ZBP1 induces inflammatory signaling via RIPK3 and promotes SARS-CoV-2-induced cytokine expression

COVID-19 caused by the SARS-CoV-2 virus remains a threat to global health. The disease severity is mediated by cell death and inflammation, which regulate both the antiviral and the pathological innate immune responses. ZBP1, an interferon-induced cytosolic nucleic acid sensor, facilitates antiviral responses via RIPK3. Although ZBP1-mediated cell death is widely described, whether and how it promotes inflammatory signaling is unclear. Here, we report a ZBP1-induced inflammatory signaling pathway that depends on ubiquitination and RIPK3’s scaffolding ability independently of cell death. In human cells, ZBP1 associates with RIPK1 and RIPK3 as well as ubiquitin ligases cIAP1 and LUBAC. RIPK1 and ZBP1 are ubiquitinated to promote TAK1- and IKK-mediated inflammatory signaling. Additionally, RIPK1 recruits the p43/41-caspase-8-p43-FLIP heterodimer to suppress RIPK3 kinase activity, which otherwise promotes inflammatory signaling in a kinase activity-dependent manner. Lastly, we show that ZBP1 contributes to SARS-CoV-2-induced cytokine production. Taken together, we describe a ZBP1-RIPK1-RIPK3-mediated inflammatory signaling pathway relayed by the scaffolding role of RIPKs and regulated by caspase-8. Our results suggest the ZBP1 pathway contributes to inflammation in response to SARS-CoV-2 infection.


Introduction
Viruses can cause a range of severe acute and chronic diseases and represent a global health threat, as illustrated by Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) that caused the Coronavirus disease  pandemic. Many viral diseases are characterized by over-reaction of the immune system that contributes to disease development (Fajgenbaum & June, 2020). This is seen in COVID-19 where patients with severe symptoms suffer from a "cytokine storm", which is featured by elevated excessive circulating cytokine levels that correlate with lethality (Zhou et al, 2020;Chen et al, 2020;Lucas et al, 2020).
Inflammation and cell death underlie antiviral innate immune responses and contribute to pathological inflammatory conditions when deregulated. Receptor interacting protein (RIP) kinases (RIPKs), engaged downstream of immune receptors, are central regulators of cell death and inflammatory signaling pathways and contribute to host immune defenses against viruses and bacteria (Newton, 2020;He & Wang, 2018;Topal & Gyrd-Hansen, 2021).
The formation of non-degradative ubiquitin (Ub) chains linked via lysine 63 (K63-Ub) and methionine 1 (M1-Ub) within receptor signaling complexes facilitates the activation of the kinases TAK1 and IKKα/β, which in turn activate MAP kinase signaling and NF-kB signaling to stimulate the expression of pro-inflammatory cytokines and chemokines (reviewed in Hrdinka and Gyrd-Hansen, 2017).
In this study, we identify RIPK1 and RIPK3 as scaffolding kinases that mediate ZBP1-triggered inflammatory signaling independently of cell death. ZBP1-RIPK3-RIPK1 inflammatory signaling is dependent on K63-Ub and M1-Ub and Ub ligases cIAPs and LUBAC but does not require the kinase activity of RIPK1 and RIPK3. Inhibition of caspase-8 activity exposes a RIPK3 kinase activity-mediated inflammatory signaling pathway, akin to the pathway triggered by TNF when cIAPs and caspase-8 are inhibited. Finally, we provide evidence that ZBP1 contributes to the production of cytokines and chemokines during SARS-CoV-2 infection.

