2025 Theses Doctoral

Dissecting the human immune "variome"

Walsh, Zachary Hudson

Single-nucleotide variants (SNVs) can profoundly shape T cell biology, with far-reaching implications for both cancer immunotherapies and the pathogenesis of inborn errors of immunity (IEI). T cells are the essential substrate for effective cellular immunotherapies, including tumor-infiltrating lymphocyte (TIL) therapy and chimeric antigen receptor (CAR) T cells. While transformative for subsets of patients with solid and hematologic malignancies, most individuals do not achieve durable benefit, underscoring the urgent need to enhance these approaches.

Concurrently, the expansion of next-generation sequencing has uncovered thousands of variants in genes implicated in IEI. Yet, the majority remain “variants of uncertain significance” (VUS), creating diagnostic and therapeutic ambiguity for patients. The central premise of this work is that accurate classification of IEI variants will accelerate precision diagnosis and therapy, and that lessons from these “experiments of nature” can be harnessed to engineer superior T cell therapies.

To address this, we developed and applied massively parallel CRISPR-based base editor screens in primary human T cells, targeting >100 IEI-associated genes and systematically mapping variant effects on hallmarks of T cell–mediated antitumor immunity. This strategy revealed gain-of-function (GOF) variants in genes such as 𝘗𝘐𝘒3𝘊𝘋 and 𝘗𝘐𝘒3𝘙1, whose introduction enhanced the efficacy of diverse T cell immunotherapies. We then conducted saturation base editing screens of 𝘗𝘐𝘒3𝘊𝘋 and 𝘗𝘐𝘒3𝘙1, genes in which GOF variants can cause activated PI3K-δ syndrome (APDS), an IEI characterized by immunodeficiency, lymphoproliferation and autoimmunity. These studies definitively classified hundreds of VUS, enabling accelerated clinical diagnosis and precision therapy with the PI3Kδ inhibitor Leniolisib.

Beyond PI3K signaling, we established a framework for dissecting complex variant-to-phenotype relationships using 𝘊𝘈𝘙𝘋11, a central TCR signaling transducer. Employing new base editing strategies to introduce homozygous and heterozygous mutations, we coupled variant status with cellular phenotypes using single-cell allele-integrated multi-omics (sc-AIMseq). These studies revealed how autosomal random monoallelic expression may govern variable penetrance and expressivity in patients and families. Strikingly, selected 𝘊𝘈𝘙𝘋11 variants also promoted favorable T cell states that can be harnessed for therapeutic engineering.

Together, these studies illustrate how systematic interrogation of human genetic variation bridges fundamental insights into monogenic disorders with translational opportunities for precision genome engineering. The methodological framework we developed provides a blueprint for extending such approaches well beyond immunology and cancer, with broad potential across medicine.