lab-ledA negative-sense cytoplasmic RNA virus, vesicular stomatitis virus, can be engineered to produce mature Dicer-dependent miR-124 that loads into Argonaute 2, silences targets, reaches many tissues in mice, and persists after the vector itself is cleared.
Ryan A Langlois; Jillian S Shapiro; Alissa M Pham; Benjamin R tenOever
2012 · Molecular Therapy · primary research
- Senior authors
- Benjamin R tenOever
- Correspondence
- Benjamin R tenOever
Research areas & themes
Citation
Langlois RA, Shapiro JS, Pham AM, tenOever BR. In Vivo Delivery of Cytoplasmic RNA Virus-derived miRNAs. Molecular Therapy. 2012. Volume 20, issue 2, pages 367-375.
DOI 10.1038/mt.2011.244. PMID 22086233. PMCID PMC3277236.
One-sentence contribution
A negative-sense cytoplasmic RNA virus, vesicular stomatitis virus, can be engineered to produce mature Dicer-dependent miR-124 that loads into Argonaute 2, silences targets, reaches many tissues in mice, and persists after the vector itself is cleared.
Executive summary
MicroRNA biogenesis is understood to begin in the nucleus, where Drosha and DGCR8 cleave the primary transcript. Reports that positive-sense cytoplasmic RNA viruses could yield mature microRNAs, termed virtrons, raised the possibility of a nuclear-independent route, but that claim had been challenged as an artifact of rapidly dividing transformed cells whose transient nuclear envelope breakdown might give viral transcripts access to the microprocessor. This study addresses both a biological question and a delivery problem. Vesicular stomatitis virus, a cytoplasmic virus of negative polarity, was engineered to express the murine pri-miR-124 locus as an independent transcript inserted between the glycoprotein and polymerase genes, and was compared directly with Sindbis virus and influenza A virus carrying the same locus. The recombinant vesicular stomatitis virus produced mature miR-124 that was abolished in Dicer-deficient fibroblasts, accumulated to an estimated 25,000 to 35,000 copies per cell, associated with Argonaute 2, and repressed reporters carrying miR-124 target sites by roughly 60 to 90 percent without impairing an unrelated microRNA. In mice, the vector delivered miR-124 to lung, spleen, liver, kidney and heart, and in immunocompetent animals the microRNA persisted after viral leader RNA had disappeared, with a measurable reduction in induction of the miR-124 target Ptbp1. Cytoplasmic microRNA production therefore occurs in animals and supplies a transient delivery platform.
Scientific context
Canonical microRNA biogenesis requires nuclear cleavage of the primary transcript by Drosha and DGCR8, export by Exportin 5, and cytoplasmic cleavage by Dicer with TRBP, after which the duplex is loaded into an Argonaute protein and one strand is selected as the guide while the other, the star strand, is normally degraded. Bona fide virus-encoded microRNAs had been described only in large DNA viruses. Two reports, one from another group working with a flavivirus replicon and one from this laboratory working with Sindbis virus, had recovered mature microRNAs from positive-sense cytoplasmic RNA viruses, but the mechanism was undefined and an alternative explanation had been raised in the literature, that transformed cells dividing rapidly allow viral transcripts to reach the nuclear microprocessor. Separately, delivery remains the principal obstacle to therapeutic RNA interference. DNA virus vectors had been reported to cause toxicity through saturation of nuclear export and of Argonaute, and lentiviral delivery integrates. The paper therefore sets out to test whether cytoplasmic microRNA production is a genuine process by demonstrating it in animals and with a virus of the opposite polarity.
Central question
Can a cytoplasmic RNA virus of negative polarity be engineered to generate a functional microRNA, and does cytoplasmic microRNA production occur in animal tissue rather than only in transformed cells in culture.
