Population genomics in the arboviral vector Aedes aegypti reveals the genomic architecture and evolution of endogenous viral elements

Abstract

Horizontal gene transfer from viruses to eukaryotic cells is a pervasive phenomenon. Somatic viral integrations are linked to persistent viral infection whereas integrations into germline cells are maintained in host genomes by vertical transmission and may be co-opted for host functions. In the arboviral vector Aedes aegypti, an endogenous viral element from a nonretroviral RNA virus (nrEVE) was shown to produce PIWI-interacting RNAs (piRNAs) to limit infection with a cognate virus. Thus, nrEVEs may constitute a heritable, sequence-specific mechanism for antiviral immunity, analogous to piRNA-mediated silencing of transposable elements. Here, we combine population genomics and evolutionary approaches to analyse the genomic architecture of nrEVEs in A. aegypti. We conducted a genome-wide screen for adaptive nrEVEs and searched for novel population-specific nrEVEs in the genomes of 80 individual wild-caught mosquitoes from five geographical populations. We show a dynamic landscape of nrEVEs in mosquito genomes and identified five novel nrEVEs derived from two currently circulating viruses, providing evidence of the environmental-dependent modification of a piRNA cluster. Overall, our results show that virus endogenization events are complex with only a few nrEVEs contributing to adaptive evolution in A. aegypti.

Keywords: Aedes aegypti; endogenous viral elements; mosquito genomes; piRNA cluster.

FIGURE 1

Atlas of Aedes aegypti nrEVEs. (a) Violin plot showing nrEVEs identified in the A. aegypti genome (AaegL5 assembly). (b) Scatter plot representing the amino acid identity of each nrEVE and its best hit retrieved by blastx searches against the NR database grouped by viral family. Whiskers represent the median and the interquartile range. Red dots are the novel nrEVEs discovered in the geographical populations. (c) Bar plots showing the type of the closest transposable element (TE) upstream and downstream of all nrEVEs (upper panel), nrEVEs grouped by their viral origin (middle panel), and nrEVEs grouped by their location within (IN) or outside (OUT) piRNA clusters. Abbreviations: LTR (long terminal repeat), UD (unclassified TEs). (d) Pie charts indicating the transposon composition of TEs in the whole genome or piRNA clusters [Colour figure can be viewed at wileyonlinelibrary.com]

FIGURE 2

Distribution and piRNA coverage of nrEVEs on reference viral genome. Flaviviridae‐derived nrEVEs (Flavi‐nrEVE) aligned to the Xishuangbanna flavivirus genome (NC_034017.1). Flavi5, Flavi7, Flavi16 and Flavi17 are composed of repeated and not contiguous parts of the viral genome (composite nrEVEs) and thus have been fragmented to map to the corresponding viral sequence. Stars indicate stop codons or small indels that interrupt the viral open reading frame and dotted white boxes indicate large deletions that generate stop codons. Top panels indicate piRNAs mapping to the indicated positions in soma (orange) and ovaries (blue), respectively [Colour figure can be viewed at wileyonlinelibrary.com]

FIGURE 3

Aedes aegypti piRNA clusters. (a) Expression of piRNA clusters in germline and somatic tissues. piRNA coverage per million mapped small RNAs (rpm) plus a pseudo‐count of 1 is plotted in order to include values of zero. Colour indicates the likelihood of a cluster being expressed with the same strand bias in both tissues. (b) Bar plots showing the distribution of nrEVEs from different viral families within (IN) and outside (OUT) piRNA clusters. (c) Table listing the top‐10 most highly expressed piRNA clusters based on their highest overall expression in either germline or soma, with at least 5% uniquely mapping piRNA reads. (d) Coverage plot of a piRNA cluster with strand bias towards expression from one strand (uni‐strand). (e) Coverage plot of a piRNA cluster without strong strand bias (dual‐strand). Log₂ coverage in both germline (ovaries) and somatic tissues is shown. Genes are indicated with black arrows, transposons are indicated with light grey (plus strand) or red (minus strand) boxes, and nrEVEs are depicted with dark grey boxes (plus strand) [Colour figure can be viewed at wileyonlinelibrary.com]

