A Plant-Derived Nucleic Acid Reconciles Type I IFN and a Pyroptotic-like Event in Immunity against Respiratory Viruses

This article is featured in In This Issue, p.2189

The mammalian innate immune response to pathogen invasions is a complex cascade of events that is initiated upon sensing of pathogen-associated molecular patterns (PAMPs). Although some viruses display PAMPs on their genetic materials, which are recognized as foreign by host immune receptors, others generate such structures or patterns during replication. Recognition of PAMPs, such as dsRNA structures, by TLRs or RIG-I–like receptors activates signal-transduction pathways that culminate in the production of antiviral effector molecules capable of stemming virus spread (1–3). Type I IFN (IFN-I) is considered the main antiviral cytokine capable of inducing a large number of protective molecules (4–6). Activation of antiviral immune responses is also associated with a range of inflammatory signals that trigger the activation of caspases, IL-1β production, and inflammatory cell recruitment (7–10).

The cascade of events involved in the immune response differs slightly between tissues and organs. The respiratory tract, for example, is home to numerous immune cells. Upon infection and inflammation, circulating immune cells are quickly recruited to help mount a strong immune reaction (7, 9). Although the central role of IFN-I has long been recognized in the antiviral response, recent reports have shed light on the contribution of inflammatory events (8, 11–18). Particular attention has been given to the recently observed inflammasome activation driven by viral RNA (and analogs) recognition by immune receptors. It is now accepted that recognition of viral RNA species in the respiratory tract activates caspases and cell death as part of the host immune response (8, 11, 14, 16). However, direct evidence linking caspase-1 activation by viral RNA analogs to immunity has been lacking.

For decades, simulation of virus infection and replication has been achieved by the use of synthetic RNA mimicking PAMPs. Since the late 1960s, the use of such analogs has helped to elucidate key components of antiviral immunity. Furthermore, activation of the antiviral immune response has been achieved by the use of artificial dsRNA, represented by polyinosinic-polycytidylic acid (poly I:C) (19–23). Historically, immune protection induced by such artificial analogs has been attributed almost exclusively to IFNs, without considering inflammatory events, therefore raising questions about their accuracy to mimic viral RNA. Although attempts to stabilize poly I:C for accurate in vivo and clinical applications have been sought (21, 24), the artificial nature of the molecule remains a concern to many investigators. In fact, poly I:C is composed of nonphysiological polyinosinic acid and polycytidylic acid.

To address this issue, we investigated other immune-stimulating molecules. In this report, we focused on a natural long (>10 kbp) dsRNA, a genome of Endornavirus, which is found in several plants (25–30) and some species of fungi (31). In plants, endornaviruses are transmitted vertically through pollens and ova without causing any notable disease. This plasmid-like dsRNA is found in several edible plants, such as some species of rice and green bell peppers. We used dsRNA extracted from rice bran (rb-dsRNA) in this study. Our findings suggest that rb-dsRNA possesses the ability to consistently stimulate the induction of murine IFN genes, as well as IFN-stimulated genes, in vitro and in vivo. In the respiratory tract, alveolar macrophages (AMΦs) appear to be the main producers of IFN-I upon stimulation with rb-dsRNA. Importantly, the rb-dsRNA sensing activated antiviral protection against respiratory viruses, such as influenza A virus (IAV) and murine parainfluenza type I virus (Sendai virus [SeV]), in a manner that is dependent on IFN-I and caspase 1. rb-dsRNA provides the advantage of being of natural origin, suggesting that it could be a safer alternative to synthetic analogs. Furthermore, because of its natural nucleotide sequence (32), we propose that rb-dsRNA could be a more accurate mimic of viral RNA. Additionally, this study reconciled IFN-I and an inflammatory event in antiviral immune protection driven by a nucleic acid mimicking replicating viruses.

