ISG15 negatively regulates RIPK3-mediated cell death and viral pathogenesis Authors: Yi-Chieh Perng, Jessica MC Warsaw, Sachendra S Bais, Bradley E Hiller, Marissa C Locke, David J Morales-Heil, Chaoqun Li, Alissa R Young, Kristen J Monte, Robert E Schmidt, Douglas R Green, Yi-Nan Gong, Deborah J Lenschow Published: 2026 ABSTRACT SUMMARYNecroptosis, a form of programmed, inflammatory necrosis, plays an important role in viral-host defense and inflammation. The receptor-interacting protein kinase 3 (RIPK3)/mixed lineage kinase domain-like pseudokinase (MLKL) pathway mediates necroptosis. Yet, the mechanisms that control necroptosis to limit immunopathology are poorly understood. Here, we report that interferon-stimulated gene 15 (ISG15) negatively regulates RIPK3-mediated cell death, including necroptosis, and limits immunopathology during chikungunya virus (CHIKV) infection. ISG15-deficient mice infected with CHIKV display increased levels of necroptosis, resulting in elevated proinflammatory cytokine and chemokine production, leading to increased lethality. This dysregulated host response is fully prevented when MLKL or RIPK3 is ablated in Isg15−/− mice. Mechanistically, ISG15 non-covalently associates with the RIPK3 necrosome in an RIP homotypic interaction motif (RHIM)-dependent manner, regulating necroptosis downstream of CHIKV infection, tumor necrosis factor (TNF), lipopolysaccharide (LPS), and poly(I:C) stimulation. These results demonstrate a role for ISG15 in limiting immunopathology during infection by modulating necroptosis-dependent inflammation and pathogenesis. INTRODUCTION During an infection, tissue damage can be caused by the pathogen itself or by immune-system-driven mechanisms, termed immunopathology. Regulated cell death is inextricably linked to tissue damage, immunopathology, and inflammation during viral infection.1-3 It can be beneficial by eliminating infected cells to prevent viral dissemination but also detrimental if excessive or unregulated, leading to tissue damage and inflammation.1,2 Several forms of regulated cell death have been described, including apoptosis and necroptosis. Apoptosis is a caspase-dependent form of cell death and can be induced through intrinsic pathways caused by cellular stress or extrinsic pathways caused by ligation of death receptors on the cell surface. Apoptosis is typically considered “immunologically silent” cell death, with minimal release of pro-inflammatory cytokines or damage-associated molecular patterns (DAMPs).4,5 In contrast, necroptosis is a form of programmed, caspase-independent cell death associated with cytokine and DAMP production.6-11 It has been implicated in the pathogenesis of several diseases, including viral infections, neurodegenerative diseases, ischemia-reperfusion injury, and anti-tumor immunity.8,12-15 Several stimuli can induce necroptosis, including viral infection and activation of tumor necrosis factor (TNF) receptor 1 (TNFR1).4,7 Activation of TNFR1 signaling under conditions of absent/inactivated caspase-8 or surplus receptor-interacting protein kinase 3 (RIPK3) results in formation of the RIPK3 necrosome complex with the association and phosphorylation of RIPK1 and RIPK3 and subsequent phosphorylation of mixed lineage kinase domain-like pseudokinase (MLKL). Phosphorylated MLKL (pMLKL) causes loss of membrane integrity and massive release of DAMPs, cytokines, and chemokines.7,16-21 Inappropriate or excessive production of DAMPs can initiate a self-amplifying cycle of cell death and cytokine/DAMP release termed necroinflammation, which can lead to systemic inflammation, tissue loss, and multi-organ failure.1,5,22-25 Therefore, careful regulation of the induction and resolution of inflammatory cell death is critical during the host response to pathogens.26,27 Understanding these mechanisms could yield insight into new therapeutic opportunities for infectious and inflammatory disease. Interferon-stimulated gene 15 (ISG15) is rapidly and robustly induced by type I interferons (IFN-I) and has been implicated in the host response to pathogen infection.28-30 ISG15 is a ubiquitin homolog that can exist in three distinct states—released into the extracellular space, conjugated to target proteins, or free within the cell.30 All three forms have been implicated in its many actions, including its antiviral and antibacterial activity. Examples include conjugation of ISG15 (termed ISGylation) to host and viral proteins to limit pathogen replication, release of ISG15 from cells to stimulate IFN-γ production, and non-covalent interactions of intracellular ISG15 with proteins to regulate cellular functions such as IFN-I signaling.29,31-35 Mice lacking ISG15 are susceptible to many viruses (reviewed in Perng and Lenschow30). In some infections, ISG15 can directly disrupt viral replication, leading to increased tissue viral load in Isg15−/− mice.36 However, for other viruses, a direct antiviral effect has not been observed. In mouse models of influenza A and Sendai virus infection, Isg15−/− mice were more susceptible to infection than wild-type (WT) mice, despite similar viral loads.37,38 During these infections, Isg15−/− mice exhibited greater distal airway damage than WT mice.37 In a model of neonatal chikungunya virus (CHIKV) infection, Isg15−/− mice displayed increased susceptibility, with all Isg15−/− mice dying several days post-infection (dpi).39 Isg15−/− mice developed an exaggerated immune response, with increased pro-inflammatory cytokines and chemokines compared to WT mice.39 Together, these observations suggest that ISG15 can play an immunomodulatory role, protecting the host from tissue damage, inflammation, and lethality during viral infection. Here, we describe a function for ISG15 as a negative regulator of RIPK3-mediated cell death, including necroptosis, and demonstrate its critical role in regulating immunopathology and inflammation during CHIKV infection. Mice lacking ISG15 exhibited increased necroptosis and elevated, persistent proinflammatory cytokine and chemokine production, leading to increased lethality. The exacerbated immunopathology and lethality observed in Isg15−/− animals during CHIKV infection were fully rescued by Ripk3 or Mlkl deficiency. Mechanistic studies revealed that ISG15 directly interacts with the RIPK3 necrosome complex in a conjugation-independent manner to suppress cell death signaling downstream of TNF-, lipopolysaccharide (LPS)-, and poly(I:C)-induced cell death and during CHIKV infection. These results demonstrate a previously undefined role for ISG15 in limiting immunopathology during infection by modulating necroptosis-dependent inflammation and pathogenesis. RESULTS ISG15 deficiency results in increased cell death after CHIKV infection Previous work has demonstrated that ISG15 protects mice from lethality during CHIKV infection. Yet, the precise mechanism is unknown.37-39 CHIKV has been shown to induce apoptosis in fibroblasts in vitro, and a study found that some necroptotic genes were upregulated during infection, although whether CHIKV directly induces necroptosis is unclear.40 Since dysregulation of viral-induced cell death by ISG15 could impact disease pathogenesis, we examined whether ISG15 regulated the type or magnitude of cell death induced during CHIKV infection. WT and Isg15−/− murine embryonic fibroblasts (MEFs) were infected with CHIKV and analyzed for cell death. We found that CHIKV infection of MEFs resulted in minimal cell death in either genotype during the first 18 h post-infection (hpi) (Figure 1A). However, by 24 hpi, cells lacking ISG15 began to show increased cell death compared with WT cells, which continued to increase through 48 hpi, despite both genotypes displaying similar viral titers at all time points analyzed (Figures 1A and 1B). An analysis for cell death markers revealed modest levels of the P17 subunit of caspase-3 (apoptosis) and pMLKL (necroptosis) in lysates from WT-infected cells, while lysates from CHIKV-infected Isg15−/− MEFs displayed significantly increased levels of both pMLKL and caspase-3 (P17) (Figures 1C and 1D). Therefore, ISG15 deficiency results in increased apoptotic and necroptotic cell death during CHIKV infection in vitro. ISG15 deficiency results in increased cell death after CHIKV infection WT and Isg15−/− MEFs were infected with CHIKV (MOI 2.0). (A) Quantification of SYTOX Green+ WT and Isg15−/− MEFs over time. Data are means of six samples ± SEM (two-way ANOVA, Tukey’s test; *p < 0.05, ****p < 0.0001). (B) Supernatants were harvested, and viral titers were determined by plaque assay. Data are mean ± SEM (two-way ANOVA, Sidak’s test). (C) Representative blot of cell lysates from MEFs immunoblotted for caspase-3 (p17 subunit), pMLKL, total MLKL (tMLKL), ISG15, and β-actin. (D) Quantification of caspase-3 (p17), pMLKL, and tMLKL was normalized to β-actin. Ratios were calculated from three