Characterisation of Epstein-Barr virus (EBV) genomes in EBV-associated lymphoproliferative diseases

Abstract

Epstein-Barr virus (EBV) infection can cause EBV-associated lymphoproliferative diseases (EBV LPDs). The role of viral variants in the pathogenesis of EBV LPDs is not well understood. In this study, four cohorts of paediatric patients including chronic active EBV infection (CAEBV), EBV-associated haemophagocytic lymphohistiocytosis (EBV-HLH), infectious mononucleosis (IM) and post-transplant lymphoproliferative disorder (PTLD) were followed longitudinally with blood and saliva samples collected at serial time points. The EBV genomes harboured in peripheral blood mononuclear cells (PBMCs), plasma and saliva were sequenced by next-generation sequencing. In total, 263 EBV genomes from 4 CAEBV, 24 EBV-HLH, 66 IM and 27 PTLD patients were successfully sequenced with saliva samples having the highest sequencing success rate. In the subsequent analyses, CAEBV and EBV-HLH cases were combined under the disease category of EBV T/NK-LPDs as defined by the World Health Organization pathological classification. Most of the viral genomes were Type I EBV genomes as demonstrated by the principal component analysis (PCA), where Type I and Type II/intertypic EBV genomes were separated in PC1 and PC2. Both the number of variants of EBV genomes and type of codon changes found in the patients of each disease cohort were highly similar to those harboured in the saliva of healthy control subjects (n = 173). Significantly more non-synonymous codon changes were observed in the latent than lytic genes indicating that the latent EBV genes are under positive evolutionary selection. Differences in allele frequency between cases and controls at each variant site were calculated. Both EBV T/NK-LPDs and PTLD, but not IM, have regions of great differences in allele frequency compared with the controls. To verify this finding and control for population stratification in the analysis, EBV genome-wide association studies between each disease cohort and 173 control subjects were performed using the logistic regression model. A risk variant at position 3380 (p.Gly549Ser) in the BNRF1 gene was detected in EBV T/NK-LPDs with weak linkage disequilibrium (LD) with other variants in the viral genome whereas risk variants at positions 64217 (p.Ala33Ala) and 64132 (p.Ser5Asn) in the BORF2 gene were detected in PTLD where the reference alleles were the risk alleles. In these risk loci, LD is relatively strong with the variants in the BOLF1 and BORF2 regions. Moreover, 10 - 15% of the EBV genomes isolated in EBV T/NK-LPDs and PTLD contain large genomic deletions. Intriguingly, analysis of the intra-host samples of EBV T/NK-LPDs and PTLD demonstrated that these deletions were found in the plasma and PBMC but not in the saliva samples. The analysis of intra-host samples further demonstrated that the risk alleles of the EBV genomes did not concentrate in a specific tissue compartment or appear at specific time points. In conclusion, the data indicated that specific viral variants may play a role in the pathogenesis of EBV T/NK-LPDs and PTLD but not in that of IM.