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 7
Ch
ap
te
r 7
General discussion
The underlying molecular culprits in celiac disease
Despite the considerable progress that has been made in dissecting the genetic architecture of celiac disease (CeD), the mechanisms behind the disease’s etiology remain unknown1.
Interestingly, although genome-wide association studies (GWAS) have suggested that immune genes such as IL21, CD28, IL18R1, IL18RAP, RUNX3, CCR1 and SH2B3 are, apart from HLA-DQ2/8, the major contributors to CeD pathogenesis1, these studies have
also revealed that a small group of genes, including, Lipoma-preferred partner (LPP) and
C1orf106, are potentially involved in cell–cell interactions and intestinal barrier function and
might reveal another mechanism apart from a dysregulated immune system2. The bias toward
immune genes suggested by GWAS results may reflect the used platform for association studies, namely the Immunochip, a custom-designed DNA single nucleotide polymorphisms (SNPs) chip that included SNPs covering some 200 mainly immune-related loci3. Performing
association studies that cover the entire genome may reveal associations to non-immune genes in autoimmune diseases like CeD.
Various gene prioritization strategies – including mining for the overlap between SNPs identified by CeD GWAS with annotated publicly available datasets of regulatory elements, co-expression analysis and expression quantitative trait loci (eQTL) analysis – have indicated that CeD SNPs are enriched in B cell enhancers and that T helper cells (Th1, Th2 and Th17), NK and B cells are implicated in CeD pathogenesis. These analyses have also revealed a set of 15 interconnected genes that converge in the interferon gamma (IFNg pathway2,
thereby confirming earlier studies suggesting a role for this pathway in CeD pathogenesis4.
More recently, the latest GWAS study by Ricaño-Ponce et al5 not only confirmed the link
between the IFNg signaling pathway and CeD, it also revealed a novel association with the NFkB signaling pathway that mediates critical inflammatory events. These results indicate that several biological inflammatory pathways operating in different cell types may contribute to CeD.
In addition to host genetic factors, there is evidence indicating that environmental factors other than dietary gluten, such as viral infections and the gut microbiome, may also be associated with CeD6. For example, a decrease in the number of gram-positive bacteria and a concomitant
increase in the number of gram-negative bacteria has been observed in CeD individuals7.
Moreover, recent studies have reported that Proteobacteria and Firmicutes are the most abundant bacteria phyla present in children and adults with CeD, respectively8. Although it
remains unclear whether these changes in the microbiome are the cause or the consequence of CeD, one proposed mechanism is the immunomodulatory properties of bacteria or their products9.
Together these observations reinforce the notion that CeD is truly a multifactorial disease in which different cell types and multiple genetic and environmental risk factors jointly contribute to disease pathology. Although GWAS have pointed to several molecular culprits, – genes, biological pathways and major cell types – associated with CeD1, most of these findings lack
experimental validation. This thesis describes research that was performed to address this knowledge gap. We investigated the functional and transcriptional consequences of genetic and environmental perturbations relevant in the context of CeD in a cell-type-specific manner. For this we integrated the results of in silico approaches with data from wet lab experiments (see Fig. 1 and General Introduction).
Impaired intestinal barrier function
The intestinal epithelial barrier is one of the molecular culprits in CeD etiology implicated by GWAS. This barrier serves as a filter, allowing the selective passage of nutrients, electrolytes and water, but it also has an essential function in the defense against pathogens and other harmful agents. A “leaky gut” has been has been suggested as an underlying causal mechanism in the development of immune-mediated diseases including CeD for many years10. One of the
first unbiased genetic studies in CeD revealed association to the gene myosin IXB11, and
suggested a primary impairment of the intestinal barrier. Under physiological circumstances, the intestinal epithelium barrier is maintained due to the presence of tight junctions (TJs) that act as a seal between adjacent epithelial cells and control paracellular transport10. More than
20 years ago it was already shown that the TJ integrity is compromised in patients with CeD, resulting in increased intestinal permeability that may allow the passage of gliadin, a peptide contained within gluten, into the lamina propria12,13. These gliadin molecules in turn then lead
to an immune response that is mediated via the activation of gluten-specific CD4+ T cells14.
This immune response ultimately results in the secretion of high levels of the pro-inflammatory cytokine IFNg and activation of intraepithelial cytotoxic lymphocytes (IE-CTLs), two processes that both perpetuate barrier dysfunction4,15. This suggests that, in CeD, primary defects in
enhanced intestinal permeability and a deregulated immune response go hand in hand and augment each other.
To gain more insight into the function and impact of one of the CeD-associated GWAS genes prioritized to play a role in intestinal barrier function, we focused on LPP. This gene is located within the locus that shows the strongest association to CeD outside the HLA region16. Because
the level of expression of LPP is decreased in intestinal biopsies of CeD patients compared to intestinal biopsies of controls2, we used Caco-2 cells as a model for the intestinal epithelial
barrier and performed LPP knockdown (KD) in these cells. Here we noted that reduced expression of LPP impairs some processes essential for intestinal homeostasis, including cell proliferation and barrier permeability (Chapter 2), a condition similar to the intestinal phenotype observed in CeD patients13,17. Most strikingly, upon LPP KD we observed severe defects in
cell polarization in our 3D Caco-2 spheroids. Our results are in line with the observation of proliferative abnormalities in embryonic fibroblasts derived from LPP knockout (KO) mice18.
Ch
ap
te
r 7 Moreover, our results also match the polarization defects seen in both LPP KO embryos of Zebra fish (a recognized model for studying cell polarity and organ morphogenesis)19 and the
intestinal epithelial cells of CeD individuals20,21. Together these results indicate that LPP could
be fundamental for maintaining the intestinal architecture and the ability to repair the tissue damage caused by the abnormal immune response induced by gluten.
