Dynamic Transcriptional Programs During Single NK Cell Killing: Connecting Form to Function in Cellular Immunotherapy

Introduction

Natural killer (NK) cell-based therapies are a promising new method for treating indolent cancer, however engineering new therapies is complex and progress towards therapy for solid tumors is slow. New methods for determining the underlying intracellular signaling driving the killing phenotype would significantly improve this progress.

Methods

We combined single-cell RNA sequencing with live cell imaging of a model system of NK cell killing to correlate transcriptomic data with functional output. A model of NK cell activity, the NK-92 cell line killing of HeLa cervical cancer cells, was used for these studies. NK cell killing activity was observed by microscopy during co-culture with target HeLa cells and killing activity subsequently manually mapped based on NK cell location and Annexin V expression. NK cells from this culture system were profiled by single-cell RNA sequencing using the 10× Genomics platform, and transcription factor activity inferred using the Viper and DoRothEA R packages. Luminescent microscopy of reporter constructs in the NK cells was then used to correlate activity of inferred transcriptional activity with killing activity.

Results

NK cells had heterogeneous killing activity during 10 h of culture with target HeLa cells. Analysis of the single cell sequencing data identified Nuclear Factor Kappa B (NF-κB), Signal Transducer and Activator of Transcription 1 (STAT1) and MYC activity as potential drivers of NK cell functional phenotype in our model system. Live cell imaging of the transcription factor activity found NF-κB activity was significantly correlated with past killing activity. No correlation was observed between STAT1 or MYC activity and NK cell killing.

Conclusions

Combining luminescent microscopy of transcription factor activity with single-cell RNA sequencing is an effective means of assigning functional phenotypes to inferred transcriptomics data.