ZBP1-dependent inflammatory cell death, PANoptosis, and cytokine storm disrupt IFN therapeutic efficacy during coronavirus infection

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus responsible for coronavirus disease 2019 (COVID-19), continues to cause significant morbidity and mortality in the ongoing global pandemic. Understanding the fundamental mechanisms that govern innate immune and inflammatory responses during SARS-CoV-2 infection is critical for developing effective therapeutic strategies. While IFN-based therapies are generally expected to be beneficial during viral infection, clinical trials in COVID-19 have shown limited efficacy and potential detrimental effects of IFN treatment during SARS-CoV-2 infection. However, the underlying mechanisms responsible for this failure remain unknown. In this study, we found that IFN induced ZBP1-mediated inflammatory cell death, PANoptosis, in human and murine macrophages and in the lungs of mice infected with β-coronaviruses, including SARS-CoV-2 and mouse hepatitis virus (MHV). In patients with COVID-19, expression of the innate immune sensor ZBP1 was increased in immune cells from those who succumbed to the disease compared with those who recovered, further suggesting a link between ZBP1 and pathology. In mice, IFN-β treatment following β-coronavirus infection increased lethality, and genetic deletion of Zbp1 or its Zα domain suppressed cell death and protected the mice from IFN-mediated lethality during β-coronavirus infection. Overall, our results identify that ZBP1 induced during coronavirus infection limits the efficacy of IFN therapy by driving inflammatory cell death and lethality. Therefore, inhibiting ZBP1 activity may improve the efficacy of IFN therapy, paving the way for the development of new and critically needed therapeutics for COVID-19 as well as other infections and inflammatory conditions where IFN-mediated cell death and pathology occur.


INTRODUCTION
Coronavirus disease 2019 , caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, has led to substantial mortality, morbidity, and global socioeconomic and psychological distress (1,2). Severe cases of COVID-19 are characterized by acute respiratory distress syndrome, multiorgan failure, and death (3,4). Innate immunity is the first line of host defense against invading viruses, including SARS-CoV-2. Conserved cellular receptors in innate immune cells detect viral DNA and RNA and rapidly induce interferon (IFN) production, which subsequently up-regulates numerous IFN-stimulated genes (ISGs) to limit viral replication (5,6). However, SARS-CoV-2 can evade antiviral responses by actively interfering with IFN production (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17), and the sera of patients with mild or moderate COVID-19 has similar levels of type I and III IFNs compared with those of healthy patients (18). To counteract these evasion mechanisms, IFN treatment has been considered as a therapeutic option for patients with COVID- 19. In preclinical studies, treatment with exogenous type I IFNs or agonists that induce type I IFN reduces viral load when administered before SARS-CoV-2 infection (19)(20)(21)(22), but this effect is limited once infection is established (23). These observations suggest that IFN-based therapies have prophylactic, but limited therapeutic, potential. Moreover, the use of IFN-based therapy in the treatment of patients with COVID-19 has yielded mixed results (24,25). The World Health Organization (WHO) Solidarity Trial Consortium clinical trial found that type I IFN treatment is not beneficial in treating SARS-CoV-2 infection (24). In addition, a retrospective study found that administration of IFN-2b within 5 days of hospitalization is associated with a decrease in mortality, whereas later administration is associated with increased mortality (25). Patients with severe COVID-19 are also reported to have increased IFN levels in their sera compared with those with mild infection (26). Together, these findings indicate that IFN-based therapy may only be beneficial to a certain subset of patients when administered early during the infection and needs to be balanced to avoid pathogenic effects. However, determining the optimal time frame for safe and effective IFN therapy is challenging. Therefore, gaining further insights into the molecular mechanisms of IFN signaling during SARS-CoV-2 infection is essential to inform clinical protocols and improve the efficacy of IFN therapy in COVID- 19.
IFN signaling has multiple roles in response to viral infection. In addition to inducing ISGs that have antiviral properties (27), IFN signaling also up-regulates ISGs that trigger cell death (28,29). In COVID-19, there is a positive feedback loop between excessive cell death and IFN signaling that leads to cytokine storm, contributing to multiorgan damage as well as clinical features of disease and mortality (4,30). Therefore, it is important to understand the connection between IFN, cell death, and pathology during SARS-CoV-2 infection to identify ways to maximize the efficacy of IFN-based therapeutic strategies.
One critical ISG that plays a role in cell death during other viral infections is Z-DNA-binding protein 1 (ZBP1), and its complex roles in innate immunity and inflammatory cell death have recently been found. During viral infection or IFN- treatment, ZBP1 is up-regulated in an IFNAR-dependent manner (29,31). Sensing of viral and endogenous Z-nucleic acids by the ZBP1 Z domain allows ZBP1 to interact with RIPK3 via RHIM-RHIM homotypic interactions, thereby activating inflammatory cell death, PANoptosis (29,(31)(32)(33)(34). In contrast, up-regulation of ZBP1 by IFN- treatment alone is not sufficient to induce cell death (29). Whether ZBP1 has a role in driving the pathology during severe COVID-19 or whether it has an impact on the efficacy of IFN-based therapies is not known. Here, using CRISPR-CAS9 knockout (KO) screening and the analysis of multiple publicly available RNA sequencing (RNA-seq) datasets, we identified ZBP1 as a critical cytosolic sensor that drives inflammatory cell death in response to IFN therapy during -coronavirus infections, including SARS-CoV-2 and mouse hepatitis virus (MHV) infections. IFN- treatment during SARS-CoV-2 and MHV infections induced robust ZBP1-dependent inflammatory cell death, PANoptosis, in macrophages. In addition, single-cell transcriptomic analysis showed that ZBP1 was more abundantly expressed in immune cells from patients who succumbed to COVID-19 than in those who recovered without needing hospitalization. Furthermore, genetic deletion of Zbp1 or its Z domain protected mice from the lethality induced by IFN treatment during -coronavirus infection. These findings suggest that inhibiting ZBP1 activity could improve the efficacy of IFN therapy during -coronavirus infection. This improved understanding of the fundamental mechanisms engaged by innate immune receptors in inducing inflammatory responses and PANoptosis during SARS-CoV-2 infection is essential to inform the development of effective therapeutic strategies for the treatment of not only COVID-19 but also other infections and inflammatory diseases.

