Although pilot clinical trials have shown adoptive NK cell transfer can result in tumor regression in humans with cancer, additional insight from animal models are needed to optimize NK cell proliferation in vivo as well as to improve their homing and retention in the tumor microenvironment. Mice have provided fundamental insights into NK cell biology, although significant divergence from humans as a consequence of evolution limits their usefulness to optimize adoptive transfer of ex vivo expanded NK cells in humans. Rhesus macaques (RM) express orthologues to most human MHC class I and II genes, and unlike mice, express KIRs and CD56, making them phenotypically and functionally similar to human NK cells. Further, unlike murine NK cells, cultured macaque NK cells can be expanded ex vivo for prolonged periods of time augmenting their cytotoxicity. Although recent studies have provided detailed characterization of freshly isolated RM NK cells, little information exists on the phenotype and function of these cells following ex vivo expansion. We investigated the feasibility of expanding RM NK cells ex vivo using culture conditions similar to those used to expand clinical grade human NK cells and characterized the cytotoxic function of subsets of these cells based on CD16 and CD56 expression.

NK cells were enriched from RM PBMCs by either bead immunodepletion of CD3+ cells or by flow sorting) CD3-/CD20-/CD14-/CD16+ single positive (CD16SP) or CD3-/CD20-/CD14-/CD16-/CD56+ single positive (CD56SP) or CD3-/CD20-/CD14-/CD16+/CD56+ double positive (CD16/56DP) NK cells. NK cells were expanded in vitro by co-culture with an irradiated human EBV lymphoblastoid cell line (EBV-LCL) in media containing 500IU/uL of human IL-2. In contrast to humans, where resting NK cells from the peripheral blood are mostly CD16/56DP with smaller SP populations, the majority of NK cells from RM peripheral blood were CD16SP (Figure). Despite this difference, RM NK cells appeared phenotypically similar to human NK cells following ex vivo expansion, and had similar expansion kinetics, typically yielding a 100-fold increase in cells that were 99% CD3- and mostly CD16/56DP or CD56SP with a minor fraction of CD16SP cells. Using NKG2A as a pan NK cell marker as well as expression levels of CD16 and CD56, we evaluated for changes in the cytotoxic function associated with ex vivo expansion in 5 distinct NK cell subsets; CDSP16bright, CD16SPdim, CD56SPbright, CD56SPdim, and CD16/56DP. After 7 days expansion, CD16SP populations contracted substantially in contrast to CD16/56DP and CD56SP populations which increased. After 14 days of expansion, the CD16SP population represented <5% of the expanded NK cells, while the CD16/56DP and the CD56SP populations represented approximately 60% and 40% of NK cells respectively, with distinctions between CD16 and CD56 bright and dim populations being lost (figure). To assess the cytotoxicity of specific subsets of expanded vs fresh NK cells, we evaluated CD107a expression at baseline and following co-culture with K562 cells (2hrs;1:1 ratio). Among fresh NK cells, the highest induction of CD107a was in the CD16SPdim subpopulation. In contrast, following ex vivo expansion, CD107a induction was highest in the dominant CD16/56DP population followed in decreasing order by the CDSP16bright and CD56SPbright populations. To confirm these data, we assessed target killing by expanded NK cells using a flow-based cytotoxicity assay measuring Annexin V and 7AAD expression on K562 cells following co-culture with one of 3 different flow-sorted NK populations; a) CD16SP or b) CD56SP or c) CD16/56DP NK cells. Similar to prior data, we found the dominant CD16/56DP populations to have the greatest cytotoxicity against K562 cells (25%) followed by the CD16SP (17%) and CD56SP (15%) populations.


These data show RM NK cells can be expanded ex vivo using culture conditions similar to those used to expand clinical grade human NK cells. RM NK cells expanded using EBV-LCL feeder cells are phenotypically similar to expanded human NK cells and based on CD16 and CD56 expression, contain subpopulations that have varying degrees of cytotoxicity against K562 targets. The similarities between expanded RM and human NK cells establish this animal model as having excellent potential to optimize adoptive NK cell transfer in humans. We aim to establish this model and test expansions in vivo.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.