed with phosphate-buffered saline (PBS, Gibco Life Technologies) and incubated in FCS-and antibiotic-free DMEM for 12 hours.
To investigate the possible role of several bacteria in the regulation of epithelial differentiation, LS174T cells were treated with heat-inactivated E. coli strains Symbioflor G1, G2 and G3, Escherichia coli K-12, E. coli Nissle 1917, Lactobacillus fermentum and acidophilus, Bifidobacterium longum, breve and adolescentis as well as Bacteroides vulgatus for 3 and 12 hours. LS174T cells were also incubated for 3 hours with E. coli Nissle 1917 wild type and E. coli Nissle 1917 mutant strains EcNDfliA, EcNDfliC, EcNDflgE, EcNDfim, EcNDfoc and EcNDcsgBA, which were kindly provided by T. Oelschlaeger (Institute for Molecular Biology of Infection, University of Würzburg, Germany). The characteristics of the used bacterial strains are summarized in table 1. All E. coli strains and B. vulgatus were grown under aerobic, Lactobacilli and Bifidobacteria under anaerobic conditions as described previously [30]. For experiments, bacteria were heat-inactivated in a water bath at 65uC for 1 hour, washed with PBS and adjusted to a density of 3610 8 cells/ml with FCS-and antibiotic-free DMEM.
The possible involvement of Notch signalling was investigated by the treatment of LS174T cells with the c-secretase (Notch) inhibitor dibenzazepine (DBZ, Axon Medchem, Groningen, Netherlands) in a concentration of 1 mM (in 0.1% DMSO in DMEM) for 3, 6, 12 and 24 hours in absence or presence of E. coli Nissle 1917.
After incubation, LS174T cells were rinsed in PBS and mRNA (for PCR measurements) or total protein (for Western blot experiments) was isolated as described below. All cell culture experiments were performed for at least 3 independent times in triplicates.

Mouse Tissue
Colonic samples (mRNA and tissue from whole mouse colonic mucosa) of mice housed in germfree conditions at the University of Ulm, mice housed in specific pathogen free (SPF) conditions at the University of Cologne and germfree mice that were transferred to the SPF facility in Cologne for cohousing with SPF mice for 4 weeks (conventionalized) (all C57Bl/6), were kindly provided by M. Pasparakis (Institute for Genetics, Centre for Molecular Medicine University of Cologne).
All animal procedures were conducted in accordance with European, national and institutional guidelines and protocols and were approved at 06.08.2008 from the ethics committee of Tübingen (Regierungsprä sidium AZ 35/9185.81-3), Research-Nr. 929.

RNA Isolation and Reverse Transcription
Treated LS174T cells were washed with PBS and harvested by scraping. Total RNA from the cell lysates was isolated and DNAse digestion was performed as recommended by the supplier to avoid genomic DNA contamination (RNeasy Mini Kit, Qiagen, Hilden, Germany). Subsequently 1 mg of total RNA was reverse transcribed into cDNA with oligo (dT) primers and 15 U/mg AMV Reverse Transcriptase (Promega, Madison, USA) according to standard procedures. RNA preparations were used for PCR analysis.

Quantitative Real-time Reverse Transcriptase PCR
For mRNA quantification, real-time PCR was performed in a SYBR Green fluorescence temperature cycler (LightCyclerH, Roche Diagnostics, Mannheim, Germany). Single-stranded cDNA (or gene-specific plasmids as controls) corresponding to 10 ng of RNA served as a template for PCR with specific oligonucleotide primer pairs (table 2) as described previously [31]. All primers were checked for specific binding to the sequence of interest using BLAST. Plasmids for each product were synthesized with the TOPO TA Cloning Kit (Invitrogen, Carlsbad, CA, USA) according to the supplier's protocol. PCR-amplified DNA fragments were confirmed by sequencing. The correctly sequenced plasmids were serially diluted for internal standard curves. The mRNA data were normalized to the mRNA of ß-actin.

