PKM2-dependent glycolysis promotes NLRP3 and AIM2 inflammasome activation

Abstract

Sepsis, severe sepsis and septic shock are the main cause of mortality in non-cardiac intensive care units. Immunometabolism has been linked to sepsis; however, the precise mechanism by which metabolic reprogramming regulates the inflammatory response is unclear. Here we show that aerobic glycolysis contributes to sepsis by modulating inflammasome activation in macrophages. PKM2-mediated glycolysis promotes inflammasome activation by modulating EIF2AK2 phosphorylation in macrophages. Pharmacological and genetic inhibition of PKM2 or EIF2AK2 attenuates NLRP3 and AIM2 inflammasomes activation, and consequently suppresses the release of IL-1β, IL-18 and HMGB1 by macrophages. Pharmacological inhibition of the PKM2-EIF2AK2 pathway protects mice from lethal endotoxemia and polymicrobial sepsis. Moreover, conditional knockout of PKM2 in myeloid cells protects mice from septic death induced by NLRP3 and AIM2 inflammasome activation. These findings define an important role of PKM2 in immunometabolism and guide future development of therapeutic strategies to treat sepsis.

Figure 1. Pharmacological inhibition of PKM2 impairs inflammasome activation.

LPS (500 ng ml⁻¹, 3 h)-primed mouse BMDMs and human PMA-differentiated THP1 cells were respectively treated with inflammasome activators (ATP (5 mM, 30 min), poly(dA:dT) (1 μg ml⁻¹, 8 h), MDP (200 ng ml⁻¹, 8 h) or flagellin (200 ng ml⁻¹, 8 h)) in the absence or presence of shikonin at the same time (1 and 5 μM). IL-1β, IL-18 and HMGB1 (a,b) in supernatants and caspase-1 activity (c) in whole-cell extract were assayed with ELISA or activity assay kit (n=3, *P<0.05, ANOVA LSD test). In parallel, the levels of IL-1β and cleaved caspase-1 (p20) in culture supernatants (SN) and the precursors of IL-1β (pro-IL-1β) and caspase-1 (pro-caspase-1) in lysates of BMDMs were assayed using western blot (d). As a control, the cationic lipid transfection reagent LyoVec did not induce the cleavage of IL-1β and caspase-1 in BMDMs. All quantification data expressed as means±s.e.m of three independent experiments. Western blot data are representative of two independent experiments.

Figure 2. Genetic inhibition of PKM2 suppresses inflammasome activation.

(a) Western blot analysis of PKM2 expression in BMDMs and PMA-differentiated THP1 cells after knockdown of PKM2 by specific shRNA for 48 h. (b,c) LPS (500 ng ml⁻¹, 3 h)-primed BMDMs (b) and PMA-differentiated THP1 cells (c) were treated with various inflammasome activators (ATP (5 mM, 30 min), poly(dA:dT) (1 μg ml⁻¹, 8 h), MDP (200 ng ml⁻¹, 8 h) or flagellin (200 ng ml⁻¹, 8 h)). Extracellular levels of IL-1β, IL-18 and HMGB1 and cellular levels of caspase-1 were assayed using ELISA or activity assay kit (n=3, *P<0.05, ANOVA LSD test). (d,e) In parallel, TNF levels were assayed with ELISA (d); IL-1β and cleaved caspase-1 (p20) in culture supernatants (SN) and the precursors of IL-1β (pro-IL-1β) and caspase-1 (pro-caspase-1) in lysates of BMDMs were assayed using western blot (e). (f) Native gel electrophoresis was performed using whole-cell extracts from LPS treatment alone (500 ng ml⁻¹, 3 h) or LPS (500 ng ml⁻¹, 3 h)-primed BMDMs after treatment with ATP (5 mM, 30 min) or poly(dA:dT) (1 μg ml⁻¹, 8 h). (g) Shikonin (5 μM) did not increase PKM2-shRNA-mediated inhibition of IL-1β release in LPS (500 ng ml⁻¹, 3 h)-primed BMDMs following treatment with ATP (5 mM, 30 min). (h) Q-PCR analysis of gene expression in indicated BMDMs following treatment with LPS (500 ng ml⁻¹) for 3 h (n=3, *P<0.05, t-test). All quantification data expressed as means±s.e.m of three independent experiments. Western blot data are representative of two independent experiments.

Figure 3. PKM2-dependent glycolysis is required for inflammasome activation.