independent experiments by densitometry. Data are mean ± SEM (two-way ANOVA, Sidak’s test; **p < 0.01, ****p < 0.0001). ISG15 negatively regulates necroptosis in both human and murine cells Programmed cell death can be induced by multiple stimuli, including viruses, Toll-like receptor ligands, and signaling through TNF family members. Signaling through TNF receptors is one of the best-characterized pathways, resulting in the induction of either apoptosis or necroptosis, depending on the cellular context.2,4,41 Therefore, we evaluated the ability of ISG15 to regulate TNF-induced cell death. WT and Isg15−/− MEFs were stimulated with a high dose of TNF, a known inducer of apoptosis, and we observed similar levels of cell death in each genotype (Figure S1A). Additional triggers of apoptosis, including staurosporine and the SMAC mimetic LCL-161, also induced similar levels of cell death in Isg15−/− MEFs as in WT, suggesting that ISG15 does not modulate apoptosis induced by these stimuli (Figures S1B and S1F). In contrast, when cells were stimulated with TNF (T) in the presence of pan-caspase inhibitor Z-VAD-FMK (zVAD, Z), a known inducer of necroptosis (TZ), Isg15−/− MEFs exhibited a dramatic increase in cell death as compared to WT (Figures 2A and 2B). We next stimulated MEFs with TNF plus an SMAC mimetic, LCL-161 (S), in the presence or absence of zVAD to induce either apoptosis (TS) or necroptosis (TSZ). We observed that cell death was similar between TS-stimulated WT and Isg15−/− cells (Figure 2C). In contrast, addition of zVAD resulted in a marked increase in cell death in Isg15−/− MEFs as compared to WT (Figure 2D). In human HT-29 cells, in which ISG15 was deleted by CRISPR, we also observed increased necroptosis after stimulation with TSZ (Figure 2F) or SZ (Figure S1D) but no difference in apoptosis following stimulation with TS alone (Figure 2E). We also stimulated MEFs with staurosporine to induce apoptosis (Figures S1B and S1F) and staurosporine + zVAD to induce necroptosis (Figure S1G) and found that ISG15 suppressed necroptosis but not apoptosis. Finally, a similar increase in cell death was observed in TZ-treated Isg15-deficient murine lung epithelial spheroids (Figure 2G). Together, these results suggest that both human and murine ISG15 modulate cell death by known inducers of necroptosis in multiple cell types. Of note, the increased cell death in Isg15−/− MEFs and HT-29 cells was also fully prevented by small molecule inhibitors of RIPK1 (Nec-1s), RIPK3 (GSK’872), or MLKL (SYN-1215 or ne-crosulfonamide [NSA]) (Figure 2H and S1C-S1E). Further, we found that while TZ-stimulated WT and Isg15−/− cells exhibited similar levels of total RIPK1, RIPK3, and MLKL, we observed an earlier onset of phosphorylation of RIPK3 and MLKL in Isg15−/− MEFs as compared to WT MEFs (Figure 2I), consistent with increased necroptosis in Isg15−/− MEFs. Together, these results strongly suggest that ISG15 negatively regulates necroptosis in multiple murine and human cell types. ISG15 deficiency results in increased necroptosis (A and B) WT or Isg15−/− MEFs stimulated with TZ (50 ng/mL TNF, 50 μM zVAD) were analyzed for cell death (SYTOX Green+) (A). Representative IncuCyte images at 23 h (scale bars, 400 μm) (B). (C–F) Quantification of SYTOX Green+ WT and Isg15−/− MEFs (C and D) or HT-29 cells in which ISG15 was deleted by CRISPR (E and F). MEFs were stimulated with TS (50 ng/mL TNF, 2.5 μM LCL-161) (C) or TSZ (50 ng/mL TNF, 2.5 μM LCL-161, 50 μM zVAD) (D). HT-29 cells were stimulated with TS (10 ng/mL TNF, 100 nM LCL-161) (E) or TSZ (10 ng/mL TNF, 100 nM LCL-161, 25 μM zVAD) (F). Data are means of quadruplicate samples ± SEM (two-way ANOVA, Tukey’s test; **p < 0.01, ****p < 0.0001). (G) WT or Isg15−/− primary murine lung spheroids were stimulated with TZ (50 ng/mL TNF, 50 μM zVAD) for 45 h. Cell death was analyzed by LDH assay. Data are means of at least triplicate samples ± SD (one-way ANOVA, Tukey’s test; ****p < 0.0001). (H) Quantification of cell death (SYTOX Green+) of WT or Isg15−/− MEFs at 20 h post-treatment with TZ (50 ng/mL TNF, 25 μM zVAD) in the presence of 10 μM Nec-1s, 2 μM GSK’872, or 2 μM SYN-1215. Data are means of at least triplicate samples ± SD (two-way ANOVA, Sidak’s test; ****p < 0.0001). (I) Immunoblotting of RIPK1, RIPK3, pRIPK3, pMLKL, tMLKL, ISG15, and β-actin in WT and Isg15−/− MEFs at the indicated times post-TZ stimulation (50 ng/mL TNF, 25 μM zVAD). See also Figure S1. ISG15 regulates necroptosis independent of protein ISGylation ISG15 is a ubiquitin-like protein that mediates its functions through ISGylation or the actions of unconjugated ISG15.30 To determine which form of ISG15 modulates necroptosis, we monitored cell death in cells lacking UbE1L, the E1 enzyme required for ISGylation.36,42 Following TZ treatment, Isg15−/−, but not Ube1l−/− MEFs, exhibited increased necroptosis as compared to WT MEFs (Figure 3A), characterized by increased levels of pMLKL in cell lysates (Figure 3B), indicating that the phenotype is independent of conjugation. Furthermore, deletion of ISG15, but not UBE1L, in human HT-29 cells resulted in increased necroptosis following TSZ stimulation (Figures 3C and 3D; Figure S2A). Similar results were also observed in E1A/RAS-transformed MEFs (Figure S2B). Isg15−/−-transformed MEFs were then transduced with a lentivirus expressing either WT ISG15 (ISG15GG) or ISG15 in which the terminal di-glycine was mutated to di-alanine (ISG15AA), abrogating its ability to conjugate onto target proteins (Figure S2C).43 Reconstitution of ISG15-null cells with either ISG15GG or ISG15AA normalized the cell death phenotype observed in Isg15−/− cells (Figure 3E). Taken together, these results indicate that both murine and human ISG15 inhibit necroptosis in a conjugation-independent manner. ISG15 regulates necroptosis independent of protein ISGylation (A and B) WT, Ube1l−/−, or Isg15−/− MEFs were stimulated with TZ (50 ng/mL TNF, 25 μM zVAD). (A) Quantification of SYTOX Green+ over time and at 20 h post-stimulation. (B) Cell lysates at 6 h post-stimulation were immunoblotted for pMLKL, β-actin, and ISG15. (C and D) HT-29 cells in which hISG15 or hUBE1L has been deleted by CRISPR were stimulated with TSZ (10 ng/mL TNF, 100 nM LCL-161, 25 μM zVAD) and were (C) analyzed for cell death (SYTOX Green+) over time or at 12 h post-stimulation. (D) Cell lysates at 6 h post-stimulation were immunoblotted for pMLKL, β-actin, and ISG15 expression. (E) E1A/RAS-transformed WT and Isg15−/− MEFs were reconstituted with empty vector, WT ISG15 (LRLRGG), or conjugation-deficient ISG15 (LRLRAA). Quantification of SYTOX Green+ cells over time or at 24 h post-stimulation with TSZ (20 ng/mL TNF, 100 nM LCL-161, 25 μM zVAD). Data are means of at least quadruplicate samples ± SD (one-way ANOVA, Tukey’s test; ****p < 0.0001). See also Figure S2. ISG15 modulates necroptosis by targeting RIPK3 We next determined where ISG15 acts on the necroptosis pathway. Stimulation with TNF alone recruits adapter proteins to form TNF complex I, including TNFR1-TRADD-TRAF2, leading to cell survival through nuclear factor κB (NF-κB) activation or apoptosis through Fas-associated death domain protein (FADD) recruitment.41 Under certain conditions, such as caspase inhibition with zVAD, cytosolic complex IIb (necrosome) is formed, which includes RIPK1, RIPK3, and pMLKL, ultimately resulting in necroptosis.41 We found that TNF stimulation of either WT or ISG15-deficient MEFs resulted in no difference in activation of the NF-κB signaling pathway (Figure S3A) or induction of cell death (Figure S1A), suggesting that TNF complex I was unaffected by ISG15. We next evaluated the impact of ISG15 on other signaling pathways. TLR3 ligand poly(I:C) or TLR4 ligand LPS, in the presence of zVAD, induce necroptosis via the TLR-TRIF-RIPK3-MLKL axis, bypassing complex I.44,45 We therefore stimulated WT and Isg15−/− MEFs with either poly(I:C) (Figure 4A) or LPS (Figure 4B) plus zVAD and found that this resulted in increased cell death in ISG15-deficient cells as compared to WT (Figures 4A and 4B). Together, these results indicate that ISG15 regulates necroptosis downstream of multiple innate immune signaling pathways. ISG15 modulates necroptosis by targeting the RIPK3 necrosome complex (A and B) Quantification of SYTOX Green+ WT and Isg15−/− MEFs stimulated with 25 μM zVAD and 1,000 ng/mL poly(I:C) (A) or 1,000 ng/mL LPS (B). Data are means of quadruplicate samples ± SEM (two-way ANOVA, Tukey’s test; **p < 0.01). (C) Quantification of SYTOX Orange+ RIPK3-2Fv-NIH3T3 cells transfected with the indicated siRNA and stimulated with TZ (20 ng/mL TNF and 100 μM zVAD). Data are mean ± SEM (two-way ANOVA, Tukey’s test; ****p < 0.0001). (D) RIPK3-2Fv Mlkl−/−-NIH3T3 cells reconstituted with N-FLAG-MLKL, expressing either scramble (siSCR) or ISG15 knockdown (siIsg15) siRNA, were stimulated with 50 nM B/B. Cell lysates were