Drosha acts in the cytoplasm during infection and restricts positive-strand RNA viruses by binding structured RNA rather than by producing small interfering RNAs
lab-led
Drosha as an interferon-independent antiviral factor
lab-ledDiscovery
Loss of the nuclear RNase III enzyme Drosha, but not of Dicer, increases RNA virus replication in mammalian fibroblasts, and diverse RNA viruses drive Drosha into the cytoplasm by CRM1-dependent export in a manner that does not require new protein synthesis, RIG-I, TBK1 or type I interferon signaling.
- Senior authors
- Benjamin R. tenOever
- Correspondence
- Benjamin R. tenOever
Research areas & themes
Small RNA Biology and the Limits of Antiviral SilencingRNase III Enzymes as Interferon-Independent Antiviral Effectors
Citation
Shapiro JS, Schmid S, Aguado LC, Sabin LR, Yasunaga A, Shim JV, Sachs D, Cherry S, tenOever BR. Drosha as an interferon-independent antiviral factor. Proceedings of the National Academy of Sciences. 2014. Volume 111, issue 19, pages 7108-7113. DOI 10.1073/pnas.1319635111. PMID 24778219. PMCID PMC4024876.
One-sentence contribution
Loss of the nuclear RNase III enzyme Drosha, but not of Dicer, increases RNA virus replication in mammalian fibroblasts, and diverse RNA viruses drive Drosha into the cytoplasm by CRM1-dependent export in a manner that does not require new protein synthesis, RIG-I, TBK1 or type I interferon signaling.
Executive summary
Antiviral RNA interference in plants, nematodes and arthropods depends on Dicer-generated virus-derived small interfering RNAs, and chordates were thought to have replaced this with the protein-based type I interferon system. Reports of residual mammalian antiviral RNA interference reopened the question of whether the small RNA machinery still contributes to defense in mammalian somatic cells, and if so through which enzyme.
The authors compared virus replication in primary fibroblasts conditionally deleted for Drosha or for Dicer using Cre-expressing adenoviral vectors, and followed the subcellular behavior of Drosha during infection. Sindbis virus and vesicular stomatitis virus both replicated to higher levels in the absence of Drosha, while deletion of Dicer had no significant effect. Infection with positive sense, negative sense and segmented nuclear RNA viruses, and transfection of double-stranded RNA, all drove Drosha from the nucleus into the cytoplasm within hours. Export depended on CRM1 and did not require de novo translation, RIG-I, TBK1 or the type I interferon receptor. Cytoplasmic localization was required for full antiviral activity, and mutation of serines 300 and 302 to alanine produced a constitutively cytoplasmic Drosha with the strongest restriction of Sindbis virus.
Drosha cleaved Sindbis virus genomic RNA in vitro and altered the host transcriptome during infection, but small RNA sequencing found no virus-derived small interfering RNA signature in either fibroblasts or Drosophila cells. The antiviral activity is therefore attributed to something other than canonical RNA interference.
Scientific context
In plants, nematodes and arthropods, Dicer cleaves viral double-stranded RNA into small interfering RNAs that direct RNA-induced silencing complex cleavage of homologous viral RNA. Chordates retain much of this machinery but use it predominantly for microRNA biogenesis, with Drosha acting first in the nucleus and Dicer second in the cytoplasm. Virus-derived small interfering RNAs have been hard to detect in interferon-competent cells, yet two 2013 reports from other laboratories described antiviral RNA interference in mammalian cells, and a commentary co-authored by tenOever with Cullen and Cherry framed the open question of whether this is physiologically relevant. Separately, the tenOever laboratory had shown that a cytoplasmic RNA virus can be engineered to produce a functional microRNA and that this depends on Drosha appearing in the cytoplasm. The gap the paper addresses is whether either RNase III enzyme contributes to antiviral defense in ordinary somatic cells and what the responsible activity is.
Central question
Do the mammalian RNase III enzymes Drosha and Dicer contribute to the cellular response to RNA virus infection in somatic cells, and if the contribution is Drosha-dependent, does it operate through the production of virus-derived small interfering RNAs or through some other activity?
