Impaired RIPK1 ubiquitination sensitizes mice to TNF toxicity and inflammatory cell death

Receptor-interacting protein 1 (RIP1; RIPK1) is a key regulator of multiple signaling pathways that mediate inflammatory responses and cell death. TNF-TNFR1 triggered signaling complex formation, subsequent NF-κB and MAPK activation and induction of cell death involve RIPK1 ubiquitination at several lysine residues including Lys376 and Lys115. Here we show that mutating the ubiquitination site K376 of RIPK1 (K376R) in mice activates cell death resulting in embryonic lethality. In contrast to Ripk1K376R/K376R mice, Ripk1K115R/K115R mice reached adulthood and showed slightly higher responsiveness to TNF-induced death. Cell death observed in Ripk1K376R/K376R embryos relied on RIPK1 kinase activity as administration of RIPK1 inhibitor GNE684 to pregnant heterozygous mice effectively blocked cell death and prolonged survival. Embryonic lethality of Ripk1K376R/K376R mice was prevented by the loss of TNFR1, or by simultaneous deletion of caspase-8 and RIPK3. Interestingly, elimination of the wild-type allele from adult Ripk1K376R/cko mice was tolerated. However, adult Ripk1K376R/cko mice were exquisitely sensitive to TNF-induced hypothermia and associated lethality. Absence of the K376 ubiquitination site diminished K11-linked, K63-linked, and linear ubiquitination of RIPK1, and promoted the assembly of death-inducing cellular complexes, suggesting that multiple ubiquitin linkages contribute to the stability of the RIPK1 signaling complex that stimulates NF-κB and MAPK activation. In contrast, mutating K115 did not affect RIPK1 ubiquitination or TNF stimulated NF-κB and MAPK signaling. Overall, our data indicate that selective impairment of RIPK1 ubiquitination can lower the threshold for RIPK1 activation by TNF resulting in cell death and embryonic lethality.

Persistent pathway activation and deubiquitination of RIPK1 can induce complex II formation. Complex II is a cytosolic complex consisting of caspase-8, FADD (Fasassociated protein with death domain), FLIP (FLICE inhibitory protein) and RIPK1 [20,26]. This complex can activate caspase-3 and -7 resulting in apoptosis, but will recruit RIPK3 to form complex IIb if caspases are inhibited or absent, leading to necroptosis [1,2]. RIPK1 is activated by autophosphorylation and becomes ubiquitinated at the same time [15,16]. Subsequently, active RIPK1 binds to RIPK3, which leads to RIPK3 autophosphorylation and phosphorylation of pseudokinase MLKL (mixed lineage kinase domain like) [27][28][29]. Phosphorylated MLKL oligomerizes and translocates to the plasma membrane inducing membrane perturbations that result in cell lysis [30].
RIPK1 has been implicated in numerous inflammatory and neurodegenerative pathologies in animal disease models, and by demonstration of RIPK1 pathway activation in patients [31,32]. Multiple studies have shown that genetic inactivation or chemical inhibition of RIPK1 kinase activity is protective in animal disease models of gut and skin inflammation, rheumatoid arthritis, Alzheimer's disease, multiple sclerosis and Parkinson's disease, to name a few [31,32]. RIPK1 inhibition is also protective in acute disease models such as the TNF-induced systemic inflammatory response syndrome (SIRS) model [31,33]. All these studies suggest that RIPK1 is an attractive drug target in inflammatory and neurodegenerative diseases [32,34].
While it is clear that the kinase activity of RIPK1 is dispensable for organismal homeostasis or reproduction, the physiological importance of RIPK1 ubiquitination is less clear. To assess the biological relevance of prosurvival and pro-cell death ubiquitination of RIPK1, we generated knock-in mice where K376 or K115 was mutated to arginine. RIPK1(K376R) reduced K11-linked, K63-linked, and linear RIPK1 ubiquitination, and promoted the assembly of death-inducing cellular complexes. Apoptosis in embryos caused lethality and was dependent on the catalytic activity of RIPK1, TNFR1, and caspase-8/ RIPK3. RIPK1(K115R) was not lethal, but moderately enhanced responsiveness to TNF-induced death. Together, our data suggest that selective impairment of RIPK1 ubiquitination promotes activation of RIPK1, cell death, and embryonic lethality.