Six- to eight-week-old female mice were used throughout the study. IAV (A/Puerto Rico/8/34 H1N1) was propagated in fertilized chicken eggs, and the titer was quantified using standard plaque assay with MDCK cells. SeV (Cantell strain) was propagated in 293T cells, and titer was quantified using a standard plaque assay with Vero cells. The following Abs were used: purified rat anti-mouse CD16/CD32 clone 2.4G2 (BD Biosciences), anti-mouse F4/80 clone BM8 (BioLegend), anti-mouse SiglecF clone M1304A01 (BioLegend), anti-mouse CD11c clone N418 (BioLegend), anti-mouse CD11b clone M1/70 (BioLegend), anti-mouse Ly-6G clone 1A8 (BioLegend), anti-mouse Ly-6C clone HK-1.4 (BioLegend), anti-influenza A NP clone A1 (Millipore), anti-human influenza A, B, rabbit polyclonal (Takara Bio), anti–caspase-1 p10 clone M-20 (Santa Cruz Biotechnology), and anti-GSDMDC1 clone A-7(Santa Cruz Biotechnology).

Mice were anesthetized with pentobarbital (i.p.) before intranasal (i.n.) administration of 30 μl of inoculum per mouse. Respective inocula contained 20 μg of rb-dsRNA, IAV (Puerto Rico/8/34), SeV (Cantell strain), any other treatment (e.g., clodronate liposome), or the same volume of PBS as control.

Mice were infected with 1000 PFU IAV, except when otherwise indicated. For survival experiments, 200 PFU IAV was used. For SeV, 3000 PFU was used throughout the study.

The caspase-1–specific inhibitor Ac-YVAD-cmk (InvivoGen) and IL-1R antagonist (IL1RA), also called Anakinra (MyBioSource), were administered several times before i.n. administration of rb-dsRNA or PBS. Briefly, mice received two i.p. injections (10 mg/kg) of Ac-YVAD-cmk or IL1RA or vehicle (12 h apart [for Ac-YVAD-cmk] or 8 h apart (for IL1RA)] 12 h (for Ac-YVAD-cmk) or 8 h (for IL1RA) before administration of the first rb-dsRNA dose. These treatment regimens were optimized according to the manufacturer’s instructions for Ac-YVAD-cmk and as previously described for IL1RA (33–35).

Total RNA was extracted from cells or tissues using TRIzol reagent (Invitrogen, Life Technologies) and subjected to DNase I treatment before 200 ng of total RNA was reverse transcribed with MultiScribe Reverse Transcriptase (Applied Biosystems). Gene expression was quantified by real-time quantitative PCR (RT-qPCR).

Lungs were excised, washed, incubated, and homogenized in 5 ml of buffer containing collagenase and DNase I. The lung homogenate was passed through a 70-μm cell strainer, centrifuged at 300 × g, and resuspended in the appropriate buffer for Ab staining as follows. Cells were incubated for 30 min with purified rat anti-mouse CD16/CD32 clone 2.4G2 (BD Biosciences) to block nonspecific binding before cell-type–specific Abs were applied. For intracellular staining, cells were permeabilized before Ab application. Cells were washed three times and resuspended with the appropriate buffer before analysis.

Excised lungs were fixed in 4% paraformaldehyde and dehydrated in sucrose before being frozen in Tissue-Tek Cryomold (Sakura Finetek). Sections were cut and stained as follows. The immunoblocked sections were incubated with the primary Ab overnight at 4°C before the secondary Ab was applied, and the slides were observed under a confocal microscope.

Statistical analysis was performed using Stata 13.1 (StataCorp). Two-group comparisons were performed using an unpaired two-tailed Student t test. Multiple-group comparisons were performed using one-way ANOVA with the Bonferroni test. Data are expressed as mean ± SEM unless otherwise indicated, and differences are categorized as not significant, *p < 0.05, **p < 0.01, or ***p < 0.001.

Animals were stored and handled in accordance with the regulation on animal experimentation at Kyoto University and the fundamental guideline for proper conduct of animal experiment and related activities in academic research institutions under the jurisdiction of the Japanese Ministry of Education, Culture, Sports, Science, and Technology. The respective studies on C57BL/6 (B6) wild-type (WT) and knockout (KO) mice were reviewed and approved by the Animal Experimentation Committee of Kyoto University.

Supplemental Fig. 1 shows a schematic representation of the rb-dsRNA extraction protocol. Supplemental Fig. 2 shows the stimulation of murine bone marrow–derived MΦ (BM-MΦs) with rb-dsRNA and poly I:C. Supplemental Fig. 3 shows that i.n. inoculation of clodronate liposomes depletes AMΦs. Supplemental Fig. 4 shows the virus-inhibitory capacity of prophylactic and therapeutic rb-dsRNA inoculation.

