Background and aims: In the past decade, organoid models have revolutionized the study of the human gastrointestinal (GI) epithelium. Organoids can be derived from patient tissue and grown as strictly epithelial structures referred to as enteroids and colonoids. Organoids can also be generated through the directed differentiation of human pluripotent stem cells (hPSCs) into intestinal organoids (HIOs). Rare cell types are often absent in enteroids/colonoids and induced intestinal organoids grown in vitro. We hypothesized that induced expression of key transcription factors would be sufficient to specify these cell types. To test this hypothesis, we generated lentiviral constructs that allow the inducible expression of cell-type specifying transcription factors. We then transduced human enteroids/colonoids and hPSCs with these constructs and examined their ability to induced differentiated cell types.
Methods: CDNAs for KLF4 which is required for goblet cell differentiation, POU2F3 which is required for tuft cell differentiation, SPIB which is required for M-cell differentiation, and NEUROG3-T2A-NKX6-3 (NKX6-3 is required for G cell specification) were cloned into pINDUCER20. The constructs were then packaged into lentiviral particles which were then used to transduce human enteroids/colonoids and hPSCs, the latter which were differentiated into HIOs using established protocols. Multiple pulse and chase durations were tested to determine the optimal conditions for differentiation of the desire cell types. Cell types were identified by immunofluorescence staining using cell type specific antibodies and quantified using ImageJ.
Results: We found that inducible expression of the forementioned transcription factors was sufficient to specify the corresponding cell types. Induced KLF4 expression generated colonoids enriched in TFF3+ and WFDC2+ goblet cells, the latter which we could not generate by NOTCH inhibitor treatment. Induced POU2F3 expression was sufficient to induce DCLK1+ tuft cells, SPIB was sufficient to induce M-cells, and NEUROG3-T2A-NKX6-3 was sufficient to induce ectopic G-cell differentiation in colonoids.
Conclusions: We have developed a lentiviral toolkit that allows the generation of human enteroids/colonoids and induced HIOs that are enriched for differentiated cell types. This will allow the interrogation of regional differences in differentiated cell types and the specification of goblet cell, M-cell, tuft cell and enteroendocrine, cell subtypes. Using similar principles, we hope to construct a library that allows the generation of organoids enriched in differentiated cell subtypes (i.e. WFDC2+ goblet cells, or INSL5+ enteroendocrine cells). This work has the potential to treat patients by replenishing cell types that are depleted in conditions such as inflammatory bowel diseases.

KLF4 is sufficient to specify goblet cells. (A) Schematic of KLF4 construct and expected result for induced expression. (B) Immunofluorescence images of control and doxycycline treated enteroids stained for E-CAD (green), TFF3 or WFDC2 (red) and counterstained with DAPI (blue).
POU2F3 is sufficient to specify tuft cells. (A) Schematic of POU2F3 construct and expected result for induced expression. (B) Immunofluorescence images of control and doxycycline treated enteroids stained for E-CAD (green), DCLK1 (red) and counterstained with DAPI (blue).