SARS-CoV-Encoded Small RNAs Contribute to Infection-Associated Lung Pathology
SARS-CoV generates three discrete small viral RNAs during infection of mouse lung, and blocking the one derived from the nucleocapsid gene with a locked nucleic acid antagomir reduces pulmonary inflammation, tissue damage and proinflammatory cytokine expression without measurably lowering lung virus titres.
- Senior authors
- Luis Enjuanes; Isabel Sola
- Correspondence
- Luis Enjuanes
Research areas & themes
Citation
Morales L, Oliveros JC, Fernandez-Delgado R, tenOever BR, Enjuanes L, Sola I. SARS-CoV-Encoded Small RNAs Contribute to Infection-Associated Lung Pathology. Cell Host & Microbe. 2017. Volume 21, issue 3, pages 344-355.
DOI 10.1016/j.chom.2017.01.015. PMID 28216251. PMCID PMC5662013.
One-sentence contribution
SARS-CoV generates three discrete small viral RNAs during infection of mouse lung, and blocking the one derived from the nucleocapsid gene with a locked nucleic acid antagomir reduces pulmonary inflammation, tissue damage and proinflammatory cytokine expression without measurably lowering lung virus titres.
Executive summary
Severe acute respiratory syndrome coronavirus causes lethal disease characterised by an excessive inflammatory response and extensive lung damage, and the prevailing view is that much of this pathology is driven by host responses rather than by viral burden alone. Several RNA viruses had been shown to generate small noncoding RNAs from their genomes, but it was not known whether coronaviruses do so or whether such RNAs matter for disease. Deep sequencing of small RNAs from the lungs of mice infected with a mouse-adapted SARS-CoV identified three abundant species of 18 to 22 nucleotides, two from the region encoding the glutamate-rich domain of nsp3 and one from the nucleocapsid gene. Their abundance tracked viral replication rather than virulence, they were positive sense like the genome, and they were produced in cells lacking both Drosha and Dicer, so canonical microRNA processing is not required. All three repressed a luciferase reporter carrying the complementary sequence in its 3 prime untranslated region, though only the nucleocapsid-derived species did so when supplied by the virus during infection. Inhibiting that species in mice with an intranasally delivered locked nucleic acid reduced gross lung pathology, histological inflammation and the expression of CCL2, interleukin 6 and CXCL10, while leaving lung titres unchanged, which the authors present as a contribution of the small RNA to immunopathology that is separable from replication.
Scientific context
DNA viruses and nuclear-replicating RNA viruses were already known to encode microRNAs, and improving sequencing sensitivity had extended the catalogue of small viral RNAs to cytoplasmic RNA viruses including influenza virus, enterovirus 71, hepatitis A virus and several flaviviruses. Two functional roles had been documented for such species in other systems, regulation of the viral life cycle and contribution to pathogenesis. A theoretical objection had been raised in the field, that excising a small RNA from a cytoplasmic RNA virus genome would damage that genome, but engineered viruses producing functional microRNAs without a replication penalty had weakened that argument. Against this background no coronavirus small RNA had been described. Separately, SARS-CoV pathology was already understood to involve dysregulated proinflammatory cytokine expression, with the envelope protein established as a virulence factor acting through several routes including nuclear factor kappa B activation, inflammasome activation and p38 signalling, and with an envelope-deleted virus attenuated in vivo. The study sits at the junction of these two lines and asks whether a viral small RNA is another contributor to that inflammatory pathology.
Central question
Does SARS-CoV generate small noncoding RNAs during infection, and if so do they contribute to the inflammatory lung pathology that characterises severe disease.
