Fecal luminal factors from patients with irritable bowel syndrome induce distinct gene expression of colonoids

Abstract Background Alteration of the host‐microbiota cross talk at the intestinal barrier may participate in the pathophysiology of irritable bowel syndrome (IBS). Therefore, we aimed to determine effects of fecal luminal factors from IBS patients on the colonic epithelium using colonoids. Methods Colon‐derived organoid monolayers, colonoids, generated from a healthy subject, underwent stimulation with fecal supernatants from healthy subjects and IBS patients with predominant diarrhea, phosphate‐buffered saline (PBS), or lipopolysaccharide (LPS). Cytokines in cell cultures and fecal LPS were measured by ELISA and mRNA gene expression of monolayers was analyzed using Qiagen RT2 Profiler PCR Arrays. The fecal microbiota profile was determined by the GA‐map™ dysbiosis test and the fecal metabolite profile was analyzed by untargeted liquid chromatography/mass spectrometry. Key results Colonoid monolayers stimulated with fecal supernatants from healthy subjects (n = 7), PBS (n = 4) or LPS (n = 3) presented distinct gene expression profiles, with some overlap (R2Y = 0.70, Q2 = 0.43). Addition of fecal supernatants from healthy subjects and IBS patients (n = 9) gave rise to different gene expression profiles of the colonoid monolayers (R2Y = 0.79, Q2 = 0.64). Genes (n = 22) related to immune response (CD1D, TLR5) and barrier integrity (CLDN15, DSC2) contributed to the separation. Levels of proinflammatory cytokines in colonoid monolayer cultures were comparable when stimulated with fecal supernatants from either donor types. Fecal microbiota and metabolite profiles, but not LPS content, differed between the study groups. Conclusions Fecal luminal factors from IBS patients induce a distinct colonic epithelial gene expression, potentially reflecting the disease pathophysiology. The culture of colonoids from healthy subjects with fecal supernatants from IBS patients may facilitate the exploration of IBS related intestinal micro‐environmental and barrier interactions.


| INTRODUC TI ON
The intestinal epithelial barrier is composed of a single layer of cells, which allows the absorption of microbial and dietary metabolites and limits the access of harmful antigens and commensal bacteria to the underlying tissues. 1,2 Therefore, the loss of homeostatic hostmicrobiota cross talk may compromise the integrity of the intestinal epithelial barrier and lead to gastrointestinal diseases such as irritable bowel syndrome (IBS). 1,3 Indeed, there are reports of an impaired intestinal epithelial barrier, 4,5 altered mucosal expression of antibacterial genes 6 and imbalance of microbiota composition 7,8 in at least subgroups of patients with IBS. Furthermore, it has been proposed that IBS patients with diarrhea have an altered fecal metabolite profile compared to healthy subjects. [9][10][11] Hitherto, functional and structural properties of the intestinal epithelial barrier of patients with IBS have been explored using cell lines or tissue samples. Stimulation with plasma, soluble mediators from colonic biopsies as well as luminal proteases from patients with IBS negatively influenced the integrity of the colon cancer derived Caco-2 cell line. [12][13][14] Further, intestinal barrier permeability in IBS has been functionally addressed ex vivo by Ussing chamber experiments and characterized by the expression of adhesion proteins in short-term cultures of intestinal biopsies. 13,15 However, the often used Caco-2 cell line, an immortalized line originally derived from a single donor, provide a limited and simplistic physiological representation of the gut, 16 whereas access to primary intestinal epithelial cells and biopsies may be limited, and when available only suitable for short-term experiments. 17 During recent years, intestinal organoids have evolved as an attractive alternative for studying physiology, cell-cell or microbe-cell interactions. Compared to traditional in vitro methods, intestinal organoids provide a cell culture system more accurately mimicking the intestinal epithelium 18,19 with long-term viability. 20 Organoids generated from tissue-resident stem cells under specific culture conditions resemble the organ from which they have been derived and after which they are named. 21 Organoids from colonic tissue, termed as colonoids, grow as spheres with the apical surface facing inwards 19 and can be seeded as monolayers for stimulation assays. 18,19 Traditionally, organoids are developed from the study subject (patients/healthy) of interest. 22 However, the impact of the local intestinal microenvironment for intestinal epithelial barrier configuration and function is lost in this set-up. Hence, model systems to specifically study interactions between the local intestinal microenvironment and the epithelium, providing possibilities to better understand the complex pathophysiology of IBS, is much warranted.
We hypothesized that supernatants from fecal samples, containing microbial ligands, metabolites, and other luminal factors, provide stimuli similar to that present in the intestinal lumen, which will reproduce the intestinal microenvironment of the donor who provided the fecal material. Therefore, with the aim to identify IBS-specific regulation of colonic epithelium, we determined the effects of fecal luminal factors from IBS patients in colonoids established from a healthy donor.

