New mitochondrial DNA synthesis enables NLRP3 inflammasome activation

Abstract

Dysregulated NLRP3 inflammasome activity results in uncontrolled inflammation, which underlies many chronic diseases. Although mitochondrial damage is needed for the assembly and activation of the NLRP3 inflammasome, it is unclear how macrophages are able to respond to structurally diverse inflammasome-activating stimuli. Here we show that the synthesis of mitochondrial DNA (mtDNA), induced after the engagement of Toll-like receptors, is crucial for NLRP3 signalling. Toll-like receptors signal via the MyD88 and TRIF adaptors to trigger IRF1-dependent transcription of CMPK2, a rate-limiting enzyme that supplies deoxyribonucleotides for mtDNA synthesis. CMPK2-dependent mtDNA synthesis is necessary for the production of oxidized mtDNA fragments after exposure to NLRP3 activators. Cytosolic oxidized mtDNA associates with the NLRP3 inflammasome complex and is required for its activation. The dependence on CMPK2 catalytic activity provides opportunities for more effective control of NLRP3 inflammasome-associated diseases.

Figures

Extended Data Fig. 1 ~. TFAM is required for ox-mtDNA generation and NLRP3 inflammasome activation.

a, Inflammasome activator-induced changes in mitochondrial membrane potential (Ψm) in non-or LPS-primed wild-type BMDMs were measured by TMRM fluorescence. Data are mean ± s.d. (n = 3 biological replicates). b, Relative mtROS amounts were measured by MitoSOX fluorescence in non-or LPS-primed wild-type BMDMs after stimulation with different inflammasome activators. Data are mean ± s.d. (n = 3 biological replicates). c, Amounts of 8-OH-dG in mtDNA isolated from the mitochondrial fraction of non-or LPS-primed wild-type BMDMs that were treated with different inflammasome activators. Data are mean ± s.d. (n = 3 biological replicates). d, Cytosolic release of mtDNA, determined by qPCR with primers specific for mtDNA (non-NUMT) and nDNA (B2m), in non-or LPS-primed wild-type BMDMs after treatment with different inflammasome activators. Data are mean ± s.d. (n = 4 biological replicates). e, Immunoblot analysis of TFAM in Tfamff/f and TfamΔMye BMDMs. Results are typical of three independent experiments. f, Relative total mtDNA amounts in Tfamf/f and TfamΔMye BMDMs determined by qPCR with primers specific for mtDNA (D-loop, non-NUMT) and nDNA (Tert, B2m). Data are mean ± s.d. (n = 3 biological replicates). g, Relative mtROS amounts were measured by MitoSOX fluorescence in LPS-primed Tfamf/f and TfamΔMye BMDMs after stimulation with indicated NLRP3 activators. Data are mean ± s.d. (n = 3 biological replicates). h, Amounts of 8-OH-dG in mtDNA isolated from the mitochondrial fraction of LPS-primed Tfamff/f and TfamΔMye BMDMs that were stimulated with various NLRP3 activators. Data are mean ± s.d. (n = 3 biological replicates). i, Immunoblot analysis of Casp1 p20 and mature IL-1β (p17) in culture supernatants of Tfamf/f (W) and TfamΔMye (K) BMDMs that were stimulated with LPS plus different inflammasome activators. Results are typical of three separate experiments. j, Immunoblot analysis of pro-IL-1β, NLRP3, ASC and pro-Casp1 in the lysates of Tfamff/f and TfamΔMye BMDMs before and after LPS priming. Results are typical of three separate experiments. k, l, Amounts of IL-1β (k) and TNF (l) in culture supernatants of LPS-primed Tfamf/f and TfamΔMye BMDMs that were stimulated with various NLRP3 activators. Data are mean ± s.d. (n = 3 biological replicates). m, Amounts of IL-1β in culture supernatants of LPS-primed Tfamf/f and TfamΔMye BMDMs that were stimulated with H2O2 in the absence and presence of nigericin. Data are mean ± s.d. (n = 3 biological replicates). **P < 0.01; ***P < 0.001; two-sided unpaired t-test.

