Hydrogen Peroxide Is Crucial for NLRP3 Inflammasome-Mediated IL-1β Production and Cell Death in Pneumococcal Infections of Bronchial Epithelial Cells

Abstract

Epithelial cells play a crucial role in detection of the pathogens as well as in initiation of the host immune response. Streptococcus pneumoniae (pneumococcus) is a typical colonizer of the human nasopharynx, which can disseminate to the lower respiratory tract and subsequently cause severe invasive diseases such as pneumonia, sepsis, and meningitis. Hydrogen peroxide (H₂O₂) is produced by pneumococci as a product of the pyruvate oxidase SpxB. However, its role as a virulence determinant in pneumococcal infections of the lower respiratory tract is not well understood. In this study, we investigated the role of pneumococcal-derived H₂O₂ in initiating epithelial cell death by analyzing the interplay between 2 key cell death pathways, namely, apoptosis and pyroptosis. We demonstrate that H₂O₂ primes as well as activates the NLRP3 inflammasome and thereby mediates IL-1β production and release. Furthermore, we show that pneumococcal H₂O₂ causes cell death via the activation of both apoptotic as well as pyroptotic pathways which are mediated by the activation of caspase-3/7 and caspase-1, respectively. However, H₂O₂-mediated IL-1β release itself occurs mainly via apoptosis.

Introduction

Streptococcus pneumoniae (pneumococcus), a Gram-positive bacterium and a typical asymptomatic colonizer of the human upper respiratory tract, remains the most commonly identified cause of community-acquired pneumonia [1]. Pneumococci are equipped with a variety of virulence factors allowing them to circumvent the host immune response. In many cases, initial mild manifestations may result in severe invasive diseases such as pneumonia, meningitis, and sepsis [2, 3]. One of the most important and well-studied secreted virulence factors of pneumococci is the pore-forming cytolysin pneumolysin (Ply) [4]. Moreover, pneumococci secrete high amounts of hydrogen peroxide (H₂O₂) as a product of the pneumococcal carbohydrate-metabolizing enzyme pyruvate oxidase SpxB [5]. It is believed that secreted H₂O₂ diffuses through the host cell membrane and causes oxidative damage [6]. Consequently, Spellerberg and colleagues have shown that S. pneumoniae mutants deficient in H₂O₂ production are characterized by a reduced virulence in vivo [5]. Furthermore, it was demonstrated that pneumococci-derived H₂O₂ is a major vasodilator and contributes to the cerebral hyperemia during early stages of meningitis and therapies with antioxidants attenuate the pathophysiological responses [7, 8]. In vitro studies with cells of the respiratory tract showed that both H₂O₂ and Ply attenuate ciliary function by (i) slowing down the ciliary beat frequency and (ii) impairing the structural integrity of human ciliated epithelium [9, 10]. Furthermore, studies on rat alveolar epithelial cells have demonstrated that cell injury induced by S. pneumoniae mutants that lacked Ply activity was comparable to that of the parental strain. H₂O₂ was identified as a major factor responsible for the cell cytotoxicity [11].

Epithelial cells span the entire length of the respiratory tract and were initially presumed to solely serve as a physical barrier for inhaled particles or microorganisms [12]. However, many studies have shown their significance as important immune sentinels [13, 14]. They possess a diverse repertoire of pattern-recognition receptors that assist in detection of various pathogen and danger-associated molecular patterns [15]. Once the healthy host cells encounter a death-inducing stimulus, they can initiate a variety of molecular cell death pathways. Apoptosis is a controlled form of cell death, which features the activation of initiator caspases like caspases-2, 8, 9, and 10. Subsequently, these caspases activate effector caspases such as caspases-3, 6, and 7 [16]. Once activated, the effector caspases cleave various intracellular substrates, including α-fodrin, gelsolin, p21-activated kinase 2, and inhibitor of caspase-activated DNase/DNA fragmentation factor-45 [17]. The final stage of apoptotic cell death includes the formation of apoptotic bodies which expose surface molecules. Professional phagocytes including macrophages recognize the apoptotic bodies and ingest them, making this form of cell death less inflammatory in nature [16].

Pyroptosis is another form of programmed cell death, and in contrast to apoptosis, it is more of a pro-inflammatory nature. Pyroptosis is driven by the activation of a diverse range of inflammasomes. Inflammasomes are multiprotein complexes that assemble based on the detection of danger signals [18]. Of the various inflammasomes, NLRP3 inflammasome is the most unique as it is activated by a wide range of stimuli, such as whole pathogens [19], extracellular ATP [20], and pathogen-associated RNA and toxins [21, 22]. The NLRP3 inflammasome is activated in a two-step process, namely priming and activation [23, 24]. The priming step is essential to increase the intracellular concentrations of NLRP3 and pro-interleukin (IL)-1β. The transcription factor NF-κB is activated upon detection of a priming signal via toll-like receptors, tumor necrosis factor receptors, or interleukin-1 receptors. NF-κB further upregulates the gene expression of NLRP3 and pro-IL-1β. However, the priming step itself does not result in IL-1β secretion. Next, the exposure to an activating signal leads to the assembly and activation of the NLRP3 inflammasome [18]. Active NLRP3 inflammasome converts recruited procaspase-1 to bioactive caspase-1, which further cleaves inactive pro-IL-1β and pro-IL-18 into mature IL-1β and IL-18, respectively. Furthermore, caspase-1 cleaves 50-kDa gasdermin-D (GSDMD) into an active 30-kDa GSDMD-N (N-terminal domain of GSDMD) fragment. GSDMD-N integrates into the cell membrane and forms pores through which IL-1β and IL-18 are secreted into the extracellular space. This form of controlled cell lysis is termed pyroptotic cell death [25, 26, 27].

