Gluten affects epithelial differentiation-associated genes in small intestinal mucosa of coeliac patients

Abstract

In coeliac disease gluten induces an immunological reaction in genetically susceptible patients, and influences on epithelial cell proliferation and differentiation in the small-bowel mucosa. Our aim was to find novel genes which operate similarly in epithelial proliferation and differentiation in an epithelial cell differentiation model and in coeliac disease patient small-bowel mucosal biopsy samples. The combination of cDNA microarray data originating from a three-dimensional T84 epithelial cell differentiation model and small-bowel mucosal biopsy samples from untreated and treated coeliac disease patients and healthy controls resulted in 30 genes whose mRNA expression was similarly affected. Nine of 30 were located directly or indirectly in the receptor tyrosine kinase pathway starting from the epithelial growth factor receptor. Removal of gluten from the diet resulted in a reversion in the expression of 29 of the 30 genes in the small-bowel mucosal biopsy samples. Further characterization by blotting and labelling revealed increased epidermal growth factor receptor and beta-catenin protein expression in the small-bowel mucosal epithelium in untreated coeliac disease patients compared to healthy controls and treated coeliac patients. We found 30 genes whose mRNA expression was affected similarly in the epithelial cell differentiation model and in the coeliac disease patient small-bowel mucosal biopsy samples. In particular, those genes involved in the epithelial growth factor-mediated signalling pathways may be involved in epithelial cell differentiation and coeliac disease pathogenesis. The epithelial cell differentiation model is a useful tool for studying gene expression changes in the crypt–villus axis.

Introduction

Coeliac disease is a common autoimmune-mediated enteropathy characterized by an immune response to ingested wheat gluten and related prolamins from rye and barley, which leads to small-intestinal mucosal inflammation and morphological damage together with symptoms of malabsorption [1]. The classical gluten-induced small-intestinal lesion shows mucosal villous atrophy with crypt hyperplasia, where decreased differentiation and increased proliferation of epithelial cells are seen [2]. Elimination of gluten from the diet results in clinical improvement and the small-intestinal mucosa heals without scarring, but the disease will relapse if gluten is reintroduced into the diet. The mucosal damage in coeliac disease is considered to be induced by both innate and adaptive immune responses to ingested gluten [3]. However, the mechanisms responsible for the remodelling of the small-intestinal mucosa leading to tissue injury, increased epithelial cell proliferation and failure in differentiation have remained obscure [3].

The small-intestinal mucosa is covered by a single layer of epithelial cells which are descendants of stem cells located in the crypt region. Epithelial cells first proliferate and then begin to migrate along the crypt–villus axis towards the villous tip. During this migration the cells differentiate to absorptive enterocytes. In the upper third of the villi epithelial cells reach terminal differentiation which is coupled with cell cycle arrest and apoptosis [4]. In apoptosis enterocytes form into vacuoles and shrink, chromatin is condensed and finally the cell is exfoliated into the intestinal lumen [5]. The homeostasis and function of the intestinal epithelium are dependent upon an equilibrium between cellular proliferation and loss is important to relation of gut health and oncogenesis [5]. Epithelial cells are attached to the basement membrane, which lies between the epithelium and mesenchymal cells. The lamina propria mesenchymal cells [6] − regulated by various hormonal, paracrine and exogenous factors [6] − control growth, motility and morphogenesis, proliferation and differentiation of epithelial cells by secreting soluble agents such as transforming growth factor β (TGF-β) [7]. Cellular phenotype and differentiation are defined by the genes expressed in individual cells [8].

We have recently developed a three-dimensional epithelial cell differentiation model in which the phenotypically chloride-secreting crypt-like T84 epithelial cells are differentiated to intestinal enterocyte-like cells when cultured in collagen gel with TGF-β [9]. In coeliac disease ingested gluten affects the differentiation of small-intestinal mucosal epithelial cells and causes changes in the mucosal architecture, seen as crypt hyperplasia [2]. In our model the undifferentiated T84 cells are thought to mimic epithelial crypt cells in the small-intestinal mucosa and, similarly, TGF-β-differentiated T84 cells, to mimic epithelial cells in the villous region in the small-intestinal mucosa [9–13].