Extended Data Fig. 2 ~. ox-mtDNA activates the NLRP3 inflammasome.

a, Immunoblot analysis of NLRP3, AIM2 and Polγ in shCtrl-or specific shRNA-transduced BMDMs. Data are typical of three separate experiments. b, Amounts of IL-1β in culture supernatants of LPS-primed TfamΔMye BMDMs that were transfected with mtDNA, methylated mtDNA, ox-mtDNA, and methylated ox-mtDNA. All mtDNAs were 90-bp long and of an identical sequence. Data are mean ± s.d. (n = 3 biological replicates). c, Representative fluorescent microscopy images of wild-type BMDMs that were co-stained for 8-OH-dG, ASC and DAPI before and after stimulation with LPS plus the indicated inflammasome activators. Results are typical of three independent experiments. Scale bars, 5 μm.

Extended Data Fig. 3 ~. LPS induces mtDNA replication.

a, Time-course analysis of total mtDNA amounts in wild-type BMDMs after LPS (200 ng ml−1) stimulation. Data are mean ± s.d. (n = 3 biological replicates). Results are typical of three separate experiments. b, Representative fluorescent microscopy images of EdU-labelled shCtrl-or shPolg-transduced wild-type BMDMs that were stimulated with or without LPS (200 ng ml−1) for 6 h. Scale bars, 5 μm. Results are typical of three separate experiments. c, Percentages of cells with mtDNA replication as determined in b. Data are mean ± s.d. (n = 3 different microscopic fields per group; original magnification, ×40). d, Immunoblot analysis of mitochondrial resident proteins, TOM20 and VDAC, in wild-type BMDMs before and after LPS stimulation. Results are typical of three separate experiments. e, Representative fluorescent microscopy images of the mitochondrial resident protein ATP5B and DAPI staining in wild-type BMDMs before and after LPS stimulation. Results are typical of three independent experiments. Arrows indicate fragmented mitochondria. ***P < 0.001, two-sided unpaired t-test.

Extended Data Fig. 4 ~. MyD88 and TRIF mediate LPS-induced IRF1 and CMPK2 expression and NLRP3 inflammasome activation.

a, Amounts of 8-OH-dG in mtDNA isolated from mitochondrial fractions of wild-type, Myd88−/− and Trif−/− BMDMs that were primed with LPS for different durations followed by stimulation with ATP. Data are mean ± s.d. (n = 3 biological replicates per time point). b, c, Quantification of fluorescent microscopy images of ASC puncta (b) and ASC puncta positive for EdU (c) in wild-type, Myd88−/− and Trif−/− BMDMs that were primed with LPS for different durations, followed by ATP stimulation. Data are mean ± s.d. (n = 3 different microscopic fields per group; original magnification, ×40). d–f, Immunoblot analysis of NLRP3, pro-IL-1β, IRF1 and CMPK2 in wild-type, Myd88−/− and Trif−/− BMDMs before and after LPS stimulation. Data are typical of three separate experiments. *P < 0.05; **P < 0.01; ***P < 0.001; two-sided unpaired t-test.

Extended Data Fig.5 ~. Effects of IRF1 on priming, mitochondrial damage and non-canonical inflammasome activation.

