Human NK cells have been shown to be heterogeneous and differ in their ability to kill target cells. NK cells with the ability to sequentially kill multiple targets are called serial killers. The serial killing activity of NK cells is essential for their cytotoxic function since the majority of kills can be performed by a minority of cells. Consequently, serial killing NK cells are of particular interest for therapeutic applications. However, while it is known that only a fraction of NK cells perform serial killing, it is currently not possible to predict which NK cells will engage in serial killing. Additionally, current methods to identify serial killing NK cells rely mostly on low throughput, time consuming microscopy or micro-fluidic setups which makes it hard to integrate them into a normal workflow.

To investigate serial killing NK cells, we established a staining protocol that can differentiate between the timing and the count of multiple NK cell degranulation events occurring during target cell co-culture. NK cells are analyzed via flow cytometry which enables us to combine degranulation data with other phenotypic readouts. Our method can reproducibly identify donors with high proportions of serial degranulating NK cells. Multiple degranulation events were paralleled by a stepwise loss of granzyme B. Loss of CD16 was associated with the number of degranulation events, whereas the upregulation of CD69 correlated with the timing of degranulation. RNAseq analysis of sorted NK cells that degranulated multiple times compared to ones that did not degranulate identified several differentially expressed genes. However, most of these genes were activation induced and therefore they cannot be used to identify serial killers beforehand. Overall, our data support a stochastic model of serial killing compared to a predetermined one.