Novel virus discovery and genome reconstruction from field RNA samples reveals highly divergent viruses in dipteran hosts

Abstract

We investigated whether small RNA (sRNA) sequenced from field-collected mosquitoes and chironomids (Diptera) can be used as a proxy signature of viral prevalence within a range of species and viral groups, using sRNAs sequenced from wild-caught specimens, to inform total RNA deep sequencing of samples of particular interest. Using this strategy, we sequenced from adult Anopheles maculipennis s.l. mosquitoes the apparently nearly complete genome of one previously undescribed virus related to chronic bee paralysis virus, and, from a pool of Ochlerotatus caspius and Oc. detritus mosquitoes, a nearly complete entomobirnavirus genome. We also reconstructed long sequences (1503-6557 nt) related to at least nine other viruses. Crucially, several of the sequences detected were reconstructed from host organisms highly divergent from those in which related viruses have been previously isolated or discovered. It is clear that viral transmission and maintenance cycles in nature are likely to be significantly more complex and taxonomically diverse than previously expected.

Figure 1. Common nucleotide polymorphisms observed in AACV and CAZV assemblies.

Only nucleotide variations with a frequency in the mapped reads of at least 20% of the most common nucleotide present at that position are shown. The nucleotide present in the chosen reference assembly is shown first.

Figure 2. Analysis of CBPV and AACV sequences.

(A) Map of the CBPV RNA1 genome segment. The region covered by contig KF298264 (AACV) is indicated by the orange bar. The read coverage density is indicated in green. (B) Analysis of variability at synonymous sites in an alignment of the currently available full-length CBPV sequences (EU122229 and EU122231) and AACV (KF298264). Shown are the degree of variability at synonymous sites in a 75-codon sliding window, relative to the average in the ORF1-ORF3 frameshift fusion (obs/exp), and the corresponding statistical significance (p-value). (C) Positions of stop codons in the three forward reading frames in the three sequences (KF298264 - top row of triangles in each panel; CBPV EU122231 and EU122229 - bottom two rows of triangles in each panel). (D, E and F) Corresponding figures for RNA2 (KF298265 - top row of triangles in each panel; CBPV EU122232 and EU122230 - bottom two rows of triangles in each panel). Note that AACV RNA2 lacks a homolog of the CBPV ORF1, and has a shorter 3' UTR than CBPV, as indicated by gaps in the orange bar.

Figure 3. CBPV and AACV 5' and 3' UTR sequences.

Initiation and termination codons are highlighted in green and red respectively. CBPV and AACV sequences share common 5' and 3' motifs. GxGGGAA (blue-grey), AUAAGUC (orange) and other motifs are present in the 3' UTR of both RNA1 and RNA2 of both viruses, whereas a GU...AAACxU motif (salmon) is present at the 5' end of CBPV RNA1 and RNA2 and AACV RNA2. The absence of this motif at the 5' end of the AACV RNA1 sequence suggests that this sequence is incomplete (see text).

Figure 4. Bayesian maximum likelihood (ML) phylogenetic tree for AACV RdRp and related sequences.

The tree is midpoint-rooted and, for clarity, only posterior probability values >80% are shown.

Figure 5. Predicted frameshift sites.

(A) In AACV, CBPV and LSVs, +1 frameshifting for RdRp expression is predicted to occur on a conserved UUU_CGU motif (highlighted in cyan; P-site slippage on UUU_C), similar to the site of +1 frameshifting in influenza A virus PA-X expression [32]. KF298264 – AACV; EU122229 and EU122231 – CBPV; HQ871931 – LSV1; HQ888865 – LSV2. (B) In entomobirnaviruses, -1 frameshifting for VP4N-X expression is predicted to occur on a conserved U_UUU_UUA motif (highlighted in orange; tandem P- and A-site slippage), which is a particularly slippery site for -1 frameshifting [38]. This may be further supplemented by some level of tandem P- and A-site -1 slippage on U_CCU_UUU (ESV, CYV, MXV, CAZV) or A_AAU_UUU (DXV) (grey and orange highlighting). Note the increased nucleotide conservation downstream of the U_UUU_UUA motif, consistent with overlapping features. KF298271 – CAZV; JN589003 – ESV; JQ659254 – CYV; JX403941 – MXV; U60650 – DXV.

Figure 6. Analysis of entomobirnavirus sequences.

(A) Map of ESV genome segment A. The region covered by contig KF298271 (CAZV) is indicated by the orange bar. The read coverage density is indicated in green. (B) Analysis of variability at synonymous sites in an alignment of the currently available entomobirnavirus sequences. Shown are the degree of variability at synonymous sites in a 45-codon sliding window, relative to the average in the pVP2-VP4-VP3 ORF (obs/exp) and the corresponding statistical significance (p-value). (C) Positions of stop codons in the three forward reading frames in the five sequences (from top to bottom in each panel: U60650 - DXV, KF298271 - CAZV, JX403941 - MXV, JQ659254 - CYV, JN589003 - ESV). (D, E and F) Corresponding figures for segment B (from top to bottom in each panel: AF196645 - DXV, KF298272 - CAZV, JX403942 - MXV, JQ659255 - CYV, JN589002 - ESV).

Figure 7. Alignments of entomobirnavirus 5' and 3' UTR sequences, indicating that the CAZV sequences are nearly complete.

Initiation and termination codons are highlighted in green and red respectively. Note that both entomobirnavirus segments have upstream AUG codons, although it is not known whether these are utilised. ESV - JN589003/JN589002, CYV - JQ659254/JQ659255, MXV - JX403941/JX403942, CAZV - KF298271/KF298272.

Figure 8. Bayesian maximum likelihood (ML) phylogenetic tree for CAZV RdRp and related sequences.

The tree is midpoint-rooted and, for clarity, only posterior probability values >80% are shown.

Figure 9. Bayesian maximum likelihood (ML) phylogenetic tree for the flavivirus-like KF298267 and related sequences.

The tree is midpoint-rooted and, for clarity, only posterior probability values >80% are shown.

Figure 10. Bayesian maximum likelihood (ML) phylogenetic tree for the bunyavirus-like KF298274 and related sequences.

The tree is midpoint-rooted and, for clarity, only posterior probability values >80% are shown.

Figure 11. Neighbour-joining (NJ) phylogenetic tree for the orbivirus-like KF298266 and KF298273 and related sequences.

Nodes with <80% bootstrap support have been collapsed.

Figure 12. Codon usage statistics for the RdRp ORFs in the narnavirus-like contigs KF298275 and KF298276.

In-frame forward read-direction stop codons (red) are necessarily absent. Reverse complements of in-frame but reverse read-direction stop codons are highlighted in orange; a single UUA codon corresponds to the UAA stop codon of the >1000-codon reverse-read-direction ORF in KF298276.

Figure 13. Neighbour-joining phylogenetic tree for the RdRp amino acid sequences of the narnavirus-like KF298275 and KF298276 and related sequences.

Nodes with <80% bootstrap support have been collapsed.