RNase III nucleases from diverse kingdoms serve as antiviral effectors

Abstract

In contrast to the DNA-based viruses in prokaryotes, the emergence of eukaryotes provided the necessary compartmentalization and membranous environment for RNA viruses to flourish, creating the need for an RNA-targeting antiviral system. Present day eukaryotes employ at least two main defence strategies that emerged as a result of this viral shift, namely antiviral RNA interference and the interferon system. Here we demonstrate that Drosha and related RNase III ribonucleases from all three domains of life also elicit a unique RNA-targeting antiviral activity. Systemic evolution of ligands by exponential enrichment of this class of proteins illustrates the recognition of unbranched RNA stem loops. Biochemical analyses reveal that, in this context, Drosha functions as an antiviral clamp, conferring steric hindrance on the RNA-dependent RNA polymerases of diverse positive-stranded RNA viruses. We present evidence for cytoplasmic translocation of RNase III nucleases in response to virus in diverse eukaryotes including plants, arthropods, fish, and mammals. These data implicate RNase III recognition of viral RNA as an antiviral defence that is independent of, and possibly predates, other known eukaryotic antiviral systems.

Extended Data Figure 1. Characterization of RNaseIII^−/− cells

a, Sequence alignment of genetic alterations in the two alleles encoding Drosha in RNaseIII−/− cells. The deletion and insertion result in a frameshift and early stop codon. b, Graph depicts the ten most abundant miRNAs in each condition, parental 293T (WT) or RNase III−/− either mock or SINV-infected for 24 hrs as determined by Illumina small RNA deep sequencing. c, qPCR analysis of DGCR8 mRNA levels in mock treated and SINV-infected (MOI=0.1, 8hpi) NoDice and RNaseIII^−/− cells. d, Transcriptome profiling and correlation analyses of NoDice cells at baseline and RNaseIII^−/− cells transfected with GFP-tagged human Drosha for 72 hrs. Graph depicts data from two biological replicates per condition.

Extended Data Figure 2. Drosha depletion does not alter the response to IFN-I

a, qPCR analysis of IFIT1 mRNA levels in NoDice and RNaseIII^−/− fibroblasts treated with IFNβ (100 u/ml, 8 hrs). b, Western blot of NoDice and RNaseIII^−/− fibroblasts infected with SINV for 1 hr prior to administration of indicated amounts of IFNβ for 24 hrs. c–e, Northern blot (c) and western blot (d) of primary Rnasenf/f ear-derived fibroblasts treated with indicated AdVs for 2–3 days and then treated with either 100U IFNβ for 6 hrs (b) or infected with SINV for 24 hrs (d). e, mRNA-Seq of total RNA from samples in (c). Heatmap depicts known mouse ISGs and IFN down regulated genes with a log2 fold change greater than 1, as defined by the Interferome database (www.interferome.org).

Extended Data Figure 3. Characterization of Drosha-2A cells

a, Immunofluorescence of human fibroblasts stably expressing GFP-tagged Drosha-WT or Drosha-2A (S300A/S302A). b, The indicated Flag-tagged proteins were immunoprecipitated from whole cell extracts (WCE) and incubated at 37 °C with in vitro transcribed genome of SIN124. Production of pre-miR-124 was determined by small RNA northern blot. c, Indicated cell types were infected with SINV at an MOI of 0.001 and viral titers were determined at 16 hpi. Shown is average and standard deviation of three independent experiments, with p<0.05 as determined using a two-tailed student’s t-test.

Extended Data Figure 4. Drosha-RB-RIIIDmut recognizes stem-loop structures in SINV RNA

a, Immunoprecipitation of exogenously expressed Flag-tagged proteins. Shown is protein expression in the whole cell extract (WCE) and after immunoprecipitation (IP) with a Flag-specific antibody. b, Cells were transfected with Flag-tagged SeV-N or Drosha-RBmut and infected at 36 hpt with SINV at an MOI of 3. 8 hpi, Flag-tagged proteins were immunoprecipitated and bound RNA was isolated to perform qPCR. Graph shows SINV RNA levels relative to input and normalized to tubulin. The average of three independent experiments is shown. Error bars depict standard deviation and * denotes p<0.05 using a one-tailed student’s t-test. c, Prediction of the structure of the 5′ 200 nts of the SINV genome using RNAfold. d, EMSA was performed with the indicated immunoprecipitated proteins and radio-labeled in vitro transcribed RNA comprising the 5′ 200 nts of the SINV genome. Unbound genome is indicated as ‘Free RNA’.

Extended Data Figure 5. Using virus engineering to discern Drosha’s antiviral mechanism

a, Schematic of the SINV replicon encoding Gaussia luciferase in place of the structural polyprotein used in Figs. 3e–g. b, Schematic of the SINV temperature sensitive mutant (SIN-RdRpts). Star denotes ts point mutant. c, NoDice and RNaseIII^−/− cells were infected with virus depicted in b, at an MOI of 10 and incubated at 40 °C, a temperature at which the mutant viral RdRp is completely inactive. Levels of genomic (g) SINV RNA were determined by qPCR at indicated times post infection. Data is representative of two independent experiments where each condition was done in triplicate. d, Schematic of the SINV encoding Firefly luciferase in the nsP3 region and an inactive RdRp (SIN-nsP3Luc) e, Graph depicts levels of in vitro translation of Firefly luciferase produced from virus in d, in the presence of membrane fractions from control or Drosha-2A. The data shown is the average of three independent experiments.

