Article Text

PDF

Original article
Genomic ATG16L1 risk allele-restricted Paneth cell ER stress in quiescent Crohn's disease
  1. J Jasper Deuring,
  2. Gwenny M Fuhler,
  3. Sergey R Konstantinov,
  4. Maikel P Peppelenbosch,
  5. Ernst J Kuipers,
  6. Colin de Haar,
  7. C Janneke van der Woude
  1. Department Gastroenterology and Hepatology, Erasmus MC University Medical Centre Rotterdam, Rotterdam, The Netherlands
  1. Correspondence to Dr G M Fuhler, Department of Gastroenterology and Hepatology, Erasmus MC Rotterdam, ‘s-Gravendijkwal 230, Rotterdam 3015 CE, The Netherlands; g.fuhler{at}erasmusmc.nl

Abstract

Objective Although genome wide association studies have partly uncovered the genetic basis of Crohn's disease (CD), it remains a challenge to link genetic polymorphisms to functional intestinal phenotypes. Paneth cells are specialised antimicrobial epithelial cells localised to the small-intestinal crypt-base. Here, we investigate whether genomic variations in ATG16L1 affect Paneth cell function.

Design Genomic variation of ATG16L1 (T300A, rs2241880) was determined in DNA from 78 patients with CD and 12 healthy controls. Paraffin-embedded ileal biopsies from patients with genotype AA (n=17), GA (n=38) and patients with the GG allele (n=23) were stained for GRP78, phospho-EIF2α, lysozyme, cleaved-caspase 3, phosphohistone H3, phospho-IκB, p65, phospho-p38MAPK and PHLDA1. Microbial composition of biopsies was assessed by PCR. Disease phenotype was scored.

Results In patients with quiescent disease but with an ATG16L1 risk allele, the endoplasmic reticulum (ER) stress markers GRP78 and pEIF2α were highly expressed in Paneth cells. Other CD risk gene variations did not correlate with Paneth cell ER stress. Functionally, patients with ER-stressed Paneth cells showed no changes in intestinal epithelial cells proliferation or apoptosis, Paneth cell or stem cell numbers, p65, phospho-IκB and phospho-p38 staining. However, a significantly increased presence of adherent-invasive Escherichia coli was observed in biopsies from patients with ER-stressed Paneth cells. Phenotypically, patients with GRP78 positive Paneth cells have relatively less colonic disease over ileal disease (−21%, p=0.04), more fistulas (+21%, p=0.05) and an increased need for intestinal surgery (+38%, p=0.002).

Conclusions The ATG16L1 T300A polymorphism defines a specific subtype of patients with CD, characterised by Paneth cell ER stress even during quiescent disease. Paneth cell ER stress correlates with bacterial persistence, and is thus likely to modulate antimicrobial functionality of this cell type in patients with CD.

  • IBD - GENETICS
  • IBD BASIC RESEARCH
  • CROHN'S DISEASE
  • SMALL BOWEL DISEASE
  • SMALL INTESTINAL BIOPSY

Statistics from Altmetric.com

Significance of this study

What is already known about this subject?

  • A polymorphism of the ATG16L1 gene (T300A, rs2241880) is associated with an increased susceptibility to Crohn's disease (CD).

  • Epithelial specific ATG16L1 mice tend to have comparable Paneth cell morphology to ATG16L1 T300A in patients with CD.

  • Hallmarks for crinophagy are found in Paneth cells of paediatric patients with CD.

  • Autophagy and endoplasmic reticulum (ER) stress are closely intersecting pathways.

  • Inflammation-associated protein folding difficulties (ER stress) dramatically influence the expression of ER depending proteins.

What are the new findings?

  • Patients with quiescent CD and healthy controls carrying the ATG16L1 T300A risk allele have increased ER stress in their Paneth cells.

  • The ER-stressed Paneth cells do not modulate epithelial cell proliferation or apoptosis, or the number of crypt-base located stem cells.

  • The presence of adherent-invasive Escherichia coli is increased in patients with CD with ER-stressed Paneth cells.

How might it impact on clinical practice in the foreseeable future?

  • The realisation that ER stress in ATG16L1 risk allele carrying patients may play a role in impaired bacterial clearance provides further understanding of the pathology of this disease and may aid design of personalised therapy.

Introduction

Inflammatory bowel disease (IBD) is the result of a deregulated immune response to gut microbiota. Genetic and environmental factors play an important role in the development of IBD. The most frequent forms of IBD are Crohn's disease (CD) and ulcerative colitis. By identifying single nucleotide polymorphisms (SNPs) associated with an increased risk for developing CD, genome wide association studies (GWAS) have now uncovered approximately 10% of the disease variance of CD, resulting in the identification of 140 genetic locations in the human genome apparently linked to this disease.1 ,2 The main challenge now is to link these genetic variants to altered functionality in SNP carrying individuals, thus explaining their association with disease susceptibility.

