Celiac disease
Zorro Manrique, Maria Magdalena
DOI:
10.33612/diss.122712049
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from
it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2020
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Zorro Manrique, M. M. (2020). Celiac disease: From genetic variation to molecular culprits. University of
Groningen. https://doi.org/10.33612/diss.122712049
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
CHAPTER 2
The role of celiac disease
associated protein LPP in
intestinal barrier function.
Maria Zorro
1, Vinod Kumar
1, Lucja M. Jarosz
2, Luz Maria Medrano
1, Joram
Mooi-weer
1, Michael Schumann
3, Jan F. Richter
4, Cisca Wijmenga
1, Sven C.D. van
IJzen-doorn
2, Iris Jonkers
1, Sebo Withoff
1.
Ch
ap
te
r 2
Abstract
Intestinal epithelial barrier function is critical for homeostasis of digestive and immunological processes. Deregulation of the epithelial barrier has been associated with several diseases, in particular celiac disease (CeD). Although the majority of genes associated with CeD are related to immune function, genome-wide association studies, eQTL mapping and pathway analyses have also implicated a number of genes that play a role in cell–cell interaction. We hypothesized that these “non-immune” CeD genes exert their effect by deregulation of intestinal barrier homeostasis and therefore investigated the potential role of LPP, one of the “non-immune” genes most strongly associated to CeD, in barrier function. Using Caco-2 cells as a model of the intestinal barrier, we evaluated the functional impact of reduced
LPP expression. Cells with LPP knockdown (KD) display a reduction in proliferation and their capacity to form lumen-like structures in 3D culture assays. Microscopic analysis revealed increased permeability at focal sites in transwell assays in KD but not in control cell lines. Finally, RNA-seq analysis revealed that exposure to Interferon-g leads to downregulation of metabolic pathways and a concomitant upregulation in immune genes in the KD cell line, suggesting that LPP might also play a role in the immune response elicited by epithelial cells. The downregulation of LPP observed in intestinal biopsies of CeD patients supports our hypothesis that LPP downregulation contributes to the deregulation of epithelial cell function in CeD both at barrier- and immune-response-level.
Introduction
Celiac disease (CeD) is a chronic autoimmune disorder triggered by ingestion of gluten in genetically susceptible individuals. In CeD, the oral intake of gluten causes an abnormal activation of both the innate and adaptive immune response that results in cell damage and
villus atrophy in the small intestine1. Although the HLA-DQ2 and -DQ8 alleles are the most
important genetic risk factors, genome-wide association studies (GWAS) have uncovered an
additional 42 loci outside the HLA-locus that also contribute to CeD susceptibility2. Many of
the associated loci encompass genes that regulate T and B cell biology, as well as genes involved in the IFNg signaling pathway. This is not surprising given that deregulation of the
immune system and IFNg signaling are involved in disease onset3,4. More recently, however,
eQTL mapping and pathway and co-expression analyses of GWAS loci have also prioritized genes involved in cell–cell interaction and intestinal barrier function as potential causal genes contributing to CeD susceptibility. We hypothesized that these non-immune genes, exemplified by LPP and C1orf106, contribute to CeD pathophysiology by disrupting cell–cell interaction,
leading to increased permeability, also called ‘leaky gut’, one of the hallmarks of CeD5,6.
Barrier function may therefore already be impaired in CeD individuals before development of
CeD5, which is in line with a role for genetic factors in barrier function.
It has been shown by others that CeD biopsies exhibit alterations in the expression of tight
junction proteins7, actin cytoskeleton disturbances8 and increased intestinal epithelial cell
apoptosis9. It is currently unclear if these alterations cause CeD or are a consequence of
disease-associated inflammation. Interestingly, one of the strongest non-HLA CeD-associated
genetic variants (odds ratio ~1.26)10 mapped to a 2.8 kb region in intron 2 of the LPP gene
(chr3: 188117070–188119901, NCBI build 37)11. Although the function of LPP remains poorly
characterized, transcriptional analysis of intestinal biopsies showed that its expression is
downregulated in CeD patients compared to healthy individuals4. A recent study demonstrated
that LPP co-localizes with E-cadherin in adherent junctions and focal adhesions. Other researchers have described alterations in intracellular distribution of E-cadherin in cell
junctions upon LPP knockdown12, and LPP knockdown led to disturbances in the intracellular
distribution pattern of LPP and focal adhesion proteins after exposure to gliadin peptides, the
degradation products of dietary gluten13. These data are in line with our in silico predictions
indicating a putative role for LPP in cell–cell interaction and support the hypothesis that downregulation of LPP expression contributes to enhanced intestinal permeability in CeD patients.
Here, we aim to show that reduced expression of LPP leads to decreased barrier homeostasis in Caco-2 cells, a well characterized in vitro model for studying intestinal barrier function. We generated Caco-2 cells that stably expressed a short hairpin RNA (shRNA) sequence targeting LPP mRNA. We then assessed the effects of LPP knockdown on the proliferative
Ch
ap
te
r 2
capacity of the cells, on the permeability of Caco-2 monolayers in 2D transwell experiments, and on lumen formation in 3D spheroid cultures. Furthermore, we evaluated the transcriptional response after a challenge with IFNg. What we found was that reduced LPP expression led to decreased cell proliferation and polarization. Moreover, LPP knockdown increased the local permeability and the response to inflammatory signals. Our results provide new insights into a potential role for LPP in both epithelial integrity and in immune function.
Materials and Methods
Cell culture. A human Caco-2 cell line (colonic epithelial cancer cells; ATCC), a gift from
Dr. Sven van IJzendoorn from the Cell Biology department of the University Medical Center
Groningen, was cultured at 37oC in maintenance media containing MEM medium (GIBCO),
10% fetal bovine serum (Sigma-Aldrich), 1% non-essential amino acids (Invitrogen) and 100 U/ml penicillin-streptomycin (GIBCO). For passaging, cells were treated for 5 minutes with 1X trypsin (Invitrogen) and passaged twice per week.