ZBP1 and RIPK3 stimulate inflammatory signaling independently of cell death
ZBP1-induced signaling is mediated by RIPK3 and is dependent on RHIM interactions (Kaiser et al, 2008;Rebsamen et al, 2009). To investigate the ability of ZBP1 to stimulate inflammatory signaling versus cell death, we generated HT29 cells with doxycycline (Dox)-inducible expression of FLAG-tagged wild-type (WT) ZBP1 (ZBP1 WT ) or a mutant form in its Z-form nucleic acid binding (Zα)-domains (ZBP1 Zα1α2mut ) ( Figure 1A). Dox treatment induced the expression of ZBP1 WT and ZBP1 Zα1α2mut in a dose-dependent manner, albeit ZBP1 Zα1α2mut expressed at higher levels than ZBP1 WT ( Figure 1B).
In accordance with the ability of ZBP1 to stimulate ligand-dependent cell death (Jiao et al, 2020;Wang et al, 2020), treatment with 500 ng/ml Dox resulted in the loss of viability after 48 h in HT29 cells expressing ZBP1 WT but not ZBP1 Zα1α2mut ( Figure 1C). Contrary to previous studies in murine systems (Upton et al, 2012;Wang et al, 2020;Zhang et al, 2020;Jiao et al, 2020), the cell death induced by ZBP1 in HT29 cells was predominantly apoptosis, as viability was restored by the caspase inhibitor zVAD-fmk (zVAD) alone or in combination with inhibition of RIPK3 (GSK'872), RIPK1 (Nec1s) or MLKL (necrosulfonamide; NSA), whereas GSK'872, Nec1s or NSA alone did not restore viability ( Figure S1A). Treatment with 50 ng/ml or 100 ng/ml Dox did not result in the loss of viability although ZBP1 expression was induced ( Figures 1C and S1B). Based on this, we treated the cells with increasing concentrations of Dox up to 50 ng/ml and measured cytokine production after 24 h. Despite no loss of cell viability, the induced expression of ZBP1 stimulated the production of chemokines CXCL8 and CXCL1 in a Dox concentration-dependent manner, which was also dependent on the ligand-binding ability of ZBP1 ( Figure 1D). The Dox-induced expression of ZBP1 Zα1α2mut was higher than ZBP1 WT , yet chemokine production was significantly lower, underscoring the importance of ligand-binding for ZBP1 signaling at low expression levels ( Figure 1D). Similarly, transient expression of ZBP1 in HEK293T cells induced NF-kB activity in a ligand-dependent manner ( Figure S1C). In line with previous reports (Rebsamen et al, 2009;Kaiser et al, 2008), ZBP1-induced inflammatory signaling relied on RIPK3's RHIM, since ZBP1 WT induced NF-κB activity in cells expressing full-length RIPK3 (RIPK3 FL ) but not a Cterminal truncated RIPK3 lacking the RHIM (RIPK3 dC ) ( Figure S1C). Similarly, ZBP1-induced inflammatory signaling in HT29 cells was mediated by RIPK3 ( Figures 1E and S1D). Notably, at higher Dox concentrations, both ZBP1 WT and ZBP1 Zα1α2mut induced the production of chemokines in HT29 cells, either when treated with equal amount of Dox ( Figure S1E) or when induced to express at the same level ( Figure S1F), albeit the chemokine production induced by ZBP1 Zα1α2mut was significantly lower than that by ZBP1 WT when expressing at the same levels ( Figure S1F). This suggests that ZBP1, when highly expressed, can trigger inflammatory signaling independently of ligand-binding (Maelfait et al, 2017).
Dimerizer-induced RIPK3 oligomerization in HT29 cells stimulated inflammatory signaling after 30-60 min as determined by phosphorylation of the NF-κB subunit RelA/p65 and MAPKs p38 and JNK as well as the expression and production of the chemokine CXCL8 ( Figures 1F-1H). We obtained similar results after dimerization or oligomerization of RIPK3 in HCT116, U2OS and HEK293T cells ( Figures S2B-H). Notably, contrary to murine cells where oligomerization of RIPK3 induces both necroptosis and inflammatory signaling (Orozco et al, 2014;Yatim et al, 2015), RIPK3 oligomerization did not result in loss of viability of HT29 or HCT116 cells within 24 h ( Figures 1I and S2I) although the HT29 cells were sensitive to TNFinduced cell death when pre-treated with Smac mimetic Compound A (CpA; (Vince et al, 2007)) (apoptosis) or CpA + zVAD (necroptosis) (Figures S2J).
In addition to CXCL8 and CXCL1, we found that ZBP1 expression in HT29 cells stimulated the secretion of the chemokines CXCL10/IP-10, CCL20, CXCL7 and other proinflammatory mediators ( Figures S3A and S3B). This suggests that ZBP1 signaling may stimulate the chemoattraction of neutrophils and other immune cells. Indeed, conditioned media from Doxtreated HT29 / Tet-On-ZBP1 cells stimulated chemotactic migration of THP1 monocytic cells, neutrophil-like differentiated HL60 cells and primary human donor neutrophils ( Figures 1J-1L and S3C). To determine if undetected cell death after Dox treatment might contribute to chemotaxis, HT29 cells were treated with zVAD and Nec1s in combination with Dox for chemotaxis. ZBP1-dependent chemotactic migration of primary donor neutrophils was unaffected by treatment with both inhibitors, indicating that the migration was stimulated by ZBP1-dependent inflammatory signaling and not by cell death ( Figure 1M).