Paneth cells are specialised intestinal epithelial cells that are found at the crypt base of the small intestine. They are characterised by large secretory granules and an extensive endoplasmic reticulum (ER). Initially described as innate immune cells producing antimicrobial products, Paneth cells have recently been suggested to constitute a cardinal component of the intestinal stem cell niche.3 As Paneth cells contribute to controlling the luminal flora as well as repairing the intestinal barrier following an insult, genomic alterations that impede the Paneth cell-compartment functionality can potentially increase the propensity to develop CD.

CD-linked SNPs that might affect Paneth cells include those in NOD2 (CARD15), associated with reduced intracellular microbial sensing, a decrease in Paneth cell-produced α-defensins and reduced antimicrobial defence.4 ,5 ,6 Furthermore, genetic variation in the XBP1 gene may lead to increased ER stress as a consequence of a defective unfolded protein response (UPR) in highly secretory intestinal epithelial cells (IECs) like goblet and Paneth cells, thereby affecting their function.7 Finally, it is described that a SNP in the ATG16L1 gene (T300A, G is risk allele, rs2241880) may confer its risk for CD development by influencing Paneth cell morphology and function.8 Overall, Paneth cell function seems to be especially susceptible to the presence of a variety of SNPs strongly associated with CD, suggesting that the affected pathways, bacterial/viral sensing (NOD2), ER stress (XBP-1, ORMDL3) and autophagy (ATG16L1, IRGM), are of crucial importance for normal Paneth cell function. The latter is essential for maintaining intestinal homoeostasis.3 This notion is further supported by studies in mice sporting epithelial-specific deficiencies in these pathways, which show abnormal Paneth cell phenotype and function, in some cases leading to spontaneous ileitis.7 ,8 In contrast, although a recent study described activation of autophagy and crinophagy with concomitant reduced numbers of secretory granules in the Paneth cell compartment of patients with CD, these features correlated to neither disease activity nor ATG16L1 and IRGM genotype.9 It is thus fair to say that the link between genotype and Paneth cell functionality in CD remains poorly understood.

An underexplored possible connection between the Paneth cell compartment, CD-linked genetic polymorphisms and intestinal inflammation is ER stress. ER stress has been shown to induce autophagy as part of the UPR, and impairment in the autophagy pathway can lead to ER stress and the induction of the UPR.10 Activation of NOD2 by its ligand muramyl dipeptide (MDP) or bacteria induces NF-κB mediated induction of autophagy,4 whereas autophagy has been shown to play an important role in the uptake and delivery of NOD2 signals.5 We and others have recently shown that elevated ER stress negatively influences normal IEC function.7 ,11 ,12 However, the possible connection between genotype and Paneth cell ER stress remains unclear.

For these reasons, we investigated whether evidence for the convergence of ER stress and inflammatory pathways could be detected in Paneth cells of patients with CD and how these pathways would relate to genetic polymorphisms in CD. In this study we demonstrate that patients with CD and healthy controls with at least one risk allele for ATG16L1 (T300A, rs2241880) show elevated ER stress in their Paneth cells. Combined with an altered mucosal bacterial composition in patients with CD with ER-stressed Paneth cells, these data suggest that ATL16L1 risk status may identify a subpopulation of patients with CD in which elevated Paneth cell ER stress can contribute to disease.

Methods

Patient material

Formalin-fixed paraffin-embedded (FFPE) small intestine biopsies were collected from the pathology archive from in total 78 patients with CD: 17 with ATG16L1 (T300A) allele AA, 38 with the GA allele and 23 with the GG allele, and 12 healthy controls: 3 with ATG16L1 (T300A) allele AA, 3 with the GA allele and 6 with the GG allele. Healthy individuals have no medical or family history of IBD. All obtained biopsies were endoscopically and histologically (Geboes criteria) rated as not actively inflamed. The demographic and disease-specific characteristics were subtracted from electronic patient files according to the Montreal classification13 and are presented in online supplementary tables S2 and S3.

Genomic DNA analysis

The patients with CD were genotyped using genomic DNA isolated from peripheral blood mononuclear cells using Wizard Genome DNA purification kit (Promega Benelux B.V.). The healthy individuals were genotyped using genomic DNA isolated from FFPE biopsies. Detailed protocol is described in online supplementary methods. The following genetic loci associated with IBD and ER stress were analysed: ATG16L1 (rs10210302),14 NOD2 (rs2066844, rs2066845, rs2066847),15 XBP1 (rs35873774).7 KBiosciences UK ltd, Hertfordshire, UK, performed the SNP analysis. All genotypes are presented in online supplementary table S1.

Histology

For this we used serial sections (4 μm) of intestinal FFPE tissue blocks. Paneth cells were stained using Periodic acid-Schiff (PAS)—alcian blue staining, a detailed protocol is provided in the online supplementary content.

