Altered mitochondrial homeostasis during systemic lupus erythematosus impairs NET formation rendering neutrophils ineffective at combating Staphylococcus aureus

Abstract

Inflammation involves a delicate balance between pathogen clearance and limiting host tissue damage, and perturbations in this equilibrium promote disease. Patients suffering from autoimmune diseases, such as systemic lupus erythematosus (SLE), have higher levels of serum S100A9 protein and increased risk for infection. S100A9 is highly abundant within neutrophils and modulates antimicrobial activity in response to bacterial pathogens. We reasoned that increased serum S100A9 in SLE patients reflects accumulation of S100A9 protein in neutrophils and may indicate altered neutrophil function. Herein, we demonstrate elevated S100A9 protein within neutrophils from SLE patients and MRL/lpr mice associates with lower mitochondrial superoxide, decreased suicidal neutrophil extracellular trap formation, and increased susceptibility to Staphylococcus aureus infection. Furthermore, increasing mitochondrial superoxide production restored the antibacterial activity of MRL/lpr neutrophils in response to S. aureus. These results demonstrate that accumulation of intracellular S100A9 associates with impaired mitochondrial homeostasis thereby rendering SLE neutrophils inherently less bactericidal.

Introduction

Staphylococcus aureus is a Gram-positive bacterium that is the leading cause of soft tissue infections (1, 2), bacterial endocarditis (3–5), and the second-most frequent agent of hospital-associated pneumonia (6) and bloodstream infections (7–9). Neutrophil-mediated inflammation is a hallmark of the innate immune response to S. aureus (10, 11). Calprotectin is a heterodimer formed by the S100A8 and S100A9 proteins (12, 13) and comprises approximately 50% of the cytoplasmic protein in neutrophils (14), and significantly influences disease outcome during S. aureus infection. When S100A9-deficient (A9^−/−) mice, which lack calprotectin, are systemically infected with S. aureus they exhibit increased bacterial burdens in the liver and kidney (15, 16) but have lower bacterial burdens in the heart with increased overall survival (17, 18), compared to wild-type (WT) mice. While an emphasis has been placed on the ability of calprotectin to sequester nutrient metals from bacterial pathogens (15, 16, 19–24), an intracellular function for S100A9 in regulating the antibacterial activity of neutrophils was described recently; whereby loss of S100A9 coincides with increased mitochondrial superoxide (O₂⁻) production (18).

In response to pathogens, neutrophils undergo two distinct mechanisms of neutrophil extracellular trap (NET) formation consisting of a meshwork of DNA and antimicrobial peptides and proteins. Suicidal NET formation occurs within 2–4 hr of activation resulting in the terminal release of NETs through rupture of the cell membrane (25), while vital NET formation occurs within 5–60 min of engaging S. aureus whereby NETs are exocytosed and leave the membrane intact (26). The formation of NETs plays an integral role in combating S. aureus infections by restricting bacterial dissemination and eliciting direct bactericidal activity (27, 28) as well as augmenting macrophage bactericidal activity (18). Increased production of mitochondrial O₂⁻ in A9^−/− neutrophils causes accelerated and more robust suicidal NET formation, and impaired primary degranulation in response to S. aureus (18). Consistent with this, A9^−/− mice exhibit increased survival with reduced bacterial burdens specifically in the heart (17, 18).

Chronic inflammation contributes to autoimmune diseases including systemic lupus erythematosus (SLE). SLE is an autoimmune disease with genetic and environmental components leading to tissue-damaging inflammation (29, 30). Neutrophils contribute to SLE pathology by increasing the burden of nuclear antigens through NET formation (31, 32) and heightens the risk for cardiovascular disease in SLE patients (33–35). It has been observed by multiple groups that SLE disease activity correlates with increased serum S100A9 protein levels (34, 36, 37); however, the source of S100A9 during autoimmune disease has not been identified. We reasoned that increased serum concentration of S100A9 is indicative of increased S100A9 protein within neutrophils. Since S100A9 levels affect neutrophil bactericidal activity (18), this could contribute to the correlation between autoimmune diseases and S. aureus being a frequent cause of infections in patients with SLE (38–41). Herein, we demonstrate that neutrophils from MRL/lpr mice and SLE patients accumulate increased levels of S100A9 protein coinciding with altered mitochondrial homeostasis and reduced O₂⁻ production in neutrophils. As a result, suicidal NET formation is decreased in response to S. aureus by murine and human SLE neutrophils, which impairs bactericidal activity and renders the host more susceptible to S. aureus infection.

Materials and Methods

Mice

C57BL/6J were purchased from Jackson Labs (JAX mice stock no. 000664). MRL/MpJ-Tnfrs6^lpr/J mice were purchased from Jackson Labs (JAX mice stock no. 000485) or obtained from Dr. B. Vilen. B6.S100a9^−/− mice were maintained in an accredited animal facility. Mice (2–5 per cage) were housed in specific pathogen-free conditions and randomly assigned to experimental groups. Food and water were provided ad libitum. All animal experiments were approved and performed in compliance with the Vanderbilt Institutional Animal Care and Use Committee (IACUC).

Reagents

Antibodies specific for Ly6G, CD3, CD19, Ly6C, CD15, CD16, rabbit IgG, and annexin V were from Biolegend; CD11b was from Tonbo; CD63 was from BD Biosciences; myeloperoxidase, histone citrulline, dsDNA, and mouse IgG-alkaline phosphatase were from Abcam; S100A9 (murine and human) was from Cell Signaling. Anti-nucleosome antibody (PL2–3) was a gift from Dr. B. Vilen. Streptavidin Alexa 488 was from Biolegend. Live/dead stain, mitoSOX, mitoTracker, and BacLight® Red Bacterial Stain were from Invitrogen; Helix NP Blue (Sytox) was from Biolegend. Murine Fc-blocking antibody (2.4G2) was from Tonbo; human Fc-blocking antibody was from BD Biosciences. Paraformaldehyde (PFA) was from Electron Microscopy Sciences. Rotenone was from Sigma. Histopaque® 1119 and Histopaque® 1077 were from Sigma. Dulbecco’s Modified Eagle’s Medium (DMEM) and phosphate buffered saline (PBS) was from Gibco and fetal bovine serum (FBS) from Atlanta biologicals were used for all tissue culture.

