SARS-CoV-2 infection induces ZBP1-dependent PANoptosis in bystander cells

Significance

The severity of COVID-19 is closely linked to an excessive inflammatory response. However, the underlying mechanisms by which severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection provokes this intense inflammatory reaction are not fully understood. In this study, we demonstrate that SARS-CoV-2-infected cells secrete 2′3′cGMP-AMP, tumor necrosis factor-alpha (TNF-α), and interferon-beta (IFN-β), which act on bystander cells, triggering Z-DNA binding protein 1 (ZBP1)-dependent PANoptosis. This process contributes to the persistence of the inflammatory response. Additionally, inhibition of the cyclic GMP-AMP synthase pathway or knockout of stimulator of interferon genes reduces PANoptosis and alleviates the pathology associated with SARS-CoV-2 and influenza A virus infections. Our study reveals that PANoptosis in bystander cells during SARS-CoV-2 infection is a key mechanism driving pathological damage.

Abstract

Virus-induced excessive inflammatory response is a key contributor to pathology in respiratory viral infections. However, the underlying mechanisms by which viral infection provokes intense inflammatory reaction and how sustained inflammation leads to tissue damage are not fully understood. Using severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection as an example, our research demonstrates that SARS-CoV-2 infection can induce PANoptosis in bystander cells, contributing to the persistence of inflammatory responses. Specifically, the activation of cyclic GMP-AMP synthase (cGAS)-stimulator of interferon genes (STING) in infected cells leads to the secretion of 2′3′cGMP-AMP, TNF-α, and IFN-β. These molecules activate STING-induced autophagy-mediated ADAR1 degradation, resulting in the accumulation of Z-nucleic acid, which subsequently triggers ZBP1-dependent PANoptosis in bystander cells. Additionally, inhibiting the cGAS pathway or knocking out STING effectively reduces PANoptosis and alleviates the pathology associated with SARS-CoV-2 and influenza A virus infection in mouse models. Overall, our findings reveal the unexpected role of PANoptosis in bystander cells during SARS-CoV-2 infection as a mechanism driving pathological damage and persistent inflammatory responses. Targeting this process may offer a promising strategy to mitigate tissue damage in COVID-19 as well as other viral infections and inflammatory conditions.

The inflammatory responses induced by viral infections cast complex and usually dual effects. It plays an important immunoprotective role in the early stages of viral infection, but excessive or sustained inflammatory responses can lead to tissue damage and aggravation of disease (1). Pathogenic respiratory viral infections including severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and influenza A virus (IAV) cause excessive and sustained inflammatory responses, leading to a “cytokine storm” that leads to high morbidity and mortality (2–4). In addition, long-term inflammation also leads to depletion of the immune system and decline in immune function (5). Inflammation is intricately linked to cell death induced by viral infections. Uncontrolled cell death can amplify inflammatory responses and lead to severe tissue damage (6). Blocking inflammatory cell death could prevent lung injury in severe coronavirus disease 2019 (COVID-19) and influenza. Therefore, understanding the mechanisms by which viral infections trigger uncontrolled inflammatory responses and cell death is essential for developing targeted therapies for severe inflammatory diseases caused by viral infections.

SARS-CoV-2 infection induces inflammatory cell death and notably increases various inflammatory cytokines and can trigger acute respiratory distress syndrome (ARDS) in severe cases (6–8). Research has indicated that the severity of COVID-19 and the risk of mortality are associated more with excessive type I interferon (IFN) and inflammation rather than the viral load itself (6, 9–11). In addition, Reports indicate that over 65 million individuals who have recovered from COVID-19 experience post-acute sequelae of SARS-CoV-2 infection (PASC), a number that is steadily increasing (12). Also known as Long COVID, PASC is characterized by persistent, recurrent, or new symptoms that appear 30 d or more following the initial infection. Factors such as severe symptoms during the acute phase of the disease may heighten the risk of developing PASC in patients (13, 14). Persistent high levels of inflammatory cytokines, such as type I IFN, IL-1β, IL-6, and TNF-α, have been observed in patients with PASC (15, 16).

Overactivation of the immune system in response to pathogenic infection can lead to a sharp increase in the level of proinflammatory cytokines in the circulatory system, and cause systemic inflammation and multiple organ failure (17). Activation of the cGAS–STING pathway triggered by SARS-CoV-2 and IAV infection is a primary driver of a delayed pathological type I IFN response (18–21). Furthermore, the inflammatory response characteristic of COVID-19 patients persists even after the virus has been cleared (22). It has been documented that SARS-CoV-2 and IAV can suppress and evade the innate immune response in infected cells (23). However, the role of uninfected bystander cells and their potential impact by infected cells remains poorly understood. Given that uninfected bystander cells possess fully functional innate immune machinery capable of sensing and responding to signals from dead or damaged infected cells, we hypothesize that these bystander cells, when affected by infected cells, may mount a delayed and uncontrolled innate immune and inflammatory response. This response could lead to subsequent pathological damage (24).

In this study, we demonstrate that cGAMP, produced by SARS-CoV-2-infected cells, is transported to bystander cells. In these cells, activated STING subsequently induces autophagy, leading to the degradation of adenosine deaminase acting on RNA 1 (ADAR1). Concurrently, TNF-α and IFN-β, secreted by the infected cells, promote the accumulation of Z-nucleic acid (Z-NA) in bystander cells in the absence of ADAR1, triggering Z-DNA binding protein 1 (ZBP1)-dependent PANoptosis. Furthermore, we confirm that inhibiting cGAS activation or knocking out STING can significantly reduce PANoptosis (particularly bystander cell PANoptosis) and lung damage in mouse models of SARS-CoV-2 or IAV infection. Overall, our findings reveal mechanisms of bystander cell death and excessive inflammation in SARS-CoV-2 infection, offering promising avenues for treatment strategies in infectious and inflammatory diseases.

Results

SARS-CoV-2 Infection Triggers PANoptosis of Bystander Cells.