Caco-2 LPP knockdown, scrambled and LPP rescue cell lines. LPP Knockdown cell
lines (KD) were developed by transfecting pRS vectors containing four different shRNA that targeted LPP (see details in Supplementary Fig. 1) using the Amaxa cell nucleofector Kit T for Caco-2 (Lonza), according to the manufacturer’s instructions. This generated four different sublines. Untransfected Caco-2 cells (wild type (WT)) or cells transfected with a non-targeting scrambled shRNA (SCR) were generated as controls. To generate stable cell lines, transfected cells were selected by culturing for 2 months in increased concentrations of puromycin (GIBCO), up to a final concentration of 10 mg/ml. Knockdown efficiency was
determined by RT-qPCR and western blot (every ~3-4 weeks). An LPP-rescue cell line was
created by expressing an LPP cDNA sequence engineered to be resistant to the LPP shRNA. In brief, shRNA-resistant cDNA was created by introducing nine silent point mutations in the
LPP shRNA targeting region of LPP cDNA using the QuikChange II site-directed mutagenesis
kit (Stratagene). The recombinant LPP cDNA was cloned into the pLenti-C-Myc-DDK-IRES-Neo expression vector (ORIGENE) and transfected into the LPP KD cell line with the strongest reduction in LPP (KD cell line containing the shRNA #2) using the Amaxa cell nucleofector Kit T (Lonza), according to the instructions of the supplier. Selection was started 48 hours (hrs) post transfection by adding increasing concentrations of G418 sulfate (Geneticin, GIBCO) to the culture medium, up to a final concentration of 400 mg/ml. The level of LPP was tested by real time (RT)-qPCR and western blot. The rescue cell line was passaged twice a week in maintenance medium with puromycin (10 mg/ml) and Geneticin (400 mg/ml).
RT-qPCR. Total RNA was extracted using the mirVana™ miRNA isolation kit (Ambion)
according to the manufacturer’s protocol. cDNA was prepared by RevertAid H Minus First Strand cDNA Synthesis Kit (Thermo Scientific). We used primers for LPP, HPRT, MX1 and IL15R as follows: LPP 5’ CCAACAATGTCTCACCCATC, LPP 3’ ACTACCGGGGCAAACTTTTT,
MX1 5’ ATCCAGCCACCATTCCAAGG, MX1 3’ TTTGCGATGTCCACTTCGGA, IL15R
5’ AGAGAGCCTCTCCCCTTCTG, IL15R 3’ TTGAAGGCATCAGCTGGGAG. Reverse transcription was performed using the Fast-Real Time PCR system 7900 ht (Applied Biosystems) using the SYBR green mix (Bio-Rad). Data were analyzed with the SDS software package V2.3 (Applied Biosystems). Expression values were normalized to the endogenous
control (HPRT) to calculate the relative LPP, MX1 or IL15R expression level using the 2-ΔΔCT
method. Relative LPP values were then converted to percentages by setting the SCR as a reference (100%).
Western blot. Caco-2 cells were homogenized in lysis buffer (PBS containing 2% SDS and
1X complete protease inhibitor cocktail (Roche)). Protein concentration was determined with the bicinchoninic acid BCA Protein Assay Kit (Thermo Scientific). 20 mg of protein was loaded on SDS-polyacrylamide electrophoresis gel and transferred to nitrocellulose membrane (Bio-Rad). Blots were probed overnight with mouse anti-LPP antibody (1:1,000, clone 8B3A11, ABCAM) or for 1 hr with mouse anti-actin antibody (1:5,000, clone C4M, MP Biomedicals), followed by incubation with goat anti-mouse horseradish-peroxidase–conjugated secondary antibodies (1:10,000, Jackson Immuno Research). Bands were visualized using Lumi-Light Western Blotting Substrate (Roche). Images were captured by the Chemidoc MP imaging system (Bio-Rad). The signal intensity of each band of interest was quantified using Image Lab™ software (Bio-Rad). Specific intensities were normalized using the actin signal and expressed as a percentage by setting the SCR cell line as reference (100%).
3D cultures. Cells were suspended in DMEM supplemented with 2% v/v matrigel (BD
Biosciences) and seeded at a density of 6000 cells/well onto matrigel pre-coated 8-well Lab-Tek Chambered coverglasses (Nunc). After 1 week, bright field microscopy pictures were taken with the Evos FL cell imaging system (Thermo Fisher Scientific). The number and size of spheroids and the presence of a lumen in the center of each spheroid was quantified manually using the ImageJ program (NIH). At least 100 spheroids were counted per condition. To further analyze spheroid morphology, their nuclei and actin were stained (see details in the section below).
Immunofluorescence microscopy. Cells were fixed in 4% paraformaldehyde at room
temperature for 20 or 30 minutes (for 2D or 3D cultures, respectively), and permeabilized
with 0.2% Triton-X100 (Sigma-Aldrich) for 10 or 20 minutes at 37oC (for 2D or 3D cultures,
respectively). Nonspecific background was blocked with 3% FBS at 37oC. Cells were incubated
with primary mouse anti-LPP (1:250, clone 8B3A11, Abcam) or E-cadherin (1:100, clone
5H6L18, Invitrogen) at 37oC for 1.5 hrs. The cells were washed with PBS and incubated with
secondary Alexa-Fluor488 or cy3-conjugated goat anti-mouse or anti-rabbit antibodies (1:250, Jackson ImmunoResearch Laboratories). Actin was labeled with TRITC-phalloidin (10mg/ml,
Sigma-Aldrich) and the nuclei with DAPI (1mg/ml, SIGMA) at 37oC for 30 min. Images were
acquired with a Leica TCS SP8 confocal microscope (Leica) and analyzed using Leica LAS AF software (Leica).
Ch
ap
te
r 2
Cell proliferation. Caco-2 cells were plated at 250,000 cells/ml in 25 cm2 flasks (Greiner) and
grown for 72 hrs. Next, cells were trypsinized and quantified microscopically using trypan blue (GIBCO) staining.
Transepithelial resistance. Caco-2 cells were seeded in 0.4 mm polycarbonate transwells
in 12-well plates (Corning) at a density of 50,000 cells/well and maintained for 21 days. The medium was replaced 2-3 times per week. Transepithelial resistance (TEER) values were measured twice per week with an EVOM2 Ohm meter (World Precision Instruments), according to the manufacturer’s instructions.
Paracellular permeability assay. The permeability across Caco-2 monolayers was
assessed by measuring the flux of fluorescein isothiocyanate–labeled dextran (FD4, Sigma-Aldrich) from the apical to the basal compartment of the transwell. After 21 days, the medium in the basal compartment was replaced by Hank’s balanced salt solution (HBSS) to mimic physiological or “normal conditions”, or by 2.5 mM EDTA in calcium- and magnesium-free PBS (“stress conditions”). 500ml of FD4 solution (250 mg/ml) were added in each apical compartment. Aliquots were taken from the basal compartment at 30 min intervals over a
2.5-hr period at 37oC in sterile conditions. The amount of FD4 in the basal compartment
was determined in a BioTEK Synergy HT plate reader (Biotek) using a calibration curve. The
apparent permeability coefficient (Papp) was calculated by the following formula: Papp = dQ/
dt×1/A×Co, where Papp = apparent permeability in cm/s, dQ/dt = the rate of appearance of FD4
in the basal compartment in μg/s, Co = the initial FD4 concentration in the apical compartment
in μg/ml and A = the surface area of the insert in cm2.