immunoblotted for pMLKL and tMLKL. (E) RIPK3-2Fv-NIH3T3 or MLKL1-181-2Fv-NIH3T3 cells were transfected with the indicated siRNA. Transfected cells were treated with 50 nM B/B, then Annexin V+ and SYTOX Green− cells (orange box) were sorted, cultured with washout ligand (3 μM) for 6 h, then quantified by FACS. “Resuscitated” cells are circled in red. (F) Quantification of cell resuscitation. Data are means of at least triplicate samples ± SD (two-way ANOVA, Tukey’s test; ****p < 0.0001). See also Figure S3. We next used a previously described dimerization assay where cell death can be induced independently of upstream signaling and caspase inhibition.46 In this assay, RIPK3 was fused to two tandem FKBP-12 (2Fv)-binding domains, creating RIPK3-2Fv. Then, a dimerizer ligand B/B (AP20187) triggers oligomerization and activation of RIPK3-2Fv, leading to necroptosis independent of upstream signaling and caspase inhibition.46,47 RIPK3-2FV-NIH3T3 cells in which ISG15 expression had been small interfering (siRNA)-silenced were treated with TZ, and we observed increased cell death as compared to cells treated with scrambled siRNA (siSCR) (Figure 4C; Figures S3B and S3C). We then stimulated RIPK3-2Fv-NIH3T3 Mlkl−/− cells, which had been reconstituted with an N-flag-tagged MLKL, with B/B. Importantly, N-flag-tagged MLKL can still be phosphorylated by RIPK3 but cannot induce cell death, more readily allowing for detection of pMLKL in cells.48 We observed that levels of pMLKL were augmented following B/B-induced RIPK3 dimerization when ISG15 was silenced as compared to cells treated with siSCR (Figure 4D). Taken together, these results suggest that ISG15 regulates necroptosis at either the level of the RIPK3 necrosome complex (IIb) or downstream. To further define where ISG15 regulates necroptosis, another dimerizable construct was made by fusing the N-terminal region of MLKL (residues 1–181) to two tandem Fv-binding domains, creating MLKL1–181-2Fv. Then, ISG15 was silenced in either RIPK3-2Fv-NIH3T3 or MLKL1-181-2Fv-NIH3T3 cells, and a “resuscitation assay” was performed. Previous work demonstrated that following MLKL activation, there is a rapid Ca2+ influx into cells, followed by phosphatidylserine exposure on the cell surface, prior to complete membrane disruption. These dying necroptotic cells, which stain Annexin-V+ SYTOX Green−, can be resuscitated via Ca2+-influx-activated ESCRT-III-mediated plasma membrane repair if either RIPK3 or MLKL is inactivated by a dimerizer competitor washout (Figure 4E).49 Cells that are resuscitated become Annexin-V− SYTOX Green−, while cells that are not resuscitated become Annexin-V+ SYTOX Green+ (complete necroptosis with membrane rupture). RIPK3-2Fv-NIH3T3 or MLKL1-181-2Fv-NIH3T3 cells were treated with B/B, then Annexin V+ SYTOX Green− dying (but not lysed) cells were sorted and cultured with washout ligand for 6 hours. When ISG15 was silenced in RIPK3-2Fv-NIH3T3 cells, more cells underwent complete membrane rupture (Annexin-V+ SYTOX Green+), and resuscitation (Annexin-V− SYTOX Green−) was strongly in-hibited as compared to the siSCR control (Upper panels Figures 4E and 4F). In contrast, when ISG15 was silenced in MLKL1-181-2Fv-NIH3T3 cells, the number of cells that underwent complete membrane rupture (Annexin-V+ SYTOX Green+) or were resuscitated (Annexin-V− SYTOX Green−) was unaffected as compared to the siSCR control (lower panels, Figures 4E and 4F). These results suggest that ISG15 acts at the level of the RIPK3 necrosome rather than at the level of MLKL or events downstream that affect ESCRT-mediated repair of membranes.49 ISG15 modulates RIPK3-dependent apoptosis RIPK3 can also induce apoptosis independent of its pro-necroptotic kinase activity. In cells lacking MLKL, oligomerization of RIPK3-2Fv can induce apoptosis through a second pathway, resulting from the interaction of RIPK3 with RIPK1 via their RHIMs to induce a ripoptosome-like RIPK1, FADD, and cFLIPL-caspase-8 complex.50 RIPK3-2Fv-NIH3T3 Mlkl−/− cells were treated with B/B, and we observed increased apoptosis when ISG15 was silenced. Cell death was measured by Annexin V+ staining (Figure 5A), caspase-3 activity (Figure 5B), and cell morphology (Figure 5C). The addition of zVAD attenuated the increased cell death (Figure 5A). Some RIPK3 kinase inhibitors, including GSK’872, can induce a conformational change in RIPK3 that results in RHIM-mediated interactions with RIPK1, leading to caspase-8-mediated apoptosis.51 Treatment of RIPK3-2Fv-NIH3T3 Mlkl−/− cells with GSK’872 also resulted in increased apoptosis when ISG15 was silenced (Figures 5A-5C). Finally, a similar increase in cell death was observed in Isg15−/− primary MEFs treated with high doses of GSK’872 (Figure 5D). Together, these data strongly suggest that ISG15 regulates cell death through its action on the RIPK3-containing complex. ISG15 modulates cell death by targeting RIPK3 (A and B) RIPK3-2Fv-NIH3T3 Mlkl−/− cells were transfected with the indicated siRNA, then treated with 50 nM B/B or 20 μM GSK’872 in the presence or absence of 100 μM zVAD. Quantification of cell death (Annexin V+) (A) or apoptosis (caspase-3 glo) (B) was assessed at 8 h post-stimuli. (C) IncuCyte images of apoptotic cells from (B) (scale bars, 50 μm). White arrows indicate apoptotic cells. (D) WT or Isg15−/− MEFs were stimulated by 20 μM GSK’872 for 2 h, and cell death was quantified (Annexin V+). Data are means of at least triplicate samples ± SD (two-way ANOVA, Sidak’s test; ****p < 0.0001). (E and F) Cell lysates of untreated WT MEFs were immunoprecipitated with anti-ISG15 (E) or anti-RIPK3 (F) antibodies, followed by immunoblotting with indicated antibodies. Anti-GFP antibody was the pull-down control. (G) Cell lysates of untreated Ube1l−/− MEFs were immunoprecipitated with anti-ISG15 antibody, followed by immunoblotting with the indicated antibodies. (H) HEK-293T cells overexpressing either FLAG-tagged human RIPK1 or human RIPK3 with hISG15 were co-immunoprecipitated with anti-FLAG antibody, followed by immunoblotting with anti-human ISG15 antibody. For (E–H), the data are representative images of three independent experiments. Indeed, co-immunoprecipitation of ISG15 revealed that RIPK1 and RIPK3 interact with ISG15 in resting, WT MEFs (Figure 5E). Reciprocal co-immunoprecipitation using an anti-RIPK3 antibody also revealed this interaction (Figure 5F). This interaction was independent of ISGylation, since RIPK3 was similarly co-immunoprecipitated with ISG15 in Ube1l−/− MEFs (Figure 5G). Finally, human ISG15 also associated with human RIPK1 and RIPK3 in human HEK-293T cells that overexpressed these proteins (Figure 5H), demonstrating a cross-species conserved interaction between ISG15 and necrosome components RIPK1 and RIPK3. Regulation of RIPK3-mediated cell death by ISG15 requires an intact RHIM Next, we investigated how ISG15 regulates RIPK3’s activity. In the necrosome, both RIPK1 and RIPK3 can form oligomers via their RHIM, a conserved sequence approximately 18–22 amino acids in length with a core tetrad sequence (V/I)-Q-(V/I/L/C)-G, such as VQVG (hRIPK3) or VQIG (mRIPK3).52 RIPK3 homo-oligomerization or hetero-oligomerization with RIPK1 occurs via their RHIMs and is necessary for necroptosis.46 To evaluate the role that ISG15 might play in necrosome formation, WT or Isg15−/− MEFs were stimulated with TZ. At 3 h post-stimulation, RIPK3 co-immunoprecipitation revealed increased levels of RIPK1 associated with RIPK3 in Isg15−/− MEFs as compared to WT MEFs (Figure 6A). In addition, increased levels of pRIPK1 were co-purified with RIPK3 at 3 and 6 h post-stimulation in Isg15−/− MEFs as compared to WT MEFs (Figure 6A). These data suggest that ISG15 may regulate RHIM-mediated formation of RIPK1/3 oligomers, “the necrosome.” ISG15 regulates necrosome assembly in a RIPK3 RHIM-dependent manner (A) Immunoprecipitation using anti-RIPK3 antibody to pull down the necrosome from cell lysates of WT and Isg15−/− MEFs stimulated with TZ (40 ng/mL TNF, 20 μM zVAD). Lysates were immunoblotted with the indicated antibodies. (B and C) RIPK3-2Fv-NIH3T3 or RIPK3(ΔRHIM)-2Fv-NIH3T3 cells were transfected with either siIsg15 or siSCR. Resuscitation experiments were performed as described in Figure 4B. “Resuscitated” cells are circled in red. (C) Quantification of cell resuscitation. Data are means of at least triplicate samples ± SD (one-way ANOVA, Tukey’s test; ****p < 0.0001). (D) RIPK3-2Fv-NIH3T3 or RIPK3(ΔRHIM)-2Fv-NIH3T3 cells were transfected with the indicated siRNA and then treated with 20 μM GSK’872. Quantification of cell death (Annexin V+) was assessed at 8 h post-stimulation. Data are means of at least triplicate samples ± SD (two-way ANOVA, Sidak’s test; *p < 0.05, ****p < 0.0001). See also Figure S4. To dissect the specific role of ISG15 in RHIM-mediated oligomerization, we again employed the RIPK3-2Fv system. In this system, oligomerization of RIPK3-2Fv