PANoptosis has been identified as playing a significant role in coronavirus infections (3). To investigate the status of PANoptosis caused by SARS-CoV-2, we utilized established macrophage and alveolar epithelial cell models, specifically ACE2-THP-1 and Calu-3 cells (4, 25). In our experiments, the cells infected with SARS-CoV-2 exhibited robust cleavage of the pyroptotic effector GSDME and GSDMD, activation of apoptotic effectors caspase-3 and -7, and significant phosphorylation of MLKL, a marker for necroptosis (Fig. 1A and SI Appendix, Fig. S1A). Furthermore, the activation of the cGAS–STING pathway by SARS-CoV-2 infection is a primary driver of the pathological type I IFN response in COVID-19 (18, 20, 21). We observed strong phosphorylation of STING Ser366 in SARS-CoV-2 infected ACE2-THP-1 and Calu-3 cells (Fig. 1B and SI Appendix, Fig. S1B), indicating that the activation of cGAS–STING is associated with the occurrence of PANoptosis.

Fig. 1.

SARS-CoV-2 infection triggers STING activation and PANoptosis in bystander cells. (A) Immunoblot analysis of pro- (P53) and activated (P34) GSDME, pro- (P55) and activated (P33) GSDMD, pro- (P35) and cleaved (P17) CASP3, pro- (P35) and cleaved (P20) CASP7, phosphorylated RIP3 (p-RIP3) and phosphorylated MLKL (p-MLKL) in ACE2-THP-1 cells infected with SARS-CoV-2 (MOI = 0.5) at indicated time points. (B) Immunoblot analysis of p-STING and p-IRF3 in ACE2-THP-1 cells infected with SARS-CoV-2 (MOI = 0.5) at the indicated time points. (C) ACE2-THP-1 cells and THP-1 cells were cocultured (1:1) and then infected with SARS-CoV-2 (MOI = 0.5). After 48 h of infection, the cells were stained with cleaved Caspase 3 (cle-Casp3; Green), SARS-CoV-2 NP (Red), and DAPI (Blue). (D–G) THP-1 cells were cultured for 1, 2, or 2.5 d with heat-inactivated culture supernatants of ACE2-THP-1 cells infected with (CoV-2-sup) or without (Mock-sup) SARS-CoV-2 for 3 d and then analyzed. Experimental design (D). The percent analysis and representative images of cell death in THP-1 cells treated with CoV-2-sup for 2.5 d (E). Immunoblot analysis of THP-1 cells treated with CoV-2-sup for 2.5 d (F) or indicated time points (G) using the indicated antibody. Asterisks denote a nonspecific band. Data are mean ± SD (n = 6 in E per group). Two-tailed unpaired t test with Welch’s correction (E). ****P < 0.0001. [Scale bars, 10 μm (C), 50 μm (C), 100 μm (E).] Data are from at least three independent experiments.

After coculturing ACE2-THP-1 cells with THP-1 cells and infecting them with SARS-CoV-2, we observed cells double-positive for SARS-CoV-2 nucleocapsid protein (NP) and cleaved caspase-3 (cle-Casp3), as well as NP-negative but cle-Casp3-positive cells (Fig. 1C). This indicates that bystander cells undergo cell death during SARS-CoV-2 infection. To investigate whether bystander cell death is influenced by SARS-CoV-2 infected cells and contribute to prolonged inflammation, we utilized THP-1 and ZBP1-HT-29 cells, a model system for studying cell death. These cells were incubated with culture supernatants derived from SARS-CoV-2 infected ACE2-THP-1 cells (CoV-2-sup) 3 d postinfection (Fig. 1D and SI Appendix, Fig. S1C). The SARS-CoV-2 in the culture supernatant was inactivated (SI Appendix, Fig. S1D). The results clearly demonstrated that heat-inactivated media from SARS-CoV-2-infected cells induced death in bystander cells (Fig. 1E and SI Appendix, Fig. S1E), characterized by cleavage of GSDME and GSDMD, activation of caspase-3/-7, and phosphorylation of MLKL, along with phosphorylation of STING (Fig. 1 F and G and SI Appendix, Fig. S1 F and G). Thus, our findings indicate that SARS-CoV-2 infection triggers PANoptosis and STING activation in bystander cells, potentially contributing to the virus’s capacity to induce prolonged inflammation.

Z-nucleic Acid Induces ZBP1-Dependent PANoptosis in SARS-CoV-2 Bystander Cells.

PANoptosis is triggered by specific sensors that recognize pathogen-associated molecular patterns (PAMPs), damage-associated molecular patterns (DAMPs), or cytokines (26–30). To date, ZBP1, AIM2, and NLRP12 have been identified as potential sensors for PANoptosis (3, 26, 29–31). We respectively knocked down ZBP1, AIM2, and NLRP12, and found that knocking down ZBP1 significantly inhibited cell death induced by CoV-2-sup (Fig. 2A and SI Appendix, Fig. S2A). Furthermore, ZBP1 expression has been suggested to correlate with COVID-19 pathology (3). To determine which domain of ZBP1 mediates PANoptosis in bystander cells, we constructed THP-1 and HT-29 cells that stably express HA-tagged wild-type (WT) ZBP1 (HA-ZBP1), ZBP1 lacking the Zα1 domain (HA-ZBP1 ΔZα1), ZBP1 lacking both Zα1 and Zα2 domains (HA-ZBP1 ΔZα1 + ΔZα2), and ZBP1 carrying two key amino acid substitutions at the Z-nucleic acid (Z-NA) binding site (HA-ZBP1 Mut Zα2) (SI Appendix, Fig. S2 B and C). After coculture with CoV-2-sup, the cell death rate of HA-ZBP1 ΔZα1 + ΔZα2 and HA-ZBP1 Mut Zα2 was significantly lower than that of HA-ZBP1 and HA-ZBP1 ΔZα1 (Fig. 2B and SI Appendix, Fig. S2D), suggesting that the Z-NA binding ability of ZBP1 is crucial for mediating PANoptosis in bystander cells. RIP3 and Caspase-8 is involved in the assembly of PANoptosomes triggered by ZBP1 (2, 26, 32–34). Coimmunoprecipitation assays revealed robust interactions between ZBP1, RIPK3, and Caspase-8 in CoV-2-sup-treated cells, with immunofluorescence analysis further demonstrating their colocalization (Fig. 2 C and D). These findings indicate that ZBP1, RIP3, and Caspase-8 are assembled into PANoptosomes, thereby mediating PANoptosis in SARS-CoV-2 bystander cells.

Fig. 2.

