Genome-wide patterns of intrahuman dengue virus diversity reveal associations with viral phylogenetic clade and interhost diversity

Abstract

Analogous to observations in RNA viruses such as human immunodeficiency virus, genetic variation associated with intrahost dengue virus (DENV) populations has been postulated to influence viral fitness and disease pathogenesis. Previous attempts to investigate intrahost genetic variation in DENV characterized only a few viral genes or a limited number of full-length genomes. We developed a whole-genome amplification approach coupled with deep sequencing to capture intrahost diversity across the entire coding region of DENV-2. Using this approach, we sequenced DENV-2 genomes from the serum of 22 Nicaraguan individuals with secondary DENV infection and captured ∼75% of the DENV genome in each sample (range, 40 to 98%). We identified and quantified variants using a highly sensitive and specific method and determined that the extent of diversity was considerably lower than previous estimates. Significant differences in intrahost diversity were detected between genes and also between antigenically distinct domains of the Envelope gene. Interestingly, a strong association was discerned between the extent of intrahost diversity in a few genes and viral clade identity. Additionally, the abundance of viral variants within a host, as well as the impact of viral mutations on amino acid encoding and predicted protein function, determined whether intrahost variants were observed at the interhost level in circulating Nicaraguan DENV-2 populations, strongly suggestive of purifying selection across transmission events. Our data illustrate the value of high-coverage genome-wide analysis of intrahost diversity for high-resolution mapping of the relationship between intrahost diversity and clinical, epidemiological, and virological parameters of viral infection.

Fig 1

Schematic representation of the distinction between intrahost diversity and consensus-level (interhost) diversity. Viral genomes and genomic polymorphisms are represented by lines and symbols, respectively.

Fig 2

Nucleotide diversity in individuals infected with DENV-2. (A) Proportion (p) of loci that demonstrate diversity relative to the total number of sequenced nucleotides in each sample, categorized by the identity of the consensus nucleotide. (B) Proportion of specific nucleotide changes observed in each sample (P < 0.0001; Kruskal-Wallis). (C) Proportion of mutations that occur at codon position 1, 2, or 3 in each sample. P < 0.0001 (Kruskal-Wallis). **, P < 0.0001 for individual comparisons (one-way chi-square test).

Fig 3

Wide range of DENV-2 intrahost diversity in secondary dengue cases in humans. (A and B) Total number of loci with variant alleles (line plots) and distribution of percent diversity (i.e., percentage of reads that are distinct from the consensus; box plots) at various loci in codon space (A) or amino acid space (B) for each sample. NI-1 and NI-2B refer to viruses from Nicaraguan clades 1 and 2B, respectively.

Fig 4

Gene-specific and domain-specific trends in DENV-2 intrahost diversity. (A and B) Average percent codon diversity (A) and percent amino acid diversity (B) (average percent divergence) per gene per sample. Averages are significantly different at P = 0.0005 (A) and 0.0055 (B) (Kruskal-Wallis), with individual comparisons significant at P < 0.0001 (**) for E and NS2B (A) by one-way chi-square test, P = 0.0005 (*) for NS5 and NS2B (A) by one-way chi-square test, and P = 0.0006 (*) for NS5 and NS2B (B) by one-way chi-square test. (C and D) Distribution of loci with intrahost (C) codon [p (codon)] or (D) amino acid [p (amino acid)] diversity between ectodomains EDI to EDIII and the C-terminal (C-term) region of the E gene in each sample. In panel C, P = 0.0027 (*) for comparisons between EDII and EDIII (one-way chi-square test). In panel D, P = 0.0010 (*) for comparisons between EDII and EDIII as well as EDII and C-term (one-way chi-square test). (E and F) Proportion of codons (E) and amino acids (F) with diversity [p (divergence)] in the three ectodomains (EDI to EDIII) and the C-terminal region of the Envelope gene calculated for each sample. For panel F, significant differences are observed in p (divergence) at the amino acid level across the domains (P = 0.0419 by Kruskal-Wallis), with comparison between EDII and EDIII significant at P = 0.0094 (*; one-way chi-square test test). The gray line (C to F) shows the corresponding value at the consensus level (values on the left y axis for A to D and on the right y axis for E and F [p (divergence-interhost)].

Fig 5

Nicaraguan clade NI-1 viruses demonstrate higher diversity than clade NI-2B viruses. All parameters were calculated per gene or per domain for each sample. Only significant differences between clades are shown. (A) Proportion of codons [p (divergence)] that exhibit diversity. Differences in p (divergence) values are significant at P = 0.0012 (**; NS4A) and 0.0155 (*; NS4B). (B) Percent codon diversity summed for all loci with diversity (sum percent divergence). Significant differences in sum percent divergence values are observed for E (*, P = 0.0343), NS4A (**, P = 0.0003), and NS4B (**, P = 0.0049). (C) Average percent codon diversity. Differences in average percent divergence values are significant at P = 0.0009 (**; NS4A) and P = 0.0062 (*; NS4B). (D) p (divergence) values for codon diversity in E ectodomains EDII and EDIII. P = 0.0138 (*; EDII) and 0.0047 (**; EDIII). All P values were calculated using one-way chi-square test tests. *, P < 0.05; **, P < 0.005 (highly significant because it passes the Bonferroni correction threshold).

Fig 6

Purifying selection at the intrahost-interhost interface. All parameters were calculated per sample. (A) Proportion (p) of all synonymous (Syn) or nonsynonymous (Non.Syn) mutations that occur in both intrahost and interhost data sets or in the intrahost data set only. P values were calculated using the paired t test (*, P < 0.05; **, P < 0.005). (B and C) Relationship between the abundance of intrahost variants and the appearance of these variants in consensus genomes. (B) Percent intrahost codon diversity, categorized by whether the secondary variant is identical (Intrahost & Interhost) or nonidentical (Intrahost only) in intrahost and interhost data sets. Data are shown for loci that demonstrate synonymous mutations only. Differences are significant at P < 0.0001 (**; one-way chi-square test). (C) Percent intrahost amino acid diversity, categorized by whether the intrahost mutation is also found in interhost consensus genomes (Intrahost & Interhost) or is distinct from variation observed at the interhost level (Intrahost only). Data are shown for loci that can tolerate nonsynonymous mutations. Differences are significant at P = 0.0002 (**; one-way chi-square test). (D) Proportion (p) of predicted tolerated nonsynonymous substitutions relative to all nonsynonymous substitutions in intrahost data sets. The gray reference line indicates the proportion of predicted tolerated mutations at the consensus level in 161 Nicaraguan isolates.