co-ledEngineering target sites for miR-192, a microRNA abundant in human and mouse respiratory tissue but absent from the ferret respiratory tract, into the influenza A virus hemagglutinin segment attenuates the virus in mice while leaving replication and transmission in ferrets intact, providing a genetic layer of biocontainment for transmission studies.
Ryan A Langlois; Randy A Albrecht; Brian Kimble; Troy Sutton; Jillian S Shapiro; Courtney Finch; Matthew Angel; Mark A Chua; Ana Silvia Gonzalez-Reiche; Kemin Xu; Daniel Perez; Adolfo García-Sastre; Benjamin R tenOever
2013 · Nature Biotechnology · methods/resource
- Senior authors
- Benjamin R tenOever
- Correspondence
- Daniel Perez; Adolfo García-Sastre; Benjamin R tenOever
Research areas & themes
Citation
Langlois RA, Albrecht RA, Kimble B, Sutton T, Shapiro JS, Finch C, Angel M, Chua MA, Gonzalez-Reiche AS, Xu K, Perez D, García-Sastre A, tenOever BR. MicroRNA-based strategy to mitigate the risk of gain-of-function influenza studies. Nature Biotechnology. 2013. Volume 31, issue 9, pages 844-847. DOI 10.1038/nbt.2666. PMID 23934176. PMCID PMC3808852.
One-sentence contribution
Engineering target sites for miR-192, a microRNA abundant in human and mouse respiratory tissue but absent from the ferret respiratory tract, into the influenza A virus hemagglutinin segment attenuates the virus in mice while leaving replication and transmission in ferrets intact, providing a genetic layer of biocontainment for transmission studies.
Executive summary
Demonstrations that a small number of amino acid changes in hemagglutinin can make H5N1 influenza A virus transmissible between ferrets raised the question of how experiments of that kind can be conducted with an additional safeguard beyond physical containment. The authors asked whether the natural divergence in microRNA expression between the experimental host and humans can be used to build a virus that behaves normally in the model species but is crippled in human cells.
Small RNA deep sequencing of human A549 lung cells, primary ferret lung and MDCK cells identified candidate microRNAs abundant in human cells and scarce in the two carnivore-derived sources. Northern blotting narrowed the candidates and miR-192 was selected, since it was also detectable in mouse lung and in human bronchial, alveolar and nasal epithelium. Four fully complementary miR-192 target sites were placed in a duplicated packaging region downstream of the hemagglutinin stop codon, leaving the hemagglutinin coding sequence untouched.
The targeted virus was silenced in cells expressing miR-192 and replicated normally in cells that do not. In mice it caused no morbidity or mortality even at ten times the lethal dose of the parental virus. In ferrets an H3N2 version carrying the same modification infected, replicated and transmitted by direct and by respiratory contact indistinguishably from controls, and sequencing recovered no escape variants from either animal system.
Scientific context
Reports that airborne transmission of H5N1 between ferrets can be conferred by a small number of hemagglutinin substitutions prompted a broad debate about whether such gain-of-function experiments should be performed at all, including a voluntary moratorium. Physical containment at enhanced biosafety level 3 was the existing safeguard. Separately, several groups had shown that microRNA target sites inserted into a viral genome can restrict a virus to tissues lacking the relevant microRNA, work cited here from the tenOever laboratory on hematopoietic-specific targeting of influenza A virus and on species-specific attenuation, and from other laboratories on flaviviruses and on attenuated virus vaccine design. What had not been done was to exploit the difference in microRNA expression between the experimental animal and the human, so that the same virus is permissive in the model and restricted in the species of concern. The paper frames this as adding a molecular layer to existing physical precautions rather than replacing them.
Central question
Can a microRNA expressed in human and mouse respiratory tissue but absent from the ferret respiratory tract be used to restrict influenza A virus in human cells without compromising the replication and transmission behavior that makes the ferret a useful transmission model?
Experimental strategy
The strategy has a discovery step and a validation step, and the validation is deliberately two-sided. Discovery was unbiased profiling of the small RNA pool in human lung epithelium, primary ferret lung and MDCK cells, followed by northern blot confirmation, because deep sequencing read counts across species can be misleading when reference annotation differs. The two non-human sources are both from the order Carnivora, which the authors note, and MDCK cells matter because influenza A virus stocks are grown in them, so a candidate microRNA has to be absent there for the virus to be propagable.