LPP and immune modulation
In a previous study using a gene reporter assay in the cervical epithelial cell line HeLa,
Friederich et al. observed that LPP has transcriptional activation capacity22. The authors also
found that, although LPP was located in the cytoplasm under resting conditions, it accumulated in the nucleus in response to treatment with Leptomicin B22, a selective inhibitor of the transport
of proteins containing a nuclear export sequence. This is interesting because the capacity to
Therapeutic interventions
Intestinal epithelial cells (IEC) Mechanical barrier
immune functions
eQTL analysis+ GWAS (Novel candidate genes)
TRAFD1 master regulator
IE-CTLs Tissue alarmins (transcriptional reprogramming)
RP11-291B21.2 (naïve and effector cell-Activation)
SuRE, CRISPR Single cell methods
CeD-in-a-chip Molecular culprits LPP A B C D E Gluten Microbiome Cytokines IE-CTLs IEC IEC Chip IE-CTLs Other Immune cells IFN response MHCI signaling Common set of IFN response genes
Fig.1. Major conclusions and future directions towards a better understanding of molecular mechanisms under-lying CeD. We found that intestinal epithelial cells are not just a barrier that allows the passage of substances through
the lumen, they also modulate the immune response in the gut (Chapter 2) (A). By integrating eQTL analysis with GWAS
data, we observed that the most likely causal genes converge into innate and adaptive immune pathways. Interestingly, ~10% of these causal genes are ncRNAs, indicating that a broad range of biological processes are perturbed in CeD through the actions of coding and non-coding genes and that their effect occurs in particular cell types (Chapter 3). We also identified TRAFD1 as a novel regulator of genes that control IFNg signaling, a biological pathway known to influence the activation of immune cells (B). We noted that the tissue alarmins IL-15 and IFNb dramatically shape the response
of thousands of genes and key biological pathways (proliferation, IFN signaling, protein synthesis) in IE-CTLs, which are key effectors for epithelial cell destruction. This suggests that endogenous factors have a major impact on tissue destruction (Chapter 4). We also identified a novel lncRNA, RP11-291B21.2, as a regulator of the activation of IE-CTLs (Chapter 6), a process that might be tightly tuned to allow effector immunity while avoiding tissue damage (C). Together
our findings indicate that several coding and non-coding genes, cell types and tissue signals contribute to CeD (A-C).
However, further efforts are required to fully understand how and to what extent they are connected and how they influence disease outcome before we can translate them into targets for therapeutic intervention. Through advances in transcriptome analysis, genome editing and new culturing methods that emulate complex environmental and tissue architecture (organ-on-a-chip), this could soon become a reality (D, E).
shuttle between the nucleus and cytoplasm is characteristic of protein complexes that regulate transcription, indicating that LPP could participate in regulation of gene expression.
We performed RNA-seq analysis on LPP-KD cells expanded in transwells (2D model) and exposed to IFNg. We selected this cytokine because it has been found to be upregulated in the intestinal mucosa of CeD individuals23. In addition, IFNg has demonstrated a remarkable
capacity to disrupt the TJs of human-derived intestinal epithelial monolayers24 and the
3D-architecture of mouse-derived intestinal organoids25. Most importantly, although the IFNG
locus itself is not associated with CeD, Kumar et al. found that several CeD candidate genes converge into the IFNg signaling pathway2, which suggests that the deregulation of IFNg is a
cause rather than a consequence of the disease.
These findings motivated us to explore the effects of IFNg in our Caco-2 model. We observed an enhanced proinflammatory response upon IFNg stimulation as compared to control Caco-2 cells. Among the genes upregulated in LPP-KD Caco-2 cells upon IFNg treatment, we found innate immune genes such as CCL9, that encodes a chemokine that promotes the recruitment of monocytes, and also members of the Toll-like receptor (TLR) signaling cascade (MYD88 and MAP3K8). TLRs, which are expressed in immune cells and intestinal epithelial cells, are essential for triggering the production of inflammatory mediators upon detection of pathogens and danger signals. It is therefore possible that epithelial cells with reduced LPP respond hyperactively to “common” environmental signals such as gliadins and microbiota, resulting in an enhanced inflammatory response in the intestine of CeD-affected individuals (Fig. 2). Amongst other significantly upregulated genes in the LPP KD cell line, we found CD274, which encodes the regulatory molecule Programmed death-ligand (PD-L1), and SERPING1, a negative regulator of the protein C1 of the complement system. The interaction of PD-L1 commonly expressed in the surface of antigen presenting cells with its receptor PD-1, which is expressed in T cells, acts as a natural break to reduce over-activation and maintain self-tolerance. Interestingly, the PD-L1/PD-1 signaling pathway can inhibit the activation of T cells, thus contributing to the immune-scape of cancer cells26, and currently the use of monoclonal
antibodies anti-PD-1 constitutes an effective therapeutic approach against melanoma27. In
the case of CeD, the CD274 and SERPING1 loci have been recently found to be associated with the disease (this thesis, Chapter 3). So far, the link between CeD and SERPING1 is not clear. However, the expression of PD-L1, which is encoded by CD274, was found to be considerably higher in the intestinal epithelial cells and lamina propria of CeD patients compared to controls28. This upregulation may be a mechanism to avoid being targeted by
IE-CTLs, which are able to recognize and kill stressed or infected cells. In contrast, PD-1 was not expressed in these patients, suggesting that alterations in the PD-L1/PD1 signaling pathway could be relevant in the inflammatory response and loss of tolerance to gluten seen in CeD. Because the expression of CD274 was significantly upregulated in LPP-KD Caco-2 cells under inflammatory conditions, it is conceivable that LPP deregulation in the intestinal mucosa
Ch
ap
te
r 7
contributes to defects in the PD-L1/PD1 signaling pathway by modulating PD-L1 expression in the presence of an inflammatory environment dominated by IFNg (Fig. 2). Although our 2-D model allowed us to gain insight in to the regulatory function of LPP and generate some hypotheses about it is role in CeD pathogenesis, a model using 3D organoids, such as that proposed by Bardenbaucher et al 25, could help us to look at the impact of LPP-KD on cell–cell
adhesion at a better resolution.
