MIND bomb 2 prevents RIPK1 kinase activity-dependent and -independent apoptosis through ubiquitylation of cFLIPL

Mind bomb 2 (MIB2) is an E3 ligase involved in Notch signalling and attenuates TNF-induced apoptosis through ubiquitylation of receptor-interacting protein kinase 1 (RIPK1) and cylindromatosis. Here we show that MIB2 bound and conjugated K48– and K63–linked polyubiquitin chains to a long-form of cellular FLICE-inhibitory protein (cFLIPL), a catalytically inactive homologue of caspase 8. Deletion of MIB2 did not impair the TNF-induced complex I formation that mediates NF-κB activation but significantly enhanced formation of cytosolic death-inducing signalling complex II. TNF-induced RIPK1 Ser166 phosphorylation, a hallmark of RIPK1 death-inducing activity, was enhanced in MIB2 knockout cells, as was RIPK1 kinase activity-dependent and -independent apoptosis. Moreover, RIPK1 kinase activity-independent apoptosis was induced in cells expressing cFLIPL mutants lacking MIB2-dependent ubiquitylation. Together, these results suggest that MIB2 suppresses both RIPK1 kinase activity-dependent and -independent apoptosis, through suppression of RIPK1 kinase activity and ubiquitylation of cFLIPL, respectively.

4.C-FLIP seem to undergo K48 and K63 ubiquitinylation by MIB2 (Highlighted in Figure 4). Moreover, the authors applied a very elegant quantitative mass spectrometry approach in order to quantify the degree of the ubiquitinylation of cFLIPL. However, it has to be said that surprisingly, the absence of MIB2 decreased K48 and K63 ubiquitinylation of cFLIPL only two fold. Hence, my question is how the two times difference in the degree of ubiquitilation would have such dramatic influence on the functional outcome? 5.Along these lines, in HeLa cells cFLIP is expressed only to the lower levels. Hence, I do not see how this low level expression of cFLIPL and the two-times decrease of its K63 and K48 levels can lead to such a drastic differences in apoptosis and formation of complex II upon KO of MIB2 6. Figure 7: Apparently TNF/BV6 treatment does not cause the formation of complex II in HeLa cells, which contradicts several reports. More detailed titration and kinetics have to be done, as it is rather not likely. Many other groups reported the assembly of the complex II in the similar conditions and the cells are dying which means they have to have complex II 7. Figures 6,7, the levels of cFLIPL are not changed plus minus MIB2, this indeed rules out any regulation of apoptosis via expression levels of cFLIP and probably leaves the room only for checking out the role for K63 chains.
Summarizing, there is no evidence(experimental or theoretical) presented that it is indeed ubiquitilation of c-FLIP leads to the different sensitivity of HeLa MIB2 KO cells I would rather assign the observed effects to the effects of MIB2 on RIPK1. Hence the functional role of c-FLIP ubiquitinylation by MIB2 still have to be uncovered Minor comments: 8.Extended figure 3: in the quantification of c-FLIP levels the loading control is overexposed, hence, it is difficult to draw the conclusion. C-FLIPs is always quicker degraded, which was also observed in both cell lines ( KO MIb2 versus wt) so one can not draw a conclusion that this is yet another evidence of MIB2 action 9.In TNF signaling complex I is not termed ‚death inducing signaling complex' but just complex I. With complex IIa and IIb, there are certainly some different views on this nomenclature in the literature, as some people name complex IIb as necrosome or RIPK1/RIPK3 complex and complex IIa as ripoptosome, but in a particular review that was cited by authors indeed complex IIb is assigned to FADD/caspase-8/c-FLIP complex and complex IIa to TRADD/caspase-8/FADD complex. I personally prefer the first nomenclature.
Reviewer #2 (Remarks to the Author): Nakabayashi report c-FLIP-long as a novel binding partner and substrate of the E3 ligase MIB-2. MIB-2-dependent ubiquitylation of c-FLIP-long is shown to suppress TNF-induced apoptosis. The conclusions are largely justified by the data, although the notion that ubiquitylation is affecting c-FLIP stability is less certain. For example, unmodified c-FLIP may just be more prone to aggregation than degradation in extended fig. 3. Do they get the same result if they examine the insoluble fraction for c-FLIP?
My other main criticisms are: 1) It is not clear anywhere in the text, methods, or legends if experiments exclusively use human rather than mouse proteins. 2) Throughout the figures, the efficiency of each IP is not confirmed by WB for the protein that is immunoprecipitated. These quality control WBs should be included in every experiment.
3) The authors should indicate exactly which cysteines are mutated in R1M and R2M MIB-2. 4) Fig. 3C should include WBs using linkage-specific Ub antibodies (K48, K63, and K11 antibodies). The linkages generated in this more defined in vitro setting would be useful to compare to those generated in cells. 5) The c-FLIP input WB in Fig. 5f should be less cropped so that the reader can see the loss of ubiquitylation on the 3A mutant (as compared to the WT control). i.e. are the slower migrating forms absent?
In this manuscript, Nakabayashi et al. show that E3 ligase Mind bomb 2 (MIB2) can bind and generate K48/K63-linked ubiquitin chains on cFLIPL. Previous study has shown MIB2 could ubiquitinate RIPK1 to inhibit RIPK1 kinase activity, thereby suppress TNF-induced apoptosis (Feltham et al., 2018). However, the authors found cells expressing MIB2-binding defective mutant of cFLIPL are also hypersensitive to TNF-induced apoptosis. Therefore, they draw the conclusion that MIB2 inhibits TNF-induced cell death through both ubiquitination of cFLIPL and inhibition of RIPK1 kinase activity. Overall, this study identified cFLIPL as a new substrate of E3 ligase MIB2 based on in vitro biochemistry data, and extended the current understanding of the function of MIB2 in the regulation of TNF-induced apoptosis. However, more experiments need to be performed to determine the molecular mechanism of how MIB2-mediated ubiquitination of cFLIPL suppresses TNF-induced apoptosis independent of RIPK1 kinase activity inhibition.
Specific comments: 1. Line 65, TAK1 inhibitors are thought to inhibit the phosphorylation but not ubiquitination of RIPK1.
2. In Fig. 1c, d, total HA-Ub and IP-Flag level were missing.
3. Extended Fig. 2c, the data is not sufficient to demonstrate that MIB2 ubiquitinates multiple lysine residues of cFLIPL. Since MIB2 could generate both K63-and K48-linked ubiquitin chains on cFLIPL, it may not be easy to detect the change of total ubiquitination level. The authors should also detect the K63-and K48-linked ubiquitination level in various mutants. Fig. 3a, b, the MIB2 on the stability of cFLIPL seems to have no much significance. This experiment is important and needs to be improved. Since the authors show MIB2 could generate K63-and K48-linked ubiquitination on cFLIPL, it is critical to make sure the role of MIB2 on the stability of cFLIPL, since MIB2 has been shown to be involved in TNF-induced apoptosis through regulating RIPK1. Therefore, the author should examine the stability of cFLIPL in WT and MIB2 KO cells under different apoptosis inducers (TNF alone; CHX alone; TNF+CHX; TNF+BV-6). In addition, cFLIPL is degraded very faster under apoptosis condition. The dose and time point of stimulators should be modified. Fig. 5a-5e, deficiency of MIB2 causes TNF-induced apoptosis due to enhanced RIPK1 kinase activity has been reported earlier (Feltham et al., 2018). It is critical to discriminate the relationship between MIB2-mediated ubiquitination of cFLIPL and MIB2-mediated suppression of RIPK1 kinase activity. Therefore, the authors should use RIPK1 kinase inhibitor or genetically knockout RIPK1 in MIB2-KO cells to determine the contribution of ubiquitination of cFLIPL to the apoptosis in MIB2-KO cells. In addition, the authors need to examine the RIPK1 kinase activity in cFLIPL-3A-reconstituted MEFs to determine whether MIB2-mediated ubiquitination of cFLIPL could suppress RIPK1 kinase activation.

