Ubiquitin-Mediated Regulation of RIPK1 Kinase Activity Independent of IKK and MK2

1Inflammation and cell death are ancient processes of fundamental biological importance that enable survival and adaptation during infection and injury. Tumor necrosis factor (TNF) is a potent inflammatory cytokine that triggers, through its type 1 receptor (TNF-R1), either pro-survival/inflammatory or pro-death signaling pathways in a ubiquitin (Ub)- and phosphorylation-dependent manner. TNF can regulate tissue homeostasis in at least three different ways: (1) activation of nuclear factor κB (NF-κB) and MAPK/JNK-transcriptional programs, (2) induction of caspase-8-dependent apoptosis, or (3) stimulation of receptor-interacting protein kinase (RIPK)-mediated necroptosis.

2Binding of TNF to TNF-R1 results in the formation of two signaling complexes. Upon TNF ligation, a protein complex assembles on the cytoplasmic tail of TNFR1. This complex, frequently referred to as complex-I, consists of TNF-R1, the adaptors TRADD and TRAF2, the kinase RIPK1, and the E3 ubiquitin (Ub) ligases cellular inhibitor of apoptosis 1 (cIAP1) and cIAP2. Within this complex, RIPK1 and other proteins are rapidly conjugated with M1, K11, K48, and K63 Ub linkage types, and cIAP-mediated conjugation of Ub to RIPK1 allows recruitment of the kinase complex TAK1/TAB2/TAB3 and the E3 ligase linear Ub chain assembly complex (LUBAC, composed of HOIL/HOIP/SHARPIN). LUBAC-mediated linear ubiquitylation of different components of complex-I (RIPK1, TRADD, and TNF-R1) subsequently reinforces complex-I and allows efficient recruitment and activation of IKK (composed of NEMO/IKKα/IKKβ), which in turn drives activation of NF-κB. While the synthesis of M1- and K63-linked poly-Ub chains play key roles in Ub-dependent assembly of complex-I and the induction of NF-κB target genes that drive inflammation and cell survival following TNF stimulation, the role of K11 and K48 poly-Ub remains largely uncharacterized.

3TNF-induced cell death is mediated by an RIPK1-containing secondary complex that is frequently referred to as complex-II or necrosome. It is thought that the Ub chains conjugated to RIPK1 by cIAP1/2 and LUBAC in complex-I constitute one of the decisive factors preventing RIPK1 from forming complex-II and limiting its killing potential. Consistently, genetic deletion of cIAPs completely abrogates RIPK1 ubiquitylation, leading to exaggerated complex-II formation and RIPK1-mediated cell death in response to TNF. The interpretation of the role of RIPK1 ubiquitylation in suppressing the cytotoxic potential of RIPK1 is complicated by the fact that loss of cIAPs not only abrogates RIPK1 ubiquitylation, but also interferes with recruitment of LUBAC, TAK1, and IKK. In particular, loss of TAK1 recruitment prevents activation of MK2 and IKK, which in turn regulate the cytotoxic potential of RIPK1 via direct phosphorylation. Thus, loss of cIAPs not only interferes with activation of NF-κB, but also abrogates MK2- and IKK-mediated suppression of RIPK1.

4While it is beyond doubt that cIAPs suppress TNF-induced cell death, how this is achieved remains unclear. The main problem in dissecting cIAP-mediated regulation of TNF-induced cell death has been the fact that the signaling aspect of Ub (recruitment/activation of TAK1, IKK, MK2, and NF-κB-mediated gene induction) and the direct Ub-dependent anti-apoptotic function of cIAPs cannot be separated. We now identified a point mutation in cIAP1 that selectively sensitizes cells to TNF-induced cell death, without interfering with TNF-mediated activation of NF-κB, and IKK- and MK2-mediated phosphorylation of RIPK1. This mutation affects the evolutionarily conserved ubiquitin-associated (UBA) domain of cIAP1. Mice with a knockin mutation in the UBA domain develop normally but are acutely sensitive to TNF-induced systemic inflammatory response syndrome (SIRS), which is caused by enhanced sensitivity to TNF-mediated cell death. Our data are consistent with the notion that the UBA domain is required for Ub-mediated regulation of RIPK1 kinase activity. We find that cIAP1 represses RIPK1 kinase auto-activation via UBA-dependent ubiquitylation of an expanded repertoire of Ub-acceptor lysines of RIPK1. In addition, we find that cIAP1 with a functional UBA domain increases the number of ubiquitylation sites on RIPK1. Moreover, it enhances K48-linked poly-ubiquitylation of RIPK1. Together, this destabilizes active RIPK1 via proteasomal degradation. In the absence of a functional UBA domain, fewer K residues are ubiquitylated and fewer K48-linked chains are present on RIPK1. Together, this causes lethal accumulation of active RIPK1 kinase in response to TNF in cIAP1UBAmut cells. Our data demonstrate that cIAP-mediated ubiquitylation of RIPK1 directly regulates its kinase activity, independently of the recruitment of IKK and TAK1 kinase complexes.

5The BIR1 and RING domains of cIAP1/2 are required for TNF signaling, but little is known about the role of the UBA domain. Because UBA domains often regulate protein activity via protein-protein interactions, we conducted a yeast two-hybrid experiment with the UBA containing C-terminal portion of cIAP1 (cIAP1U/C/R) to establish a UBA interactome of cIAP1 (Figures 1A and 1B). This identified known as well as novel cIAP1-binding proteins (Figure 1B). Surprisingly, out of the 137 clones identified, TRAF2 was isolated 107 independent times. While previous work established that TRAF2 binds to the BIR1 of cIAP1 and cIAP2 (Figure 1A), our data suggest that TRAF2 also associates with the C-terminal portion of cIAP1.