Experimental strategy
The design turns on polarity, replication compartment and organism. Vesicular stomatitis virus replicates exclusively in the cytoplasm and is of negative polarity, so a microRNA recovered from it cannot be attributed to the positive-strand replication strategy of Sindbis virus, and its transcripts carry a 5' cap and a poly(A) tail like cellular messenger RNA, which makes nuclear import improbable. The murine pri-miR-124 locus was inserted as an independent transcription unit between the glycoprotein and polymerase genes with flanking polymerase recognition signals. A scrambled noncoding insert provided the matched control. Sindbis virus and influenza A virus carrying the same locus, both previously generated in the laboratory, allow the three replication strategies to be compared side by side in a single experiment, with influenza serving as the nuclear reference case. Dicer-deficient fibroblasts test the cytoplasmic half of the pathway. Small RNA deep sequencing quantifies abundance, strand composition and the frequency of the product relative to the whole microRNA pool. Argonaute 2 immunoprecipitation tests RISC loading, and reporters bearing perfect target sites for either the guide or the star strand separate intended from unintended silencing. Interferon alpha receptor knockout mice remove the antiviral response so that different vectors can be compared at matched replication levels and so that systemic spread reveals the accessible tissue range, while wild-type mice test whether the approach works in an immunocompetent host and whether the payload outlives the vector.
Key findings
- Insertion of the pri-miR-124 hairpin did not prevent rescue of recombinant vesicular stomatitis virus, and infection produced a small RNA matching plasmid-derived miR-124 in size, with no change to endogenous miR-93 and no induction of endogenous miR-124 by the scrambled control virus at comparable replication levels (Figures 1b and 1c).
- All three engineered viruses produced miR-124 in wild-type fibroblasts, and production of both viral miR-124 and endogenous miR-93 was abolished in Dicer-deficient fibroblasts while viral transcript levels were unchanged, establishing that biogenesis of the viral product requires Dicer and that its loss is not secondary to reduced replication (Figures 2a and 2b).
- Deep sequencing recovered approximately 80,000 to 400,000 reads mapping to the 583 nucleotide viral pri-miR-124, with mature miR-124 reaching as much as 2 percent of all profiled microRNAs and an estimated 25,000 to 35,000 copies per cell, against fewer than 15 copies per cell in mock-infected fibroblasts (Figure 3a).
- Star strand accumulation was pronounced and restricted to the cytoplasmic vectors, reaching as much as 40 percent of the reads mapping to pri-miR-124, and was not seen for endogenous miR-93 or miR-20a (Figures 3a and 3b). Plasmid-driven miR-124 also produced detectable star strand but at significantly lower levels than the cytoplasmic viruses, which the authors interpret as evidence that overexpression alone does not account for the effect and that star strand accumulation may be a characteristic of cytoplasmic processing.
- Argonaute 2 immunoprecipitation recovered virus-derived miR-124 regardless of which virus produced it, with no reduction in Argonaute-associated miR-93, and overexpression of miR-124 from any of the three vectors did not impair repression of a miR-142 reporter by cotransfected miR-142 (Figures 4a and 4b). The authors read this as indicating that the vectors are unlikely to add toxicity through competition for the silencing machinery.
- All three viruses repressed a luciferase reporter bearing perfect miR-124 target sites by roughly 60 to 90 percent, and also repressed a reporter carrying the Scp1 3' untranslated region with its endogenous response elements. The star strand reporter was also significantly repressed by the cytoplasmic vectors, though at substantially lower levels, which the authors present as a real off-target risk that could be reduced by altering duplex thermodynamics or the star strand sequence (Figure 4c and Supplementary Figure S3).
- In interferon alpha receptor knockout mice, intranasal infection with each miR-124 virus gave high lung levels of the microRNA within one to two days, and intravenous delivery of the vesicular stomatitis vector produced detectable miR-124 in lung, spleen, liver, kidney and heart by two days, tracking viral transcript levels (Figures 5a and 5b).
- In wild-type mice, viral leader RNA appeared by one day and fell sharply after four days, while miR-124 remained at high relative levels at five days, which the authors take as evidence that the delivered small RNA persists through clearance of the vector (Figure 6a).