5.
6. Fig. 6c, 6f, although reconstitution of MIB2 KO cells with MIB2-MT (Ligase activity defective) was comparable to the apoptosis in MIB2 KO cells, it is not convincing to conclude defective ubiquitination of cFLIPL mediated by MIB2 is required for TNF-induced apoptosis in MIB2 KO cells, since MIB2 could also ubiquitinate RIPK1. More experiments are needed to determine whether RIPK1 kinase-dependent cell death is dominant in MIB2-MT reconstitution cells. First, the authors should examine whether MIB2-MT can still interact and ubiquitinate RIPK1. Second, the cell death assay should add a control using RIPK1 kinase inhibitor (eg, Necrostatin-1) to examine the contribution of RIPK1 kinase activity in the cell death of MIB2-MT reconstitution cells. 7. Fig. 6h, the cFLIPL level in the immunoprecipitates was missing; and an IgG control should be added to exclude the non-specific binding.
8. Extended Fig. 5a, it is weird that IP-C, cFLIPL blotting has IgGs, but in IP-H, cFLIPL blotting nearly has no IgGs. 9. Line 367, a spelling error "compalex" 10. Line 342-343, although TNF plus CHX is initially thought to induce RIPK1-independent cell death. However, under some conditions such as TBK1 deficiency, it could also induce RIPK1 kinasedependent cell death (Xu et al., 2018). Since MIB2 could suppress RIPK1 kinase activation, TNF plus CHX may also induce RIPK1 kinase-dependent cell death in MIB2 KO cells. The authors should examine RIPK1 autophosphorylation and use RIPK1 kinase inhibitor to do the TNF plus CHX stimulation in MIB2 KO cells.
11. Line 378-382, K63-linked ubiquitination of RIPK1 has been reported to suppress RIPK1 kinase activity by recruiting TAK1 kinase (Tang et al., 2019). The authors should examine the TAK1, IKK, TBK1 and MK2 recruitment in TNFR1 complex1 in MIB2-KO cells to determine the exact molecular mechanism by which MIB2-mediated ubiquitination suppresses RIPK1 kinase activity.
In this manuscript, Nakabayashi et al. show that E3 ligase Mind bomb 2 (MIB2) can bind and generate K48/K63-linked ubiquitin chains on cFLIP L . Previous study has shown MIB2 could ubiquitinate RIPK1 to inhibit RIPK1 kinase activity, thereby suppress TNFα-induced apoptosis (Feltham et al., 2018). However, the authors found cells expressing MIB2-binding defective mutant of cFLIP L are also hypersensitive to TNFα-induced apoptosis. Therefore, they draw the conclusion that MIB2 inhibits TNF-induced cell death through both ubiquitination of cFLIP L and inhibition of RIPK1 kinase activity. Overall, this study identified cFLIP L as a new substrate of E3 ligase MIB2 based on in vitro biochemistry data, and extended the current understanding of the function of MIB2 in the regulation of TNFα-induced apoptosis. However, more experiments need to be performed to determine the molecular mechanism of how MIB2-mediated ubiquitination of cFLIP L suppresses TNFα-induced apoptosis independent of RIPK1 kinase activity inhibition.
Specific comments: 1. Line 65, TAK1 inhibitors are thought to inhibit the phosphorylation but not ubiquitination of RIPK1.
2. In Fig. 1c, d, total HA-Ub and IP-Flag level were missing.
3. Extended Fig. 2c, the data is not sufficient to demonstrate that MIB2 ubiquitinates multiple lysine residues of cFLIP L . Since MIB2 could generate both K63-and K48-linked ubiquitin chains on cFLIP L, it may not be easy to detect the change of total ubiquitination level. The authors should also detect the K63-and K48-linked ubiquitination level in various mutants. Fig. 3a, b, the MIB2 on the stability of cFLIP L seems to have no much significance. This experiment is important and needs to be improved. Since the authors show MIB2 could generate K63-and K48-linked ubiquitination on cFLIP L , it is critical to make sure the role of MIB2 on the stability of cFLIP L , since MIB2 has been shown to be involved in TNFα-induced apoptosis through regulating RIPK1. Therefore, the author should examine the stability of cFLIP L in WT and MIB2 KO cells under different apoptosis inducers (TNF alone; CHX alone; TNF+CHX; TNF+BV-6). In addition, cFLIP L is degraded very faster under apoptosis condition. The dose and time point of stimulators should be modified. Fig. 5a-5e, deficiency of MIB2 causes TNFα-induced apoptosis due to enhanced RIPK1 kinase activity has been reported earlier (Feltham et al., 2018). It is critical to discriminate the relationship between MIB2-mediated ubiquitination of cFLIP L and MIB2-mediated suppression of RIPK1 kinase activity. Therefore, the authors should use RIPK1 kinase inhibitor or genetically knockout RIPK1 in MIB2-KO cells to determine the contribution of ubiquitination of cFLIP L to the apoptosis in MIB2-KO cells. In addition, the authors need to examine the RIPK1 kinase activity in cFLIPL-3A-reconstituted MEFs to determine whether MIB2-mediated ubiquitination of cFLIP L could suppress RIPK1 kinase activation. 6. Fig. 6c, 6f, although reconstitution of MIB2 KO cells with MIB2-MT (Ligase activity defective) was comparable to the apoptosis in MIB2 KO cells, it is not convincing to conclude defective ubiquitination of cFLIP L mediated by MIB2 is required for TNFα-induced apoptosis in MIB2 KO cells, since MIB2 could also ubiquitinate RIPK1. More experiments are needed to determine whether RIPK1 kinase-dependent cell death is dominant in MIB2-MT reconstitution cells. First, the authors should examine whether MIB2-MT can still interact and ubiquitinate RIPK1. Second, the cell death assay should add a control using RIPK1 kinase inhibitor (eg, Necrostatin-1) to examine the contribution of RIPK1 kinase activity in the cell death of MIB2-MT reconstitution cells. 7. Fig. 6h, the cFLIP L level in the immunoprecipitates was missing; and an IgG control should be added to exclude the non-specific binding.

