IL-1β plays a beneficial role in the in vivo
innate immune response to IAV infection 
. Our findings reveal a novel regulation of the inflammasome in lung epithelial cells, which are both the primary target of IAV infection in vivo
and major players in the outcome of infection 
. In IAV-infected macrophages, IL-1β secretion is NLRP3-, ASC-, and caspase 1-dependent; it has also been suggested to be RIG-I-independent 
. Conversely, in primary human lung epithelial cells, RIG-I, TLR3, and NLRP3 are partially redundant for IL-1β activation through a caspase 1–dependent mechanism.
Several lines of evidence support our conclusion that RIG-I is a pivotal inflammasome activator in IAV-infected human lung epithelial cells. First, although there was significant inhibition in TLR3 and NLRP3 knockdown cells, knockdown of RIG-I had the most dramatic effect on caspase 1 activation and IL-1β secretion, independent of the donor and of the viral strain used. Second, inhibition of the RIG-I signaling partners MAVS, TRIM25, and Riplet also resulted in significant inhibition of caspase 1 activation and IL-1β secretion. Third, WT RIG-I overexpression resulted in a significant upregulation of IL-1β secretion in response to IAV infection in reconstituted inflammasome experiments. In contrast, the S183I loss-of-function variant of RIG-I 
resulted in inhibition of IL-1β secretion. Fourth, TLR3 and NLRP3 upregulation in IAV-infected NHBE cells is RIG-I dependent, as demonstrated both by RIG-I knockdown experiments and RIG-I overexpression assays. Thus, RIG-I operates upstream of TLR3- and NLRP3-dependent inflammasome activation. Fifth, stimulation of RIG-I is the major inducer of type I IFN responses in lung epithelial cells, as described here and previously 
; also type I IFN precedes and induces the IL-1β response and caspase 1 activation, as demonstrated by knockdown of IFN-β and IFNAR1 expression. Sixth, the functional demonstration that RIG-I directly activates the inflammasome in addition to its effect on the type I IFN response was provided by treatment of primary cells with IFN-β, which partially rescued the knockdown effect of RIG-I on the IL-1β response and cleaved caspase 1 secretion. These results mirror our observation with IFNAR1 and IFN-β knockdowns and support our hypothesis that RIG-I is critical for IFN-β-dependent IL-1β response, as IFN-β significantly increased inflammasome activation. It also supports the functional role of RIG-I per se
in inflammasome activation, as IFN-β had only a partial rescue effect. Moreover, we observed that, in primary human lung epithelial cells, RIG-I directly interacted with caspase 1 and that IAV infection induces RIG-I binding to ASC, as shown in the context of VSV in a human cell line 
. Thus, our data demonstrate that RIG-I directly promotes caspase 1 inflammasome activation in IAV-infected primary human lung epithelial cells and amplifies the IL-1β response through a type I IFN positive feed-back loop as shown in .
Proposed model of RIG-I as a pivotal regulator of the inflammasome response against IAV in lung epithelial cells.
