GATA6 Deficiency Leads to Epithelial Barrier Dysfunction and Enhances Susceptibility to Gut Inflammation

Abstract

Background and Aims

Intestinal barrier dysfunction is a hallmark of inflammatory bowel diseases [IBD], but the mechanisms that lead to such a defect are not fully understood. This study was aimed at characterising the factors involved in the defective barrier function in IBD.

Methods

Transcriptome analysis was performed on colon samples taken from healthy controls [CTR] and IBD patients. Expression of GATA-binding factor 6 [GATA6], a transcription factor involved in intestinal epithelial cell differentiation, was evaluated in colon samples taken from CTR and IBD patients by real-time polymerase chain reaction [PCR] and immunohistochemistry. Intestinal sections of wild-type and Gata6del mice, which exhibit a conditional Gata6 deletion in intestinal epithelial cells and which are either left untreated or receive subcutaneous indomethacin or rectal trinitrobenzene sulphonic acid, were stained with haematoxylin and eosin. In parallel, some Gata6del mice received antibiotics to deplete intestinal flora. Mucosal inflammatory cell infiltration and cytokine production were evaluated by flow cytometry and real-time PCR, respectively, and tight junction proteins were examined by immunofluorescence. Intestinal barrier integrity was assessed by fluorescein isothiocyanate [FITC]-dextran assay.

Results

Multiple genes involved in cell commitment/proliferation and wound healing were differentially expressed in IBD compared with CTR. Among these, GATA6 was significantly decreased in the IBD epithelium compared with CTR. In mice, conditional deletion of GATA6 in the intestinal epithelium induced primarily epithelial damage, diminished zonula occludens-1 expression, and enhanced intestinal permeability, ultimately resulting in bacteria-driven local immune response and enhanced susceptibility to gut inflammation.

Conclusions

Reduced expression of GATA6 promotes intestinal barrier dysfunction, thus amplifying intestinal inflammatory pathology.

Ulcerative colitis, Crohn’s disease, intestinal epithelium

1. Introduction

The mammalian intestine is lined by highly differentiated intestinal epithelial cells [IECs], which along with specialised cells responsible for the secretion of mucus [goblet cells], anti-microbial peptides [Paneth cells], and enteric hormones [enteroendocrine cells], are constantly renewed every 4–5 days through a process of cell division, maturation, and migration.¹ Renewal relies on proliferative cells [stem cells] that reside at or near the base of the crypts of the intestinal glands and have the ability to self-renew and to give rise to rapidly cycling daughter cells, called transit amplifying [TA] cells.^2–4 TA cells undergo a limited number of cell divisions before becoming terminally differentiated cells and migrate up, with the exception of Paneth cells, the crypt-villus axis.^2–4

Altogether, the epithelial cells contribute to form a tight, although not impermeable, barrier between the external luminal environment and the internal immune system, with the task to defend against potentially harmful molecules and microorganisms, while leaving access to essential nutrients and solutes.⁵ Intestinal permeability is also finely regulated by junctional protein complexes [i.e., tight junctions, gap junctions, adherens junctions, and desmosomes], which seal the intercellular space between neighbouring IECs, thus controlling passage of molecules through the paracellular spaces.^6,7 Defects in the epithelial barrier function can lead to excessive invasion of luminal antigens into the subepithelial tissues and induction of unwanted immune reactions, with consequent development of inflammatory pathologies.^8,9 This occurs, for example, in the gut of patients with ulcerative colitis [UC] and patients with Crohn’s disease [CD], the main inflammatory bowel diseases [IBD] in human beings.^8–11 Although it remains unclear whether epithelial barrier dysfunction represents a primary defect in the aetiology of IBD or develops as a consequence of the ongoing mucosal inflammation, several studies in murine models of colitis have shown that changes in expression and/or distribution of tight junction proteins lead to disrupted epithelial barrier function, thereby promoting immune cell-mediated tissue damage.^12–16

However, there is still a gap in our understanding of the molecular mechanisms involved in the control of epithelial barrier function during normal homeostasis and IBD. In this study, we aimed at characterising factors/mechanisms associated with defective barrier function in IBD.

2. Methods

2.1. Patients

Intestinal biopsies were taken from inflamed colonic mucosa of 15 CD patients and 24 UC patients. Seven CD patients were receiving mesalazine, one patient was on salazopyrine, one patient was receiving ustekinumab, five patients were on infliximab, and one patient received no therapy. Twelve UC patients were receiving mesalazine, three patients were on vedolizumab, three patients were on steroids, three patients were on infliximab, and three patients received no therapy. Biopsy samples were also taken from normal colon of 28 subjects undergoing colonoscopy for colon cancer screening. The human studies were approved by the local ethics committee [protocol number: 116/20], and each patient gave written informed consent.

