Introduction

Inflammation is an important innate immune response that is usually activated by the ligation of pattern recognition receptors (PRRs) leading to upregulation of a broad range of pro-inflammatory factors, including cytokines and chemokines, and migration of leukocytes from the circulation to the site of tissue damage [1]. The clearance of these cells, once the infection or injury has been resolved, is crucial for the tissue healing process. An acute inflammatory response lasts only a few days, whilst a response of longer duration is referred to as chronic inflammation.

B-cell lymphoma 2 (BCL-2) family of proteins are the critical regulators of the intrinsic apoptotic pathway [2]. However, A1 (in humans called BFL-1), remains a relatively poorly characterised anti-apoptotic protein [3]. BCL2A1 (encoding A1) was first identified as an early response gene induced in bone marrow-derived macrophages upon treatment with granulocyte-macrophage colony-stimulating factor (GM-CSF) and lipopolysaccharide (LPS). This study also demonstrated that in mice, A1 expression is restricted to cells of the haematopoietic compartment [3]. In humans, BFL-1 expression appears to be more widespread, but it is still predominantly found in haematopoietic cells [4]. A1/BFL-1 expression can be induced by inflammatory cytokines, such as tumour necrosis factor-alpha (TNFα) and IL-1β [5]. The identification of BCL2A1 as an NF-kB target gene [6] and its expression in inflammatory cells suggest a role for A1 during inflammation.

In mice, studies of A1 are complicated by the presence of three functional isoforms (A1-a, -b, - and -d) and one pseudogene (A1-c) [7]. Mice lacking the A1-a isoform displayed only minor defects in neutrophils and mast cells [8, 9]. Recently we generated mice lacking all three functional isoforms of A1 (A1−/− mice) and found that they do not display any major abnormalities [10]. Here, we utilised A1−/− mice to assess an in vivo role of A1 in response to specific inflammatory challenges.

Neutrophils in the blood are the first-line defence against infection. They are continuously released from their bone marrow reservoir, into the blood and are recruited to sites of infection. Circulating neutrophils have a short lifespan of only ~7 h in the absence of infection [11]. This cellular lifespan is extended after exposure to GM-CSF, granulocyte-colony stimulating factor (G-CSF) or diverse pathogen or damage-associated molecular patterns, such as LPS [12, 13]. Prolonged neutrophil survival and activity can be detrimental and their death is required for resolution of inflammation [14].

The limited lifespan of neutrophils is partly due to their expression of pro- and anti-apoptotic proteins. Mcl-1 is critical for neutrophil survival under steady-state [15]. Human neutrophils also express other anti-apoptotic proteins, including BFL-1 [16]. Similar to Mcl-1, A1 is also a short-lived pro-survival protein [17, 18]. Here, we examined the role of A1 in neutrophil survival in mouse models of inflammation in vivo.

Results

A1 protein is strongly upregulated upon pro-inflammatory stimuli in vivo

A1 protein expression was analysed by Western blotting of lung tissues and peritoneal lavage cells from wild-type (WT) and A1 knockout (A1−/−) mice that had been challenged with the indicated inflammatory stimuli using our recently developed monoclonal antibody [19]. Intranasal administration of LPS or Pseudomonas aeruginosa (PA) significantly upregulated A1 expression in the lungs of WT mice (Fig. 1a, b). Similarly, A1 expression was induced in the peritoneal cells of WT mice following intraperitoneal (i.p.) injection of LPS or caecal slurry (CS) containing bacteria (Fig. 1a, b). The spleen and bone marrow were used as distal organ controls (Fig. 1a). LPS and many other constituents from bacteria stimulate PRRs which lead to the activation of NF-kB signalling and hence, the upregulation of A1 expression. A1 expression was either absent or very low in the lung, spleen, peritoneal cells and bone marrow of naïve mice and was only increased in the bone marrow in response to LPS (Fig. 1a). We also assessed A1 expression following non-microbial stimuli using the monosodium urate (MSU)-induced peritonitis or ovalbumin (OVA)-induced lung inflammation models. This revealed robust induction of A1 in both settings (Fig. 1c).

Fig. 1: A1 protein expression in tissues and cells upon induction of inflammation by microbial and non-microbial stimuli.
figure 1

WT and A1 knockout (A1−/−) mice were challenged with microbial components by either intranasal installation or i.p. injection of a LPS, b P. aeruginosa (PA) or caecal slurry (CS). Twenty-four hours after LPS or PA treatments, lung tissues were harvested. Peritoneal lavage cells were harvested 4 h after LPS treatment and 18 h after CS treatment. c WT and A1−/− mice challenged with non-microbial components. Mice were i.p. injected with monosodium urate (MSU) crystals and peritoneal lavage was obtained at 24 h post-injection. Mice were made to inhale Ovalbumin (OVA) and at the endpoint, lung tissues were harvested. The lung and peritoneal lavage cells were analysed by Western blotting for A1 protein expression. Probing for HSP70 was used as a loading control.