We extracted and purified rb-dsRNA from Japanese rice Koshihikari cultivar, as illustrated in Supplemental Fig. 1. rb-dsRNA had a similar size compared with previously reported Endornavirus RNA (13.9 kbp, Fig. 1A) (29, 30). To confirm the rb-dsRNA identity, the purified nucleic acid was digested with a range of nucleases (Fig. 1B). The extracted nucleic acid did not contain any obvious trace DNA because it was resistant to DNase I, was sensitive to RNase III, but was resistant to S1 nuclease. RNase III and S1 nuclease have specific activities against dsRNA and single-stranded nucleic acids, respectively. The purified nucleic acid was sensitive to RNase A treatment in a low-salt condition, but it was relatively resistant in a high-salt condition. These results confirm that the purified nucleic acid is a long dsRNA molecule.

To assess rb-dsRNA’s capacity to stimulate immune responses similarly to poly I:C, a murine MΦ cell line (Raw 264.7) was stimulated (without any transfection reagent) with rb-dsRNA or poly I:C as a positive control. rb-dsRNA exhibited a strong capacity to induce the production of IFN-β (Fig. 1C) and a chemokine, IP-10 (Fig. 1D). These data demonstrate that rb-dsRNA possesses a strong immune-stimulating capacity. Therefore, we tested it in a range of in vivo studies.

Intranasal administration of rb-dsRNA in mice resulted in induction of antiviral cytokine and chemokine genes. Most notably, mRNA levels of IFN-β and IFN-α4 (Fig. 2A), members of the IFN-I family, were upregulated very quickly, reaching a peak of expression at ∼8 h. Several other IFN-stimulated genes (ISGs) were also upregulated. Expression of BST2 (also known as tetherin) (Fig. 2A), a host factor blocking the release of newly formed enveloped viruses (36), was gradually increased. Expression of ISGs reported to inhibit virus entry, replication, and maturation were also notably increased (Fig. 2B). IFN-induced transmembrane protein-3 is known to inhibit entry and replication of IAV and several other viruses (37, 38). ISG56 is known to inhibit replication and translation of several viruses (37, 39, 40), whereas Pai-1 inhibits extracellular maturation of IAV (41). Elevated expression of chemoattractants with C-X-C-motifs (NAP-3, MIP-2α, and IP-10) suggested that rb-dsRNA can provoke leukocyte recruitment (Fig. 2C). Expression levels of inflammatory cytokines, such as IL-1β, IL-6, and TNF-α (Fig. 2D), were transiently increased, suggesting activation of inflammatory reactions. Taken together, i.n. administration of rb-dsRNA mimicked virus-derived dsRNA by quickly upregulating the expression of a wide range of cytokines and chemokines associated with immune responses.

To investigate the activation of signaling pathways, we used KO mouse models, as depicted in Fig. 3A, and bone marrow MΦs derived from the respective models. At 8 and 12 h after i.n. rb-dsRNA administration, mice were analyzed for mRNA expression and protein secretion, respectively. These time points corresponded to the mRNA (Fig. 2) and protein (data not shown) peaks of IFN-I and most cytokines tested. Previous reports suggested that MDA5 preferentially senses long dsRNA ligands, whereas RIG-I recognizes shorter RNAs with different signatures (42, 43). Mice lacking IPS-1 or MDA5 showed reduced IFN-β mRNA induction compared with WT mice (Fig. 3B); however, these expression levels were significantly lower than those in WT mice. This suggests that rb-dsRNA was sensed by MDA5 and triggered signaling through IPS-1. In contrast, mice lacking TRIF, an adaptor molecule for TLR3 signaling, also exhibited reduced IFN-β expression, suggesting the involvement of TLR3. To further confirm the involvement of IPS-1 and TRIF signaling pathways, we used mice lacking TRIF and MDA5 (TRIF/MDA5 double KO [DKO]). Upregulation of IFN-β mRNA and protein in bronchoalveolar lavage fluid (BALF) was completely abolished in DKO mice (Fig. 3C). Consistent with these results, rb-dsRNA–induced expression of IP-10 and IL-1β was also abrogated in DKO mice (Fig. 3D, 3E). Furthermore, we investigated the involvement of RIG-I in the rb-dsRNA–induced immune response (Fig. 3F). Using primary mouse embryonic fibroblasts (MEFs), we confirmed that rb-dsRNA does not activate RIG-I, as indicated by the unchanged induction of IFN-β and MDA5 genes in RIG-I–KO MEFs. However, cells deficient in MDA5 lost their ability to activate IFN-β and its stimulated genes MDA5 and RIG-I. This observation supported the possibility that cytosolic rb-dsRNA activates MDA5 with no evidence of RIG-I activation.