2.2. Mice

Gata6del mice were obtained by crossing VillinCreER^T2-positive mice [kindly provided by S. Vetrano, Milan, Italy] and Gata6^loxP/loxP mice [Jackson Laboratory, Bar Harbor, Maine, USA]. Mice were weaned at 3 weeks of age and put into new cages [randomised to littermates]; therefore, each experimental group contained mice from at least two different litters. Conditional deletion of the Gata6 gene in the intestinal epithelium cells [Villin-positive cells] of Gata6del mice was obtained upon treatment with intraperitoneal injections of tamoxifen [TMX] [1 mg/mouse, #T5648 Sigma, St Louis, MO] for 5 consecutive days. Gata6^loxP/loxP VillinCreER^T2-negative mice served as controls. Mice were then sacrificed 4 weeks after the last injections and tissue harvested. Mice were hosted in the University of Rome ‘Tor Vergata’ animal facility [Rome, Italy] and Plaisant animal facility [Castel Romano, Rome, Italy]. Female mice [6–8 weeks old] were used in all the experiments. In some experiments, 2 weeks after the last injection with TMX, Gata6del mice were exposed to a cocktail of antibiotics [vancomycin #V2002, ampicillin #A0166, neomycin #N1876, and metronidazole #M3761, 1 g/l each, Sigma] for 2 weeks. Mice were then sacrificed at Day 28 and ileum and colon samples collected. All the in vivo experimental procedures were approved by the animal ethics committee according to Italian legislation on animal experimentation [authorisation n° 203/2017-PR] and in compliance with European rules [2010/63/UE].

2.3. Transcriptome analysis

Total RNA was extracted from human colon biopsy samples and murine ileum and colon samples using the RNeasy Mini Kit [Qiagen] and digested with DNase [Qiagen]. Samples with quantified complementary DNA were sequenced in the Microarray Unit of the Consortium for Genomic Technologies [Milan, Italy] by hybridisation to Affymetrix Human GeneChip Clariom S microarrays or Affymetrix Mouse GeneChip Clariom S microarrays, as previously described.¹⁷ Transcripts were selected based on fold change value of 1.5 or higher, which was generated from the comparison between Gata6del ileum vs CTR ileum, Gata6del colon vs CTR colon, UC patients vs healthy controls.

2.4. Experimental gut inflammation

A solution of 1% of 2,4,6-trinitrobenzenesulphonic acid [TNBS, #92822 Sigma] in acetone/olive oil [4:1 ratio] was applied to the shaved skin of Gata6del and control mice [150 µl/mouse] 4 weeks after the last injection with TMX. On Day 7, mice received TNBS intrarectally [1.75 mg/mouse in 35% ethanol] and were then sacrificed after 3 days. Gata6del and control mice were treated with a subcutaneous injection of indomethacin [5 mg/Kg, #I7378 Sigma] and were sacrificed 24 h later.

2.5. In vivo intestinal permeability

In vivo intestinal epithelial barrier permeability was assessed using the fluorescein isothiocyanate [FITC]-labelled dextran method. Mice were fasted for 4 h and received 12 mg/mouse of FITC-labelled dextran [4kDa, #46944 Sigma] by oral gavage. Blood samples were collected from the submandibular vein [cheek punch] after 90 min and fluorescence intensity was measured in the serum [excitation, 490nm; emission, 520nm; DXT880 Multimode detector, Beckman Coulter].

2.6. Histopathological scoring

Cryosections of murine ileum and colon samples were stained with haematoxylin and eosin [H&E] and scored in blinded fashion on the basis of changes in the epithelium and cell infiltration, as previously described.^18,19 Liver cryosections isolated from Gata6del and control mice were also stained with H&E to assess hepatocyte damage and cell infiltration.

2.7. Immunohistochemistry and immunofluorescence

Cryosections of human colon biopsies and murine ileum and colon samples were stained with primary antibody against GATA6 [1:1000, #5851 Cell Signaling Technology, Danvers, MA]. Isotype control-stained sections were prepared under identical immunohistochemical conditions using a rabbit normal IgG control antibody [R&D Systems]. Positive cells were visualised using MACH4 Universal HRP-Polymer kit with DAB [Biocare Medical #M4BD534G, Pacheco, CA] and analysed by LEICA DMI4000 B microscope with LEICA application suite software [V4.6.2]. To detect zonula occludens-1 and claudin-2 proteins, murine ileum and colon cryosections were fixed with 4% paraformaldehyde for 10 min and permeabilised with 0.1% Triton X-100 for 20 min at room temperature. Sections were then blocked for 1 h at room temperature [BSA 1%, Tween 0.1%, glycine 2%] and rabbit primary antibody against zonula occludens-1 [1:50 #40–2200, Invitrogen, Carlsbad, CA] or mouse primary antibody against claudin-2 [1:50 #AB53032, Abcam, Cambridge, UK] were incubated overnight at 4°C. After washing with PBS 1X, the secondary antibody goat anti-rabbit Alexa488 [1:1500, #A11008 Invitrogen] or goat anti-mouse Alexa488 [1:1000, #A11017 Invitrogen] were applied for 1 h at room temperature. Slides were then washed with PBS 1X, mounted using the prolong gold antifade reagent with DAPI [#P36931 Invitrogen], and analysed by LEICA DMI4000 B microscope with LEICA application suite software [V4.6.2] [Leica, Wetzlar, Germany].

2.8. Isolation of lamina propria mononuclear cells and splenocytes

Lamina propria mononuclear cells [LPMC] were isolated from Gata6del mice either exposed or not exposed to antibiotics [Gata6del + Abs] and control mice after 4 weeks following the last injection of TMX. Briefly, the dissected murine colon was freed of mucus and epithelial cells by DTT and EDTA and then digested with DNase [0.1 mg/ml, #11284932001 Roche] and liberase [0.1 mg/ml, #05401127001 Roche]. LPMCs were filtered through a 70-µm cell strainer, spun at 1500 rpm for 5 min, and resuspended in complete RPMI 1640 medium [10% FBS, 1% P/S].

