Necroptosis-blocking compound NBC1 targets heat shock protein 70 to inhibit MLKL polymerization and necroptosis

Significance

Necroptosis is a regulated form of necrotic cell death implicated in many human diseases, including infection, inflammation, neurodegeneration, and cancer. TNF-induced necroptosis results in the formation of MLKL tetramers, which further polymerize to form disulfide bond-dependent amyloid-like fibers to promote cell death. Here we report the identification of a necroptosis-blocking compound, NBC1, which covalently conjugates two cysteines of chaperone Hsp70. Importantly, Hsp70 requires these two cysteines to promote polymerization of MLKL tetramers in an ATP-independent manner. NBC1 blocks MLKL polymerization and subsequent cell death. This work reinforces the importance of MLKL polymer formation and identifies chaperone Hsp70 as an obligatory facilitator for MLKL polymerization, providing further insights for intervention of necroptosis-associated diseases.

Abstract

Necroptosis is a regulated necrotic cell death pathway involved in development and disease. Its signaling cascade results in the formation of disulfide bond-dependent amyloid-like polymers of mixed lineage kinase domain-like protein (MLKL), which mediate proinflammatory cell membrane disruption. We screened compound libraries provided by the National Cancer Institute and identified a small-molecule inhibitor of necroptosis named necroptosis-blocking compound 1 (NBC1). Biotin-labeled NBC1 specifically conjugates to heat shock protein Hsp70. NBC1 and PES-Cl, a known Hsp70 substrate-binding inhibitor, block the formation of MLKL polymers, but not MLKL tetramers in necroptosis-induced cells. In vitro, recombinant Hsp70 interacts with the N-terminal domain (NTD) of MLKL and promotes NTD polymerization, which has been shown to mediate the cell killing activity. Furthermore, the substrate-binding domain (SBD) of Hsp70 is sufficient to promote MLKL polymerization. NBC1 covalently conjugates cysteine 574 and cysteine 603 of the SBD to block its function. In addition, an SBD mutant with both cysteines mutated to serines loses its ability to promote MLKL polymerization. Interestingly, knockdown of Hsp70 in cells leads to MLKL destabilization, suggesting that MLKL might also be a client protein of Hsp70. In summary, using NBC1, an inhibitor of necroptosis, we identified Hsp70 as a molecular chaperone performing dual functions in necroptosis. It stabilizes MLKL protein under normal condition and promotes MLKL polymerization through its substrate-binding domain during necroptosis.

Necroptosis is a regulated immunogenic necrotic cell death process (1). Morphologically, it is characterized by organelle swelling, plasma membrane rupture, and release of damage-associated molecular patterns (DAMPs). It has been implicated in a variety of pathological conditions, including infection, inflammation, ischemic injuries, cancer, and neurodegeneration (2–7).

A multitude of pathophysiologic stimuli have been shown to induce necroptosis, including death ligands, such as tumor necrosis factor α (TNFα), Fas ligand, or TNF-related apoptosis inducing ligand (TRAIL), or pathogen recognition receptors, such as Toll-like receptors 3 and 4 (TLR3, TLR4) or Z-DNA-binding protein 1 (ZBP1/DAI) (2–7). The best studied pathway is TNFα-mediated necroptosis. Following TNF binding to its receptor and concurrent inhibition of caspase 8, receptor interacting protein kinases 1 and 3 (RIPK1/3) interact through their RIP homotypic interaction motif (RHIM), activate via phosphorylation, and form an amyloid-like structure (8–13). RIPK3 recruits mixed lineage kinase domain-like protein (MLKL) to form the necrosome (14, 15). Phosphorylation of MLKL by RIPK3 induces a conformational change of MLKL, causing MLKL to form tetramers and translocate to the membrane fractions, resulting in cell death (16–21). Recently, we demonstrated that MLKL tetramers further polymerize to form disulfide bond-dependent amyloid-like fibers, which are essential for necroptosis execution. An MLKL cysteine mutant that fails to form a disulfide bond also fails to activate necroptosis efficiently. Moreover, compound necrosulfonamide (NSA) covalently conjugates cysteine 86 of human MLKL to block MLKL polymerization and necroptosis without blocking tetramer formation, suggesting that tetramer formation is not sufficient for cell killing, while polymers are necessary (22–24). However, how MLKL polymer formation is regulated is not known.

It is not surprising that molecular chaperone proteins have been implicated in the necroptosis pathway, since many different complexes form during the process. For example, heat shock protein 90 (Hsp90) and its cochaperone CDC37 have been shown to be involved in necroptosis at different steps (25–29). Hsp90 is an abundant and highly conserved molecular chaperone with a diverse set of client proteins, many of which are members of the kinome. Interactions are dependent on recognition of the kinase or pseudokinase domain by cochaperone CDC37. It has been reported that the Hsp90/CDC37 complex interacts with RIPK3 and is required for RIPK3 activation. Chemical inhibition of Hsp90 prevents RIPK1 interaction with RIPK3 and blocks phosphorylation of RIPK3 and MLKL, abrogating necroptosis (25, 27). Hsp90/CDC37 also interacts with MLKL to promote MLKL oligomerization and membrane translocation (26). Interestingly, Hsp90 inhibitors prevent necroptosis induced by TNF, but fail to block necroptosis induced by the overexpression of the N-terminal domain (NTD) of MLKL (26).

Through an unbiased small-molecule screen, we have identified a chemical inhibitor of necroptosis that targets an additional molecular chaperone, heat shock protein 70 (Hsp70). Hsp70 stabilizes MLKL and promotes MLKL polymerization. Unlike Hsp90, Hsp70 interacts with the NTD of MLKL, and inhibition of Hsp70 blocks necroptosis induced by the dimerization of the NTD. This work highlights the complex and important role of heat shock proteins in necroptosis.

Results

Identification of Necroptosis-Blocking Compound NBC1.