IFN response is delayed during infection with -coronavirus
Understanding the differentially regulated pathways in patients with varying severities of COVID-19 is critical to identifying the molecular mechanisms that drive pathogenesis of severe SARS-CoV-2 infection. Using a publicly available dataset, we performed pathway analysis to determine the differentially regulated pathways in noncritically ill (NCI) and critically ill (CI) patients with COVID-19 (35). We found that the IFN- and IFN- response pathways were highly enriched in both NCI and CI patients with COVID-19 compared with healthy patients (fig. S1A). Moreover, among the hallmark pathways (36,37) (see Materials and Methods), IFN- response was the topmost enriched pathway in CI patients compared with NCI patients with COVID-19 (Fig. 1A). We also analyzed the IFN responses to determine differentially regulated ISGs during SARS-CoV-2 infection. Several ISGs were up-regulated in CI patients with COVID-19 compared with healthy patients or NCI patients with COVID-19 ( fig. S1B). Among the ISGs up-regulated, most of them, including BATF2, PARP9, DDX60, IFIT3, and IFI44, have been reported to have antiviral properties (27,38,39), whereas some are involved in inducing cell death, such as EIF2AK2 (28) and ZBP1 (29,31,32). However, pathway analysis of whole blood transcriptomes showed that IFN responses in patients with COVID-19 were not as robust as in patients with influenza or sepsis ( fig. S2) (40). Because IFNs can have both beneficial and detrimental effects depending on the context (41), we sought to determine the effect of IFNs on cell death during infections with -coronaviruses and influenza A virus (IAV). Murine bone marrow-derived macrophages (BMDMs) infected with MHV, a prototypical virus of the -coronavirus genus that mimics many of the key aspects of human -coronavirus biology, showed delayed dynamics of IFN- release and signal transducer and activator of transcription 1 (STAT1) activation and reduced ISG up-regulation compared with IAV-infected cells (Fig. 1, B and C). We also found that SARS-CoV-2-infected human THP-1 cells showed delayed IFN responses compared with IAV-infected cells (Fig. 1, B and C). In addition, analysis of public datasets (42,43) showed early induction of both IFN- and IFN- responses in the lung tissue-derived epithelial cell line Calu-3 upon infection with IAV compared with Calu-3 cells infected with SARS-CoV-2 (Fig. 1D). Moreover, patients with COVID-19 are known to exhibit markedly delayed IFN induction (40). Together, these data suggest that IFN responses are delayed during -coronavirus infection.

