Necroptosis: MLKL Polymerization

Abstract

Necroptosis is a subtype of regulated necrosis that occurs when caspases are inhibited or fail to activate. Stimulus of cell death receptors results in a signaling cascade that triggers caspase independent, immunogenic cell death. The core pathway relies on receptor interacting protein kinase (RIPK) 1 and 3, which interact through their receptor homotypic interacting motif (RHIM) domains, and form amyloid-like structures termed the necrosome. RIPK3 recruits and phosphorylates mixed lineage kinase domain-like pseudokinase (MLKL), the terminal mediator in the necroptotic pathway. MLKL polymerizes to form a second amyloid-like structure that causes cell membrane disruption resulting in cell death. Although the core necroptosis pathway has been elucidated, the details of MLKL membrane translocation and membrane disruption remain an open area of research.

Introduction

Necroptosis is one of the regulated cell death pathways^1–4. Phenotypically necroptosis resembles necrosis, but results from a specific caspase independent, kinase dependent signaling cascade, culminating in plasma membrane rupture and immunogenic cell death⁵. The inflammatory response generated from expulsion of cellular contents and cytokines stimulates both innate and adaptive immune responses. Necroptotic cell death has been described in TNF mediated systemic inflammatory response syndrome (SIRS), ischemic reperfusion injury, neurodegeneration, and pathogen infection, particularly viral disease^6–9. The relevance of necroptosis in viral disease is perhaps the most compelling^10–16. Yet, it is increasingly clear that programmed cell death in infectious and non-infectious disease states is unlikely to be unilateral. Recent evidence suggests that multiple parallel cell death pathways may be occurring concurrently^15,17–19. Beyond the pathophysiologic questions associated with necroptosis, the specific mechanism of how MLKL causes membrane permeabilization and cell death remains unresolved. Illumination of the cryptic aspects of the terminal necroptotic pathway will offer insights into how necroptosis can be medicinally manipulated.

Necroptosis signaling pathway

Cell extrinsic signals of necroptosis include ligands of the death receptor family (TNF, FAS or TRAIL), viral double stranded DNA or lipopolysaccharide stimulation of TLR 3 and 4, respectively, inflammatory mediators like interferon and viral stimulus of DNA-dependent activator of interferon regulatory factor^18,20. The best studied is TNF mediated necroptosis. The death receptors of the TNF family contain a cytoplasmic death domain that induces regulated cell death. Ligand binding to the death receptor TNFR1 induces receptor trimerization. This conformational change enhances binding with the death domain (DD) adaptor protein RIPK1. RIPK1 binding leads to either cell survival through inflammatory cytokine signaling, or regulated cell death depending on the context. When ubiquitination of RIPK1 is disrupted through loss of the E3 ligases²¹, cell death signaling proceeds. Subsequently, three cell death outcomes are possible: RIPK1-dependent or independent apoptosis, or necroptosis. Of the cell death pathways, apoptosis predominates when caspase 3 and 8 are active, and necroptosis progresses when caspase activity is inhibited, but recent evidence suggests that caspase inhibition is not imperative for necroptosis and both cell death pathways can progress concurrently dependent on cellular context²². The fate decision is dictated by TAK1 phosphorylation of Ser231 in the intermediate domain of RIPK1²³. Sustained phosphorylation of the intermediate domain of RIPK1 stimulates interaction with RIPK3 and necroptotic cell death, as long as RIPK3 intracellular level is sufficient^5,24. In the canonical necroptosis pathway RIPK1 binds to RIPK3, a serine/threonine kinase, through their C-terminal Receptor Homotypic Interacting Motif (RHIM) domains, which results in RIPK3 phosphorylation. Initially, RIPK1 kinase activity was believed responsible for RIPK3 phosphorylation but other kinases may be involved. RIPK3 activation via phosphorylation is imperative to necroptosis and seeds synthesis of amyloid-like polymers termed the necrosome²⁵ (Figure 1A).