In parallel, murine spleens were taken from Gata6del and control mice after 4 weeks following the last injection of TMX, cut in small pieces, and gently pressed through a 70-µm cell strainer. Splenocytes were then collected, washed with complete medium, and resuspended in ACK buffer to lyse red blood cells for 1 min. Lysis was then stopped by adding cold medium. Cells were spun at 1500 rpm for 5 min and resuspended in complete RPMI 1640 medium [10% FBS, 1% P/S].

2.9. Cell culture

The human normal colonic epithelial cell line HCEC-1CT was obtained from EVERCYTE GmbH [Vienna, Austria] and maintained in ColoUp medium [#MHT-039 EVERCYTE GmbH]. HCEC-1CT were transfected with 50 nM of GATA6 or control siRNA [#AM16704 Ambion, Austin, TX] using Lipofectamine 3000 reagent [Invitrogen, Carlsbad, CA]. After 24 h, cells were rested in ColoUp medium for additional 8 h and collected for protein extraction or assessment for human ZO-1 expression by immunofluorescence [1:100, #610967 BD Biosciences].

2.10. Chromatin immunoprecipitation [ChIP] assay

The human normal colonic epithelial cells HCEC-1CT were cross-linked with 1% formaldehyde for 15 min at room temperature, harvested in lysis buffer [2×10⁷ cells/mL], and sonicated using a Branson Sonifier 150 [20% amplitude, six cycles consisting of 15-s pulses followed by 45-s rest periods]. ChIP was performed using protein A agarose beads [16–125, Millipore], in combination with either anti-GATA6 antibody [used 1:50, #5851 Cell Signaling Technology, Danvers, MA] or normal rabbit IgG, as previously described.²⁰ DNA was purified through a commercially available kit [cat. 69504, QIAGEN] and amplified by quantitative polymerase chain reaction [qPCR] using the following primers: human ZO1_prom Fw 5’-AAGAAAACCCGACCTACTACGC-3’, human ZO1_prom Rev 5’-TGCTGTCTTTGGAGGAGTGG-3’, NEG Fw 5’-CATTGGGAAGTGATGATGTGATCT-3’, NEG Rev 5’-GTCCTCTCTGCCATCTTCACTCA-3’.

2.11. Flow cytometry

LPMC and splenocytes were stimulated with PMA [10 ng/mL, #P1585 Sigma], ionomycin [1 µg/mL, #I0634 Sigma], and brefeldinA [10 µg/mL, #00-4506-51 eBioscience] for 4 h. Cells were then collected and stained with the LIVE/DEAD Fixable Dead Cell Stain kit [#L34957 Thermo Fisher] for 30 min in ice, washed, and stained with anti-mouse CD45-APCCy7 [#557659 BD Biosciences, San Diego, CA], anti-mouse CD11c-PECy7 [#558079 BD Biosciences], anti-mouse CD11b-PerCP-Cy 5.5 [#550993 BD Biosciences], anti-mouse CD11b-PECy7 [#600112U100 eBioscience], anti-mouse F4/80-APC [#17480182 eBioscience], and anti-mouse Gr1-PerCP-Cy 5.5 [#552093 BD Biosciences]. Cells were fixed and permeabilised with Cytofix/Cytoperm, [# 554714 BD Biosciences] and intracellular cytokines were detected with the following antibodies: anti-mouse IL-6-FITC [#1170682, eBioscience, San Diego, CA] and anti-mouse TNF-α-PE [#12732182 eBioscience]. Appropriate isotype-matched controls were included. The Gallios flow cytometer [Beckman Coulter, Brea, CA] was used for the acquisition and the Kaluza software [Beckman Coulter] was used for analysis.

2.12. Quantification of faecal lipocalin-2

Faecal samples were weighted and resuspended in PBS 1X containing 0.1% Tween at a final concentration of 100 mg/ml. Samples then were vortexed for 20 min, centrifuged for 10 min at 14 000 g and 4°C, and supernatants collected and stored at -80°C. Lcn-2 protein levels were quantified using the Duoset murine LCN-2 enzyme-linked immunosorbent assay [ELISA] kit [#DY1857 R&D Systems], and optical density was read at 450 nm [DXT880 Multimode detector, Beckman Coulter].

2.13. Real-time PCR

Total RNA was isolated from human colon biopsies and murine ileum and colon samples using RNeasy Mini Kit [#74104 Qiagen, Hilden, Germany] and digested with DNase [Qiagen]. A constant amount of RNA [1 μg/sample] was retrotranscribed into complementary DNA [cDNA] using Oligo[dT] primers and M-MLV-reverse transcriptase [#28025021 Thermo Fisher Scientific, Waltham, MA]. Real-time PCR was performed for human GATA6, murine TNF-α, murine IL-6, and murine IFN-γ. RNA expression was calculated relative to the β-actin gene using the ΔΔCt algorithm. Primer sequences were as follows: mouse Tnf-α Fw 5’-ACCCTCACACTCAGATCATC-3’, Rev 5’-GAGTAGACAAGGTACAACCC-3’; mouse Il-6 Fw 5’-AGCCA GAGTCCTTCAGAGAG-3’, Rev 5’-GATGGTCTTGGTCCTT AGCC-3’; mouse Ifn-γ Fw 5’-CAATGAACGCTACACACTGC-3’, Rev 5’-CCACATCTATGCCACTTGAG-3’; human Gata6 Fw 5’-CTC TACAGCAAGATGAACGG-3’, Rev 5’-AAGGTGGTAGTTGTGG TGTG-3’.