(a) LPS (500 ng ml⁻¹, 3 h)-primed BMDMs and PMs were treated with inflammasome activators (ATP (5 mM, 30 min) or poly(dA:dT) (1 μg ml⁻¹, 8 h)) in the absence or presence of shikonin (5 μM) or 2DG (2 mM). PEP and lactate levels were assayed using a commercial kit (n=3, *P<0.05). (b,c) Knockdown of PKM2 by shRNA prevented ATP (5 mM, 30 min) and poly(dA:dT) (1 μg ml⁻¹, 8 h)-induced increases in PEP and lactate levels (b), as well as mRNA expression of glycolysis-related genes (c) in LPS (500 ng ml⁻¹, 3 h)-primed BMDMs and PMs (n=3, *P<0.05 versus control shRNA group, ANOVA LSD test). In parallel, mRNA expression of indicated genes was assayed in BMDMs and PMs following treatment with LPS (500 ng ml⁻¹, 3 h) alone (c). (d–h) Knockdown of SLC2A1, LDHA or PDK1 by shRNA (d) inhibited ATP-(5 mM, 30 min) or poly(dA:dT) (1 μg ml⁻¹, 8 h)-induced lactate production (e), IL-1β (f) and HMGB1 (g) release and caspase-1 activation (h) in LPS (500 ng ml⁻¹, 3 h)-primed BMDMs (n=3, *P<0.05 versus control shRNA group, ANOVA LSD test). In parallel, mRNA expression of indicated genes was assayed in BMDMs following treatment with LPS (500 ng ml⁻¹, 3 h) alone (d). (i,j) Knockdown of PKM2, LDHA or SLC2A1 by shRNA increased mitochondrial membrane potential (ΔΨm) loss (i) and mitochondrial ROS production (MitoSOX) (j) with or without ATP (5 mM, 30 min) and poly(dA:dT) (1 μg ml⁻¹, 8 h) treatment in LPS (500 ng ml⁻¹, 3 h)-primed BMDMs (n=3, *P<0.05 versus control shRNA group, ANOVA LSD test). (k) Western blot analysis of MAP1LC3B-II and SQSTM1 in indicated LPS-primed BMDMs after treatment with ATP (5 mM, 30 min) and poly(dA:dT) (1 μg ml⁻¹, 8 h). All quantification data expressed as means±s.e.m of three independent experiments. Western blot data are representative of two independent experiments.

Figure 4. PKM2 contributes to glycolysis in caspase-1-deficient macrophages.

(a) LPS (500 ng ml⁻¹, 3 h)-primed WT (caspase-1^+/+) and caspase-1^−/− BMDMs were treated with inflammasome activators (ATP (5 mM, 30 min), poly(dA:dT) (1 μg ml⁻¹, 8 h), MDP (200 ng ml⁻¹, 8 h) or flagellin (200 ng ml⁻¹, 8 h)). The levels of PEP and lactate were assayed using a commercial kit (n=3, *P<0.05, ANOVA LSD test). (b) Inhibition of PKM2 using shikonin (5 μM) or shRNA significantly inhibited ATP- (5 mM, 30 min) or poly(dA:dT) (1 μg ml⁻¹, 8 h)-induced PEP and lactate production in LPS (500 ng ml⁻¹, 3 h)-primed caspase-1^−/− BMDMs (n=3, *P<0.05, ANOVA LSD test). (c) In parallel, ATP-induced interaction between NLRP3 and PYCARD and poly(dA:dT)-induced interaction between AIM2 and PYCARD were assayed with immunoprecipitation (IP). (d,e) Knockdown of PKM2 did not increase LDH release and ΔΨm loss in LPS (500 ng ml⁻¹, 3 h)-primed caspase-1^−/− BMDMs following treatment with ATP (5 mM, 30 min) or poly(dA:dT) (1 μg ml⁻¹, 8 h) (n=3, *P<0.05, ANOVA LSD test). All quantification data expressed as means±s.e.m of three independent experiments. Western blot data are representative of two independent experiments.

Figure 5. PKM2-dependent glycolysis promotes EIF2AK2 phosphorylation.

