The H7N9 influenza A virus infection results in lethal inflammation in the mammalian host via the NLRP3-caspase-1 inflammasome

The avian origin influenza A virus (IAV) H7N9 has caused a considerable number of human infections associated with high rates of death since its emergence in 2013. As a vital component of the host innate immune system, the nucleotide-binding domain leucine-rich repeat containing receptor, pyrin domain containing 3 (NLRP3) inflammasome plays a critical role against H1N1 viral infection. However, the function of NLRP3 inflammasome in host immunological responses to the lethal H7N9 virus is still obscure. Here, we demonstrated that mice deficient for NLRP3 inflammasome components, including NLRP3, caspase-1, and Apoptosis-associated speck-like protein containing a CARD (ASC), were less susceptible to H7N9 viral challenge than wild type (WT) controls. Inflammasome deficiency in these animals led to significantly milder mortality and less pulmonary inflammation compared with WT mice. Furthermore, IL-1 receptor deficient mice also exhibited a higher survival rate than WT controls. Thus, our study reveals that the NLRP3 inflammasome is deleterious for the host during H7N9 infection in mice, which is due to an overwhelming inflammatory response via caspase-1 activation and associated IL-1 signal. Therefore, fine-tuning the activity of NLRP3 inflammasome or IL-1 signaling may be beneficial for the host to control H7N9 associated lethal pathogenesis.

In March 2013, the novel avian-origin influenza A virus (IAV) H7N9 was identified in China. H7N9 exhibits low pathogenicity in birds, but has an ability to bind certain mammalian cell receptors, and replicates at temperature close to the normal body temperature of mammals [1][2][3] . The infected patients often develop severe symptoms, including pneumonia and acute respiratory distress syndrome (ARDS) 4 . The H7N9 infection had caused 1230 confirmed human cases, with 428 deaths by 22 February 2017 [data from FAO (Food and Agriculture Organization of the United Nations)]. The severe symptoms and high mortality associated with H7N9 infection pose a big threat to the public health. Although recent studies of H7N9 have achieved great progress, clear definition of the host immunological responses to H7N9 is still lacking 1,5,6 .
Influenza virus infection mainly induces two types of innate immune responses 7 . A rapid antiviral response, i.e. type I interferon production, to facilitate intracellular control of viral replication; and a proinflammatory response characterized by secretion of cytokines and chemokines, which promotes recruitment of various immune cells for viral clearance 8 . As a key mediator of inflammation, the proinflammatory cytokine IL-1β plays a crucial role in the pathogenesis of various diseases 9 . The maturation and secretion of IL-1β require a protein Figure 1. H7N9 RNA activates the NLRP3 inflammasome in BMDCs. (a) BMDCs were primed with 500 ng/ml LPS and transfected with different dose of H7N9 RNA for 12 hours, the supernatant were harvest and the levels of IL-1β and TNF-α were measured by ELISA. (b) BMDCs from wild type and Nlrp3 −/− , Asc −/− , Casp1/11 −/− mice were primed as in (a) and transfected with 0.5 ug/ml H7N9 RNA for 12 hours, supernatant were collected for ELISA as indicated. (c) Supernatants or cell lysates from H7N9 RNA-stimulated BMDCs were harvested and the expression of proteins was analysed by immunobloting. The upper lanes, marked as pellet, showed the complex called inflammasome, which activates caspase-1 to cleave proIL-1β to its mature form 10 . More importantly, caspase-1 and the NLRP3 inflammasome have been found essential for protecting mice against H1N1 IAV infection [11][12][13] .
In a previous study, we have demonstrated that H7N9 influenza virus isolated from human patients could establish successful infection of mice 14,15 . Employing this murine model, in the current study we demonstrate a detrimental role for the NLRP3-caspase-1 inflammasome and associated IL-1 signal in H7N9 infection of mice. Our data suggest that dampening the NLRP3 inflammasome activity or inhibiting IL-1 signaling should be beneficial for the host during a lethal H7N9 challenge.