| Study subjects and sample collection
Patients with IBS with predominant diarrhea and high severity of symptoms, based on the Bristol Stool Form scale 23

and IBS Severity
Scoring System (IBS-SSS), 24 respectively, were selected among the baseline data of participants from a previously reported intervention study. 25  • Here, we present a unique model of colonoid monolayers to evaluate IBS-specific regulation of colonic epithelium, which may provide improved understanding of the pathophysiology of IBS.
criteria. 26 The healthy subjects were randomly selected from a previous study 8 and had no previous or current history of gastrointestinal diseases. 8 All study participants fulfilled the inclusion criteria described in Data S1. Prior to visiting the clinic, study subjects collected fecal samples at home and kept them in the freezer until transportation. Samples were then stored on site at −80°C until preparation of fecal supernatants. Further, a healthy subject provided sigmoid colonic biopsies (25-35 cm proximal from the anus), without prior bowel preparation, using standard biopsy forceps (Olympus, 3.3 mm). All subjects gave written informed consent prior to participation in the corresponding studies. 8,25 The protocols were approved by the Regional Ethical Review Board at the University

| Preparation of fecal supernatants
Feces were weighed and dissolved in two weight volumes of ice-cold phosphate-buffered saline (PBS), prior centrifugation for 10 min at 1,600 g. The liquid phase was then ultra-centrifuged at 35,000 g for 2 h at 4°C. The collected fecal supernatant was stored at −80°C until use.

| Establishment of human colonoid cultures
Colonic crypts were isolated from sigmoidal colonic biopsies from a healthy subject following protocols established elsewhere 27 with minor modifications. Briefly, first crypts and later colonoids, were maintained embedded in 40 µl solid Matrigel ® Matrix (Corning ® ) containing 1 µM Jagged-1 peptide (AnaSpec), in 24-well plates (Nuclon ™ Delta Surface, Thermo Fisher Scientific) and were cultured in expansion medium as described previously. 27 The expansion medium was complete medium (Advanced Dulbecco's modified Eagle medium (DMEM)/Ham's F-12, 100 U penicillin/streptomycin, 10 mM HEPES and 0.2 mM GlutaMAX (purchased from Gibco ® , Life Technologies ™ ) supplemented with 1× B27 supplement, 1× N2 supplement, 50 ng/ ml human epidermal growth factor (purchased from Gibco ® ), 1 μg/ ml [Leu-15] Gastrin (AnaSpec), 500 nM A 83-01, 10 μM SB202190, 10 nM prostaglandin E2 and 1mM N-acetylcysteine (from Sigma-Aldrich), 100 μg/ml Primocin (InvivoGen), 10 μM CHIR99021 (Sigma-Aldrich) and 10 μM Y-27632 dihydrochloride (Tocris). 27 In addition, the media contained 50% Wnt3A conditioned medium, 20% R-spondin conditioned medium and 10% Noggin conditioned medium. The cell lines used to supply Wnt3A and Noggin were kind gifts from Professor Hans Clevers (Hubrecht Institute); the Cultrex ® R-spondin1 cell line was purchased from Trevigen, Inc. The expansion medium, without CHIR99021 and Y-27632, 27 was replaced every 2-3 days. Colonoids were maintained at 37°C, 5% CO 2 . When denser colonoid cultures had been established, colonoids were passaged every 7 days. All passages were performed by using Corning Cell Recovery Solution (Corning ® ) and dissolving the Matrigel on ice similar to reference, 27 without using an orbital shaker. The dissolved matrigel was collected into basal media (DMEM/Ham's F-12 with 1× Glutamax, 10 mM HEPES and 10% heat-inactivated fetal bovine serum (FBS)). The colonoid structures were disrupted by pipetting and approximately 50 colonoids were seeded in each well. After each passage, cells were cultured in expansion media as described above. 27 Three-dimensional (3D) colonoids generated from biopsies from a healthy subject were seeded as monolayers on transwell membranes and grown for 3 days until confluence. After 3 days of differentiation, colonoid monolayers were fixed for immunofluorescence staining or stimulated with fecal supernatants for 24 h. Stimulation with phosphatebuffered saline (PBS) and lipopolysaccharide (LPS) were used as negative and positive controls, respectively. After stimulation, RNA was isolated for gene expression analysis and culture supernatants collected for detection of cytokines. 3D, 3-dimensional; 2D, 2-dimensional; RNA, ribonucleic acids replaced with fresh differentiation medium after 2 days and colonoid monolayers were then kept in culture for one additional day to reach 3 days of differentiation.