It must be noted that although previous studies indicate the importance of pneumococcal H₂O₂ in causing cell damage, not much is known about its impact on the activation of cell death pathways. A study by Erttmann and Gekara [28] suggested that H₂O₂ blocks ATP-mediated NLRP3 inflammasome activation in mouse macrophages. However, this study analyzed exclusively very early events (30 min) of pneumococcal infections and was limited to inflammasome activation and not cell death. In contrast, studies on human macrophages suggest that NLRP3 is activated in reactive oxygen species-sensitive manner including H₂O₂ [29, 30]. Furthermore, it was shown that the addition of H₂O₂ alone to human THP-1 macrophages is sufficient to trigger NLRP3-caspase 1-mediated IL-1β release [30].

Here, we aimed to analyze the impact of pneumococci-derived H₂O₂ on cell death pathways. We show that H₂O₂ damages bronchial epithelial cells and induces caspase-1 activation, resulting in IL-1β release. However, GSDMD-N, a cleavage product of caspase-1, was not detected and therefore, IL-1β is mainly released through the apoptotic pathway.

Materials and Methods

Bacterial Strains

Encapsulated S. pneumoniae wild-type strains TIGR4, 19F, and D39 and nonencapsulated S. pneumoniae D39Δcps and its isogenic mutants ΔcpsΔply, ΔcpsΔspxB, and ΔcpsΔplyΔspxB were cultured on Columbia blood agar plates (Oxoid) and in Todd-Hewitt broth supplemented with 0.5% (w/v) yeast extract (THY; Roth). Streptococcus pyogenes strain 5448 was cultured overnight in THY [31]. Staphylococcus aureus strain LUG2012 (USA300 lineage) was cultured overnight at 37°C in casein hydrolysate and yeast extract medium with agitation [32]. All static bacterial cultures were maintained at 37 °C and 5% CO₂. For growth analysis, pneumococcal strains were cultured in THY, and the growth behavior was monitored by measuring the optical density (OD_{600 nm}).

S. pneumoniae Mutant Generation

Genomic DNA of the D39Δcps strain served as a template for the knock-out of the ply gene (SPD1726). Briefly, SPD1725 was amplified using the following primers: 5′-CGGGATCCGCAAATAAAGCAGTAAATGAC-3′ and 5′-GACTCTAGAGGATCCCCGGGTCAACAGACACTCATCCACATTC-3′. The resulting PCR product was digested with SacII, ligated into the pGEM-T Easy vector, and transformed into E. coli DH5α. To delete the ply gene, the recombinant plasmid was used as a template for an inverse PCR with the following primers: 5′-CTATT­TGGGGATCGATTCTCTATCCTCAGG-3′ and 5′-CCGGAAG­CTTAACAGCGTCTACGCTGACTGTATA-3′. The chloramphenicol (Cm) resistance gene cassette was amplified using the following primers: 5′-CGCGAAGCTTCGAAAATTTGTTTGATTTTTAATGG-3′ and 5′-GATAATCGATCGGGTTCCGAGG­CTCAACGTCAA-3′. Both PCR products were digested with ClaI and HindIII, ligated, and transformed into E. coli DH5α. The resulting recombinant plasmid pGEM-TΔply::Cm was transformed into S. pneumoniae D39Δcps as described previously [33, 34]. Recombinant S. pneumoniae clones were selected on blood agar containing chloramphenicol (5 μg/mL). Replacement of the ply gene was analyzed via PCR using the following primers: 5′-CGGGATCCGCAAATAAAGCAGTAAATGAC-3′ and 5′-GCGGTACCCT­AGTCATTTTCTACCTGAG-3′.

Genomic DNA of the D39ΔspxB mutant, which was kindly provided by K. Mühlemann (University of Bern, Bern, Switzerland) [35], served as a template for spxB gene knock-out. Briefly, the mutated spxB gene region containing the erythromycin (ermB) resistance gene cassette was amplified using the following primers: 5′-GGAGAACGTTTCCAATTCTATG-3′ and 5′-GACCGGATT­GCTCCGATCTT-3′. The resulting 3.1-kb fragment was digested with XcmI, ligated into the pGXT plasmid (Promega), and transformed into E. coli DH5α. The purified pGXT-spxB:erm plasmid was used for transformation of D39Δcps and ΔcpsΔply as described previously [33, 34]. Bacteria were grown for 2 h at 37°C and plated on blood agar plates containing 5 μg/mL erythromycin. The resulting S. pneumoniae D39 spxB-deficient mutants were screened by colony PCR using the following primers: 5′-GCGCGCTAGCACTCAAGGGAAAATTACTGC-3′ and 5′-GCGCGAGCTCTTATTTAATTGCGCGTGATTG-3′. Gene organization in D39Δcps and respective mutant strains are depicted in online suppl. Figure 1; see www.karger.com/doi/10.1159/000517855 for all online suppl. material.