Results
H7N9 RNA induces NLRP3 inflammasome activation. It has been reported that H1N1 RNA plays a critical role in activating the NLRP3 inflammasome 11 . To assess whether the NLRP3 inflammasome could be activated during H7N9 challenge, we transfected LPS-primed BMDCs (bone marrow derived dendritic cells) with H7N9 RNA and measured the production of IL-1β. As expected, robust IL-1β secretion was detected in the H7N9 RNA-transfected samples, while the production of an inflammasome-independent cytokine TNF-α was unaltered (Fig. 1a), suggesting that H7N9 RNA-induced IL-1β production was likely dependent on inflammasome activity. To ensure the role of inflammasome in H7N9 RNA-mediated induction of IL-1β, we generated BMDCs from Nlrp3 −/− , Asc −/− and Casp1/11 −/− mice. Transfection of LPS-primed BMDCs from such mice led to significantly compromised IL-1β secretion compared to wild type controls (Fig. 1b). And western blotting revealed that caspase-1 activation and ASC oligomerization were also decreased in BMDCs from Nlrp3 −/− and Asc −/− mice (Fig. 1c), which might account for the compromised IL-1β production. Thus, our results suggest that H7N9 RNA activates caspase-1 via the NLRP3 inflammasome in mouse dendritic cells.
Deficiency of NLRP3 or caspase-1 protects mice against H7N9 infection associated morbidity and mortality. Next, we investigated the function of NLRP3 inflammasome during H7N9 infection in vivo.
Deficiency of NLRP3 or caspase-1 decreases pulmonary inflammation during H7N9 challenge. To study the pathological changes in the H7N9 infected Nlrp3 −/− , Casp1/11 −/− and WT mice, lung tissues were collected and H&E stained samples were examined by microscopy. On 3 d.p.i, infiltration of inflammatory cells was observed in the infected lungs of all strains of mice. There was an obvious thickening of alveolar septum, destruction of partial alveolar structure, as well as pulmonary septal rupture ( Fig. 3a and c). However, infiltration of inflammatory cells to the lungs of WT mice was much more severe than that in Nlrp3 −/− or Casp1/11 −/− mice, especially in the bronchus. In addition, the destruction of alveolar structure in the lungs of Nlrp3 −/− and Casp1/11 −/− mice was weaker than that in WT mice ( Fig. 3a and c). On 7 d.p.i, all strains of mice showed more severe infiltration of inflammatory cells to the lungs comparing to that on 3 d.p.i ( Fig. 3b and c). Nonetheless, the pulmonary pathology of Nlrp3 −/− and Casp1/11 −/− mice was still milder than that in WT mice on 7 d.p.i ( Fig. 3b and c). Collectively, these results suggested a role for NLRP3 and caspase-1 in promoting pathological changes in the lungs of mice infected with H7N9.

Deficiency of caspase-1 or NLRP3 results in attenuated proinflammatory cytokine and chemokine production upon H7N9 infection.