ZBP1-RIPK3 inflammatory signaling is mediated by IKK and TAK1 independently of the kinase activity of RIPK3 or RIPK1
To gain insights into the mechanism underpinning ZBP1 and RIPK3-induced inflammatory signaling, HT29 cells were treated with kinase inhibitors in combination with Dox-induced expression of ZBP1. Treatment with Nec1s, GSK'872 and NSA did not lead to a significant decrease in ZBP1-induced production of CXCL8 and CXCL1 in HT29 cells or NF-κB activation in RIPK3-expressing HEK293FT cells (Figures 2A and 2B), indicating that inflammatory signaling by ZBP1 occurs independently of the kinase activity of RIPK3 and RIPK1.
Accordingly, inhibition of RIPK3, RIPK1 or MLKL had no effect on the ZBP1-stimulated chemotactic migration of neutrophils ( Figure 2C).
The involvement of M1-Ub implied that LUBAC contributes to ZBP1-RIPK3-dependent inflammatory signaling. Indeed, RIPK3 oligomerization-induced signaling was substantially reduced in HCT116 cells deficient for HOIP (Hrdinka et al, 2016), the catalytic subunit of LUBAC, as compared to WT cells ( Figure 4A). Also, ZBP1-induced NF-κB activity was substantially impaired by siRNA-mediated knockdown of HOIP in HEK293FT cells ( Figure 4B). K63-Ub assembled by cIAPs facilitates the recruitment of LUBAC to the TNF receptor signaling complex (Haas et al, 2009), which prompted us to investigate their involvement in ZBP1-RIPK3 inflammatory signaling. Depletion of cIAPs by CpA completely inhibited ZBP1 and RIPK3-oligomerization-induced inflammatory signaling and chemokine production S4B and S4C). In addition to depletion of cIAPs, CpA antagonizes the function of XIAP in NOD2 signaling when used at 1 µM or above (Damgaard et al, 2013). However, RIPK3 oligomerization-induced inflammatory signaling was comparable in WT and XIAPknockout HCT116 cells ( Figure S4D), indicating that the inhibition of ZBP1 and RIPK3 inflammatory signaling by CpA reflects a requirement for cIAPs. Consistent with our signaling data, analysis of the ZBP1 complex immunoprecipitated from Dox-treated HT29 cells showed that ZBP1, in addition to RIPK1 and RIPK3, coimmunoprecipitated K63-and M1-Ub, cIAP1, LUBAC (HOIP, HOIL-1, SHARPIN), the LUBACassociated DUBs OTULIN and CYLD,TAK1,and IKKb (Figures 4G and 4H). This suggests that ZBP1 forms a pro-inflammatory receptor signaling complex that, akin to other immune receptor complexes, consists of adaptor kinases, K63-and M1-Ub ligases and DUBs, and the ubiquitin-dependent kinases TAK1 and IKKb.

RIPK1 is required for ZBP1 signaling
Prompted by our data showing ZBP1-induced RIPK1 ubiquitination, we sought to determine the functional role of RIPK1 in the ZBP1 complex. For this, we generated RIPK1-KO HT29 single cell clones and integrated the Dox-inducible ZBP1 expression system. Strikingly, both ZBP1-induced chemokine production and cell death were blocked in the absence of RIPK1, demonstrating a central role of RIPK1 in mediating signaling responses by ZBP1 in HT29 cells (Figures 5A and 5B,S5A and S5B).
The association of RIPK3 with ZBP1, especially its phosphorylated form, was greatly reduced in the absence of RIPK1, indicating a primary function of RIPK1 in stabilizing the ZBP1-RIPK3 interaction ( Figure 5C). Intriguingly, the association of ZBP1 with LUBAC, cIAPs, CYLD and OTULIN was not affected in RIPK1-KO cells as compared to WT cells, while there was a reduction in co-immunoprecipitated Ub-conjugates with ZBP1 ( Figure 5C). Correspondingly, the ubiquitination on ZBP1 was substantially reduced in RIPK1-KO cells ( Figure 5D). This supported the notion that the ubiquitination of RIPK1 and ZBP1 is important for ZBP1dependent inflammatory signaling, although it is not clear how RIPK1 regulates ZBP1 ubiquitination.

RIPK1 links caspase-8 to RIPK3 to suppress kinase activity-dependent inflammatory signaling
Strikingly, p43/41-caspase-8 and p43-cFLIPL, processed as part of a catalytically active heterodimer, were co-immunoprecipitated with ZBP1 in a RIPK1-dependent manner ( Figure   5C). Since RIPK1 links caspase-8 to RIPK3 to suppress necroptosis, we speculated that caspase-8 might regulate ZBP1-RIPK3 inflammatory signaling. Indeed, zVAD treatment enhanced the production of CXCL8 by Dox-induced ZBP1 expression in HT29 cells and chemokine expression following dimerizer treatment of HCT116/RIPK3-2xFV cells ( Figures   6A and 6B). Also, shRNA-mediated knockdown of caspase-8 in HCT116/RIPK3-2xFV cells led to increased dimerizer-induced chemokine expression ( Figure 6C). Strikingly, the enhanced signaling in the absence of caspase-8 activity was inhibited by GSK'872, Nec1s and NSA, as well as by TAK1 and IKK inhibitors ( Figures 6A-6C and S6A). These data show that in the absence of caspase-8 activity, ZBP1-induced inflammatory signaling is, in part, mediated by RIPK3 and RIPK1 kinase activity as well as MLKL.
This was reminiscent of previous reports showing that necroptotic signaling stimulated by TNF in the presence of Smac mimetics and caspase inhibitor zVAD (TSZ) induces RIPK1-RIPK3-MLKL-dependent expression of chemokines concomitantly with cell death (Zhu et al, 2018).
In concordance, TSZ (TNF+CpA+zVAD) stimulated strong CXCL8 expression in HCT116 / RIPK3-2xFV cells but not in parental HCT116 cells, whereas TNF-, TNF+CpA-, and TNF+zVAD-induced CXCL8 expression was largely independent of RIPK3 ( Figure 6D). No cell death was detected following TSZ treatment at the time of the assay, demonstrating that the RIPK3-inflammatory signaling pathway is engaged by TNF without cell death ( Figure S6B). To determine if RIPK1 is required for caspase-8 to suppress RIPK3 kinase activity-mediated signaling, we treated HCT116 / RIPK1-KO cells with zVAD while inducing RIPK3 oligomerization. Whilst RIPK1 was not needed for RIPK3 oligomerization-induced chemokine expression, RIPK1 was required for the enhancement of chemokine expression by zVAD ( Figure 6E). Under caspase-inhibited conditions, RIPK3 oligomerization led to the accumulation of phosphorylated RIPK3 (P-RIPK3) that migrated slower on SDS-PAGE than the unmodified RIPK3 ( Figure 6F). This slower-migrating P-RIPK3 likely represents the kinase-activated form of RIPK3, which, in the absence of zVAD, is suppressed by caspase-8 (He et al, 2009;Choi et al, 2018). This is supported by the appearance of a ~30 kDa RIPK3 fragment after dimerizer-treatment of cells expressing RIPK3-2xFV (Figures 1F and S2H;(Feng et al, 2007)). Consistent with the idea that RIPK1 links caspase-8 to RIPK3, dimerizer treatment of RIPK1-KO cells induced the accumulation of the slower-migrating P-RIPK3 form even without caspase inhibition ( Figure 6F).
Together, these data show that caspase-8 activity restricts ZBP1 inflammatory signaling by inhibiting RIPK3 kinase activity-mediated signaling and suggests that RIPK1 mediates the suppression of kinase-activated RIPK3 ( Figure 6G).