Protein Preparation
LS174T cells were washed with PBS, harvested by scraping and centrifuged for two times at 1300 rpm for 5 minutes. Whole cell lysates were extracted with lysis buffer containing 20 mM Tris- HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Trition X-100, 25 mM Sodiumpyrophosphat, 1 mM Glycerolphosphat, 1 mM Na 3 VO 4 , 6M Urea and 1% Protease Inhibitor Cocktail (Sigma-Aldrich Chemie GmbH, Steinheim, Germany). Mouse colonic tissue was extracted with lysis buffer (see above) and homogenized by FastPrep instrument. Mouse and cell culture lysates were centrifuged at 13000 rpm at 4uC for 20 minutes and the supernatant was collected. Total protein amount was measured with the Bicinchoninic Acid Protein Assay (Smith) as described previously [32]. Isolated proteins were used for Western blot experiments.

Immunostaining and Goblet Cell Count
For Muc1 and Muc2 staining, LS174T cells were seeded in 2well Chamber Slides (Nalge Nunc International Corp., Naperville, IL, USA) at a density of 0.65610 6 per well. Cells were incubated with E. coli Nissle 1917 for 6 hours, washed with PBS and fixed with 100% ethanol for 10 minutes at 220uC. Then, ethanol was removed and the slides were rinsed for two times with TBST. Immunostaining and visualisation was performed as previously described [33]. Anti-Muc1 antibody (VU4H5, Santa Cruz Biotechnology, Heidelberg, Germany) was diluted 1:20 and Anti-Muc2 (NCL-Muc2, Leica Biosystems Newcastle Ltd, Balliol Business Park West, United Kingdom) antibody 1:200 in DAKO REAL Antibody dilution buffer (Dako, Glostrup, Denmark). Cells were also counterstained with hematoxylin.
The number of goblet cells was determined in sections from mouse colonic tissue (germfree: n = 6, SPF housed: n = 4, conventionalized: n = 4) following a standard Alcian Blue staining by blindly counting the Alcian Blue positive vacuoles in a total of 10 crypts per mice.

Statistics
Quantitative real-time PCR and Western blot results were analysed using the Mann-Whitney test. Values of p,0.05 were considered to be statistically significant. All statistical analyses were performed and all graphs were generated with the GraphPad Prism version 5.0 software. Data are presented as means with standard error of the mean (SEM).
To clarify whether the effects on HBD2 and Muc1 expression are caused by bacterial treatment or indirectly by changes in Hes1 and Hath1 expression, we blocked the Notch pathway in LS174T cells using the gamma-secretase inhibitor DBZ up to 24 hours with and without E. coli Nissle. The DBZ treatment led to a strong downregulation of Hes1 (3 h: 0.34-fold, p = 0.01; 6 h: 0.11-fold, p = 0.0003; 12 h: 0.09-fold, p,0.0001 and 24 h: 0.11-fold, p = 0.001) followed by a delayed Hath1 upregulation (3 h: 0.93fold, n.s.; 6 h: 1.19-fold, p = 0.0206; 12 h: 2.01-fold, p = 0,0297 and 24 h: 2.44-fold, p = 0.0032), without affecting HBD2, Muc1 or Muc2 expression. Thus, no significant differences in mRNA expression of these products in untreated and DBZ treated cells as well as no differences between E. coli Nissle only and E. coli Nissle+DBZ treated cells were observed (Fig. S3 and S4) (Fig. 5A-D). These results clearly demonstrate that the flagellin of E. coli Nissle 1917 is essential for regulating Hes1, Hath1, HBD2 and Muc1 mRNA expression.

