Sindbis virus induces the production of a novel class of endogenous siRNAs in Aedes aegypti mosquitoes

Abstract

Small RNA regulatory pathways are used to control the activity of transposons, regulate gene expression and resist infecting viruses. We examined the biogenesis of mRNA-derived endogenous short-interfering RNAs (endo-siRNAs) in the disease vector mosquito Aedes aegypti. Under standard conditions, mRNA-derived endo-siRNAs were produced from the bidirectional transcription of tail-tail overlapping gene pairs. Upon infection with the alphavirus, Sindbis virus (SINV), another class of mRNA-derived endo-siRNAs was observed. Genes producing SINV-induced endo-siRNAs were not enriched for overlapping partners or nearby genes, but were enriched for transcripts with long 3' untranslated regions. Endo-siRNAs from this class derived uniformly from the entire length of the target transcript, and were found to regulate the transcript levels of the genes from which they were derived. Strand-specific quantitative PCR experiments demonstrated that antisense strands of targeted mRNA genes were produced to exonic, but not intronic regions. Finally, small RNAs mapped to both sense and antisense strands of exon-exon junctions, suggesting double-stranded RNA precursors to SINV-induced endo-siRNAs may be synthesized from mature mRNA templates. These results suggest additional complexity in small RNA pathways and gene regulation in the presence of an infecting virus in disease vector mosquitoes.

Figure 1. Steady-state, but not dsSINV-induced, endo-siRNAs derive primarily from overlapping gene pairs in Ae. aegypti

(A) The four categories of overlapping gene considered in this study. (B) The percentage of each overlapping gene category was calculated for the entire Ae. aegypti mRNA gene set (genome), as well as for genes producing endo-siRNAs in uninfected (−), dsSINV-infected, or dsSINV-NoVB2-infected Ae. aegypti. Bars represent the percentage of genes annotated as non-overlapping; tail-tail (T-T); head-head (H-H); complete (C); or non-exon overlapping (non-exon). Percentages above each bar represent the sum of all four categories of overlap. Numbers above horizontal bars indicate the total number of genes used in the analysis. (C) For the genes identified as “non-overlapping”, the distance to the nearest mRNA-encoding gene on the opposing strand was calculated for both the 5' (upstream) and 3' (downstream) directions.

Figure 2. dsSINV-induced, mRNA-derived endo-siRNAs derive from abundant transcripts with unusually long 3'UTRs

(A) Normalized expression of Ae. aegypti gene transcripts (All) or transcripts corresponding to dsSINV-induced endo-siRNA genes. Horizontal bar indicates the mean; error bars indicate 95% confidence intervals. Region denoted by bracket indicates the top 200 expressed genes. Predicted length of mRNA transcripts in the current annotation (Genome) or in dsSINV-induced endo-siRNA producing genes either in total (B), without 3'UTR (C) or 3'UTR alone (D). Statistical significance was determined using the Mann Whitney test [pairwise comparisons for A, B and C], or the Kruskal-Wallis test followed by Dunn's Multiple Comparison test [three-way comparison for D].

Figure 3. dsSINV-induced endo-siRNAs derive from the entire mRNA

Endo-siRNA distributions on genomic intervals for a tail-tail overlapping gene pair which produces small RNAs in the absence of dsSINV (A), or from genes which produce endo-siRNAs only upon dsSINV infection (B). Y-axis indicates the number of unique-mapping 21 nt small RNAs per 20 nt bin, with neighboring bins overlapping by 10nt. Gene structure is indicated by block arrows (exons) and bent lines (introns). Bar graphs indicate the normalized distribution of unique-mapping 21 nt small RNAs per nucleotide of exon.

Figure 4. dsSINV-induced endo-siRNAs do not appear to derive from antisense or run-on transcription

Each bar graph displays the quantification of the top or bottom (coding) strand transcripts corresponding to gene AAEL010787 in uninfected (−) Ae. aegypti, or in Ae. aegypti four days after infection with dsSINV or dsSINV-NoVB2 viruses. Assays corresponding to exonic (block arrows), intronic (solid lines), or antisense strand (dotted line) regions are indicated.

Figure 5. dsSINV-induced endo-siRNAs map to antisense strands of exon-exon junctions

(A) Schematic representation of the exon-exon junction sequences used to map small RNA reads from each library, where the last 25 bp of each exon combined with the first 25 bp of the following exon for each gene in the gene set. Small RNAs with a 5' mapping position of 11–20 were counted as being derived from the exon-exon junction, with all other mapping positions considered being from the ends of their respective exons. (B) Length distribution of small RNAs mapping to the sense (s) or antisense (as) strand of exon-exon junctions or ends. (C) The ratio of sense (s) to antisense (as) and junction-mapping (J) to end-mapping (E) reads was plotted for each library for 21nt small RNAs. (D, E) Antisense small RNAs mapping to exon-exon junctions of genes AAEL010787 and AAEL004865; exons are indicated with block arrows with the corresponding bases capitalized. Introns are indicated with thin lines and lower case bases. Only the ends of introns are shown, with (…) indicating a jump in sequence. Small RNAs (shown as cDNA versions, for simplicity) mapping to the top or bottom strand of each sequence are indicated, with alignment gaps denoted by (---).

Figure 6. dsSINV-induced endo-siRNAs restrict the steady-state level of full-length target transcripts in an RNAi-dependant manner

(A) Northern blot analysis was performed using total RNA derived from uninfected (−) Ae. aegypti, or Ae. aegypti infected with dsSINV or dsSINV-NoVB2 at 1, 2, 3, or 4 days post infection, as indicated. (B) Phosphorimager-based quantification of each blot, with RNA levels derived from the time of injections (Dy0) set to 100%. Where multiple isoforms are present, quantification was based on the longest isoform.