Is there evidence that vertebrate somatic cells use small RNA silencing against viruses?
Plants, nematodes and arthropods process viral double-stranded RNA into small interfering RNAs that guide cleavage of viral transcripts, and viruses of those hosts encode suppressors of that pathway. Vertebrates detect the same double-stranded RNA but convert detection into a transcriptional program built on pattern recognition receptors, type I and type III interferons, and interferon-stimulated genes. Whether vertebrates also retain a functioning small RNA arm was genuinely open, and the question was reopened in 2013 by two reports in Science, from the Voinnet group and from the Ding group, arguing for antiviral RNA interference in mammalian cells. Deep sequencing had made the dispute harder rather than easier, because virus-derived small RNAs are detectable in infected mammalian cells but their size distribution, Dicer dependence and function were unresolved.
Three contributions, of different kinds and with different weight.
The first is an argument from quantity. tenOever 2013 in Nature Reviews Microbiology, a single-author review, proposes three constraints on any antiviral role for chordate microRNAs. Copy number, with a conservative estimate of roughly 100 copies per cell needed for activity on a host transcript and an inferred order of magnitude more for the far more numerous viral transcripts. Silencing capacity, since partial complementarity gives less than twofold repression and forecloses the catalytic cleavage that virus-derived interfering RNAs achieve. Kinetics, since most viral life cycles run under twelve hours while average protein half-life under stress exceeds ten hours. The review presents these as argument rather than as data, and it uses them to set aside a body of reports that measure microRNA induction as fold change from a negligible baseline. The thresholds are reasoning, and the review says so.
The second is a set of evidentiary criteria, and it was not led by this laboratory. Cullen 2013 in Cell Host and Microbe is a Minireview led by Bryan Cullen with Sara Cherry and tenOever, classified collaborative, and its arguments belong to those three authors jointly with Cullen as first and corresponding author. It accepts the mouse embryonic stem cell evidence, where encephalomyocarditis virus yields Dicer-dependent small RNAs of the expected size and phasing and a nodamura virus lacking its B2 protein is partly rescued in cells lacking all four Argonautes. It finds the somatic cell and in vivo case unproven, on a specific ground. The inference there depends on treating nodavirus B2 and Ebola virus VP35 as selective inhibitors of small interfering RNA production, when both bind double-stranded RNA and also antagonise interferon, and the B2 point mutation used abolishes double-stranded RNA binding as well. The piece names the experiment that would settle it, genetic rescue of a suppressor-deficient virus in cells whose silencing machinery is genetically compromised together with direct evidence that the small RNAs silence viral messenger RNA, and states it had not been done.
The third is experimental and is the laboratory's own answer. Backes 2014 reframed the question as a fitness experiment. Rather than removing silencing components from the host, which changes the cell before infection begins, the authors armed vesicular stomatitis virus with vaccinia VP55, the poly(A) polymerase subunit that tails and destroys RISC-loaded small RNAs, and compared it with a virus carrying influenza NS1 as a positive control for what disabling a real antiviral system looks like. The VP55 virus destroyed host microRNAs efficiently in fibroblasts, in primary bone marrow derived macrophages and in mouse lung, and gained no replication advantage in any of them. In wild-type mice it was attenuated by about a log, which messenger RNA sequencing attributes to derepression of interferon-stimulated transcripts normally held down by microRNAs. In mice lacking both type I and type III interferon receptors, the armed virus, the NS1 virus and the control reached comparable titres, which removes the redundancy objection that a silencing contribution might be hidden beneath interferon. The authors state plainly that absence of a biological activity is difficult to prove and that they regard the result as a strong argument rather than a demonstration.
The reagent came before the question was posed in this form. Backes 2012, co-led with Sara Cherry, identified VP55 as a poxvirus enzyme that adds short nontemplated adenosine tracts to microRNA guide strands after strand selection and thereby marks them for host-supplied decay, with a 3 prime terminal 2 prime O-methyl group conferring complete protection. That paper reads the existence of such an enzyme as evidence that host microRNAs impose a cost on the virus, and reconstituting the dominant fibroblast microRNAs during infection reduced virus yield by roughly half. Backes 2014 then used the same enzyme to argue the opposite conclusion about antiviral silencing, and the two readings sit together only because the 2014 study locates the cost in derepressed interferon-stimulated genes rather than in direct silencing of the virus.
The position also shifted in strength. tenOever 2013 argues from thresholds that chordate microRNAs cannot be antiviral. Cullen 2013 declines to settle the somatic case either way and sets criteria instead. Backes 2014 supplies a direct test with a negative result. The sequence is a claim, a demand for evidence, and then evidence, and it is worth noting that the intermediate step was a collaborative piece led by another laboratory that was more cautious than this laboratory's own review of the same year.
Backes 2012 is also assigned to the theme on the functional limits of microRNAs, where the tailing enzyme becomes a tool rather than a finding about poxvirus biology.
The negative answer here is the premise of the reconstruction theme, where Benitez 2015 asks whether the system that mammals do not use could have worked. It is also the premise of the evolutionary theme, since a system that is absent invites an explanation for its absence. Shapiro 2014 and Aguado 2017 in the RNase III theme complicate the negative answer in a specific way, by showing that a component of the silencing machinery restricts viruses without producing small interfering RNAs at all.
Engineering vesicular stomatitis virus to eliminate RISC-loaded small RNAs attenuates rather than enhances replication in mice, and confers no replication advantage even when interferon signaling is removed, arguing that small RNA silencing does not contribute to mammalian antiviral defense.
A critical appraisal of the evidence for antiviral RNA interference in mammals, accepting that mouse embryonic stem cells generate virus-derived small interfering RNAs while finding the somatic cell case unproven, because the viral proteins invoked also antagonize interferon.
Argues that chordates, unlike plants, nematodes and arthropods, do not use small RNAs as antiviral defence, and that this absence of interplay between host microRNAs and RNA viruses is what leaves the microRNA machinery available for engineering viral tropism and small RNA delivery.
Poxviruses degrade host microRNAs through the catalytic subunit of their own poly(A) polymerase, VP55, which adds short nontemplated adenosine tails to argonaute-loaded guide strands and thereby marks them for cellular decay, while small RNAs carrying a 3 prime terminal 2 prime O-methyl group are spared.