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
Evidence for a cytoplasmic microprocessor of pri-miRNAs
lab-ledDiscovery
Primary microRNA transcripts generated in the cytoplasm by a recombinant Sindbis virus are cleaved without any nuclear involvement yet still require Drosha, which relocalises from nucleus to cytoplasm on infection while the endogenous microRNA profile of the cell remains largely unchanged.
- Senior authors
- Benjamin R. tenOever
- Correspondence
- Benjamin R. tenOever
Research areas & themes
Citation
Shapiro JS, Langlois RA, Pham AM, tenOever BR. Evidence for a cytoplasmic microprocessor of pri-miRNAs. RNA 2012, volume 18, issue 7, pages 1338-1346.
DOI 10.1261/rna.032268.112. PMID 22635403. PMCID PMC3383965.
One-sentence contribution
Primary microRNA transcripts generated in the cytoplasm by a recombinant Sindbis virus are cleaved without any nuclear involvement yet still require Drosha, which relocalises from nucleus to cytoplasm on infection while the endogenous microRNA profile of the cell remains largely unchanged.
Executive summary
Canonical microRNA biogenesis is compartmentalised. A primary transcript is cleaved in the nucleus by the microprocessor, made up of the RNase III enzyme Drosha and the double-stranded RNA binding protein DGCR8, and the resulting precursor is exported and cleaved again by Dicer in the cytoplasm. Cytoplasmic RNA viruses can be engineered to produce functional microRNAs, which raises the question of how a primary transcript that never enters the nucleus is processed. Using recombinant Sindbis viruses encoding miR-124, miR-122 or both in tandem, the authors show that mature microRNAs are produced from a range of hairpin sequences and transcript positions, load into Argonaute 2 and silence a reporter carrying target sites. Blocking cell division by serum starvation, which removes the possibility that nuclear envelope breakdown grants access to the nuclear microprocessor, did not change the conversion of cytoplasmic primary transcript into mature microRNA, and in situ hybridisation showed the viral transcript confined to the cytoplasm, unlike the nuclear signal from an influenza A virus engineered to express the same hairpin. Genetic tests across conditional knockout fibroblasts showed a complete requirement for Drosha, a requirement for Dicer only at the final maturation step, no requirement for TRBP2, PACT or AGO2, and a partial effect of DGCR8 loss confined to mature microRNA accumulation. Immunofluorescence showed endogenous Drosha redistributed to the cytoplasm after Sindbis virus infection. Small RNA sequencing showed the endogenous microRNA profile essentially unchanged despite this redistribution and despite virus-derived miR-124 reaching several percent of cellular microRNA content.
Scientific context
At the time of writing, canonical biogenesis was well described as a stepwise nuclear then cytoplasmic pathway, and a set of noncanonical routes had been catalogued, most of which bypass the microprocessor while retaining Dicer. Mirtrons arise by splicing and debranching, and miR-451 is processed by the catalytic activity of Argonaute 2 rather than by Dicer. A separate line of work from several laboratories, including earlier reports from this group and from Rouha and colleagues, had shown that cytoplasmic RNA viruses can be engineered to yield functional microRNAs, and initial characterisation had suggested those transcripts were processed independently of the canonical microprocessor while still requiring Dicer. The paper states that the mechanism and characteristics of cytoplasmic biogenesis had not been thoroughly elucidated, and that is the gap it sets out to close.
Central question
What cellular machinery cleaves a primary microRNA transcript that is generated and retained in the cytoplasm, and does the use of that machinery perturb normal microRNA biogenesis in the infected cell?
Experimental strategy
The system is a recombinant Sindbis virus, an alphavirus whose entire replication cycle is cytoplasmic, carrying a microRNA locus so that the primary transcript is produced where the nuclear microprocessor is not. Varying the payload across a second hairpin and a tandem arrangement tests whether processing depends on the particular sequence, structure or position within the transcript. A matched comparison with an influenza A virus engineered to produce the same hairpin provides a nuclear-replicating control for the in situ hybridisation, so that cytoplasmic confinement of the viral transcript is demonstrated rather than assumed. Serum starvation with CFSE dilution as a division readout removes mitotic nuclear envelope breakdown as a route of access to nuclear components. The genetic requirements are then addressed directly with a panel of fibroblast lines lacking individual biogenesis factors, using conditional deletion with adenoviral Cre for the essential factors Drosha and DGCR8 and confirming deletion by loss of the abundant endogenous miR-93. Function is read out with Argonaute 2 immunoprecipitation and a Gaussia luciferase reporter bearing perfect target sites. Finally, small RNA deep sequencing asks whether any of this disturbs the cell's own microRNA population.
Key findings
- A recombinant Sindbis virus expressing miR-122 produced abundant mature 22-nucleotide microRNA at 16 hours, comparable to the earlier miR-124 virus, and a tandem construct carrying both loci in a single transcript yielded both mature species (Figure 1A). Observation aside, the authors read this as evidence that cytoplasmic processing is not restricted to one hairpin sequence or structure.
