Colitis Linked to Endoplasmic Reticulum Stress Induces Trypsin Activity Affecting Epithelial Functions

Abstract

Background and Aims

Intestinal epithelial cells [IECs] from inflammatory bowel disease [IBD] patients exhibit an excessive induction of endoplasmic reticulum stress [ER stress] linked to altered intestinal barrier function and inflammation. Colonic tissues and the luminal content of IBD patients are also characterized by increased serine protease activity. The possible link between ER stress and serine protease activity in colitis-associated epithelial dysfunctions is unknown. We aimed to study the association between ER stress and serine protease activity in enterocytes and its impact on intestinal functions

Methods

The impact of ER stress induced by Thapsigargin on serine protease secretion was studied using either human intestinal cell lines or organoids. Moreover, treating human intestinal cells with protease-activated receptor antagonists allowed us to investigate ER stress-resulting molecular mechanisms that induce proteolytic activity and alter intestinal epithelial cell biology.

Results

Colonic biopsies from IBD patients exhibited increased epithelial trypsin-like activity associated with elevated ER stress. Induction of ER stress in human intestinal epithelial cells displayed enhanced apical trypsin-like activity. ER stress-induced increased trypsin activity destabilized intestinal barrier function by increasing permeability and by controlling inflammatory mediators such as C-X-C chemokine ligand 8 [CXCL8]. The deleterious impact of ER stress-associated trypsin activity was specifically dependent on the activation of protease-activated receptors 2 and 4.

Conclusions

Excessive ER stress in IECs caused an increased release of trypsin activity that, in turn, altered intestinal barrier function, promoting the development of inflammatory process.

1. Introduction

Inflammatory bowel diseases [IBD], including Crohn’s disease [CD] and ulcerative colitis [UC], are chronic idiopathic inflammatory disorders of the gastrointestinal tract. Genetic susceptibility together with environmental factors lead to an exaggerated gut immune response towards commensal microbiota, triggering chronic inflammation of the bowel.^1,2 Over the past 10 years, genetic studies have identified up to 242 susceptibility loci involved in either casual variants, or in very-early-onset development of IBD, including X-box binding protein 1 [XBP1], a gene involved in endoplasmic reticulum stress [ER stress] resolution.^3,4

The ER is an organelle responsible for the synthesis of proteins and lipids, maturation of proteins and regulation of intracellular calcium. Alteration of ER functions leads to cellular dysfunction termed ER stress.^5–7 ER stress triggers the unfolded protein response that consists in three stress sensors: inositol-requiring kinase/endonuclease 1 [IRE1], pancreatic ER kinase [PERK] and activating transcription factor 6 [ATF6]; these are all essential for cells to adapt to a harmful environment. Under ER stress, activated IRE1 removes a 26-bp nucleotide intron from mRNA encoding XBP1.⁸ The XBP1 spliced form then regulates expression of key genes implicated in tissue homeostasis and control of inflammation such as: regulation of protein folding, protein quality control and degradation, and phospholipid synthesis.^5,8,9

Previous studies have shown that ER stress is linked to intestinal inflammation.^10–14 Genetic deletion of Xbp1 in intestinal epithelial cells caused spontaneous inflammation in mice.¹³ In humans, colonic mucosa of IBD patients exhibits a dysregulation of ER stress, with high levels of the XBP1 splicing gene. Levels of glucose-regulated protein 78-kDa [GRP78], a marker of ER stress induction and XBP1 splicing genes, are increased in the inflamed gut of CD patients.^13–15 UC patients display increased mRNA levels of spliced XBP1, glucose-regulated protein 94-kDa [GRP94] and ATF6 in uninflamed colonic mucosa, while GRP78 expression is upregulated in both uninflamed and inflamed mucosa.^12–14 Although the consequences of ER stress on intestinal inflammation are well characterized, the mechanisms by which ER stress leads to inflammatory signs in IBD has only partially been revealed. What mediators released upon ER stress by intestinal epithelial cells participate in IBD inflammatory signs? Colonic tissues and stools of IBD patients display increased amounts of proteases, including metalloproteinase and serine proteases.^16–19 However, the cellular origin of this activity, the nature of proteases released and the factors involved in the induction of proteolytic activity are unclear. Recent studies have identified the colonic epithelium as an important source of serine protease activity: elastolytic and trypsin-like activity.^18,20 Serine proteases can signal through different mechanisms, including the degradation or activation of molecules or even through the cleavage and activation of protease-activated receptors [PARs]. These receptors modulate a number of cellular signals, including ion transport, barrier function and inflammatory mediator release.^21–25

In this study, we hypothesized that active epithelial proteases could be released upon ER stress induction and thereby participate to generate inflammatory features in the context of IBD. We identified that trypsin activity in tissues from IBD patients is associated with the epithelium. Moreover, we deciphered the molecular mechanism by which ER stress induced proteolytic activity and alters intestinal epithelial cell biology.