Statistical Analysis. GraphPad Prism V7.0 (GraphPad software, San Diego, CA) was used
to assess statistical differences across groups. Detailed information (type of statistical tests applied, p values and number of replicates) is provided in the text of the figure legends.
Stimulation experiments. Caco-2 cells were cultured in 0.4 mm polycarbonate Transwells
in 12-well plates (Corning) at a density of 75,000 cells/well. The medium was replaced 2-3 times per week. After 2 weeks, the cells were treated in the basal compartment with 60 ng/ml of IFNg (PeproTech) for 3, 8 or 40 hrs. Untreated cells were included as controls. Following incubation, cells were gently washed, collected in cell lysis buffer from the mirVana™ miRNA
isolation kit (AMBION) and stored at -80oC for further use. To pre-assess the responsiveness
of Caco-2 cells to IFNg stimulation, relative levels of MX1 and IL15R mRNA were measured by RT-qPCR.
RNA sequencing. RNA from unstimulated and IFNg-stimulated cells was isolated with the
mirVana™ miRNA isolation kit (Ambion), following the manufacturer’s protocol. Sample concentration and integrity was measured using the Nanodrop 1000 Spectrophotometer (Thermo Scientific) and the High-sensitivity RNA analysis kit (Bio-Rad EXPERION). The
libraries were prepared with the Truseq RNA preparation kit (Illumina), using the manufacturer’s protocol. The pooled libraries were sequenced on a HiSeq2500 instrument (Illumina), using default parameters (single read, 1x50bp). The trimmed fastQ files where aligned to build 37
of the human reference genome using Hisat (version 0.5.1)14, allowing for two mismatches
and sorted using SAMtools V1.215. Gene quantification was performed by HTSeq/0.6.1p116
using mode=union, stranded=no. The Ensembl V75 database was used for gene annotation. Differentially expressed genes (DEGs; those with |log2 FC>1| and adjusted p value ≤ 0.01)
were extracted by the DESeq2 R package17. Principal component analysis (PCA) plots and
heatmaps were generated by R base functions and the pheatmap and ggplot2 packages.
Results
LPP knockdown impairs proliferation and lumen formation in Caco-2 cells
Given the potential role of LPP in cell–cell adhesion, we wanted to know if LPP downregulation interferes with cell proliferation and cell polarity. Therefore, we knocked down LPP in Caco-2 cells using shRNA molecules. We tested four different shRNAs against different exons of LPP (Supplementary Fig. 1) and observed that the cells transfected with shRNA #2 presented the most efficient and consistent knockdown (Supplementary Fig. 2). We therefore selected this cell line for further characterization and coined it KD. Our Caco-2 KD cell line displayed a significant decrease of LPP at mRNA-level (~75%) and protein-level (~90%) compared to the controls (WT or SCR, Fig. 1A-C). Furthermore, a significant reduction (~50%) of proliferative capacity was observed in KD compared to controls (Fig. 1D). Of note, the treatment with the non-targeting scrambled shRNA did not affect LPP expression or cell proliferation, indicating that LPP remains at physiological levels and that the transfection conditions were not toxic to the cells.
To determine the potential effect of LPP knockdown on cell polarization, we grew the cells for 1 week in the presence Matrigel, a commercial hydrogel containing cellular matrix proteins,
to induce spontaneous formation of 3D spheroids with a central lumen18. We noted that all
three cell lines (WT, SCR and KD) developed spheroids similar in numbers (~10 spheroids/ microscopic field, Fig. 1E) and size (~100 mM, Fig. 1F). However, the percentage of spheroids with a hollow central lumen was significantly lower in KD compared to controls (~15% vs ~50%; Fig. 1G). To exclude the possibility that reduced proliferation of the KD cells simply caused a delay in the lumen formation, rather than an impairment of the process, we followed up several of the 3D cultures for 2 weeks but did not observe any changes (data not shown). This suggests that the impairment of lumen formation is independent of proliferative capacity. Many types of epithelial cells are polarized, which means that they present three domains: an apical domain facing the lumen, a basal domain underlying the epithelial cells from the
connective tissue and a lateral domain between the apical and basal domains19. Microscopic
analysis of the spheroids indicated that most control spheroids consisted of a single layer of cells lining a central lumen. The WT and SCR spheroids were characterized by nuclei
Ch
ap
te
r 2
positioned towards the basal side of the spheroid and a thick layer of actin at the luminal side (Fig. 1H, I). In contrast, the position of the nuclei and the deposition of actin fibers appeared to be random in most of the KD spheroids (Fig. 1J), indicating defects in apico-basal cell polarity in the KD cells leading to failure of spheroid formation.
LPP knockdown does not affect the barrier function as measured by TEER and FD4 flux
We evaluated TEER, a measurement of the electrical resistance across a cell monolayer
that reflects tight junction integrity20, and observed a time-dependent increase in TEER
in the control cell lines, with values after 3 weeks significantly higher in the SCR cell line compared to WT. In contrast, the KD cells displayed minimal TEER increase during the first 2 weeks, followed by a strong increase after 3 weeks (Fig. 2A). However, this increase was not statistically different between SCR and KD. This dramatic change in TEER in the KD cell line could be caused by a delay in monolayer formation during the first 2 weeks. Surprisingly, we performed nucleic staining after 1, 2 and 3 weeks and observed confluent monolayer formation in all cases at all time points (data not shown), indicating that the lower TEER in Figure 1. LPP knockdown impairs proliferation and lumen formation in Caco-2 cells. Caco-2 controls (WT, SCR)
or LPP knockdown (KD) cell lines were grown under standard conditions (A-D) or in the presence of Matrigel (E-J). After
3 days, LPP expression was analyzed by western blot (A-B) or qPCR (C). Cell proliferation was assessed by trypan blue exclusion (D). Spheroid characteristics: number (E), size (F), percentage of hollow lumens (G). Confocal images of representative 3D cultures (H-J) showing nuclei (DAPI) and actin distribution (phalloidin). Scale bar represents 100 mM. Data are expressed as a percentage by taking the SCR control cell line as reference (100%). Bar plots show mean ± SEM. Graphs are representative of at least three independent experiments. p values were obtained using a one-tailed Student t test, **** p<0.0001.
the KD cell cultures is not caused by a decrease in proliferation or incomplete confluency. It is important to note that the large variability in the replicate data, particularly in the KD cells, that is illustrated by the high coefficient of variation (CV) (Fig. 2B) makes it difficult to conclusively state that TEER is affected by LPP knockdown. In contrast, the variability
decreased gradually to acceptable levels (CV<20%)21 in the controls. Together this implies
that additional experiments are needed to assess tight junction integrity more accurately. Next, we evaluated the paracellular permeability in the cell cultures by measuring the flux of FD4 from the apical compartment to the basal compartment of the transwell system. We could not detect FD4 in the basal compartment after 2.5 hrs under normal conditions (HBSS) in any of the Caco-2 cell lines (data not shown), implying that there were no defects in paracellular permeability under “physiological conditions”. Conversely, the flux of FD4 towards
the basal compartment increased from 0 to ~1x105 cm/seg after treatment with EDTA (Fig.
