Description

Single cell RNA-sequencing using droplet microfluidics reveals co-evolutionary dynamics of leukemic-immune subpopulations during response and resistance to immunotherapy

Pavan Bachireddy^1,2, Vinhkhang Nguyen^1,2, Katherine G. Tooley^1,2, Ilke Akartuna³, David A. Weitz³, Peter V. Kharchenko² and Catherine J. Wu^1,2

¹Dana-Farber Cancer Institute, Boston, MA; ²Harvard Medical School, Boston, MA;

³School of Engineering and Applied Sciences, Harvard University, Cambridge, MA

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is one of the earliest forms of successful cancer immunotherapy in patients, curing many hematologic malignancies. Unfortunately, very few therapeutic options exist for patients whose disease recurs after allo-HSCT. Used over 20 years, donor lymphocyte infusion (DLI) is a potentially curative human immunotherapy for these patients, with a spectrum of response rates across various leukemias and lymphomas. To identify biologic response predictors and elucidate the basis of DLI effectiveness, we recently analyzed bone marrow infiltrating T cells and identified gene expression signatures of T cell exhaustion that predicted DLI response. However, critical questions remain regarding the specific identities of the unique T cell subsets that are exhausted as well as the molecular states of immune cells associated with resistance. To systematically interrogate the co-evolutionary dynamics between heterogeneous leukemic and infiltrating immune subpopulations during periods of response and resistance, we have used a droplet microfluidics platform, inDrops, to comprehensively evaluate the transcriptomes of 36,000 single cells from paired peripheral blood and bone marrow samples before and during DLI response as well as 11 years later during DLI relapse in a single patient. These cryopreserved, primary human samples afforded a spatiotemporal analysis of incipient response and subsequent relapse to DLI within the same patient, between two tissue compartments, over a decade. Focusing the analysis on CD8+ T cells from both timepoints, statistical clustering methods revealed robust separation between CD8+ T cells during CLL response and resistance, suggesting broad shifts in the transcriptional programs of CD8+ T cells associated with different clinical outcomes. The major shifts from ‘response’ to ‘resistance’ phenotypes in gene expression reflected silencing of genes promoting metabolic and chromatin reprogramming for T cell effector memory differentiation and proliferation as well as upregulation of genes promoting T cell dysfunction. Thus single cell transcriptome analysis can reveal novel dynamic changes in immune cell heterogeneity during immunotherapeutic response and resistance.