We performed a forward small-molecule screen using libraries provided by the National Cancer Institute’s Developmental Therapeutics Program Open Chemical Repository to identify inhibitors of TNFα-induced necroptosis. Using a phenotypic cell death assay, 2,675 small molecules were evaluated. We initiated the screen with the colon cancer cell line HT-29, which undergoes TNFα-mediated necroptosis using conventional stimuli: TNFα (T) to activate TNFR1, Smac mimetic (S) to inhibit cIAP-mediated ubiquitination of RIPK1, and ZVAD-FMK (Z), the pan-caspase inhibitor (10). RIPK1 inhibitor necrostatin-1 (Nec-1) and MLKL inhibitor NSA were used as positive controls (9, 14, 30). Successful candidate compounds from the primary screen were further tested in NTD-DmrB cells, which stably express a tet-inducible truncated MLKL transgene containing the N-terminal domain (NTD; amino acids 1 to 190) fused to a chemically induced dimerization domain (DmrB) with C-terminal 3×FLAG tag (22). Using NTD-DmrB cells bypasses the proximal necroptosis signaling cascade and identifies inhibitors that act downstream of MLKL dimerization. A third-tier assay used mouse fibroblast L929 cells. Because cysteine 86 targeted by NSA is not conserved in the Mus musculus MLKL homolog, NSA is ineffective in murine cells (14), ensuring that any novel MLKL inhibitors would have a mechanism unique from NSA. From the tiered screen, a single small molecule effectively blocked necroptosis in all cell lines. It was originally named NSC632841 in PubChem, with an International Union of Pure and Applied Chemistry (IUPAC) name of (3E,5E)-3,5-dibenzylidene-1-prop-2-enoylpiperidin-4-one (Fig. 1A). We renamed it NBC1 for “necroptosis-blocking compound 1.” NBC1 had variable EC₅₀ values in HT-29 (0.72 µM), NTD-DmrB (3.4 µM), and L929 cells (8.6 µM), which may reflect cell type and species differences (Fig. 1 B–D).

Fig. 1.

Identification of a necroptosis-blocking compound, NBC1, from the NCI Open Chemical Repository Collection. (A) Chemical structure of NBC1. Potential Michael acceptor sites are labeled with a red asterisk. (B) NBC1 dose–response curve in HT-29 cells. Necroptosis was induced with TNFα (20 ng/mL), Smac mimetic (100 nM), and ZVAD-FMK (20 µM), abbreviated as T/S/Z. Cells were treated for 16 h, and cell viability was measured by the CellTiter-Glo assay. Viable cells expressed as a percentage of dimethyl sulfoxide (DMSO)-treated cells. Data are presented as mean ± SD of technical triplicates. Identical concentrations of T/S/Z were used in the following experiments unless otherwise indicated. (C, Top) Domain structures of MLKL and NTD-DmrB, which is the N-terminal domain (NTD) of MLKL (amino acids 1 to 190) fused to a chemically induced dimerization domain (DmrB) with 3×FLAG-tag at the C terminus. (C, Lower) NBC1 dose–response curve in NTD-DmrB cells. Necroptosis was induced with dimerizer and ZVAD-FMK (D/Z). Cells were treated for 6 h, and cell survival was assessed by the CellTiter-Glo assay. Viable cells expressed as a percentage of DMSO-treated cells. Data are presented as mean ± SD of technical triplicates. Identical concentrations of D/Z were used in following experiments. (D) NBC1 dose–response curve in mouse fibrosarcoma L929 cells. Necroptosis was induced with TNFα (5 ng/mL) and ZVAD-FMK (T/Z). Cells were treated for 16 h, and cell viability was measured by the CellTiter-Glo assay. Viable cells expressed as a percentage of DMSO-treated cells. Data are presented as mean ± SD of technical triplicates. Identical concentrations were used in the following experiments unless otherwise indicated.

NBC1 Requires Michael Acceptors to Block Necroptosis, but Does Not Target the Cysteines of NTD.

NBC1 contains three α,β-unsaturated enone moieties, presumably acting as Michael acceptors, which are highly reactive and known to covalently conjugate active cysteines in target proteins (31). We have shown previously that NSA contains two Michael acceptors, both of which are required to efficiently conjugate cysteine 86 of human MLKL (10, 23). NBC1 analogs with different amounts of Michael acceptors were generated to perform structure–activity relationship studies (Fig. 2A). Compounds with two Michael acceptors (NBC1-D1, NBC1-D2) retained significant activity, while compound NBC1-D3 with only one Michael acceptor lost all activity (Fig. 2B). This result suggests that at least two Michael acceptors are required for NBC1 to block necroptosis. When the target site of NSA, cysteine 86 of human MLKL, was mutated to serine in the NTD-C86S-DmrB cells, NSA no longer inhibited necroptosis, while NBC1 still did (Fig. 2C). We have shown previously that the NTD-4CS-DmrB mutant with all four cysteines (C18, C24, C28, and C86) mutated to serines does not activate necroptosis efficiently (22). However, when NTD-4CS-DmrB was expressed more than 10 times higher in a stable line than in the NTD-DmrB line (Fig. 2D, compare lanes 1 and 6), about 40% of cells could be induced to undergo necroptosis (Fig. 2E). NBC1 was still protective in the NTD-4CS-DmrB^high cells, indicating that it did not target any cysteine moiety in the NTD-DmrB protein (Fig. 2E).

Fig. 2.

NBC1 requires Michael acceptors for its function, but does not target the cysteines of MLKL. (A) Chemical structures of NBC1 and its derivatives (NBC1-D1, NBC1-D2, NBC1-D3). Potential Michael acceptor sites are labeled with red asterisks. (B) Dose–response curve of NBC1 and derivatives in NTD-DmrB cells (*P < 0.05). (C) Necroptosis survival assay in NTD-DmrB and NTD-C86S-DmrB cells. The mutation of cysteine 86 to serine in the NTD-C86S-DmrB cells renders NSA ineffective (*P < 0.05). (D) Immunoblot comparison of MLKL expression in NTD-4CS-DmrB^high and NTD-DmrB cells. NTD-4CS-DmrB has all four cysteines in NTD mutated to serines. Increasing amounts of whole-cell lysate were loaded as indicated for SDS/PAGE and subjected to Western blotting with FLAG and LDH antibodies. Both NTD-DmrB and NTD-4CS-DmrB are tagged with C-terminal 3×FLAG epitope. (E) Necroptosis survival assay in NTD-4CS-DmrB^high cells (*P < 0.05).

