Hematopoietic-specific targeting of influenza A virus reveals replication requirements for induction of antiviral immune responses

lab-led

Influenza A virus engineered to be silenced by the hematopoietic microRNA miR-142 shows that replication inside antigen presenting cells is dispensable for CD8 T cell priming and clearance but required for much of the RIG-I-dependent type I interferon response in vivo.

Ryan A. Langlois; Andrew Varble; Mark A. Chua; Adolfo García-Sastre; Benjamin R. tenOever

2012 · Proceedings of the National Academy of Sciences · primary research

Senior authors

Benjamin R. tenOever

Correspondence

Benjamin R. tenOever

Research areas & themes

Citation

Langlois RA, Varble A, Chua MA, García-Sastre A, tenOever BR. Hematopoietic-specific targeting of influenza A virus reveals replication requirements for induction of antiviral immune responses. Proceedings of the National Academy of Sciences. 2012. 109(30), 12117-12122.

DOI 10.1073/pnas.1206039109. PMID 22778433. PMCID PMC3409765.

One-sentence contribution

Influenza A virus engineered to be silenced by the hematopoietic microRNA miR-142 shows that replication inside antigen presenting cells is dispensable for CD8 T cell priming and clearance but required for much of the RIG-I-dependent type I interferon response in vivo.

Executive summary

Influenza A virus replicates chiefly in respiratory epithelium, yet it also enters dendritic cells and macrophages, and the functional consequence of that second compartment has been hard to isolate. Loss-of-function approaches act on the host and therefore remove the cell type along with its other roles. This study instead alters the pathogen. Four perfect target sites for miR-142, a microRNA expressed in cells of hematopoietic origin, were placed into a duplicated packaging region of the influenza nucleoprotein segment, producing a replication-competent virus (NP142t) whose transcription is destroyed specifically in immune cells while remaining intact in epithelial cells and fibroblasts. Silencing was verified at the level of small viral RNA, viral cRNA and mRNA, nucleoprotein protein, multicycle growth, surface hemagglutinin on lung CD45+CD11c+ cells, and infectious virus recovered from the lung draining lymph node. Mice infected with the targeted virus lost weight, cleared virus, and generated nucleoprotein- and polymerase acidic-specific CD8 T cells indistinguishably from mice given the control virus. What changed was innate signaling. Interferon beta and IRF-7 induction fell in cultured macrophages and in whole lung, and interferon beta induction required RIG-I in both macrophages and fibroblasts. The authors read this as evidence that replication intermediates formed in hematopoietic cells are the dominant in vivo trigger of type I interferon, while cross-presentation suffices for cytotoxic T lymphocyte responses.

Scientific context

Antigen presenting cells can acquire influenza antigen by direct infection, by phagocytosis of dying infected epithelial cells, or by membrane exchange that transfers peptide-loaded MHC class I. All three routes had been reported to support presentation by CD103+ and CD8α+ dendritic cell subsets, so which route dominates during a natural infection remained contested. In parallel, RIG-I had been established as the sensor driving interferon induction in infected epithelial cells, but whether Toll-like receptors carry the larger share of interferon induction from hematopoietic cells was unresolved. The paper frames both questions as consequences of one limitation, namely that the usual experimental tools delete host genes or host cells and so cannot ask what a single infected compartment contributes while leaving the rest of the animal intact. Prior work cited by the authors had shown that inserting complementary microRNA target sites into a viral genome converts the host microRNA into an effective silencing agent, including miR-93 targeting of influenza nucleoprotein and miR-142 restriction of dengue virus dissemination.

Central question

Is influenza A virus replication within cells of hematopoietic origin required for the induction of type I interferon and for the generation of virus-specific CD8 T cell responses in vivo, or can antigen and pathogen-associated molecular patterns supplied by infected epithelial cells substitute for both?

Experimental strategy

The design manipulates the pathogen rather than the host so that one variable, viral replication inside immune cells, can be removed without deleting a cell type, a receptor, or a signaling adaptor. Placing miR-142 target sites in the nucleoprotein segment exploits the restriction of miR-142 expression to hematopoietic lineages, and duplicating the segment five packaging sequence creates a genuine 3' untranslated region so that the inserts do not compromise packaging. A length-matched random insert (NPctrl) provides the control virus, which equates for genome length and for any nonspecific effect of the insertion itself. Small RNA deep sequencing of infected human pulmonary epithelial cells first establishes that infection does not itself reshape the host microRNA profile, which is the assumption the whole approach rests on. Silencing is then verified across an engineered cell line expressing exogenous miR-142, primary cells that do or do not express it endogenously, and the intact animal. With the tool validated, pathogenesis, adaptive immunity, and innate signaling are read out in parallel in the same infection model, and the sensor requirement is tested by repeating the interferon readout in RIG-I deficient and littermate control primary cells.

