Reduced Nucleoprotein Availability Impairs Negative-Sense RNA Virus Replication and Promotes Host Recognition

lab-led

Restricting nucleoprotein availability in negative-sense RNA virus infections both blocks full-length genome replication and increases production of aberrant replication products that are sensed through RIG-I and MAVS, so a strong interferon response accompanies rather than follows successful infection.

Benjamin E. Nilsson-Payant; Daniel Blanco-Melo; Skyler Uhl; Beatriz Escudero-Pérez; Silke Olschewski; Patricia Thibault; Maryline Panis; Maria Rosenthal; César Muñoz-Fontela; Benhur Lee; Benjamin R. tenOever

2021 · Journal of Virology · primary research

Senior authors

Benjamin R. tenOever

Correspondence

Benjamin R. tenOever

Research areas & themes

Citation

Nilsson-Payant BE, Blanco-Melo D, Uhl S, Escudero-Pérez B, Olschewski S, Thibault P, Panis M, Rosenthal M, Muñoz-Fontela C, Lee B, tenOever BR. Reduced nucleoprotein availability impairs negative-sense RNA virus replication and promotes host recognition. Journal of Virology. 2021. Volume 95, issue 9, article e02274-20.

DOI 10.1128/jvi.02274-20. PMID 33568513. PMCID PMC8104106.

Nilsson-Payant and Blanco-Melo contributed equally, with author order determined by drawing straws.

One-sentence contribution

Restricting nucleoprotein availability in negative-sense RNA virus infections both blocks full-length genome replication and increases production of aberrant replication products that are sensed through RIG-I and MAVS, so a strong interferon response accompanies rather than follows successful infection.

Executive summary

Negative-sense RNA viruses carry their own polymerase and wind their genomes around a nucleoprotein scaffold to form ribonucleoprotein complexes. Nucleoprotein was known to regulate the balance between transcription and genome replication and to shield viral RNA from host nucleases to differing degrees across virus families, but its role in preventing host detection during infection had not been separated from its role in replication. The authors used recombinant influenza A virus and Sendai virus carrying microRNA target sites in the untranslated region downstream of the nucleoprotein open reading frame, an arrangement that degrades nucleoprotein messenger RNA without targeting genomic viral RNA. Silencing nucleoprotein abolished detectable viral protein and genome replication yet produced a strongly elevated interferon response relative to the amount of virus present. Sequencing showed reduced genome coverage together with read enrichment at segment termini and increased noncanonical junction reads for influenza, and 3 prime terminal enrichment indicative of copy-back products for Sendai. Northern blotting revealed accumulation of mini-viral RNA that tracked with interferon beta induction, a relationship reproduced in a reconstituted replication complex where nucleoprotein was titrated against fixed polymerase. Reporter assays in knockout cells placed the response through RIG-I and MAVS rather than MDA5. Knockdown of nucleoprotein across seven negative-sense RNA viruses spanning six families gave the same pairing of lost replication with induced IFIT1, while equivalent knockdown of SARS-CoV-2 nucleocapsid reduced replication without inducing interferon. Nucleozin, which targets nucleoprotein, elicited interferon while the polymerase inhibitor baloxavir marboxil did not.

Scientific context

Nucleoprotein had two established roles that had largely been studied separately. As a replication factor it acts as an elongation factor for the polymerase, and for influenza it is dispensable for templates up to about 76 nucleotides while supporting diminished synthesis on templates up to about 125 nucleotides, which is understood to prioritise viral protein synthesis early and to delay genome replication until enough protein is present to package new material. For paramyxoviruses and rhabdoviruses the polymerase favours transcription when nucleoprotein is scarce and replication when it is abundant. As a protective scaffold, its effectiveness varies. Ribonucleoproteins of human parainfluenza virus 5 and Rift Valley fever virus resist ribonucleases, vesicular stomatitis virus sequesters RNA in a deep cavity, rabies virus ribonucleoproteins are only partially resistant, influenza genomes contain both shielded and nucleoprotein-free structured regions, and Ebola virus nucleoprotein appears less protective. Separately, work by other groups had established defective viral genomes as important agonists, including mini-viral RNAs of under 100 nucleotides consisting only of the terminal promoter regions, which are recognised by RIG-I and drive exaggerated interferon responses, and which accumulate when polymerase is in relative excess. What had not been tested was whether deliberately lowering nucleoprotein, rather than raising polymerase, produces the same outcome, and whether the relationship holds across the phylum.