NBC1 Specifically Conjugates Hsp70.

To identify the molecular target of NBC1, biotinylated NBC1 (biotin-NBC1) and negative analog (biotin-NBC1-D3) were generated (Fig. 3A). Biotin-NBC1 maintained its ability to block necroptosis but required a higher concentration (Fig. 3B). Following treatment of NTD-DmrB cells with increasing concentrations of biotin-NBC1, a unique avidin-positive band was identified at ∼72 kDa (Fig. 3C, arrowhead). Streptavidin precipitation of biotin-NBC1–treated NTD-DmrB cell lysates isolated this 72-kDa protein (Fig. 3D, arrowhead). Mass spectrometry identified the labeled protein as the molecular chaperone heat shock protein 70 (Hsp70; HSPA1A). This interaction was confirmed in biotin-NBC1–treated NTD-DmrB cells (Fig. 3E). Notably, MLKL was not identified in the mass spectroscopy analysis, nor did MLKL interact with biotin-NBC1 in vivo. The conjugation was covalent, as boiling failed to disrupt biotin-NBC1 binding to Hsp70 in vivo (Fig. 3E) or recombinant human Hsp70 in vitro (Fig. 3F). Interestingly, higher biotin-NBC1 concentrations resulted in lower signals in Hsp70 immunoblot (Fig. 3F, compare lanes 2, 4, and 6), suggesting that biotin-NBC1 might block the antibody (BD 610608) binding to its antigen, which is aa 429 to 640, containing its substrate binding domain (SBD). Furthermore, biotin-NBC1 did not conjugate another important chaperone Hsp90 in vitro (SI Appendix, Fig. S1A).

Fig. 3.

NBC1 specifically conjugates Hsp70. (A) Chemical structures of biotinylated NBC1 (biotin-NBC1) and its negative analog (biotin-NBC1-D3). Potential Michael acceptor sites are labeled with red asterisks. (B) Necroptosis survival assay in NTD-DmrB cells. Cells were treated as in Fig. 1C (*P < 0.05). (C) Identification of a unique biotin-NBC1–conjugated protein. NTD-DmrB cells were treated with increasing concentrations of biotin-NBC1 for 16 h. Whole-cell lysates were separated by SDS/PAGE, followed by avidin-HRP Western blotting. (D) Identification of Hsp70 as an NBC1 target protein. NTD-DmrB cells were treated with or without biotin-NBC1, and biotin-labeled proteins were precipitated with monoavidin beads as described in Materials and Methods. Eluates were separated by SDS/PAGE followed by silver staining. The band marked by the arrow was excised for mass spectrometry and identified as Hsp70. (E) NTD-DmrB cells were treated with DMSO or biotin-NBC1 (20 µM), and 1 mg of whole-cell lysates were subjected to monoavidin bead precipitation. Eluates and 20 µg whole-cell lysates were subjected to Western blotting with antibodies against FLAG and Hsp70 (Enzo, ADI-SPA-812). (F) In vitro conjugation assay. Recombinant human Hsp70 (350 nM) was incubated with increasing concentrations of biotin-NBC1 or biotin-NBC1-D3 overnight at 4 °C. Samples were boiled in SDS loading buffer and subjected to Western blotting with avidin-HRP and Hsp70 antibody (BD no. 610608).

Chemical Inhibition of Hsp70 with Known Inhibitor PES-Cl Blocks Necroptosis.

Hsp70 has two highly conserved domains, an N-terminal nucleotide-binding domain (NBD), which regulates the affinity of substrate binding, and a C-terminal substrate-binding domain (SBD) (32–34) (Fig. 4A). In order to confirm a role for Hsp70 in necroptosis, two chemical inhibitors with unique mechanisms of action were used in cell death assays. VER-155008 (VER), which acts as an ATP competitive inhibitor to block NBD function (35, 36), did not protect cells from necroptosis in any cell lines tested, including HT-29 cells, L929 cells, NTD-DmrB cells, and NTD-4CS-DmrB^high cells. On the contrary, PES-Cl, which binds to the SBD and disrupts interaction with substrate proteins (37), was protective (Fig. 4 B–E). This suggests that blockade of Hsp70’s SBD but not NBD is critical to its protective effect. Hsp70 inhibitors are known to disrupt autophagy and induce accumulation of p62 oligomers. NBC1, but not its negative analog NBC1-D3, also induced p62 aggregation and LC3-II accumulation (Fig. 4F), suggesting that it indeed inhibits Hsp70.

Fig. 4.

Known Hsp70 inhibitor PES-Cl blocks necroptosis. (A) Diagram of Hsp70 domain structure. Hsp70 contains a nucleotide-binding domain (NBD) at amino acids 1 to 385 and a substrate-binding domain (SBD) at amino acids 390 to 616. Compound VER-155008 (VER) inhibits NBD through ATP competition, and compound PES-Cl blocks substrate binding. (B) Necroptosis survival assay in HT-29 cells. X-axis represents the concentration of compounds (*P < 0.05). (C) Necroptosis survival assay in L929 cells (*P < 0.05). (D) Necroptosis survival assay in NTD-DmrB cells (*P < 0.05). (E) Necroptosis survival assay in NTD-4CS-DmrB^high cells (*P < 0.05). (F) NBC1 effect on p62 aggregation and LC3 accumulation. NTD-DmrB cells were treated with DMSO, PES-Cl (10 µM), NBC1 (10 µM), or NBC1-D3 (10 µM) for 16 h. Cell lysates were subjected to Western blotting with p62, LDH, and LC3 antibodies.

Inhibition of Hsp70 by NBC1 Blocks MLKL Polymerization.