Key findings

1. Deep sequencing of infected human pulmonary epithelial cells showed no significant change in host microRNA levels during infection (Fig. S1). The authors take this as validation that microRNA-based targeting can be used as a tool without the infection itself moving the target microRNA.
2. In MDCK cells engineered to express miR-142, the NP142t virus lost small viral RNA accumulation (Fig. 1A), viral mRNA and cRNA (Fig. S2C), and detectable nucleoprotein (Fig. 1B), while parental MDCK cells supported both viruses. The loss of both mRNA and cRNA is interpreted as rapid targeting of the incoming genome.
3. Multicycle growth curves showed no difference between NP142t and NPctrl in the absence of miR-142 and more than three logs of attenuation in miR-142-expressing cells (Fig. 1C), indicating the insert imposes no intrinsic fitness cost.
4. Replication of NP142t was abolished in murine bone marrow derived macrophages, which express endogenous miR-142 (Fig. 1D), and was unchanged in primary lung fibroblasts, which do not (Fig. 1E).
5. In infected mice, surface hemagglutinin was lost on lung CD45+CD11c+ dendritic cells during NP142t infection and was unchanged on CD45- cells (Fig. 2A), and infectious virus was reduced in the lung draining lymph node while lung titers were unchanged (Fig. 2B). Because only hematopoietic cells traffic to the node during infection, the authors use the node as an independent in vivo measure of silencing.
6. Weight loss, recovery, and the kinetics of lung viral titer and clearance were indistinguishable between the two viruses (Fig. 2C and 2D), and sequencing confirmed that the absence of a phenotype was not caused by escape mutants (Fig. S3).
7. Direct antigen presentation was lost. JAWS II antigen presenting cells, which express miR-142 (Fig. 3A), activated nucleoprotein-specific CD8 T cell hybridomas after NPctrl infection but only to the level of naive or ultraviolet-inactivated virus controls after NP142t infection (Fig. 3B).
8. Despite that loss, the frequency and number of nucleoprotein- and polymerase acidic-specific tetramer positive CD8 T cells were not significantly different between viruses in C57BL/6 mice at 9 days (Fig. 3C and 3D) or in BALB/c mice at 10 days (Fig. S4), and interferon gamma production after peptide restimulation showed a downward trend that did not reach significance (Fig. 3E and 3F). The authors interpret the combination as evidence that cross-presentation or cross-dressing is sufficient for protective cytotoxic T lymphocyte responses.
9. Interferon beta and IRF-7 induction were reduced in bone marrow derived macrophages infected with NP142t and were unaffected in primary lung fibroblasts (Fig. 4A and 4B), that is, the innate defect tracked with miR-142 expression rather than with the virus itself.
10. Whole lung interferon beta and IRF-7 messenger RNA were significantly reduced during NP142t infection (Fig. 4C and 4D) even though lung tissue is predominantly nonhematopoietic. This is the observation the authors treat as the central result, and they read it as evidence that the hematopoietic compartment contributes disproportionately to the in vivo interferon response.
11. Interferon beta induction was abolished in RIG-I deficient macrophages and fibroblasts for both viruses (Fig. 4E and 4F), while transfected double-stranded RNA still induced interferon beta in those cells (Fig. S5), showing the cells retain the capacity to produce interferon.

Mechanistic model

The data support a model in which the accumulation of influenza replication intermediates inside hematopoietic cells, rather than uptake of material from infected epithelium, is the dominant in vivo trigger of RIG-I-dependent type I interferon, while antigen for CD8 T cell priming reaches the presenting cell through exogenous routes that do not require the presenting cell to be productively infected. Two of the links in this model are shown directly. Silencing removes replication in hematopoietic cells, and whole lung interferon falls. Interferon beta induction by either virus requires RIG-I in the cells tested. The connection between those two observations is inferred rather than demonstrated, because the RIG-I requirement was established in cultured primary cells and not in a hematopoietic-restricted knockout animal, so the study does not establish that the interferon lost in the lung is produced by hematopoietic cells through RIG-I. The specific replication intermediate that RIG-I engages is not identified here. The authors also note that they cannot explain, from these data, why blocking replication in a numerically minor compartment produces so large a deficit in total interferon, and they raise the possibility that NS1-mediated suppression of sensing in epithelial cells biases the response toward the immune compartment. That possibility is offered as a proposal and is not tested.

Conceptual or technical advance

The work makes viral tropism an experimental variable that can be set independently of host genotype. Because a tissue-restricted microRNA is used to silence an essential viral gene, the same animal carries both a permissive and a nonpermissive compartment, and the contribution of one compartment can be read out against an otherwise identical infection. That turns questions previously approachable only by gene deletion or cell depletion, both of which remove functions beyond infection, into gain-of-specificity questions. The validation chain also matters practically, since silencing is verified at the level of small viral RNA, mRNA, cRNA, protein, growth, surface antigen, and organ titer, and escape mutants are excluded by sequencing, which sets the standard for what a microRNA-restricted virus has to demonstrate before its phenotypes can be interpreted.