Bacterial strains

All experiments utilized the S. aureus strain USA300 (LAC) (42). The S. aureus stock was maintained as a −80 °C and 2 days prior to each experiment, was freshly streaked onto tryptic soy agar (TSA; 2% agar) plates each experiment. Single colonies of S. aureus were transferred to tryptic soy broth (TSB), and grown as liquid cultures overnight at 37 °C with 180 rpm shaking. For all in vitro experiments, overnight cultures were diluted into non-heat inactivated FBS at a 1:10 ratio on ice, 2 hr prior to culturing with neutrophils.

Infections

Retroorbital infections were performed as described (18, 43). In brief, S. aureus was streaked onto TSA, two-days prior to infection. Single colonies were transferred from the plate and grown overnight in TSB. On the day of the infection, overnight cultures were sub-cultured 1:100 into 5 mL TSB. S. aureus was grown to mid-exponential phase, washed, and resuspended in 100 μL ice-cold PBS. All infections used 10–13 week-old female mice. Mice were anesthetized by intraperitoneal injection of 2,2,2-tribromoethanol diluted in PBS. Systemic infections were induced by intravenous injection of the retroorbital sinus with 2×10⁷ CFU S. aureus. Mice were monitored prior to humane euthanasia by inhalation of CO₂ at 4 dpi.

Organ harvest for CFU and neutrophil characterization

Organs were harvested as previously described (18). Tissue from the hearts, livers, and kidneys were processed through a 35 μm filter into FACS media (PBS, 2% FBS, 0.02% NaAz) to create single cell suspensions. Scalpels were used to cut the heart tissue into smaller pieces to facilitate processing of the tissue through the filter. In addition, special care was taken to flush the heart chambers with PBS to remove any clotted blood and ensure isolated immune cells are from the tissue. Aliquots were taken from each organ homogenate for CFU enumeration by spot plating or to quantify total S100A9 protein levels in the tissue by ELISA. The remaining organ homogenate was pelleted and the supernatant was isolated to quantify NET abundance by ELISA. The remaining pellet was red blood cell lysed per manufacturer’s instructions (Biolegend) and transferred in FACS media into separate aliquots to be stained ex vivo for flow cytometry.

Assessment of autoimmunity in mice

Three days prior to infection murine levels of proteinuria and serum autoantibody were quantified. Urine protein was assessed according to Uristix 4 (Siemens) instruction. Blood was collected by tail nick and serum levels of IgG reactive to nucleosome and dsDNA was quantified by ELISA.

Neutrophil isolation

Neutrophils were isolated as previously described (18). Bone marrow was flushed from the tibias and femurs of mice and single cell suspensions were created. Polymorphonuclear granulocytes were isolated using density centrifugation by isolating the neutrophils at the interface between the layers of Histopaque® 1119 and Histopaque® 1077. Neutrophils were resuspended in D10 media (DMEM, 10% FBS) and rested on ice for at least 1 hr before transferring to ultra-low cluster round bottom 96-well plates (Costar) in D10 media. Neutrophils were incubated (37 °C, 5% CO₂) for 1 hr prior to experiments. The isolated cells were 85–95% neutrophils (CD11b⁺Ly6G⁺).

Bacterial killing assay

Bacterial killing assays were performed as previously described (18). Neutrophils were cultured with S. aureus at a MOI of 1 (50,000 CFU per 50,000 neutrophils) in D10 media (37 °C, 5% CO₂). At each time point (30 min, 2, 6, 12, and 24 hr), samples were serially diluted and spot-plated onto TSA. Plates were cultured overnight (37 °C), and CFU enumerated the next day. Percent growth was quantified by dividing the CFU of the S. aureus-neutrophil co-culture by the S. aureus alone culture.

Confocal imaging

NET imaging was conducted as previously described (18). Neutrophils were stained in with CellMask Deep Red in PBS for 20 min at room temperature and washed in D10 media prior to plating. Glass bottom Petri dishes (MatTek Corporation; Ashland, MA) were charged using a 0.01% poly-L-lysine solution (Sigma Aldrich) and neutrophils were incubated in the dishes for 2 hr (37 °C, 5% CO₂) in D10 to allow cells to properly attach to the dish. S. aureus was stained with BacLight® Green for 30 min on ice and added to the dish at an MOI of 1. Helix NP Blue was added 20 min prior to fixation to stain for extracellular DNA (NET structures). Fixation of the neutrophils occurred in 4% paraformaldehyde at room temperature and incubated for 15 min at 4 °C. Dishes were aspirated and samples preserved using Prolong Gold Antifade Mountant. All microscopy was conducted using a Zeiss 880 microscope with AiryScan and Zeiss Zen Software. Images were analyzed using ImageJ.

Flow cytometry

All flow cytometry experiments were conducted as previously described (18). All data were collected using a BD LSRII flow cytometer with FACSDIVA software and analyzed using FlowJo (FlowJo LLC, Ashland, OR). Samples were gated forward scatter height (FSC-H) by forward scatter area (FSC-A) and side scatter height (SSC-H) by side scatter area (SSC-A) to remove doublet populations. The singlet population was gated SSC-A by FSC-A to isolate the granulocyte population. The resulting cell population was then assessed for assay-specific fluorescent markers as described below:

Exogenous Treatments:

Rotenone (0.5 μM) was added 15 min hr prior to stimulation with S. aureus.

Degranulation Flow:

Single cell suspensions isolated from the tissue, bone marrow isolated neutrophils, or peripheral blood PMNs from patients (referred to as cells for the remainder of the flow cytometry section) were stained with a live/dead dye to monitor membrane integrity 20 min prior to fixation. Cells were fixed in 4% room temperature paraformaldehyde and placed on ice for 15 min. Cells were pelleted, aspirated, and resuspended in Fc block in FACS media for 30 min on ice. Cells were washed and stained with antibodies recognizing appropriate neutrophil surface markers (murine: CD11b, Ly6G; human: CD15, CD16), anti-CD63 (primary), and anti-CD35 (secretory) or appropriate isotype controls in FACS media for 30 min on ice. Cells were washed following staining and resuspended in FACS media for analysis. Cells were gated for live cells (Live/Dead-negative) and then neutrophils (CD11b- and Ly6G-positive or CD15- and CD16-positive). The median fluorescence intensity (MFI) was quantified for surface CD63 and CD35 relative to staining with an isotype control antibody.