To further validate the cross-talk between RIG-I, type I IFN, and IL-1β during IAV infection, we investigated the impact of influenza NS1 variants with a known effect on virulence and on the type I IFN response in ferrets 
. It is of note that the virus carrying NS1 from the highly pathogenic 1918 influenza strain significantly inhibited the IL-1β response in ferrets and in human primary lung epithelial cells, as compared to NS1 from a seasonal IAV with moderate virulence 
. At least in primary lung epithelial cells, the 1918 NS1 also inhibited both type I IFN and RIG-I upregulation in response to IAV infection, suggesting that its effect on IFN induction contributes to inhibition of the IL-1β response. Furthermore, at a low MOI 1918 NS1 and USSR NS1 are similarly expressed in primary lung epithelial cells, but 1918 NS1 forms more prominent complexes with RIG-I than USSR NS1. This suggests that 1918 NS1 could induce a stronger inhibition of RIG-I activity than USSR NS1, which explains its inhibitory effect on the type I IFN and IL-1β responses. At a high MOI, 1918 NS1 was more abundant than USSR NS1, which may also contribute to its increased inhibitory effect. NS1 has been shown to inhibit type I IFN secretion by interacting with TRIM25 
, an ubiquitin ligase critical for RIG-I function; as we show here, TRIM25 is essential for IL-1β secretion. However, the mechanism underlying the differences in inhibition observed for NS1 from the 1918 and USSR viral strains remains unclear, since the previously identified TRIM25-inhibiting residues E96, E97, R38, and K41 are conserved 
. In addition, NS1 is a multi-functional protein that interferes with the type I IFN response at several levels 
. Similar to the regulation of RIG-I and type I IFN expression that we observed, macaques infected with the reconstructed 1918 pandemic virus showed considerably lower induction of these genes at day 3 post-infection in comparison with macaques infected with seasonal H1N1 
. Thus, we cannot exclude that the inhibitory effect associated with the 1918 NS1 protein could be, at least in part, due to the impact of NS1 on other cellular components involved in type I IFN signaling 
. Further studies involving NS1 site directed mutagenesis and reassortant viruses are necessary to determine whether the 1918 NS1 inhibits TRIM25-dependent RIG-I activation or other cellular components to limit antiviral and inflammasome responses, and which amino acid residues are involved in this function. Altogether, our data on 1918 NS1 raise new questions about the involved mechanism(s), but also demonstrate for the first time the impact of this negative regulator of type I IFN signaling on the lung epithelial cells IL-1β response (). Furthermore, the data support our demonstration of a positive effect of type I IFN on the IL-1β response of lung epithelial cells with an approach using modified virus complementary to the specific targeting of host proteins with siRNA.
The influenza virus M2 protein triggers NLRP3-dependent, but likely RIG-I–independent, inflammasome activation in macrophages 
. Conversely, in lung epithelial cells, NS1 inhibits RIG-I–dependent inflammasomes activation. These results underscore the complexity and specificity of virus interactions with different types of cells during pathogenesis. This suggests that the involvement of different PRRs in inflammasome activation in different cell types and their partial redundancy may play an important role in counteracting viral immune interference, for instance, through NS1 in lung epithelial cells. In line with this idea, 1918 NS1 strongly inhibited IL-1β secretion in lung epithelial cells, which is primarily RIG-I dependent, but did not affect the IL-1β response in macrophages, which is NLRP3 dependent (data not shown). One could hypothesize that the RIG-I–independent inflammasome response of macrophages confers resistance to viruses with an NS1 similar to 1918 NS1, which inhibits the IL-1β response in lung epithelial cells, but not in cells of myeloid origin.
In myeloid cells and in a mouse model of Candida albicans
infection, the presence of type I IFNs inhibits the IL-1β response regulated by the activation of the NLRP3 inflammasome and increases mouse susceptibility to C. albicans
. These results are consistent with reported type I IFN anti-inflammatory effects, which could be due to inhibition of IL-1β production in some autoimmune diseases 
. Our results indicate an entirely different paradigm and suggest that conversely such treatment would be deleterious for diseases implicating the lung epithelium, including COPD or asthma 
. However, type I IFN is required for activation of the inflammasome in macrophages in response to cytosolic Francisella novicida
and Listeria monocytogenes
, but not vacuole-localized Salmonella typhimurium
. The type I IFN response to these cytosolic bacteria induces inflammasome-mediated cell death in macrophages 
. Thus, the action of type I IFNs on IL-1β may be dependent on cell type and on the pathogen. Further IAV infection experiments in different cell types, including peripheral blood mononuclear cells and dendritic cells, will be important in determining whether the RIG-I–mediated type I IFN production that we observed is part of the general host response to IAV, or whether it is specific to lung epithelial cells. It is conceivable that the differential regulation of IL-1β by type I IFNs in non-immune versus immune cells reflects the different functions of type I IFNs, which are both antiviral and regulatory during the early infection of lung epithelial cells. At this time-point, it is important both to control virus growth rapidly and to produce sufficient IL-1β to promote cell recruitment 
. However, at the point at which the macrophages become infected, the immunoregulatory role of type I IFNs may become predominant, to prevent potentially deleterious overproduction of IL-1β by macrophages or dendritic cells. Further studies of the cross-regulation of type I IFN and inflammasome responses are required to determine and contrast the beneficial and deleterious roles of type I IFN in inflammasome activation for virus control and clearance from the host.