2C), which is known to disrupt tight junctions22. We noted that the permeability is significantly
higher in the KD and WT compared to the SCR cell line. However, due to the differences in permeability among the controls, it is difficult to conclude that there is a substantial increase in the permeability in the KD cells under the conditions that we tested. Finally, even though the permeability assay we performed has been extensively used in industry and research to
provide an estimation of global permeability22, this strategy does not consider the presence
of focal places where the tight junctions and the permeability are defective. We therefore
performed a pilot experiment using a novel method called the sandwich23 that assesses the
Figure 2. LPP knockdown does not affect barrier function of Caco-2 cell monolayers. Caco-2 cells were expanded for 21 days in transwell supports. TEER values were measured twice per week (A). The coefficients of variation of TEER measurements demon-strate the variability of the assay (B). Permeability (apparent
per-meability coefficient (Papp)) of Caco-2 monolayers under stress
conditions (EDTA) (C). Plots show the mean ± SD (if applicable). Data represent one of at least three individual experiments.
Ch
ap
te
r 2
permeability in focal sites by immobilizing labeled molecules close to their site of passage. Here we observed more locations with local disruptions of permeability in the KD cells compared to controls after exposure to EDTA (Supplementary Fig. 3). Although this finding implies a marked susceptibility of the KD cell line to alterations in both tight junction formation and permeability in response to environmental stress, further replications should be done to confirm this result.
Non-physiological overexpression of LPP in the KD cell line does not fully reconstitute Caco-2 cell proliferation and does not rescue lumen formation.
To confirm that the impaired proliferation and polarization we observe in Caco-2 cells is due to LPP reduction, we generated a rescue cell line by expressing LPP cDNA in the KD subline insensitive to the shRNA. As expected, we observed a significant increase in LPP expression in the rescue cell line compared to the KD subline (Fig. 3A, B). Furthermore, we noted a moderate but significant increase in the proliferation rate of the rescue cell line compared to the KD cell line (Fig. 3C), indicating that LPP rescue at least partially restores cell proliferation.
Figure 3. Non-physiological overexpression of LPP in the KD cell line does not fully reconstitute Caco-2 cell pro-liferation and does not rescue lumen formation. Caco-2 WT, SCR, KD or rescue cell lines were grown under standard culture conditions. After 3 days, LPP expression was analyzed by western blot (A) or qPCR (B). Cell proliferation was quantified by trypan blue exclusion (C). Representative confocal images of 3D (D, F, G) or 2D (E) cultures were taken to assess the percentage of hollow lumens (D), the cell morphology and the distribution of LPP (E-F), actin or E-cadherin (G). Bar plots are depicted as percentage by setting the SCR control as reference (100%). Bar plots show mean ± SEM. Data is representative of at least three independent experiments. ***p < 0.001, **** p <0.0001; one-tailed Student’s t-test.
One possible explanation for the incomplete restoration of proliferative capacity in the KD upon reintroduction of LPP could be that expression of LPP rescue cDNA is variable in the KD cell line. Indeed, immune-staining of LPP showed that LPP levels were very heterogeneous and that some cells presented much higher levels than control cells (Fig. 3E, F). Thus, the high expression of LPP in a subpopulation of cells might explain the overall observation of a two-fold increase in LPP mRNA and protein levels (Fig. 3A, B) in the rescue cell line. Lastly, we assessed the polarization capacity of the rescue cell line by staining rescue spheroids for actin and E-cadherin. This experiment showed that the rescue cells, like the KD cells, still display an altered distribution of actin and E-cadherin throughout the cells (Fig. 3G). Taken together, these results indicate that our attempt to rescue the knockdown phenotype resulted in overexpression of LPP in some cells of the population, but not in all. As a consequence, the polarization defect in the KD cell line could not be rescued, suggesting that LPP expression may be tightly regulated to ensure that it localizes properly in the cell–cell junctions and to allow adequate proliferation and polarization of the intestinal cells.
IFNg induces a distinct transcriptional signature in Caco-2 cell lines and an enhanced inflammatory response upon LPP knockdown
To explore the major transcriptional changes upon LPP knockdown occurring under physiological and inflammatory conditions, we evaluated the gene expression responses of unstimulated and IFNg-stimulated Caco-2 cells at different time points (3, 8 and 40 hrs; Fig. 4A) using RNA-seq. The optimal IFNg dose was determined in a previous titration assay experiment where we selected the minimal concentration that induced a stable decrease in TEER in the WT cell line as a proxy for alterations in tight junction integrity (Fig. 4B). IFNg stimulation led to a time-dependent increase in the expression of the IL-15 receptor gene (IL15R) and the MX1 gene (Fig. 4C, D), two genes highly expressed in intestinal biopsies
of CeD patients24,25, which indicates that our model responds to CeD-related inflammatory
signals. Of note, MX1 expression was higher (~40x) in the KD cell line compared to controls after 3, 8 and 40 hrs of IFNg stimulation. MX1 transcripts decreased after 3 hrs of stimulation in the KD cell line, but remained ~10x higher compared to WT and SCR. MX1 plays a pivotal
role in the antiviral immune response26 and in inflammatory-mediated disease etiology25,27.
The remarkable induction of this gene that we observed in the KD cell line suggests that reduced expression of LPP contributes to an exacerbated inflammatory response in intestinal epithelial cells.
PCA of the global gene transcription response indicates that all the cell lines in the panel respond to IFNg in a time-dependent fashion (Fig. 4E). Although the PCA clearly discriminates between the WT, SCR or KD cell lines, it should be noted that the distance between the samples shows major differences between the controls at baseline (unstimulated) and after IFNg stimulation, suggesting that the overall transcriptome in the SCR cell line does not resemble the WT. These discrepancies were surprising given the previously described similarity between the controls in phenotypic characteristics such as proliferation rate and polarization capacity.
Ch
ap
te
r 2
To test whether the transcriptomic differences in the controls were due to alterations in LPP expression, we plotted the expression profile of LPP over time (Fig. 4F). In general, we observed similar expression patterns in the control cells. Despite small fluctuations, the LPP level was 1.2 (log2) times lower in the KD cell line compared to controls. Together these data suggest that all the cell lines are responsive to IFNg, which is a disease-associated inflammatory stimulus or stressor.