It is reported that the caspase-1, activated by inflammasome, plays a critical role in murine proinflammatory responses to IAV infection 11,12,16 . To investigate the role of caspase-1 in H7N9 infection related inflammation, cytokines and chemokines in the supernatants of lung homogenates were analyzed (Fig. 5). In Casp1/11 −/− lung tissues, the levels of IL-1β (118.8 ± 30.68 pg/g tissue) and IL-18 (1.168 ± 0.3761 ng/g tissue) were significantly lower than those in WT tissues (IL-1β: 309.5 ± 28.54 pg/g; IL-18: 2.856 ± 0.3617 ng/g, respectively) on 7 d.p.i (P < 0.01) (Fig. 5a). Other cytokines such as IL-12/23p40, IL-1α, IFN-γ and IFN-β also exhibited decreased level in Casp1/11 −/− lung tissues compared with WT control on either 3 d.p.i or 7 d.p.i, although the differences were not significant (Fig. 5a). The level of IL-6 fluctuates and was lower in the lungs from Casp1/11 −/− mice on 3 d.p.i, but higher on 7 d.p.i compared to WT counterparts, without significant differences in both cases (Fig. 5a). Other cytokines such as IL-4, TNF-α and IL-17 did not show significant differences between tissues from Casp1/11 −/− and WT mice (Fig. 5a). As for the detected chemokines, MIP-1α was significantly lower in Casp1/11 −/− lung tissues than that in WT control on 3 d.p.i (P < 0.05). The levels of other chemokines, including MIP-1β, IP-10, CXCL1, CCL2 and GM-CSF were just slightly lower in the lungs of Casp1/11 −/− mice than that in WT counterparts on either 3 d.p.i or 7 d.p.i. Therefore, caspase-1 was involved for an optimal production of IL-1β, IL-18 as well as MIP-1α in the lung in response to H7N9 challenge. To note, the production of above mentioned cytokines and chemokines from Nlrp3 −/− mice showed a similar pattern as that from Casp1/11 −/− mice (Fig. 5b), indicating that deficiency of NLRP3 inflammasome genes reduced the secretion of several critical proinflammatory mediators in the lung of mice upon H7N9 IAV challenge. In addition, the production of these cytokines and chemokines in the serum of Nlrp3 −/− , Casp1/11 −/− and WT mice showed a similar trend with that from the lung homogenates ( Supplementary Fig. S1).   ASC and IL-1 receptor mediated signal promoted mortality and weight loss of mice after H7N9 infection. The adaptor protein ASC plays a vital role in the NLRP3 inflammasome formation, which is required for caspase-1 activation and maturation of IL-1β. The mature IL-1β binds to IL-1 receptor 1, the functional receptor of IL-1β, induces downstream signal transduction and executes inflammatory responses. Thus, we set out to determine whether ASC and IL-1R1 play a role similar as that of NLRP3 and caspase-1 during H7N9 IAV infection. To this end, we infected Asc −/− and Il1r1 −/− mice with H7N9 virus and monitored their weight change and mortality daily. Our results demonstrated that the mortality of Asc −/− and Il1r1 −/− mice was lower compared with that of the WT mice (Fig. 6a). The WT mice also suffered more weight loss than these knock-out animals (from 7 to 10 d.p.i) after H7N9 challenge. Of note, the difference for body weight loss between WT and Asc −/− mice was statistically significant (P < 0.05) on 8 and 9 d.p.i. (Fig. 6b).

Deficiency of ASC or IL-1 receptor decreases pulmonary inflammation upon H7N9 infection.
To analyze the pulmonary inflammation of Asc −/− and Il1r1 −/− mice, lung tissues of these mice were harvested for H&E staining and microscopy. On 3 d.p.i, more inflammatory cells were recruited to the alveoli and blood vessels in the lungs of WT mice compared with that of the Asc −/− or Il1r1 −/− mice, and more inflammatory cells infiltrated into the bronchus resulting in a local blockade (Fig. 7a and c). Moreover, the pulmonary exudation was even milder in the Il1r1 −/− mice compared with WT mice (Fig. 7a and c). On 7 d.p.i, all mice suffered more severe inflammation than that observed on 3 d.p.i. Furthermore, there were more inflammatory cells infiltrated in the alveoli of WT mice than in Asc −/− or Il1r1 −/− mice (Fig. 7b and c). Moreover, the damage in the lung of Asc −/− mice was milder compared with that in WT mice, and there was less inflammatory exudation in Il1r1 −/− mice than in WT controls ( Fig. 7b and c). Therefore, our results suggest that signals mediated by ASC and IL-1 receptor lead to more severe disease upon H7N9 IAV challenge.

Discussion
Aberrant and excessive inflammation during H7N9 viral infection is linked to severe morbidity and mortality in human patients 17 . A recent study demonstrated the histological distribution of inflammation-related genes expression in the lungs of BALB/c mice infected with H7N9 influenza A virus 18 . Significantly higher expression levels of NLRP3, RIP3, IL-1β and TNF-α in the lung of H7N9-infected mice were noted, indicating possible role of these molecules in driving lung pathogenesis 18 . For example, the involvement of TNF-α activated signaling pathway during influenza virus infection was clearly demonstrated 19,20 . However, the exact roles of other molecules involved in inflammatory response during H7N9 challenged remained elusive. Specifically, the role of inflammasome signaling had not been defined before our present work.