ZBP1 mediates SARS-CoV-2-induced cytokine production
Lastly, we investigated the contribution of ZBP1-induced inflammatory responses during virus infection. COVID-19, caused by the SARS-CoV-2 virus, is characterized by lower antiviral responses and high inflammatory chemokine and cytokine production (Blanco-Melo et al, 2020). Analysis of publicly available RNA-sequencing datasets (Blanco-Melo et al, 2020) showed a substantial increase in ZBP1 expression in post-mortem patient lung biopsies compared to healthy controls ( Figure 7A). In addition, analysis of a single-cell RNAsequencing dataset (Ren et al, 2021) showed that ZBP1 expression was significantly higher in COVID-19 patients in the progressive disease stage than those in the convalescent stage or healthy controls ( Figure 7B). Monocytes and neutrophils have been found to be two major sources of cytokines in the peripheral blood that contribute to the "cytokine storm" in critically ill COVID-19 patients (Ren et al, 2021;Vanderbeke et al, 2021). Analysis of the single-cell RNAseq dataset (Ren et al, 2021) revealed an increase in ZBP1 expression in neutrophils and all major inflammatory subsets of monocytes in COVID-19 patients in progressive stage compared to convalescent stage patients ( Figure S7A). Taken together, these analyses indicate a potential role of ZBP1 in fueling the cytokine storm in COVID-19 patients.
To investigate the role of ZBP1 in SARS-CoV-2 infection, Calu-3 human lung adenocarcinoma cells were infected at two different multiplicities of infection (MOI), and the expression profiles of a panel of inflammatory genes were determined over time. This showed that SARS-CoV-2, in a dose-dependent manner, caused the upregulation of ZBP1, cytokines IL-6 and TNF, and chemokines CXCL10, CXCL8 and CXCL1 between 48 h and 72 h after infection, and led to the secretion of IL-6 and CXCL10 ( Figures 7C and 7D). In concordance, analysis of the singlecell RNA sequencing dataset revealed positive correlations between virus load and the expression levels of ZBP1 and various cytokines in virus-positive cells from bronchoalveolar lavage fluid (BALF) and sputum samples ( Figure 7E).
We then established three independent Calu-3 cell lines with stable knockdown of ZBP1 (shZBP1) and control cells expressing a non-targeting shRNA (shMM; Figure 7F, top-left panel). Infection with SARS-CoV-2 (MOI=2) showed that knockdown of ZBP1 attenuated the expression of the measured cytokine and chemokine genes, as well as the production of CXCL10 and to a lesser degree of IL-6 ( Figures 7F and 7G). The attenuation of the proinflammatory response after infection correlated with the efficacy of the individual shRNAs to silence ZBP1 expression; shZBP1-50 resulted in a substantially stronger attenuation than did shZBP1-52 or shZBP1-53 ( Figures 7F and 7G). The effect of ZBP1 knockdown was less prominent for interferon-induced genes ( Figure S7B), and the intracellular virus amount 3 days after infection was similar in all cell lines ( Figure S7C). Of note, whilst ZBP1 expression can trigger cell death, the corresponding markers remained negative by Western blotting analysis in Calu-3 cells up to 3 days after SARS-CoV-2 infection ( Figure S7D). Together, this suggests that ZBP1, subsequent to its transcriptional upregulation by SARS-CoV-2 infection, stimulates the production of inflammatory chemokines and cytokines that contribute to the severity of COVID-19 symptoms.
Altogether, our data reveal that ZBP1 can activate Lys63-and Met1-Ub-dependent inflammatory signalling mediated by a kinase-independent scaffolding function of RIPK3 and RIPK1, and suggest that this pathway contributes to inflammatory responses triggered by SARS-CoV-2.