6To narrow down the region within the UBA-CARD-RING fragment that mediates TRAF2 binding, we determined the ability of truncation mutants to interact with TRAF2. The UBA domain readily interacted with TRAF2 (Figures 1C and S1A), and point mutations in the conserved MGF motif of the hydrophobic patch of the UBA domain (MF > AA) abrogated TRAF2 binding. Consistent with the notion that cIAP1 binds TRAF2 through multivalent interactions via its BIR1 as well as UBA domains, we found that point mutations in either the BIR1 (ER > AA) or UBA (MF > AA) did not abolish the interaction between cIAP1 and TRAF2 in yeast two-hybrid experiments (Figures 1D, 1E, and S1B).

7We next mapped the region of TRAF2 that bound to the UBA domain. Surprisingly, the cIAP-interacting motif (CIM) in the TRAF-N domain, which is required for TRAF2 to interact with the BIR1 of cIAP1, was also indispensable for UBA binding. Accordingly, deletion of the CIM completely abrogated the interaction between TRAF2 and the UBA domain of cIAP1 (Figures 1F and S1C). These data indicate that cIAP1 contains two surfaces on very distinct spatially separated domains that bind to the same short TRAF2 motif. Because TRAF2 forms trimers, we cannot discern whether the UBA and BIR1 bind to the very same CIM of one TRAF2 molecule or to different CIMs of adjacent molecules.

8To independently corroborate the interaction, we performed isothermal titration calorimetry (ITC) using recombinant cIAP2. cIAP2 was used instead of cIAP1 because structural information of TRAF2/cIAP2 is available, and previous ITC measurements indicate that the BIR1 domains of cIAP1 and cIAP2 bind TRAF2 equivalently. Of note, cIAP1’s UBA is 87% similar to that of cIAP2. While the interaction between TRAF2 and the BIR1 domain of cIAP2 exhibits a dissociation constant of 1.7 μM, we found that the BIR1 in conjunction with the UBA domain bound to TRAF2 with a significantly higher affinity (0.43 μM) (Figures 2A and 2B). The increase in affinity can be explained by direct binding of the UBA domain to TRAF2, with a dissociation constant of 0.48 mM (Figure 2C).

9To determine the importance of the UBA domain for TRAF2 binding, we used Flp-InTM-RexTM-HEK293 (referred to as Flp-In) cells. Prior to transgene insertion, isogenic parental HEK293Flp-In;shcIAP1 cells bearing a doxycycline-inducible mir30-based short hairpin RNA (shRNA) against the 3′ UTR of endogenous cIAP1 were generated and reconstituted with either wild-type (WT) cIAP1 or the indicated mutants (Figure 2D). Because HEK293Flp-In cells do not express detectable levels of cIAP2 (data not shown), this system ensures single-copy insertion and equal expression levels of untagged cIAP1 proteins without interference from endogenous cIAPs. Expression of the doxycycline-inducible cIAP1 shRNA in parental cells reduced cIAP1 to an almost undetectable level and resulted in concomitant activation of the non-canonical NF-κB pathway (Figure 2E). Cells reconstituted with either a WT or a UBA mutant version of cIAP1 exhibited comparable levels of cIAP1, indicating that the UBA mutation did not affect protein stability. Moreover, cIAP1MF > AA suppressed activation of the non-canonical NF-κB pathway (Figure 2E) and underwent SMAC mimetic (SM)-induced auto-ubiquitylation and degradation, indicating that cIAP1MF > AA is able to ubiquitylate NIK and itself. Biotinylated SM readily co-purified TRAF2 with WT cIAP1 (Figure 2F). In contrast, and consistent with earlier reports, we found that mutation of the BIR1 (cIAP1ER > AA) almost completely abolished the binding of cIAP1 to TRAF2 (Figure 2F). Interestingly, mutation of the UBA domain, via either alteration of the MGF motif (MF > AA) or substitutions of E401 and N428 to RR (EN > RR), which disrupt UBA-mediated protein:protein interactions, likewise impaired TRAF2 binding (Figures 2F and 2G). While cIAP1-BIR1ER > AA and -UBAMF > AA mutants retained some binding to TRAF2, combined mutation (ER > AA/MF > AA) completely abrogated the interaction between cIAP1 and TRAF2 (Figure 2F). Together, these data corroborate the notion that TRAF2 interacts with cIAP1 via its BIR1 and UBA domain.

10To study the function of the UBA domain of cIAP1 in vivo, we generated a conditional knockin mouse bearing the MF > AA mutation in the absence of cIAP2 (Figure 3A). Previous work indicated that cIAP1 and cIAP2 function redundantly to each other. Therefore, we generated the conditional cIAP1UBAmut mouse from an embryonic stem cell (ESC) clone that previously had been targeted at the cIAP2 locus. These doubly targeted animals (cIAP2−/−cIAP1UBAmut) are subsequently referred to as cIAP1UBAmut. cIAP1UBAmut mice were weaned at the expected Mendelian ratio (Figure 3B) and were indistinguishable from their WT counterparts (Figures S2A and S2B). Additionally, these mice had an overtly normal immune system (Figure S2C). Primary mouse embryonic fibroblasts (MEFs) from WT and cIAP1UBAmut embryonic day 13.5 (E13.5) embryos exhibited the same cIAP1 protein levels (Figure 3C), indicating that the UBA mutation had no impact on the stability of cIAP1. As expected, these cells exhibited undetectable levels of cIAP2 mRNA (Figure S2D). To verify whether the UBA mutation affected cIAP1’s E3 ligase function, we tested the ability of SM to stimulate auto-ubiquitylation and degradation of cIAP1UBAmut. We found that the behavior of cIAP1UBAmut was indistinguishable from its WT counterpart (Figure S2E).