In present study we sought to identify novel genes whose expression in epithelial cells is induced by gluten in coeliac disease. This was conducted by combining the cDNA microarray data obtained from patient small-bowel mucosal biopsy samples [14] with the experimental epithelial cell differentiation model [13]. Such a comparison seems eligible, as three-dimensionally cultured intestinal epithelial T84 cells have been used successfully as a small-intestinal crypt–villus axis model [9–13]. In addition, we have shown previously a similarity in mRNA and protein expression between the model and the small-intestinal epithelium [11]. In this study, the genes which exhibited significant alteration in their expression in both data sets were chosen for further characterization and validation. The products of these genes were investigated further using both our epithelial cell differentiation model and patient small-bowel biopsy samples to elucidate their role in pathways known to function in epithelial cell proliferation and differentiation.

Materials and methods

Sample material

Patient samples.

Small-intestinal mucosal biopsies for mRNA and protein expression studies were taken from the distal duodenum from coeliac disease patients and control subjects during upper gastrointestinal endoscopic examination at the Department of Gastroenterology and Alimentary Tract Surgery, Tampere University Hospital. Small-bowel mucosal biopsy samples were collected from three sample groups: adults with untreated coeliac disease who had positive serum IgA-class endomysial antibodies (EMA) and severe small-bowel mucosal villous atrophy and crypt hyperplasia, adults with treated coeliac disease who had been on a strict gluten-free diet for at least 1 year, and healthy control subjects (Table 1). Treated coeliacs and healthy controls had normal mucosal villus structure and were EMA-negative. Intestinal samples for cDNA microarray and quantitative reverse transcription–polymerase chain reaction (qRT–PCR) were snap-frozen in liquid nitrogen and immediately processed. Intestinal mucosal biopsy samples for Western blotting were embedded in Tissue Tek (Miles Inc., Elkhart, IN, USA) and frozen. Samples for immunohistochemical studies were embedded in paraffin [14].

Table 1.

Small-bowel mucosal biopsy samples used for cDNA microarray and quantitative real-time polymerase chain reaction (mRNA samples); for Western blotting (protein samples); and immunohistochemistry.

			Age (years)
Sample	Group	No. of patients (female)	Median	Range
mRNA	Untreated CD	4 (2)	39	19–62
	Treated CD	4 (2)	48	19–62
	Healthy control	4 (2)	45	42–57
Western blotting	Untreated CD	3 (3)	23	20–40
	Treated CD	4 (3)	44	32–68
	Healthy control	4 (3)	46	34–54
Immunohistochemistry	Untreated CD	9 (7)	59	28–68
	Treated CD	5 (4)	55	37–70
	Healthy control	5 (4)	47	33–76

CD, coeliac disease.

Epithelial cell differentiation model.

Human intestinal epithelial T84 cells (CCL 2′48; ATCC, Rockville, MD, USA) were cultured three-dimensionally within type I collagen gel as described previously [9]. Briefly, the cells were cultured in Dulbecco's modified Eagle medium and Ham's F-12 (1 : 1) (Gibco brl, Paisley, Scotland, UK) supplemented with 5% heat-inactivated fetal calf serum (Gibco brl) and antibiotics (500 IU/ml penicillin and 100 μg/ml streptomycin; Gibco brl). T84 cells were induced to differentiate by adding 20 ng/ml human recombinant TGF-β1 (hTGF-β1; R&D Systems Europe, Oxon, UK). During the differentiation process epithelial cells form luminal structures which secrete alkaline phosphatase into the lumen [9]. The gene expression in undifferentiated epithelial cells was compared to that of TGF-β-differentiated epithelial cells. Cell cultures were grown for 7 days. All experiments were carried out in triplicate [13].

Isolation of RNA

The mRNA was extracted from biopsy and cell culture samples into ice-cold TRIzol reagent (Invitrogen, Grand Island, NY, USA) according to the manufacturer's protocol. In the case of biopsy samples RNA was extracted as described previously [14]. All samples were subjected to DNAse I (Roche Diagnostics, Indianapolis, IN, USA). Total RNA was quantified by spectrophotometry and quality-checked by agarose gel electrophoresis.

