Whole Exome Sequencing Analysis in Chronic Lymphoproliferative Disorder of NK Cells (CLPD-NK) Patients Fails to Detect Significant Viral Load

Background. Chronic Lymphoproliferative Disorder of Natural Killer cells (CLPD-NK) is a rare disease characterized by chronic NK cell lymphocytosis, typically equipped with a restricted pattern of activating killer immunoglobulin like receptors. Several lines of evidence point to a putative role of viral agents in the pathogenetic mechanisms accounting for NK cell proliferation. A role of Epstein Barr Virus (EBV) has been claimed in the aggressive form of NK cell proliferation [i.e. Aggressive NK Cell Leukemia (Kawa Ha et al, JCI 1989)]. At variance, in the indolent CLPD-NK, EBV DNA was seldom detected by PCR (Loughran TP Jr et al, Blood 1993), whereas we provided serologic evidence for previously experienced viral infections in the majority of CLPD-NK patients (Zambello R et al, Leukemia 1995). However, the role of infectious agents in the pathogenesis of chronic antigen stimulation of NK cells in CLPD-NK patients is still elusive.

Methods. Several studies showed that whole exome sequencing (WES), as well as RNA-seq, can identify pathogens, particularly viruses, in cancer clones enabling the definition of possible integration sites in the host genome or their presence as episomal forms. In the present study, we searched viral sequences in WES data of a cohort of 10 CLPD-NK patients. High-depth (average 51 M reads) WES data of highly purified (>97% purity) leukemic NK clone and autologous granulocytes, as control, from CLPD-NK patients were analyzed for pathogen discovery. Three different methods for virus detection were used: i) Centrifuge (Kim et al, Genome Research 2016), a metagenomic tool performing a wide search against GenBank viral and bacterial sequences, which has been recently used to detect EBV load in ANKL and NK T cell Lymphoma (NKTCL) tumor clones (Dufva et al. Nat Comm 2018); ii) VirusSeq (Chen et al., Bioinformatics 2013), a software to identify viruses and eventually their integration sites, grounding on its custom and relatively wide viral sequence database; and iii) ViFi (Nguyen et al. Nucleic Acids Res, 2018), which can find viral integration in the human genome, with an analysis focused on a narrow set of viruses (EBV, HPV, HBV) with well-established oncogenic potential.

Results. First, as positive control two datasets were used: RNA-seq data of an HBV positive sample with known site of integration (provided by VirusSeq) and WES data of three NKTCL tumor and control clones with high, moderate or low EBV load in the tumor samples previously detected by Dufva et al. (Dufva O et al, Nat Comm 2018), publicly available at NCBI SRA. HBV presence was detected in the positive sample by all the three methods and our reanalysis of NKTCL samples with Centrifuge confirmed published results, detecting EBV in all tumoral samples and in none of the matched control samples. Next, analysis of samples from the 10 CLPD-NK patients with the three methods did not identify a significant viral load neither in tumor nor in control specimens. Neither VirusSeq nor ViFi found viruses in CLPD-NK samples, whereas, according to Centrifuge analysis, the human disease virus with the highest load in most patients was Varicella Zoster (Human alphaherpesvirus 3) but with low read count and, most relevant, without specificity for the cancer clone.

Conclusions. Our very thorough WES analysis did not identify a significant and meaningful viral load in CLPD-NK samples, neither in leukemic clones nor in control cell population. Although not ruling out that a viral infection might play a role as starting event for inciting NK cell proliferation, our data do not support a role of viral infections in maintaining the proliferation in CLPD-NK.