Bioprinted 3D vascularized tissue model for drug toxicity analysis

Background

Three-dimensional (3D) tissue models used for drug testing need perfusable vessels to carry nutrients, oxygen and drugs, and an endothelial lining would let a model reproduce how a drug crosses the vessel wall before reaching tissue. The authors review earlier bioprinted vascular approaches and note their drawbacks: carbohydrate glass filaments have a short lifespan, cell-laden gelatin printing has low resolution and suits only constructs thicker than 1 cm, and pluronic multilayer systems are complex and consume more bioprinting material.

Design and approach

The construct uses a sacrificial bioprinting technique developed previously by the authors’ group: an agarose fiber is gelled in a glass capillary at 4 °C for 10 s, printed into a HepG2/C3A-laden GelMA prepolymer and, after 20 s of UV crosslinking, pulled out by hand to leave a smooth channel. The authors cite reproducible channels of 100–1000 μm, removal without dissolving a sacrificial material and a smooth lumen for an endothelial monolayer as advantages of this method. HepG2/C3A cells were chosen as a stable human hepatocyte source compared with primary hepatocytes. After screening agarose at 4%, 6% and 8% and GelMA at 5%, 8% and 10%, the authors used 6% agarose, because 4% was too soft and 8% too stiff, and 8% GelMA with 20 s of UV exposure. HUVECs were seeded into the channel and cultured for 48 h, and the construct was mounted in a sealed bioreactor where HUVEC medium flowed through the channel while HepG2/C3A medium surrounded the hydrogel.

How it was evaluated

The hydrogel was characterized by scanning electron microscopy, and diffusion of 40 nM rhodamine 6G was imaged at 1, 15 and 30 min in channels of 100, 500 and 1000 μm under static conditions and under flow at 50 μl/h, and in 500 μm channels with and without the HUVEC layer. The endothelium was examined by confocal microscopy with F-actin and CD31 staining. Drug effects were tested by perfusing 30 mM APAP through the channel for 48 h, by treating HUVECs in well plates with 15 and 30 mM APAP for 24 and 48 h, by live/dead staining of the vascularized constructs, and by PrestoBlue metabolic assays of HepG2/C3A constructs with and without the HUVEC layer. All experiments used human cells in vitro.

Key results

The 8% GelMA had a Young’s modulus of 12.1 ± 1.1 kPa, higher than the 5.5 ± 1.6 kPa reported for healthy human liver; the authors state that this was the lowest GelMA concentration and UV exposure that gave appropriate cell growth. The average pore size was 143.2 ± 6.4 μm, and dye diffused throughout the channel faster under flow than under static conditions.

HUVECs clogged 100 μm channels and needed up to 5 days to cover 1000 μm channels, whereas they uniformly covered 500 μm channels after 48 h, aligned along the channel axis and showed strong CD31 staining. By 7 days, endothelial cells had invaded and sprouted into the surrounding GelMA. Because the porosity of the scaffold was the same with and without the endothelium, the authors attribute the slower dye diffusion seen with the HUVEC layer (p < 0.001) to the endothelial barrier itself.

Perfusion of 30 mM APAP for 48 h disrupted the endothelial layer, which showed discontinuous F-actin and nuclear staining, and static treatment with 15 or 30 mM APAP reduced HUVEC metabolic activity in all conditions, indicating that the damage did not depend on flow. Without drug, cells in both zones of constructs about 3 mm thick remained more than 80% viable. With APAP, cell death concentrated near the channel: viability in the zone within about 300 μm of the channel fell to 20% after 48 h, cells in the zone about 600 μm away were less affected, and HepG2/C3A cells near the channel survived better than the HUVECs.

HepG2/C3A-only constructs showed lower metabolic activity than constructs with a HUVEC-lined channel after 24 and 48 h of 30 mM APAP. The authors interpret this as a protective effect of the endothelial layer and propose that the HUVECs delay drug diffusion as a barrier, metabolize part of the drug, or both.

Open questions

The authors describe the model as preliminary. The GelMA remained stiffer than healthy liver tissue, and they suggest that co-culture of human primary hepatocytes with liver endothelial cells would better represent in vivo conditions.

How it connects to other research

The related Nature Materials study, which the authors cite, printed carbohydrate glass filaments into a hydrogel laden with rat primary hepatocytes and dissolved them to form vessels; this model instead pulls out a printed agarose fiber to form a single endothelialized channel and uses it to test how the endothelium changes drug transport and toxicity.