Celiac disease-on-chip: Modeling a multifactorial disease in vitro

Pathogenesis of CeD

Current models for CeD

To date, there is no model system that fully recapitulates the complexity of CeD. Current in vitro models include immortalized cell lines and mucosal biopsies. The immune system has been investigated using cell lines of monocytes, such as THP-1, or intestinally derived T cells.^37,38 Existing data on epithelial barrier function are largely based on intestinal mucosal biopsies or Caco-2, a tetraploid human colonic epithelial cancer cell line. Immortalized cell lines do not represent the genetics of CeD and have poor genomic integrity (Table 1). Patient-derived intestinal biopsy material does contain the CeD-associated genetic background and directly reflects the disease phenotype, but is scarce because of its invasive nature. Biopsies also have limited proliferative capacity, and individual cell types are difficult to study within a heterogeneous biopsy. Conventional systems to measure barrier function and transport, like transwell systems, do not recapitulate the intestinal physiology (e.g. IECs fail to form villus-like structures or produce mucus), and co-cultures with microbial cells are difficult in these static systems because of rapid overgrowth and contamination. ³⁹ Studying CeD in vivo is dependent on humanized mouse models that express human HLA-DQ8 or HLA-DQ2.^40–42 These models have shown that the presence of gluten-specific CD4 + T cells is not sufficient to induce CeD-like pathogenesis and that triggers of the innate immune system, particularly IL-15 overexpression, are essential for inducing intestinal damage upon gluten exposure. However, mice are not ideal models because of differences in intestinal tract physiology, ⁴³ immune system ⁴⁴ and microbiome composition. ⁴⁵ To further elucidate the mechanisms underlying CeD, it is essential to capture the entire pathobiology of CeD using multicellular and human-based models.

OPEN IN VIEWER

Table 1.

Possible biological systems for modeling complex diseases: advantages and disadvantages.

Biological system	Critical factors for modeling complex diseases					Other advantages (+) and disadvantages (–)
	Patient genetics	Availability of material	Genetic engineering	Heterogeneity/ complexity	Can be combined with other cell types from same donor a
Immortalized cell lines	No	High	Established	Single cell type	No	+ Potentially easier to handle − Poor genomic integrity
Intestinal biopsy	Yes	Low b	Difficult	Multiple cell types	Yes	+ Disease phenotype − Inflamed tissue − Limited lifespan
Humanized mouse models	No	High	Established	Multiple cell types	Yes	+ Whole organism: presence of hormone, neurologic, and metabolic signals from other organs or cell systems − Requires thorough understanding of induction of disease − Difficult to translate to humans because of interspecies differences in physiology, pharmacology and cellular processes
Induced pluripotent stem cells	Yes	High	Established	Single cell type	Yes	+ Suitable for genotype selection (patient and control cases) − Requires knowledge of differentiation to relevant tissue

hiPSC and organoid technology

hiPSCs can be generated from different types of somatic cells taken from any donor. hiPSCs can divide indefinitely and have the potential to differentiate into any of the cell types found in the human body. In 2006, Yamanaka and colleagues demonstrated for the first time that human and mouse fibroblasts could be reprogrammed to a pluripotent state, resembling embryonic cells in culture.^46,47 Pluripotency was achieved by viral overexpression of only four transcription factors: Oct4, Sox2, Klf4 and c-Myc. With the development of improved protocols, hiPSC lines can now be efficiently generated from urine-derived epithelial cells and blood-derived erythroblasts, among others. ⁴⁸

Using knowledge on embryonic development, hiPSCs can be differentiated into human intestinal organoids (HIOs): miniature parts of the gut that are cultured in a dish. The first HIOs were grown from intestinal crypts derived from human biopsy material. ⁴⁹ When cultured in an extracellular matrix (ECM) gel in the presence of specific growth factors, it is possible to maintain the stem cell niche and the proliferative and differentiation capacity of crypt cells in vitro, allowing them to grow out into complex three-dimensional (3D) “budding” structures. These structures contain multiple functional IEC subtypes that can be kept in culture for prolonged periods of time. ⁵⁰ The generation of HIOs from hiPSCs is more complex and leads to a less mature differentiated phenotype. ⁵¹ However, embryonal development of intestinal tissue can be mimicked by exposing hiPSCs to a series of specific growth factors in a strict time-dependent manner. ⁵²

