The PAZ domain of Aedes aegypti Dicer 2 is critical for accurate and high-fidelity size determination of virus-derived small interfering RNAs

Abstract

The exogenous siRNA (exo-siRNA) pathway is a critical RNA interference response involved in controlling arbovirus replication in mosquito cells. It is initiated by the detection of viral long double-stranded RNA (dsRNA) by the RNase III enzyme Dicer 2 (Dcr2), which is processed into predominantly 21 nt virus-derived small interfering RNAs (vsiRNAs) that are taken up by the Argonaute 2 (Ago2) protein to target viral single-stranded RNAs. The detailed understanding of Dicer structure, function and domains owes much to studies outside the context of viral infection and studies in model organisms, and as such how Dcr2 domains contribute to detecting viral dsRNA to mount antiviral responses in infected mosquito cells remains less well understood. Here, we used a Dcr2 reconstitution system in Aedes aegypti derived Dcr2 knockout (KO) cells to assess the contribution of the PIWI-Argonaute-Zwille (PAZ) domain to induction of the exo-siRNA pathway following infection with Semliki Forest virus (SFV; Togaviridae, Alphavirus). Amino acids critical for PAZ activity were identified, and loss of PAZ function affected the production of 21 nt vsiRNAs-with enrichment of 22 nt SFV-derived small RNAs observed-and silencing activity. This study establishes PAZ domain's functional contribution to Dcr2 processing of viral dsRNA to 21 nt vsiRNAs.

Keywords: Dcr2; RNA interference; arbovirus; host response; mosquito cell; small RNA.

FIGURE 1.

Domain features of Dcr2. (A) Schematic representation of D. melanogaster and Ae. aegypti Dcr2, including functional domains: helicase domain, domain of unknown function (DUF), PAZ domain, RNase IIIA and IIIB domains, and dsRNA-binding domain (dsRBD); aa, amino acid. (B) Multiple sequence alignment of insect Dcr2 referring to Aaeg Dcr2 AAW48725, including the selected mutations of highly conserved amino acids for potential PAZ domain loss of function. Accession numbers for sequences used are indicated in Supplemental Table S1. Alignment was produced using Benchling (https://www.benchling.com/).

FIGURE 2.

Mutation and expression of Aaeg Dcr2. (A) Table showing produced Aaeg Dcr2 PAZ domain mutants M1–M4, the respective introduced loss-of-function mutations, as well as corresponding amino acid residues in Dm Dcr2; aa, amino acid. (B) Assessment of (myc-tagged) Aaeg Dcr2 expression by western blot analysis. AF319 cells were transfected with either pPUb plasmids expressing WT Dcr2 or Dcr2 PAZ domain mutants M1–M4, using pPUb-myc-eGFP as control. Anti-myc and anti-α-tubulin (control) antibodies were used. Representative of three independent repeats (other repeats in Supplemental Fig. S2).

FIGURE 3.

Functional analysis of Aaeg Dcr2 PAZ domain mutants. (A) Antiviral activity of Aaeg Dcr2. AF319 cells were transfected with expression plasmids encoding myc-tagged WT Dcr2 or Dcr2 PAZ domain mutants M1–M4 or eGFP (negative control). These cells were infected with SFV-FFLuc (MOI = 0.1) 24 h posttransfection (hpt). FFLuc levels were measured at 48 h postinfection (hpi) and are shown as mean ± SEM relative light units compared to eGFP control set to 1 from three independent repeats, performed in technical triplicate with (*) P < 0.05 according to Student's t-test. (B) Silencing activity of mutant Aaeg Dcr2. AF319 cells were cotransfected with (i) myc-tagged pPUb plasmids expressing WT Dcr2 or Dcr2 PAZ domain mutants M1–M4 (with eGFP as control), (ii) FFLuc and RLuc (internal control) reporter plasmids, and (iii) dsRNA targeting FFLuc (FFLuc dsRNA, blue) or eGFP (eGFP dsRNA; nonsilencing control, gray). At 24 hpt, FFLuc and RLuc levels were measured and are shown as mean ± SEM relative light units (FFLuc/RLuc) normalized to dseGFP from three independent repeats, performed in technical triplicate, with (*) P < 0.05, (**) P < 0.01, (***) P < 0.001 versus controls according to two-way ANOVA.

FIGURE 4.

Aaeg Dcr2 PAZ domain mutations affected 21 nt vsiRNA magnitude and length. (A) Small RNA sequencing of AF319 cells, transiently expressing Dcr2 PAZ domain mutants M1–M4 or WT Dcr2, or eGFP (as negative control), and infected with SFV (MOI = 1) at 48 hpi. Histogram of small RNA reads 18–30 nt in length, mapped to the SFV (SFV4 GenBank ID: KP699763) genome (positive numbers) and antigenome (negative numbers), with colors indicating first base nucleotide prevalence per size, shown as mean percentage of mapped reads (y-axis, percentage reads) from two independent experiments ±SD. Mapping of SFV-derived 21 nt vsiRNAs, mapped along the SFV genome (magenta) or antigenome (cyan) (y-axis, vsiRNA reads per million) ±SD. (B) Fold change of magnitude of SFV-derived small RNAs of lengths 20–22 nt of Dcr2 PAZ domain mutants M1–M4 compared to WT Dcr2.

FIGURE 5.

Aaeg Dcr2 PAZ domain mutations result in dysregulated vsiRNA duplex overlap lengths. (A) Heat maps showing mean overlap probability z-scores of 18–30 nt SFV-derived small RNAs from AF319 transiently expressing Dcr2 PAZ domain mutant M1–M4 or WT Dcr2, or eGFP (control) from two independent repeats, as initially characterized in Figure 4. Differing lengths of vsiRNAs, analyzed individually, are shown horizontally, and nucleotide overlaps are listed vertically. The red arrow labeled Dcr2 indicates the expected 2 nt overlap from dsRNA cleavage with cells boxed in black. The black arrow labeled pp (ping-pong) shows the expected 10 nt overlap from potential ping-pong amplification in the piRNA pathway. (B) Number of overlapping pairs per million mapped reads of 19–23 nt virus-derived small RNAs from AF319 transiently expressing Dcr2 PAZ domain mutants M1–M4 or WT Dcr2, or eGFP (control) infected with SFV. Data are shown as the mean of two independent repeats with the range of values.

FIGURE 6.

Mutations in the PAZ domain of Aaeg Dcr2 change the position and magnitude of cleavage over the SFV genome and virus-derived small RNA populations compared to WT Dcr2. Samples from Figure 4 were analyzed as follows. (A) Scatter plots showing normalized differential coverage (d_i) for each position in the SFV genome. Each panel compares WT Dcr2 with no Dcr2, an eGFP expressing Dcr2 negative treatment, or Dcr2 PAZ domain mutants M1–M4. The linear regression line is depicted within each plot, with the R² value indicating the fit of the model. The number of data points (n) with d_i > 0 over the SFV genome used for the calculation is shown on each graph. (B) Heatmap displaying pairwise R² values from the linear regressions, comparing the similarity of differential coverage profile between treatments, as indicated. (C) Multidimensional scaling (MDS) plots representing the relationships between different size classes of vsiRNA reads from WT Dcr2, eGFP control, and Dcr2 PAZ domain mutants M1–M4. Each plot corresponds to a specific size class of SFV-derived small RNA: 20, 21, 22, and 23 nt, as well as a combined plot of 20–23 nt. The axes represent the leading two log fold change (logFC) dimensions and the percentage of variation indicated.