Microbiota also Regulate Hes1, Math1 and KLF4 in vivo
As our in vitro observations suggested an effect of bacteria on intestinal epithelial cell differentiation, we next evaluated the in vivo relevance of our findings by using germfree animals to assess the role of the intestinal microflora in the regulation of epithelial cell differentiation in mice. Therefore, mRNA expression of mHes1, Math1 (the mouse homolog of Hath1), mKLF4, mMuc1 and mMuc2 was analyzed in the colon of germfree mice as compared to mice reared with specific pathogen free (SPF) intestinal microflora (m…mouse). In the colon of SPF mice, mRNA expression of mHes1 (0.77-fold, p = 0.04), Math1 (0.58fold, p = 0.006) and mKLF4 (0.73-fold, p = 0.011) was lower in comparison to germfree mice ( Fig. 6A-C). These results are consistent with our in vitro observations suggesting an inhibitory effect of bacteria on expression of these differentiation genes. In order to further evaluate this, we also analyzed colons from germfree mice that regained their microbiota by cohousing with SPF mice for 4 weeks. These conventionalised mice showed even more decreased mRNA levels for mHes1 (0.51-fold p = 0.006), Math1 (0.48-fold, p = 0.006) and mKLF4 (0.57-fold p = 0.011) as compared to the germfree mice ( Fig. 6A-C), again confirming the inhibitory effect of the microbiota on expression of these differentiation factors. Surprisingly however, this microbiota effect was not reflected in the expression levels of mMuc1 (Fig. S6A) and mMuc2 (Fig. S6B) since they were not significantly altered in germfree versus colonized mice.
In accordance with the mRNA data we also found a diminished colonic Math1 and mKLF4 protein expression by Western blot analysis in SPF housed and conventionalized mice as compared to the germfree animals (Fig. 7). In contrast, mHes1 protein seems to be unchanged in these three groups (Fig. 7). However, the diminished expression of the goblet cell differentiation markers Math1 and mKLF4 does not lead to a reduced number of goblet cells as shown in Fig. 8.