11To test the binding of cIAP1UBAmut to TRAF2, we purified cIAP1 using biotinylated SM. Significantly less TRAF2 was co-purified with cIAP1 in cIAP1UBAmut cells, while no TRAF2 binding was observed in TRAF2−/− cells (Figure 3D). Importantly, however, while this immunoprecipitation (IP) setting reveals a weakened association between cIAP1UBAmut and TRAF2, cIAP1UBAmut and TRAF2 are perfectly capable of interacting with each other in a physiological meaningful way because the UBA mutation of cIAP1 does not phenocopy loss of TRAF2 (Figure 3E). Accordingly, RIPK1 is readily poly-ubiquitylated in complex-I from cIAP1UBAmut cells, while RIPK1 ubiquitylation was lost in Traf2 knockout (KO) cells (Figure 3E). Moreover, cIAP1:TRAF2-mediated regulation of NIK, and suppression of non-canonical NF-κB, was normal in cIAP1UBAmut cells (Figure 3F). In contrast, depletion of cIAP1WT and cIAP1UBAmut by SM or depletion of TRAF2 by TWEAK activated non-canonical NF-κB (Figure 3F). As TRAF2 is essential to bring cIAPs to NIK, these data demonstrate that cIAP1UBAmut:TRAF2 association is sufficiently strong in vivo to target NIK for degradation. Additionally, a functional UBA domain of cIAP1 was dispensable for timely TNF-induced phosphorylation of p65, degradation of IκB, phosphorylation of MAPKs (Figures 3G and S2F), and the production of cytokines in primary MEFs, bone marrow-derived macrophages (BMDMs), and keratinocytes (Figures 3H–3K). Taken together, our data demonstrate that cIAP1UBAmut retains E3 ligase activity, and a functional UBA domain is dispensable for embryonic development or routine tissue homeostasis. Additionally, we conclude that a functional UBA domain of cIAP1 is not required for Ub-dependent formation of complex-I, activation of the canonical NF-κB pathway, and suppression of non-canonical NF-κB signaling.

12Next, we tested the response of cIAP1UBAmut mice to TNF challenge. Injection of TNF provokes systemic inflammation that is driven by RIPK1 kinase-dependent cell death and resembles clinical SIRS. Strikingly, cIAP1UBAmut mice were much more sensitive to TNF-induced SIRS than WT and cIAP2−/− counterparts. Accordingly, following administration of a dose of murine TNF as low as 4 μg/20 g of body weight, cIAP1UBAmut mice exhibited a dramatic drop in body temperature and significant increase in morbidity (Figures 4A and 4B). TNF-treated cIAP1UBAmut mice also had significantly elevated levels of aspartate transaminase (AST), alanine transaminase (ALT), and lactate dehydrogenase (LDH) in the plasma, indicating liver and tissue damage (Figures 4C and 4D). Consistently, livers from cIAP1UBAmut mice had higher numbers of TUNEL-positive cells than cIAP2−/− or WT littermate control mice (Figures 4E and 4F). Collectively, these data demonstrate that a functional UBA domain in cIAP1 is required to protect mice from the lethal effects of TNF.

13Next, we examined the role of the UBA domain in repressing TNF-induced cell death in primary BMDMs, mouse dermal fibroblasts (MDFs), and MEFs. TNF treatment did not induce substantial cell death in either WT or cIAP2-deficient cells; however, it was a potent cell death stimulus in cIAP1UBAmut cells (Figures 5A, 5C, 5D, S3A, and S3B). This cell death was RIPK1 kinase dependent because treatment with the selective RIPK1 inhibitor GSK′963 suppressed TNF killing. Likewise, primary cIAP1UBAmut BMDMs were exquisitely more sensitive to RIPK1-mediated TNF-induced necroptosis than BMDMs from either WT littermates or single targeted cIAP2−/− animals (Figure 4A). Consistently, increased association of RIPK1 with RIPK3 was detected by proximity ligation assay (PLA) (Figure 5B). Primary MDFs and MEFs instead seemed to die by apoptosis because RNAi-mediated depletion of MLKL had no apparent effect on TNF-induced cell death in cIAP1UBAmut MEFs (Figure S3D). Indeed, TNF stimulation of cIAP1UBAmut MDFs and MEFs caused elevated levels of caspase activity (Figure S3E), which was accompanied with enhanced complex-II formation (Figures 5D and 5F), and cleavage and activation of caspase-8 and caspase-3 (Figure S3F). Importantly, cIAP1UBAmut MEFs, prepared from multiple embryos within the same litter and between litters, behaved in the same manner (Figures S3A–S3C). Together, these data suggest that the UBA domain allows cIAP1 to inhibit TNF-induced death.

14Because ubiquitylation of RIPK1 and other components of complex-I represses RIPK1-dependent formation of complex-II, we analyzed complex-I formation in MDFs and MEFs. As previously reported, deficiency of cIAP1 and cIAP2 prevented ubiquitylation of RIPK1 and the recruitment of the LUBAC components SHARPIN and HOIL-1 5 min after TNF stimulation (Figure 6A). We found no evidence for defective ubiquitylation of RIPK1 in complex-I in single-targeted cIAP2−/−, but cIAP1-deficient MEFs had significantly lower molecular weight-modified forms of RIPK1 and substantial levels of non-modified RIPK1. cIAP1UBAmut cells, however, displayed a marked and reproducible decrease in the extent of high molecular weight RIPK1 ubiquitylation (red arrows) in complex-I in both MEFs and MDFs (Figures 6A, 6B, and S4A), compared to cIAP2−/− and WT cells. Importantly, the extent of non-modified RIPK1 in complex-I was indistinguishable between WT, cIAP2−/−, and cIAP1UBAmut cells (Figures 6A and 6B), suggesting that equivalent amounts of RIPK1 undergo Ub modifications in these three genotypes. Because the same amount of RIPK1 is being ubiquitylated in WT, cIAP2−/−, and cIAP1UBAmut cells, but overall ubiquitylation of RIPK1 seems to be affected in cIAP1UBAmut cells compared to WT and cIAP2−/− cells, we conclude that RIPK1 undergoes Ub modifications that are distinct from those observed in WT, cIAP1−/−, or cIAP2−/− cells. Although the UBA mutation resulted in reduced levels of ubiquitylated RIPK1 in complex-I, this had no apparent effect on the kinetics of the recruitment of other components of the TNF-RSC such as SHARPIN and HOIL-1 (Figures 5A and 5B).