Central question

Does nucleoprotein availability govern host detection of negative-sense RNA virus infection independently of its role in supporting replication, and if so, does limiting nucleoprotein generate the aberrant replication products that pattern recognition receptors sense?

Experimental strategy

The design turns on separating protein availability from genome availability. Recombinant influenza A and Sendai viruses carry a cassette of microRNA target sites downstream of the nucleoprotein open reading frame, positioned so that nucleoprotein messenger RNA is degraded in cells expressing those microRNAs while genomic viral RNA is not targeted. Matched viruses with nonfunctional target sites provide the control, so the comparison is between two infections that differ in nucleoprotein supply alone. Three layers of evidence are then stacked. Sequencing of viral reads reports genome coverage, terminal enrichment, and noncanonical junctions as proxies for defective genome production, with the authors noting that junction-spanning reads undercount the true defective population. Northern blotting against the conserved 5 prime promoter detects all replication products including species smaller than the 106 nucleotide U6 loading control, which allows mini-viral RNA to be seen directly alongside interferon beta induction measured on the same samples. To remove any strain-specific explanation, the influenza replication complex is reconstituted from plasmids with a truncated 200 nucleotide segment 6 template, fixed polymerase, and a titration of nucleoprotein, with a catalytically inactive polymerase as control, so the relationship between nucleoprotein dose, product size, and interferon can be read in isolation from infection. Sensing is then assigned genetically using A549 reporter cells deficient in RIG-I, MDA5, or MAVS. Generality is tested by small interfering RNA knockdown of nucleoprotein across seven negative-sense viruses covering Orthomyxoviridae, Paramyxoviridae, Pneumoviridae, Rhabdoviridae, Filoviridae and Arenaviridae, with SARS-CoV-2 nucleocapsid included as a positive-sense counterpoint. Finally the principle is tested pharmacologically by comparing a nucleoprotein-directed inhibitor with a polymerase-directed one.

Key findings

1. Targeting nucleoprotein messenger RNA abolished detectable viral protein through late time points for both influenza A virus and Sendai virus, yet IFIT1 protein rose markedly relative to the nontargeted controls (Figures 1C and 1D).

2. Messenger RNA sequencing showed induction of type I interferon response genes including the receptors DDX58, IFIH1 and TLR3, the transcription factors IRF1 and IRF7, and effectors including BST2, IFIT1 to IFIT3, OAS1 and MX1 (Figures 1E and 1F). For Sendai virus the absolute induction was equal or slightly lower than control, but viral transcripts were reduced roughly 300-fold, so the response is large relative to the replication present. Gene set enrichment confirmed antiviral categories (Figure 1G).

3. Viral read coverage and total relative viral RNA were significantly reduced for both targeted viruses (Figures 2A, 2B, 2E and 2F). For influenza, coverage of segments 1 to 3 and segment 8 was enriched at both termini, a pattern characteristic of defective interfering particles (Figures 2A and 2C), and noncanonical junction reads increased significantly (Figure 2D). For Sendai virus, reads were enriched at the 3 prime terminus of the antigenome, which the authors read as copy-back defective genomes typical of paramyxoviruses (Figure 2G).

4. Northern blotting against the conserved 5 prime promoter showed steady accumulation of full-length viral RNA and small viral RNA in control infections. With nucleoprotein targeted, full-length viral RNA was undetectable while species larger than small viral RNA and smaller than the 106 nucleotide loading control accumulated strongly, matching the published description of mini-viral RNA (Figure 3A). Their appearance as early as three hours post-infection coincided with strong interferon beta messenger RNA induction (Figure 3B).

5. In the reconstituted system, increasing nucleoprotein against fixed polymerase increased full-length viral RNA and decreased mini-viral RNA, while catalytically inactive polymerase or absent nucleoprotein gave no product at all, consistent with the published requirement for nucleoprotein on templates above roughly 76 nucleotides (Figure 3C). Interferon beta induction tracked with the presence of mini-viral RNA (Figure 3D). Because this system contains no NS1, the authors take it as showing that loss of the interferon antagonist is not what produces the response, while acknowledging that in infection its absence contributes.

6. In interferon-stimulated response element reporter cells, the response to nucleoprotein-targeted influenza was strong in wild-type cells, lost in cells lacking RIG-I and lost in cells lacking MAVS, and not significantly affected by loss of MDA5 (Figure 3E).