Z-NA triggers ZBP1-dependent PANoptosis in SARS-CoV-2 bystander cells. (A) THP-1 cells expressing control shRNA (sh-Ctrl), shRNA targeting ZBP1 (sh-ZBP1), sh-AIM2, or sh-NLRP12 with the CoV-2-sup for 2.5 d were analyzed for cell death rate by Sytox Green. (B) HA-tagged WT ZBP1(HA-ZBP1), ZBP1 lacking its Zα1 domain (HA-ZBP1 ΔZα1), ZBP1 lacking its Zα1 and Zα2 domain (HA-ZBP1 ΔZα1 + ΔZα2), and ZBP1 of carrying two key amino acid substitutions at Z-NA binding site (HA-ZBP1 Mut Zα2) THP-1 (ZBP1-KO) cells were treated for 2.5 d with the CoV-2-sup. The cell death percentage analysis was performed using Sytox Green. (C) HA-ZBP1 HT-29 cells were cultured for 3 d with or without the CoV-2-sup and then the cell lysates were subjected to co-IP with anti-Casp8 or anti-HA antibodies and were immunoblotted with anti-HA, anti-Casp8, or anti-RIP3 antibodies, respectively. (D) THP-1 cells were transfected with ZBP1-mcherry (Red), Casp8-GFP (Green), and RIP3-BFP (Blue) expression plasmid. After 12 h, the cells were treated with MOCK-sup or CoV-2-sup for 56 h before observation under a confocal microscope. The Right panel shows the ZBP1-mCherry (Red), Casp8-GFP (Green), and RIP3-BFP (Blue) pixel intensity. (E and F) Time course of Z-NA accrual in THP-1 treated with the CoV-2-sup. The cell was detected the presence of Z-NA using against Z-NA antibodies (clone Z22) (E). After the CoV-2-sup treatment, THP-1 were exposed to the indicated nucleases and stained for Z-NA using against Z-NA antibodies (clone Z22) (F). Fluorescence intensity (arbitrary units, a.u.) of Z-NA signals in E (Bottom) and F (Right), respectively. Data are mean ± SD (n = 6 in A and B; n = 10 in E and F per group). One-way ANOVA test (A, B, E, and F). ****P < 0.0001. [Scale bars, 10 μm (D–F).] Data are from at least three independent experiments.

To further confirm that the accumulation of Z-NA in SARS-CoV-2 bystander cells is key to inducing PANoptosis, we treated THP-1 and ZBP1-expressing HT-29 cells with CoV-2-sup and detected the presence of Z-NA using antibodies (clone Z22). The results showed that Z-NA signals appeared 2 d posttreatment and gradually increased (Fig. 2E and SI Appendix, Fig. S2E). Further analysis showed that RNase A treatment abolished most Z-NA signals, while complete Z-NA elimination required combined treatment with RNase A and DNase I (Fig. 2F and SI Appendix, Fig. S2F). These results indicate that the accumulated Z-NA primarily consists of Z-RNA, with a minor Z-DNA component. Collectively, these data demonstrate that the accumulation of Z-NA in SARS-CoV-2 bystander cells triggers ZBP1-dependent PANoptosis.

cGAMP Secreted From Infected Cells Activates STING in Bystander Cells to Induce PANoptosis.

To elucidate the relationship between cGAS activation and cell death during SARS-CoV-2 infection, ACE2-THP-1 and Calu-3 cells were infected with SARS-CoV-2 after transfection with cGAS-targeting siRNA (sicGAS) or treatment with G140 (a human-cGAS-specific small-molecule inhibitor) (SI Appendix, Fig. S3C). Compared to the uninfected group, SARS-CoV-2 infection induced apparent PANoptosis, which was strongly inhibited by sicGAS and G140 (SI Appendix, Fig. S3A). Interestingly, the culture supernatant from SARS-CoV-2-infected cells treated with sicGAS or G140 lost the ability to induce PANoptosis and STING activation in bystander cells (SI Appendix, Fig. S3B). This suggests that CoV-2-sup contains a substance dependent on cGAS activation to trigger PANoptosis and STING activation in bystander cells. To further determine the role of cGAS/STING in bystander cell PANoptosis, cGAS or STING knockout (KO) cells were analyzed (SI Appendix, Fig. S3D). Interestingly, STING-deficient cells were completely protected from PANoptosis upon treatment with CoV-2-sup, whereas cGAS deficiency only slightly attenuated PANoptosis (Fig. 3A and SI Appendix, Figs. S3E and S4A). Furthermore, CoV-2-sup promoted STING and TBK1 phosphorylation in THP-1 and ZBP1-expressing HT-29 cells in a STING-dependent but cGAS-independent manner (Fig. 3B and SI Appendix, Figs. S3F and S4B), indicating that CoV-2-sup might directly activate STING in bystander cells.

Fig. 3.

cGAMP secreted from SARS-CoV-2 infected cells activate STING in bystander cells to induce PANoptosis. (A and B) Ctrl sg, cGAS sg1, or STING sg1 THP-1 cells were cultured for 2.5 d (A) or 1 d (B) with CoV-2-sup. Then, the cell lysates were analyzed via immunoblotting using the indicated antibodies. (C) ACE2-THP-1 and Calu-3 were infected with SARS-CoV-2 (MOI = 0.5) for 3 d. The cGAMP in the culture supernatant of cells was quantified by ELISA. (D) Ctrl KO, SLC19A1 KO, ABCC1 KO, LRRC8A KO, and SLC46A2 KO THP-1 were cultured for 2.5 d with CoV-2-sup. The cell death percentage analysis was performed using Sytox Green. (E and F) Ctrl sg, SLC19A1 sg1, ABCC1 sg1, LRRC8A sg1, and SLC46A2 sg1 THP-1 cells were cultured for 2.5 (E) or 1 (F) d with CoV-2-sup. The cell lysates were analyzed via immunoblotting using the indicated antibodies. Asterisks denote a nonspecific band. Data are mean ± SD (n = 5 in C; n = 6 in D per group). One-way ANOVA test (C and D). ***P < 0.001; ****P < 0.0001. Data are from at least three independent experiments.

