PELI1 functions as a dual modulator of necroptosis and apoptosis by regulating ubiquitination of RIPK1 and mRNA levels of c-FLIP

PELI1 Deficiency Protects Against Necroptosis.

Because RIPK1 is subject to extensive ubiquitination modifications in TNF-RSC (complex I) and complex IIb in response to TNFα (14), we considered the possibility that RIPK1 might be modulated by additional E3 ubiquitin ligases beyond cIAP1/2, the only E3 ubiquitin ligase characterized as involved in regulating the activation of RIPK1. We conducted a _targeted screen of E3 ubiquitin ligases reported to interact with RIPK1 (Fig. S1A). In this screen, we identified PELI1 as an E3 ubiquitin ligase whose knockdown protected cells against necroptosis.

To confirm the role of PELI1 in mediating necroptosis, we generated WT and Peli1^−/− MEFs (mouse embryonic fibroblasts cells) derived from WT and Peli1^−/− mice (13) and tested their sensitivity to necroptosis. We found that Peli1^−/− MEFs were highly resistant to necroptosis induced by TNFα/SM164/zVAD, TNFα/5z7/zVAD, or TNFα/CHX/zVAD (Fig. 1 A and B and Fig. S1B). The resistance of Peli1^−/− MEFs to necroptosis induced by TNFα/SM164/zVAD and TNFα/5z7/zVAD was complete and could not be further enhanced by the treatment of Nec-1s (Fig. 1 C–E). We also tested the role of PELI1 in human HT29 cells and mouse RGC5 cells, using lentiviral vector-mediated shRNA _targeting Peli1. Knockdown of Peli1 in HT29 protected against TNFα/SM164/zVAD-induced or TNFα/CHX/zVAD-induced necroptosis (Fig. 1 F and G). Furthermore, knockdown of Peli1 in RGC5 cells protected against necroptosis induced by TNFα/SM164/zVAD (Fig. S1C). In addition, we checked the effect of knocking down Peli1 in murine fibrosarcoma L929 cells and found that knockdown of Peli1 protected L929 cells from necroptosis induced by zVAD (Fig. S1D).

Taken together, these data suggest that PELI1 deficiency protects against necroptosis.

PELI1 Is Indispensable for the Formation of RIPK1-RIPK3 Necrosome.

In WT MEFs, RIPK1 is activated after treatment of TNFα/SM164/zVAD to lead to the formation of complex IIb, also called necrosome, which includes FADD, caspase-8, and RIPK3, which mediates the activation of RIPK3 and the subsequent phosphorylation and activation of MLKL (14–18). The activated MLKL then oligomerizes to execute cell death by disrupting the integrity of plasma membranes (17, 19–21).

In Peli1^−/− MEFs stimulated by TNFα/SM164/zVAD to induce necroptosis, the phosphorylation of S166 RIPK1, a marker for its activation (6, 8, 22), occurred on schedule as that of WT, suggesting PELI1 deficiency does not affect the activation of RIPK1 in MEFs (Fig. 2 A and B). In contrast, the phosphorylation of S232 RIPK3 and S345 MLKL, the markers for the activation of RIPK3 and MLKL, respectively, were blocked in Peli1^−/− MEFs. Furthermore, the oligomerization of MLKL, detectable by Western blotting using nonreducing conditions, was also blocked in Peli1^−/− MEFs (Fig. 2C). Thus, although PELI1 is important for the activation of RIPK3 and MLKL to promote necroptosis, it is not directly involved in mediating the activation of RIPK1.

We next analyzed the formation of complex IIb by immunoprecipitation, using an anti-FADD. We found that the binding of RIPK1, including p-S166 RIPK1, with FADD in cells stimulated by TNFα/SM164/zVAD was not affected by PELI1 deficiency (Fig. 2B). In contrast, the binding of RIPK3 with RIPK1 and FADD was blocked by PELI1 deficiency (Fig. 2 A and C).

Because the complex IIb/necrosome is known to be present in an insoluble amyloid-like conformation (23), we next analyzed the translocation of RIPK1 and RIPK3 into the insoluble fraction on induction of necroptosis. We found that although the translocation of modified RIPK1 was largely normal at early points (15 min⁻¹⋅h) after induction of necroptosis, the presence of highly modified RIPK1 in insoluble fraction at 2 h, the time at which complex IIb formed, was reduced in Peli1^−/− MEFs (Fig. 2C). Furthermore, the translocation of RIPK3 into the insoluble fraction was blocked by PELI1 deficiency. These data suggest PELI1 is involved in promoting the binding of RIPK1 to RIPK3 to mediate the formation of complex IIb.

