Organoid Models of Human Gastrointestinal Development and Disease

Abstract

We have greatly advanced our ability to grow a diverse range of tissue-derived and pluripotent stem cell-derived gastrointestinal (GI) tissues in vitro. These systems, broadly referred to as organoids, have allowed the field to move away from the often non-physiologic, transformed cell lines that have been used for decades in GI research. Organoids are derived from primary tissues and have the capacity for long-term growth. They contain varying levels of cellular complexity and physiologic similarity to native organ systems. We review the latest discoveries from studies of tissue-derived and pluripotent stem cell-derived intestinal, gastric, esophageal, liver, and pancreatic organoids. These studies have provided important insights into GI development, tissue homeostasis, and disease and might be used to develop personalized medicines.

The human gastrointestinal (GI) tract comprises this foregut, midgut, and hindgut—each region gives rise to different tissues and organs ^1–4. The foregut endoderm gives rise to the epithelium of the oral cavity, pharynx, esophagus, stomach, liver, pancreas, and proximal duodenum. The midgut and hindgut endoderm give rise to the epithelium of the distal duodenum, jejunum, ileum, colon, rectum and anal canal, as well as the epithelial lining of the bladder and urethra ^4–36. In vitro organoid cultures are generated from stem or progenitor cells and have the capabilities for long-term growth, cellular diversity, function, and spatial organization specific to the organ they represent and they have been used to study endoderm-derived GI organs ^37–54. Although they were initially developed as a way to culture primary mouse GI tissues, organoid technologies now enable the growth of diverse primary human tissues in vitro ^{43, 44, 47, 50, 52, 55–60}.

Before the development of human GI organoids, the GI tract was modelled using cell lines and explanted tissue, both of which have significant limitations (reviewed in ⁶¹). Cell lines, for example, are often derived from malignant specimens or are immortalized by viral infection, cell fusion, or introduction of oncogenes, which limit their use in developmental, physiologic, and regenerative studies ^62–67. Explant cultures, which have organotypic properties such as complex 3-dimensional (3D) architecture and cellular heterogeneity, have important roles in studies of development and physiology, but are limited by their short-term nature ^{53, 68–75}. Recent advances in maintaining GI tissues in long-term in vitro cultures have allowed researchers to overcome many of these barriers, invigorating basic and translational research ^{37–52, 54}. For example, long-term organoid cultures have been used to examine the stem cell niche, cellular differentiation, interactions between cells, and physiologic functions ^{37, 38, 40–42, 44, 46–51, 54, 57, 58, 60, 76–85}. They have also been used to model malignancy, infection, and inflammation, as well as in toxicology studies.

Many research articles and reviews have highlighted the use of organoid systems to characterize the adult stem cell niche and gut morphogenesis, and will not be covered in detail in this review (see ^{1, 3, 40–43, 56, 58, 59, 70, 86–95}). We aim to define and discuss the variety of long-term in vitro GI models that have been created, and how they have increased our understanding of development, genetic diseases of the GI tract, malignancy, host–pathogen interactions, tissue reprogramming, and regeneration.

Defining Organotypic GI Culture Models

The term organotypic has referred to tissue explanted from an organ that continues to function as it would in vivo ^{71, 96, 97}. However, with recent advances in de novo differentiation of 3D organ-like tissues from human pluripotent stem cells, and in long-term culture of human primary tissues, the umbrella of the term organotypic has expanded ⁹⁸. We use the term organotypic to define short-term explant models and long-term in vitro culture systems that recapitulate the complex cellular diversity and function of the native organ. Long-term organotypic culture systems include primary (tissue-derived) epithelium-only organoids ^{37, 38, 40–42, 44, 45, 49, 51, 54, 60, 70}, tissue-derived epithelial–mesenchymal organoids ^{46, 48, 70, 99}, and pluripotent stem cell (PSC)-derived organoids ^{47, 50, 80, 100, 101} (see Figure 1).

Figure 1. Composition, Cellular Sources and Nomenclature for GI Organotypic Culture Systems.

GI organotypic cultures consist of short-term explant cultures and long-term organoid cultures. The 3 types of GI organoid cultures include 1) tissue-derived epithelial, 2) tissue-derived epithelial–mesenchymal, and 3) PSC-derived organoids. Tissue-derived epithelium and PSC-derived intestinal organoids are shown as examples of each type of tissue.

The ability to grow long-term intestinal organotypic cultures in vitro was made possible through identification of conditions that mimic the intestinal stem cell niche ^{48, 49}. In 2009, Sato et al isolated intestinal crypts from mice that gave rise to epithelium-only organoids when cultured with laminin-rich Matrigel and growth factors, which reflected the intestinal niche in vivo ⁴⁹. These epithelial organoids could be generated from a single intestinal stem cell, identified by LGR5 expression ⁴⁹. Using another strategy, Ootani et al generated epithelium–mesenchyme organoids by culturing minced fragments of small intestine or colon in a collagen matrix exposed to an air–liquid interface. These fragments gave rise to highly proliferative epithelium, with supportive myofibroblasts located along the basolateral aspect of the epithelial cells. The resulting epithelial–mesenchymal organoids could be maintained in vitro for up to 1 year ⁴⁸. For long-term growth, intestinal organoids, in both studies, required media containing growth factors that stimulate Wnt signaling; this recapitulates in vivo environment of the intestinal crypt, which contains high levels of Wnt signaling.