In in vivo settings, because mice were administered rb-dsRNA without a transfection reagent, we speculated that the response is mediated by phagocytic cells in the lungs. MΦ populations are considered professional phagocytes that are constantly present in tissues as sentinels against foreign invasion. We differentiated MΦs from murine bone marrow cells in the presence of M-CSF and examined their responsiveness to rb-dsRNA. IFN-β was undetectable in TRIF-KO cells, whereas MDA5 KO partially impaired the induction, and DKO cells did not respond to these dsRNA, suggesting a dominant contribution of the TLR3–TRIF signaling axis for IFN-β gene induction in these cells (Supplemental Fig. 2A). Production of the IFN-stimulated protein IP-10 was attenuated in TRIF- or MDA5-KO MΦs, with TRIF deficiency showing the strongest reduction, whereas DKO MΦs barely responded to rb-dsRNA or poly I:C (Supplemental Fig. 2B).

rb-dsRNA is recognized by TLR3 and MDA5, but not RIG-I, and it subsequently induces IFN production through TRIF and IPS-1 adaptors, respectively. At the cellular level, these results also suggest the possible implication of MΦs in the activation of the immune response by rb-dsRNA in murine lungs.

MΦs, particularly AMΦs, have been shown to act as sentinels for tissue homeostasis by driving virus-induced and RNA-induced IFN-I production in the lungs. By doing so, they act as primary activators of antiviral innate immunity in the lungs (7, 44). To identify cells producing IFN-I upon i.n. rb-dsRNA administration, we discriminated the two MΦ populations in the lung: interstitial MΦs (IMΦs), located in the parenchymal space, and AMΦs, found in the airway space (45, 46). We isolated SiglecF⁺ F4/80⁺ (AMΦ) and SiglecF⁻ F4/80⁺ (IMΦ) MΦs by MACS. RT-qPCR revealed that AMΦs were the main lung MΦ population exhibiting upregulation of IFN-I upon administration of rb-dsRNA (Fig. 3G, 3H). Taking advantage of their strategic localization, we performed a depletion experiment to directly assess the contribution of AMΦs to IFN-I and cytokine expression. MΦ depletion by i.n. administration of clodronate liposomes resulted in depletion of AMΦs, with no obvious reduction in the percentage of SiglecF⁻ F4/80⁺ (IMΦs/monocytes) (Fig. 3I, Supplemental Fig. 3). In fact, IMΦs/monocytes were increased, presumably as a result of monocyte infiltration in response to MΦ depletion. Depletion of AMΦs significantly reduced rb-dsRNA–induced IFN-I and the inflammatory cytokine IL-1β (Fig. 3J).

Taken together, these data suggest that lung MΦs are a major source of IFN-I induced by rb-dsRNA and that AMΦs are the MΦ subpopulation responsible for IFN-I production.

Because rb-dsRNA activates a cascade of events associated with the antiviral innate immune response in the lung, we assessed its ability to induce immune protection against IAV (H1N1, Puerto Rico/8/34) and SeV. Mice were administered rb-dsRNA i.n. before and after the IAV challenges to assess its prophylactic and therapeutic potential, respectively. Although both treatment regimens demonstrated rb-dsRNA’s capacity to stem virus proliferation in the lungs (Supplemental Fig. 4A, 4B), survival of mice infected with a highly lethal amount of virus was not statistically significant in an experimental setup using five or six mice (data not shown). To overcome this, mice were treated twice (before and after) throughout this study, as depicted in Fig. 4A, to characterize the antiviral activity induced by rb-dsRNA. This treatment regimen was used throughout the study to characterize antiviral activity induced by rb-dsRNA. Mice treated with rb-dsRNA exhibited significantly high survival upon IAV infection (Fig. 4B). IAV is known to cause lung injury that can culminate in hypoxia, respiratory failure, and death (18, 47–50). Treatment with rb-dsRNA reduced IAV-induced acute inflammation, which was correlated with a higher body weight (Fig. 4C, 4D). In SeV-challenged mice, rb-dsRNA treatment significantly reduced morbidity (Fig. 4E).

