Endoplasmic Reticulum Stress Activates the Inflammasome via NLRP3- and Caspase-2-Driven Mitochondrial Damage

Abstract

Endoplasmic reticulum (ER) stress is observed in many human diseases, often associated with inflammation. ER stress can trigger inflammation through nucleotide-binding domain and leucine-rich repeat containing (NLRP3) inflammasome, which might stimulate inflammasome formation by association with damaged mitochondria. How ER stress triggers mitochondrial dysfunction and inflammasome activation is ill defined. Here we have used an infection model to show that the IRE1α ER stress sensor regulates regulated mitochondrial dysfunction through an NLRP3-mediated feed-forward loop, independently of ASC. IRE1α activation increased mitochondrial reactive oxygen species, promoting NLRP3 association with mitochondria. NLRP3 was required for ER stress-induced cleavage of caspase-2 and the pro-apoptotic factor, Bid, leading to subsequent release of mitochondrial contents. Caspase-2 and Bid were necessary for activation of the canonical inflammasome by infection-associated or general ER stress. These data identify an NLRP3-caspase-2-dependent mechanism that relays ER stress to the mitochondria to promote inflammation, integrating cellular stress and innate immunity.

Figure 1. IRE1 modulates infection-induced inflammasome activation via TXNIP

(A) Xbp1 splicing, a marker of IRE1 activation, in RB51-infected BMDM. qRT-PCR samples were treated with PstI to distinguish between spliced (184 bp) and unspliced variants (119 bp following PstI digestion) (B) Caspase-1 cleavage in RB51-infected BMDM in absence or presence of 4μ8c. Immunoblots are representative of n≥3 independent experiments performed and imaged in parallel with identical parameters using a LiCor Odyssey imaging system. (C) IL-1β ELISA analysis of supernatants from RB51-infected BMDM treated with or without TUDCA (chemical chaperone, 300 μM) and 4μ8c (IRE1 inhibitor, 50 μM). Error bars represent mean ± SD of n≥3 independent experiments. *** represent p-value <0.0001, n.s. = not significant. (D) qRT-PCR analysis of IL-1β transcript in RB51-infected BMDM in the absence or presence of 4μ8c. (E) Serum IL-1β in mice treated with 5% DMSO (control, n = 15) or 4μ8c (n = 15) and infected with RB51 (i.p., CFU 1 × 10⁸). Data in (E) were pooled from 2 independent experiments. (F) IL-1β ELISA analysis of supernatants from RB51-infected BMDM transfected with non-targeted and Xbp1 siRNA. (G) IL-1β ELISA analysis of supernatants from RB51-infected BMDM transfected with non-targeted and Txnip siRNA. (H) CM-H₂DCFDA was used to measure ROS during RB51 infection in presence of 4μ8c or Txnip siRNA. Rotenone serves as a positive control for ROS induction. Inset in (F,G,H) demonstrates silencing efficiency by immunoblot. UNT, TM, and L+A represent untreated, tunicamycin (10 μg/mL, positive control for ER stress activation) and LPS+ATP (200 ng/mL and 1mM respectively; positive control for inflammasome activation). See also Supplemental Figure S1.

Figure 2. ER stress-induced mitochondrial dysfunction drives IL-1β production

(A) The ROS indicator dye, CM-H₂DCFDA, was used to measure ROS in WT and mCAT BMDM infected with RB51. Rotenone serves as a positive control for ROS induction. (B) IL-1β concentrations in supernatants from RB51-infected WT and mCAT BMDM. (C) qPCR analysis of mtDNA release into cytosol during RB51 BMDM infection. (D) IL-1β concentrations in RB51-infected BMDM treated with or without cyclosporin A (CsA, 10 μM). Error bars in (A – D) represent mean ±SD of n≥3 independent experiments. *** represent p-value <0.0001, n.s. = not significant. UNT, TM, and L+A represent untreated and tunicamycin (10 μg/mL, positive control for ER stress activation), and 200 ng/mL LPS + 1mM ATP (positive control for inflammasome activation). See also Supplemental Figure S2.

Figure 3. NLRP3 is required for an ER stress-induced feed-forward loop of mitochondria damage

≥ Immunoblot of NLRP3 at the mitochondrial fraction of (A) WT and mCAT BMDM and (B) non-targeted (NT) and Txnip-silenced BMDM. (C) Quantitative PCR of mtDNA in cytosolic extracts from RB51-infected WT and Nlrp3^−/−, Asc^−/−, and z-YVAD-CHO (caspase-1 inhibitor, 2 μM) treated BMDM. (D) IL-1β concentrations in RB51-infected WT and Nlrp3^−/−, Asc^−/−, and z-YVAD-CHO (caspase-1 inhibitor, 2 μM) inhibited BMDM. Error bars represent mean ± SD of n≥3 independent experiments. *, **, and *** represent p-values of <0.05, <0.001, and <0.0001 respectively. n.s. = not significant. UNT, TM, and LPS+ATP represent untreated, tunicamycin (positive control for ER stress induction, 10 μg/mL), and LPS+ATP (positive control for inflammasome activation, 200 ng/mL and 1mM) respectively. Immunoblots in (A) and (B) are representative of n≥3 independent experiments that were performed and imaged in parallel with identical parameters using a LiCor Odyssey imaging system. See also Supplemental Figure S3.

