The adaptor MAVS promotes NLRP3 mitochondrial localization and inflammasome activation

Abstract

NLRP3 is a key component of the macromolecular signaling complex called the inflammasome that promotes caspase 1-dependent production of IL-1β. The adaptor ASC is necessary for NLRP3-dependent inflammasome function, but it is not known whether ASC is a sufficient partner and whether inflammasome formation occurs in the cytosol or in association with mitochondria is controversial. Here, we show that the mitochondria-associated adaptor molecule, MAVS, is required for optimal NLRP3 inflammasome activity. MAVS mediates recruitment of NLRP3 to mitochondria, promoting production of IL-1β and the pathophysiologic activity of the NLRP3 inflammasome in vivo. Our data support a more complex model of NLRP3 inflammasome activation than previously appreciated, with at least two adapters required for maximal function. Because MAVS is a mitochondria-associated molecule previously considered to be uniquely involved in type 1 interferon production, these findings also reveal unexpected polygamous involvement of PYD/CARD-domain-containing adapters in innate immune signaling events.

Figure 1. NLRP3 is cytosolic in the resting state and localizes to mitochondria upon activation

(A–B) Representative confocal immunofluorescence images (A) and quantification (B) of NLRP3 colocalized with Mitotracker in HEK-293T cells stably expressing NLRP3. Expression of NLRP3 was induced with doxycycline (DOX; 1 μg/ml), and cells were either untreated or treated with 15 μM nigericin for 45 min prior to imaging. (C–D) Representative immunofluorescence images (C) and quantification (D) of NLR colocalized with Mitotracker in HEK-293T cells expressing NLRP3 or NLRP4 under transient low expression conditions. Cells were transfected with 100 ng plasmid DNA for 8 h, and were either untreated or treated with 15 μM nigericin for 45 min prior to imaging. (E) Subcellular fractionation of WT and ASC KO BMDMs. BMDMs were either untreated (Mock) or primed with LPS (1 μg/ml for 4h) prior to stimulation with 7.5 μM nigericin for 30 min (Nig). Mitochondrial (M) and cytosolic (C) fractions were fractionated and analyzed for expression of NLRP3 by immunoblot. Purity of the fractions was assessed by blotting for Complex 1 (mitochondrial protein) and ERK-2 (cytosolic protein). Imaging data are representative of several images from 3 independent experiments. Data points on graphs represent independent fields (with 3–5 cells per field). Error bars are mean ± SEM. p-values from an unpaired t-test (two-tailed) are shown. Mito, Mitotracker; coloc, colocalization channel. See also Figure S1 and Table S1.

Figure 2. NLRP3 possesses an N—terminal sequence that controls mitochondrial association and ASC speckle formation

(A) Panel of N-terminal deletions of NLRP3. ‘----‘ indicates the deleted sequence. Structured region of the NLRP3 PYD encompasses residues 6–90. (B) Representative immunofluorescence images of HEK-293-ASC-YFP cells expressing WT and N-terminal deletions of NLRP3 under transient low expression conditions (100 ng plasmid DNA for 8h), and activated with 15 μM nigericin for 45 min prior to imaging. Representative images of unstimulated cells are shown in Figure S2. Imaging data are representative of several images from 4 independent experiments. Mito, Mitotracker. See also Figure S2.

Figure 3. Effect of N—terminal deletions of NLRP3 on mitochondrial recruitment, ASC speckle formation, and IL-1β secretion

(A–B) Quantification of ASC speckle formation in HEK-293-ASC-YFP cells transiently transfected with low levels of WT or N-terminal deletions of NLRP3 (as in Figure S2 and Figure 2B), and either untreated (A) or treated with 15 μM nigericin (B) for 45 min prior to imaging. (C) Quantification of colocalization of WT and N-terminal deletion mutants of NLRP3 with Mitotracker in HEK-293-ASC-YFP cells transfected as above, and either untreated or treated with 15 μM nigericin for 45 min prior to imaging. (D) Minimum N-terminal sequence (2-7) necessary for NLRP3 mitochondrial association and ASC aggregation is indicated in red. (E) ELISA for IL-1β in supernatants of PMA-differentiated (non-LPS primed) THP-1 cells stably expressing NLRP3 WT or NLRP3 Δ2-21, and stimulated with the indicated NLRP3 activators. Expression of NLRP3 was induced for 12 h with DOX (1 μg/ml), prior to differentiation with 100 nM PMA for 3 h. (F) Subcellular fractionation of HEK-293T cells stably expressing either NLRP3 WT or NLRP3 Δ2-21. NLRP3 expression was induced with 1 μg/ml DOX for 6 h prior to stimulation with nigericin for 45 min. Data points in (A) and (B) represent the percentage of all transfected cells (pooled from multiple fields) containing a fully developed, incompletely developed, or no ASC speckle. At least 20 cells were evaluated for each construct. Data points in (C) represent independent fields (3–5 cells per field). Data are representative of 3 independent experiments. Error bars are mean ± SEM, and p-values from an unpaired t-test (two-tailed) are shown. ND; not detectable. See also Figure S3.