Phagocytosis and ROS flow:

S. aureus was stained using BacLight® prior to culturing with neutrophils. Cells were live/dead stained as described in ‘Degranulation Flow’ except DHR123 (3 μM) was also introduced to the cultures to quantify total ROS production. Cells were fixed, blocked, and stained for neutrophils (anti-CD11b and -Ly6G) as described in ‘Degranulation Flow’. Following staining, cells were washed, and resuspended in FACs media for analysis. Cells were gated for live cells (Live/Dead-negative) and then neutrophils (CD11b- and Ly6G-positive). The media MFI was quantified for fluorescent S. aureus and DHR123 and normalized to unstimulated neutrophils.

NET Flow:

To quantify the percentage of neutrophils undergoing NET formation, cells were live/dead stained as described above except Helix NP Blue (2.5 nM) was introduced during the live/dead stain. Cells were fixed, blocked, and stained for neutrophils (murine: anti-CD11b and -Ly6G; human: anti-CD15 and -CD16) as described in ‘Degranulation Flow’. During the staining for neutrophils, cells were also stained with anti-H3Cit and -MPO antibodies. Cells were washed and stained with fluorescent streptavidin and anti-rabbit IgG in FACS media for 30 min on ice. Following staining, cells were washed, and resuspended in FACS media for analysis. Cells were gated for neutrophils and then live or dead cells. Neutrophils positive for extracellular dsDNA (Helix NP), MPO, and H3Cit were defined as having undergone NET formation. If the membrane of the cell undergoing NET formation was intact then the cell was defined as having undergone vital NET formation, and if the membrane of the cell undergoing NET formation was permeabilized then the cell was defined as having undergone suicidal NET formation. Representative gating is provided in Supplemental Fig. 2B. The percentage of neutrophils undergoing vital or suicidal NET formation was quantified relative to the total number of neutrophils.

Apoptosis flow:

Cells were live/dead stained, fixed, and blocked, as described in ‘Degranulation Flow’. Cells were washed and stained and stained for neutrophils (anti-CD11b and -Ly6G) as described above except fluorescent staining by annexin V was used to detect apoptotic cells. Following staining, cells were washed, and resuspended in FACs media for analysis. Cells were gated for neutrophils (CD11b- and Ly6G-positive) and the percentage of cells in early (Live/Dead-negative, annexin V-positive) and late (Live/Dead-posative, annexin V-positive) apoptosis was quantified.

Mitochondrial O₂⁻ Flow:

To quantify mitochondrial O₂⁻ production, cells were live/dead stained as described in ‘Degranulation Flow’ except mitoSOX red (5 μM) and mitoTracker deep red (500 nM) were introduced during the live/dead stain. Cells were fixed, blocked, and stained for neutrophils (murine: anti-CD11b and -Ly6G; human: anti-CD15 and -CD16) as described in ‘Degranulation Flow’. Following staining, cells were washed and resuspended in FACS media for analysis. Cells were gated for live cells and then neutrophils. mitoSOX and mitoTracker MFI were normalized to untreated or mock WT MFI.

A9 Protein Flow:

To quantify the abundance of intracellular S100A9, cells were live/dead stained, fixed, blocked, and stained for neutrophils (murine: anti-CD11b and -Ly6G; human: anti-CD15 and -CD16) as described in ‘Degranulation Flow’. For peripheral blood samples, T cells (CD3⁺), B cells (CD19⁺), monocytes (Ly6C⁺), were also stained. Following staining, cells were washed and stained for S100A9 or an isotype control in permeabilization buffer (PBS, 0.05% saponin, 0.5% BSA) for 30 min on ice. To detect S100A9, cells were washed and stained with fluorescently labeled anti-rabbit IgG for in permeabilization buffer for 30 min on ice. Following staining, cells were washed and resuspended in FACS media for analysis. The MFI was quantified for surface S100A9 relative to staining with an isotype control antibody.

Mitochondrial and Glycolytic Stress Test - Seahorse

Seahorse plates were coated in Cel-Tak (Corning) for 20 min, aspirated, washed using sterile water, and aspirated prior to the addition of neutrophils. Neutrophils were resuspended in either mitochondria (Seahorse Agilent pH 7.4 media, 1 mM pyruvate, 10 mM glucose, 2 mM glutamine) or glycolysis (Seahorse Agilent pH 7.4 media, 2 mM glutamine) assay media and 2×10⁵ neutrophils were added to each well. No neutrophils were added to the four wells in the corners of the plate as a background control. S. aureus was diluted (2×10⁴CFU/μL) in serum and heat-killed for 10 min at 80 °C. For the mitochondrial stress test, the cartridge was sequentially loaded with heat-killed S. aureus (MOI=1), oligomycin (15 μM), FCCP (15 μM), and rotenone/antimycin A (5 μM) (diluted in mitochondria assay media) for injection into the culture. For the glycolysis stress test, the cartridge was sequentially loaded with heat-killed S. aureus (MOI=1), glucose (100 mM), oligomycin (15 μM), and 2-deoxy-d-glucose (500 mM) (diluted in glycolysis assay media) for injection into the culture. To normalize across wells by cell density, the Seahorse plate was imaged using a BioTek Cytation 5. Oxygen consumption and extracellular acidification rates were collected using a Seahorse XFe96 Analyzer and analyzed using Wave Desktop software (Agilent Technologies).