These achievements laid the groundwork for studies describing differentiation and growth of human embryonic and induced PSC (ESC and iPSC)-derived organoids ^{47, 50, 80, 100, 101}. In these studies, ESCs or iPSCs were differentiated into intestine or gastric tissue using step-wise differentiation that mimicked stages of embryonic development (reviewed in ^{1, 3}). This culture system, similar to culture of tissue-derived epithelial and epithelial-mesenchymal organoids, required a 3D extracellular matrix with media that supported high levels of Wnt signaling.

Similar 3D cultures, in environments with high levels of Wnt signaling, have more recently enabled the culture of primary human esophageal, gastric, liver, pancreatic, and colon organoids (see Figure 2) ^{37, 38, 44, 60}. We discuss briefly advances in pancreas and liver organoids, but will focus on the implantation of PSC-derived and tissue-derived epithelial esophageal, gastric, and intestinal organoids for studies of development, disease, and regenerative medicine.

Figure 2. GI organ diversity represented in organoid systems.

GI organoids have been generated from esophagus, liver, biliary tree, stomach, pancreas, small intestine, and large intestine. (Note: GI tract schematic modified from reference 4).

Pancreas and Liver

The distal foregut gives rise to organ buds that form the liver and the dorsal and ventral pancreas. During development, the hepatoblast gives rise to ductal epithelium and hepatocytes (reviewed in ¹⁰²). There is controversy over the process, but in adults, hepatocytes mediate regeneration after liver damage, and a bi-potent progenitor in the ductal epithelium^{89, 103} has been reported to give rise to hepatocytes and ducts. Interestingly, this bi-potent progenitor also generates epithelial liver organoids in mice.^{42, 43, 89} Culture of these cells with hepatocyte growth factor, epidermal growth factor (EGF), R-spondin 1, and fibroblast growth factor 10 produces liver organoids that express duct markers. However, inhibition of Notch and transforming growth factor-β (TGFB) promoted hepatocyte differentiation ⁴². Liver organoids expressing ductal markers have not demonstrated cholangiocyte function.

In addition to tissue-derived liver organoids, hPSC-derived live buds have recently been described. These were generated by combining human iPSC-derived hepatoblasts ¹⁰⁴ with human umbilical endothelial vein cells and human mesenchymal stem cells to develop self-organizing 3D tissues ⁵². Transplantation of the liver buds into mice resulted in vascularization and maturation of the tissue. Furthermore, transplanted liver buds were capable of rescuing drug-induced liver failure. Similar to tissue-derived epithelial liver organoids, however, these liver buds did not generate functional ductal epithelium. More recently, ESCs and iPSCs have been used to generate functional ductal epithelium characterized by bile acid transfer, cystic fibrosis transmembrane conductance regulator (CFTR)-mediated fluid secretion, and responses to the hormones, secretin, and somatostatin ^{58, 59}. It will be interesting to determine if inclusion of PSC-derived cholangiocyte organoids in PSC-derived liver buds results in organization of functional hepatocyte and ductal structures.

Similar to liver epithelial organoids, human pancreatic epithelial organoids can be generated from embryonic progenitors or adult progenitors that arise after injury ⁴¹. Pancreatic duct ligation results in Wnt activity and the development of LGR5+ progenitors in the pancreatic ductal epithelium, which can generate mouse pancreatic epithelial organoids. In vitro and upon transplantation after reaggregation with embryonic pancreas, pancreatic organoids primarily expressed ductal markers. However, 5% of re-aggregated transplanted organoids gave rise to monohormonal insulin-, glucagon-, and somatostatin-producing endocrine lineages. On the other hand, mouse pancreatic epithelial organoids generated from mouse embryonic tissue can give rise to acinar and endocrine structures in vitro ⁴⁰

These findings provide evidence for the limited differentiation potential of organoids generated from adult pancreatic progenitors, compared to organoids derived from embryonic pancreatic progenitors. Human epithelial pancreatic organoids have also been generated from pancreatic ductal cells. Although organoids from adult progenitors are important in the study of pancreatic ductal adenocarcinoma, these organoids require ectopic expression of PDX1, MAFA, and neurogenin-3 (NEUROG3 or PMN) for differentiation into insulin-producing cells. Further studies are required to elucidate conditions for endocrine differentiation of human pancreatic organoids in the absence of genetic modification ^{56, 105}.

Development

Although mouse models and genetic analyses have greatly increased our understanding of embryonic GI development, mouse development does not always recapitulate human development, particularly in development of the intestinal stem cell niche ^106–111. For example, Paneth cells, which contribute to maintenance of the adult stem cell niche ¹¹², develop in the first trimester of human gestation. However, Paneth cells do not differentiate until post-natal day 14 in mice ^{106, 107}. Furthermore, EGF increases intestinal epithelial maturation in mice but decreases maturation of human fetal intestine in vitro ^108–111. GI organoids generated from human specimens are therefore a valuable adjunct to mouse developmental studies, and can be used to study developmental events that are difficult to study in vivo.

Small intestine epithelial organoids generated from fetal mice have been used to study the fetal intestinal niche ^{57, 81}. In contrast to adult epithelial organoids, which demonstrate budding morphology, fetal epithelial organoids frequently formed spherical organoids or spheroids. This tendency to form spheroids, however, decreased with age such that intestine from post-natal day 15 mice did not generate spheroids ⁸¹. Interestingly, mRNA expression of several important intestinal genes differed markedly between fetal spherical and budding organoids, such that fetal budding organoids more closely resembled adult budding organoids ⁸¹.