| Stimulation of colonoid monolayers with fecal supernatants
The experimental design is schematically shown in Figure 1.

| Immunofluorescence and imaging
Triple immunofluorescence staining was performed against muc-2, phospho-ezrin or wheat germ agglutinin, in combination with phalloidin and Hoechst, in filter inserts in a 24-well plate as previously Changes in brightness/contrast and reduction of background noise were applied to emphasize the qualitative analysis of the images.
Images of the human healthy colonoids during culture were taken with Leica DM IL LED Inverted Microscope with ICC50 HD Camera (Type: 11 090 137 001) using 4× objective.

| Gene expression analysis
The material from 2 to 3 monolayers, which underwent the same stimulus, was pooled prior to RNA extraction. Briefly, total RNA was was used for normalization. A complete list of the genes targeted in the custom array is shown in Data S1.

| Measurement of cytokines and LPS
Levels of IL-1β, TNFα and IL-8 in culture supernatants were measured using V-PLEX custom human biomarker plate from MSD ® Multi-Spot Assay System (Meso Scale Diagnostic) following the manufacturer's instructions. Culture supernatants obtained from both transwell compartments were evaluated. LPS levels were measured in fecal supernatants using the LPS ELISA kit (Aviva System Biology), following the manufacturer's instructions.

| Untargeted liquid chromatography-mass spectrometry analysis
The luminal factors of fecal supernatants were analyzed at Chalmers Mass Spectrometry Infrastructure using a non-targeted liquid chromatography-mass spectrometry (LC-MS) approach. Briefly, the analyses were carried out using a reversed-phase chromatography (RP) and hydrophilic interaction chromatography (HILIC) with positive and negative electrospray ionization polarities, as described elsewhere. 29 The samples of each study group were analyzed in separate batches, which included its own quality control samples.
The analytical workflow named "notame", described in, 30 was used to pre-process the acquired data and included drift correction within and between batches, data imputation using missForest R package and clustering of features to remove weak and repeated features. 30 Log 10 transformation was applied prior between-batch-correction to reduce possible batch effects caused by the instrument.

| Fecal microbiota DNA analysis
The microbiota profile was determined at Genetic Analysis AS using the GA-map ™ dysbiosis test. In short, total bacterial genomic DNA is extracted with magnetic beads from homogenized and mechanically disrupted fecal samples. PCR is used to amplify the 16S rRNA gene (V3-V9 regions) followed by probe labeling. The DNA probes, labeled with nucleotides, hybridize to complementary probes coupled to magnetic beads, that allow signal detection by BioCode 1000A 128-Plex Analyzer (Applied BioCode). 31 Fifty-four DNA probes allow targeting ≥300 bacteria belonging to different taxonomic levels. This test generates as a result a bacterial profile based on fecal bacterial abundance (or probe signal intensity). 31

| Data analysis
Multivariate data analyses were used to evaluate the relationship between the different study groups (Y-variables) and gene expres- indicating discrimination and predictability, respectively. R 2 Y values ≥0.5 correspond to good fit (max. R 2 Y = 1). Q 2 values ≥0.4, but no more than 0.3 away than R 2 Y, are considered adequate for biological variables. Univariate analyses were applied to identify differences between specific variables. The statistical tests performed (parametric or non-parametric) were based on the distribution of the data determined by Shapiro-Wilk tests and histograms.
p < 0.05 were considered statistically significant. More details can be found in Data S1.