Earlier we suggested that both barrier leakage and a hyperinflammatory reaction to microbiota and gluten peptides could contribute to CeD in a noxious cycle. In Chapter 2 we showed that both processes might even be featured within one tissue type, the intestinal epithelia, via the role of LPP. As the intestinal barrier is formed from different cell types, it would be very interesting to study the effects of intestinal barrier genes in different intestinal cell populations and perhaps under other intestinal inflammatory conditions. Indeed, some reports have indicated that the dysfunction of Paneth cells, which are essential for preventing the translocation of the commensal microbiome through the production of antimicrobial peptides, can be involved in both the alteration of the intestinal microbiota and the development of an abnormal intestinal immune response in patients with Crohn’s disease (a form of inflammatory bowel disease (IBD)). A prior study indicated that the number of Paneth cells is reduced in the intestinal mucosa of CeD patients29, suggesting that these cells are affected in disease
pathogenesis. Furthermore, it was recently reported that the release of the antimicrobial peptides in Paneth cells depends on the presence of IFNg, rather than on stimulation of these cells by bacteria30. 1 2 3 IE-CTL IEC APC BC GS-TC + Passage/response to gluten/ microbiome Gluten Microbiome IL-15 IFN-I IL-21 TG2 Anti-gluten Anti-TG2 MHCI IFNg LPP KD + PD-L1 + TLR + CCL9 + Barrier defects +Recruitment +inflammation +BC survival Transcriptional reprogramming CCR1 CCR2 IL18R +Th-1 polarization LncRNAs TRAFD1 +IFN signaling MHC-I Antigen processing & presentation LncRNAs +Inflammation? -PD1 HLA DQ2/8
Fig. 2. The current CeD puzzle. A model for the role of the intestinal barrier, innate immunity and genes that may play
a role in these is presented in this figure, based on current literature and findings in this thesis. Blue and black letters mark genes and biological pathways that are linked to CeD by genetics or by in vitro observation/biopsies, respectively. The symbols + and - indicate whether certain gene/biological process is increased/activated or decreased/suppressed, respectively. The dotted lines separate the different cell types into: 1) The intestinal epithelial cells (IEC) in close
inter-action with the IE-CTLs; 2) the innate immune cells (monocytes and dendritic cells, that act as antigen presenting cells
Thus far the biology of LPP had remained largely unknown. However, SNPs in the LPP locus have been found to be associated with two other autoimmune diseases, vitiligo31 and
rheumatoid arthritis (RA)32. Similar to CeD, affected individuals in both diseases present with
increased IFNg expression in the skin33 and synovial fluid34, respectively. Because vitiligo occurs
specifically in skin epithelial cells, this raises the possibility that alterations in the expression of
LPP could fuel the local inflammatory response induced by IFNg. On the other hand, although
the link between RA and other systemic autoimmune diseases that involve LPP is perhaps less obvious, a recent report indicated that the loss of intestinal barrier integrity or “a leaky” gut increases the traffic of antigens towards the lamina propria, leading to a subsequent activation of autoreactive immune cells and the development of autoimmune diabetes in mice35. LPP
could thus contribute to autoimmune diseases through its non-immune (cell–cell adhesion) or immune functions (response to inflammatory stimuli), or both, and further studies are needed to interrogate this.
CeD is not just a CD4+ T cell–mediated disease
During my thesis work, novel loci were associated with CeD1 and large datasets (e.g. the
BIOS cohort)36 became available for eQTL analysis, allowing us to prioritize causal SNPs by
eQTL mapping. We therefore measured eQTL effects using RNA-seq data of whole blood from approximately 4000 healthy individuals comprising the BIOS cohort. The SNPs exerting eQTL effects were further prioritized using an in silico approach encompassing four different statistical methods (Chapter 3). This identified 118 causal candidate genes whose expression is regulated in cis (within 1.5Mb± of the disease locus) by CeD risk-SNPs. Subsequent co-expression and reactome pathway analysis grouped these 118 genes into four different clusters of coexpressed genes, three of which matched to pathways. One cluster containing genes such as CD28 and CTLA4 was associated with T cell co-stimulation, a process essential for T cell activation37. This result confirms previous findings describing the central
role gluten-specific CD4+ T cells play in CeD pathogenesis. The other two clusters encompass
adaptive and innate cytokine signaling genes (e.g. IL21, IL18RAP, IL18R1) and chemokines (e.g. CCR1, CCR2), pointing to the involvement of both adaptive and innate cells and their cytokines. In agreement with this result, upregulation of adaptive and innate cytokines and chemokines has been observed in the intestinal mucosa of CeD patients38.