We next analyzed the mechanism that might explain the inability of RIPK1 to bind to RIPK3 in TNFα/SM164/zVAD-treated Peli1^−/− MEFs. Because PELI1 is an E3 ubiquitin ligase, we compared the ubiquitination patterns of RIPK1 in complex IIb in WT and Peli1^−/− MEFs. Using anti-K63 and M1-ubiquitin chain-specific antibodies (24), we found that K63 ubiquitination of RIPK1 in complex IIb was inhibited, whereas M1 ubiquitination of RIPK1 increased in Peli1^−/− MEFs after 2 h of treatment of TNFα/SM164/zVAD (Fig. 2D). In contrast, we found that PELI1 deficiency resulted in no difference in K63 or M1 ubiquitination of RIPK1 in the TNF-RSC (complex I) (Fig. 2E). These results suggest that PELI1 is involved in regulating RIPK1 ubiquitination in complex IIb.

PELI1 Is Recruited into TNF-RSC in a RIPK1-Dependent Manner.

We next determined whether PELI1 might be recruited into TNF-RSC. We found that in cells treated with FLAG-TNFα for 30 min⁻¹⋅h, but not at 5 min, immunoprecipitation by anti-FLAG revealed the presence of PELI1 (Fig. 3A). This recruitment of PELI1 into complex I was independent of RIPK1 kinase activity, as the pretreatment with RIPK1 kinase inhibitor Nec-1s had no effect on the recruitment of PELI1 into complex I (Fig. 3B). However, the recruitment of PELI1 into complex I was blocked in Ripk1^−/− MEFs (Fig. 3C). Thus, the recruitment of PELI1 into complex I requires RIPK1, but not its kinase activity.

The recruitment of PELI1 into complex I was also blocked in MEFs deficient for TRADD, a key adaptor protein that directly binds to the intracellular death domain of TNFR1 (25) (Fig. 3D). Because TRADD is involved in mediating the recruitment of cIAP1/2, we next pretreated cells with cIAP1/2 inhibitor SM164, which led to the degradation of cIAP1/2, and found that cIAP1/2 deficiency also inhibited the recruitment of PELI1 into complex I (Fig. 3E). The LUBAC complex, consisting of HOIP/HOIL/SHARPIN, is involved in mediating M1 ubiquitination of TNF-RSC, including RIPK1 (26, 27). In Hoip^−/− cells stimulated by TNFα, no PELI1 was detected in complex I, suggesting that the recruitment of PELI1 into complex I requires M1 ubiquitination (Fig. 3F). In contrast, the recruitment of PELI1 into complex I in Abin-1^−/− MEFs was increased (Fig. 3G). Because K63 ubiquitination of RIPK1 in complex I under ABIN-1-deficient conditions was increased (28), this result suggests that the recruitment of PELI1 into complex I is also regulated by K63 ubiquitination of RIPK1. However, the recruitment of PELI1 into complex I was not altered in Cyld^−/− or Nemo^−/− MEFs (Fig. S2).

Taken together, these results suggest that both K63 and M1 ubiquitination of RIPK1 are involved in regulating the recruitment of PELI1 into complex I.

PELI1 Mediates K63 Ubiquitination of RIPK1 in a Kinase Activity-Dependent Manner.

RIPK1 contains a kinase domain at N terminus, a death domain (DD) at its C terminus, and an intermediate domain. To investigate the domain of RIPK1 involved in mediating its binding with PELI1, we overexpressed full-length and truncation mutants of RIPK1, HA-RIPK1-ΔKD, HA-RIPK1-ΔC, and HA-RIPK1-ΔDD with that of FLAG-tagged PELI1 in 293T cells, and characterized their interactions by immunoprecipitation. We found that HA-RIPK1, HA-RIPK1-ΔKD, and HA-RIPK1-ΔDD, but not HA-RIPK1-ΔC, could bind to PELI1 (Fig. 4A). Thus, PELI1 interacts with the intermediate domain of RIPK1, and the interaction of RIPK1 and PELI1 was not inhibited by Nec-1s (Fig. 4B).

Because PELI1 was reported to be able to mediate K63 ubiquitination of RIPK1 to regulate the activation of NF-κB in response to TLR stimulation (13), and we found that K63 ubiquitination of RIPK1 in complex IIb in response to the treatment of TNFα/SM164/zVAD was strongly defective in PELI1-deficient cells (Fig. 2D), we next examined whether PELI1 was involved in mediating K63 ubiquitination of RIPK1 during necroptosis. Furthermore, as the K115 residue of RIPK1 was reported to be a dominant ubiquitination site of RIPK1 in necroptosis (29), we considered the possibility that PELI1 might ubiquitinate RIPK1 on K115.