FIGURE 4

Novel viral integrations in wild‐collected Aedes aegypti mosquitoes. (a) Scheme of the novel nrEVEs with similarity to cell fusing agent virus (CFAV) and Aedes anphevirus (AeAV) identified in the genome of wild‐collected mosquitoes. CFAV‐EVEs are mapped to the genome of CFAV Galveston strain (NCBI Reference Sequence: NC_001564.2) and the AeAV‐like nrEVE to the genome of AeAV strain MRL‐12 (MH037149.1). Dotted lines represent part of CFAV‐EVE‐3 that integrated in the opposite direction compared to the CFAV genome. (b) Scheme of the integration points and endogenized sequences. nrEVE sequences are represented by grey boxes; flanking TE sequences (if any) are represented with colours as indicated. Dotted TE boxes indicate flanking TEs for which we could not distinguish if they integrated together with the viral sequences or were already present in the integration point. (c) Frequency distribution of novel nrEVEs tested with PCR in 24 mosquitoes from each site (Kenya, Ghana, Gabon, American Samoa and Mexico) [Colour figure can be viewed at wileyonlinelibrary.com]

FIGURE 5

Acquisition of a novel nrEVE by a piRNA cluster. Coverage plot of piRNA cluster 3p23.4. Genes are indicated with black arrows, and transposons and nrEVEs are indicated with light and dark grey (plus strand) or light and dark red (minus strand) boxes, respectively. Arrows indicate variably distributed nrEVEs which are present only in some individual mosquitoes [Colour figure can be viewed at wileyonlinelibrary.com]

FIGURE 6

Genome‐wide screen of reference nrEVEs. (a) Outline of the distribution of reference nrEVEs annotated in the AaegL5 assembly. Fixed nrEVEs have been observed in all populations and variably distributed (VD)‐nrEVEs have been observed only in some populations. Mesoni‐ and Chuvi‐nrEVEs are not included in this analysis. (b) Convex logistic principal component analysis (PCA) of VD‐nrEVE frequencies based on their geographical origin. Each dot indicates an individual mosquito, colour‐coded based on geographical location. (c) Whisker plots comparing the level of nucleotide polymorphism among Aedes aegypti populations in conserved genes (CG), fast‐evolving genes (VG), fixed nrEVEs and VD‐nrEVEs. Each dot represents the average value of an individual mosquito, boxes span the interquartile range, marked lines within the boxes represent the median, and whiskers represent the minimum and the maximum. Dotted lines are colour‐coded according the population they represent and depict the median value of the level of polymorphism of fast‐evolving genes. (d) Composite likelihood ratio (CLR) signal around the indicated nrEVEs in mosquito populations. CLR signal around Rha44 in Ghana, Rha53 in American Samoa and Rha69 in Mexico is higher than the 99th percentile of their corresponding window distribution and thus indicative of positive selection. (e) Comparison of the number of singletons (i.e., SNPs found only in one individual) vs. SNPs in Flavi‐ or Rhabdo‐nrEVEs in each population. Statistical differences were established by the Wilcoxon rank sum test (ns, not significant; *p < .05; **p < .01; ***p < .0001) [Colour figure can be viewed at wileyonlinelibrary.com]

FIGURE 7

nrEVEs with signal of selection. Tajima’s D and G12 values in the indicated nrEVEs in geographical populations of Aedes aegypti. Red indicates lower Tajima’s D or higher G12 values than the cutoffs for each chromosome in each population (Tables S2 and S3). Blue boxes indicate high Tajima’s D values (i.e., values higher than the 95% percentile for each chromosome in each population). Light grey boxes indicate nonsignificant tests. Only nrEVEs fixed in all tested populations and with either a significant Tajima’s D or a significant G12 values in at least one of the tested populations are shown [Colour figure can be viewed at wileyonlinelibrary.com]