For immunohistochemistry, all the primary antibodies and specific protocol details are summarised in the online supplementary files and in supplementary table S4. The slides were scored in a blinded fashion by light microscopy (Zeiss, Axioskop, Oberkochen, Germany). Intestinal biopsies with high numbers of GRP78-positive and p-EIF2α-positive Paneth cells were scored as positive and biopsies with basal levels of GRP78-stained and p-EIF2α-stained Paneth cells as negative. Representative examples of negative and positive score samples are presented in online supplementary figure S1. Positive cleaved-caspase 3 and phosphohistone (p-histone) H3 cells of at least 20 intestinal crypts per biopsy were counted. Paneth cell specific lysozyme, p65, p-IκB, p-p38MAPK, and stem cell specific PHLDA1 staining were examined by immune reactivity intensity.

Cell culture and β-Catenin reporter assay

Wnt3a producing (Wnt3a) and control cells (L-cells) were kindly provided by Dr R Smits (Erasmus MC, Rotterdam, The Netherlands). Cells were stimulated for 18 h with 2 µM Tunicamycin to induce ER stress, and the conditioned medium was used in a β-catenin reporter assay as previously described.16 XBP1 mRNA splicing, GRP78 mRNA and CCAAT/enhancer-binding protein (C/EBP) homologous protein (CHOP) mRNA expression were measured as ER stress indicators.11

Microbiome analysis

DNA was extracted from previously determined Paneth cell ER-stress positive and ER-stress negative FFPE biopsy tissue samples, using Genomic DNA isolation kit (Promega) as per manufacturer's instructions. Resulting DNA was eluted in 50 µL of sterile distilled water and stored at −20°C. The primers used in the study are shown in online supplementary table S5. Standard PCR reaction conditions were 95°C for 5 min, followed by 40 cycles of 94°C for 1 min, 55°C for 1 min, 72°C for 1 min and finally 2 min at 72°C. Reactions were carried out on an iCycler 1 (BioRad thermocycler).

Statistics and software

Statistical analyses were performed using the SPSS V.11.0 software package for Windows. Data on different patient groups were compared using the Mann-Whitney U test, χ2 test, Fisher's exact test and the McNemar's test. Multivariate analyses relating ER-stress to different genotypes are performed using SPSS V.11.0. A two-tailed p value <0.05 was accepted as statistically significant. Images were composed using Adobe Photoshop CS6.

Results

Increased ER stress in Paneth cells of ATG16L1 mutated patients

To investigate the relationship between genetic polymorphisms and ER stress in Paneth cells, cross sections of terminal ileum biopsies from patients with quiescent CD were stained with multiple histological markers. The Paneth cell specificity of GRP78 is presented in online supplementary figure S2. Only in patients with an ATG16L1 risk allele an association with GRP78 positivity (ER stress) in the Paneth cell compartment was detected (figure 1). GRP78 staining extended to all Paneth cells of the entire biopsy in 37% of the patients with CD carrying one risk allele (GA) and to 44% of those carrying two risk alleles (GG), whereas no GRP78+ Paneth cells were observed in patients without any risk allele (AA) (figure 1A,D). Furthermore, p-EIF2α positive Paneth cells were observed in 43% of the ATG16L1 risk allele carriers (GA and GG), but in none of the non-risk allele patients with CD (AA) (figure 1B,E). These two ER stress markers show a good correlation, with all non-risk allele carriers being negative for GRP78 and p-EIF2α, and 12 out of 14 GG patients showing identical patterns of GRP78 and p-EIF2α expression (see online supplementary table S6). In patients carrying one of the ATG16L1 risk alleles, 17 out of 23 patients showed perfect correlation between GRP78 and p-EIF2α staining, with p-EIF2α expression in the remainder of the patients being more prevalent than GRP78 staining. No histological differences were detected between the three different ATG16L1 genotypes using the PAS—alcian blue and lysozyme staining (figure 1C, and see online supplementary figure S2A). Several CD-associated gene mutations can (in-) directly affect autophagy and Paneth cell function.7 ,8 As such, this might explain the subgroup of patients with ER-stress positive Paneth cells in the ATG16L1 G allele carriers. However, excluding the patients with NOD2 (rs2066844, rs2066845), XBP1 (rs11175593) and IRGM (rs13361189) mutations did not affect the percentage of patients with GRP78+ Paneth cells (figure 2A). In addition, multivariate analysis on the patients with GRP78+ Paneth cells regarding the ER stress associated genes did not reveal significant associations other than ATG16L1 (rs2241880, p<0.002, table 1). Similar results were obtained when stratifying patients according to p-EIF2α staining, confirming the close correlation between these two ER stress markers (figure 2B and table 1)

Table 1

Results of multivariate analysis (ER-stressed Paneth cells vs ER stress related SNPs)