In a separate study, exogenous Wnt ligand increased maturation of fetal spheroids into budding organoids ⁵⁷. Epithelial spheroids generated from human fetal small intestine and PSC-derived human intestinal organoids were more closely related to immature mouse fetal spheroids than to adult tissue ⁵⁷. These studies exploited the fetal state captured by the intestinal organoid culture system to demonstrate that Wnt signaling stimulates a transition of the fetal organoids grown in vitro from spheroid to budding structure. Furthermore, mouse fetal spheroids that readily engrafted into areas of colitis, induced by dextran sodium sulfate (DSS), appeared to adopt a colonic fate, indicated by expression of the colonic enzyme carbonic anhydrase-II and lack of expression of small intestine enzymes such as lysozyme and alkaline-phosphatase ⁵⁷. These data indicate that mouse fetal small intestine spheroids, in contrast to adult organoids ¹¹³, appear to retain plasticity reflecting their immature status. It is unclear if human fetal intestinal spheroids possess similar cellular plasticity.

In addition to using primary human fetal tissue to study human developmental processes, we can use human PSC-derived tissues to study these processes, as they closely resemble fetal tissue. However, human PSC-derived tissues cannot be used to study physiology and function of adult tissues. For example, hepatocytes or liver organoids generated from PSCs were more closely related to early fetal liver rather than adult liver, based on expression of important regulators of liver development ^{52, 104, 114}. Similarly, the maturational status of early iterations of PSC-derived insulin producing cells represented fetal pancreatic β cells, based on their transcriptional profile and lack of response to glucose ¹¹⁵.

The transcriptome of PSC-derived gastric and lung organoids more closely resembles that of fetal compared to adult lung ^{47, 116}. In addition, RNA sequence-based transcriptional profiling confirmed that human PSC-derived small intestine organoids are most similar to fetal tissue ⁸². This study demonstrated that features of human fetal to adult maturation included increased expression of the intestinal stem cell marker OLFM4, as well as increased Paneth cell differentiation and expression of brush border enzymes on enterocytes. Transplantation of PSC-derived organoids into immunocompromised mice resulted in the maturation of these fetal-like tissues into tissues resembling those of adults. Their intestinal architecture included prominent crypt-villus structures; increased expression of OLFM4; brush border enzymes; and other markers of mature small intestine such as dipeptidyl peptidase 4, glucose transporter type 2, and villin^{82, 117}.

Given the diverse array of in vitro grown human tissues (primary or PSC-derived, fetal or adult, epithelium only or epithelium plus mesenchyme), it will be important to consider the experimental strengths and limitations of each system when designing experiments and implementing such systems. For example, tissue-derived epithelial intestinal organoids can be generated in less than 2 weeks, whereas PSC-derived intestinal organoids require 5 weeks. Fetal epithelial spheroids generated from early intestine progress into budding organoids with low efficiency, whereas fetal epithelial spheroids from perinatal tissue progress to budding organoids with high efficiency ^{57, 81}. Furthermore, mouse epithelial organoids require induction of inflammation for transplantation, whereas human epithelial organoids have n successfully transplanted. Fetal epithelial spheroids might therefore be best suited for in vitro physiologic studies that characterize the early fetal intestine. In contrast, PSC-derived intestinal organoids undergo maturation with high efficiency when transplanted in vivo, and are therefore amenable to in vivo studies of human development.

Models of GI Disease

GI organoids have been used to model human genetic diseases, infections, inflammatory bowel diseases (IBD), and malignancies. Because of our ability to grow robust and diverse patient-derived organoids, we can use these tissues to conduct drug screens and identify personalized approaches for treating disease.

Monogenic diseases that affect the epithelium

Epithelial organoids have been used to assess function of the CFTR—mutations in this transporter can cause CF ^{78, 118}. Although CF is most commonly associated with the pulmonary epithelium, CFTR mutations result in accumulation of thick secretions in the pulmonary and GI tracts. Moreover, simple rectal biopsies are relatively non-invasive compared to bronchoscopy or endoscopy, and can be conducted at a routine doctors visit. Dekkers et al identified forskolin-induced swelling of normal epithelial intestinal organoids as a physiologic function of non-mutant CFTR. Strikingly, epithelial organoids generated primarily from patients with CF did not undergo forskolin-induced swelling ⁷⁸. However, forskolin-induced swelling was restored in patient-derived organoids upon incubation with pharmacologic agents that improved function of the CFTR. Furthermore, correction of CFTR mutations in epithelial organoids from the small intestine of patients with CF, using the CRISPR/Cas9 genome editing system, reduced forskolin-induced swelling ¹¹⁸. Epithelial organoid cultures can therefore model disease physiology, and genome editing can be used to repair genetic mutations in organoids.

GI epithelial organoids have also been used to study microvillus inclusion disease (MVID) and multiple intestinal atresia (MIA) ^{119, 120}. MVID is frequently caused by mutations in myosin Vb (MYO5B) that result in inappropriate localization of apical proteins and enterocyte polarization ^{121, 122}. Wiegerinck et al identified a mutation in syntaxin 3 (STX3) by whole-exome sequencing in patient with MVID but no mutation in MYO5B. Duodenal epithelial enteroids from these patients had partial loss of microvilli, inappropriate accumulation of vesicles, and absence of STX3 protein ¹²⁰. Patients with MIA can develop atresia from the stomach to rectum ¹²³. Using a combination of genome-wide linkage analysis and whole-exome sequencing, Bigorge et al and Avitzur et al identified previously uncharacterized mutations in the tetratricopeptide repeat domain–7A (TTC7A) of patients with MIA and combined immunodeficiency. Intestinal epithelial organoids generated from a subset of these patients demonstrated disturbed polarity and disrupted epithelial architecture ¹¹⁹. Epithelial organoids generated from patient specimens can therefore model and increase our understanding of numerous genetic diseases of the GI tract. Epithelial organoids could be of particular use helping us identify the GI component of complex genetic diseases that affect multiple organ systems.