Given that the receptors for cytokines/chemokines, and their receptor molecules, are broadly expressed in the cells making up the intestinal barrier –epithelial cells, IE-CTLs, dendritic cells, monocytes and B cells – this finding shows that the genetic involvement in CeD immunology is more complex than initially suspected. Rather than just one cell being causal, it is the interplay of different immune cells and genetic risk factors that contributes to CeD pathology. It could be that IL-21 is expressed by gluten-specific CD4+ T cells and provides help to the B cells that
play a role in the disease etiology39. B cells produce the autoantibodies detected in CeD and
have been postulated to serve as antigen-presenting cells40. In the same way, IL-18, which
Ch
ap
te
r 7
to its receptor ((IL18R) encoded by IL18RAP and IL18R1) is thought to contribute to the polarization of T cells towards a Th1 profile41, that match the phenotype of the gluten-specific
CD4+ T cells that produce IFNg. Finally, CCR1 and CCR2, which are expressed in monocytes,
dendritic cells, neutrophils and NK cells42, could mediate the recruitment and activation of the
above-mentioned cells at sites of inflammation and thereby sustain the tissue damage (Fig. 2).
The size of our RNA-seq dataset also allowed us to investigate the effect of risk-SNPs on genes far away or even on other chromosomes, so called trans-eQTL analysis. In total we measured 497 trans-eQTL effects. We then performed mediation analysis, which yielded a subset of 40 genes that appeared to be under control of one master regulator: TRAFD1.
TRAFD1 is a poorly characterized gene that seems to act as a regulator of the NFkB
signaling pathway43, a pivotal mediator of inflammatory responses. Pathway analysis on this
subset of 40 genes suggested that they are mainly involved in IFN signaling (IFNg signaling/ IFN signaling) and Major Histocompatibility Complex Class I (MHC-I) antigen processing/ presentation. Three of these trans-eQTL genes, STAT1, CD274 and PDCD1LG2, are also under control of cis-eQTL SNPs. These results show that CeD risk-SNPs can modulate IFN signaling and MHC-I pathways through both cis and trans modulatory mechanisms, indicating that larger numbers of SNPs deregulate multiple cis and trans genes, converging on a limited number of pathways, and could possibly have an additive effect.
The MHC-I pathway is necessary for recognizing antigens via T cell receptor (TCR) in IE-CTLs44, suggesting a genetic and causal link between IE-CTLs and CeD. IE-CTLs are effector
memory cells that display an “activated yet resting”45 state in the epithelial layer, which means
that they await additional signals to become fully activated. Given the enrichment of genes regulated by TRAFD1 in the IFNg signaling pathway, IFN signaling could be one of the signals needed to activate the IE-CTLs, endowing them with killing properties to target the intestinal epithelial cells. Interestingly, STAT1 is a central mediator of the IFN signaling pathway46,
whereas PD-L1 and PD-L2 (encoded by CD274 and PDCD1LG2, respectively) are two co-inhibitory molecules that bind to PD-126. Thus, it is plausible that in addition to a deregulation
in IFN signaling, an abnormal interaction between PD-1 in the surface of T cells and its ligands could reinforce the activation of both the gluten-specific CD4+ T cells and the IE-CTLs leading
to CeD-specific pathology. IE-CTL activation in CeD
Under physiological circumstances, IE-CTL numbers are low; there are thought to be approximately 11 IE-CTLs present per 100 epithelial cells47. Unfortunately, under circumstances
that are not fully understood, their numbers can increase abnormally, contributing to the development of organ-specific autoimmune disorders such as type I diabetes (T1D)48 and
CeD49. Currently, little is known about how cytokines modulate the activation of IE-CTLs, mostly
because the affected tissues are difficult to access and biopsies yield only a limited number of cells50. To overcome these limitations partly, we took advantage of a co-culture system
in which IE-CTLs from intestinal biopsies are expanded51, allowing for the characterization
of their transcriptional and epigenetic response. We stimulated the cells using two tissue alarmins, IL-15 and interferon-b (IFNb)52, and an adaptive cytokine, IL-2153, that have been
found to be upregulated in tissues of several organ-specific autoimmune disorders54, including
CeD52–55. We found that the tissue alarmins produced by epithelial cells and antigen-presenting
cells promote a massive transcriptional response, whereas the impact of the IL-21 produced by gluten-specific CD4+ T cells was minor. This indicates that although gluten-specific CD4+
T cells are key players in CeD, the signals released by the stressed tissue, alarmins, have a significant impact on IE-CTL activation (Chapter 4 and Fig. 2).
Remarkably, all three stimuli promoted the expression of a common set of interferon signaling genes (e.g. IFI3, MX1, IFI16) and immune cell activation genes (e.g. GZMB, TNF), implying that these genes may cooperate to license IE-CTLs to kill. Of note, STAT1, GB1P and PARP9, which are proinflammatory genes regulated by TRAFD1, were also induced by these three stimuli, indicating that tissue alarmins and adaptive cytokines may contribute to CeD by modulating the expression of CeD candidate genes, particularly in IE-CTLs. Most importantly, we found that the common set of IFN genes was enriched in GWAS loci associated to multiple autoimmune diseases (including IBD, T1D, psoriasis, RA, multiple sclerosis (MS)) suggesting a link between this gene set, the abnormal activation of cytotoxic lymphocytes (CTLs)/IE-CTLs and the concomitant cell destruction observed in individuals with organ-specific autoimmune disorders.
Taken together, several lines of evidence – the enrichment in autoimmune disease GWAS loci we saw in the group of IFN-signaling genes induced by cytokines in IE-CTLs (Chapter 4), our evidence that IFN-signaling is a causal pathway in CeD (Chapter 3), and previous reports showing that treatment with IFN type I (IFNa) to counter viral infections can trigger CeD56 – constitute intriguing findings that prioritize the generation of therapeutic interventions
for autoimmune diseases by targeting the IFN expression pathway.