ELISA

Serum autoantibodies:

Vinyl 96-well plates were charged using 0.01% poly-L-lysine solution (Sigma Aldrich), and the plate was coated with dsDNA (Sigma Aldrich) or histones (Immunovision) and dsDNA. Total IgG reactive to nucleosome with histones (4 mg/well; ImmunoVision) and dsDNA (1 mg/well; Sigma) was captured to charged vinyl plates. Captured antibodies were detected with anti-mouse IgG-alkaline phosphatase and p-nitrophenyl phosphate (Sigma). A standard curve was generated using anti-dsDNA (Abcam) and anti-nucleosome (PL2–3) antibodies. Absorbance at 405 nm for each well was quantified using a BioTek Cytation 5. Samples and standards were read in duplicate and DNA abundance was quantified relative to the standard.

NET quantification:

For the NET ELISA a similar procedure was followed as described above. Vinyl 96-well plates were charged and for samples, wells were coated with anti-neutrophil elastase or -MPO antibody. For the DNA standard, dsDNA was titrated using 2-fold dilutions starting at 10 mg/mL. DNA in the wells was detected using an anti-dsDNA antibody (mouse IgG) followed by a secondary anti-mouse antibody conjugated with alkaline phosphatase and 1 mg/mL p-nitrophenyl phosphate. Absorbance at 405 nm for each well was quantified using a BioTek Cytation 5. Samples and standards were read in duplicate and DNA abundance was quantified relative to the standard.

S100A9 quantification:

Organ homogenates were lysed (PBS, 1% Igepal, protease inhibitors) and pelleted to remove cellular debris. The resulting cell lysates or serum from the blood were diluted 1:100 and S100A9 protein levels were quantified per manufacturer’s instructions (Abcam, #ab213887).

Human samples

SLE patients and healthy controls were enrolled following informed consent as approved by the Vanderbilt University Medical Center institutional review board. HC were matched to SLE patients as close as possible by age and gender. HC donors reported no history of autoimmune disease and had not been sick within 2 weeks of donating blood. Disease activity was measured by the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI), and patients with a score ≥ 6 were defined as active and scores < 6 as inactive. Paired healthy control and SLE peripheral blood samples were collected with sodium heparin (BD Biosciences) and PMNs were isolated using density centrifugation at the interface between the layers of Histopaque® 1119 and Histopaque® 1077.

Statistics

Specific statistical details for each experiment are located in the corresponding figure legend. Error bars for experiments represent variation of the mean across mice. A minimum of three experimental replicates were performed for each assay and the specific number of replicates is noted in the corresponding figure legend. All P values were calculated using a one- or two-way ANOVA (Sidak’s multiple comparisons test or Tukey multiple comparisons test), unpaired t-test, or log-rank (Mantel-Cox) test when applicable. Statistical work was performed using Prism 6 software (GraphPad) and significance is indicated on the graphs as follows: *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001, ns=not significant.

Results

MRL/lpr mice are more susceptible to systemic S. aureus infections

To identify whether altered neutrophil function contributes to the heightened susceptibility of SLE patients to S. aureus infections (38–41), MRL/MpJ-Tnfrs6^lpr/J (MRL/lpr) mice were employed as they spontaneously generate systemic autoimmunity and are commonly used as a murine model of SLE (44). Therefore, we systemically infected 10–13-week-old female WT and MRL/lpr mice with S. aureus and monitored disease progression over four days. For all infections, MRL/lpr mice with elevated autoantibody titers and mild glomerulonephritis (proteinuria score ≤ 2) were used (Supplemental Fig. 1A–B). Over the four-day infection, MRL/lpr mice were more susceptible to systemic S. aureus infection compared to age-matched WT mice (Fig. 1A), despite comparable weight loss during infection (Supplemental Fig. 1C). Coinciding with decreased survival, MRL/lpr mice had increased bacterial burdens in the heart compared to WT (Fig. 1B) four-days post infection (dpi); however, bacterial burdens in the liver and kidney were comparable. These results demonstrate that MRL/lpr mice are more susceptible to systemic S. aureus infections.

FIGURE 1.

MRL/lpr mice are more susceptible to S. aureus infections with increased bacterial burdens in the heart. (A) During the infection, mouse survival was monitored. Each point represents the percentage of living mice (3 infections; mock n=14, WT; inf. n=19, MRL/lpr; inf. n=21). (B) At 4 dpi, organs were homogenized and (B) colony forming units (CFU) enumerated using spot plating. Each point represents a single mouse (3 infections; mock n=6, WT; inf. n=14, MRL/lpr; inf. n=9). (A) Log-rank (Mantel-Cox) test and (B) one-way analysis of variance (ANOVA) with Tukey multiple comparisons test (*P≤0.05, ****P≤0.0001, ns=not significant).

Increased levels of S100A9 in the serum correlate with SLE disease activity (34, 36, 37); therefore, we reasoned that S100A9 protein levels are elevated in SLE neutrophils. Consistent with what is observed in SLE patients, serum levels of S100A9 are elevated in MRL/lpr mice (Fig. 2A). Coinciding with increased serum S100A9 levels, S100A9 protein abundance within MRL/lpr neutrophils in the blood was higher than WT neutrophils (Fig. 2B, Supplemental Fig. 1D). Intracellular S100A9 levels were not detectable by flow cytometry in other immune cell populations (Fig. 2B, Supplemental Fig. 1D). These results demonstrate that increased serum S100A9 corresponds with MRL/lpr neutrophils accumulating and releasing higher levels of S100A9 protein.

FIGURE 2.

Intracellular S100A9 protein is elevated in MRL/lpr neutrophils. Mice were systemically infected (inf.) with S. aureus (CFU=2×10⁷). 3 days prior to infection, (A) S100A9 levels were quantified from the serum of the mice by ELISA and (B) intracellular S100A9 protein levels were quantified in immune cells isolated from the blood by flow cytometry. (B) Median fluorescence intensity (MFI) normalized by isotype control antibody. Each point represents (A) the mean result (technical duplicate) of serum isolated from a single mouse or (B) immune cells isolated from a single mouse (1 infection, n=5). At 4 dpi, organs were homogenized and (C) intracellular abundance of S100A9 in neutrophils (Ly6G⁺CD11b⁺) was quantified by flow cytometry and (D) total S100A9 protein in cell lysates from the tissue was quantified by ELISA. (C) MFI normalized by isotype control antibody. Each point represents (C) a single mouse (3 infections; mock n=6, WT; inf. n=14, MRL/lpr; inf. n=9) or (D) the mean result (technical duplicate) of cell lysate isolated from tissue from a single mouse (1 infection; n=5). (A-B, D) Unpaired t-test or (C) one-way ANOVA with Tukey multiple comparisons test (**P≤0.01, ***P≤0.001, ****P≤0.0001, ns=not significant).

