Bacterial Pore-Forming Toxins Promote the Activation of Caspases in Parallel to Necroptosis to Enhance Alarmin Release and Inflammation During Pneumonia

Pore-forming toxins are the most common virulence factor in pathogenic bacteria. They lead to membrane permeabilization and cell death. Herein, we show that respiratory epithelial cells (REC) undergoing bacterial pore-forming toxin (PFT)-induced necroptosis simultaneously experienced caspase activation independently of RIPK3. MLKL deficient REC treated with a pan-caspase inhibitor were protected in an additive manner against PFT-induced death. Subsequently, cleaved versions of caspases-2, -4 and -10 were detected within REC undergoing necroptosis by immunoblots and monoclonal antibody staining. Caspase activation was observed in lung samples from mice and non-human primates experiencing Gram-negative and Gram-positive bacterial pneumonia, respectively. During apoptosis, caspase activation normally leads to cell shrinkage, nuclear condensation, and immunoquiescent death. In contrast, caspase activity during PFT-induced necroptosis increased the release of alarmins to the extracellular milieu. Caspase-mediated alarmin release was found sufficient to activate resting macrophages, leading to Interleukin-6 production. In a mouse model of Gram-negative pneumonia, deletion of caspases -2 and -11, the mouse orthologue of caspase-4, reduced pulmonary inflammation, immune cell infiltration and lung damage. Thus, our study describes a previously unrecognized role for caspase activation in parallel to necroptosis, and indicates that their activity plays a critical pro-inflammatory role during bacterial pneumonia.

can induce apoptosis through its permeabilization of the mitochondrial inner membrane, causing a metabolic breakdown in the cell and the release of cytochrome C, leading to the activation of caspase-9 12 . Alternatively, exposure to Fas ligand, triggers apoptosis through caspase-8 mediated extrinsic apoptosis 13 . Pertinent to this manuscript, apoptosis can also be initiated by less studied caspases such as caspase-2, the most evolutionarily conserved caspase, and caspase-10 upon contact with bacterial pathogens 11,14 . How this occurs is still unclear. PFTs have also been shown to induce programmed modes of necrotic death, specifically pyroptosis and necroptosis 7 . Pyroptosis is mediated by the activation of the canonical caspase-1 or the non-canonical caspase-4 15,16 . Caspase-4 and its murine orthologue, caspase-11, have been shown to mediate epithelial defenses against enteric bacterial pathogens by activating the inflammasome 16 . To date, the complete functions of caspases -2, -4, and 10 remain undefined 10,11,17,18 .
Necroptosis, a form of programmed necrosis, is highly inflammatory due to the purposeful release of cytosolic molecules containing danger-associate-molecular-patterns, i.e. alarmins. Necroptosis is modulated by receptor-interacting serine-threonine kinases -1 and -3 (RIPK1, RIPK3) and occurs when caspase-8, a pivotal apoptotic caspase, is inhibited. Briefly, caspase-8 inactivates both RIPK1 and RIPK3 by sequestration or proteolytic cleavage 19,20 . When cellular distress signals are present and caspase-8 is inactive, RIPK3 binds to RIPK1 forming the necroptosome. The necroptosome phosphorylates mixed lineage kinase domain-like protein (MLKL), which then integrates and damages cell membranes, leading to necrotic death [19][20][21] . Importantly, necroptosis is understood and is by definition exclusive of caspase activity 19,21 . This is supported by our own studies that showed no impact of caspase-inhibition on PFT-mediated necroptotic death of macrophages 22 . In addition, it has also been reported that caspase-8 activity suppresses RIPK3 activation 19 . Nonetheless, a recent report showed that during influenza A respiratory tract infection, caspase-8 induced apoptosis occurred in parallel to necroptosis in a RIPK3 dependent manner 23 . Thus, the distinction between apoptosis and necroptosis may not be so clear cut.
In recent work, we have shown that PFTs activate the necroptosis pathway in macrophages 22 and respiratory epithelial cells (REC) 24 and this contributes to pulmonary injury during bacterial pneumonia. PFT-induced necroptosis required RIPK1, RIPK3 and MLKL, yet was activated independently of death receptor engagement through non-canonical means, i.e. ion dysregulation as the result of membrane permeabilization 24 . One important distinction between macrophage and REC death is that only partial protection against PFTs was observed in REC following necroptosis inhibition. In contrast, PFT-induced death in macrophages was unequivocally blocked by necroptosis inhibition. Based upon these results, we hypothesized that a second cell death signaling pathway was active in PFT-exposed REC.