Enteric anendocrinosis is an extremely rare disease caused by mutations in NEUROG3 that is characterized by severe malabsorptive diarrhea and a lack of intestinal enteroendocrine cells ^{124, 125}. PSC-derived intestinal organoids with adenoviral knockdown of NEUROG3 have fewer enteroendocrine cells, similar to patients with anendocrinosis⁵⁰. Researchers recently disrupted NEUROG3 in ESCs¹²⁶, but these cells could still form mature pancreatic endocrine cells. We anticipate the PSC-derived human intestinal organoids will be devoid of enteroendocrine cells, although additional studies are required.

IBD

IBD is characterized by mucosal inflammation of the GI tract that is caused by a combination of genetic susceptibility, immune dysregulation, disruptions in the microbial communities, and environmental factors ¹²⁷. Organoid cultures have not yet been used to directly model IBD, but have been used to study cell death, mucosal integrity, and the effects of inflammatory cytokines ^{118, 128–130}. For example, IBD specimens were found to contain foci of hypoxic inflammation ¹³¹ and levels of hypoxia inducible factor 2α (HIF2A or EPAS) were increased in the intestinal epithelial cells from mice with colitis and patients with IBD ¹²⁹. Disruption of Hif2α in mice reduced features of colitis induced by DSS. Intestinal epithelial cells from the HIF2A-null mice expressed lower levels of inflammatory cytokines, including tumor necrosis factor (TNF), following DSS induction of colitis, compared to mice with that express HIF2A. Similarly, under hypoxic conditions, PSC-derived intestinal organoids increased expression of TNF ¹²⁹.

IBD can also lead to fibrosis in the intestine, and PSC-derived organoids have been used to model a fibrosis-like response to TGFB stimulation that could be reversed by anti-fibrotic drugs ¹³². Abnormal Paneth cell phenotypes are associated with specific susceptibility alleles for Crohn’s disease ¹³³. Administration of interferon-γ (IFNG), an inflammatory cytokine, to cultures of mouse small intestinal organoids caused Paneth cell degranulation, associated with apoptosis ¹³⁴. In the healthy intestine, Paneth cells are quickly replaced. However, continuous exposure to IFNG causes progressive loss of Paneth cells, which could contribute to development of IBD ¹³⁴. Further studies, potentially with epithelial organoids from patients with IBD, are necessary to determine if Paneth cell degranulation is involved in the pathogenesis of IBD.

IBD is associated with increased risk for colorectal cancer. Oxidative stress could be involved in the pathogenesis of IBD and/or IBD-associated cancer. Deletion of the antioxidant selenoprotein P exacerbated DSS-induced colitis and tumorigenesis ¹³⁵. Small intestine epithelial organoids generated from these mice had increased baseline proliferation but decreased survival when exposed to the oxidative stressor hydrogen peroxide ¹³⁵. It will be interesting to determine if this paradoxical relationship also occurs in epithelial organoids generated from patients with IBD susceptibility alleles.

Cell death has been analyzed in mouse small intestine epithelial organoids; findings might be applied studies of IBD ¹²⁸. Future studies of organoid models will facilitate the exploration of individual and combinatorial contributions of Paneth cell and goblet cell dysfunction, the gut microbiome, and cytokine levels to the pathogenesis of IBD.

Interactions between GI tissues and microbes

Many different types of microbes can cause GI diseases or contribute to their development in susceptible individuals. Studies of GI organoids might increase our understanding of susceptibility to and pathogenesis of microbial infections and possible therapies. Gastric organoids are of particular importance because the recapitulate in vivo physiology and respond to infections like in vivo tissues; they might be used to study malignant transformation secondary to infection.

In the stomach, histamine stimulates parietal cells to secrete acid. Gastric epithelial organoids co-cultured with immortalized stomach mesenchymal cells reduced intra-luminal pH in response to histamine; this effect was reversed by the proton pump inhibitor omeprazole¹³⁶. Helicobacter pylori cause most cases of peptic ulcer disease by inducing hypersecretion of acid and have been associated with gastric cancer. Proliferation increases in PSC-derived gastric organoids infected with H pylori but not with variants that lack its virulence factor, CagA⁴⁷.

Injection of human gastric epithelial organoids with H pylori leads to activation of the transcription factor NF-κB, which is activated in gastric cancer cells. This led to increased expression of interleukin-8, an inflammatory cytokine ¹³⁷. A separate study that found activation of NF-κB following injection of H pylori into mouse gastric epithelial organoids; ⁸⁴ it reported upregulation of Sonic hedgehog, a protein involved in vertebrate development and the immune response to H pylori ⁸⁴. In a similar study, Wroblewski et al showed that H pylori infection of mouse gastric epithelial organoids resulted in mislocalization of occludin at the tight junction, previously observed in vivo^{138, 139}. Similar to findings from PSC-derived gastric organoids, this study demonstrated that only H pylori variants that expressed CagA increased proliferation in gastric organoids¹³⁸. Together, these studies have shown that H pylori infection can be modeled gastric organoids and can be used to study gastric transformation.