- Infection with the control virus raised the miR-124 target transcript Ptbp1 roughly 30-fold over naive lung, whereas the miR-124 virus gave less than a fivefold increase, while induction of the unrelated control transcript Tnfa did not differ significantly between the two viruses (Figure 6b). Because whole lung includes uninfected cells and microRNA repression is modest, the authors interpret the reduced induction as strong evidence of functional silencing in vivo rather than as a measurement of knockdown magnitude in infected cells.
Mechanistic model
The study does not establish the mechanism of cytoplasmic microRNA biogenesis, and the discussion opens by saying that it remains unclear. What the data constrain is the following. The cytoplasmic step requires Dicer, since the product is lost in Dicer-deficient cells while viral transcription is unaffected. The product enters Argonaute 2 and silences targets, so it completes the functional pathway. Because the vesicular stomatitis virus pri-miR-124 transcript carries a cap and a poly(A) tail and so resembles a cellular messenger RNA, the authors argue that return of that transcript to the nucleus of primary cells is highly improbable, and they infer from this that either the cytoplasm contains components capable of the first processing step or hairpin formation recruits canonical machinery out of the nucleus. Neither alternative is tested here, and the enzyme performing the pri to pre conversion is not identified.
Two features are advanced as interpretive links rather than demonstrated mechanism. First, the shared properties of virtrons and mirtrons, namely star strand accumulation and independence from DGCR8 reported in the laboratory's earlier work, are read as suggestive of noncanonical biogenesis, with the alternative explanation of RISC saturation from overexpression argued against by the lower star strand levels seen with plasmid expression. Second, the persistence of miR-124 in lung after loss of viral leader RNA is attributed to the long half-life of microRNAs, which is cited rather than measured here.
Conceptual or technical advance
Two things become testable. The demonstration in animals, and in a negative-sense cytoplasmic virus, argues against the proposal that cytoplasmic microRNA production is an artifact of rapidly dividing transformed cells, and so makes the search for a nuclear-independent processing activity, and for endogenous small RNAs made by that route, a defined question rather than a disputed observation.
Practically, the work delivers a vector class with a distinct profile. A cytoplasmic RNA virus does not integrate, does not depend on nuclear export for the primary processing step, produces a high per cell copy number, distributes to many tissues from an intravenous route, and leaves a payload that outlasts the infection. The discussion sets out the resulting design space, including replacement of the natural hairpin with artificial microRNAs engineered to release perfectly complementary small interfering RNAs, combination with microRNA target site technology to restrict tropism, use in oncolytic settings, and, notably, the proposal that a virus could deliver a library of artificial microRNAs only to infected cells so that selection identifies host restriction factors. All of these are proposed applications, not results of this paper. The vectors are framed as suited to conditions with acute phenotypes lasting less than about a week.
Relationship to the broader research program
This paper sits between the demonstration that a nuclear RNA virus can be made to produce a microRNA and the later use of engineered viruses as screening instruments. It extends the platform from influenza to two cytoplasmic viruses, moves it from cell culture into mice, and adds quantification by deep sequencing, Argonaute association and in vivo target measurement to the characterization toolkit.
Category 3 synthesis. Read with Varble and colleagues in 2010, which established the influenza vector, and Varble and colleagues in 2013, which delivers artificial small RNA libraries from a virus and selects on them in animals, the discussion here contains the explicit statement of the screening idea that the 2013 paper implements. That trajectory across three papers is visible when they are placed side by side and is not a claim made by any one of them. The interest in whether RNA viruses interact productively with the host small RNA machinery, and in what that implies about antiviral RNA interference in mammals, recurs across the corpus and is treated conceptually in the 2016 perspective.
- Varble and colleagues, 2010, Engineered RNA viral synthesis of microRNAs, cited as reference 14 and supplying the influenza vector used here. Predecessor and methodological foundation.
- Shapiro and colleagues, 2010, Noncanonical cytoplasmic processing of viral microRNAs, cited as reference 10, from the same laboratory and sharing an author, supplying the Sindbis vector and the prior evidence of DGCR8 independence. Predecessor.
- Perez and colleagues, 2009, MicroRNA-mediated species-specific attenuation of influenza A virus, cited as reference 23. Methodological foundation for microRNA target site engineering.