In addition, the lack of SpxB and Ply was confirmed via Western blot analyses. Briefly, D39Δcps mutants were cultivated in THY medium until the late exponential growth phase was reached, bacteria were lysed, and protein extracts were separated via SDS-PAGE and transferred onto a nitrocellulose membrane. For detection of SpxB and Ply [36], mouse polyclonal antibodies (in-house production by the laboratory of S.H.) were used.

Measurement of H₂O₂ Production by Pneumococci

S. pneumoniae strains were cultured in a chemically defined medium (RPMI1640 containing 2 mM L-glutamine medium [HyClone] supplemented with 30.5 mM glucose, 0.65 mM uracil, 0.27 mM adenine, 1.1 mM glycine, 0.24 mM choline chloride, 1.7 mM NaH₂PO₄⋅H₂O, 3.8 mM Na₂HPO₄⋅2H₂O, and 27 mM NaHCO₃) at 37 °C and 5% CO₂ without agitation (OD_{600 nm} of 0.80–1.0) [37]. Bacteria were pelleted at 10,000 × g for 2 min and the culture supernatants were collected. Five hundred microliters of the culture supernatant was mixed with 500 μL Titanium-(IV)-oxide-sulfate-sulfuric acid solution (Sigma-Aldrich). The samples were incubated at room temperature in the dark for 5 min and the OD of the resulting solution was determined at 407 nm. A standard curve containing defined concentrations of H₂O₂ was used to determine H₂O₂ concentrations in bacterial culture supernatants.

Infections, Stimulations, and Cell Culture Conditions

Bronchial epithelial 16HBE14o-cells (16HBE) were used in all infection and stimulation experiments. Cells were cultured in a minimum essential medium (Gibco) supplemented with 10% (v/v) fetal bovine serum (Life Technologies), 2 mM L-glutamine (Invitrogen), 10 mM HEPES, and 1% (v/v) minimal essential amino acids (both GE Healthcare) in fibronectin-coated flasks (Corning) at 37°C and 5% CO₂ atmosphere.

For pneumococcal infections or H₂O₂ stimulations, 3 × 10⁵ cells/well were seeded in 24-well tissue culture plates (Greiner) and maintained at 37°C and 5% CO₂ atmosphere. All experiments were performed in minimum essential media. The cells were either infected with bacteria (MOI 50) or stimulated with different concentrations of H₂O₂ for 2, 4, or 6 h. To activate NLRP3 inflammasome, additional experiments included 4 h of 100 ng/mL LPS or 100 ng/mL TNFα cell stimulations prior to an infection or H₂O₂ stimulation. To inactivate H₂O₂-mediated actions, 300 U/mL of catalase was added to the experiments.

To assess bacterial infectivity, 16HBE cells were infected with pneumococci for 2, 4, or 6 h. At indicated time points, the cells were washed, detached, lysed, and the viable counts of pneumococci (colony-forming units) released from lysed cells were determined by plating serial dilutions on blood agar.

Caspase-1 and Caspase-3/7 Activities

All infection and stimulation experiments were performed as described above. Four hours post infections/stimulation, active caspase-1 and caspase-3/7 were labeled with cell-permeable fluorescent probes FAM-YVAD-FMK FLICA or FAM-DEVD-FMK FLICA (both Bio-Rad) according to the manufacturer's instructions. The cell nuclei were stained with Nuclear-ID Red DNA stain (Enzo Life Sciences). The staining was visualized using an Axio Observer Z1 microscope (Zeiss). To quantify the amount of caspase-positive cells, random images were taken and caspase-activated cells are presented as a percentage of caspase-1- or caspase-3/7-active cells in relation to the total cell number.

Cytotoxicity Analysis and IL-1β ELISA

To determine the cytotoxic responses, supernatants of bacterial infections or H₂O₂ cell stimulations were collected and LDH release from cells was quantified using CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega) according to the manufacturer's instructions. The IL-1β release was quantified using Human IL-1 beta ELISA (R&D Systems) according to the manufacturer's instructions.

Inhibition Studies

The irreversible caspase-1 inhibitor Ac-YVAD-cmk (Invivogen), caspase-3/7 inhibitor Ac-DEVD-CHO (Enzo Life Sciences), and NLRP3 inhibitor MCC950 (Invivogen) were used in this study. All inhibitors were dissolved in DMSO and were added to the 16HBE cells 1 h prior to the other stimuli. The following concentrations were used: 50 μM (Ac-YVAD-cmk and Ac-DEVD-CHO) and 200 nM (MCC950). Cell treated with 1% DMSO served as a vehicle control.

