DMEM, Opti-MEM, l-glutamine, streptomycin, penicillin, and fetal bovine serum (FBS) were obtained from Thermo/Fisher/Gibco (Carlsbad, CA, USA). LPS (Escherichia coli O111:B4) (L4391), ATP (P8232), dimethyl sulfoxide, Hoechst 33342, propidium iodide (PI), Tween-80, and compound C were bought from Sigma-Aldrich (St. Louis, MO, USA). Metformin was obtained from MedChem Express (Princeton, NJ, USA), dissolved in PBS at 300 mM and stored at −20°C. The working solution of metformin was freshly prepared. Antibodies against IL-1β (#12242), HMGB1 (#3935), AMPKα (#5832), phospho(p)-AMPK (Thr172) (#2535), β-tubulin (#2128), ASC (#67824), and horseradish peroxidase-conjugated goat anti-mouse/rabbit IgG were bought from Cell Signaling Technology (Danvers, MA, USA). The antibody to NLRP3 (AG-20B-0014) was purchased from Adipogen AG (Liestal, Switzerland). The antibody against caspase-1p10 (sc-514) was obtained from Santa Cruz (Santa Cruz Biotechnology, Dallas, TX, USA). Anti-mouse CD11b FITC (11-0112) and anti-mouse Ly-6G (Gr-1) PE (12-9668) were obtained from eBioscience (San Diego, CA, USA). C57BL/6 mice (6–8 weeks of age) were purchased from the Laboratory Animal Center of Southern Medical University (Guangzhou, China). All animal experiments were performed according to the guidelines for the care and use of animals approved by the Committee on the Ethics of Animal Experiments of Jinan University.

Mouse macrophage cell line J774A.1 was obtained from the Kunming Cell Bank of Type Culture Collection Chinese Academy of Sciences (Kunming, China). The cells were maintained in complete DMEM medium (containing 10% FBS, 100 µg/ml penicillin, 100 µg/ml streptomycin, and 2 mM l-glutamine) and cultured at 37°C in a humidified incubator with 5% CO₂. The cells were subcultured every 2–3 days.

Bone marrow-derived macrophages from C57BL/6 mice were differentiated as reported previously (3). Briefly, mouse bone marrow cells were differentiated in DMEM supplemented with 10% FBS and 20% M-CSF-conditioned medium from L929 cells for 6 days. The BMDMs were then cultured overnight in 24-well plates at 1.8 × 10⁵ cells/well in 0.5 ml medium.

Soluble IL-1β in cell culture medium and orbital serum were determined by cytometric bead array (CBA) mouse IL-1β flex set (#560232) according to the manufacturer’s instructions (BD Biosciences, San Jose, CA, USA).

Cells were washed twice with PBS-F (PBS containing 0.1% NaN₃ and 3% FBS), then stained with FITC-labeled anti-CD11b and PE-labeled anti-mouse Ly-6G monoclonal antibodies for 30 min. Red blood cells, if there were, were lysed. After washing with PBS-F, the cells were fixed with 4% paraformaldehyde in PBS and then analyzed by a flow cytometer (FACSCalibur; Becton Dickinson).

Cell pyroptosis was measured as described previously (24, 25). Briefly, cells were seeded in 24-well plates and cultured for 24 h. The cells were then primed with 500 ng/ml LPS for 4 h followed by treatment with indicated concentrations of metformin in Opti-MEM for 1 h. Next, indicated concentrations of ATP were added to the culture medium for indicated time periods. Subsequently, the cells were stained with staining solution containing PI (2 µg/ml) and Hoechst 33342 (5 µg/ml). Dead cells (PI permeable) were observed under a Zeiss Axio Observer D1 microscope and fluorescence images were captured by the Zeiss AxioCam MR R3 cooled CCD camera controlled with ZEN software (ZEISS, Germany).

Precipitation of soluble proteins was performed as previously described (26). In brief, proteins existing in cell culture supernatants were precipitated overnight with 7.2% trichloroacetic acid and 0.15% sodium deoxycholate solutions. After washing three times with cold acetone, the precipitated proteins were redissolved in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer and subjected to western blot analysis.

