Boosting the IL‐22 response using flagellin prevents bacterial infection in cigarette smoke‐exposed mice

Summary

The progression of chronic obstructive pulmonary disease (COPD), a lung inflammatory disease being the fourth cause of death worldwide, is marked by acute exacerbations. These episodes are mainly caused by bacterial infections, frequently due to Streptococcus pneumoniae. This susceptibility to infection involves a defect in interleukin (IL)‐22, which plays a pivotal role in mucosal defense mechanism. Administration of flagellin, a Toll‐like receptor 5 (TLR‐5) agonist, can protect mice and primates against respiratory infections in a non‐pathological background. We hypothesized that TLR‐5‐mediated stimulation of innate immunity might improve the development of bacteria‐induced exacerbations in a COPD context. Mice chronically exposed to cigarette smoke (CS), mimicking COPD symptoms, are infected with S. pneumoniae, and treated in a preventive and a delayed manner with flagellin. Both treatments induced a lower bacterial load in the lungs and blood, and strongly reduced the inflammation and lung lesions associated with the infection. This protection implicated an enhanced production of IL‐22 and involved the recirculation of soluble factors secreted by spleen cells. This is also associated with higher levels of the S100A8 anti‐microbial peptide in the lung. Furthermore, human mononuclear cells from non‐smokers were able to respond to recombinant flagellin by increasing IL‐22 production while active smoker cells do not, a defect associated with an altered IL‐23 production. This study shows that stimulation of innate immunity by a TLR‐5 ligand reduces CS‐induced susceptibility to bacterial infection in mice, and should be considered in therapeutic strategies against COPD exacerbations.

The progression of COPD is marked by acute exacerbations frequently caused by bacterial infections. Susceptibility to infection during COPD involved a defect in interleukin (IL)‐22. In mice exposed to cigarette smoke, systemic stimulation of the immune system by the TLR5 agonist, flagellin stimulates the immune response both in the spleen and the lung, leading to the production and the circulation of IL‐22. This cytokine participates to activation of airway epithelial cells and increased the secretion of antibacterial peptide. Subsequently, this protects against bacterial expansion and lung tissue alterations in cigarette smoke‐exposed mice.

Introduction

Chronic obstructive pulmonary disease (COPD) is characterized by a progressive and irreversible decline in lung function [1]. Being the fourth leading cause of death worldwide[2], it is caused mainly by chronic exposure to cigarette smoke (CS) or pollutants [3]. Inhalation of CS essentially leads to the activation of epithelial cells and macrophages, both responsible for the mobilization of effector and immunomodulatory cells, including neutrophils and invariant natural killer T (iNK T) cells [4, 5]. The associated chronic inflammatory response progressively leads to airway remodeling, impaired mucociliary clearance and lung parenchymal destruction, further culminating in irreversible airflow limitation [6]. Together, these components are involved in the increased susceptibility of COPD patients to bacterial and viral airway infections [7, 8]. Streptococcus pneumoniae, Moraxella catarrhalis and non‐typable Haemophilus influenzae are the main bacterial triggers of acute exacerbations of COPD, which have a strong impact on health status, exercise capacity, lung function and mortality [9, 10].

Control of airway bacterial infections involves innate immune components [such as surfactant activities, anti‐microbial peptides and pathogen recognition receptors (PRR)] and adaptive immune responses [including mucosal immunoglobulin (Ig)A and bronchus‐associated lymphoid tissue]. Among the PRRs, Toll‐like receptors (TLR), sense lipoproteins (TLR‐2), lipopolysaccharides (TLR‐4), flagellin (TLR‐5) or DNA (TLR‐9) are responsible for the mobilization of effector cells [11]. During COPD, bacterial infections (both in patients and experimental models) are characterized by an increased influx of immune cells, including neutrophils, alveolar macrophages (AM), dendritic cells (DC) and T lymphocytes [4, 12, 13]. However, this response is not effective enough to clear the pathogens. In this context, we recently reported a defective production of T helper type 17 (Th17) cytokines [interleukin (IL)‐17/IL‐22] in response to bacteria both in COPD patients and mice chronically exposed to CS [14, 15], knowing that IL‐17 and IL‐22 are essential to bacterial clearance through their ability to recruit neutrophils and to induce the production of anti‐microbial peptides and the expression of tight junction molecules [16, 17]. Additionally, we have already demonstrated that the supplementation of CS‐exposed mice with recombinant IL‐22 improves the clearance of bacteria and prevents associated lung damage and inflammation, suggesting that the modulation of this cytokine might have therapeutic potential in COPD exacerbations [14]. Several reports have shown that activation of innate receptors, including TLR, is able to elicit a protective immune response against infections [18, 19]. Among these, systemic administration of flagellin, the main component of bacterial flagella, induces immediate production of Th17 cytokines through the activation of DC and type III innate lymphoid cells in blood and lung compartments by a TLR‐5‐dependent mechanism [20]. Moreover, its local administration mainly provokes a rapid neutrophil infiltration into the airways associated with the clearance of bacteria, including S. pneumoniae [21, 22, 23]. Protection against S. pneumoniae was also associated with the early production of both interferon (IFN)‐γ and the synthesis of various IFN‐dependent genes [24, 25, 26].

In this study, we hypothesized that flagellin could inhibit the development of bacteria‐induced COPD exacerbations by inducing an appropriate Th17 response. Here, we report that preventive and delayed administration of recombinant flagellin (FliC) reduces the susceptibility to S. pneumoniae in CS mice as well as subsequent lung inflammation and remodeling. We also show that IL‐22 is involved in the protective effects of FliC. Finally, in‐vitro stimulation of peripheral blood mononuclear cells (PBMCs) indicate that FliC induced IL‐22 production in non‐smokers, while an inhibitory effect of active smoking was observed.

Material and methods

Animals

Male C57BL/6J 6–8‐week‐old mice were purchased from Janvier Labs (Le Genest‐St‐Isle, France). Wild‐type (WT) mice were exposed daily to cigarette smoke (CS) [5 days/week for 12 weeks (whole body exposition chamber; Emka, Scireq, Montreal, QC, Canada)] to induce COPD‐like pathogenesis [5]. Research cigarettes 3R4F were obtained from the University of Kentucky Tobacco and Health Research Institute (Lexington, KY, USA). The control group was exposed to ambient air (Air‐mice). All procedures were performed according to the Pasteur Institute, Lille, Animal Care and Use Committee guidelines (agreement number no. AF16/20090) and were approved by the local ethics committee.

Bacterial challenge and flagellin treatment

Streptococcus pneumoniae serotype 1 (clinical isolate E1586) was obtained from the National Reference Laboratory, Ministry of Health, Montevideo, Uruguay [25]. S. pneumoniae was grown to log‐phase (optical density at 600 nm = 0·5) in Todd–Hewitt yeast broth (THYB) (Sigma‐Aldrich, St Louis, MO, USA) and 2% fetal calf serum (FCS), washed with sterile phosphate‐buffered saline (PBS) and stored at −80°C in THYB plus 12% glycerol for up to 3 months. For bacterial challenge, working stocks were thawed, washed with sterile PBS and diluted to the appropriate concentration. The number of bacteria in stocks was confirmed by seeding blood agar plates with serial dilutions. CS‐exposed mice were anesthetized with ketamine (80 mg/kg) and xylazine (16 mg/kg) and intranasally (i.n.) challenged with 50 µl PBS or 4 × 10⁵ colony‐forming units (CFU) of S. pneumoniae. Experiments were repeated at least three times (n ≥ three mice per group) unless specified otherwise.

CS and S. pneumoniae trigger recruitment of inflammatory cells directly responsible for lung damage and airways obstruction. Because FliC also induces inflammation, in the absence of bronchodilators we chose the intraperitoneal (i.p.) route for FliC administration. Throughout the study, we used a recombinant flagellin originated from the Salmonella enterica serovar Typhimurium that was purified and depleted in endotoxin, as previously described (FliC) [27]. As a preventive approach, 5 µg of FliC (warmed up at 65°C/10 min to unfold the protein) was administrated i.p. just before and 48 h after bacterial challenge. To evaluate delayed protection, FliC was administered 6 h after S. pneumoniae infection, as Air‐mice quickly clear the bacteria. For IL‐22 neutralizing experiments, mice received 200 µg of neutralizing anti‐IL‐22 (AM22, a generous gift of Dr J. C. Renauld) or control isotype (mouse IgG2a) antibodies i.p. 5 min before infection, as reported [20].

Sample collection and processing

Mice were euthanized 1 or 3 days after infection. Bronchoalveolar lavage (BAL) fluids, lungs, spleens and blood samples were collected and kept on ice until the processing or frozen immediately in liquid nitrogen.

BAL was performed by instilling five times 0·5 ml (final volume 2·5 ml) of sterile PBS 2% FCS via a 1‐ml sterile syringe with a 23‐gauge lavage needle into a tracheal incision. BAL samples were used for cytokine analysis by enzyme‐linked immunosorbent assay (ELISA), flow cytometry analysis and CFU numbering. Lung tissues were collected aseptically for CFU counts, cytokine measurement, histological analysis and pulmonary cell characterization (by flow cytometry and ex‐vivo restimulation for 72 h). Ninety‐six‐well plates were coated with anti‐CD3 (30 µl, 0·2 mg/ml solution; eBiosciences, San Jose, CA, USA) for restimulation of 5 × 10⁵ cells/well/200 µl medium (RPMI‐1640 with 10% FCS, 200 U/ml penicillin/streptomycin, 2 mM L‐glutamine). Blood samples were used to determine CFU counts and sera were harvested for ELISA. Total spleen cells were also isolated and analyzed by flow cytometry and ex‐vivo stimulation, 5 × 10⁵ cells/well/200 µl medium, either unstimulated or with heat‐killed (HK) S. pneumoniae [multiplicity of infection (MOI) 2] and/or FliC (1 µg/ml). Supernatants were collected after 72 h for ELISA and mouse tracheal epithelial cell (mTEC) stimulation.

Mouse tracheal epithelial cells (mTEC) culture and activation with spleen cell supernatants

To be protective in the lung, systemically injected FliC requires mobilization of immune cells present in different compartments, including the spleen and soluble factors in the bloodstream. Here we investigated the indirect cross‐talk between FliC‐stimulated splenocytes and mTEC. Trachea from three to four mice were digested with pronase (1·6 mg/ml; Roche, Indianapolis, IN, USA) in Dulbecco’s modified Eagle’s medium (DMEM)/F12 (gibco, Carlsbad, CA, USA) and 200 U/ml penicillin/streptomycin (gibco) for 1 h at 37°C. The reaction was then stopped with FCS (gibco). After DNaseI digestion (Roche Diagnostics), cells were incubated for 4 h on a Nunc delta surface dish (Thermo Fisher Scientific, Fremont, CA, USA) to allow fibroblasts to attach. Non‐adherent epithelial cells were recovered and cultured in complete DMEM/F12 medium plus [DMEM/F12 with 2% Ultroser G (PALL Life Sciences, New York, NY, USA), 200 U/ml penicillin/streptomycin and 2 mM L‐glutamine] on human collagen G matrix (Biochrom, Berlin, Germany) precoated 24‐well culture plates. Confluent (80%) mTEC were starved overnight with DMEM/F12 medium plus containing 0.5% Ultroser G before stimulation. For mTEC stimulation, splenocyte supernatants were harvested as described above (see Sample collection and processing). We first set the optimal dilution and duration of mTEC stimulation with spleen cell supernatants. Spleen mTEC cross‐talk was assessed by stimulating mTEC with 10% (v/v) supernatants from splenocytes restimulated ex‐vivo with HK‐S. streptococcus and FliC of non‐infected CS‐exposed mice for 24 h.

Anti‐IL‐22 antibodies (10 µg/ml, AM22) were added to splenocyte supernatant 15 min before addition to mTEC. The experiment was repeated six times with three different supernatants.

Flow cytometry

BAL, lung and spleen total cells were stained with Turk’s solution and counted with optical microscope. Cells were then incubated with the appropriate panel of antibodies (Supporting information, Table S1) for 30 min in PBS 2% FCS. Data were acquired with LSR Fortessa (BD Biosciences) and analyzed with FlowJo™ software version 7.6.5 (Stanford, CA, USA). Gating strategies were previously reported by Sharan et al. [15], and phenotypes of cell populations are shown in Table 2 and Supporting information, Fig. S1a,b. Absolute cell numbers were calculated according to the total cell number and the frequency of CD45⁺ immune cells.

Histological analysis

To study lung remodeling, lungs were inflated and formalin‐fixed. Lungs were paraffin‐embedded; cross‐sections were cut and stained with hematoxylin and eosin. Random areas were selected for disease scoring (Supporting information, Table S4). Lung injury scoring included both lung remodeling and inflammation. More specifically, the extent of lung injury, alveolar wall thickness, presence of hyaline membrane, neutrophilic alveolitis, bronchial epithelial degeneration, neutrophilic and lymphocytic peribronchitis, vasculitis, emphysema and hemorrhage for a cumulative score from 0 to 30 were considered.

Activation of peripheral blood mononuclear cells (PBMC)

Smoking cessation is the first recommendation in COPD management and therapy[28]. To test if active smoking impairs FliC immunostimulation, peripheral blood was collected from healthy non‐smokers (n = 10, mean age = 40·2 years, sex ratio, F/M = 4/6) and active smokers (n = 6, mean age = 39·1, sex ratio, F/M = 4/2, smoking history: 36·4 ± 4·6 packs/year) (authorization of the local committee of person protection: CPP 2008‐A00690‐55). Written informed consent was received from participants prior to blood sampling. PBMC were purified on Ficoll Paque gradient (GE Healthcare, Chicago, IL, USA) and an aliquot was collected for mRNA preparation. Cells (3 × 10⁶ in 1 ml) were stimulated with FliC (10 µg/ml) or phytohemagglutinin (PHA) (1 µg/ml) (Difco, Detroit, MO, USA) as a positive control in RPMI‐1640 (gibco, Invitrogen, Carlsbad, CA, USA), supplemented with 10% FCS, 200 U/ml penicillin/streptomycin and 2 mM L‐glutamine. Supernatants were harvested after 24 h incubation at 37°C for ELISA.

Cytokine quantification by ELISA

Levels of murine IFN‐γ, IL‐1β, IL‐23, IL‐17, IL‐22 and tumor necrosis factor alpha (TNF‐α) were quantified in lung tissue lysates and BAL using commercial ELISA kits (Invitrogen/R&D Systems, Minneapolis, MN, USA). Similarly, levels of IFN‐γ, IL‐17 and IL‐22 were measured by ELISA in the sera, the supernatants of ex‐vivo‐stimulated murine lung and spleen cells and the supernatants of human PBMC (R&D Systems, Abingdon, UK). The concentrations of human IL‐1b, IL‐6 and IL‐23p19 were measured by ELISA (Invitrogen). Levels of murine granulocyte–macrophage colony‐stimulating factor (GM‐CSF), IL‐6 (Invitrogen) and chemokine (C‐X‐C motif) ligand 1 (CXCL1) (R&D Systems) were assayed in mTEC and splenocyte supernatants.

Reverse transcription–polymerase chain reaction (RT–PCR) quantification of mRNA expression

Quantitative RT–PCR (qRT–PCR) was performed to quantify mRNA of interest. The primers used in this study are described in Supporting information, Table S3. Results were expressed as mean ± standard error of the mean (s.e.m.) of the relative gene expression calculated for each experiment in folds (2^−ΔΔCt) using Gapdh as house‐keeping gene for murine genes. For the analysis of TLR expression in human PBMC, results were reported as ΔCt using β‐actin as house‐keeping gene in order to compare the expression in smokers and non‐smokers.

Statistical analysis

The data are expressed as mean ± s.e.m. Results were statistically analyzed with prism software (GraphPad version 5) using the non‐parametric Mann–Whitney U‐test when comparing only two groups. For comparison of more than two groups, the Kruskall–Wallis test followed by Dunn’s test was used for non‐parametric conditions, while one‐way analysis of variance (anova) (followed by Bonferroni’s correlation t‐test) was used when data fulfilled Gaussian distribution and variance equality. Data are expressed in terms of probability (P). Differences were considered significant when P < 0·05 [P = non‐significant (n.s.); *P < 0·05; **P < 0·01; ***P < 0·001].

Results

Preventive administration of FliC accelerates the clearance of S. pneumoniae in CS‐exposed mice and prevents lung lesions

We have previously reported that mice chronically exposed to CS developed the major COPD features and were more susceptible to S. pneumoniae infection [5]. As S. pneumoniae could exacerbate the CS‐induced lesions in mice, we evaluated the capacity of FliC to prevent such a response. Mice were therefore treated with FliC i.p. (Fig. 1a). As previously reported [14], CS‐exposed mice exhibited a higher bacterial load than Air‐mice in the BAL, lung and blood (Fig. 1b). Interestingly, administration of FliC strongly reduced the bacterial load in BAL and lungs at days 1 and 3 post‐infection (p.i.), both in Air and CS‐exposed mice, and limited bacterial dissemination in the blood at day 3.

Fig. 1.

Treatment with flagellin protects cigarette smoke (CS)‐exposed mice against Streptococcus pneumoniae. (a) To assess the impact of smoking and recombinant flagellin (FliC) on S. pneumoniae infection, C57BL6/J mice were first exposed to cigarette smoke (CS) for 12 weeks, followed by intranasal challenge with S. pneumoniae [4 × 10⁵ colony‐forming units (CFU)]. Mice received FliC intraperitoneally just before infection and at day 2 post‐infection. Samples were harvested at days 1 or 3 post‐infection. Control mice were exposed to ambient air (Air‐mice) and received S. pneumoniae and FliC as CS‐exposed mice. (b) Bacterial load (CFU) in bronchoalveolar lavage (BAL), lung and blood. (c) Lung histopathology after hematoxylin and eosin staining. A representative photograph of each condition is shown. Three independent experiments were performed with at least four mice per group for each repeat. Pooled data (total n ≥ 12) are expressed as mean ± standard error of the mean (s.e.m.). Statistical analysis was performed using the non‐parametric Mann–Whitney U‐test. *P < 0·05; **P < 0·01; ***P < 0·001.

Uncontrolled lung infection leads to tissue remodeling due to vascular dilatation, cell infiltration and alveolar wall destruction. The effect of FliC was next assessed on the lung remodeling. At day 1 p.i. S. pneumoniae induced a mild inflammatory cell recruitment, mainly in peribronchial and perivascular areas (Fig. 1c). Alterations of the alveolar walls and the bronchial epithelial barrier were also observed 3 days p.i. in CS‐exposed mice, but not in Air‐mice. FliC decreased the lung inflammation and strongly limited lesions of the alveolar walls and bronchial epithelium. Improved histopathological scores at day 3 in CS‐exposed S. pneumoniae mice were observed (Table 1). Of note, FliC had no striking effect on lung remodeling in non‐infected mice.

Altogether, these data demonstrate that preventive administration of FliC resulted in a more effective control of S. pneumoniae infection and S. pneumoniae‐induced lung remodeling in CS‐exposed mice.

Table 1.

Histopathological scores in CS‐exposed mice at day 3 post‐infection

	PBS	S. pneumoniae	S. pneumoniae + FliC
Inflammation	1·8 ± 0·4	8·1 ± 1·5	4·5 ± 0·8
Lung remodeling	1·2 ± 0·2	3·7 ± 0·5	1·7 ± 0·5
Total	2·8 ± 0·2	11·9 ± 1·8	6·2 ± 1

The final score in the table (0–30) is a cumulation of inflammation (0–17) and tissue lesion (0–13) scores, as described in Supporting information, Table S4. Means ± standard error of the mean (s.e.m.) are shown for five to seven mice in each group. CS = cigarette smoke; PBS = phosphate‐buffered saline; FliC = recombinant flagellin.

FliC changes the pulmonary immune response to S. pneumoniae in CS‐exposed mice

To control the infection in the lung, i.p.‐administered FliC should stimulate the immune response systemically as well as at the site of infection. To assess this hypothesis, we characterized the impact of FliC treatment on the immune response occurring in the lungs, BAL and spleens. CS‐S. pneumoniae mice exhibited a higher pulmonary inflammation depicted by increased total cell numbers in BAL and lung tissue, associated with a trend for increased neutrophil numbers in BAL (Fig. 2a and Supporting information, Fig. S1c, P = n.s.). FliC increased neutrophil numbers in the lung but not the BAL of CS‐S. pneumoniae mice (Supporting information, Fig. S1d). In contrast, as previously described [29], FliC increased the numbers of total cells and neutrophils in the BAL and lung of Air‐S. pneumoniae mice at day 1 post‐infection (Supporting information, Fig. S1c,d). Because DCs and AM are key players in mucosal immune defenses, we assessed their numbers and activation in the lung. FliC had no impact on myeloid DC and AM maturation, as shown by I‐Ab and CD86 expression (Supporting information, Fig. S1e and data not shown). However, FliC led to a reduced number of DC in the BAL at day 3 post‐infection (Fig. 2b). As a consequence, FliC also significantly reduced the number of conventional αβT cells in BAL and lung from CS‐S. pneumoniae mice. There was only a trend for the number of NK T cells (Fig. 2c). Administration of FliC alone in CS‐exposed mice significantly increased the total cell number in the lung as well as the neutrophil numbers both in the BAL and the lung (Supporting information, Fig. S2a–d). This was related to an increased level of CXCL1 both in the BAL fluid and lung lysates (Supporting information, Fig. S2e–f).

Fig. 2.

Flagellin modulates the inflammation in the lungs of Streptococcus pneumoniae‐challenged cigarette smoke (CS)‐exposed mice. (a) Total cell numbers in bronchoalveolar lavage (BAL) and lungs. Cell numbers are determined using Turks’ solution staining and light microscopy. (b) Absolute number of neutrophils, alveolar macrophages and dendritic cells in BAL and lungs. (c) Absolute number of natural killer (NK) T and conventional [T cell receptor (TCR)‐αβ+] cells in BAL and lungs. Data are from samples collected at day 3 after S. pneumoniae challenge. Three independent experiments were performed with at least three mice per group for each repeat. Pooled data (total n ≥ 9) are expressed as mean ± standard error of the mean (s.e.m.). Statistical analysis was performed using the non‐parametric Mann–Whitney U‐test. *P < 0·05; **P < 0·01.

To evaluate the systemic effects of i.p.‐injected FliC, we investigated the immune response in the spleen. FliC did not affect DC, monocyte and macrophage numbers, but resulted in the maturation of macrophages and DC depicted by the increased CD86 and I‐Ab (Fig. 3a). FliC injection was also associated with the activation of conventional T cells (increased expression of CD25 and CD69) (Fig. 3b). Administration of FliC also resulted in higher IL‐22 levels in S. pneumoniae plus FliC‐stimulated splenocyte supernatants, whereas IFN‐γ and IL‐17 levels were not affected by FliC (Fig. 3c).

Fig. 3.

Systemic administration of recombinant flagellin (FliC) induces early cell activation and interleukin (IL)‐22 production in the spleen of cigarette smoke (CS)‐exposed mice. (a) Macrophages, myeloid dendritic cells, inflammatory monocytes numbers and activation (expression of activation markers CD86 and I‐Ab) were assessed by flow cytometry. (b) Conventional T cells [T cell receptor (TCR)‐αβ⁺] number and expression of the activation markers (CD25 and CD69) were determined by flow cytometry. (c) Measurement of interferon (IFN)‐γ, IL‐17 and IL‐22 production by splenocytes stimulated ex‐vivo with Streptococcus pneumoniae or FliC + S. pneumoniae for 72 h. MFI = median fluorescence intensity. Three independent experiments were performed with at least three mice in each group. Pooled data (total n ≥ 9) are expressed as mean ± standard error of the mean (s.e.m.). Statistical analysis was performed using the non‐parametric Mann–Whitney U‐test. *P < 0·05.

It is known that FliC is able to promote the production of anti‐bacterial peptides, which are critical for pathogen clearance [30, 31]. Therefore, we investigated the expression of anti‐microbial peptides in our model. S. pneumoniae challenge significantly increased the expression of Defb3, Camp, Reg3g (data not shown) and Defb2, but not S100a8 and S100a9 in the lung of both Air‐ and CS‐exposed mice. (Fig. 4a and data not shown). FliC significantly increased the mRNA expression of S100a9 (Fig. 4a) as well as S100a8 at day 1 (data not shown), and the protein levels at day 3 post‐infection in the BAL of both Air and CS‐exposed mice (Fig. 4b).

Fig. 4.

Flagellin up‐regulates the production of anti‐microbial peptides in the lung of cigarette smoke (CS)‐exposed mice infected with Streptococcus pneumoniae. (a) Reverse transcription–polymerase quantitative chain reaction (RT–qPCR) quantification of mRNA expression for Defβ2 and S100a9 in the lung tissue. (b) Concentrations of S100A8 and S100A9 proteins measured by enzyme‐linked immunosorbent assay (ELISA) in the bronchoalveolar lavage (BAL). Three independent experiments were performed with at least three mice per group for each repeat. Pooled data (total n ≥ 9) are expressed as mean ± standard error of the mean (s.e.m.). Statistical analysis was performed using the non‐parametric Mann–Whitney U‐test. *P < 0·05; **P < 0·01; ***P < 0·001.

Overall, FliC transiently amplifies the recruitment of neutrophils in the BAL of CS‐S. pneumoniae‐mice and activates spleen antigen‐presenting cells (APC). These data also suggest that the beneficial effect of FliC on the outcome of infection involves both the lung and spleen compartments, and is associated with the production of anti‐microbial peptides.

IL‐22 is involved in FliC‐mediated protection in infected CS‐exposed mice

We showed that FliC modulates immune cells in a systemically manner. Next, we evaluated Th17 (IL‐17 and IL‐22) and Th1 (IFN‐γ) cytokines, important players in the control of lung inflammation and bacterial infection [16, 17]. Infection with a sublethal dose of S. pneumoniae did not significantly modulate the expression of IL‐17 and IL‐22 at day 1 in Air‐ and CS‐exposed mice, while IFN‐γ was slightly enhanced (Fig. 5a,b). FliC enhanced IL‐22 production at day 3 p.i. in Air‐S. pneumoniae and CS‐S. pneumoniae mice, but not IL‐17 and IFN‐γ. FliC also tended to enhance (P = n.s.) IL‐1β, IL‐23 (important for Th17 polarization) and TNF‐α (as a Th1 cytokine) in the BAL of Air‐mice (Supporting information, Fig. S3a). Treatment with FliC also resulted in higher IL‐22 levels in ex‐vivo‐stimulated lung cell supernatants (Fig. 5b). IL‐17 and IFN‐γ levels were not affected by FliC.

Fig. 5.

Cigarette smoke (CS)‐exposed mice displayed increased ability to produce interleukin (IL)‐22 in the lung following recombinant flagellin (FliC) administration. The ability of ambient air (Air‐mice) and CS‐exposed mice to produce T helper type 17 (Th17) cytokine was measured by enzyme‐linked immunosorbent assay (ELISA). (a) Interferon (IFN)‐γ, IL‐17 and IL‐22 levels in the bronchoalveolar lavage (BAL). (b) IFN‐γ, IL‐17 and IL‐22 levels in supernatants of anti‐CD3‐stimulated lung cells for 72 h. Three independent experiments were performed with at least four mice per group for each repeat. Pooled data (total n ≥ 12) are expressed as mean ± standard error of the mean (s.e.m.). Statistical analysis was performed using the non‐parametric Mann–Whitney U‐test. *P < 0·05.

The systemic effect of FliC was strengthened by measuring Th17 and Th1 cytokines in the serum. Interestingly, FliC treatment in non‐infected CS‐exposed mice was associated with a higher concentration of IL‐22, without affecting IL‐17 and IFN‐γ levels (Supporting information, Fig. S3b,c). In order to establish the link between the activation of spleen cells and the anti‐bacterial response in the lung, we performed mouse tracheal epithelial cells (mTEC) stimulation with supernatants from FliC + S. pneumoniae‐stimulated spleen cells. The supernatants of FliC + S. pneumoniae‐stimulated spleen cells significantly induced high production of CXCL1, IL‐6, GM‐CSF and calgranulins (S100A8 and S100A9), as well as mRNA expression of the anti‐microbial peptide CAMP (Fig. 6a–c) by mTEC. In our experimental conditions, FliC alone had no effect on mTEC for indicated readout (data not shown). CXCL1, IL‐6 and GM‐CSF levels in spleen cell supernatants were much lower than in activated‐mTEC cultures and were not modulated by co‐stimulation with FliC and S. pneumoniae (Fig. 6d). IL‐22 neutralizing antibodies blocked FliC‐related production of S100A8 and S100A9 (P < 0·05 for S100A8), while it had no effect on CXCL1, IL‐6 and GM‐CSF expression.

Fig. 6.

Soluble factors produced by recombinant flagellin (FliC)‐stimulated splenocytes promote the activation of airway epithelium by an interleukin (IL)‐22‐independent mechanism. Murine tracheal epithelial cells (mTEC) were activated with supernatants of splenocytes (stimulated or not with Streptococcus pneumoniae and FliC) for 24 h. The role of IL‐22 was analyzed by addition of neutralizing anti‐IL‐22 antibody (10 μg/ml) in spleen cell supernatants. Cytokines were measured in the supernatants of mTEC and mRNA were isolated from the cell layer for reverse transcription–polymerase quantitative chain reaction (RT–qPCR). (a) Levels of chemokine (C‐X‐C motif) ligand 1 (CXCL1), IL‐6 and granulocyte–macrophage colony‐stimulating factor (GM‐CSF) measured by enzyme‐linked immunosorbent assay (ELISA) were reported as mean ± standard error of the mean (s.e.m.)]. (b) mRNA expression of CAMP were evaluated by RT–qPCR (mean ± s.e.m.). (c) Levels of S100A8 and S100A9 measured by ELISA were reported as mean ± s.e.m. (d) CXCL1, IL‐6 and GM‐CSF levels in splenocyte supernatants activated with FliC, S. pneumoniae or FliC + S. pneumoniae were reported as mean ± s.e.m. For comparison concerns, the same scale range was used in (a) and (d). These experiments were repeated six times with duplicates. Statistical analysis was performed using the non‐parametric Mann–Whitney U‐test. *P < 0·05; **P < 0·01; ***P < 0·001.

It has already been described that the preventive effect of FliC against S. pneumoniae infection is mediated by IL‐22 over‐expression in a non‐CS context [20]. In order to define the role of IL‐22 in the protective effects of FliC in CS‐exposed mice, anti‐IL‐22 blocking antibodies or isotype control antibodies were administered before FliC treatment (Fig. 7a). Blockade of IL‐22 abrogated the protective effect of FliC (Fig. 7b–d). This leads to increased bacterial count in the BAL, lungs and blood at day 3 post‐infection (P < 0·05). Administration of anti‐IL‐22 antibody decreased IL‐17 levels in the BAL at both days 1 and 3 (Fig. 7e,f). In contrast, an increased number of neutrophils was observed at day 3, with similar trends for DC (Fig. 7g).

Fig. 7.

Interleukin (IL)‐22 is important in flagellin‐mediated protection during Streptococcus pneumoniae infection. (a) Cigarette smoke (CS)‐exposed mice were intraperitoneally injected with anti‐IL‐22 neutralizing antibodies just before recombinant flagellin (FliC) treatment and challenged with S. pneumoniae for 1 or 3 days. (b–d) Colony‐forming unit (CFU) count in the bronchoalveolar lavage (BAL) (b), lung (c) and blood (d), (e–f) IL‐17 production in BAL (e) and by lung cells stimulated ex‐vivo with anti‐CD3 for 72 h (f). (g) Absolute numbers of alveolar macrophages, neutrophils and dendritic cells were reported in the BAL and the lungs. The data are expressed as mean ± standard error of the mean (s.e.m.) of two independent experiments with at least three mice per group for each report (total n ≥ 6). Statistical analysis was performed using the non‐parametric Mann–Whitney U‐test. *P < 0·05.

These results indicate that FliC amplifies the production of IL‐22 by lung and spleen cells from CS‐exposed mice, and suggest that the protective effect of FliC in CS‐S. pneumoniae mice requires systemic and local (lung) IL‐22 production.

Delayed administration of flagellin accelerates the clearance of S. pneumoniae in CS‐exposed mice

For clinical purposes, it was essential to determine if the FliC‐induced protection in CS‐S. pneumoniae mice is also effective when administered in infected animals. To this end, FliC was administrated 6 h after S. pneumoniae challenge (Fig. 8a). As seen in Fig. 8b, FliC administration resulted in a huge reduction of the bacterial load in the BAL and lungs of CS‐exposed mice. This was associated with increased levels of Il‐17 and IL‐22 in the BAL (Fig. 8c). Priming of lung cells was also exhibited by a significant increase of S100A9 concentrations in the BAL, but not S100A8 (Fig. 8d).

Fig. 8.

Delayed treatment with flagellin limits the susceptibility of cigarette smoke (CS)‐exposed mice to Streptococcus pneumoniae. (a) Experimental protocol. CS‐exposed mice received recombinant flagellin (FliC) 6 h after S. pneumoniae challenge and were euthanized 3 days later. (b) Colony‐forming unit (CFU) counts in the bronchoalveolar lavage (BAL) and lung. (c) Concentrations of interleukin (IL)‐17 and IL‐22 in the BAL and lung tissue lysates. (d) Concentrations of S100A8 and S100A9 in the BAL. (e) Total cell number in the BAL and lung. (f) Absolute number of neutrophils, alveolar macrophages (AM) and inflammatory monocytes in the BAL and the lung. Three independent experiments were performed with at least four mice per group for each repeat. Pooled data (total n ≥ 12) are expressed as mean ± standard error of the mean (s.e.m.). Statistical analysis was performed using the non‐parametric Mann–Whitney U‐test. *P < 0·05; **P < 0·01; ***P < 0·001.

FliC did not modulate the number of total cells, AM, DC and lymphocytes in CS‐exposed mice at day 3 (Fig. 8e and data not shown). In contrast, the number of neutrophils and inflammatory monocytes were higher in the BAL, while there was a decrease in the lung tissue compared to non‐infected and infected CS‐exposed mice (Fig. 8f).

Similarly, intranasal administration of FliC was also able to significantly reduce the bacterial load in the BAL and the lung at day 1 and with a trend at day 3 p.i. (Supporting information, Fig. S4). These data show that delayed treatment or intranasal administration of flagellin is also able to limit the susceptibility of CS‐exposed mice to S. pneumoniae.

Active smoking impairs IL‐22 induction by FliC in human peripheral blood mononuclear cells (PBMC)

Cigarette smoking is the main factor in COPD physiopathology, and acute exacerbations worsen the disease. Thus, the first recommendation for the management of COPD and exacerbations is smoking cessation [28]. To investigate the interest of FliC treatment in the clinical management of COPD, we evaluated the consequences of active smoking on the immune response of PBMC following FliC treatment. To assess smoker cells responsiveness, PBMC were stimulated with PHA as positive control. Both smokers and non‐smokers responded to PHA, as concentrations of IL‐17, IL‐22, IFN‐γ, IL‐1β and IL‐6 levels were significantly increased. In contrast with non‐smokers, IL‐23 levels were not significantly higher after the addition of PHA in PBMC from smokers. Interestingly, FliC stimulation of PBMC from non‐smokers significantly enhanced the production of IL‐22 and IFN‐γ (but not IL‐17) (Fig. 9a) associated with a higher production of IL‐1β, IL‐6 and IL‐23 (Fig. 9b). PBMC from active smokers have an altered response to FliC with a defective production of IL‐22 and IFN‐γ (Fig. 9a). Whereas the secretion of IL‐1β and Il‐6 was similarly induced by FliC in smokers, the secretion of IL‐23 was not amplified in contrast with non‐smokers. Moreover, we have evaluated the mRNA expression of TLR‐5 and TLR‐3 (as a control) in the blood mononuclear cells from these subjects. The mRNA expression of both TLRs is similar in both groups (Supporting information, Fig. S5). This indicates that the induction of IL‐22 following FliC stimulation is abrogated by active smoking, probably due to a defective production of IL‐23 instead, whereas it is not related to TLR‐5 down‐regulation.

Fig. 9.

Flagellin stimulates interleukin (IL)‐22 and IL‐23 production by human peripheral blood mononuclear cells (PBMCs) in healthy subjects but not in active smokers. PBMCs from healthy non‐smokers (n = 10) and active smokers (n = 6) were stimulated in vitro with recombinant recombinant flagellin (FliC) or phytohemagglutinin (PHA) for 24 h. (a) The production of T helper type 17 (Th17) cytokines (IL‐17 and IL‐22) and Th1 cytokine [interferon (IFN)‐g] was evaluated in supernatants by enzyme‐linked immunosorbent assay (ELISA). (b) The production of pro‐Th17 cytokines (IL‐1β, IL‐6 and IL‐23) was measured in the supernatant by ELISA. Boxes show the interquartile range, whereas the error bars represent the minimum and maximum values; the line in the box shows the median. Statistical analysis was performed using the non‐parametric Mann–Whitney U‐test. *P < 0·05.

Discussion

As reported in COPD patients, CS‐exposed mice are characterized by an altered clearance of bacteria, an increased inflammatory reaction, some lung damage and an increased mucus secretion [4, 32]. The inflammatory reaction associates the mobilization of effector cells (neutrophils and macrophages) and a strong proinflammatory cytokine burst in CS‐exposed mice. Unfortunately, the immune system remains unable to clear the bacteria. In this study, we demonstrated that treatment with flagellin, a TLR‐5 ligand, is able to improve the ability of CS‐exposed mice to clear S. pneumoniae. Similar results were also observed in non‐typeable Haemophilus influenza (NTHi)‐challenged CS‐exposed mice (unpublished data). It has been reported that local treatment with flagellin improved the clearance of a lethal dose of S. pneumoniae in non‐CS‐exposed mice with increased neutrophil mobilization [23, 33]. As the pathophysiology of COPD exacerbation episodes implicated a deleterious effect of neutrophils in the tissue damage, we chose i.p. rather than the intratracheal administration to prevent uncontrolled neutrophil recruitment in the airways. Immune protection of flagellin systemic treatment involved increased IL‐22 production in BAL of non‐CS, Air‐S. pneumoniae mice [20, 34, 35]. Our data show similar results in CS‐S. pneumoniae mice, with improved ability of anti‐CD3 stimulated lung cells to produce IL‐22. We have already shown that CS exposure altered S. pneumoniae‐induced IL‐22 production in mice NK, NK T and innate lymphoid cells (ILC) [14]. In our model, modulation of IL‐22 production is related to the activation of spleen cells (macrophages, DC and lymphocytes). Both macrophages and DC express TLR‐5 and they can respond to flagellin treatment [34, 36] and initiate IL‐22‐producing cells (NK, NK T, ILC and conventional T cell) polarization. We demonstrated that FliC stimulated spleen cells released soluble factors promoting cytokine (IL‐6) chemokine (CXCL1), growth factor (GM‐CSF) and anti‐microbial peptide (CAMP) production by mTEC. FliC also increased IL‐22 levels in the serum of CS‐S. pneumoniae mice, suggesting that this cytokine can reach lung cells.

IL‐22 neutralizing antibodies abrogated the protective effect of FliC, suggesting that IL‐22 plays a key role in FliC‐induced protection in CS‐S. pneumoniae mice. These data are in line with our previous report showing that the supplementation with recombinant IL‐22 is able to accelerate the clearance of the bacteria and limit the consequences of bacterial infection (inflammation and tissue damage) in CS‐exposed mice [14]. Calgranulins (S100A8, S100A9) are involved in S. pneumoniae killing by modulating APC migration [37] and starving pathogen from essential ion nutrients (Ca²⁺, Zn²⁺ and Mn²⁺) [38]. Different results were reported in non‐CS influenza A virus (IAV)‐infected mice, where recombinant IL‐22 prevents S. pneumoniae superinfection and promotes epithelial wall repair, but does not modulate anti‐microbial peptides [39]. Our data also suggest that epithelial wall repair is shown by reduced bacteria translocation in the blood and improved histological score of lung damage.

Neutrophil mobilization is involved in FliC protection, as we observed an early (day 1 post‐infection) increase of neutrophil count in the lung of CS‐S. pneumoniae mice. This recruitment can be related to the direct activity of flagellin on the production of chemokines and proinflammatory cytokines [35]. In fact, in non‐infected CS‐exposed mice FliC induced CXCL1 production, neutrophil recruitment in BAL and lung (Supporting information, Fig. S3). Total cell counts also increased, but not significantly. Neutrophils are recruited and maintained by CXCL1 and GM‐CSF; they are an important source of calgranulins that are released during bacterial infection [38].

IL‐22 production is hampered in bacteria‐infected PBMC from COPD [14]. Our experiments revealed that FliC boosts IL‐22 production by non‐smoking PBMC, suggesting the interest of this treatment in humans. However, active smoking abrogates FliC‐induced IL‐22 and IL‐23 production, probably due to CS‐dependent inhibition of APC function through oxidative stress [40]. In parallel, we also demonstrated that the mRNA expression of TLR‐3 and TLR‐5 is not altered in smokers compared to non‐smoker healthy subjects, showing that the defective response is not directly related to TLR‐3 and TLR‐5 signaling itself. Our data strengthen the importance of smoking cessation in COPD management, and suggest that the treatment with FliC could be associated with or preceded by anti‐oxidative therapy in active smokers, in order to reduce the detrimental effect of smoke. Surprisingly, flagellin induce IL‐22 production in smoking mice but not in humans, suggesting that human immune cells might be more susceptible to the inhibitory effect of CS or that mice develop resistance after chronic exposure to CS.

In our model, the infection progresses very quickly, making the investigation of curative approaches challenging at late time‐points. Here, we showed that FliC treatment remains efficient even 6 h after S. pneumoniae challenge. Similarly, recent data suggest that FLiC treatment can be efficient curatively, either alone or with antibiotics, against S. pneumoniae infection [23, 41].

Despite FliC protection shown in CS mice, additional experiments remain necessary before therapeutic use in COPD patients. In fact, it is worth noting that our murine experimental model did not fully reproduce all clinical features observed in severe COPD patients, exhibiting large emphysema and/or co‐morbidities. In these patients, it would be necessary to combine FliC with antibiotics. Such combined therapy has an additive anti‐bacterial effect as reported in post‐Influenza A virus pneumococcal infection [20, 23, 41]. These data suggest that flagellin can be associated with anti‐biotherapy for curative treatment of COPD exacerbations.

In conclusion, we demonstrate that flagellin treatment is able to control S. pneumoniae infection in CS‐exposed mice to limit subsequent lung inflammation and remodeling. FliC protection is IL‐22‐dependent, and leads to the recruitment of neutrophils, activation of APCs, lymphocytes and modulation of anti‐microbial peptides (S100A8 and S100A9) (Fig. 10). FliC restored an efficient bacterial clearance and limited the inflammatory reaction, so this approach could be a step forward towards the treatment of COPD exacerbations.

Fig. 10.

Graphical abstract. The progression of chronic obstructive pulmonary disease (COPD) is marked by acute exacerbations frequently caused by bacterial infections. Susceptibility to infection during COPD involved a defect in interleukin (IL)‐22. In mice exposed to cigarette smoke, systemic stimulation of the immune system by the Toll‐like receptor (TLR)‐5 agonist flagellin stimulates the immune response both in the spleen and the lung, leading to the production and the circulation of IL‐22. This cytokine participates in activation of airway epithelial cells and increases the secretion of anti‐bacterial peptide. Subsequently, this protects against bacterial expansion and lung tissue alterations in cigarette smoke‐exposed mice.

Disclosures

All authors declare that they have no competing interests.

Author contributions

Conception and design: B. K., M. P. C., R. P., C. C., J. C. S., F. T., M. P. and Ph. G.; analysis and interpretation: B. K., M. P. C., M. P. and Ph. G.; histopathological analysis: Pi. G. and Ph. G.; Experiments with human MNC: F. H.; drafting the manuscript for important intellectual content: B. K., M. P. C., R. P., C. C., J. C. S., F. T., M. P. and Ph. G. All authors gave their agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Supporting information

Fig. S1. Flic modulated the recruitment of inflammatory cells in the lung whereas it did not affect cell activation in Sp challenged CS mice.

Fig. S2. Flagellin induced recruitment of neutrophils into the airways and the lung of CS‐Sp‐mice.

Fig. S3. Flagellin administration to CS‐Sp‐mice did not prime the secretion of inflammatory cytokines in the lung.

Fig. S4. Intranasal administration of Flagellin decreased bacterial load in CS‐exposed mice against Sp.

Fig. S5. TLR expression in peripheral blood mononuclear cells did not differ among non‐smokers and active smokers.

Table S1. Antibody panels for antigen presenting cells (APC) and T Lymphocytes labelling for flow cytometry.

Table S2. Phenotypes of immune cell populations.

Table S3. Primer sequences for RT‐qPCR in mice.

Table S4. Lung injury scoring criteria.