The abundance of intracellular S100A9 influences neutrophil function (18). Therefore, mice were systemically infected with S. aureus and intracellular S100A9 levels within neutrophils were quantified by flow cytometry. Neutrophils in the heart of mock and infected MRL/lpr mice had increased intracellular S100A9 protein levels relative to WT (Fig. 2C, Supplemental Fig. 1E), which coincided with a 0.8-log₁₀ increase S. aureus burdens (Fig. 1B). Neutrophils in the liver and kidney, organs that had no change in bacterial burdens, had comparable intracellular S100A9 protein levels in WT and MRL/lpr mice (Fig. 2C). In confirmation, the concentration of S100A9 was quantified in the cell lysate acquired from organ homogenates by enzyme-linked immunosorbent assay (ELISA). S100A9 protein levels were higher in cell lysates obtained from the hearts of MRL/lpr mice compared to WT (Fig. 2D), which is consistent with MRL/lpr neutrophils retaining higher levels of intracellular S100A9 in the heart. Levels of S100A9 protein in cell lysates from the liver and kidney of WT and MRL/lpr mice were comparable (Fig. 2D). Thus, the accumulation of intracellular S100A9 may negatively impact neutrophil function, contributing to the increased bacterial burdens in the heart and heightened susceptibility of MRL/lpr mice to S. aureus infections.

MRL/lpr neutrophils produce lower levels of mitochondrial O₂⁻ in response to S. aureus

Dysregulation of mitochondrial homeostasis in immune cells has been linked to SLE pathology (45–49), but whether similar defects are observed in SLE neutrophils responding to bacterial pathogens has not been studied in detail. To assess mitochondrial function, neutrophils were isolated from the bone marrow by density centrifugation. Bone marrow neutrophils from MRL/lpr mice showed an accumulated of intracellular S100A9 protein compared to WT by flow cytometry (Fig. 3A, Supplemental Fig. 2A). Since intracellular levels of S100A9 affects mitochondrial O₂⁻ production and downstream neutrophil function (18), mitochondrial respiration was monitored in neutrophils by quantifying the oxygen consumption rate (OCR) using extracellular flux analysis (Fig. 3B). Heat-killed S. aureus was used for these experiments to avoid artifacts associated with oxygen consumption by live bacteria. Proton leak has been described as a mechanism to compensate for the electrons that slip from the electron transport chain and contributes to mitochondrial O₂⁻ production (50–52). Unstimulated MRL/lpr neutrophils exhibited an 8.2-fold higher proton leak when compared to WT, but this phenotype was lost following stimulation with S. aureus (Fig. 3C). Basal respiration of MRL/lpr neutrophils was significantly higher than WT neutrophils in the absence of heat-killed S. aureus, and a similar trend was observed following stimulation with heat-killed S. aureus (Fig. 3C). In addition, maximal respiration and ATP production were elevated in MRL/lpr neutrophils, although they also did not reach statistical significance. Instead, non-mitochondrial oxygen consumption was elevated in MRL/lpr neutrophils when stimulated with heat-killed S. aureus (Fig. 3C), indicative of activity by peroxisomes and/or non-mitochondrial NADPH oxidases (53). Finally, glycolysis and glycolytic capacity were comparable between WT and MRL/lpr neutrophils, as measured by extracellular acidification rates (ECAR; Supplemental Fig. 2B–C). These results suggest that while electron transfer is less efficient in MRL/lpr neutrophils, stimulation with heat-killed S. aureus increases the efficiency of electron transfer and simultaneously diverts oxygen consumption to non-mitochondrial processes.

FIGURE 3.

MRL/lpr neutrophils have altered mitochondrial homeostasis with reduced mitochondrial O₂⁻ production in response to S. aureus. Neutrophils were isolated from the bone marrow, (A) the intracellular abundance of S100A9 in neutrophils (Ly6G⁺CD11b⁺) was quantified by flow cytometry. Median fluorescent intensity (MFI) normalized by isotype. Each point represents neutrophils isolated from a single mouse (2 experiments, n=4). (B-C) Isolated neutrophils were stimulated with heat-killed S. aureus (HK-SA; MOI=1), and the oxygen consumption rate (OCR) was quantified by Seahorse analysis. A mitochondrial stress test was performed using the sequential addition of oligomycin (oligo.; 15 μM), FCCP (15 μM), and rotenone/antimycin A (Rot./Ant.; 5 μM). OCR normalized by cell number. Each point represents (B) the mean of all mean results of neutrophils isolated from a single mouse ±standard error or (C) the mean result of neutrophils isolated from a single mouse (5 experiments; n=5). (D) Neutrophils pre-treated with rotenone (Rot.; 0.5 μM) for 15 min were cultured with S. aureus (MOI=10) and mitochondrial O₂⁻ production was quantified by flow cytometry. MitoSOX MFI normalized to MitoTracker MFI. Each point represents neutrophils isolated from a single mouse (2 experiments; n=4). (E) Mice were systemically infected (inf.) with S. aureus (CFU=2×10⁷) and at 4 dpi, organs were homogenized and mitochondrial O₂⁻ production by neutrophils was quantified by flow cytometry. MitoSOX MFI normalized to MitoTracker MFI. Each point represents a single mouse (3 infections; mock n=6, WT; inf. n=14, MRL/lpr; inf. n=9). (A, C, E) One-way or (D) two-way ANOVA or with Tukey multiple comparisons test (*P≤0.05, ***P≤0.001, ****P≤0.0001, ns=not significant).