15Next, we addressed whether the weakened association of cIAP1UBAmut with TRAF2 might lead to decreased recruitment of cIAP1 to complex-I and decreased ubiquitylation mediated by cIAP1 (Figures 6A and 6B). To this end, we increased the amount of cIAP1UBAmut to the levels of WT cIAP1 in complex-I using a doxycycline-inducible reconstitution approach. This allowed us to induce the expression of cIAP1UBAmut so that a comparable amount of cIAP1UBAmut protein was present in complex-I as in WT or cIAP2 KO cells (Figure 6C). Even though the levels of cIAP1UBAmut in complex-I were comparable to cIAP1 WT, this did not “normalize/correct” the ubiquitylation pattern of RIPK1, nor did it have an impact on the recruitment of unmodified RIPK1 in complex-I (Figure 6C). This demonstrates that the different smearing pattern of ubiquitylated RIPK1 in cIAP1UBAmut cells is not due to impaired recruitment of cIAP1 to complex-I, and, therefore, is not TRAF2 binding dependent but merely UBA dependent.

16To provide a robust analysis of the composition of Ub linkage types on RIPK1 in cIAP2−/− and cIAP1UBAmut cells, we employed absolute quantification (AQUA)-based mass spectrometry analysis of Ub chain composition on complex-I-derived RIPK1. To this end, we performed two consecutive IPs, first of complex-I (FLAG-TNF) and then of RIPK1 (Figure 7A), followed by AQUA-based absolute quantification of chain types using mass spectrometry. Our analysis revealed that K48-linked chains on RIPK1 were reproducibly less abundant (>10% reduction) in cIAP1UBAmut,cIAP2−/− mutant cells compared to cIAP2−/− (Figures 7A and S5A). Besides K63-linked Ub chains, no other linkage types were reproducibly detected on RIPK1 in MEFs, although other groups previously reported the presence of K11- and M1-linked chains on RIPK1, using different cell types and conditions. Calculations from the AQUA-based mass spectrometry experiment indicated that the majority of Ub is conjugated in the form of mono-Ub moieties rather than chains, and the actual chains on RIPK1 are surprisingly short, even in the control situation (Figure S5A). Given that the overall smearing pattern is reduced in cIAP1UBAmut cells, this suggests not only that K48 ubiquitylation is affected, but also that the occupancy of Ub-acceptor K residues is altered.

17To study the occupancy of Ub-acceptor lysines of RIPK1, we analyzed ubiquitylated RIPK1 from complex-I using Ub chain restriction (UbiCRest) (Figure 7B). To this end, we used a combination of deubiquitylating enzymes (vOTU/OTULIN) that remove all chain types but leave the most proximal Ub attached to RIPK1. vOTU hydrolyzes all Ub-linkage types except M1-linked chains, which can be cleaved by OTULIN. Incubation with vOTU/OTULIN revealed a reduction in the Ub-site occupancy in cIAP1UBAmut compared to cIAP2−/− (Figure 7B). The reduced Ub occupancy of RIPK1 might help to explain the shift toward the lower molecular weights of ubiquitylated RIPK1 in cIAP1UBAmut cells.

18Because K48-linked chains as well as poly-mono-ubiquitylation can target proteins for degradation, we tested whether the combined reduction in poly- and mono-ubiquitylation of RIPK1 affects the stability of RIPK1 in complex-I. Using tandem Ub-binding entities (TUBEs), which allow isolation of poly-ubiquitylated proteins, we found that the levels of ubiquitylated RIPK1 dramatically accumulated in cIAP1UBAmut cells compared to WT cells over a 6-hr time period following TNF treatment (Figure 7C). Importantly, TNF-induced accumulation of ubiquitylated RIPK1 in cIAP1UBAmut cells coincided with a significant increase in formation of complex-II and activation of caspase-8 (Figure 7C). This demonstrates that the UBA domain of cIAP1 represses lethal accumulation of RIPK1, most likely by facilitating efficient poly-mono as well as K48-mediated ubiquitylation and degradation of RIPK1, which would lower the number of “seeding” molecules for formation of complex-II. Importantly, while ubiquitylated RIPK1 accumulated over time in TNF-treated cIAP1UBAmut cells, treatment of cIAP1UBAmut cells with proteasome inhibitors (MG132) did not result in a further increase in RIPK1 accumulation (Figure 7D), corroborating the notion that the stabilization effect is due to the UBA mutation. In contrast, proteasome inhibition in WT cells resulted in a substantial accumulation of RIPK1 in the ubiquitylated proteome (Figure S5A), indicating that under normal conditions RIPK1’s stability is regulated, at least in part, in a Ub- and proteasome-dependent fashion. Of note, because RIPK1 forms amyloid-like structures upon activation, the TUBE-based experiment also purifies non-ubiquitylated RIPK1 that is bound to ubiquitylated RIPK1. Moreover, using a broad-spectrum DUB inhibitor (PR619), we found no evidence for a role of the UBA domain in shielding ubiquitylated RIPK1 from DUB digestion (Figures S5B and S5C).