Experimental strategy
The design separates the two enzymes genetically and then asks what the responsible enzyme is doing. Conditional floxed alleles for Drosha and for Dicer in primary fibroblasts, disrupted by replication-incompetent adenoviral vectors delivering Cre, allow a clean loss-of-function comparison in the same cell background, with loss of the Drosha- and Dicer-dependent miR-93 as the functional readout of deletion and the Drosha- and Dicer-independent U6 as the control. Multicycle growth curves at low multiplicity then report on replication.
Sindbis virus was chosen because it had been shown to tolerate engineered microRNA production, which argues it does not encode a suppressor of RNA silencing, and the authors confirmed this by showing that inserted miR-124 target sites silence the virus when the microRNA is supplied. Vesicular stomatitis virus provides a negative sense comparator and a translation-independent test, since it packages its own polymerase and can therefore make transcripts under cycloheximide.
Localization was addressed with immunofluorescence across a virus panel, with subcellular fractionation after CRM1 knockdown, and with a genetic epistasis panel using Ddx58, Tbk1 and Ifnar1 deficient cells, reading out cytoplasmic Drosha activity through processing of a virus-encoded cytoplasmic primary microRNA rather than through imaging alone. Phosphosite mutants at serines 300 and 302 test whether localization and antiviral activity are coupled. Mechanism was probed by in vitro cleavage of viral genomic RNA with purified Drosha and by small RNA and messenger RNA sequencing with and without Drosha, in mammalian cells and in Drosophila cells.
Key findings
-
Drosha was nuclear in mock-treated murine fibroblasts and accumulated in the cytoplasm within six hours of infection with Sindbis virus, vesicular stomatitis virus or an NS1-deficient influenza A virus, and after transfection of double-stranded RNA (Figure 1A). The generality across genome types and the response to a pathogen-associated molecular pattern alone are direct observations.
-
Drosophila DL1 cells infected with a Sindbis virus encoding primary miR-124 produced mature miR-124, and depleting Drosha abolished mature microRNA production while increasing unprocessed virus-derived primary microRNA and viral RNA (Figure 1B and 1C). The authors read this as cytoplasmic Drosha activity being conserved in arthropods.
-
Sindbis virus carrying two or four miR-124 target sites was silenced only when miR-124 was supplied exogenously, indicating the virus does not encode a suppressor of RNA silencing (Figure 2A).
-
Deletion of Drosha in conditional knockout fibroblasts raised Sindbis virus titers significantly across a multicycle growth curve and raised capsid protein levels, while deletion of Dicer did not significantly change titers or protein (Figure 2C to 2F). Vesicular stomatitis virus behaved the same way (Supplementary Figure 2). This is the central loss-of-function result.
-
Knockdown of CRM1 prevented cytoplasmic accumulation of Drosha after infection and abolished processing of the cytoplasmic virus-derived primary microRNA while leaving endogenous miR-93 intact (Figure 3A to 3C). Cycloheximide did not prevent production of vesicular stomatitis virus-derived miR-124 despite loss of viral G protein (Figure 3D). Together these place the source of cytoplasmic Drosha in active nuclear export rather than in retention of newly made protein.
-
Cytoplasmic Drosha activity was retained in cells lacking RIG-I, TBK1 or IFNAR1 (Supplementary Figure 3). The authors interpret this as evidence that the response is a separate, interferon-independent arm. The experiment shows that these three components are not required, and does not exclude other sensing routes.
-
GFP-tagged Drosha that is wild type or phosphomimetic at serines 300 and 302 was nuclear, while the alanine mutant was constitutively cytoplasmic. Infection moved the wild-type protein to the cytoplasm but not the phosphomimetic form (Figure 4A). In a transfected genomic RNA assay, wild-type and phosphomimetic Drosha attenuated Sindbis virus by about one log while the constitutively cytoplasmic alanine mutant attenuated it by more than two logs (Figure 4B and 4C). The authors infer a virus-inducible phosphatase acting on these residues. No phosphatase is identified and the inference is not tested.
-
Purified Drosha cleaved Sindbis virus genomic RNA in vitro, producing the expected pre-miR-124 product of about 55 nucleotides from the embedded hairpin and also generating species of about 20 to 25 nucleotides from the nonstructural polyprotein region, while purified GFP did not (Figure 5A).
-
Small RNA deep sequencing of infected fibroblasts recovered over 675,000 reads across the Sindbis virus genome with no enrichment of the 21 nucleotide species characteristic of virus-derived small interfering RNAs, and loss of Drosha increased total viral small RNA reads without changing their profile or genomic distribution. The same was true in Drosophila cells (Figure 5B and 5C). This is the principal negative result and it is what separates the observed activity from canonical antiviral RNA interference.