a, Immunoblot analysis of NLRP3, ASC, pro-caspase-1, pro-IL-1β, NEK7 and IRF1 in the lysates of wild-type and Irf1−/− BMDMs before and after LPS stimulation. Data are typical of three separate experiments. b, Amounts of TNF in culture supernatants of LPS-primed wild-type and Irf1−/− BMDMs that were stimulated with the indicated inflammasome activators. Data are mean ± s.d. (n = 3 biological replicates). c, Percentages of Ψm preservation were measured by TMRM fluorescence in LPS-primed wild-type and Irf1−/− BMDMs that were stimulated with indicated NLRP3 activators. Data are mean ± s.d. (n = 3 biological replicates). d, Relative amounts of mtROS were measured by MitoSOX fluorescence in LPS-primed wild-type and Irf1−/− BMDMs that were stimulated with indicated NLRP3 activators. Data are mean ± s.d. (n = 3 biological replicates). e, Immunoblot analysis of caspase-11 and IRF1 in lysates from wild-type and Irf1−/− BMDMs before and after LPS stimulation. Results are typical of three independent experiments. f, Amounts of IL-1β in culture supernatants of LPS-primed wild-type and Irf1−/− BMDMs that were further transfected with FuGENE-complexed LPS. Data are mean ± s.d. (n = 3 biological replicates).

Extended Data Fig. 6 ~. IRF1 mediates LPS-induced CMPK2 expression.

a, Relative amounts of Cmpk2 mRNA in wild-type and Irf1−/− BMDMs before and after LPS stimulation. Data are mean ± s.d. (n = 3 biological replicates per time point). b, Immunoblot analysis of CMPK2, VDAC and tubulin in mitochondrial and cytosolic fractions of wild-type BMDMs after LPS stimulation. Results are typical of three separate experiments. c, Chromatin immunoprecipitation analysis of IRF1 recruitment to the Cmpk2 promoter. Data are mean ± s.d. (n = 4 biological replicates for wild-type and Irf1−/− groups; n = 5 and 6 biological replicates for wild-type + LPS and Irf1−/− + LPS groups, respectively). Cxcl10, a known IRF1-target gene, was included as a positive control. d, Relative mRNA amounts of dGK (also known as Dguok), Tk2, Ak2, Nme4 and Polg in wild-type BMDMs before and after 6 h LPS stimulation. Data are mean ± s.d. (n = 3 biological replicates). e, Immunoblot analysis of the enzymes encoded by the genes in d in the lysates of wild-type BMDMs before and after LPS stimulation. Results are typical of three independent experiments. f, Immunoblot analysis of CMPK2 and NME4 in shCtrl-or specific shRNA-transduced BMDMs. Data are typical of three independent experiments. *P < 0.05; **P < 0.01; two-sided unpaired t-test.

Extended Data Fig. 7 ~. CMPK2 deficiency does not affect inflammasome subunit expression nor NLRP3 activator-induced mitochondrial damage.

a, Immunoblot analysis of pro-IL-1β, NLRP3, ASC, pro-Casp1 and NEK7 in the lysates of wild-type (shCtrl) and CMPK2-deficient (shCmpk2) BMDMs before and after LPS priming. Results are typical of three separate experiments. b, NLRP3 activator-induced changes in Ψm in LPS-primed shCtrl and shCmpk2 BMDMs were measured by TMRM fluorescence. Data are mean ± s.d. (n = 3 biological replicates). c, Relative amounts of mtROS measured by MitoSOX fluorescence in LPS-primed shCtrl and shCmpk2 BMDMs after stimulation with the indicated NLRP3 activators. Data are mean ± s.d. (n = 3 biological replicates). d, Representative fluorescent microscopy images of EdU-labelled wild-type BMDMs transduced with either shCtrl-or shCmpk2-encoding lentiviruses that were stimulated with or without LPS (200 ng ml−1) for 6 h. Scale bars, 5 μm. e, Percentages of cells with mtDNA replication as determined in d. Data are mean ± s.d. (n = 3 different microscopic fields per group; original magnification, ×40). f, Relative amounts of total mtDNA in wild-type and Irf1−/− BMDMs before and after treatments with the indicated TLR agonists. Data are mean ± s.d. (n = 3 biological replicates). g, Relative amounts of total mtDNA in shCtrl-or shCmpk2-encoding lentivirus-transduced wild-type BMDMs before and after treatments with the indicated TLR agonists. Data are mean ± s.d. (n = 3 biological replicates). *P < 0.05; **P < 0.01; ***P < 0.001; two-sided unpaired t-test.