Figure 1

Intestinal crypt characterisation of inactive patients with Crohn's disease (CD) with the ATG16L1 (T300A) GA or GG risk alleles. (A) Representative images of small intestinal cross sections of inactive patients with CD Immunohistochemistry (IHC) stained for GRP78. The red-brown colour indicates a positive GRP78 staining. Specificity of the Paneth cell GRP78 staining is presented in online supplementary figure S2A. Original magnification 1000×. (B) Representative images of small intestinal cross sections of inactive patients with CD IHC stained for phosphorylated EIF2α (pEIF2α). Red-brown precipitate indicates positive pEIF2α staining. Original magnification 1000×. (C) Identical cross sections as in A now stained for PAS—alcian blue and lysozyme. The blue colour indicates mucus, and the crypt based pink-red colour represents granules present in Paneth cells. For the lysozyme staining the immune reactivity in cells at the crypt base indicates the Paneth cells. Multiple examples of the lysozyme staining are presented in online supplementary figure 2A,B. Original magnification 1000×. (D) GRP78 immune-reactivity scored and presented in graph. The graph represents the percentage biopsies with GRP78 positive and negative Paneth cells per genotype. The tissue specimens where scored blinded as positive or negative according to the presence of GRP78 immune-reactivity,*** p<0.001. (E) pEIF2α immune reactivity scored and presented in graph. The graph represents the percentage biopsies with GRP78 positive and negative Paneth cells per genotype. The tissue specimens where scored blinded as positive or negative according to the presence of pEIF2α immune-reactivity, **p<0.01.

Figure 2

ATG16L1 specificity of Paneth cell stress. (A) Other endoplasmic reticulum (ER) stress related genes excluded. Same graph as in figure 1D (GRP78), excluding the patients with mutations in NOD2 (rs2066844, rs2066845), XBP1 (rs11175593) and IRGM (rs13361189), ***p<0.001. (B) Other ER stress related genes excluded. Same graph as in figure 1E (pEIF2α), excluding the patients with mutations in NOD2 (rs2066844, rs2066845), XBP1 (rs11175593) and IRGM (rs13361189), ***p<0.001. (C) GRP78 staining in other parts of the small intestine. Tissue specimens stained for GRP78 from the same GG patient as used in figure 1A, now originating from the duodenum and jejunum. Original magnification 1000×. (D) Colonic Crohn's disease (CD) excluded. Same graph as (B), excluding the patients with primary colonic CD, ***p<0.001.

To establish whether the presence of GRP78+ Paneth cells was a local phenomenon or a constitutive property of the entire small intestine, biopsies from the duodenum and jejunum were investigated as well. Patients with CD with a terminal ileal GRP78+ Paneth cell phenotype, also presented with a thus-affected duodenum and jejunum (figure 2C).

The disease location could provide yet another factor that may be related to the ER-stressed phenotype of Paneth cells. Stratifying our data based on disease location (small intestine vs colonic CD), we observed that GRP78+ and p-EIF2α+ Paneth cells are a feature that is more associated with small intestinal CD (figure 2D, see online supplementary figure S3A and tables 2 and 3).

Table 2

Disease characteristics with or without ER stressed-Paneth cells

Table 3

Disease characteristics per genotype

To determine if ER stressed-Paneth cells are specific for patients with CD, we also stained small intestinal biopsies from healthy controls. Proteins GRP78 and p-EIF2α were increased in expression in controls with the ATG16L1 risk allele (figure 3A). GRP78 expression was enhanced in 60% of controls with the GG alleles (p<0.01, figure 3B). In 33% of the controls with the GA alleles and 50% of those with the GG alleles, increased p-EIF2α protein expression was observed (p<0.05, figure 3C). These data serve to illustrate that the ATG16L1 risk allele is associated with an increased incidence of ER stress in Paneth cells, in mucosae from healthy individuals and patients with quiescent CD. Paneth cells have a clear antimicrobial function, and are important in regulating the intestinal stem cell niche.3 Therefore, we further investigated the biological effects of the ER stress phenotype in the Paneth cells of patients with CD.

Figure 3

Paneth cell endoplasmic reticulum stress in healthy controls. (A) Representative images of small intestinal cross sections of healthy controls IHC stained for GRP78 and pEIF2α. Red-brown staining indicates positive GRP78 staining. Original magnification 1000×. (B) GRP78 expression score. GRP78 expression was scored as in figure 1. The GRP78 score is presented per ATG16L1 genotype. (C) pEIF2α expression score. pEIF2α expression was scored as in figure 1. The pEIF2α score is presented per ATG16L1 genotype.

Changes in mucosal bacterial composition in patients with ER-stressed Paneth cells

The presence of ER stress in Paneth cells from ATG16L1 T300A carrying individuals in itself is not sufficient to confer CD, as we also observed ER stress in ATG16L1 risk allele carrying healthy controls. Hence, we speculated that an aberrant microbial composition, as has been reported in CD, may contribute to the disease. We therefore characterised the presence of several bacterial strains in biopsies from patients with CD and healthy controls. Increased incidence of adherent-invasive Escherichia coli (AIEC), Listeria monocytogenes, Mycobacterium avium (subspecies paratuberculosis) and Salmonella spp. have all been reported in CD.17 ,18 ,19 Our studies show that AIEC was present in biopsies from 100% of patients with CD with ER stress positive Paneth cells, with only 1 of 13 ER stress negative CD patients showing positivity for this bacterium (table 4 and see online supplementary table S7, example AIEC PCR see online supplementary figure S4). In addition, L monocytogenes, M avium (subspecies paratuberculosis) and Salmonella were only detected in a subset of ER-stress positive patients with CD. In contrast, none of these bacterial strains were observed in biopsies from healthy controls, irrespective of the presence of ER stress in their Paneth cells. Thus, it is conceivable that in combination with an altered microbiome, ER-stressed Paneth cells in patients with CD are more susceptible to bacterial insult, which is subsequently not cleared properly.