Mice
The Genentech institutional animal care and use committee responsible for ethical compliance approved all animal protocols. Tnfr1 −/− , Casp8 +/− , Ripk3 −/− and Rosa26-Cre. ER T2 mice were described before [33,35,36]. Ripk1 K115R knockin mice were generated by CRISPR/Cas9 technology [37,38]. The guide target sequence was: 5' GAA AGG AAG GAT AAT CGT GG 3' with protospacer adaptive motif (PAM): AGG and an oligonucleotide donor 5'CTT  CAG GTC CTT GTG TAT CAC ACC TTT GTC ATG  TAA GTA GCA CAT GCC TTC TAT tgc ttc tac aat TAT  CCT TCC tcg CAA TGA AAG TGG GAC ATC TAT CTG  GAA TAA CAC ATT AAG TCT ATG AAG TGA AGA  GGC AAT CTA ACA GGC AAG AGC 3' (Integrated  DNA technologies) were used. Cas9 mRNA (Thermo Fisher; A29378), sgRNA (Synthego) and oligonucleotide template were used to modify zygotes. After zygote microinjection and embryo transfer, genomic DNA was prepared from tail tip biopsies of potential G0 founders. G0 mosaic founders were analyzed for the top 15 off-target loci per sgRNA (obtained by the CRISPR design tool from Benchling). Founders without mutations were selected for mating with wild type C57BL/6N mice for germline transmission of the gene-edited chromosome.
The same approach was chosen for Ripk1 K376R knockin strain. The guide target sequence was: 5' CGA-GAATGATCGCAGTGTGC 3'; PAM: AGG and the oligonucleotide donor (5'ATT CTG CCT TGG CTG CGG  TTT TGT CTG TTT CTC TGC AAA TAT TCC AAA  AGC ATG ATA GCT GGC TTC CTC TTG CAG tcg tgc  CTG tac ACT gcg atc ATT CTC GTC CTG TGG GTA  CTC TGG GGA GGA AGA AAA CCA GGA CTC CTC  CAC AGG ACC 3') was used.
For timed pregnancies, the day of vaginal plug detection was set as embryonic day 0.5 (E0.5). Pregnancies were confirmed by ultrasound or by weight gain of the dams. RIPK1 kinase inhibitor GNE684 was dosed by oral gavage twice day at 10 mg/ml in 10% DMSO + MCT. The mice were dosed with 50 mg/kg GNE684.
Systemic inflammatory response syndrome (SIRS) was induced by intravenous (iv) injection of 500 μg of mouse TNF (Genentech) [39] per kg body weight. Calculations to determine group sizes were not performed, mice were grouped according to genotypes and the studies were unblinded. Body temperature was monitored using a rectal probe and a digital thermometer. Mice were euthanized if their body temperature was below 25°C or if severely lethargic.
The Rosa26-Cre.ER T2 allele was maintained heterozygous. Nuclear translocation was induced by intraperitoneal injection of tamoxifen (80 mg/kg body weight) for three consecutive days. Tamoxifen (Sigma-Aldrich) was solubilized in sunflower seed oil (Sigma-Aldrich).
Cell death was analyzed using Incucyte ZOOM and S3 (Essen BioSciences) using Sytox Green nucleic acid stain (Life technologies). 200 μg/ml digitonin (Sigma Aldrich) was used as a positive control to achieve complete cell lysis.

Statistical analysis
The number of independent experiments performed is indicated in the figure legends (at least two). The variance was assumed to be similar between the compared groups and that groups have normal distribution. The statistical significance was analyzed by the indicated tests. One-way ANOVA, Two-way ANOVA and Mantel-Cox (log rank) were performed using the GraphPad Prism software.
