(193an) Engineering the Liver Diverticulum from Human Pluripotent Stem Cells

We hypothesize that the structures and cues that initiate 3D liver formation can be mimicked with human pluripotent stem cells (hPSC). Thus we aimed to engineer an in vitro model of the liver diverticulum (LD), a key structure that: 1) arises in mouse development (E9.5) and human development (d26) and 2) forms the 3D liver. From inside the gut to outside, the LD is composed of a single layer of hepatic endoderm (HE), encased by a single layer of endothelial cells, and surrounded by the septum transversum mesenchyme (STM). 3D liver formation starts when the hepatic endoderm (HE) delaminates, joins with the endothelial cells, and migrates into the STM. We first determined the appropriate LD dimensions with an online mouse database. Next, to model the LD, we used a commercially available cell migration system, which enabled selective cell seeding and cell adhesion of two migrating cell types, on 0.22 cm² square islands separated by 500 μm. HepG2 liver cancer cells, a model of the HE, and human mesenchymal stem cells (hMSC), a model of the STM, were cultivated and analyzed using time lapse microscopy. At 8 hours we observed that 60% of MSC and 30% of HepG2 cells on the opposing edges of each cell island, had migrated to the opposing side, and at 24 hours, the cells had intermixed (n=2). In a second LD model, we wished to specifically evaluate HepG2 motility towards MSC, but not the reverse. Using a Transwell cell invasion assay, HepG2 were seeded on a matrigel-coated insert (8 μm pore) and inserted above cultured MSC. After 24 hours, there was a 70% increase in liver cell invasion through the insert compared to the “no MSC” condition (n=2). Overall, both 2D assays showcase migration of HepG2 in proximity of MSC’s. We then explored further 3D systems to self-assemble the LD with HepG2 (HE), MSC (STM), and human umbilical cord vein endothelial cells (HUVEC). Dispase-treated HepG2 cell fragments (100 µm x 100 µm) were initially seeded within hydrogels composed of Matrigel (MG). We added HUVEC 48 hours later, and performed microscopy. After 12 hours, HUVEC cell migration resulted in 80% of the HepG2 fragments having interconnected tubular, cord-like HUVEC structures, but not in no MG controls. In our final model of LD, we further refined this system. Singular HepG2 cell fragment was seeded into a single well of a 384-well plate. HUVEC were added to coat the fragment surface, and the system was submerged in a hydrogel (MG) containing MSC. We observed that average HUVEC/HEPG2 fragment area increased by 22%. This was due to 3-4 multicellular protrusions characteristic of collective cell migration (n=2) towards MSC, as occurs in the LD. Ultimately, our data suggests that HepG2 do migrate towards MSC in 2D systems, and that HepG2 and HUVEC together migrate towards MSC in 3D systems. Future studies will encompass PSC-Heps in a micro-engineered liver tissue, for an in-vivo liver cell therapy model.