Whole cell lysates and western blotting were performed as previously described (27). Cells were lysed with 2× SDS-PAGE loading buffer, and the total proteins were separated by SDS-PAGE and then transferred onto polyvinylidene difluoride membranes (Hybond-P; GE Healthcare Life Sciences, Piscataway, NJ, USA). The membrane was blocked and incubated with primary antibody overnight, followed by incubation with horseradish peroxidase-conjugated goat anti-rabbit IgG or goat anti-mouse IgG. Bands on the membrane were revealed by a BeyoECL Plus kit (Beyotime, P0018) and recorded on X-ray films (Kodak, 6535873). Images were captured by using FluorChem 8000 Imaging System (AlphaInnotech, San Leandro, CA, USA), and densitometry of each band was quantified using AlphaEaseFC software (AlphaInnotech).

Bone marrow-derived macrophages were planted in glass-bottomed dishes (5 × 10⁵ cells/dish) and cultured at 37°C overnight. Then the cells were primed with 500 ng/ml LPS for 4 h, followed by treatment with 2 mM metformin or vehicle (PBS) for 1 h. Subsequently, 2 mM ATP was added to the culture medium for 30 min. After these treatments, the cells were fixed in 4% paraformaldehyde for 15 min and permeabilized with 2 ml cold methanol (−20°C) for 10 min. Then the cells were incubated with ASC antibody (1:300) overnight, followed by staining with CF568-conjugated goat-anti-rabbit IgG (#20103) (Biotium, Hayward, CA, USA) for 1 h. Finally, Hoechst 33342 solution (5 µg/ml in PBS) was added to stain the nuclei for 10 min. The cells were observed under a Zeiss Axio Observer D1 microscope with a Zeiss LD Plan-Neofluar 40×/0.6 Korr M27 objective (Carl Zeiss MicroImaging GmbH, Göttingen, Germany). Fluorescence images were captured with a Zeiss AxioCam MR R3 cooled CCD camera controlled with ZEN software (ZEISS).

The siRNA (5′-GCA GAA GUG UGU AGA GCA A-3′) duplexes targeting mouse AMPKα1 and negative control (NC) siRNA were designed and synthesized by RiboBio (Guangzhou, China). Knockdown of AMPKα1 was performed according to the instructions provided by the manufacturers. Briefly, transfection reagent Lipofectamine RNAiMAX (Invitrogen, 13778-075) and AMPKα1 siRNA were diluted in Opti-MEM, respectively, and then mixed together. The NC siRNA mixture was also prepared. Then the cell culture medium was replaced with Opti-MEM, and the siRNA mixture was added at a final siRNA concentration of 100 nM. Six hours later, the cells were cultured in normal medium containing 10% FBS. After being cultured for additional 72 h, the cells were ready for experiments.

Bacterial sepsis mouse model was established as previously described (9, 27). Briefly, C57BL/6 mice were acclimated for 1 week. Bacteria (E. coli DH5α) were cultured and proliferated in Luria Broth medium. Then the viable bacteria were resuspended in PBS at 2 × 10⁹ CFU/ml. Each mouse was intraperitoneally injected with 0.5 ml bacterial suspension. Metformin (dissolved in PBS) or vehicle (PBS) was given intragastrically 1 h later at a dose of 250 mg/kg body weight for once. The mice were given ad libitum access to drinking water or water containing 1.5 mg/ml metformin. Mouse survival was monitored every 6 h for 4 days.

Mice were sacrificed 8 h after bacterial infection. The liver was harvested, fixed in 4% neutral formaldehyde. Paraffin slice of the liver was stained with hematoxylin and eosin. Images were captured under the Zeiss Axio Observer D1 microscope armed with a color CCD (ZEISS).

All experiments were performed independently at least three times unless otherwise stated, with one representative experiment shown. Data are presented as mean ± SD. Statistical analysis was performed using GraphPad Prism 4.0 (GraphPad; San Diego, CA, USA). One-way analysis of variance followed by Tukey’s multiple comparison test was used to analyze the statistical significance among multiple groups and unpaired Student’s t-test between two groups. Kaplan–Meier survival curves were adopted for analysis of mice survival, and the statistical difference between two groups was determined using the log-rank (Mantel–Cox) test. P values < 0.05 were considered statistically significant.

