Positive Charges in the Brace Region Facilitate the Membrane Disruption of MLKL-NTR in Necroptosis

Abstract

Necroptosis is a type of programmed cell death executed through the plasma membrane disruption by mixed lineage kinase domain-like protein (MLKL). Previous studies have revealed that an N-terminal four-helix bundle domain (NBD) of MLKL is the executioner domain for the membrane permeabilization, which is auto-inhibited by the first brace helix (H6). After necroptosis initiation, this inhibitory brace helix detaches and the NBD can integrate into the membrane, and hence leads to necroptotic cell death. However, how the NBD is released and induces membrane rupture is poorly understood. Here, we reconstituted MLKL_2-154 into membrane mimetic bicelles and observed the structure disruption and membrane release of the first brace helix that is regulated by negatively charged phospholipids in a dose-dependent manner. Using molecular dynamics simulation we found that the brace region in an isolated, auto-inhibited MLKL_2-154 becomes intrinsically disordered in solution after 7 ns dynamic motion. Further investigations demonstrated that a cluster of arginines in the C-terminus of MLKL_2-154 is important for the molecular conformational switch. Functional mutagenesis showed that mutating these arginines to glutamates hindered the membrane disruption of full-length MLKL and thus inhibited the necroptotic cell death. These findings suggest that the brace helix also plays an active role in MLKL regulation, rather than an auto-inhibitory domain.

Keywords: MD simulation; MLKL; auto-inhibitory; brace helix.

Figure 1

Negatively charged lipids promote the membrane-association of MLKL_2–154. (A) The overlaid 2D ¹H-¹⁵N TROSY-HSQC spectra of MLKL_2–154 in DMPC bicelles with (red) and without (blue) 20% cardiolipin. (B) The peak of six residues of MLKL_2–154 in DMPC bicelles supplemented with different concentrations of cardiolipin. The first left panel shows the same spectral region as the right panels for each residue. Panels 2–5 are spectra-recorded at increasing concentrations of 2%, 6%, 10%, and 20% cardiolipin. Residue names marked by an asterisk and colored in orange are new resonances which appeared upon the addition of cardiolipin. (C) Peak intensity changes (I_20%–I_0%) between MLKL_2–154 in DMPC bicelles with (I_20%) and without (I_0%) 20% cardiolipin. The orange dots denote the newly emerged resonances. The assignment of resonance peaks was conducted as described below.

Figure 2

The distortion and the release of the auto-inhibitory region of MLKL_2–154. (A) Backbone assignment of 0.6-mM U-[¹⁵N, ¹³C, ²H] MLKL_2–154 in DMPC bicelles with 10% cardiolipin. The resonances in red are the new set of resonances. (B) Primary and secondary structure alignment of MLKL_2–154 in DMPC bicelles with 10% cardiolipin. The coils above the sequence indicate the helical regions of MLKL_2–154 NMR structure (PDB code: 2MSV). Asterisks (*) below the sequence indicate the MLKL_2–154 residues with NMR chemical-shift assignment and dots (·) represent helical regions of MLKL_2–154 predicted by TALOS+ based on chemical-shift values. The new set of resonances were colored in red and marked by squares (□). (C) Residue-specific PREs were measured by recording 2D ¹H-¹⁵N TROSY-HSQC spectra when titrated with water-soluble paramagnetic agent Gd-DOTA (upper panel) and membrane-embedded paramagnetic agent 16-DSA (lower panel). I/I₀ vs. Gd-DOTA or 16-DSA concentrations were plotted for each residue, in which I and I₀ are the intensity of a peak in the presence and absence of the paramagnetic agent, respectively.

Figure 3

The MD simulations of MLKL_2–154 in pure water and in membranes. (A) The starting protein structure for 2MSV. The H6 helix (from residues 133 to 150) is present at that point in time and lasted for the first 7 nanoseconds. The RMSF is used for characterizing local changes along the protein chain, along with SSE (Secondary Structure Elements) of MLKL_2–154 highlighted in which α-helical regions are highlighted in pink. (B) The average structure over the 8th nanosecond, the residues 133–139 are no longer the SSE composition. (C) Over the entire simulation time (100 ns), H6 helix is missing, however, the helical structures of the NBD are still stable. (D) The initial position of MLKL_2–154 (PDB code: 2MSV) on the DMPC membrane predicted by the PPM Server. K16, R17, K50 and R51 are engaged initially opposite to the brace. (E) Over 100 ns simulation time, the positively charged cluster in the auto-inhibitory region got close to the membrane with R152 and R153 inserting into the membrane.

Figure 4

The arginine cluster acts as a critical element for the release of the auto-inhibitory region. (A) Superimposed 2D ¹H-¹⁵N TROSY-HSQC spectra of WT MLKL_2–154 and mutants (R145E/R146E, R152E/R153E, 4RE) in solution (blue) and in DMPC bicelles with 10% cardiolipin or DMPG (red), respectively. (B) MLKL positive mutants cannot be concentrated in membrane fraction after phase separation. HT29-MLKL KO were lentvirally reconstituted with different MLKL mutants. Cells were treated with T/S/Z for 6 h. T/S/Z treatment indicates TNF (20 ng/mL), Smac (100 nM), zVAD (20 μM). Results are reported from one representative experiment from at least three independent repeats. (C) MLKL mutants suppressed T/S/Z induced necroptosis. HT29-MLKL KO cell line was treated with T/S/Z for 10 h. T/S/Z treatment indicates TNF (20 ng/mL), Smac (100 nM), zVAD (20 μM). p values were determined by unpaired two-tailed Student’s t-test with Welch’s correction. **** p < 0.0001. All results are reported from one representative experiment from at least three independent repeats.