can be mediated by both Fv-B/B-Fv interactions and through RHIM interactions (Figure S4). The addition of washout ligand can effectively disassemble Fv-B/B-Fv interactions, but not RHIM interactions, resulting in sustained necroptotic signaling until the RHIMs cease interacting (Figure S4A). In contrast, in cells expressing a dimerizable RIPK3 mutant lacking the RHIM (RIPK3(ΔRHIM)-2Fv), oligomerization is only mediated by Fv-B/B-Fv interactions (Figure S4B). The addition of washout ligand disrupts Fv-B/B-Fv interaction, and since there are no RHIM interactions, this results in immediate cessation of necroptotic signals and cell survival. Consistent with the previous experiment (Figure 4E), in cells expressing RIPK3-2Fv, B/B induced oligomerization of both Fv-B/B-Fv and RHIMs of RIPK3, driving phosphatidylserine exposure and resulting in Annexin-V+ SYTOX Green− cells. Washout then deactivated Fv-binding domain interactions. However, signaling through the complex was maintained by RIPK3-RHIM interactions. Under these conditions, siSCR-silenced, but not ISG15-silenced, cells were resuscitated (upper panels, Figures 6B and 6C; Figure S4A). In contrast, in cells expressing the RIPK3(ΔRHIM)-2Fv mutant, B/B induced oligomerization of only the Fv-B/B-Fv domains.46 Washout then deactivated Fv-binding domain-mediated oligomerization, disrupting necroptotic signaling and resuscitating both WT and ISG15-deficient cells (lower panels, Figures 6B and 6C; Figure S4B). Finally, treatment of RIPK3-2Fv-Mlkl−/−-NIH3T3 cells with the RIPK3 inhibitor GSK’872 induced increased apoptosis when Isg15 was silenced as compared to siSCR-treated cells. However, apoptosis was not induced in RIPK3(ΔRHIM)-2Fv-Mlkl−/−-NIH3T3 cells, with or without Isg15 silencing (Figure 6D). Altogether, these results demonstrate that ISG15 regulates necroptosis at the level of RIPK3 by interfering with assembly of the RIPK3 oligomer and requires an intact RHIM to mediate its effects. ISG15 restricts CHIKV-induced hyperinflammation and viral pathogenesis by inhibiting MLKL-mediated necroptosis CHIKV is a mosquito-transmitted re-emerging alphavirus that clinically presents as febrile inflammatory arthritis and myositis.53 In neonates, CHIKV can also present with systemic cytokine responses and encephalitis.54 Previous work has shown that neonatal mice lacking ISG15 rapidly succumbed to CHIKV infection and displayed markedly elevated cytokine and chemokine levels, although the reason for this response is unknown.39 We observed increased cell death in CHIKV-infected Isg15−/− MEFs (Figure 1). To evaluate further, we generated MEFs lacking both ISG15 and either RIPK3 or MLKL. Upon CHIKV infection, the increased cell death observed in Isg15−/− MEFs was rescued when either RIPK3 or MLKL was deleted (Figure 7A). We therefore sought to determine whether ISG15-mediated regulation of necroptosis during CHIKV infection contributes to disease pathogenesis. ISG15 restricts virus-mediated hyperinflammation and pathogenesis by modulating MLKL-mediated necroptosis during CHIKV infection in vivo (A) Quantification of SYTOX Green+ WT, Isg15−/−, Isg15−/−;Ripk3−/−, and Isg15−/−;Mlkl−/− MEFs infected with CHIKV (MOI 2.0). Cell death was analyzed at 48 h post-infection. Data are means of quadruplicate samples ± SD (one-way ANOVA, Tukey’s test; ****p < 0.0001). (B–D) WT (n = 38), Ube1l−/− (n = 22), Isg15−/− (n = 17), Ripk3−/− (n = 15), Mlkl−/− (n = 20), Isg15−/−;Ripk3−/− (n = 19), and Isg15−/−;Mlkl−/− (n = 27) mice were infected with 2 × 105 PFU CHIKV-LR2006 intradermally at 9 days of age. (B) Injection site tissue was collected at 4 days post-infection (dpi) and immunoblotted with anti-ISG15 or anti-pMLKL antibodies. (C) Mice were monitored for lethality until 20 dpi (Mantel-Cox and Gehan-Breslow-Wilcoxon tests; ****p < 0.0001). (D) Sera were collected at 10 dpi. Cytokines and chemokines were measured by Bio-Plex assay. Each graph represents at least two independent experiments with 5–7 pups per genotype. Data are means ± SEM (mixed-effects analysis, Tukey’s test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). See also Figure S5. Neonatal mice were subcutaneously infected with CHIKV-LR, and at 4 dpi, tissue from the injection site was analyzed for evidence of necroptosis. While pMLKL was not detected in tissues isolated from WT mice, we detected significant levels of pMLKL in tissues from Isg15−/− mice (Figure 7B), suggesting that the loss of ISG15 increases necroptosis during CHIKV infection. To determine whether this increased necroptosis was responsible for the heightened inflammatory responses and lethality, we infected 9-day-old neonates with CHIKV-LR. While WT mice displayed no clinical signs of disease and all survived infection, 100% of Isg15−/− mice succumbed to infection, with the majority dying between 8 and 12 dpi (Figure 7C). As previously reported, this phenotype was independent of ISG15 conjugation since mice lacking UbE1L also displayed no lethality (Figure 7C).39 We next crossed Isg15−/− mice to either Ripk3−/− or Mlkl−/− mice. While neither Ripk3−/− nor Mlkl−/− mice displayed increased lethality as compared to WT mice following CHIKV infection, we strikingly found that both Isg15−/−;Ripk3−/− and Isg15−/−;Mlkl−/− mice were fully protected from CHIKV-induced lethality (Figure 7C), suggesting that the increased necroptosis induced by CHIKV infection in Isg15−/− mice is responsible for the increased lethality. Proinflammatory cytokines, including TNF and interleukin (IL)-6, along with several chemokines, were induced to similar levels in all genotypes early in infection (Figure S5A).39 Strikingly, at 10 dpi, Isg15−/− mice continued to show markedly elevated levels of serum IL-6, TNF, and chemokines, including CCL-2, CCL-5, CCL-11, and CXCL-10 compared to WT mice (Figure 7D). In contrast, IL-1β, IL-1α, leukemia inhibitory factor (LIF), and IL-13 were modestly induced and did not differ among genotypes (Figure 7D; Figure S5A). Like the lethality phenotype, the elevated cytokine and chemokine levels observed in Isg15−/− mice were reduced to WT levels in Isg15−/−;Mlkl−/− mice (Figures 7D; S5A), indicating that the exaggerated cytokine responses observed in Isg15−/− mice were also due to increased necroptosis. Further characterization revealed that viral titers in the serum of infected mice were markedly elevated at 2 dpi and were largely cleared by 7 dpi in all genotypes (Figure S5B). In the spleen, viral titers were higher in Isg15−/− mice than in WT mice early on, but replicating virus was cleared by 10 dpi. In contrast, in the muscle, Isg15−/− mice had significantly elevated viral titers as compared to WT mice at 4 , 7, and 10 dpi and failed to clear replicating virus. Most striking was the presence of replicating virus in the brains of only the Isg15−/− mice, which was detected as early as 2 dpi and continued to increase throughout infection. In all tissues, the increase in viral titers observed in Isg15−/− mice was normalized upon Mlkl deletion (Figure S5B). Analysis of brain sections at 10 dpi for CHIKV staining revealed foci of viral RNA+ cells with limited tissue damage throughout the cortex of Isg15−/− mice, while both WT and Isg15−/−;Mlkl−/− mice displayed no evidence of CHIKV viral RNA staining (Figures S6A-S6C). While a robust cellular infiltrate was not detected in Isg15−/− mice at this time point, staining for microglial activation with the Iba1 marker revealed profuse microglial activation in Isg15−/− brains, while minimal Iba1 staining was detected in WT or Isg15−/−;Mlkl−/− brains (Figures S6D-S6I).55 Iba1 staining revealed a striking pattern of microglial activation not only in areas that coincided with viral RNA staining but also in areas within the parenchyma in which viral RNA was not detected (Figures S6J-S6O). Together, these results demonstrate that, in the absence of ISG15, neonatal CHIKV infection induces an exaggerated necroptotic response. This results in increased viral replication, persistence, and dissemination, especially in the muscle and brain, and an exaggerated inflammatory response, including elevated, sustained proinflammatory cytokines and chemokines, which ultimately results in increased lethality. In all cases, deletion of MLKL from the ISG15 background prevented this phenotype. These findings suggest that ISG15 functions as an important negative regulator of MLKL-mediated necroptosis and plays a key role in maintaining host homeostasis during viral infection by modulating necroptosis-dependent pathogenesis. DISCUSSION Type I interferons (IFN-I) are key inflammatory mediators produced by the host during viral infections that carry out their functions through the induction of hundreds of interferon-stimulated genes (ISGs). Recent studies have found that the presence of neutralizing autoantibodies against IFN-I can underlie severe forms of viral infections, including Ross River virus, a