Virus discovery by deep sequencing and assembly of virus-derived small silencing RNAs

Abstract

In response to infection, invertebrates process replicating viral RNA genomes into siRNAs of discrete sizes to guide virus clearance by RNA interference. Here, we show that viral siRNAs sequenced from fruit fly, mosquito, and nematode cells were all overlapping in sequence, suggesting a possibility of using siRNAs for viral genome assembly and virus discovery. To test this idea, we examined contigs assembled from published small RNA libraries and discovered five previously undescribed viruses from cultured Drosophila cells and adult mosquitoes, including three with a positive-strand RNA genome and two with a dsRNA genome. Notably, four of the identified viruses exhibited only low sequence similarities to known viruses, such that none could be assigned into an existing virus genus. We also report detection of virus-derived PIWI-interacting RNAs (piRNAs) in Drosophila melanogaster that have not been previously described in any other host species and demonstrate viral genome assembly from viral piRNAs in the absence of viral siRNAs. Thus, this study provides a powerful culture-independent approach for virus discovery in invertebrates by assembling viral genomes directly from host immune response products without prior virus enrichment or amplification. We propose that invertebrate viruses discovered by this approach may include previously undescribed human and vertebrate viral pathogens that are transmitted by arthropod vectors.

Fig. 1.

Position and distribution of FHV and SINV siRNA contigs assembled from small RNAs sequenced from Drosophila S2 cells infected with the B2-deletion mutant of FHV (11) (A), a transgenic C. elegans strain in the RNAi-defective 1 (rde-1) mutant background carrying an FHV RNA1 replicon in which the coding sequence of B2 was replaced by that of GFP (29) (B), and adult mosquitoes infected with SINV (22) (C). Note that the length of RNA genomes was not drawn to scale.

Fig. 2.

Discovery of dsRNA viruses DTV (A) and DBV (B) and +RNA viruses DTrV (C) and MNV (D) from S2-GMR cells by virus discovery by deep sequencing and assembly of total small RNAs (vsSAR). Red bars refer to the virus-specific contigs initially identified by % sequence similarities of their encoded proteins to a viral protein in the databases. The contigs of DTV, DTrV, and MNV showed the highest similarities to Penaeid shrimp infectious myonecrosis virus (PsIMV), EEV, and WNV, respectively. However, four different members in the Birnaviridae were identified as the closest to DBV contigs: (a) Infectious bursal disease virus (IBDV), (b) DXV, (c) Marine birnavirus (MAV), and (d) Blotched snakehead virus (BSV). Gray bars refer to the contigs that were assembled from small RNAs of S2-GMR cells and subsequently mapped to specific viruses after the complete genomes were obtained. Note that the length of RNA genomes was drawn to scale and the ORFs encoded by the partial genome of DTrV (3,005 nucleotides in length) and MNV (1,130 nucleotides in length) were incomplete.

Fig. 3.

S2-GMR cells contained four infectious RNA viruses. (A) DTV, DBV, DXV, and ANV were all detected by RT-PCR in noncontaminated S2 cells 4 days after inoculation with the supernatant of the S2-GMR cells. Healthy S2 cells and the S2-GMR cells were used as controls. Primers used for RT-PCR were expected to yield specific products of 1,087, 1,030, 865, and 1,212 base pairs in length from DTV, DBV, DXV, and ANV, respectively. (B) Detection of an abundant DI-RNA derived from ANV RNA2 in S2-GMR cells (Right) that were much less abundant in S2 cells after inoculation with the supernatant of S2-GMR cells (Left) by Northern blot hybridizations using a probe recognizing the 3′-terminal 120 nucleotides of RNA2. (C) Structure of the cloned DI-RNA of ANV (Upper) and mapping of the perfectly matched 21-nt siRNAs sequenced from S2-GMR cells to the positive (blue) and negative (red) strands of ANV RNA2 (20-nt windows) (Lower).

Fig. 4.

Size distribution (A) and aggregate nucleotide composition (B) of virus-derived small RNAs in Drosophila OSS cells. For each viral genome or genome segment, the percent of either sense (red bar) or antisense (blue bar) viral small RNAs of distinct sizes over total reads with a length of 18–31 nucleotides with a perfect match is shown in A. Percent aggregate nucleotide compositions for all viral reads of 21-nt siRNA or 27-nt + 28-nt piRNAs were calculated, with the total numbers of reads in each size shown in parentheses.