5.
8. Extended Fig. 5a, it is weird that IP-C, cFLIP L blotting has IgGs, but in IP-H, cFLIP L blotting nearly has no IgGs. 9. Line 367, a spelling error "compalex" 10. Line 342-343, although TNF plus CHX is initially thought to induce RIPK1-independent cell death. However, under some conditions such as TBK1 deficiency, it could also induce RIPK1 kinase-dependent cell death (Xu et al., 2018). Since MIB2 could suppress RIPK1 kinase activation, TNF plus CHX may also induce RIPK1 kinase-dependent cell death in MIB2 KO cells. The authors should examine RIPK1 autophosphorylation and use RIPK1 kinase inhibitor to do the TNF plus CHX stimulation in MIB2 KO cells.
11. Line 378-382, K63-linked ubiquitination of RIPK1 has been reported to suppress RIPK1 kinase activity by recruiting TAK1 kinase (Tang et al., 2019). The authors should examine the TAK1, IKK, TBK1 and MK2 recruitment in TNFR1 complex1 in MIB2-KO cells to determine the exact molecular mechanism by which MIB2-mediated ubiquitination suppresses RIPK1 kinase activity. This is a very important study, which identifies a new E3 ligase, MIB2, that ubiquitinylates cFLIPL and shows that it specifically binds to this protein at the C-terminal part as well as it ubiquitinylates this protein at the Cterminal and very N-terminal part. Importantly, they compare the effects of this kinase with ITCH and interestingly find that ITCH seems to have very minor effects on the ubiquitinylation of cFLIPL. They very elegantly show that this is a specific ligase for cFLIPL because it does not interact with cFLIPS. Moreover, they show that MIB2 binds to the caspase-like part of cFLIPL and ubiquitinylates the very C-terminal part of cFLIPL. Further, the authors analyse the role of this ubiquitinylation event in TNF-induced cell death. The study is supported by a number of very sophisticated biochemical experiments. Despite all these very important points, I have several major comments on the manuscript and on the mechanism suggested. RESPONSE: Thank you very much for your appreciation of our study and critical comments to improve our manuscript.
Major comments: 1. Ubiquitinylation of cFLIPL seems to be essential for the prevention of its interaction with procaspase-8 and formation of the DED filaments.
However, the authors identify the very C-terminal part of cFLIPL as a main target of ubiquitinylation. However, this part should play a secondary role in the recruitment of c-FLIPL to the DED filament, as the DED part of cFLIPL plays an essential role here. This hypothesis does not fit to the current knowledge on the DED filament formation RESPONSE: Thank you for pointing out a critical issue. Indeed, N-terminal DED domains of cFLIPL are required for DED filament formation with caspase 8. However, C-terminal domain of cFLIPL also mediates the interaction with caspase 8 (Shu et al., 1997;Yu et al., 2009). Moreover, as reported in previous studies (Hughes et al., 2016;Yu et al., 2009), substitution of glutamine at 468 with glutamic acid (Q468D) in the C-terminal caspase-like domain of cFLIPL fails to promote the processing of recombinant caspase 8 in a cell-free system. These results suggest that the C-terminal part of cFLIPL is involved in a proper heterodimer formation with caspase 8, and subsequent caspase 8 activation.
Notably, multiple lysine residues in the C-terminal part of cFLIPL, including K460, K462, K473, and K474, are adjacent to Q468 ( Supplementary Fig. 4a, b); thus, ubiquitylation of these lysine residues by MIB2 might block complex IIb maturation through steric hindrance. Consistent with the hypothesis, cFLIPL mutants lacking MIB2-dependent ubiquitylation, such as cFLIPL-3A or cFLIPL-K9R cannot completely restore anti-apoptotic function to levels found in WT cFLIPL (Fig. 9d, h). These results further substantiate a crucial role for Cterminal caspase-like domain of cFLIPL in attenuating caspase 8 activation. We made a new Fig. 10 to represent our model and mentioned the results in the text (line 206-211, 341-378, 402-411, 465-478, 518-527).