2.14. Analysis of 16S rRNA gene amplicon sequencing

Raw sequences were quality checked using the FastQC software, and primers and adapters were eventually removed using Cutadapt. Preprocessed reads were analysed using the QIIME 2 pipeline.²¹ In detail, reads were chimera checked and clustered in Amplicon Sequence Variants [ASV] using the DADA2 algorithm.²² The taxonomic assignment of representative sequences obtained by DADA2 was performed using the q2-feature-classifier and the Silva database vr. 138.^23,24 Statistical analysis of ASV tables was performed in R, using the following packages: phyloseq, DESeq2, vegan, and ggplot.^25,26 ASV with a low frequency were removed from the analyses²⁴ and were normalised using DESeq2.²⁷ Normalised data were used to measure alfa and beta diversity metrics. Using the tax_glom[] function in the phyloseq package, data were summed up to different taxonomic levels [phylum, class] and compared among samples.

2.15. Western blotting

Cells were lysed on ice in buffer containing 10 mM HEPES [pH 7.9], 10 mM potassium chloride [KCl], 0.1 mM ethylenediaminetetraacetic acid [EDTA], 0.2 mM ethylene glycol-bis [β-aminoethyl ether]-N,N,N’,N’-tetraacetic acid [EGTA], and 0.5% Nonidet P40 supplemented with 1 mM dithiothreitol [DTT], 10 mg/ml aprotinin, 10 mg/ml leupeptin, 1 mM phenylmethylsulphonyl fluoride [PMSF], 1 mM Na3VO4, and 1 mM sodium fluoride [NaF]. Lysates were clarified by centrifugation and separated on sodium dodecyl sulphate [SDS]-polyacrylamide gel electrophoresis. Blots were incubated with antibodies against GATA6 [1:1000, #5851 Cell Signaling Technology, Danvers, MA, USA], ZO-1 [1:1000, #610967 BD Biosciences], and β-actin antibody [1:5000, #A544 Sigma], followed by a secondary antibody conjugated to horseradish peroxidase [1:20000, Dako, Santa Clara, CA, USA].

2.16. Statistical analysis

Parametric data were analysed using the two-tailed Student’s t test for comparison between two groups or one-way analysis of variance followed by Bonferroni’s post hoc test for multiple comparisons. Non-parametric data were analysed using the Mann‐Witney U test for comparison between two groups or Kruskal‐Wallis test for multiple comparisons. Significance was defined as p-values <0.05.

3. Results

3.1. GATA6 expression is decreased in IBD

To evaluate whether active inflammation in IBD is marked by changes in the expression of molecules involved in the control of epithelial cell function, we performed microarray analysis comparing mucosal samples taken from involved areas of UC patients and unaffected mucosa of normal controls. Several genes involved in epithelial cell commitment, cell proliferation, and wound healing were differently expressed between the two sets of samples [Figure 1A]. The most downregulated genes in UC were genes involved in cell commitment, whereas several genes involved cell proliferation and wound healing showed to be upregulated in UC [Figure 1A]. Among the genes controlling epithelial cell commitment, GATA6 emerged as a potential candidate for further analysis, as this transcription factor is known to regulate the differentiation and function of all four types of intestinal epithelial cells.^28,29 To confirm our data, we analysed GATA6 RNA transcripts in additional mucosal samples taken from UC patients, CD patients, and controls [Figure 1B]. A significant decrease in GATA6 RNA was seen in both UC and CD as compared with controls [Figure 1B]. Immunohistochemical analysis showed that GATA6 was mainly expressed by IECs in the human colon but was downregulated in IBD epithelium [Figure 1C].

Figure 1.

GATA6 expression is decreased in IBD patients. [A] Heat map showing microarray based differential expression, log2 [fold change] of genes related to cell commitment, cell proliferation, and wound healing of mucosal samples taken from involved areas of UC patients and normal colonic mucosa of controls. [B] GATA6 RNA expression was assessed by real-time PCR in biopsy samples taken from inflamed colonic mucosa of seven CD patients and 14 UC patients and normal mucosa of 18 controls [CTR]. Data are expressed as mean. Differences among groups were compared using one-way analysis of variance [ANOVA] followed by Bonferroni’s post hoc test [**p ≤0.01, ***p ≤0.001]. Each point in the graph indicates the GATA6 RNA expression in a colonic sample of a single patient. [C] Representative immunohistochemical pictures showing GATA6 in colon biopsy samples from IBD patients and controls. The picture is representative of four experiments analysing colonic samples taken from eight CTR, eight CD patients, and eight UC patients. Right panel shows the percentage of GATA6-expressing cells per total epithelial cells. Data indicate mean, and differences among groups were compared using one-way analysis of variance [ANOVA] followed by Bonferroni’s post hoc test [***p ≤0.001]. Scale bar: 75 µm. IBD, inflammatory bowel disease; UC, ulcerative colitis; CD, Crohn’s disease; PCR, polymerase chain reaction.