IFN treatment contributes to cell death and lethality during -coronavirus infection
Because IFN responses are reduced in patients with mild and moderate COVID-19 (18,40), IFN therapy has been suggested as a treatment strategy for patients with COVID-19. However, IFN-based therapy is beneficial only when administered before the onset of clinical symptoms (25). On the other hand, delayed IFN- treatment is pathogenic in a mouse model of Middle East respiratory syndrome coronavirus (MERS-CoV) infection (44), and administration of IFN-2b later than 5 days after hospitalization in patients with SARS-CoV-2 is associated with increased mortality (25), suggesting that the timing of the IFN response relative to virus replication dictates the outcome. To investigate the beneficial or pathological effects of IFN therapy during -coronavirus infection, we evaluated the effect of IFN treatment during infection in mice. Because infection with -coronaviruses induced delayed IFN responses (Fig. 1, B to D) and because we wanted to mimic the increased IFN responses observed in CI patients with COVID-19 (Fig. 1A), we treated MHV-infected or SARS-CoV-2-infected wild-type (WT) mice with recombinant IFN- and examined the pathology and lethality. MHV-infected mice were dosed with IFN- on days 1 and 3 after infection, and SARS-CoV-2-infected mice were dosed on days 2 and 4 after infection. Whereas 50 to 60% of WT mice infected with the median lethal dose (LD 50 ) of MHV survived ( Fig. 2A and fig. S3A), all the MHV-infected WT mice receiving IFN- succumbed to infection within 7 days ( Fig. 2A). Moreover, 63% of the SARS-CoV-2infected WT mice receiving IFN- succumbed to infection within 5 days (Fig. 2B), suggesting that IFN- treatment contributes to lethality during -coronavirus infection.
Several studies have indicated that IFNs potentiate inflammatory cell death, PANoptosis, leading to cytokine storm and organ damage (4,45). Consistent with this, at day 3 after MHV infection, we observed increased production of inflammasome-dependent cytokines interleukin-1 (IL-1) and IL-18 as well as other pro-inflammatory cytokines in the bronchoalveolar lavage fluid (BALF) of WT mice after treatment with IFN- compared with phosphate-buffered saline (PBS) (Fig. 2C). The lungs of IFN--treated mice showed widespread septal thickening with scattered syncytial cells and extensive necrosis and loss of bronchiolar epithelium in affected areas during MHV infection ( Fig. 2D  and fig. S3B, arrows). The increased production of inflammatory cytokines was accompanied by an increase in alveolar, perivascular, and peribronchiolar infiltrates of immune cells, such as macrophages and neutrophils in the lungs of MHV-infected WT mice treated with IFN- ( Fig. 2D and fig. S3, C and D). In addition, there was augmented cell death, as evidenced by the presence of an increased number of TUNEL + (terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick end labeling-positive) cells, in the lungs of MHV-infected mice treated with IFN- compared with the infected mice treated with PBS or mock-treated mice (fig. S3, E and F), which likely accounts for the extensive necrosis and loss of bronchiolar epithelium observed in these mice. TUNEL positivity, desquamation of bronchial epithelium, marked inflammatory responses, and necrosis were also found in lung samples from human patients who succumbed to SARS-CoV-2 infection ( fig. S4).
To further evaluate the molecular mechanisms of the IFN-induced pathology during infection, we tested whether the increased cell death observed in the lungs of mice treated with IFN- during MHV infection could be recapitulated in vitro. We found that BMDMs infected with MHV showed delayed cell death compared with cells infected with IAV ( Fig. 2, E and F). Moreover, treatment with IFN- or IFN-, but not IL-6, IL-1, or tumor necrosis factor (TNF), significantly increased cell death at 24 hours after MHV infection ( fig. S5). Whereas addition of IFN- after infection did accelerate MHV-induced cell death, consistent with our in vivo findings, the supplementation of IFN- did not change the dynamics of cell death induced by IAV infection (Fig. 2, E and F). Together, these data indicate that IFN treatment robustly potentiates the ability of -coronaviruses to induce cell death, contributing to cytokine storm, lung damage, and lethality.   (42,43). Data are representative of at least three independent experiments (B and C). ****P < 0.0001. Analysis was performed using the two-way ANOVA (B). Data are shown as means ± SEM.

IFN treatment potentiates inflammatory cell death, PANoptosis, during -coronavirus infection
Mechanistically, pyroptosis is executed by gasdermin family membermediated pore formation, which can be induced through inflammasome activation and caspase-1 cleavage of gasdermin D (GSDMD) (46,47). Apoptosis is induced by the initiator caspases caspase-8/10 or caspase-9, which activate executioners caspase-3 and caspase-7 to drive cell death (48,49). Necroptosis involves RIPK3-mediated mixed lineage kinase domain like pseudokinase (MLKL) oligomerization to induce MLKL pores in the membrane and execute cell death (50,51). Studies have recently identified another cell death pathway called PANoptosis, a unique inflammatory cell death pathway that integrates components from other cell death pathways. The totality of biological effects in PANoptosis cannot be individually accounted for by pyroptosis, apoptosis, or necroptosis alone.