SDS-PAGE and Immuno-Detection of Gasdermin D

16HBE cells (1 × 10⁶/well) were seeded in 6-well tissue culture plate and stimulated with 150 μM H₂O₂, 150 μM H₂O₂, and 300 U/mL catalase or left untreated. After 6 h of incubation, cells were washed and lysed using NP-40 lysis buffer (150 mM NaCl, 1% [v/v] NP40, 50 mM Tris-HCl, pH 8.0-containing protease inhibitor mixture [Cell Signaling Technology]). Samples were normalized to equal amounts of protein (20 μg) and boiled in sample buffer (100 mM Tris [pH 6.8], 2% [w/v] SDS, 10% [v/v] β-mercaptoethanol, 20% [v/v] glycerol, and 0.05% [w/v] bromophenol blue). As molecular mass marker, prestained protein standards (Bio-Rad) were used. The samples were separated by 12% SDS-PAGE and transferred to a PVDF membrane. The membranes were blocked with 3% (v/v) skim milk prior to primary antibody incubations. Antibody incubations were performed according to manufacturer's guidelines. The following antibodies were used: Gasdermin D (E8G3F) Rabbit mAb, Cleaved Gasdermin D (E7H9G) Rabbit mAb, β-Actin (D6A8) Rabbit mAb (all from Cell Signaling Technology), and secondary anti-rabbit IgG horseradish peroxidase linked Fab fragment (GE Healthcare). Quantification of the Western blots was performed using ImageJ software.

Quantitative Reverse Transcription PCR Analysis

Total RNA was isolated from cells using RiboPure RNA Purification Kit (Ambion) according to manufacturer's instructions. cDNA synthesis was performed using Superscript first-strand synthesis system for RT-PCR (Invitrogen). Quantitative reverse transcription PCR was performed using the SYBR GreenER Kit (Bio-Rad). All samples were normalized to levels of β-actin transcription. The following primers were used: NLRP3-for, 5′-GCAAAAAGAGATGAGCCGAAGT-3′; NLRP3-rev, 5′-GCTGTCTTCC­TGGCATATCACA-3′; IL-1β-for, 5′-CACGATGCACCTGTA­CGATCA-3′; IL-1β-rev, 5′-GTTGCTCCATATCCTGTCCCT-3′; h-betaAct-for, 5′-CTCTTCCAGCCTTCCTTCCT-3′; h-betaAct-rev, 5′-AGCACTGTGTTGGCGTACAG-3′.

Statistical Analyses

Statistics were performed using GraphPad Prism version 7.0 (GraphPad). Statistical significance of differences was determined using Kruskal-Wallis test with Dunn's post-test. A p value <0.05 was considered significant.

Results

H₂O₂ Produced by Pneumococci Induces Cell Death and IL-1β Release

Infections of rat alveolar epithelial cells have demonstrated that Ply-deficient S. pneumoniae mutants are as cytotoxic as their parental strains [11]. To assess whether H₂O₂ production by pneumococci can be directly linked to the cell-damaging events and potentially to inflammasome activation, human bronchial epithelial cells were infected with S. pneumoniae D39∆cps in the presence or absence of catalase. Potential catalase-mediated bactericidal effects were excluded (online suppl. Fig. 1). Since NLRP3 inflammasome activation involves a priming step for NF-κB activation and subsequent transcription of pro-IL-1β and NLRP3 [38], cells were also stimulated with LPS or TNFα prior to infections. Four and six hours postinfection, cytotoxicity as well as IL-1β release was determined (Fig. 1). While 4 h of pneumococcal infection remained almost nontoxic to the cells, a significant increase in cytotoxicity was observed at 6 h postinfection (Fig. 1a). The cytotoxicity was significantly diminished in infections which were supplemented with catalase (Fig. 1a). IL-1β was only detected in cell culture supernatants at 6-h postinfection and only in congruence with the high cell cytotoxicity (Fig. 1b). Both assessed parameters, cytotoxicity and IL-1β release, did not require a priming step.

Fig. 1.

H₂O₂-mediated cytotoxicity and IL-1β release by human bronchial epithelial cells. Unprimed, LPS-, or TNF-primed human bronchial epithelial cells were infected with D39∆cps at MOI 50 in the presence or absence of catalase for 4 and 6 h. Cytotoxicity (a) and IL-1β release (b) were evaluated at the indicated time points. Bars (a) denote mean values ± standard deviation (SD). The data in (b) are displayed as box plots. The level of significance was determined using Kruskal-Wallis test with Dunn's post-test (n ≥ 4; *p < 0.05; **p < 0.01; ***p < 0.001).

Active SpxB and Resulting H₂O₂ Production Are Responsible for the Cell Cytotoxicity

Pneumococci exploit pyruvate oxidase enzyme SpxB to produce H₂O₂ [5]. To study the impact of H₂O₂ on cell death, isogenic spxB-deficient mutants were constructed. To exclude Ply-mediated cytotoxic effects, ply-deficient mutants were also included in this study. Single ply and spxB as well as double mutants were generated in D39∆cps background. In addition to the nucleic acid techniques (Methods section), the knock-outs were also verified via immunoblots targeting Ply and SpxB proteins (online suppl. Fig. 3). Next, functional assays were performed. Growth analyses confirmed that single as well as double knock-outs of spxB or/and ply in S. pneumoniae D39∆cps did not affect bacterial growth behavior in complex THY media (Fig. 2a). Furthermore, H₂O₂ production by all 4 strains was verified. To exclude potential capsule-mediated effects on H₂O₂ release, encapsulated wild-type strains were also included in the analysis. SpxB-deficient mutants (∆cps∆spxB and ∆cps∆ply∆spxB) produced significantly lower amounts of H₂O₂ as compared to D39∆cps or D39∆cps∆ply mutants (Fig. 2b). The presence or absence of capsule did not play a role in pneumococcal H₂O₂ production and release (Fig. 2b).