2.Why procaspase-8 can not form DED filaments without cFLIP and bind to FADD? There are many studies that show that this is possible, i.e. Hughes et al, 2016, Mol Cell and others.
RESPONSE: Indeed, in the absence of cFLIP, caspase 8 undergoes homodimer and forms DED filaments, resulting in induction of apoptosis (Dillon et al., 2012;Panayotova-Dimitrova et al., 2013;Piao et al., 2012;Yeh et al., 2000). However, as reported by others (Hughes et al., 2016;Yu et al., 2009), in the presence of cFLIPL, caspase 8 preferentially interacts with cFLIPL, but not caspase 8 itself. This might be because the closed L2' loop conformation of caspase 8 causes steric clash when the proform of caspase 8 undergoes homodimer (Yu et al., 2009). Thus, ubiquitylated cFLIPL might preferentially interact with caspase 8, but block higher-ordered oligomer formation and caspase 8 activation. Given that immunoprecipitates with anti-FADD and anticaspase 8 antibodies did not contain caspase 8 and FADD, respectively, in WT HeLa cells treated with TNF/BV6 (0.1 µM) (see Fig. 7a, b, and Supplementary  Fig. 9a, b), FADD and caspase 8 did not appear to form a stable complex in the presence of ubiquitylated cFLIPL. We made a new Fig. 10 to represent our model and mentioned it in the text (line 402-411, 507-527).

Then if we assume that cFLIPL is not recruited to the DED filaments,
there is also an essential role which belongs to c-FLIPS in controlling the activity of the filament, cFLIPS will be still present in the system. The role of both isoforms have to be considered. Finally, cFLIPL plays both antiand pro-apoptotic role in the DED filament, which is not accounted in this model.

RESPONSE:
Thank you for pointing out a critical issue. cFLIPL and cFLIPs are expressed in human cells such as HeLa and HCT116 cells. Overexpression of cFLIPs and viral homologs of cFLIPs (v-FLIPs) blocks death receptor-induced apoptosis in murine and human cells (Thome et al., 1997). However, expression levels of cFLIPs were comparable to those of cFLIPL in HeLa and HCT116 cells ( Supplementary Fig. 8). Moreover, cFLIPs was also recruited to complex IIb along with cFLIPL (Rebuttal Figure 1) (Hillert et al., 2020). It is unclear whether cFLIPs is recruited to complex IIb in the absence of cFLIPL. Although overexpression of cFLIPs blocks apoptosis, but increases susceptibility to necroptosis in vitro and in vivo (Oberst et al., 2011;Shindo et al., 2019), it is unclear whether endogenous cFLIPs in the absence of cFLIPL plays a major role in cell fate decision. Relative contribution of cFLIPL and cFLIPs to TNFinduced apoptosis may be evaluated by generating cells specifically lacking cFLIPL or cFLIPs by CRISPR/Cas9 technology. However, such experiments are beyond the scope of the present study and will be addressed in the future study. We only discussed these points in the rebuttal letter. Regarding the function of cFLIPL, Hughes et al. elegantly showed a model that low concentrations of cFLIPL promote activation of caspase 8 through promoting processing of caspase 8, whereas high concentrations of cFLIPL suppress caspase 8 activation (Hughes et al., 2016). This conclusion is drawn by the experiments using in vitro translated cFLIPL in a cell-free system. In sharp contrast, germline deletion of Cflar results in embryonic lethality due to enhanced apoptosis and necroptosis (Yeh et al., 2000), indicating that cFLIP is a genuine apoptosis inhibitor in vivo.
To integrate these apparently inconsistent observations obtained from in vitro and in vivo experiments, we surmised that post-transcriptional modification of cFLIPL that occurs in cells but not in a cell-free system might critically affect the function of cFLIPL. Q468D mutant of cFLIPL fails to form a proper oligomer with caspase 8, preventing caspase 8 processing (Hughes et al., 2016;Yu et al., 2009). These results indicate that the C-terminal part of cFLIPL is required for cFLIPL-dependent caspase 8 activation. In this respect, MIB2-dependent ubiquitylated lysine residues of cFLIPL, including K460, K462, K473, and K474, are closely located in Q468. Thus, it is plausible that Ubiquitylated cFLIPL and caspase 8 heterodimers fail to form higher-ordered oligomers through steric hindrance, thereby blocking complex IIb maturation and caspase 8 activation. Consistent with this idea, cFLIPL mutants lacking MIB2-dependent ubiquitylation cannot completely restore anti-apoptotic function to the levels found in WT cFLIPL. Thus, these results suggest that in addition to relative amounts of cFLIPL vs. caspase 8 described above, posttranscriptional modification of cFLIPL may critically determine whether cell live or die. Thus, our study has provided a new avenue leading to complete understanding of regulation of anti-apoptotic and proapoptotic functions of cFLIPL. We mentioned these points in the text and made new Figs. 9 and 10 to include these results (line 341-378, 402-411, 507-527).

4.C-FLIP seem to undergo K48 and K63 ubiquitinylation by MIB2
(Highlighted in Figure 4). Moreover, the authors applied a very elegant quantitative mass spectrometry approach in order to quantify the degree of the ubiquitinylation of cFLIPL. However, it has to be said that surprisingly, the absence of MIB2 decreased K48 and K63 ubiquitinylation of cFLIPL only two fold. Hence, my question is how the two times difference in the degree of ubiquitylation would have such dramatic influence on the functional outcome?
RESPONSE: Thank you for pointing out a critical issue. Although total amounts of K48's and K63's ubiquitylation of cFLIPL was only reduced to half, it is plausible that a decrease in ubiquitylation of critical lysine residues to prevent a higher-ordered oligomer formation and caspase 8 activation (Fig. 10e, f). Indeed, reconstitution of Cflar-/-MEFs with cFLIPL-3A or cFLIPL-K9R (both mutants lack MIB2-dependent ubiquitylation) failed to suppress TNF/BV6induced apoptosis comparable to the levels in cFLIPL WT cells (Fig. 9). Moreover, there may be some threshold of ubiquitylated cFLIPL that can suppress complex IIb formation in response to TNF along with concentrations of BV6. Indeed, TNF and low concentrations of BV6 (0.1 µM) induced apoptosis in MIB2 KO cells, but not WT cells, whereas TNF and high concentrations of BV6 (1.0 µM) induced apoptosis in both WT and MIB2 KO cells (Fig. 6c-j). These results indicate that the MIB2-dependent ubiquitylation of cFLIPL may critically determine the threshold for whether cells live or die (line 230-240, 341-378, 507-527).  (Fig. 6, 7, and Supplementary Fig. 9). In sharp contrast, high concentrations of BV6 (1 µM) with TNF induced complex IIb formation and cell death in wild-type cells ( Fig. 6c and Rebuttal Fig. 2). Although low concentrations of BV6 (0.1 µM) was sufficient to completely deplete cIAP1/2 ( Fig. 6f, 7a, b, Supplementary Fig. 6), high concentrations of BV6 (1 µM) were required to induce the complex IIb formation and apoptosis in wild-type cells ( Fig. 6c and Rebuttal Fig. 2). Elucidation of the mechanisms underlying BV6 concentration-dependent induction of complex IIb is a very important issue but is beyond the scope of the present study. We mentioned it in the text (line 228-257).

