Development of a canine model to evaluate CAR-T cell subsets

Abstract

Immunotherapy is a rapidly expanding therapeutic modality in the oncology clinic. Immunotherapy with chimeric antigen receptor (CAR) T cells has demonstrated success in human blood cancers, although relapse occurs in ~50% of patients. In the context of solid tumors, CAR-T cells show only transient effects. As a "living therapy", development of the next generation CAR-T cells requires a deeper understanding of how CAR biology, T cell biology, and tumor biology interact over the course of disease. CAR constructs are made up of an extracellular antigen binding domain fused with one or more intracellular signaling domains. The extracellular domain is frequently built from the fused heavy and light chains of an antibody, called a single chain fragmented variable (scFv) domain, and confers the CAR its tumor specificity. Intracellular signaling domains typically contain the zeta chain of CD3 along with one or more costimulatory domains, often CD28 and/or 4-1BB. Mechanisms of relapse and treatment failure for CAR-T cell therapy are related to loss of CAR-T targeting (e.g. target antigen loss) and dysfunction of CAR-T cells (e.g. immunosuppressive microenvironment). These mechanisms of resistance to CAR-T cell therapy have largely been resolved using in vitro and mouse models, where methods to overcome resistance have also been developed. While our mechanistic understanding of tumor and immune biology has been propelled to a place where beneficial therapies are being employed, there is a need for more clinically relevant models to help streamline clinical trials of promising next-generation therapies. Dogs provide an immune-intact animal model that can help address some fundamental questions of CAR-T cell biology in a clinically relevant animal model. Cancers in dogs share many pathobiological features to their human counterparts and receive similar treatments including chemotherapy, radiation, and surgery. Dogs' intact immune system, large size, and relatively outbred genetics provide an ideal environment to evaluate novel cellular therapies such as CAR T cell immunotherapies. This work sought to develop a model of CAR-T cell therapy in dogs to better support pre-clinical evaluation of novel CAR constructs and to better understand how CAR and T cell biology interact in the context of cancer. The first aim of this dissertation was to develop a canine CAR-T model for functional assessments of canine CAR-T cells. The second aim of this dissertation was to explore cell subsets using single cell sequencing. We first sought to establish a canine CAR-T cell in vitro model system. We hypothesized that the disialoganglioside GD2 would be a viable CAR tumor target in canine osteosarcoma (OS). The primary CAR used in these studies was a GD2 directed CAR which is currently undergoing human clinical trials. Using the same antibody clone from which the GD2 CAR is derived, we demonstrated that GD2 is expressed on a subset of canine OS and melanoma cell lines. We demonstrated that canine primary T cells are refractory towards lentiviral (HIV-1) based transduction, although they are susceptible to gammaretroviral (MSCV) transduction. Canine GD2+ cell lines were susceptible to GD2 CAR-T cells as assessed by IFN-g and IL-2 cytokine release, Incucyte live image microscopy, and luciferase killing assays. GD2 CARs with either CD28 or 4-1BB costimulatory domains were effective in killing GD2+ OS. The second aim of this proposal sought to elucidate and define canine immune cell subsets using single cell RNA sequencing (scRNASeq). We hypothesized that a phenotypically distinct population of CAR-T would preferentially expand and persist throughout CAR-T cell manufacturing and tumor engagement. To the aim of identifying specific immune subsets, we first sought to create an atlas of cells present in the canine lymph node (LN). In addition to peripheral blood and primary tumor lymphocytes, LNs are an additional key site for modulating cancer immunity. While the immune subsets in the peripheral blood and tumor have been described in dogs with OS, an atlas of canine LNs has not been made. We generated an immune atlas of cells with the LNs from healthy dogs and the regional LNs of OS bearing dogs. In addition to annotations of cell clusters present in l LNs, comparisons between dog populations revealed fewer B cells recovered in the OS group along with higher HLA-DQB2 gene expression across a variety of cell types. Taken together, these aims provide a basis for investigation of CAR-T cell therapy at single cell resolution. In the first aim, we developed canine GD2 directed CAR-T cells which could effectively kill GD2+ tumor in vitro. The second aim developed a workflow using scSeq to identify canine immune cell populations and provides a LN cell atlas to supplement those for the peripheral blood and tumor tissue. This platform will be used for translational oncology studies to assess novel CAR constructs, investigate CAR-T cell biology, and allow for optimization of next generation CAR-T cells.