Intestinal organoids have also been used to study pathogens ^{130, 140–143}. Colonization of mouse small intestine epithelial organoids with Salmonella enterica serovar Typhimurium (referred to hereafter as Salmonella) reduced their growth, disrupted tight junctions, activated inflammatory signaling via NF-κB, and reduced expression of Lgr5¹⁴⁰. This study revealed the complexity of the response to infection, even in the absence of immune cells.

Another study capitalized on the absence of immune cells by studying the effect of α-defensins—antimicrobial peptides produced in response to Salmonella infection ¹⁴¹. Small intestine epithelial organoids made from mice with wild-type or mutated α-defensins were injected with Salmonella. Absence of mature α-defensin reduced intra-luminal bacterial killing, which could be partially restored by expression of human α-defensin. This study demonstrated the contribution of α-defensins to the immune response to Salmonella, which had been a challenge to establish in vivo due to the commensal microbiota and cellular immune response.

Forbester et al demonstrated that injection of Salmonella into the lumen of human PSC-derived organoids resulted in epithelial invasion, but this could be inhibited with injection of a mutant strain of Salmonella with reduced invasive capacity ¹⁴². Researchers have also studied human PSC-derived organoids infected with Clostridium difficile—the leading cause of healthcare-associated diarrhea, on intestinal epithelium ¹³⁰. The obligate anaerobe was able to survive up to 12 hours after intra-luminal injection, indicating low intra-luminal oxygen concentration despite ambient oxygen culture conditions. Furthermore, purified toxin A and toxin B, determinants of C difficile virulence, also disrupted paracellular barrier function whereas a non-toxogenic C difficile strain did not. PSC-derived intestinal organoids can therefore be used to study anaerobic pathogen infections and possible treatments.

GI organoids have been an indispensable tool for studies of factors that contribute to complex and poorly understood diseases such as necrotizing enterocolitis (NEC). NEC, which affects premature, formula-fed infants, results in bowel necrosis. Sodhi et al demonstrated that lack of Toll-like receptor-4 (TLR4) protected neonatal mice from hypoxia-induced NEC ¹⁴⁴. Interestingly, small intestine epithelial organoids generated from these mutant mice had increased in goblet cell differentiation. Administration of the Notch inhibitor, dibenzapine, protected wild-type mice from NEC.

Researchers observed an increase in p53 up-regulated modulator of apoptosis (PUMA), which regulates apoptosis, upon activation of TLR4 ¹⁴⁵. Activation of TLR4 in mouse small intestine organoids also resulted in PUMA upregulation, as well as reduced proliferation and increased apoptosis. However, apoptosis was not observed in organoids that lacked PUMA. Furthermore, PUMA was upregulated in intestinal crypts of wild-type mice and patients with NEC, but not TLR4-null mice. The use of human neonatal epithelial organoids and PSC-derived organoids to study the effects of microbes on the intestinal epithelium in the presence and absence of TLR4 signaling could increase our understanding of the pathogenesis of NEC. Finkbeiner et al demonstrated that clinical rotavirus isolates can replicate within the epithelial and mesenchymal cell compartments of PSC-derived intestinal organoids. Epithelial and PSC-derived GI organoids can therefore be used to study host–pathogen interactions and develop new therapies for GI infections.

In addition to pathogens, GI organoids have been used to understand the influence of indigenous (commensal) microbes on epithelial health. Stimulation of mouse small intestine crypts with muramyl dipeptide, a common bacterial peptidoglycan motif, increased organoid formation¹⁴⁶. This survival advantage was lost with deletion of NOD2, a receptor that directly senses bacterial products. Mice given muramyl dipeptide before intestinal insult with doxorubicin had increased crypt survival, not seen in NOD2-knockout mice. Lukovac et al demonstrated that presence of certain commensal bacteria led to expression of genes involved in lipid metabolism in the intestinal epithelia ¹⁴⁷.

Cancer

GI cancer organoids can model the pathophysiology of GI malignancies and their response to treatment. Colonic epithelial cells accumulate genetic changes that transform them into adenocarcinomas—intestinal organoids are powerful tools for studying this process. In epithelial organoids generated from intestinal adenomas of APC-deficient mice, R-spondin-1 was not required for constitutive Wnt signaling ⁶⁰. Lentiviral knockdown of APC in mouse small intestine epithelial organoids confirmed Wnt activation and perturbed differentiation; these organoids generated subcutaneous tumors when injected into immunocompromised mice ¹⁴⁹. Activation of Kras in these organoids further increased the tumor burden ¹⁴⁹.

APC loss of heterozygosity was modeled using epithelial organoids generated from the small intestine of Apc^Min/+mice, a model of familial adenomatous polyposis ^{150, 151}. These organoids developed a cystic morphology and acquired DNA damage¹⁵⁰, believed to result from increased activity of the repair protein RAD21. Apc^Min/+ organoids with Rad21 haploinsufficiency had reduced loss of heterozygosity at Apc. Interestingly, when these organoids were incubated with TGFB, apoptosis increased, associated with increased levels of BCL2-like 11 (apoptosis facilitator) (BIM) ¹⁵². Furthermore, in patient-derived colon adenocarcinoma organoids, a BH3 mimetic agent caused cell death within the organoid. Strikingly mutations in Kras, in conjunction with Apc deletion, delayed apoptosis and decreased expression of Bim. Studies in intestinal organoids support the concept that loss of contributes to colorectal carcinogenesis.