The HIO system still has limitations when it comes to studying multifactorial diseases.^53,54 HIOs are inconsistent in size and shape and are cultured in a static system (embedded in extracellular matrix) that does not recapitulate the intestine's physical environment (including fluid flow and peristaltic movement). The closed configuration of HIOs renders them less ideal for studying transport over the intestinal barrier or interactions with commensal microbes or pathogens (Figure 2). Apical access can be achieved by microinjection, ⁵⁵ but this technique is labor intensive and technically challenging. The wide range of organoid sizes complicates this procedure even more and makes it nearly impossible to standardize the cell:stimulus ratio. Additionally, dead cells accumulate in the enclosed lumen of the HIO, ultimately impairing the viability of the system. Lastly, physiological interactions with other components of the intestine (e.g. immune and vascular system) are difficult to emulate within the extracellular matrix, while the matrix is necessary to generate and maintain HIOs. These limitations can be overcome by an organ-on-chip system. OPEN IN VIEWER

Intestine-on-chip

Organ-on-chip systems are microfluidic devices in which cells are cultured in continuously perfused microchannels engineered to mimic the physical microenvironment of tissues and organs. ⁵³ A current model makes use of a chip containing two parallel hollow channels approximately 1 mm wide separated by a porous ECM-coated membrane ⁵⁶ (Figure 3). In this device, a monolayer of IECs can be grown on the upper surface of the membrane separating both channels, while endothelial cells can be grown on the other side, representing blood vessels. The culture media for the cells is delivered via the upper and lower channel, which can also be used to introduce metabolites, cytokines, microbial cells and/or immune cells into the system. The system also provides mechanical forces to simulate the physical microenvironment of the intestine through fluid flow that introduces shear stress on the cells and two vacuum compartments on the sides that create a peristalsis-like motion. Remarkably, these mechanical forces induce epithelial cells to spontaneously form polarized 3D villus-like structures that contain cells expressing markers characteristic of differentiated IECs (i.e. adsorptive enterocytes, mucus-producing goblet, Paneth and enteroendocrine cells).^39,57–59 The resulting epithelial layer exhibits basic functional properties, such as mucus production, high barrier resistance, activity of brush border and drug-metabolizing enzymes, and high efficiency in nutrient uptake because of the increased intestinal surface. These characteristics allow for studies focusing on digestion and nutrient uptake, barrier integrity and drug metabolism,^39,57,58 and for co-cultures with commensal microbial cells for extended periods of time (up to weeks).^39,60 OPEN IN VIEWER

In accordance with the morphological changes, the transcriptional profile of epithelial cells cultured in the dynamic chip system is very different from that of cells cultured in static Transwell systems or compared with HIO. In fact, the intestine-on-chip profile most resembles the profile of the corresponding in vivo intestinal segment.^58,60

The material most often used for chip fabrication, polydimethylsiloxane, is fully transparent, making the chip readily amenable to microscopy. For research purposes, sophisticated intestine-on-chip systems can be engineered to contain sensors, for example to measure TEER. ⁶¹ Integrated sensors are a major step forward because they allow for continuous monitoring of the system, something that is very difficult and laborious in conventional culture systems.