3.2. GATA6 deficiency in intestinal epithelial cells promotes epithelial damage and enhances local immuno-inflammatory response

GATA6 is expressed in differentiated and proliferating epithelial cells and known to regulate the function of such cells in the mature mouse intestine.^28,29 In order to investigate the effect of GATA6 deficiency on intestinal homeostasis and development of gut inflammation, we used Gata6^loxP/loxP, VillinCreER^T2-positive [Gata6del] mice where the Gata6 gene is specifically deleted in the intestinal epithelium.^28,29 Immunohistochemistry analysis 4 weeks after treatment revealed complete absence of GATA6 in both adult distal ileum and colonic tract [Figure 2A‐B]. Upon dissection, animals exhibited no macroscopic abnormalities, but histological analysis revealed that the villi of the ileum were shorter compared with controls [Figure 2A‐B]. Moreover in both ileum and colon, loss of GATA6 associated with distortion of gland architecture, marked depletion of mucus, oedema, more pronounced infiltration of inflammatory cells, and greater epithelial damage, resulting in a higher histological score of intestinal inflammation as compared with wild-type mice [Figure 2A‐B].

Figure 2.

Loss of GATA6 in intestinal epithelial cells promotes epithelial damage and enhances local immune-inflammatory response. [A‐B] Representative H&E staining, histological score, and immunohistochemistry staining for GATA6 in ileum [A] and colon [B] sections taken from wild-type mice [CTR] and Gata6^loxP/loxP, VillinCreER^T2-positive [Gata6del] mice after 4 weeks post-treatment with TMX. The figure is representative of three experiments in which seven mice/group were analysed. Data indicate mean, and differences between groups were compared using Student’s t-test [*p ≤0.05, **p ≤0.01]. Scale bar: 75 µm. [C] Heat map showing microarray-based differential expression, log2 [fold change] of genes related to immune-inflammatory response of ileum and colonic samples taken from CTR and Gata6del mice 4 weeks post-treatment with TMX. [D] TNF-α, IL-6 and IFN-γ RNA expression was assessed by real-time PCR in ileal and colonic samples. Data were generated using 10–12 mice/group [*p ≤0.05]. Each point in the graph indicates the expression of the specific RNA transcript in the ileum or colon of a single mouse; horizontal bars indicate the median value. Differences between groups were compared using the Mann‐Whitney U test. [E] Dot-plots and histograms showing the percentages of dendritic cell [CD11b⁺CD11c⁺]-, neutrophil [CD11b⁺Ly6G⁺]-, and macrophage [CD11b⁺F4/80⁺]-producing TNF-α and/or IL-6 in the lamina propria of CTR and Gata6del mice. Data were generated analysing three pools of lamina propria cells isolated from a total of nine mice/group. Data indicate mean, and differences between groups were compared using Student’s t-test [*p ≤0.05, **p ≤0.01]. [F] Histograms showing the percentages of dendritic cell [CD11b⁺CD11c⁺]-, neutrophil [CD11b⁺Ly6G⁺]-, and macrophage [CD11b⁺F4/80⁺]-producing TNF-α and/or IL-6 in the spleen of CTR and Gata6del mice. Data were generated analysing three pools of splenocytes isolated from a total of nine mice/group. [G] Scatter plot showing levels of faecal Lcn-2 protein in CTR and Gata6del mice. Data were generated using of eight or nine mice/group. Horizontal bars indicate median value, and differences were compared using the Mann‐Whitney U test [***p ≤0.001]. PCR, polymerase chain reaction; H&E, haematoxylin and eosin; CTR, control; TMX, tamoxifen.

To begin to dissect the mechanisms by which loss of GATA6 alters the intestinal homeostasis, we performed a microarray analysis of ileal and colonic samples taken from CTR and Gata6del mice [Figure 2C]. Several genes involved in the innate immune response were upregulated in both the ileum and the colon of Gata6del mice as compared with wild-type mice [Figure 2C]. Additionally, analysis of RNA expression of inflammatory cytokines in mucosal samples by real-time PCR showed enhanced expression of TNF-α and IL-6 in both the ileum and the colon of Gata6del mice as compared with controls [Figure 2D], whereas the content of IFN-γ RNA transcripts in both ileal and colon samples did not differ in the two groups of mice [Figure 2D]. Next, we isolated lamina propria mononuclear cells [LPMC] from ileal and colon samples and analysed the cell sources of the above cytokines by flow cytometry [Figure 2E]. Although we employed several cell separation techniques, ileal LPMC exhibited a marked baseline mortality, making difficult any further cytokine analysis. In colon samples of Gata6del mice, TNF-α was mainly produced by CD11c⁺CD11b⁺cells, whereas IL-6 was mainly expressed by CD11c⁺CD11b⁺ cells and CD11b⁺F4/80⁺ cells [Figure 2E]. To examine whether GATA6 loss associates with enhanced systemic immune response, we looked at the frequency of TNF-α- and/or IL-6-producing immune cells in the spleens of CTR and Gata6del mice. Loss of GATA6 did not alter the percentage of TNF-α- and/or IL-6-expressing immune cells [Figure 2F]. Moreover, TNF-α was undetectable in serum samples of both wild-type and Gata6del mice [data not shown]. Histological examination of liver sections failed to identify hepatocyte damage or increased infiltration of the liver with inflammatory cells in Gata6del mice [Supplementary Figure 1, available as Supplementary data at ECCO-JCC online]. Analysis of lipocalin-2 [Lcn-2], a secretory protein mainly produced by neutrophils, in the faeces showed enhanced excretion of the protein in Gata6del mice as compared with controls [Figure 2G]. Altogether, these data indicate that selective loss of GATA6 in intestinal epithelial cells is accompanied by various degrees of epithelial damage and enhanced local immunoinflammatory response.