Here, using the gene knockout mice we demonstrated that activation of the NLRP3 inflammasome was detrimental to the host upon H7N9 IAV infection. We noted that WT mice lost more body weight than Casp1/11 −/− , Asc −/− and Nlrp3 −/− mice and their overall survival rate was lower during H7N9 infection (Figs 2 and 6). Pathology analysis showed that there were inflammatory responses in the infected lungs of all strains of mice, both at 3 d.p.i and at 7 d.p.i; and the pulmonary inflammations at 7 d.p.i. were more severe than that at 3 d.p.i.; moreover, the pulmonary inflammations of the WT mice were stronger than Casp1/11 −/− , Asc −/− and Nlrp3 −/− mice (Figs 3 and 7). Flow cytometric analysis showed that inflammatory cells has started to infiltrate into the alveolar spaces in the BALF at 3 d.p.i. It seemed that CD11b + or F4/80 + (high-expressed on microphages) or CD11c + (high-expressed on dendritic cells) cells in the BALF of WT mice were significantly more than those in Casp1/11 −/− mice at 3 d.p.i. There were also Ly6c + (expressed on monocytes), Ly6g + (high-expressed on neutrophils) and CD3 + (expressed on T cells) in the BALF of WT mice exhibited higher level, though it was not significant differences (Fig. 4). However, except the CD11b + cells were retained and B220+ (expressed on B lymphocytes) cells were increased, the percentage of the most infiltrated cells were decreased in the BALF at 7 d.p.i. (Fig. 4), which seem to be "contrast" to the severer pulmonary inflammation pathology observation at 7 d.p.i. (Figs 3 and 7).   . 3a), and macrophages/dendritic cells and a few other cells started to be recruited to the infected area by chemokine MIP-1α (Figs 4a-c and 5a). The production of IL-1β was still at low level at this stage (Fig. 5a). With the development of disease, at 7 d.p.i., severe inflammation occurred in the infected lung tissue (Fig. 3b). With the migration of the lymphocytes, especially B cells, to the inflammatory area (Fig. 4g), the total number of infiltrated inflammatory cells were increased but the percentage of monocytes, macrophages, dendritic cells were decreased ( Supplementary Fig. S2, Fig. 4). The infiltrated inflammatory cells produced large amount of IL-1β (Fig. 5), which accelerated the pulmonary inflammation. From 7 d.p.i. to 14 d.p.i, the body weight of survived WT mice recovered completely, whereas Casp1/11 −/− mice could not returned to the initial level (0 d.p.i) (Fig. 2d). Microscopic analyses revealed that on 14 d.p.i, most tissues from WT mice exhibited complete recovery, while the Casp1/11 −/− tissues still manifested strong inflammation (data not shown). Thus, WT mice suffered more due to the lung damage and inflammation in the early phase (before 7 d.p.i) of H7N9 infection and either died of organ failure, or recovered at the later phase. In contrast, Casp1/11 −/− mice did not develop as strong inflammation and lung damage as seen in WT counterparts in the early phase, consequently fewer of them died. However, Casp1/11 −/− mice failed to recover completely at the later phase. These results might suggest a potentially dual role for caspase-1 in H7N9 infection, detrimental role to the host upon H7N9 infection before 7 d.p.i. and possibly necessary role for recovery from the lung inflammation after 7 d.p.i. phase, likely due to a contribution to tissue regeneration. But it should be kept in mind that the comparison between 20% surviving WT and 60% surviving Casp1/11 −/− is not justified, and more stringent experiments are needed for further explorations on the role of caspase-1 to host recovery.