Viral infections have been reported to simultaneously trigger necroptosis and RIPK3/caspase-8 mediated apoptosis within mammalian fibroblasts and lung epithelial cells 23 . Following influenza challenge, RIPK3 has been shown to activate caspase-8 16 . To test for this, we transfected A549 cells with siRNA targeting MLKL and caspase-8 and challenged these with Sma. No additive protection was observed in the MLKL and caspase-8 knockdown cells (Fig. 1f, Supplementary Fig. 2). Importantly, no activation of caspases-1, -3 or -8 was observed in Sma challenged A549 cells; conversely this was observed when A549 cells were challenged with positive control cyclohexamide or LPS/nigericin ( Supplementary Fig. 3). Simultaneous siRNA knockdown of RIPK3 and MLKL also did not have an additive protective effect following Sma challenge (Fig. 1f). Together, these results suggest a different form of caspase-associated cell death was occurring in PFT-exposed RECs, one independent of canonical intrinsic or extrinsic apoptosis.
To verify that caspase activation and necroptosis occurred within the same cell, A549 cells were first examined using propidium iodide (PI), which enters necrotic cells, and then caspase-active specific fluorescent-labeled inhibitors of caspases (FLICA) 25 after challenge with rPly. We observed that both A549 and NHBECs were PI positive and pan-caspase FLICA positive, providing evidence of ongoing caspase activation and necrosis within the same cell. PI positive cells were also FLICA positive for caspase-2 and -10, but not caspase-1,  Supplementary Fig. 6). Because FLICA may be non-specific, we subsequently examined cells with fluorophore-labeled monoclonal antibodies against cleaved caspase-2, -3, -4, -10, and pMLKL. Consistent with prior results, we observed simultaneous caspase-2, -4 and -10 activation with necroptosis in Sma (Fig. 2e). This too was observed in rPly or α-toxin challenged cells ( Supplementary Fig. 7).
Caspase activation during PFT-mediated death changes the morphological characteristics of necroptosis. Canonical necroptosis results in cell membrane dissolution, release of cytoplasmic components, and preservation of the nucleus 19 . In contrast, apoptosis involves cell shrinkage, nuclear fragmentation, and the formation of apoptotic bodies 10 . We sought to determine if caspase activity affected the morphological characteristics of cell death in A549 cells treated with pneumolysin in the presence or absence of Zvad (Fig. 3). In cells where caspase activity was not inhibited, transmission electron microscopy images showed less membrane blebbing or "bubbles" than necroptotic cells. In addition, we also observed increased mitophagy, paraptotic-like vacuolation, and heterochromatin condensation in comparison to those undergoing solely necroptosis. In contrast, Zvad treated cells solely undergoing necroptosis showed mitochondrial swelling, singular heterochromatin condensation, and perinuclear electron dense fibers. Interestingly, perinuclear localization of electron dense fibers occurred in greater magnitude in cells treated with Zvad than those with intact caspase activity.
Caspase activation parallel to necroptosis enhances alarmin release from necroptotic cells and promotes the host inflammatory response. Due to the observed differences described above, we also hypothesized that caspase activation parallel to necroptosis modulates the alarmins that are released from dying cells. To test this, we pretreated A549 cells with Zvad or the NSA, and measured levels of known alarmins released into the cell supernatants following challenge with rPly. We found that caspase activity promoted the release of S100A9 and Hsp60 into the cell supernatants (Fig. 4a). To assess the effect of caspase-mediated alarmin release on bystander immune cells, we challenged mock or Zvad treated A549 cells with rPly and collected their supernatants. Subsequently, the activity of residual rPly was neutralized with a monoclonal antibody 26 and the supernatants from these cells were used to challenge human macrophages (THP-1 cells). We found that IL-6 production by human THP-1 macrophages was reduced when exposed to supernatants from cells that had caspase activity inhibited during PFT-induced necroptosis (Fig. 4b).
Validating this in vivo, we observed increased levels of cleaved caspase-2, cleaved caspase-11, and pMLKL in lung samples from mice challenged intratracheally with Sma (Fig. 4c, Supplementary Fig. 8a). Whereas in lung samples from non-human primates with experimental Spn pneumonia we also observe increased levels of cleaved caspase-2, -4 and -10 along with pMLKL (Fig. 4d, Supplementary Fig. 8b). Critically, mice deficient in caspase-2 or caspase-11 that had been challenged with Sma showed a stark reduction in lung consolidation when compared to wildtype, similar to mice deficient in MLKL (Fig. 5a). Immunofluorescent staining against Ly6G, a canonical marker of mature neutrophils, showed a substantial decrease in granulocyte infiltration into the lungs (Fig. 5b,  Supplementary Fig. 8c) that coincided with reduced Interleukin-1α and KC levels in lung homogenates (Fig. 5c). Thus, caspase activation worsened severity of pneumonia.