Relationship to the broader research program

Several threads that recur across this corpus intersect here. The use of small RNA biology as an engineering tool rather than only as an object of study appears in the microRNA targeting strategy, and the reference list points to earlier laboratory work using miR-93 to attenuate influenza in a species-specific manner and miR-142 to restrict dengue virus dissemination. Small viral RNA, a product characterized in earlier laboratory work on the influenza transcription to replication switch, is repurposed here as a sensitive readout of ongoing polymerase activity. The question of which cells produce interferon during infection, and which sensor drives it, is continuous with laboratory work on the interferon enhanceosome and on transcription factor redundancy in establishing the antiviral state, both of which are cited. Placing this paper beside later corpus entries that engineer recombinant viruses to deliver or restrict RNA cargo would be category 3 synthesis, and it should be assembled only when those records exist, since it cannot be read off this paper alone.

Related publications

• Perez et al. 2009, microRNA-mediated species-specific attenuation of influenza A virus. Methodological foundation, cited as the prior demonstration that microRNA target sites in nucleoprotein render influenza replication inversely proportional to the cognate microRNA.
• Perez et al. 2010, influenza A virus-generated small RNAs regulate the switch from transcription to replication. Methodological foundation, the source of the small viral RNA readout and of the northern blot protocol used here.
• Pham, Langlois and tenOever 2012, replication in cells of hematopoietic origin is necessary for dengue virus dissemination. Companion, the same miR-142 restriction strategy applied by the same group to a different virus and a different question.
• Schmid et al. 2010, transcription factor redundancy ensures induction of the antiviral state. Predecessor, laboratory work on interferon induction cited in the discussion of redundant sensing.
• tenOever et al. 2007, multiple functions of the IKK-related kinase IKKepsilon in interferon-mediated antiviral immunity. Predecessor, laboratory work on the interferon induction pathway cited for the enhanceosome step.

Limitations and boundaries

The findings are established in mice infected intranasally with a single laboratory influenza A strain, and the adaptive immune readouts cover two mouse backgrounds at single late time points rather than a full kinetic series. Silencing removes productive replication in hematopoietic cells, so the model cannot distinguish a requirement for viral replication from a requirement for the downstream products of replication, and it cannot address any contribution of abortive infection that still generates sensed RNA below the detection thresholds used. The RIG-I requirement was tested in cultured primary macrophages and fibroblasts, not in animals with a hematopoietic-restricted deletion, so the in vivo sensor attribution is inferred. Because miR-142 is expressed across hematopoietic lineages, the approach silences replication in dendritic cells, macrophages, and other immune cells together and cannot assign the interferon deficit to a single subset. Interferon was measured as messenger RNA for interferon beta and IRF-7 rather than as secreted protein, and the interferon gamma restimulation difference was a nonsignificant trend that the authors explicitly decline to interpret as a functional defect. Finally, the authors note that the magnitude of the interferon deficit relative to the small size of the infected hematopoietic compartment is not explained by the data presented.

Audience summaries

25 words

An influenza virus silenced only inside immune cells still primed normal CD8 T cells and was cleared normally, but induced far less type I interferon.

75 words

Influenza infects airway epithelium and also immune cells, and separating the two contributions has been difficult. By inserting target sites for a blood-lineage microRNA into an essential viral gene, this work built a virus that cannot replicate in immune cells while growing normally in epithelium. Infected mice cleared the virus and made normal antiviral T cells, but produced substantially less type I interferon, implicating infected immune cells as the main source of that alarm signal.

150 words

Whether influenza A virus needs to replicate inside dendritic cells and macrophages, as opposed to simply supplying them with antigen, bears on both how T cells are primed and where interferon comes from. This study engineered a recombinant virus carrying four target sites for miR-142, a microRNA confined to hematopoietic cells, in a duplicated untranslated region of the nucleoprotein segment. The virus grew normally in epithelial cells and fibroblasts and was silenced completely in macrophages, dendritic cells, and the lung draining lymph node, with no escape mutants detected. Mice infected with it showed unchanged weight loss, lung titers, clearance kinetics, and tetramer-positive CD8 T cell responses, despite a measurable loss of direct antigen presentation in culture, which the authors read as sufficiency of cross-presentation. Interferon beta and IRF-7 induction, however, fell in macrophages and in whole lung, and interferon beta induction required RIG-I in the primary cells tested.

Discoveries supported by this paper

Documented publication relationships

• Replication in Cells of Hematopoietic Origin Is Necessary for Dengue Virus Dissemination — companion. Direction: this paper → linked paper. stated in the Related publications section of 2012-langlois-hematopoietic-specific-targeting-o
• Transcription Factor Redundancy Ensures Induction of the Antiviral State — predecessor. Direction: this paper → linked paper. stated in the Related publications section of 2012-langlois-hematopoietic-specific-targeting-o

Pathogens

Technologies