T cell resistance to glucocorticoids in graft-versus-host disease

Abstract Glucocorticoid (GC)-resistant, acute graft-versus-host-disease (GVHD) is a major cause of mortality following allogeneic hematopoietic stem cell transplantation. Second line treatment with Ruxolitinib (RUX) can be effective, but only a minority of patients attain durable complete remissions. Biopsies from the gut in patients with refractory GVHD are frequently characterised by large clonal expansions of CD8+ T cells. Experimental models of GVHD have suggested that these T cell effector expansions may be sustained by progenitor-type T cells expressing the transcription factor TCF1, which possess greater inherent proliferative and self-renewal potential than effector cells. We hypothesised that a similar progenitor-type population can be found in humans and is resistant to GC treatment. Using clinically relevant concentrations of methylprednisolone and several independent methods for human T cell stimulation, we determined output population architecture and function with a combination of multiparameter flow cytometry, bulk and single cell transcriptional profiling. Under conditions of repetitive anti-CD3/CD28 stimulation over a three-week period, GC-treated CD8+ T cells showed greater proliferation and overall expansion than control cells. GC-treated CD8+ T cells showed greater retention of cell markers (e.g., IL-7RA, CD27) and transcriptional profiles (e.g., KLF2, LTB, BCL2) associated with a 'less-differentiated’ state. In vitro assays indicated intact cytokine-generating function and killing capacity of the GC-resistant CD8+ T cells versus controls. These GC-driven changes to T cell proliferation and phenotype were largely restricted to CD8+T cells and occurred independently of the presence of CD4+ T cells. To track the population architecture of GC-treated CD8+ T cells compared to controls, we performed scRNAseq of 3-week expanded cultures. The GC-treated CD8+ T cell population showed a distinct differentiation trajectory composed primarily of two IL7RA-expressing clusters: a major effector memory-like cluster expressing cytotoxic genes (e.g., PRF1) as well as known GC-regulated genes (e.g., GILZ); and a smaller cluster composed of progenitor-like cells, expressing TCF7 (encoding TCF1) and LEF1. To test how this GC-induced shift in population architecture emerged during culture, we tracked cell proliferation and cellular state according to the individual input differentiation status of CD8+ T cells (TN, TCM, and TEM(RA)) prior to activation; these experiments showed that GC-induced changes in proliferation and cell state were primarily derived from the expanding TN-origin cells. To test the clinical relevance of these findings in vivo, we interrogated patient scRNAseq data derived from n=19 biopsies of acute lower gut GVHD, treated with (n=14) or without GC (n=5) at the time of biopsy. We identified strong enrichment for our TCF-1+ progenitor-like cluster in patients with severe GVHD treated with GC (median 35.1% versus 9.5% of CD8+ T cells, p<0.05). To address how JAK1/2 inhibition would influence the GC-resistant phenotype, we applied RUX at clinically relevant concentrations and showed that its addition blocked proliferation in GC-treated cultures to a greater extent than control cells. However, RUX simultaneously further enriched for the TCF-1hi IL-7Rahi progenitor-like population which additionally displayed elevated expression of the anti-apoptotic molecule, BCL-2. We reasoned that BCL-2 inhibition would have the potential to separately target the progenitor population. To test this hypothesis, we treated control and GC-treated T cell cultures with clinically relevant concentrations of Venetoclax (VEN), alone or in combination with RUX. VEN treatment alone did not block proliferation but led to reduced frequencies of progenitor-like cells arising in the presence of GC. When combined, VEN and RUX led to both reduced proliferation and reduced progenitor frequency in parallel to increased activated Caspase-3 expression by CD8+ T cells, particularly in the presence of GC. Taken together, these data show that treatments used for GVHD including GC and RUX may select for progenitor-like CD8+ T cells with the potential to sustain tissue injury and lead to refractory disease. Strategies combining drug targets which can inhibit T cell proliferation whilst preventing the retention of multipotent and proliferative progenitor populations may be required to overcome the limitations of current GVHD treatment strategies.