19TNF-induced activation of IKK and MK2 directly suppresses the kinase activity and cytotoxic potential of RIPK1. In particular, MK2 directly phosphorylates mouse RIPK1 at serine (S)321 and S336 (S320 and S335 in human), which in turn suppress RIPK1 auto-activation at S166. Because cIAP1UBAmut cells are sensitized to RIPK1 kinase-mediated cell death in response to TNF, we addressed whether the UBA mutation affects MK2-mediated suppression of RIPK1 kinase activity. RIPK1 IP from TNF-treated cIAP2−/− and cIAP1UBAmut cells revealed a strong increase in auto-phosphorylation at S166 in cIAP1UBAmut cells, which is entirely consistent with the notion that TNF causes auto-activation of RIPK1 and RIPK1 kinase-dependent cell death in these cells (Figure 7E). However, activation of MK2 and MK2-mediated phosphorylation of S321 appeared normal in cIAP1UBAmut cells, demonstrating that activation of RIPK1 kinase activity in cIAP1UBAmut cells was MK2 independent. Activation of IKK and NF-κB-mediated expression of target genes was equivalent in WT, cIAP2−/−, and cIAP1UBAmut cells (Figures 3G–3J), suggesting that IKK-mediated regulation of RIPK1 was not perturbed in cIAP1UBAmut cells. Consistent with this view, we find that inhibition of IKK with TPCA-1 further sensitized cIAP1UBAmut cells to TNF killing (Figure 7F). Together, our data are consistent with a model whereby cIAP1 regulates RIPK1 kinase activity not only by activating downstream kinases, such as IKK and MK2, but also by directly repressing RIPK1 kinase activity via Ub-dependent inactivation.

20Ub-mediated inactivation of RIPK1 has long been postulated to contribute to the regulation of cytokine-induced cell death. However, detailed insights into the determinants and the actual molecular and functional consequences of RIPK1 ubiquitylation have not been demonstrated. Here, we show that the UBA domain of cIAP1 interacts with TRAF2 and is required for proper regulation of RIPK1 kinase activity. In the absence of a functional UBA domain, more active RIPK1 kinase accumulates in response to TNF, causing RIPK1 kinase-mediated cell death and systemic inflammatory response syndrome.

21UBA-dependent ubiquitylation of RIPK1 seems to regulate RIPK1 kinase activity through two potentially interconnected mechanisms: first, the UBA contributes to optimal Ub occupancy of RIPK1. This is evident as fewer lysine residues are conjugated to Ub in cells from cIAP1UBAmut animals. While this reduction in ubiquitylation of RIPK1 has no effect on its scaffolding function, such as activation of TAK1, IKK, MK2, and NF-κB, and the production of cytokines, reduced RIPK1 ubiquitylation compromises regulation of RIPK1 kinase activity, leading to enhanced RIPK1 auto-phosphorylation and formation of complex-II. Second, the UBA domain of cIAP1 also influences the cytotoxic potential of RIPK1 by targeting it for proteasomal degradation.

22Although the overall difference in K48 ubiquitylation of RIPK1 5 min after TNF treatment was reproducibly small, this causes a significant alteration of RIPK1’s stability as active, ubiquitylated RIPK1 accumulates over time in cIAP1UBAmut cells. Multiple Ub modifications, which include short Ub chains, constitute a powerful degradation signal. Hence, it is likely that the combined reduction in poly- and mono-ubiquitylation of RIPK1 ultimately contributes to the increased protein stability of RIPK1 in complex-I from cIAP1UBAmut cells, leading to exacerbated RIPK1 kinase activity in response to TNF. Thus, under normal conditions, the conjugation of Ub to RIPK1 might impede its auto-activation and, in addition, mark it for proteasomal degradation, thereby limiting accumulation of cytotoxic RIPK1. Nevertheless, cIAP1 clearly also suppresses the cytotoxic potential of RIPK1 in a UBA-independent manner. Thus, cIAP1 also represses RIPK1 auto-activation by facilitating Ub-mediated recruitment of LUBAC, and activation of IKK and MK2. This enables IKK- and MK2-mediated regulation of RIPK1 via inhibitory phosphorylation. Our current data are consistent with a revised model for TNF signaling whereby RIPK1’s kinase activity is suppressed through both Ub-mediated phosphorylation of RIPK1 by IKK and MK2 as well as by direct Ub-mediated inactivation and degradation of RIPK1. However, we cannot formally exclude the possibility that the UBA also enables the recruitment of a yet to be identified kinase, which might repress RIPK1.

23Unexpectedly, we find that the UBA domain of cIAP1 contributes to efficient TRAF2 binding. Consistent with the notion that cIAP1 interacts with TRAF2 through a multivalent interaction via its BIR1 as well as UBA domain, we find that point mutations in either the BIR1 or UBA domain weaken the interaction between cIAP1 and TRAF2. Although cIAP1UBAmutant proteins bind less well to TRAF2 in coIP studies under resting conditions, under in vivo settings this association is sufficient to maintain TNF-induced activation of NF-κB or support TRAF2/TRAF3-mediated degradation of NIK. Importantly, despite the fact that UBA mutant cIAP1 interacts less efficiently with TRAF2 under IP conditions, the reduced ubiquitylation of RIPK1 in complex-I observed in cIAP1UBAmutant cells is not due to impaired cIAP1 recruitment. This is evident as elevating the recruitment of cIAP1UBAmut into complex-I does not “normalize/correct” the ubiquitylation pattern of RIPK1. Further evidence is provided by the fact that the UBA mutation of cIAP1 does not phenocopy loss of TRAF2. Accordingly, RIPK1 is readily ubiquitylated in complex-I in cells from cIAP1UBAmut animals. In contrast, RIPK1 ubiquitylation in complex-I is completely lost in Traf2−/− cells. Taken together, our data demonstrate that cIAP1UBAmut retains E3 ligase activity, and the UBA domain is dispensable for embryonic development, routine tissue homeostasis, or NF-κB regulation. Thus, the elevated sensitivity to the cytotoxic potential of TNF cannot be explained by the regulation of NF-κB, IKK, or MK2. We propose that the UBA mutation provides rare insight into the protein architecture of complex-I. We suggest that the UBA contributes to the proper positioning of the RING domain of cIAP1 within complex-I so that the Ub-loaded E2 enzyme can optimally transfer ubiquitin to RIPK1.