2. Methods

2.1. Patients

Colonic tissues were obtained from individuals referred to the Gastroenterology Unit of the Centre Hospitalier de Toulouse [France] [Supplementary Table 1]. Biopsies were collected during colonoscopy for clinical evaluation of the disease of established and well-characterized CD [n = 28] and UC [n = 10] patients or taken from individuals undergoing colon cancer screening who were otherwise controls [n = 14] [Supplementary Table 1]. Biopsies from controls and non-inflammatory area from IBD patients did not show macroscopic or microscopic signs of inflammation [data not shown]. Written and verbal informed consent was obtained before enrolment in the study, and the Ethics Committee approved the human research protocol [Comité d’Ethique sur les Recherches Non Interventionnelles, NCT01990716]. Biopsies were embedded in optimal cutting temperature [OCT] compound [Dako] at −186°C and stored for in situ trypsin activity.

2.2.In situ zymography of colonic biopsies

Frozen OCT sections of colonic tissues from patients [8-μm thickness] were permeabilized with PBS 2% Tween-20, rinsed with washing solution [PBS] and incubated at 37°C overnight with N-p-Tosyl-GPR-amino-4-methylcoumarin hydrolchloride [50 µg/mL, Sigma] together with 0.3% low-melting-point agarose, as previously described.²⁰ Nuclei were stained with Topro3 [Invitrogen]. All sections were visualized with an LSM710 confocal microscope [Carl Zeiss] and analysed by observers blinded to patient subgroup, with Zen 2009 software [Carl Zeiss].

2.3. Cell culture and reagents

Caco-2 cells were purchased from the DSMZ collection and HT29-MTX cells were kindly provided by T. Lesuffleur [INSERM Unit 843, Paris France]. They were grown in GlutaMAX DMEM [Gibco] supplemented with 10% of heat-inactivated fetal bovine serum [Biowest], 1% non-essential amino acids and 1% antibiotics [100 U/mL penicillin, 100 mg/mL streptomycin [Gibco] at 37°C in a 5% CO₂ water-saturated atmosphere.²⁶ The medium was changed three times per week and cells were passaged once a week. Cells were grown to confluence as a monolayer in transwell inserts, 12-well plates of 12-mm, polyester membrane, 0.3-µm pore size [Costar] for 20 days.

ER stress was induced by adding Thapsigargin or Tunicamycin [both10 µg/mL, Sigma] in the culture media as previously described.²⁷ Before stimulation, cells were washed twice with Ca²⁺/Mg²⁺-free PBS [Sigma]. Trypsin-like activity was inhibited by adding serine protease inhibitor AEBSF [200 µM, Sigma], or trypsin-like protease inhibitor Leupeptin [50 µM] to the culture medium. Caco-2 cells were pre-treated for 45 min with PAR antagonists, PAR2 [GB83, 10 µM, Axon Medchem] and PAR4 [ML354, 0.5 µM, Tocris Bioscience], and ML-7 [20 µg/mL, Sigma] before Thapsigargin stimulation.

2.4. Organoid culture

Biopsies from control patients from our human biological banking numbered DC-2015–2443 were washed with Dulbecco’s PBS [Gibco] and incubated in an antibiotic solution containing gentamicin [50 µg/ml, Sigma], normocin [100 µg/ml, InvivoGen] and amphotericin [2.5 µg/ml, Gibco] four times for 5 min. Next, biopsies were incubated in a 10 mM DTT [Roche] solution, before transfer in 8 mM EDTA [Ambion] solution under gentle rocking for 1 h at 4°C. After removal from the EDTA solution, biopsies were re-suspended in PBS to isolate intestinal crypts. Heat inactivated fetal bovine serum [FBS, Life Technologies] was added [5%] and crypts were centrifuged at 40 g for 5 min. Crypts were washed [three times] with a solution containing DMEM/F12 [Invitrogen] supplemented with Glutamax‐CTS [Invitrogen] and HEPES [Gibco], counted and centrifuged once again. Next, hES Matrigel [Corning] was added to the pellets; 25 µL of Matrigel containing 50 crypts was plated in each well of a pre‐warmed 48‐well plate [Greiner] with culture media: 50% advanced DMEM/F12, 50% conditioned medium (supernatants from L Wnt‐3A cells [ATCC CRL‐2647]], 2 mM Glutamax‐CTS, 10 mM HEPES, B27 minus Vitamine A 1× [Life Technologies], N2 1× [Invitrogen], N‐acetylcysteine 1 mM [NAC; Sigma], nicotinamide 10 mM [Sigma], recombinant human epithelial growth factor 50 ng/mL [Gibco], human Noggin 100 ng/ mL[Tebu], human R‐spondin 1 µg/mL [Bio-techne], gastrin 10nM [Sigma], SB202190 10 µM [Sigma], LY2157299 0.5 µM [Axon MedChem] and Prostaglandin E2 0.01 µM [Sigma]). Cultures were incubated in a humidified incubator at 37°C and 5% CO₂. Culture medium was changed every 3 days without NAC and LY2157599. Following 15 days of culture, ER stress was induced by 10 µg/mL of Thapsigargin [Sigma] diluted in 0.3% DMSO into culture media without FBS for 6 and 16 h.