LncRNAs: novel CeD candidate genes and regulators of T cell activation
We also observed that the massive changes in gene expression induced by tissue alarmins were not dependent on changes in H3K27ac, a histone mark of actively transcribed genes, indicating that additional regulatory mechanisms wait to be identified (Chapter 4). Considering that the expression of coding genes peaked at 3 and 4 hours after cytokine stimulation, while the expression of non-coding RNAs (ncRNAs) was dramatically reduced at the same time points, and keeping in mind that non-coding transcripts are essential in gene regulation57, it
is plausible that ncRNAs may be one of the mechanisms that control the dramatic changes in gene expression observed in IE-CTLs exposed to inflammatory conditions that resemble the gut environment conditions of CeD individuals. Therefore, ncRNA could be considered an additional piece of the CeD puzzle (Fig. 2).
Ch
ap
te
r 7
Indeed, approximately 10% of the candidate genes regulated in cis by CeD risk-SNPs (cis-eQTL) do not encode proteins, but are rather long non-coding RNA genes (lncRNAs) (Chapter 3). Although the function of lncRNAs is not well known in general57, the co-expression
pattern of all the cis-eQTL genes indicated that the lncRNAs could participate in immune-activation processes and cytokine and chemokine signaling (Fig. 2). Because cytokines and chemokines influence the activation and recruitment of immune cells58, one can hypothesize
that alterations in the expression of these lncRNA could drive major influx of inflammatory cells to the intestine, thereby worsening the ongoing activation of innate (e.g. dendritic cells, monocytes) and adaptive immune cells (e.g. CD4+ T cells and IE-CTLs) and intestinal barrier
abnormalities. Although we do not know yet how these cis-eQTL lncRNAs increase the susceptibility to CeD, our results provide a basis for future exploration.
At the beginning of my studies we identified a non-characterized lncRNA, RP11-291B21.2, based on its unique expression pattern in RNA-seq data from intestinal- and blood-derived immune cells (Chapter 6). In silico guilt-by-association analysis using Genenetwork59
suggested a role for RP11-291B21.2 in maintaining the naïve status of CTLs. Subsequent analysis of RP11-291B21.2 expression patterns in different functional subpopulations of CTLs (e.g. naïve, effector memory and central memory) from blood confirmed this prediction. Furthermore, a marked reduction in the expression of both effector molecules (e.g. IFNg, tumor necrosis factor and CD107a) and proinflammatory genes (e.g. IL8, FOS and JUNB) was observed upon RP11-291B21.2 knockdown in IE-CTLs, which are effector memory cells awaiting activation. This indicated that RP11-291B21.2 inhibits the proinflammatory response on fully differentiated CD8+ T cells.
Our results show that RP11-291B21.2 has a dual function. It both maintains a resting status in the naïve cells and controls the activation of effector-memory CTLs, two processes essential for avoiding excessive inflammation and tissue damage. Additional experiments are now needed to elucidate how RP11-291B21.2 achieves this task and if this lncRNA could have a role in disease mechanisms. We used anti-CD3 antibodies to test the effect of RP11-291B21.2 in IE-CTL activation, however it would be more physiologically relevant to test if the regulatory effects of this lncRNA occur under inflammatory conditions relevant for CeD, for example upon IL-15, IFNg or IFNb stimulation. Nonetheless, RP11-291B21.2 clearly has a complex and specific function in CD8+ T cell activation.
Future directions
Over the past few years GWAS and complementary in silico analysis have allowed the prioritization of many candidate genes. These studies have shown that most of these genes are expressed and regulated in a cell- and context-specific manner. Nevertheless, the technical and biological constraints on obtaining disease-relevant tissues and cell types have limited the study of these genes to blood samples. While blood is a tissue that contains many immunological cell types, it is still a mixture, which makes it difficult to explore the effects of
the individual cell types. Recent advances in sequencing, genome editing and cell culture have now enabled the development of technical methods that will facilitate the study of gene function and gene regulation in relevant cell-types and under basal and perturbed conditions because these new methods require smaller amounts of input material and, most importantly, the cells can be maintained under conditions that closely resemble the intestinal environment. One of these new methods is single-cell (sc) RNA-seq, a technique which allows us to look at the transcription level at single-cell resolution. scRNA-seq can be applied, for example, to intestinal biopsies to study the cellular composition of the intestine from CeD patients and controls. scRNA-seq can also uncover different activation statuses within a cell population, as well as novel cell types and biological pathways that remained unrecognized with the use of previous RNA-seq platforms60.
Most CeD-associated SNPs overlap with non-coding elements bearing regulatory elements1,
and this means that assessing open chromatin at single-cell level would be very useful for predicting the effects of SNPs in individual cells. The application of the Assay for Transposase-Accessible Chromatin using single-cell sequencing (scATAC-seq) allows the identification of parts of the genome that are more accessible to regulatory proteins in thousands of single cells in parallel. The technical advantages and high sensitivity of single-cell methods can help integrate layers of genetic information with regulatory and expression information, and thereby develop a more complete picture of the transcriptional and regulatory landscape of disease-relevant cell types. This new information may translate into a better understanding of disease pathology and lead to new treatments or diagnostic methods specifically tailored to the cell type and conditions most relevant to the disease.