The engineering step placed target sites in the hemagglutinin segment rather than in nucleoprotein or NS1, which earlier studies had used, specifically so that the segment carrying the transmission-determining protein cannot be separated from the safety element by reassortment. Duplicating the 5 prime packaging sequence creates an untranslated region after the stop codon in which target sites can sit without altering the hemagglutinin protein or the packaging signal.
Validation used a low-pathogenicity H5 construct with the polybasic cleavage site removed for the cell culture and mouse work, which permits lethality studies under appropriate containment, and the seasonal H3N2 Wyoming strain for ferret transmission, since aerosol transmission of that strain in ferrets is established. Escape was assessed by plaque purification and sequencing from both animal systems.
Key findings
-
Small RNA sequencing of A549 human lung cells, primary ferret lung and MDCK cells identified miR-138, miR-193b and miR-192 as candidates abundant in human cells and low or absent in the other two (Figure 1). Northern blotting contradicted the sequencing for miR-193b, which was substantially expressed in ferret lung, leaving miR-138 and miR-192. The discrepancy is reported directly and the authors treat the blot as the arbiter.
-
miR-192 but not miR-138 was detectable in murine lung at levels the authors judged sufficient for silencing, and miR-192 was robustly expressed in human bronchial and alveolar lines and in primary human nasal epithelial cells, with nasal cells carrying approximately twofold more than alveolar cells (Supplementary Figure 1).
-
In a single cycle of infection, hemagglutinin and nucleoprotein expression from the wild-type, scrambled control and 192t viruses were comparable in MDCK cells, while hemagglutinin from the 192t virus was ablated in MDCK cells engineered to express miR-192 and in A549 cells (Figure 2b). The authors read this as sequence-specific post-transcriptional silencing of the hemagglutinin segment.
-
Over multiple cycles, wild-type and control viruses grew equivalently regardless of miR-192, while 192t replication was blunted in MDCK cells expressing miR-192 and in A549 cells (Figure 2c).
-
Mice infected intranasally with 100 plaque forming units of the 192t H5 HAlo virus showed no weight loss and no mortality, while wild-type and control infections were uniformly lethal by day eight. A tenfold higher dose of 192t still produced no morbidity or mortality (Figure 3a and 3b). Pulmonary titers and viral hemagglutinin and nucleoprotein protein were reduced at days three and five (Figure 3c and 3d).
-
All 17 plaques recovered from mouse lung at day five retained perfect miR-192 target sites, and virus sequenced from ferrets at day seven after aerosol transmission showed no excision events. The observation is an absence of detected escape at these sampling depths. The authors' statement that these results demonstrate the safety of the platform goes beyond what the sampling establishes and should be read as their interpretation.
-
An H3N2 Wyoming virus engineered the same way showed roughly 100-fold reduced growth in A549 cells relative to control (Supplementary Figure 4).
-
In ferrets, wild-type, control and 192t H3N2 viruses all produced high nasal wash titers from day one that declined over ten days in directly inoculated animals, and all three transmitted to naive animals by direct contact and by respiratory contact (Figure 4). This is consistent with the absence of miR-192 in ferret lung and is the central requirement for the approach to be usable.
Mechanistic model
The silencing mechanism itself is not dissected here and the paper does not set out to establish one. The working model, supported by the cell culture data and consistent with the earlier work the authors cite, is that endogenous miR-192 loaded into the RNA-induced silencing complex recognizes the fully complementary sites in the hemagglutinin messenger RNA and directs its post-transcriptional destruction, so that hemagglutinin protein fails to accumulate and multicycle replication collapses in cells that express the microRNA. Placement downstream of the stop codon and upstream of the duplicated packaging signal is intended to leave protein sequence and segment packaging unaffected, and the equivalence of the engineered and wild-type viruses in miR-192-negative cells is the evidence offered that this intention was met.
What the data constrain is the phenotype in each host. What they do not address is whether silencing is by cleavage or by translational repression, whether the negative-sense genomic RNA is engaged at all, and what fraction of the attenuation in mice is cell-intrinsic silencing as opposed to altered kinetics of immune engagement.
Conceptual or technical advance
Biocontainment becomes a property that can be written into the viral genome rather than only into the building. The specific technical contributions are the choice of a microRNA defined by a cross-species expression difference between the experimental animal and the species of concern, and the placement of target sites in the hemagglutinin segment through duplication of the packaging signal, which couples the safety element to the segment that carries the transmission phenotype under study. The authors propose that the approach generalizes to other pathogens for which gain-of-function work is contemplated, naming Ebola virus, SARS coronavirus and henipaviruses, and that targeting multiple segments or multiple microRNAs would further reduce escape. Both of these are stated as prospects and are not tested here.