3.3. Loss of GATA6 leads to impaired epithelial barrier function

Given the increased gut inflammation and local immune activation noted in Gata6del mice, we performed in vivo experiments assessing gut barrier function in control and Gata6del mice. Intestinal permeability, as determined by serum FITC-dextran concentration 90 min after oral gavage, was significantly increased in Gata6del mice compared with controls [Figure 3A]. Next, we reasoned that the enhanced intestinal permeability seen in Gata6del mice could be secondary to changes in the expression of tight junction proteins. These include a series of transmembrane proteins [e.g., occludins, claudins] which form fibrils that cross the plasma membrane and interact with the actin cytoskeleton within the adjoining cell through plaque proteins [e.g. zonula occuldens-1, ZO-1].^6,7,30 Zonula occludens-1 is known to regulate the paracellular permeability to specific ions and large molecules.^6,7,30 Immunofluorescence analysis revealed a marked reduction of ileal and colonic ZO-1 expression in Gata6del mice as compared with controls, whereas the expression of claudin-2 [CLDN-2], which forms charge-selective paracellular pores, was not altered [Figure 3B]. To determine whether the reduced ZO-1 expression seen in Gata6del mice was a direct consequence of the effect of GATA6 on ZO-1 expression,²⁰ we evaluated ZO-1 in a human normal colonic epithelial cell line HCEC-1CT transfected with a siGATA6. Western blot and immunofluorescence analysis showed that silencing of GATA6 reduced ZO-1 content [Figure 3C-D]. Moreover, chromatin immunoprecipitation assay revealed that GATA6 binds ZO-1 promoter in HCEC-1CT cells [Supplementary Figure 2, available as Supplementary data at ECCO-JCC online], suggesting that GATA6 directly regulates ZO-1 expression and intestinal permeability.

Figure 3.

Loss of GATA6 impairs epithelial barrier function. [A] Levels of serum FITC-dextran in CTR and Gata6del mice 90 min after oral gavage. Data were generated using six or seven mice/group. Data indicate mean, and differences between groups were compared using Student’s t-test [*p ≤0.05]. [B] Immunofluorescence analysis of zonula occludens-1 [ZO-1] and claudin-2 [CLDN-2] [green] in ileum and colon sections isolated from CTR and Gata6del mice. The images are representative of four or five mice/group. Scale bar: 25 µm [C] Representative western blotting for GATA6, ZO-1, and β-actin in HCEC-1CT cells transfected with either Gata6 or control small interfering RNA [CTR or Gata6 siRNA]. Images are representative of three independent experiments [D] Immunofluorescence analysis of ZO-1 [green] in HCEC-1CT cells treated as in C. Scale bar: 25 µm. CTR, control.

3.4. GATA6 loss promotes mucosal dysbiosis and bacteria-driven local immune response

Next, we assessed whether Gata6del mice exhibited changes of gut microbiota composition. The bacterial communities from colonic samples derived from wild-type and Gata6del mice were analysed by 16S rRNA sequencing. GATA6 loss associated with some significant bacterial changes in terms of phyla and related classes [Figure 4A‐B]. In particular, Gata6del mice exhibited a significant increase in the phylum Bacteroidetes and increased levels of Bacteroidia and Clostridia classes [Figure 4A‐B]. These data suggest that loss of GATA6 significantly impacts on the composition of gut microbiota in mice. To better understand the role of microbiota in the intestinal morphological and immunological changes seen in Gata6del mice, we depleted gut microbiota with broad-spectrum antibiotics in drinking water for 2 weeks. After antibiotic treatment, Gata6del mice continued to exhibit the main epithelial alterations, including changes in ZO-1, as compared with controls [Figure 4C-D]. In contrast, antibiotics abrogated the induction of inflammatory cytokines by LPMC [Figure 4E] and downregulated faecal level of Lcn-2 [Figure 4F].

Figure 4.