-
RNA sequencing of infected cells with and without Drosha showed more than 25 messenger RNAs induced greater than fivefold in the absence of Drosha, a comparable number downregulated, and more than 25 noncoding RNAs of unknown function upregulated, with Itga2 and Hspa1a confirmed by quantitative PCR (Figure 5D and Supplementary Figure 4).
Mechanistic model
The study does not establish a definitive mechanism for the antiviral effect, and the authors say as much, describing their account as speculation. What the data constrain is the following. Drosha limits RNA virus replication in primary fibroblasts, this restriction does not require Dicer and does not correlate with production of canonical virus-derived small interfering RNAs, and full restriction requires that Drosha be in the cytoplasm, where infection places it through CRM1-dependent export that is triggered independently of RIG-I, TBK1 and type I interferon signaling. Two non-exclusive activities are consistent with the data. Drosha can cleave structured regions of the viral genome directly, which would be expected to expose unprotected ends and promote decay, and Drosha shapes the host transcriptome during infection in ways that include induction and repression of dozens of transcripts.
Several things are explicitly unresolved. The authors do not know whether the transcriptome effect originates in the nucleus or the cytoplasm. They do not identify the phosphatase implied by the serine 300 and 302 data, nor the sensing pathway that triggers export. They raise as an open question how cytoplasmic Drosha distinguishes viral from host RNA, suggesting adenine and uracil content or the primary sequence determinants recently described for primary microRNA hairpins, and they note it would be interesting to know whether viral genomes avoid such motifs. None of that is tested here. The claim that the response is conserved from arthropods to chordates rests on the Drosophila microRNA processing and small RNA data and is an extrapolation from those to a conserved defense.
Conceptual or technical advance
A microRNA biogenesis enzyme is shown to have a function in virus restriction that is separable from its canonical pathway, since Dicer loss has no equivalent effect and no small interfering RNA signature appears. This makes the subcellular relocalization of a nuclear RNase III enzyme a measurable, virus-inducible event with a defined transport requirement and a defined phosphorylation switch, and it supplies a tractable assay for that event in the form of processing of a virus-encoded cytoplasmic primary microRNA. It also sharpens the ongoing debate about mammalian antiviral RNA interference by showing that a genuine small RNA machinery contribution to antiviral defense can exist without the small interfering RNA products that the debate has centered on.
Relationship to the broader research program
The paper grows directly from earlier tenOever laboratory work showing that cytoplasmic RNA viruses can be engineered to make microRNAs, which required the unexpected observation of Drosha outside the nucleus, reported by Shapiro and colleagues in 2010 and 2012. The engineered microRNA target site approach used here as a control for suppressor of RNA silencing activity is the same tool the laboratory developed for tropism restriction and biocontainment. Alongside Schmid and colleagues 2010, which asked what antiviral transcription persists without interferon signaling, this paper asks what antiviral activity persists without interferon signaling at the level of RNA, and shares an author between the two. Category 3 synthesis, visible only when these papers are placed side by side, is a sustained interest in arms of the antiviral response that operate outside the interferon axis and in repurposing the RNA silencing machinery both as a tool and as a subject.
Related publications
- Shapiro and colleagues 2010, RNA, Noncanonical cytoplasmic processing of viral microRNAs. Predecessor. Established that a cytoplasmic RNA virus can yield a functional microRNA, which is the observation that led here.
- Shapiro and colleagues 2012, RNA, Evidence for a cytoplasmic microprocessor of pri-miRNAs. Predecessor. Identified the Drosha translocation event whose generality and function are examined here.
- Schmid and colleagues 2010, Journal of Biological Chemistry. Conceptual extension. Shares the question of interferon-independent antiviral responses and shares an author.
- Langlois and colleagues 2012, Molecular Therapy, In vivo delivery of cytoplasmic RNA virus-derived miRNAs. Methodological foundation. Source of the VSV124 virus used in the translation-independence experiment.
- Perez and colleagues 2009 and Langlois and colleagues 2013. Methodological foundation. Established the engineered microRNA target site approach used here to test for a suppressor of RNA silencing.