Next, we examined the IAV titer in the lung (Fig. 5A). rb-dsRNA treatment reduced the IAV titer at 1 and 3 d postinfection (p.i.); however, at 7 d p.i., the infectious viral titer declined, and the difference was indistinguishable between treated and untreated animals. Similarly, viral RNA quantification showed suppressed viral replication at 1 and 3 d p.i. but not at 7 d p.i. (Fig. 5B). Inhibition of IAV replication at early time points was also apparent in histochemical analyses (Fig. 5C). The number of IAV Ag⁺ cells was clearly reduced at 1 d p.i. in rb-dsRNA–treated mice, whereas this change was less obvious at 7 d p.i. Interestingly, SeV-infected mice exhibited sustained inhibition of viral replication up to 7 d p.i. (Fig. 5D). These results suggest that treatment with rb-dsRNA enhanced antiviral immunity and reduced viral load in the lungs early in infection.

To confirm whether rb-dsRNA–driven antiviral protection is directly linked to TLR3/TRIF and MDA5/IPS-1 recognition of the nucleic acid and subsequent IFN production, we tested mice deficient in the TRIF and IPS-1 signaling pathways (TRIF/MDA5 DKO) and the IFN-I receptor (IFNAR1 KO). Survival was examined in mice treated with rb-dsRNA and challenged with IAV. In TRIF/MDA5-DKO mice, rb-dsRNA did not confer resistance to IAV infection, in marked contrast to WT mice treated with rb-dsRNA (Fig. 6A). In contrast, IFNAR1-KO mice exhibited reduced, but partial, protection (Fig. 6B). We further examined lung tissues of these mice. The inhibitory effect of rb-dsRNA treatment on viral load and viral RNA accumulation in the lungs was absent in TRIF/MDA5-DKO mice (Fig. 7A–C), suggesting that the inability to sense rb-dsRNA abolished the activation of antiviral protection against IAV. Consistent with the survival result (Fig. 6B), IFN-I receptor deficiency resulted in partial rb-dsRNA–induced antiviral protection (Fig. 7A, 7C). Recent studies have established the important role of IFN-λ (IL-28) in immunity against influenza (51, 52). Therefore, we speculated that the residual rb-dsRNA–induced virus inhibition could be linked to IFN-λ. Mice lacking IL-28α signaling capacity (IL-28αR KO) did not exhibit any reduced ability to inhibit virus spread (Fig. 7D).

These results suggested that, although there was a residual antiviral immunity, IFN-I induced via TRIF and IPS-1 by rb-dsRNA was an important contributor to the observed antiviral protection, whereas IFN-III was not. To completely characterize the observed antiviral immunity, we further investigated other possibilities.

Considering that IFN-I only played a partial role in antiviral immune protection induced by rb-dsRNA, we investigated the contribution of other dsRNA-induced events. In addition to induction of IFNs, recognition of viral RNA and analogs is associated with inflammatory signatures, such as recruitment of inflammatory cells (10) and activation of the inflammasome (8, 11, 14), or alternative inflammatory and anticancer events (16, 53). These responses have been shown to contribute to immunity.

We sought to determine whether rb-dsRNA also induced cell death, particularly pyroptotic events in vitro. When mice were subjected to i.n. rb-dsRNA treatment, MΦs in BALF exhibited robust cell death (Fig. 8A). Next, we stimulated BM-MΦs and AMΦs in vitro. Stimulation of primary MΦs indicated that rb-dsRNA mediated spontaneous lactate dehydrogenase (LDH) release in the supernatant (Fig. 8B), an event associated with pyroptosis. Unlike rb-dsRNA, poly I:C did not result in LDH release. In fact, in previous reports, in vitro LDH release induced by poly I:C on primary MΦs required the addition of ATP (14).