GATA6 loss promotes mucosal dysbiosis and bacteria-driven local immune response. [A‐B] Relative abundance of phyla and classes of colonic mucosa-associated microbiota in Gatadel and control [CTR] mice. Horizontal bars indicate median value. Data were generated using eight or nine mice/group and differences between the two groups were compared using the Mann‐Whitney U test [*p ≤0.05, **p ≤0.01]. [C] Representative H&E staining and histological score of ileum and colon sections taken from Gata6del and CTR mice treated or not with antibiotics for 2 weeks [Gata6del + Abs]. Results are generated using four or five mice/group. Data indicate mean, and differences among groups were compared using one-way analysis of variance [ANOVA] followed by Bonferroni’s post hoc test [*p ≤0.05, **p ≤0.01]. Scale bar: 75 µm. [D] Immunofluorescence analysis of ZO-1 [green] in ileum and colon sections isolated from mice treated as in C. The figure is representative of four or five mice/group. Scale bar: 25 µm. [E] Histograms showing the percentages of dendritic cell [CD11b⁺CD11c⁺]-, neutrophil [CD11b⁺Ly6G⁺]-, and macrophage [CD11b⁺F4/80⁺]-producing TNF-α and/or IL-6 cytokines in the lamina propria of Gata6del mice treated or not with the antibiotic cocktail [Gata6del + Abs]. Data were generated using lamina propria cells isolated from three pools from a total of nine mice/group. Data indicate mean, and differences between groups were compared using Student’s t-test [*p ≤0.05, **p ≤0.01]. [F] Scatter plot showing levels of faecal Lcn-2 protein in CTR and Gata6del mice. Data were generated using four or five mice/group and are indicated as mean; statistical difference was analysed using one-way analysis of variance [ANOVA] followed by Bonferroni’s post hoc test [***p ≤0.001, ****p ≤0.0001]. H&E, haematoxylin and eosin; CTR, control.

3.5. Gata6del mice are more susceptible to experimental gut inflammation

In a final set of experiments, we assessed whether loss of GATA6 enhanced the susceptibility of mice to experimental gut inflammation. Initially, we showed that Gata6del mice developed a more severe colitis when challenged with low doses of TNBS, as shown by more pronounced body weight loss, infiltration of inflammatory cells, and epithelial damage, and consequently a higherer histological score of intestinal inflammation as compared with wild-type mice [Figure 5AB]. Consistently, Gata6del mice exhibited higher levels of TNF-α and IFN-γ RNA transcripts as compared with wild-type mice [Figure 5C]. To confirm these data, we used another model of intestinal inflammation induced by a single subcutaneous injection of indomethacin. Gata6del mice exhibited a more pronounced ileal mucosal injury and increased levels of TNF-α RNA transcripts compared with controls [Figure 5D-E]. Altogether, these data indicate that selective loss of GATA6 in intestinal epithelial cells exacerbates gut inflammation.

Figure 5.

Gata6del mice are more susceptible to experimental gut inflammation. [A] Control [CTR] and Gata6del mice received one intrarectal administration of TNBS [1.75 mg/mouse in 35% ethanol]. Body weight was recorded every day until sacrifice [Day 3]. Data were generated using seven mice/group and expressed as mean. Differences between groups were compared using Student’s t-test [**p ≤0.01]. [B] Representative H&E staining and histological score of colon sections taken from control [CTR] and Gata6del mice treated as in A. Data were generated using seven mice/group and expressed as mean. Differences between groups were compared using Student’s t-test [*p  0.05]. Scale bar: 100 µm. [C] TNF-α and IFN-γ RNA expression in colonic samples taken from mice treated as in A were assessed by real-time PCR. Data were generated using seven mice/group [*p ≤0.05]. Each point in the graph indicates the TNF-α or IFN-γ RNA expression in the colonic sample of a single mouse; horizontal bars indicate median value. Differences between groups were compared using the Mann‐Whitney U test. [D] Representative H&E staining and histological score of ileum sections taken from CTR and Gata6del mice treated with a single subcutaneous injection of indomethacin [5 mg/Kg]. Data were generated using six mice/group and expressed as mean. Differences between groups were compared using Student’s t-test [**p ≤0.01]. Scale bar: 100 µm. [E] TNF-α RNA expression in ileal samples was assessed by real-time PCR. Data were generated using six mice/group. Each point in the graph indicates the TNF-α RNA expression in the ileum of a single mouse; horizontal bars indicate the median value. Differences between groups were compared using the Mann‐Whitney U test [*p ≤0.05]. PCR, polymerase chain reaction; H&E, haematoxylin and eosin; CTR, control.

4. Discussion

Intestinal epithelial barrier impairment is a hallmark of IBD, and such a dysfunction is believed to make a valid contribution to the amplification of the IBD-associated immuno-inflammatory response.^8–11 Indeed, various compounds that target the epithelium and restore barrier function are now considered good candidates for IBD treatment.^31–34 This study was undertaken to identify factors/mechanisms underlying barrier impairment in IBD. By a microarray analysis of colon samples taken from inflamed mucosa of IBD patients and normal colon samples of controls, we initially identified several molecules involved in the regulation of epithelial barrier function, which were differently expressed during gut inflammation. Among these we focused our attention on GATA6, as this transcription factor is known to regulate differentiation and proliferation of mature intestinal epithelial cells.^28,29 Real-time PCR and immunohistochemistry of IBD and control samples showed that GATA6 deficiency occurred in both CD and UC, and this was evident at both RNA and protein levels, indicating that IBD is marked by defective epithelial expression of GATA6.

To assess the functional effect of the reduced GATA6 expression on the intestinal homeostasis, we used mice with an inducible deletion of GATA6 in the ileal and colonic epithelium [Gata6del], and evaluated whether GATA6 loss was followed by changes in the intestinal morphology, immune response, and local microflora. Gata6del mice exhibited epithelial damage and a pronounced infiltration of the mucosa with inflammatory cells, thereby resulting in a higherer histological score of intestinal inflammation as compared with wild-type mice, but neither overt clinical symptoms nor macroscopic lesions. GATA6 loss was accompanied by enhanced innate immune and cytokine responses in both ileum and colon. This is in line with studies by Beuling and colleagues, showing that GATA6, but not GATA4, is expressed in the mouse terminal ileum and that conditional deletion of GATA6 gene results in a decrease in cellular proliferation in crypts, a decrease in entero-endocrine cell allocation, a conversion of Paneth cells into goblet-like cells at the base of crypts, and an alteration in the expression of specific absorptive enterocyte genes.^28,29 However, if such defects are due either primarily to the lack of GATA6 and/or to secondary to changes in the local immune-inflammatory response and/or luminal flora, remains unknown.