Discussion
REC play many critical immunological roles during bacterial infection; they serve as a physical barrier, secrete mucus and anti-microbial molecules, and produce chemokines that recruit immune cells to the site of infection 27,28 . In the recent past we have shown that PFTs cause necroptosis of REC and this is, in large part, responsible for the pulmonary injury that occurs during pneumonia 24 . Herein, we demonstrate that upon challenge with PFTs, RECs undergo necroptosis with parallel caspase activity and this potentiates the inflammatory response by enhancing the release of alarmins from dying cells. Our results raise questions and open new lines of investigation that can improve our understanding of basic mechanisms of infectious disease and the host response. For example, what other cell types experience parallel caspase activation to necroptosis following exposure to a PFT and why is this not the case for macrophages? Is there a PFT-mediated necroptosis-caspase signaling intersection where MLKL potentiates caspase activity, albeit these pathways seem independent? These are critical question that now require detailed exploration.
One overriding question is that it is unknown how these caspases are being activated. Caspase-2 can function as an initiator and/or effector caspase, a capability that matches its ancient and conserved function 17 . Caspase-2 might, in some fashion, recognize the binding of PFTs to the plasma membrane or the associated consequences 14 .
In this manner, caspase-2 may aggregate, oligomerize, and autoprocess itself upon PFT challenge of cells 29 . It is important to note that caspase-2 has been shown not to process canonical apoptosis effectors such as caspases -3, -6 or -7 30 . Instead it leads to the induction of apoptosis by increasing the release of death mediators from the mitochondria 30 . In addition, caspase-2 has been shown to be involved in cell death mediated by cellular stress such as DNA damage, ER stress, or metabolic catastrophe 31 . A review of the literature also suggests that targeting of the mitochondria by MLKL, or even by the PFT itself, may activate mitochondrial localized caspase-2 32 . Other substrates released from damaged mitochondria include BID, ICAD, and PKCδ which promote mitochondrial outer membrane permeabilization activation and DNA fragmentation 29 , and all of which are known caspase-2 substrates or initiating signals 29 . For these reasons, we speculate that RIPK activation might further promote caspase activation upon MLKL-driven mitochondrial damage 33 although this association would not be required for necroptosis.
Our data using human cells and baboon samples show that caspase-10 is active upon PFT challenge. Caspase-10 is a homolog of caspase-8, and is highly conserved throughout multiple species, however absent in mice. Caspase-10 has been shown to have a key role in death receptor-mediated apoptosis 34 . Importantly, it has been shown that caspase-10 reduces death-inducing signaling complex (DISC) association and activation of caspase-8, by this it negatively regulates caspase-8 mediated death 35 . Notably, for necroptosis to occur, caspase-8 has to be inhibited, suggesting that upon PFT challenge the activation of caspase-10 might promote necroptosis and at the same time serve as an initiator for other death mediators. In addition, caspase-10 activation through a membrane receptor could also explain the activation of caspase-2. However, the latter has yet to be elucidated.
Caspase-4 is an inflammatory caspase, recently linked to the non-canonical activation of the inflammasome and the release of IL-1β. This is known to occur after cells are exposed to intracellular lipopolysaccharide 36 . PFTs might be a way to introduce LPS into the cytosol during Gram-negative infections. However, we observed caspase-4 activation in cells challenged with purified PFTs (tested to be endotoxin free). One possible reason caspase-4 is activated in an LPS-independent manner may be endoplasmic reticulum stress caused by PFT-induced ion dysregulation, cellular ROS generation, and uncontrolled metabolic changes 5,24 . Endoplasmic reticulum stress has been implicated as an activator of caspase-4 in other model systems 37,38 . Regardless, activation of caspase-4, and presumably the inflammasome, would allow for respiratory epithelial cells to produce and release pro-inflammatory mediators during PFT-mediated necroptosis. Whether this occurs is currently being investigated.
