An Apoptotic Caspase Network Safeguards Cell Death Induction in Pyroptotic Macrophages

Summary Pyroptosis has emerged as a key mechanism by which inflammasomes promote host defense against microbial pathogens and sterile inflammation. Gasdermin D (GSDMD)-mediated cell lysis is a hallmark of pyroptosis, but our understanding of cell death signaling during pyroptosis is fragmented. Here, we show that independently of GSDMD-mediated plasma membrane permeabilization, inflammasome receptors engage caspase-1 and caspase-8, both of which redundantly promote activation of apoptotic executioner caspase-3 and caspase-7 in pyroptotic macrophages. Impaired GSDMD pore formation downstream of caspase-1 and caspase-8 activation suffices to unmask the apoptotic phenotype of pyroptotic macrophages. Combined inactivation of initiator caspase-1 and caspase-8, or executioner caspase-3 and caspase-7, is required to abolish inflammasome-induced DEVDase activity during pyroptosis and in apoptotic Gsdmd−/− cells. Collectively, these results unveil a robust apoptotic caspase network that is activated in parallel to GSDMD-mediated plasma membrane permeabilization and safeguards cell death induction in pyroptotic macrophages.


INTRODUCTION
Pyroptosis is initiated downstream of inflammasome assembly in activated innate immune cells (Broz and Dixit, 2016;Lamkanfi and Dixit, 2014). It has emerged as a powerful defense mechanism of the host against microbial pathogens (de Vasconcelos and Lamkanfi, 2020). It also drives detrimental autoinflammation, sepsis, and non-alcoholic steatohepatitis (NASH) by promoting passive secretion of interleukin-1b (IL-1b) and alarmins Xiao et al., 2018;Xu et al., 2018). Pyroptosis induction by inflammasomes is considered a linear pathway in which murine inflammatory caspase-1 and caspase-11 and human caspase-1, caspase-4, and caspase-5 cleave gasdermin D (GSDMD) to release the N-terminal GSDMD N domain that forms higher-order oligomeric pores in the plasma membrane to induce osmotic swelling and early cell lysis (Aglietti et al., 2016;Ding et al., 2016;Kayagaki et al., 2015;Sborgi et al., 2016;Shi et al., 2015). This is in marked contrast to apoptosis, in which parallel maturation of apoptotic executioner caspase-3 and caspase-7 by initiator caspase-8 and caspase-9 results in cleavage of hundreds of substrates that orchestrates the coordinated disassembly of the cell without spilling the intracellular content in the extracellular environment (Nagata and Tanaka, 2017).
A wealth of recent findings suggests extensive cross-talk between inflammatory and apoptotic caspases (de Vasconcelos and Lamkanfi, 2020;Fritsch et al., 2019;Newton et al., 2019;. We and others previously demonstrated that apoptosis-associated speck-like protein containing a CARD (ASC) specks serve as cytosolic scaffolds for inflammasome-mediated caspase-8 activation and induction of apoptosis in caspase-1-deficient macrophages in response to stimuli of the Nlrc4, Nlrp1b, AIM2, or Nlrp3 inflammasome pathways (Lee et al., 2018;Pierini et al., 2012;Puri et al., 2012;Sagulenko et al., 2013;Van Opdenbosch et al., 2017). Moreover, an early study showed that caspase-1 activates the apoptotic executioner caspase-7 in wild-type macrophages in response to stimuli of the Nlrp3 and Nlrc4 inflammasomes (Lamkanfi et al., 2008). Yet, the molecular mechanisms in inflammasome-activated macrophages that regulate the switch from pyroptosis to apoptosis signaling remains unclear.
Here, we demonstrate that pyroptosis induced by the Nlrp1b, Nlrc4, and Nlrp3 inflammasomes in wild-type macrophages exhibits hallmark apoptotic features, including activation of apoptotic caspase-3 and caspase-7, DEVDase activity, and cleavage of apoptotic substrates. We show that inflammasome receptors independently engage caspase-1 and caspase-8, both of which redundantly promoted activation of apoptotic executioner caspase-3 and caspase-7 in parallel to GSDMDmediated plasma membrane permeabilization in wild-type macrophages. Combined inactivation of initiator caspase-1 and caspase-8, or executioner caspase-3 and caspase-7, was required to abolish inflammasome-induced DEVDase activity during pyroptosis as well as in apoptotic Gsdmd À/À cells. Notably, impaired GSDMD pore formation downstream of caspase-1 and caspase-8 activation sufficed to unmask the apoptotic phenotype of pyroptotic macrophages. Collectively, these results unveil a robust apoptotic caspase network that is activated in parallel to GSDMD-mediated plasma membrane permeabilization to safeguard cell death induction in pyroptotic macrophages.
