Ontogeny As a Critical Determinant of Natural Killer Cell Potential and Function

Hematopoietic development during mammalian embryogenesis is comprised of a restricted primitive program of primitive erythroid, megakaryocytic, and macrophage lineages, and a definitive program of definitive erythroid, myeloid and lymphoid potential emerging from hematopoietic stem cell (HSC)-independent and dependent processes. Interestingly, progenitors of natural killer (NK) cells, but not B- or T-cells, have been found in the early human yolk sac, suggesting that NK cells may arise from HSC-independent sources.

NK cells recognize and kill virally infected cells and tumor cells, making them a highly desirable cell-type for adoptive immunotherapy. To bypass donor-related issues, human pluripotent stem cell (hPSC)-derived NK cells offer the possibility of uniform activity in a renewable "off-the-shelf" cell product. As the differentiation of hPSCs recapitulates early developmental processes, we sought to characterize the developmental origin of hPSC-derived NK cells. Studies in mice indicate that NK cells in the adult are derived from hematopoietic stem cells (HSCs) that commit to a lymphoid differentiation pathway. However, while NK cells, like HSCs, have been found in the fetal liver, the developmental origin of the fetal NK cell lineage remains poorly understood.

We have developed a stage-specific hPSC differentiation method that separates WNT-independent (WNTi) hematopoietic progenitors that harbor "primitive" hematopoietic potential from WNT-dependent (WNTd) erythro-myeloid-(T-)lymphoid "definitive" hematopoietic progenitors. Using this system, we find that CD34+ cells from both populations harbor NK cell potential. NK cells from hPSC WNTi progenitors (WNTi-NK cells) mature rapidly, are significantly more granular, and express very high levels of CD16 in comparison to their hPSC WNTd counterparts (WNTd-NK cells) and cord blood-derived NK (cbNK) cells. Further, WNTi CD34+ progenitors always gave rise to a granulocyte population alongside NK cells, suggesting they may be derived from a myeloid progenitor. Both WNTi-NK and WNTd-NK cells robustly respond to tumor targets, antibody-dependent cell-mediated cytotoxicity (ADCC), and PMA/ionomycin stimulation in comparison to cbNK cells. In all cases, WNTi-NK cells exhibited a strong bias for cytolytic degranulation over cytokine production, while WNTd-NK cells were biased for IFNg secretion. Similarly, WNTi-NK cells exhibit superior ADCC-mediated cell killing of Raji cells.

We then turned to the well-characterized murine embryo to determine whether HSC-independent NK cell progenitors are developmentally conserved. Assessing NK cell potential via explant culture, we found that as early as E7.5, yolk sac explants give rise to NK cells, as well as primitive and definitive erythroid progenitors. Further, we find that murine E9.5 yolk sac kit+CD41+CD16/32+ erythro-myeloid progenitors (EMP) give rise to NK cells ex vivo. Similar to hPSC WNTi-NK cells, EMP-derived NK cells were larger and more granular, and emerged alongside a granulocyte population in explant culture. Thus, the murine yolk sac harbors unique NK cell potential, from a committed myeloid progenitor, prior to HSC emergence.

Collectively, these studies suggest that ontological origin is an unexpectedly important consideration in the design of hPSC-derived NK cell-based therapeutics, and raise new questions regarding the potential of early hematopoietic progenitors in the mammalian embryo.