Another promising tool for studying how CeD genetic variants affect gene expression is Survey of Regulatory Elements (SuRE) SNP61. Although existing in silico analyses can
pinpoint potential risk SNPs, they lack sufficient resolution to identify causal SNPs. In contrast, SuRE can identify regulatory regions (enhancer and promoter activity) and systematically screen millions of human SNPs for their potential effects on regulatory activity in specific cell types62. The SNPs uncovered by SuRE, and perhaps intriguing candidate genes regulated
by these SNPs, could be further validated by applying the Clustered Regularly Interspaced Short Palindromic Repeats/associated protein 9 (CRISPR/Cas9) method. Due to its specificity and flexibility, CRISPR/Cas9 and its variants (e.g. CRISPRi) can be applied to knockout or overexpress a single gene or multiple genes simultaneously63. In addition, CRISPR base
editing could be used to replace a protective SNP with a risk SNP using, thus allowing us to directly asses the effects of genetic variation on gene expression and biological pathways. CRISPR editing and other gene-editing techniques should be applied in the most relevant and context-specific manner possible. One option to validate candidate genes and cell-specific eQTLs would be to use biopsy-derived cells. However, this approach is difficult and cannot capture the cell–cell interactions that we know are present in tissues. In contrast,
organ-on-Ch
ap
te
r 7
a-chip, a new multi-cellular model that mimics the tissue microenvironment using cell lines derived from primary cells, is an ideal platform to overcome this issue. With the advent of the human induced pluripotent stem cell technology, it is now possible to differentiate cells into virtually any cell type64. As a consequence, the starting material to perform functional
experiments can be taken from a less invasive source than intestinal biopsies, e.g. blood or urine. The pluripotent cells from these biological fluids can be reprogrammed towards intestinal epithelial cells or other cell types that can then be seeded onto the chip. In the case of intestinal epithelial cells, the mechanical forces that are applied to the chip induce the formation of a 3D structure that resembles the gut architecture, with the epithelial cells forming a villus-like conformation, and this model even shows the development of accessory cells (e.g. goblet cells, Paneth cells)65.
Another advantage of the chip culture system is that it supports the simultaneous co-culture of different cell types that bear the genetic makeup of the donor64. Since the gut encompasses
many cell types, and the effects of gene regulation are known to be cell-type- and context-specific1, the CeD model on a chip should ideally contain most of the molecular culprits (Fig.
1). These include an epithelial layer, along with the main immune cell types (e.g. CD4+ T cells,
IE-CTLs and B cells); environmental factors including gluten peptides and cytokines (e.g. IL-15, IFNb, IL-21 and IFNg, alone or in combination); and the microbiome (e.g. Proteobacteria8)
or microbial products (e.g. short-chain fatty acids such as propionate and butyrate) that have been shown to influence immune-cell activation and disease outcome9. The microbiome can
even be added without the risk of bacterial overgrowth because of the microfluidic system that is coupled to the chip66.
Once the model is established, the integrity of TJs can be measured more accurately than in cell culture systems through the incorporation of electrodes and by testing cell–cell adhesion structures with fluorescent dyes64. Moreover, the subsequent application of modern gene
editing (e.g. CRISPRi) and omics approaches (e.g. proteomics, scRNA-seq) will allow us to more efficiently assess the impact of targeting individual or multiple candidate genes on the overall pattern of gene expression in a more physiological culture system.
Together, these novel techniques and disease-relevant models will hopefully reveal a more complete picture of the CeD puzzle, leading to a better understanding of disease pathology and eventually to clues that can lead us towards the development of new treatments to reduce the abnormal inflammation seen in CeD.
The current CeD puzzle
By piecing together all the information we have gathered during my PhD we propose a model that may serve as a starting point for future research. In this model prominent roles are found for interferong signaling and MHC-I antigen processing/presentation (driven by TRAFD1), T cell costimulation/activation, cytokine and chemokine signaling (represented by candidate
genes such as CCR1, CCR2, IL21), lncRNAs, epithelial cells, adaptive (gluten-specific CD4+
T cells, IE-CTLs) innate immune cells (monocytes, dendritic cells, neutrophils and NK) and environmental signals such as tissue alarmins (IL-15 and IFNb) (Fig. 2). Together these biological pathways and cell types may interact to activate and perpetuate the intestinal inflammatory cascade in CeD.
Ch
ap
te
r 7
References
1. Withoff, S., Li, Y., Jonkers, I. & Wijmenga, C. Understanding Celiac Disease by Genomics. Trends Genet.32, 295-308 (2016).
2. Kumar, V. et al. Systematic annotation of celiac disease loci refines pathological path-ways and suggests a genetic explanation for increased interferon-gamma levels. Hum. Mol. Genet. 24, 397–409 (2015).
3. Cortes, A. & Brown, M. A. Promise and pitfalls of the Immunochip. Arthritis Res. Ther. 134, 101 (2011).
4. Wapenaar, M. C. et al. The interferon gamma gene in celiac disease: Augmented expression correlates with tissue damage but no evidence for genetic susceptibility. J. Autoimmun. 23, 183-90 (2004).
5. Ricaño-Ponce, I. et al. Immunochip me-ta-analysis in European and Argentinian populations identifies two novel genetic loci associated with celiac disease. Eur. J. Hum. Genet. (2019).
6. Marasco, G. et al. Gut Microbiota and Celiac Disease. Dig. Dis. Sci. 61, 1461–1472 (2016). 7. Girbovan, A., Sur, G., Samasca, G. & Lupan,
I. Dysbiosis a risk factor for celiac disease. Med. Microbiol. Immunol. 206,83-91 (2017). 8. Krishnareddy, S. The Microbiome in Celiac
Disease. Gastroenterol. Clin. North Am. 48, 115-126 (2019).
9. Ratajczak, W. et al. Immunomodulatory poten-tial of gut microbiome-derived short-chain fatty acids (SCFAs). Acta biochim. Pol. 66,1-12 (2019).