2.5. Trypsin-like activity

Trypsin-like activity was measured in basal and apical supernatants as previously described²⁸ with the substrate N-p-Tosyl-GPR-amino-4-methylcoumarin hydrochloride [0.1 mM, sigma] in 50 mM Tris, 10 mM CaCl₂ buffer [pH 8]. The rate of hydrolysis was measured by the change in fluorescence [360/460 nm excitation/emission wavelengths] every 30 s for 15 min at 37°C on a NOVOstar microplate reader 96-well plate. Activity was standardized to the rate generated by known concentration of porcine pancreas trypsin [Sigma].

2.6. Real-time RT-PCR analysis

Total RNA was extracted with the NucleoSpin RNA/Protein Kit [Macherey-Nagel] and converted to complementary DNA using the Maxima First Strand cDNA Synthesis Kit for reverse transcription-quantitative PCR [RT-qPCR] [Thermo Scientific].²⁹ Real-time PCR was performed using SYBR Green Master I Kit [Roche], sense- and antisense-specific primers [see Supplementary Table 4] in a LightCycler 480 Instrument [Roche]. After amplification, the relative expression of mRNA was determined with the 2^-DDCt method by using hGAPDH as a reference gene.

2.7. Measurement of paracellular permeability

Paracellular permeability of differentiated Caco-2 monolayers was monitored as previously described.³⁰ Briefly, dextran-fluorescein isothiocyanate [FITC] [4 kDa, Sigma] was added to the apical compartment of a transwell, with the corresponding reagents.³¹ Paracellular permeability was monitored by the passage of dextran from the apical to the basolateral medium, as previously described.²⁶

2.8. Measure of CXCL8, β-defensin-2 [βDF2], Trefoiled factor 3 [TFF3] and mucin2 [MUC2] protein expression

Supernatant from cells was stored at −80°C. Concentrations of CXCL8, βDF2, TFF3 and MUC2 were measured by enzyme-linked immunosorbent assay from, respectively, BD Biosciences, Elabioscience, Boster and St John’s Laboratory according to the manufacturers’ instructions.³²

2.9. Mass spectrometry analysis

Protein samples were loaded on a 1D SDS-PAGE gel [0.15 × 8 cm] and the electrophoretic migration was stopped as soon as the proteins entered the separating gel, in order to isolate all proteins in a single gel band. The corresponding gel slice was excised and washed twice with 100 mM ammonium bicarbonate and once with 100 mM ammonium bicarbonate/acetonitrile [1:1]. Proteins were in-gel digested using 1 µg of modified sequencing-grade trypsin [Promega] in 50 mM ammonium bicarbonate overnight at 37°C for sample digestion. For double digestion, sample was first digested using 1 µg GluC overnight at 37°C and then the sample was digested using 1 µg tryspin overnight at 37°C. The resulting peptides were extracted from the gel by successive incubations in 50 mM ammonium bicarbonate and 10% formic acid/acetonitrile [1:1]. The extracted fractions were pooled with the initial digestion supernatant, dried in a speed-vac, and resuspended with 5% acetonitrile and 0.05% trifluoroacetic acid [TFA] for MS analysis.

Peptides were analysed by nanoLC-MS/MS using an UltiMate 3000 RSLCnano system coupled to a Q-Exactive-Plus mass spectrometer [Thermo Fisher Scientific]. Five microlitres of each sample was loaded on a C-18 precolumn [300 µm ID × 5 mm, Dionex] in a solvent made of 5% acetonitrile and 0.05% TFA and at a flow rate of 20 µL/min. After 5 min of desalting, the precolumn was switched online with the analytical C-18 column [75 µm ID × 15 cm, Reprosil C18] equilibrated in 95% solvent A [5% acetonitrile, 0.2% formic acid] and 5% solvent B [80% acetonitrile, 0.2% formic acid]. Peptides were eluted using a 5–50% gradient of solvent B over 125 min at a flow rate of 300 nL/min. The Q-Exactive-Plus was operated in a data-dependent acquisition mode with the XCalibur software. Survey scan mass spectra were acquired in the Orbitrap in the 350–1500 m/z range with the resolution set to a value of 70 000. The ten most intense ions per survey scan were selected for higher-energy C-trap dissociation fragmentation. Dynamic exclusion was employed within 30 s to prevent repetitive selection of the same peptide. One injection was performed for each sample.