It has been reported that cGAMP produced in virus-infected cells can be transmitted to bystander cells through specific ion channels, subsequently inducing STING activation (35, 36). In this study, cGAMP was indeed detected in the supernatants of SARS-CoV-2-infected ACE2-THP-1 and Calu-3 cells by ELISA (Fig. 3C). The transporters SLC19A1, ABCC1, LRRC8A, and SLC46A2 have all been linked to the transmembrane transport of cGAMP (36–39). To confirm the role of cGAMP import in inducing PANoptosis in bystander cells, KO cell lines for these transporters were generated and analyzed (SI Appendix, Fig. S3I). The cell death rates of SLC19A1 KO and LRRC8A KO cells were significantly lower than those of Ctrl KO cells after being cultured with CoV-2-sup (Fig. 3D and SI Appendix, Fig. S4C). Compared to Ctrl KO, the phosphorylation of STING, IRF3, and MLKL, as well as the cleavage of GSDME, GSDMD, and caspase-3/−7, decreased only in SLC19A1 KO or LRRC8A KO cells upon treatment with CoV-2-sup (Fig. 3 E and F and SI Appendix, Figs. S3 G and H and S4 D and E). SLC19A1 appears to play a more prominent role in THP-1 cells, whereas LRRC8A is more critical in ZBP1-expressing HT-29 cells (Fig. 3 E and F and SI Appendix, Figs. S3 G and H and S4 D and E). Recent studies have demonstrated that extracellular vesicle (EV)-mediated transfer of gasdermin pores drives bystander cell death (40). To determine whether EV-transported gasdermin pores contribute to SARS-CoV-2-induced bystander cell death, we depleted EVs from CoV-2-sup via ultracentrifugation. The results showed that both EV-depleted and whole supernatants induced comparable levels of bystander cell death (SI Appendix, Fig. S3J). Collectively, these data suggest that cGAMP secreted by SARS-CoV-2-infected cells is transferred to bystander cells through SLC19A1 and LRRC8A, subsequently activating STING to trigger PANoptosis of bystander cells.

STING-Induced Autophagic Degradation of ADAR1 Is Essential for Bystander cell PANoptosis.

Now that STING activation has been demonstrated to be critical for bystander cell PANoptosis, we sought to determine the specific mechanism by which STING mediates this process. Two of the most well-characterized functions of STING are 1) initiating a downstream type I IFN response and 2) inducing autophagy (41–43). To validate the first possibility, we utilized two well-defined STING mutants: S366A, which abolishes the recruitment of IRF3, and R238A, which disrupts cGAMP binding (41, 44). Reconstitution of WT STING restored CoV-2-sup-induced PANoptosis in STING KO THP-1 and ZBP1-expressing HT-29 cells (Fig. 4 A and B and SI Appendix, Fig. S5 A and B). The S366A mutant only slightly attenuated CoV-2-sup-induced PANoptosis (Fig. 4A and SI Appendix, Fig. S5A), suggesting that the downstream type I IFN response contributes to PANoptosis, but is not essential for its occurrence. However, the R238A mutant completely abolished CoV-2-sup-induced PANoptosis (Fig. 4A and SI Appendix, Fig. S5A), indicating that STING’s binding activity to cGAMP was crucial.

Fig. 4.

STING-induced autophagic degradation of ADAR1 is essential for bystander cell PANoptosis. (A and B) THP-1 (STING KO) cells that stably express WT or mutant STING(S366A) or STING(R238A) were cultured for 2.5 (A) or 1 (B) d with or without CoV-2-sup and then the cell lysates were analyzed via immunoblotting using the indicated antibodies. (C) ZBP1 HT-29 (Ctrl KO) and ZBP1 HT-29 (STING KO) cells were cultured for 2 d with or without CoV-2-sup and then the cell lysates were subjected to co-IP with anti-ADAR1 antibodies and were immunoblotted with anti-ADAR1, anti-STING, or anti-LC3 antibodies, respectively. (D) ZBP1 HT-29 cells were cultured for 3 d with or without CoV-2-sup and the indicated inhibitors, namely, MG132 (10 μM), CQ (10 μM), or Z-VAD-FMK (20 μM). The cell lysates were analyzed via immunoblotting using the indicated antibodies. (E) Ctrl KO, STING KO (Ctrl knockdown, Ctrl KD), and STING KO (ADAR1 KD) ZBP1 HT-29 cells were cultured for 3 d with Mock-sup or CoV-2-sup and then the cell lysates were analyzed via immunoblotting using the indicated antibodies. Asterisks denote a nonspecific band. Data are from at least three independent experiments.

ADAR1 inhibits Z-NA accumulation and negatively regulates ZBP1-induced cell death (45, 46). Our data showed that the reduction of STING and ADAR1 occurred in CoV-2-sup-treated cells (Fig. 1 F and G and SI Appendix, Fig. S1 F and G). However, reexpression of the R238A mutant, but not WT STING or the S366A mutant, prevented the reduction of STING and ADAR1 (Fig. 4A and SI Appendix, Fig. S5A). Additionally, CoV-2-sup stimulated the conversion of LC3 into its lipidated form (LC3-II) in cells reconstituted with WT STING or the S366A mutant, but not with the R238A mutant (Fig. 4B and SI Appendix, Fig. S5B). This evidence suggests that STING-induced autophagy might be the key mechanism underlying bystander cell PANoptosis. We further confirmed that ADAR1 interacted with STING and LC3 in Ctrl KO cells after being cultured with CoV-2-sup, but not in STING KO cells (Fig. 4C). Immunoblotting results revealed that STING-mediated degradation of ADAR1 was completely inhibited by the autophagy inhibitor chloroquine (CQ), but not by the proteasome inhibitor MG132 or the apoptotic inhibitor Z-VAD-FMK (Fig. 4D), confirming that ADAR1 degradation was mediated by STING-induced autophagy. Furthermore, effective ADAR1 knockdown in STING KO cells (SI Appendix, Fig. S5C) restored CoV-2-sup-induced PANoptosis (Fig. 4E), reinforcing that the autophagic degradation of ADAR1 induced by STING is an important switch for bystander cell PANoptosis.

SARS-CoV-2 Promotes Bystander Cell PANoptosis Via TNF-α and IFN-β.