LPP knockdown affects inflammatory and metabolic genes in response to IFNg
To identify the main biological pathways affected by LPP knockdown, we performed an unsupervised hierarchical cluster analysis of all the IFNg-dependent DEGs. A total of 2,212 DEGs were identified and clustered based on log2-fold changes relative to unstimulated controls. In agreement with the PCA, clustering analysis revealed that each cell line exhibits a characteristic response upon stimulation over time and that the WT and SCR cell lines show striking differences despite their phenotypic resemblance and consistent LPP expression in time (Fig. 5A). We identified 15 clusters, of which seven were similar between the control cell lines but different in the KD at one or more time points (clusters 1, 2, 5, 6, 7, 9 and 12; indicated with black boxes in Fig. 5A). These seven clusters represent pathways that are uniquely perturbed upon LPP knockdown. Pathway analysis on these DEGs revealed that these clusters encompass genes associated with a range of biological processes (Fig. 5B). Figure 4. IFNg induces a distinct transcriptional signature in Caco-2 cell lines and an enhanced inflammatory response upon LPP knockdown. Caco-2 cell lines were seeded in transwells. After 2 weeks, cells were left untreated (unstimulated (0 hrs)) or incubated with IFNg 50, 200, 500 ng/ml (A) or 60 ng/ml (C-F). (A) IFNg titration using TEER (arbitrary units) as proxy. Each concentration was tested on duplicated. (B) Experimental setup for C-F. Relative expres-sion of MX1 (C) and IL-15R (D) determined by qPCR. Bars indicate the mean ± SEM of technical triplicates in one of the duplicates of the assay. (E) PCA showing the top variable gene expression patterns determined by RNA-seq in the different cells lines and time points. (F) Line plot depicting LPP expression pattern (Log 2 counts, each point in duplicate).
Clusters 1, 7 and 12 are upregulated in the KD cells and represent a heat/stress response (cluster 1), mTOR signaling (cluster 7) and immune response (cluster 12). The downregulated DEG clusters 5 and 9 encompass genes involved in the metabolism of proteins and lipids, respectively. Of note, no overrepresented biological pathways were found for clusters 2 and 6, which were downregulated in the KD upon 40 hrs of IFNg stimulation.
Because a deregulated immune response is key in CeD pathophysiology, and LPP had not previously been associated with immune response, we examined the genes in cluster 12 more closely. Among the most statistically significant DEGs in this cluster (top 20 genes, ranked by p value), we observed IFN-inducible genes (IRF9, IFI27), genes encoding antiviral proteins (IFITM3), chemokines (CXCL9) and transcription factors (FOXC1) (Fig. 5C). Notably, most of these genes were lowly expressed in the controls after 8 hrs of stimulation, but showed strong upregulation in the KD cell line, suggesting an increased pro-inflammatory response upon LPP knockdown. Taken together, our data suggest that reduced LPP expression could contribute to perturbations in gene expression in intestinal epithelial cells exposed to inflammatory stress, particularly in genes involved in immune and metabolic processes.
Figure 5. LPP knockdown affects inflammatory and metabolic genes in response to IFNg. (A) Unsupervised hier-archical clustering of all DEGs relative to the untreated cell lines (|log2 FC|>1, adjusted p value ≤ 0.01) as detected by RNA-seq. Fifteen biologically relevant clusters were identified. Black boxes mark clusters in which the controls remained similar between them but were different compared to the KD. The color key (scaled log 2FC) indicates an increase (red)
or decrease (blue) in gene expression, respectively. (B) Bar plot illustrating the most significant biological pathways
(determined by Toppgene, https://toppgene.cchmc.org) in DEG clusters, if found (FDR≤ 0.05). (C) Heatmap showing the FC expression 8 hrs after IFNg stimulation in the top 20 DEGs in cluster 12. Darker blue represents upregulated genes. Lighter blue represents downregulated genes.
Figure 5. condition time time 3h 8h 40h condition kd scr wt −2 −1 0 1 2 A Scaled Log2 FC 1 condition time time 3h 8h 40h condition kd scr wt −2 −1 0 1 2 2 3 4 5 6 7 9 8 10 11 12 13 14 15 condition time time 3h 8h 40h condition kd scr wt −2 −1 0 1 2 0 5 10 15 9-monocarboxilicmetabolic process 5-Protein synthesis 12-defense response 7-Negative regulation TORC1 signaling 1- cellular response to heat/unfolded protein
-Log10 (value) 0 5 10 15 U p-re gu la te d D EG s cl ust er D ow n-re gu la te d D EG s cl ust er s 1 Response to heat/stress 7 TORC1 signaling 12 Immune response 5 Protein synthesis 9 Lipid metabolism WT SCR KD Scaled Log2 FC CD274 CXCL9 GRAMD2A SERPING1 CSF1 IFI27 IFITM1 C1S TRIM31 SEMA3F CFB IRF9 FOXC1 C7orf49 RAB27A PCDH7 IFITM3 TRIM16 XDH C Figure 5. condition time time 3h 8h 40h condition kd scr wt −2 −1 0 1 2 A B Scaled Log2 FC 1 condition time time 3h 8h 40h condition kd scr wt −2 −1 0 1 2 2 3 4 5 6 7 9 8 10 11 12 13 14 15 condition time time 3h 8h 40h condition kd scr wt −2 −1 0 1 2 0 5 10 15 9-monocarboxilicmetabolic process 5-Protein synthesis 12-defense response 7-Negative regulation TORC1 signaling 1- cellular response to heat/unfolded protein
-Log10 (value) 0 5 10 15 U p-re gu la te d D EG s cl ust er D ow n-re gu la te d D EG s cl ust er s 1 Response to heat/stress 7 TORC1 signaling 12 Immune response 5 Protein synthesis 9 Lipid metabolism WT SCR KD Scaled Log2 FC CD274 CXCL9 GRAMD2A SERPING1 CSF1 IFI27 IFITM1 C1S TRIM31 SEMA3F CFB IRF9 FOXC1 C7orf49 RAB27A PCDH7 IFITM3 TRIM16 XDH C condition time time 3h 8h 40h condition kd scr wt −2 −1 0 1 2 Scaled Log2 FC 1 condition time time 3h 8h 40h condition kd scr wt −2 −1 0 1 2 2 3 4 5 6 7 9 8 10 11 12 13 14 15 condition time time 3h 8h 40h condition kd scr wt −2 −1 0 1 2 0 5 10 15 9-monocarboxilicmetabolic process 5-Protein synthesis 12-defense response 7-Negative regulation TORC1 signaling 1- cellular response to heat/unfolded protein
-Log10 (value) 0 5 10 15 U p-re gu la te d D EG s cl ust er D ow n-re gu la te d D EG s cl ust er s 1 Response to heat/stress 7 TORC1 signaling 12 Immune response 5 Protein synthesis 9 Lipid metabolism WT SCR KD Scaled Log2 FC CD274 CXCL9 GRAMD2A SERPING1 CSF1 IFI27 IFITM1 C1S TRIM31 SEMA3F CFB IRF9 FOXC1 C7orf49 RAB27A PCDH7 IFITM3 TRIM16 XDH C
Ch
ap
te
r 2
Discussion
Of the more than 40 non-HLA loci associated with CeD, the LPP locus is one of the strongest susceptibility loci when it comes to odds ratio, and we have previously shown that LPP
expression is decreased in CeD biopsies4. Now, by using Caco-2 cells WT and Caco-2
sublines with LPP knockdown, we have examined the role of LPP in intestinal barrier integrity. By combining 2D and 3D culture approaches, we observed an impairment in proliferation and polarization upon LPP knockdown. Gene expression profiling by RNA-seq upon IFNg stimulation revealed that immune and metabolic genes are among those most affected in the LPP KD cell line. Our data demonstrate the value of using in silico predictions to infer biological functions for poorly characterized proteins and further our understanding of CeD etiology by suggesting there is an intricate interplay between barrier function and immune function.