ZBP1 contributes to pathology and lethality during -coronavirus infection
Cytosolic sensors that are involved in sensing and inducing inflammatory cell death during -coronavirus infection remain largely unknown. To identify the innate immune sensors that are required for cell death during -coronavirus infection, we performed a whole-genome CRISPR-CAS9 KO screen in murine immortalized BMDMs (iBMDMs) using MHV. After generating a pool of cells with individual genes deleted by CRISPR, we infected them with MHV for 24 hours and analyzed the surviving pool of cells to identify the genes that were enriched or depleted in this population. The enriched genes in the surviving pool of cells could be potentially important to play a positive role in inducing cell death during -coronavirus infection. Because IFN- treatment contributed to lethality in mice, we then focused our analysis on the ISGs that were significantly up-regulated in the surviving pool ( Fig. 4A and table S1). Of these ISGs, ZBP1 has been reported to trigger inflammatory cell death in several contexts (29,31,34,53,66,68), including viral infection with IAV. Moreover, ZBP1 was also identified among the ISGs up-regulated in CI patients with COVID-19 ( fig. S1B). Therefore, we hypothesized that ZBP1 may be involved in lethality observed in patients with severe COVID-19. To further explore whether ZBP1 expression may correlate with pathology, we reanalyzed publicly available single-cell RNA-seq datasets to examine the expression of ZBP1 in immune cells from healthy patients compared with cells from patients with stable COVID-19, who did not require hospitalization, and with cells from those with progressive COVID-19, which led to lethality (69). We observed that the expression of ZBP1 was significantly increased in neutrophils, dendritic cells, macrophages, basophils, natural killer (NK) cells, CD4 + T cells, and regulatory T (T reg ) cells from patients with COVID-19 compared with healthy controls (fig. S6A). Moreover, patients with progressive COVID-19, who succumbed to infection, had increased ZBP1 expression in their immune cells, particularly NK cells, CD4 + memory and naïve T cells, effector memory CD8 + T cells, and B memory and naïve cells, compared with patients with stable COVID-19 ( fig. S6B). Together, these data suggest a positive correlation between ZBP1 expression and disease and lethality in patients with COVID-19. Next, we sought to determine the physiological role of ZBP1 in driving lethality and cell death during -coronavirus infection in the presence or absence of IFN- treatment. Although IFN- treatment substantially increased the lethality of MHV infection in WT mice, mice that were deficient in ZBP1 were similarly susceptible to MHV infection-induced lethality whether or not they received IFN- treatment (Fig. 4B). We found that ZBP1 expression was greatly enhanced in the lungs of MHV-infected mice after IFN- treatment compared with the lungs of untreated infected mice (Fig. 4C). Moreover, we observed decreased production of inflammasome-dependent and inflammasome-independent proinflammatory cytokines in the BALF of MHV-infected Zbp1 −/− mice compared with those of WT mice after IFN- treatment ( fig. S7). The regulatory role of ZBP1 in vivo was further confirmed by in vitro findings that the robust cell death observed in MHV-infected BMDMs in response to IFN treatment was reduced in ZBP1-deficient cells (Fig. 4, D and E). Sensing of viral or endogenous Z-RNA by the Z domains of ZBP1 triggers NLRP3 inflammasome activation, inflammatory cell death, and perinatal lethality in mice, indicating that the Z domains are crucial to regulate the immune responses driven by ZBP1 (32)(33)(34). Therefore, we investigated the role of the Z2 domain of ZBP1 in driving the cell death induced by MHV with IFN- treatment. Similar to Zbp1 −/− BMDMs, MHV-infected cells lacking the Z2 domain of ZBP1 showed reduced cell death compared with those of WT BMDMs after treatment with IFN- (Fig. 4F). However, IFN- treatment did not affect the ZBP1-dependent cell death in BMDMs during IAV infection (Fig. 4G), likely because IAV infection naturally induces more rapid IFN production than -coronavirus infection, making the additional IFN ineffective. Although Zbp1 −/− and Zbp1 Z2 mice showed increased susceptibility to IAV infection with respect to WT mice, IFN- treatment did not change the outcome in WT mice during IAV infection (Fig. 4, H and I).
Together, our results show that ZBP1 drives cell death, cytokine storm, and lethality in -coronavirus-infected mice in response to IFN- treatment.
ZBP1 triggers inflammatory cell death, PANoptosis, and cytokine storm during -coronavirus infection ZBP1 has been shown to trigger cell death by activating PANoptosis in cells during infection with viruses and fungi (31,32,53,63,66). Combined with our observation that PANoptosis was occurring in the lungs of MHV-infected mice in response to IFN- treatment (Fig. 3, A to C), this suggests that ZBP1mediated PANoptosis may contribute to lung pathology, cytokine storms, and lethality in MHV-infected mice during IFN- treatment. In support of this hypothesis, we found that the robust activation of PANoptotic markers was reduced in the lungs of MHV-infected Zbp1 −/− mice compared with those of WT mice during IFN- treatment (Fig. 5, A to C). In addition, consistent with the reduced cell death observed in Zbp1 −/− and Zbp1 Z2 BMDMs (Fig. 4, D to F), these cells also showed impaired activation of PANoptotic effectors in response to IFN- treatment (Fig. 5, D to F). Similarly, ZBP1-silenced THP-1 cells infected with SARS-CoV-2 showed reduced activation of PANoptotic effectors when treated with IFN- compared with control-infected and IFN--treated cells ( fig. S8, A and B). Together, these results indicate that ZBP1, and specifically its Z2 domain, is required for -coronavirus-induced inflammatory cell death in response to IFN- treatment. ZBP1 activation leads to its interaction with RIPK3 and recruitment of caspase-8 and caspase-6 to form a cell death signaling scaffold, termed the ZBP1-PANoptosome, that drives NLRP3 inflammasome activation and cell death (31,32,63,70). We therefore assessed whether RIPK3 and caspase-8 had any role in the -coronavirus-induced cell death pathway during IFN- treatment. Similar to Zbp1 −/− or Zbp1 Z2 BMDMs, cells lacking RIPK3 showed reduced cell death and activation of PANoptotic molecules compared with those of WT BMDMs during MHV infection in response to IFN- treatment ( fig. S8, C to G). We did not observe significant differences in the extent of cell death and activation of PANoptosis between Ripk3 −/− and Ripk3 −/-Casp8 −/− BMDMs ( fig. S8, C to G), indicating that the phenotype observed in Ripk3 −/-Casp8 −/− BMDMs is largely attributed to the RIPK3 deficiency because caspase-8 is upstream of RIPK3. Together, these findings suggest that ZBP1-mediated activation of RIPK3-dependent PANoptosis is responsible for the cell death induced by IFN- treatment during -coronavirus infection.

