A Modality-Agnostic Potency Assay Enabling Both Ex Vivo and In Vivo Genome Editing Therapeutics for Sickle Cell Disease

ABSTRACT / PROJECT SUMMARY Sickle-cell disease (SCD) is an autosomal recessive disorder that causes considerable morbidity and mortality, affecting an estimated 100,000 individuals in the US, and millions more worldwide. Multiple editing-based cures for SCD are currently in clinical development, however, there are no clinical-grade laboratory tests available capable of characterizing the biophysical and rheological properties of RBCs derived from genome-edited SCD HSPCs. Assays capable of characterizing RBC quality are urgently needed to assess the potency of emerging editing-based genomic therapies for SCD, regardless of the editing modality. One of the central challenges that has impeded the development of a highly performant potency assay for evaluating the functional efficacy of editing-based genomic therapies for SCD has been the lack of laboratory technologies capable of sensitively, accurately, and precisely capturing the biophysical properties of SCD-RBCs utilizing only a small number of cells. In recent years, several innovations have emerged that now make the development and analytical validation of a potency assay for editing-based SCD genomic therapies feasible. One of these has been the advent of microfluidics-based diagnostic devices capable of functionally characterizing the health of RBCs at unprecedented levels of resolution and sensitivity. This proposal seeks to leverage (1) an existing suite of these aforementioned next-generation RBC biophysical and functional characterization devices, (2) conventional hematologic assays, and (3) a well-established machine learning approach to develop and analytically validate a first-in-kind potency assay for editing-based therapies for SCD. To achieve this, we will first construct a panel of comprehensively profiled, gold-standard reference samples of HSPCs that simulate a representative range of genome editing outcomes in SCD and prepare data for machine learning training. A machine learning model will then be trained to predict the percentage of RBCs that functionally exhibit a non-SCD phenotype. Once trained, we will validate the performance of the new potency assay, a panel of HSPCs affected by SCD will be therapeutically edited using at least three different modalities (e.g. homology-directed repair, base-editing, etc.).