Extended Data Figure 6. Localization of and miRNA production from cytoplasmic viruses in diverse eukaryotes

a, Zebrafish embryos were inoculated with SINV for 24 hrs and then analyzed by fluorescence in situ hybridization (FISH) using a probe complementary to the capsid region of the genome. b, qPCR analysis of zebrafish embryos treated with indicated morpholinos for 2 days (n=4). c, Arabidopsis thaliana protoplasts were mock or TCV-infected for 40 hrs and then analyzed by FISH using a Cy3-labelled probe complementary to bases 1210–1259 of the TCV genome. d, Quantification of mature miR-124 production from recombinant TCV was performed using the TaqMan miRNA assay on RNA from Fig. 4b. All samples were normalized to endogenous snoR66. Quantifications of each sample were performed in triplicate and error bars denote standard deviation from two biological replicates.

Extended Data Figure 7. The impact of diverse RNase III members on virus infection

a, Schematic depicting core domains of human Drosha, C-terminal region of C. intestinalis Drosha, or full length RNase IIIs of S. pombe, M. maripaludis, and S. pyogenes. Domains depicted include: Proline-rich, P-rich; arginine serine-rich, RS-rich; Conserved Central Domain, CED; RNAseIII domain (RIIID) and double stranded RNA binding domain (dsRBD). b, Western blots from BSR-T7 cells, co-transfected with the indicated RNase III-expression plasmids and SeV rescue plasmids encoding SeV-GFP genome, SeV-N, SeV-P, and SeV-L genes. RNase III expression was determined at 48 hpt and virus replication at 72 hpt. c, Western blot of DL1 cells treated with indicated dsRNA for 3 days and subsequently infected with SINV (MOI=1) for 96 hrs. d, 293T (WT) or RNaseIII^−/− cells were infected with SINV for 24 hrs. Graphs depict the number of SINV reads mapping to indicated positions along the viral genomes from the small RNA deep sequencing performed in Extended Data Fig. 1b.

Figure 1. Drosha mediates miRNA-independent antiviral activity

a, Northern blot (NB) of RNA from NoDice and RNaseIII^−/− cells reconstituted with indicated plasmids. Blots probed for miR-93 and U6. b, Western blot (WB) of whole cell extract from NoDice and RNaseIII^−/− cells infected with SINV (MOI=0.01) at 4, 8, and 12 hpi. Blot probed for SINV Capsid (SIN-C) and pan-Actin (Actin). c–f, WBs of Ross River virus (MOI=0.1) (c), Langat virus (MOI=0.1) (d), Influenza virus (MOI=1) and (e), GFP-encoding Sendai virus (MOI=1) (f). Protein levels were assessed at 24 hpi using virus-specific or GFP antibodies as indicated. g, RNA-Seq correlation analyses of NoDice and RNaseIII^−/− cells at baseline. h, As described in g, except cells were treated with dsRNA for 8 hrs. Graphs in g and h depict data from biological replicates.

Figure 2. The RNA-binding domain of Drosha is essential for virus inhibition

a, Core domains and variants of Drosha including: Proline-rich, P-rich; arginine serine-rich, RS-rich; Conserved Central Domain, CED; RNAseIII domain (RIIID) and double-stranded RNA binding domain (dsRBD). Black boxes indicate point mutations (mut) b, NB and WB of RNaseIII^−/− cells transfected with the indicated Drosha variants and cytoplasmic miR-124. WB depicts Drosha, Dicer, and pan-Actin. c, WB and d, titers from RNaseIII^−/− cells expressing the indicated transcripts and SINV RNA 24 hpt. Graph denotes average titers obtained with error bars generated using standard deviation from three independent experiments. ‘ns’ denotes not significant, all other conditions had a p-value of <0.05. e, WB as described in c with Drosha RBmut. f, WB of input and Flag-immunoprecipitated (IP) fractions derived from 293T cells expressing indicated transcripts and Flag-DGCR8.

Figure 3. Cytoplasmic Drosha binds stem-loop structures in viral RNA to inhibit RdRp activity

a, RNA hairpins (HP-1/-2) enriched by Drosha-RBmut-based SELEX as predicted by RNAfold. b, same as a, using GFP as control (Ctl-1/02) bait. c, RNA-based EMSA of hairpins in a, and b, using recombinant Drosha-RB. d, EMSA as in c, performed with immunoprecipitated GFP, Sendai nucleoprotein (SeV-NP), or Drosha-RBmut. e, NB of RNA from NoDice and RNaseIII^−/− transfected a SINV-based replicon denoting genomic (g) and subgenomic (sub-g) SINV RNA. f–g, luciferase (f) and antigenome expression (g) of replicon as described in e. Data is representative of independent experiments where each condition was done in triplicate. Error bars denote standard deviation. h, WB of cytoplasmic membrane fractions from control or Drosha-2A cells expressing SINV replicase-components. i, in vitro minus strand RNA synthesis assay utilizing membrane fractions from h.

Figure 4. RNase III inhibition of virus replication is highly conserved

a, NB of RNA from zebrafish embryos treated with indicated morpholinos, and inoculated 2 days later with wild type SINV (WT) or a strain encoding an artificial miRNA (amiRNA). Blot depicts amiRNA and U6 at 40hrs post infection b, NB of RNA from Arabidopsis thaliana protoplasts treated with a Turnip crinkle virus (TCV) containing a scrambled sequence (Scbl) or miR-124 (124). c and d, WB from RNaseIII^−/− cells, co-transfected with the indicated plasmids and either in vitro transcribed SINV gRNA or Langat virus rescue plasmid. e and f, Small RNA-Seq of DL1 cells treated with indicated dsRNA for 3 days and subsequently infected with SINV (MOI=1) (e) or DCV (MOI=7) (f). Graphs depict the number of reads mapping to indicated positions along the viral genomes.