Caractérisation des variants manquants dans les maladies rénales héréditaires et développement d'une thérapie basée sur l'ARN dans les organoïdes rénaux

Chronic kidney disease (CKD) affects approximately 10% of the general population. Despite the prevalence of CKD, its cause remains unknown in at least 20% of the patients. Treatment options are also limited due to the complexity underlying kidney biology and the lack of in vitro models that faithfully mimic the mature kidney. My research focused on addressing two main objectives. First, we aimed to improve the genetic diagnosis for patients with CKD, and secondly, we developed a robust in vitro kidney organoid model to better understand the disease and validate new therapeutic strategies. First, given the challenges in genetically diagnosing patients with tubulointerstitial disease associated with pathogenic variants the MUC1 gene (ADTKD-MUC1), I developed a computational tool to genotype the coding variable number tandem repeat (VNTR) of the MUC1 gene. I introduced VNtyper, a Python tool utilizing an alignment-free genotyping method for fast and accurate genotyping of the coding-VNTR. I analyzed targeted sequencing data from two cohorts: 108 index cases with previously detected MUC1 variations and 4040 patients tested with a hereditary renal disease panel. VNtyper re-diagnosed patients with unknown pathogenic variations in the first cohort and diagnosed 35 new patients previously overlooked. The second hereditary kidney disease we focused on is X-linked Alport syndrome (XLAS), characterized by glomerular basement membrane (GBM) disruption a leading to podocyte injury. It arises from genetic variations in the COL4A5 gene, encoding the '5 chain of the collagen IV ['5(IV)]. Clinical manifestations are hematuria, proteinuria and kidney failure. '5(IV) associates with '3(IV) and '4(IV) chains, encoded by COL4A3 and COL4A4 respectively (responsible for autosomal forms of the disease), to form heterotrimers that are crucial for the structure and function of the GBM network. By applying targeted RNA-seq approach to 19 patients with a typical AS phenotype and no identified pathogenic variants, we evidenced splice variants in all patients. Since these variations are amenable to antisense oligonucleotide therapy (ASO), I designed multiple ASOs and tested them in patient-derived fibroblasts. To achieve the second objective of my research, I applied CRISPR/Cas9 technology to independently introduce two deep-intronic variations and a V5-tag in the COL4A5 gene in male human induced pluripotent stem cells (hiPSCs), and differentiate them into kidney organoids. I developed two XLAS kidney organoid models: one with a severe mutation resulting in the presence of a mutant transcript and the total absence of the wild-type transcript, and another with a mild mutation allowing the residual expression of the wild-type transcript together with the mutant transcript. To quantify mutant and wild-type transcripts of COL4A5 in the organoids, I carried out targeted RNA-seq and fragment analysis. To further characterize the developmental state of our organoid model throughout the differentiation time course, single-cell RNA-seq and proteomics studies were performed on isogenic control and mutant organoids at day 22 and 38, to monitor organoid maturation, and particularly GBM maturation. Immunofluorescence studies of collagen IV ['1 to '6] and laminin ['5, '1, and '2] chains confirmed the maturation status of the GBM. We also developed an AI-assisted Fiji macro to quantify the variations of '5(IV) chain expression in the basal membranes. Finally, since 2'O-méthyl (2'OMe) ASO delivery to the kidney organoid model was not very efficient, I switched to phosphoroamidate morpholino oligomer (PMO) backbone while keeping the sequence, and developed a framework for efficient transfection and restoration of canonical splicing in the kidney organoids. Using this protocol, I successfully restored COL4A5 protein expression and localization in the BMs of kidney organoids.