- Varble and tenOever, 2011, Implications of RNA virus-produced miRNAs, cited as reference 43. Review or synthesis from the same laboratory.
- Stojdl and colleagues, 2003, on vesicular stomatitis virus strains defective in innate immune shutdown, cited as reference 49 and the source of the rescue procedure, with tenOever as an author. Methodological foundation.
- Varble and colleagues, 2013, An In Vivo RNAi Screening Approach to Identify Host Determinants of Virus Replication. Conceptual extension, implementing the library screening idea raised in this discussion.
- tenOever, 2016, The Evolution of Antiviral Defense Systems. Review or synthesis addressing the relationship between RNA viruses and small RNA pathways.
Limitations and boundaries
The mechanism of cytoplasmic pri-microRNA processing is explicitly unresolved, and the argument that the transcript does not reach the nucleus rests on its resemblance to a cellular messenger RNA rather than on a localization experiment. Only one microRNA locus, murine miR-124-2, was tested, in one insertion position per vector, so generality across hairpins and insertion sites is untested, as is the behavior of the artificial microRNAs the discussion proposes for therapeutic use.
Much of the in vivo work uses interferon alpha receptor knockout mice, chosen deliberately so that vectors could be compared without the confound of differential replication, which means that the tissue distribution seen after intravenous delivery is obtained in the absence of type I interferon signaling and does not establish the range accessible in an immunocompetent animal. In wild-type mice only vesicular stomatitis virus and only the lung were examined for function. The functional in vivo readout is a reduction in virus-induced Ptbp1 induction measured in whole lung, which contains uninfected cells, so it demonstrates activity without quantifying silencing in infected cells, and it is a transcript measurement rather than a protein measurement. Persistence of miR-124 beyond clearance is inferred from loss of viral leader RNA rather than from a direct measure of infectious virus. The star strand is loaded and does repress its target, so off-target silencing is a demonstrated property of these vectors rather than a hypothetical one, and the proposed fixes are untested here. No toxicity, pathogenesis or dose ranging study is reported, and the therapeutic framing throughout the discussion is extrapolation.
Audience summaries
25 words
A virus that never enters the nucleus was engineered to make a working microRNA, delivering it to many mouse tissues and leaving it behind after clearance.
75 words
MicroRNA production was thought to begin in the nucleus. Vesicular stomatitis virus, which replicates only in the cytoplasm, was engineered to carry a microRNA precursor. It produced abundant mature miR-124 that required Dicer, loaded into Argonaute 2, and silenced targets. In mice the vector delivered the microRNA to lung, spleen, liver, kidney and heart, and the microRNA remained after the virus was cleared, reducing induction of a known target gene.
150 words
Whether microRNAs can be generated outside the nucleus was contested, since the two reported examples came from positive-sense cytoplasmic viruses in transformed cells. Vesicular stomatitis virus, a cytoplasmic virus of negative polarity, was engineered to express the murine pri-miR-124 locus as an independent transcript and compared with Sindbis and influenza vectors carrying the same locus. Mature miR-124 was produced by all three, was lost in Dicer-deficient cells without a change in viral transcription, reached an estimated 25,000 to 35,000 copies per cell, associated with Argonaute 2, and repressed target reporters by roughly 60 to 90 percent without impairing an unrelated microRNA. The cytoplasmic vectors also produced substantial star strand, which repressed its own reporter and represents a genuine off-target liability. In mice the vector reached several organs and the microRNA outlasted the infection, reducing induction of Ptbp1. The processing enzyme responsible for the cytoplasmic step is not identified.
Discoveries supported by this paper
Discoverylab-led for Perez 2009, Varble 2010, Langlois 2012 in PNAS, Langlois 2012 in Molecular Therapy, Pham 2012, Varble 2013, Schmid 2014, Benitez 2015 on in vivo screening, Møller 2018 and the tenOever 2019 review, and co-led for Langlois 2013, whose ferret transmission work sits with the Perez and García-Sastre groups. Daniloski 2021 in Cell is co-led with the Sanjana laboratory
Documented publication relationships
Pathogens
Technologies