A9^−/− neutrophils produce higher levels of mitochondrial O₂⁻ in response to S. aureus (18); therefore, we hypothesized that the accumulation of intracellular S100A9 in MRL/lpr neutrophils may impair mitochondrial O₂⁻ production. To quantify mitochondrial O₂⁻ generation, neutrophils were stained with mitoSOX, a fluorescent dye that localizes to mitochondrial structures and fluoresces upon oxidation by O₂⁻. Consistent with increased proton leak (Fig. 3C), unstimulated MRL/lpr neutrophils produced higher levels of mitochondrial O₂⁻ compared to WT (Fig. 3D). However, MRL/lpr neutrophils produced lower levels of mitochondrial O₂⁻ 60 min post inoculation compared to WT neutrophils (Fig. 3D, Supplemental Fig. 2D). This finding is consistent with the reduced proton leak observed in MRL/lpr neutrophils following stimulation with heat-killed S. aureus (Fig. 3C). Treating MRL/lpr neutrophils with rotenone, an inhibitor of complex I of the electron transport chain, was sufficient to increasing mitochondrial O₂⁻ production to levels comparable to WT (Fig. 3D, Supplemental Fig. 2D). These results demonstrate that MRL/lpr neutrophils produce lower levels of mitochondrial O₂⁻ in response to S. aureus, which coincides with the accumulation of intracellular S100A9 in MRL/lpr neutrophils disrupting mitochondrial homeostasis.

To determine whether the lower production of mitochondrial O₂⁻ observed in vitro was present in MRL/lpr mice, mice were infected and the production of mitochondrial O₂⁻ by neutrophils was quantified by flow cytometry 4 dpi. Consistent with increased S100A9 protein, MRL/lpr neutrophils in the heart produced 2.4-fold less mitochondrial O₂⁻ during infection than WT (Fig. 3E, Supplemental Fig. 2E). Mitochondrial O₂⁻ production by neutrophils in the kidney and liver were comparable (Fig. 3E, Supplemental Fig. 2E). These results suggest that accumulation of S100A9 coinciding with lower mitochondrial O₂⁻ production, impacts neutrophil antibacterial activity in the heart and rendering MRL/lpr mice more susceptible to infection.

Decreased mitochondrial O₂⁻ production by MRL/lpr neutrophils impairs suicidal NET formation

Neutrophils utilize multiple antimicrobial processes when engaging S. aureus. Upon engaging S. aureus, neutrophils couple phagocytosis with the respiratory burst (54–56). Phagocytosis of fluorescently labeled S. aureus by WT and MRL/lpr neutrophils was comparable (Supplemental Fig. 2F). In contrast to what was observed with MRL/lpr macrophages (57), MRL/lpr neutrophils produce lower levels of ROS compared to WT 120 min post inoculation (Supplemental Fig. 2G). However, this phenotype was not present in vivo as neutrophils isolated from the heart and kidney of infected WT and MRL/lpr mice produced comparable levels of ROS, while MRL/lpr neutrophils constitutively produced higher levels of ROS in the liver independent of infection (Supplemental Fig. 2H). These results suggest that impaired ROS production does not account for the increased susceptibility of MRL/lpr mice during infection.

Since the production of mitochondrial O₂⁻ promotes suicidal NET formation in response to S. aureus (18), neutrophils were stimulated with S. aureus and cultures were imaged for NET structures after 240 min. NET structures were readily apparent in cultures with WT neutrophils, but NET structures were smaller and less frequent in cultures with MRL/lpr neutrophils (Fig. 4A). To quantify NET formation, we used a flow cytometry-based technique (Supplemental Fig. 3A) that has been previously described in detail (18). In response to S. aureus, MRL/lpr neutrophils showed impaired suicidal NET formation (3.8-fold; Fig. 4B, Supplemental Fig. 3B) and sustained vital NET formation (6.2-fold; Fig. 4C, Supplemental Fig. 3C) compared to WT neutrophils 240 min post inoculation. While the abundance of intracellular S100A9 and mitochondrial O₂⁻ production influences release of primary granules (18), primary degranulation was comparable between WT and MRL/lpr neutrophils (Supplemental Fig. 3D). In addition, alternative cell death pathways were similar between WT and MRL/lpr neutrophils except at 240 min post inoculation where a higher percentage of WT neutrophils displayed signs of late apoptosis compared to MRL/lpr (Supplemental Fig. 3E). These results demonstrate that MRL/lpr neutrophils have sustained vital NET formation but impaired suicidal NET formation in response to S. aureus.

FIGURE 4.

Decreased mitochondrial O₂⁻ production prevents suicidal NET formation in MRL/lpr neutrophils. (A) Neutrophils were cultured with S. aureus (MOI=10) and representative images of neutrophils (red) stimulated for 4 hr with a nuclease-deficient strain of S. aureus (green) are provided. Extracellular DNA was stained using Helix NP Blue. (B-C) Neutrophils pre-treated with rotenone (Rot.; 0.5 μM) for 15 min were cultured with S. aureus (MOI=10) and the percentage of neutrophils undergoing (B) suicidal (dead: extracellular dsDNA⁺MPO⁺H3Cit⁺) and (C) vital (live: extracellular dsDNA⁺MPO⁺H3Cit⁺) NET formation in response to S. aureus was quantified by flow cytometry. Each point represents neutrophils isolated from a single mouse (2 experiments; n=4). (D) Neutrophils were cultured with S. aureus (SA; MOI=1) and restriction of S. aureus growth after 6 hr was quantified by colony forming unit (CFU) spot plating. Percent growth of S. aureus calculated relative to S. aureus growth in the absence of neutrophils. Each point represents the mean result (technical triplicate) of neutrophils isolated from a single mouse (2 experiments; n=3). (E-F) Mice were systemically infected (inf.) with S. aureus (CFU=2×10⁷) and at 4 dpi, organs were homogenized and (E) the percentage of neutrophils undergoing suicidal NET formation was quantified by flow cytometry. Each point represents a single mouse (3 infections; mock n=6, WT; inf. n=14, MRL/lpr; inf. n=9). (F) Supernatant from homogenized organs were used to quantify NET abundance in the tissues by ELISA. Capture antibodies directed towards myeloperoxidase (MPO) or neutrophil elastase (NE). NET DNA abundance was quantified using a DNA standard and normalized to the mass of the tissue. Each point represents the mean result (technical duplicate) from supernatants isolated from a single mouse (1 infection; n=5). Two-way ANOVA with (B-C) Tukey or (D) Sidak’s multiple comparisons test, (E) one-way ANOVA or with Tukey multiple comparison test, or (F) unpaired t-test (*P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001, ns=not significant).

