HDAC6 mediates an aggresome-like mechanism for NLRP3 and pyrin inflammasome activation

Abstract

Inflammasomes are supramolecular complexes that play key roles in immune surveillance. This is accomplished by the activation of inflammatory caspases, which leads to the proteolytic maturation of interleukin 1β (IL-1β) and pyroptosis. Here, we show that nucleotide-binding domain, leucine-rich repeat, and pyrin domain-containing protein 3 (NLRP3)- and pyrin-mediated inflammasome assembly, caspase activation, and IL-1β conversion occur at the microtubule-organizing center (MTOC). Furthermore, the dynein adapter histone deacetylase 6 (HDAC6) is indispensable for the microtubule transport and assembly of these inflammasomes both in vitro and in mice. Because HDAC6 can transport ubiquitinated pathological aggregates to the MTOC for aggresome formation and autophagosomal degradation, its role in NLRP3 and pyrin inflammasome activation also provides an inherent mechanism for the down-regulation of these inflammasomes by autophagy. This work suggests an unexpected parallel between the formation of physiological and pathological aggregates.

Fig. 1.. NLRP3 and pyrin inflammasomes, but not the AIM2 inflammasome, co-localize with the MTOC.

(A) The NLRP3, pyrin, and AIM2 inflammasome pathways triggered by nigericin or MSU, TcdB, and dsDNA, respectively. As shown below, NLRP3 and pyrin inflammasome puncta localize at the MTOC. Inflammasome activation culminates in pro-caspase-1 and pro-IL-1β processing. Upward arrows indicate processing sites. (B) Immunofluorescence images showing the co-localization of NLRP3 and ASC puncta with the centrosomal markers ninein and GTU in THP-1 cells. Blue represents nuclear staining by Hoechst 33342. (C) Line scan of intensity distribution profiles of puncta a, b, and c from (B). (D–E) Live-cell images of iBMDM-Casp-1 (D) and iBMDM-IL-1β (E) at 30 min (top panel) and 60 min (bottom panel) post-nigericin stimulation, showing co-localization of inflammasome puncta (depicted by mNeonGreen) with the MTOC (depicted by SiR-Tubulin staining that labels the microtubule network). (F) FRET analysis of caspase-1 cleavage and IL-1β processing at MTOC as a function of time for areas inside and outside the puncta in iBMDM-Casp-1 (left) and iBMDM-IL-1β (right) cells. FRET was calculated by dividing the FRET channel fluorescence intensity (donor excitation with acceptor emission) with mTurquoise2 channel fluorescence intensity (donor excitation with donor emission). Values are mean±SD for n=10–15 cells. (G–H) Recruitment of IL-1β to a region in proximity to the MTOC imaged using 3D lattice light-sheet microscopy (LLSM). iBMDM-IL-1β cells stained with SiR-Tubulin were exposed to nigericin for 12 min (G), and 23 min (H). (a–c) Representative images deconvolved using the Richardson–Lucy algorithm corresponding to a single optical plane section. The arrows highlight the MTOC and the nearby locations where IL-lβ was recruited. (d–f) Enlarged images of the regions indicated by the arrows. (I–J) Lack of co-localization of AIM2 inflammasome puncta with the MTOC in iBMDM-Casp-1 (I) and iBMDM-IL-1β (J) cells activated by dsDNA for 6 hours. (K–L) Co-localization of pyrin inflammasome puncta with the MTOC in iBMDM-Casp-1 (K) and iBMDM-IL-1β (L) cells activated by TcdB toxin for 1 hour. Images are representative of three or more independent experiments and arrowheads indicate puncta or MTOC (B, D–E, G–L). Scale bars: 10 μm (B, D–E), 5 μm (Ga–c, Ha–c, I–L), and 1 μm (Gd–f, Hd–f).

Fig. 2.. Microtubule retrograde transport is required for the activation of NLRP3 and pyrin inflammasomes, but not the AIM2 inflammasome.