- In the tandem virus, miR-124 accumulated to lower levels than in the single-hairpin virus, which the authors attribute to its position in the transcript rather than to any demonstrated positional mechanism. No intermediate cleavage products were detectable, which is interpreted as rapid processing or degradation of the cytoplasmic primary transcript.
- Argonaute 2 immunoprecipitation recovered both miR-122 and miR-124 regardless of transcript source, while a GFP control precipitate recovered neither, and the lower miR-124 yield from the tandem virus was mirrored in Argonaute association and in silencing of a luciferase reporter carrying four perfect miR-122 sites (Figure 1, B and C).
- Serum starvation produced a complete block in division as judged by sustained CFSE signal at 24 and 48 hours. Infection of arrested cells gave reduced viral replication but the ratio of cytoplasmic primary transcript to mature miR-124 was unchanged at roughly one to one and a quarter (Figure 2, A and B). The authors state this strongly suggests cell division is not required for cytoplasmic microRNA synthesis.
- RNA in situ hybridisation with a probe against the primary miR-124 transcript showed nuclear accumulation in cells infected with the influenza A virus construct and an absence of nuclear signal with cytoplasmic abundance in cells infected with the Sindbis construct (Figure 2D).
- Conditional deletion of Drosha abolished both precursor and mature virus-derived miR-124 (Figure 3). This is the central genetic result and it is a direct demonstration that a nominally nuclear enzyme is required for cleavage of a transcript that never reaches the nucleus.
- Loss of DGCR8 reduced mature miR-124 and led to accumulation of the precursor without preventing conversion of primary transcript to precursor (Figure 3). The authors explicitly present two competing readings, that DGCR8 is dispensable for cleavage but needed for its accuracy, or that the cytoplasmic activity is DGCR8-independent and the phenotype is an indirect consequence of losing endogenous microRNAs, and they state that future work is needed to distinguish them.
- Cells lacking Dicer accumulated both the cytoplasmic primary transcript and a precursor of about 60 nucleotides at levels and size comparable to wild-type cells, with only the mature species absent (Figure 3). This places Dicer at the second cleavage only and argues it is not involved in the first.
- Loss of TRBP2 or of PACT left all three RNA species unchanged, and loss of AGO2 left precursor and mature miR-124 unchanged (Figure 3). The authors note that the two double-stranded RNA binding proteins share partial homology and may therefore be redundant rather than simply dispensable.
- Immunofluorescence for endogenous Drosha showed nuclear staining in mock-treated fibroblasts and a pronounced redistribution to the cytoplasm after infection with either the parental or the microRNA-expressing Sindbis virus (Figure 4A). The relocalisation is therefore a response to infection and not to the presence of a cytoplasmic hairpin substrate.
- Small RNA deep sequencing of mock-treated and infected fibroblasts showed the annotated microRNA profile largely unchanged by infection and by expression of a cytoplasmic primary transcript, while virus-derived miR-124 reached 4.3 percent of total cellular microRNA at 24 hours, with estimated concentrations up to about 55,000 copies per cell (Figure 4, B and C). The authors suggest the stability of the endogenous profile reflects microRNA half-lives longer than 24 hours, which is an interpretation and not a measurement made here.
Mechanistic model
The data constrain the pathway but do not establish a complete mechanism, and the authors say as much for the steps that remain open. What is shown is that a cytoplasmic primary transcript is cleaved into a precursor in a Drosha-dependent and Dicer-independent step, that the precursor is then converted to a mature microRNA in a Dicer-dependent step, that the product loads into Argonaute 2 and silences targets, and that Drosha itself moves out of the nucleus on infection. The model the authors propose is that virus infection redistributes Drosha to the cytoplasm where it acts on highly structured cytoplasmic RNAs, giving the enzyme a second function alongside its canonical microprocessor role. What is not established is the composition of the cytoplasmic cleaving activity. The paper cannot say whether DGCR8 is a genuine partner in the cytoplasm, and it raises the alternative that an uncharacterised and possibly virus-inducible double-stranded RNA binding protein substitutes for it. Nor is the trigger or the mechanism of Drosha relocalisation determined, although the authors note prior work that Drosha nuclear localisation is phosphorylation-dependent. The speculation that Drosha relocalisation reflects a role as a virus restriction factor independent of microRNA biogenesis, and the evolutionary framing that follows it, are explicitly offered as speculation.
Conceptual or technical advance
The work converts an operational observation, that engineered cytoplasmic viruses yield functional microRNAs, into a defined set of genetic requirements, and in doing so reassigns Drosha from a strictly nuclear enzyme to one whose localisation is conditional on infection. It also revises the earlier reading of these transcripts as microprocessor-independent, since Drosha proves to be strictly required. Practically, the demonstration that a cytoplasmic RNA virus can deliver a microRNA to high copy number without disturbing the host microRNA profile matters for the use of such viruses as small RNA delivery vehicles, and the panel of biogenesis knockout fibroblasts combined with recombinant Sindbis viruses provides a general assay for dissecting noncanonical processing.