Experimental strategy
Discovery was carried out in infected animals rather than in culture, on the reasoning that small noncoding RNA populations are cell specific and that lung tissue provides the context in which pathology arises. Two viruses were compared, a virulent mouse-adapted strain that reproduces severe human disease and an attenuated envelope-deleted derivative, at two time points, which allows small RNA abundance to be separated from virulence and attached instead to replication. A viral sequence yielding no reads was designated as a negative control species and carried through the subsequent assays, which controls for the specificity of the RT-qPCR detection. Cell line work then tested whether production depends on host species or cell type and whether it requires the canonical microRNA machinery, using 293T cells and a matched line lacking both RNase III enzymes made permissive by transient receptor expression. Whether the viral sequences could be processed as microRNA precursors in the absence of infection was tested by placing flanking genomic sequence into introns of reporter plasmids. Function as a silencing guide was assessed with luciferase reporters bearing perfect complementary targets in the 3 prime untranslated region, first with synthetic mimics in uninfected cells and then with the reporter delivered into infected cells so that the virus itself supplies the small RNA, with a cellular microRNA reporter included to confirm that infection does not disable the silencing machinery. Finally the species that performed in both settings was inhibited in mice with a chemically modified locked nucleic acid given intranasally before infection, with readouts spanning weight, viral genomic RNA, subgenomic messenger RNA, lung titre, blinded gross and histological pathology scoring, cytokine and interferon-stimulated gene expression, and immunohistochemistry for viral antigen distribution.
Key findings
- Small RNA sequencing from infected mouse lung at two and four days recovered viral reads at under 0.1 percent of the total, distributed broadly across the genome as expected for breakdown products but with distinct peaks. Three species accounted for roughly 18 percent of viral reads, two mapping within the nsp3 glutamate-rich domain coding region at positions 3,052 and 3,184 and one within the nucleocapsid gene at position 28,461, all positive sense and 18 to 22 nucleotides (Figure 1B, Table 1). Read counts were highest for the virulent virus at two days, matching the titre peak (Figure 1A).
- RT-qPCR confirmed the three species and the absence of the control sequence in lung, and detected them to differing degrees in serum, where the second nsp3 species predominated while the other two were near mock levels despite comparable lung abundance (Figures 1C and 1D). The authors offer differential stability or active transport to blood as alternative explanations and do not distinguish between them.
- Production was much higher in mouse DBT cells expressing the receptor than in human Calu-3 2B4 cells, in proportion to the titres reached in each, and scaled with genomic RNA and infectious titre over time (Figures 2A and 2B, Figure S1). An envelope ion channel mutant that grows like wild type produced comparable svRNA levels (Figure S2B). The authors conclude that biogenesis is not cell or species restricted but depends on the extent of replication.
- Genomic sequences flanking the small RNAs, placed in introns of reporter plasmids, were not processed into the mature species in uninfected cells (Figure S4), indicating a requirement for viral factors or for host factors induced during infection.
- In 293T cells lacking Drosha and Dicer, genomic RNA accumulated as in the parental line and titres were slightly lower, and svRNA production was not overtly impaired (Figures 3A and 3B). Canonical microRNA processing is therefore dispensable.
- Synthetic mimics of all three species repressed a matched luciferase reporter to 40 to 60 percent of control, comparable to a cellular microRNA mimic, and antisense inhibitors restored activity partially for the first nsp3 species and fully for the other two (Figure 4A). A cellular microRNA reporter was silenced to under 3 percent in both mock and infected cells, showing that infection leaves the silencing machinery functional at the time point tested (Figure 4B). When the reporter was introduced into infected cells so that the virus supplied the small RNA, only the nucleocapsid-derived species produced significant silencing, and less than its mimic did (Figure 4C).
- Antisense inhibition in cell culture reduced the nucleocapsid-derived species to about five percent of control and had no significant effect on nsp3 or nucleocapsid protein levels (Figure 2C, Figures S3B and S3C). Titres fell modestly, most clearly for that species with more than one log at seventy-two hours (Figure 2D). The authors note that the RT-qPCR readout of inhibition can give false positives through interference between the oligonucleotide and the assay and therefore treat the functional assays as the confirmation.
- Intranasal locked nucleic acid given a day before infection reduced the nucleocapsid-derived species to about ten percent of control at two and four days without adverse effects over ten days (Figure 5A). Weight loss was modestly and persistently reduced but not prevented (Figure 5B). Genomic RNA and nucleocapsid subgenomic messenger RNA fell roughly two-fold at two days while lung titres were not significantly changed (Figures 5C to 5E).
- Blinded scoring showed significantly reduced gross lung pathology and reduced histological inflammation, with less alveolar and bronchiolar infiltration and edema, in treated animals (Figures 6A to 6D). Expression of CCL2, interleukin 6 and CXCL10 fell significantly, while ISG15 and MX1 rose and interferon gamma was unchanged (Figure 6E). Immunohistochemistry showed the same tissue distribution of viral antigen in both groups (Figure S5).