To assess whether DEVDase activity accompanies pyroptosis induced through additional inflammasome pathways, we stimulated wild-type B6 macrophages with FlaTox, a synthetic fusion of the Bacillus anthracis (B. anthracis) lethal factor N-terminal region fused to Legionella pneumophila flagellin (LFn-FlaA) that selectively activates the Nlrc4 inflammasome when targeted to the cytosol with B. anthracis protective antigen (PA) (Van Opdenbosch et al., 2017;von Moltke et al., 2012). Increased DEVDase activity in FlaTox-stimulated B6 BMDMs ( Figure 1C) occurred concomitant with plasma membrane permeabilization as measured by PI staining (Figure S1E). Loss of Nlrc4 abrogated FlaTox-induced DEVDase activity and PI staining ( Figures 1C and S1E), demonstrating that DEV-Dase activity is induced following Nlrc4 activation.
Consistent with Nlrp1b-and Nlrc4-mediated pyroptosis being associated with increased DEVDase activity, western blot analysis confirmed cleavage of well-established apoptosis markers in pyroptotic cell lysates ( Figure 1D). Caspase-mediated cleavage of ROCK1 in a 30-kDa fragment renders the protein constitutively active and drives apoptotic membrane blebbing (Coleman et al., 2001;Sebbagh et al., 2001). We observed a ROCK1 cleavage fragment in pyroptotic cell lysates of LeTx-and FlaTox-treated B6 Nlrp1b+ macrophages that was similarly sized to the ROCK1 cleavage fragment of staurosporinetreated macrophages ( Figure 1D). Consistent with published reports (de Vasconcelos et al., 2019;Yu et al., 2014), we also observed substantial proteolytic maturation of the proapoptotic Bcl2 protein BID into a fragment that appeared of similar size as the tBID cleavage product in apoptotic tumor necrosis factor a (TNF) + cycloheximide (CHX)-treated macrophages ( Figure 1D).
Impaired GSDMD Pore Formation Downstream of Caspase-1 and Caspase-8 Activation Unmasks Apoptosis Based on our detection of DEVDase activity in pyroptotic macrophages and the observation that caspase-1 and caspase-8 redundantly activate executioner caspase-3 and caspase-7, we hypothesized that GSDMD-induced pore formation and cell lysis may mask a background apoptotic program in pyroptotic cells. In order to eliminate confounding effects of GSDMD-mediated cell lysis, we further dissected the function and signaling mechanism of this caspase cascade in a GSDMD-deficient background. Consistent with previous reports showing that canonical inflammasome activation triggers an alternative cell death response in GSDMD-deficient macrophages (de Vasconcelos et al., 2019;Gonç alves et al., 2019;He et al., 2015;Kayagaki et al., 2015;Tsuchiya et al., 2019), release of the lytic cell death marker lactate dehydrogenase (LDH) was blunted in culture media of LeTx-treated B6 Nlrp1b+ Gsdmd À/À cells and Fla-Tox-stimulated Gsdmd À/À BMDMs ( Figures 3A and 3B). However, careful analysis of DIC micrographs showed cells with a shrunken and blebbing appearance that are reminiscent of apoptosis and distinct from the classical swollen morphology of pyroptotic macrophages ( Figures 3C and 3D). In agreement, flow cytometric analysis of LeTx-and FlaTox-induced cell death in respectively B6 Nlrp1b+ Gsdmd À/À and Gsdmd À/À BMDMs identified a population of $45% apoptotic cells that was positive for the early apoptosis marker Annexin-V while being impermeable to PI (Annexin-V + /PI À ). In contrast, GSDMD-proficient pyroptotic macrophages displayed Annexin-V and PI co-staining (Annexin-V + /PI + ) ( Figures 3E and 3F). A kinetic analysis of DEVDase activity further corroborated these results. The number of DEVDasepositive cells was comparable in pyroptotic B6 Nlrp1b+ and apoptotic B6 Nlrp1b+ Gsdmd À/À macrophages following LeTx or FlaTox stimulation ( Figure 3G). DEVDase activity was delayed in apoptotic macrophages relative to pyroptotic cells ( Figures  3G and 3H), although this could at least partially be due to less efficient cytosolic uptake of the fluorogenic substrate in early Article ll OPEN ACCESS apoptotic cells. Notably, BMDMs from mice expressing an inactive GSDMD I105N mutant that is impaired in inducing pyroptotic cell lysis (Kayagaki et al., 2015) phenocopied Gsdmd À/À macrophages ( Figure 3I), demonstrating that inflammasome-mediated apoptosis induction was not unique to GSDMD-deficient macrophages and that impaired GSDMD pore formation downstream of caspase-1 and caspase-8 activation suffices to unmask inflammasome-induced apoptotic hallmarks in macrophages. Unlike B6 Nlrp1b+ Gsdmd À/À macrophages, Gsdmd À/À macrophages (that lack expression of a LeTx-responsive Nlrp1b allele) failed to induce apoptosis ( Figures S2A and S2B) as well as PI staining and DEVDase activity ( Figure S2C) in response to LeTx intoxication. Paralleling results in LeTx-intoxicated B6 Nlrp1b+ macrophages ( Figures 1B and S1D), the proteasome inhibitor MG132 inhibited the induction of DEVDase activity and PI staining in LeTx-stimulated B6 Nlrp1b+ Gsdmd À/À macrophages ( Figure S2D). Thus, expression of a functional Nlrp1b allele is required for LeTx-induced apoptosis in GSDMD-deficient macrophages.