2.10. Protein identification and quantification

Raw MS files were processed with the MaxQuant software [version 1.5.2.8] for database searches with the Andromeda search engine and quantitative analysis. Data were searched against human entries of the Swissprot protein database. Carbamidomethylation of cysteines was set as a fixed modification whereas oxidation of methionine and protein N-terminal acetylation were set as variable modifications. The specificity of trypsin or tryspin-GluC digestion was set and two missed cleavage sites were allowed. The precursor mass tolerance was set to 20 ppm for the first search and 4.5 ppm for the main Andromeda database search. The mass tolerances MS/MS mode was set to 20 ppm. Minimum peptide length was set to seven amino acids, and the minimum number of unique peptides was set to one. Andromeda results were validated by the target-decoy approach using a reverse database at both a peptide and protein false discover rate (FDR) of 1%. For label-free relative quantification of the samples, the ‘match between runs’ option of MaxQuant was enabled with a time window of 0.7 min, to allow cross-assignment of MS features detected in the different runs.

The ‘LFQ’ metric from the MaxQuant ‘protein group.txt’ output was used to quantify proteins. To characterize the effect of Thapsigargin treatment, treated/non-treated ratios were calculated. Potential proteins of interest were selected based on a ratio >2 or <0.5.

2.11. Statistical analysis

Results are expressed as mean ± SEM except for immunostaining and in situ zymography quantification of biopsies, where each dot represents one patient. Statistical analyses were performed using GraphPad Prism 5.00 [GraphPad software] software package for PC. Multigroup comparisons were performed using a one-way analysis of variance followed by a Bonferroni correction for multiple tests. Two-group comparisons were performed using an unpaired t test not assuming the Gaussian distribution. The Gaussian distribution was tested by a Kolmogorov–Smirnov test. A value of p < 0.05 was considered statistically significant. All p-values were two-sided.

3. Results

3.1. Trypsin activity in IBD patient colon biopsies is associated with the epithelium

Trypsin-like activity was quantified after incubation of colonic biopsies for 1 h in Hank’s balanced salt solution [Figure 1A] [see Supporting Table 1 for patient information]. Trypsin-like activity was constitutively released from colonic biopsies. However, tissues from inflamed areas of IBD patients (both CD [n = 28] and UC [n = 10]) released higher amounts of trypsin activity in comparison to non-inflamed areas or Control [n = 14] [Figure 1A]. Additionally, in situ zymography using a substrate for trypsin-like activity was performed in biopsies and confirmed a stronger trypsin-like activity in CD and UC patients, compared to Control [Figure 1B]. Photomicrographs of in situ zymography also revealed that most of the trypsin-like activity detected in biopsies from CD and UC patients was associated with the epithelium [Figure 1B left panel]. However, in UC colonic tissues, non-epithelial cells, probably immune cells, also showed increased trypsin-like activity. Quantification in tissue sections showed that both overall [Figure 1B right panel] and epithelium-associated trypsin-like activity [data not shown] were significantly increased in tissues from CD and UC patients. These results suggest that intestinal epithelial cells are a major source of trypsin-like activity, and are particularly increased in IBD.

3.2. ER stress in intestinal epithelium provokes the apical release of trypsin-like activity

We investigated whether the induction of ER stress in human intestinal epithelial cells could upregulate the release of trypsin-like activity. Monolayers of differentiated Caco-2 cells cultured in transwells were stimulated with Tunicamycin or Thapsigargin, two ER stress inducers. Both inducers triggered the release of trypsin-like activity, specifically at the apical compartment, compared to non-stimulated cells, with an activity peak 6 h post-stimulation [Figure 1C], while activity levels were completely inhibited by AEBSF incubation, a large-spectrum serine protease inhibitor [data not shown]. Increased trypsin proteolytic activity due to ER stress activation was inhibited by the addition of 4-PBA, a chemical ER stress inhibitor [Figure 1D]. ER stress gene markers [XBP1s, ATF4 and ATF6] confirmed ER stress induction after 6 h of treatment [Figure 1E]. We also confirmed that both Tunicamycin and Thapsigargin did not increase lactate deshydrogenase [LDH] activity, suggesting lack of cell lysis for all culture conditions [data not shown].

3.3. Trypsin genes are associated with intestinal epithelial ER stress

Hence, we investigated by RT-qPCR the amount of mRNA from trypsin genes present in unstimulated vs ER stress conditions, and which unfolded protein response (UPR) axis, under Thapsigargin, is responsible for trypsin activity increase. In ER stress conditions, Caco-2 cells showed a significant increase in mRNA expression of PRSS1 [gene of Trypsin-1] and PRSS2 [gene of Trypsin-2] at 6 h of stimulation, but a decreased expression of PRSS3 [gene of Trypsin-3] [Figure 2A]. To understand by which UPR axis [XBP1s, ATF4 or ATF6] trypsin proteolytic activity was increased, we used Spearman’s correlation coefficient to correlate the expression of ER stress markers: XBP1s, ATF4 or ATF6, with trypsin gene candidates: PRSS1, -2 and -3 [Supporting Figure 1A]. Correlations showed an overall activation of the three UPR pathways associated with a boost of trypsin activity [Supporting Figure 1A].