Extended Data Fig. 8 ~. dNTP availability controls LPS-induced mtDNA synthesis and NLRP3 inflammasome activation.

a, Relative total mtDNA amounts in shCtrl and shNme4 BMDMs before and after LPS priming. Data are mean ± s.d. (n = 3 biological replicates). b, Amounts of 8-OH-dG in mtDNA isolated from the mitochondrial fraction of LPS-primed shCtrl and shNme4 BMDMs that were stimulated with various NLRP3 activators. Data are mean ± s.d. (n = 3 biological replicates). c, Amounts of 8-OH-dG in cytosolic mtDNA from LPS-primed shCtrl and shNme4 BMDMs that were stimulated with various NLRP3 activators. Data are mean ± s.d. (n = 3 biological replicates). d, e, Amounts of IL-1β (d) and TNF (e) in supernatants of LPS-primed shCtrl and shNme4 BMDMs that were stimulated with various inflammasome activators. Data are mean ± s.d. (n = 3 biological replicates). f, Immunoblot analysis of SAMHD1 and Polγ in wild-type and Trif−/− BMDMs that were stimulated with LPS for different durations as indicated. Results are typical of three independent experiments. g, Relative total mtDNA amounts in wild-type and Samhd1−/− BMDMs before and after LPS stimulation. Data are mean ± s.d. (n = 3 biological replicates). h, i, Amounts of IL-1β (h) and TNF (i) in the culture supernatants of LPS-primed wild-type and Samhd1−/− BMDMs that were stimulated with inflammasome activators as indicated. Data are mean ± s.d. (n = 3 biological replicates). j, Amounts of NLRP3 activator-induced IL-1β in culture supernatants of LPS-primed wild-type and Samhd1−/− BMDMs with or without Polgexpression. Data are mean ± s.d. (n = 3 biological replicates). *P < 0.05; **P < 0.01; ***P < 0.001; two-sided unpaired t-test.

Extended Data Fig. 9 ~. CMPK2 expression restores NLRP3 inflammasome activation in IRF1-deficient macrophages.

a, Immunoblot analysis of IRF1, CMPK2, pro-IL-1β, NLRP3, ASC and pro-Casp1 in lysates of wild-type and Irf1−/− BMDMs before and after transduction with a wild-type CMPK2-encoding lentivirus. Results are typical of three independent experiments. b, NLRP3 activator-induced changes in Ψm in LPS-primed CMPK2-transduced wild-type and Irf1−/− BMDMs were measured by TMRM fluorescence. Data are mean ± s.d. (n = 3 biological replicates) and analysed by two-sided unpaired t-test (not significant). c, Relative amounts of mtROS measured by MitoSOX fluorescence in LPS-primed control (vector)-or CMPK2-transduced wild-type and Irf1−/− BMDMs before and after stimulation with NLRP3 activators. Data are mean ± s.d. (n = 3 biological replicates) and analysed by two-sided unpaired t-test (not significant). d, Immunoblot analysis of CMPK2 in Irf1−/− BMDMs that were transduced with wild-type or mutant (CMPK2(D330A)) CMPK2-encoding lentiviruses. Results are typical of three separate experiments.

Extended Data Fig. 10 ~. IRF1 is required for in vivo mtDNA replication and NLRP3 inflammasome activation.