Studies on the role of NLRP3-caspase-1 inflammasome during IAV infection produced conflicting results. Some studies have demonstrated that NLRP3-caspase-1 inflammasome is protective against H1N1 infection through viral clearance or virus induced pulmonary necrosis 11, 12 . Inflammation is essential to protect host from infection, but can also cause tissue damage or dysfunction of important organs and lead to death 21 . Dysregulation of the proinflammatory response may result in a "cytokine storm" that contributes to a severe viral pneumonia and serious complications in the lung 21 . The highly pathogenic avian influenza virus H5N1 causes hypercytokinemia and severe tissue damage 22 . It is reported that production of proinflammatory cytokines and chemokines, including IL-6 and IP-10, is remarkably increased in H7N9 infected patients 17,23 , and H7N9 infection associated inflammation in the lungs resulted in rapidly progressive pneumonia and development of ARDS in the majority of hospitalized patients 24,25 . In addition, the H5N1 IAV infected patients who progressed severe lung injury have elevated levels of CCL2, IP-10, CXCL9 and IL-8 compared with the seasonal H1N1 infected patients 26 . Therefore, for the highly pathogenic influenza virus the hyperactivated proinflammatory responses may be detrimental for the host. Moreover, a recent study demonstrates that NLRP3 inflammasome plays both a protective and detrimental role during H1N1 and H3N2 infection of mice depending on the disease phase 13 . In our study, we found the inflammasome activation was deleterious in the case of H7N9 infection, possibly due to overwhelming and fatal acute inflammation in the lungs following H7N9 infection. Pathology and infiltrating cellular analyses as well as cytokine and chemokine analyses all supported this possibility (Figs 3-5). Pathology results revealed that on 3 d.p.i, there were more inflammatory cells around the blood vessels and bronchus in WT mice compared with Casp1/11 −/− mice (Fig. 3). The flow cytometric analyses further demonstrated the abundance of inflammatory cells in the BALF of WT mice compared to Casp1/11 −/− counterparts (Fig. 4), which were consistent with the observed pulmonary pathological changes.
Both Casp1/11 −/− and Nlrp3 −/− mice produced relatively less level of IFN-β in the lung compared with control mice. Although no significant difference was noted, these data indicated a possible role for the NLRP3-caspase-1 inflammasome signaling in affecting the type I interferon production. Theoretically, type I interferon response plays a role in limiting viral replication, however, we did not observe any significant difference in viral load between Casp1/11 −/− or Nlrp3 −/− versus WT mice. It is likely that the H7N9 could employ strategies to counteract the generation and function of type I interferon.
Chemokines are the driving factors promoting the recruitment of inflammatory cells. MIP-1α (CCL3) can activate granulocytes and lead to acute neutrophilic inflammation. It also induces the synthesis and release of IL-1β, IL-6 and TNF-α from fibroblasts and macrophages 27 . MIP-1α level is increased in H7N9 and H5N1 infected mice, but not in H9N2 infected animals 28 . We also noticed a clear increase of MIP-1α in WT mice upon H7N9 infection (Fig. 5). Interestingly, this elevation was abrogated in the early phase (3 d.p.i) in the absence of caspase-1 (Fig. 5). We reasoned that the MIP-1α production may be, directly or indirectly, influenced by NLRP3 or caspase-1-dependent IL-1β and/or IL-18 secretion.
IL-1β showed significant reductions in Nlrp3 −/− and Casp1/11 −/− mice compared with WT on 7 d.p.i (Fig. 5). IL-1β is a critical proinflammatory cytokine that is increased during infections with H1N1 and H5N1 influenza viruses 29 . IL-1β also mediates the recruitment of monocytes and neutrophils into the lung 30 . Deficiency of IL-1β or its receptors manifests different responses to respiratory challenge with influenza virus [31][32][33] . Hence, the function of IL-1β in controlling IAV infections remains obscure. Our data showed that Il1r1 −/− mice exhibited attenuated tissue pathogenesis upon H7N9 infection (Figs 6 and 7).