DISCUSSION
Vaccines remain the best approach to end the COVID-19 pandemic (71). However, the efficacy of the vaccines is challenged by the emergence of new SARS-CoV-2 variants (72). Use of drugs-including the antivirals remdesivir, the nirmatrelvir and ritonavir combination (Paxlovid), and molnupiravir, as well as anti-inflammatory or immunomodulatory agents such as dexamethasone, tocilizumab, and baricitinib-has been shown to improve outcomes in patients with COVID-19 (73)(74)(75)(76)(77). However, these agents typically provide limited benefits to a subset of patients, and global availability is inadequate; the morbidity and mortality caused by COVID-19 remain high (1). Given the potency of IFNs in controlling viral replication, several clinical trials using IFNs alone or in combination with other antiviral drugs have been initiated to treat patients infected with SARS-CoV-2. Coronaviruses, including SARS-CoV-2 and MHV, induce a delayed IFN response (18,19), because of which the viruses likely undergo active replication to establish infection. However, preclinical data have shown that this delayed response can be overcome by treatment with exogenous IFNs or IFN-inducing agonists (19). The observed benefits of early type I IFN responses induced by some agonists in preclinical models suggest that IFN-based therapies may be effective for the prevention and treatment of COVID-19 (19-23, 78, 79). However, the use of IFN therapy, which acts by decreasing viral replication, has shown mixed responses in clinical trials (24,25,(80)(81)(82)(83)(84)(85). A recent double-blinded and placebo-controlled clinical trial (ACTT-3) showed no significant difference in patient outcomes with the combination of IFN--1a plus remdesivir compared with remdesivir alone, suggesting no clinical benefits of IFN therapy in patients with COVID-19 (24). Moreover, patients who required supplemental oxygen or noninvasive ventilation tended to develop adverse outcomes, particularly worsening respiratory function, in response to IFN--1a administration in this trial (24). It is possible that IFN treatment may increase the inflammatory response, leading to more severe respiratory disease in these patients. In vivo, the lungs of MHV-infected mice had increased immune cell infiltration, inflammatory cytokine levels, and PANoptosis in response to IFN treatment compared with PBS treatment, and the IFN-treated mice were more likely to succumb to infection. Similar survival was observed in SARS-CoV-2-infected mice, where IFN treatment significantly increased mortality. In vitro, MHV-infected BMDMs or SARS-CoV-2-infected monocytes showed increased PANoptosis after IFN treatment compared with mock treatment. BMDMs or THP-1 cells were treated with IFN- after 5 hours of MHV or SARS-CoV-2 infection. However, the effect of IFN- treatment at different time points after -coronavirus infection has not been investigated in this study, and the outcomes may change depending on when IFN- treatment is given. Moreover, the dose of IFN- is another factor that may influence the outcome. In addition, because epithelial cells are the primary cells that are infected by -coronaviruses, future work to determine the activation of PANoptosis in cell types other than myeloid cells, such as epithelial and endothelial cells, would provide additional information about the cell types that respond to IFNs and are responsible for driving detrimental effects during -coronavirus infections. In addition, PANoptosis in macrophages activates both GSDMD and GSDME pore-forming molecules during SARS-CoV-2 infection, but it is possible that these molecules may have PANoptosis-independent functions in certain cell types. For instance, high expression of GSDMD has been associated with neutrophil extracellular trap (NET) structures in the lungs of patients with COVID-19. Pharmacological inhibition of GSDMD through treatment with the inhibitor disulfiram reduces NET release and organ damage in a mouse model of SARS-CoV-2 infection, suggesting that GSDMD-dependent NETosis plays a critical role in COVID-19 immunopathology (86). These functions require further study.
IFNs, particularly IFN-, synergize with TNF to induce inflammatory cell death, leading to organ and tissue damage in cytokine storm-associated mouse models (4,45). Moreover, IFNs can damage the lung epithelial barrier and impair lung epithelial regeneration during viral infection (87). In addition, IFN treatment induces sustained expression of ISGs (19), which include both antiviral and inflammatory cell death-driving molecules, such as ZBP1. Not all patients infected with SARS-CoV-2 showed increased IFN responses and ZBP1 expression. However, we observed that CI patients with more severe COVID-19 showed increased IFN responses and ZBP1 expression compared with noncritical patients with COVID-19, suggesting that IFNs and ZBP1 are associated with disease progression and mortality. The loss of ZBP1 substantially protected the -coronavirusinfected mice from lung damage, cytokine storm, and lethality during IFN treatment, highlighting the detrimental effect of ZBP1 in disease progression. The use of IFN therapy in patients infected with SARS-CoV-2 would therefore likely mimic the increased and sustained ZBP1 expression observed in CI patients with COVID-19, which could result in multiorgan damage, cytokine storm, and mortality in patients. In contrast to its role in response to IFN treatment during -coronavirus infection, we observed that ZBP1 was critical to the host defense against IAV infection irrespective of IFN treatment. This is likely due to the ability of IAV to induce rapid IFN responses and ZBP1 expression, which facilitates viral clearance via molecular mechanisms that include ZBP1-dependent inflammatory cell death (31,88,89). However, in addition to clearing the virus, dysregulated inflammatory cell death leads to cytokine storm and organ damage (4,45). MHV-infected mice succumbed to infection upon IFN treatment despite the fact that IFN should inhibit viral replication, suggesting that inflammatory cell death, rather than the virus, is responsible for driving pathological effects. However, these consequential pathological effects of IFN therapy during -coronavirus infection can be blocked by inhibiting ZBP1 activity. Therefore, inhibiting ZBP1 activity may improve the beneficial effects of IFN- therapy during -coronavirus infection. Moreover, the associated side effects and inconvenience associated with exogenous IFN administration can be overcome by administering agonists that engage DNA sensing pathways, such as cGAS and STING, or RNA sensing pathways, such as MDA5, RIG-I, and MAVS, to produce IFNs endogenously (19)(20)(21)(22). The sustained and late IFN production induced by the cGAS-STING pathway also contributes to collateral host responses and tissue damage leading to lethality in SARS-CoV-2-infected mice (90), which could be due to ZBP1-dependent inflammatory cell death, and these factors would need to be considered when designing therapeutic strategies.
Overall, our results suggest that IFN-based treatments for COVID-19 face therapeutic challenges due to ZBP1-mediated inflammatory cell death that contributes to cytokine storm, tissue damage, and, ultimately, lethality. Combination strategies that block ZBP1 while also providing IFN treatment may therefore be beneficial for patient outcomes and should be pursued for further research and development. This improved understanding of the mechanistic basis of innate immune sensing, IFN-mediated pathology, and inflammatory cell death during SARS-CoV-2 infection informs COVID-19 therapeutic development and provides fundamental molecular details that should be evaluated across interferonopathies.