Next, we examined at which step inhibition of Hsp70 blocks necroptosis. In HT-29 cells, T/S/Z-induced MLKL phosphorylation (Fig. 5A) and MLKL tetramer formation (Fig. 5B) were not affected by cotreatment with NBC1, suggesting that Hsp70 functions downstream of MLKL tetramer formation. NBC1 also did not affect membrane translocation of MLKL (Fig. 5C). To determine if Hsp70 could regulate polymerization of MLKL, we conducted semidenaturing detergent agarose gel electrophoresis (SDD-AGE), which is widely used to detect SDS-resistant amyloid or amyloid-like polymers (38). NBC1 but not NBC1-D3 inhibited MLKL polymerization (Fig. 5D, lane 3), suggesting that Hsp70 is required for MLKL tetramers to further polymerize to form functional polymers (Fig. 5E). Similarly, in NTD-DmrB cells, NBC1 did not inhibit D/Z-induced tetramer formation (Fig. 5F) or membrane translocation (Fig. 5G), but blocked polymer formation (Fig. 5H), confirming the essential role of Hsp70 in MLKL polymerization (Fig. 5I). Necroptotic cells expose phosphatidylserine (PS) on the outer plasma membrane after MLKL activation (39). In both HT-29 and NTD-DmrB cells, NBC1 blocked signal-induced PS externalization, visualized by Annexin V-FITC staining (SI Appendix, Fig. S1 B–E), indicating that NBC1 indeed inhibits MLKL activation.

Fig. 5.

Hsp70 is required for MLKL polymer formation. (A) HT-29 cells were induced with or without T/S/Z for 6 h in the presence of DMSO, NBC1 (10 µM), or NBC1-D3 (10 µM). Whole-cell lysates were separated by SDS/PAGE and subjected to Western blotting with indicated antibodies. Antibody against phosho-ser358 of MLKL is labeled as p-MLKL. (B) HT-29 cells were treated as in A. Whole-cell lysates were separated by nonreducing SDS/PAGE and subjected to Western blotting with MLKL antibody. (C) HT-29 cells were treated as in A, and cytosol and crude membrane fractions were obtained. Western blotting was performed with indicated antibodies. LAMP1 is a lysosomal membrane protein, and 14–3-3 is a cytosolic protein. (D) HT-29 cells were treated as in A. Whole-cell lysates were separated by SDD-AGE and subjected to Western blotting with MLKL antibody. Under this condition, the MLKL monomer was barely detected, while the polymers presented a strong characteristic spread-out pattern in necroptotic samples. (E) Working model in HT-29 cells. Necroptosis induction promotes the formation of necrosome and MLKL tetramers. Hsp70 facilitates the polymerization of MLKL tetramers into functional polymers. Compound NBC1 inhibits Hsp70 SBD function to block MLKL polymer formation and necroptosis. (F) NTD-DmrB cells were induced with or without D/Z for 4 h in the presence of DMSO, NBC1 (10 µM), or NBC1-D3 (10 µM). Whole-cell lysates were separated by nonreducing SDS/PAGE (Upper) or reducing SDS/PAGE (Lower), followed by Western blotting with FLAG antibody. (G) NTD-DmrB cells were treated as in F, and cytosol and crude membrane fractions were obtained. Western blotting was performed with indicated antibodies. (H) NTD-DmrB cells were treated as in F. Whole-cell lysates were separated by SDD-AGE and subjected to Western blotting with FLAG antibody. (I) Working model in NTD-DmrB cells. Dimerizer induces the formation of NTD-DmrB tetramers. Hsp70 facilitates the polymerization of NTD-DmrB tetramers into functional polymers. Compound NBC1 inhibits Hsp70 SBD function to block MLKL polymer formation and necroptosis.

Hsp70 Promotes MLKL Polymerization In Vitro.

Previously, we have developed a recombinant system to detect in vitro polymerization of MLKL (22, 23). Next we examined how Hsp70 affects MLKL polymerization. Recombinant GST-NTD readily interacted with Hsp70, which was blocked by NBC1, but not NBC1-D3 (Fig. 6A). At low concentration (0.1 µM), GST-NTD did not from polymers after incubation (Fig. 6B, lane 1). Coincubation with Hsp70, but not bovine serum albumin (BSA), greatly enhanced GST-NTD polymerization (Fig. 6B, lane 2 and 3). Adding ATP to the reaction had no effect (Fig. 6B, lanes 4, 5, and 6), suggesting that NBD function is not required. Importantly, NBC1 but not NBC-D3 inhibited the effect of Hsp70 on GST-NTD polymerization (Fig. 6C).

Fig. 6.

Hsp70 promotes MLKL polymerization in vitro. (A) Immunoprecipitation with recombinant proteins. Recombinant Hsp70 (2 µM) was incubated overnight at 4 °C with DMSO, NBC1 (10 µM), or NBC1-D3 (10 µM) and then dialyzed against PBS buffer. Recombinant GST-NTD (C-terminal 3×FLAG-tagged) was then incubated with the treated Hsp70 samples and subjected to anti-FLAG immunoprecipitation, followed by Western blotting with Hsp70 antibody (Enzo, ADI-SPA-812). (B) In vitro MLKL polymerization assay. Recombinant GST-NTD (0.1 µM) was incubated alone, with Hsp70 (0.7 µM), or with BSA (0.7 µM) at room temperature for 3 h in the absence or presence of ATP (1 mM). The samples were separated by SDD-AGE (Upper) or SDS/PAGE (Middle), followed by Western blotting with FLAG antibody. Hsp70 and BSA were visualized by Coomassie blue staining (Lower). (C) Hsp70 was incubated with NBC1 and NBC1-D3 as described in A. GST-NTD (0.1 µM) was then incubated with BSA (0.7 µM) or treated Hsp70 (0.7 µM) at room temperature for 3 h. The samples were separated by SDD-AGE (Upper) or SDS/PAGE (Middle), followed by Western blotting with FLAG antibody. Hsp70 and BSA were visualized by Coomassie blue staining (Lower). (D) BSA, recombinant Hsp70, NBD, and SBD were visualized by Coomassie blue staining. (E) GST-NTD was incubated with BSA, Hsp70, NBD, and SBD as described in B. The samples were separated by SDD-AGE (Upper) or SDS/PAGE (Lower), followed by Western blotting with FLAG antibody. (F) Biotin-NBC1 or biotin-NBC1-D3 was incubated with the SBD and cysteine-to-serine mutants of the SBD (C574S, C603S, C574/603S), followed by Western blotting with avidin-HRP (Upper). Recombinant proteins were visualized by Coomassie blue stain (Lower). (G) GST-NTD was incubated with indicated recombinant proteins as in B. The samples were separated by SDD-AGE (Upper), and protein loading was visualized by Coomassie blue staining (Lower). (H) Working model. Hsp70 interacts with the NTD of MLKL tetramer and uses cysteine 574 and 603 to protect active cysteines in the tetramer, allowing new disulfide bonds to form properly between tetramers to promote polymerization.