Using primary MΦs obtained from caspase 1–deficient mice, we found that rb-dsRNA–induced LDH release is dependent on caspase 1 (Fig. 8C). To further examine caspase 1 involvement, mice were treated with the caspase 1 inhibitor Ac-YVAD-cmk and stimulated with rb-dsRNA. Mice administered rb-dsRNA showed an increased level of LDH in the BALF, and this increase was abrogated by caspase 1 inhibition (Fig. 8D). Likewise, PI⁺ and annexin V⁺ cells induced by rb-dsRNA were decreased by caspase 1 inhibition (Fig. 8E). Immunoblot analysis of BALF cells confirmed caspase 1 activation (Fig. 8F), and IL-1β release in the BALF of rb-dsRNA–treated WT mice was confirmed by ELISA (Fig. 8J). dsRNA have previously been shown to activate caspase 1 in MΦs, subsequently resulting in inflammasome formation (14). To reveal the signaling cascade responsible for caspase 1 activation, we treated different KO mice with rb-dsRNA and examined activation of caspase 1. rb-dsRNA–driven activation of caspase 1 was independent of the MDA5/IPS-1 axis (Fig. 8G), but it required TLR3/TRIF signaling (Fig. 8H, 8I). MDA5 appeared to play a role, although it was not significant, in IL-1β production, whereas TRIF-deficient mice completely lost the ability to produce detectable amounts of IL-1β in BALF (Fig. 8J). Considering that MDA5 did not contribute to caspase 1 activation, we speculated that its contribution to IL-1β production was at the mRNA level. RT-qPCR revealed that, although the TRIF signaling pathway was the dominant contributor to mRNA induction of IL-1β, MDA5 signaling also partially participated in the transcriptional activation of IL-1β mRNA (Fig. 8K). These results suggest that, in addition to IFN-I gene induction and production, rb-dsRNA triggers caspase 1 activation in MΦs through a TRIF signaling adaptor. This cascade of inflammatory events subsequently resulted in the release of LDH and secretion of IL-1β, an event linked to pyroptosis. Release of LDH and IL-1β has been reported to depend on gasdermin D (54–57). To confirm whether gasdermin D cleavage is involved in our observations, we investigated it in BALF cells. Intranasal administration of rb-dsRNA led to the cleavage of gasdermin D downstream of TRIF (Fig. 8L), suggesting its involvement in dsRNA-induced pyroptosis.

To further investigate whether rb-dsRNA–driven caspase 1 activation contributes to antiviral protection in vivo, we treated mice with the caspase 1 inhibitor and challenged them with IAV. IAV replication was inhibited by rb-dsRNA treatment, and inhibition of caspase 1 significantly reduced rb-dsRNA–driven antiviral activity (Fig. 9A). Consistent with these results, rb-dsRNA–induced IAV inhibition was reduced in caspase 1–KO mice (Fig. 9B) without significantly affecting IFN-β induction (Fig. 9C). These results suggest that the caspase-1–dependent inflammatory response contributes to antiviral immune activity in vivo independently of IFN.

To reconcile IFN-I signaling and caspase 1 activation in rb-dsRNA–driven antiviral protection, we administered the caspase 1 inhibitor to IFNAR1-KO mice. Partial protection by rb-dsRNA in IFNAR1-KO mice was abrogated by caspase 1 inhibition (Fig. 9D), suggesting an IFN-independent, caspase 1–dependent role in antiviral protection. This observation prompted us to investigate the possible involvement of IL-1β. Inhibition of IL-1 signaling using the IL1RA Anakinra suggested that the IFN-independent, caspase 1–dependent residual antiviral protection in IFNAR1-KO mice was independent of IL-1 signaling (Fig. 9E). Interestingly, IFNAR1-KO mice lost the ability to produce detectable amounts of IL-1β in BALF (Fig. 9F) and exhibited significantly decreased induction of IL-1β mRNA (Fig. 9G) (58). This suggested even more that a contribution by IL-1β to antiviral protection in IFNAR1-KO mice was unlikely. Furthermore, the survival rate of IFNAR1-KO mice treated with rb-dsRNA was significantly lower in the presence of the caspase 1 inhibitor and was comparable to TRIF/MDA5-DKO mice (Fig. 9H).