Study design
The aim of this study was to investigate the molecular mechanism underlying the failure of IFN therapy during SARS-CoV-2 infection. We used MHV, a prototypical virus of the -coronavirus genus that mimics many of the key aspects of human -coronavirus biology, as well as SARS-CoV-2 to infect WT mice in the presence and absence of IFN- and determined pathology and lethality. We also performed a CRISPR screen to identify key molecules involved in cell death in response to MHV infection. Furthermore, we investigated the effect of IFN treatment on mouse BMDMs infected with MHV and human THP-1 cells infected with SARS-CoV-2 in inducing inflammatory cell death, PANoptosis. Sample sizes used in each experiment are detailed in the figure legends. (92), and Ripk3 −/-Casp8 −/− (93) mice have been described previously. All mice were bred and maintained in a specific pathogen-free facility at the Animal Resources Center at St. Jude Children's Research Hospital and were backcrossed to the C57BL/6 background (J substrain) for at least 10 generations. Both male and female mice were used in this study; 6-to 12-week-old age-and sex-matched mice were used for in vivo and in vitro studies and were randomly assigned to different treatment groups/stimulations. Cohoused animals were used for in vivo MHV infections. Mice were maintained with a 12-hour light/ dark cycle and were fed standard chow. Animal studies were conducted under protocols approved by the St. Jude Children's Research Hospital committee on the Use and Care of Animals or with the approval of the Institutional Animal Care and Use Committee of University of Tennessee Health Science Center (UTHSC IACUC; protocol #20-0132).

Virus culture
The MHV (A59 strain from R. Channappanavar) (94) was amplified in 17Cl-1 cells as previously described (95). Briefly, 17Cl-1 cells were inoculated with MHV-A59 at a multiplicity of infection (MOI) of 0.1. At 48 hours after infection, the whole flask with media and cells was frozen at −80°C and then thawed at 37°C. After repeating the freezing-thawing cycle twice, the supernatant was collected and centrifuged at 2000g for 10 min to remove the cell debris. Then, the virus was purified by ultracentrifugation at 30,000 rpm for 1 hour, after which the pellets were resuspended with fresh medium. The virus titer was measured by plaque assay in 17Cl-1 cells. SARS-CoV-2 for in vitro analyses (strain USA-WA1/2020 isolate; BEI Resources, catalog no. NR-52281) was obtained from BEI Resources (www. beiresources.org) and amplified in Vero E6 cells. SARS-CoV-2 for in vivo analyses (isolate hCoV-19/USA/MD-HP01542/2021; lineage B.1.351, NR-55282) was obtained from BEI Resources. SARS-CoV-2 B.1.351 (Beta) seed stocks were amplified in Vero E6/TMPRSS2 cells in infection media comprising minimum essential medium with Earle's salts and l-glutamine with 2% FBS and 1% penicillinstreptomycin. Amplified stock virus was stored at −80°C until used. The IAV (A/Puerto Rico/8/34, H1N1 [PR8]) was prepared as previously described (63) and propagated from 11-day-old embryonated chicken eggs by allantoic inoculation. IAV titer was measured by plaque assay in Madin-Darby canine kidney (MDCK) cells.