7. Small interfering RNA knockdown of nucleoprotein reduced the fraction of infected cells for influenza A virus, human parainfluenza virus 3, vesicular stomatitis virus and Ebola virus, with respiratory syncytial virus and Lassa virus not significantly changed by that measure, while mean viral protein fluorescence intensity fell significantly for all (Figures 4A and 4B). Immunoblotting showed loss of viral protein together with IFIT1 induction for all seven negative-sense viruses tested (Figures 4C to 4I).

8. Knockdown of the SARS-CoV-2 subgenomic nucleocapsid transcript reduced viral replication, confirmed as a significant drop in the percentage of viral reads, without inducing IFIT1 (Figures 4J and 4K). The authors offer two candidate explanations, that positive-sense genomes are noninflammatory because they are capped and polyadenylated, or that coronavirus replication is sequestered in lipid-enclosed vesicles, and do not distinguish between them.

9. Nucleozin, which targets nucleoprotein, and baloxavir marboxil, which targets the polymerase PA subunit, both blocked influenza replication in a concentration-dependent manner, but only nucleozin also induced IFIT1 (Figure 3F). The authors propose on this basis that nucleoprotein-directed drugs could offer bystander priming of neighbouring cells, which is an extrapolation from cell culture.

Mechanistic model

The mechanism the data support runs as follows. Nucleoprotein acts as an elongation factor, so when it is scarce the polymerase can initiate but cannot processively copy full-length templates. Short products, in particular mini-viral RNAs under about 100 nucleotides, fall below the length at which nucleoprotein is required and are therefore preferentially synthesised and amplified under exactly the conditions that prevent full-length replication. These short products carry 5 prime triphosphate promoter ends and are potent agonists, and the reporter genetics place their detection through RIG-I signalling to MAVS. The net result is an inverted relationship in which less virus produces more interferon. Several elements remain interpretation rather than demonstration. The authors state explicitly that it is unclear how much of the host response in their experiments is driven by mini-viral RNA as opposed to longer defective genomes, and they note that their junction-read quantification substantially undercounts total defective genomes because a read can only be classified as defective when it spans a noncanonical junction. Their extension of the model to explain why high multiplicity passage yields defective interfering particles, through nucleoprotein demand outstripping supply late in infection or under host shutoff, is an interpretation offered in discussion rather than a result. The absence of an interferon response after SARS-CoV-2 nucleocapsid knockdown is presented as a contrast whose cause is not resolved. The evolutionary argument in the discussion, that the fragility of the nucleoprotein balance may explain why negative-sense RNA viruses are less diverse than positive-sense ones, is explicitly speculative.

Conceptual or technical advance

The work reframes nucleoprotein as a single control point where replication competence and immune invisibility are coupled, so that any perturbation reducing it necessarily trades one for a worse outcome in the other. That has a direct practical consequence. Two drugs that block influenza replication equally well differ in whether they also engage host defences, which suggests that the target chosen determines whether an antiviral additionally primes uninfected cells, and the authors note that nucleoprotein has been comparatively neglected as a target for clinically important negative-sense viruses. The microRNA targeting arrangement also supplies a way to remove a viral protein during genuine infection without altering the genome that the polymerase copies, which is what allows protein supply and genome supply to be separated. The demonstration that the same pairing holds across six negative-sense families while failing for a coronavirus nucleocapsid sets a boundary on the principle that is useful for vaccine and adjuvant design.

Relationship to the broader research program

The recombinant influenza and Sendai viruses used here come from the laboratory's established practice of inserting microRNA target sites into viral genomes to control the expression of individual viral gene products, an approach developed earlier for species-restricted and tissue-restricted attenuation and used elsewhere in the corpus as a conditional genetics tool. Marked as category 3 synthesis, the corpus-level pattern is that the same insertion strategy recurs as an instrument across very different questions, here to titrate a structural protein during infection, elsewhere to silence a herpesvirus gene only in one lineage. A second recurring thread is the treatment of aberrant viral RNA as the actual substrate of innate detection, which connects this study to the laboratory's longstanding interest in what viral RNA species the cell sees and in small RNAs generated during influenza replication. The inclusion of SARS-CoV-2 alongside the negative-sense panel reflects the period in which the work was completed and the laboratory's parallel coronavirus programme.