Additional experiments are also required to examine whether our results can be generalized to the native epithelium. One approach would be to use cells cultured in an air-liquid interface, which produces polarized pseudostratified mucociliary cells resembling natural airway epithelial cells. Under these culture conditions, well-differentiated epithelial cells produce a reduced inflammatory response and are more resistant to rhinovirus infection than undifferentiated cells 
. The enhanced resistance to IAV infection of these differentiated cells could contribute to the control of the inflammatory response of epithelial cells. Furthermore, it has been shown that cytokine secretion in response to IAV infection is polarized towards the apical and basolateral membranes, whereas IFNAR expression is restricted to the basolateral membrane 
. The localization of IFNAR limits its stimulation by type I IFN to the basolateral side, which could further reduce the extent of the positive-feed back loop acting on IL-1β secretion in differentiated epithelial cells. However, yet another scenario is conceivable: in vivo
, epithelial cells are in close proximity to dendritic cells, which usually reside near the basolateral membrane. In response to virus infection, dendritic cells produce large amounts of type I IFNs, which could act in a paracrine manner to efficiently amplify the production of IL-1β by the airway epithelium. Of note, however, upon IAV infection IFNAR-deficient or STAT1-deficient differentiated mouse epithelial cells show a significant reduction of IL-1β production compared to wild-type controls 
. Thus, although the extent of the IL-1β response in differentiated epithelial cells infected with IAV cells remains to be established in vivo
, the available data indicate that, at least in mouse cells, IL-1β and type I IFN production are positively correlated in differentiated epithelial cells, in a similar fashion to what we have observed in undifferentiated primary human epithelial cells.
Our results showing that type I IFN controls inflammasome activation and production of bioactive IL-1β add to our knowledge of how airway epithelial cells respond to infection and regulate lung innate and adaptive immunity 
. Whereas a large body of evidence supports the critical role of IFNAR/STAT1 axis for both lung epithelial immunity as well as the subsequent adaptive immunity 
, the mechanisms that mediate these responses are yet not entirely clear. IL-1β is central in the communication between epithelial cells and monocytes during the initiation of inflammation and development of the adaptive response. However, IL-1β can also act in an autocrine manner on epithelial cells infected with rhinovirus to elicit secretion of CXCL8, a potent neutrophil attractant, and control of viral replication 
. In addition to long-range signals, epithelial cell IL-1β may have a role in regulating the local homeostasis of the lung epithelium. Epithelial cells express and secrete a battery of immune modulators the role of which is to limit innate inflammation in this environmentally exposed site, including surfactant proteins A and D and mucin 1, which suppress the activity of alveolar macrophages 
. Furthermore, alveolar macrophages display a unique response to PRR agonists, characterized by a strong inflammatory response but lacking autocrine/paracrine IFN-β secretion and STAT-1 activation 
. Thus, it is tempting to speculate that lung epithelial cells, which display a distinctive RIG-I–mediated type I IFN antiviral responses and inflammasome activation, actively participate in the orchestration of lung immune response by providing simultaneously IFN-β and IL-1β paracrine stimulation to alveolar macrophages. Such epithelial responses to infection could provide rapid and robust local immunity to rapidly curb viral load, stimulate immune cells neighboring the affected site while limiting inflammation to the local environment. To clarify these mechanisms conditional or cell type-specific knock-out mice of IL-1β will be required.
The targeting of RIG-I–like receptors (RLR) with specific agonists has been proposed as a prophylaxis and treatment for IAV, as it may increase antiviral innate and adaptive immunity, leading to more rapid elimination of the virus 
. Our results suggest that these therapeutic strategies will not only initiate a powerful antiviral response, but also trigger a potent IL-1β response via RIG-I. Further studies of the tissue specificity of the RIG-I–dependent inflammasome may be useful to determine if other epithelial tissues share the same specificity that we observed in respiratory epithelial cells. This will improve our understanding of whether RLR agonists may be a useful addition to our antiviral therapeutic arsenal, and if they may be instrumental in the next generation of vaccine adjuvants.