- The authors state that the genomic sequences giving rise to these species are conserved in the human Urbani strain and in several SARS-related bat coronaviruses but not in MERS-CoV, with the nucleocapsid-derived sequence fully conserved among the SARS-related bat viruses. This is a sequence comparison rather than an experimental demonstration that human SARS-CoV produces the same species.
Mechanistic model
The study does not establish a mechanism. It shows that three small viral RNAs exist during infection, that they can act as silencing guides against perfectly complementary targets, and that removing one of them reduces inflammatory pathology in mice. It does not identify how they are generated, what host transcripts they act on, or by what route the nucleocapsid-derived species influences the inflammatory response.
On biogenesis, the data exclude canonical Drosha and Dicer processing and exclude processing of the flanking genomic sequence as a precursor in uninfected cells. The authors offer two candidate routes without testing either, an argonaute 2-dependent route of the kind described for miR-451, and cleavage by a viral enzyme, with the coronavirus endoribonuclease nsp15 named as a possibility. They also note the precedent of flavivirus subgenomic RNA generated by the cellular exoribonuclease XRN1 stalling at structured RNA.
On function, the model proposed is that the nucleocapsid-derived species contributes to lung pathology by silencing host messenger RNAs that normally restrain the inflammatory response, acting through RNA interference. The evidence offered is that the species silences a reporter carrying its complement and that inhibiting it lowers cytokines. No endogenous target is identified, and the authors state that it is not known whether these species engage targets through a seed sequence, which they give as the reason target prediction is difficult. They also raise RNA to protein interactions as an alternative mode of action. The rise in ISG15 and MX1 on inhibition is interpreted as those genes acting as immunomodulators limiting proinflammatory output rather than as a direct antiviral effect, an interpretation supported by analogy to published Chikungunya virus work rather than by experiment here.
The paper argues from the absence of effects on nsp3 and nucleocapsid protein levels, and the absence of any titre increase on inhibition, that producing these small RNAs does not meaningfully damage the genome and that they do not silence the negative-sense replication intermediate to which they are complementary. That is an inference from negative results.
Conceptual or technical advance
The work extends the catalogue of virus-derived small RNAs to coronaviruses and does so from infected tissue rather than cell culture, which is what allows a pathology endpoint to be attached to the finding. It separates two things that are usually confounded in virology, the amount of virus and the severity of the inflammatory response, by showing a reagent that lowers the second without lowering the first in the lung. That makes the small RNA a candidate virulence determinant of a kind that would be missed by any screen scored on replication. Practically, it puts forward an antagomir directed at a viral rather than a host sequence as an intervention aimed at immunopathology, and it provides the sequences, detection assays and inhibitors needed to test the idea in other coronaviruses.
Relationship to the broader research program
The tenOever contribution here is stated in the author contributions as reagents, conceptual advice and manuscript writing, with the project conceived and the experiments designed by the Enjuanes and Sola group at the National Center of Biotechnology in Madrid. The connection to the tenOever laboratory's own program runs through the small viral RNA concept and the tools used to test it. The paper repeatedly frames itself against the influenza small viral RNA described by the tenOever group in 2010, adopting the same term, the same argument that non-uniform genomic distribution distinguishes regulatory species from degradation products, and the same locked nucleic acid inhibition strategy, while noting the contrasts, that influenza small viral RNAs come from noncoding segment ends and act on the viral life cycle whereas these come from coding regions and act on host pathology. It also uses the intronic reporter assay for testing whether viral sequence can be processed as a microRNA precursor, a method from the same laboratory's engineered-virus work. Setting this paper beside the 2010 influenza study and the laboratory's later coronavirus work indicates that virus-derived small RNAs serve different purposes in different families, which is a category 3 synthesis available only from reading those papers together.
Related publications
- Perez and colleagues, 2010, conceptual and methodological foundation, from the tenOever laboratory. Introduced influenza small viral RNA, the term svRNA, the hotspot argument distinguishing regulatory species from breakdown products, and the locked nucleic acid inhibition approach, all of which this paper adopts and explicitly compares itself to.
- Varble and colleagues, 2010, methodological foundation, from the tenOever laboratory. Source of the intronic reporter constructs used to test whether the viral sequences can be processed as microRNA precursors in the absence of infection, and cited for the demonstration that engineered small RNAs can be excised from a cytoplasmic virus genome without self-silencing.