Cysteines 574 and 603 of Hsp70 Are Targeted by NBC1.

When recombinant NBD and SBD were examined in the in vitro polymerization assay, SBD, but not NBD, promoted GST-NTD polymerization, similar to the full-length Hsp70 (Fig. 6 D and E). There are two cysteines (C574 and C603) in the SBD, which localize in the α-helical lid subdomain. When the two cysteines were mutated to serines in the SBD, the binding affinity of biotin-NBC1 with the mutant proteins drastically decreased, especially with the double mutant C574S/C603S (Fig. 6F). Furthermore, the C574S/C603S mutant completely lost the ability to activate GST-NTD polymer formation (Fig. 6G, lane 6), suggesting that these two cysteines in the SBD are essential for promoting MLKL polymerization (Fig. 6H).

Hsp70 Knockdown Blocks Necroptosis and Destabilizes MLKL.

Heat shock protein 70 is an inducible molecular chaperone responsible for protein folding and stability (32, 33). In order to evaluate whether reduction of Hsp70 inhibited necroptosis, Hsp70 was silenced with small interfering RNA (siRNA). Concurrent treatment with the pan-caspase inhibitor ZVAD-FMK was required to block apoptosis induced with Hsp70 knockdown. In NTD-DmrB cells, knockdown of Hsp70 with two different siRNA oligos effectively blocked necroptosis (Fig. 7A). However, Hsp70 knockdown also resulted in NTD-DmrB protein level decrease (Fig. 7B). Quantitative PCR revealed that mRNA level of NTD-DmrB was not reduced by Hsp70 knockdown, indicating that the decrease of NTD-DmrB protein is posttranscriptional (Fig. 7C). Since NTD-DmrB is a transgene, endogenous MLKL was examined next. In HeLa cells stably expressing RIPK3, knockdown of Hsp70 inhibited necroptosis (Fig. 7D), and endogenous MLKL level also decreased (Fig. 7E). Furthermore, when Hsp70 was knocked down with a short hairpin RNA (shHsp70) in HT-29 cells, necroptosis was inhibited and again MLKL level decreased (Fig. 7 F and G). These results suggest that MLKL is likely a client protein of Hsp70 and loss of Hsp70 protein results in destabilization of MLKL. Due to the high sequence homology between Hsp70 and the constitutively expressed Hsc70, we next assessed the role of Hsc70. Knockdown of Hsc70 did not protect against necroptosis and did not decrease MLKL expression (Fig. 7 H and I), suggesting that this effect is specific to Hsp70. Interestingly, when Hsp70’s cochaperone heat shock protein 40 (Hsp40; DNAJB1) was silenced, necroptosis was not affected and MLKL level did not decrease (SI Appendix, Fig. S1 F and G), suggesting that an alternate Hsp70 cochaperone might be involved in maintaining MLKL stability.

Fig. 7.

Hsp70 knockdown destabilizes MLKL. (A) NTD-DmrB cells were transfected with siRNAs against luciferase, MLKL, or Hsp70 for 72 h. Cells were then treated with D/Z to induce necroptosis for 6 h, followed by the CellTiter-Glo assay (*P < 0.05). (B) NTD-DmrB cells were transfected with siRNAs as in A. Cell lysates were subjected to Western blotting with Hsp70, FLAG, and β-tubulin antibodies. (C) NTD-DmrB cells were transfected with siRNAs as in A. The mRNA level of Hsp70 and NTD-DmrB is assessed by qPCR. All mRNA expression is normalized to actin, and siLuc is set to 1. Data are presented as mean ± SD of triplicate wells (*P < 0.05). (D) HeLa/TO-RIPK3 cells were transfected with siRNAs as in A. Cells were treated with T/S/Z to induce necroptosis for 16 h, followed by the CellTiter-Glo assay (*P < 0.05). (E) HeLa/TO-RIPK3 cells were transfected with siRNA and treated as in D. Cell lysates were subjected to Western blotting with Hsp70, MLKL, RIPK3, and LDH antibodies. (F) HT-29 cells were transduced with lentivirus harboring shRNA against Hsp70 and selected for 5 d. Parental cells as well as two different pools of knockdown cells (shHsp70.1 and shHsp70.2) were induced with T/S/Z for 16 h, followed by the CellTiter-Glo assay (*P < 0.05). (G) Hsp70 knockdown was performed as in F, and Western blotting was performed with Hsp70, MLKL, and LDH antibodies. (H) NTD-DmrB cells were transfected with indicated siRNAs for 72 h. Cells were treated with D/Z to induce necroptosis for 6 h, followed by the CellTiter-Glo assay (*P < 0.05). (I) NTD-DmrB cells were transfected with indicated siRNAs for 72 h, followed by Western blotting with Hsc70, FLAG, and β-tubulin antibodies.

Discussion

Previously we have reported that MLKL forms disulfide bond-dependent amyloid-like polymers to activate necroptosis (22). However, how MLKL polymer formation is regulated is not well understood. In this study, using a necroptosis-blocking compound, NBC1, we identified chaperone Hsp70 as an obligatory facilitator for MLKL polymerization. NBC1 contains three Michael acceptors, which could covalently conjugate highly reactive cysteines in target proteins. Structure–activity studies revealed that NBC1 required at least two Michael acceptors to effectively block necroptosis. With a series of in vitro and cell-based assays, we identified chaperone Hsp70 as its direct target. Mutagenesis experiments further demonstrated that cysteines 574 and 603 of Hsp70 were covalently conjugated by NBC1. Hsp70 markedly promoted the efficiency of MLKL in vitro polymerization, and this activity was inhibited by NBC1. Importantly, NBC1 blocked MLKL polymerization and necroptosis in living cells.