Fig. 2.

Pneumococcal H₂O₂ is responsible for bronchial epithelial cell lysis. a Monitoring of growth behavior of S. pneumoniae D39∆cps, ∆cps∆ply, ∆cps∆spxB, and ∆cps∆ply ∆spxB mutant strains in THY medium (n = 3). b H₂O₂ production by pneumococci was determined by colorimetric analysis of bacterial culture supernatants (n = 4). 16HBE cells were infected with pneumococcal strains, and bacterial infectivity (c, e) and cytotoxicity toward the cells (d, f) were assessed at indicated time points (n ≥ 4). The data in (b, c, e) are displayed as box plots. Bars in (d, f) denote mean values ± SD. The level of significance was determined using Kruskal-Wallis test with Dunn's post-test (n ≥ 4; n.s., not significant; *p < 0.05; **p < 0.01; ***p < 0.001). CFU, colony-forming unit; SD, standard deviation; THY, Todd-Hewitt broth supplemented with 0.5% (w/v) yeast extract.

To determine the role of an active SpxB in pneumococcal infections of bronchial epithelial cells, 16HBE cells were infected with pneumococcal wild-type strains as well as D39∆cps and its isogenic mutants lacking spxB, ply, or both genes for 2, 4, or 6 h. Total bacterial infectivity (Fig. 2c, e) as well as cytotoxicity toward the cells (Fig. 2d, f) was assessed. After an initial increase of bacterial counts from 2 to 4 h of infection, a significant drop in infection rates for unencapsulated spxB-positive strains was observed (Fig. 2c). In contrast, the bacterial counts of spxB mutants remained at a constant level between 4 and 6 h of infection (Fig. 2c). Analyses of the cell viability revealed that 16HBE infections with pneumococci harboring an active spxB gene were characterized by a time-dependent cell cytotoxicity (Fig. 2d). In contrast, mutant strains lacking spxB were not cytotoxic (Fig. 2d). To assess if direct contact is required for H₂O₂-mediated damage of the cells, infections with encapsulated wild-type strains, which are known to adhere to a lesser extent as compared to the nonencapsulated strains, were also performed. After 4 h of infection, total infectivity of D39 and TIGR4 was approximately 10-fold lower compared to the D39∆cps strain (Fig. 2c, e). 19F strain adhered to the cells even at much lower rates (Fig. 2e). However, the encapsulated wild-type strains were as cytotoxic as the nonencapsulated mutants (Fig. 2d, f). Taken together, these results indicate that cell cytotoxicity at the early stages of infection can be primarily attributed to the influence of H₂O₂ produced by pneumococci.

H₂O₂ Activates Caspase-1 and Caspase-3/7 in 16HBE Cells

Both H₂O₂-mediated cytotoxicity toward 16HBE cells and subsequent IL-1β release suggest the involvement of NLRP3 inflammasome in pneumococcal infections. To assess the NLRP3 axis prior to the excessive cell lysis, unprimed 16HBE cells were infected with pneumococcal strains for 4 h and caspase-1 activation was determined via fluorescent probing with FAM-YVAD-FMK FLICA. In addition, caspase-3/7 activation as a marker for apo­ptosis was analyzed. Although initial data indicated that LPS or TNF priming is dispensable, LPS was kept through the entire series of experiments as a generally accepted inflammasome model control. Only minor cytotoxic events were observed in infections with D39 SpxB-expressing strains (Fig. 3a). Activated caspase-1 was detected in 10–20% of the infected cells (Fig. 3b, d). In contrast, spxB mutants did not cause cell lysis (Fig. 3a) and 16HBE cells remained negative for active caspase-1 (Fig. 3b, d). Analyses of active caspase-3/7 revealed a similar pattern. While infections with SpxB-expressing D39 strains activated caspase-3/7, infections with spxB mutants did not (Fig. 3c, d). The observed infection phenotype was independent of priming (Fig. 3 and online suppl. Fig. 4a).

Fig. 3.

S. pneumoniae strains with functional SpxB activate caspase-1 and caspase-3/7. Unprimed or primed 16HBE cells were infected with indicated pneumococcal strains for 4 h and cytotoxicity (a) as well as caspase-1 and caspase-3/7 (b–d) activation was assessed. The infected cells were stained using fluorescent inhibitor probe FAM-YVAD-FMK (b, d left panel) and FAM-DEVD-FMK (c, d right panel) to microscopically visualize active caspase-1 and caspase-3/7, respectively. Nuclear-ID stain was used to visualize cell nuclei. Caspase-1- and caspase-3/7-positive cells were counted and are presented as a percentage of positive cells in relation to the total number of cells (b, c). Bars (a) denote mean values ± SD. The data in (b, c) are displayed as box plots. The level of significance was determined using Kruskal-Wallis test with Dunn's post-test (n = 4; *p < 0.05; **p < 0.01; ***p < 0.001). SD, standard deviation.