cDNA synthesis and array hybridization

Gene expression was monitored using a Human GeneFilter GF200 (Research Genetics, Huntsville, AL, USA) consisting of 5188 test sequences, 96 control points and 192 housekeeping genes. Arrayed sequences contained both genes with known or predicted function and expressed sequence tags (ESTs) with unknown function. Probe preparation and microarray hybridization were performed following the manufacturer's (Research Genetics) protocol using 1·5 μg total RNA as template, 10 μl (10 mCi/ml) ³³P-deoxycytidine triphosphate (dCTP) (Amersham Pharmacia Biotech UK Limited, Buckinghamshire, UK), a 1·5 μl deoxyribonucleoside [deoxyribonucleoside (dNTP) mix containing deoxyadenosine triphosphate (dATP)], thymidine triphosphate (dTTP), deoxyguanine triphosphate (dGTP) at 20 mM (Finnzymes, Espoo, Finland), a 1·5 μl reverse transcriptase (Life Technologies, Inc., Frederick, MD, USA) and 1·0 μl dithiothreitol (DTT) (Life Technologies). ³³P-labelled cDNAs were synthesized for 90 min at 37°C. ³³P-cDNA products were separated from unincorporated nucleotides by chromatography on Bio-Spin 6 (Bio-Rad, Solna, Sweden). The filter was prehybridized with human 5 μg Cot-1 (Life Technologies, Inc.) DNA and 5 μg poly dA (Research Genetics) in MicroHyb solution (Research Genetics) to minimize non-specific labelling. The denatured ³³P-cDNA was added and incubated for 12 h at 42°C in a roller bottle. The arrays were washed at high stringency [twice in 2× sodium cloride sodium citrate (SSC) 1% sodium dodecyl sulphate (SDS) and 0·5× SSC] and signals were detected from storage phosphor screens by a Storm 860 phosphoimager (Molecular Dynamics, Amersham Biosciences, Buckinghamshire, UK) with 50-micron resolution as described previously [14]. The label was removed from the filter by boiling in 0·5% SDS, the efficiency of removal being ensured with storage phosphor screens by Storm 860 phosphoimager after exposure for 12 h.

Processing and statistical analysis of the microarray data

Microarray filter images were aligned and spot intensities analysed with Pathways Software (Research Genetics). Handling of the raw data, normalization and improvement of fidelity by setting a cut-off value proceeded as described previously [13,14]. In order to eliminate the variation arising from patient or culture conditions, only samples hybridized on the same filter were compared with each other. Furthermore, only genes having the change in the same direction in all sample pairs within a sample group and having a mean ratio of 1·25 were considered significant. By taking into further analysis only those genes which had changed in the same direction in all sample pairs, we sought to concentrate on changes applying to the whole group and not to a single patient or single cell culture experiment. We chose to arrange the affected genes in the form of signalling cascades according to the Kyoto Encyclopedia of Genes and Genomes (KEGG) [15] in homo sapiens (http://www.genome.jp/dbget-bin/) [16–20]. Epithelial growth factor receptor (EGFR) has been proposed recently to be involved in gluten-induced mechanisms [21]. We therefore chose three genes, EGFR, β-catenin and Wiskott–Aldrich syndrome protein family member 1 (WAVE1) along this EGFR signalling network for further characterization and validation.

qRT–PCR

Confirmation of microarray results with qRT–PCR was performed as described previously [14]. Primers were designed with the assistance of the Primer 3 program (available at http://www-genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi) and were chosen according to the requirements presented previously [14]. The primers and probe sequences for the genes are given in Table 2. Thermocycling reaction and crossing-point determination were performed in a LightCycler apparatus using the LightCycler − FastStart DNA Master SYBR Green I Kit (Roche Diagnostics, Mannheim, Germany) as described previously elsewhere [14]. The expression level of EGFR, β-catenin and WAVE1 were normalized by the expression values of the housekeeping gene glyceraldehyde-3-phosphate-dehydrogenase (GAPDH). After PCR, every sample was also run in 1·5% agarose gel electrophoresis to ensure that a correct-sized product was amplified in the reaction.

Table 2.

List of used primers, their temperature and MgCl optimum and length of product.

Name	Direction	Sequence	Size	MgCl μmol	Annealing temp
GAPDH	F	ATG CCA GTG AGC TTC CCG TTC AGC	199	2	70
GAPDH	R	TGG TAT CGT GGA AGG ACT CAT GAC
EGFR	F	TGACTGCTGCCACAACCAGT	230	3	62
EGFR	R	AGCCGTGATCTGTCACCACA
CTNNB1	F	AGAGATGGCCCAGAATGCAG	303	3	62
CTNNB1	R	ATGTGAAGGGCTCCGGTACA
WAVE1	F	GGTCCAGAGCTGGCTGAAGA	250	3	63
WAVE1	R	TCGGTTTTGCATCTCCTGCT

GAPDH: glyceraldehyde-3-phosphate-dehydrogenase; EGFR: epithelial growth factor receptor; CTNN: β-catenin; WAVE1: Wiskott–Aldrich syndrome protein family member 1.