RIPK3 as a scaffolding kinase for inflammatory signaling
RIPK3 was initially reported to stimulate both inflammatory signaling and cell death, but RIPK3-deficient cells showed normal NF-κB signaling in response to the stimulation of TNFR1, B and T cell receptors, and Toll-like receptors (TLRs) 2 and 4, excluding the role of RIPK3mediated inflammatory signaling in those contexts (Yu et al, 1999;Newton et al, 2004;Sun et al, 1999). Here, we demonstrate that RIPK3 is a bona fide inflammatory mediator by identifying it as an essential adaptor protein in ZBP1-dependent inflammatory signaling in human cells.
Interestingly, the role of RIPK3 in inflammatory signaling depends on its RHIM domain and does not require its kinase activity, which is in contrast to its role in necroptosis where kinase activity is essential. This indicates that RIPK3 acts as a scaffold and not as a kinase in ZBP1induced inflammatory signaling. This is reminiscent of the scaffolding role of other receptor-associated kinases in mediating inflammatory signaling, including RIPK1 in TNFR1 signaling, RIPK2 in NOD2 signaling, and IL-1R-associated kinases (IRAKs) in IL-1R signaling (Ea et al, 2006;Hrdinka et al, 2018;Ordureau et al, 2008;Koziczak-Holbro et al, 2008). While these kinases serve as scaffolds for the formation of K63-and/or M1-Ub, ubiquitination of RIPK3 was not consistently detected in response to ZBP1 expression. Instead, ZBP1 and RIPK1 were both modified with K63-and/or M1-Ub. Together with the requirement of RIPK1 and the RHIM of RIPK3 for ZBP1 inflammatory signaling, this suggests that RIPK3 may mediate signaling by RHIM-mediated recruitment and/or stabilization of RIPK1 in the ZBP1 complex, whereas RIPK1 and ZBP1 are the primary ubiquitination targets to facilitate NF-κB activation. Since the deletion of RIPK1 reduced the association of RIPK3, in particular its phosphorylated form, with ZBP1, RIPK1 may also contribute to the stabilization of a RHIM-mediated ZBP1-RIPK3-RIPK1 complex.
This would be concordant with previous reports showing that a RIPK1-RIPK3 interaction precedes formation of RIPK3 oligomers during the activation of RIPK3 (Li et al, 2012;Wu et al, 2014). Of note, our study of the ZBP1 signaling complex relied on Dox-induced overexpression of ZBP1, which precludes a detailed time-resolved study of the assembly of the signaling complex and of ubiquitination dynamics of complex components in response to ligand-binding. Such investigations will be warranted when cognate ZBP1 ligands are better defined in order to gain detailed insights into the assembly of the ZBP1 signaling complex.
It was intriguing that RIPK3, in addition to its scaffolding role, also promoted inflammatory signaling by ZBP1 in a kinase activity-dependent manner when caspase-8 activity was inhibited. This is reminiscent of the inflammatory signaling pathway described for TNF-induced necroptosis (TSZ treatment) in HT29 cells, which is mediated by RIPK3 and RIPK1 kinase activity and by MLKL (Zhu et al, 2018). This suggests that caspase-8 represents a checkpoint switch for RIPK3 kinase activity-mediated signaling also in the context of ZBP1 by constantly suppressing the kinase activity-dependent inflammatory signaling pathway after the engagement of RIPK3. While it remains to be defined how the RIPK3 kinase activitydependent signaling leads to inflammatory gene activation, our observations indicate that the default role of RIPK3 in ZBP1 inflammatory signaling is as a scaffolding kinase and that the kinase activity-dependent pathway is activated when caspase-8 activity is antagonized, such as during infection by viruses encoding caspase inhibitors.
It is not fully understood how caspase-8 regulates RIPK3 kinase activity-dependent signaling but our data suggest that caspase-8 activity suppresses kinase-activated RIPK3. This notion is supported by the RIPK1-dependent co-immunoprecipitation of both the p43/41-caspase-8-p43-FLIP heterodimer and phosphorylated RIPK3 with ZBP1, and is in concordance with previous findings that kinase-activated RIPK3 is recognized and depleted from the cytosolic pool (Choi et al, 2018).

ZBP1-mediated inflammatory signaling and cell death responses
Since the discovery that ZBP1, via RIPK3, induces necroptosis during murine cytomegalovirus (MCMV) infection, its role in cell death during infection and embryonic development has been well established (Kuriakose & Kanneganti, 2018;Upton et al, 2012;Lin et al, 2016;Thapa et al, 2016;Wang et al, 2020;Newton et al, 2016;Jiao et al, 2020). Our study expands the understanding of ZBP1's function as we uncover that ZBP1, in human cell lines, contributes to SARS-CoV-2-induced chemokine and cytokine expression, and that ZBP1, in a ligand binding-dependent manner, triggered RIPK3-mediated inflammatory signaling at a lower expression threshold and at earlier time points than needed for stimulation of cell death. It is tempting to speculate that ZBP1, akin to other innate immune receptors, induces ubiquitindependent NF-κB signaling as a first line of defense to recruit innate immune cells (e.g. neutrophils and monocytes), and that triggering of cell suicide is a backup mechanism invoked during pathological conditions where ZBP1 expression is highly induced by interferons.
Curiously, the major mode of cell death in ZBP1-expressing HT29 cells appeared to be RIPK1mediated apoptosis whereas previous studies in murine systems showed that RIPK1 restricts ZBP1-RIPK3-induced necroptosis during development and in skin inflammation (Lin et al, 2016;Newton et al, 2016;Devos et al, 2020). Whether this represents a difference between mice and human or is specific to the experimental systems is interesting and should be addressed in future studies.