Figure 1.

(A) Overview of the TNF-induced necroptosis pathway. Upon treatment with TNF, Smac-mimetic and a pan-caspase inhibitor Z-VAD-FMK, receptor-interacting protein kinase 1 and 3 (RIPK1, RIPK3) hetero-oligomerize into amyloid-like fibers which recruit mixed lineage kinase-like protein MLKL to form the necrosome. MLKL contains an N-terminal domain (NTD) and a kinase-like domain. After being phosphorylated by RIPK3 in the kinase-like domain, MLKL undergoes conformational changes and forms disulfide bond-dependent tetramers. MLKL tetramers subsequently translocate to membrane fractions and further polymerize to form unique amyloid-like fibers. How MLKL polymers disrupt membrane integrity is still not clear. Importantly, disulfide bond formation of MLKL is tightly controlled by the cytosolic oxidoreductase thioredoxin 1 (Trx1). Inhibition of Trx1 with compound PX-12 in some cells leads to spontaneous activation of MLKL polymerization and subsequent necroptosis. Compound necrosulfonamide (NSA) specifically conjugates cysteine 86 of human MLKL and cysteine 32 of Trx1 to block MLKL tetramer polymerization and necroptosis. (B) Overview of necroptosis activation by induced-dimerization of MLKL-NTD. In a cell line where NTD of MLKL is fused to a dimerization domain called DmrB, addition of compound Dimerizer leads to disulfide bond-dependent tetramer formation of the fusion protein. The tetramers further polymerize to form amyloid-like fibers, leading to membrane disruption and necroptosis. Similar to the full-length MLKL, compound NSA also blocks the polymerization step of NTD-DmrB. This cell line bypasses the requirement of the upstream activators, such as TNF, TNFR, RIPK1, and RIPK3 for necroptosis activation, providing a powerful tool to study steps downstream of MLKL oligomerization.

The terminal mediator: MLKL

RIPK3 recruits mixed lineage kinase domain-like pseudokinase (MLKL), the obligatory effector of necroptosis, to the necrosome and phosphorylates human MLKL at T357 and S358^26,27. Upon phosphorylation of MLKL by RIPK3, MLKL undergoes a conformational change that triggers MLKL tetramer formation which further polymerize to form unique amyloid-like fibers required for cell death^28–34 (Figure 1A). The structural stability of polymeric MLKL relies on intermolecular disulfide bonds. Under steady-state conditions, the cytoplasmic thiol oxidoreductase thioredoxin-1 (Trx1) system maintains cellular redox balance by catalyzing disulfide exchange reactions, to limit disulfide bond formation between target proteins, such as monomeric MLKL. The small molecule necroptosis inhibitor necrosulfonamide (NSA) enhances this interaction by cross-linking Trx1 cysteine-32 to cysteine-86 of human MLKL, which blocks MLKL polymerization and membrane permeabilization.³⁵ When cellular homeostatic mechanisms such as the Trx1 system are overwhelmed or inhibited, necroptosis ensues (Figure 1A).