Protein extraction and Western blotting

Protein expressions of the selected proteins were assessed by Western blotting. Proteins from the frozen biopsy samples embedded in Tissue Tek were extracted as described elsewhere [22]. Briefly, the biopsy samples were cut out of the Tissue Tek, placed in ice-cold buffer [50 mM potassium phosphate pH 7·4, 150 mM NaCl, 5 mM ethylenediamine tetraacetic acid (EDTA)] and homogenized with an Ultra Turrax T8 tissuemizer (IKA Labortechnik, Staufen, Germany). The protein concentration was measured by Bio-Rad protein assay (Bio-Rad, Sweden).

All samples were denatured and run in Tris–glycine precast gels (Invitrogen, Paisley, UK). Proteins were blotted according to the manufacturer's protocol (Invitrogen) on a nitrocellulose filter (Hybond C-Extra; Amersham Biosciences Ltd, Little Chalfont, Buckinghamshire, UK). Immunoblotting was performed for three selected proteins: EGFR, detected with polyclonal EGFR antibody (dilution 1 : 750) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), β-catenin, detected with monoclonal β-catenin antibody (1 : 1000) (BD Transduction Laboratories, San Jose, CA, USA) and WAVE1, detected with polyclonal WAVE1 antibody (1 : 750) (Santa Cruz Biotechnology, Inc.) on the basis of changes detected in their expression at mRNA level. β-actin was chosen as internal control to detect the amount of loading (1 : 1500) (monoclonal β-actin antibody; Sigma-Aldrich, St. Louis, MO, USA). Anti-goat horseradish peroxidase (HRP)-conjugate (Dako, Glostrup, Denmark) was used to detect EGFR and WAVE1, and anti-mouse HRP-conjugate (Dako) for β-catenin and β-actin. Blotting was performed as described previously [13]. The relative amounts of EGFR, β-catenin and WAVE1 were calculated from scanned images [grey-scale, 1200 dpi, tagged image file format (TIF)] with an Amersham Image Quant TL (Amersham Biosciences, Piscataway, NJ, USA). The calculated values are from three separate experiments.

Immunohistochemical studies

Paraffin sections from human small-intestinal mucosa biopsy samples were cut and mounted on slides treated with Vectabond TM reagent (Vector Laboratories Inc., Burlingame, CA, USA) and immunostained with monoclonal antibodies against EGFR (at a concentration of 1 mg/ml) (Novocastra Laboratories Ltd, Newcastle upon Tyne, UK) and β-catenin (dilution 1 : 1000) (BD Transduction Laboratories, Lexington, KY, USA), and polyclonal antibody against phospho-β-catenin (1 : 500) (Ser33/37/Thr41) (Cell Signalling Technology Ltd, Danvers, MA, USA) as described elsewhere [13]. The sections were counterstained with haematoxylin.

Statistical analysis

The Microsoft Excel program was used to calculate median and range of patients' ages. Mean value and standard deviation from microarray, qRT–PCR and Western blotting results were also calculated using Microsoft Excel.

Ethical considerations

The Ethical Committee of Tampere University Hospital approved this study protocol, and all subjects gave written informed consent.

Results

Microarray results

A total of 30 genes or ESTs of the 5188 sequences spotted were seen to be affected similarly when data from small-intestinal mucosal biopsy samples from untreated and treated coeliac patients and healthy controls were compared with the epithelial model of undifferentiated and TGF-β-differentiated T84 epithelial cells. It is of note that the expression of 29 of the 30 affected genes was reversed with a gluten-free diet, indicating that the detected expression changes were triggered by gluten. These altered genes, grouped into ontological classes according to their known or predicted functions, were found to belong to seven ontology groups (Table 3). Nine of the 30 commonly affected genes have an impact on cell communication and signal transduction (Table 3, group A), five encode proteins which participate in metabolism (Table 3, group B), four exert effects on cell cycle control and DNA processing (Table 3, group C) and three are associated with transcription and translation (Table 3, group D). Only one gene, WAVE1, known to regulate the actin cytoskeleton, belonged to the gene group which controls cellular organization (Table 3, group E). Lysosomal H⁺transporting ATPase (ATP6V0B) was the only gene which affects transport facilitation (Table 3, group F). Seven of the affected genes had no function known hitherto (Table 3, group G).

Table 3.

List of 30 genes undergoing a significant change in their mRNA expression and with alteration in the same direction both in coeliac disease small-bowel mucosal biopsy samples and in three-dimensional epithelial cell culture. The mRNA expression was detected with the cDNA microarray. Numbers are mean values of expression ratios from three separate experiments.