Table 4

Microbiome composition of patients with CD ER stress (+) and ER stress (−) and healthy controls

ER-stressed Paneth cells do not affect the intestinal stem cell compartment

Recently, it has become clear that the Paneth cell is an important component of the intestinal stem cell niche through for example, Wnt3a production. To see if ER stress could affect the expression and excretion of these growth factors, we chemically induced ER stress in Wnt3a producing L-cells. L-cells stimulated for 18 h with 2µM Tunicamycin showed increased XBP1 mRNA splicing (+150%), as well as elevated levels of GRP78 (15-fold) and CHOP mRNA (10-fold), indicating ER stress in these cells (see figure 4A,B). Using a β-catenin reporter assay, we detected high levels of active Wnt3a in the supernatant of Wnt3a-expressing L-cells (figure 4C). However, Tunicamycin-induced ER-stress significantly reduced the levels of Wnt3a in the supernatant by 10-fold (p<0.001) (figure 4C). Hence, we provide mechanistic evidence that ER stress can affect production of growth factors needed for stem-cell homoeostasis. To investigate the effect of ER-stressed Paneth cells on the intestinal stem cell compartment in patients, the amount of stem cells (PHDLA1+ cells) and the number of Paneth cells per crypt were measured in biopsies. The number of stem cells per crypt did not differ between the groups, AA(−) 4.0; GA(−) 3.4; GA(+) 4.0; GG(−) 3.6; GG(+) 3.2; p=0.79, (figure 5A,B). Furthermore, the amount of Paneth cells per crypt was also not affected by ATG16L1 mutation, AA(−) 6.2; GA(−) 6.5; GA(+) 6.6; GG(−) 6.7; GG(+) 5.6; p=0.92 (see online supplementary figure S3B). In the same way, the amount of Paneth or stem cells is similar when stratifying biopsies according to the presence or absence of ER-stressed Paneth cells (see online supplementary figures S3C,D). From these data we conclude that the ER-stressed Paneth cells do not affect the small-intestinal stem cell compartment and thus the effects seen on crypt inflammation most likely can be attributed to diminished antimicrobial activity of the Paneth cell, whereas the effects on stem cell niche function are less important.

Figure 4

In vitro Wnt3a production in L-cells with and without endoplasmic reticulum stress. (A) XBP1 mRNA splicing was determined from L-cells and Wnt3a producing L-cells, unstimulated, stimulated with Dimethyl Sulfoxide (DMSO), or with 2µM Tunicamyci for 18 h. The ratios of unspliced (U) to spliced (S) XBP1 were plotted. Ctr depicts unstimulated cells, DMSO indicates vehicle control stimulated cells. The gel is a representative image of two independent experiments and the graph represents the mean±SD band intensity of the two independent experiments. (B) GRP78 and CHOP mRNA expression. The same mRNA samples as used in (A) were used to measure the mRNA expression of GRP78, CHOP and Wnt3a by quantitative RT-PCR. N.D. is not detectable. (C) Wnt3a protein production. Conditioned medium from the cells as used in (A) was used in the β-catenin reporter assay. β-catenin reporter activity is measured and presented in a graph as relative fluorescence units, ***p<0.001.

Figure 5

Detection of intestinal stem cells, and intestinal epithelial cells apoptosis and proliferation. (A) Cross sections as used in figure 1A, now stained for PHLDA1. The red-brown immune reactivity represents positive expression. Original magnification 1000×. Endoplasmic reticulum (ER) stress (−) represents the patient with Crohn's disease (CD) group without ER-stressed Paneth cells and ER stress (+) represents the patient with CD group with ER-stressed Paneth cells. (B) Quantification of (A), number of stem cells per crypt. All crypts base PHLDA1 positive cells from at least 20 crypts of each tissue specimen were counted. AA(−): ATG16L1 (T300A) AA allele without ER-stressed Paneth cells, GA(−): GA allele without ER-stressed Paneth cells, GA(+): GA allele with ER-stressed Paneth cells, GG(−): GG allele without ER-stressed Paneth cells, GG(+): GG allele with ER-stressed Paneth cells. The graphs represent the mean±SD and n.s. is not significant. ER stress was determined by GRP78 staining; similar results were obtained when patients were stratified according to p-EIF2α staining (not shown). (C) Cleaved-caspase 3 as apoptosis marker. Cross sections as used in figure 1A, now stained for cleaved-caspase 3. ER stress (−) represents the patients with CD group without ER-stressed Paneth cells and ER stress (+) represents the patients with CD group with ER-stressed Paneth cells. Original magnification 1000×. (D) Apoptosis, only in the villi tips. The same tissue specimens as used in (A) are presented. Immune reactivity represents apoptotic cells, cleaved-caspase 3. Original magnification 200×, magnification of enlarged section 400×. (E) Proliferation in the transamplifying cells. Cross sections as used in figure 1A, now stained for phosphohistone H3 (p-histone H3). Original magnification 1000×. (F) Quantification of crypt cell proliferation. All p-histone H3 positive and negative cells from 20 crypts of each tissue specimen were counted. AA(−): ATG16L1 (T300A) AA allele without ER-stressed Paneth cells, GA(−): GA allele without ER-stressed Paneth cells, GA(+): GA allele with ER-stressed Paneth cells, GG(−): GG allele without ER-stressed Paneth cells, GG(+): GG allele with ER-stressed Paneth cells. The graph represents the mean±SD *p<0.05.