| Demographic information
In total, 7 healthy subjects (5 females) and 9 IBS patients (6 females) with predominant diarrhea provided fecal samples from which supernatants were prepared. The IBS patients were older than the

| Establishment and characterization of colonoid monolayers
The cultures of colonoids comprised spheres with relatively thin walls and a dark core ("lumen") containing debris. Colonoids in varying sizes were located in different focal planes throughout the extracellular matrix that supported their growth (Figure 2A). The colonoid structures were established as a monolayer comprising polarized, organized, and differentiated intestinal epithelial cells expressing phospho-ezrin and displaying a glycocalyx ( Figure 2B,C). Filamentous actin staining identified epithelial cells tightly connected by junctional complexes ( Figure 2B-D). Moreover, the differentiated colonoid monolayers contained dispersed mature mucin-producing goblet cells ( Figure 2D).

| Addition of fecal supernatants from healthy subjects modifies gene expression profile of colonoid monolayers
The addition of fecal supernatants from healthy subjects, LPS or PBS to the monolayers resulted in distinct clusters, although with some overlap, based on gene expression ( Figure 3A). The three different culture conditions gave rise to distinct gene expression profiles and 27 out of the 75 genes contributed to the separation between the groups ( Figure 3B,C). The expression of CCL20, DSC2 and ICAM1 was lower in monolayers stimulated with fecal supernatants from healthy subjects compared to LPS-stimulated monolayers ( Figure 3C, Table S1). In contrast, the expression of PECAM1 was higher in monolayers stimulated with fecal supernatants from healthy subjects compared to LPS-stimulated monolayers. Further, the expression of PVRL1 was higher in monolayers stimulated with fecal supernatants from healthy subjects compared to PBS-stimulated monolayers ( Figure 3C, Table S1).

| Fecal supernatants from IBS patients induce a distinct gene expression of colonoid monolayers
Next, we compared the effects of fecal supernatants from healthy subjects and patients with IBS on gene expression of colonoid monolayers. Addition of fecal supernatants from the two study groups to colonoid monolayers induced distinct gene expression clusters with only minor overlap ( Figure 4A). In total, 22 out of the 75 genes were differently regulated by the addition of fecal supernatants from the two study groups ( Figure 4B). Genes related to immune response such as CD1D, IRF7, TNFSF13, IRF5, TLR9, LYZ, ICAM1 and CX3CL1 showed higher expression in colonoid monolayers stimulated with fecal supernatants from IBS patients, while TLR5 expression was higher in colonoid monolayers stimulated with fecal supernatants from healthy subjects ( Figure 4C). Also, the addition of fecal supernatants from IBS patients upregulated the gene expression of the pro-inflammatory cytokines IL-1β and TNFα and en-  Figure 5).

F I G U R E 3
Gene expression of colonoid monolayers stimulated with fecal supernatants from healthy subjects or control stimuli. Colonoid monolayers were stimulated for 24 h with fecal supernatants from healthy subjects (n = 7, light gray dots), LPS (n = 3, dark gray diamonds) or PBS (n = 4, white 4-point stars). Gene expression was analyzed by PCR arrays for genes involved in antibacterial and inflammatory response and barrier function. (A) A principal component analysis (PCA) plot based on 75 expressed genes (11 unique genes excluded as they were below detection limit). (B) Score scatter plot and (C) loading scatter plot from an orthogonal partial least squares-discriminant analysis (OPLS-DA) using a variable influence on projection (VIP) cut-off >1.1. HEALTHY, colonoid monolayers stimulated with fecal supernatants from healthy subjects; LPS, lipopolysaccharide-stimulated colonoid monolayers; PBS, PBS-stimulated colonoid monolayers F I G U R E 4 Gene expression of colonoid monolayers stimulated with fecal supernatants from healthy subjects or IBS patients. Colonoid monolayers were stimulated for 24 h with fecal supernatants from healthy subjects (n = 7 gray dots) or IBS patients (n = 9 black dots). Gene expression was analyzed by PCR arrays for genes involved in antibacterial and inflammatory response and barrier function. (A) A principal component analysis (PCA) based on 75 genes (11 unique genes excluded as they were below detection limit). (B) Score scatter plot and (C) loading column plot from an orthogonal partial least squares-discriminant analysis (OPLS-DA) using a VIP cut-off >1.15. HEALTHY, colonoid monolayers stimulated with fecal supernatants from healthy subjects; IBS, colonoid monolayers stimulated with fecal supernatants from IBS patients. Between group-comparisons were tested with Mann-Whitney U test; asterisks identify statistically significant p values: *p < 0.05; **p < 0.01

