Genome-wide analysis of wild-type Epstein-Barr virus genomes derived from healthy individuals of the 1,000 Genomes Project

Abstract

Most people in the world (∼90%) are infected by the Epstein-Barr virus (EBV), which establishes itself permanently in B cells. Infection by EBV is related to a number of diseases including infectious mononucleosis, multiple sclerosis, and different types of cancer. So far, only seven complete EBV strains have been described, all of them coming from donors presenting EBV-related diseases. To perform a detailed comparative genomic analysis of EBV including, for the first time, EBV strains derived from healthy individuals, we reconstructed EBV sequences infecting lymphoblastoid cell lines (LCLs) from the 1000 Genomes Project. As strain B95-8 was used to transform B cells to obtain LCLs, it is always present, but a specific deletion in its genome sets it apart from natural EBV strains. After studying hundreds of individuals, we determined the presence of natural EBV in at least 10 of them and obtained a set of variants specific to wild-type EBV. By mapping the natural EBV reads into the EBV reference genome (NC007605), we constructed nearly complete wild-type viral genomes from three individuals. Adding them to the five disease-derived EBV genomic sequences available in the literature, we performed an in-depth comparative genomic analysis. We found that latency genes harbor more nucleotide diversity than lytic genes and that six out of nine latency-related genes, as well as other genes involved in viral attachment and entry into host cells, packaging, and the capsid, present the molecular signature of accelerated protein evolution rates, suggesting rapid host-parasite coevolution.

Keywords: EBV; Illumina reads; human herpesvirus 4; recombination; selection; whole-genome analysis.

Fig. 1.—

Boxplot of the coverage for all LCLs in the whole EBV genome (right) and in the B95-8-specific deletion (left). The inner panel displays a zoom in the coverage scale that shows that coverage at B95-8 is nonzero, suggesting the presence of natural EBV in some LCLs.

Fig. 2.—

Scatter plot where each dot indicates the number of variants called in each LCL. Light blue dots represent LCLs with strong evidence of the presence of natural EBV strains (from ten different individuals). The size of the dot indicates the mean allele balance found in each LCL. Five LCLs clearly show a large amount of genetic variation, as they harbor the greatest proportion of natural EBV.

Fig. 3.—

Genome Browser snapshot of the reads from the three individuals from which we obtained high-quality EBV genomes aligned against EBV type 1 (top panel) and type 2 (bottom panel) reference genome. The snapshot shows the EBNA-3region, which was used to discriminate EBV type 1 from type 2. Large regions with no coverage are observed in the type 2 mappings, demonstrating that all three cell lines are infected only by type 1 EBV strains.

Fig. 4.—

PCA using all the SNVs in eight complete EBV strains. PC1 discriminates between type-1 and type-2 strains, while PC2 separates African from Asian strains. A PC3 (marginally significant) does not place strains across a clear geographical axis, as NA18384 and MUTU have Kenyan origin. NA19384, NA19315, and NA19114 in green; MUTU in red, GD1 and AKATA in gray; type 2 EBV in blue; and type 1 reference genome in black.

Fig. 5.—

Phylogenetic tree representing genetic distance between our three EBV strains and published sequences. Numbers in internal branches indicate the consistency of the branch out of 1,000 bootstraps. Color gradient represents bootstrap support. Trees are rooted using the midpoint. (a) Phylogenetic relationships using the whole EBV genome. (b) Using only the B95-8-specific deletion, which in the reference EBV genome corresponds to a Raji EBV strain fragment from Nigeria. (c) Excluding latency genes, which contain most intertypic EBV variation.

Fig. 6.—

Recco analysis on the three African EBV strains. We consider one sequence at a time to be the recombinant product of the rest of sequences. Gray regions are masked due to repeats. The scale color is proportional to genetic distance based on identity, red indicates short distance, and yellow indicates large distance. The first row is the one considered the recombinant product in all panels.

Fig. 7.—

Circos plot showing genome-wide single-nucleotide variant diversity in the three LCLs with the highest natural EBV load. Each green circle corresponds to one individual. Red dots indicate variable positions. Blue dots indicate variants present in all three EBV sequences. The innermost circle shows diversity on 500-bp windows considering all variants together. In the outer circle, latency genes (red) and lytic genes (blue) are represented. Light gray shadows indicate repetitive regions, and dark gray sector indicates the location of the B95-8-specific deletion.