Although the degradation of ADAR1 was essential for bystander cell PANoptosis, no increase in mortality or Z-NA accumulation was observed in THP-1 cells that had only undergone ADAR1 knockdown (SI Appendix, Fig. S6 A, B, and H). The previous report demonstrated that, in the absence of cytokine stimulation, macrophages with ADAR1 deletion do not exhibit significant cell death within a short period (34), and Z-NA could only be detected in ADAR1 KO cells after 7 d of culture (46). This led us to speculate that, in addition to cGAMP, which activates STING and induces subsequent ADAR1 degradation, CoV-2-sup might contain other components that promote cell death. Several studies have reported a relationship among the severity of COVID-19, the occurrence of PASC, and the influx of inflammatory cytokines (15, 16, 47, 48). At the same time, inflammatory cytokines have been suggested to trigger inflammatory cell death (3, 6). To identify the proinflammatory cytokines most closely related to the pathogenesis of SARS-CoV-2 infection and the occurrence of PASC, we reanalyzed publicly available circulating cytokine datasets from healthy volunteers and patients with moderate or severe COVID-19 (48). We also reorganized data from two reports on the changes in circulating cytokines in individuals who recovered from SARS-CoV-2 infection either without PASC (no PASC) or with ongoing PASC (15, 16). We observed that IFN-α2, IFN-β, IL-1α, IL-1β, IFN-γ, IL-2, IL-6, IL-8, IL-15, IL-18, IL-33, IFN-λ1, and TNF-α were significantly increased in severe COVID-19 patients or in patients with PASC (Fig. 5 A–C).

Fig. 5.

SARS-CoV-2 promotes bystander cell PANoptosis via TNF-α and IFN-β. (A) Heatmap depicting the levels of pro-inflammatory cytokines in serum of patients with COVID-19 (48). (B and C) Heatmap depicting the levels of IFN (B) (15) and pro-inflammatory cytokines (C) (16) in serum of people who recovered from SARS-CoV-2 infection either without PASC (no PASC), PASC patients for 8 mo after SARS-CoV-2 infection (ongoing PASC) and unexposed healthy people (Unexposed healthy controls, UHC). (D–F) Receptor knockdown: WT and cytokine receptor-knockdown THP-1 cells were cultured with CoV-2-sup for 2.5 d; Inhibitor: THP-1 cells were pretreated for 1 h with indicated cytokine-inhibitors [TNF-α inhibitor (TNFi), 500 μg/mL etanercept; type I IFN neutralizing antibody (IFNi), 1:200; IL-6 inhibitor (IL6i), 50 μg/mL tocilizumab] prior to CoV-2-sup treatment for 2.5 d. Experimental design (D). The cell death percentage analysis was performed using Sytox Green (E). The cell lysates were analyzed by immunoblotting with indicated antibodies (F). Asterisks denote a nonspecific band. Data are mean ± SD (n = 6 in E). One-way ANOVA test (E). ****P < 0.0001. Data are from at least three independent experiments.

To identify the contributors to SARS-CoV-2 bystander cell death, we first excluded cytokines (e.g., IFN-γ, IL-2, IL-33, IFN-λ1, etc.) that are not produced or secreted by THP-1 cells. For the remaining candidate cytokines, we performed multistrategy screening using inhibitors, neutralizing antibodies, and shRNA-mediated receptor knockdown to pinpoint their roles in bystander cell death (Fig. 5D and SI Appendix, Fig. S6 I and J). Among these, SARS-CoV-2-induced bystander cell death was significantly inhibited in the presence of TNF-α inhibitors (TNFi) or type I IFN neutralizing antibody (IFNi) (Fig. 5E). The cleavage of GSDME and GSDMD, activation of caspase-3/-7, and MLKL phosphorylation were significantly attenuated or completely abolished in bystander cells treated with TNFi or IFNi (Fig. 5F). Consistently, in SARS-CoV-2-infected ACE2-THP-1 cells treated with TNFi or IFNi, PANoptosis was markedly attenuated (SI Appendix, Fig. S6D). Continuous increases in TNF-α and IFN-β was detected in the supernatants of ACE2-THP-1 cells infected with SARS-CoV-2 (SI Appendix, Fig. S6C). It has also been confirmed that the heat-inactivated CoV-2-sup contains active IFN-I and TNF-α (SI Appendix, Fig. S6 E–G). Additionally, we observed that the interaction between ZBP1, Caspase-8, and RIP3 decreased significantly after TNFi treatment and completely disappeared after IFNi treatment (SI Appendix, Fig. S6K). These results confirmed that TNF-α and IFN-β in CoV-2-sup can promote PANoptosis, with IFN-β playing a more crucial role.

To further investigate whether STING activation and ZBP1-dependent PANoptosis in SARS-CoV-2 bystander cells contribute to the amplification of cytokines, we cultured STING KO and ZBP1 KO THP-1 cells with CoV-2-sup. The concentrations of TNF-α and IL-6 in the STING KO and ZBP1 KO groups were significantly lower than those in the control group (SI Appendix, Fig. S6L). Additionally, the concentration of IFN-β in the STING KO group was significantly lower than in the other two groups (SI Appendix, Fig. S6L). These results confirm that STING activation and ZBP1-dependent PANoptosis in SARS-CoV-2 bystander cells play a potentially crucial role in maintaining or amplifying inflammatory responses in COVID-19.

Cotreatment with cGAMP, TNF-α, and IFN-β Triggers ZBP1-Dependent PANoptosis.

We demonstrated that bystander cell PANoptosis was dependent on STING-induced ADAR1 degradation, as well as the presence of TNF-α and IFN-β (Figs. 4 and 5). To further verify this mechanism in a context unrelated to SARS-CoV-2 infection, control cells and ADAR1 knockdown cells were stimulated with TNF-α, IFN-β, alone or in combination (SI Appendix, Fig. S6H and S7I). Results showed that THP-1 cells only with ADAR1 knockdown did not exhibit significant cell death (Fig. 6A), which is consistent with the data presented above and previous reports (SI Appendix, Fig. S6A) (34, 46). However, as in other studies (45), ZBP1 HT-29 cells with ADAR1 knockdown displayed noticeable cell death. Importantly, under ADAR1 knockdown conditions, the cell death induced by cotreatment with TNF-α and IFN-β was significantly higher than that induced by untreated or single treatments ((Fig. 6 A and B and SI Appendix, Fig. S7 A and B). This cell death was characterized by cleavage of GSDME and GSDMD, activation of caspase-3/-7, as well as MLKL phosphorylation (Fig. 6 B and C and SI Appendix, Fig. S7B). Consistently, Z-NA accumulation was not observed in WT THP-1 cells treated with TNF-α and IFN-β (SI Appendix, Fig. S7C).

Fig. 6.