K115 ubiquitination of RIPK1 has been linked to necroptotic cell death signaling [15,17]. To investigate the role of this ubiquitination site in vivo, CRISPR RIPK1 K115R KI mice were generated (see Methods). Ripk1 K115R/ K115R (K115R) mice were born at normal Mendelian frequencies (Fig. 1f). Aging for 15 months did not result in any genotype-related differences between WT and K115R mice, but comparable age-related findings including neoplasia were identified in both genotypes (Fig. 1g). In line with these findings, serum IL-6 levels were not different between WT and K115R animals (Fig. 1h). When K115R mice were challenged by intravenous injection of TNF (500 μg/kg body weight) in a TNF-induced systemic inflammatory response syndrome (SIRS) model, they exhibited a significantly higher morbidity (Mantel-Cox test p = 0.0095) (Fig. 1i) and a more pronounced drop in body temperature than WT littermates (Fig. S2A). K115R animals had slightly but not significantly elevated serum levels of IL-6 after 4 h (p = 0.25), whereas levels of other cytokines and chemokines such as RANTES, KC, and IFNγ were not altered (Fig. S2B). While RIPK1 K115R BMDMs (bone marrow-derived macrophages) exhibited normal ubiquitination of RIPK1 and normal NF-κΒ and MAPK signaling in response to TNF, the cells showed enhanced necroptosis signaling in response to TNF and zVAD (Fig. S2C-G). Necroptosis mediated by LPS and zVAD, or poly(I:C) and zVAD, was also slightly enhanced (Fig. S2G). Based on these results, RIPK1 ubiquitination at K115 seems to play a minor role in TNF-induced cell death signaling in vitro or in vivo.

Inhibition of RIPK1 kinase activity blocks cell death and delays lethality in RIPK1(K376R) embryos
To investigate the role of RIPK1 catalytic activity in the lethality of RIPK1 K376R embryos, we dosed pregnant females with the RIPK1 kinase inhibitor GNE684 (ref. [31]). Pregnant females were dosed by oral gavage twice daily (50 mg/kg BID) starting at E9.5 (Fig. 2a) and embryos were necropsied at E12.5. As expected, K376R embryos from pregnant females treated with vehicle alone showed devascularization and blood vessel breakdown (Fig. 2b). Interestingly, K376R embryos exposed to GNE684 showed a complete rescue of the macroscopic phenotype (Fig. 2b). Accordingly, vehicle-treated K376R embryos had endothelial cells containing cleaved caspase-3, whereas GNE684-treated K376R embryos did not (Fig. 2c). Placentas of GNE684-treated embryos also lacked detectable cleaved caspase-3 (Figs. 2d and S1E). pRIPK3 positive cells were significantly increased in placentas of vehicletreated K376R embryos compared to WT or treated K376R placentas (Figs. 2e and S1F). Thus, the kinase activity of RIPK1 appears crucial for RIPK1(K376R) to induce cell death and embryonic lethality at E12.5.
We also analyzed levels of proinflammatory cytokines and chemokines in E11.5 livers by quantitative PCR analysis. Several proinflammatory cytokine and chemokine genes (Tnf, Cxcl2, Cxcl1, and Cxcl2), as well as the NF-κB target genes TNFAIP3, Bcl3, and RelB were expressed at higher levels in K376R liver when compared to WT livers (Figs. 2f and S1G). Expression of these genes in K376R embryos treated with GNE684 was comparable to that seen in WT embryos (Figs. 2f and S1G), suggesting that cell death drives proinflammatory gene expression.

RIPK1(K376R) impairs TNF-induced NF-κB and MAPK signaling and promotes cell death complex formation
To investigate the importance of RIPK1 lysine 376 for TNF-induced NF-κB and MAPK activation, as well as complex I formation, we treated primary MEFs (mouse embryo fibroblasts) derived from WT or K376R embryos with FLAG-TNF (Fig. 3a, b). RIPK1(K376R) decreased phosphorylation of IκBα, RelA/p65, JNK, and p38 when compared to RIPK1 WT (Fig. 3a). NF-κB activation was not affected in Ripk1 K376R/+ primary MEFs compared to WT (Fig. S3A). Ubiquitination of RIPK1 and its recruitment to the TNFR1 receptor complex was also impaired in RIPK1 (K376R) MEFs compared to WT MEFs (Fig. 3b). Accordingly, LUBAC components HOIP and SHARPIN as well as NEMO were only weakly recruited to the TNFR1 complex in K376R cells (Fig. 3b). TRADD recruitment to the complex was unchanged, which is expected given that TRADD is recruited via death domain interactions independently of RIPK1 ubiquitination (Fig. 3b). Subcellular fractionation analyses of RelA/p65 in immortalized MEFs treated with TNF confirmed the NF-κB activation defects observed in K376R primary MEFs. Translocation of phosphorylated p65 into the nucleus was impaired in K376R MEFs after TNF treatment (Fig. S3B). Interestingly, expression of select NF-κB target genes in primary MEFs after 4 h of TNF treatment was not greatly affected by RIPK1(K376R) (Fig. S3C, top row). Impaired TNF-induced gene expression was more apparent upon normalization to unstimulated cells of the same genotype (Fig. S3C, bottom row), possibly because of higher baseline gene expression in unstimulated K376R MEFs.
