Ebola Virus Epidemiology, Transmission, and Evolution during Seven Months in Sierra Leone

Abstract

The 2013-2015 Ebola virus disease (EVD) epidemic is caused by the Makona variant of Ebola virus (EBOV). Early in the epidemic, genome sequencing provided insights into virus evolution and transmission and offered important information for outbreak response. Here, we analyze sequences from 232 patients sampled over 7 months in Sierra Leone, along with 86 previously released genomes from earlier in the epidemic. We confirm sustained human-to-human transmission within Sierra Leone and find no evidence for import or export of EBOV across national borders after its initial introduction. Using high-depth replicate sequencing, we observe both host-to-host transmission and recurrent emergence of intrahost genetic variants. We trace the increasing impact of purifying selection in suppressing the accumulation of nonsynonymous mutations over time. Finally, we note changes in the mucin-like domain of EBOV glycoprotein that merit further investigation. These findings clarify the movement of EBOV within the region and describe viral evolution during prolonged human-to-human transmission.

Graphical abstract

Figure 1

Within and between Country Genomic Relationships of Ebola Virus Makona (A) Phylogenetic and temporal placement of recently sequenced Ebola virus (EBOV) within Sierra Leone. New EBOV genomes (232 genomes, dark blue), sampled from June 16 through December 26, 2014, provide a high-resolution view of the accumulated genetic diversity and fill in the missing ancestry between EBOV Makona genome data sets. The maximum clade credibility (MCC) tree was inferred using Bayesian evolutionary analysis by sampling trees (BEAST), with tips anchored to sampling date. Tips are labeled for EBOV from five non-African health-care workers (HCWs) infected in Sierra Leone and treated in Europe (sequenced by other groups, light green). Previously described nested EBOV Makona lineages SL1, SL2, and SL3 Gire et al. (2014), as well as a new lineage SL4, are labeled at their most-recent common ancestor (MRCA) nodes. (B) Lack of EBOV Makona SL3 spread to Liberia or Mali. Shown is a median-joining haplotype network constructed from a coding-complete EBOV genome alignment including 340 EBOV Makona sequences. Each colored vertex represents a sampled viral haplotype, with colors indicating countries of origin. Colors are as in (A), with the exception that the distinction is no longer made between older (Gire) and newer (Park) Sierra Leonean data sets (both are now dark blue), and two additional countries are shown (Liberia in yellow, Mali in red). The size of the each vertex is relative to the number of sampled isolates. Hatch marks indicate the number of mutations along each edge. See also Figures S1 and S2.

Figure 2

Evidence for Host-to-Host Transmission of Multiple Ebola Virus Makona Genomes (A) Certain intrahost variants (iSNVs) appear in samples throughout the 2013–2015 EVD epidemic, suggesting that iSNVs can be transmitted between patients. Variants shared between two or more samples are shown as rows of connected points; each row is a genomic position (ordered by position along the genome, top to bottom), and each point indicates the presence of the iSNV in a patient. (B) Phylogenetic placement of derived alleles at genomic position 18,911 implies both repeated transmission within clades as well as some amount of recurrent mutation. Colored tips are sized according to frequency of iSNV at position 18,911. Tips with small black points are those with iSNV calls at any position; other tips represent samples with no iSNV calls. This figure shows only the portion of the tree relevant for this analysis; large branches with no SNPs or iSNVs at position 18,911 are not shown. See also Figure S3.

Figure 3

Ebola Virus Evolution during a Prolonged EVD Epidemic (A) Estimates of EBOV evolutionary rates at three timescales: decades (yellow, all known EVD outbreaks), months (blue, Baize + Gire + Park), and weeks (red: Baize + Gire). (B) Purifying selection. We estimated nonsynonymous (red) and synonymous (blue) substitution rates on external (unique to an isolate, potential dead end) and internal (shared by multiple isolates, evidence of human-to-human transmission) branches. Nonsynonymous mutations accumulate faster on external branches than on internal branches. For synonymous mutations, the difference between external and internal branches is less pronounced. (C) Enrichment for nonsynonymous mutations at shorter timescales. Intrahost (all variants that appear within a single host at less than 100% frequency); unique interhost (SNPs fixed in exactly one individual); shared interhost (SNPs fixed in two or more individuals); shared between EVD outbreaks (internal branch SNPs on a between-outbreak tree). See also Figure S4.