24As TNF is a key player in the cytokine network that supports inflammation-associated cancer and cancer-related inflammation, it will be important to gain a better understanding of the checkpoints that control life and death decisions in response to TNF. A better understanding of such checkpoints could lead to new approaches for the treatment of chronic inflammatory diseases that are fueled by aberrant RIPK1-induced cell death, and/or reveal novel strategies for anti-cancer immunotherapies that harness RIPK1’s ability to trigger immunogenic cell death.

25Further information and requests for reagents may be directed to and will be fulfilled by the Lead Contact, Pascal Meier (pmeier@icr.ac.uk).

26The cIAP2 and cIAP1 genes are positioned 10 Kb apart on the same chromosome. Hence, they recombine as a single genomic locus. The cIAP2FRT/FRTcIAP1minigene mouse was generated by electroporating the cIAP1 targeting vector into C57BL6-derived ES cells that were previously targeted on the cIAP2 locus. For the cIAP1 targeted allele the M396A and F398A mutations were introduced into exon 4 and a F3 site-flanked PGK-Hygro selection cassette was inserted upstream of the genetically altered exon 4. A mini-gene corresponding to exon 4-7 of cIAP1 and a BGH poly-A signal were placed upstream of the selection cassette. The cDNA mini-gene is flanked by loxP sites. Mice carrying the cIAP2FRT/FRTcIAP1minigene alleles were crossed to transgenic mice expressing the FLPo recombinase to generate the cIAP2−/−cIAP1minigene animals. The PGK-Hygro resistance cassette on the cIAP1 targeted allele was also deleted by the FLPo-mediated recombination. The cIAP2−/−cIAP1minigene mice were subsequently crossed to transgenic mice expressing the Cre recombinase to delete the mini-gene and generate the cIAP2−/−cIAP1UBAmut animals. cIAP2−/−cIAP1loxP/loxP mice were previously described. Mice were kept according to the UK Home Office regulations. In vivo experiments were conducted according to institutional, national and European animal regulations. Animal protocols were approved by the Ethics Committee of Ghent University.

27Experiments in mice were performed at the Department of Pharmacology of the Faculty of Veterinary Medicine of the Ghent University, Belgium, according to institutional, national and European regulations. Animal protocols were approved by the ethics committee of Ghent University. mTNF was diluted in endotoxin-free PBS and injected intravenously (i.v.) in a volume of 0.2 ml. Rectal body temperature was recorded with a digital thermometer (model 2001; Comark Electronics). Plasma samples and tissue samples of liver were collected at designated times after injection. Blood was obtained by cardiac puncture.

28Primary and immortalized MEFs, primary MDFs and Flp-InTM-RexTM-HEK293 cells were cultured in Dulbecco’s modified Eagle’s Medium (DMEM) supplemented with 10% Fetal Bovine Serum (FBS), penicillin and streptomycin under 10% CO2.

29Primary Mouse Embryonic Fibroblasts (MEFs) were generated from E13.5 embryos. After removing the placenta, yolk sac, head and the dark red organs, embryos were finely minced and digested for 20 min in 0.25% trypsin. Single cell suspension was then obtained by pipetting up and down the digested embryos. Mouse Dermal Fibroblasts (MDFs) were isolated as described in. To generate Bone Marrow Derived Macrophages (BMDMs), bone marrow cells from tibia and femur of 2 month old mice were seeded in non-coated Petri dishes and cultured for 6 days in Dulbecco’s modified Eagle medium + 10% fetal bovine serum + 20% (v/v) L929 mouse fibroblast conditioned medium. Keratinocytes were isolated as described in. Splenocytes were isolated from 2 month old mice. Mouse spleens were mashed through a cell strainer into the Petri dish using the plunger end of a syringe. Cells were then washed once in cold PBS and treated with 1X Red Blood Cell Lysis Buffer (BioLegend, Cat N 420301) for 5 min on ice. Cells were then washed again in PBS and counted.

305 × 105 splenocytes were resuspended in 500 μL of cold PBS and stained with DAPI (1/5000) for 20 min on ice. Cells were then washed with cold PBS and resuspended in 50 μL Staining Buffer. 2 μL of blocking antibody (Anti-Mouse CD16/CD32) were added, and cells were kept on ice for 10 min. 50 μL of Staining Buffer containing the desired antibodies were then added and cells were kept on ice for 30 min. Cells were washed in cold PBS, resuspended in 1 mL of cold PBS and analyzed by FACS.

31For the generation of the 293 stable cell lines, where endogenous cIAP1 was reconstituted with WT and mutant cIAP1, Flp-InTM T-RExTM HEK293shcIAP1 cells were created. First Flp-InTMT-RExTM-HEK293 cells (Invitrogen) were transduced with lentiviral particles targeting the 3′ UTR of the cIAP1 mRNA. To this end, we used pTRIPZ-shcIAP1 (Open Biosystems), which allows Doxocycline-inducible expression of a miR30-based shcIAP1 RNA. After puromycin selection, individual clones were tested for cIAP1 knockdown efficiency. Such cells were further tested functionally using TNF signaling and NIK activation as readouts. Next, the respective cIAP1 constructs were cloned into pcDNA5.1/FRT/TO (Invitrogen). The empty pcDNA5.1/FRT/TO-2HA-Strep plasmid was used as a control. To generate site-specific single copy insertions, pcDNA5.1FRT/TO-based plasmids were co-transfected with pOG44 into Flp-InTMT-RExTM-HEK293shcIAP1 cells. After selection with hygromycin, stable cells were selected. Endogenous cIAP1 suppression and expression of WT or mutant versions of cIAP1 were simultaneously induced by treating cells with 100 ng/ml Doxycycline for at least 48 hours.