- Cullen, Cherry and tenOever 2013, Cell Host and Microbe. Review or synthesis. Frames the question of whether RNA interference is a physiologically relevant mammalian antiviral response, which this paper addresses.
- Sabin and colleagues 2013, PLoS ONE. Companion. Fly work from a co-author's laboratory on Dicer-2 processing of viral RNA species, cited as agreeing with the Drosophila results here.
Limitations and boundaries
The mammalian work is confined to fibroblasts, predominantly primary mouse lung fibroblasts and transformed human and hamster lines, and the authors qualify their Dicer conclusion as holding at least in the context of primary fibroblasts. No animal infection is performed, so the physiological weight of the effect in a whole organism is not addressed. The replication phenotype rests on two cytoplasmic RNA viruses, Sindbis virus and vesicular stomatitis virus, and the influenza A virus experiment addresses only Drosha localization, not replication. The magnitude of the effect on titer is modest relative to the effect of ectopic expression of the cytoplasmic Drosha mutant, and the ectopic experiments involve overexpression in transfected cells with viral genomic RNA rather than natural infection. Deletion of Drosha removes all microRNAs as well as the nuclease activity, so secondary consequences of microRNA loss cannot be excluded as contributors, and the authors' own RNA sequencing confirms that Drosha shapes the transcriptome even without infection. The in vitro cleavage assay uses purified enzyme and naked RNA, which does not establish that the same cleavage occurs on replicating genomes within cells. The absence of a virus-derived small interfering RNA signature is a negative result at the sequencing depth and in the cell types used, and other laboratories reported such signatures in different systems. Finally, the interferon independence claim rests on three specific genetic lesions and does not exclude other sensors or signaling routes.
Audience summaries
25 words
Removing Drosha, an enzyme that normally makes microRNAs in the nucleus, lets RNA viruses grow better, and infection sends the enzyme into the cytoplasm.
75 words
Insects and plants fight viruses with small interfering RNAs made by Dicer, a defense thought lost in mammals. In mouse fibroblasts, deleting Dicer did not change virus growth, but deleting Drosha did. Infection with several unrelated RNA viruses, or double-stranded RNA alone, pushed Drosha out of the nucleus by an export route that does not need interferon signaling. Drosha cut viral RNA and reshaped host gene expression without producing the classic small interfering RNAs.
150 words
Antiviral RNA interference is well established outside chordates, and whether mammals retain a functional version remains contested. Using primary fibroblasts with conditional deletion of Drosha or Dicer, the authors found that Sindbis virus and vesicular stomatitis virus replicated to higher titers without Drosha, while Dicer loss had no significant effect. Positive sense, negative sense and segmented nuclear RNA viruses, and transfected double-stranded RNA, all triggered accumulation of Drosha in the cytoplasm. Export required CRM1 and proceeded without new protein synthesis and in cells lacking RIG-I, TBK1 or the type I interferon receptor. A Drosha mutant fixed in the cytoplasm by alanine substitution at serines 300 and 302 restricted Sindbis virus most strongly, tying localization to activity. Purified Drosha cleaved viral genomic RNA in vitro and Drosha shaped the infected host transcriptome, yet small RNA sequencing in fibroblasts and in Drosophila cells revealed no virus-derived small interfering RNA signature.
Discoveries supported by this paper
Documented publication relationships
- Transcription Factor Redundancy Ensures Induction of the Antiviral State — conceptual extension.
- Noncanonical cytoplasmic processing of viral microRNAs — predecessor.
- Evidence for a cytoplasmic microprocessor of pri-miRNAs — predecessor.
- Transcription Factor Redundancy Ensures Induction of the Antiviral State — conceptual extension.
- In Vivo Delivery of Cytoplasmic RNA Virus-derived miRNAs — methodological foundation.
- A Small-RNA Enhancer of Viral Polymerase Activity — methodological foundation.
- Is RNA Interference a Physiologically Relevant Innate Antiviral Immune Response in Mammals? — review or synthesis.
- RNase III nucleases from diverse kingdoms serve as antiviral effectors — predecessor.
Pathogens
Technologies
Bulk RNA sequencingPlaque assay and TCID50 titrationImmunofluorescence and confocal microscopySmall interfering RNA and short hairpin knockdownSmall RNA deep sequencingSmall RNA northern blotIn vitro reconstitution with purified componentsConditional and inducible gene knockoutSubcellular fractionation