Human 3D Gastrointestinal Microtissue Barrier Function As a Predictor of Drug-Induced Diarrhoea

An important consideration in the development of a novel therapeutic agent is the potential for off‑target effects on the GI tract. The cells lining the crypts of the small intestine are the most rapidly proliferating in the body and therefore sensitive to the effects of many therapeutic agents. Damage to stem cells within the intestinal crypts leads to cell and villus loss, ulceration and inflammation of the epithelium, compromising the integrity of the GI tract. Gastrointestinal toxicities such as nausea, vomiting, constipation and diarrhoea are some of the most common adverse drug events observed in phase 1 clinical trials.

In most instances these toxicities are not life-threatening and therefore treated with anti‑diarrhoeal co‑therapy and/or dose reduction resulting in significant impact on patient quality of life. Addressing GI‑Toxicity problems at the preclinical stage of drug development would help alleviate these adverse events and lead to lower drug attrition rates. Currently, assessment of GI safety is conducted in vivo in higher‑order species due to the lack of translational concordance with clinical data in rodent studies. The cost and ethical implications of higher‑order species testing limits the capacity for earlier screens during the development process with the final candidate typically being the only drug tested.

Long-term culturing of intestinal stem cells as organoids has the potential to transform the throughput and translatability of preclinical GI‑Toxicity studies. Organoids from intestinal stem cells differentiate into the diverse cell types observed within the GI‑tract and self-organise into villus and crypt-like structures making them ideal for screening therapeutics for cytotoxicity.

In this study, the predictive potential of changes in microtissue barrier function was assessed for drug screening potential on therapeutics known to be associated with clinical diarrhoea. The dynamics of response were modelled to generate a strategic guide for GI‑Tox mitigation. Small intestinal epithelial cells were seeded on a support layer of fibroblasts and development and differentiation was induced by transwell culturing with an air liquid interface with cultures only being fed basolaterally.

The mature microtissues consisted of a columnar epithelial monolayer with nuclei at the base similar to enterocytes. A connective tissue layer containing fibroblasts and myofibroblasts supported the formation of raised or fold-like structures and higher resolution images and villin immunostaining demonstrated the presence of microvilli-like projections and a brush border.

The presence of tight junctions between adjacent enterocyte cells was demonstrated by claudin‑1 and cytokeratin expression. Small clumps or pairs of LGR5+ or OLFM4‑positive stem cells were observed at irregular intervals throughout the layers but the majority of cells were ki67‑stained proliferating cells.

Microtissue barrier function, analysed by transepithelial electrical resistance (TEER), was lower than TEER of Caco‑2 cultures, possibly due to the more diverse cell types present, which lack tight junctions. These lower TEER values are more consistent with those observed in native tissue.

Microtissue barrier function was assessed for its efficacy in predicting clinical diarrhoea by testing a list of 39 widely prescribed drugs. Drugs were selected by diarrhoea incidence in the general population and the mechanism by which diarrhoea is caused.

In blinded assays, 4 concentrations of each drug were tested (1‑100μM) using the microtissue TEER assay at 48 and 96hrs drug exposure. IC15 and IC25 were calculated and TEER values were normalised to clinical exposure associated with diarrhoea (Total plasma Cmax), which was extracted from published data.

96h‑IC15, being the time/concentration with minimum disruption to barrier function was chosen as the threshold for analysis. 15 out of 17 non-diarrhoeagenic drugs had TEER IC15/exposure ratios above 80, whereas diarrhoeagenic drugs were more potent with 11 of 14 having TEER IC15/exposure ratios less than 80. 8 drugs had no effect at the concentrations tested.

The predictive accuracy was 84% and a blinded repeat experiment resulted in an 83% predictive accuracy score, whereas a similar TEER predictability test with these drugs on Caco‑2 cells displayed 77% accuracy but only 57% sensitivity.

Four additional compounds, AZD3409, AZD8931, AZD7140, and AZD3, were selected for testing as clinical trials with these compounds were affected by diarrhoea either limiting dose or as the most severe adverse event. All 4 of these compounds had TEER IC15/Clinical Cmax ratios of 8 consistent with their diarrhoeagenic effects.

The authors speculate that TEER may be a useful method in cases where toxicity is caused by an on-target mechanism. Controlling adverse event risk in these cases may be done with dosing schedules that maximise therapeutic efficiency, a procedure that can utilise mathematical modelling. AZD1 (Dose-limited by diarrhoea), AZD2 (Not dose-limited) and AZD8391 (positive control) were tested on microtissues for 3 days followed by washout, TEER was measured each day.

A decrease in TEER was observed during treatment with AZD1 and AZD8391, AZD2 did not affect TEER. After washout AZD8931-treated microtissue recovered quickly but AZD1 recovered slowly over 17 days. A second treatment resulted in no recovery for AZD1 whereas AZD8391 recovered slower than the single treatment time.

As this data was based on continuous exposure, artificially high levels of drug may lead to higher levels of toxicity that is irrelevant in the clinical setting. Using SimCYP software, intracellular enterocyte exposure of AZD1 after an oral 75 mg dose was estimated to last for 150 mins with a concentration between 5‑100μM. Treatment of microtissues with 5‑100μM AZD1 for 150 mins at 3 dosing schedules demonstrated that TEER disruption was delayed in an exposure-related fashion and that TEER recovery at the highest does takes longer than the longest off-treatment time.

A mathematical model for AZD1 exposure using a non-autonomous system of differential equations was developed with the assumption that damage to the barrier accumulates and TEER is only affected after a threshold of barrier damage is reached. A consistent value of 0.48+/-0.01 was calculated that linked damage and barrier function to TEER measurements at different doses and schedules. This model was further developed to include damage recovery after TEER impairment.

New preclinical approaches are required for earlier testing of therapies for GI‑toxicity. In this study the authors describe an in vitro human 3D GI microtissue and validated it for GI‑toxicity using a panel of known drugs that do/do not cause diarrhoea. Microtissues form structures consistent with gut epithelium and can be generated from higher-order species and more importantly from human cells. With moderate throughput the assay can quantify the dynamics of toxicity, both onset and recovery. Epithelial barrier function as the main endpoint assay is a direct correlation of tissue damage and may lead to improved clinical translation enabling earlier screening reducing the impact of GI toxicities in clinical trials.

Want to ask a question about


Epistem's Organoid and GI‑Toxicity Models

We are committed to providing innovative services to accelerate drug development and based on the Hans Clevers models, Epistem offers three dimensional intestinal organoid models in a variety of species (mouse, rat, dog and human). This well-established model can be used for determining the effect of novel therapeutic agents on normal intestinal tissue to predict the in vivo response, typically off target/off tissue toxicities. Multiple therapeutic agents can be quickly and efficiently screened, which requires only small volumes of agent and is possible where in vivo screening may be unfeasible.

To support our organoid models we offer rodent models of GI‑toxicity, including radiation and chemotherapy evaluation. In addition, our analytical services include measures of enterocyte loss / recovery (including characterisation of the target cell population and response kinetics), diarrhoea severity, mucosal permeability, cytokine (tissue and plasma) changes and inflammatory cell responses via FACS analyses and complete blood count measures.

Peters et al., 2019. Human 3D Gastrointestinal Microtissue Barrier Function As a Predictor of Drug-Induced Diarrhea. Toxicological Sciences, 168(1), 3-17 DOI: 10.1093/toxsci/kfy268

About Epistem

Epistem's contract research service is committed to providing reliable, innovative and transferable pre-clinical models and services to support decision making throughout the drug discovery and development pipeline.

Tel: +44 (0)161 850 7600  Email: