Key Points
Neutrophils produce high levels of Cxcl2 in response to whole bacteria.
Cxcl2 production depends on sensing of bacterial nucleic acids by TLR7, 9, and 13.
Cxcl2 acts autocrinously to enhance antibacterial activities and its own production.
Visual Abstract
Abstract
The influx of neutrophils to infection sites is a fundamental step in host defenses against the frequent human pathogen group B Streptococcus (GBS) and other extracellular bacteria. Using a mouse model of GBS-induced peritonitis, we show in this study that the chemokines Cxcl1 and Cxcl2 play distinctive roles in enhancing the recruitment and the antibacterial activities of neutrophils in a manner that is linked to differences in the cellular sources of these mediators. Cell depletion experiments demonstrated that neutrophils make a significant contribution to the in vivo production of Cxcl2 but not Cxcl1. In vitro, neutrophils responded weakly to LPS but released high levels of Cxcl2 after stimulation with GBS or other bacteria. Neutrophil-derived Cxcl2 acted in an autocrinous manner to increase its own production and to enhance antibacterial activities, including the release of oxygen radicals. In both neutrophils and macrophages, the production of Cxcl1/2 largely required the presence of functional UNC93B1, a chaperone protein involved in signaling by endosomal TLRs. Moreover, the phenotype of UNC93B1-defective phagocytes could be recapitulated by the simultaneous absence of TLR7, 9, and 13 but not by the absence of individual TLRs. Collectively, our data show that neutrophils recognize Gram-positive and Gram-negative bacteria by means of multiple phagosomal TLRs, resulting in de novo synthesis of Cxcl2, amplification of neutrophil recruitment, and potentiation of their antibacterial activities. These data may be useful to devise alternative therapeutic strategies aimed at enhancing the recruitment and the functional activities of polymorphonuclear leukocytes during infections caused by antibiotic-resistant bacteria.
Introduction
Neutrophils are the most abundant nucleated cells in the blood and are the body’s main guardians against bacterial and fungal infections. These leukocytes have the ability to kill micro-organisms either intracellularly, by means of phagocytosis, or extracellularly, after the release of an arsenal of antimicrobial products. Although they mostly receive attention for their phagocytic and killing abilities, neutrophils can perform a variety of other important functions, including tissue remodeling, Ag presentation, recruitment of other leukocytes, and polarization of T cell responses (1–6). Neutrophils are produced in the bone marrow and are rapidly recruited from the blood into tissues in response to a variety of chemoattractants, such as formylated peptides released by bacteria and damaged host cells or mediators produced by resident cells during inflammation (7, 8). These mediators include lipid metabolites, such as leukotriene B4 (LTB4) and chemokines, particularly those belonging to the CXCL8 family, which all bind to CXCR1 and CXCR2 receptors and share a common glutamic acid–leucine–arginine (ELR)+ motif in their structure (9, 10). The human chemokine CXCL8 (IL-8) is believed to play a crucial role in neutrophil recruitment. A CXCL8 orthologue does not exist in mice, although many members of the CXCL family have been identified in these animals.
The murine chemokines Cxcl1 (formerly named keratinocyte-derived chemokine [KC]) and Cxcl2 (or MIP-2) are highly homologous, bind to the same receptor (CxcR2) and are believed to be functionally similar to human CXCL8 (11). Activation of CxcR2, a G protein–coupled receptor (or GPCR), by Cxcl1/2 leads to downstream signaling through the vasodilator-stimulated phosphoprotein (VASP), PI3K, and RAC, which localize asymmetrically along the cell and orchestrate the directed migration of neutrophils (12). Notably, Cxcl1/2 has the ability to bind glycosaminoglycans in the extracellular matrix, thus generating chemoattractant gradients over distances of hundreds of microns along tissues (13). Both resident cells (including macrophages, epithelial and endothelial cells) and inflammatory leukocytes can produce Cxcl1/2 in the presence of injury and/or infection. Moreover, upregulation of these chemokines can be induced by a number of proinflammatory cytokines, including IL-1β, IL-1α, TNF-α, and G-CSF (14).
Extracellular encapsulated bacteria, such as streptococci and staphylococci, are extremely frequent human pathogens that typically cause the formation of neutrophil-rich exudates, including pus. Among these bacteria, group B streptococci (GBS) are emerging as an increasing cause of serious diseases, including sepsis, meningitis, arthritis, osteomyelitis, pneumonia, and skin/soft tissue infections (15). GBS persists as the leading cause of sepsis and meningitis in neonates (16, 17), whereas invasive infections in adults, especially in the elderly and in patients with underlying chronic diseases, are constantly increasing (18, 19). The mechanisms underlying innate immune defenses against GBS have been fairly well characterized over the last 25 years using murine infection models (20, 21). Various TLRs, including TLR2/7/13 are activated by these pathogens, leading to cytokine responses that ultimately result in potentiation of phagocytic killing and bacterial clearance (22–26). Neutrophil infiltrates are prominent at sites of GBS infection (27–29), and these cells play a dominant role in host defenses against these bacteria, as suggested by both clinical (30, 31) and experimental (32, 33) observations. Despite this, little is known of the mechanisms governing neutrophil recruitment to sites of infection. The release of Cxcl1/2 is promoted, in part, by IL-1R signaling during experimental GBS disease (20, 32), but their role during infection by these and other bacterial pathogens is incompletely understood. In particular, although the LPS component of Gram-negative bacteria can directly stimulate the production of Cxcl1/2 in macrophages through the activation of TLR4 (34), whether and by which mechanism chemokine responses can be directly induced by GBS or other bacteria is presently unclear.
In the current study we sought to identify the cellular sources of Cxcl1/2 and the mechanisms involved in the production of these chemokines during GBS infection. We found that GBS can directly induce the production of Cxcl1/2 in both macrophages and neutrophils by mechanisms involving the nucleic acid–sensing receptors TLR7/9/13. Moreover, we show that neutrophils are able to enhance their own recruitment to infection sites and their antibacterial activities using a positive feedback mechanism driven by high-level Cxcl2 release.
Materials and Methods
Mice
C57BL/6 wild-type (WT) mice and IL-1R−/− mice were purchased from Charles River Laboratories. Mice lacking single TLRs (TLR2, 3, 7, or 9) or TLR adaptors (MyD88, TRAM, TRIF, or MAL) were obtained from S. Akira (Osaka University, Osaka, Japan). The 3d mutant mice bearing the H412R mutation in the chaperone protein UNC93B1 were obtained from B. Beutler (University of Texas Southwestern Medical Center, Dallas, TX). Heterozygous TLR13−/+ mice were provided by the Knockout Mouse Project Repository (www.komp.org) and the Mouse Biology Program (www.mousebiology.org) at the University of California Davis. Subsequently, TLR13−/− mice were bred in the Animal Facility of the Department of Pathology of the University of Messina (Messina, Italy) as described previously (35). TLR7, 9, 13−/− triple-knockout (KO) mice were generated in the above animal facility by crossing TLR9−/− with TLR7−/− and TLR13−/− mice. All KO mice, bred on a C57BL/6J background, were born and developed normally. All mice used in the current study were housed under specific pathogen–free conditions in individually ventilated cages.
Bacterial strains and culture
GBS WT strain H36B (serotype Ib) was used for most experiments. A clinical Pseudomonas aeruginosa isolate (B18 strain), Staphylococcus aureus strain Newman, Streptococcus pneumoniae strain D39, and encapsulated Escherichia coli strain K1 E-R8 were also used in selected experiments. Bacteria were grown in Todd Hewitt broth at 37°C with 5% CO2 to the midlog phase, washed twice in nonpyrogenic PBS (0.01 M phosphate, 0.15 M NaCl [pH 7.4]; EuroClone) and resuspended to the desired concentration in PBS.
Murine peritonitis model
Six-week-old female mice were injected i.p. within 0.2 ml of PBS containing the indicated doses of GBS grown to the midlog phase. In some experiments, mice were treated with penicillin (500 IU i.p.) at 1 h after GBS challenge to prevent bacterial overgrowth. For each experiment, the actual number of injected bacteria was determined by colony counts. In experiments dealing with LPS- or zymosan-induced peritonitis, E. coli K12 ultrapure LPS (InvivoGen) or zymosan (InvivoGen) were injected i.p. at the indicated doses in 0.2 ml of PBS. Peritoneal lavage fluids (PLF) were collected at various times after challenge to measure host cell numbers by flow cytometry, cytokine concentrations, and bacterial CFUs. In preliminary experiments, mice were injected i.p. with 0.2 ml of PBS vehicle, and PLF samples were collected at 3, 6, and 24 h after injection. Nonsignificant elevations in cell numbers were detected in these samples over baseline values, whereas Cxcl1/2 concentrations were always below the detection limit (data not shown). For these reasons vehicle controls were not run in subsequent studies.
To ablate macrophages, mice were injected i.p. with a suspension of clodronate liposomes (1 mg clodronate in 0.2 ml of PBS) or control PBS liposomes (36). Briefly, phosphatidylcholine (86 mg) and cholesterol (19 mg; both from Sigma-Aldrich) were dissolved in chloroform (5.0 ml) in a 1-l flask. The organic phase was removed under a flow of nitrogen. The lipid film formed at the bottom of the flask was dispersed at room temperature for 15 min with 10 ml of plain PBS or PBS containing 0.7 mol/l clodronate (Roche). The resulting suspension was incubated for 1 h at room temperature and then sonicated for 10 min in a water bath. After incubation at room temperature for an additional hour, liposomes were diluted in 50 ml of PBS and centrifuged twice at 100,000 × g for 30 min at 20°C. The final pellet was washed once and resuspended in 20 ml of PBS. Neutrophil depletion was achieved by i.v. injection of 100 μg of rat monoclonal anti-mouse Ly-6G Ab (clone 1A8) or rat IgG2a control (isotype control), both from BD Pharmingen, at 24 h before i.p. inoculation with 2 × 105 CFU of GBS.
Flow cytometry
Flow cytometry analysis of leukocyte subsets in PLF was performed on a FACS Canto II flow cytometer (BD Biosciences) as previously described (37). Briefly, all cells were incubated with 0.5 μg of Mouse Fc Block (Purified Rat Anti-Mouse CD16/CD32, clone 2.4G2; BD Pharmingen) for 20 min before staining for 20 min with Abs directed against F4/80 for macrophages (F4/80 mAb BM8 eFluor 450; eBioscience), Ly-6G for neutrophils (PE Rat Anti-Mouse Ly-6G, clone 1A8; BD Pharmingen) or CD11c for dendritic cells (PE Armenian Hamster Anti-Mouse CD11c, clone N418; BD Pharmingen) using the respective isotype Abs as controls. Cell percentages/counts were determined in PLF using the gating strategy depicted in Supplemental Fig. 1A and BD TruCount tubes (BD Biosciences). Data analysis was performed using Flowing Software 2.5.1.
Isolation and stimulation of bone marrow–derived neutrophils and macrophages
Neutrophils were obtained from the bone marrow of WT and immune-deficient mice using Percoll density gradient centrifugation as previously described (38) with some modifications. Briefly, after removing the femurs and the tibias and cutting off the epiphyses of the bones, bone marrow cells were spun out of the bones by centrifugation and resuspended in Dulbecco’s PBS without Ca2+ and Mg2+ (DPBS; EuroClone). Cell aggregates were carefully dissociated with a sterile Pasteur Pipette, and the suspension was then flushed onto a 50-ml conical tube through a 70-μm cell strainer (EuroClone) to remove any remaining bone pieces. Cells were collected by centrifugation at 400 × g for 15 min and resuspended in 5 ml of DPBS. For isolation of neutrophils, bone marrow cells were carefully layered on the top of a 62.5% Percoll (GE Healthcare Life Sciences) layer. After centrifugation at 1060 × g for 30 min at 4°C, pellets (neutrophils plus RBCs) were washed twice with DPBS, and erythrocytes were hypotonically lysed by shaking the cell button gently for 20 s with 20 ml of 0.2% NaCl. Isotonicity was then restored with 20 ml of 1.6% NaCl. After two washing steps with DPBS, neutrophils were suspended in RPMI 1640 containing 10% (v/v) FCS. The viability of cells obtained via this procedure was routinely >90%, as assessed by trypan blue exclusion assay. Isolated cells (15–20 ± 0.6 × 106 cells per mouse) were stained with May/Grunwald/Giemsa (39), and ∼90% of them were morphologically mature neutrophils (bands and segmented). Moreover, purity of neutrophil populations was >95%, as determined by flow cytometry using the neutrophil-specific marker Ly-6G. In selected experiments, Percoll-isolated cells were further treated with the anti-CD115 MACS Bead Separation Kit (Miltenyi Biotec) to remove contaminating macrophages, as described (40).
To obtain bone marrow–derived macrophages (BMDMs), marrow cells were cultured for 6–7 d in RPMI 1640 supplemented with 10% FCS, penicillin (50 IU/ml), and streptomycin (50 μg/ml). Medium was supplemented with 100 ng/ml M-CSF (PeproTech) to obtain BMDMs.
Isolated bone marrow–derived neutrophils and macrophages (5 × 105 per well in 0.2 ml of RPMI 1640 supplemented with 10% FCS) were seeded in microtiter plates and stimulated with GBS grown to the midlog phase at the indicated multiplicities of infection (MOI). All infections were carried out by centrifuging cell suspensions for 10 min at 400 × g to facilitate bacteria/neutrophil interactions. After incubation for 1 h at 37°C with 5% CO2, penicillin (250 IU/ml) and streptomycin (250 μg/ml) were added to kill extracellular bacteria. In preliminary experiments, we verified that this antibiotic treatment did not affect viability of intracellular bacteria, as shown by CFUs in cell lysates. Control wells were stimulated with E. coli K12 ultrapure LPS (InvivoGen). We chose to use LPS as a control for experiments involving both Gram-positive and Gram-negative bacteria because other isolated bacterial components, such as GBS peptidoglycan, induced little chemokine production in neutrophils, as determined in preliminary studies. In some experiments, neutrophils and BMDMs were treated with actinomycin D (5 μg/ml) or cycloheximide (5 μg/ml), both from Sigma-Aldrich. Cell culture supernatants were collected at 24 h and stored at −80°C for cytokine measurements.
Cxcl2 mRNA expression analysis
Cytokine measurement
Reactive oxygen species measurement
Statistical analyses
Differences in cytokine levels and cell counts were assessed by Student t test or one-way ANOVA and Bonferroni multiple-comparison posttest. Differences in bacterial CFU counts were assessed by the Mann–Whitney U test. Differences were considered statistically significant when p values were lower than 0.05 (p < 0.05). Statistical analyses were performed with GraphPad Prism 5.0 (GraphPad Software, San Diego, CA).
Ethics statement
All studies were performed in strict accordance with the European Union guidelines for the use of laboratory animals. The procedures were approved by the Animal Welfare Committee of the University of Messina and by the Ministero della Salute of Italy.
Results
Neutrophil recruitment to sites of GBS infection requires both Cxcl1 and Cxcl2
In initial experiments, we analyzed inflammatory cell influx in response to the i.p. injection of graded doses of live GBS. Under these conditions, the influx of leukocytes in the peritoneal cavity increased with increasing bacterial doses (Fig. 1A). Moreover, there was a positive correlation between the number of inflammatory cells and the number of bacteria found in PLF samples (Fig. 1B). Because individual mice differed in their ability to control infection and in the accompanying severity of inflammation, we sought to improve the reproducibility of the assay by using antibiotics to prevent bacterial overgrowth. We found that the administration of penicillin at 1 h after bacterial challenge was associated with a reproducible cell influx that reached peak levels at 6 h (Fig. 1C). Under these conditions, 80–90% of PLF cells were represented by neutrophils at 1–12 h post challenge. In contrast, a significant influx of macrophages and dendritic cells was observed only at 24 h (Fig. 1C). Next, it was of interest to compare this live bacteria model with classical peritonitis models using the LPS or zymosan cell wall components of, respectively, Gram-negative bacteria and fungi. To this end, in preliminary experiments, we injected mice i.p. with LPS or zymosan and found that the doses inducing maximal neutrophil influx were 10 ng and 0.2 mg, respectively (data not shown). Next, when comparing our live bacteria model with these classical peritonitis models, it was found that GBS were considerably more efficient at recruiting neutrophils than LPS or zymosan, even when these stimuli were used at the optimal doses (Fig. 1D). To gain insights into the role of Cxcl1/2, the main neutrophil-attracting chemokines, we measured their concentrations in GBS-induced peritoneal exudates. Cxcl1/2 elevations were detectable early after bacterial challenge and peaked at 3 h, thus associating in timing with neutrophil influx (Fig. 1E, 1F). Notably, Cxcl2 but not Cxcl1 levels were significantly higher in GBS-induced exudates relative to those observed with the classical proinflammatory stimuli LPS or zymosan (Fig. 1E, 1F). Therefore, this first set of data indicated that GBS induced a pronounced neutrophil influx that was proportional to the number of bacteria present at the infection site and was associated with robust production of Cxcl2.
Cell influx and chemokine concentrations in the peritoneal cavity of mice inoculated with GBS. (A) Cell influx at different times after i.p. administration of GBS at the indicated bacterial doses (CFU). Data represent means + SDs of three duplicate determinations, each conducted in a different animal. (B) Correlation between the numbers of inflammatory cells and bacterial CFUs in PLF samples; data were taken from the experiments shown in (A). (C) Kinetics of recruitment of various cell types after i.p. administration of GBS (2 × 107 CFU) followed 60 min later by penicillin (500 IU); Ly-6G+ neutrophils, F4/80+ macrophages, CD11c+ dendritic cells. Data represent means + SDs of three duplicate determinations, each conducted in a different animal. (D) Kinetics of neutrophil influx in the peritoneal cavity after i.p. challenge with GBS (2 × 107 CFU followed 1 h later by 500 IU of penicillin), LPS (10 ng), or zymosan (0.2 mg). Data represent means ± SDs of three duplicate determinations, each conducted in a different animal. Cxcl1 (E) and Cxcl2 (F) protein levels in PLF samples at the indicated times after i.p. injection of GBS (2 × 107 CFU followed 1 h later by 500 IU of penicillin), LPS (10 ng), or zymosan (0.2 mg). Means ± SDs of three duplicate determinations, each conducted in a different animal. *p < 0.05, **p < 0.01, ***p < 0.001 versus GBS as determined by one-way ANOVA and Bonferroni posttest.
To explore the role of these chemokines in neutrophil recruitment, we injected mice i.p. with neutralizing anti-Cxcl1, anti-Cxcl2 mAbs, or control IgG before GBS challenge. Either anti-chemokine Ab could significantly reduce GBS-induced neutrophil recruitment (Fig. 2A). Moreover, the combined administration of anti-Cxcl1 and anti-Cxcl2 did not further reduce neutrophil influx relative to the levels observed with either Ab alone (Fig. 2A, 2B). Of note, none of the mAb treatments affected the number of circulating neutrophils, thus excluding depletion of these cells as a cause of their reduced influx into the peritoneal cavity (data not shown). Next, we asked whether blocking the activity of Cxcl1/2 in this model would have an impact on antibacterial host defenses. Therefore, we injected mice with neutralizing anti-Cxcl1, anti-Cxcl2 mAb, or control IgG before challenge with GBS and measured bacterial numbers in PLF samples collected at 3 h post challenge. Either Cxcl1 or Cxcl2 blockade resulted in an increase in bacterial numbers, although this effect was significantly more pronounced using anti-Cxcl2 compared with anti-Cxcl1 (Fig. 2C). These data suggest that Cxcl1 and Cxcl2 play distinct, nonredundant roles and act cooperatively in GBS-induced neutrophil recruitment. Moreover, Cxcl2 plays an important role in potentiating antibacterial defenses.
Effect of chemokine neutralization and macrophage depletion on neutrophil recruitment and antibacterial defenses. Neutrophil counts (A and B) were determined in peritoneal exudates of mice at the indicated times after i.p. challenge with GBS (2 × 107 CFU followed 1 h later by 500 IU of penicillin). Neutralizing Abs directed against Cxcl1, Cxcl2, or control IgG were administered i.p. as a single or combined treatment at 3 h before GBS challenge. (C) Bacterial numbers (CFUs) in PLF samples obtained at 3 h after GBS challenge (2 × 107 CFU). Neutralizing Abs directed against Cxcl1, Cxcl2, or control IgG were administered i.p. as a single or combined treatment at 3 h before GBS challenge. Data are expressed as the means ± SDs of five duplicate observations, each conducted on a different animal. Kinetics of neutrophil influx (D) and Cxcl1/2 production (E and F) in PLF samples after treatment with clodronate or control liposomes and subsequent i.p. challenge with GBS (2 × 107 CFU followed 1 h later by 500 IU of penicillin). Means + SDs of three duplicate determinations, each conducted in a different animal. *p < 0.05, **p < 0.01, ***p < 0.001 versus control as determined by unpaired t test (A, B, and D–F) or Mann–Whitney U test (C).
Neutrophils are a significant source of Cxcl2 during GBS infection
Because macrophages are strong chemokine producers in response to a number of stimuli, including LPS (34), we investigated the in vivo impact of these cells on Cxcl1/2 production during GBS infection using a macrophage ablation strategy. To this end, mice were treated with clodronate liposomes (36), which resulted in a 95% reduction in macrophage numbers in the peritoneal cavity but not in any significant reduction in blood neutrophil numbers (data not shown). Clodronate liposome treatment was associated with markedly decreased levels of both Cxcl1 and Cxcl2 in PLF samples, concomitantly with a reduction in GBS-induced neutrophil influx, whereas control liposomes had no effect (Fig. 2D–F), suggesting that macrophages are key to GBS-induced chemokine production. However, because neutrophils can also produce chemokines, it could not be discerned from our data whether the observed clodronate-induced reduction in chemokine levels resulted from macrophage ablation per se or from the reduced neutrophil numbers in the exudates or from both. To gain further insights, we examined the effects of in vivo neutrophil depletion on GBS-induced Cxcl1/2 release. Animals were injected with anti–Ly-6G clone 1A8 mAb, which is highly specific for neutrophils (41), or with an equal amount of isotype control Ig. Anti–Ly-6G treatment was sufficient to reduce neutrophil blood counts to <2% by 24 h, and these low numbers persisted for at least 72 h after treatment (Supplemental Fig. 1B). Next, anti–Ly-6G– or control Ig-treated mice were inoculated i.p. with GBS, and cell influx and chemokine concentrations were measured in PLF over time. As expected, neutrophils were almost completely absent in peritoneal exudates from anti–Ly-6G–treated but not control Ig-treated mice after GBS challenge (Fig. 3A). Under these conditions, no significant reduction was observed in the levels of Cxcl1 in the neutrophil-depleted mice relative to Ig-treated controls (Fig. 3B). In contrast, Cxcl2 concentrations were significantly decreased in the neutrophil-depleted animals (Fig. 3C). Collectively, these data suggest that neutrophils significantly contribute to Cxcl2 responses to GBS, although they are apparently dispensable for the secretion of Cxcl1.
Effect of neutrophil depletion on chemokine production. Kinetics of neutrophil influx (A) and Cxcl1/2 production (B and C) in PLF samples after administration of anti–Ly-6G or isotype control mAbs and subsequent i.p. challenge with GBS (2 × 107 CFU followed 1 h later by 500 IU of penicillin). Means ± SDs of three duplicate determinations, each conducted in a different animal. *p < 0.05, **p < 0.01 versus isotype control as determined by unpaired t test.
Cxcl1/2 are synthesized de novo following GBS recognition
To investigate whether macrophages or neutrophils release chemokines directly in response to GBS, we used ex vivo isolated cell populations. We preliminarily investigated whether Cxcl1/2 are constitutively stored inside neutrophils because these cells mostly function by releasing a large number of preformed proinflammatory and antimicrobial products in response to activating stimuli. However, cell lysates from peritoneal macrophages or bone marrow–derived neutrophils contained undetectable levels of chemokines, as measured by ELISA (data not shown). In further experiments, bone marrow–derived neutrophils and macrophages were exposed in vitro to various doses of GBS, and chemokine concentrations were measured in supernatants at 24 h after stimulation. As a positive control, we used LPS, a known inducer of Cxcl1/2 in macrophages (42). Both GBS and LPS induced high-level, dose-dependent production of Cxcl1/2 in macrophages (Fig. 4A, 4B). In contrast, although neutrophils were relatively weak producers of Cxcl1 (Fig. 4A), they could mount robust Cxcl2 responses to GBS (Fig. 4B). Surprisingly, such responses were one-to-two orders of magnitude higher than those induced by LPS. It is unlikely that chemokine production by neutrophil preparations is accounted for by contamination with mature macrophages or dendritic cells because F4/80+ or Cd11c+ cells were not present in purified neutrophil preparations (Supplemental Fig. 1C). To further exclude the possible contribution of other mononuclear phagocytes, cells expressing CD115 (a marker of monocytes and macrophage/dendritic cell precursors) were removed with anti-CD115 Abs. However, removal of CD115+ cells did not affect chemokine production in response to GBS (Supplemental Fig. 1D). When phagocytes were pretreated with cycloheximide, an inhibitor of protein synthesis, Cxcl1/2 release was markedly reduced in both macrophages and neutrophils, thus indicating that the chemokines are produced de novo in response to GBS (Supplemental Fig. 2A, 2B). Moreover, blocking transcription with actinomycin D during the first hour of stimulation delayed production of both chemokines (Supplemental Fig. 2C, 2D). Collectively, these data indicated that GBS can directly induce de novo production of Cxcl1/2 in myeloid cells and that neutrophils are as efficient as macrophages in mounting Cxcl2 responses to these bacteria but not to LPS.
Cxcl2 acts autocrinously to amplify its own production and neutrophil antibacterial activities. (A and B) In vitro chemokine production in bone marrow–derived neutrophils and macrophages (both 5 × 105 cells) stimulated with GBS. Cxcl1 (A) and Cxcl2 (B) were measured at 24 h after treatment with increasing MOIs (2, 5, 10, 20) of GBS or increasing LPS doses (1, 10, 100, and 1000 ng/ml). *p < 0.05, **p < 0.01, ***p < 0.001 as determined by unpaired t test. (C) Real time-quantitative PCR assessment of Cxcl2 mRNA levels in GBS-stimulated (MOI 5) neutrophils in the presence of neutralizing anti-Cxcl1 or anti-Cxcl2 Abs. (D) Effect of exogenous Cxcl2 on GBS-induced Cxcl2 transcription. Recombinant chemokines (1.25 μg/ml final concentration) were added to neutrophil cultures at the time of GBS (MOI 5) addition. (E) Dose–response curve of Cxcl1- or Cxcl2-induced Cxcl2 mRNA expression in unstimulated neutrophils. (F) Bacterial-killing assay in the presence of neutralizing anti-Cxcl1 or anti-Cxcl2 Abs. Normal mouse IgG was used as a control. Data are expressed as percentages of the number of CFUs in the presence of neutrophils relative to the number of CFUs in the absence of neutrophils. In (A)–(F), data are expressed as means + SD of data from three independent experiments conducted in duplicate. *p < 0.05, **p < 0.01 versus control IgG, one-way ANOVA and Bonferroni posttest. (G) Effect of neutralizing anti-chemokine Abs on the production of ROS in neutrophils. ROS production by GBS-stimulated neutrophils (MOI 5) pretreated with neutralizing Abs directed against Cxcl1, Cxcl2, or with control IgG. Fluorescence histograms and δ median fluorescence intensities (ΔMFI) in unstimulated and GBS-stimulated cells. The shadowed area indicates the shift to the right of the histogram of stimulated cells relative to unstimulated cells. Shown are data from a representative experiment of six showing similar results. ΔMFI values from the other five experiments are shown in Supplemental Fig. 2E.
Neutrophil-derived Cxcl2 acts autocrinously to activate neutrophil functions
Next, we investigated the mechanisms underlying the ability of neutrophils to produce high levels of Cxcl2 in the presence of GBS. Because Cxcl2 was recently found to act synergistically with immune complexes to induce the expression of more Cxcl2 (43), we asked whether a similar mechanism was effective during GBS stimulation. Therefore, we looked at the effects of neutralizing anti-Cxcl1 or anti-Cxcl2 Abs on Cxcl2 mRNA levels in GBS-stimulated neutrophils. We found that Cxcl2 but not Cxcl1 blockade decreased Cxcl2 transcription (Fig. 4C). Moreover, the addition of recombinant Cxcl2 markedly increased Cxcl2 transcription in the presence of GBS (Fig. 4D). However, the addition of exogenous Cxcl2 by itself induced only a moderate increase in Cxcl2 mRNA levels (Fig. 4E). Collectively these data indicate that, albeit capable of inducing by itself only slight elevations in the transcription of its own gene, Cxcl2 can strongly synergize with bacterial stimuli to enhance its production.
In view of the potent activity of endogenous Cxcl2 in promoting GBS clearance in our peritonitis model (Fig. 2C), we investigated whether this chemokine can promote key antibacterial functions in neutrophils. We initially observed that the ability of neutrophils to control in vitro bacterial growth was reduced in neutrophils pretreated with neutralizing anti-Cxcl2 but not anti-Cxcl1 (Fig. 4F). Moreover, Cxcl2 blockade also significantly reduced GBS-induced ROS production (Fig. 4G, Supplemental Fig. 2E). Collectively, these data indicate that during stimulation with GBS neutrophil-derived Cxcl2 activates an autocrinous loop, leading to the production of more Cxcl2 and to enhanced antibacterial activities of these cells.
Chemokine production in response to Gram-positive and Gram-negative bacteria requires endosomal TLRs
Because previous studies have demonstrated a prominent role of TLRs in GBS recognition and proinflammatory cytokine responses (44, 45), we sought to investigate whether these receptors are required for GBS-induced Cxcl1/2 production in macrophages and neutrophils. To this end, we used cells isolated from the bone marrow of mice with genetic defects in TLR adaptor/chaperone proteins or in single TLRs. The release of both Cxcl1 (Fig. 5A) and Cxcl2 (Fig. 5B) in response to GBS absolutely required MyD88, a TLR adaptor protein, but not other TLR adaptors, such as MAL, TRIF, or TRAM (46, 47).
Concentrations of Cxcl1 and Cxcl2 in supernatants of BMDM and neutrophil cultures after GBS stimulation. Supernatants were collected at 24 h post infection of 5 × 105 cells with GBS (MOI 5). Cells were obtained from WT mice or mice lacking the TLR adaptors MyD88, MAL, TRIF, or TRAM (A and B); animals lacking IL-1 (IL-1R) or IL-18 (IL-18R) receptors (C and D); and mice with genetic defects in single or multiple endosomal TLRs (E and F). Means + SD of three independent experiments conducted in duplicate. *p < 0.05, **p < 0.01, ***p < 0.001 versus WT mice as determined by one-way ANOVA and Bonferroni posttest.
Because MyD88 is involved in the transduction of signals originating not only from TLRs but also from IL-1R and IL-18R, in further experiments, we stimulated cells lacking IL-1R or IL-18R to ascertain whether IL-1/18 signaling might be involved in GBS-induced chemokine production. Macrophages but not neutrophils lacking IL-1R produced moderately lower chemokine levels (Fig. 5C, 5D). Moreover, absence of IL-18R in either cell type had no effect on chemokine production. Taken together, these data indicated that lack of IL-1/18Rs could not account for the total abrogation of Cxcl1/2 production observed in MyD88-deficient cells. We next tested cells from 3d mice, which have defective signaling of endosomal TLRs, such as TLR3/7/9/13, because of a mutation in UNC93B1, a chaperone protein required for the localization of TLRs to endosomal compartments (48). It was found that neutrophils from 3d mice were significantly impaired in their ability to produce Cxcl1/2 in response to GBS (Fig. 5E, 5F).
Moreover, 3d neutrophils were also impaired in their ability to respond to other Gram-positive and Gram-negative pathogens, including Streptococcus pneumoniae, Staphylococcus aureus, E. coli and P. aeruginosa, in terms of Cxcl2 production (Fig. 6). In contrast, a lack of single endosomal TLRs or of TLR2, a receptor involved in sensing GBS lipoproteins (49), as well as the simultaneous absence of TLR7 and 9 did not impair chemokine release (Fig. 5E, 5F, Supplemental Fig. 2F). Notably, however, the simultaneous absence of TLR7, 9, and 13 reconstituted the phenotype observed in the 3d mice (Fig. 5E, 5F, Supplemental Fig. 2F). Altogether, these data suggest that Cxcl1/2 production in neutrophils in response to bacterial infection requires MyD88-dependent transduction of signals originating from the simultaneous activation of TLR7, 9, and 13 (Fig. 7).
Concentrations of Cxcl2 in supernatants of bone marrow–derived neutrophils from mice with functionally defective UNC93B1 chaperone protein (3d mice). Supernatants were collected at 24 h post infection with graded doses (MOI 2.5, 5, 10, and 20) of P. aeruginosa strain B18 (A), Streptococcus pneumoniae strain D39 (B), E. coli strain K1 E-R8 (C), and Staphylococcus aureus strain Newman (D). Means + SD of three independent experiments conducted in duplicate. *p < 0.05, **p < 0.01, ***p < 0.001 versus WT mice as determined by one-way ANOVA and Bonferroni posttest.
Proposed mechanisms of Cxcl2 production in response to GBS in neutrophils. GBS are disrupted in phagolysosomes, where bacterial ssRNA, 23S rRNA, and DNA interact with TLR7, 13, and 9, respectively. This induces the formation of a multiprotein complex containing MyD88 and IRAK2/4 (Myddosome), leading to phosphorylation of TAK1, activation of downstream kinases/transcription factors, and Cxcl2 transcription. The chemokine is secreted extracellularly, where it activates its receptor (CXCR2), initiating an autocrine amplification loop, which might involve MAPK phosphorylation.
Discussion
Much evidence has been gathered over the last two decades for the fundamental role of Cxcl1 and Cxcl2 in driving neutrophil trafficking to sites of inflammation. Some degree of redundancy was initially suggested by the ability of both Cxcl1 and Cxcl2 to potently induce neutrophil migration when exogenously applied in vivo and by the fact that the two chemokines share the same receptor (50, 51). However, this view has been recently challenged using various well-characterized models, including immune complex–induced arthritis and TNF-α–induced inflammation, in which Cxcl1 and Cxcl2 act sequentially to promote neutrophil transmigration from blood to tissues (43, 52, 53). Because chemokines can display different functional profiles in response to different inflammatory stimuli (54), in the current study we focused on the functional activities of Cxcl1 and Cxcl2 in a mouse model of bacterial infection. We found that Cxcl1 and Cxcl2 play distinct, nonredundant roles in governing neutrophil recruitment to infection sites and that this is linked to differences in the cellular sources of these chemokines. Specifically, Cxcl1 production was almost entirely accounted for by resident macrophages, whereas neutrophils were required for the release of optimal levels of Cxcl2. The latter acted as a potent stimulus not only for the recruitment of other neutrophils but also for the activation of crucial effector functions of these cells, including bactericidal activity and production of ROS. These effects were observed using a model of peritonitis induced by live bacteria and would have been missed using classical inflammation models involving isolated microbial products, such as LPS or zymosan-induced peritonitis. Indeed, considerably higher levels of Cxcl2 but not Cxcl and higher neutrophil numbers were measured in the whole bacteria model, relative to the LPS/zymosan models. Cell depletion experiments indicated that neutrophils were responsible for the increased Cxcl2 production and, when stimulated in vitro with streptococci or other bacteria, these cells produced more than 10-fold higher amounts of Cxcl2 relative to the values observed after LPS stimulation. Notably, high-level Cxcl2 secretion was linked with the ability of this chemokine to act autocrinously on neutrophils to stimulate its own production in the presence of bacteria. Indeed, although capable of inducing, by itself, only slight elevations in the transcription of its own gene, Cxcl2 significantly synergized with bacteria in amplifying its own production. Studies are underway to identify the molecular mechanisms underlying these synergistic effects, focusing, in particular, on the ability of Cxcl2 to induce the activation of p38 and other mitogen-associated protein kinases as well as c-Jun activation and binding of this transcription factor to AP1 motifs in the cxcl2 promoter (Fig. 7).
The capacity of neutrophils for producing functionally relevant Cxcl2 levels, as shown in this study, is in line with recent findings obtained in various models of inflammation, including inflammation induced by immune complexes or TNF-α (41, 43). Collectively, our data provide further evidence to support the idea that high-level Cxcl2 production contributes to the ability of the neutrophils that first reach foci of inflammation to recruit and activate other neutrophils (52, 55, 56). Intravital microscopy studies have shown that these cells form tight aggregates or “swarms” around one or few leading neutrophils that undergo activation-associated death upon contact with small areas of tissue lesion or infection (57, 58). The process (called swarming) can be divided into a recruitment phase, in which distant neutrophils are attracted toward “leading” ones, and an aggregation phase, in which neutrophils undergo adhesive homotypic cell interactions at the center of the cluster. Interestingly, Cxcl2 expression was detected at the center of the cluster in a model of sterile inflammation and the Cxcl1/2 receptor CXCR2 was found to be required for the aggregation but not for the recruitment phase of swarming, whereas the lipid mediator LTB4 was required for both phases (58). Although well-defined neutrophil clusters can be visualized in a number of bacterial infections (58–65), the role played by Cxcl1/2 in homotypic cell adhesion/activation is unclear, and future studies will be needed to address this point. Recent observations in a model of Staphylococcus aureus dermatitis indicate that, early after bacterial recognition, macrophages produce the lipid mediator LTB4, which is required for neutrophil chemotaxis, abscess formation, and bacterial clearance (66). Intriguingly, in the latter study, LTB4 was required for the production of Cxcl1 and other chemokines but not for the production of Cxcl2, indicating that LTB4 and Cxcl2 are produced by different pathways during bacterial infection and may have distinctive functions.
In view of its importance in terms of host defenses, we analyzed Cxcl2 production in neutrophils by focusing on the signaling pathways activated by GBS and other bacteria. We established in initial experiments that Cxcl1/2 are not stored in preformed granules in neutrophils but are de novo synthesized upon stimulation. A variety of stimuli, including proinflammatory cytokines (e.g., IL-1α, TNF-α, and IL-17A and F), growth factors (e.g., G-CSF) and ROS, can activate Cxcl1/2 genes (67). It was previously found that GBS stimulates various cell types to release IL-1β, which can, in turn, enhance the production of Cxcl1 in macrophages (32, 35). Moreover, Staphylococcus aureus infection can induce transcription of the gene encoding CXCL8 (the human orthologue of Cxcl1/2) in keratinocytes via an autocrinous mechanism involving IL-1α release (68). For these reasons, we verified the possibility that IL-1 signaling is at least partially responsible for the synthesis of Cxcl2 in neutrophils. This, however, was not the case because IL-1R–deficient neutrophils were not defective in GBS-stimulated Cxcl2 production in response to GBS. It also unlikely that TNF-α, a potent neutrophil activator, participates in GBS-induced Cxcl2 production because neutralization with anti–TNF-α did not affect Cxcl2 expression (data not shown).
We show, in this study, instead, that Cxcl1 and 2 are produced downstream of TLRs in both macrophages and neutrophils in response to a wide range of bacteria. This was evidenced by complete abrogation of chemokine production in the absence of the TLR adaptor Myd88 but not in the absence of other TLR adaptors, such as MAL, TRIF and TRAM. Thus, our data suggest that, similar to TNF-α, IL-1β, IL-12, IL-18, and IFN-β (20, 24, 26, 69, 70), Cxcl1 and 2 are primary mediators that are directly produced by immune cells after TLR-mediated GBS recognition. This was confirmed by showing that GBS-induced chemokine production specifically requires the simultaneous presence of TLR7, 9, and 13. Together with previous evidence (32), our data highlight the presence of a multimodal bacterial detection system, whereby each of these TLRs recognizes a distinct type of nucleic acid released from the pathogen in phagosomes. We show, in this study, that TLR7, 9, and 13 can compensate for the absence of each other in terms of Cxcl1/2 induction. Such a redundancy seems crucial to prevent pathogens from establishing infection by escaping single immune recognition mechanisms. This is likely achieved by the ability of MyD88-dependent endosomal TLRs to activate very similar gene induction programs (Fig. 7), as shown, for example, in the case of TLR7 and 13 (71). Studies are underway to clarify whether residual Cxcl1/2 responses observed in TLR7/9/13 triple KO cells (see for example Supplemental Fig. 2F) might be sustained by TLR2, which also participates in GBS recognition by sensing secreted lipoproteins. It should also be mentioned that our data do not exclude a possible role of other surface TLRs, such as TLR4 and 5, although the ability of GBS to directly stimulate these receptors has not been previously described. Our data showing that endosomal TLRs are required for Cxcl1/2 induction by GBS are in general agreement with the emerging role played by these receptors, particularly those recognizing bacterial RNA, in immune detection of Gram-positive bacteria in both murine and human cells (72–75). Notably, we found, in this study, that neutrophils require nucleic acid–sensing TLRs for responding not only to Gram-positive but also to Gram-negative bacteria, at least in terms of Cxcl2 production. This is at variance with the dominant role played by LPS-induced TLR4 activation in other in vitro and in vivo responses to Gram-negative bacteria (76, 77). However, we found, in this study, that LPS is a weak stimulus for Cxc1/2 induction in neutrophils, unveiling the contribution of nucleic acid–sensing TLRs in chemokine response to Gram-negative bacteria. Our data are in general agreement with two recent studies showing a significant contribution of TLR8 in induction of at least some cytokines (particularly IL-12p70 and IFN-β) in human monocyte-derived macrophages in response to both Gram-positive and Gram-negative bacteria (78). TLR8 is an endosomal receptor that, like mouse TLR13, can sense bacterial RNA (74, 79–81). Whereas TLR8 is largely nonfunctional in rodents (82), TLR13 is absent in humans. Moreover, murine myeloid cells express TLR7, 9, and 13, whereas human monocyte-derived macrophages express TLR8 but not other endosomal receptors (83). Despite these species-related differences, the studies cited above and the present one provide insights into the functional role of an evolutionarily conserved strategy for detecting the presence of bacteria by means of TLR-dependent sensing of phagosomal nucleic acids.
In conclusion, we show, in this study, that bacteria stimulate high-level production of Cxcl1/2 in phagocytes through a mechanism involving multiple endosomal TLRs and that these chemokines play a crucial role in host defenses. In particular, neutrophil-mediated Cxcl2 participates in a feed-forward mechanism that amplifies both neutrophil recruitment to sites of bacterial infection and the bactericidal activities of these cells. These data may be useful to develop much-needed therapeutic agents capable of enhancing bactericidal immune responses during infections caused by antibiotic-resistant pathogens.
Disclosures
The authors have no financial conflicts of interest.
Acknowledgments
We thank Dr. Antonella Ferrara for conceiving and drawing the cartoon, shown in Fig. 7, and the graphical abstract.
Footnotes
The online version of this article contains supplemental material.
Abbreviations used in this article:
- BMDM
- bone marrow–derived macrophage
- DPBS
- Dulbecco’s PBS without Ca2+ and Mg2+
- GBS
- group B streptococcus
- KC
- keratinocyte-derived chemokine
- KO
- knockout
- LTB4
- leukotriene B4
- MOI
- multiplicity of infection
- PLF
- peritoneal lavage fluid
- ROS
- reactive oxygen species
- WT
- wild-type.
- Received August 28, 2019.
- Accepted November 19, 2019.
- Copyright © 2020 by The American Association of Immunologists, Inc.
References
In this issue