Our results confirm the previous finding that disulfide bond-dependent MLKL polymer formation is required for necroptosis. First, NTD-4CS-DmrB requires more than 10 times overexpression to achieve about half of the cell killing activity of wild-type NTD-DmrB protein (Fig. 2 C–E), suggesting that disulfide bond formation is important for MLKL function. Second, compounds that block MLKL polymerization also block necroptosis, although they might employ different mechanisms. For example, NSA conjugates cysteine 86 of MLKL, while NBC1 conjugates cysteines 574 and 603 of Hsp70 to block MLKL polymerization. Interestingly, neither NSA nor NBC1 blocks MLKL tetramer formation, suggesting that formation of MLKL tetramers is not sufficient for cell killing, while MLKL polymer formation is required.

The best known function of Hsp70 is to facilitate proteins folding into their native state by transiently interacting with short hydrophobic peptide segments to prevent aggregation or to refold aggregated proteins. This is extensively studied in pathological amyloid assembly and disassembly (40–42). For instance, Hsp70 in combination with Hsp110 and Hsp40 can function as disaggregase to disassemble Parkinson’s-linked α-synuclein amyloid fibrils to nontoxic monomers or small oligomers (41). However, in the case of MLKL, Hsp70 serves to promote polymerization rather than disassemble the polymers, suggesting a unique mechanism.

There are at least two major differences between the aggregation-prevention and disassembling function of Hsp70 and its MLKL polymerization-promoting function. First, our results indicate that Hsp70 promotes MLKL polymerization in an ATP-independent manner, while ATP binding and hydrolysis by the NBD are essential for Hsp70’s substrate binding and release cycle to perform its aggregation-prevention and disassembling function (32, 33). For MLKL, addition of ATP had no effect for Hsp70-assisted in vitro polymerization (Fig. 6B). Additionally, SBD alone was sufficient to promote MLKL in vitro polymerization (Fig. 6 E and G). Furthermore, the NBD inhibitor VER-155008 did not block necroptosis, while SBD inhibitors NBC1 and PES-Cl blocked MLKL polymerization and necroptosis (Figs. 4 and 5). Finally, knockdown of the Hsp70 cofactor Hsp40, which stimulates Hsp70’s ATPase activity, had no effect on necroptosis (SI Appendix, Fig. S1F). These results suggest that Hsp70 uses its SBD to bind MLKL tetramers and promote MLKL polymerization without requiring NBD function. Second, cysteines 574 and 603 of the SBD are important for Hsp70 to promote MLKL polymerization, while these cysteines have not been reported to be important for its aggregation-prevention and disassembling function. The SBD is divided into two subdomains, a two-layer β-sandwich, containing the peptide binding pocket, and α-helical lid subdomain, composed of five helices. Both cysteine residues are located in the α-helical lid domain (amino acids 508 to 641), which regulates the kinetics of substrate binding and can induce conformational changes in the bound substrate (32, 33). The lid does not need to fold over the substrate, allowing great flexibility for multiple sizes of substrate to bind (34). Our results show that NBC1 directly conjugated cysteines 574 and 603 to block Hsp70’s ability to promote MLKL polymerization (Fig. 6F). Moreover, recombinant SBD with these two cysteines mutated to serines lost its ability to activate MLKL polymerization (Fig. 6G). It is very interesting that the two cysteines of Hsp70 play a vital role for promoting MLKL polymerization, given the fact that MLKL polymer formation depends on proper disulfide bond formation among its tetramer subunits.

These results lead us to propose the following working hypothesis for how Hsp70 promotes MLKL polymerization (Fig. 6H). Phosphorylation of MLKL induces its conformational changes to form disulfide bond-dependent tetramers. Hsp70 interacts with exposed short hydrophobic peptides in the newly formed tetramer, and uses cysteines 574 and 603 to protect and maybe activate other cysteines in the tetramer. Binding of Hsp70 with the tetramer could also shield the tetramer from the reducing power of thioredoxin 1, which we have previously shown directly interacts with MLKL and keeps MLKL in a reduced state (23). The Hsp70-associated MLKL tetramer is then delivered to the growing MLKL polymer and forms proper disulfide bonds with the polymer. Once the tetramer incorporates into the polymer, it dissociates from Hsp70, possibly because of steric hindrance, and the released Hsp70 will start a new cycle again. This also explains why ATP is not needed for Hsp70 substrate dissociation. More experiments are needed to test this hypothesis. Promoting MLKL polymerization by Hsp70 is similar to a previous report that chaperones including Hsp70 could bind to toxic Aβ₄₂ oligomers and promote their assembly into larger, less toxic species. Interestingly, Hsp70 also functions equally effectively with or without ATP in that regard (43). Taken together, these observations suggest that, apart from its aggregation-prevention and disassembling function, Hsp70 can promote polymerization of small oligomers in an ATP-independent manner under specific circumstances. Understanding the mechanism of this function might also provide more insights into pathological aggregation-induced neurodegenerative diseases.

Knockdown of Hsp70 destabilized MLKL protein but did not disrupt mRNA levels (Fig. 7), consistent with MLKL’s client protein status. Yet, inhibition of Hsp70 with NBC1 did not significantly modulate MLKL protein levels in HT-29 or NTD-DmrB cells up to 48 h (SI Appendix, Fig. S2 A and B). This may be due to the more profound effects of Hsp70 knockdown on protein stability than chemical inhibition alone. In fact, inhibition of Hsp70 in silenced cells did not further decrease NTD-DmrB protein level (SI Appendix, Fig. S2C). It is also possible that NBC1 might not affect Hsp70 interaction with nascent MLKL protein during translation. Chemical inhibition of Hsp90 also does not consistently reduce levels of its client proteins RIPK1, RIPK3, or MLKL, and this is reportedly dependent on specific inhibitors and the cell lines utilized (25–29). The percentage of cell survival and effective concentration of NBC1 varied between cell lines. Mouse L929 cells had a 10-fold higher EC₅₀ than HT-29 cells, illustrating potential species and tissue differences that may be relevant to Hsp70 regulation in necroptosis.