- Shapiro and colleagues, 2010, predecessor, from the tenOever laboratory. Cited among the demonstrations that cytoplasmic RNA viruses can produce functional small RNAs.
- Benitez and colleagues, 2015, methodological foundation. Source of the Drosha and Dicer deficient cells used to test biogenesis requirements.
- DeDiego and colleagues, 2007 and 2014, Nieto-Torres and colleagues, 2014, and Jimenez-Guardeño and colleagues, 2014, predecessor, from the Enjuanes laboratory. Establish the envelope protein as a SARS-CoV virulence factor driving inflammatory pathology, and supply the attenuated viruses used here.
- Roberts and colleagues, 2007, methodological foundation, from another laboratory. Source of the mouse-adapted MA15 virus that underlies the animal model.
- Parameswaran and colleagues, 2010, Weng and colleagues, 2014, Shi and colleagues, 2014, Bidet and colleagues, 2014, and Roby and colleagues, 2014, predecessor, from other laboratories. The prior reports of small RNAs from cytoplasmic RNA viruses against which the coronavirus findings are positioned.
Limitations and boundaries
No endogenous target of any of the three small RNAs is identified, so the proposed silencing of host inflammatory regulators is a hypothesis rather than a result, and the authors state that the absence of information about seed-based targeting makes prediction difficult. The mechanism of biogenesis is unknown and the candidates offered are untested. Silencing activity in the context of infection was seen only for the nucleocapsid-derived species, and the authors note that the two nsp3 species may be compartmentalised away from plasmid-derived targets or competitively blocked, which is an explanation rather than a demonstration. Inhibition of the second nsp3 species could not be confirmed by RT-qPCR at all. In vivo the effect on pathology is measured after a single intranasal dose given before infection, the reduction in weight loss is modest, and the treatment did not improve survival, which the authors attribute to an unoptimised dose and schedule. Several statistical claims rest on a p value threshold of 0.1 rather than 0.05, including the effects on genomic RNA and on cytokine expression, and the histopathology used three animals per condition. All in vivo work uses one inbred mouse strain and a mouse-adapted virus, and human relevance rests on sequence conservation rather than on demonstration that human SARS-CoV infection of human tissue produces these species at comparable levels, with the human lung cell line tested here producing them only at low levels alongside low titres. The dissociation between pathology and titre is established for lung titre at the days sampled and does not exclude effects on replication elsewhere or at other times. Finally, the observed rise in ISG15 and MX1 on inhibition is interpreted through analogy to another virus system rather than tested.
Audience summaries
25 words
SARS-CoV makes three short RNAs during infection. Blocking the one from its nucleocapsid gene eased lung inflammation and damage in mice without reducing virus levels.
75 words
Sequencing small RNAs from the lungs of infected mice showed that SARS-CoV produces three discrete short RNAs from its own genome, two from the nsp3 region and one from the nucleocapsid gene. Their production does not need the cell's usual microRNA processing enzymes. Each can silence a matched reporter. Giving mice a chemical inhibitor of the nucleocapsid-derived RNA before infection reduced lung inflammation, tissue damage and inflammatory cytokines while lung virus titres stayed the same.
150 words
Severe SARS-CoV disease is driven substantially by a dysregulated inflammatory response rather than by viral burden alone. Deep sequencing of small RNAs from the lungs of mice infected with mouse-adapted SARS-CoV identified three abundant species of 18 to 22 nucleotides, two from the glutamate-rich domain of nsp3 and one from the nucleocapsid gene. Their abundance scaled with replication rather than with virulence, and they were still produced in cells lacking both Drosha and Dicer, so canonical microRNA biogenesis is not required. Synthetic mimics of all three silenced reporters bearing their complements, but only the nucleocapsid-derived species did so when supplied by the virus during infection. A locked nucleic acid inhibitor given intranasally before infection cut that species to ten percent, significantly reduced gross and histological lung pathology, and lowered CCL2, interleukin 6 and CXCL10, while lung titres were unchanged. No host target was identified, so the mechanism remains open.
Documented publication relationships
- Engineered RNA viral synthesis of microRNAs — methodological foundation.
- Noncanonical cytoplasmic processing of viral microRNAs — predecessor.