We have demonstrated that reduced expression of LPP leads to defects in cell polarization which, in turn, could lead to alterations in tissue architecture. These processes could lead to a decreased capacity of intestinal epithelial cells to heal upon exposure to inflammatory damage or to decreased intestinal barrier integrity that contributes to leakage. Previous genetic association studies pointed to two cell polarity regulators, PARD3 and MAGI-2, as
potentially contributing to increased CeD susceptibility28. In CeD patients, Par-3 (encoded by
the PARD3 gene) has been found to be abnormally distributed to intestinal cell membranes
rather than to the apical junctions, thereby leading to barrier defects6,29. Similarly, we found
that LPP knockdown leads to abnormalities in actin and E- cadherin distribution, two essential components of the cell–cell adhesion protein complex. C1orf106, another CeD-associated gene that we predicted to be involved in cell-cell adhesion, has recently been knocked
out in mice. Indeed, in C1orf106-/- mice, the regulation of adherent junctions and intestinal
permeability is affected30. The striking phenotypes observed in our study and those of others
strengthen the hypothesis that intestinal barrier function is causally impaired in CeD patients and that this is partially due to genetic deregulation of barrier function genes.
We also observed that LPP knockdown impairs the proliferative capacity of cells. Although this result contrasts with the results of prior studies showing increased proliferation in
CeD-biopsy-derived enterocytes31,32, a recent study described that CeD-derived organoids display a
reduction in proliferation when compared to organoids generated from healthy individuals33. It
is not known whether these CeD organoids also exhibit decreased LPP levels, but the results obtained with this state-of-the-art organoid model are in line with our hypotheses. These organoid models now constitute a great opportunity to study the mechanisms underlying the defects in proliferation and polarization caused by reduced LPP expression (or of other proteins involved in cell-cell interaction) in intestinal barrier integrity.
Defects in tight junction formation have a detrimental effect on both TEER and cell permeability. Unfortunately, due to the variability in our data and the discrepancy between global and local permeability (Fig. 2 and Supplementary Fig. 3), we were not able to reliably measure the effect of reduced LPP expression on these two biological parameters. We think it is likely that the accuracy of our results might have been affected by biological and technical factors, including the inherent variability across cell lines (higher after ~11 days of culture in the KD cell line, as shown in Fig. 2B); the presence of scattered or irregular defects in the tight junction
across the monolayer23, or perhaps the fact that we measured TEER with unfixed electrodes
(chopstick), which detect electrical signals from small sections of the monolayers. Therefore, we propose the use of fixed electrodes (e.g. Endohm) that capture electrical resistance signals
from a larger surface34 or the assessment of the electrical impedance in monolayers cultured
directly in the Electric Cell-substrate Impedance Sensing system35 to reduce the variability
and improve TEER analysis. With regard to the cell permeability, after exposure to EDTA we observed an increase in focal leakage in the KD monolayers compared to controls. Although this finding implies a marked increase in the susceptibility of the KD cell line to environmental
stress (for which EDTA is a model)22, further replication experiments and perhaps the use of
different incubation periods with the stressor agent should be done to confirm our results. The rescue assay constitutes the standard for validation of shRNA silencing experiments.
However, expressing a rescue construct at physiological levels is challenging36. Indeed, in
our rescue experiment, LPP expression was very heterogenous and only partially rescued the proliferation phenotype in the culture and not the cell polarization phenotype. Given that protein overexpression is associated with endoplasmic reticulum stress, toxicity and barrier
function alterations, among other effects37,38, LPP overexpression could have prevented the
rescue of the cell polarization defect. We propose that rescue should be tried with a weaker promoter, followed by selection of clones that display an LPP level similar to that of the controls. Our transcriptional analyses uncovered differences between the KD cell line and controls. Surprisingly, despite the similarities in the phenotypic assessment between the parental WT cell line and the SCR cell line, transcriptional profiling by RNA-seq revealed major differences. Although the SCR sequence should not have any functional effect, based on in silico sequence similarities, unforeseen interactions or even the presence of foreign RNA may have provoked a response in the SCR cell line. In future experiments, multiple SCR sequences may have to be tested to create the most suitable control for the KD cell line. Here, we tried to exclude the effect of the non-targeting shRNA in the analysis of the gene response to detect the “true effects” caused by LPP knockdown by analyzing in detail the genes particularly affected in KD but not in controls.
We noted an upregulation of genes involved in the immune response and in mTOR signaling and a downregulation of genes involved in the metabolism of proteins and lipids. This is an interesting finding as inflammatory signals may trigger rapid metabolic changes to assure an appropriate adaptation to the environment and mTOR is known to be a major regulator of
Ch
ap
te
r 2
metabolic processes. LPP has been suggested to have a transcription co-activating function. It is therefore feasible that, because of this, several genes and biological pathways are affected upon LPP knockdown. Early studies have demonstrated that LPP can shuttle between the
nuclei and cell adhesion sites39, indicating that its involvement in both cell–cell interaction
and gene regulation is plausible. Although immunostaining of LPP shows that the protein is largely cytoplasmic (Fig. 3E, F), which contrasts with its role as co-activator, we cannot exclude the low-level presence of LPP under unstimulated conditions or the shuttling of LPP to the nucleus under stimulatory conditions. Finally, in the light of the recent understanding that
epithelial cells play important roles in immune homeostasis40, our finding that immune genes
are active in Caco-2 cells is also feasible. Interestingly, the mTOR signaling pathway has also been implicated in the impairment of autophagy and increased inflammatory response seen in patients with inflammatory bowel disease, another immune-mediated disease characterized
by a deregulated expression of pro-inflammatory cytokines (e.g. IFNg)41 and chemokines
(e.g. CXCL9)42 in the intestinal tissue. Thus, our finding that LPP knockdown leads to
reduced expression of mTOR and metabolic genes, together with an increased expression of inflammatory genes, suggest a potential contribution of mTOR in the immune activation of intestinal epithelial cells via LPP and constitutes an interesting finding to further explore in the context of CeD.