(A–C) NLRP3 inflammasome activation under various inhibition conditions analyzed by caspase-1 processing (p20) (A), quantification of propidium iodide (PI) permeability by flow cytometry (B), and secreted IL-1β quantified by ELISA (C). Colchicine and nocodazole: microtubule polymerization inhibitors; Ciliobrevin A: dynein ATPase inhibitor; Rocilinostat, tubacin and tubastatin A: HDAC6 inhibitors. (D) Caspase-1 processing (p20) upon NLRP3 inflammasome activation with pre-treatment of increasing concentrations of tubacin (left to right: 5 μM, 10 μM, 20 μM and 40 μM) or the NLRP3 inhibitor MCC950 (left to right: 0.1 μM, 0.5 μM, 1 μM, 10 μM and 20 μM). Anti-acetylated α-tubulin and anti-β-actin immunoblots are shown for tubulin acetylation and as the loading control, respectively. (E–G) Pyrin inflammasome activation under various pharmacological conditions analyzed by caspase-1 processing (p20) (E), PI permeability (F), and secreted IL-1β (G). (H–J) AIM2 inflammasome activation under various pharmacological conditions analyzed by caspase-1 processing (p20) (H), PI permeability (I) and secreted IL-1β (J). Data are presented as the mean±SD for 3–4 wells from three or more independent experiments.

Fig. 3.. HDAC6 is required for NLRP3 inflammasome activation.

(A) Immunoblotting shows the absence of HDAC6 protein in CRISPR/Cas9 Hdac6^−/− iBMDMs as compared to WT iBMDMs. Loading control was provided by the anti-β-actin antibody. The loss of HDAC6 leads to an increase in acetylated α-tubulin depicted using anti-acetylated α-tubulin antibody. (B–D) Compromised NLRP3 inflammasome activation in Hdac6^−/− iBMDMs challenged with nigericin, shown for caspase-1 processing (B), PI permeability (C), and secreted IL-1β (D). Data are presented as the mean±SD for triplicate wells from three or more independent experiments (C, D). (E) Domain architecture of human HDAC6 with important mutations (DA, Ub1 and Ub2) labeled. DA: H216A/H611A on catalytic residues, deacetylase mutant; Ub1: mutations H1160A/H1164A on zinc-coordinating residues; Ub2: mutation W1182A on the surface that binds ubiquitin. (F) Rescue of nigericin-mediated caspase-1 processing in Hdac6^−/− iBMDMs by reconstituting with WT human HDAC6. (G) Analysis of nigericin-mediated caspase-1 processing in Hdac6^−/− iBMDMs reconstituted with WT HDAC6, and the DA, Ub1 and Ub2 mutants. (H) Sensitivity to rocilinostat in Hdac6^−/− iBMDMs reconstituted with WT HDAC6, but not the DA mutant, as depicted by inhibition of p20 processing. (I) Rescue of nigericin-induced punctum formation in Hdac6^−/− iBMDM-Casp-1 cells transfected with WT HDAC6-mRuby3. Arrowheads indicate puncta. Cells containing puncta had HDAC6 expression (a and b), whereas cells that do not contain puncta did not have HDAC6 expression (c and d). (J) Rescue of nigericin-induced punctum formation in Hdac6^−/− iBMDM-Casp-1 cells stably reconstituted with WT and DA mutant of HDAC6-mRuby3, but not with Ub1 and Ub2 mutants of HDAC6-mRuby3. Arrowheads indicate puncta. HDAC6 WT reconstituted cells failed to form puncta upon pretreatment by rocilinostat. (K–L) Inflammasome puncta formation and its link to autophagy analyzed by immunofluorescence of ASC and the autophagy marker LC3b before (K) and after (L) NLRP3 inflammasome stimulation. Arrowheads indicate puncta. Images are representative of three or more independent experiments. Scale bars: 10 μm.

Fig. 4.. HDAC6 is required for pyrin inflammasome activation, but not AIM2 inflammasome activation.