The increased gut inflammation and local immune activation seen in Gata6del mice prompted us to examine whether loss of GATA6 was followed by perturbation of epithelial barrier function. In vivo intestinal permeability to oral FITC-dextran was increased in Gata6del mice as compared with controls. We did not observe enhanced translocation of bacteria to the intestinal lamina propria of Gata6del mice [personal unpublished observations], suggesting that the permeability defect is not sufficient to afford increased translocation of bacteria, while permitting passage of smaller molecules. Consistent with these findings was the demonstration that the enhanced intestinal permeability seen in Gata6del mice was associated with defective expression of the regulator of intestinal tight junction proteins ZO-1. A significant downregulation of ZO-1 was also seen in cultured human colon epithelial cells following Gata6 silencing, thus suggesting a direct control of ZO-1 expression by GATA6. This hypothesis was supported by the demonstration that GATA6 binds ZO-1 promoter in human colon epithelial cells, confirming previous observations in pancreatic cancer cells or multiple epithelial genes, including protocadherins and tight junction components.²⁰

Loss of GATA6 was associated with significant changes in the composition of bacteria adherent to the colonic mucosa, including an increase in the phylum Bacteroidetes and increased levels of Bacteroidia and Clostridia classes. These findings recapitulate major bacterial alterations seen in other models of gut inflammation, in which changes in the number of Bacteroidia and Clostridia can occur even before the onset of colitis.³⁵ Interestingly, depletion of gut microbiota with broad-spectrum antibiotics in Gata6del mice abrogated the local cytokine response without changing the main epithelial alterations, including downregulation of ZO-1. Taken together, these observations indicate that selective loss of GATA6 in intestinal epithelial cells promotes primarily epithelial damage and alters barrier integrity, thereby facilitating bacteria-driven local immune-inflammatory response.

In a situation of defective epithelial barrier function as seen in Gata6del mice, one would expect that spontaneous, and perhaps chronic, disease might develop due to increased bacterial translocation across the epithelium. However, despite defects in barrier function, Gata6del mice do not develop clinical/macroscopic signs of spontaneous ileitis or colitis, even in long-lived mice [28 weeks old, personal unpublished observations], which suggests that protective mechanisms were probably in place to counter-regulate inflammatory pathways and prevent spontaneous disease. However, exposure of such mice to dangerous clues could favour development of pathology. Indeed, loss of GATA6 amplified the local inflammatory cytokine response and intestinal damage induced by administration of both indomethacin and low-doses of TNBS.

Most of the spontaneous epithelial alterations and those seen following induction of gut inflammation in Gata6del mice bear similarities to those reported in mice lacking other transcription factors specific to the intestinal epithelium, which are down-regulated in IBD tissue [Figure 1A]. For instance, mice deficient in Cdx2 exhibit dysbiosis in association with high intestinal permeability and aberrant mucosal inflammatory response, culminating in enhanced susceptibility to experimental colitis.³⁶ Similarly, IEC-specific deletion of serine-threonine kinase 11 [STK11], also called liver kinase B1 [LKB1], resulted in a definitive shift in the composition of the microbial population in the mouse intestine with predominance of colitogenic species and, hence, increased susceptibility to dextran sodium sulphate-induced colitis.³⁷ Notably, the ability of STK11 in suppressing colitogenic microbiota was mediated by modulation of IL-18 expression.³⁷ Epithelial deficiency of the transcription repressor hairy and enhancer of split 1 [Hes1] leads to intestinal microbial dysbiosis associated with decreased expression of commensal-dependent antimicrobial genes and impaired resistance against pathogenic bacterial colonisation. Moreover, Hes1 deficiency enhanced susceptibility to experimental intestinal inflammation.³⁸ These observations indicate that multiple IEC-intrinsic transcription factors contribute to maintain gut homeostasis, even though each of them has a unique role in regulating intestinal epithelial function.

In conclusion, our data show that human IBD is marked by defective epithelial expression of GATA6, and selective loss of GATA6 in murine intestinal epithelium leads to altered barrier function, which promotes bacteria-driven local immune response thereby amplifying gut pathology.

Funding

The authors declare that there are no sources of funding to be acknowledged.

Conflict of Interest

We have nothing to disclose.

Author Contributions

FL: study concept and design, technical and material support, acquisition of data, analysis and interpretation of data, drafting of the manuscript, statistical analysis. CS: analysis and interpretation of data, technical and material support, critical revision of the manuscript. GB, DP: analysis and interpretation of data, acquisition of data, technical and material support. CM, EF, ET, EL, IM, DDF, ADG, AC, AO: acquisition of data, technical and material support. AD, IM: analysis and interpretation of data, critical revision of the manuscript. GM: supervised the project. All authors had access to the all data and have reviewed and approved the final manuscript.