Back

Scalable expansion of human iNKT cells: single-cell profiling and in vivo control of GvHD with preserved GvL activity

Brouard, J.; Caraiman, C.; FIEVET, G.; Monchablon, L.; Coman, T.; Hergalant, S.; Pagliuca, S.; MOULIN, D.; Rubio, M. T.

2026-09-16 immunology
10.64898/2026.09.11.750594 bioRxiv
Show abstract

Invariant natural killer T (iNKT) cells can limit graft-versus-host disease (GVHD) after hematopoietic stem cell transplantation (HSCT), but their scarcity in peripheral blood limitstherapeutic development. Current clinical-grade human iNKT expansion protocols mainly rely on IL-2, require prior iNKT-cell sorting, last 6-8 weeks, and predominantly expand CD4+ iNKT cells, whereas human CD4- iNKT cells are more strongly associated with GVHD control in patients and uniquely regulate antigen-presenting cells and T-cell activation. We developed a scalable culture system to preferentially expand human CD4- iNKT cells directly from total peripheral blood mononuclear cells (PBMCs) using alpha galactosylceramide (-GalCer) and optimized cytokine conditions. IL-15 was the most effective cytokine. The optimized 14-day protocol generated a mean of 3.8x107 iNKT cells from 2x107 PBMCs, including 74% CD4- iNKT cells. Single-cell transcriptomic profiling identified eight major iNKT subsets, differentiation trajectories during expansion, and distinct IL-2- versus IL-15-associated transcriptional programs. IL-15-expanded iNKT cells induced apoptosis of monocyte-derived dendritic and leukemic cells in vitro, controlled xeno-GVHD, and preserved graft-versus-leukemia (GVL) activity in preclinical mouse models. This platform enables reproducible production of human CD4- iNKT cells at clinically relevant scale and position IL-15-expanded iNKT cells as a compelling immunotherapy candidate for allo-HSCT.

Matching journals

The top 14 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.