				Duodenal biopsy samples
Name	Gene ontology class	GenBank number	Chromosomal location	Untreated coeliac versus healthy control	Untreated coeliac versus treated celiac	Treated coeliac versus healthy control	Experimental epithelial cell differentiation model Undifferentiated versus TGF-β differentiated
RIT1: ras-like without CAAX 1	A	AA027840	1q21.3	1·36	1·60		1·42
DGKD: diacylglycerol kinase, delta (130 kDa)	A	AA280691	2q37.1	1·38	1·53		1·76
CTNNB1: catenin (cadherin-associated protein), beta 1, 88 kDa	A	AA442092	3p21	1·93	1·66		1·56
TM4SF1: transmembrane 4 superfamily member 1	A	AA487893	3q21-q25	1·52			1·65
PIK3R1: phosphoinositide-3-kinase, regulatory subunit, polypeptide 1	A	R54050	5q12-q13	1·41	1·54		2·07
EGFR: epidermal growth factor receptor (erythroblastic leukaemia viral (v-erb-b) oncogene homolog, avian)	A	R35665	7p12	1·35	1·26		2·111·36
PKIA: protein kinase (cAMP-dependent, catalytic) inhibitor alpha	A	AA281667	8q21.11	2·08			1·4
RNTRE: related to the N terminus of tre	A	AA281057 AA281137	10p13	1·461·36	1·531·42		1·381·4
EFNB2: ephrin-B2	A	AA461424	13q33	1·4	1·41		1·321·41
DHCR24: 24-dehydrocholesterol reductase	B	AA482324	1p33-p31.1	2·24	2·12		1·58
ALDH7A1: aldehyde dehydrogenase 7 family, member A1	B	AA101299	5q31	1·34	1·53		1·74
HIBADH: 3-hydroxyisobutyrate dehydrogenase	B	N77326	7p15	1·28			1·54
DIO2: Deiodinase, iodothyronine, type II	B	R62242	14q24.2-q24.3	0·71	0·67	1·27	2·75
FTCD: Formiminotransferase cyclodeaminase	B	W00987	21q22.3	1·39			1·32
BAP1: BRCA1 associated protein-1 (ubiquitin carboxy-terminal hydrolase)	C	H09065	3p21	1·39	1·29		1·591·54
MTCP1 X: mature T-cell proliferation 1	C	AA029842	Xq28	1·34			1·32
RLF: rearranged L-myc fusion sequence,Similar to the Zn-15 transcription factor	C	R26070	1p32	1·46	1·48		2·04
ROK1: ATP-dependent RNA helicase	C	W73792	17q21.1	1·29			2·41
ZNF161: zinc finger protein 161	D	AA232647	17q23.3	2·15	2·01		1·491·36
MRPL4: Mitochondrial ribosomal protein L4	D	AA490981	19p13.2	1·69			1·51
SF3A1: splicing factor 3a, subunit 1, 120 kDa	D	T72698	22q12.2	1·25			1·44
WAVE1: WAS protein family, member 1	E	N59851	6q21-q22	1·43	1·31		1·61
ATP6V0B: ATPase, H + transporting, lysosomal 21 kDa, V0 subunit c	F	AA480826	1p32.3	1·26			1·45
C LY6G5B: lymphocyte antigen 6 complex, locus G5B	G	R69566	6p21.3	1·28			1·28
D123: D123 gene product	G	AA448289	10p13	1·65	1·54		1·51
CDNA FLJ43113 fis, clone CTONG2028208	G	R31512	12p13	1·29			1·64
TMEM78B: Transmembrane protein 87A	G	N63753	15q14	1·46	1·34		1·93
FTS: fused toes homologue (mouse)	G	W52803	16q12.1	1·26			1·52
EST: LOC326624: Rab37-like	G	R06033	17q25	1·42			1·67
TMEM50B: transmembrane protein 50B	G	N90335	21q22.11	1·50			1·77

Ontology classes used here are: (A) signal transduction, (B) metabolism (other than energy metabolism), (C) cell cycle control and DNA processing, (D) transcription and translation, (E) control of cellular organization, (F) transport facilitation, (G) genes with unknown function. Italic type: up-regulated mRNA expression; bold type: down-regulated mRNA expression. Empty cells indicate insignificant or no change.