Similarly, mRNA extracted from IBD colonic biopsies displayed upregulation of PRSS2, while PRSS3 was downregulated compared to Control. PRSS1 did not reach significance [Figure 2B]. Of note, PRSS2 expression was upregulated [60-fold] in UC biopsies. The ER stress marker XBP1 was upregulated only in UC, compared to CD and Control biopsies [Figure 2C]. In contrast, mRNA expression of ER stress markers ATF4 and ATF6 were similar in Control, CD and UC biopsies [Figure 2C]. Together, these data indicated a correlation between an increased expression of PRSS1 and PRSS2 in intestinal epithelial cells submitted to ER stress, or in IBD patients, with an increased trypsin-like activity released under similar conditions by human intestinal epithelium. By using Spearman’s coefficient, we investigated which UPR axis was responsible for the upregulation of trypsin genes. Human colonic biopsies displayed enhanced trypsin activity associated with increased XBP1s mRNA expression. Moreover, PRSS2 was the only trypsin-like gene associated with enhanced levels of XBP1s [Supporting Figure 2], whereas PRSS3 expression was negatively correlated with XBP1s mRNA levels. Furthermore, as previously observed in stressed Caco-2 cells, organoid cultures from Control patients stimulated with Thapsigargin for 16 h exhibited a strong increase of PRSS1 and PRSS2 mRNA expression, while mRNA levels of PRSS3 were not affected [Supporting Figure 3].

Finally, we tried to confirm the protease[s] underlying the increased activity detected in ER stress-stimulated supernatants of Caco-2 cells by MS. Unbiased quantitative MS-based proteomics analysis could not identify with sufficient confidence any serine proteases that would be upregulated in ER stress condition compared to unstimulated Caco-2 cells. This can be explained by the fact that enzymes might be very active, but not necessarily present in large amounts. However, these analyses revealed the existence of a large spectrum of serine protease inhibitors [serpins] of varying abundances being either over- or under-represented in stress conditions, with an overall downregulation of secreted serpin levels [Table S2]. Among these, serpin A1, serpin F2 and serpin H1 are among the most abundant serpins detected. Quantitative proteomics analyses revealed, for the first time, that Caco-2, under ER stress, can modulate the apical secretion of a large spectrum of serpins, with a potential increase of serine protease activity, as a consequence. Notably, the decrease of serpin A1, also known as alpha1 anti-trypsin, could directly impact the elevated level of trypsin-like activity measured in ER stress conditions. As a whole, our results indicate that ER stress triggers an inbalance in proteolytic activity by upregulating tripsin gene expression, while downregulating endogenous inhibitors.

3.4. ER stress-induced trypsin-like activity modifies epithelial biology

We investigated whether ER stress induction or its associated release of proteolytic activity were able to disrupt barrier function of differentiated Caco-2 monolayers. Figures 3 and 4 show that the induction of ER stress via Thapsigargin exposure on Caco-2 cells increased paracellular permeability [Figures 3A and 4A], mRNA expression of antimicrobial peptides [HBD2 and TFF3], mucins [MUC2] and the inflammatory chemokine and cytokine CXCL8 and TNFα [Figures 3B and 4B]. It also increased CXCL8 protein secretion both apically and baso-laterally in Caco-2 monolayers and apical protein secretion of βDF2, TFF3 and MUC2 [Figures 3C and 4C]. These abnormalities were restored when Caco-2 cells were pre-incubated with 4-PBA [an ER stress inhibitor] before being challenged with Thapsigargin for 6 h [Figure 3A–C]. These results confirmed that the epithelial alterations associated with Thapsigargin exposure were indeed due to ER stress induction.

Apical addition of protease inhibitors [AEBSF or Leupeptin] inhibited some of the ER stress-induced epithelial alterations, such as increased permeability [Figure 4A] and mRNA expression of HBD2 and TFF3 [Figure 4B]. In contrast, addition of AEBSF or Leupeptin to Thapsigargin-treated cells did not modify ER stress-induced MUC2, CXCL8 and TNFα mRNA overexpression [Figure 4B] or protein expression [Figure 4C], with the exception of CXCL8 protein, which was upregulated [Figure 4C]. These data showed that ER stress-induced proteolytic activity contributed to the loss of epithelial barrier integrity and to the upregulation of HBD2 and TFF3 expression, while MUC2, CXCL8 and TNFα mRNA overexpression seemed to be directly regulated by the induction of ER stress.