To identify whether impaired production of mitochondrial O₂⁻ in MRL/lpr neutrophils prevents suicidal NET formation, neutrophils were treated with rotenone to increase mitochondrial O₂⁻ production (Fig. 3D, Supplemental Fig. 2D). In addition to increasing mitochondrial O₂⁻ levels, rotenone accelerated and enhanced suicidal NET formation such that rotenone-treated MRL/lpr neutrophils were comparable to WT (Fig. 4B, Supplemental Fig. 3B). Coinciding with increasing suicidal NET formation at 240 min post inoculation, the percentage of MRL/lpr neutrophils undergoing vital NET formation was decreased to levels comparable to WT (Fig. 4C, Supplemental Fig. 3C). In addition, rotenone treatment decreased primary degranulation in both WT and MRL/lpr neutrophils (Supplemental Fig. 3D). These results demonstrate that impaired production of mitochondrial O₂⁻ by MRL/lpr neutrophils prevents suicidal NET formation, while sustaining vital NET formation long term.

Impairing suicidal NET formation decreases bactericidal activity of neutrophils in response to S. aureus (18, 27, 28). To determine whether the impaired suicidal NET formation observed in MRL/lpr neutrophils similarly restricts bactericidal activity, neutrophils were cultured with S. aureus and bacterial killing was quantified by dilution spot plating. Coinciding with altered NET formation, MRL/lpr neutrophils failed to restrict S. aureus growth as well as WT neutrophils at 2 and 6 hr post inoculation (Fig. 4D). These results demonstrate that MRL/lpr neutrophils have a significant loss in bactericidal activity.

To determine whether the impaired suicidal NET formation observed in vitro was present in MRL/lpr mice, mice were infected and the percentage of neutrophils undergoing NET formation was quantified by flow cytometry 4 dpi. Coinciding with higher bacterial burdens (Fig. 1B), suicidal NET formation was diminished 2.3-fold in the heart of MRL/lpr mice but not significantly different in the liver and kidney relative to WT (Fig. 4E, Supplemental Fig. S3F). Consistent with a lower percentage of neutrophils undergoing suicidal NET formation in the heart, a decreased abundance of NETs was quantified from the supernatant of heart tissue from MRL/lpr mice compared to WT by ELISA (Fig. 4F). Finally, vital NET formation and primary degranulation were comparable in the heart and liver, but decreased in the kidney of A9^−/− mice compared to WT (Supplemental Fig. 3G–H). These data demonstrate that the decrease in suicidal NET formation in the heart during SLE may provide a favorable environment for S. aureus and contribute to the increased bacterial burdens and enhanced mortality during S. aureus infections.

Altered neutrophil activity in patients with SLE

MRL/lpr mice are genetically identical and subjected to a controlled environment; however, multiple genetic and environmental factors contribute to SLE in patients. Therefore, we sought to determine how pervasive the impaired mitochondrial O₂⁻ and NET formation phenotypes are in patients with SLE. Polymorphonuclear (PMN) cells were isolated from the peripheral blood of healthy control (HC) and SLE patients (Table 1) and co-cultured with S. aureus. Overall, the level of intracellular S100A9 protein was 1.7-fold higher in neutrophils from patients with SLE compared to HC (Fig. 5A, Supplemental Fig. 4A). Similar to MRL/lpr mice, neutrophils from patients with SLE produced less mitochondrial O₂⁻ relative to HC following co-culture with S. aureus (Fig. 5B, Supplemental Fig. 4B). Furthermore, after a 240 min culture with S. aureus, samples from patients with SLE showed a 5.5-fold increase in the percentage of neutrophils undergoing vital NET formation (Fig. 5C, Supplemental Fig. 4C) and a 3.3-fold decrease in suicidal NET formation (Fig. 5D, Supplemental Fig. 4D). Primary degranulation was comparable between SLE and HC neutrophils (Supplemental Fig. 4E). Consistent with impaired suicidal NET formation, PMNs from patients with SLE failed to restrict S. aureus growth to the same levels as PMNs from HC (Fig. 5E), suggesting SLE PMNs have impaired bactericidal activity. Disease activity, as measured by SLE disease activity index (SLEDAI; 58), and patient treatments did not account for the altered response of neutrophils from patients with SLE to S. aureus (Table 1, Fig. 5). These results demonstrate that the diminished mitochondrial O₂⁻, suicidal NET formation, and antimicrobial activity identified in neutrophils from MRL/lpr mice and patients with SLE, regardless of disease activity. Thus, accumulation of intracellular S100A9 coinciding with impairing suicidal NET formation may contribute to the increased incidence of S. aureus infections in patients with SLE.

TABLE 1.

Distribution of patient samples.

	Healthy (n=12)	All SLE (n=12)	Active SLE (SLEDAI≥6) (n=7)	Inactive SLE (SLEDAI<6) (n=5)
Age
	41.7 (24–64)	46.8 (34–57)	50.0 (45–57)	44.9 (34–52)
Sex
Female	12	12	7	5
Male	0	0	0	0
Race
White	8	7	4	3
Black	4	5	3	2
Ethnicity
hispanic	1	1	1	0
nonhispanic	11	11	6	5
Medications
PDS	-	4	4	0
HCQ	-	9	7	2
cDMARD*	-	8	5	3
bDMARD	-	1	1	0
NSAIDs	-	3	1	2

cDMARD = methtrexate, leflunamide, azathioprine,mycophenolate mofetil, tacrolimus

bDMARD= adalimumab

NSAIDs = aspirin, diclofenac, nabumetone

FIGURE 5.