ZBP1 and SARS-CoV-2
Our SARS-CoV-2 experimental data and analysis of RNAseq datasets from COVID-19 patients show that ZBP1 is upregulated by SARS-CoV-2 infection and that its expression correlates with the expression of pro-inflammatory chemokines and cytokines, including CXCL10/IP-10 and IL-6 ( Figure 7). IL-6 and CXCL10 are upregulated in critically ill COVID-19 patients as part of the cytokine storm, and IL-6 levels correlate with mortality (Ruan et al, 2020;Huang et al, 2020;Chen et al, 2020). This suggests that ZBP1-mediated inflammatory signaling may contribute to an unbalanced immune response in COVID-19 patients, characterized by reduced antiviral signaling and enhanced inflammatory levels (Blanco-Melo et al, 2020).
SARS-CoV-2-induced ZBP1 expression in Calu-3 cells did not lead to detectable cell death, supporting the notion that ZBP1, as a primary response, stimulates inflammatory signaling in human cells. However, our observation does not exclude the possibility of ZBP1-induced cell death during SARS-CoV-2 infections in vivo, where ZBP1 expression may be further induced by interferons produced by immune cells (Fu et al, 1999;Takaoka et al, 2007). Conversely, it is plausible that ZBP1-induced inflammatory signaling contributes to pathological inflammatory conditions where ZBP1-mediated cell death has been reported (Upton et al, 2012;Kuriakose et al, 2016;Thapa et al, 2016;Newton et al, 2016;Lin et al, 2016;Jiao et al, 2020). Irrespective, our data suggest that antagonizing ZBP1 signaling pharmacologically could attenuate pathological inflammation.

Lead Contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact Prof. Mads Gyrd-Hansen (mgyrd@sund.ku.dk).

Materials Availability
Plasmids and cell lines generated in this study are available from lead contact with a completed Materials Transfer Agreement.

Cell lines
All cell lines were cultured at 37°C and 5% CO2 in growth medium supplemented with 10% v/v fetal bovine serum (FBS, Labtech FCS-SA), 60 μg/ml penicillin and 100 μg/ml streptomycin Cell lines were not authenticated.

Isolation of primary neutrophils
Primary human neutrophils were obtained from healthy donors with their written informed consent by The Oxford Radcliffe Biobank with project number ORB 20/A136. The study is authorized by South Central -Oxford C Research Ethics Committee (Ref# 19/SC/0173).
Neutrophils were isolated from 50 ml Ficoll-layered blood cone using EasySep Human Neutrophil Isolation Kit (StemCell 17957) following manufacturer's instructions.

Generation of cell lines stably expressing RIPK3 variants
For the construction of HT29-and HCT116-based RIPK3-expressing cell lines, LZRS-zeobased retroviral vectors were used (Rodriguez et al, 2016). To produce retroviral particles, Phoenix-Ampho cells were plated at a density of 3.5x10 6 cells in 10 cm dishes in 15 ml complete growth media. The next day, transfection was carried out with a mixture of 1.5 ml OptiMEM, 36 μl FuGENE HD (Promega E2311), 1.2 μg pMD.G (VSVG) plasmid and 10.8 µg of the retroviral vector, and media was replaced 24h after transfection. Two batches of retroviral particles were collected, respectively at 48 and 72 hours post-transfection. Retroviral particles were precipitated by incubating 0.45 µm syringe-filtered culture media in 150 mM NaCl and 5% PEG-8000 at 4°C overnight, followed by centrifugation at 3500 xg for 15 min.
Pellets were resuspended in 150 µl sterile PBS and stored at -80°C until use.
For retroviral transduction, between 1 and 3x10 5 cells were seeded into 6-well plates. The next day, cells were transduced using 1 ml virus-containing supernatant or 25 µl precipitated viral particles in the presence of 10 µg/ml polybrene in a total of 2 ml complete growth medium.
Cells were incubated with virus-containing supernatant for 24 h before replaced with complete growth medium to rest. 48-72 hours after transduction, cells were selected with 250 ng/µl zeocin (Invitrogen R250) until two passages after the complete elimination of non-transduced cells in the control well.

Generation of HCT116 cells with inducible GFP-SUB expression
To generate HCT116 / Tet-On-GFP-SUB / RIPK3-2xFV cells, HCT116 cells were first transduced with lentiviral particles generated from pLenti-CMV-Blast plasmids carrying the Tet Repressor gene. During selection with 5 µg/ml blasticidin (Thermo Fisher A1113903), single clones were isolated, and clone C4 with high levels of Tet-Repressor expression was used for the next steps.