TNF-α and IFN-β induce ZBP1-dependent PANoptosis in ADAR1 knockdown THP-1 cells without SARS-CoV-2 infection. (A) Ctrl-sh and ADAR1-sh THP-1 cells were treated with TNF-α (50 ng/mL) alone, IFN-β (100 ng/mL) alone, or the combination of TNF-α (50 ng/mL) and IFN-β (100 ng/mL). Then cell death rate was detected at indicated time by Sytox Green. (B) ADAR1-sh THP-1 treated with TNF-α (50 ng/mL) alone, IFN-β (100 ng/mL) alone, or cotreatment with TNF-α (50 ng/mL) and IFN-β (100 ng/mL) for 48 h. The cell lysates were analyzed via immunoblotting using the indicated antibodies. (C) After ADAR1-sh THP-1 cotreated with TNF-α (50 ng/mL) and IFN-β (100 ng/mL) for indicated time, the cell lysates were analyzed via immunoblotting using the indicated antibodies. (D) ADAR1-sh THP-1 treated with TNF-α (50 ng/mL) alone, IFN-β (100 ng/mL) alone, or cotreatment with TNF-α (50 ng/mL) and IFN-β (100 ng/mL) for 48 h. The cell was detected the presence of Z-NA using against Z-NA antibodies (clone Z22). Fluorescence intensity (arbitrary units, a.u.) of Z-NA signals in D (Right). (E) Ctrl-sh and ADAR1-sh THP-1 cells were treated with TNF-α (50 ng/mL) alone, IFN-β (100 ng/mL) alone, or the combination of TNF-α (50 ng/mL) and IFN-β (100 ng/mL) for 48 h. The cell lysates were subjected to co-IP with anti-Casp8 or anti-ZBP1 antibodies and were immunoblotted with anti-Casp8, anti-ZBP1, or anti-RIP3 antibodies, respectively. (F) (ADAR1-sh ZBP1-KO) THP-1 cells with HA-ZBP1, HA-ZBP1 ΔZα1, HA-ZBP1 ΔZα1 + ΔZα2, or HA-ZBP1 Mut Zα2 were cotreated with TNF-α (50 ng/mL) and IFN-β (100 ng/mL) for 48 h. The cell death was quantified by Sytox Green. Data are mean ± SD (n = 6 in A and F; n = 10 in D per group). Two-tailed unpaired t test with Welch’s correction (A and D). Two-way ANOVA test (D). One-way ANOVA test (F). ****P < 0.0001. [Scale bars, 10 μm (D).] Data are from at least three independent experiments.

To distinguish the relative contributions of TNF-α and IFN-β to PANoptosis, it was found that, the levels of Z-NA accumulation in ADAR1 knockdown cells were similar between the IFN-β group and the combination group; however, TNF-α treatment alone did not induce Z-NA accumulation (Fig. 6D). Additionally, TNF-α and IFN-β together induced a strong interaction between ZBP1, Caspase-8, and RIP3 in ADAR1 knockdown cells (Fig. 6E). In contrast, TNF-α alone did not induce this interaction, and IFN-β alone resulted in a weaker interaction (Fig. 6E). These data confirm that TNF-α and IFN-β cotreatment induces robust PANoptosis in ADAR1 knockdown cells, with IFN-β driving Z-NA accumulation and TNF-α cooperating with IFN-β to enhance the formation of the PANoptosome. To determine whether the PANoptosis induced by TNF-α and IFN-β in ADAR1 knockdown cells is dependent on ZBP1, we expressed WT or mutant ZBP1 in ADAR1 knockdown cells (SI Appendix, Fig. S7 J and K), and subsequently cotreatment with TNF-α and IFN-β. The cell death rates in the HA-ZBP1 ΔZα1 + Zα2 and HA-ZBP1 Mut Zα2 groups were significantly lower than those in the HA-ZBP1 group (Fig. 6F and SI Appendix, Fig. S7D). These findings confirm that, even without SARS-CoV-2 infection, cotreatment with TNF-α and IFN-β can trigger ZBP1-dependent robust PANoptosis in ADAR1 knockdown cells. Similar to TNF-α and IFN-β, the combination of TNF-α and IFN-α also induces comparable cell death (SI Appendix, Fig. S7E). In contrast to the previously reported cell death induced by TNF-α and IFN-γ (6), cell death induced by TNF-α and IFN-β depends on the downregulation of ADAR1 and the expression of ZBP1 (SI Appendix, Fig. S7 F and G).

In addition, the transfection of cGAMP into cells triggers the degradation of ADAR1 and ZBP1-dependent cell death, which can be inhibited by CQ (SI Appendix, Fig. S8 A and B). Furthermore, the cell death could also be suppressed by cotreatment with TNFi and IFNi, and could be magnified by exogenous addition of TNF-α and IFN-β (SI Appendix, Fig. S8C). Moreover, the direct addition of the three cytokines (cGAMP, TNF-α, and IFN-β) to cells induced STING activation and downregulated ADAR1 expression, thereby triggering robust ZBP1-dependent PANoptosis (SI Appendix, Fig. S8 D and E). The kinetics of cell death induced by costimulation with cGAMP, TNF-α, and IFN-β were proportional to the concentrations of the three cytokines (SI Appendix, Fig. S8F). Notably, even when the concentrations of cGAMP, TNF-α, and IFN-β were reduced to levels comparable to those measured in CoV-2-sup, significant cell death was still observed (SI Appendix, Fig. S8F). These data suggested that cotreatment with cGAMP, TNF-α, and IFN-β triggers ZBP1-dependent PANoptosis.

Inhibition of cGAS Provides Protection Against Lung Damage Induced by SARS-CoV-2 or IAV Infection.