member of the Togaviridae family closely related to CHIKV, and the development of encephalitis as seen in West Nile virus and tick-borne encephalitic viruses.56-58 Previously, it was shown that IFN-I signaling can also sensitize cells to necroptosis.59-61 However, the ISGs that mediate these effects are poorly characterized. In this study, we found that ISG15 acts as a key negative regulator of necroptosis in multiple murine and human cells. This previously unrecognized immunomodulatory function of ISG15 prevents excessive necroptosis and inflammation during CHIKV viral infection, including the development of encephalitis. We observed that CHIKV induced necroptosis and apoptosis in MEFs. While induction of apoptosis by alphaviruses has previously been defined, induction of necroptosis has not been directly demonstrated.62-64 The mechanism by which CHIKV activates these cell death pathways remains unknown and warrants further investigation. Interestingly, despite observing both apoptosis and necroptosis induced by CHIKV, deleting Mlkl in Isg15−/− MEFs normalized cell death to WT levels (Figure 7A). CHIKV infection has been shown to delay apoptosis via viral-induced autophagy, and recent studies have shown that activated MLKL blocks autophagosome formation.62,65 Increased activated MLKL in CHIKV-infected Isg15−/− MEFs could attenuate viral-induced autophagy through this mechanism and induce apoptosis. Therefore, deletion of MLKL in Isg15−/− MEFs normalized both necroptosis and apoptosis, suggesting that the increased apoptosis is due to an MLKL-mediated process, such as blocked formation of the autophagosome. We next determined that ISG15 regulates necroptosis at the level of the RIPK3 necrosome complex. ISG15 interacts with the RIPK3 necrosome and appears to inhibit its assembly. Whether this is mediated by a direct interaction between ISG15 and RIPK3, by ISG15-regulated modification of RIPK3, such as its ubiquitination, which is needed for necrosome assembly, or by alteration of necrosome composition by ISG15 remains an important question to be investigated. ISG15 regulation of RIPK3 does depend on the RIPK3 RHIM, as ISG15 promotes disassembly of RHIM-mediated interactions. In the presence of ISG15, upstream necroptotic signals—such as those propagated through RIPK1—must exceed the threshold imposed by ISG15’s negative regulation to initiate downstream signaling. Consequently, in the absence of ISG15, RIPK1–RIPK3 necrosome formation is accelerated and enhanced (Figure 6A). This model may also explain why RIPK3-2Fv-expressing cells exhibit only moderate MLKL phosphorylation with B/B (Figure 4D), which can be further augmented upon ISG15 silencing. These findings suggest that Fv-B/B-Fv-induced RIPK3 oligomerization in this system still needs to over-come ISG15-mediated suppression of RHIM-mediated complex formation. While it remains to be determined whether ISG15 directly binds the RHIM, such binding is not required for its regulatory function on RHIM. In all conditions we tested, ISG15 does not appear to regulate apoptosis when the RIPK3 RHIM is not involved. Staurosporine-induced apoptosis is known not to require RIPK3,66 whereas apoptosis induced by TS may or may not involve RIPK3, given contrasting studies on its requirement in the literature.67-69 However, even if TS-induced apoptosis requires RIPK3, the RIPK3 RHIM is known not to be required.67 Our data show that ISG15-mediated regulation of RIPK3 depends on the RIPK3 RHIM. Therefore, the absence of ISG15 regulation of TS-induced apoptosis supports our conclusion that ISG15 regulates RIPK3-RHIM-mediated cell death. Further investigation will be needed to define the impact of ISG15 on regulation of additional apoptotic stimuli involving RIPK1 or RIPK3, along with the role of their respective RHIMs. The regulation of necroptosis by ISG15, both in vitro and in vivo, was mediated by unconjugated ISG15. Unconjugated ISG15 can function through non-covalent interactions with proteins, as seen in human ISG15 binding to Usp18 to limit its degradation,31 or it can be released from cells and function as a cytokine, stimulating IFN-γ production by binding to LFA-1.70 During CHIKV infection of mice, we cannot differentiate between intracellular ISG15 or released ISG15 regulating necroptosis. However, since LFA-1 expression is restricted to lymphocytes, the increased necroptosis observed in the cell lines evaluated was not due to binding of ISG15 to its known receptor. While an alternative receptor for released ISG15 could exist, we favor its regulation of cell death through non-covalent interactions with the RIPK3 necrosome as revealed in our biochemical experiments. Notably, many viruses, including coronaviruses, encode deISGylases that cleave and release ISG15 from ISGylated proteins.30 This viral-mediated deISGylation facilitates intracellular accumulation and release of unconjugated ISG15.32 While the impact of these viral evasion strategies on targets of ISGylation has been explored,71 an important consequence of these immune evasion strategies is an increase in levels of intracellular ISG15, which could impact additional cellular functions such as IFN signaling and RIPK3-mediated inflammation and cell death. RIPK3 plays a critical role in the induction of necroptosis but also controls non-necroptotic inflammation, including activation of the NLRP3 inflammasome and NF-κB signaling.72 During West Nile virus infection, RIPK3 restricts viral pathogenesis by promoting neuroinflammation, independent of cell death.12 In contrast, during CHIKV infection, we observed increased immunopathogenesis and viral dissemination to the brain that was detrimental to host survival in Isg15−/− mice. This increased pathogenesis was completely dependent upon RIPK3-mediated necroptosis, as deletion of RIPK3 or MLKL fully prevented the increased pathogenesis observed in Isg15−/− mice. Necroptosis is known to damage the blood-brain barrier during systemic inflammation due to ischemic injury or sepsis.73,74 If increased necroptosis caused damage to the blood-brain barrier in Isg15−/− mice, this could provide a route by which CHIKV enters the brain. Given recent studies highlighting autoantibodies that target type I IFNs, and their association with an increased risk of encephalitis, studying the role of ISG15 in these other viral models will be important. In addition, these findings suggest that maintaining strict control of necroptosis induction and subsequent inflammation is critical for host homeostasis during pathogen infection. Given the many cellular processes that RIPK3 regulates, it will be important to determine if ISG15 regulates additional RIPK3-dependent processes. Our study elucidates the role of ISG15 as a negative regulator of necroptosis to restrict immunopathogenesis. Interestingly, Isg15−/− mice develop normally without chronic inflammatory symptoms, suggesting that ISG15 limits host damage caused by pathogen-induced necroptosis. However, patients with ISG15 deficiency have recently been identified and have a systemic IFN-I inflammatory signature with necrotizing skin lesions, among other symptoms.75,76 Our data show that both murine and human ISG15 modulate necroptosis. However, murine ISG15, unlike human ISG15, does not negatively regulate IFN-I signaling.31 This observation explains why patients with ISG15 deficiency have systemic IFN-I signatures, while this is not observed in ISG15−/− mice. However, the etiology of the necrotic skin lesions is unknown. Our work suggests that dysregulated cell death pathways may be present in patients with ISG15 deficiency, which could contribute to pathogenesis of the necrotic skin lesions. Since murine ISG15 does not regulate IFN-I signaling, it may be an ideal tool for studying pathogenesis mediated by the ISG15-necroptosis axis. Further studies will help to elucidate the role of ISG15-mediated regulation of the RIPK3-MLKL signaling axis in the pathogenesis of other diseases and may reveal novel therapeutic interventions for diseases characterized by excessive necroptosis and inflammation. Limitations of the study In this study, we evaluated several stimuli of apoptosis and necroptosis. We show that ISG15 regulates cell death that requires the RHIM of RIPK3. Further investigation is needed to define the impact of ISG15 on the regulation of additional apoptotic and necroptotic stimuli and to determine if ISG15 regulates other forms of inflammatory cell death. While we demonstrate, using co-IP, that ISG15 interacts with the RIPK3 necrosome, we do not know whether ISG15 directly binds RIPK3 or other components of the complex. Also, the precise mechanism by which ISG15 regulates necrosome assembly remains unresolved and warrants further investigation. Finally, additional disease models in which necroptosis is implicated should be studied to define the impact of ISG15 in a variety of