We also monitored the release of trypsin-like activity on the apical compartment of HT29-MTX cells [a muco-secreting cell line], another human intestinal epithelial cell line, after ER stress induction by Thapsigargin for 2, 4 and 6 h [Figure 5A]. A polarized secretion of trypsin activity by intestinal epithelial cells was observed upon ER stress induction. In these cells, ER stress also induced increased mRNA levels of MUC5 and HBD2 [Figure 5B]. Relative expressions of both MUC5 and HBD2 were restored to basal levels when ER stress-stimulated HT29-MTX cells were incubated in the presence of AESBF [Figure 5B]. Unlike Caco-2 cells, Thapsigargin-treated HT29-MTX released higher amounts of CXCL8 only in the apical compartment [Figure 5C]. Therefore, in two different human intestinal epithelial cell lines, we demonstrated that many but not all of the effects associated with ER stress and causing epithelial dysfunctions were due to an excessive release of trypsin proteolytic activity.

3.5. ER stress-associated trypsin activity alters barrier function of Caco-2 monolayers by activating PAR2 and PAR4

We also searched for the mechanism of action of trypsin-like activity on ER stress-associated altered barrier function, investigating the potential role of PAR2 and PAR4, two receptors known to be activated by trypsins.^33,34 ER stress induced by Thapsigargin increased mRNA levels of PAR2 and PAR4 [Figure 6A], a feature also found in IBD patients.³⁵ Pre-treatment with a PAR2 antagonist [GB83] and/or PAR4 antagonist [ML354] blocked the increase of ER stress-induced paracellular permeability in Caco-2 cells [Figure 6B]. This confirmed the trypsin-like-induced increased permeability in Caco-2 cells. PAR2 is known to increase myosin Light Chain Kinase (MLCK) phosphorylation that leads to epithelial cell cytoskeleton contraction and enhanced mucosal permeability.³⁶ Pre-incubation with ML-7, an MLCK inhibitor, also restored normal barrier function, inhibiting the effects of apical trypsin-like activity on increased epithelial permeability [Figure 6B]. These data show that ER stress-induced trypsin activity modulates intestinal permeability through the activation of PAR2 and PAR4 that, in turn, phosphorylate MLCK, a protein involved in epithelial cell contraction. As observed with the serine protease inhibitor AEBSF, PAR2 and PAR4 antagonists also enhanced protein secretion of CXCL8 [Figure 6C]. Finally, pre-treatment with a PAR2 antagonist [GB83] and/or PAR4 antagonist [ML354] did not modulate the increase of apical trypsin activity induced by ER stress in Caco-2 cells [Figure 6D].

3.6. Apical trypsin activity does not modify ER stress processes

ER stress markers [XBP1s, ATF4, ATF6 and CHOP], increased by Thapsigargin, were not modified by the addition of AEBSF either on Caco-2 [Figure 7] or on HT29-MTX [data not shown]. These results suggest that ER stress-associated proteolytic activity does not affect or control the induction of ER stress by feedback mechanisms.

4. Discussion

Abnormal induction of ER stress and increased serine protease activity in colonic tissues have been observed in IBD patients,^17,37 as well as in mouse colitis models.^{10,12,27,38,39} In this study, we confirmed that trypsin activity was amplified in colonic tissues from IBD patients, both UC and CD, compared to Controls. We also showed that the IBD patient tissue-associated trypsin activity was detected mostly in the epithelial layer. ER stress enhanced apically released trypsin activity by intestinal epithelial cells, as demonstrated in the Caco-2 cell line. We showed that trypsin activity was responsible for ER stress-induced alterations of epithelial homeostasis, through a mechanism involving the activation of PAR2 and PAR4.

Previous studies reported that colonic tissues and luminal contents in inflammatory mouse models and IBD patients display excessive serine protease activity.^17,22,38 Moreover, large amounts of thrombin and aminopeptidase B were found in colonic tissue supernatant from CD and UC, respectively.¹⁸ Although researchers are aware of high amounts of cellular serine protease activity, some studies have highlighted the epithelium as an important cellular source of proteolytic activity, with a particular focus on thrombin.^40,41 The role of proteases released from infiltrated immune cells has been suggested, as well as a possible pancreatic or even a microbial source.^22,42 In this study, we demonstrated that colonic mucosa from IBD patients releases higher trypsin proteolytic activity than Controls, and this activity was associated with the epithelium. Although UC biopsies also displayed trypsin activity in non-epithelial cells—probably immune cells—most of the proteolytic activity overlaps within epithelial cells.