Based on our findings, cGAMP, TNF-α, and IFN-β in CoV-2-sup are the core components triggering bystander cell PANoptosis (Figs. 4 and 5). Our data show that cGAS inhibition significantly reduces cGAMP, TNF-α, and IFN-β in the culture supernatant of SARS-CoV-2-infected cells (SI Appendix, Fig. S9A). Therefore, we hypothesized that the inhibition of the cGAS–STING pathway could reduce PANoptosis and mitigate SARS-CoV-2-induced lethality and lung injury in vivo. We first confirmed that the cGAS inhibitor G140 can effectively inhibit PANoptosis in ACE2-THP-1 cells (Fig. 7A). To further investigate the role of cGAS–STING inhibition in SARS-CoV-2 infection in vivo, we selected another cGAS inhibitor, RU.521, which has superior efficacy in inhibiting the enzymatic activity of mouse cGAS compared to G140 (49, 50). K18-hACE2 mice were then injected with Vehicle or RU.521, once daily for 5 consecutive days starting 1 d after SARS-CoV-2 infection (Fig. 7B). Immunofluorescence staining confirmed SARS-CoV-2 infection through visualization of viral NP protein in lung tissues (Fig. 7C). Importantly, Z-NA, cleaved caspase-3, and IL-1β staining was also observed in cells that did not stain with NP antibodies (Fig. 7C), indicating cell death and an inflammatory response in bystander cells. Furthermore, RU.521 treatment notably inhibited the increase of TNF-α and IFN-β in bronchoalveolar lavage fluid (BALF) of SARS-CoV-2-infected mice (SI Appendix, Fig. S9B). It also significantly reduced the cleavage of GSDME and GSDMD, activation of caspase-3/-7, as well as MLKL phosphorylation in lung tissue resulting from SARS-CoV-2 infection (SI Appendix, Fig. S9C). Moreover, immunofluorescence and immunohistochemistry data showed that RU.521 treatment significantly inhibited cleaved caspase-3, phosphorylated MLKL, and IL-1β in the lungs induced by SARS-CoV-2 infection (Fig. 7D). Histological examination of the lungs revealed that the RU.521 treatment greatly alleviated SARS-CoV-2-induced immune cell infiltration and alveolar septa expansion (Fig. 7E). Remarkably, mice treated with RU.521 exhibited effective protection from death due to SARS-CoV-2 infection, with a 67% increase in survival (Fig. 7F). Together, these data strongly suggest that Z-NA accumulation and PANoptosis occurred in SARS-CoV-2 bystander cells in vivo. The inhibition of the cGAS can effectively reduce PANoptosis, which is critical for protecting against SARS-CoV-2-mediated pathology and mortality in vivo (Fig. 8).

Fig. 7.

Inhibition of cGAS provides protection against lung damage induced by SARS-CoV-2 infection. (A) ACE2-THP-1 was infected with or without SARS-CoV-2 (MOI = 0.5) for 3 d, after being treated with Vehicle or G140 (5 μM) for 4 h. Then, the cell lysates were analyzed via immunoblotting using the indicated antibodies. (B–F) Mice were Mock or SARS-CoV-2 infection (intranasal; 3 × 10³ PFU per mouse) and intraperitoneally administered with vehicle (n = 6) or RU.521 (5 mg/kg; n = 6) (once daily for 5 consecutive days starting 1 d after SARS-CoV-2 infection). Schematic of experimental design (B). Paraffin-embedded lung sections from Mock- and SARS-CoV-2-infected mice were stained with Z-NA (Green), cleaved Caspase 3 (cle-Casp3) (Green) or IL-1β (Green) and SARS-CoV-2 NP (Red) and DAPI (blue) (C). Paraffin-embedded lung sections from mock group, SARS-CoV-2 vehicle group, and SARS-CoV-2 RU.521 group mice were stained with cleaved Caspase 3 (cle-Casp3) (Green), IL-1β (Green) and DAPI (blue), or subjected to immunohistochemistry with p-MLKL antibodies (D, Top). Quantitative analysis of cleaved Caspase 3 (cle Casp3) and IL-1β positive cells per field and the average optical density (AOD) of p-MLKL (D, Bottom). Hematoxylin and eosin (H&E) images of lungs from Mock-, Vehicle-, RU.521-group mice (E). The survival curves of mice with the indicated treatments (F). Asterisks denote a nonspecific band. Data are mean ± SD (n = 6 in C, D per group). P values were calculated using the log-rank test (F). One-way ANOVA test (D). **P < 0.01; ***P < 0.001; ****P < 0.0001. [Scale bars, 20 μm (C and D), 500 μm (E, Top), 50 μm (E, Bottom).] Data are from at least three independent experiments.

Fig. 8.

Schematic overview of the mechanisms underlying SARS-CoV-2-induced STING activation and ZBP1-dependent PANoptosis in bystander cells. The underlying mechanisms by which SARS-CoV-2 infection provokes this intense inflammatory reaction and how sustained inflammation leads to tissue damage are not fully understood. This study demonstrates that SARS-CoV-2 infection induces STING activation and ZBP1-dependent PANoptosis in bystander cells, contributing to the persistence of inflammatory responses. Specifically, we found that cGAS–STING activation in SARS-CoV-2 infected cells leads to the secretion of cGAMP, TNF-α, and IFN-β. The translocation of cGAMP to bystander cells is through the transporter of SLC19A1 or LRRC8A. This translocation activates STING in the bystander cells, leading to STING-induced autophagy-mediated degradation of ADAR1. The down-regulation of ADAR1, combined with IFN-β-induced accumulation of Z-NA and TNF-α-induced assembly of the PANoptosome complex, triggers ZBP1-dependent PANoptosis in bystander cells. cGAS inhibitors RU.521 effectively inhibit PANoptosis, thereby providing protection against lung injury caused by SARS-CoV-2.

IAV can trigger ZBP1-dependent PANoptosis following infection (30). Furthermore, IAV infection activates the cGAS pathway by inducing the release of mtDNA (19, 51). Validation revealed that IAV infection activates STING and induces PANoptosis, while cGAS inhibition reduces the levels of PANoptosis (SI Appendix, Fig. S10 A and B). To further investigate the roles of RU.521 during IAV infection in vivo, mice were administered with RU.521 daily for 5 consecutive days, starting 1 d postinfection with a nonlethal dose of IAV (SI Appendix, Fig. S11A). Immunofluorescence analysis of lung sections from the infected mice revealed cleaved caspase-3 and IL-1β staining in cells not labeled with IAV hemagglutinin (HA) antibodies, indicating cell death and an inflammatory response in bystander cells (SI Appendix, Fig. S11B). During the first 7 d postinfection, RU.521 treatment significantly mitigated weight loss in the infected mice (SI Appendix, Fig. S11C). Additionally, the treatment notably inhibited the increase of TNF-α and IFN-β in bronchoalveolar lavage fluid (BALF), which was induced by IAV infection (SI Appendix, Fig. S11D). RU.521 also effectively reduced the cleavage of GSDME and GSDMD, activation of caspase-3/-7, MLKL phosphorylation, and IL-1β release in lung tissue caused by the infection (SI Appendix, Fig. S11 E and F). Similar to observations with SARS-CoV-2 infection, RU.521 treatment significantly alleviated IAV-induced inflammatory cell infiltration and alveolar septal expansion (SI Appendix, Fig. S11G). These findings demonstrate that inhibiting PANoptosis through a cGAS–STING inhibitor is an effective strategy for preventing IAV infection-mediated pathology in vivo.