diseases. STAR★METHODS EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS Compounds, antibodies, and reagents Antibodies used in this study were: cleaved caspase 3 (5A1E, Cell Signaling Technology, 9664), phosphor-MLKL (mouse, EPR9515(2), ab196436, abcam), MLKL (mouse, Sigma, MABC604), phosphor-MLKL (human, ab187091), actin (ab6276, abcam), ISG15 (generated in-house), RIPK1 (BD 610458), RIPK3 (ab62344, abcam), phospho-RIPK3 (ab205421, abcam), phosphor-RIPK1 (Cell Signaling Technology 31122), phosphor-NF- kB (Cell Signaling Technology, 3033S), IBA1 (Proteintech, 10904-1-AP), UBE1L (Proteintech, 15818-1-AP), and appropriate secondary antibodies from LICORbio. The reagents LCL-161, staurosporine, GW806742X (SYN-1215), zVAD-FMK, Nec-1s, and GSK’872 were purchased from MedChemExpress. Other reagents include ne-crosulfonamide (Invitrogen), poly(I:C) (Invivogen), LPS (Invitrogen), murine TNF (R&D Systems), human TNF (R&D Systems), mouse IFN-β (Biolegend), human IFN-β (PeproTech). The B/B dimerizer and washout ligand were purchased from Takara Bio. Plasmids Wild-type murine ISG15 (ISG15GG) and conjugation-deficient ISG15 (ISG15AA) were cloned into the pTRIP-eYFP vector. Cell lines Primary MEFs were generated by harvesting embryos from WT, Ube1l−/−, Isg15−/−, Mlkl−/−, Isg15−/−;Mlkl−/−, Ripk3−/−, and Isg15−/−;Ripk3−/− female mice 13–15 days after the appearance of the copulation plug. WT and Isg15−/− MEFs were transformed by Adenovirus E1A to acquire immortalized cells (WT and Isg15−/− E1A/RAS transformed MEFs). NIH3T3cells were purchased from ATCC (#CRL-1658). HEK-293T cells were purchased from ATCC (#CRL-3216). The cells mentioned above were cultured at 37°C, 5% v/v CO2 in a humidified incubator in Dulbecco’s Modified Eagle medium (Corning) supplemented with 10% FBS (HyClone), 1% Penicillin-Streptomycin (Corning), 10 mM HEPES (Corning), 1X non-essential amino acid (Sigma), and 2 mM L-glutamine (Corning). The human colorectal carcinoma line HT-29 was obtained from ATCC (#HTB-38) and maintained in McCoy’s 5a medium (Gibco) supplemented with 10% FBS (HyClone), 1% Penicillin-Streptomycin (Corning), 10 mM HEPES (Corning), 1X non-essential amino acid (Sigma), and 2 mM L-glutamine (Corning). Primary murine lung spheroids were derived from WT or Isg15−/− mice and cultured in 50% 3dGRO L-WRN conditioned media (Sigma-Aldrich) containing 10 μM ROCK inhibitor (R&D Systems, #1614) and 10 μM TGFβ inhibitor (R&D Systems, #1254).81 NIH3T3 cells and their derivatives are from historical stocks in Dr. Douglas R. Green’s lab. Mycoplasma contamination was detected using an in-house PCR assay. All other immortalized cell lines were tested for mycoplasma contamination upon arrival in the lab and confirmed to be mycoplasma negative. The cell lines were not authenticated by STR profiling. For primary MEFs, all cell lines used in all experiments were used within 5 passages, with growth rate and viability monitored throughout. CRISPR generation of ISG15−/− HT-29 cell line To generate the ISG15−/− HT-29 cell line, a sequence for CRISPR interference targeting human ISG15 was found and validated for off-target effects by using the CRISPR guide design tools at CRISPR.mit.edu. The sequence preceding the PAM motif used is as follows: GCACGCCTTCCAGCAGCGTC. The sense and antisense oligos were designed to incorporate the restriction enzyme site Bbs1 of pX458 bicistronic expression vector expressing Cas9 and synthetic single-guide RNA.82 The oligo DNAs were annealed, phosphorylated, and incorporated into the pX458 vector linearized with the Bbs1 restriction enzyme. HT-29 cells were transfected with this ISG15-targeting pX458 construct with Lipofectamine 2000 reagent (Thermo Scientific) according to the manufacturer’s instructions. The cells were selected by puromycin and sorted for the GFP+ population. The sorted GFP+ cells were subsequently re-plated to a 96-well plate at a density of one cell per well. Individual colonies were expanded, and the expression of ISG15 was confirmed with an anti-murine ISG15 antibody. The deletion of the ISG15 loci in the genome of selected clones was confirmed by targeted sequencing (Genome Engineering & Stem Cell Center (GESC@MGI), Washington University). The clones were tested for mycoplasma contamination and confirmed to be negative. CRISPR generation of UBE1L−/− HT-29 cell line The GESC@MGI generated the UBE1L−/− HT-29 cell line. Targeted genomic deletions were generated using a CRISPR/Cas9-based genome editing approach. Guide RNAs were designed using established algorithms to reduce predicted off-target effects and were synthesized as single guide RNAs. Cells were electroporated with Cas9–sgRNA ribonucleoprotein complexes using the Lonza 4D-Nucleofector Core and X units, then allowed to recover as pooled populations. Genomic editing outcomes were assessed by next-generation sequencing (NGS) using primers spanning the targeted deletion region. Cell pools containing the desired deletions were subjected to single-cell sorting, expanded to clonal populations, and re-genotyped to confirm deletion status. The clones were tested for mycoplasma contamination and confirmed to be negative. Deletion of UBE1L at the protein level in individual clones was confirmed by immunoblotting with an anti-human UBE1L antibody. siRNA knockdown “On-target” siRNA (smart-pool) against murine ISG15 or scramble siRNA were purchased from Dharmacon. siRNA transfection was performed by Lipofectamine RNAiMax (Life Technologies) on NIH3T3cells. RNAi efficiency was assessed by real-time PCR at 60 h post-transfection. Resuscitation assay was performed at 72 h post-transfection. Lentiviral transduction to reconstitute ISG15 in transformed MEFs Lentivirus supernatants were prepared after co-transfection into HEK-293T cells of a pTRIP series plasmid (pTRIP-ISG15AA-eYFP to express conjugation-deficient murine ISG15 (ISG15AA), pTRIP-ISG15GG-eYFP to express wild-type murine ISG15 (ISG15GG), pTRIP-FLUC-eYFP as a control plasmid) with pVSV-G (expressing the vesicular stomatitis virus envelope protein G (a gift from Bob Weinberg (Addgene plasmid # 8454)) and pCMV.DR.8.91 (expressing all necessary lentivirus helper functions) using polyethylenimine reagent (Polysciences). WT and Isg15−/− E1A/RAS transformed MEFs were transduced with lentivirus expressing ISG15AA, ISG15GG, or control plasmid. Single clones were sorted by eYFP and subsequently re-plated to a 96-well plate at a density of one cell per well. Individual colonies were expanded, and the expression of ISG15 was confirmed by immunoblotting with an anti-murine ISG15 antibody. HEK-293T cells HEK-293T cells were seeded in a 10-cm cell culture plate at a 50–70% confluency one day prior to transfection. 12–24 μg of plasmid DNA and 48 μL linear polyethylenimine-based transporter 5 (Polysciences) were mixed with 1 mL OPTI-MEM (Invitrogen) and incubated at room temperature for 20 min. The mixture was then added to 10 mL complete medium containing 2% fetal bovine serum and applied to the cells. Cell lysates were harvested at 72–96 h post-transfection for subsequent analyses as out-lined above. Mice Mice were bred and maintained at Washington University School of Medicine in accordance with all federal and University guidelines. WT C57BL6/J mice were purchased from Jackson Laboratory and bred at Washington University. Isg15−/− mice (provided by Klaus-Peter Knobeloch, University Clinic Freiburg, Germany), Ube1l−/− mice (provided by Dong-Er Zhang, University of California, San Diego), Mlkl−/− (provided by James Murphy, Walter and Eliza Hall Institute of Medical Research), and Ripk3−/− mice (provided by Vishva Dixit, Genentech) were generated as previously described.18,78-80 The Isg15−/−;Mlkl−/− and the Isg15−/−;Ripk3−/− mice were generated by crossing the Isg15−/− mouse with Mlkl−/− or Ripk3−/− mice. Ripk3−/−, Isg15−/−, Ube1l−/−, Mlkl−/−, and Ripk3−/− mice were fully backcrossed to C57BL6/J (>99.72, 99.93, 99.90%, and 99.66%, respectively) by congenic SNP analysis through Taconic Laboratories. All mouse experiments were performed with random assignment, and controls were age- and sex-matched. Both male and female mice were used, and mice were infected at nine days of age. Viral infection studies All cell and animal work with live CHIKV was performed in a biosafety level 3 facility and followed guidelines established by the Environmental Health and Safety Committee at Washington University School of Medicine. Animal studies were carried out in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocols were approved by the Institutional Animal Care and Use Committee at the Washington University School of Medicine (assurance number A3381—01). In vivo: Recombinant CHIKV La Réunion strain (LR2006) was used for neonatal infection. Viral stocks were generated as previously described.39 Neonates were infected at nine days of age with 2 × 105 PFU viruses diluted in 10 μL of phosphate-buffered