Figure 4. The mitochondrial adapter MAVS mediates NLRP3 mitochondrial localization

(A, C) Representative confocal immunofluorescence images (A) and quantification (C) of NLRP3 co-localized with Mitotracker in HEK-293T cells treated with control siRNA or MAVS siRNA prior to over-expression of NLRP3. siRNA transfected into cells appears as punctate spots shown in blue. (B, D) Representative immunofluorescence images (B) and quantification (D, left) of HEK-293T cells treated with control siRNA or MAVS siRNA prior to transient low expression of NLRP3 and stimulation with 15 μM nigericin for 45 min. Immunoblot and corresponding densitometric quantification for efficiency of MAVS knock-down are shown (D, right). (E) Immunoblot showing NLRP3 association with MAVS in BMDMs upon treatment with activators of NLRP3, but not activators of NLRC4 or NOD2/NLRP1. In some experiments MAVS expression in lysates decreases upon stimulation with ATP for unknown reasons. (F) Immunoblot showing progressive loss of MAVS association by N-terminal deletion mutants of NLRP3 in HEK-293T cells. ‘|’ indicates removal of an irrelevant lane(s) here and throughout. Cells were transiently transfected with low levels of FLAG tagged WT or N-terminal deletions of NLRP3, and treated with 15 μM nigericin for 45 min prior to lysate preparation. (G) Immunoblot showing MAVS association of NLRP4 with the added N-terminus of NLRP3 (NLRP4 + 2-11) in HEK-293T cells. Cells were transiently transfected with the indicated FLAG tagged constructs (1 μg DNA/10⁶ cells for 20 h) prior to lysate preparation. (H) Immunoblot showing association of the indicated in vitro translated FLAG-tagged NLR proteins with purified MAVS. Immunoblots are representative of 2 independent experiments. Nig, nigericin; flg, flagellin; MDP, muramyldipeptide. Imaging data are representative of several images from 2 independent experiments. Data points on graphs in (C) and (D) represent individual fields with 3–5 cells per field (C) or 30–40 cells per field (D). Error bars are mean ± SEM, and p-values from an unpaired t-test (two-tailed) are shown. See also Figure S4 and Movies S1–S2.

Figure 5. Silencing MAVS in BMDMs reduces IL-1β secretion in response to NLRP3 activators

(A–E) ELISA for IL-1β in supernatants of control or MAVS-silenced, LPS-primed BMDMs treated with the indicated concentrations of ATP and nigericin for 20 min (A, B), transfected poly I:C and poly dA:dT for 6 h (C, E), or transfected flagellin for 2 h (D). (F) Immunoblot for efficiency of MAVS knock-down in BMDMs. MAVS-2 siRNA specifically silenced MAVS and was used for assessing IL-1β secretion shown in A–E. (G) Quantitative RT-PCR for IFN-β (G, left) and pro-IL-1-β (G, right) expression in human monocytes treated with NLRP3 activator LPS-ATP or RIG-I activator Influenza A strain HKX31 (multiplicity of infection=1). mRNA expression was normalized to the housekeeping gene GAPDH. ELISA and RT-PCR data are representative of 2 independent experiments. Error bars are mean ± SEM, and p-values from an unpaired t-test (two-tailed) are shown. * p<0.05; ** p<0.01; *** p<0.001; ns, not significant; ND, not detectable.

Figure 6. MAVS is required for optimal inflammasome activation by non-crystalline NLRP3 activators in vitro.

(A–F) ELISA showing IL-1β secretion from WT and MAVS −/− macrophages. LPS-primed BMDMs were treated with the indicated concentrations of ATP (A) or nigericin (B) for 20 min, alum crystals for 4 h (C), transfected poly I:C for 6 h (D), flagellin for 2 h (E) or poly dA:dT for 6 h (F). (G) Immunoblots showing processing of IL-1β and caspase-1 and ASC oligomerization in WT and MAVS −/− macrophages. LPS-primed BMDMs were treated with transfected poly I:C (2 μg/10⁶ cells) for 6 h, ATP (2.5 mM) and nigericin (7.5 μM) for 20 min, alum crystals (200 μg/ml) for 6 h or transfected flagellin (1 μg/10⁶ cells) for 2 h. Cell culture supernatants (Sup), cell lysates, and cross-linked pellets from whole cell lysates were analyzed by immunoblotting as indicated. Different doses of each NLRP3 activator were examined and representative blots at sub-saturating doses are shown. ‘*’ indicates non-specific bands observed at higher exposures. (H) Immunoblot for NLRP3 following stimulation with LPS (1 μg/ml for 4 h). ‘*’ indicates a non specific band. Data are representative of 2 independent experiments. Error bars are mean ± SEM, and p-values from an unpaired t-test (two-tailed) are shown. * p<0.05; ** p<0.01; *** p<0.001; ns, not significant; ND, not detectable. Casp-1, caspase-1. See also Figure S5.

Figure 7. MAVS is essential for NLRP3 inflammasome function in vivo during acute tubular necrosis

(A) Percentage weight loss of WT, MAVS −/−, and ASC −/− mice injected intra-peritoneally with 250 mg folic acid/kg body weight (n=5 mice per group). (B) Measurement of BUN (blood urea nitrogen) in sera of mice with ATN. (C) Representative hematoxylin-eosin stained images from kidneys of WT, MAVS −/−, and ASC −/− mice at 36 h post-folic acid treatment. (D–E) Representative confocal immunofluorescence images (D) and quantification (E) of neutrophil influx at the cortico-medullary junction in kidneys of mice with ATN. (F–G) Representative confocal immunofluorescence images (F) and quantification (G) of IL-1β positive cells at the cortico-medullary junction. The sum of neutrophils (E) or IL-1β positive cells (G) counted in five independent high power fields (200x) at the cortico-medullary junction of each mouse kidney is shown. Data points in (B, E and G) represent individual mice. Error bars are mean ± SEM, and p-values from a two-tailed t-test are shown. Results are representative of 3 independent experiments. FA, folic acid. See also Figure S6.