Although many studies have focused on how inflammasomes are assembled and activated, little is known about the regulatory signaling of NLRP3 inflammasome activation upon extracellular ATP stimulation. As dramatic metabolic changes occur during macrophage activation (22), it is of great interest whether AMPK, a key regulator of energy metabolism, regulates ATP-induced inflammasome activation in LPS-primed macrophages. To explore this problem, LPS-activated BMDMs were treated with serial concentrations of metformin, a well-known AMPK agonist, followed by ATP treatment. First, we observed the pyroptotic cells (PI permeable) and calculated their ratio to the total cells (indicated by nuclear dye Hoechst 33342). As expected, ATP induced pyroptosis in the BMDMs; and metformin treatment further increased the pyroptosis induced by LPS and ATP treatment (Figures 1A,B). However, metformin alone did not induce cell death in both control and LPS-treated cells without ATP treatment (Figure S1A in Supplementary Material). Similar results were observed in mouse macrophage J774A.1 cell line (Figures 1C,D; Figure S1B in Supplementary Material). Considering that metformin is an AMPK agonist, these results suggested that AMPK signaling might have contributed to ATP-induced pyroptosis in LPS-activated macrophages.

It has been observed that the activity of AMPK can be dampened in macrophages by LPS, free fatty acid, or other stimulators (28). But recent studies showed that AMPK signaling is dramatically increased in LPS-activated macrophages after ATP treatment (21, 22), accompanied by pyroptosis and mature IL-1β release (22). Consistent with these studies, western blot analysis demonstrated that LPS (alone) treatment led to a marked increase of NLRP3 and pro-IL-1β protein expression in BMDMs, but it did not induce the release of active caspase-1p10, mature IL-1β, and HMGB1 (Figures 2A–D, supernatants). Upon ATP triggering, however, active caspase-1p10 fragment, HMGB1, and mature IL-1β (17 kDa) were released into the supernatants (Figures 2A–D). Moreover, the release of these contents from the cells in response to ATP stimulation can be further promoted by metformin treatment (Figures 2A–D). As the release of mature IL-1β is dependent on pyroptosis (4, 10), our results suggested that the augmentation of AMPK signaling by metformin enhanced LPS + ATP-induced inflammasome activation and pyroptosis.

Next, we performed two additional assays to corroborate the augmentation effect of metformin on ATP-induced inflammasome activation. First, the subcellular distribution of ASC in BMDMs was observed by immunofluorescence microscopy. ASC is an adaptor protein that mediates the recruitment of caspase-1 to the NLRP3 inflammasome to form a large speck in the cell, and thus activates caspase-1 (29). The results showed that ASC was evenly distributed in LPS and vehicle-treated BMDMs (Figure 3A). After ATP treatment, ASC specks could be observed in ~20% of LPS-primed cells (Figures 3A,B), indicating the activation of inflammasomes as having been indicated previously (30). Furthermore, metformin treatment increased the ratio of cells containing ASC specks induced by ATP to >40% (Figures 3A,B). Second, the CBA assay was used to detect soluble IL-1β in the supernatants. The results showed that LPS + ATP treatment induced a significant release of IL-1β from BMDMs, which could be greatly promoted by metformin treatment (Figure 3C).

In J774A.1 cells, we also observed that LPS induced the expression of NLRP3 and pro-IL-1β (Figure 4A, cell lysates). Upon ATP treatment, both active caspase-1p10 and HMGB1 could be released into the supernatants, and metformin dose-dependently increased their release (Figures 4A–C). Although soluble IL-1β was hardly detectable in the precipitated proteins of the culture supernatants by western blotting, CBA assays could detect such low levels of IL-1β and the result showed that LPS + ATP induced the release of IL-1β and metformin dose-dependently enhanced its release (Figure 4D).

It seemed that metformin did not influence the constitutive expression of pro-caspase-1 (45 kDa) and the induced-expression of pro-IL-1β (31 kDa) and NLRP3 (120 kDa) by LPS in the two cell types, nor did it induce the release of mature IL-1β, active caspase-1p10, and HMGB1 as well as the formation of ASC specks in the absence of ATP treatment (Figures 2–4).