Along these lines, in HeLa cells cFLIP is
Rebuttal Fig. 2 High, but not low concentrations of BV6 along with TNF induces complex II formation. WT HeLa cells were stimulated with TNF (10 ng/ml) in the absence or presence of the indicated concentrations of BV6. TNFinduced complex IIb was immunoprecipitated with anti-caspase 8 antibody and detected by anti-RIPK1 antibody. Asterisk indicates cross-reacted bands. Results are representative of two independent experiments. RESPONSE: Thank you for your thoughtful comments. Although MIB2 ubiquitylated cFLIPL, MIB2-dependent ubiquitylation did not promote degradation of cFLIPL, but slightly increased the stability of cFLIPL ( Supplementary Fig. 8). As suggested, the most important finding in the present study is that MIB2-dependent ubiquitylation of cFLIPL blocks complex IIb maturation, thereby suppressing caspase 8 activation. cFLIPL mutants lacking MIB2-dependent ubiquitylation, such as cFLIPL-3A and cFLIPL-K9R cannot completely restore anti-apoptotic function to levels found in WT cFLIPL after TNF/BV6 stimulation (Fig. 9d, h). Moreover, despite the lack of RIPK1 activation, TNF/BV6 induced apoptosis in Cflar-/-cells expressing cFLIPL-3A or cFLIPL-K9R, but not cFLIPL WT (Fig. 9i). Consistent with these results, RIPK1 inhibitor did not block TNF/BV6-induced apoptosis in Cflar-/-cFLIPL-3A or cFLIPL-K9R cells (Fig. 9j). Collectively, these results strongly support that ubiquitylation of cFLIPL suppresses RIPK1 kinase activity-independent cell death. We made a new Fig. 9 to include these results and mentioned these points in the text (line 341-378, 507-527).

RESPONSE:
Indeed, as described above, we also found that MIB2 suppressed RIPK1dependent apoptosis in a ubiquitin ligase-activity dependent manner. In contrast to a precious study, TNF-induced RIPK1 ubiquitylation was not impaired in MIB2 KO cells. In addition, our present study strongly supports the idea that MIB2 suppresses RIPK1 kinase activity-independent cell death through ubiquitylation of cFLIPL. The following results strongly support the idea. First, TNF/BV6-induced cell death was enhanced in MIB2 KO cells, but RIPK1 inhibitor did not block or only partially blocked cell death enhancement in MIB2 KO HCT116 and MIB2 KO HeLa cells, respectively. Second, RIPK1 inhibitor did not block TNF/CHX-induced cell death in MIB2 KO cells. Third, reconstituting Cflar-/-MEFs with cFLIPL mutants lacking MIB2-dependent ubiquitylation (cFLIPL-3A and cFLIPL-K9R) failed to restore cell viability to levels comparable for cells reconstituted with WT cFLIPL. Fourth, TNF/BV6 treatment of cells expressing cFLIPL-3A or cFLIPL-K9R, but not cFLIPL WT, resulted in cell death without inducing RIPK1 phosphorylation. Consistent with this result, RIPK1 inhibitor did not block TNF/BV6-induced apoptosis in Cflar-/-cFLIPL-3A or cFLIPL-K9R cells. Together, these findings suggest that ubiquitylation of cFLIPL might mediate MIB2-dependent suppression of TNF-induced apoptosis in a RIPK1 kinase activity-independent manner. We made new Figures 6, 8, 9, 10, and mentioned these points in the text (line 460-478)

Minor comments: 8.Extended figure 3: in the quantification of c-FLIP levels the loading control is overexposed, hence, it is difficult to draw the conclusion. C-FLIPs is always quicker degraded, which was also observed in both cell lines ( KO MIb2 versus wt) so one can not draw a conclusion that this is yet another evidence of MIB2 action.
RESPONSE: As suggested, we performed additional experiments to investigate degradation of cFLIPL after stimulation with different stimuli in more detail.
Signal intensities of cFLIPL, cFLIPs, and tubulin were determined by Image J, and ratios of cFLIPL/tubulin and cFLIPs/tubulin were calculated. Relative intensities of each ratio at the indicated times compared to the ratio at 0 (100 %) was calculated. Degradation of cFLIPL in MIB2 KO cells was slightly accelerated but not delayed compared to WT cells after CHX or TNF/CHX stimulation ( Supplementary Fig. 8). Rapid degradation of cFLIPL and cFLIPs in MIB2 KO cells after TNF/BV6 stimulation might in part be the result of a secondary effect of apoptosis of MIB2 KO cells. We mentioned these in the text and included the results to make a new Supplementary Fig. 8 (line 273-295).
9.In TNF signaling complex I is not termed ‚death inducing signaling complex' but just complex I. With complex IIa and IIb, there are certainly some different views on this nomenclature in the literature, as some people name complex IIb as necrosome or RIPK1/RIPK3 complex and complex IIa as ripoptosome, but in a particular review that was cited by authors indeed complex IIb is assigned to FADD/caspase-8/c-FLIP complex and complex IIa to TRADD/caspase-8/FADD complex. I personally prefer the first nomenclature. RESPONSE: Indeed, there is some discrepancy of nomenclatures of TNFR1 complex IIa, IIb, necrosome, and ripoptosome. To avoid confusion, we mentioned that "the death-inducing signaling complex, which is referred to as complex IIb or ripoptosome" in the text (line 70).
Nakabayashi report c-FLIP-long as a novel binding partner and substrate of the E3 ligase MIB-2. MIB-2-dependent ubiquitylation of c-FLIP-long is shown to suppress TNF-induced apoptosis. The conclusions are largely justified by the data, although the notion that ubiquitylation is affecting c-FLIP stability is less certain. For example, unmodified c-FLIP may just be more prone to aggregation than degradation in extended Fig. 3. Do they get the same result if they examine the insoluble fraction for c-FLIP?