Related publications

• Aguado et al. 2018 (PNAS), methodological foundation. Same laboratory. Cited as the source of the IAV-NPT, IAV-NPC, SeV-NT and SeV-NC recombinant viruses and of the microRNA silencing cassette used throughout, and of their rescue procedure.
• Perez et al. 2010 (PNAS), predecessor. Same laboratory. Characterised influenza A virus-generated small RNAs and their role in the switch from transcription to replication, the small viral RNA species visible alongside mini-viral RNA in the Northern blots here.
• Perez et al. 2012 (Journal of Virology), predecessor. Same laboratory. Small-RNA enhancement of viral polymerase activity, part of the same line of work on influenza replication products.
• Te Velthuis et al. 2018, predecessor. Established mini-viral RNA as a RIG-I agonist generated when polymerase is in relative excess, the prior observation that the present work approaches from the opposite direction by limiting nucleoprotein.
• Benitez et al. 2015, conceptual extension. Same laboratory. In vivo RNA interference screening identifying MDA5, cited here in the context of pattern recognition receptor contributions to influenza sensing.
• Blanco-Melo et al. 2020 (Cell), companion. Same laboratory with overlapping authorship, and the source of the A549-ACE2 system used for the SARS-CoV-2 arm.

Limitations and boundaries

Nearly all of the work is in A549 lung epithelial cells or in transfected HEK-293T cells, with no primary cells, airway models, or animals, so the consequences of limited nucleoprotein in tissue or in an infected host are untested. Nucleoprotein levels are reduced by microRNA targeting or by small interfering RNA rather than titrated to defined amounts during infection, so the study describes a scarcity regime rather than a dose relationship, with the sole exception of the plasmid-based reconstitution, which uses a truncated 200 nucleotide template and therefore does not model a real genome. Quantification of defective genomes by junction-spanning reads is acknowledged by the authors to be a substantial undercount, and the relative contribution of mini-viral RNA against longer defective genomes to the interferon response is explicitly left unresolved. Most sequencing rests on two or three biological replicates at single time points and high multiplicity. The cross-family survey uses different reporter viruses, different readouts of replication, and a single knockdown condition per virus, and for respiratory syncytial virus and Lassa virus the percentage of infected cells did not change significantly even though protein intensity did. Sensing was assigned in one reporter cell line for influenza only, so RIG-I dependence is not established for the other viruses. The negative result with SARS-CoV-2 rests on one siRNA against the subgenomic nucleocapsid transcript in one cell line. The proposed therapeutic advantage of nucleoprotein-directed inhibitors rests on a single comparison of two compounds in cell culture with IFIT1 as the readout, with no demonstration of bystander protection or of benefit in vivo. The paper states that it shows for the first time that these findings apply broadly across negative-sense RNA viruses, and that priority claim is the paper's own.

Audience summaries

25 words

Starving a negative-sense RNA virus of its nucleoprotein stops it replicating but makes it far more visible, because the polymerase then churns out short immune-triggering fragments.

75 words

Negative-sense RNA viruses wrap their genomes in nucleoprotein, which the polymerase also needs to copy full-length templates. Silencing nucleoprotein in influenza and Sendai virus infections blocked replication yet produced a much stronger interferon response, because the polymerase generated short defective genomes that RIG-I detects. The same pairing held for seven viruses across six families but not for SARS-CoV-2 nucleocapsid, and a nucleoprotein-directed drug triggered interferon where a polymerase-directed drug did not.

150 words

Nilsson-Payant and colleagues used recombinant influenza A and Sendai viruses carrying microRNA target sites downstream of the nucleoprotein open reading frame, which degrades nucleoprotein messenger RNA without targeting genomic RNA. Silencing abolished detectable viral protein and full-length genome replication while strongly inducing interferon-stimulated genes relative to the viral material present. Sequencing showed terminal read enrichment and increased noncanonical junctions, and Northern blotting showed accumulation of mini-viral RNA tracking with interferon beta. Titrating nucleoprotein against fixed polymerase in a reconstituted replication complex reproduced the inverse relationship between full-length product and mini-viral RNA, and reporter cells lacking RIG-I or MAVS lost the response while MDA5-deficient cells did not. Knockdown of nucleoprotein in seven negative-sense viruses across six families consistently paired lost replication with IFIT1 induction, whereas SARS-CoV-2 nucleocapsid knockdown reduced replication without inducing interferon. The authors propose nucleoprotein as a drug target that would also engage host defences.

Discoveries supported by this paper

Documented publication relationships

• A Small-RNA Enhancer of Viral Polymerase Activity — predecessor. Direction: this paper → linked paper. stated in the Related publications section of 2021-nilsson-payant-reduced-nucleoprotein-availability
• Imbalanced Host Response to SARS-CoV-2 Drives Development of COVID-19 — companion. Direction: this paper → linked paper. stated in the Related publications section of 2021-nilsson-payant-reduced-nucleoprotein-availability

Pathogens

Technologies