Defective recruitment of RIPK1(K376R) into complex I suggested that TNF-mediated cell death might be enhanced. Indeed, treatment with TNF for up to 6 h induced caspase-3 cleavage and autophosphorylation of RIPK3 in K376R MEFs, but not in WT MEFs (Fig. 3c). Signaling in response to TNF and zVAD (TZ) was also enhanced by RIPK1 (K376R) based on increased phosphorylation of RIPK1, RIPK3, and MLKL, and more robust complex II formation in RIPK1 K376R MEFs when compared to WT MEFs (Fig. 3c). Again, we observed comparable signaling in Ripk1 K376R/+ and WT MEFs (Fig. S3D). GNE684 blocked these signaling events, indicating their dependence on RIPK1 activation (Figs. 3c and S3E). Consistent with these data, RIPK1 K376R MEFs were more sensitive to either TNF-or TZ-induced cell death when compared to WT MEFs (Fig. 3d). Interestingly, addition of the IAP antagonist BV6 (ref. [40]) enhanced TNF-or TZ-induced cell death in WT, but not K376R MEFs (Fig. 3d). Indeed, WT and RIPK1 K376R MEFs exhibited comparable cell death in response to TNF, BV6, and zVAD (TBZ; Fig. 3d). Accordingly, TBZ induced comparable phosphorylation of RIPK1, RIPK3, and MLKL in WT and K376R MEFs (Fig.  S3E). Thus, depletion of c-IAP1/2 by BV6 does not augment cell death in the absence of K376 RIPK1 ubiquitination site. K376R MEFs were also more sensitive to LPS and zVAD-induced cell death when compared to WT MEFs (Fig. S4A, B).

RIPK1(K376R) sensitizes adult mice and cells to TNF-induced SIRS and necroptosis
To study the role of RIPK1 K376 ubiquitination in adult mice, the Ripk1 K376R allele was combined with a conditional Fig. 1 RIPK1 K376R  e Quantification of IHC for pRIPK3 positive cells in placentas of Ripk1 +/+ and Ripk1 K376R/K376R at E12.5. Representative images are provided in S1F. f RT-qPCR analysis of indicated cytokines and chemokines using E11.5 embryonic liver total RNA [n = 3] after dosing for three days with vehicle or 50 mg/kg GNE684. Data of three independent experiments are shown as mean with SD. P values, oneway ANOVA followed by Tukey's multiple comparison test (** > 0.0021, *** > 0.0002).
Given that K376R cells exhibited decreased TNFinduced RIPK1 ubiquitination and complex I recruitment, we used ubiquitin linkage-specific antibodies to determine which ubiquitin chain linkages were reduced on RIPK1 (K376R). Surprisingly, there was an overall decrease in K11-linked, K63-linked, and linear ubiquitination on RIPK1(K376R) when compared with WT RIPK1 (Figs. 6e and S6C). These findings suggest that K376 of RIPK1 could be a site for ubiquitination by multiple linkages, all of them likely contributing to TNF stimulated NF-κB and MAPK signaling as well as complex I stability.