In transformed cells, antagonism of Hsp70 and its constitutively active family member Hsc70 often leads to apoptotic cell death (44, 45). Previous publications have shown a divergent function for Hsp70 in necroptosis. Chemical inhibition of Hsp70 with the allosteric inhibitor JG-98 and analogs destabilizes RIPK1 regulators c-IAP1/2, XIAP, and cFLIP_S/L, causing apoptosis, or necroptosis if caspases are inhibited, in breast cancer cell lines (MDA-MB-231, MCF7, SK-BR-3) and Jurkat cells (46). However, the effects are variable in A549 (lung), T-47D (breast), and HT-29 (colon) cells, and cotreatment with JG-98, ZVAD-FMK, and NSA is not protective, suggesting cell line-specific differences (46). Similarly, VER and NBC1 treatment caused moderate cell number reduction in both HT-29 and NTD-DmrB cells, which was not rescued by ZVAD-FMK and NSA treatment, suggesting that the cytotoxicity of these Hsp70 inhibitors is not through apoptosis or necroptosis in these two cell lines (SI Appendix, Fig. S3 A and B). Furthermore, both compounds did not induce detectable degradation of c-IAP1/2 or XIAP in HT-29 cells (SI Appendix, Fig. S3C). These results suggest that Hsp70 inhibition-induced cytotoxicity is compound-specific and cell line-specific. Interestingly, NBC1 and PES-Cl attenuated apoptosis induced by cotreatment of TNF and TAK1 inhibitor 5Z-7, while VER had no apparent effect (SI Appendix, Fig. S3 D–G), suggesting that SBD, but not NBD, might be involved in activating apoptosis. Since caspase-8 activation requires its tandem death-effector domain (tDED) to assemble into filament structure (47), it is tempting to speculate that Hsp70 might also be involved in the polymerization of caspase-8, similar to its role in facilitating MLKL polymerization. Future experiments will be needed to address this possibility.

In the context of necroptosis, the role of Hsp70 is distinct from Hsp90. The Hsp90/CDC37 complex often associates with kinases through recognition of the kinase domain by CDC37. They indeed bind both RIPK3 and MLKL and facilitate MLKL tetramer formation (25–29). On the contrary, Hsp70 associates with the NTD of MLKL tetramers and promotes polymerization. Moreover, inhibition of Hsp90 blocks MLKL tetramer formation (25–27), which is not affected by Hsp70 inhibitors (Fig. 5). Importantly, Hsp90 inhibitor 17-AAG is unable to block necroptosis induced by NTD-DmrB expression, while Hsp70 inhibitor NBC1 could (SI Appendix, Fig. S4). These results suggest that Hsp90/CDC37 and Hsp70 might actually cooperate and function in sequential order to promote necroptosis. Hsp90 interacts with RIPK3 through its cochaperone CDC37 and assists RIPK3 with MLKL phosphorylation. Hsp90/CDC37 then binds the pseudokinase domain of MLKL and facilitates phosphorylated MLKL in tetramer formation. Subsequently, Hsp70 interacts with the NTD of MLKL tetramers and promotes MLKL tetramers to further polymerize, leading to eventual cell death. Many important questions remain regarding MLKL polymerization and cell death induction. What role does membrane association play in MLKL polymerization? Does it need recruiting factors on the membrane other than lipids? Most interestingly, how do MLKL polymers rupture membranes?

Materials and Methods

General Reagents.

Screening compounds were obtained from the National Cancer Institute (NCI)/Division of Cancer Treatment and Diagnosis (DCTD)/Developmental Therapeutics Program (DTP; https://dtp.cancer.gov/). The Chuo Chen lab synthesized NBC1 analogs and biotinylated compounds (SI Appendix, Supplemental Material and Methods). Recombinant human TNFα, Smac-mimetic, and anti-RIPK3 antibody were prepared as previously described (10). The following reagents and antibodies were used: ZVAD-FMK (ApexBio), dimerizer (Clonetech, no. 635058), necrosulfonamide and PES-Cl (Millipore), VER-155008, anti-FLAG M2 antibody and affinity gel (Sigma), anti-human MLKL (GeneTex, GTX107538), anti–phospho-S358 of MLKL (Abcam, ab187091), anti-RIPK1 (BD, no. 551042), anti-lactate dehydrogenase (LDH) (Abcam, ab53292), anti-Hsp70 (BD, no. 610608; Enzo, ADI-SPA-812-F), anti-Hsp40 (Cell Signaling, no. 4868), anti-Hsc70 (Santa Cruz, K-19), anti-p62 (Santa Cruz, SC-28359), anti-LC3 (Cell Signaling, no. 4108), anti-LAMP1 (Santa Cruz, SC-17768), anti-14–3-3 (Santa Cruz, SC-629), anti–c-IAP1 (BD, no. 556533), anti–c-IAP2 (BD, no. 51–90000062), anti-XIAP (BD, no. 610717), anti–caspase-8 (Cell Signaling, no. 9746), anti-PARP1 (BD, no. 556362), avidin-HRP (Cell Signaling, no. 3999S), and monoavidin agarose (Pierce, no. 20228). T/S/Z treatment includes TNF (20 ng/mL), Smac-mimetic (100 nM), and ZVAD-FMK (20 μM). L929 cells were treated with TNF (2 ng/mL) and ZVAD-FMK (20 μM). NTD-DmrB cells were treated with dimerizer (20 nM) and ZVAD-FMK (20 μM). For compound treatment, unless otherwise stated, NBC1, the negative analog, and PES-Cl were used at 10 μM, and necrosulfonamide (NSA) was used at 5 μM.

Cell Culture and Stable Cell Lines.

HT-29, HeLa, and L929 cells were cultured in DMEM (high glucose) supplemented with 10% FBS. NTD-DmrB, NTD-4CS-DmrB, and HeLa/TO-RIPK3 cell lines were reported before (22, 23, 30). They were generated in the background of HeLa-TetR cells that expressed the Tet repressor (TetR), and the transgene expression was induced with 50 ng/mL doxycycline (Dox) for 24 h.

Small Interfering RNA (siRNA) Transfection.