In order to directly assess the role of caspase-8, we generated Ripk3 À/À Casp8 À/À macrophages in a B6 Nlrp1b+ Gsdmd À/À background. A dynamic analysis of DEVDase activity and PI incorporation demonstrated that these cell death markers were comparably induced in caspase-8/Ripk3-deficient and control B6 Nlrp1b+ Gsdmd À/À macrophages ( Figures 4F and 4G). Additionally, LeTx-and FlaTox-induced maturation of caspase-1 and apoptotic executioner caspase-3 and caspase-7 in these cells was comparable to levels seen in caspase-8/Ripk3-sufficient B6 Nlrp1b+ Gsdmd À/À BMDMs ( Figure 4H). These findings demonstrate that in marked contrast to caspase-1-deficient macrophages (Lee et al., 2018;Van Opdenbosch et al., 2017), caspase-8 is dispensable for activation of caspase-3 and caspase-7 in macrophages lacking GSDMD expression. A functional implication of these results is that unlike in Values represent mean ± SD of technical duplicates of a representative experiment from three biological repeats. All scale bars represent 10 mm.
Nlrp3-Inflammasome-Induced Apoptosis in Gsdmd À/À Macrophages Lipopolysaccharide (LPS) priming is required for ATP-and nigericin-induced activation of the Nlrp3 inflammasome in BMDMs (Bauernfeind et al., 2009;Le Feuvre et al., 2002;Song et al., 2017). As expected, micrographs of LPS+ATP-and LPS+nigericin-stimulated wild-type macrophages displayed a characteristic pyroptotic morphology featuring a swollen cytosol and rounded nuclei ( Figure 5A). As with pyroptosis induction by the Nlrp1b and Nlrc4 inflammasomes (Figure 1), Nlrp3-driven pyroptosis in LPS+ATP-and LPS+nigericin-stimulated wild-type macrophages was accompanied by a sharp rise in DEVDase activity concomi- tant with PI staining (Figures 5B and 5C). The morphology of ATP-stimulated Gsdmd À/À macrophages differed considerably, with cytosolic shrinkage and formation of apoptotic bodies evident within minutes after ATP stimulation ( Figure 5A). Furthermore, DEVDase activity in LPS+ATP-stimulated Gsdmd À/À macrophages preceded the induction of PI staining by $1 h ( Figure 5B), consistent with the induction of secondary necrosis upon prolonged in vitro incubation of apoptotic cells. Unexpectedly, LPS+nigericin-stimulated Gsdmd À/À macrophages had a swollen appearance suggestive of necrotic cell death ( Figure 5A), although a delayed induction of DEVDase activity that slightly preceded the induction of PI staining was observed ( Figure 5C), suggesting that inflammasome-mediated apoptotic morphological changes may have been masked by osmotic disbalance directly mediated by the ionophore. In agreement, Nlrp3-mediated pyroptosis in wildtype BMDMs was associated with prominent maturation of caspase-1 and caspase-7, whereas Gsdmd À/À macrophages additionally triggered robust cleavage of caspase-3 and caspase-8 following treatment with LPS+ATP or LPS+nigericin ( Figure 5D). These results extend our observations on pyroptotic DEVDase activity to the Nlrp3 inflammasome and show that Nlrp3 activation in Gsdmd À/À cells promotes induction of apoptotic cell death markers, akin to the Nlrp1b and Nlrc4 pathways.

