The discovery, distribution, and diversity of DNA viruses associated with Drosophila melanogaster in Europe

Abstract

Drosophila melanogaster is an important model for antiviral immunity in arthropods, but very few DNA viruses have been described from the family Drosophilidae. This deficiency limits our opportunity to use natural host-pathogen combinations in experimental studies, and may bias our understanding of the Drosophila virome. Here, we report fourteen DNA viruses detected in a metagenomic analysis of 6668 pool-sequenced Drosophila, sampled from forty-seven European locations between 2014 and 2016. These include three new nudiviruses, a new and divergent entomopoxvirus, a virus related to Leptopilina boulardi filamentous virus, and a virus related to Musca domestica salivary gland hypertrophy virus. We also find an endogenous genomic copy of galbut virus, a double-stranded RNA partitivirus, segregating at very low frequency. Remarkably, we find that Drosophila Vesanto virus, a small DNA virus previously described as a bidnavirus, may be composed of up to twelve segments and thus represent a new lineage of segmented DNA viruses. Two of the DNA viruses, Drosophila Kallithea nudivirus and Drosophila Vesanto virus are relatively common, found in 2 per cent or more of wild flies. The others are rare, with many likely to be represented by a single infected fly. We find that virus prevalence in Europe reflects the prevalence seen in publicly available datasets, with Drosophila Kallithea nudivirus and Drosophila Vesanto virus the only ones commonly detectable in public data from wild-caught flies and large population cages, and the other viruses being rare or absent. These analyses suggest that DNA viruses are at lower prevalence than RNA viruses in D.melanogaster, and may be less likely to persist in laboratory cultures. Our findings go some way to redressing an earlier bias toward RNA virus studies in Drosophila, and lay the foundation needed to harness the power of Drosophila as a model system for the study of DNA viruses.

Keywords: DNA virus; Drosophila; adintovirus; bidnavirus; densovirus; endogenous viral element; filamentous virus; galbut virus; nudivirus.

Figure 1.

Genome structures and read depth. The plots show annotated coding DNA sequences (CDS, red and blue arrows), and terminal Inverted repeat (yellow boxes) for each of the near-complete virus genomes discussed. The read depth (pale blue) is plotted above the genome on a log scale for the population with the highest coverage in the DrosEU dataset. The five largest viruses (top) are plotted according to the 20 kbp scale bar, and the other viruses (bottom) are plotted according to the 2 kbp scale bar. The nudiviruses are circular, and have been arbitrarily linearised for plotting. Drosophila Esparto nudivirus was completed using public dataset (SRR3939042). Note that Drosophila Vesanto virus segments S07 and S11 were absent from the illustrated sample (lower right).

Figure 2.

Phylogenetic relationships. (A) Nudiviruses, hytrosaviruses, filamentous viruses, nucleopolyhedrosis viruses and nimaviruses, inferred from six concatenated protein coding genes. Note that these lineages are extremely divergent, and the alignment is not reliable at deeper levels of divergence. (B) Densoviruses, inferred from NS1. (C) Bidnaviruses (sometimes labelled ‘densovirus’) and adintoviruses (including representative polintons), inferred from DNA Polymerase B. (D) Pox and entomopoxviruses, inferred from three concatenated protein coding genes. All phylogenies were inferred from protein sequences by maximum likelihood, and scale bars represent 0.5 amino-acid substitutions per site. In each case, trees are mid-point rooted, viruses reported from Drosophila are shown in red, and sequences identified from virus transcripts in publicly available transcriptome assemblies are shown in blue, labelled by host species. The nudivirus from Phortica variegata was derived from PRJNA196337 (Vicoso and Bachtrog 2013). Alignments and tree files with bootstrap support are available in Supplementary Material.

Figure 3.

Drosophila Vesanto virus segment copy-number. (A) Heatmap showing the relative number of sequencing reads from each of the twelve Vesanto virus segments (columns), for each of the population samples (rows). Populations are included if at least one segment appeared at 1 per cent of the fly genome copy-number. Rows and columns have been ordered by similarity (dendrogram) to identify structure within the data. Colours show copy-number relative to the highest-copy segment, on a log scale. (B) Correlations in copy-number among the segments, with ‘significant’ correlations (P < 0.05, no corrections) shown with coloured ellipses, according to the direction (red positive, blue negative) and strength of correlation. The absence of strong negative correlations between segments encoding homologous proteins (e.g. S01, S03, S11, which all encode genes with homology to DNA Polymerase B) may indicate that these segments do not substitute for each other.

Figure 4.

Geographic distribution of DNA virus reads in European D.melanogaster. Maps show the spatial distribution of virus read copy-number (relative to fly genomes) on a non-linear colour scale. Data are shown for the five viruses that were each detected more than once (rows), separated by year and whether flies were collected relatively ‘early’ or ‘late’ in the season (columns).

Figure 5.

Geographic variation in estimated prevalence: Drosophila Kallithea nudivirus (A), Drosophila Linvill Road denosovirus (B), and the galbut virus EVE (C and D). Sampling sites are marked as white dots, and the colour gradient illustrates predictions from the INLA model, but with scale transformed to the predicted individual-level prevalence (%), assuming independence among individuals and population samples of size 40. Only Drosophila Kallithea nudivirus, Drosophila Linvill Road densovirus, and the galbut virus EVE displayed a significant spatial component, and only the EVE differed between seasons.

Figure 6.

Drosophila melanogaster harbours an endogenous genomic copy of galbut virus. (A) Maps show the spatial distribution of the DNA reads from the galbut EVE, as a percentage of fly genomes (maximum 13.8%) on colour scale. Rows show years of sampling, and columns show ‘early’ or ‘late’ samples in each year (B) The relationship between the galbut EVE and galbut virus sequences detectable in public datasets, illustrated by a Bayesian maximum clade-credibility tree inferred under a strict clock, with median-scaled node dates. The 95 per cent highest posterior density for the root date of extant galbut viruses is shown in blue (230–1,060 years before present), and the 95 per cent highest posterior density for the inferred date of insertion, is shown in red (20–290 years before present).