32The GSK’963 RIPK1 kinase inhibitor was provided by GSK. The following antibodies were used: α-RIPK1 (BD Biosciences, 610459), α-HOIL (gift from Henning Walczak), α-cIAP1 (Enzo, ALX-803-335-C100), α-TNFR1 (Abcam, 19139), α-Actin (Santa Cruz Biotechnology, sc-1615), α-P-p65 (Cell Signaling, 3033), α-p65 (Cell Signaling, 8242), α-IkBα (Santa Cruz, sc-371), α-P-p38 (Cell Signaling, 9215), α-p38 (Cell Signaling, 9212), α-P-JNK (Cell Signaling, 9255), α-JNK (Santa Cruz Biotechnology, sc-571), α-P-ERK (Cell Signaling, 9101), α-ERK (gift from Chris Marshall) α-caspase-8 (Cell Signaling, 9429), α-FLAG [M2] (SIGMA, M8823), α-Ub (Dako, Z0458), α-FLIP (Adipogene, AG-20B-0056), α-FADD (Santa Cruz Biotechnology, sc-6036), α-RIPK3 (ProSci, 2283), α-Tubulin (SIGMA, T9026), α-SHARPIN (Proteintech, 14626-1-AP), α-TRAF2 (Cell Signaling, 4712), α-CD8-PE-Cy7, GR-1-PE-Cy7, CD11c-FITC, CD4-FITC, CD11b-Cy5, B220-FITC (gift from Henning Walczak), α-CD69-PE (eBioscience, 12-0691-82), α-CD3-APC (eBioscience, 47-0032-82), and α-CD16 (eBioscience, 14-0161-82).

33The UbiCRest analysis with linkage selective DUBs was performed essentially as described in. Briefly, the release fraction (see complex-I purification) was incubated with the following DUBs: 1 μM OTULIN, 0.5 μM vOTU, 1.5 μM USP21. The reaction was conducted in the presence of 1 mM DTT for 30 min at 37°C. Reactions were stopped with SDS sample buffer, and the ubiquitylation status analyzed by western blotting.

34Cells were lysed in DISC lysis buffer (20 mM Tris-HCL pH7.5, 150 mM NaCl, 2 mM EDTA, 1% Triton X-100, 10% glycerol) supplemented with protease inhibitors, 1 mM DTT, PR619 (10 μM) and GST-TUBE (50 μg/ml; 50 μg TUBE/mg protein lysate). Cell lysates were rotated at 4°C for 20 min then clarified at 4°C at 14,000 rpm for 10 min. 20 μL GST beads were added and immunoprecipitations were performed overnight. Beads were washed 4x in wash buffer (50 mM Tris pH 7.5, 150 mM NaCl, 0.1% Triton X-100, and 5% glycerol) + PR619 (10 μM), and bound proteins eluted by boiling in 50 μ l 1x SDS loading dye.

35Cells were seeded in 15 cm dishes and treated as indicated with 3x FLAG-hTNF (5 μg/ml). To terminate stimulation, media was removed and plates were washed with 50 mL of ice cold PBS. Plates were frozen at −80°C until all time points were acquired. Plates were thawed on ice and cells were lysed in 1% Triton X-100 lysis buffer (30 mM Tris-HCl pH 7.4, 120 mM NaCl, 2 mM EDTA, 2 mM KCl, 10% glycerol and 1% Triton X-100, supplemented with protease inhibitors and PR619 (10 μM). Cell lysates were rotated at 4°C for 20 mins then clarified at 4°C at 14,000 rpm for 30 mins. Proteins were immunoprecipitated from cleared protein lysates with 20 μL of α-FLAG M2 beads (SIGMA) with rotation overnight at 4°C. For the 0 hr sample 5 μg/ml of FLAG-TNF were added post-lysis. 4x washes in 1% Triton X-100 buffer with PR619 (10 μM) were performed, and samples eluted by boiling in 60 μL 1x SDS loading dye. For complex-II purification cells were seeded in 10 cm dishes and treated as indicated using media containing 1x FLAG-TNF (100 ng/ml) and zVAD (10 μM). Cells were lysed on ice in 1% Triton X-100 lysis buffer (30 mM Tris-HCL pH 7.4, 120 mM NaCl, 2 mM EDTA, 2 mM KCl, 1% Triton X-100 supplemented with protease inhibitors and 10 μM PR619). Cell lysates were rotated at 4°C for 20 mins then clarified at 4°C at 14,000 rpm for 10 mins. 20 μL of protein G Sepharose (SIGMA), blocked for 1 hr with lysis buffer containing 1% BSA, were bound with FADD antibody (1.5 μg antibody/mg protein lysate) and were rotated with cleared protein lysates 4 hr at 4°C. 4x washes in lysis buffer were performed, and samples eluted by boiling in 80 μL 1x SDS sample buffer.