No difference in IEC apoptosis or proliferation in patients with ER-stressed Paneth cells

In apparent agreement with this notion were experiments in which we investigated the functional effects of the observed ER-stressed Paneth cells in vivo. We first determined the amount of apoptotic (cleaved-caspase 3) and proliferating (p-histone H3) IEC per crypt. No cleaved-caspase 3 was detected in the crypts of any of the biopsies, irrespective of Paneth cell ER-stress (figure 5C). In all the biopsies, again irrespective of the presence of ER-stressed Paneth cells, identical numbers of cleaved-caspase 3 positive cells were detected at the top of the villi (figure 5D).

The amount of p-histone H3 positive cells per crypt did not significantly differ between the genotypes and phenotypes (AA(−) 1.3; GA(−) 1.1; GA(+) 1.2; GG(−) 1.5; GG(+) 0.9; p=0.84, see figures 5E,F). These results show that ER-stressed Paneth cells are neither apoptotic themselves nor do they seem to affect apoptosis or proliferation of other IECs in the crypt, and demonstrate that intrinsic defects in Paneth cell functionality probably underlie the ATG16L1 SNP phenotype.

ER stress in Paneth cell is not associated with NF-κB activation

A distinctive subset of patients with quiescent CD, characterised by the ATG16L1 risk allele, demonstrated restricted Paneth cell ER stress, warranting further exploration of the exact characteristics of this phenotype. Activation of Paneth cells by bacterial ligands may enhance the demand of secretory antimicrobial peptides, potentially leading to ER stress.20 Therefore it is possible that the ER stress in the Paneth cells results from constitutive subclinical inflammation in the affected patients. To investigate this we stained and screened all biopsies for the sentinel inflammatory transcription factor NF-κB using its nuclear localisation as an activation marker, as well as its upstream activator IκB, phosphorylation of which corresponds to NF-κB activation.

No nuclear or cytoplasmic p65 expression was found in either ER-stressed or apparently normal Paneth cells (figure 6A). In addition, p-IκB levels were similar between ER stress positive and negative Paneth cells (figure 6C,D). Thus the Paneth cell ER stress is not a secondary consequence of an ongoing inflammatory reaction, but probably represents an intrinsic property of the Paneth cell compartment in ATG16L1 risk allele patients. This is corroborated by the fact that ATG16L1 T300A carrying healthy controls, in which inflammation is almost guaranteed to be absent, also show ER-stressed Paneth cells (figure 3). Interestingly however, increased cytoplasmic p65 expression is seen in the stem cells and the cells of the transamplifying domain from biopsies with ER-stressed Paneth cells (ER stress (+)) (p<0.05, figure 6B), suggesting that the decreased Paneth cell functionality, as a consequence of ER stress, leads to increased inflammation elsewhere in the small intestinal crypt. Another hallmark for inflammation and a general cellular stress regulator is p-p38MAPK. The connection between NFκB and p38 in inflammation is well documented, and p38 phosphorylation is one of the targets of CD treatments such as anti-TNFα and tacrolimus.21–24 We investigated the phosphorylation of p38 in biopsies from CD patients with ER stress (+) and ER stress (−) Paneth cells, and in corroboration with the data above, found no differences between these two groups (figure 6E,F). Based on these data, there is no indication of the presence of subclinical inflammation in ER stress (+) CD patients.

