The influence of human genetic variation on Epstein-Barr virus sequence diversity

Abstract

Epstein-Barr virus (EBV) is one of the most common viruses latently infecting humans. Little is known about the impact of human genetic variation on the large inter-individual differences observed in response to EBV infection. To search for a potential imprint of host genomic variation on the EBV sequence, we jointly analyzed paired viral and human genomic data from 268 HIV-coinfected individuals with CD4 + T cell count < 200/mm³ and elevated EBV viremia. We hypothesized that the reactivated virus circulating in these patients could carry sequence variants acquired during primary EBV infection, thereby providing a snapshot of early adaptation to the pressure exerted on EBV by the individual immune response. We searched for associations between host and pathogen genetic variants, taking into account human and EBV population structure. Our analyses revealed significant associations between human and EBV sequence variation. Three polymorphic regions in the human genome were found to be associated with EBV variation: one at the amino acid level (BRLF1:p.Lys316Glu); and two at the gene level (burden testing of rare variants in BALF5 and BBRF1). Our findings confirm that jointly analyzing host and pathogen genomes can identify sites of genomic interactions, which could help dissect pathogenic mechanisms and suggest new therapeutic avenues.

Figure 1

Illustration of EBV sequence variation. (A) The EBV genome is about 170 Kbp long and contains 83 genes, for a total of 4392 amino acid residues. As an example, we focus on the BNRF1 gene and on two amino acid changes: Glu → Ala and Ser → Leu. We know for each sample the genomic variants across the whole genome, as illustrated with the colored nucleotides. Using the nucleotide information and a reference genome we can compute the amino acid changes. (B) We compare each individual (ID) to reference data and encode an amino acid as 1 if that individual has a non-synonymous change, and a 0 if not. This process returns us a matrix containing binary values, with individuals as row, and amino acids as columns. In our example, individual 2 has an amino acid change Glu → Ala and individual 3 an amino acid change Ser → Leu. (C) To transform the data into outcomes for the G2G analysis we can use the amino acid matrix as it is (EBV amino acids dataset) or remove all amino acid columns that appear in more than 1 individual and then pool amino acids per gene (1 = variant present, EBV genes dataset).

Figure 2

Significant associations—(A): BALF5, (B): BBRF1, (C): BRLF1:p.Lys316Glu. The x-axis represents the chromosomal position and the y-axis displays the -log10(p-value). Colour alternates between chromosomes. Regions that contain statistically significant SNP are presented in red (top SNP ± 400 Kbp). The light grey dashed line represents the GWAS significance threshold of 5 × 10^–8, the dark grey dashed line the G2G threshold of 1.09 × 10^–10. This figure was produced using R.

Figure 3

Locuszoom plots. Locuszoom plots for the three EBV association signals highlighted in red in Fig. 2 (A: BALF5, B: BBRF1, C: BRLF1:p.Lys316Glu). This Figure was produced using LocusZoom.