362 × 105 cells (MEFs and MDFs) were seeded in six well plates and 24 hr later they were treated as indicated for an additional 24 hr. Hoechst 33342 (10 μg/ml) and Propidium Iodide (PI) (1 μg/ml) were added. After 2 mins 10 to 15 images per well were taken with a fluorescent inverted microscope and the ration of dead/live cells were counted manually. 5 × 104 BMDMs were seeded in 96 well plates and 24 hr later they were treated as indicated for an additional 24 hr. Hoechst (0.5 μg/ml) and PI (1 μg/ml) were added and the ratio dead/live cells was measured using the Celigo imaging system.

37Various human cIAP2 and TRAF2 segments were cloned into pSUMO vector respectively to produce N-terminally His-SUMO tagged proteins. The constructs were then transformed into BL21 (DE3) cells and cultured in LB medium at 37°C, respectively. Protein expression was induced overnight at 20°C with 0.5 mM IPTG when OD600 reached 0.8. Cells were lysed in buffer containing 25 mM Tris-HCl at pH 8.0, 150 mM NaCl, 10 mM imidazole and 5 mM β-mercaptoethanol. The recombinant proteins were affinity-purified by Ni-Sepharose beads (GE Healthcare Life Sciences). The SUMO tag was removed by overnight digestion with homemade ULP1 protease at 4°C. The untagged proteins were further purified by HiTrap Q anion exchange and Superdex 200 gel filtration chromatography (GE Healthcare Life Sciences). Isothermal titration calorimetry measurements were performed at 16°C, using a MicroCal ITC200 microcalorimeter (MicroCal). For the TRAF2:cIAP2 interactions, the calorimetric titrations were performed by injecting 2 μL of cIAP2 protein solution (2–4 mM) into a sample cell containing 200 μl 0.20 mM TRAF2 protein in 25 mM Tris-HCl at pH 8.0, 150 mM NaCl. A total of 20 injections were performed with a spacing of 150 s and a reference power of 6 μcal/s. Binding isotherms were plotted and analyzed using Origin Software (MicroCal).

382 × 105 cells (MEFs) were plated in 6-well plates and treated as indicated in 2 mL for the indicated times. After treatment, media was removed, and 300 μL 1% DISC lysis buffer (20 mM Tris-HCL pH 7.5, 150 mM NaCl, 2 mM EDTA, 1% Triton X-100, 10% glycerol) was added to each well, cells were scraped and lysates were left on ice for 5 min. 50 μL of lysate per condition were transferred into a 24 well plate and 450 μL DEVDase assay mix (20 mM Ac-DEVD-AMC (Sigma), 1 mM DTT, 50 mM Tris pH 7.5, 150 mM NaCl, 0.1% Triton X-100, and 5% glycerol) was added to each well (NB: cell lysates were not cleared). Plates were wrapped in foil and reactions allowed to proceed by incubation at room temperature for up to 24 hr. DEVDase activity was read at 380 nM excitation/460 nM emission.

39The yeast strain Y2HGold (Clontech) was co-transformed with the respective bait and prey plasmids. Positive transformants were selected on minimal SD-Leu-Trp medium (Formedium). Three single colonies for each bait and prey co-transformation were patched out on fresh SD-Leu-Trp plates and grown for 2 days at 30°C. Each patch was resuspended in 180 μL of sterile water in a 96-well plate and replica plated onto non-selective (SD-Leu-Trp) or selective medium (SD-Leu-Trp-His), containing the indicated concentration of 3-amino-1,2,4-triazole (3-AT, Formedium). Yeast plates were incubated at 30°C for 1 week. UbcH5b prey vector was provided by Rachel Klevit.

40BL21 cells were transformed with pGEX6p-1-3XFLAG-TNF plasmid. One colony was picked and incubated o/n in 100 mL LB medium with 100 μg/ml ampicillin. Next day 900 mL of LB medium without AMP were added, cells were grown for 1 h at 37°C and then 0.5 mM IPTG was added for further 4 h. Bacteria were spun for 15 min at 4000 rpm, the supernatant was discarded and the pellet was lysed in 10 mL Triton X-100 lysis buffer (10 mM Tris pH 7.5, 150 mM NaCl, 10% glycerol and 1% Triton X-100 with complete protease inhibitor cocktail). The lysate was sonicated and left for 15 min at 4°C and clarified for 30 min at 140,000 rpm at 4°C. Lysate was rotated with glutathione Sepharose beads for 4 hr at 4°C and 3x washed with IPPG150 buffer (0,1% Triton X-100, 50 mM Tris pH 7.5, 150 NaCl, 5% glycerol) were performed, followed by a final wash in PreScission cleavage buffer (50 mM Tris pH 7, 150 mM NaCl, 1 mM EDTA and 1 mM DTT). For PreScission cleavage 500 μL cleavage buffer and 30 μL PreScission enzyme were added to the beads at 4°C o/n. The beads were spun, and supernatants were collected and passed through Thermo Scientific buffer exchange columns to remove bead contamination. Recombinant TNF concentration was determined on Coomassie Blue-stained polyacrylamide gel using known amount of BSA as standard, and quantitated using Image Lab software.

41PLA was performed according to the manufacturer’s protocol using the Duolink Detection Kit (SIGMA). Cells were examined with a confocal microscope (Objective 40x, Zeiss LSM 710).

42Ub chains were separated on a NuPAGE 4%–12% gradient gel (Invitrogen) before in-gel digestion with trypsin and the addition of Ub-AQUA peptide internal standards according to. 10 μL of each sample was directly injected onto an EASY-Spray reverse-phase column (C18, 3 μm, 100 Å, 75 μm × 15 cm) using a Dionex UltiMate 3000 high-pressure liquid chromatography system (Thermo Fisher Scientific) and analyzed on a Q-Exactive mass spectrometer (Thermo Fisher Scientific) using parallel reaction monitoring (PRM), similar to. Data were analyzed further according to

43Raw data have been deposited to Mendeley Data and are available at https://doi.org/10.17632/f559yfv4h4.1.