STING KO mice were infected with a lethal dose of IAV to further validate the occurrence of bystander cell death and its role in viral infection in vivo. STING-KO mice exhibited a significantly higher survival rate (increased by 50%) and markedly mitigated body weight loss compared to WT mice (SI Appendix, Fig. S12 A and B). Further analyses revealed that bystander cell death in the lungs of STING-KO mice was reduced, accompanied by decreased levels of TNF-α and IFN-β, attenuated PANoptosis, and alleviated pulmonary pathological damage (SI Appendix, Fig. S12 C–G). Collectively, these results provide strong evidence for the critical role of STING-dependent bystander cell death in virus-induced lung injury.

Discussion

Extended or persistent cell death can lead to excessive inflammation, tissue damage, and even host death (52, 53). ZBP1-dependent cell death has been reported to be a key pathogenic factor in SARS-CoV-2 (3). Our finding that ZBP1-dependent PANoptosis occurs in bystander cells of SARS-CoV-2-infected individuals provides evidence for this view. As a highly contagious virus, SARS-CoV-2 has developed various strategies to combat and evade antiviral IFN responses (54–56). Consequently, relatively low viral component levels can effectively inhibit IFN response in the early stages of SARS-CoV-2 infection (57). However, as the infection progresses and viral components targeting IFN accumulate, the virus loses control over the IFN response (20, 58). This delayed but strong type I IFN response has been observed in severe COVID-19 patients and animal models of SARS-CoV-2 infection (3, 18, 59, 60). Our study found that cGAMP produced by cGAS activation in SARS-CoV-2-infected cells was transferred to bystander cells, thereby activating the STING pathway in those bystander cells. Furthermore, we found that this process is a key prerequisite for inducing PANoptosis in bystander cells. Due to the lack of viral components that inhibit the IFN response in bystander cells, this may explain the delayed but strong IFN response in severe COVID-19 patients. Additionally, it suggests that immune activation of bystander cells as an effective antiviral strategy also carries the risk of inducing cell death and inflammatory tissue damage.

We subsequently found that cGAMP transferred to bystander cells activates STING, which induces autophagic degradation of ADAR1, thereby creating conditions for Z-NA accumulation and the occurrence of PANoptosis in bystander cells. Building on this, the combined action of TNF-α and IFN-β can induce (accelerate) the occurrence of ZBP1-dependent PANoptosis. Our study reveals a mechanism for triggering PANoptosis, in which cotreatment with cGAMP, TNF-α, and IFN-β induces ZBP1-dependent PANoptosis. In this process, cGAMP induces STING-mediated ADAR1 degradation, IFN-β facilitates Z-NA accumulation, and TNF-α promotes the assembly of the PANoptosome complex. This provides a theoretical foundation for viewing “delayed but strong type I IFN response” as a pathological factor during viral infection (3, 18). Additionally, we found that STING and PANoptosis activation in SARS-CoV-2 bystander cells promoted the amplification and maintenance of inflammatory cytokines and the inflammatory response. This may explain the prolonged high expression of cytokines such as TNF-α, IFN-β, and IL-6 in some COVID-19 patients (15, 16).

In our study, activation of cGAS was central to triggering bystander cell PANoptosis during SARS-CoV-2 infection. We used RU.521, a cGAS inhibitor, to inhibit PANoptosis and protect against tissue damage caused by SARS-CoV-2 or IAV infection. RU.521 appears to have a better protective effect on survival rates and lung pathological damage in SARS-CoV-2-infected mice than H-151, a previously studied STING inhibitor (18). The likely reason for this difference is that H-151 cannot prevent the production and secretion of cGAMP in SARS-CoV-2-infected cells, nor can it prevent autophagy induced by STING in bystander cells. This further indicates that the production of cGAMP and autophagy induced by STING play an important role in the pathogenesis of SARS-CoV-2 infection. Studies have demonstrated that eliminating cell death induced by IAV infection through the deletion of ZBP1 or caspase-6 increases the risk of infection-related death (3, 30), due to impaired clearance of infected cells (viral clearance) (2). Therefore, blocking inflammatory death in bystander cells may provide protection against virus-induced lung injury. Consistent with this hypothesis, we observed that STING KO mice infected with IAV exhibited significantly reduced bystander cell death in the lungs compared to WT mice, accompanied by alleviated pulmonary damage and markedly improved survival rates. These findings robustly validate the critical role of the bystander cell PANoptosis axis characterized in this study during viral pathogenesis, while concurrently delineating actionable therapeutic avenues for intervention.

Overall, we identified a mechanism of disease pathology during COVID-19, wherein cytokines released by SARS-CoV-2-infected cells triggered PANoptosis of bystander cells. Subsequently, we clarified the specific molecular mechanisms, provided multiple potential therapeutic targets, and determined an effective therapeutic regimen. This insight into the mechanistic underpinnings of the innate immune response, cytokine-mediated pathology, and inflammatory cell death during SARS-CoV-2 infection informs therapeutic strategy development for COVID-19. Additionally, the therapeutic regimen we provide is equally effective against IAV infection, suggesting that this mechanism may be important for viral infections in general, as well as other inflammatory diseases. Meanwhile, this interaction between infected and bystander cells in host defense may have far-reaching implications for infectious diseases and vaccine development.

Materials and Methods

Animals, cell culture and transfection, cell stimulation, virus culture, focus forming assay (FFA), plasmid construction, stable cell line generation, siRNA transfection, dual-luciferase reporter assays, cell death rate analysis, cytokine analysis, quantification of cGAMP by ELISA, coimmunoprecipitation assay, immunoblotting, immunofluorescence microscopy, immunohistochemistry, histopathological analysis, animal experiment, statistical analysis, and study approval are described in SI Appendix, Materials and Methods.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.