CRISPR-Cas9 genome editing has been used to mutate tumor suppressor genes such as APC, SMAD4, and TP53, and the oncogenes KRAS and PI3KCA, in colon epithelial organoids grown from normal human colon tissues ¹⁵³. Organoids in which all 5 genes were mutated grew independent of growth factors and had gene expression signatures similar to those of epithelial organoids generated from human colon adenomas. Organoids containing the 5 mutations formed tumors when transplanted into mice, but with lower frequency than colorectal adenocarcinoma organoids. Transplantation of engineered and patient-derived colorectal adenocarcinoma organoids under the kidney capsule of immunocompromised mice resulted in tumor formation, although patient-derived adenoma organoids did not engraft. ¹⁵³. Organoids containing inactivating mutations of APC, SMAD4, and TP53 and activating mutations in KRAS also grew independently of growth factors and formed solid tumors following subcutaneous injection into immunocompromised mice ¹⁵⁴. In contrast, patient-derived, control organoids did not engraft after subcutaneous injection. Epithelial organoids can therefore be used to assess the combinatorial effects of mutations associated with malignancy.

Consistent with studies linking mutations in the Wnt signaling pathway to human colon adenocarcinoma, organoids with high expression of the Wnt target gene, EphB2, have a transcriptional signature similar to that of intestinal stem cells and are more likely to form tumors following subcutaneous injection into immunocompromised mice ¹⁵⁵. Similarly, mouse small intestinal cells expressing high levels of EPHB2 generated organoids more efficiently than small intestinal cells with low expression of EPHB2 ¹⁵⁵. These data indicate that epithelial organoids can be used to assess the proliferative capacity of a population of cells, and can also be used to test the efficacy of different chemotherapeutic agents ¹⁵⁶.

Human colonic organoids generated from adenocarcinomas and normal tissues can be used to screen drugs and other molecules for their effects on tissues from different patients, with different genomes. Studies of the response of colon adenocarcinoma organoids to small molecules associated loss-of-function mutations in TP53 with resistance to the MDM2 inhibitor nutlin-3a. Organoids from colonic tumors with KRAS mutations were resistant to the anti-EGFR inhibitors cetuximab and BIBW2992 ¹⁵⁷. GI organoids can therefore be used to bridge pharmacologic studies on short-term cultures with xenotransplant studies, to develop individualized treatment regimens. Organoids might be used in high-throughput screens of drugs and pharmacologic agents for those with effects in the patients from which organoids were generated.

Similar to organoids derived from epithelial tissues, epithelial–mesenchymal organoids have been generated from tumor samples ^{99, 158}. Li et al generated mouse colonic epithelial-mesenchymal organoids with deletion of Apc, activating mutations in Kras, and lentiviral knockdown of p53 and SMAD4. Interestingly, the epithelial–mesenchymal organoids generated from small intestine were more susceptible to transformation than organoids from the colon. The small intestine organoids required alteration of only 2 genes whereas colon organoids required all 4 alterations for transformation.

Adenocarcinoma of the stomach carries is the second most common cause of cancer worldwide and patients have poor prognoses ¹⁵⁹. Given the poor response of gastric cancer to treatment, there is a need determine which patients are most likely to respond to specific drugs. Mouse epithelial–mesenchymal organoids from normal tissue have been transformed into gastric cancer organoids through alterations to Kras or p53 ⁹⁹. Gastric organoids with activating mutations of Kras or loss of p53 become dysplastic and generate adenocarcinomas in mice⁹⁹.

Gastric epithelial organoids have been used to model hereditary gastric cancer and determine the role of TGFB receptor 2 (TGFBR2) in metastasis. Mutations in the cadherin 1, type 1 gene (CDH1 or ECAD) are associated with hereditary gastric cancer. Nadauld et al performed genome sequence analysis of a patient who developed gastric cancer and later developed metastases. This study identified TGFBR2 as genetic factor that increased risk for metastatic disease. Knockdown of TGFBR2 in mouse gastric epithelial-mesenchymal organoids with disruptions in Cdh1 and p53 increased tumorigeniciy when organoids were transplanted back into mice ¹⁵⁸.

Whole-genome sequencing and epigenetic profiling of gastric cancer and adjacent normal tissue specimens identified mutations in RHOA ¹⁶⁰. Mouse intestinal epithelial organoids with lentiviral overexpression of RHOA mutants survived in the absence of the ROCK inhibitor, which prevents anoikis, after passage. RHOA mutations therefore seem to promote survival of these cells and contribute to gastric carcinogenesis. Studies of organoids generated from patients’ tumors could provide more information about the oncogenic effects of RHOA mutations.

Human gastric cancer epithelial organoids have also been generated by culture of primary cells from resected gastric adenocarcinomas. Human gastric organoids derived from normal and tumor tissues have similar morphologies in culture. However, orthotopic tumors derived from organoids transplanted into mice engraft within the gastric epithelium and generate metaplastic phenotypes similar to those of the patients’ tumor specimens (unpublished results, Bertaux-Skeirik and Zavros). This approach might be used to develop an organoid-based platform for testing the specific responses of individual patients’ tumors to different therapeutic agents.