10. Mu, Q., Kirby, J., Reilly, C. M. & Luo, X. M. Leaky gut as a danger signal for autoimmune diseases. Front. Immunol. 23 (2017). 11. Monsuur, A. J. et al. Myosin IXB variant
increases the risk of celiac disease and points toward a primary intestinal barrier defect. Nat. Genet. 37, 1341–1344 (2005).
12. Lammers, K. M. et al. Identification of a novel immunomodulatory gliadin peptide that causes interleukin-8 release in a chemokine receptor CXCR3-dependent manner only in patients with coeliac disease. Immunology. 132, 432-40 (2011).
13. Van Elburg, R. M., Uil, J. J., Mulder, C. J. J. & Heymans, H. S. A. Intestinal permeability in patients with coeliac disease and relatives of patients with coeliac disease. Gut. 34, 354-7 (1993).
14. Nilsen, E. M. et al. Gluten specific, HLA-DQ restricted T cells from coeliac mucosa produce cytokines with Th1 or Th0 profile dominated by interferon γ. Gut. 37,766-76 (1995). 15. Jabri, B. & Sollid, L. M. T Cells in Celiac
Dis-ease. J. Immunol. 198, 3005–3014 (2017).
16. 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).
17. Freire, R. et al. Human gut derived-organoids provide model to study gluten response and effects of microbiota-derived molecules in celiac disease. Sci. Rep.9, 7029 (2019). 18. Vervenne, H. B. V. K. et al. Targeted disruption
of the mouse Lipoma Preferred Partner gene. Biochem. Biophys. Res. Commun. 379,368-73 (2009).
19. Hilke B.V.K. Vervenne a, et al. Lpp is involved in Wnt/PCP signaling and acts together with Scrib to mediate convergence and extension movements during zebrafish gastrulation. Dev. Biol. 320, 267–277 (2007).
20. Schumann, M. et al. Cell polarity-determining proteins Par-3 and PP-1 are involved in epi-thelial tight junction defects in coeliac disease. Gut. 61, 220-8 (2012).
21. Schumann, M., Siegmund, B., Schulzke, J. D. & Fromm, M. Celiac Disease: Role of the Epithelial Barrier. Cell. Mol. Gastroenterol. Hepatol. 3, 150-162 (2017).
22. Petit, M. M. R. et al. LPP, an actin cytoskele-ton protein related to zyxin, harbors a nuclear export signal and transcriptional activation capacity. Mol. Biol. Cell. 11. 117-29 (2000). 23. Kontakou, M. et al. Detection of interferon
gamma mRNA in the mucosa of patients with coeliac disease by in situ hybridisation. Gut. 35, 1037-41 (1994).
24. Yu, M. et al. Aryl hydrocarbon receptor acti-vation modulates intestinal epithelial barrier function by maintaining tight junction integrity. Int. J. Biol. Sci. 14, 69-77 (2018).
25. Bardenbacher, M. et al. Permeability analyses and three dimensional imaging of interferon gamma-induced barrier disintegration in intes-tinal organoids. Stem Cell Res. 35: 101383 (2019).
26. Dong, Y., Sun, Q. & Zhang, X. PD-1 and its ligands are important immune checkpoints in cancer. Oncotarget. 8, 2171-2186 (2017). 27. Kwok, G., Yau, T. C. C., Chiu, J. W., Tse, E.
& Kwong, Y. L. Pembrolizumab (Keytruda). Hum. Vacc. and Immunother. 12, 2777-2789 (2016).
28. Candelaria Ponce de León, Miguel An-gel López-Casado, Pedro Lorite, Teresa Palomeque, M. I. T. Dysregulation of the PD-1/PD-L1 pathway contributes to the patho-genesis of celiac disease. Cell. Mol. Immunol. Vol. 16, 777–779 (2019).
29. Di Sabatino, A. et al. Distribution, proliferation, and function of paneth cells in uncomplicated and complicated adult celiac disease. Am. J. Clin. Pathol. 130, 34-42 (2008).
release of antimicrobial products is directly controlled by immune cell-derived IFN-γ. J. Exp. Med. 211, 1393-405 (2014).
31. Jin, Y. et al. Variant of TYR and autoimmunity susceptibility loci in generalized vitiligo. N. Engl. J. Med. 362, 1686–97 (2010).
32. Coenen, M. J. H. et al. Common and different genetic background for rheumatoid arthritis and coeliac disease. Hum. Mol. Genet. 18, 4195-203 (2009).
33. Yang, L. et al. Interferon-gamma inhibits melanogenesis and induces apoptosis in melanocytes: A pivotal role of CD8+ cytotoxic T lymphocytes in vitiligo. Acta Derm. Venereol. 95, 664-70 (2015).
34. Reyes-Pérez IV, Sánchez-Hernández PE, Muñoz-Valle JF, Martínez-Bonilla GE, García-Iglesias T, González-Díaz V, García-Arellano S, Cerpa-Cruz S, Polan-co-Cruz J, R.-D. M. Cytokines (IL-15, IL-21, and IFN-γ) in rheumatoid arthritis: association with positivity to autoantibodies (RF, anti-CCP, anti-MCV, and anti-PADI4) and clinical activity. Clin Rheumatol. 38, 3061-3071 (2019). 35. Sorini, C. et al. Loss of gut barrier integrity
triggers activation of islet-reactive T cells and autoimmune diabetes. Proc. Natl. Acad. Sci. 116, 15140-15149 (2019).
36. Zhernakova, D. V. et al. Identification of con-text-dependent expression quantitative trait loci in whole blood. Nat. Genet. 49, 139-145 (2017).