Interestingly, in a recent study, similar increased trypsin-like activity has been detected in human intestinal epithelial cells from irritable bowel syndrome patients.²⁰ Trypsin-like activity has been associated with an increased Trypsin-3 expression and secretion, and could be upregulated by stress hormones or bacterial motifs.²⁰ In a previous study, we found that Caco-2 express the three forms of trypsin: PRSS1, PRSS2 and PRSS3.²⁰ However, here we showed that PRSS3 expression is downregulated in IBD intestinal mucosa, as well as in ER-tress induced Caco-2. Of note, cationic [PRSS1] and anionic [PRSS2] trypsins share 96% homology. However, in the present study, we were not able to unequivocally identify the trypsin form[s] responsible for the activity released in response to ER stress. By using Spearman’s correlation coefficient, we have tried to associate the expression of UPR axis markers: XBP1s, ATF4 or ATF6, with trypsin gene forms: PRSS1, -2 and -3. Caco-2 cell correlations were not strong enough to determine the ER stress axis that triggers trypsin activity or the trypsin gene candidate responsible for the release of yrypsin-like activity. In human colonic biopsies, Spearman’s correlations revealed increased trypsin activity, specifically PRSS2 levels, with XBP1s mRNA expression. Human organoid culture stimulated with Thapsigargin for 16 h also displayed increased levels of PRSS1 and PRSS2. Similarly, Caco-2 cells, colonic biopsies and human stimulated organoids displayed decreased PRSS3 mRNA levels. The lack of specific tools hampered us from determining which enzyme was specifically responsible for ER stress-induced proteolytic activity. Indeed, commercially available antibodies were tested for western blot analysis, but they failed to detect selective bands and demonstrated strong cross-reactions between the different forms of pure trypsin. MS analysis was therefore used but failed to detect trypsin-like proteases in supernatants of intestinal epithelial cells stimulated to induce ER stress. This might be due to the low quantity of the active protease[s] in the supernatants of ER stress-stimulated cells with regard to other proteins. Indeed, proteases are extremely efficient catalysers of peptide bond breakdown and therefore are expressed in very low levels. No selective commercially available inhibitor could discriminate between the activity of the three epithelial forms of trypsin: Trypsin-1, Trypsin-2 and Trypsin-3. Furthermore, we tried to apply a short hairpin RNA [shRNA] method targeting Trypsin-1, Trypsin-2 and Trypsin-3 expression in Caco-2 cells; given the high level of homology between these genes, all PRSS gene expressions [PRSS1, PRSS2, PRSS3] were affected, preventing the selective inhibition of only one form. Unexpectedly, MS-based quantitative proteomics revealed the presence of serpins, several under-represented upon ER stress induction [SerpinE1, SerpinA6, Serpin1, SerpinA5, SerpinA1, SerpinF2, SerpinA3, SerpinG1, SerpinA4 and SerpinC1], others over-represented [Serpin H1-, Control-specific-S Serpin A10 and S SerpinD1-, and ER stress-specific-SerpinB1 and SerpinB6]. SerpinA1, also known as an Alpha-1-antitrypsin, is the most abundant serpin found in the plasma.⁴³ It interacts not only with trypsin but also with chymotrypsin and human leukocyte elastase, and its deficiency has been linked to lung and liver disease.^44,45 This suggests that the ER stress-associated increase in trypsin-like activity could also be due to a decreased presence of endogenous inhibitors. It is interesting to note that no endogenous inhibitor of Trypsin-3 has yet been described, and that Trypsin-3 is actually known for being able to degrade protease inhibitors.^46,47 At the mRNA level, Trypsin-3 was down-regulated in response to ER stress, and if this reflects also a down-regulation of active Trypsin-3 protein, it could contribute to the increased presence of serpins.

Beside their role as serine proteases inhibitors, serpins are involved in many biological functions, including: coagulation, inflammatory response, immunity and cancer prevention.⁴⁸ For example, SerpinA5 is known to be a potent antimicrobial factor; a reduced expression induced by ER stress could therefore participate in the inflammatory process in vivo.⁴⁹ In contrast, although plasma levels of SerpinA1 are correlated with UC severity,⁵⁰ we found a reduced secretion by Caco-2 induced by ER stress. In UC and experimental colitis, increased levels of SerpinH1 in intestinal mucosa are strongly involved in carcinoma development and fibrosis.^51,52 Similarly to the increased level of SerpinB1 in colonic samples from UC patients, ER stress enhances its secretion on Caco-2 monolayers.⁵³

Abnormal induction of ER stress in intestinal epithelial cells is associated with IBD pathophysiology.³⁷ Previous studies have shown aberrant induction of ER stress in the colonic mucosa of UC patients.^12,13 Likewise, our results confirm that colonic biopsies of UC, but not CD, carry an abnormal induction of ER stress by upregulation of XBP1s. IBD colonic tissue also displayed upregulation of Trypsin-1 and Trypsin-2, but not Trypsin-3 mRNA. We revealed that the release of trypsin activity by intestinal epithelial cells was a consequence and not a cause of ER stress. Indeed, in our study, most of the disruptive effects of ER stress on intestinal epithelial barrier were mediated by the apical release of trypsin activity. Specifically, the upregulation of βDF2 and TFF3 mRNA expression, protein level, as well as increased permeability induced by ER stress, were caused by apical release of trypsin-like activity. In contrast, increased mRNA expression of MUC2, CXCL8 and TNFα caused by ER stress was not linked to serine protease activity. This finding constitutes a breakthrough in understanding ER stress pathways and effectors in the context of epithelial barrier function.