Epithelial organoids have also been generated from Barrett’s esophagus (BE) and gastric cancer specimens ^{60, 160}. BE predisposes esophageal tissues to dysplasia and adenocarcinoma. BE can be identified based on the presence of intestinal goblet cells. Esophageal epithelial organoids generated from patients with BE could be induced to undergo goblet cell differentiation ⁶⁰. Given recent developments in the ability to culture mouse esophageal epithelial organoids, it might be possible to develop normal human esophageal epithelial organoids, which could be compared to organoids form BE or esophageal cancer cells.

Epithelial organoids generated from colon cancer tissues have many features of the tumors from which they were derived, are easily expandable, and can be used in screens for drug efficacy and toxicity. Tissue epithelial–mesenchymal organoids and PSC-derived organoids might also be used in screens of pharmacologic agents for activity against gut tumors.

Regenerative Medicine

Factors such as ease of culture, expandability, capacity to model heterogeneous cell fates, and suitability to genetic manipulation make GI organoids a useful tool. Long-term organotypic cultures have been used to study the stem cell niche, and recently as guides for normal development and sources of replacement tissues. Generation of autologous tissue derived from biopsy specimens or iPSCs offers advantages over conventional organ transplants, which are limited in availability and require immunosuppression.

Endocrine differentiation of gut organoids

GI endocrine cells secrete hormones; disruption of to process can cause disease and death. For example, autoimmune destruction of insulin-producing pancreatic β cells results in type I diabetes, and loss of enteroendocrine cells leads to anendocrinosis and life-threatening diarrhea ^{124, 125}. Dysregulation of gut hormones have also been implicated in obesity and metabolic syndrome. Gut organoids can be used to study the pathogenesis of these diseases, and perhaps more importantly, as an alternative tissue source for therapeutics ¹⁶¹.

Ectopic expression of NGN3 or NEUROG3 in mouse epithelial mesenchymal organoids, human PSC-derived intestinal organoids, and human PSC-derived gastric organoids increased numbers of enteroendocrine cells, as it does in mice^{47, 48, 50, 162, 163, 164}. Furthermore, lentiviral knockdown of ARX in PSC-derived intestinal organoids led to reduction of CCK and secretin, similar to mice with intestine-specific disruption of Arx ⁷⁷.

PSC-derived intestinal organoids have also been used to study trans-differentiation, or conversion from 1 cell lineage into another, of insulin-producing cells^{165, 166}. Bouchi et al identified the transcription factor FOXO1 as a marker of enteroendocrine cell progenitors and a regulator of cell fate decision. Overexpression of a dominant-negative form of FOXO1 increased insulin expression, indicating transdifferentiation into insulin-producing cells. Ectopic expression of the dominant-negative form of FOXO1 also resulted in upregulation of transcription factors important for the development of insulin-producing pancreatic β cells, such as NEUROG3, MAFA, NEUROD1, and NKX6.1. Remarkably, expression of the dominant-negative form of FOXO1 also resulted in secretion of c-peptide, a byproduct that forms when proinsulin is cleaved into insulin in response to insulin secretagogues such as glucose, arginine, and potassium chloride.

Insulin-producing cells in PSC-derived organoids can detect insulin secretagogues and secrete c-peptide. When organoids that express the dominant-negative form of FOXO1 were transplanted subcutaneously into humanized mice, however, human c-peptide from could not be detected in serum. This result is likely due to low yield of insulin-producing cells after FOXO1 inhibition (approximately 0.05% of cells in the organoid). Genes can be expressed from adenoviral vectors in organoids for short-term experiments, but expression is transient. Bouchi et al used lentiviral vectors to knockdown FOXO1 in PSC-derived organoids. Similar to their results with adenoviral inhibition of FOXO1, lentiviral knockdown resulted in an increase in insulin-producing cells.

In another study showing that PSC-derived intestinal organoids can be transdifferentiated into insulin-producing cells, lentivirus was used to ectopically and constitutively express transcription factors necessary for pancreatic β-cell development, including PMN ¹⁶⁶. PMN-expressing organoids produced insulin mRNA and protein. Moreover, 22% of cells produced c-peptide, indicating that PMN overexpression results in more efficient generation of insulin-producing cells than FOXO1 inhibition. It is not clear whether insulin-producing cells generated by PMN overexpression can reduce features of diabetes in animal models.

Bouchi et al also confirmed the presence of enteroendocrine cells in PSC-derived organoids that produce glucagon-like peptide-1, gastric inhibitory polypeptide, somatostatin, ghrelin, and cholecystokinin. As such, PSC-derived organoids can be used to study regulators of hormone secretion and the effects of hormone secretion on other tissue types, in vitro and in vivo. PSC-derived organoids might also be used in bariatric surgery research. Weight loss following bariatric surgery does not occur via reduction of enteric surface area or carrying capacity, but rather via biochemical changes ¹⁶⁷. PSC-derived organoids, which are in transplanted in discontinuity with the gut, permit delineation of enteroendocrine hormone production due to enteral content vs changes due to biochemical modulators. They might be used to identify biochemical that obviate the need for bariatric surgery.

Gut regeneration

Short-bowel syndrome is a devastating disease that affects children and adults. Many patients must have had more than 50% of their small bowel resected due to congenital anomalies, infection or inflammation, or intestinal ischemia. As a result, they do not maintain fluid, electrolyte, or nutrient homeostasis. Supportive therapy, including parenteral nutrition, and curative therapy, such as small-bowel transplantation, cause complications with significant morbidity and mortality. Human intestinal tissue was successfully integrated into the intestines of mice. However, given that patients with short-bowel syndrome lack sufficient intestinal length, integration of tissue into existing bowel is not feasible, and therefore, alternative sources to replace missing tissue are needed¹⁶⁸.