To confirm that the activation of caspases and thereby an increase in cytotoxicity were associated with H₂O₂, 16HBE cells were stimulated with pure H₂O₂ for 4 or 6 h. A dose-dependent increase in cytotoxicity toward the cells was observed (online suppl. Fig. 5). Addition of catalase to the stimulations reverted the cytolytic events (online suppl. Fig. 5). H₂O₂ concentrations of 100 and 150 μM, which were approximately of equivalent cytotoxicity as compared to the D39∆cps strain infections, were used to study the activation of caspases in primed and unprimed cells. Minor cytotoxicity was only observed in stimulations with 150 μM H₂O₂ (Fig. 4a). Cells with active caspase-1 were mostly detected in these stimulations (Fig. 4b, d). Furthermore, caspase-3/7 activation followed exactly the same pattern (Fig. 4c, d; online suppl. Fig. 4b). Addition of catalase to the stimulations reduced the H₂O₂-mediated cytotoxicity and activation of caspase-1 and caspase-3/7 (Fig. 4 and online suppl. Fig. 5). These results confirm that H₂O₂ induces cell death and caspase-1 as well as caspase-3/7 activation in 16HBE cells.

Fig. 4.

H₂O₂ activates caspase-1 and caspase-3/7 in bronchial epithelial cells. Unprimed or primed 16HBE cells were stimulated with 150 and 100 μM H₂O₂ in the presence or absence of catalase for 4 h and cytotoxicity (a) as well as caspase-1 and caspase-3/7 (b–d) activation was assessed. The stimulated cells were stained using fluorescent inhibitor probe FAM-YVAD-FMK (b, d) and FAM-DEVD-FMK (c, d) to microscopically visualize active caspase-1 and caspase-3/7, respectively. Nuclear-ID stain was used to visualize cell nuclei. Caspase-1- and caspase-3/7-positive cells were counted and are presented as a percentage of positive cells in relation to the total number of cells (b, c). Bars (a) denote mean values ± SD. The data in (b, c) are displayed as box plots. The level of significance was determined using Kruskal Wallis test with Dunn's post-test (n = 4; *p < 0.05; **p < 0.01; ***p < 0.001). SD, standard deviation.

H₂O₂ Activates Inflammasome Signaling Resulting in IL-1β Release

Since active caspase-1 cleaves pro-IL-1β into mature form, which is mostly released via pyroptosis, IL-1β release post bacterial infections or H₂O₂ stimulations was determined in culture supernatants. To confirm that the priming step is dispensable, TNF and LPS stimulation were performed prior to infections. In line with the cytotoxicity data (Fig. 2d), IL-1β was exclusively detected after 6 h of infections with SpxB-positive strains (Fig. 5a; online suppl. Fig. 6a). H₂O₂ stimulations of 16HBE cells confirmed this phenotype. The highest amounts of IL-1β were detected 6 h post stimulations (Fig. 5b; online suppl. Fig. 6b). Addition of catalase neutralized H₂O₂-mediated IL-1β release.

Fig. 5.

SpxB-mediated H₂O₂ production induces IL-1β release after 6 h of stimulation. Unprimed, LPS-, or TNF-primed human bronchial epithelial cells were infected with D39∆cps and the isogenic mutants at MOI 50 (a, c, d), or stimulated with 150 and 100 μM H₂O₂ in the presence or absence of catalase (b–d). IL-1β release was evaluated at 6 h post infection (a) or stimulation (b). Relative mRNA expression of NLRP3 (c) and pro-IL-1β (d) was quantified in infected and stimulated cells. The data in (a, b) are displayed as box plots. Bars in (c, d) denote mean values ± SD. The level of significance was determined using Kruskal-Wallis test with Dunn's post-test (n ≥ 4; *p < 0.05; **p < 0.01; ***p < 0.001). SD, standard deviation.

Our data clearly indicate that H₂O₂-mediated cytotoxicity and IL-1β release do not require a priming step. To further delineate the role of H₂O₂ in NLRP3 inflammasome activation, NLRP3 and pro-IL-1β mRNA expression in the infected and H₂O₂-stimulated cells was analyzed (Fig. 5c, d). Cells stimulated with LPS and ATP served as a positive control. Upregulation in the expression of both genes, NLRP3 and pro-IL-1β, was detected in cells infected with D39∆cps and D39∆cps∆ply as compared to the spxB mutant infections (Fig. 5c, d). Furthermore, both genes were upregulated in response to H₂O₂ stimulations (Fig. 5c, d). Addition of catalase reversed this effect. These results confirm that H₂O₂ alone is sufficient for priming and activating the NLRP3 inflammasome in 16HBE cells.