We confirmed that overexpression of PELI1 can increase both WT and K63 ubiquitination of RIPK1 (Fig. S3A). Furthermore, the K63 ubiquitination of RIPK1 promoted by PELI1 was significantly reduced in the K115R mutant compared with that of WT, with or without the treatment of SM-164, which promotes the degradation of cIAP1/2 (30) (Fig. 4C). Interestingly, although the treatment of Nec-1s did not disrupt the binding of PELI1 and RIPK1 (Fig. 4B), it inhibited the K63 ubiquitination of RIPK1 mediated by PELI1 (Fig. 4C). However, Nec-1s did not reduce the PELI1-mediated K63 ubiquitination of S161E RIPK1, a mutation that keeps RIPK1 in constitutively active conformation and resistant to the inhibition by Nec-1s (6) (Fig. 4D). In contrast, the expression of PELI1 did not promote the ubiquitination of caspase-8, the other component of complex IIb that is also known to be subject to ubiquitination (31) (Fig. S3B). These results suggest that PELI1 is involved in mediating the K63 ubiquitination of activated RIPK1 on K115 in necroptotic cells induced by TNFα. Furthermore, as K63 ubiquitination of WT RIPK1, but not the S161E mutant of RIPK1 by PELI1, is blocked by Nec-1s, these results suggest that inhibition of PELI1-mediated K63 ubiquitination of RIPK1 by Nec-1s may, at least in part, explain the resistance of Nec-1s-treated cells to necroptosis, as PELI1 deficiency offers strong resistance to necroptosis.

PELI1 Inhibits RIPK1-Dependent and RIPK1-Independent Apoptosis by Regulating c-FLIP.

To investigate the role of PELI1 in apoptosis, we next investigated the sensitivity of PELI1 knockout cells to the induction of apoptosis. Although Peli1^−/− MEFs were resistant to necroptosis, knockout of Peli1 in both MEFs and BV2 cells sensitized to apoptosis induced by TNFα/SM164 or TNFα/5z7 (Fig. 5 A–C). Apoptosis of Peli1^−/− MEFs and Peli1^−/− BV2 cells induced by TNFα/SM164 or TNFα/5z7 was fully blocked by Nec-1s (Fig. 5 A–C). In addition, Peli1^−/− MEFs were also sensitized to apoptosis induced by TNFα/CHX, which was not inhibitable by Nec-1s (Fig. 5 D and E). Thus, PELI1 deficiency sensitizes to apoptosis mediated by both RIPK1-dependent and RIPK1-independent pathways.

We characterized the expression of key genes involved in regulating TNFα-mediated apoptosis in WT and Peli1^−/− cells. The cleavage of caspase-8, caspase-3, and PARP-1, the important hallmarks of apoptosis, induced by TNFα/5z7 and TNFα/CHX were elevated in Peli1^−/− MEFs compared with that of WT (Fig. 6 A and B), consistent with increased sensitivity of Peli1^−/− MEFs to both RIPK1-dependent and RIPK1-independent apoptosis (Fig. 5). In contrast, we found that the protein levels of caspase-8, caspase-3, FADD, ITCH, c-Jun, and PARP-1 were not different in Peli1^−/− MEFs compared that of WT. Interestingly, the levels of c-FLIP, a key regulator of TNFα-mediated caspase-8 activation, were reduced in Peli1^−/− MEFs (Fig. 6A). Consistent with the important role of c-FLIP in regulating both RIPK1-dependent and RIPK1-independent apoptosis, knockdown of c-FLIP sensitized cells to apoptosis induced by TNFα/CHX and TNFα/5z7 (Fig. S4).

We next explored the mechanism by which PELI1 might regulate the levels of c-FLIP. Because the expression of PELI1 has been found to be positively coregulated with that of c-Myc in diffuse large B-cell lymphoma (32) and c-Myc binds to and represses the c-FLIP promoter (33), we considered the possibility that PELI1 might regulate the levels of c-Myc. We found indeed that the levels of c-Myc were increased in Peli1^−/− MEFs (Fig. 6A). To examine whether the regulation of c-FLIP and c-Myc by PELI1 might occur at the transcriptional levels, we measured and compared the mRNA levels of c-FLIP and c-Myc by qPCR. We found that, consistent with their protein levels, the levels of c-Myc mRNA were increased, whereas that of c-FLIP decreased, in Peli1^−/− MEFs (Fig. 6C). These results suggest the possible contribution of elevated c-Myc to reduce the expression of c-FLIP in Peli1^−/− MEFs.

