Vascularized, Immune-Infiltrated, and Perfusable Kidney Organoid-on-Chip Models

Abstract

The cost of bringing new drugs to market has risen exponentially during the last sixty years. However, this trend, known as Eroom’s law, is beginning to turn around due to an improved understanding of disease biology. The United States Food and Drug Administration (FDA) Modernization Act was recently passed, which allows the use of human cell-based assays and computational models as preclinical data to further reduce drug development costs and improve their safety. Here, we report the development of vascularized, immune-infiltrated, and perfusable kidney organoid-on-chip models for drug testing and disease modeling. These models are expected to be widely adopted given that roughly 25 % of new drugs fail in the Phase III clinical trials due to nephrotoxicity. We first established a vascularized kidney organoid-on-chip model. Kidney organoids are seeded on an adherent extracellular matrix (ECM) and then subjected to superfusive flow on a millifluidic chip. Under flow, their endogenous pool of endothelial progenitor cells is expanded by nearly four-fold. These organoids contain a pervasive network of microvessels that are surrounded by mural cells. By contrast, the microvasculature that arises in kidney organoids cultured under static conditions is less well developed and recedes over time. Vascularized kidney organoids on chip also exhibit more mature podocyte and tubular compartments with enhanced cellular polarity and adult gene expression compared to static controls. Glomerular vascular development progressed through an intermediate stage akin to those involved in the formation of embryonic kidneys including the presence of capillary loops abutting primitive foot processes. Interestingly, however, vascular invasion of glomeruli was disrupted by the addition of exogenous VEGF. Next, we used this vascularized kidney organoid-on-chip model to unravel disease pathology and explore repurposed drugs for treating autosomal recessive polycystic kidney disease (ARPKD). First, polycystin-1 (PKHD1)-mutant organoids were generated by our collaborators in the Morizane lab (Massachusetts General Hospital) to mimic the monogenetic cause of the disease. These PKHD1-mutant organoids were then seeded onto our millifluidic chip and subjected to flow. Under these conditions, distal nephron dilation was observed. Transcriptomics discovered 229 signal pathways that were not identified in static controls. Mechanosensing molecules, RAC1 and FOS, emerged as potential therapeutic targets, which were validated by patient kidney samples by the Morizane lab. Based on this insight, we tested two FDA– approved and one investigational new drug that target RAC1 and FOS using our organoid-on-a-chip model, each of which suppressed cyst formation. To expand our efforts further, we created an immune-infiltrated vascularized kidney organoid-onchip model, which was used to investigate new drug candidates for cancer immunotherapy. We focused on, peptide-MHC targeted T cell bispecific antibodies (TCBs), which have recently emerged as a new class of biotherapeutics with improved specificity. These TCBs simultaneously bind to target peptides presented by the highly polymorphic and species-specific major histocompatibility complex (MHC) encoded by the human leukocyte antigen (HLA) allele present on target cells and to the CD3 co-receptor expressed by human T lymphocytes. To study their “on-target, off-tumor” effects, we introduced peripheral blood mononuclear cells (PBMCs) along with either non-targeting (control) or targeting TCB-based tool compounds to our model. The target consists of the RMF peptide derived from the intracellular tumor antigen Wilms' tumor 1 (WT1) presented on HLA-A2 via a bivalent T-cell receptor-like binding domain. Using our model, we measured TCB-mediated CD8+ T cell activation and killing of RMF-HLA-A2- presenting cells in the presence of PBMCs and multiple tool compounds. DP47, a non-pMHC targeting TCB that only binds to CD3 (negative control), does not promote T cell activation and killing. Conversely, the non-specific ESK1-like TCB (positive control) promotes CD8+ T cell expansion accompanied by dose-dependent T cell mediated killing of multiple cell types, while WT1-TCB* recognizing the RMF-HLA-A2 complex with high specificity, leads solely to selective killing of WT1-expressing podocytes within kidney organoids under flow. Given the limitations of this model, it remains undetermined whether an intravenously administered TCB would reach WT1 expressing cells in the kidney under physiological conditions and lead to “on-target, off-tumor” effects. Our 3D kidney organoid model offers a new platform for preclinical testing of cancer immunotherapies and investigating tissue-immune system interactions. A major drawback of our vascularized kidney organoid-on-chip models is their limited and uncontrollable microvessel perfusion. To address this deficiency, we created a new chip design containing an embedded macrovessel that can be directly anastomosed with the organoid microvessels resulting in a perfusable kidney organoid-on-chip model. To create this model, kidney organoids are seeded in one channel, while endothelial cells are seeded and formed into a confluent monolayer in the other channel. Both channels are surrounded by an ECM and are individually addressable by flow. We find that endogenous endothelial cells migrate from the vascularized kidney organoids through the ECM towards the macrovessel, where they form lumen-on-lumen anastomoses. Both fluorescently labeled dextrans of varying molecular weights and red blood cells were introduced through the macrovessel, perfused through the interconnected microvessel network, and transported to renal glomerular epithelia. In summary, we have created vascularized, immune-infiltrated, and perfusable kidney organoidon- chip models for drug testing, disease modeling, and fundamental studies of renal development. Looking ahead, the integration of these three capabilities would result in the most sophisticated, physiologically relevant model of the human kidney to date.