Discussion
The current study focused on the regulatory effects of intestinal bacteria on the expression of the intestinal epithelial differentiation factors Hes1, Hath1 and KLF4. Moreover the bacterial effects on mucins Muc1 and Muc2, as well as the defensin HBD2 were investigated as well.
We found a bacterial regulation of the transcription factors Hes1, Hath1 and KLF4 in the colon adenocarcinoma cell line LS174T, especially by E. coli K-12 and E. coli Nissle 1917. These changes in mRNA expression were confirmed for Hes1 and Hath1 (and also in trend for KLF4) on the protein level by Western blot experiments following stimulation with E. coli Nissle 1917. Notably, in case of Hes1, a higher molecular weight band appeared following treatment with E. coli Nissle 1917. This double band did not occur in the mouse colon where only a single band was observed, therefore we conclude that the double band is an in vitro phenomenon of cell culture and has likely no physiological relevance in vivo. However, to our knowledge, this is the first study that shows bacteria to regulate these three epithelial differentiation factors in colonic epithelial cells in vitro. Prior observations demonstrated Hes1 to be induced by Porphyromonas gingivalis lipopolysaccharides in the mouse osteoblastic cell line MC3T3E-  1 and in primary mouse bone marrow stromal cells [35]. Moreover Mycobacterium bovis led to increased Hes1 transcripts in peritoneal mice macrophages [36] whereas Salmonella typhimurium causes a decrease of Hes1 expression in PS cells [37]. In case of Hath1, there are no data available concerning the interaction with bacteria at all, whereas KLF4 was shown to be induced in macrophages again by P. gingivalis lipopolysaccharides [38]. In the present study Muc1 expression of LS174T cells was also significantly induced by several bacteria, such as E. coli Nissle 1917 and E. coli K-12, whereas Muc2 expression was unchanged following incubation with all bacteria strains tested. Accordingly, immunostaining on LS174T cells showed a clear induction of Muc1 protein following stimulation with E. coli Nissle 1917 as compared to untreated cells, whereas Muc2 protein was unaffected. Several, in part conflicting studies, focused on the impact of bacteria on the expression profiles of the mucins Muc1 and Muc2 in intestinal epithelial cells. For instance, HT29 cells treated with E. coli Nissle 1917 did not alter the mRNA and protein expression of these two mucins. In contrast, the probiotic cocktail VSL#3 induced Muc2 secretion in HT29 cells [39] but not in LS174T cells [40]. Moreover, L. acidophilus enhanced Muc2 transcripts in HT29 cells [41], whereas another group could not confirm these data [42]. A recent study showed a strong up-regulation of Muc2 in LS174T cells after treatment with flagellin from Salmonella typhimurium [43]. Differences between these prior observations and our data could be explained by the use of different cell lines (e.g. HT-29 vs. LS174T cells) and possibly also by different bacterial preparations (e.g. living vs. heat-inactivated bacteria).
In addition, we found HBD2 transcripts to be upregulated by E. coli Nissle 1917, E. coli K-12, and other bacteria. This is in principle consistent with prior data where HBD2 mRNA was demonstrated to be induced in Caco-2 cells following a treatment with E. coli Nissle 1917, uropathogenic E. coli, as well as L. fermentum, L. acidophilus and VSL#3, but not with E. coli K-12 [30]. Moreover, in LS174T cells the expression of HBD2 was shown to be elevated once treated with E. coli D21, Micrococcus luteus and Salmonella typhimurium [44]. Notably, the effect of L. fermentum on HBD2 expression was clearly higher in Caco-2 than in LS174T cells. This also implies that HBD2 regulation varies in different cell lines.
Overall, cell culture experiments showed a stronger downregulation of the columnar cell differentiation marker Hes1, as compared to the secretory cell differentiation marker Hath1. Therefore, it could be speculated that specific bacteria such as E. coli Nissle 1917 and E. coli K-12 may influence the differentiation of specific cell lineages with a shift towards the goblet cell lineage. Nevertheless, the interplay of underlying mechanisms and the exact consequences of the effects on the differentiation markers need further study.
Previously, the induction of HBD2 by E. coli Nissle 1917 was demonstrated to be dependent on flagellin [34]. Since Hes1, Hath1 and Muc1 were also regulated by E. coli Nissle 1917, we analyzed the role of flagellin with respect to these three factors. In contrast to E. coli Nissle 1917 wild type, Hes1 and Hath1 mRNA was not downregulated by the flagellin mutant strains EcNDfliA, EcNDfliC and EcNDflgE. Accordingly, Muc1 expression was enhanced in E. coli Nissle 1917 wild type, but not in EcNDfliA, EcNDfliC and EcNDflgE. This implies that Hes1, Hath1 and Muc1 are regulated by E. coli Nissle 1917 flagellin, similar to HBD2.
To elucidate the effect of the intestinal microflora in vivo, we analysed the expression of mHes1, Math1 and mKLF4 in the colon of germ free mice compared to SPF and conventionalized mice. Similar to the cell culture data, we observed a significantly lower Math1 and mKLF4 mRNA and protein expression in colonized mice compared to germ free mice, whereas mHes1 expression was reduced on mRNA but not on protein level. This difference in mHes1 expression could be a result of posttranscriptional regulation mechanisms which need further investigations.
Several arguments underline that intestinal bacteria play a crucial role in IBD pathogenesis: Inflammation in IBD is located in areas with a high density of bacteria (mostly colon and/or terminal ileum) [45]; germ free mice do not develop colitis [46]; exposure of fecal stream to the terminal ileum worsen inflammation [47]; antimicrobial peptides are insufficiently expressed in CD, and mutations of human receptors recognizing luminal bacteria, such as NOD2 [48,49] and TLR dysfunction [50,51] are linked to a higher risk of IBD development. Moreover, the intestinal microflora is altered in IBD as compared to healthy controls. Numerous studies described changes in the composition of the microflora between CD, UC and healthy patients [52][53][54], and mucosa-associated and even intracellular bacteria were found in both types of IBD [17,18]. Recent studies showed UC to be associated with goblet cell [21] and ileal CD with Paneth cell differentiation defects [20]. In addition, mice with an epithelialspecific defect leading to reduced Hes1 expression were recently shown to spontaneously develop colitis [55]. Considering these observations, our data suggest that in addition to the genetic predisposition, the luminal microbiota may directly affect epithelial differentiation and its defensive role.
There are also reasons to suggest bacteria to be involved in colon cancer pathogenesis: intestinal cancer is mostly found in the colon, the segment with the highest number of bacteria [56], some bacteria can induce malignancies, e.g. H. pylori and gastric neoplasia [57,58], and, moreover, patients with colon cancer have adherent bacteria [27] as well as more circulating antibodies against specific bacteria (e.g. S. gallolyticus) compared to healthy controls [59]. On the other hand, several studies reported that probiotics, such as L. acidophilus NCFM, suppress carcinogenesis [60,61]. In most colorectal cancers Notch signaling was found to be activated [62,63], whereas Hath1 and KLF4 were decreased [64][65][66][67]. It may be speculated that the downregulation of the important epithelial cell differentiation factors Hath1 and KLF4 could play a role in this regard.
Taken together, the current study shows that intestinal bacteria regulate the epithelial differentiation factors Hes1, Hath1 and KLF4 in vitro and in vivo. This could be involved in IBD and colon cancer pathogenesis although further details remain to be elucidated.