a, b, 12-week-old wild-type or Irf1−/− mice were injected intraperitoneally with LPS (50 mg per kg of body weight) and their sera were collected 3 h later and analysed by ELISA for IL-1β (a) and TNF (b). Results are mean ± s.d. (n = 6 and 7 for WT and Irf1−/− mice, respectively). c, Survival of wild-type or Irf1−/− mice that were injected intraperitoneally with LPS (50 mg per kg body weight; n = 10 and 11 for WT and Irf1−/− mice, respectively). d, Relative amounts of total mtDNA in peritoneal infiltrates of wild-type or Irf1−/− mice before and after LPS (50 mg per kg body weight) injection. Data are mean ± s.d. (n = 6 in PBS-treated groups; n = 12 in LPS-treated groups). e, Peritoneal IL-1β in wild-type or Irf1−/− mice 4 h after intraperitoneal injection of alum (1 mg) or PBS. Data are mean ± s.d. (n = 5 in PBS-treated groups; n = 6 in alum-treated groups). f, g, Alum-induced peritoneal infiltration of neutrophils (CD11b+Ly6G+F4/80−) (f) and monocytes (CD11b+Ly6C+Ly6G−) (g) in wild-type and Irf1−/− mice 12 h after alum (1 mg) or PBS injection. Data are mean ± s.e.m. (n = 3 for PBS-treated groups and n = 6 for alum-treated groups). ***P < 0.001; two-sided unpaired t-test (a, b, d–g) and log-rank test (c). h, A working model to illustrate how TLR-mediated priming controls mtDNA replication and NLRP3 inflammasome activation. Whereas IRF1 acts positively to induce the transcription of CMPK2, which supplies rate-limiting dCDP for mtDNA synthesis, TRIF-dependent signalling also acts negatively to limit dNTP supply through the induction of SAMHD1.

Fig. 1 ~. Newly synthesized mtDNA is required for NLRP3 inflammasome activation.

a, TfamΔMye BMDMs transduced with shRNA against Nlrp3, Aim2 (shNlrp3, shAim2) or control shRNA (shCtrl) and primed with LPS were incubated with synthetic mtDNA, ox-mtDNA, nuclear DNA (nDNA) or oxidized nuclear DNA (ox-nDNA), and the release of IL-1β (left) and TNF (right) was measured 4 h later. b, Relative total mtDNA amounts were quantified by quantitative PCR (qPCR) with primers specific for the mitochondrial D-loop region or a region of mtDNA that is not inserted into nuclear DNA (non-NUMT) and primers specific for nDNA (Tert, B2m) in wild-type BMDMs before and after LPS priming. c, Fluorescent microscopy of EdU-labelled newly synthesized mtDNA in wild-type BMDMs before and after LPS priming. Scale bars, 5 μm. Images are representative of three independent experiments. d, Relative total mtDNA amounts in wild-type BMDMs transduced with Polg shRNA (shPolg#1 and shPolg#2) or control shRNA, before and after LPS priming. e, IL-1β (left) and TNF (right) release by shCtrl-or shPolg-transduced LPS-primed BMDMs treated with different inflammasome activators. f, Relative total mtDNA amounts in wild-type (WT), Myd88−/− and Trif−/− (Trif is also known as Ticam 1) BMDMs after LPS priming. Data in a, b, d–f are mean ± s.d. (n = 3 biological replicates). *P < 0.05; **P < 0.01; ***P < 0.001; two-sided unpaired t-test. NS, not significant.

Fig. 2 ~. IRF1-dependent mtDNA synthesis, ox-mtDNA generation and NLRP3 inflammasome activation.

a, Relative total mtDNA amounts in wild-type and Irf1−/− BMDMs that were stimulated with LPS. b, Representative images showing EdU incorporation into mtDNA in wild-type and Irf1−/− BMDMs incubated without or with LPS for 6 h. Scale bars, 5 μm. c, Percentages of cells undergoing mtDNA synthesis, as determined in b (n = 3 different microscopic fields per group; original magnification, ×40). d, IL-1β release by LPS-primed wild-type and Irf1 BMDMs that were treated with different inflammasome activators. e, Immunoblot analysis of cleaved caspase-1 (Casp1 p20) in culture supernatants of wild-type and Irf1−/− BMDMs that were treated with LPS plus different inflammasome activators. Data are representative of three independent experiments. exp., exposure; K, knockout (Irf1−/−); W, wild type. f, g, Amounts of 8-OH-dG in mtDNA within mitochondria (f) or in the cytosol (g) of LPS-primed wild-type and Irf1−/− BMDMs treated with NLRP3 activators. Data in a, d, f and g are mean ± s.d. (n = 3 biological replicates). **P < 0.01; ***P < 0.001; two-sided unpaired t-test.

