Mitochondrial apoptosis, considered "immunologically silent," can also activate IL-1β: BAX/BAK engages two pro-inflammatory pathways simultaneously through caspase-3/7

The traditional view considers mitochondrial apoptosis immunologically silent: apoptotic cells do not immediately rupture their membranes and are cleared by phagocytes. However, death receptor- and TLR-associated apoptosis is not always quiet: caspase-8 can sometimes directly cleave IL-1β. Vince and colleagues therefore asked a specific question: Can BAX/BAK-mediated mitochondrial apoptosis activate IL-1β in macrophages? If so, is this carried out by NLRP3, caspase-8, or cell lysis caused by gasdermin pore formation?

Step 1: Establishing a system that specifically initiates BAX/BAK apoptosis (Figure 1). The authors inhibited two anti-apoptotic proteins separately and together in macrophages. Inhibiting BCL-XL alone did not kill the cells; rapid death occurred only when MCL-1 was also depleted. Cells with a Bax/Bak double knockout resisted all the combinations. These results show that death induced by this system specifically proceeds through BAX/BAK, and all subsequent conclusions are based on this system.

Step 2: Confirming the inflammatory response and positioning upstream and downstream events (Figures 2–3). The authors first primed cells with LPS and then initiated BAX/BAK apoptosis. Caspase-1 and IL-1β were markedly activated, while TNF secretion remained unchanged: the cells' ability to produce and secrete proteins was not impaired, and the affected pathway was specifically the IL-1β pathway. The authors then treated cells with high potassium and MCC950. Both inhibited IL-1β but could not prevent apoptosis. These results show that cell death and inflammatory activation are two separable events, with NLRP3 downstream of BAX/BAK and dependent on potassium efflux.

Figure 3Complete original figure

View complete figure ↗

Compare figure ↗

Step 3: Distinguishing two parallel pathways (Figures 4–6). The authors knocked out NLRP3 or caspase-1 and found that IL-1β was only partially reduced: the canonical pathway could not account for all of it. Although GSDMD was cleaved, knocking it out did not affect IL-1β release. The authors then examined caspase-8. BAX/BAK activation was accompanied by IAP protein loss and caspase-8 activation, which also occurred without LPS, excluding interference from TLR stimulation. IL-1β was markedly reduced in caspase-8-deficient cells, while caspase-1 activation continued as usual. These results show that caspase-8 forms a second IL-1β-processing pathway parallel to NLRP3, rather than acting upstream of NLRP3.

Figure 6Complete original figure

View complete figure ↗

Compare figure ↗

Step 4: Identifying the common upstream component (Figure 7). Knocking out caspase-3 alone produced no phenotype. Only the caspase-3/7 double knockout blocked both pathways simultaneously, while TNF secretion and the response to nigericin remained normal. GSDME was likewise cleaved and likewise dispensable. These results show that caspase-3/7 is functionally redundant and is the common upstream component linking apoptosis to inflammation.

Figure 7Complete original figure

View complete figure ↗

Compare figure ↗

Conclusion: After BAX/BAK activation, caspase-3/7 activates NLRP3 through potassium efflux along one pathway and activates caspase-8 through IAP loss to directly cleave IL-1β along the other; gasdermin cleavage is merely an accompanying event.

1. Assess death and inflammation separately: Inhibiting the inflammasome does not mean saving the cells. The measurement panel should include death readouts (PI/LDH), cleavage of the different caspases, ASC specks, extracellular and intracellular IL-1β, and a control involving an unrelated cytokine such as TNF.
2. "Cleaved" does not mean "required": Both GSDMD and GSDME are cleaved, yet neither knockout affects the outcome. Cleavage is only a clue to involvement; knockout is needed to determine the functional role.
3. Redundancy requires a double knockout to reveal it: A caspase-3 single knockout has no phenotype, which can easily lead to the mistaken conclusion that it is "unimportant". Double knockouts should be considered for executor molecules in the same family.

• Comparison | Mitochondrial apoptosis and NLRP3 (Allam 2014, EMBO Rep): The Allam paper shows that canonical NLRP3 stimuli do not require mitochondrial apoptosis. This paper shows that an apoptotic system in which BCL-XL and MCL-1 are both inhibited can indeed drive IL-1β. Whether something is "required" must be stated together with the stimulus context; the two papers do not contradict each other.
• Comparison | RIPK3 activates NLRP3 without requiring MLKL (Lawlor 2015, Nat Commun): Within the same context of limited IAP activity, the Lawlor paper emphasizes the RIPK3–caspase-8–NLRP3 axis, while this paper emphasizes the BAX/BAK–caspase-3/7 axis. The two pathways complement each other.
• Comparison | GSDMD is an executor of pyroptosis (He 2015, Cell Res): The He paper establishes the central role of the caspase-1–GSDMD axis in canonical inflammasome signaling. This paper shows that GSDMD is cleaved during BAX/BAK apoptosis but is not required: the cell-death context determines the release mechanism.
• Extension | TLR-induced death in IFN-γ-primed macrophages (Simpson 2022, Immunity): This paper shows that MCL-1 and BCL-XL jointly restrain BAX/BAK in macrophages. The Simpson paper shows that IFN-γ/LPS reduces BCL-2 through caspase-8 and reduces MCL-1 through iNOS, releasing this restraint from two directions.

(The comparisons across papers above are a synthesis, rather than a direct conclusion of any one original paper.)