RESPONSE:
We appreciate positive comments on our manuscript and pointing out a critical issue. As suggested, we prepared insoluble fractions by adding sample buffer to the pellets of cell lysates after lysis with RIPA buffer followed by sonication. Then, solubilized insoluble fractions were subjected to SDS-PAGE. As shown in Rebuttal Fig. 3, we could not detect cFLIPL in insoluble fractions. Thus, it is unlikely that unmodified cFLIPL forms aggregates, thereby disappearing from cell lysates after treatment. We only mentioned it in the rebuttal letter. with TNF/CHX for the indicated times. Soluble and insoluble fractions were prepared and analyzed by immunoblotting with the indicated antibodies. P indicates positive control samples of cell lysates of untreated WT HeLa and HCT116 cells.

Rebuttal
Regarding the stability of cFLIPL, we assume that slightly increased stability of cFLIPL does not play a major role in preventing TNF-induced cell death ( Supplementary Fig. 8). But rather, ubiquitylation of cFLIPL in the Cterminal caspase-like domain suppresses caspase 8 activation through binding and preventing higher-ordered oligomer formation and caspase 8 activation. We mentioned it in the text and present a model in Fig. 10e, f ( line 402-411, 507-527).

My other main criticisms are: 1) It is not clear anywhere in the text, methods, or legends if experiments exclusively use human rather than mouse proteins.
RESPONSE: Sorry for the lack of description. We used all human cDNAs except for ITCH (from mouse cDNA) for constructing expression vectors in the experiments. We mentioned it in the Methods (line 569-570).

2) Throughout the figures, the efficiency of each IP is not confirmed by WB for the protein that is immunoprecipitated. These quality control WBs should be included in every experiment.
RESPONSE: Sorry for not including critical control experiments. As suggested, we included the results showing the efficiency of immunoprecipitated proteins with the indicated antibodies for IP experiments.

3) The authors should indicate exactly which cysteines are mutated in R1M and R2M MIB-2.
RESPONSE: Sorry for the lack of a detailed description of MIB2 mutants. We mutated catalytic cysteine(s) at 900 and 977 to serine(s) in the RING finger domains of MIB2. We mentioned these in the text, Methods, and Figure legends (line 147-148, 589-592, 942-943) . Fig. 3C should include WBs using linkage-specific Ub antibodies (K48, K63, and K11 antibodies). The linkages generated in this more defined in vitro setting would be useful to compare to those generated in cells.

4)
RESPONSE: As suggested, we performed Western blotting with anti-K48-or anti-K63-linked polyubiquitin chain-specific antibodies using products from in vitro ubiquitylation assays. Indeed, anti-K11-linked polyubiquitin chain-specific antibody had been commercially available from Merck (MABS107) until recently. However, the vendor has stopped supplying the antibody. Thus, we only included the results with anti-K48-or anti-K63-linked polyubiquitin chainspecific antibodies in Fig. 4g and mentioned it in the text (line 183-189). Fig. 5f should be less cropped so that the reader can see the loss of ubiquitylation on the 3A mutant (as compared to the WT control). i.e. are the slower migrating forms absent? RESPONSE: Thank you for picking up this. Since we did not co-transfect HA-Ub along with WT or cFLIPL-3A and MIB2 in old Fig. 5f, we did not detect a slower migrating form of cFLIPL in the uncropped blot (Rebuttal Fig. 4). To show a defect in MIB2-dependent ubiquitylation of cFLIPL-3A, we transfected WT and cFLIPL-3A along with MIB2 and HA-Ub. As shown in Fig. 9b, MIB2-dependent ubiquitylation of cFLIPL-3A was greatly reduced in HEK293 cells compared to the levels in WT cFLIPL. We mentioned it in the text (line 351-353). Previous study has shown MIB2 could ubiquitinate RIPK1 to inhibit RIPK1 kinase activity, thereby suppress TNFa-induced apoptosis (Feltham et al.,  2018). However, the authors found cells expressing MIB2-binding defective mutant of cFLIPL are also hypersensitive to TNFa-induced apoptosis. Therefore, they draw the conclusion that MIB2 inhibits TNFinduced cell death through both ubiquitination of cFLIPL and inhibition of RIPK1 kinase activity. Overall, this study identified cFLIPL as a new substrate of E3 ligase MIB2 based on in vitro biochemistry data, and extended the current understanding of the function of MIB2 in the regulation of TNFa-induced apoptosis. However, more experiments need to be performed to determine the molecular mechanism of how MIB2mediated ubiquitination of cFLIPL suppresses TNFa-induced apoptosis independent of RIPK1 kinase activity inhibition. Thank you for your positive comments on our study. As suggested, we performed substantial amounts of experiments to address the comments raised by the reviewers. Now the results presented here strongly support that MIB2dependent ubiquitylation of cFLIPL suppresses RIPK1 kinase activityindependent cell death.

5) The c-FLIP input WB in
Specific comments: 1. Line 65, TAK1 inhibitors are thought to inhibit the phosphorylation but not ubiquitination of RIPK1.

RESPONSE:
We appreciate pointing out our misunderstanding. As suggested, we changed the sentence to "IAP inhibitors and TAK1 inhibitors that block ubiquitylation and phosphorylation of RIPK1, respectively" (line 67-68). Fig. 1c, d, total HA-Ub and IP-Flag level were missing.