Fig. 3 ~. CMPK2 controls mtDNA synthesis, ox-mtDNA generation and NLRP3 inflammasome activation.

a, Time-course analysis of CMPK2 accumulation in wild-type and Irf1−/− BMDMs before and after LPS priming. b, Immunoblot analysis of cleaved caspase-1 (Casp1 p20) and mature IL-1β (p17) in the supernatants of shCtrl (wild-type, W) and shCmpk2 (knockdown, K) BMDMs that were stimulated with LPS plus the indicated inflammasome activators. Data in a and b represent three independent experiments. c, d, IL-1β (c) and TNF (d) secretion by LPS-primed BMDMs transduced with control shRNA or Cmpk2 shRNA (shCmpk2#1 and shCmpk2#2), and incubated with different inflammasome activators. e, Relative total mtDNA amounts in control and Cmpk2 shRNA BMDMs before and after LPS stimulation. f, g, Amounts of 8-OH-dG in mtDNA within mitochondria (f) or in the cytosol (g) of LPS-primed BMDMs transduced with control shRNA or Cmpk2 shRNA that were treated with different NLRP3 activators. Data in c–g are mean ± s.d. (n = 3 biological replicates, except for n = 4 in e). **P < 0.01; ***P < 0.001; two-sided unpaired t-test.

Fig. 4 ~. CMPK2 catalytic activity is required for NLRP3 inflammasome activation.

a, Relative total mtDNA amounts in LPS-treated wild-type and Irf1−/− BMDMs transduced with CMPK2-encoding or empty lentiviruses. b, Top, immunoblot analysis of Casp1 p20 in the supernatants of CMPK2 lentivirus-transduced LPS-primed wild-type and Irf1−/− BMDMs that were stimulated with the indicated NLRP3 activators. Data are representative of three independent experiments. Bottom, IL-1β and TNF secretion by the above cells. c, Relative total mtDNA amounts in LPS-treated wild-type and Irf1−/− BMDMs transduced with either wild-type CMPK2-or CMPK2(D330A)-encoding lentiviruses. d, Top, immunoblot analysis of Casp1 p20 in supernatants of wild-type CMPK2-or CMPK2(D330A)-encoding lentiviruses transduced LPS-primed wild-type and Irf1−/− BMDMs that were stimulated with NLRP3 activators. Data are representative of three independent experiments. Bottom, IL-1β and TNF secretion by the above cells. Data in a, b (bottom), c and d (bottom) are mean ± s.d. (n = 3 biological replicates). **P < 0.01; ***P < 0.001; two-sided unpaired t-test.

Fig. 5 ~. Newly synthesized mtDNA associates with the NLRP3 inflammasome complex.

a, Inflammasomes from BMDMs after indicated treatments were immunoprecipitated with ASC antibodies. The immunocomplexes were spotted on a nitrocellulose membrane, UV-crosslinked and probed with antibodies to 8-OH-dG and BrdU, or separated by SDS-PAGE and immunoblotted with antibodies to pro-Casp1, NLRP3, AIM2 and ASC. Data are representative of three independent experiments. b, Representative fluorescent microscopy images of EdU-labelled BMDMs that were co-stained for ASC, EdU and DAPI before and after stimulation with LPS plus different inflammasome activators. Arrows indicate co-localization of EdU and ASC signals. Scale bars, 5 μm. c, Percentages of cells with ASC and EdU co-localization as determined in b. Data are mean ± s.d. (n = 3 different microscopic fields per group; original magnification, ×40). d, Relative D-loop (left) and Cox1 (right) mtDNA amounts in inflammasome complexes isolated as in a. Data are mean ± s.d. (n = 3 biological replicates). **P < 0.01; ***P < 0.001; two-sided unpaired t-test.