Discussion RIPK1 ubiquitination plays a critical role in the spatial and temporal regulation of TNF stimulated inflammatory and cell death signaling [11]. We analyzed the physiological relevance of two described ubiquitination sites on RIPK1, K115, and K376 (ref. [12,[14][15][16][17]). Mice expressing only RIPK1(K376R) died during embryonic development around E12.5 with massive cell death in the embryo and yolk sac, as well as upregulation of proinflammatory cytokines and chemokines. Inhibition of RIPK1(K376R) by GNE684 prolonged embryo survival, as did genetic deletion of TNFR1. However, only deletion of caspase-8 and RIPK3 enabled animals to reach adulthood. Together, these data indicate that the absence of K376 RIPK1 ubiquitination causes lethality requiring the kinase activity of RIPK1, TNFR1, and caspase-8/RIPK3 mediated cell death. Consistent with these findings, K376R MEFs and K376R/-BMDMs showed increased sensitivity to TNF and LPS stimulated cell death, while K376R/-mice succumbed rapidly to TNF-induced hypothermia. Given that lysine 376 of RIPK1 is ubiquitinated within minutes of TNF binding to TNFR1 [12,13], absence of this site could affect RIPK1 ubiquitination within complex I and subsequent NF-κB and MAPK activation. Indeed, stimulation of RIPK1 K376R cells with TNF lead to reduced RIPK1 ubiquitination with K11-linked, reduced expression of NF-κB target genes in K376R MEFs, levels of inflammatory cytokines were higher in K376R/-BMDMs and in K376R embryonic livers. It is possible that cell death releases DAMPs (danger-associated molecular patterns), thus stimulating inflammatory signaling in cells still alive. Taken together, these data suggest that complex I destabilization leading to cell death is the major driver of the RIPK1 K376R phenotype.
Interestingly, BV6 did not increase TNF-induced cell death in K376R cells, which is consistent with deletion of the RIPK1 E3 ligases c-IAP1/2 mainly impacting ubiquitination on K376 of RIPK1. Accordingly, wild type cells with K376 available for ubiquitination were sensitized to TNF-induced death by BV6. These data reiterate the importance of c-IAP1/2 mediated RIPK1 ubiquitination for TNFR1 complex stability. Recently, two other studies reported a similar phenotype of RIPK1 K376R knockin mice [44,45]. However, those reports focused exclusively on K63-linked RIPK1 ubiquitination, while we found that K11-linked and linear RIPK1 ubiquitination were also severely impacted. Reduced LUBAC recruitment and linear RIPK1 ubiquitination are especially relevant, considering the critical importance of this modification for the stability of complex I and for preventing cell death [20,46]. We also explored elimination of the K376 RIPK1 ubiquitination site in adult mice and found that this site was not critical for adult homeostasis, at least in the short-term. Still, mice with compromised ubiquitination at K376 of RIPK1 were exquisitely sensitive to TNF-induced hypothermia, arguing that proper TNF signaling requires K376 RIPK1 ubiquitination.
We also characterized the role of K115 RIPK1 ubiquitination [15,17] and found that elimination of this site mildly sensitized mice or cells to TNF. TNF-induced ubiquitination and signaling in RIPK1 K115R cells appeared normal, likely because RIPK1 undergoes ubiquitination at multiple sites during proinflammatory and cell death signaling [15]. Even though abolition of the K376 ubiquitination site did not completely prevent RIPK1 ubiquitination, it did promote the transition of RIPK1 from complex I to the cell death-promoting complex II. The K115 site evidently does not have such importance and its elimination moderately affects TNFinduced cell death. Further studies on additional RIPK1 ubiquitination sites will be needed to fully explore cell death-associated RIPK1 ubiquitination. Given that therapeutic targeting of RIPK1 is being tested in multiple clinical trials [34,47], understanding how ubiquitination and other mechanisms regulate activation of RIPK1 may help identify human diseases that would benefit from RIPK1 inhibition.
Author contributions DV supervised all studies. MK performed the majority of cellular and signaling experiments. TG and MK performed ubiquitin-chain specific immunoprecipitations. MK and DD performed in vivo experiments. JDW performed histological analysis and LGK performed microscopy. CY and MR-G designed and generated the knock-in mouse strains. KN contributed to experimental designs. DV and MK wrote the paper with input from all authors.

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.