Bioprinted 3D Primary Human Intestinal Tissues Model Aspects of Native Physiology and ADME/Tox Functions

Abstract

The human intestinal mucosa is a critical site for absorption, distribution, metabolism, and excretion (ADME)/Tox studies in drug development and is difficult to recapitulate in vitro. Using bioprinting, we generated three-dimensional (3D) intestinal tissue composed of human primary intestinal epithelial cells and myofibroblasts with architecture and function to model the native intestine. The 3D intestinal tissue demonstrates a polarized epithelium with tight junctions and specialized epithelial cell types and expresses functional and inducible CYP450 enzymes. The 3D intestinal tissues develop physiological barrier function, distinguish between high- and low-permeability compounds, and have functional P-gp and BCRP transporters. Biochemical and histological characterization demonstrate that 3D intestinal tissues can generate an injury response to compound-induced toxicity and inflammation. This model is compatible with existing preclinical assays and may be implemented as an additional bridge to clinical trials by enhancing safety and efficacy prediction in drug development.

Keywords: Bioengineering; In Vitro Toxicology Including 3D Culture; Tissue Engineering.

Graphical abstract

Figure 1

Architecture of 3D Intestinal Tissue (A) A bilayered architecture is achieved by bioprinting an interstitial layer containing adult human intestinal myofibroblasts (IMF) followed by adult human intestinal epithelial cells (hIEC). The vimentin-expressing interstitial cells and CK19-expressing epithelial compartments remain separate over 17 days in culture (B, C). Epithelial cells express markers necessary for barrier formation such as E-cadherin (D), ZO-1 (D, inset), and apically expressed villin (E). Mucus production from goblet cells (* denotes secreted mucus, arrow indicates goblet cells) is observed throughout culture (F, G). Lysozyme-stained Paneth cells (H) and chromogranin-expressing enteroendocrine cells (I) are also present.

Figure 2

Gene Expression Comparisons of Native Intestine, 3D Intestinal Tissue, and Caco-2 Monolayers (A) Heatmap clustering shows a closer relationship between native intestine and 3D intestinal tissue than Caco-2 monolayers (native intestine N = 3 donors, 3D intestine n = 3 biological replicates per donor, and Caco-2, n = 3 biological replicates). (B) General markers for intestinal barrier function (CDH1, VIL1) are similar among the three groups. Epithelial subtype markers are highest for native and 3D intestinal tissue. Drug-inducible transcription factor NR1I2 (PXR) is similar for native intestine and 3D intestinal tissue. Native intestine and 3D intestinal tissue express all metabolic enzymes analyzed, including Phase I cytochrome P450s, the majority of which Caco-2 cells lack. Transporters are expressed by all three groups, with variation in the level. For efflux transporters, 3D intestine is most similar to native intestine, whereas Caco-2 cells underexpress ABCB1 (P-gp) and ABCG2 (BCRP). Uptake transporters SLC15A1 (PEPT1) and SLCO2B1 (OATP2B1) are similarly expressed by native and 3D intestine, but not by Caco-2 monolayers. Data in (B) are expressed as mean ± SD.

Figure 3

Cytochrome P450 Metabolism in 3D Intestinal Tissue (A) CYP2C9 basal activity was validated by a luciferin activity assay and was reduced by inhibitor sulfaphenazole (n = 6). CYP3A4 activity was shown by luciferin activity assay (n = 3–4) (B, C) and midazolam hydroxylation (n = 5) (D, E), both of which were inhibited by ketoconazole. (C, E) CYP3A4 induction via rifampicin treatment was detected by increased luciferin activity and midazolam metabolism in 3D intestine (n = 5). (C) A comparison of CYP3A4 activity in 3D intestinal tissue with 2D Caco-2 monolayer shows low activity of CYP3A4 in Caco-2 monolayers with no rifampicin induction (n = 4–11 for 3D intestine, n = 3 for 2D Caco-2). (F) Rifampicin treatment increased gene expression of PXR-regulated CYPs, ABCB1 (P-gp), and UGT1A1 but not CK19 and ECAD (n = 4). Level of significance: ****p < 0.0001, ***p < 0.001, **p < 0.01 by two-way ANOVA. Data are expressed as mean ± SD.

Figure 4

Barrier Function of 3D Intestinal Tissue (A) Transepithelial electrical resistance (TEER) was measured for 3D intestinal tissue from day 6 to day 21, showing increase early in culture, with maintenance of barrier function within physiological levels (dotted lines) after day 10 (n = 24). (B) Permeability of test compounds for 3D intestine, measured in the apical to basal direction, shows distinction between low- (mitoxantrone [n = 4]), moderate- (topotecan [n = 4], digoxin [n = 6]), and high- (propranolol [n = 5]) permeability compounds. (C) Comparison of TEER values shows more physiological TEER values for 3D intestine (n = 24) and 3D Caco-2 (n = 6) versus 2D Caco-2 (n = 5), whereas IMF alone does not show barrier function (n = 19). (D) Comparison of lucifer yellow permeability values between models shows low permeability for all epithelial-cell-containing models and high permeability for IMF only (n = 4 for 2D and 3D Caco-2 and IMF only, n = 28 for 3D intestine). Data are expressed as mean ± SD.

Figure 5

P-gp and BCRP Transporter Function in 3D Intestinal Tissue (A and B) (A) P-gp and (B) BCRP are apically localized in 3D intestinal tissue, similar to native intestine. Expression in Caco-2 monolayers was in patches at the apical surface. (C) P-gp substrate digoxin had greater permeability in the B to A direction with efflux ratio of 2.1. In the presence of P-gp inhibitor zosuquidar, the efflux ratio of digoxin reduced to 1.2 (n = 6). (D) BCRP/P-gp substrate topotecan had an efflux ratio of 8.8 under control conditions. BCRP inhibition by Ko143 reduced B to A transport and the efflux ratio decreased to 3.6. Dual inhibition of P-gp and BCRP resulted in ablation of transport with an efflux ratio of 1.4 (n = 3–4). Level of significance: ****p < 0.0001 by two-way ANOVA. Data in (C) and (D) are expressed as mean ± SD.

Figure 6

Indomethacin Toxicity in 3D Intestinal Tissue (A) TEER measurements of tissues following 24-hr incubation with ethanol vehicle or varying doses of indomethacin show dose response decrease in TEER with increasing indomethacin. (B) LDH activity increased with increasing indomethacin dose, suggesting increased cytotoxicity. (C) Prostaglandin E₂ synthesis decreased to similar levels for all indomethacin doses tested, confirming drug activity (n = 6–7, A–C). (D) Histology of indomethacin-treated tissues shows disruption of the epithelium and distorted nuclear staining at higher doses of indomethacin, accompanied by a reduction in E-cadherin, a marker of barrier function. Level of significance: ****p < 0.0001 by one-way ANOVA. Data in (A–C) are expressed as mean ± SD.

Figure 7

TNF-α-Induced Toxicity in 3D Intestinal Tissue (A) Tissues treated with TNF-α for 24 hr showed increased epithelial disorganization compared with controls. (B) Increased LDH activity correlated with changes in cell morphology following TNF-α treatment (n = 5–6). (C) A subset of genes related to inflammation, TNF-α, IL-6, IL-8, CCL2, and ICAM, were upregulated in response to TNF-α treatment (n = 3). Level of significance: ***p < 0.001 by t test in (B); **p < 0.01, ***p < 0.001, and ****p < 0.0001 by two-way ANOVA in (C). Data in (B) and (C) are expressed as mean ± SD.