Each of the 3 types of gut organoids have been successfully transplanted into mice, so these tissues might be used to engineer small intestine ^{82, 117, 99, 57, 113, 54}. Mice with DSS-induced colitis given enemas containing epithelial colon organoids had higher body weights 2 weeks after transplantation than mice that did not receive the organoids. The organoids engrafted and were functional 4 weeks after transplant ⁵⁴. Similarly, mouse fetal and adult small intestine epithelial organoids engrafted into the distal colon after induction of inflammation ^{57, 113}.

Orthotopic transplant of mouse intestinal organoid tissue has required induction of inflammation. Human epithelial organoids transplanted under the kidney capsule or subcutaneously do not engraft in the absence of transforming mutations ^{153, 154}. In contrast, human PSC-derived intestinal organoids have been transplanted under the kidney capsule or in the omentum without the use of inflammatory agents such as DSS ^{117, 82}. Additional studies are necessary to determine if human epithelial organoids can engraft in the colon in the presence of inflammation; this might be a new therapeutic approach for IBD ¹²⁷.

Intestinal organoids have not yet been used to lengthen small intestine. Most bowel-lengthening studies use organoid units, generated from fragments of full-thickness jejunum or ileum. These contain a mesenchymal core surrounded by a polarized intestinal epithelium. After siting onto polymer scaffolds, organoid units were first successfully transplanted into the omentum of syngeneic Lewis rats to create tissue-engineered small intestine (TESI) ¹⁶⁹. TESI have been appended to the anastomosis in rat models of small bowel resection ¹⁷⁰. Interestingly, inclusion of TESI in the anastomosis increased bowel length 9 months after bowel reconstruction and caused faster weight gain after reconstruction ^{170, 171}.

TESI has also been generated from human small intestinal tissue and successfully transplanted into immunocompromised mice ¹⁶⁸. The major limitation of this approach, however, is the inability to expand TESI in vitro. In contrast, reconstituted epithelial–mesenchymal organoids, epithelial organoids, and PSC-derived organoids can all be expanded in vitro. More studies are needed to determine whether these approaches can be used to increase bowel length. Reconstituted epithelial–mesenchymal organoids and PSC-derived organoids are likely to be effective because of their mesenchymal support structure. However, synthetic scaffolds could provide the necessary support for transplanted epithelial organoids.

LIMITATIONS

Despite the strengths that organoids offer over traditional cell lines, their use is restricted. For example, although organoids are complex and often retain multiple cell lineages, they lack many of the cellular inputs present in an in vivo system (eg, neural, endothelial, or immune cells). We can increase the complexity of these systems through co-culture with additional cell types, but organoids should still be viewed as a reductionist system.

Organoids are also limited by significant variations among laboratories in methods and tissue sources. This issue is less of a concern with PSC-derived organoids, because laboratories generally access the same cell lines and follow well-established methods to differentiate organoids. Conversely, tissue-derived organoids are generated from patients using growth methods that have been optimized in each laboratory. This heterogeneity affects reproducibility if the derivation and characterization of the lines are not well described. For example, changing growth conditions for tissue-derived intestinal epithelial organoids can skew the differentiation state from stem cell-enriched to a balance of stem cells and differentiated cell types.

A practical limitation of organoid models is their cost of maintenance. Many commercial growth factors that are required for culture—these can be expensive and have poor bioactivity. These limitations also restrict assays that require significant scale-up, such as drug screens, providing barriers to their use in personalized medicine approaches. Many groups have turned to cell lines that generate growth factors in conditioned media, to reduce costs and improve bioactivity, but this introduces additional experimental variation.

Future Directions

Organoid cultures have been developed for multiple GI organs. These systems are multicellular, similar to in vivo models, but have the flexibility of in vitro systems. As such, gut organoids have become a powerful tool with steadily increasing applications.

Intestinal organoid cultures continue to offer insights into the stem cell niche. Researchers have recently begun to focus on generating patient-specific tissues. Individualized organoids can be used to identify intestine-specific pathophysiologies in complex disease processes, such as lack of forskolin-induced swelling in organoids with mutated CFTR. Patient-specific organoids can be used to study mechanisms of diseases such as MVID and MIA. Organoids generated from normal and tumor specimens can be used to characterize the events that occur during transformation, growth, and progression of malignancies such as colorectal or gastric cancers. Finally, organoids generated from patient biopsies could be used to regenerate intestinal tissues for patients with disorders such as short bowel syndrome or IBD and avoid the adverse effects of immune rejection or immune suppression. Finally, organoids can be used in screens of pharmacologic agents that have effects on cells and tissues from individual patients, to advance personalized medicine.

List of abbreviations

Bim

BH3-only protein Bim_EL isoform

DSS

Dextran sodium sulfate

ECAD

Epithelial cadherin

EGF

Epithelial growth factor

ESC

Embryonic stem cell

GI

Gastrointestinal

iPSC

Induced pluripotent stem cells

IFNG

Interferon-γ

NEC

Necrotizing enterocolitis

PSC

Pluripotent stem cell

PUMA

p53 up-regulated modulator of apoptosis

TESI

Tissue engineered small intestine

TGFB

Transforming growth factor β

TLR4

Toll-like receptor

TGFBR2

Transforming growth factor-beta receptor 2

TNF

Tumor necrosis factor