Nine of the 30 significantly altered genes are associated, directly or indirectly, with the EGFR signalling network [16–20] (Fig. 1). These comprised EGFR, the gene related to the N terminus of tre (RN-tre), diacyl glycerol kinase D (DGKD), phosphoinositide-3-kinase regulatory subunit 1 (PI3KR1), β-catenin (CTNNB1), ephrin B2 (EFNB2), WAVE1, fused toes homologue (FTS) and mature T cell proliferation gene (MTCP1) (Fig. 1). In validation of the data we focused particularly on the genes which lay directly or indirectly on the EGFR pathway. These were EGFR itself, β-catenin and WAVE1. The expression and localization of the protein products of these genes were characterized further.

Fig. 1.

Schematic illustration of mRNA expression levels significantly altered both in duodenal biopsy samples from untreated coeliac disease patients and in undifferentiated epithelial T84 cells in the epithelial cell differentiation model. Expression levels of epidermal growth factor receptor (EGFR), related to the N terminus of tre (RN-tre), ephrin B2, Wiskott–Aldrich syndrome protein family member 1 (WAVE1), diacylglycerol kinase D (DGKD), phosphoinositide-3-kinase regulatory subunit 1 (PI3KR1), fused toes homologue (FTS), mature T cell proliferation (MTCP1) and β-catenin were increased in untreated coeliac disease and in undifferentiated cell culture samples. Red colour indicates a significant up-regulation of mRNA expression both in duodenal biopsy samples and in the epithelial cell differentiation model. White indicates genes not mutually altered. Yellow circles indicate the outcome of the pathway. The figure was drawn according to the Kyoto encyclopedia of genes and genomes (KEGG) (http://www.genome.jp/dbget-bin/).

Expression of EGFR

The expression of EGFR mRNA and protein was found to be increased consistently when biopsy samples from untreated coeliac disease small intestine were compared to treated coeliac disease and healthy control samples (Fig. 2). Immunostaining of small-intestinal mucosal biopsy specimens using monoclonal EGFR antibody revealed that apical staining in epithelial cells was more pronounced in the crypt region in untreated (Fig. 3d) compared to treated coeliac disease patients (Fig. 3e) or healthy controls (Fig. 3f). The localization and intensity of the EGFR labelling in the surface epithelium in untreated coeliac disease patients (Fig. 3a) and labelling in the villus region in treated coeliac patients (Fig. 3b) and healthy control samples (Fig. 3c) were comparable with each other. The staining was concentrated more prominently in cell junctions in the villus region and surface epithelium (Fig. 3a–c) than in the crypt region (Fig. 3d–f). In the epithelial cell differentiation model the expression of EGFR was diffuse in undifferentiated cell clusters (Fig. 3h), whereas in TGF-β-differentiated luminal formations the labelling was concentrated more on the apical side of the differentiated cells (Fig. 3g).

Fig. 2.

Levels of epidermal growth factor receptor (EGFR) mRNA [microarray and quantitative real-time polymerase chain reaction (qRT–PCR)] and protein (Western) expression in small-bowel mucosal biopsy samples from untreated coeliac disease (CD) patients, treated CD and healthy controls. Bars indicate the standard deviation of three separate experiments (a). A representative Western blot subjected to densitometric analysis. β-actin was used as loading control (b).

Fig. 3.

Immunostainings of epidermal growth factor receptor (EGFR) in small-bowel mucosal biopsy samples. There is more intense apical EGFR labelling in the crypt area in untreated coeliac patients (d) than in treated coeliac patients (e) or healthy controls (f). In the surface epithelium in untreated coeliac disease patients (a) the staining is similar to the labelling in the villus region in treated coeliac patients (b) or healthy controls (c). The boxes highlight the apical area in epithelial cells where the difference between the samples is most evident. In the epithelial cell differentiation model there was more diffuse labelling in undifferentiated T84 epithelial cells (h) than in transforming growth factor-β-differentiated cells (g).

Expression of β-catenin

Small-intestinal biopsy samples from untreated coeliac disease patients contained more β-catenin mRNA and protein compared to samples from treated coeliac patients and healthy controls (Fig. 4). In the small-bowel mucosa β-catenin was localized in the nuclear region in the proliferative area in the crypts (Fig. 5d–f) in all groups. Nuclear β-catenin staining extended further up in the crypt in untreated (Fig. 5d) than in treated coeliac disease biopsy samples (Fig. 5e) and healthy controls (Fig. 5f). Junctional staining was similar in all groups (Fig. 5a–c) in the upper parts of the crypts and in the villus region. The nuclei were devoid of staining. In the epithelial cell differentiation model there was more nuclear β-catenin label in undifferentiated cell clusters (Fig. 5h) than in TGF-β-differentiated cells (Fig. 5g), where the staining was intense at the junctions of differentiated epithelial cells.

Fig. 4.