Increased S100A9 in neutrophils from patients with SLE correlates with decreased mitochondrial O₂⁻ and impaired suicidal NET formation. PMNs were isolated from human peripheral blood of healthy control (HC) and SLE patients, and (A) the intracellular abundance of S100A9 was quantified by flow cytometry. MFI normalized by isotype. Each point represents PMNs isolated from a single donor (12 experiments; n=12). (B-D) PMNs were cultured with S. aureus (MOI=10) and (B) mitochondrial O₂⁻ production in PMNs (CD15⁺CD16⁺) and the percentage of PMNs undergoing (C) vital (Live: extracellular dsDNA⁺MPO⁺H3Cit⁺) and (D) suicidal (Dead: extracellular dsDNA⁺MPO⁺H3Cit⁺) NET formation was quantified by flow cytometry. (B) mitoSOX MFI normalized by mitoTracker MFI. Each point represents PMNs isolated from a single donor (12 experiments; n=12). (E) PMNs were cultured with S. aureus (SA; MOI=1) and restriction of S. aureus growth after 6 hr was quantified by CFU spot plating. Percent growth of S. aureus calculated relative to S. aureus growth in the absence of PMNs. Each point represents the mean result (technical triplicate) of PMNs isolated from a single donor (12 experiments; n=12). ▲=Active disease (SLEDAI≥6), ▽=Inactive disease (SLEDAI<6). (A) Unpaired t-test or (B-E) two-way ANOVA with Sidak’s multiple comparisons test (*P≤0.05, **P0.01, ****P≤0.0001, ns=not significant).

Discussion

We report that neutrophils from MRL/lpr mice and SLE patients accumulate higher levels of intracellular S100A9 protein and produce lower levels of mitochondrial O₂⁻ in response to S. aureus. As a result of decreased production of mitochondrial O₂⁻, MRL/lpr neutrophils have impaired suicidal NET formation but heightened vital NET formation in response to S. aureus. In mice, the decrease in suicidal NET formation correlates with a specific increase in bacterial burdens in the heart, and decreased survival during systemic S. aureus infection. Consistent with prior observations (18), these findings demonstrate that elevated intracellular concentrations of S100A9 renders neutrophils less bactericidal towards S. aureus and may contribute to the increased susceptibility of SLE patients to bacterial infection.

Mitochondrial dysfunction has been associated with SLE pathology (45–49), and occurs in a mammalian target of rapamycin (mTOR)-dependent manner prior to overt immune dysregulation and tissue damage in MRL/lpr mice (59). We observe a comparable phenotype whereby MRL/lpr neutrophils produce higher levels of mitochondrial O₂⁻ (Fig. 2A) with increased proton leak (Fig. 1C) prior to stimulation with S. aureus. However, S. aureus stimulation results in a decrease in proton leak coinciding with increased oxygen consumption by non-mitochondrial processes (Fig. 1C), most likely reflecting increased NADPH oxidase activity. Similarly, MRL/lpr macrophages maintain a prolonged oxidative burst following phagocytosis of immunoglobulin G (IgG)-containing immune complexes thereby disrupting lysosome degradation in an mTOR-dependent manner (57, 60). As a result, MRL/lpr macrophages accumulate high levels of nuclear antigen on the cell surface that precedes and promotes many lupus-associated pathologies (57, 61, 62). While the accumulation of nuclear antigen has been observed on B cells, T cells, monocytes, dendritic cells, and macrophages (62), it is unclear whether accumulation occurs on neutrophils. Taken together, the accumulation of IgG-containing immune complexes that dysregulate macrophages during lupus may play a similar role in driving mitochondrial dysfunction and the elevated NET formation observed during SLE pathology.

The incidence of S. aureus infections in patients with SLE (38–41) may be partially attributed to immunosuppressants; however, MRL/lpr mice are susceptible to S. aureus infections independent of immunosuppressants (Fig. 3). Conversely, S. aureus can exacerbate SLE pathologies. Reflecting the importance of immune activation driven by S. aureus, chronic exposure to superantigens from S. aureus in WT mice results in a disease mimicking lupus (63), whereby type-1 interferon production associated with S. aureus infection and SLE antagonizes cutaneous lupus erythematosus lesions and facilitates colonization of the lesion by S. aureus (64). Our data demonstrate that decreased mitochondrial O₂⁻ and suicidal NET formation by neutrophils from MRL/lpr mice and patients with SLE could render neutrophils unsuitable to combat S. aureus (Fig. 2–4). As a result, this may create a feed-forward mechanism whereby autoimmunity renders the host more susceptible to infection, while increased S. aureus colonization promotes heightened SLE disease activity.

The observation that neutrophils from MRL/lpr mice and SLE patients undergo less suicidal NET formation in response to S. aureus was initially surprising. Whether altered NET formation contributes to SLE pathology by providing a source of nuclear antigen has been debated (31, 32, 65, 66); however, it has been consistently observed that murine and human SLE neutrophils undergoing higher levels of NET formation during SLE pathology (45–49, 59). Furthermore, increased NET deposition in the kidney contributes to the inflammatory response and type I interferon production associated with kidney nephritis (32, 67). In contrast, neutrophils from MRL/lpr mice and SLE patients exhibit impaired suicidal NET formation in response to S. aureus (Fig. 4A–B, 5D) with reduced NET formation in heart (Fig. 4E–F). NET formation in the kidney was comparable between WT and MRL/lpr mice, despite the MRL/lpr mice having elevated autoantibody titers and mild glomerulonephritis (Supplemental Fig. 1A–B). These results suggest that the signal transduction associated with autoantibodies triggering NET formation during SLE (68) may differ or conflict with pathogen-mediated NET formation. Therefore, while the kidney provides a unique metabolic/immunologic niche that promotes NET formation during SLE, these same autoimmune signals may impair suicidal NET formation in the heart during infection. In addition, we have determined that murine and human SLE neutrophils sustain heightened vital NET formation and fail to undergo suicidal NET formation (Fig 4B–C, 5C–D). The molecular mechanism regulating vital NET formation is unclear and may account for discrepant reports regarding the role for NET formation in SLE (31, 32, 65, 66). Identifying whether vital and suicidal NETs are immunogenically distinct is necessary to unravel the complexities of NET formation in SLE.

Key points.

• Altered mitochondrial homeostasis in SLE neutrophils impairs superoxide production.

• NET formation by SLE neutrophils is decreased in response to S. aureus.

• SLE neutrophils are less bactericidal towards S. aureus.