In
RESPONSE: As suggested, we included the results in Fig. 1c and 1d. 3. Extended Fig. 2c, the data is not sufficient to demonstrate that MIB2 ubiquitinates multiple lysine residues of cFLIPL. Since MIB2 could generate both K63-and K48-linked ubiquitin chains on cFLIPL, it may not be easy to detect the change of total ubiquitination level. The authors should also detect the K63-and K48-linked ubiquitination level in various mutants.
RESPONSE: Thank you for your valuable suggestion. As suggested, we performed Western blotting with anti-K48-or anti-K63-linked polyubiquitin chain-specific antibodies using various mutants. Combined mutations of the indicated lysine residues in the C-terminal caspase-like domain did not change the signals reacted with anti-K48-or anti-K63-linked polyubiquitin chainspecific antibodies, suggesting that both K48-linked and K63-linked polyubiquitin chains were anchored to each lysine ( Supplementary Fig. 3c). Moreover, we also performed Western blotting with anti-K48-or anti-K63linked polyubiquitin chain-specific antibodies using products of in vitro ubiquitylation assays (Fig. 4g). We mentioned these results in the text (line 183-189, 202-211). Fig. 3a, b, the MIB2 on the stability of cFLIPL seems to have no much significance. This experiment is important and needs to be improved. Since the authors show MIB2 could generate K63-and K48linked ubiquitination on cFLIPL, it is critical to make sure the role of MIB2 on the stability of cFLIPL, since MIB2 is involved in TNFa-induced apoptosis through regulating RIPK1. Therefore, the author should examine the stability of cFLIPL in WT and MIB2 KO cells under different apoptosis inducers (TNF alone; CHX alone; TNF+CHX; TNF+BV-6). In addition, cFLIPL is degraded very faster under apoptosis condition. The dose and time point of stimulators should be modified.

Extended
RESPONSE: As suggested, we performed additional experiments to investigate kinetics of degradation of cFLIPL and cFLIPs after stimulation with different stimuli in more detail. Signal intensities of cFLIPL, cFLIPs, and tubulin were determined by Image J, and cFLIPL/tubulin and cFLIPs/tubulin ratios were calculated. Relative percentages of each ratio at the indicated times compared to the ratio at 0 (100%) were calculated. Degradation of cFLIPL in MIB2 KO cells was slightly accelerated but not delayed compared to those in WT cells after CHX or TNF/CHX stimulation ( Supplementary Fig. 8). This rapid degradation might in part be the result of a secondary effect of apoptosis of MIB2 KO HeLa cells. We mentioned these results in the text and made a new Supplementary Fig. 8 (line 273-295). Fig. 5a-5e, deficiency of MIB2 causes TNFa-induced apoptosis due to enhanced RIPK1 kinase activity has been reported earlier (Feltham et al., 2018). It is critical to discriminate the relationship between MIB2-mediated ubiquitination of cFLIPL and MIB2-mediated suppression of RIPK1 kinase activity. Therefore, the authors should use RIPK1 kinase inhibitor or genetically knockout RIPK1 in MIB2-KO cells to determine the contribution of ubiquitination of cFLIPL to the apoptosis in MIB2-KO cells. In addition, the authors need to examine the RIPK1 kinase activity in cFLIPL-3A-reconstituted MEFs to determine whether MIB2-mediated ubiquitination of cFLIPL could suppress RIPK1 kinase activation.

5.
RESPONSE: Thank you for your thoughtful suggestion. We stimulated WT and MIB2 KO cells with TNF/BV6 (0.1 µM) in the absence or presence of Nec-1s. To our surprise, Nec-1s only partially blocked TNF/BV6-induced apoptosis in MIB2 KO HeLa cells and did not block in MIB2 KO HCT116 cells (Fig. 6g, h). Moreover, TNF/CHX-induced cell death was enhanced in MIB2 KO HeLa cells, especially at low concentrations of TNF (Fig. 6i, j), whereas these cell death enhancements were not suppressed in the presence of Nec-1s. Consistently, we found that TNF/CHX did not induce phosphorylation of RIPK1 in WT and MIB2 KO HeLa and HCT116 cells (Fig. 6k). These results suggest that MIB2 blocked, at least in part, RIPK1 kinase activity-independent cell death (line 252-257).
To further substantiate a role for ubiquitylation of cFLIPL in suppression of TNF-induced cell death, we also generated a cFLIPL mutant where MIB2dependent ubiquitylated lysine residues were mutated to arginines (cFLIPL-K9R). We then stimulated Cflar-/-MEFs reconstituted with WT cFLIPL, cFLIPL-3A, or cFLIPL-K9R with TNF and low concentrations of BV6 (0.1 µM). TNF/BV6 did induce cell death in Cflar-/-cFLIPL-3A and cFLIPL-K9R cells, but not WT cFLIPL cells (Fig. 9d, h). Notably, TNF/BV6 did not induce RIPK1 phosphorylation in either transfectant (Fig. 9i), and RIPK1 inhibitor did not block TNF/BV6-induced cell death in Cflar-/-cFLIPL-3A and cFLIPL-K9R cells (Fig.  9j). Together, MIB2 blocks RIPK1 kinase activity-independent cell death through ubiquitylation of cFLIPL. We mentioned these points in the text (line 341-378), and made new Figures 6 and 9 to include these results. Fig. 6c, 6f,  RESPONSE: Thank you for the thoughtful suggestion. Regarding the contribution of MIB2 to ubiquitylation of RIPK1, our date did not support the previous study (Feltham et al., 2018). First, ubiquitylation of RIPK1 in complex 1 in MIB2 KO cells was comparable to the levels in WT cells after TNF stimulation ( Fig. 5a and Supplementary Fig. 5b). Second, in the presence of IAP inhibitor that depletes cIAP1/2, ubiquitylation of RIPK1 in complex 1 was greatly reduced in both WT and MIB2 KO cells ( Fig. 6f and Supplementary Fig. 6), indicating that cIAP1/2 play a major role in ubiquitylation of RIPK1. Third, MIB2 constitutively bound RIPK1, and RIPK1 in the MIB2-containing complex was not ubiquitylated before TNF stimulation but underwent rapid ubiquitylation in MIB2 KO HA-MIB2 WT HeLa cells 10 min after TNF stimulation (Fig. 10b). Indeed, we cannot formally exclude the possibility that contribution of MIB2 to RIPK1 ubiquitylation is cell-type specific. We made new Figures to include these results and mentioned it in the text (line 213-226, 241-251, 390-401).