Genetic association studies and pathway analysis approaches have indicated a role for the immune system in susceptibility to CeD. As GWAS studies are hypothesis generating studies, they have also indicated that genes involved in cell–cell adhesion and barrier function could also play a role in CeD etiology. By performing LPP knockdown in Caco-2 cells we found evidence that LPP impacts cell proliferation and polarization. We hypothesize that this may lead to decreased barrier integrity and decreased effectivity to repair tissue damage inflicted by gluten-induced inflammation. By exposing Caco-2 cells to inflammatory conditions relevant to CeD, we found that LPP is also a potential regulator of the immune function of intestinal epithelial cells. However, the mechanism by which LPP affects these processes still needs
to be elucidated. With the advent of novel technologies for gene editing (CRISPR)43 and
development of state-of-the-art in vitro culture platforms that allow for mixed cellular cultures
(organ-on-chip)44, new avenues are opening up that will enable the exploration of these
hypotheses.
Acknowledgements
We thank Kate McIntyre for editorial assistance. This work was supported by a European Research Council Advanced Grant (FP/2007–2013/ERC grant 2012–322698) and an NWO Spinoza Prize (SPI 92-266) to C.W. We would like thank Rutger Modderman, Astrid Maatman, L.M.M. and J.M. for technical support. We also thank M.S. and J.F.R. for performing the sandwich assay and S.C.D.V.I. and L.M.J. for technical and academic support and for allowing us to use of the facilities at the Cell Biology Department of the University Medical Center Groningen.
References
1. Lindfors, K. et al. Coeliac disease. Nat. Rev. Dis. Prim. 5, 3 (2019).
2. Trynka, G. et al. Dense genotyping identifies and localizes multiple common and rare vari-ant association signals in celiac disease. Nat. Genet. 43, 1193–1201 (2011).
3. Withoff, S., Li, Y., Jonkers, I. & Wijmenga, C. Understanding Celiac Disease by Genomics. Trends in Genetics. 32, 295–308 (2016). 4. Kumar, V. et al. Systematic annotation of celiac
disease loci refines pathological pathways and suggests a genetic explanation for increased interferon-gamma levels. Hum. Mol. Genet. 24, 397–409 (2015).
5. van Elburg, R. M., Uil, J. J., Mulder, C. J. & Heymans, H. S. Intestinal permeability in pa-tients with coeliac disease and relatives of patients with coeliac disease. Gut. 34, 354–7 (1993).
6. Schumann, M., Siegmund, B., Schulzke, J. D. & Fromm, M. Celiac Disease: Role of the Epi-thelial Barrier. Cell. Mol. Gastroenterol. Hepa-tol. 3, 150–162 (2017).
7. Szakál, D. N. et al. Mucosal expression of claudins 2, 3 and 4 in proximal and distal part of duodenum in children with coeliac disease. Virchows Arch. 456, 245–50 (2010).
8. Holmgren Peterson, K., Magnusson, K. E., Stenhammar, L. & Fälth-Magnusson, K. Con-focal laser scanning microscopy of small-in-testinal mucosa in celiac disease. Scand. J. Gastroenterol. 30, 228–34 (1995).
9. Ciccocioppo, R. et al. Increased Enterocyte Apoptosis and Fas-Fas Ligand System in Celi-ac Disease. Am. J. Clin. Pathol. 115, 494–503 (2001).
10. Kumar, V., Wijmenga, C. & Withoff, S. From genome-wide association studies to disease mechanisms: celiac disease as a model for autoimmune diseases. Semin. Immunopathol. 34, 567–80 (2012).
11. Almeida, R. et al. Fine mapping of the celiac disease-associated LPP locus reveals a po-tential functional variant. Hum. Mol. Genet. 23, 2481–2489 (2014).
12. Van Itallie, C. M. et al. Biotin ligase tagging identifies proteins proximal to E-cadherin, in-cluding lipoma preferred partner, a regulator of epithelial cell–cell and cell–substrate adhe-sion. J. Cell Sci. 127, 885–895 (2014). 13. Nanayakkara, M. et al. A Celiac Cellular
Phe-notype, with Altered LPP Sub-Cellular Distri-bution, Is Inducible in Controls by the Toxic Gliadin Peptide P31-43. PLoS One 8, e79763 (2013).
14. Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 29, 15–21 (2013). 15. Li, H. et al. The Sequence Alignment/Map
for-mat and SAMtools. Bioinforfor-matics. 25, 2078–
2079 (2009).
16. Anders Simon, P. P. T. and H. W. HTSeq—a Python framework to work with high-through-put sequencing data. Bioinformatics .31, 166– 169 (2015).
17. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).
18. Juuti-Uusitalo, K. et al. Differential effects of TNF (TNFSF2) and IFN-γ on intestinal epithe-lial cell morphogenesis and barrier function in three-dimensional culture. PLoS One. 6, e22967 (2011).
19. Chang, B., Svoboda, K. K. H. & Liu, X. Cell po-larization: From epithelial cells to odontoblasts. Eu J Cell Biol. 98, 1-11 (2019).
20. Srinivasan, B. et al. TEER measurement tech-niques for in vitro barrier model systems. J. Lab. Autom. 20, 107–26 (2015).
21. Jelliffe, R. W., Schumitzky, A., Bayard, D., Fu, X. & Neely, M. Describing Assay Preci-sion-Reciprocal of Variance Is Correct, Not CV Percent: Its Use Should Significantly Improve Laboratory Performance. Ther. Drug Monit. 37, 389–94 (2015).
22. Li, C., Liu, T., Cui, X., Uss, A. S. & Cheng, K.-C. Development of In Vitro Pharmacokinetic Screens Using Caco-2, Human Hepatocyte, and Caco-2/Human Hepatocyte Hybrid Sys-tems for the Prediction of Oral Bioavailability in Humans. J. Biomol. Screen. 1084–1091 (2007).
23. Reiche, J., Schumann, M. & Richter, J. F. The Sandwich Assay: A Method for Subcellular Visualization of Paracellular Macromolecule Passage in Epithelial Sheets. Curr. Protoc. cell Biol. 78, 20.10.1-20.10.13 (2018).