Levels of β-catenin mRNA and protein expression determined by microarray, quantitative real-time polymerase chain reaction (qRT–PCR) and Western blotting (a). mRNAs and proteins were extracted from small-bowel mucosal biopsy samples from untreated coeliac disease patients (CD) and treated CD patients and from healthy controls. Bars indicate the standard deviation of three separate experiments. A representative Western blot subjected to densitometric analysis. β-actin was used as loading control (b).

Fig. 5.

Immunostainings with β-catenin in small-bowel mucosal biopsy samples in untreated coeliac disease (a,d), in treated coeliac disease samples (b,e) and healthy controls (c,f). In all biopsy samples β-catenin was localized in the nuclear region (black arrows) in the proliferative area (d,e,f) while in the untreated coeliac biopsy samples (d) positive cells extended further up in the crypt than in treated coeliac disease biopsy samples (e) and healthy controls (f). Open arrows indicate cells showing only junctional staining. In the epithelial cell differentiation model the undifferentiated T84 epithelial cells (h) show more nuclear and junctional labelling than transforming growth factor-β-differentiated cells (g).

Microarray results and immunostaining showed there to be more β-catenin in the untreated coeliac disease biopsy samples than in those from control and treated coeliac disease patients. This prompted us to study whether degradation of β-catenin was also increased. In immunohistochemical staining there were only a few phospho-β-catenin-positive cells in the crypts. However, the crypts in untreated coeliac biopsy samples (Fig. 6a) contained clearly more positive cells than those from treated coeliac patients (Fig. 6b) and healthy controls (Fig. 6c). In the epithelial cell differentiation model, undifferentiated cells contained more phospho-β-catenin label (Fig. 6d) than TGF-β-differentiated cells (Fig. 6e).

Fig. 6.

Immunostainings with phospho-β-catenin in untreated (a) and treated coeliac disease (b) and healthy control (c) small-bowel mucosal samples. There are only few stained cells (indicated with arrows) in all samples. There are more stained cells in the extended crypts (a) in untreated coeliac disease patients than in treated coeliac patients (b) and healthy controls (c). In the epithelial cell differentiation model there is more phospho-β-catenin label in undifferentiated T84 epithelial cells (d) than in TGFβ-differentiated cells (e).

Expression of WAVE1

The expression of WAVE1 mRNA and WAVE1 protein was found to be increased when small-bowel mucosal biopsy samples from untreated coeliac disease were compared to those from treated coeliacs and healthy controls (Fig. 7a). The expression changes in WAVE1 were concordant in mRNA and protein level (Fig. 7a).

Fig. 7.

The expression of WAVE1 mRNA and protein in small-bowel mucosal biopsy samples from untreated coeliac disease patients (CD), treated CD patients and healthy controls, determined by microarray, quantitative real-time polymerase chain reaction (qRT–PCR) and Western blotting. Bars indicate the standard deviation of three separate experiments. A representative Western blot subjected to densitometric analysis. β-actin was used as loading control (b).

Discussion

In the present study we compared gene expression changes in intestinal biopsy samples to the epithelial cell differentiation model. Our aim was to identify which of the genes induced by gluten in coeliac disease patient biopsy samples are expressed by epithelial cells when they are differentiated. In this study we employed the epithelial cell culture system in which TGF-β1 is used as differentiative agent [9]; TGF-β1 is involved the normal development and differentiation of the intestine [23]. Furthermore, TGF-β1 has been associated more recently with some fine regulation of the immune response [24] and innate immunity in the pathogenesis of coeliac disease [25].

Comparison of the data obtained produced a set of 30 genes whose mRNA expression was altered in both patient and cell culture samples. Expression of 29 of the affected genes was normalized during the healing of the small-intestinal mucosa, suggesting that the detected changes were due to the reaction to gluten, and not a primary defect. Interestingly, nine affected genes, EGFR, RN-tre, DGKD, PI3KR1, β-catenin, ephrin B2, WAVE1, FTS and MTCP1, are involved directly or indirectly in the intracellular EGFR signalling pathway [16–20] (Fig. 1). There are four such pathways starting from EGFR: phosphoinositide-3-kinase (PI3K)/serine/threonine kinase protein kinase B (PKB, also known as Akt), c-Src, Ras/MAPK [26,27] and JAK/STAT [27]. As we saw no altered gene expressions on the c-Src, Ras/MAPK or JAK/STAT pathways, we concentrated on the PI3K pathway (Fig. 1).