Production of lentiviral particles
For the production of lentiviral particles, HEK293FT cells were plated at a density of 3.5x10 6 cells in 10 cm dishes in 15 ml complete growth media. The next day, they were transfected with a mixture of 1.5 ml OptiMEM, 36 μl FuGENE HD, 6 μg psPAX2 vector, 1.5 μg pMD.G (VSVG) and 4.5 μg lentiviral vector. 24 h after transfection, transfection reagent-containing media were replaced with 10 ml complete growth media to allow the secretion of lentiviral particles for 72 h. Virus-containing supernatant was filtered, and lentiviral particles were precipitated as described for retroviral particles or directly frozen at -80°C for preservation.

Generation of HT29 cells with Dox-inducible expression of ZBP1
For lentivirus production, HEK293T cells were transfected with C-terminally FLAG-tagged wild-type or Zα1α2-mutant human ZBP1-expressing transducing vectors in the doxycyclineinducible Tet-On pDG2 backbone (De Groote et al, 2016) together with the pCMV delta R8.91 gag-pol-expressing packaging plasmids and pMD2.G VSV-G-expressing envelope plasmid.

Generation of shRNA-mediated stable knockdown cell lines using lentiviral particles
Lentiviral particles were used to construct stable cell lines knocked down against mismatch control sequence (shMM, Sigma SHC002) or target genes. shRNA plasmids used for this study are pLKO.1-based targeting the following sequences: For the transduction of lentiviral particles, HCT116 / RIPK3-2xFV cells were seeded in 12-well plates at a density of 0.5-1x10 5 cells per well and transduced by incubation with 10-20 μl precipitated virus in 1 ml complete growth medium containing 10 μg/ml polybrene for 24 h.
Calu-3 cells were seeded in 10 cm dishes at a density of 7.5x10 5 cells per dish, and transduced by incubation with 750 µl virus-containing supernatant in 10 ml complete growth medium containing 4 μg/ml polybrene for 24 h. 72 h after transduction, cells were selected with puromycin as described above.
Calu-3 epithelial lung cancer cells were cultured in DMEM (Lonza) supplemented with 10% heat inactivated fetal calf serum, 200 IU/mL penicillin, 100 μg/mL streptomycin and 600 μg/mL L-glutamine prior to infection experiments. Calu-3 cells were seeded in flat-bottom 12-well plates (2x10 5 cells/well in 1 ml media) or flat-bottom 6-well plates (5x10 5 cells/well in 2 ml media). Upon reaching confluency of 50-70% 24-48 h after seeding, the cells were infected with SARS-CoV2 B.1 at MOIs of 0.5 or 2.0. The plates were incubated at 37°C and tilted every 15 minutes for 1 h to allow for virus adsorption. After 1 h, the media was replaced with fresh media. Supernatant and cell lysates were harvested after 16, 24, 48 and 72h respectively.
To inactivate any live virus prior to biochemical analyses, cell-free supernatant was incubated in 0.5% Triton-X (Sigma Aldrich 11332481001) for 30 minutes at room temperature. Cells for qPCR were lyzed in RNeasy lysis buffer (Qiagen 79254) containing 1% β-Mercaptoethanol and incubated for 10 minutes at room temperature. Cells for Western blotting were lyzed in RIPA buffer (Thermo Fisher 89901), supplemented with with 4x XT sample buffer (BioRad 161-0791) and XT Reducing Agent (BioRad 161-0792), and heated at 95°C for 5 minutes.
Virus-inactivated samples were used for subsequent analysis by ELISA, qPCR and Western blotting.
Supernatant was collected by centrifuging at 300 xg for 5 min or by filtering through 0.2 µm CA syringe filter. THP1 or primary neutrophils were resuspended in FBS-free RPMI media supplemented with 0.5% BSA to an approximate of 1 million/ml, and 0.1 ml were plated into each upper chamber of a 24-well plate (Corning 2421). 0.5 ml conditioned media or HT29 chemotaxis buffer were plated into the lower chamber. The chambers were incubated at 37°C , 5% CO2 for 3 hours or as indicated in the figure. Migrated cells were collected from the lower chambers and mixed with Countbright Absolute counting beads (Thermo Fisher C36950) to determine cell density together with the plating population to obtain an accurate plating number. The volume of collected cells was determined by the weight difference of the collection tube before and after collection, assuming density of 1 g/ml. The migration percentage is calculated by dividing migrated cell numbers by plated cell numbers.