IL-1β Release Caused by H₂O₂ Is a Result of NLRP3 Inflammasome Activation but Not Pyroptosis

Caspase-1 activation, cell death, and subsequent IL-1β release are the hallmarks of inflammasome signaling. To confirm that these H₂O₂-induced responses were indeed mediated via the inflammasome pathway, cells were pretreated with caspase-1 inhibitor Ac-YVAD-cmk or an inhibitor of NLRP3 conformational change and oligomerization MCC950. Since caspase-3/7 activation was also observed, Ac-DEVD-CHO was used to block apoptosis. As previously shown, both D39∆cps infections and H₂O₂ stimulations of cells for 6 h resulted in cell death and subsequent IL-1β release (Fig. 6a-d). Blocking of NLRP3, caspase-1, caspase-3/7, or both caspases resulted in significantly diminished cell death and IL-1β amounts in cell culture supernatants (Fig. 6a-d). These results indicate that H₂O₂ leads to NLRP3 inflammasome activation and IL-1β production; however, the final IL-1β release is also dependent on the apoptotic cell death pathway. To verify these results, it was essential to determine the GSDMD cleavage products, including full-length GSDMD (50 kDa) and fragments of GSDMD cleaved by caspase-1 (30 kDa) or caspase-3/7 (43 kDa) (Fig. 6e, f), in infected and H₂O₂-stimulated cells. While the 43-kDa GSDMD is inactive, the 30-kDa GSDMN-N is mainly responsible for the cell membrane perforation and pyroptosis. Western blot analyses revealed that full length and small amounts of the inactivated 43-kDa fragment of GSDMD are present in both infected and H₂O₂-stimulated cells (Fig. 6e, f; online suppl. Fig 7). No or traces of the 30-kDa product were detected in 5 or 4 h infections, respectively (Fig. 6e; online suppl. Fig 7). However, 30-kDa GSDMD-N was not detected in any of the pneumococcal infections if a specific antibody was used (online suppl. Fig. 7). To show that GSDMD can be cleaved to GSDMD-N (30 kDa) in 16HBE cells, infections with S. pyogenes 5448 and S. aureus LUG2012 were also performed (Fig. 6g). While GSDMD-N was not detected in 5448 infections, it was readily detectable in LUG2012 infections (Fig. 6g). Lack of the active 30-kDa fragment in pneumococcal infections and inhibition studies suggest that the IL-1β release is independent of GSDMD.

Fig. 6.

H₂O₂-mediated IL-1β release is a result of NLRP3 inflammasome activation. Unprimed human bronchial epithelial cells were infected with D39∆cps at MOI 50 (a, b), or stimulated with 150 μM H₂O₂ (c, d). Cells were treated with Ac-YVAD-cmk, Ac-DEVD-CHO, and MCC950, 1 h prior to the other stimuli. Cytotoxicity (a, c) and IL-1β release (b, d) were evaluated at the 6-h time point. The data are displayed as box plots. The level of significance was determined using Kruskal-Wallis test with Dunn's post-test (n ≥ 4; *p < 0.05; **p < 0.01; ***p < 0.001). Western blot analyses of D39∆cps-infected (e) and H₂O₂-stimulated (f) 16HBE cells. e 16HBE cells were infected with different D39 mutant strains or (f) stimulated with H₂O₂ for indicated time points, cells were lysed, and 20 μg of total protein was separated via SDS-PAGE. Representative images of GSDMD and β-actin as a loading control from 5 independent experiments are displayed (n = 5). g Western blot analyses of Streptococcus pyogenes 5448 or Staphylococcus aureus LUG2012-infected 16HBE cells. Original uncropped versions of the blots are shown in online suppl. Fig. 7. GSDMD, gasdermin-D.

Discussion/Conclusion

Epithelial cells of the respiratory tract are the first responders to an infection or pathogenic stimuli. They respond by secreting various effector molecules such as cytokines and antimicrobial peptides that facilitate an influx of professional phagocytes to engulf the pathogens or infected and dying cells [39]. Here, we report that pneumococci-derived H₂O₂ induces epithelial cell cytotoxicity by priming the cells as well as activating the NLRP3 inflammasome, resulting in caspase-1 activation and IL-1β release. Nevertheless, the final IL-1β release is mainly a result of an apoptotic cell death.

Pneumococci are frequent asymptomatic colonizers of the nasopharyngeal cavity and in certain stress condition; they can disseminate to the lower respiratory tract and cause invasive diseases, including pneumonia [40]. Two major secreted cytotoxins, Ply and H₂O₂, are mainly responsible for the lytic injury of a wide range of human cells [41, 42]. As a pore-forming toxin, Ply is implicated in many cell-damaging processes [43, 44, 45]. Previous studies have shown that Ply activates the NLRP3 inflammasome [46]. However, it is noteworthy to mention that Ply is not actively secreted by S. pneumoniae due to a lack of an N-terminal signal peptide needed for secretion [47]. Instead, Ply is released via autolysis at the late stage of growth [48]. It has also been shown that Ply-deficient bacteria are as cytotoxic as their respective wild-type strains in the early phase of infection [4]. Therefore, it is plausible to expect that Ply-specific inflammasome activation and cell death occur at the later stages of infection. In contrast to Ply, only 1 study analyzed H₂O₂-mediated cell-damaging events in pneumococcal infections. A previous study showed that pneumococci-derived H₂O₂ suppresses ATP-mediated inflammasome activation in mouse bone marrow-derived macrophages. The authors proposed that pneumococci utilize H₂O₂ to suppress the immune system and through this mechanism, the bacteria might colonize and coexist with the host [28]. In the present study, the cytotoxic impact of pneumococcal H₂O₂ on bronchial epithelial cells was analyzed. In contrast to the study mentioned above, infections with H₂O₂-producing pneumococci were highly cytotoxic to the epithelial cells. This result is in line with other studies showing that H₂O₂ is of cytotoxic nature [49, 50]. Current literature implicates pneumococcal H₂O₂ as a major cause of lung tissue damage through the apoptotic pathways [51, 52]. However, little is known about its inflammasome-activating properties and pyroptosis. Notably, we find that caspase-1 and caspase-3/7 are activated by H₂O₂, indicating that inflammasome activation occurs and potentially both apoptotic and pyroptotic cell death pathways are simultaneously initiated in the presence of H₂O₂. Various inflammasome pathways have been studied so far, of which the NLRP3 inflammasome is the most well characterized. It can be activated by a wide range of stimuli, making it very diverse in nature [53]. In contrast to monocytes, the NLRP3 and pro-IL-1β concentrations in resting epithelial cells are inadequate to initiate activation of the inflammasome, making the priming step mandatory [54]. Here, we used LPS and TNF because they are commonly accepted priming agents in inflammasome-related studies [23, 28, 54]. During a pneumococcal infection in the lung, the role of priming can be also provided by IL-1β itself [54, 55]. Furthermore, pneumococcal lipoproteins can mediate a TLR2 response [56] and could also play a role in priming. However, our results demonstrate that the priming is dispensable for H₂O₂-mediated IL-1β production in pneumococcal infections. This fact is further supported by previous studies which have shown that H₂O₂ can directly activate NF-κB through the NF-κB-inducing kinase [57, 58, 59].

Furthermore, our data show that caspase-1 or NLRP3 inhibition significantly reduced the H₂O₂-mediated cytolytic injury as well as IL-1β release, indicating that H₂O₂ is a potent activator of the NLRP3 inflammasome. However, inhibition of the apoptotic pathway followed exactly the same pattern. In support of this, 2 studies demonstrated that H₂O₂ induces mitochondrial damage which in turn triggers both apoptotic signals and NLRP3-dependent IL-1β secretion [51, 60]. In the next step, we aimed to analyze whether H₂O₂ mediates pyroptosis. The key molecule in this process is GSDMD. For many years, a linear model of pyroptosis was proposed. Activated caspase-1 cleaves both pro-IL-1β and GSDMD to IL-1β and GSDMD-N, respectively. The GSDMD-N perforates the lipid bilayer and IL-1β is released through these pores [25]. Although our data demonstrate that H₂O₂ induces caspase-1 activation and IL-1β production, IL-1β release itself occurs in a GSDMD-independent fashion. Only an inactive form of GSDMD (43 kDa), a product of caspase-3 cleavage [61], was detected post H₂O₂ stimulations. In line with this, it was suggested that prolonged stimulation of the inflammasome can potentially result in GSDMD-independent cell death and IL-1β release [62].

Nevertheless, cross-talks between the various cell death pathways in a complex cellular environment can still be expected. Recently, it was demonstrated that caspase-1 can also redundantly promote activation of apoptotic executioner caspase-3 and caspase-7 in macrophages [63]. Furthermore, caspase-3 can activate GSDME/DFNA5 to form membrane pores to induce secondary necrotic/pyroptotic cell death [64, 65]. Even more, cross-talks between pyroptosis and necroptosis have also been reported [66]. Zhao and colleagues have shown that H₂O₂ can induce necrosis through the RIP1/RIP3-PARP-AIF pathway [67]. Furthermore, mixed lineage kinase domain-like, a marker for necroptosis, is able to activate the NLRP3 inflammasome and simultaneously rupture the cell membrane, resulting in GSDMD-independent IL-1β release [66].

In conclusion, this study strongly supports the concept that pneumococci-derived H₂O₂ has a detrimental influence on human bronchial epithelial cells. Although IL-1β release is mainly dependent on the apoptotic pathway, H₂O₂ mediates caspase-1-dependent IL-1β production. It is noteworthy to mention that in a complex cellular environment, it is plausible to expect cross-talks between the various cell death pathways. Therefore, to delineate the complex nature of cell death induced by H₂O₂, further experimental studies pertaining to other cellular pathways such as necroptosis are warranted.

Statement of Ethics

The article is exempt from ethical committee approval as all work was performed in vitro using a cell line and no primary human samples were used.

Conflict of Interest Statement

The authors declare no conflict of interest.

Funding Sources

This research was supported by the Federal Excellence Initiative of Mecklenburg Western Pomerania and European Social Fund Grant KoInfekt (ESF_14-BM-A55-0001_16 to S.H.) and the German Research Foundation (DFG; grant no.: 407176682 to N.S.).

Author Contributions

S.S., S.H., and N.S. designed the study. S.S., L.H.J., P.S., and G.B. performed the experiments. S.S., L.H.J., and N.S. analyzed the data. S.S. and N.S. wrote the manuscript. S.S., L.J., P.S., G.B., S.H., and N.S. read, edited, and reviewed the manuscript.

Supplementary Material

Supplementary data