How caspase activity acts on organelle disassembly to impact alarmin release is not clear. Yet, caspase-inhibited REC challenged with rPly had obvious differences in cellular traits during death versus controls, and released less alarmins. That caspase-activation further promoted necroptosis-driven inflammation in the airway was also unambiguous. This was evidenced in vitro, where naïve macrophages treated with cell supernatants from caspase inhibited REC responded less vigorously than controls, as well as in vivo, were we observed that caspase-2 and caspase-11 KO mice had reduced tissue damage and levels of measurable inflammatory chemokines and cytokines. The recognition that caspase activation contributes to pulmonary inflammation marks the involved caspsases as potential targets for adjunct therapeutic intervention along with antimicrobials. In this instance not to clear the pathogen, but to reduce lung injury due to the excessive host response.
Herein, we have discovered that caspase activity, other than caspase-8, can occur parallel to the necroptosis machinery. Caspase activation may occur as an evolutionary conserved response to PFTs during pulmonary infections 14,29,39,40 . In the airway, caspase activity during necroptosis was observed to exacerbate pulmonary inflammation by modulating the physical properties of cell death, which in turn impacted alarmin release. The almost universal presence of PFTs in bacterial pathogens makes this discovery highly significant and identifies caspases as potential targets for novel intervention strategies. Numerous unanswered question remain as to how caspase activation is initiated and whether it is intertwined with necroptosis signaling. Which other cells and diseases might include a version of this form of cell death is also unknown and requires further investigation.

Methods
Ethics Statement. Animal experiments were performed using protocols approved by the Institutional Animal Care and Use Committee at the University of Alabama at Birmingham (Protocol # 20270). Animal care and protocols followed the NIH Guide for the Care and Use of Laboratory Animals. week-old wildtype C57BL/6 mice from Jackson Laboratories (Sacramento, California) were used. MLKL knockout mice in the C57BL/6 background were obtained from Dr. Warren Alexander 41 . Caspase-2 KO (007899) and Caspase-11 KO (024698) mice were obtained from The Jackson Laboratory (Sacramento, California). Oropharyngeal aspiration was performed on each mouse as previously described with an inoculum of 100 μl of a ~1.0 × 10 6 CFU dose 42 . Briefly, after being anesthetized (2% vaporized isoflurane) each mouse was hung upright by its incisors, the tongue gently pulled outward with blunt forceps, and the respective inoculum pipetted into the pharynx accompanied by coverage of the nares to achieved forced inhalation. All mice experiments were performed with protocols approved by the University of Alabama at Birmingham Institutional Animal Care and Use Committee and in agreement with the NIH Guide for the Care and Use of Laboratory Animals. Healthy baboons, with a median age 11 (Interquartile radio [IQR], 10-19) years old were intrabronchialy challenged with Spn (1 × 10 9 CFU) using a bronchoscope 43,44 . Between 5-7 days after infection during severe pneumonia animals were sacrificed and lung tissue was collected. All baboon experiments were performed using protocols approved by the Southwest National Primate Research Center Institutional Animal Care and Use Committee and in agreement with the NIH Guide for the Care and Use of Laboratory Animals.

Infections of Mice and Baboons.
Microscopy and Image Capture. Images were captured using a Zeiss AxioXam MRm Rev3 and/or MRc cameras attached to a Zeiss AxioImager Z1 epifluorescent microscope (Carl Zeiss, Thornwood, NY) or a Leica LMD6 with DFC3000G-1.3-megapixel monochrome camera (Leica Biosystems, Buffalo Grove, IL).

Transmission electron microscopy (TEM).
Published protocols were used for tissue processing 45 . TEM images were obtained at the University of Alabama at Birmingham electron microscopy core, using FEI-Tecnai T12 Spirit 20-120kv, equipped with an AMT digital camera. Silencing RNA and CRISPR-Cas9. Commercially available siRNA targeting RIPK1, RIPK3, MLKL, caspase-1, caspase-2, caspase-4, caspase-8 and caspase-10 (Santa Cruz, Dallas, TX) were used to transfect A549 alveolar epithelial cells following the manufacturer instructions. Commercially available CRISPR-Cas9 plasmids to knock out MLKL (Santa Cruz, Dallas, TX) were used following the manufacturer instructions. Statistical analysis. Prism 7 (Graph Pad Software, La Jolla CA) was used for graph development and statistical analysis. Mann-Whitney U tests were applied for two-group comparisons, and nonparametric Kruskal-Wallis test and Dunn's post hoc analysis were used for multiple-group comparisons.
Data availability. The data generated in the current study are available from the corresponding author on reasonable request.