Relationship to the broader research program
This paper sits within a sustained line of work in the laboratory on engineering RNA viruses to express small RNAs and on what that engineering reveals about host RNA biology. Its own reference list points to earlier reports from the group on noncanonical cytoplasmic processing of viral microRNAs, on engineered RNA viral synthesis of microRNAs, on influenza A virus-generated small RNAs, on microRNA-mediated species-specific attenuation of influenza A virus, and on in vivo delivery of cytoplasmic virus-derived microRNAs. Setting this paper beside the later laboratory work on RNA interference and virus interactions would be category 3 synthesis and is not attempted from this paper alone.
Related publications
- Shapiro and colleagues, 2010, predecessor. Cited here as the report of noncanonical cytoplasmic processing of viral microRNAs that this study sets out to characterise mechanistically, and as the source of the recombinant Sindbis virus cloning strategy.
- Varble and colleagues, 2010, methodological foundation. Cited as the origin of the engineered RNA viral microRNA approach and specifically as the source of the influenza A virus expressing miR-124 used here as the nuclear-replicating comparator.
- Langlois and colleagues, 2012, companion. Cited as the accompanying demonstration that cytoplasmic virus-derived microRNAs can be delivered in vivo.
- Perez and colleagues, 2009 and 2010, predecessor. Cited as earlier work from the group on influenza A virus-generated small RNAs and on microRNA-mediated attenuation.
- Schmid and colleagues, 2010, methodological foundation. Cited for immunoblotting procedure and part of the same laboratory's output.
Limitations and boundaries
The primary transcripts studied are virus-derived and engineered, so the findings describe what happens to a structured, capped and polyadenylated hairpin-bearing transcript produced at high abundance by an alphavirus, not a natural cytoplasmic pri-microRNA. All experiments are in cultured cells, chiefly BHK cells and mouse embryonic fibroblasts, at multiplicities of infection from one to ten and over roughly 16 to 24 hours, so nothing is established for primary tissue, for animals or for longer timescales. Two viruses are used and one of them, influenza A virus, only as a localisation comparator. The DGCR8 result is not resolved and the authors present two incompatible explanations for it. The negative results for TRBP2, PACT and AGO2 come from single knockout lines and cannot exclude redundancy, a point the authors make for the two double-stranded RNA binding proteins. The conditional knockouts require six days of Cre expression before infection, during which endogenous microRNAs are depleted, so indirect effects on the small RNA machinery cannot be excluded for any of the deletion phenotypes. The stability of the endogenous microRNA profile is measured at a single 24-hour timepoint by sequencing and is consistent with, but does not demonstrate, the proposed explanation based on microRNA half-life. The mechanism and the functional consequence of Drosha relocalisation are not determined.
Audience summaries
25 words
Microbes that never enter the nucleus can still make microRNAs, because virus infection moves the nuclear enzyme Drosha into the cytoplasm, where it cleaves viral hairpin transcripts.
75 words
MicroRNAs are normally cut first in the nucleus by Drosha and then in the cytoplasm by Dicer. Using a Sindbis virus engineered to carry microRNA hairpins, the authors show that processing happens entirely in the cytoplasm, does not need cell division, and yet still depends absolutely on Drosha. Infection itself moves Drosha out of the nucleus. Despite this relocation, the cell's own microRNA population is essentially unchanged, and the viral microRNA silences targets normally.
150 words
Recombinant Sindbis viruses carrying miR-124, miR-122 or both were used to generate primary microRNA transcripts confined to the cytoplasm. Mature products formed from all constructs, loaded into Argonaute 2 and silenced a target reporter. Serum starvation that fully blocked division did not alter processing efficiency, and in situ hybridisation placed the viral transcript in the cytoplasm, in contrast to the nuclear signal from an influenza A virus expressing the same hairpin. Conditional knockout fibroblasts showed an absolute requirement for Drosha, a requirement for Dicer only at the precursor to mature step, no requirement for TRBP2, PACT or AGO2, and a DGCR8 effect limited to mature microRNA accumulation that the authors leave unresolved. Endogenous Drosha redistributed to the cytoplasm after infection with either parental or microRNA-expressing virus. Small RNA sequencing showed the host microRNA profile largely unchanged even as virus-derived miR-124 reached several percent of total cellular microRNA.
Discoveries supported by this paper
Documented publication relationships
- Noncanonical cytoplasmic processing of viral microRNAs — predecessor.
- Engineered RNA viral synthesis of microRNAs — methodological foundation.
- Transcription Factor Redundancy Ensures Induction of the Antiviral State — methodological foundation.
- RNA viruses and the host microRNA machinery — methodological foundation.
- The Mammalian Response to Virus Infection Is Independent of Small RNA Silencing — methodological foundation.
- Drosha as an interferon-independent antiviral factor — predecessor.
- In Vivo RNAi Screening Identifies MDA5 as a Significant Contributor to the Cellular Defense against Influenza A Virus — predecessor.
- RNase III nucleases from diverse kingdoms serve as antiviral effectors — predecessor.