hiPSCs, HIOs and intestine-on-chip to model CeD

In contrast to monogenic diseases in which a single gene is involved, genetic modeling of complex diseases like CeD requires the inclusion of the many disease-associated genetic risk factors that need to be studied in the disease-relevant cell or tissue. ¹⁹ Combining hiPSC and HIO technology, in vitro models of the intestine can be created from cells that contain the spectrum of CeD-associated genetic risk factors (Table 1). Because hiPSC lines can be generated from relatively easily accessible somatic cells such as urine-derived epithelial cells, skin-derived fibroblasts or blood-derived erythroblasts, ⁴⁸ there is no dependency on intestinal biopsy material obtained by invasive endoscopic procedures (in the case of CeD). This facilitates the collection of starting material from both patients and healthy individuals. Varied genetic backgrounds can then be studied to contrast the disease genetic background with low risk backgrounds (Figure 4). To study specific elements of the disease process, like barrier function, genetic engineering can be used to perturb the system by creating extreme genotypes (i.e. gene knock-out by CRISPR/Cas9 technology). These technologies could be used to generate isogenic hiPSC lines that contain identical genetic background, except for, for instance, one repaired CeD-associated genetic risk factor. Such isogenic lines may reveal the functional consequences of a single genetic variation associated with CeD. Using hiPSCs as a starting point, the effect of a disease-associated genotype can be evaluated in multiple disease-relevant cell types, either individually or in combination, in an intestine-on-chip. This model is unique because it integrates (1) the CeD-associated genetic background, (2) the interaction between disease-relevant cell types, (3) any relevant environmental stimuli and (4) the physical microenvironment of the intestine in a complex yet controllable manner. Very recently, proof-of-concept was provided for an hiPSC-derived intestinal epithelial-layer-on-chip. ⁵⁹ This system now needs to be adapted toward a more CeD-relevant model that includes hiPSC-derived endothelial ⁶² and immune cells.^63–65 OPEN IN VIEWER

Improved understanding of CeD etiology

A CeD intestine-on-chip model can help address significant questions. It will allow the investigation of the interaction between IELs and IECs in the presence or absence of triggering environmental factors. In particular, the IL-15 expression by IECs implicated in activation of IELs ¹⁴ can be monitored in response to these different stimuli. A possible primary defect in intestinal barrier function, which in turn alters gluten transport, can be addressed using different assays in a simple system in which iPSC-derived IECs are present outside the immune context (Figure 5(a)). With this system, genes involved in the process can be identified. The role of the gut microbiome in CeD pathogenesis can also be studied. One can envision that the microbiota affects barrier function, but also that CeD-associated genetics affect microbiome homeostasis by altering the immune response. Finally, the effect of different environmental factors can be studied by introducing them into the system, for instance introducing viral ligands, metabolites produced by CeD-associated microbiota, or ATIs. The complexity of the system can be also adjusted to fit the research question, ranging from one cell type to more complex systems (Figure 5(b)). OPEN IN VIEWER

Development and testing of novel treatments

A lifelong adherence to a gluten-free diet has a profound impact on everyday life, which makes treatments to inhibit the strong pro-inflammatory immune response to gluten very valuable. A physiologically relevant CeD intestine-on-chip model can be used to test novel drug candidates and existing drugs for repositioning. By using patient-derived hiPSCs, differences in genetic background that may affect drug efficacy can be taken into account. To be used for drug screening and/or addressing pharmacogenetic questions, high-throughput systems should be developed, as current devices are still low throughput and costly. ⁶⁶ Nevertheless, an intestine-on-chip has great potential for personalized medicine, providing a model that can include an individual's genetic background, relevant cell types and environmental triggers.

Toward a patient-on-chip

Although CeD is regarded as a disease of the intestine, the disease presents systemically. ⁶ To capture the extraintestinal phenotypes, different organ-on-chip systems could be coupled in the future. In the context of CeD, a first expansion might be to couple an intestine-on-chip to a brain-on-chip. The intestine-brain-axis is of particular interest because the clinical spectrum of CeD includes behavioral changes such as anxiety, depression and fatigue.^67,68 The mechanism underlying this “cross-talk” between intestine and brain is poorly understood, but proposed explanations include the interaction of gluten peptides with endorphin receptors in the brain, the migration of activated immune cells to the brain ⁶⁹ and detrimental effects of circulating microbial metabolites^70,71—all processes that could be tested by linking organ-on-chip systems.