DISCUSSION
Macrophages that lack caspase-1 or express the catalytically inactive caspase-1 C284A mutant switch to caspase-8-mediated apoptosis (Pierini et al., 2012;Sagulenko et al., 2013;Van Opdenbosch et al., 2017). GSDMD-deficient macrophages were also suggested to switch cell death modes. Legionella pneumophila infection was reported to trigger caspase-7-mediated pore formation in GSDMD-deficient macrophages (Gonç alves et al., 2019), and other inflammasome stimuli were shown to induce caspase-3-mediated apoptosis and deafness associated tumor suppressor/gasdermin E (DFNA5/GSDME)-mediated secondary necrosis downstream of caspase-1 (Tsuchiya et al., 2019), or to switch to caspase-1-mediated cleavage of caspase-3 and caspase-7 and apoptotic cell death (Mahib et al., 2020;Taabazuing et al., 2017). However, the mechanisms by which inflammasomes regulate the switch from pyroptosis to apoptosis signaling remain unclear. Rather than switching cell death modes, our observations strongly suggest that an apoptotic program is readily activated concomitant with induction of GSDMD pores in wild-type macrophages and not only in the context of genetic deletion of caspase-1 or GSDMD. The presented findings give rise to a mechanistic model of pyroptosis in which caspase-1 cleaves GSDMD for cell lysis in parallel to caspase-1 and caspase-8 redundantly activating caspase-3 and caspase-7, both of which promote apoptotic DEVDase activity independently of each other in pyroptotic cells ( Figure S5). In support of this model, we showed that impaired GSDMD pore formation downstream of caspase-1 and caspase-8 activation by the Nlrp1b and Nlrc4 inflammasomes sufficed to unveil apoptotic morphological features in GSDMD I105N mutant macrophages. This novel paradigm of pyroptosis, in which activation of the apoptotic machinery is an intrinsic component of pyroptotic cell death signaling ( Figure S5), has the merit that it elegantly explains how inflammasome-induced apoptotic hallmarks ensue through distinct signaling pathways in macrophages lacking caspase-1 or GSDMD, respectively. In addition to the previously reported roles of ASC-mediated caspase-8 activation (Gonç alves et al., 2019;Pierini et al., 2012;Sagulenko et al., 2013;Van Opdenbosch et al., 2017), we now provided genetic evidence that also caspase-1 plays a critical role in inflammasome-induced apoptosis signaling. Indeed, combined deletion of caspase-1 and caspase-8 (or ASC in lieu of caspase-8) proved essential to blunt DEVDase activity and apoptosis induction in GSDMD-deficient cells. Moreover, we posit that executioner caspase-3 and caspase-7 act redundantly for inflammasome-induced apoptosis signaling, because combined deletion of caspase-3 and caspase-7 was necessary to blunt inflammasome-induced DEVDase activity and induction of cell death in GSDMD-sufficient and GSDMDdeficient macrophages, respectively. The redundancy we have uncovered between caspase-1 and caspase-8 as initiator caspases; and between GSDMD, and caspase-3 and caspase-7 in the execution phase of pyroptosis likely serves to ensure a commitment to cell death induction in inflammasome-activated macrophages. This built-in redundancy in the pyroptotic caspase cascade may have evolved to ensure the robustness of pyroptosis as an anti-microbial host defense mechanism. GSDMD-targeting pathogens remain to be discovered, but cowpox viruses express a cytokine response modifier (CrmA) that efficiently targets caspase-1 and caspase-8 (Zhou et al., 1997). This may have been an evolutionary more effective strategy for cowpox viruses to curb inflammasome-induced cell death than selective inhibition of caspase-1 or caspase-8. Moreover, the relative expression levels of caspase-1 and GSDMD, as well as other factors that regulate the kinetics of caspase-1-mediated GSDMD pore formation, may alter the balance of this integrated pyroptotic cell death program in favor of apoptosis as the default morphological outcome in non-myeloid cell types. For instance, it has been suggested that inflammasome-induced apoptosis may be the default inflammasome cell death mode in cell types that express no or low levels of GSDMD, such as primary cortical neurons, mast cells, keratinocytes, and endothelial cells (Sollberger et al., 2015;Tsuchiya et al., 2019;Xi et al., 2016). Although our studies have focused on macrophages (which express abundant levels of GSDMD), they suggest a mechanistic model of inflammasome-induced apoptosis that also may operate in cell types with low GSDMD levels. Finally, our results predict and clarify the mechanisms by which selective pharmacological GSDMD inhibitors will convert pyroptotic cell lysis into an apoptosis response that may curb detrimental inflammatory cytokine secretion in infectious and autoinflammatory diseases (de Vasconcelos and Lamkanfi, 2020;Kanneganti et al., 2018;. In conclusion, the presented work transforms understanding of pyroptosis from a linear signaling axis into an integrated cell death signaling network ( Figure S5). Future studies should address whether caspase-3/7-mediated substrate cleavage in pyroptotic cells contributes to the quality of the instigated inflammatory and immune responses and how they impact on the resolution of infections.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:  Received: January 21, 2020 Revised: April 2, 2020 Accepted: July 2, 2020 Published: July 28, 2020

Lead Contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Mohamed Lamkanfi (mohamed.lamkanfi@ugent.be).

Materials Availability
This study did not generate new unique reagents.

Data and Code Availability
This study did not generate unique datasets or codes.