| The fecal microbiota and metabolite profiles distinguish healthy subjects and IBS patients
The microbiota profile of the fecal samples used for preparing fecal supernatants differed between healthy subjects and IBS patients ( Figure 6A,B). The bacterial taxa contributing most to the separation were Proteobacteria, Pseudomonas spp., Dorea spp. and Ruminococcus gnavus (p < 0.05), and were more abundant in IBS patients, whereas Bacteroides pectinophilus was more abundant in healthy subjects (p < 0.05). Additionally, the fecal supernatants from healthy subjects and IBS patients, respectively, presented distinct profiles based on the 9699 spectral features detected in an untargeted LC-MS analysis ( Figure 6C). Two hundred of the detected compounds contributed to the separation between the two groups ( Figure 6D). Among the top five metabolites in either direction contributing most to the separation, only xanthine has been annotated and was more abundant in fecal supernatants from IBS patients compared with healthy subjects. However, neither metabolite nor microbiota profiles were influenced by the age or sex of the donors ( Figure S2).

| DISCUSS ION
In this work we describe that components of fecal supernatants, a overlooked. 32 In this study, we therefore exposed colonoids estab-  43 Similarly, another study provided evidence of impaired barrier function when adding fecal supernatants from IBS patients to intestinal biopsies mounted in Ussing chambers. 13 In the above men- conditions. Whether addition of fecal supernatants from IBS patients and healthy subjects, respectively, gives rise to functional differences related to epithelial integrity of the colonoids in our model system remains to be further elucidated. Still, our results support the notion that fecal supernatants from IBS patients have a distinct effect on the intestinal epithelium, regulating genes maintaining epithelial barrier integrity.
In line with a previous study, 14  The fecal microbiota profiles differed between IBS patients and healthy subjects. The taxa Proteobacteria, Pseudomonas spp., Dorea spp. and Bacteroides pectinophilus were found to be linked to IBS patients in our study, have previously been associated with IBS-D and symptoms, including intestinal permeability. [48][49][50][51][52][53][54] Further, the mucin degrader Ruminoccocus gnavus has been associated to IBS severity. 55,56 An altered microbiota composition may result in altered metabolite composition, reflecting the overall function and metabolism of the bacterial community. Indeed, similar to previous reports by us and others, in this study IBS patients presented with distinct fecal metabolite profiles, which were, however, not influenced by age or sex. [9][10][11] Analyzing the fecal supernatants with LC-MS allowed us to determine the presence of almost 10,000 spectral features, of which 200 were found significant for separation of IBS from the healthy controls. The possibility to identify compounds from untargeted LC-MS analysis is, however, resource intensive and the ability to, with reasonable certainty, annotate the top metabolites driving the separation between study groups was limited with the available resources in this project. The only unique annotated top metabolite, the purine-derived metabolite xanthine, previously demonstrated to promote intestinal barrier development, 57 was found in higher levels in IBS patients, similar to a previous study by our group. 11 Besides, although the vast majority of metabolite compounds have yet to be annotated, it is still more likely that the combination of compounds rather than the presence or absence of specific compounds give rise to the IBS-specific gene regulation of colonoid monolayers. IBS patients have demonstrated higher concentration of fecal proteases, 13 which may cause alterations of the epithelial permeability in Caco-2 cells 12 and humanized mice. 13 Additionally, LPS content of fecal samples has been reported to be higher in IBS patients than in healthy subjects. 58 However, the different effect on gene regulation of colonoid monolayers cultured with fecal supernatants from IBS patients and healthy subjects seen in our study was not driven by LPS, as the level of LPS did not differ between the two groups. Altogether, our data strongly implies that the gene expression profile of colonoid