We observed that rb-dsRNA upregulates chemoattracting cytokine genes and causes cell death in the alveolar compartment (Fig. 8). This observation prompted us to examine cell populations in the lungs of rb-dsRNA–treated and/or IAV-infected mice. Leukocytes were enzymatically isolated from the whole lung and monitored based on their respective cell markers (Fig. 10A, Supplemental Fig. 3).

Notably, a dramatic decline in AMΦ frequency was induced within 12 h after rb-dsRNA treatment. This frequency increased gradually, returning to the original level after 6 d. IAV infection also induced a decrease in AMΦs, reaching the lowest level after 3 d, without a notable recovery up to 6 d. rb-dsRNA–treated and IAV-infected mice showed a sharp decrease in AMΦs, similar to mice treated with rb-dsRNA alone; however, unlike IAV-infected mice, the frequency of AMΦ partially recovered (Fig. 10B).

In contrast to AMΦs, transient neutrophil accumulation was induced by rb-dsRNA; it returned to normal levels after 3 d. IAV infection alone induced moderate neutrophil accumulation at 72 h p.i., and the level was maintained at 144 h p.i. rb-dsRNA–treated and IAV-infected mice showed a sharp increase in neutrophils, similar to mice treated with rb-dsRNA alone; however, unlike IAV-infected mice, their number continued to decline until 6 d p.i. (Fig. 10C).

Similar to neutrophils, monocyte accumulation was induced by rb-dsRNA and IAV infection in a transient and persistent manner, respectively. Unlike for neutrophils, rb-dsRNA–treated and IAV-infected mice exhibited a biphasic increase in monocytes; this resulted in a similar monocyte level among untreated and rb-dsRNA–treated and IAV-infected mice at 6 d p.i. (Fig. 10D).

These results illustrate that rb-dsRNA treatment and viral infection differentially promoted dynamic immune cell population change in the lungs. This suggests that rb-dsRNA–driven immune protection could extend beyond a direct antiviral effect by modulating the immune population in the lungs, thereby improving lung function and host survival.

In this article, we introduced a novel immune stimulant that can protect mice from lethal airway viral infection. This ligand is of natural origin; therefore, it might be a safer in vivo alternative to widely considered synthetic analogs. Because long dsRNA generated during viral replication is often infectious and hard to obtain in large quantities, the enzymatically synthesized dsRNA, poly I:C, has been used for immunological investigation for many years. For this purpose, polynucleotide phosphorylase, an enzyme polymerizing nucleoside diphosphates into polynucleotide in a template-independent manner, was used. By annealing the resultant homopolymer, dsRNAs, such as poly A:U, poly C:G, and poly I:C, are obtained. Of these dsRNAs, poly I:C exhibited, by far, the strongest immune-stimulatory effect, indicating the importance of its unique nucleotide sequence; however, its mechanism is still unclear. Therefore, it is suggested that it may not mimic the precise biological activity of natural virus–derived dsRNA. The use of natural dsRNA as an immunological stimulant has been sought; however, previous reports (59, 60) were limited to in vitro assessment with remote physiological relevance. We revealed that the plant-derived long dsRNA activates MDA5 in addition to TLR3.

rb-dsRNA exhibited a strong immune-stimulating capacity in mice. In addition to IFN-I, a panel of cytokines, chemokines, and antiviral proteins is induced by rb-dsRNA with respective kinetics and contributes to direct downregulation of viral replication early p.i. MDA5- and TLR3-mediated signaling commonly regulates these genes. rb-dsRNA treatment in mice suppressed subsequent IAV replication at an early time point; however, viral load was restored to a similar level as the untreated mice at a later time. In contrast, with the milder SeV, suppression of viral replication persisted. It is of note that laboratory mice lack functional mx1, an IFN-inducible gene that is critical for resistance to IAV. Therefore, the anti-IAV response mediated by IFN-I is genetically attenuated in laboratory mice. Nevertheless, the mice are capable of mounting protection against IAV by activating additional mechanisms.