24. Escudero-Hernández, C. et al. Association of the IL-15 and IL-15Rα genes with celiac dis-ease. Cytokine. 99, 73–79 (2017).
25. Discepolo, V. et al. Interplay between type-1 interferon and IL15 in celiac disease. Dig. Liver Dis. 45, e277 (2013).
26. Schattgen, S. A., Oguin, T. H. & Thomas, P. G. The antiviral molecule Mx1 positively regulates the induction of type I IFN in response to influ-enza infection. J. Immunol. 196, 202.7-202.7 (2016).
27. Psarras, A., Emery, P. & Vital, E. M. Type I interferon–mediated autoimmune diseases: pathogenesis, diagnosis and targeted therapy. Rheumatology. 56, 1662-1675 (2017). 28. Wapenaar, M. C. et al. Associations with tight
junction genes PARD3 and MAGI2 in Dutch patients point to a common barrier defect for coeliac disease and ulcerative colitis. Gut. 57, 463–467 (2007).
29. Schumann, M. et al. Cell polarity-determining proteins Par-3 and PP-1 are involved in epi-thelial tight junction defects in coeliac disease.
Ch
ap
te
r 2
Gut. 61, 220–8 (2012).
30. Mohanan, V. et al. C1orf106 is a colitis risk gene that regulates stability of epithelial ad-herens junctions. Science. 359, 1161–1166 (2018).
31. Savidge, T. C., Walker-Smith, J. A. & Phillips, A. D. Intestinal proliferation in coeliac disease: looking into the crypt. Gut. 321–333 (1995). doi:10.1136/gut.36.3.321
32. Nanayakkara, M. et al. Enterocyte proliferation and signaling are constitutively altered in celiac disease. PLoS One. 8, e76006 (2013). 33. Freire, R. et al. Human gut derived-organoids
provide model to study gluten response and ef-fects of microbiota-derived molecules in celiac disease. Sci. Rep. 9, 7029 (2019).
34. Sheller, R. A., Cuevas, M. E. & Todd, M. C. Comparison of transepithelial resistance mea-surement techniques: Chopsticks vs. Endohm. Biol. Proced. Online. 19, 4 (2017).
35. Szulcek, R., Bogaard, H. J. & van Nieuw Amer-ongen, G. P. Electric Cell-substrate Impedance Sensing for the Quantification of Endothelial Proliferation, Barrier Function, and Motility. J. Vis. Exp. 85, (2014).
36. Sigoillot, F. D. & King, R. W. Vigilance and validation: Keys to success in RNA screening. ACS Chem. Biol. 6, 47–60 (2011).
37. Xu, C., Bailly-Maitre, B. & Reed, J. C. Endo-plasmic reticulum stress: cell life and death
de-cisions. J. Clin. Invest. 115, 2656 (2005). 38. Ma, X. et al. Intestinal Epithelial Cell
Endoplas-mic Reticulum Stress and Inflammatory Bow-el Disease Pathogenesis: An Update Review. Front. Immunol. 8, 1271 (2017).
39. Petit, M. M. et al. LPP, an actin cytoskeleton protein related to zyxin, harbors a nuclear ex-port signal and transcriptional activation ca-pacity. Mol. Biol. Cell. 11, 117–29 (2000). 40. Okumura, R. & Takeda, K. Roles of intestinal
epithelial cells in the maintenance of gut ho-meostasis. Exp. Mol. Med. 49, e338 (2017). 41. Fan, J.-B. et al. Type I IFN induces protein
IS-Gylation to enhance cytokine expression and augments colonic inflammation. Proc. Natl. Acad. Sci. U. S. A. 112, 14313–8 (2015). 42. Singh, U. P. et al. Chemokine and cytokine
levels in inflammatory bowel disease patients. Cytokine. 77, 44–9 (2016).
43. Ahmadzadeh, V., Farajnia, S., Baghban, R., Rahbarnia, L. & Zarredar, H. CRISPR-Cas system: Toward a more efficient technology for genome editing and beyond. J. Cell. Biochem. (2019).
44. Moerkens, R., Mooiweer, J., Withoff, S. & Wi-jmenga, C. Celiac disease-on-chip: Modeling a multifactorial disease in vitro. United Eur. Gas-troenterol. J. 7, 467–476 (2019).
39
Supplemental information
LPP WT SC KD#1 KD#2 KD#3 KD#4 Actin A B R el at iv e LP P % m R N A le ve l WT SCR KD_1 KD_2 KD_3 KD_4 0 50 100 150 *** *** ***Supplementary Figure 2. LPP KD validation. LPP KD cell lines were generated by transfecting pRS vectors containing four different short hairpin RNA (shRNA) targeting LPP (KD#1-4). Caco-2 cells WT or SCR were included as controls. After puromycin selection, LPP expression level was determined by western Blot (A) or qPCR (B). Bar plots show mean ± SEM. Data is representative of at least three independent experiment. ***p <0.001; one-tailed Student’s t-test. Supplementary Figure 1. LPP shRNAs and targets. Diagram illustrating the transcripts, incRNAs and miRNA in the
LPP locus. The main LPP isoforms (1-3) are marked with a red star. Arrows and dashed lines indicate the exons targeted
by each shRNA. Figure adapted from https://www.ensemble.org. Color key indicates gene types and individual shRNAs targeting LPP.
Ch
ap
te
r 2
Supplementary Figure 3. Increased permeability at focal sites in the LPP KD cell line. Caco-2 cell lines were ex-panded in transwell supports for one week. The cells were incubated with of 2.5 mM EDTA for 15 minutes. Dextran-10 (red) was applied apically to detect apical-to-basolateral macromolecule passage. A basal tracer (green) was added as control stain the cell–cell contacts. Cells were fixed and mounted for microscopy. Areas of passage can be identified in red.
3 Department of Immunology, K.G. Jebsen Coeliac Disease Research Centre, University of Oslo, Oslo, Norway 4 Deutsches Rheumaforschungszentrum Berlin (DRFZ), An Institute of the Leibniz Association, Berlin, Germany 5 Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin,
and Berlin Institute of Health, Department of Gastroenterology, Infectious Diseases and Rheumatology, Berlin, Germa-ny
6 Department of Immunohematology and Blood Transfusion, Leiden University Medical Center, Leiden, the Netherlands 7 Department of Internal Medicine and Radboud Center for Infectious Diseases (RCI), Radboud University Medical
Center, Nijmegen, the Netherlands
8 Department of Computational Biology for Individualised Infection Medicine, Centre for Individualised Infection
Medi-cine, Helmholtz Centre for Infection Research, Hannover Medical School, Hannover, Germany
