PDF, English - main document Restricted access: Repository staff only until 27 November 2026. Login+Download (14MB) | Terms of use

Abstract

Natural killer (NK) cells are gaining increasing attention in cancer immunotherapy due to their reduced side-effect profile compared to T cells and compatibility with allogeneic use. However, achieving optimal persistent genetic modification of NK cells to enhance their anti-tumour efficacy has posed significant challenges. Commonly used retro- and lentiviral transduction approaches face limitations, such as low transduction efficiencies, high production costs, increasing regulatory requirements and the risk of insertional mutagenesis. To address these challenges, our group has created a DNA Vector platform which is devoid of viral components and comprises a minimal bacterial backbone. The vectors do not require integration into the target cell's genome but can replicate extrachromosomally in the nucleus of dividing cells using sequences derived from scaffold/matrix attachment regions (S/MARs). Therefore, they are promising candidates for long-term, stable modification of NK cells while preserving genomic integrity. Given the difficulty of directly engineering primary NK cells, I first introduced chimeric antigen (CAR)-encoding SMAR vectors into induced pluripotent stem cells (iPSCs) and subsequently differentiated them into NK cells. iPSCs offer a versatile platform for genetic engineering and a highly expandable source attractive for NK cell therapies. I optimised an iPSC to NK differentiation protocol and successfully generated functional NK cells from multiple iPSC lines. A new SMAR vector configuration allowed for the generation of stable CAR-iPSC lines that could be differentiated into CAR+CD34+ hematopoietic progenitor-like cells. These cells could be further differentiated into functional NK cells. I next assessed SMAR vector performance in NK-92 cells, a clinically relevant NK cell line. Strikingly, NK-92 cells from different commercial sources varied substantially in transfection efficiency, sensitivity to DNA sensing pathway inhibition, transcriptomic profile, and cytotoxic activity. Using DSMZ-derived NK-92 cells, I established stable reporter and CAR-expressing lines targeting CD19 and CEACAM5. These cell lines maintained strong transgene expression from a single vector copy for at least six months and displayed potent, antigen-specific cytotoxicity against target-expressing cancer cells. Furthermore, RNA sequencing revealed that SMAR modification exerted little impact on the transcriptome and the presence of episomal vector copies was confirmed in nine out of ten tested cell lines. Finally, I applied the vectors to primary PB NK cells, a cell type notoriously resistant to genetic engineering with both viral and non-viral methods. Developing a feeder-based expansion and electroporation protocol enabled transfection efficiencies up to 70% with CAR-expressing SMAR vectors. The resulting CAR-NK cells exhibited enhanced tumour-killing capacity compared to unmodified NK cells, although transgene expression was transient. These findings underscore the potential of our non-viral vector system and refined transfection strategies in overcoming current limitations, paving the way for more effective, safer and scalable genetic modification of NK cells for cancer immunotherapy.