Peli1^−/− Mice Are Sensitized to TNFα-Induced SIRS.

Because both necroptosis and apoptosis play important roles in TNFα injection-induced systemic inflammatory response syndrome (SIRS) in mice, we next investigated the response of Peli1^−/− mice to SIRS. We found that Peli1^−/− mice were sensitized to the lethality induced by i.v. injection of mTNFα. Peli1^−/− mice died more and earlier than that of WT at 360 μg/kg of TNFα (Fig. 7A). Consistently, the levels of c-Myc were higher and those of c-FLIP were lower in the spleens of Peli1^−/− mice (Fig. 7B). Furthermore, we detected increased cleavage of caspase-3 in the spleens of Peli1^−/− mice injected with TNFα compared with that of WT. Thus, increased activation of apoptosis is sufficient to promote the lethality of PELI1-deficient mice in response to TNFα stimulation.

Antibodies and Reagents.

The following commercial antibodies and reagents were used in this study: Pellino-1, Abcam (ab199336); TAK1, Cell Signaling Technology (5206); RIPK1, Cell Signaling Technology (3493) and BD Biosciences (610459); RIPK3, Biorad/Serotec (AHP1797); c-Myc, Santa Cruz (sc-764); c-FLIP, EnZo (ADI-AAP-440); NEMO, CST (2685); TBK1, Cell Signaling Technology (3504); FADD, Abcam (ab124812) and Santa Cruz (6036); α-Tubulin, Sigma-Aldrich (T9026); β-actin, Santa Cruz (81178); FLAG, CST (2368); 5Z-7 was from Sigma-Aldrich (O9890).

Statistics.

Data are expressed as the mean ± SEM. Pairwise comparisons between two groups were performed using the Student’s t test. Differences were considered statistically significant if *P < 0.05, **P < 0.01, ***P < 0.001 or not significant (n.s.). At least three independent repeats were included in each data point. Each experiment was repeated at least three times.

Cell Culture.

HEK293T, MEFs, RGC5, and L929 cells were cultured in DMEM (Gibco) with 10% (vol/vol) FBS (Gibco) and 100 U/mL penicillin/streptomycin and 1 mM sodium pyruvate (Invitrogen). HT29 cells were culture in McCoy’s 5A (Gibco) with 10% (vol/vol) FBS (Gibco) and 100 U/mL penicillin/streptomycin. All cells were cultured in 37 °C with 5% CO₂.

Reagents.

CellTiter-Glo luminescent cell viability kit was from Promega. R-7-Cl-O-Nec-1 (Nec-1s) was made by custom synthesis. Cell lines used in this study were tested every 3 mo for mycoplasma contamination by the MycoAlert Mycoplasma Detection Kit from Lonza.

shRNA and siRNA Sequences.

The shRNA sense sequence used for knockdown of human Peli1 is 5′-CCATGTACATGGCTATCATAA-3′, mouse Peli1 are 5′-GCACCACAAGTATAGACAGTT-3′ and 5′-GCACTGTGCATATTGCTTGTA 3′, the siRNA sense sequence used for knockdown of c-FLIP is 5′-GAUGUAUCCUCCUUAGUUU-3′.

Plasmids.

CDS domain of human Peli1(NM_020651) was cloned with a FLAG or HA tag on the N-terminal into mammalian expression vector pRRL; sequence was confirmed by DNA sequencing. Primers containing shRNA _targeting implicated genes were cloned into PLKO.1 vector individually.

Lentivirus Production and Infection.

PLKO.1 vector plasmids carrying shRNA _targeting human Peli1 or scrambled sequence were transfected individually into 293T cells with VSVG, REV, and MDL plasmids for 8 h and then cell media was changed to remove transfection reagents and was replaced with fresh media. Harvest supernatant media from transfected 293T cells 48 h after transfection and filter with 45-μM membrane. Filtered media containing lentivirus particles was used to infect _target cells.

Real-Time Quantitative PCR.

RNA was purified from cells with RNeasy kit from Qiagen, and then cDNA was made from the RNA with TaKaRa EcoDry Premix. Primers used for detection of c-FLIP are 5′-GCT CCA GAA TGG GCG AAG TAA-3′, 5′-ACG GAT GTG CGG AGG TAA AAA-3′, primers used for detection of mRNA of c-Myc are 5′-CCCTATTTCATCTGCGACGAG-3′, 5′-GAGAAGGACGTAGCGACCG-3′.