For siRNA transfection, cells were plated at 2,000 cells per well in 96-well plates and 100,000 cells per well in 6-well plates 24 h prior to transfection. Transfection was carried out as per GenMute (SignaGen) protocol with 5 nM siRNA for 96 wells and 50 nM for 6 wells. Cells were incubated in standard culture conditions for 72 h prior to treatment. The following siRNAs were used: siLuc, CGUACGCGGAAUACUUCGA; siMLKL, GCUAAGAAGAGAUAAUGAA; siHsp70-1, UGACCAAGAUGAAGGAGAU; siHsp70-2, AGGACGAGUUUGAGCACAA, siHsp40, ACCCGUCGUAUUCAAAGAUGU; siHsc70-1, CCGAACCACUCCAAGCUAU; siHsc70-2, UGACAAAGAUGAAGGAAAU; and siHsc70-3, ACGGAAAAGUCGAGAUAAU.

shRNA-mediated Hsp70 knockdown.

Stable knockdown of Hsp70 with lentivirus was performed as described before (30). Briefly, shHsp70 was cloned into the pTY-shRNA-EF1a-Hygromycin vector (gift from Zhijian “James” Chen, UT Southwestern, Dallas, TX) and cotransfected with vectors pMD2.G and psPAX2 into 293T cells to produce lentivirus. HT-29 cells were transduced with lentivirus and selected with 0.1 mg/mL hygromycin for 5 d. The sequence for shHsp70 is the same as for siHsp70-1.

Nonreducing Sodium Dodecyl Sulfate–Polyacrylamide Gel Electrophoresis (SDS/PAGE).

Nonreducing SDS/PAGE is as standard SDS/PAGE, but sample buffer excludes 2-mercaptoethanol or DTT.

Semidenaturing Detergent Agarose Gel Electrophoresis (SDD-AGE).

SDD-AGE was performed as described previously (38). Briefly, 1% agarose gel was cast in 1× TAE with 0.1% SDS. Cell lysates were loaded with sample buffer (0.5× TAE, 5% glycerol, 2% SDS, and 0.02% bromophenol blue), and the gel was run at 4 V/cm gel length in 1× TAE with 0.1% SDS. Proteins were transferred to a PVDF membrane with TBS buffer (20 mM Tris, pH 7.4, and 150 mM NaCl) using capillary transfer, followed by Western blotting.

Cell Lysates and Immunoprecipitation.

Cells were scraped and washed with PBS buffer twice, followed by lysing with five volumes of lysis buffer (50 mM Tris, pH 7.4, 137 mM NaCl, 1 mM EDTA, 1% Triton X-100, and 10% glycerol, supplemented with protease inhibitors). After 30 min incubation on ice, the cells were centrifuged at 20,000 × g for 12 min and supernatant was collected. Lysates (1 mg) were incubated with 20 μL anti-FLAG or monoavidin agarose beads at 4 °C overnight. Beads were washed five times with lysis buffer and eluted with 60 μL elution buffer (0.2 M glycine, pH 2.8) and immediately neutralized with 6 μL of 1 M Tris, pH 7.4. All procedures were done at 4 °C.

For crude membrane fractionation, cell pellets were resuspended in five volumes of buffer A (20 mM Tris, pH 7.4, 10 mM KCl, and 1 mM MgCl₂) and incubated on ice for 20 min. The cells were passed through a 22-G needle 30 times and centrifuged at 500 × g for 10 min. The supernatant was centrifuged again at 20,000 × g for 10 min and saved as cytosol fraction. The pellet was extracted with lysis buffer and centrifuged at 20,000 × g for 10 min and saved as crude membrane fraction.

Cell Survival Assay.

Cell survival was measured using CellTiter-Glo Luminescent Cell Viability Assay according to the manufacturer’s protocol (Promega). Cells were seeded at 2,000 cells per well in 96-well plates 24 h prior to treatment. Luminescence was measured using a BioTek Synergy 2 plate reader.

Recombinant Protein Purification.

The cDNAs encoding Hsp70 and its mutants were cloned into the pET21b vector. Cysteine mutants were generated by site-directed mutagenesis. His-fusion proteins were purified from BL21(DE3) Escherichia coli cells with Nickel beads as described before (23). Purified recombinant proteins were dialyzed against PBS buffer.

MLKL Polymerization.

Recombinant GST-NTD was generated as previously described (22). Recombinant GST-NTD (0.1 µM) was incubated with or without recombinant Hsp70, Hsp70 truncations or cysteine mutants, or BSA (0.7 µM) at room temperature in PBS buffer for 1 or 3 h as indicated.

In Vitro Compound Conjugation Assay.

Recombinant Hsp70, Hsp90, and Hsp70 truncations or cysteine mutants were mixed with NBC1 and NBC1 analogs and rotated at 4 °C in PBS buffer for 16 h.

Mass Spectrometry/Liquid Chromatography (MS-LC).

MS-LC was performed as previously described (30). Briefly, the protein band of interest was excised, destained, and reduced, followed by in-gel trypsin digestion. The peptides were extracted and analyzed by a QSTAR XL mass spectrometer (AB Sciex).

RNA Isolation, cDNA Generation, and qPCR.

Total RNA was isolated with Direct-zol RNA Kits (Zymo Research). cDNA was synthesized using iScript cDNA synthesis kit (BioRad, no. 1708891). Gene expression was assessed using standard qPCR approaches with iTaq universal SYBR Green supermix (no. 172-5124). Analysis was performed on a CFX Connect Real-Time PCR Detection System (BioRad). The 2^ΔΔCt method was used to analyze the relative change in gene expression normalized to actin. The following primers were used for qPCR: Hsp70-F, AGGACATCAGCCAGAACAAG-3; Hsp70-R, CTGGTGATGGACGTGTAGAAG; NTD-MLKL-F, ATGCTCCAGGACCAAGGAAAG; NTD-MLKL-R, CACTCAGCTTCCTGTTCACG; Actin-F, AACTCCATCATGAAGTGTGACG; and Actin-R, GATCCACATCTGCTGGAAGG.

Chemical Compound Synthesis.

Chemical compound synthesis is detailed in the SI Appendix.

Statistical Analysis.

Statistical analyses were performed with Excel. Student’s t test was used to determine significance in cell death, and P < 0.05 is used to determine significant differences. Data are presented as mean ± SD.

Data Availability.

All data are included in the manuscript and SI Appendix.