Figure 4

Evidence for Host Effects on Ebola Virus Makona Evolution (A) Nonsynonymous variants are enriched in the mucin-like domain of GP. Estimates of log(ω) (a.k.a., log(d_N/d_S)) per coding sequence within the Western African EVD outbreak (left) and between EVD outbreaks (right) demonstrate gene-specific patterns of natural selection. (B) Nonsynonymous variants are enriched in B cell epitopes of GP. We calculated the fractions of nonsynonymous (NS) and synonymous (S) consensus SNPs and intrahost variants (iSNVs) within experimentally determined B cell epitopes (data from ViPR; Pickett et al., 2012). Dotted line represents the fraction of GP amino acids in ViPR epitopes. Nonsynonymous SNPs (p = 0.004) and iSNVs (p = 0.037) in GP occur more frequently in epitopes than expected by chance (two-sided exact binomial test). Numbers indicate fraction of each variant type within GP epitope regions. Error bars represent binomial sampling intervals. (C) Local enrichment of T-to-C mutations within GP B cell epitopes. We observed five sequences with short stretches (<200 nucleotides) of concentrated T-to-C mutations. Of these five sequences, two (shown here, samples 20141582 and G5119.1) contain stretches of T-to-C SNPs (blue points) within GP epitopes (light blue bars). Additionally, we observe a T-to-C mutation at amino acid position 485 (blue diamond) in three samples (one shown here, G4955.1), which is otherwise completely conserved among members of all ebolavirus species (Olal et al., 2012). (D) Genome-wide increase in T-to-C mutations. We observe more T-to-C transitions within the 2013–2015 outbreak than any other transition, after correcting for nucleotide content. Error bars represent binomial sampling intervals. (E and F) Elevated T-to-C rates are genome wide but are limited to a subset of sequences. Accumulation of mutation increases linearly with time. However, some individual samples show more genetic distance than expected based on sample date. Samples with short stretches of T-to-C mutations (orange) show a significant enrichment of T-to-C mutations, as expected. Excluding these samples, the top 5% of samples by genetic distance (yellow) lack localized stretches but still show moderate enrichment of T-to-C mutations genome wide. The bottom 95% of samples (beige) show no enrichment of T-to-C mutations. Error bars represent binomial sampling intervals.

Figure S1

Phylogenetic and Temporal Context of Recent Tong et al. Samples, Related to Figure 1 (A) 175 recently published Ebola virus Makona samples from Sierra Leone (Tong et al., 2015) describe lineages that fall within the genetic diversity of our current dataset (MCC tree from BEAST, as in Figure 1). (B) They span a two month period (Sep 28 to Nov 11, 2014) that falls within the temporal sampling of our current data and shows a consistent evolutionary rate.

Figure S2

Tracing Historical Ebola Virus Makona Migrations from East to West, Related to Figure 1 (A) Nine Ebola virus (EBOV) Makona genomes (right-hand most circles) from the Freetown area with four groups of apparently ancestral EBOV genomes (middle circles)). Groups of genetically identical genomes (circles) are related to each other by simple vertical relationships (arrows). Solid circles are shown on the date of the earliest sample in the group; the circle area is proportional to the number of samples containing viruses with that genome; arrows represent a set of non-homoplasic SNPs and point from ancestral to derived alleles. Here, “SL3” and “SL4” do not refer to entire clades, but to the viruses that exactly match the canonical SL3 and SL4 genomes with no further mutations. (B) Geographic mapping of one epidemiological route that may account for four of the nine Freetown viruses shown in (A). Groups of identical viruses are shown at their first observed location.

Figure S3

Ebola Virus Makona Intrahost Single-Nucleotide Variants, Related to Figure 2 (A) Distribution of the number of iSNVs per sample. Replicate sequencing and iSNV calling was completed for 150 samples, of which 65 had no iSNV calls. Mean iSNVs per sample (including samples without iSNVs) = 2.04; mean iSNVs per sample (among samples with iSNVs) = 3.6. (B) Sample coverage by date shows the temporal distribution of samples containing Ebola virus (EBOV) genomes with and without iSNV calls. As expected, samples with iSNV calls have generally higher coverage. (C) Intermediate-frequency variants can persist over time with minimal genetic drift, as demonstrated by the iSNV at position 18,911. The existence of intermediate frequency (10%–30%) iSNVs in many different samples over time provides an argument against recurring mutations and may suggest a relatively wide transmission bottleneck between patients.

Figure S4

Increased Sampling Improves Evolutionary Rate Estimates, Related to Figure 3 Rate estimates in the recent dataset (Figure 3A) have much tighter credible intervals due to the significantly greater amount of time (total coalescent branch length) compared to the initial outbreak.