Our study suggests the importance of caspase 1–mediated events in controlling RNA virus replications and host protection independently of IFN-I and IFN-III (Fig. 7). Caspase 1 activation is specifically induced by TLR3 signaling through TRIF/RIPK/FADD, resulting in activation of the NLRP3 inflammasome (14), whereas the MDA5-mediated signal does not activate the inflammasome (8). Our results indicate that caspase 1 activation by rb-dsRNA spontaneously induces a pyroptosis-like cell death in MΦs, in contrast to poly I:C, which requires addition of ATP (14). Gasdermin D, which is the executor of pyroptosis, was also activated through TLR3/TRIF recognition and signaling (Fig. 8).

Although the antiviral activity of pyroptosis has been proposed (61, 62), no definitive mechanism has been described. It is possible that cells, particularly MΦs, eliminate pathogens by dying after engulfment of pathogens. However, this mechanism is less likely in our experimental setting because rb-dsRNA treatment alone causes depletion of AMΦs. Another possibility is that pyroptotic cells release their cytoplasmic content, thereby enhancing inflammatory signals leading to the recruitment of immune cells, such as neutrophils. Such mechanisms have been demonstrated in bacterial-induced pyroptosis (13, 15, 17, 63), whereas the observed antiviral mechanism of pyroptosis still needs to be clarified. Our results suggest that this IFN-independent immune protection induced by rb-dsRNA was independent of IL-1β signaling, because blockade of IL-1R did not alter the residual antiviral protection (Fig. 9E). Actually, lack of IFNAR1 has been reported to reduce IL-1β induction in the lungs upon RNA stimulation (Fig. 9F, 9G) or RNA virus infection (58), thereby minimizing IL-1β’s impact. Therefore, caspase 1 activation initiates an IL-1β–independent pyroptotic-like antiviral activity.

In addition to direct antiviral activity, rb-dsRNA treatment transiently reduced AMΦ frequency; however, it was completely recovered (Fig. 10B). In contrast, lethal IAV infection depletes the AMΦ population, but its recovery was not observed. This depletion by IAV infection contributes to virus-induced morbidity and mortality (49, 50, 64). Interestingly, rb-dsRNA treatment of IAV-infected mice allowed recovery of AMΦs depleted by IAV infection. Although not complete, this observation suggested that rb-dsRNA–driven repopulation of AMΦs could contribute to host survival. It is possible that this phenomenon is directly linked to viral load during the early time of infection. In contrast to AMΦs, i.n. administration of rb-dsRNA caused recruitment of neutrophils in the lungs (Fig. 10C). This accumulation of neutrophils was transient, and no infiltrated neutrophils were evident after 3 d. In contrast to rb-dsRNA, replicating IAV may constantly provide ligands and cause inflammation, resulting in prolonged infiltration of neutrophils throughout the course of infection. Although neutrophils have been shown to play a role in antiviral innate immune responses (9, 65), Ab-mediated depletion of neutrophils immediately before and after rb-dsRNA i.n. administration did not affect antiviral protection (data not shown). Similar to neutrophils, monocytes are transiently accumulated within 24 h of rb-dsRNA administration (Fig. 10D). In contrast, IAV infection gradually increased monocytes in the lungs. rb-dsRNA–treated and IAV-infected mice exhibited a biphasic accumulation of monocytes. However, there is no evident correlation between accumulation of inflammatory monocytes and host survival.

In summary, our study proposes a new TLR3 and MDA5 ligand for in vitro and in vivo studies of the immune system. It is able to activate antiviral immunity against respiratory viruses by reconciling IFN-I with direct antiviral activity and an inflammatory event, which may facilitate recovery, as well as immobilization of acquired immune responses. In the farming industry and in clinical settings, rb-dsRNA could provide a degree of protection to unvaccinated hosts in the center of a respiratory virus outbreak. This study opens doors to several nucleic acid–based therapeutic studies, including anticancer and inflammatory diseases, such as multiple sclerosis. Further studies will clarify the immune-modulating capacity of this dsRNA in body tissues in addition to the lungs.

We thank Fumitaka Miyoshi and Menjie Zhu for help with rb-dsRNA extraction, Dr. Michaela Weber for advice and support, and Dr. James Hejna for critically reading the manuscript.

The authors have no financial conflicts of interest.