In vivo infection
Six-to 12-week-old age-and sex-matched WT, Zbp1 −/− , and Zbp1 ∆Z2/∆Z2 mice were used for infections. Mice were anesthetized with Avertin (250 mg/kg) and then infected intranasally. For determination of the LD 50 dose of MHV, mice were infected with 50 l of PBS containing ~10 4 , ~10 5 , or ~ 10 6 plaque-forming units (PFU) of MHV. For subsequent infections, mice were infected with 50 l of PBS containing ~10 5 PFU of MHV or ~100 PFU of IAV. The MHV-infected mice were injected intraperitoneally with 200 l of PBS or IFN- (1 g/mouse) diluted in PBS on days 1 and 3 after infection. The mice were monitored over a period of 14 days for survival. For collection of BALF or lungs, mice were dosed with PBS or IFN- as described above on day 1, and samples were collected at day 3 after infection.
SARS-CoV-2 infections were conducted in accordance with the approval of the UTHSC IACUC (protocol #20-0132). The SARS-CoV-2-related experiments were performed in animal biosafety level 3 conditions within the UTHSC Regional Biocontainment Laboratory (RBL) according to UTHSC RBL Standard Operating Procedures and safety manuals. Each mouse was intranasally infected with SARS-CoV-2 B.1.351 variant at 1 × 10 5 PFU in a total volume of 100 l (50 l per nare) at day 0. The SARS-CoV-2-infected mice were injected intraperitoneally with 150 l of PBS or IFN- (1 g per mouse) diluted in PBS on days 2 and 4 after infection. Mice were weighed and monitored daily in the morning. The mice health status was evaluated once per day using the clinical scoring system approved by UTHSC IACUC, then twice daily upon clinical manifestation. The mice were monitored over a period of 14 days for survival.

Histopathology
For murine samples, lungs were fixed in 10% formalin and then processed and embedded in paraffin by standard procedures. Sections (5 m) were stained with hematoxylin and eosin (H&E) and examined by a pathologist blinded to the experimental groups. For immunohistochemistry, formalin-fixed paraffin-embedded lungs were cut into 4-m sections. F4/80 [Cell Signaling Technology (CST), D2SR9] and Ly-6B.2 (Novus, NBP2-13077) staining was performed according to the manufacturer's instructions. TUNEL staining was performed using the Dead-End Kit (Promega, PRG7130) according to the manufacturer's instructions.
For human lung samples, deidentified autopsy lung samples were provided by the Asian Institute of Gastroenterology as paraffinembedded tissue blocks. Sectioning (4 m) and immunohistochemistry were performed by HistoWiz with H&E, the rabbit polyclonal anti-SARS-CoV-2 nucleocapsid (N) protein antibody (GeneTex; GTX635686), and TUNEL (Promega) staining following standard protocols.

Immunoblot analysis
Immunoblotting was performed as described previously (96). Briefly, for caspase analysis, BMDMs were lysed along with the supernatant using 50 l of caspase lysis buffer (1× protease inhibitors, 1× phosphatase inhibitors, 10% NP-40, and 25 mM dithiothreitol) followed by the addition of 100 l of 4× SDS loading buffer. For signaling analysis, the BMDM supernatants were removed at the indicated time points, and cells were washed once with PBS, after which cells were lysed with radioimmunoprecipitation assay (RIPA) buffer. Proteins from lung tissues were extracted using RIPA buffer supplemented with protease and phosphatase inhibitors (Roche), and 30 g per sample was loaded on the gel. Proteins were separated by electrophoresis through 8 to 12% polyacrylamide gels. After electrophoretic transfer of proteins onto polyvinylidene difluoride membranes (Millipore, IPVH00010), nonspecific binding was blocked by incubation with 5% skim milk; then, membranes were incubated with the following primary antibodies: anti-caspase-1 (1:1000; AdipoGen, AG-20B-0042), anti-caspase-3 (1:1000; CST, #9662), anti-

Real-time cell death analysis
Real-time cell death assays were performed using an IncuCyte S3 or IncuCyte SX5 imaging system (Sartorius). BMDMs were seeded in 12-well plates (10 6 cells per well) and stimulated. After infection, 100 nM Sytox Green (Thermo Fisher Scientific, S7020) was added. The images were acquired every 1 hour at 37°C and 5% CO 2 . The resulting images were analyzed using the software package supplied with the IncuCyte imager, which counts the number of Sytox Greenpositive BMDM nuclei (Sytox + BMDM nuclei) present in each image.

