Epstein-Barr virus genetic variation in lymphoblastoid cell lines derived from Kenyan pediatric population

Abstract

Epstein-Barr virus (EBV) is associated with Burkitt's lymphoma (BL), and in regions of sub-Saharan Africa where endemic BL is common, both the EBV Type 1 (EBV-1) and EBV Type 2 strains (EBV-2) are found. Little is known about genetic variation of EBV strains in areas of sub-Saharan Africa. In the present study, spontaneous lymphoblastoid cell lines (LCLs) were generated from samples obtained from Kenya. Polymerase chain reaction (PCR) amplification of the EBV genome was done using multiple primers and sequenced by next-generation sequencing (NGS). Phylogenetic analyses against the published EBV-1 and EBV-2 strains indicated that one sample, LCL10 was closely related to EBV-2, while the remaining 3 LCL samples were more closely related to EBV-1. Moreover, single nucleotide polymorphism (SNP) analyses showed clustering of LCL variants. We further show by analysis of EBNA-1, BLLF1, BPLF1, and BRRF2 that latent genes are less conserved than lytic genes in these LCLs from a single geographic region. In this study we have shown that NGS is highly useful for deciphering detailed inter and intra-variations in EBV genomes and that within a geographic region different EBV genetic variations can co-exist, the implications of which warrant further investigation. The findings will enhance our understanding of potential pathogenic variants critical to the development and maintenance of EBV-associated malignancies.

Fig 1. Phylogenetic alignments of LCLs to EBV-1 and EBV-2.

The figure shows the alignment of the LCLs, B95.8 cell line and Jijoye cell line controls against EBV-1 and EBV-2. It is observed that B95.8 aligned closely to EBV-1, followed by LCL-3, LCL1 and LCL9, while LCL10 and Jijoye were distant from EBV-1. B95.8 is clustered with LCL3, LCLI and LCL9 are clustered together, while LCL10 and Jijoye are clustered together when compared to EBV-1. There were relatively less base changes in the middle for most of the LCLs except LCL10. Similar trend was observed on comparison with EBV-2, however, LCL10 was much closer to EBV-2 reference than Jijoye and the clustering was maintained as with EBV-1 reference. The bases were more conserved in the middle region of the genome than in the N and C terminus for LCL10 and Jijoye, for B95.8, LCL1, LCL3, and LCL9 the base changes were spread all over the genome with majority of the changes in the middle. Additionally, LCL3 was more conserved at the C-terminal end of the genome, as observed with EBV-1.

Fig 2. Non—synonymous SNPs compared with EBV-1.

When we aligned the sample sequences against EBV-1 reference gene, we observed that BOLF1, BARF0, BNRF1, BSLF1, and BRRF2 were the genes with the most SNPs. BOLF1 had most SNPs in Jijoye and LCL9; BARF0 with most SNPs in LCL1 and B95.8; BNRF1 with most SNPs in LCL3, LCL9 and Jijoye; BSLF1 with more SNPs in LCL1, LCL9 and Jijoye; and BRRF2 had the most number of non—synonymous SNPs in LCL1, LCL3, LCL9, and Jijoye. We observed that all the genes were lytic and mostly tegument and envelop.

Fig 3. Non—synonymous SNPs compared with EBV-2.

We observed that BPLF1, BLLF1, BRRF2, and EBNA1 had the most non—synonymous SNPs compared to EBV-2. For BPLF1 LCL1, LCL3 and B95.8 had the most SNPs; for BLLF1 LCL1, LCL9, LCL3 and B95.8 had the most SNPs, significantly there was no change in LCL-10 in BLLF1; for BRRF2 again most SNPs were noted in LCL1, LCL9, and LCL3; and for latent gene EBNA-1 most SNPs were observed in LCL1, LCL10, LCL9, and B95.8.

Fig 4. Amino acid changes in EBNA1, BPLF1, BLLF1 and BRRF2.

(A) EBNA1. This was the only latent gene that had amino acid variation when compared to EBV-2. Towards the N-terminal we observed changes at position 16 and 18 respectively (Q>E, E>G)) for B95.8, Jijoye, and LCL-10, at position 24, 27, 410 (D>E, A>I, A>G) for B95.8, Jijoye, LCL10, LCL9, and LCL1 respectively. At the C-terminal we noted amino acid changes at 543 (M>T), 585 (P>T), and 599 (S>N) for LCL10 alone. (B) BPLF1. When we aligned BPLF1 against EBV-2 we noticed several amino acid changes that occurred on specific samples with some toward the N-terminal that were shared by more than 3 samples. The only amino acid changes observed in LCL10 alone occurred at the C-terminal at 2935 (L>P), 2987(P>L), and 3005 (R>Q).(C) BLLF1. Comparison of BLLF1 to EBV-2 was interesting in that the amino acid changes were similar in LCL-1, -3, -9, and B95.8, but generally lacking in LCL10 and Jijoye, showing a significant difference between the EBV-1 (B95.8, LCL1,LCL3, LCL9) and EBV-2(LCL10 and Jijoye) clustered samples. (D) BRRF2. When BRRF2 was compared to EBV-2 we noted that all the amino acid variations at the N-terminal occurred in B95.8, and in LCL3 at 382 (R>C). Amino acid variations in the middle were mainly in LCL9, LCL3, and LCL1. Amino acid changes at the C-terminal at position 464 (D>E), 477 (F>L), and 537 (I>V) were observed in most of the LCLs, B95.8, and Jijoye. (E) BRRF2. On comparing BRRF2 with EBV-1, the only gene we were able to show using Illumina sequencing against EBV-1, we observed that at the N-terminal we had similar amino acid changes in Jijoye, LCL10, LCL9, LCL3, and LCL1. Other amino acid changes occurred in multiple LCLs, except for changes at the C-terminal at positions 463 (D>A) and 464 (E>D) that were both in LCL10. It is to be noted that the amino acid changes in BRRF2 against EBV-2 were significantly different from that observed against EBV-1.

Fig 5. Non-synonymous sequence changes.

(A) EBNA1 compared to EBV-2. Using heat maps in our analysis we observed that most EBNA1 mutations were substitutions with a single deletion (96733), and that the changes were similar in all LCLs except in a few cases where there were changes in LCL3 (97756) and in LCL10 (98119, 98248, and 98287) distinct from the other LCLs. (B) BPLF1 compared to EBV-2. When aligned to EBV-2 all the mutations observed in BPLF1 were substitutions. All the substitutions were generally similar in the LCLs, with some changes seen in LCL3 alone (54534, 53397, 52825, and 51854), LCL10 alone (57689, 57539, 56627, 56385, 50567, 50411, 50357, 57686, and 56 627). (C) BLLF1 compared to EBV-2. All the BLLF1 mutations observed were substitutions and they were mostly shared by all LCLs. LCL10 and Jijoye were perfect match with the reference, the rest varied with the reference genome at different sites. LCL1, LCL3, and LCL9 all had similar changes compared to reference except LCL3 at 77623, 78193, and 78000 that were similar to reference. (D) BRRF2 compared to EBV-2. We observed that when compared to EBV-2 reference, all the BRRF2 mutations were substitutions, which were similar in most LCLs except for a couple specific for LCL3 (95987 and 96329), and LCL10 (96272). (E) BRRF2 compared to EBV-1. When BRRF2 was compared to EBV-1, most of the changes were substitutions, but we had a deletion too that was detected in all samples. Again LCL3 had specific mutations (95157, 95396, 95459, and 95499) and LCL10 (95429). It is to be noted that only LCL3 had no deletion at 95201.