Interestingly, inhibition of trypsin-like proteolytic activity [AEBSF] and PAR2 and PAR4 antagonists, after ER stress, further enhanced the detection of CXCL8 protein in the apical compartment of enterocytes and muco-secreting monolayers. These data point to a receptor-mediated effect, through which PAR2 and 4 limit the production of CXCL8 induced by ER stress. Besides pro-inflammatory effects on the basolateral side attracting neutrophils, CXCL8 is also described to maintain intestinal epithelium homeostasis [intestinal maturation, migration, differentiation and wound healing].^54–56

Proteases use different mechanisms of action for cell signalling, including proteolytic cleavage of molecules or receptors to induce diverse intracellular signals. Proteolytic processing by proteases is required to mature antimicrobial peptides such as the human defensin-5 or to increase the chemotactic activity of CXCL8 [cleaved by proteinase-3] or CXCL5 [cleaved by cathepsin G].^57,58 In contrast, proteases such as cathepsin G, proteinase-3 or elastase 2 can also degrade other cytokines, including IL-6 or TNFα and proteins involved in cellular contacts.^59–63 In addition, proteases are also able to activate, by proteolytic cleavage of their extracellular N-terminal domain, the PARs.^64,65 In the intestine, PAR cleavage activates ion exchange, motility, nociception, permeability and secretion.^{20,23,66–68} Previous studies have demonstrated that both PAR2 and PAR4 are activated by trypsin-like enzymes.⁶⁹ In addition, PAR2 activation is known to increase epithelial permeability, via MLC phosphorylation.⁶⁷

Along the same lines, our data showed that pre-treatment with PAR2 and/or PAR4 antagonists inhibited the effects of ER stress on the epithelial monolayer. Moreover, Caco-2 cell monolayers pre-treated with ML-7, an MLCK inhibitor, normalized abnormal permeability, compared to Control conditions. This is in agreement with our previous finding reporting the involvement of PARs in increased permeability via MLCK. This provides further insights into the mechanisms by which ER stress-associated proteolytic activity regulates barrier function.

In conclusion, our in vitro and ex vivo data indicate that ER stress induction might trigger increased epithelial permeability in IBD, particularly in UC, through a trypsin- and PAR-dependent mechanism. Specifically, we identified that the release of trypsin-like activity and further activation of PAR2 and PAR4 are downstream events from ER stress in intestinal epithelial cells. The inhibition of trypsin-like activity might be looked at as a valid therapeutic target for intestinal inflammation in the context of UC.

Funding

This work was supported by grants from the French Ministry of Research and Technology, the National Agency for Research [ANR JCJC-11JSV1 001 01-IBDase] to CD, and the European Research Council [ERC- 310973 PIPE] to NV.

Conflict of Interest

No conflict of interest exists.

Acknowledgments

We thank the Toulouse-Purpan imaging core facility directed by Sophie Allart for help with imaging techniques. We also thank Virginie Payssan [CHU Purpan] for the biopsy collection and database management. The authors thank the Image Core Facility of the CPTP, Toulouse-Purpan, headed by Sophie Allart, the histology platform of the US 06, Toulouse-Purpan, headed by Florence Capilla, and the ANINFIMIP EquipEx facility headed by Professor E. Oswald, and supported by the French government through the Investments for the future programme [ANR-11-EQPX-0003]. C.C. was a post-doctoral fellow of the Fonds voor Wetenschappelijk Onderzoek [FWO, Belgium]. Tissue collection was sponsored by the University Hospital of Toulouse for regulatory and ethic submission, and by a grant from the delegation régionale à la recherche clinique des hôpitaux de Toulouse, through the MICILIP project. This project was also supported by the Région Midi-Pyrénées [now Occitanie], European funds [Fonds Européens de Développement Régional, FEDER], Toulouse Métropole, and by the French Ministry of Research with the Investissement d’Avenir Infrastructures Nationales en Biologie et Santé program [ProFI, Proteomics French Infrastructure project, ANR-10-INBS-08] to OB-S.

Author Contributions

NST, ADS, CRF, MQN, CB, MM, AE, CR and OBS acquired and analysed the data; LA, GP, SK, DB and EM provided human tissues; GD and CC analysed the data; FB, NV, CB and CD: study concept and design, analysis and interpretation of the data; NV, CD, FN and EM obtained funding; FB and NV: study supervision; NST, ADS and FB: drafting of the manuscript; all authors: editing of the manuscript.

References

Author notes