Phenotyping of ex vivo expanded NK cells reveals unique transcriptomic and epigenetic signatures

Abstract Background: Natural Killer (NK) cells have powerful anti-tumor effects. NK cells expanded for adoptive transfer are commonly co-cultured with irradiated “feeder” cells expressing membrane bound IL-15 or IL-21 plus supplemental cytokines to sustain an activated phenotype. There is no currently accepted standard for NK cell activation. While others have shown activation method-dependent modulation of NK cell surface receptor expression and gene expression, DNA methylation changes have not been extensively studied. Therefore, we define the unique immunophenotype, transcriptional signature, and methylation patterns of clinically relevant NK cell expansion methods with a goal of optimizing ex vivo NK cell manufacturing to produce exceptional products. Methods: From human peripheral blood mononuclear cells, CD3+ cells were depleted. The CD3- cell population was cultured with irradiated K562.mbIL15.41BBL or K562.mbIL21.41BBL cells and supplemented with IL-2 alone (NK.IL15, NK.IL21) or IL-2 + TGFβ (NK.TGFB), and cultured for 10 days. Purity was verified using flow cytometry and fold expansion measured with manual counting. A multiparameter flow (MPF) panel was used for characterization, including activating (CD56, NKp46, NKG2C, CD16, NKG2D, DNAM, NKp30, CD69), inhibitory (NKG2A, TIGIT), exhaustion (CD96), maturation (CD57), and trafficking (CD62L) NK-cell receptors. DNA and RNA were isolated. Whole genome bisulfite sequencing (WGBS) was performed and differentially methylated regions detected using BSmooth. Bulk RNA sequencing was performed and analyzed using DESeq. Cell products for scRNAseq were stained with oligo-tagged antibodies, multiplexed, processed with 10x Chromium, sequenced, and analyzed using an institutional pipeline. MPF, bulk RNAseq, and WGBS analyses were replicated using three independent cell expansions with unique NK cell donors. sc-sequencing was performed using a single donor. Results: There were no differences in NK cell purity or expansion. We distinguished NK cell subsets by MPF and found a majority of non-activated NK cells were CD56(dim)CD16+ with expanded NK cells as mostly CD56(bright) with variable CD16 expression levels. Expanded NK cells had generally higher expression of activation markers (NKG2D, DNAM, NKp30, CD69), TIGIT and CD96 compared to non-activated. Transcriptional expression of surface receptors correlated with measured protein expression for each NK cell condition. We identified expansion method-dependent differences in gene expression and global DNA methylation. NK.IL15 cells upregulated pathways associated with cell proliferation, survival, cytotoxicity, and the inflammatory cytokines, IL2 and IL8. NK.IL21 cells upregulated NFAT transcriptional programming, c-MYB and IFNα/γ signaling, and pathways associated with IL2, IL3, IL4, IL8 and IL17. NK.TGFβ cells upregulated JAK-STAT, NF-κβ and Wnt/β-catenin signaling, and pathways associated with IL2, IL8, and IL17. All expanded conditions had activation of the PI3K/AKT/mTOR pathway, important for NK cell proliferation and functionality, with unique driver genes per activation method (NK.IL15: VAV3, PLCB1; NK.IL21: NTRK2, IL2RA, MYB; and NK.TGFB: CSF1R, ITGA1,GNG8, IL7R). Eight unique NK cell clusters were identified by scRNAseq with activated NK cells as a whole with enrichment of clusters (cl) 1-4 characterized by DNA repair, oxidative phosphorylation, and immune activation pathways. NK.IL15 had expansion of cl.1, characterized by upregulation of the PDGFR-β pathway. NK.IL21 had enrichment of cl.2, with upregulation of MYC- and cell cycle-related pathways, and cl.5, with upregulation of mitochondrial genes. NK.TGFB was enriched in cl.3, defined by upregulation of PLK1 and Aurora B pathways. WGBS revealed genome-wide hypomethylation in expanded NK cells as compared to non-activated NK cells. All expanded NK cells had differentially methylated regions in CISH, NFATC2 and PIM1, genes important for immune activation. We observed unique methylation patterns in the immune regulatory genes MAPK1 and CRKL for NK.IL21 and NK.TGFB products. Conclusions: The interplay of multiple signaling pathways can comprehensively define the NK cell post-activation phenotype. Each NK cell expansion method stimulates unique drivers of immune cell activation characterized by transcriptomic and epigenetic signatures that, once related to functional behavior, can help guide development of an optimal ex vivo NK cell expansion method for cancer immunotherapy.