Figure 4. NLRP3 and caspase-2 are required for ER stress-induced inflammasome activation

Immunoblot analysis of caspase-2 in (A) with or without 4μ8c (IRE1 inhibitor, 50 μM) and (B) RB51-infected WT and Nlrp3^−/− BMDM. Immunoblot analysis of caspase-2 in the mitochondrial fraction (C) with or without 4μ8c (IRE1 inhibitor, 50 μM) and (D) RB51-infected WT and Nlrp3^−/− BMDM – ET^╪ identifies duplicate lanes of the same sample. (E) qRT-PCR analysis of mtDNA in cytosolic fractions of infected WT and Casp2^−/− BMDM. IL-1β concentrations in (F) WT and Casp2^−/− RB51-infected (i.p., CFU 1 × 10⁸) BMDM. (G) qRT-PCR analysis of Il1b transcript in WT and Casp2^−/− RB51-infected BMDM. (H) Serum IL-1β serum by ELISA in WT (n = 15) and Casp2^−/− mice (n = 15). Data in (H) were pooled from 2 independent experiments. Error bars represent mean ± SD of n≥3 independent experiments. *, **, and *** represent p-values of <0.05, <0.001, and <0.0001 respectively. n.s. = not significant. UNT, ET, TM, and L+A represent untreated, etoposide (positive control for caspase-2 activation and Bid truncation, 25 μM), tunicamycin (positive control for ER stress induction, 10 μg/mL), and LPS+ATP (positive control for inflammasome activation, 200 ng/mL and 1mM respectively). Immunoblots in (A–D) are representative of n≥3 independent experiments that were performed and imaged in parallel with identical parameters using a LiCor Odyssey imaging system. Full-length caspase-2 and TOM20 (mitochondrial specific outer membrane protein) serve as loading controls. See also Supplemental Figure S4.

Figure 5. NLRP3 controls mitochondrial dysfunction by a Bid-dependent mechanism

Bid truncation in RB51-infected (A) with or without 4μ8c (IRE1 inhibitor, 50 μM) and (B) WT and Nlrp3^−/− BMDM – ET^╪ identifies duplicate lanes of the same sample. (C) Quantitative PCR of mtDNA in cytosolic extracts from RB51-infected WT and Bid^−/− BMDM. (D) ELISA of IL-1β in supernatants from RB51-infected WT and Bid^−/− BMDM. (E) Immunoblot analysis of caspase-1 in WT and Bid^−/− BMDM infected with RB51. (F) Serum IL-1β in WT (n = 29) and Bid^−/− (n = 30) RB51-infected (i.p. CFU 1 × 10⁸) mice. The data in (F) were pooled from 2 independent experiments. Error bars represent mean ± SD of n≥3 independent experiments. * and *** represent p-values of <0.05 and <0.0001 respectively. n.s. = not significant. UNT, ET, TM, and L+A represent untreated, etoposide (positive control for Bid truncation, 25 μM), tunicamycin (positive control for ER stress induction, 10 μg/mL), and LPS+ATP (positive control for inflammasome activation, 200 ng/mL and 1mM) respectively. Immunoblots in (A), (B), and (E) are representative of n≥3 independent experiments that were performed and imaged in parallel with identical parameters using a LiCor Odyssey imaging system. Full length (FL) caspase-1 and Bid serve as loading controls. See also Supplemental Figure S5.

Figure 6. NLRP3 and caspase-2 are required for caspase-1 activation during general ER stress

(A) ELISA analysis of IL-1β in supernatants from BMDM treated with 200 ng/mL LPS and 1mM ATP; positive control for inflammasome activation), or the ER stressors BFA (brefeldin A, 20 μM), TM (tunicamycin, 10 μg/mL), and TG (thapsigargin, 10 μM). Error bars represent mean ± SD of n≥3 independent experiments. ** and *** represent p-values of <0.001 and <0.0001 respectively. n.s. = not significant. Immunoblot analysis of caspase-1 in (B) BMDM in the absence or presence of 4μ8c (IRE1 inhibitor, 50 μM), (C) WT and Nlrp3^−/− BMDM – L+A^╪ identifies duplicate lanes of the same sample, (D) WT and Casp2^−/− BMDM, and (E) WT and Bid^{− /−} BMDM. Immunoblots in (B–E) are representative of n≥3 independent experiments that were performed and imaged in parallel with identical parameters using a LiCor Odyssey imaging system. Full-length (FL) caspase-1 serves as a loading control. See also Supplemental Figure S6.