Clinical next-generation sequencing (NGS) is a pillar of diagnosis for inborn errors of immunity (IEI). However, for each pathogenic variant identified, these methods yield hundreds of variants of uncertain significance (VUS), creating ambiguity in clinical management. Here, we present a high-throughput framework to precisely generate genetic variants at endogenous loci and map them with the clinically established functional readout. Loss-of-function (LOF) variants in PIK3CD or PIK3R1 can lead to immune deficiency and/or the multisystem SHORT syndrome, while gain-of-function (GOF) variants lead to lymphoproliferation, autoimmunity, and infection—the activated PI3K-delta syndrome (APDS). In primary human T cells from multiple healthy donors, we performed saturation CRISPR base-editor screening of PIK3CD and PIK3R1 coupled with the clinical APDS-diagnostic flow cytometric assay to measure phosphorylated AKT and S6 after T cell receptor (TCR) stimulation. We successfully detected most known pathogenic variants and found >100 variants which were clearly novel GOF or LOF mutations. We individually validated 30 of these variants and show that many GOF, including those already associated with APDS, are sensitive to leniolisib, an FDA-approved PI3Kδ inhibitor for APDS. Next, we used structural modeling to map variant effects to defined regions in the PI3Kδ protein complex, identifying variant “hotspots” associated with pathogenic AKT/S6 signaling. Finally, we acquired peripheral blood samples from patients harboring germline mutations in PIK3CD or PIK3R1. We find that both exhibit pathogenic AKT/S6 signaling and leniolisib sensitivity at similar effect sizes as our screen-identified novel GOFs, emphasizing the clinical predictive value of our variant classification approach. We leveraged multiple precision genome editing approaches to correct the causative SNVs in these patient samples, improving signaling defects. Thousands of new functional annotations from our screens will be incorporated into public databases and used for variant reclassification, substantially broadening the population of previously undiagnosed patients who can immediately benefit from this precision medicine approach. This proof-of-concept study is part of the Human Immune Variome Project, an effort to functionally classify variants across hundreds of genes implicated in IEIs, to remove ambiguity related to clinical management and treatment.