37. Ford, R. M. C. and M. L. Challenges and Opportunities in Targeting the CD28/CTLA-4 Pathway in Transplantation and Autoimmunity. Expert Opin Biol Ther 17, 1001–1012 (2017). 38. Garrote, J. A. et al. Cytokine, chemokine and
immune activation pathway profiles in celiac disease: An immune system activity screen-ing by expression macroarrays. Drug Target Insights. 3, 1-11 (2008).
39. Jabri, B. & Sollid, L. M. T Cells in Celiac Dis-ease. J. Immunol. 198, 3005-3014 (2017). 40. Høydahl, L. S. et al. Plasma Cells Are the
Most Abundant Gluten Peptide MHC-express-ing Cells in Inflamed Intestinal Tissues From Patients With Celiac Disease. Gastroenterolo-gy. 156, 1428-1439 (2019).
41. León, A. J. et al. Interleukin 18 maintains a long-standing inflammation in coeliac disease patients. Clin. Exp. Immunol. (2006).
42. Esche, C., Stellato, C. & Beck, L. A. Chemo-kines: Key players in innate and adaptive immunity. J. Invest. Dermatol. 146, 479-85 (2005).
43. Mashima, R. et al. FLN29, a novel interferon- and LPS-inducible gene acting as a negative regulator of toll-like receptor signaling. J. Biol. Chem. 280, 41289-97 (2005).
44. Zhang, N. & Bevan, M. J. CD8+ T Cells: Foot Soldiers of the Immune System. Immunity, 35,
161-8 (2011).
45. Fan, X. & Rudensky, A. Y. Hallmarks of Tis-sue-Resident Lymphocytes. Cell, 164, 1198-1211 (2016).
46. Kulling, P. M. et al. Dysregulation of the IFN-γ-STAT1 signaling pathway in a cell line model of large granular lymphocyte leukemia. PLoS One. 13, e0193429 (2018).
47. Dickson, B. C., Streutker, C. J. & Chetty, R. Coeliac disease: An update for pathologists. J. Clin. Pathol. 59,1008-16 (2006).
48. Pugliese, A. Autoreactive T cells in type 1 dia-betes. J. Clin. Invest. 127, 2881-2891 (2017). 49. Jabri, B. & Sollid, L. M. Tissue-mediated
con-trol of immunopathology in coeliac disease. Nat. Rev. ImmunoL. 9, 858-70 (2009). 50. Hoytema van Konijnenburg, D. P. & Mucida,
D. Intraepithelial lymphocytes. Curr. Biol. 27, R737-R739 (2017).
51. Jabri, B. et al. Selective expansion of intraepithelial lymphocytes expressing the HLA-E- specific natural killer receptor CD94 in celiac disease. Gastroenterology. 118, 867-79 (2000).
52. Di Sabatino, A. et al. Evidence for the Role of Interferon-alfa Production by Dendritic Cells in the Th1 Response in Celiac Disease. Gastro-enterology. 133, 1175-87 (2007).
53. Fina, D. et al. Interleukin 21 contributes to the mucosal T helper cell type 1 response in coeliac disease. Gut. 57, 887-92 (2008). 54. Mention, J. J. et al. Interleukin 15: A key to
disrupted intraepithelial lymphocyte homeo-stasis and lymphomagenesis in celiac dis-ease. Gastroenterology. 125, 730-45 (2003). 55. Jabri, B. & Abadie, V. IL-15 functions as a
danger signal to regulate tissue-resident T cells and tissue destruction. Nat. Rev. Immu-nol, 15, 771-83 (2015).
56. Singh, A., Zaeri, N. & Ho, I. K. Onset of Celiac Disease after Treatment of Chronic Hepatitis C with Interferon Based Triple Therapy. Case Reports Hepatol. 763491 (2015).
57. Seo, W. & Taniuchi, I. Regulation of hema-topoiesis and immune responses by long non-coding RNAs. Int. Immunol. 29, 165-172 (2017).
58. Turner, M. D., Nedjai, B., Hurst, T. & Penning-ton, D. J. Cytokines and chemokines: At the crossroads of cell signalling and inflammatory disease. Biochimi Biophys Acta. 1834, 2563-2582 (2014).
59. Deelen, P. et al. Improving the diagnostic yield of exome- sequencing by predicting gene– phenotype associations using large-scale gene expression analysis. Nat. Commun. 10, 2837 (2019).
60. Ranzoni, A. M., Strzelecka, P. M. & Cvejic, A. Application of single-cell RNA sequencing
Ch
ap
te
r 7
methodologies in understanding haematopoi-esis and immunology. Essays Biochem. 63, 217-225 (2019).
61. Van Arensbergen, J. et al. Genome-wide mapping of autonomous promoter activity in human cells. Nat. Biotechnol. 35, 145-153 (2017).
62. Edwards, S. L., Beesley, J., French, J. D. & Dunning, M. Beyond GWASs: Illuminating the dark road from association to function. Am. J. Hum. Genet. 93, 779-97 (2013).
63. Kampmann, M. CRISPRi and CRISPRa Screens in Mammalian Cells for Precision Biology and Medicine. ACS Chem. Biol. 13,
406-416 (2018).
64. Moerkens, R., Mooiweer, J., Withoff, S. & Wijmenga, C. Celiac disease-on-chip: Mod-eling a multifactorial disease in vitro. United European Gastroenterol J. 7, 467-476 (2019). 65. Kim, H. J. & Ingber, D. E. Gut-on-a-Chip
mi-croenvironment induces human intestinal cells to undergo villus differentiation. Integr. Biol. (Camb). 5, 1130-40 (2013).
66. Kim, H. J., Huh, D., Hamilton, G. & Ingber, D. E. Human gut-on-a-chip inhabited by microbial flora that experiences intestinal peristalsis-like motions and flow. Lab Chip. 12, 2165-74 (2012).