Cell viability assay
The CellTitre-Glo ® 2.0 reagent (Promega G9242) was used to determine cell viability according to the manufacturer's protocol. Cells were seeded in opaque 96-well plates in 100 µl complete growth media at a density of 3.3x10 3 cells/well for 72-hour time courses or 1.3x10 4 cells/well for 24-hour treatment. The next day, cells were stimulated with the appropriate chemicals (doxycycline, Sigma, D3072) diluted in 5 µl OptiMEM. Relative viability is calculated by dividing the average of measurement luminescence values of technical replicates for each treatment condition by one reference condition as specified for each figure.
NF-κB induction levels were calculated by dividing the value of NF-κB luciferase activity by the value of control Renilla luciferase activity. Where indicated, NF-κB induction levels were further normalized to one reference condition.
Expression levels of proteins of interest in the passive lysis buffer lysates were determined by Western blotting as indicated.

Anti-FLAG immunoprecipitation
To perform anti-FLAG immunoprecipitation for ZBP1-associated proteins, cells were plated in 10 cm dishes at a density of 3-4x10 6 cells/dish, 3-6 dishes/condition, and stimulated with 500 ng/ml Dox the next day for 16 h. After stimulation, cells were washed twice with PBS and lysed in 500 µl/dish TBSN buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.5% NP-40) supplemented with protease inhibitor cocktail, phosphatase inhibitor cocktail and 50 mM N-Ethylmaleimide (NEM; Sigma Aldrich E1271) by ice incubation for a minimum of 15min.
Lysates were centrifuged at maximum speed for 10 minutes at 4°C to remove debris pellets.
After taking input samples, the lysate supernatant was precleared with 10 µl/dish mouse IgG-Agarose beads (Sigma A0919) for a minimum of 30min at 4°C with end-over-end rotation, before incubated with 10 µl/dish anti-FLAG M2 agarose beads (Sigma A2220) for a minimum of two hours with end-over-end rotation at 4°C. After enrichment, beads were washed five times with 1 ml TBSN buffer and eluted in 13.3 µl/dish 2xLSB by shaking at 95°C, 750rpm, for 10min.
Western blotting was used to probe for co-immunoprecipitated proteins. Equivalent of eluate materials from one 10 cm dish was loaded on each pull-down lane, and 1% input was loaded for reference except for ZBP1, for which 5% input was loaded.

Enrichment of Ub-conjugates
To enrich for ubiquitinated proteins, cells were seeded in 10 cm dishes and stimulated as Cells were lysed using 500 µl TUBE lysis buffer supplemented with protease inhibitor cocktail, phosphatase inhibitor cocktail, 50 mM NEM and 1 mM DTT following the same protocol as for FLAG immunoprecipitation. After taking input samples, centrifuged lysate supernatant was incubated with 10 µl/dish GST-1xUBA-or SUB-bound beads overnight at 4°C with end-overend rotation. After enrichment, beads were washed three times with TUBE lysis buffer and heated in 2xLSB for 10 minutes at 95°C for elution.
Equivalent of pull-down samples from one 10 cm dish and 1% input (except for ubiquitin blot, for which 5% input was loaded) were resolved on 4-12% NuPAGE gradient gels in NuPAGE MOPS-SDS running buffer and transferred onto PVDF membrane for Western blotting.

On-bead deubiquitinase treatment
On-bead deubiquitinase digestion was performed to probe for the existence and composition of ubiquitin chains in FLAG-immunoprecipitated or TUBE-enriched sections. After enrichment, beads were washed three times with lysis buffer and incubated shaking with or without 1 µM USP21 (R&D E-622-050) in 30 µl DUB buffer (20 mM HEPES, pH 7.5, 100 mM NaCl, 1 mM MnCl2, 0.01% w/v Brij-35, 5 mM DTT) for 1h at 30°C before LSB buffer was added to end the reaction.

Enzyme-linked immunosorbent assay (ELISA)
For ELISAs, cells were plated at a density of 1x10 5 per well in 24-well plates or 2x10 5 per well in 12-well plates, and stimulated the next day for 24h with the appropriate chemicals Interpolation of CT values was performed with the built-in software of the qPCR machine (Roche LightCycler 480 or Applied Biosystems 7500). Relative mRNA expression levels were calculated with the comparative CT method (Schmittgen & Livak, 2008).

SARS-CoV-2 single cell RNA-seq expression analysis
Processed single cell RNA TPM expression of (Ren et al, 2021) associated with GSE158055 were downloaded and analysed from http://covid19.cancer-pku.cn/#/summary. The patient TPM expression of a gene of interest was computed as the mean TPM expression of the patients' individual constituent single cells.

Statistical analysis for quantitative experiments
Statistical analyses were carried out in GraphPad Prism. The statistical details are described in figure legends, with n representing the number of biological replicates. Outliers were identified by ROUT method (Q=1%) and excluded from statistical analyses.
For the comparison between two conditions, an F-test was used to test for the statistical differences between the standard deviations (SDs) of the two conditions. A two-tailed unpaired Multiple t-tests were used when SDs from conditions were found to be significantly different and Brown-Forsythe and Welch ANOVA tests were not possible.
For grouped data or data from multi-factor experiment designs, two-way ANOVA and the following multiple comparison tests were used following Prism's recommendation on the choice of multiple comparison tests.
The distribution of the single cell RNA sequencing datasets were found significantly different from Gaussian distribution, and therefore they were analyzed using Kruskai-Wallis test and Dunn's multiple comparisons test.