The Unique Ontogeny Of Natural Killer Cells As Revealed By Genetic Barcoding In The Nonhuman Primate Model

Natural killer (NK) cells are cytotoxic leukocytes defined in humans and rhesus macaques by the expression of CD56 and/or CD16, and the absence of T, B, and myeloid markers. Functionally, they are defined by lysis of tumor targets in a major histocompatibility complex (MHC)–unrestricted fashion, and production of cytokines critical to innate immunity. Ex vivo expanded NK cells are being developed as adoptive transfer therapeutics for a variety of malignancies. However, the ontogeny and hematopoietic lineage relationships of human NK cells have been difficult to study, given differences in phenotype and function between human and murine NK cells, poor NK engraftment and development in human xenografts, and restricted development of NK cells from hematopoietic stem and progenitor cells (HSPC) in vitro. In addition, several phenotypic subsets of NK cells have been described, and their relationships and relative locations within the hematopoietic hierarchy are supported by limited in vivo data. We “barcoded” individual CD34+ HSPC from 3 macaques using lentiviral vectors carrying highly diverse oligonucleotides, allowing precise quantitation of clonal contributions via low cycle PCR amplification followed by high-throughput sequencing and computational analysis (Lu et al, Nat Biotech). We quantitated barcode levels, and thus individual HSPC clonal contributions, over time and between lineages, following TBI and infusion of autologous barcoded CD34+ cells. Reconstitution of the NK compartment was clonally distinct from T, B and myeloid (My) lineages. Following reconstitution of NK, T, B and My from uni-lineage progenitors at 1-2 months(m), by 3-4 m, as expected from murine and in vitro models, multi-lineage clones began to contribute and dominate, first B/My, then B/T/My. However, NK cells remained clonally distinct through 9 m, despite overall clonal diversity and marking levels in NK lineage similar to B/T and My. Even the most abundant clones in the NK lineage were not found contributing in any significant way to B/T or My lineages. Shared B/T/My/NK clones finally began to contribute more extensively at 9-14 m. We fractionated peripheral blood NK cells into 3 NK subsets and compared them to overall PB and lymph node NK cells at 4.5-6.5 m. The majority (75-90%) of PB rhesus NK cells are found in the CD16+/CD56- PB subset (corresponding to human CD16+/CD56^dim cytotoxic NK), and this subset accounted for the clonal pattern in overall NK cells (Pearson r=0.98 vs overall NK cells), and the disparity from B/T/My, with r values all <0.17 for CD16+/CD56- vs B/T or My lineages at the same time point. The minor PB (but dominant lymph node) cytokine-producing CD16-/CD56+ or CD16+/CD56+ NK subsets, previously hypothesized to be precursors for CD16+/CD56- NK cells, had clonal patterns that were more closely correlated with T/B/My lineages (r=0.37-0.62) than the CD16+/CD56- cells. The clonal correlations between the putative precursor CD16-/CD56+ cells and mature cytotoxic CD16+/CD56- cells was very low, only r=0.08. Furthermore, we expanded purified NK cells from barcoded macaques in vitro in the presence of IL-2 and irradiated EBV-LCL cells and assessed NK phenotype, function, and clonal diversity over time. 40-90X expansion was achieved by 15 days and was highly polyclonal, with 90% of the starting number of barcodes still present at the end of expansion. Clonal contribution levels between pre and post-expanded NK were highly correlated (r=0.73). CD16+/CD56+ cells became more dominant in post-exp NK cells, in contrast to the majority CD16+CD56- pre-exp. The clonal contributions in the post-exp CD16+/CD56- and CD16+/CD56+ cells correlated with each other (r=0.66), and with the starting CD16+/CD56- cells (r=0.75), but not with the starting CD16-/CD56+ (r=0.15) nor the post-exp CD16-/CD56+ cells (r=0.18). Our in vivo and in vitro results call into question the hypothesis that CD16-/CD56+ NK cells are precursors of circulating cytotoxic CD16+/CD56- NK cells, as does the complete deficiency of CD16-/CD56+ but not CD16+/CD56^dim cells in GATA2 mutant patients. Our results also reveal that the dominant blood rhesus NK cell population has a distinct ontogeny in macaques, and thus potentially in humans. NK cells expanded ex vivo using EBV-LCL cells are polyclonal, and barcoding allows dissection of events during expansion, and will permit tracking studies of expanded cells following adoptive transfer.