Figure 6

NF-κB (p65), phospho-IκB and phospho-p38 IHC. (A, C, E) Cross sections as used in figure 1A, now stained for NF-κB subunit p65, phosphorylated IκB (p-IκB) and phosphorylated p38 (p-p38), respectively. The red-brown immune reactivity represents positive expression. Original magnification 1000×. (B) Overall p65 immune reactivity of transit amplifying cells presented in a graph. All tissue specimens were scored in a blinded fashion as basal, mildly increased, increased or strongly increased according the amount of positive p65 immune reactivity. The graph represents the percentages of tissue samples with a particular score, subdivided into samples without endoplasmic reticulum (ER) stressed Paneth cells (ER stress (−)) and samples with ER-stressed Paneth cells (ER stress (+)), **p<0.01. (D) Phospho-IκB immune reactivity of Paneth cells presented in a graph. All tissue specimens were scored as percentage positive cells per crypt. The graph represents the average percentage per group, error bar is SEM. ER stress (−) and ER stress (+) samples. (F) Phospho-p38 immune reactivity presented in a graph. All tissue specimens were scored as percentage positive cells per crypt. The graph represents the average percentage per group, error bar is SEM. ER stress (−) and ER stress (+) samples .

Clinical differences in patients with ER-stressed Paneth cells

Although no obvious cellular differences were detected between the ER stress (−) biopsies and the ER stress (+) biopsies, this does not rule out the possibility of an enhanced susceptibility for developing CD or disease severity in these patients, due to impeded Paneth cell function. The clinical characteristics of 78 patients with inactive disease were analysed, 54 of whom were ER stress (−) and 24 whom were ER stress (+). In general the demographic characteristics were not significantly different between the two patient groups (see online supplementary table S3). To investigate the clinical disease severity we collected patient characteristics according to the Montreal classification (tables 2 and 3). As mentioned above, colonic CD occurred less often in ER stress (+) CD patients (17%) than ER stress (−) CD patients (38%, p=0.044) (figure 2D and table 1). We also observed a significant difference in the number of fistula between the ER stress (−) patients with CD (20%) and ER stress (+) CD patients (41%) (p=0.050). In addition, fewer ER stress (−) CD patients (54%) than ER stress (+) patients with CD (92%) underwent intestinal surgery (p=0.002), irrespective of the disease duration (see online supplementary table S3). Although not statistically significant, ER stress (−) patients tended to use more immunosuppressive medication at the time of acquisition of the biopsies. Thus, it is conceivable that immune modulatory drugs, and especially corticosteroids, may alleviate ER stress in Paneth cells. To test this hypothesis, we obtained non-inflamed biopsies from seven patients who were positive for ER stress markers in their Paneth cells while being corticosteroid free, and had subsequently undergone treatment with corticosteroids (prednisolone or budesonide), and stained these for ER stress markers. In two of these seven patients, a reversal of the ER stress phenotype was observed, demonstrating that modulation of Paneth cell ER stress with immunotherapy might indeed be possible (see online supplementary figure S5).

Discussion

Various SNPs strongly associated with CD seem to affect bacterial/viral sensing (NOD2), ER stress (XBP1, ORMDL3) and autophagy (ATG16L1, IRGM) all of which are of crucial importance to normal Paneth cell function. In this study we provide in vivo evidence for the convergence of these different pathways by the detection of ER stress in Paneth cell in a subset of the patients with CD carrying the ATG16L1 T300A risk allele (G).

A previous study performed in seven GG allele CD patients reported that the ATG16L1 mutant genotype coincided with a particular Paneth cells phenotype, comprised of granule abnormalities.8 Autophagy has been implicated as a crucial part of normal ER function and homoeostasis, helping the ER recover from stress.25 Autophagy can provide an additional degradation pathway for misfolded proteins or degrade damaged parts of the ER itself. This intimate relationship between autophagy and ER stress suggests that defects in autophagy may induce ER stress directly or hamper the recovery from stress. In the current study we show the occurrence of ER stress in Paneth cells in a subset of the patients and healthy controls carrying an ATG16L1 risk allele (G), using a previously described sustainable method.11 Our ER-stressed Paneth cell finding was not linked to SNPs in other possible ER stress-associated genes such as XBP1, NOD2 and IRGM, nor to active inflammation-induced ER stress.11 ,26 ,27 Therefore we hypothesised that the ER stress observed in Paneth cells results from a hampered recovery from stress or the increased demand for (antimicrobial) proteins downstream of NF-κB activation.20 ,28 However, using p65 and phospho-IκB staining, we found no indication of increased NF-κB pathway activation in ER-stressed Paneth cells. This conclusion was further strengthened by the lack of increased phospho-p38 expression, as well as the presence of ER-stressed Paneth cells in T300A carrying healthy controls devoid of subclinical inflammation. The overall p65 intensity was significantly higher in the ER-stress positive crypts, despite the fact that all our biopsies were histologically inactive. This can indicate that either the ER-stressed Paneth cells excrete more proinflammatory cytokines, or that due to decreased antimicrobial Paneth cell function, more microbial ligands reach the crypts.29 ,30 Indeed, the observation that the presence of AIEC was significantly associated with an ER stress positive Paneth cell phenotype in patients with CD seems to corroborate this latter hypothesis. AIEC replication was recently described to be more prevalent in ATG16L1 deficient cells.31 ,32 Our study suggests that such may be the case in patients with CD as well, and that this may be related to an increased ER stress in cells from patients carrying the ATG16L1 T300A variant.