During the preparation of this manuscript, we noticed that another group found the PB1-F2 protein derived from H7N9 virus also activates the NLRP3 inflammasome 34 . Probably, the viral particle of H7N9 contains multiple factors such as RNA and PB1-F2 that can be detected by the host inflammasome. Beside NLRP3 inflammasome, whether other inflammasomes, such as NLRP1, IPAF, AIM2, could also be involved in H7N9 infection needs to be investigated in further study.
In summary, our present work demonstrates that the NLRP3-caspase-1 inflammasome is deleterious for the mammalian host during H7N9 influenza virus infection. Therefore, fine-tuning the NLRP3-caspase-1 activity or IL-1-mediated signaling should be beneficial to control H7N9-associated lethal pathogenesis. Isolation of H7N9 virus was described before and the viral titer was measured via tissue culture infective dose (TCID 50 ) assay 15,25 . All experiments related to H7N9 virus were performed in biosafety level 3 laboratories in Shanghai Public Health Clinical Center (SPHCC), and followed the standard operating protocols approved by the Institutional Biosafety Committee at SPHCC, Fudan University.
Mice. All mice used in our experiments are on a C57BL/6 genetic background and all experiments were carried out with age and gender matched mice (8-12 weeks old, female). C57BL/6 mice were purchased from the B&K Universal Group Limited (Shanghai, China). Nlrp3 deficient mice (a gift from Dr. Warren Strober) and Asc deficient mice (a gift from Dr. Vishva M. Dixit) had been described 35,36 . Caspase-1/11 and IL-1 receptor 1 deficient mice were purchased from the Jackson Laboratory (Bar Harbor, ME, USA). Herein, these mice were described as Cells. Bone marrow derived dendritic cells (BMDCs) were generated as described 37 . Briefly, bone marrow cells were harvest and lysed with RBC lysis buffer, and seeded in 10 cm dish with 10 ml RPMI1640 with GM-CSF (20 ng/ml). The media were refreshed 3 days later and the cells were harvested for experiments on day 6.

H7N9 RNA extraction and transfection into
Mice infection. Mice were infected with 5 × 10 4 TCID 50 of H7N9 virus in a volume of 50 µl intranasally (i.n.) after sevoflurane (Hearem, Shanghai, China) inhalation anaesthesia, and monitored for clinical signs, body weight changes and physical survival during the 14 days observation period. The infected mice were sacrificed on 3, 7 or 14 days post infection for collection of samples for analysis as reported 15 . To monitor the body weight changes and survival rate, 14 to 22 mice were used in each group (n = 14~22). For the rest analysis of murine samples, more than 3 mice were used in each group.

Blood vessel region inflammation
++ Some blood vessels are cuffed by inflammatory cells ++++ Most blood vessels are surrounded by a thin layer (1-5 cells thick) of inflammatory cells ++++++ Most blood vessels are surrounded by a thick layer (>5 cells thick) of inflammatory cells

Pulmonary alveoli inflammation
++ Increased numbers of inflammatory cells in alveolar walls are evident ++++ As above, plus one to three foci per section showing alveolar exudate and atelectasis ++++++ As above, plus more than three foci per section showing alveolar exudate and atelectasis Lung tissue inflammation ++ Some blood vessels are cuffed by inflammatory cells ++++ A part of lung tissue shows consolidation and interstitial fibrous tissue proliferation ++++++ Consolidation appears in most areas of the lung

Bronchial inflammation
++ Some bronchus are cuffed by inflammatory cells ++++ Some bronchus exhibit squamous metaplasia and are blocked by inflammatory exudates ++++++ Most bronchus exhibit squamous metaplasia and are blocked by inflammatory exudates Table 1. Histology Scoring Criteria. The three levels of inflammatory response for each index, which was designated as ++, ++++ and ++++++, according to the criteria listed in the above table. The milder inflammation was designated with reduced number of "+" as indicated in respective Figures. To warrant the objectivity of the histology scoring, two pathologists from different labs read each slide in a double-blinded way and an average value was taken as the final result.