A recent study [21] by Barone and associates showed that the changes induced by gluten in the cytoskeleton could be inhibited by EGFR inhibitors and EGFR antibodies, suggesting that a peptic tryptic digest of gluten and small gluten peptides elicits EGF-like responses on epithelial cells. Activation of EGFR has been shown to induce intense apical labelling in the airway epithelium [28]. As we found increased amounts of EGFR mRNA and protein in untreated coeliac patient samples, we hypothesized that increased labelling of EGFR on the apical side of epithelial cells in the crypt region in untreated coeliac disease patients might also indicate increased activation of the receptor. The above-mentioned changes seen in the epithelium of untreated coeliac disease patients could be due to increased EGFR activation. Both our results and those of Barone and associates [21] point in that direction. It is conceivable that the increased degree of EGFR activation seen here may be a result of increased amounts of EGFR ligands [29], or molecules which mimic these, increased amounts of receptor or prolonged internalization time of EGFR as a result of up-regulation of RN-tre expression [16,30]. The increased activation might also be a result of loosening of cell junctions [28] as a direct effect of gluten on epithelial cells [31] or interferon (IFN)-γ produced by gluten-activated lymphocytes [32], which again activate EGFR [28]. Further studies are required to elucidate the role of gluten on these pathways.

The wnt signalling pathway has an essential role in maintenance of the stem cell region and in the regulation of proliferation and differentiation [33]. In the intestinal epithelium the absence of wnt signalling has been shown to result in loss of crypts and over-activity lead to adenomatous polyp formation and cancer [34]. Nuclear labelling of β-catenin, seen in the crypts of the small intestine, is a sign of activation of the wnt pathway [34]. In our untreated coeliac disease samples a prominent nuclear localization of β-catenin was observed in the small-intestinal crypts, also reported by Perry and colleagues [35]. This was foreseeable, as in coeliac disease the crypts are known to contain increased numbers of dividing epithelial cells [36]. In Western blotting we saw only slightly increased amounts of β-catenin protein in untreated coeliac disease. In another study the amount of β-catenin in biopsy samples was shown to be similar in both untreated and treated coeliac patients and in healthy controls [37]. We investigated further whether there are also elevated amounts of phospho-β-catenin. Indeed, we observed increased phospho-β-catenin labelling in the crypts in the untreated coeliac disease patients and in undifferentiated epithelial cells in the epithelial cell differentiation model. This would imply that the increased amounts of transcribed β-catenin detected in untreated coeliac disease small-bowel mucosal samples is transported efficiently to destruction. Our result is consistent with those of Ciccocioppo and associates [37], who similarly detected increased amounts of tyrosine phosphorylation of β-catenin in untreated coeliac disease patient mucosal samples.

WAVE1 is known to regulate actin polymerization by activating the ARP2/3 complex, thereby regulating membrane ruffling in migrating fibroblasts [38]. WAVE1 acts downstream from Rac [39], which is regulated by PI3K [20] and EGFR [26]. Gluten peptides and digested gliadin have been shown to induce actin reorganization and especially membrane ruffling [21,40] and this has been proposed to be the EGF-like effect of gluten [21]. As we found increased expression of WAVE1 mRNA and protein in untreated coeliac disease patients, we hypothesized that the changes in the actin cytoskeleton [40] could be induced by activation of Rac1 and WAVE1, as an EGF-like reaction to gluten (Fig. 1).

Altogether 23% of the genes found to be affected in the small-intestinal mucosal epithelium in coeliac disease patients and in undifferentiated epithelial cell culture samples had no functions characterized hitherto. Several of them, for example LY6G5B and CDNA FLJ43113 fis, are reported to be expressed in gastrointestinal tumours, the Rab37-like gene has been reported to be expressed in the small intestine and D123 regulates the cell cycle (http://www.ncbi.nlm.nih.gov/UniGene/). These sequences of unknown function are especially interesting in the search for new players possibly involved in gluten-induced mucosal injury in coeliac disease.

Conclusion

In this study we found 30 genes of known or unknown function whose mRNA expression was affected similarly in the small-intestinal mucosal epithelium in coeliac disease patients and in an in vitro three-dimensional epithelial differentiation model.

Removal of gluten from the ingested diet resulted in reversion in the transcription of 29 of 30 genes in the small-bowel biopsy samples. These genes, especially those associated with EGF-mediated signalling pathways, may be involved in epithelial cell differentiation and coeliac disease pathogenesis. Furthermore, the epithelial cell differentiation model was seen to provide a good tool for studying gene expression changes in the crypt–villus axis in coeliac small-intestinal mucosa.