(a) LPS-primed BMDMs were treated with ATP (5 mM, 30 min) in the absence or presence of shikonin (1 and 5 μM). p-EIF2AK2 was assayed (n=3, *P<0.05 versus ATP group, ANOVA LSD test). (b,c) Knockdown of PKM2 (b) and PDK1 (c) by shRNA inhibited ATP- (5 mM, 30 min) and poly(dA:dT) (1 μg ml⁻¹, 8 h)-induced p-EIF2AK2 in LPS (500 ng ml⁻¹, 3 h)-primed BMDMs (n=3, *P<0.05 versus control shRNA group, ANOVA LSD test). (d) Western blot analysis of IL-1β in culture supernatants (SN) and the precursors of IL-1β (pro-IL-1β) in lysates of LPS-primed BMDMs following treatment with lactate (5 mM, 3 h) and ATP (5 mM, 30 min). (e) LPS-primed BMDMs were treated with lactate (5 mM, 3 h) in the absence or presence of C16 (1 and 5 μM). IL-1β and HMGB1 release and p-EIF2AK2 expression were assayed (n=3, *P<0.05 versus ATP group, ANOVA LSD test). (f) Knockdown of EIF2AK2 by shRNA inhibited lactate (5 mM, 3 h)-induced IL-1β, IL-18 and HMGB1 release and caspase-1/11 activation in LPS-primed BMDMs (n=3, *P<0.05 versus control shRNA group, ANOVA LSD test). (g) Inhibition of EIF2AK2 by C16 or shRNA or knockout of caspase-1 limited MDP (200 ng ml⁻¹, 8 h) or flagellin (200 ng ml⁻¹, 8 h))-induced cytokine release in LPS-primed BMDMs (n=3, *P<0.05, ANOVA LSD test). (h) Double knockout of caspase-1 and caspase-11 inhibited lactate (5 mM, 3 h)-induced IL-1β release in LPS-primed BMDMs (n=3, *P<0.05, ANOVA LSD test). (i) A-438079 (1 μm) suppressed ATP (5 mM, 30 min), but not lactate (5 mM, 3 h)-induced p-EIF2AK2 expression in LPS-primed BMDMs. (j) Knockout of NLRP3 and AIM2 inhibited lactate (5 mM, 3 h)-induced IL-1β and HMGB1 release and caspase-1/11 activation in LPS-primed BMDMs (n=3, *P<0.05, ANOVA LSD test). All quantification data expressed as means±s.e.m of three independent experiments. Western blot data are representative of two independent experiments.

Figure 6. Pharmacological inhibition of the PKM2 pathway protects septic mice.

(a,b) p-EIF2AK2, EIF2AK2 and caspase-1 activity were assayed in isolated PMs from mice during endotoxemia or polymicrobial sepsis in the absence or presence of shikonin (8 mg kg⁻¹) or C16 (50 μg kg⁻¹). In addition, the protein levels of p-EIF2AK2 and EIF2AK2 were assayed in PMs from mice with vehicle (no LPS) injection or sham operated for CLP. (c) Mice (n=20 mice per group) were injected with a single dose of C16 (8 mg kg⁻¹), followed 30 min later by an infusion of endotoxin (LPS, 5 mg kg⁻¹, intraperitoneally), and were then re-treated with C16 12 and 24 h later. The Kaplan–Meyer method was used to compare differences in survival rates between groups (*P<0.05). (d,e) In parallel experiments, serum levels of IL-1β and HMGB1 at indicated time points were measured (n=3 animals per group, *P<0.05, t test). (f) CLP was used to induce intraabdominal sepsis in mice (n=20 group per group). Repeated administration of C16 (50 μg kg⁻¹) at 24, 48 and 72 h after CLP significantly increased survival compared with vehicle group (*P<0.05), as measured by Kaplan–Meyer test. (g,h) In parallel, the serum levels of IL-1β and HMGB1 at indicated time points were measured using ELISA (n=3 animals per group, *P<0.05, t-test). All quantification data expressed as means±s.e.m of three independent experiments. Western blot data are representative of two animals per group.

Figure 7. Conditional knockout of PKM2 in myeloid cells protects septic mice.

(a) Western blot analysis of expression of indicated proteins in BMDMs or lung isolated from myeloid cell-specific PKM2-knockout mice (PKM2^−/−) and control WT mice (PKM2^+/+). (b) Indicated mice (n=10 mice per group) were pre-injected with LPS (2 mg kg⁻¹, intraperitoneally) for 3 h and then challenged with NLRP3 activator ATP (200 mg kg⁻¹, intraperitoneally) or AIM2 activator nucleosome (20 mg kg⁻¹, intraperitoneally). Injection with LPS (2 mg kg⁻¹, intraperitoneally) alone in these mice was used as a control (n=10 mice per group). The Kaplan–Meyer method was used to compare differences in survival rates between groups (*P<0.05). (c,d) In parallel experiments, serum levels of IL-1β and HMGB1 at indicated time points were measured (n=3 animals per group, *P<0.05, ANOVA LSD test). (e) Schematic depicting PKM2-mediated glycolysis promoting NLRP3 and AIM2 inflammasome activation and proinflammatory cytokine (for example, IL-1β, IL-18 and HMGB1) release by modulating EIF2AK2 phosphorylation. All quantification data expressed as means±s.e.m of three independent experiments. Western blot data are representative of two independent experiments.