6.
To test whether HA-MIB2 MT interacts with RIPK1, we immunoprecipitated MIB2 with anti-HA antibody from MIB2 KO cells reconstituted with HA-MIB2 WT or HA-MIB2 MT. Then, immunoprecipitates were analyzed by anti-RIPK1 antibody. As shown in Fig. 10c, MIB2 MT as well as MIB2 WT constitutively bound RIPK1. Moreover, ubiquitylation of RIPK1 in the RIPK1/MIB2 complex in MIB2 KO HeLa/HA-MIB2 MT cells was comparable to the levels in MIB2 KO HeLa/HA-MIB2 WT cells, indicating that ubiquitin ligase activity is dispensable for RIPK1 ubiquitylation (Fig. 10c) Fig. 6h, the cFLIPL level in the immunoprecipitates was missing; and an IgG control should be added to exclude the non-specific binding.
RESPONSE: Thank you for pointing out a critical issue. As suggested, we included amounts of cFLIPL in the immunoprecipitates throughout the experiments. We detected more than three bands of cFLIPL in the immunoprecipitates. To verify that these bands are indeed derived from cFLIPL products, we tested whether knockdown of cFLIPL by siRNA abolished these bands in the immunoprecipitates. As shown in Supplementary Fig. 1, all bands disappeared in cFLIP knockdown HeLa cells, indicating that these bands were indeed cFLIPL. Since the molecular size of cFLIPL was identical to the second band, the upper band may be modified cFLIPL. The lower multiple bands indicated by asterisks may be degradation of products of cFLIPL. The detailed characterization of the upper band of cFLIPL will be investigated in the future study. We made a new Supplementary Fig. 1 and mentioned it in the text (line 110-117).
As suggested, we repeated similar experiments to include the results immunoprecipitated with control IgGs. We made a new Supplementary Fig. 10a to include the results. Fig. 5a, it is weird that IP-C, cFLIPL blotting has IgGs, but in IP-H, cFLIPL blotting nearly has no IgGs.

Extended
Response: Sorry for confusing you. We used mouse IgGs (for control IP) and anti-HA (3F10 from rat) antibody for immunoprecipitation. Since the immunoprecipitates were blotted with anti-cFLIPL antibody (7F10 from mouse) followed by HRP-conjugated anti-mouse IgGs antibody, heavy chains of mouse IgGs (lane C), but not rat IgG (lane H) exhibited strong signals. To avoid confusion, we repeated similar experiments using control rat IgGs and anti-HA antibody for immunoprecipitation. Now, the results clearly showed that anti-HA antibody, but not control IgGs, immunoprecipitated ubiquitylated cFLIPL before TNF stimulation, but not after TNF stimulation. We replaced old results with new ones and included them in Fig. 10a. 9. Line 367, a spelling error "complex" Response: As suggested, we corrected the typo.
10. Line 342-343, although TNF plus CHX is initially thought to induce RIPK1-independent cell death. However, under some conditions such as TBK1 deficiency, it could also induce RIPK1 kinase-dependent cell death (Xu et al., 2018). Since MIB2 could suppress RIPK1 kinase activation, TNF plus CHX may also induce RIPK1 kinase-dependent cell death in MIB2 KO cells. The authors should examine RIPK1 autophosphorylation and use RIPK1 kinase inhibitor to do the TNF plus CHX stimulation in MIB2 KO

cells.
RESPONSE: Thank you for your thoughtful comments. We stimulated WT and MIB2 KO HeLa cells with TNF/CHX in the absence or presence of Nec-1s. TNF/CHX-induced apoptosis was dramatically enhanced in MIB2 KO HeLa and MIB2 KO HCT116 cells compared to respective control wild-type cells when cells were stimulated with low concentrations of TNF (1~ 3 ng/ml). Nec-1s did not block TNF/CHX-induced apoptosis in MIB2 KO cells at any concentrations of TNF (Fig. 6i, j). Moreover, we found that TNF/CHX stimulation did not induce RIPK1 phosphorylation in either WT or MIB2 KO cells (Fig. 6k). Thus, we conclude that MIB2 suppresses RIPK1 kinase activity-independent cell death. We made a new Figure 6 to include these results and mentioned them in the text and (line 258-268).
11. Line 378-382, K63-linked ubiquitination of RIPK1 has been reported to suppress RIPK1 kinase activity by recruiting TAK1 kinase (Tang et al.,  2019). The authors should examine the TAK1, IKK, TBK1, and MK2 recruitment in TNFR1 complex1 in MIB2-KO cells to determine the exact molecular mechanism by which MIB2-mediated ubiquitination suppresses RIPK1 kinase activity.
RESPONSE: As suggested, we tested whether these kinases were recruited to the TNFR1 complex I. Consistent with a previous study (Lafont et al., 2018), TAK1, IKK, and TBK1 were recruited to the TNFR1 complex I in WT cells after TNF stimulation (Fig. 5a, Supplementary Fig. 5b). Notably, recruitment of these kinases to complex I in MIB2 KO cells after TNF stimulation was comparable to the levels in WT cells. However, MK2 was not recruited to the TNFR1 complex I in either WT or MIB2 KO cells. In contrast to the results in a previous study (Feltham et al., 2018), TNF-induced ubiquitylation of RIPK1 in MIB2 KO cells was comparable to the levels in WT cells (Fig. 5a, Supplementary Fig. 5b).
MIB2 MT as well as MIB2 WT constitutively bound RIPK1. Ubiquitylation of RIPK1 in the RIPK1/MIB2 complex in MIB2 KO HeLa/HA-MIB2 MT cells was comparable to the levels in MIB2 KO HeLa/HA-MIB2 WT cells, indicating that ubiquitin ligase activity of MIB2 is dispensable for RIPK1 ubiquitylation (Fig. 10c). In contrast, TNF/BV6 induced RIPK1 phosphorylation in MIB2 KO HeLa cells, MIB2 KO HeLa/HA-MIB2 MT, but not HA-MIB2 WT cells (Fig. 10d). Thus, ubiquitin ligase activity of MIB2 is dispensable for RIPK1 ubiquitylation, but indispensable for suppression of RIPK1 kinase activity. We made new Figures to include these results and mentioned them and discussed the mechanisms in the text (line 213-226, 390-401, 490-506) .