MLKL’s preference for the plasma membrane

MLKL contains two domains including an N-terminal domain (NTD) and a C-terminal kinase-like domain. It has been shown that the recombinant NTD is sufficient to disrupt liposomes in vitro. Furthermore, induced-dimerization of a NTD fusion protein is sufficient to induce necroptosis in cells^{18,26,29,31–33} (Figure 1B). Multiple mechanistic models have been proposed to explain MLKL induced membrane disruption, but there is currently no definitive consensus. MLKL’s binding affinity for negatively charged phospholipids has been explored in multiple in vitro studies^26,32–34. Using hydrophobic membrane lipid strips and lipid binding assays, MLKL’s binding predilection included multiple phosphoinositide species and the inner mitochondrial membrane specific cardiolipin^32,33 Phosphoinositides are acidic phospholipids that interact with proteins and regulate integral membrane proteins³⁶. Phosphoinositide species are specific to each cellular compartment membrane. PIP2, specific to the plasma membrane, attracts proteins associated with membrane budding and fusion and regulates ion channels³⁶ MLKL’s phosphoinositide binding repertoire included phosphatidylinositol 4-phosphate (PI(4)P, trans Golgi), phosphatidylinositol (4,5)-bisphosphate (PI(4,5)P_2, plasma membrane), phosphatidylinositol (3,4,5)-triphosphate (PIP_3, plasma membrane), phosphatidylinositol (3)-phosphate (PI(3)P_, late endosome and MVB), phosphatidylinositol (3,4)-biphosphate (PI(3,4)P_2, early endosome), and phosphatidylinositol (3,5)-biphosphate (PI(3,5)P_2, late endosome)^26,33. Recombinant full length and N-terminal MLKL were capable of inducing leakage of PIP₂ containing liposomes, and this occurred more efficiently than liposomes mimicking the mitochondrial membrane^32,33. MLKL’s diverse organelle membrane binding was validated through cell fractionation experiments and immunostaining²⁶.

MLKL binding to plasma membrane PIP₂ is widely accepted as causal to cell death in necroptosis^18,34, but the importance of MLKL’s affinity for the other phosphoinositides and cardiolipin remains unclear. Initially, the mitochondria was believed to play a role in necroptosis³⁷, however, multiple studies have downplayed its significance^38–40. It is significant to note that the latter studies were conducted in murine cells and the difference may reflect cell-type specificity or species divergence. NTD of mouse MLKL was shown to be sufficient to induce necroptosis in L929 cells, but not in mouse dermal fibroblasts³⁰. Species differences among the MLKL NTD clearly exist, as human and chicken NTD are unable to kill mouse dermal fibroblasts despite orthologue sequence and structural similarities^30,34.

Although the NTD of MLKL was capable of disrupting liposomes, it was not able to disrupt nanodiscs, which have a more rigid membrane surface structure due to the presence of a scaffolding protein^26,32,36. This supports insertion of the NTD, at least partially, into the lipid membranes, and this insertion may be stabilized by the pseudokinase region^{30,34,41–43}. Whether this insertion causes direct cell membrane disruption in vivo or invokes a secondary response remains contentious. Evidence also exist to supports MLKL’s membrane permealization through formation of cation channels or association with existing ion channels^31,34,44,45, however, conformation and structural imaging has not yet identified the plasma membrane permeating structure.

MLKL: gain-of-function polymer

Evolutionarily conserved, prion-like signaling proteins or protein complexes such as mitochondrial antiviral signaling protein (MAVS) and the apoptosis associated spec-like protein (ASC), differ from other prion-like proteins in that they are gain-of-function polymers and their conformation is key to their function^46,47. The identification of an MLKL amyloid-like fiber distinct from the RIPK1-RIPK3 amyloid-like signaling complex, makes it the second prion-like structure identified in the necroptosis signaling pathway^25,28. Based on biochemical evidence, generation of the polymeric form of MLKL is a required step in the necroptotic pathway, suggesting that it too is a gain-of-function polymer. Similarities between the amyloid-like MLKL polymers and the MAVS and ASC signalosomes set them apart from other typical prion domains.

For example, their monomeric forms are not enriched in glutamine and asparagine and the monomers undergo regulated prion conversion^{47, 48}. Furthermore, different from most other fibrils that contain mainly β-sheets, MAVS and ASC polymers are comprised of mostly helical structures, and the N-terminal domain of MLKL that forms polymers also contains mainly helical structures²⁸. Interestingly, MLKL oligomerization and membrane translocation requires the chaperone HSP90 and its co-chaperone Cdc37^48,49. Although the heat shock proteins are widely associated with prion templating, chaperones do not appear to be required for MAVS or ASC polymer assembly.

The exact function of the MLKL polymer is still unclear. The homology between the NTD of MLKL and the N-terminal HeLo-like domain (HELL) domain of the fungal protein HELLP identified by Daskalov et al. may offer a clue to MLKLs polymeric function.⁵⁰ The HELL domain functions analogously to the HET-S protein from the ascomycete fungus Podospora anserine, which permeabilizes plasma membranes through an N-WHUPLQDO α-helical globular, pore-forming domain (HeLo)^50–52. The HET-S/s C-terminal prion domain is required for amyloid propagation. In prion form, the C-terminal of HET-s adopts a β solenoid fold which transconforms the HET-S prion domain during cell fusion. This leads to refolding of the HET-S HeLo domain exposing a helix capable of permeating the membrane and propagating peripheral cellular migration leading to cell death via pore formation.⁵⁰ HET-S does not form a prion^53,54. The HeLo domain of Het-s, which differs from the HET-S HeLo domain by 13 residues, is non-toxic⁵². This raises the question: Which molecular species causes membrane disruption? Is the MLKL polymer a toxic amyloid-like fibril or could it serve an alternative purpose as a seed for a toxic MLKL monomer or oligomer? Alternatively, MLKL not directly involved in structure dysfunction, could it serve as an intracellular signalosome or locally transmissible prion-like protein within an extracellular vesicle pathway⁵⁵?

Could PolyQ proteins and MLKL share a similar mechanism of cytotoxicity?

Another potentially applicable model for MLKL induced cell damage and dysfunction are the polyglutamine (PolyQ) fibrils associated with multiple degenerative diseases (huntingtin in Huntington’s disease, α-synuclein in Parkinson’s disease β-amyloid in Alzheimer’s disease). The polyQ stretches shared by these aggregate-prone proteins are causal to misfolding and correlated to severity of disease^{56, 57}. Amyloidogenic protein misfolding leads to nucleation-dependent generation of polymorphic oligomers and fibrils. Oligomeric aggregates, composed of a small numbers of molecules and large numbers of non-filamentous assemblies, are believed to be the toxic species^58,59. Exposure to oligomers disrupts giant unilamellar vesicle membranes through shifts in bilayer rigidity and adhesion force, inducing leakage of intravesicular contents⁵⁹. Yet, the fibrils may also contribute to membrane damage as in the case of type 2 diabetes mellitus^58,60–62. Human islet amyloid polypeptide (hIAPP) fibril growth at the extravesicular membrane of large unilamellar vesicles causes breakdown of cell barrier function. In vitro, the rate of leakage is accelerated by the presence of preformed seeds suggesting that fibril elongation rather than intermediate oligomers are the cause of the membrane damage. Fibrils can extract lipid at the liposome contact point of distortion⁶². In vivo, this is followed by intracellular calcium dysregulation and oxidative stress^61,63.

Although MLKL lacks polyQ tracts, the polymeric expansion may represent a common mechanism for membrane breakdown in that both cellular damage scenarios require intimate association negatively charged membrane lipids and positively charged residues or areas of high positive charge density on the protein polymers. The possibility that MLKL polymers and polyQ polymers might use a common mechanism to disrupt membrane integrity is very intriguing given that there is aacumulating evidence linking necroptosis to neurodegenerative diseases⁹. Future experiments could be devised to test if the in vitro assembled MLKL polymers and polyQ polymers could disrupt the membranes of purified organelles, which may represent the cellular context better than liposomes.

Conclusion

Research in the field of necroptosis and other regulated cell death pathways has grown exponentially in the past decade. Despite the vast accumulation of data, the intricacies and inter-relatedness of necroptosis and other regulated cell death pathways hinders therapeutic application to the treatment of human disease. The identification of a polymeric MLKL fiber is an intriguing addition to the pool of prion-like protein effectors in inflammatory signaling, but it does not answer the key mechanistic question of how plasma membrane disruption occurs. Rather, it encourages further exploration of the complexities of the cell death pathways.