Interestingly, in BMDMs, AMPK activity (reflected by its phosphorylation at Thr172) was suppressed by LPS alone (Figures 5A,B); while in J774A.1 cells, AMPK activity was low in both control and LPS (alone) groups (Figures 5C,D). Upon ATP treatment, however, AMPK was reactivated in both cell types. Furthermore, metformin treatment dose-dependently increased the AMPK activation as compared to their respective control (LPS alone or LPS + ATP) (Figures 5A–D).

Altogether, these results suggested that elevated AMPK signaling was associated with inflammasome activation and pyroptosis in LPS-primed macrophages upon ATP stimulation.

To further explore whether AMPK activation contributes to ATP-induced pyroptosis in LPS-activated macrophages, BMDMs were first primed with LPS and then treated with compound C, a well-known pharmacological AMPK inhibitor (31, 32). Subsequently, the cells were treated with ATP. As shown in Figures 6A,B, compound C significantly suppressed ATP-induced pyroptosis in LPS-primed BMDMs. Supporting this, compound C also inhibited the release of IL-1β (Figure 6C) and the formation of ASC specks (Figures 6D,E) in these cells. Similar results were observed in J774A.1 cells. Compound C not only suppressed ATP-induced pyroptosis (Figures 7A,B) but also inhibited the secretion of IL-1β (Figure 7C) and HMGB1 (Figure 7D) from J774A.1 cells. Moreover, compound C significantly attenuated the effect of metformin on promoting ATP-induced pyroptosis (Figures 6A,B and 7A,B), as well as the release of IL-1β (Figures 6C and 7C) and HMGB1 (Figure 7D) in LPS-primed BMDMs and J774A.1 cells, respectively. Without ATP treatment, compound C or compound C plus metformin did not induce pyroptosis in these two macrophage types (Figures S1C,D in Supplementary Material).

To further confirm the pharmacological data, we performed further experiments in J774A.1 cells in which the expression of AMPKα had been genetically knocked down by AMPKα1-specific siRNA. When AMPKα expression was nearly reduced by 65% after siRNA knockdown (Figures 8A,B), the release of HMGB1 (Figures 8A,C, western blotting) and IL-1β (Figure 8D, CBA assay) by ATP or metformin plus ATP treatment was significantly reduced as compared to that of NC siRNA treated cells. Consistent with the reduction of these inflammasome activation markers, ATP-induced pyroptosis was also significantly decreased (Figures 8E,F). Notably, metformin-mediated augmentation of pyroptosis by ATP stimulation was also markedly attenuated by AMPKα knockdown (Figures 8E,F). Taken together, these results indicated that AMPK signaling had contributed to ATP-induced inflammasome activation and pyroptosis in LPS-primed murine macrophages.

Finally, we investigated whether metformin administration enhanced systemic inflammation in vivo using a mouse model of bacterial sepsis. Mice were first intraperitoneally injected with a sublethal dose of live E. coli according to previous studies (9, 27). The bacterial infection at a dose of 1 × 10⁹ CFU/mouse caused only ~10% mortality in the mice treated with vehicle after 24 h, but metformin administration accelerated the onset of death (within 24 h) and significantly increased the mortality of the mice to ~50% during the period of 4-day observation (Figure 9A), suggesting that metformin exacerbated the severity of bacterial sepsis. Consistent with this result, histopathological analysis showed that bacterial infection led to infiltration of inflammatory cells in the liver in both metformin and vehicle groups, but the mice in the metformin group showed more inflammatory cell infiltration and tissue injury around their hepatic vascular walls (Figure 9B). In further support of this, metformin-treated mice had a marked increase in their serum levels of IL-1β at 4-h and 8-h time points when compared with controls (vehicle treated) (Figure 9C), indicating increased inflammasome activation in metformin-treated mice upon bacterial infection. In addition, the percentages of neutrophils (CD11b⁺Gr-1⁺ cells) in the peritoneal cavity were greatly increased in metformin-treated mice in comparison to vehicle (Figures 10A,B). As indicators of inflammasome activation and pyroptosis (10), hepatic mature IL-1β and active caspase-1p10 protein levels were higher in the metformin group as compared to control and vehicle group, as revealed by western blotting (Figures 10C–E). These results suggested that bacterial infection had systemically induced inflammasome activation and IL-1β maturation, and that metformin treatment enhanced such inflammasome activation and organ injury, culminating in aggravated bacterial sepsis in mice.