Microarray and RNA-seq analysis
For analysis of gene expression in patients with COVID-19, publicly available datasets were downloaded from the Gene Expression Omnibus (GEO) database [GSE147507 (18), GSE171430 (97), GSE157103 (98), and GSE171110 (99)]. The DESeq2 v1.32.0 (100) package was used to read the count matrices and remove genes that were not expressed in at least one sample, and the limma v3.48.3 package (101) in R v4.1.1 was used to determine differential expression. Benjamini-Hochberg (102) adjusted P value < 0.05 and |log 2 FC| > 1 were then used to identify the differentially expressed genes. Genes that were differentially expressed in each of the four datasets were then used to form a consensus set of 31 genes consistently expressed differentially and with significance across the four COVID-19-related datasets. This gene set was used for subsequent analyses.
Unnormalized RNA-seq data were obtained from BioProject [PRJNA638753 (35)] to compare NCI with CI patients with COVID-19. Quality control steps were performed, including normalizing quantiles using the "normalize.quantiles" function from the preprocessCore v1.54.0 package, followed by log 2 transformation for downstream pathway enrichment analysis. Hallmark pathways were downloaded as gene sets from MSigDB (36); MSigDB defines these as follows: hallmark gene sets summarize and represent specific well-defined biological states or processes and display coherent expression and were generated by a computational methodology based on identifying overlaps between gene sets in other MSigDB collections and retaining genes that display coordinate expression (37). Gene set enrichment analysis (GSEA) (103) was performed using the fgsea v1.18.0 package (104) in R to determine the normalized enrichment scores (NES) for each pathway along with the significance of the enrichment.
The average expression profile was also determined as a z score for the 31 genes identified above, and a heatmap was generated using the ComplexHeatmap v2.8.0 package (105).
For analysis of expression in Calu-3 cells infected with SARS-CoV-2 over time, the log 2 FC information at genome-wide scale was downloaded from GEO [GSE157490 (42)], and the mRNA data for Calu-3 cells infected with IAV over time were also downloaded from GEO [GSE76599 (43)]. Quantile normalization and log 2 transformation were performed on the IAV mRNA data to obtain the log 2 FC for each gene by comparing the infection versus mock sample expression profiles. The "fgsea" function was then used to estimate the NES for IFN signaling pathways (IFN-ɑ and IFN- responses) at each available time point for the SARS-CoV-2 and IAV infection, respectively.
For comparison of gene expression in the blood transcriptome between patients with SARS-CoV-2 infection and influenza and sepsis, RNA-seq data were downloaded from GEO [GSE163151 (40)]. Quality control steps were performed, including normalizing quantiles using the normalize.quantile function from preprocessCore, followed by log 2 transformation. Comparisons were made for COVID-19 patient expression profiles versus healthy controls, sepsis versus healthy controls, and influenza versus healthy controls to determine the log 2 FC for all genes. The set of 50 aforementioned hallmark pathways and the log 2 FC information of each gene were used in the fgsea function to obtain the NES for each pathway. The average expression profile was also determined as a z score for the 31 consensus genes identified above, and a heatmap was generated.

Generation of Brie lentiviral library
The Mouse Brie CRISPR KO library was a gift from D. Root and J. Doench (Addgene, #73632 and #73633). The plasmid library was amplified and validated in the Center for Advanced Genome Engineering at St. Jude as described in the Broad GPP protocol, the only exception being the use of Endura DUOs electrocompetent cells. The St. Jude Hartwell Center Genome Sequencing Facility provided all next-generation sequencing. Single-end 100-cycle sequencing was performed on NovaSeq 6000 (Illumina). Validation to check guide RNA (gRNA) presence and representation was performed using calc_auc_v1.1.py (https://github.com/mhegde/) and count_spacers.py (106). Viral particles were produced by the St. Jude Vector Development and Production laboratory. CRISPR KO screens were analyzed using Mageck-Vispr/0.5.7 (107).
CRISPR-CAS9 library screen CAS9-expressing iBMDMs generated from Cas9-GFP knock-in mice (108) were infected with the lentiviral particles at an MOI of 0.3. Two replicates of an adequate number of cells were used as control to obtain a representation (screen depth) of >500 cells for each single-guide RNA (sgRNA) of the library, and a similar number of cells from the same batch of virus preparation were infected with MHV at an MOI of 0.2 for 24 hours. The uninfected control cell population and the surviving cells from the MHV-infected samples were subjected to CRISPR screen enrichment analysis. Total genomic DNA was isolated using NucleoSpin Blood kits (Takara Bio Inc., USA, 740954 and 740950), and the concentrations of the isolated genomic DNA samples were measured using NanoDrop (Thermo Fisher Scientific, USA).
Using the MAGeCK pipeline (109), we robustly estimated the log 2 FC with significance of the essential genes from our CRISPR screen. The genes with positive fold change were imperative for cell death. The top gene hits along with their significance from the CRISPR screen were highlighted using a volcano plot.

Statistical analysis
GraphPad Prism 8.0 software was used for data analysis. Data are presented as means ± SEM. Statistical significance was determined by t tests (two-tailed) or Mann-Whitney-Wilcoxon test (nonparametric) for two groups or one-way or two-way analysis of variance (ANOVA) with multiple comparisons test for more groups. Survival analysis was performed using the log-rank (Mantel-Cox) test. P < 0.05 was considered statistically significant where *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.