The function and differentiation of highly secretory cells such as Paneth cells and goblet cells depend strongly on a tight basal ER stress balance.7 ,8 ,12 Genetic variations (eg, NOD2, XBP1 and ATG16L1) and environmental conditions such as inflammation and ischaemia can easily disturb this balance, which can result in cellular dysfunction and even apoptosis.33 Mutations in the NOD2 gene are associated with decreased defensin production.6 In addition, in Paneth cells of severely obese individuals lower levels of defensins and lysozyme were found, which was associated with enhanced ER stress.34 In our study, no obvious difference in lysozyme expression was detected in the ER-stressed Paneth cells. The lack of low levels of inflammation that are associated with severe obesity may be one of several explanations for this difference. Although activated ER stress pathways can initiate apoptosis (reviewed12 ,33 ,35), no apoptotic (cleaved-caspase 3) Paneth cells could be found in our biopsies with or without ER stress, and similar to the results in ER stress (+) obese patients we also detected no differences in Paneth cell numbers.34 This suggests that either Paneth cells are able to cope with certain levels of ER stress, or other environmental factors influence whether ER stress can lead to cell death.

Recently Paneth cells have been described as important growth factor-producing cells,36 which maintain the intestinal crypt base structure by influencing stem cell homoeostasis.3 As we show that ER stress strongly reduced the Wnt3a production in vitro we hypothesised that ER-stressed Paneth cells could affect the intestinal stem cell population. Using a recently described human stem cell marker37 we could detect slim cylinder-shaped cells at the crypt base. However, no differences in stem cell numbers were detected between the different patient groups. This suggests that the ER-stressed Paneth cells are mildly affected, since severe Paneth cells dysfunction leads to intestinal growth malfunctions38 and spontaneous enteritis.7 ER stress (+) biopsies do not have a different amount of proliferating cells (p-h H3) compared with ER stress (−) biopsies. However, as the stem cells divide slowly, we may not have been able to pick up any subtle differences. Although no major histological changes could be detected between the tissue samples with or without ER-stressed Paneth cells, we observed that ER stress (+) patients with CD have less colonic disease than the ER stress (−) CD patients, which is in line with the observation that XBP1-/- mice develop spontaneous inflammation in their ileum, due to impeded Paneth cell function.7 The increase in ileal disease in ER stress (+) patients with CD in turn may contribute to our observation that these patients more often required intestinal surgery. Interestingly, a recent publication also associates a mutation in IRGM, another autophagy related gene, with an increased need for intestinal surgery in risk allele carrying patients with CD.39

The ER stress (+) patients have more fistulas than the ER stress (−) patients. Since microbes seem to be involved in the aetiology of these fistulas,40 our observation suggests an impeded antimicrobial function of the Paneth cells in the ER stress (+) patients with CD. This is corroborated by our finding of increased persistence of CD-associated bacteria in biopsies from these patients, which may hamper recovery to normal function after inflammation. As such, ER stress resolution or prevention should carefully be considered as CD therapy. Mouse studies have recently shown that corticosteroids can reduce ER stress and restore cellular function by decreasing GRP78 levels and XBP-1 splicing, and enhancement of protein folding.41 Our results indicate that such reversal of the ER-stressed Paneth cell phenotype with corticosteroids might be possible in some, but not all patients. The potential success of treatments in limiting Paneth cell ER stress may well depend on the origin of the ER stress. Although NF-κB-induced and p38-induced ER stress was shown to be reduced by corticosteroids,42 our data suggest that these pathways are most likely not the main cause of Paneth cell ER stress observed in our cohort of patients, calling for other ER stress modulators. Recent studies with a specific ER stress inhibitor have demonstrated an effect in ER-stressed IEC.43 However, future research is needed to explore the possibilities of ER stress inhibitors as therapeutic target for CD patients with the ATG16L1 risk allele.

Acknowledgments

We thank Dr R Smits (Erasmus MC, Rotterdam, The Netherlands) for kindly providing the L-cells and the Wnt3a producing L-cells, Dr K Biermann (Erasmus MC, Rotterdam, The Netherlands) for the histological assessments and T van den Berg (Erasmus MC, Rotterdam, The Netherlands) for assistance in collecting patient characteristics..

References

View Abstract

Review history and Supplementary material

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • JJD and GMF share first authorship.

  • CdH and CJvdW share senior authorship.

  • Contributors JJD, designed and performed study, and wrote the manuscript; GMF, analysed data, wrote the manuscript; SRK, designed and performed experiments; MPP, critical revision of the manuscript; EJK, critical revision of the manuscript; CdH, analysed data, extensive drafting of the manuscript and study supervision, CJvdW, critical revision of the manuscript and study supervision.

  • Competing interests GMF is supported by the Dutch Cancer Society (grant 2010-4737). MPP is supported by a personal ECCO grant.

  • Ethics approval This study is conducted with the approval of the Erasmus MC University Medical Centre, Rotterdam, ethics committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles