Necroptosis promotes Staphylococcus aureus clearance by inhibiting excessive inflammatory signaling

MLKL contributes to the clearance of S. aureus from infected skin

Murine models of skin infection provide the opportunity to monitor both inflammation and bacterial killing in an in vivo setting (Miller et al., 2006). To determine the role of necroptosis during skin infection, we compared the responses of Mlkl^−/− mice, which are unable to undergo necroptosis (Murphy et al., 2013), with wild type (WT) mice after a subcutaneous inoculation of USA300 S. aureus (Figure 1). In comparison to WT mice, there were significantly higher bacterial counts and more tissue damage in the Mlkl^−/− mice (Figures 1A–C). We confirmed that S. aureus infection activates necroptosis by documenting phosphorylated MLKL in infected human keratinocytes (Figure S1) and demonstrating that S. aureus-induced cytotoxicity was decreased significantly by inhibitors of MLKL (NSA) in human keratinocytes and primary cultures (Figure 1D, 1E) and by siRNA knockdown of MLKL (Figure 1F, 1G) in keratinocytes. Thus, the ability of host cells to undergo necroptosis contributed to pathogen clearance.

We next determined whether excessive inflammation was contributing to the defective bacterial clearance exhibited by the Mlkl^−/− mice. We compared the ability of these Mlkl^−/− mice to activate proinflammatory cytokine production and to recruit the immune cells necessary to clear S. aureus infection. No differences between the WT, Mlkl^−/− mice were observed after 1 day of infection either in the clearance of S. aureus or recruitment of immune cells (Figures S2A, S2B). However, by day 5, Mlkl^−/− mice had significantly more neutrophils, macrophages and γδ T cells recruited than WT mice at the site of infection (Figures 1H–J, Table S1). As early as one day post infection, Mlkl^−/− mice had increased T cell cytokine expression (IP-10/CXCL10, Rantes and MIG) (Figure S2C, Table S2), cytokines that mediate inflammatory damage in the skin (McLoughlin et al., 2006). Consistent with the heightened immune cell recruitment, by day 5, Mlkl^−/− mice had significantly increased proinflammatory cytokines (IL-6, TNF, IL-1β, KC, and MIP1α) (Figures 1K, Table S1).

The histopathological responses to infection also were different. Mlkl^−/− mice exhibited increased necrosis and adiposity, a marker of inflammation (Rajala and Scherer, 2003; Schaffler and Scholmerich, 2010), in comparison to WT infected controls (Figure 1L). Mlkl^−/− mice had hyperkeratosis and more immune cell infiltration as compared with the WT mice (Figure 1L). As increased IL-1β was observed in Mlkl^−/− infections, tissue lysates were screened for the presence of active caspase-1 (Figure 1M). Substantially more caspase-1 p20 was found at both baseline and following infection in Mlkl^−/− than in WT mice. The Mlkl^−/− mice displayed increased caspase-1 activity, increased generation of proinflammatory cytokines including IL-1β, and more neutrophils, but were nonetheless unable to clear S. aureus as efficiently as the WT mice. Together, these results suggest that necroptosis participates in the regulation of excessive inflammation that causes tissue damage during S. aureus skin infection.

Lack of caspases-1/4 is detrimental in S. aureus skin infection

We wanted to understand whether caspase-1 activation is beneficial during S. aureus skin infection. Production of IL-1β, which is processed by caspase-1, is critical for clearance of S. aureus from skin (Miller et al., 2006; Sokolovska et al., 2013; Soong et al., 2012). The contribution of the inflammasome to bacterial clearance was tested in Casp1/4^−/− mice infected with S. aureus (Figure 2). The Casp1/4^−/− knockout mice exhibited significantly decreased bacterial clearance and increased lesion sizes at 5 days after infection (Figure 2A–C). Similar to Mlkl^−/− mice, Casp1/4^−/− mice had increased numbers of neutrophils and γδ T cells but not macrophages at the site of infection (Figure 2D, E, F). Unlike Mlkl^−/− mice, which exhibited an increased in many pro-inflammatory cytokines, Casp1/4^−/− mice had increased KC only and slightly but not significantly less IL-1β post-infection (Figure 2G, Table S3). Thus, caspase 1/4 activity is necessary for S. aureus clearance, and contributes substantially to the generation of IL-1β.

Pharmacological inhibition of RIPK1 impairs clearance and increases inflammation

To verify participation of necroptosis in staphylococcal clearance, we used a pharmacological inhibitor of the RIPK1 kinase activity necrostatin-1s (Nec-1s) (Degterev et al., 2008; Takahashi et al., 2012), which also obviates unanticipated effects of the germline mutation in the Mlkl^−/− mice. After documenting that Nec-1s blocked S. aureus-induced keratinocyte cell death in vitro (Figure 3A), we treated WT mice with Nec-1s and monitored their response to S. aureus skin infection (Figure 3B–D). Similar to the Mlkl^−/− mice, the Nec-1s-treated mice had significantly larger skin lesions, greater numbers of S. aureus recovered and greater recruitment of inflammatory cells (Figures 3E–G). The overall pattern of cytokine induction suggested that Nec-1s-treated mice had increased levels of proinflammatory cytokines over the DMSO control (Figures 3H, Table S4). RIPK1 inhibition was associated with increased inflammation seen in histological sections and increased adiposity (Figure 3I). Thus, inhibition of necroptosis by Nec-1s-treatment or Mlkl^−/− deletion leads to increased inflammation but impaired staphylococcal killing.

Ripk3^−/− mice have enhanced S. aureus clearance from the skin

RIPK3 is a target of RIPK1 and a major effector of necroptosis (Christofferson et al., 2014; Vanden Berghe et al., 2014b). However, in contrast to MLKL which functions as an effector of necroptosis alone, RIPK3 can directly activate apoptosis and stimulate the NLRP3 inflammasome to generate IL-1β (Newton and Manning, 2016). After documenting that an inhibitor of RIPK3, GSK’872, blocked S. aureus induced cytotoxicity in keratinocytes (Figure 4A), we wanted to understand the role of RIPK3 during S. aureus skin infection. Using the model of cutaneous S. aureus infection, we noted significantly improved outcomes in the Ripk3^−/− mice with decreased staphylococcal burden (Figure 4B–D). Significantly fewer inflammatory cells and decreased cytokine production, including IL-1β and G-CSF, were observed following infection, with similar baseline cytokine activity in Ripk3^−/− mice compared to WT mice (Figures 4E–H, Table S5). In contrast to Mlkl^−/− mice that had increased IL-1β production (Figure 1K), the Ripk3^−/− mice had significantly less IL-1β (Figure 4H). Histopathology was notable for better preservation of skin architecture in the infected Ripk3^−/− mice (Figure 4I). The interactions of RIPK3 with additional binding partners were evident, as we observed less cell death due to apoptosis, as shown by decreased Annexin V staining in both keratinocytes and immune cells in Ripk3^−/− skin after infection (Figure 4J). Thus, deletion of Ripk3 inhibited excessive inflammatory signaling and enhanced eradication of S. aureus from infected skin.

Caspases -8, -3 and -7 are activated by S. aureus but do not contribute to cytotoxicity

Apoptosis and necroptosis share upstream effectors, with RIPK1-FADD-Caspase-8 cascade mediating apoptosis and RIPK1-RIPK3-MLKL pathway regulating necroptosis in the absence of active caspase-8 (Chan et al., 2015). Caspase-8 targets the deubiquitinase CYLD preventing RIPK1 initiation of necroptosis (Legarda et al., 2016). S. aureus has been shown previously to activate caspases-8, -3 and -7 in keratinocytes (Soong et al., 2012). Since apoptosis and necroptosis effectors are intertwined and since S. aureus has the ability to activate both mechanisms of cell death, we wanted to dissect which of the pathways predominate during infection. We observed that S. aureus activated caspases -8, -3 and -7 in macrophage and keratinocyte cell lines (Figures 5A–C). Wild type as well as toxin-deficient (agr mutant) strains of S. aureus activated caspases-3/7 to the same extent (Figure 5B). However, deletion of caspase-8, as tested in Casp8^−/−Ripk3^−/− BMDMs did not affect the overall cytotoxicity of S. aureus (Figure 5D), since they still express inflammasome components. Similarly, pharmacological inhibitors of the apoptosis executioners (caspases-3/7), which decreased the activation of these caspases (Figure 5C), did not prevent S. aureus induced cytotoxicity in THP-1 cells (Figure 5E). To document the induction of apoptosis in vivo, we demonstrated active caspase-3 in the sloughing corneocyte layer of keratinocytes on S. aureus-infected murine skin (Figure 5F). S. aureus toxins activate necroptosis (Kitur et al., 2015), as well as pyroptosis (Munoz-Planillo et al., 2009; Munoz-Planillo et al., 2013; Soong et al., 2012). Thus, while S. aureus can induce apoptosis executioners, nonetheless, cytotoxicity is mediated predominantly by the components of the inflammasome and the RIPK1/RIPK3/MLKL necroptosis pathways.

Caspase-1 signaling, but not necroptosis, contributes to S. aureus killing

In vitro assays were performed to establish whether necroptosis participates in S. aureus killing by human keratinocytes or immune cells. A gentamicin protection assay performed using human keratinocytes in primary culture demonstrated that concentrations of NSA that prevent cytotoxicity fail to inhibit S. aureus killing in contrast to the pan-caspase inhibitor, ZVAD (Figures 1E, 6A). Peritoneal exudate cells (PECs) from the Ripk3^−/− and Mlkl^−/− mice had no defect in their ability to phagocytose and kill S. aureus, whereas the Casp1/4^−/− PECs had significantly impaired bacterial killing ability (Figures 6B, 6C). These results were confirmed with bone marrow-derived macrophages (BMDMs), demonstrating a lack of effect of the Ripk3 or Mlkl mutation on either staphylococcal killing or the generation of IL-1β, in contrast to the significant contribution of caspase-1 in these processes (Figure 6D–F). Lack of caspase-1 did not completely abrogate the generation of IL-1β, suggesting that multiple pathways are involved in the production of this cytokine (Figure 6F). To determine the contribution of apoptosis to bacterial killing, we incubated BMDMs from Casp8^−/−Ripk3^−/− mice with S. aureus. Casp8^−/−Ripk3^−/− BMDMs had no defect in either bacterial uptake or killing abilities (Figure 6G, 6H). Casp8^−/−Ripk3^−/− BMDMs produce more IL-1β compared to WT BMDMs (Figure 6I), potentially because the lack of both necroptosis and apoptosis components is compensated by the upregulation of the inflammasome products. Thus, caspase-1 mediated responses, but not necroptosis or apoptosis, contribute to staphylococcal death within immune cells and keratinocytes.

MLKL and caspase-1 enhance survival in a murine model of S. aureus sepsis

Our results using the skin model of infection reflect participation of RIPK1 and MLKL in staphylococcal clearance by limiting local inflammation. As systemic inflammation is a major factor in the morbidity associated with S. aureus infections, we used a murine model of bacteremia to compare bacterial clearance when necroptosis is inhibited with Nec-1s and to highlight the participation of immune cells. Significantly greater numbers of S. aureus were recovered from the blood 24 h after infection in the Nec-1s treated mice than from controls (Figure 7A). To confirm the results obtained with Nec-1s, we used GSK3002963, a potent and specific inhibitor of RIP1 kinase activity (Berger et al., 2015), in our mouse model of sepsis. Like Nec-1s-treated mice, GSK3002963-treated mice had a decreased ability to clear S. aureus in the blood compared to mice treated with a vehicle control (Figure 7B). Mortality studies were then performed using the Mlkl^−/−, Ripk3^−/− and Casp1/4^−/− mice in a model of sepsis. We predicted that the Mlkl^−/− mice, which are unable to activate necroptosis as a mechanism to control inflammatory responses, would have worse outcomes and observed their increased mortality (with median survival of 4 days) compared to the WT mice (Figure 7C). Although the Ripk3^−/− mice are also unable to activate necroptosis, their limited ability to stimulate the inflammasome was predicted to similarly provide protection from excessive inflammation. Survival of the Ripk3^−/− mice was equivalent to that of the WT mice and significantly improved as compared to Mlkl^−/− mutants (Figure 7C). Casp1/4^−/− mice, which have a staphylococcal killing defect as well as deficient inflammasome activation had significantly increased mortality (median survival of 2 days) (Figure 7C). The outcomes from S. aureus sepsis model are consistent with those observed in the skin, in that mutations in the necroptosis pathway associated with excessive inflammatory responses, are also associated with negative outcomes.

Bacteria

S. aureus strain MRSA USA300 LAC was grown overnight in LB Broth at 37°C as previously described (Kitur et al., 2015) and suspended in PBS to achieve the required concentration. S. aureus were heat-killed by at 65°C for 90 mins.

Cell culture

Peritoneal exudate cells (PECs) were collected from mice infected intraperitoneally with 10⁷ CFU of heat-killed S. aureus 48 h and 24 h (Reyes-Robles et al., 2013). S. aureus MOI 10 were opsonized for 30 min with 50% normal mouse serum (NMS, Innovative Research) in RPMI 1640 medium (Gibco), washed and suspended in medium. PECs were infected with opsonized S. aureus MOI 1 in coated plates for 1 h and 24 h. PECs were lysed with 0.1% saponin and recovered bacteria (CFU) enumerated by serial dilutions.

Gentamicin protection assays were performed in HaCaT or BMDMs isolated from mice and differentiated in the presence of 60 ng/ml of mouse M-CSF (Peprotech). After stimulation, HaCaT 2 hours, BMDMs for 20 minutes with S. aureus MOI 100, cells were washed and treated with gentamicin (500 μg/mL) for 1 h or 24 h. Cells were washed in PBS, lysed with 0.1% saponin and bacteria enumerated by serial dilution on agar plates.

Cytotoxicity assays

HaCaT and HEKn cells were grown to confluence in Dulbecco modified Eagle medium (DMEM, Gibco) and THP-1 cells in RPMI Medium 1640 (Gibco) with 10% fetal bovine serum and 1% penicillin and streptomycin at 37°C and 5% CO₂ in a humidified environment. Cells were maintained in no antibiotic medium 24 hours prior to infection. Cells were pretreated 1 hour prior to infection with necrostatin-1 stable (Nec-1s, Enzo Life Technologies), 10 μM necrosulfonamide (NSA, Calbiochem), 10 μM GSK’872 (Calbiochem), 20 μM caspase 3/7 inhibitor (Promega) or DMSO control (v/v). For siRNA knockdown, HaCaT were grown to confluence and transfected with 100 nM siMLKL (human MLKL siRNA pools of three targets, Santa Cruz Biotechnology) or scrambled siRNA control pool using Lipofectamine RNAiMAX (Life Technologies) as previously described (Kitur et al., 2015). Cell were infected after 3 days of siRNA treatment. S. aureus multiplicity of infection (MOI) 10 were added to the cells for 4 hours for keratinocytes and 2 hours for macrophages. Lysates were used for immunoblotting and caspases-3/7 activation assays and supernatants for cytotoxicity assays. LDH cytotoxicity was performed as per manufacturer’s instructions (Roche) and as described previously (Kitur et al., 2015). Caspases-3/7 activation were performed as per manufacturer’s instructions (Caspase-Glo, Promega).

Model of skin infection

Six- to eight-week-old sex-matched wild type C57BL/6J (Jackson Laboratory), Casp1/4^−/− (B6N.129S2-Casp1^tm1Flv/J, Jackson Laboratory) (Kuida et al., 1995), Ripk3^−/− (Vishva Dixit, Genentech) (Newton et al., 2004) and Mlkl^−/− (John Silke, WEHI via Douglas R. Green, St. Jude Children’s Research Hospital) (Murphy et al., 2013) were used. Knockouts were backcrossed to C57BL/6J background and equal sex ratios were used for experiments. Mice were anesthetized, shaved and inoculated subcutaneously on the back with 2 × 10⁶ CFU/mouse of S. aureus in 100 μl of PBS or with 100 μl PBS (control group) using 26^5/8G sterile needle. Lesions were measured daily and mice were sacrificed 1 day or 5 days after infection.

Two 3-mm punch biopsy (Premier Uni-Punch) skin specimens were obtained from the lesions of infected mice, half of the tissue was fixed in 4% paraffin for 1 day and stored in 70% ethanol for histology, performed by the Histology Core at Columbia University Medical Center. Trichrome staining was done and images were taken on Zeiss Axiocam MRc 5 (Zeiss).

The other half of the specimens was homogenized using a cell strainer to obtain single cell suspension in 400 μl PBS. Recovered bacteria were enumerated by serial dilutions. Cells were spun down and homogenate used for ELISA and cells stained for FACS as described below.

Model of sepsis

1 × 10⁸ CFU of S. aureus in 100 μl PBS was delivered intravenously through the retro-orbital injection route to six- to eight-week-old anesthetized mice and survival monitored daily. For necrostatin-1 stable (Nec-1s) sepsis experiments, mice were treated with 10mg/kg Nec-1s or DMSO control 24 hours prior, during infection and 4 hours (for 24-hour infection) or 24 hours (for 48-hour infection) after S. aureus infection. For GSK3002963 sepsis experiments, mice were treated with 10mg/kg GSK3002963 reconstituted with 10% DMSO v/v and 20% ethanol v/v in PBS (or reconstitution solution for vehicle control) during infection and 4 hours after S. aureus infection. Mice were sacrificed 24 hours or 48 hours after infection and bacteria in blood enumerated and cytokines in blood analyzed as described below.

FACS analysis

Homogenized cells were prepared as previously detailed (Kitur et al., 2015). Combinations of phycoerythrin (PE) anti-CD54 (YN1/1.7.4, Biolegend), fluorescein isothiocyanate-labelled (FITC) anti-Ly-6G (Gr-1; RB6-8C5; Biolegend), peridinin chlorophyll (PerCP)-Cy5.5-labelled anti-CD11c (N418; Biolegend), allophycocyanin (APC)-labelled anti-MHC II (I-A/I-E; Biolegend) and APC-labelled anti-TCR γδ T cells (γδ T cells) were used. Neutrophils (PMNs; Ly6G⁺/MHCII⁻), macrophages (Macs; CD11C⁺/MHCII^low-mid) and γδ T cells were enumerated using FlowJo V10.

Cytokine analysis

Eve Technologies Corporation (Calgary, Canada) performed Multiplex ELISA on skin homogenate to determine mouse cytokine levels. Only cytokines that induced by infection are shown. IL-1β ELISA immunoblotting were performed as previously described (Kitur et al., 2015).

Immunoblotting and Immunohistochemistry

Cells were lysed using RIPA buffer as previously described (Kitur et al., 2015). Immuno-detection was done using anti-Caspase-1 (p20) (Casper-1, Adipogen), anti-MLKL (Abcam), anti-caspase-8 (Cell Signaling) and β-actin (Sigma-Aldrich) followed by secondary antibodies conjugated to horseradish peroxidase (Santa Cruz Biotechnology Inc.). Equal loading were confirmed by staining with 0.1% Ponceau S (w/v) in 5% acetic acid (Sigma-Aldrich).

For immunohistochemistry, sections were stained with cleaved caspase-3 (Cell Signaling) and anti-pan cytokeratin (Santa Cruz Biotechnology) followed by peroxidase staining using the Immunocruz ABC Staining kit (Santa Cruz Biotechnology Inc.). Control sections were stained with secondary antibodies only.

Ethics Statement

Animal experiments were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health, the Animal Welfare Act and U.S. federal law. The Institutional Animal Care and Use Committee (IACUC) of Columbia University approved the protocol (AC-AAAH2350 and AC-AAAG7408).

Statistical analysis

Samples without normal distribution were analyzed using the nonparametric Mann-Whitney test and mouse survival analyzed by log-rank (Mantel-Cox) test for comparison of survival curves. Samples with normal distribution were analyzed with two-tailed Student’s t test and one-way ANOVA followed by Bonferroni Corrections to correct for multiple comparisons. p < 0.05 between groups were considered significant. Outliers were determined by Grubb’s test and removed. Statistical analysis was performed using GraphPad Prism Version 4.00 (GraphPad, La Jolla, USA). Data are presented as single points with lines representing median values or as bar graphs with mean ± SEM.

For more experimental procedures, see Supplemental Experimental Procedures in Supplemental Information.