Bone marrow CD57+ LGL cells comprised of NK and CD8 T cells drive bone marrow failure in Myelodysplastic Syndromes

Abstract

Abstract Background: Autoimmune phenomena occur in 20-50% of patients with myelodysplastic syndromes (MDS), exacerbating hematopoietic suppression and increasing cardiovascular/infection/bleeding risks. Recent data from our group suggest that the canonical model of stem cell intrinsic failure in MDS is incomplete in that a major fraction of MDS patients present with a large-granular-lymphocyte (LGL) phenotype, comprised of clonal-like expansions of CD57+ T-cells and NK cells in the bone marrow (BM) that strongly correlate with autoimmune cytopenias in lower-risk (LR)-MDS. We hypothesized that CD57+LGLs contribute to BM failure (BMF) through three key mechanisms: (i) enhanced expression of the Fc receptor, CD16A and associated reactivity in response to IgG1 and IgG3 autoantibodies; (ii) ability to recognize loss of HLA class I expression on hematopoietic progenitors; and (iii) activation and proliferation in response to cytokines and chemokines. Targeting these cells has the potential to restore effective hematopoiesis. Methods: Clinical flow cytometry was used to profile presence and abundance of cells across >20 immune subsets in matching BM and blood samples (N=279 MDS). We then performed scRNA-seq, including matched TCR as well as single-nucleotide ATAC seq on BM mononuclear cells (BMMCs) from a separate cohort of six LR-MDS pts and three healthy individuals. Further, we performed mass cytometry (CyTOF) for ex-vivo profiling of BM from 19 MDS patients (N=10 LR/IR-MDS, N=9 high-risk-MDS) and age-matched controls. From this same cohort, we also performed in vitro killing assays by incucyte using isolated BM-derived LGLs. Retrospective analysis of 22 MDS patients treated with intravenous immunoglobulin (IVIG) for autoimmune cytopenias and LGL expansion was performed. TCR clonality was assessed via TCR-seq in a subset. Results: Clinical flow analysis reveal a strong association with BMF with expanded BM NK cells (>20% of lymphocytes) and/or circulating NK-LGL or CD8-LGLs (p<0.00001). Strikingly, their abundance was inversely correlated with myeloblast burden (p<0.05). In parallel, we show a major fraction of MDS patients have high abundance of BM-derived CD57+CD8 T cells and NK cells by CyTOF analysis. Further characterization by scRNA-seq and TCR-seq reveal expansion of conventional T cells or CD57+LGLs that bear a profound phenotypic relation to innate-like cells (e.g., expression of NKG2A and/or KIR receptors, CD16A, NKG2D, etc). We compared innate-like CD8 T cells between age-matched controls and LR-MDS patients, and observed a clear qualitative difference in their functional status, where LGL CD8 T cells from LR-MDS patients were strongly activated and expressed significantly higher levels of perforin, granzymes, cytokines and chemokines compared to the same subsets of cells from controls (p<0.0001). Finally, we co-cultured patient BM-derived LGLs with autologous BM cells and demonstrated in vitro killing uniquely in the presence of autologous BM-derived IgG, revealing antibody-dependent cellular cytotoxicity (ADCC) reactivity and an unappreciated axis of autoantibodies engaging LGLs as a clear mechanism of BMF. To test the hypothesis that LGL-mediated ADCC might drive BMF, we conducted a retrospective clinical study of IVIG treated MDS patients with severe BMF and abundance of LGLs (N=22) and observed an ORR=73% (16/22). In a subset of nine patients (n=7 with lower-risk MDS) who had serial blood samples available for TCR-Seq, those with confirmed clonality via TCR-Seq had an 80% hematologic and 60% hemolytic response, whereas patients without detectable clonality had only 25% response. Conclusion: Although the dataset is small and retrospective, it strongly suggests a beneficial effect of IVIG on BMF in patients with expanded NK cells and T cell clones. This clinical response to IVIG is further evidence of the importance of ADCC reactivity and underlying autoimmune signatures in driving BMF, where BM LGLs, comprised of terminally mature CD57+ NK and CD8 T cells, have capacity to kill autologous developing hematopoietic cells. Based on all of our preclinical data, we predict IVIG has a ‘diluting’ effect on the overall abundance of endogenous autoantibodies and their ability to bind antigen and induce ADCC. This strategy to target these cells has the potential to improve peripheral blood counts, patients' quality of life and improved capacity for response to immunotherapies targeting the MDS clone(s).