Relationship to the broader research program
This is a direct extension of a line of work in the tenOever laboratory on inserting microRNA response elements into influenza A virus to control where the virus can replicate. Perez and colleagues 2009 established species-specific attenuation using this approach, and Langlois and colleagues 2012 used hematopoietic-specific targeting to ask what cell types must be infected for antiviral immunity to be induced. Pham, Langlois and tenOever 2012 applied the same logic to dengue virus dissemination. The technique is used in the corpus both as a safety measure and as a tool for cell-type-restricted infection, and this paper is the biosafety application of it. Category 3 synthesis, visible across those papers together, is that the laboratory treated engineered microRNA response elements as a general control layer over viral tropism rather than as a single-purpose attenuation trick.
- Perez and colleagues 2009, Nature Biotechnology, MicroRNA-mediated species-specific attenuation of influenza A virus. Predecessor. Established species-specific microRNA targeting of influenza A virus, cited here as the direct precedent.
- Langlois and colleagues 2012, PNAS, Hematopoietic-specific targeting of influenza A virus. Predecessor and methodological foundation. Source of the deep sequencing, northern blot and microRNA-expressing MDCK methods used here, and the prior demonstration of cell-type-restricted targeting.
- Chua and colleagues 2013, Cell Reports, on influenza A virus suboptimal splicing. Companion. Cited as the source of the alternative NS1 target site placement strategy that this paper chose not to use.
- Pham, Langlois and tenOever 2012, PLoS Pathogens, on dengue virus dissemination. Application. Extends the same targeting logic to a different virus family.
Limitations and boundaries
The approach depends entirely on an expression difference that holds in the tissues that matter, and the evidence for absence of miR-192 in ferret is from lung and from a small set of samples, so restriction of the virus in other ferret tissues or at other ages is not addressed. Silencing was demonstrated with four fully complementary sites, which is the most favorable configuration, and no titration of site number or of mismatch tolerance is presented. Escape was assessed from 17 plaques at a single time point in mice and from sequencing after transmission in ferrets, which bounds the escape frequency only loosely, and the ferret arm did not use a miR-192-expressing species in which selective pressure for escape would exist. The mouse and cell culture work used an H5 construct with the polybasic cleavage site removed, so behavior of a highly pathogenic H5N1 virus carrying the same modification is not tested, and the ferret transmission work used a seasonal H3N2 strain rather than an H5N1 strain of the kind that motivated the study. The human evidence is entirely from cell lines and primary cells, since no human challenge is possible, so restriction in an infected person is an inference from those systems. Mice and ferrets were not randomized and investigators were not blinded, which the methods state. Transmission experiments were performed at biosafety level 2 plus with the seasonal strain, so the containment regime under which a gain-of-function experiment would actually run was not itself evaluated.
Audience summaries
25 words
A small RNA present in human and mouse airways but not in ferret airways was used to cripple influenza A virus in humans while preserving ferret transmission experiments.
75 words
Influenza transmission experiments rely on ferrets, which raises concern about accidental human infection. Sequencing of small RNAs across species identified miR-192 as abundant in human and mouse respiratory tissue and absent from ferret lung. Four miR-192 target sites inserted after the hemagglutinin stop codon silenced the virus in human cells and rendered it harmless in mice at ten times a lethal dose, while leaving replication and airborne transmission in ferrets unchanged.
150 words
Demonstrations that few hemagglutinin substitutions can confer ferret-to-ferret airborne transmission of H5N1 prompted debate about the biosafety of such experiments. Small RNA deep sequencing of human A549 cells, primary ferret lung and MDCK cells, followed by northern blot confirmation, identified miR-192 as expressed in human and murine respiratory tissue but not in the two carnivore-derived sources. Four perfectly complementary miR-192 sites were inserted into a duplicated packaging region downstream of the hemagglutinin stop codon, leaving the protein and packaging signal intact and tying the safety element to the segment that carries transmission determinants. The engineered virus was silenced in miR-192-expressing cells, replicated normally where the microRNA is absent, and caused no disease in mice even at tenfold the lethal dose. An H3N2 version infected, replicated and transmitted in ferrets by direct and respiratory contact like unmodified virus, and no escape variants were recovered from either animal system.
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