saline (PBS) and injected intradermally (i.d.) into the left upper pectoral area to mimic a mosquito bite. For survival experiments, pups were monitored daily for 20 days post-infection (dpi). Litters were weight matched at the initiation of the experiments (for infections at 9 days of age, pups weighed between 3.5 and 4.8g). In vitro: Primary MEFs were seeded in 96-well plates or 6-well cell culture plates at 70–90% confluency one day prior to infection. The cells were infected with virus in serum-free DMEM for 1 h, with gentle rocking, in a humidified tissue culture incubator at 37°C and 5% CO2. After infection, the inoculum was removed and replaced with growth medium supplemented with 2% FBS. Cell death was monitored in the 96-well plates via Sytox Green uptake. The 6-well plates were harvested for western blot at the indicated timepoints. Cell culture media from the 6-well plates was harvested for plaque assays at the indicated timepoints. METHOD DETAILS Co-immunoprecipitation MEFs 5 × 107 MEF cells or HEK-293T cells were lysed in 1 mL IP lysis buffer (Thermo Scientific) supplemented with protease (Roche) and phosphatase inhibitor cocktail (Millipore-Sigma). Cell lysates were incubated at 4°C for 15 min. One aliquot of the cell lysates was saved as the input control and boiled in the LDS sample buffer in the presence of sample reducing agent (Thermo Scientific). The remainder was clarified by centrifugation at 10,000×g at 4°C for 15 min. The supernatant was incubated at 4°C overnight with anti-FLAG magnetic agarose or other pulldown antibodies conjugated with magnetic beads using Dynabeads Antibody Coupling Kit (Thermo Scientific). The following day, the beads were washed four times with 1 mL PBS/150 mM NaCl (for anti-FLAG magnetic agarose) or IP lysis buffer (for other pulldown antibodies). The immunoprecipitants were eluted by boiling in reducing sample buffer for 5 min. Immunoblotting analysis Electrophoresis of proteins was performed using the NuPAGE (Invitrogen) or Bolt system (Invitrogen) according to the manufacturer’s protocol. Briefly, cell lysates were collected and lysed with either RIPA buffer (Cell Signaling) or Pierce IP lysis buffer (Thermo Scientific) in the presence of 4X NuPAGE LDS Sample Buffer (Invitrogen) and sample reducing agent (not added for samples used in non-reducing gels). Proteins were separated on a precast Bis-Tris gel and transferred onto Immun-Blot Low PVDF membranes (Bio-Rad). The membranes were blocked by Odyssey blocking buffer (Li-Cor) and were incubated with the primary antibodies at a 1:1000 dilution at 4°C for 16 h. The blots were washed 5 times with 1X TBST (Tris-Buffered Saline, 0.1% TWEEN 20) and then incubated with the appropriate secondary fluorescent antibodies (Li-Cor) at room temperature for one hour. Membranes were washed, and images were acquired with an Odyssey CLx Imaging system (Li-Cor) and analyzed with Image Studio software (Li-Cor). Cell death assays SYTOX Assay: The real-time cell death events were analyzed by using a 2-color IncuCyte ZOOM in-incubator imaging system (Essen Bioscience) or a Cytation 7 (BioTek). Dead cells were detected by uptake of the cell-impermeable dye SYTOX Green (Invitrogen). The total cell numbers were estimated by calculating the cell numbers of quadruplicate wells stained with cell-permeable dye SYTO24 or SYTO16 (Invitrogen). The data was shown as the percentage of cell death (SYTOX positive cells/SYTO24(SYTO16) positive cells). For IncuCyte quantification, the percentages of SYTOX Green+ cells were determined by IncuCyte Live Cell analysis and for Cytation 7 quantification, the percentages of SYTOX Green+ cells were determined by BioTek Gen5 software. LDH Assay: Culture supernatants were collected, and cell death was quantified by LDH Cytotoxicity Assay (Promega) according to the manufacturer’s instructions. Caspase activity assay: Cell culture caspase-3/7 activity was measured by Caspase-Glo Assay (Promega) according to the manufacturer’s instructions. Resuscitation (dimerizable RIPK3 and MLKL) assay Resuscitation assay has been described previously.49,77 In brief, RIPK3-2Fv-NIH3T3, RIPK3 (ΔRHIM)-2Fv-NIH3T3, or hMLKL1-181-2Fv-NIH3T3 cells (treated with indicated siRNA) were trypsinized and resuspended in DMEM culture media (10% FBS containing) and stimulated with 50 nM B/B for about 40 min at 37°C, with gentle flicking the tube about every 5 min. Annexin V-APC (1:40, eBio-science) and 50 nM SYTOX Green (Invitrogen) were then added to label the cells for 2 min at room temperature. Stained cells were sorted by Reflection (i-Cyt, SONY). The sorted Annexin V+, SYTOX Green− cells were re-plated with complete media alone plus 3 μM washout ligand (washout, Takara) for 7 h. The cells were then resuspended and analyzed for resuscitation efficiency. To assess resuscitation, cells were trypsinized, resuspended in 10% FBS containing DMEM, and stained with Annexin V-APC (1:50-100, eBio-science), 50 nM SYTOX Green (Invitrogen). Data were collected by flow cytometry using FACSCalibur systems (BD Biosciences) and analyzed by FlowJo Collectors’ Edition software (Tree Star). The percentages of differently labeled cells were also calculated by FlowJo (Tree Star). Relative resuscitation efficiency was calculated as the percentage of Annexin V− SYTOX Green− cells transfected with scrambled siRNA (siSCR) (100%) recovered by washout treatment. RT-qPCR For real-time PCR, total RNA was extracted using the RNeasy Mini Kit (Qiagen). Reverse transcription was catalyzed by M-MLV reverse transcriptase (Life Technology), using random hexamers (IDT). Real-time PCR was performed with SYBR Green labeling in the 7500 Fast Real-Time PCR System (Applied Biosystems). PCR conditions were 50°C for 2 min, 95°C for 10 min, and 45 cycles of 95°C for 15 s and 60°C for 1 min mRNA expression was normalized to actin. Primers used for qPCR: mISG15 qPCR F (GGTGTCCGTGACTAACTCCAT); mISG15 qPCR R (TGGAAAGGGTAAGACCGTCCT); mActin qPCR F (AGAGGGAAATCGTG CGTGAC); and mActin qPCR R (CAATAGTGATGACCTGGCCGT). Viral infection studies Viral titers: Viral titers were determined in serum and organs harvested at the indicated dpi, or in the supernatant of infected cells at the hpi indicated. Organs were harvested into 750 μL PBS and homogenized with 1.0 mm diameter zirconia-silica beads at 3,200 rpm for 1 min with a MagnaLyzer prior to plaque assay on VERO E6 cells. Serial dilutions of organ homogenates in DMEM with 1% FBS were added to VERO E6 cells (5×105 cells for 6-well plates) and incubated for 1 h at 37°C. An agar overlay was then added to the cells and incubated for 72 h at 37°C. Plates were fixed with 1% formaldehyde (>1 h at room temperature), agar plugs were removed, and plaques were visualized using a 1% crystal violet solution. Cytokine analysis: Mouse sera were collected at indicated dpi and stored at −80°C. Cytokines and chemokines were measured using Luminex multiplex technology (Bio-Rad) with the 13-plex from Millipore (MPXMCYTO-70X) according to the manufacturer’s instructions. Histological Analysis: Brains were collected at 10 dpi, fixed with 4% paraformaldehyde solution in PBS overnight, embedded in paraffin, and cut into 5μm sections. Consecutive-cut brain sections were stained with hematoxylin-eosin (Musculoskeletal Histology and Morphometry Core, Washington University in St. Louis), RNA in situ hybridization, and IBA1 immunohistochemical staining, respectively. RNA in situ hybridization was performed using RNAscope 2.5 HD Detection Reagent kit (Brown kit) according to the manufacturer’s instructions (Advanced Cell Diagnostics). Briefly, sections were deparaffinized, digested with protease, and hybridized with probes targeting CHIKV spike protein (V-CHIKV-sp) for 2 h at 42°C. Amplifiers and DAB-A/B were then hybridized sequentially at 42°C. The sections were then counterstained with 50% Hematoxylin and 0.02% Ammonia. For IBA1 immunohistochemical analysis, sections were deparaffinized, antigen retrieval was performed using Tris-based antigen retrieval solution (Vector Laboratories) at 75°C for 10 min. Sections were blocked in 5% normal goat serum (Vector Laboratories) for one hour, incubated overnight at 4°C with anti-IBA1 antibody (Proteintech, 10904-1-AP) at 1:2000 dilution, and then developed with ImmPACT DAB EqV (Vector Laboratories) for two minutes. The sections were then counterstained with hematoxylin. Images were acquired using the Olympus BX51 with an Olympus DP74 Digital Camera and blindly examined by a certified neuropathologist (Robert E. Schmidt, Washington University School of Medicine). Researchers were blinded for analysis of histopathological sections. QUANTIFICATION AND STATISTICAL ANALYSIS Experimental replicate information and sample size are indicated in the figures and figure legends. All results are given as mean ± SD or SEM and analyzed by the Mann-Whitney test performed with GraphPad Prism 10 software (GraphPad Software). For in vivo experiments, statistical significance was determined by using the Mantel-Cox and Gehan-Breslow-Wilcoxon tests. Differences with p < 0.05 were considered significant. p values were defined as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. All data collected were included for quantification, and no outliers were excluded. Supplementary Material Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2026.117360. KEY RESOURCES TABLEREAGENT or RESOURCESOURCE IDENTIFIERAntibodiesCleaved Caspase-3 (5A1E)Cell Signaling TechnologyCat# 9664; RRID:AB_2070042phospho-MLKL [EPR9515(2)] (mouse)AbcamCat# 196436; RRID:AB_2687465Total MLKL, clone 3H1Millipore SigmaCat# MABC604; RRID:AB_2820284phospho-MLKL [EPR9515(2)] (human)AbcamCat# ab187091; RRID: AB_2619685β-ActinAbcamCat# ab6276; RRID:AB_2223210ISG15Generated in-houseN/ARIPK1BDCat# 610458; RRID: AB_397831RIPK3AbcamCat# ab62344; RRID: AB_956268phospho-RIPK3AbcamCat# ab205421; RRID: AB_2868433phospho-RIPK1Cell Signaling TechnologyCat# 31122; RRID: AB_2799000phospho-NF-κBCell Signaling TechnologyCat# 3033S; RRID: AB_331284IBA1 AntibodyProteintechCat# 10904-1-AP; RRID: AB_2224377UbE1L AntibodyProteintechCat# 15818-1-AP; RRID: AB_2211764Bacterial and virus strainsChikungunya virus La Réunion strain (LR2006)Generated in-house as previously described (Werneke et al., 2011)39N/AChemicals, peptides, and recombinant proteinsDulbecco’s Modified Eagle MediumCorningCat. #10-013-CMPolyethylenimine-based transporter 5PolysciencesCat. #26008OPTI-MEM™Thermo ScientificCat. #31985062Pierce™ IP Lysis bufferThermo ScientificCat. #87787Protease inhibitor cocktailRocheCat. #11836153001Phosphatase inhibitor cocktail 2Millipore SigmaCat. #P5726Phosphatase inhibitor cocktail 3Millipore SigmaCat. #P0044Fetal Bovine SerumHarvest MidSciCat. #USDAFBSPenicillin-StreptomycinCorningCat. #30-022-ClHEPES bufferCorningCat. #25-060-ClNon-Essential Amino AcidsMillipore SigmaCat. #M7145l-glutamineCorningCat. #25-005-ClMcCoy’s 5a mediumGiibcoCat. #166000823dGRO® L-WRN Conditioned Media SupplementMillipore SigmaCat. #SCM105ROCK inhibitorR&D SystemsCat. #1614TGF beta inhibitorR&D SystemsCat. #1254Lipofectamine 2000InvitrogenCat. #11668027PuromycinMillipore SigmaCat. #P4512Necrosulfonamide (NSA)InvitrogenCat. #480073SYN-1215MedChemExpressCat. #HY-112292Poly(I:C)InvivogenCat. #tlrl-picLPSInvitrogenCat. #00-4976-03NuPAGE 4–12% Bis-Tris GelsThermoFisherCat. #NP0335BOXImmun-Blot Low PVDF membranesBio-RadCat. #1620264Intercept (TBS) Odyssey blocking bufferLi-CorCat. #927-60001SYTOX GreenInvitrogenCat. #S7020SYTO24InvitrogenCat. #S7559SYTO16InvitrogenCat. #S7578SYTOX OrangeInvitrogenCat. #S11368B/B dimerizerTakara BioCat. #635058Annexin V-APCInvitrogenCat. #50-927-9Washout ligandTakara BioCat. #635088Lipofectamine RNAiMaxInvitrogenCat. #13778150M-MLV reverse transcriptaseInvitrogenCat. #28025013ReadyMade™ Random HexamersIDTCat. #51-01-18-25SYBR™ GreenInvitrogenCat. #S7563Murine TNFR&D SystemsCat. #410-MTHuman TNFR&D SystemsCat. #210-TAStaurosporineMedChemExpressCat. #HY-15141LCL-161MedChemExpressCat. #HY-15518zVADMedChemExpressCat. #HY-16658Nec-1sMedChemExpressCat. #HY-14622AGSK’872MedChemExpressCat. #HY-101872Antigen Unmasking Solution, Tris-BasedVector LaboratoriesCat. #H-3301Normal Goat SerumVector LaboratoriesCat. #S-1000ImmPACT® DAB EqVVector LaboratoriesCat. #SK-4103Human IFN-βThermoFisher Scientific, PeproTechCat. #300-02BCMouse IFN-βBioLegendCat. #581304Pierce™ Anti-DYKDDDDK Magnetic AgaroseThermoFisher ScientificCat. #A36797Critical commercial assays13-plex Cytokine/Chemokine Luminex KitBio-RadMPXMCYTO-70XRNAscope 2.5 HD Detection Reagent KitAdvanced Cell Diagnostics, Inc.Cat. #322300Dynabeads Antibody Coupling KitThermo ScientificCat. #14311DCytoTox96 Non-Radioactive Cytotoxicity AssayPromegaCat. #G1780Caspase-Glo® 3/7 Assay SystemPromegaCat. #G8090RNeasy Mini KitQiagenCat. #74104SYBR Green PCR Master MixApplied BiosystemCat# 1609503Experimental models: Cell linesPrimary Murine Embryonic Fibroblasts (WT, Isg15−/−, Ube1L−/−, Isg15−/−;Mlkl−/−, Isg15−/−;Ripk3−/−)Generated from mice in this paperN/ANIH/3T3 cellsATCCCat. #CRL-1658; RRID: CVCL_0123HT-29 cellsATCCCat. #HTB-38; RRID: CVCL_0320Primary Murine Lung Spheroids (WT, Isg15−/−)Generated from mice in this paperN/AVero E6 cellsATCCCat. #CRL-1586; RRID: CVCL_0574E1/RAS transformed MEFs (WT, Isg15−/−)Generated in-houseN/AHEK-293T cellsATCCCat. #CRL-3216; RRID: CVCL_0063)RIPK3-2Fv-NIH3T3Yatim et al., 201577N/ARIPK3(RHIM)-2Fv-NIH3T3Yatim et al., 201577N/AhMLKL1-181-2Fv-NIH3T3Gong et al., 2017b49N/ARIPK3-2Fv-NIH3T3Mlkl−/−Gong et al., 2017b49N/AExperimental models: Organisms/strainsMouse: WT: C57BL6/JThe Jackson LabJAX #000664Mouse: Isg15−/−: C57BL6/J 18, 79-81Provided by Klaus-Peter Knobeloch, University Clinic Freiburg, Germany78N/AMouse: Ube1L−/−: C57BL6/JProvided by Dong-Er Zhang, University of California, San Diego79N/AMouse: Mlkl−/−: C57BL6/JProvided by James Murphy, Walter and Eliza Hall Institute of Medical Research18N/AMouse: Isg15−/−;Mlkl−/−: C57BL6/JThis paperN/AMouse: Ripk3−/−: C57BL6/JProvided by Vishva Dixit, Genentech80N/AMouse: Isg15−/−;Ripk3−/−: C57BL6/JThis paperN/AOligonucleotidesCRISPR interference targeting human ISG15: GCACGCCTTCCAGCAGCGTCIntegrated DNA TechnologiesN/AsiRNA against murine ISG15DharmaconN/Ascramble siRNADharmaconN/AqPCR primer mISG15 F: GGTGTCCGTGACTAACTCCATIntegrated DNA TechnologiesN/AqPCR primer mISG15 R: TGGAAAGGGTAAGACCGTCCTIntegrated DNA TechnologiesN/AqPCR primer mActin F: AGAGGGAAATCGTGCGTGACIntegrated DNA TechnologiesN/AqPCR primer mActin R: CAATAGTGATGACCTGGCCGTIntegrated DNA TechnologiesN/ARecombinant DNApX458 vectorAddgenePlasmid #48138pCMV-VSV-GAddgene, gift from Bob WeinbergPlasmid #8454pTRIP-ISG15AA-eYFPThis paperN/ApTRIP-ISG15AA-eYFPThis paperN/ApCMV.DR.8.91NovoPro LabsCat. #V007548Software and algorithmsGraphPad Prism v10GraphPadRRID: SCR_002798Image StudioLi-CorRRID: SCR_013715FlowJoTree StarRRID: SCR_008520BioTek Gen5 Software for Imaging & MicroscopyAgilentRRID: SCR_017317 KEY RESOURCES TABLE Highlights ISG15 negatively regulates RIPK3-mediated cell death in murine and human cells ISG15 negatively regulates RIPK3-mediated cell death in murine and human cells ISG15 regulates necroptosis induced by CHIKV, TNF, LPS, and poly(I:C) stimulation ISG15 regulates necroptosis induced by CHIKV, TNF, LPS, and poly(I:C) stimulation ISG15 non-covalently associates with the RIPK3 necrosome to limit its assembly ISG15 non-covalently associates with the RIPK3 necrosome to limit its assembly ISG15 restricts CHIKV-induced immunopathology by inhibiting necroptosis in vivo ISG15 restricts CHIKV-induced immunopathology by inhibiting necroptosis in vivo ACKNOWLEDGMENTS We thank Klaus-Peter Knobeloch (University Clinic Freiburg, Germany), Dong-Er Zhang (University of California, San Diego), James Murphy (Walter and Eliza Hall Institute of Medical Research), and Vishva Dixit (Genentech) for providing mice. We thank the Genome Engineering & Stem Cell Center (GESC@MGI, RRID: SCR 023243) at Washington University in St. Louis for cell line engineering. This study was supported by the following grants: NIH grant R01 AI080672 (D.J.L.), NIH grant R01 DK130476 (D.J.L.), NIH grant R01 AI044828 (D.R.G.), NIH grant R35 CA231620 (D.R.G.), NIH grant P30 CA047904 (Y.-N.G.), NIH grant T32 CA009547 (Y.-C.P.), NIH grant T32 AI007172 (J.M.C.W.), NIH grant T32 GM007200 (J.M.C.W.), NIH grant F30 DK138644 (J.M.C.W.), NIH grant F31 AI149999 (M.C.L.), NIH grant P30 AR073752 (Washington University Rheumatic Diseases Research Resource-based Center) (D.J.L.), and NIH grant P30 AR074992 (Washington University Musculoskeletal Research Center). Footnotes RESOURCE AVAILABILITY Lead contact Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Deborah Lenschow (dlenschow@wustl.edu). Materials availability All unique/stable reagents generated in this study are available from the lead contact with a completed materials transfer agreement. Data and code availability All data are available in the main text or the supplementary materials. All data reported in this paper will be shared by the lead contact upon request.This paper does not report original code.Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. All data are available in the main text or the supplementary materials. All data reported in this paper will be shared by the lead contact upon request. All data are available in the main text or the supplementary materials. All data reported in this paper will be shared by the lead contact upon request. This paper does not report original code. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. DECLARATION OF INTERESTS Over the course of this research, DRG consulted for Ventus Therapeutics, Sonata Therapeutics, ASHA Pharmaceuticals, and Mycos and received financial support/travel support from Horizon/Amgen and Boerhinger-Engleheim. REFERENCES Associated Data Supplementary Materials