A Comprehensive, Multiomics Analysis of Natural Killer-Cell Malignancies

Abstract

Natural killer/T cell lymphoma (NKTCL) and aggressive NK cell leukemia (ANKL) are rare mature NK cell malignancies which show a geographic predilection for Asia and Central/South America and are strongly associated with EBV infection. Despite recent advances in chemotherapeutical agents, the prognosis (especially for ANKL) remain poor. On the contrary, NK cell proliferations of a more indolent nature also exist. NK-large granular lymphocytic leukemia (NK-LGLL) is a disorder of NK cells with indolent clinical manifestation. Additionally, chronic active Epstein-Barr virus infection (CAEBV) affecting NK cells can have variable presentation, between indolent and aggressive. Due to their rarity, a comprehensive understanding of the events differentiating the wide spectrum of NK malignancies remains elusive. Here, we use a multi-omics approach to investigate the genetic differences underlying the different clinical manifestations of NK malignancies. To discover the pathogenesis behind indolent and aggressive NK malignancies, we studied whole-genome/exome sequencing and whole transcriptome sequencing (RNA-seq) data from over 200 NK-cases collected from 11 previous publications. We investigated driver genes from 145 WES and WGS samples, and transcriptomes from 91 samples. Single-cell RNA-seq analysis was performed to identify and characterize distinct NK-cell subsets. We used pooled CRISPR screens with a single-cell RNA sequencing readout (CROP-seq) to investigate the effect of knocking out the most frequently mutated genes on the transcriptome of NKTCL cells under the effect of commonly used chemotherapeutic agents. We also aim to construct a comprehensive host-EBV interaction map to elucidate the role of EBV in the pathogenesis of NK malignancies We validated previously found driver genes STAT3 and DDX3X as the most frequently mutated genes in NK malignancies, being mutated in 15% and 14% of samples, respectively. STAT3 mutations were found only in NKTCL and ANKL samples. Additionally, our combinatorial analysis allowed for identification of previously unappreciated putative driver genes, MSN and HLA-A, which were mutated in 8% and 4% of samples, respectively. We also discovered associations between important genes, such as significant co-occurrence of mutations in STAT3 and novel driver gene MSN (p <0.05). Samples originating from different diseases were intermingled in non-negative matrix factorization (NMF) analyses, highlighting the effect of mutational profile, rather than disease type, on sample clustering. Using RNA-seq, differentially expressed genes (Q-value < 0.01) were identified between ANKL, NKTCL and healthy samples. Gene set enrichment analysis revealed differential enrichment and depletion of biological processes in ANKL and NKTCL. Upregulation of chemokine signalling and macrophage-associated processes in NKTCL reflected the inflammatory microenvironment of the disease. Of 17 putative driver genes, CRISPR knockout of STAT3 and TP53 led to most transcriptional changes. Knockout of STAT3 caused upregulation of HLA I and II molecules and downregulation of IL10 signaling, thus promoting tumor cell immune evasion. Our preliminary results indicate that NK malignancies are, contrary to previous knowledge, a spectrum of manifestations rather than distinct diseases. Our results concerning the heterogeneous genetic landscape of NK malignancies can lead to better and faster diagnosis as well as improved treatment stratification in the future. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal