NLRP3 cages revealed by full-length mouse NLRP3 structure control pathway activation

Abstract

The NACHT-, leucine-rich-repeat- (LRR), and pyrin domain-containing protein 3 (NLRP3) is emerging to be a critical intracellular inflammasome sensor of membrane integrity and a highly important clinical target against chronic inflammation. Here, we report that an endogenous, stimulus-responsive form of full-length mouse NLRP3 is a 12- to 16-mer double-ring cage held together by LRR-LRR interactions with the pyrin domains shielded within the assembly to avoid premature activation. Surprisingly, this NLRP3 form is predominantly membrane localized, which is consistent with previously noted localization of NLRP3 at various membrane organelles. Structure-guided mutagenesis reveals that trans-Golgi network dispersion into vesicles, an early event observed for many NLRP3-activating stimuli, requires the double-ring cages of NLRP3. Double-ring-defective NLRP3 mutants abolish inflammasome punctum formation, caspase-1 processing, and cell death. Thus, our data uncover a physiological NLRP3 oligomer on the membrane that is poised to sense diverse signals to induce inflammasome activation.

Keywords: NEK7; NLRP3; TGN; TGN dispersion; cryo-EM; inflammasome; innate immunity; trans-Golgi network.

Figure 1.. Purification and Overall Structure of the NLRP3 Double Ring Cage

(A) Domain organization of NLRP3. (B) SDS-PAGE of elution and sucrose gradient fractions of NLRP3 purified in the absence of any nucleotide and Western blot (WB) of the same fractions visualizing NEK7. (C) Representative negative-staining EM images of NLRP3 oligomers purified without nucleotide (top), with ADP and MCC950 (middle) or with dATP (bottom). (D) Representative 2D class averages of NLRP3 oligomers purified without any nucleotide (top), with ADP and MCC950 (middle) or with dATP (bottom). (E and F) Cryo-EM maps of NLRP3 cage species purified in presence of ADP and MCC950 (E) or dATP (F). Maps are colored by resolution. See also Figure S1–S3.

Figure 2.. NLRP3 Cage Is Formed by LRR-LRR Interactions with PYD Inside the Cavity

(A) Atomic model of the NLRP3 cage structure. NLRP3 domains are displayed with the color scheme in Figure 1A. (B and C) Overview and detailed views of “face-to-face” (B) and “back-to-back” (C) interaction interfaces. Residues used for mutagenesis are color-coded: green for “back-to-back” and yellow or magenta for “face-to-face” interfaces. (D) Sucrose gradient profiles and representative negative-staining EM images of WT NLRP3 (black) and mutant NLRP3 color coded as in (B and C). The profiles were calculated by quantification of SDS-PAGE gels in Figure S5A. The double ring cage containing fractions are highlighted in a red square. The ratios of the fraction 3 (“monomers”) and fraction 5 (“double ring cages”) to the sum of fractions 3 and 5 are shown for each profile. Residue numbers correspond to the mouse NLRP3 sequence (B-D). (E) A comparison of the NLRP3 monomer from the double ring cage with the NLRP3–NEK7 complex (PDB ID: 6NPY) (Sharif et al., 2019). NEK7 is shown in magenta. (F) Sucrose gradient profiles of NLRP3 (solid line) and NEK7 (dashed line) in samples containing NLRP3 cage (top) and NLRP3 cage incubated with an excess of NEK7 (bottom) calculated by quantification of SDS-PAGE in Figure S5A. The double ring cage containing fractions are highlighted in a red square. (G) A negative-staining EM image of NLRP3 cages incubated with an excess of NEK7. All data in (D, F-G) is from one representative out of ≥ 3 independent experiments. See also Figure S4 and S5.

Figure 3.. PYDs Are Shielded within Two NACHT-LRR Rings to Avoid Nucleating ASC

(A and B) Nucleation of ASC^PYD filament formation by NLRP6^PYD-NACHT (positive control) or NLRP3 cage, showing fluorescence quenching curves of Alexa488-labeled ASC^PYD upon its oligomerization as a function of time (A) and the initial rates (B). RFU: relative fluorescence unit. Data are presented as mean ± s.d., n=3. Data is from one representative out of 3 independent experiments. (C) Central slices of the unsharpened maps of 7- (top) and 8-fold (bottom) NLRP3 cages obtained from the ADP + MCC950 sample. The PYD densities inside the NACHT-LRR cages are fitted with short NLRP3^PYD helices of 7 or 8 monomers, respectively. The NLRP3^PYD filament structure was modelled based on the ASC^PYD filament structure (PDB ID: 3J63) (Lu et al., 2014). The polymerization directions of the filaments are indicated with the arrows. (D) A model of a 6-fold NLRP3 cage with two NLRP3^PYD helices fitted with the polymerization directions (orange and blue arrowheads) facing the cavity (left) or the exterior (right). A distance between C-terminus of PYD (magenta) and the N-terminus of the NACHT is shown with black double-sided arrow. PYD to NACHT linker is shown with a dotted line. Polybasic region of the linker is in red. See also Figure S5.

Figure 4.. NLRP3 Double Ring Cages Are Mainly Associated with Membrane

(A) A negative-staining EM image of NLRP3 oligomers purified from the membrane extract of reconstituted HEK293T cells in the presence of dATP. (B) Whole cell lysate (Lysate), and cytosolic (Cyt) and membrane (Mem) fractions of LPS-treated WT iBMDM cells analyzed by WB using anti-NLRP3 antibody. (C) Sucrose gradient fractions of cytosolic and membrane extracts from LPS-treated WT iBMDMs analyzed by WB using anti-NLRP3 antibody. No nucleotide was added. (D) Native-PAGE of cytosolic and membrane extracts from LPS-treated WT and NLRP3^−/− iBMDMs. The final sample of the NLRP3 cages purified in the presence of dATP for structure determination was used as the control. Anti-NLRP3 and anti-β-actin antibodies were used for the WB. The relative amount of NLRP3 loaded (“load”) in each lane was visualized with anti-NLRP3 antibody (bottom). (E) In vitro lipid blot assay of double-ring and monomeric (ΔPYD) forms of NLRP3 with intact (WT) or mutated (“Linker mutant”) polybasic region. On the left, lipid scheme of the membrane with negative-charged (red) and neutral (yellow) lipid head groups is shown. On the bottom, SDS-PAGE is shown as the loading control. (F) Position and sequence of a WT polybasic region of mouse NLRP3 and its mutation (Linker mutant). Residues involved in TGN binding are colored red and their mutations blue. (G) Sucrose gradient profiles of WT and linker mutant NLRP3 calculated by quantification of SDS-PAGE in Figure S5J. The double ring cage containing fractions are highlighted in a red square. (H) Negative-staining EM images of NLRP3 samples from (G). (I) Negative staining EM images of NLRP3 oligomers purified from the membrane fraction of HEK293T cells reconstituted with OSBP-tagged linker mutant (left) and OSBP-tagged LRRm2 mutant (right). All data is from one representative out of ≥ 3 independent experiments. See also Figure S5.

Figure 5.. NLRP3 Cage Is Required for TGN Dispersion and NLRP3 Activation

(A and B) Confocal imaging of WT (A) and NLRP3^−/− (B) iBMDMs primed with LPS (top) or also treated with 20 μM nigericin for 1 h (bottom) by IF with goat anti-NLRP3 (magenta) and rabbit anti-TGN38 (green) antibodies, and DNA (Hoechst dye, blue). NLRP3 inflammasome specks are labelled with arrowheads. (C) Confocal imaging of NLRP3^−/− iBMDMs reconstituted with WT human mScarlet-NLRP3 primed with LPS (top) or also treated with 20 μM nigericin for 1 h (bottom) for TGN38 (IF, green), NLRP3 (mScarlet, magenta) and DNA (Hoechst dye, blue). (D) Confocal imaging of NLRP3^−/− iBMDMs reconstituted with WT human mScarlet-NLRP3 treated with LPS and MCC950 (top) or pre-treated with MCC950 and treated with 20 μM nigericin for 1 h (bottom) for TGN38 (IF, green), NLRP3 (mScarlet, magenta) and DNA (Hoechst dye, blue). (E-I) Confocal imaging of NLRP3^−/− iBMDMs reconstituted with WT (E) or double ring cage disrupting mutants of human mScarlet-NLRP3 (LRRm2, 3, 5 and ΔPYD) (F-I) primed with LPS (top) or also treated with 20 μM nigericin for 1 h (bottom) for 58K Golgi protein (IF, green), NLRP3 (mScarlet, magenta) and DNA (Hoechst dye, blue). (J and K) Confocal imaging of NLRP3^−/− iBMDMs reconstituted with NLRP3 linker-mutant (J) or OSBP-linker mutant (K) primed with LPS (top) or also treated with 20 μM nigericin for 1 h (bottom) for 58K Golgi protein (IF, green), NLRP3 (mScarlet, magenta) and DNA (Hoechst dye, blue). All images are maximum intensity Z projections with scale bars of 10 μm. All data is from one representative out of ≥ 3 independent experiments. See also Figure S6.

Figure 6.. Quantified Effects of NLRP3 Cage Disruption in TGN Dispersion and NLRP3 Activation

(A) Quantification of TGN dispersion based on confocal imaging for LPS- and nigericin-treated WT iBMDMs, with or without MCC950 pretreatment, NLRP3^−/− iBMDMs, NLRP3^−/− iBMDMs reconstituted with WT or mutant human mScarlet-NLRP3, with or without the Golgi-binding OSBP domain. Data are presented as mean ± s.d., n=3. (B) Cell death indicated by LDH release for WT iBMDMs, NLRP3^−/− iBMDMs, and NLRP3^−/− iBMDMs reconstituted with WT or mutant NLRP3, with or without the Golgi-binding OSBP domain. The level of LDH release is shown as a fold change between LPS-primed cells treated or not with nigericin. Data are presented as mean ± s.d., n=3. All data is from one representative out of ≥ 3 independent experiments. See also Figure S7.

Figure 7.. NLRP3 Activation Model

Proposed structural rearrangements of NLRP3 in the course of NLRP3 activation. In a resting state NLRP3 exists presumably in both monomeric and double ring cage forms. Upon activation NLRP3 gets transported to MTOC, with either “closed” (top) or “open” (bottom) NACHT domains. At MTOC the double ring cages get disrupted by NEK7 leading to a partial and then full NLRP3 oligomerization in a form of an active inflammasome complex.