(A–D) Requirement of HDAC6 and its ubiquitin binding ability in TcdB-induced pyrin inflammasome activation, shown by caspase-1 processing (p20) (A), PI permeability (B), IL-1b secretion (C), and punctum formation (D). DA: H216A/H611A on catalytic residues, deacetylase mutant; Ub1: mutations H1160A/H1164A on zinc-coordinating residues; Ub2: mutation W1182A on the surface that binds ubiquitin. (E–H) Lack of HDAC6-dependence in dsDNA-induced AIM2 inflammasome activation, shown by caspase-1 processing (p20) (E), PI permeability (F), IL-1b secretion (G), and punctum formation (H). Arrowheads indicate puncta or MTOC. Data are presented as the mean±SD for triplicate wells from three or more independent experiments (B, C, F, G). Images are representative from three or more independent experiments (D, H). Scale bars: 5 μm.

Fig. 5.. HDAC6 deficiency does not affect non-canonical and NLRC4 inflammasome activation in Nlrp3^−/− background.

(A–B) Non-canonical inflammasome activated by intracellular delivery of LPS (electroporation) quantified by PI permeability (A) and FAM-FLICA™ substrate cleavage by active caspase-11 (B). (C–D) NLRC4 inflammasome activation triggered by active FlaTox (inactive FlaTox served as a control) analyzed for caspase-1 processing (p20) (C), and PI permeability (D). Data are presented as the mean±SD for triplicate wells from three or more independent experiments. (E) Immunofluorescence analysis of NLRC4 punctum formation in Nlrp3^−/− (top) and Nlrp3^−/−Hdac6^−/− (bottom) iBMDMs upon treatment with FlaTox. Blue represents nuclear staining by Hoechst 33342. The NLRC4 inflammasome punctum represented by ASC staining is distinctly apart from the centrosomal marker ninein. Arrowheads depict puncta or MTOC. Images are representative of three or more independent experiments. Scale bars: 5 μm. (F) Summary of location of punctum formation and HDAC6-dependence in the different inflammasomes.

Fig. 6.. HDAC6 mediates microtubule transport of trans-Golgi network (TGN)-localized NLRP3.

(A–C) Immunofluorescence analysis of TGN38 and NLRP3 in Asc^−/− iBMDMs (A), WT iBMDMs (B) and Hdac6^−/− iBMDMs (C). (D) Immunofluorescence analysis of NEK7 and NLRP3 distribution in Hdac6^−/− iBMDMs. Images are representative of three or more independent experiments containing ~100 cells. Arrowheads depict puncta. Scale bars: 10 μm.

Fig. 7.. HDAC6 is required for NLRP3 inflammasome activation in mice.

(A–F) A mouse model of lethal LPS-induced endotoxemia. (A) Experimental layout with indicated timing and dose. (B–D) Effects of tubastatin A (HDAC6 inhibitor) (B), Hdac6^−/− (C), or MCC950 (NLRP3 inhibitor) (D) on IL-1β secretion. Values are mean±SD (n=3–5/group). (E) Effects of tubastatin A on acute lung injury (ALI). Representative histopathological images from harvested lung tissues are shown. Scale bar: 50 μm. (F) Quantified lung injury depicted by defined clinical parameters in ALI score. ALI scores are mean±SD (n=3–4/group). (G–K) A mouse model of MSU-induced peritonitis. (G) Experimental layout with indicted timing and dose. (H–I) Effects of Hdac6^−/− or Hdac6^−/− + MCC950 on peritoneal IL-1β production (H) and neutrophil recruitment (I) upon MSU challenge. Values are mean±SD (n=6–7/group). (J–K) Effects of MCC950 on peritoneal IL-1β production (J) and neutrophil recruitment (K) upon MSU challenge. Values are mean±SD (n=5–8/group). For (B–D, F, H–K), one-way analysis of variance (ANOVA) followed by Bonferroni’s multiple comparison test was performed for data analysis as used previously (64, 65). *: P≤0.05, **: P≤0.01, ***: P≤0.001 and ****: P≤0.0001.