Full size image

A1 has been shown to be important in regulating neutrophil survival in vitro [20, 21]. Western blot analysis of isolated neutrophils from the lungs of LPS treated mice confirmed a marked upregulation of A1 in the enriched neutrophil fraction (Supplementary Fig. S1). Collectively, these findings demonstrate that A1 is induced in neutrophils upon pro-inflammatory stimuli.

The absence of A1 does not impact neutrophil survival during LPS induced lung inflammation

Given the upregulation of A1 expression in inflammatory neutrophils in vivo in response to inflammatory insults, we tested whether A1 is critical for neutrophil survival in this setting. Upon intranasal challenge of LPS, the lungs of both WT and A1−/− mice contained increased immune cell infiltration compared to PBS-treated controls, demonstrating successful induction of lung inflammation (Fig. 2). The immune cell composition of lung tissues was further analysed by flow cytometry at 24 h post-challenge. No major differences were observed in the frequencies or numbers of B cells, T cells, macrophages or neutrophils in the lungs of WT and A1−/− mice (Fig. 2 and Supplementary Fig. S2) following LPS challenge or PBS control treatment. The gating strategy used for flow cytometry analysis of immune cells is shown in Supplementary Fig. S3. These results demonstrate that A1 is dispensable for the survival of lung-infiltrating immune cells during inflammation. LPS challenge, however, significantly reduced the number of lung B cells in both WT and A1−/− mice compared to PBS-treated mice. While we do not understand the nature of this decrease, similar observations have been made in lung cancer models [22].

Fig. 2: Analysis of immune cell populations in the lungs during LPS induced inflammation.
figure 2

WT and A1−/− mice were intranasally administered LPS (10 µg) or vehicle (PBS). After 24 h lung tissues were harvested and processed, and percentages of the indicated cell subsets were determined by flow cytometric analysis following staining for the indicated cell subset specific surface markers. Statistical significance (*P < 0.05, **P < 0.01) was determined using a student’s t-test. Each dot represents one mouse. The gating strategy used for flow cytometric analysis is shown in Supplementary Fig. S3. (n = 4–6 mice).

Full size image

We also examined the immune cell composition of lung tissues at an earlier time point (4 h post-LPS challenge) by flow cytometry (Supplementary Fig. S4). Again, no major difference was observed in the frequencies or numbers of T cells, macrophages or neutrophils in the lungs of WT and A1−/− mice at 4 h post-challenge. At this early time point, there were more B cells in LPS induced A1−/− lungs in comparison to WT lungs. This difference was no longer seen at the 24 h time point with no impact on the overall inflammatory response (Fig. 2). Inflammatory cytokines in the bronchoalveolar lavage fluid (BALF) were assessed by ELISA at 4 h following intranasal instillation of LPS. Interestingly, the levels of TNF and GM-CSF were significantly lower in the BALF of A1−/− mice compared to WT mice at 4 h post-LPS challenge (Supplementary Fig. S5).

A1 deficiency causes a minor reduction in neutrophils during Pseudomonas aeruginosa induced lung inflammation

The murine pulmonary PA infection model closely mimics bacterially-induced pneumonia in humans. Mice were intranasally administered with PA (∼1.5 × 107 CFU) to induce lung inflammation. Infected animals initially lose weight and show neutrophil-induced acute lung inflammation as the bacterial infection is rapidly cleared [23, 24]. At 24 h post-infection, the lungs were analysed by flow cytometry to identify the infiltrating haematopoietic cell subsets. Infection with PA-induced an influx of neutrophils (Fig. 3) and the frequencies of lung neutrophils from infected A1−/− mice were significantly lower than those seen in infected WT mice (Fig. 3). Observation at 24 h post-infection revealed similar weight loss but worse body condition in the A1−/− mice (Supplementary Fig. S6 a). In other mouse inflammatory models, (e.g., LPS, Mycobacterium bovis-derived Bacillus Calmette-Guerin (BCG), or Toxoplasma gondii) A1 is transiently upregulated during the first 8–16 h and at 24 h the levels of A1 is again reduced [25,26,27]. Therefore, we conducted these experiments within the 24 h time frame.

Fig. 3: Analysis of the immune cell populations in the lungs during P. aeruginosa (PA) induced inflammation.
figure 3

WT and A1−/− mice were intranasally administered P. aeruginosa (PA) or vehicle (PBS). After 24 h lung tissues were harvested and processed, and the percentages of the indicated cell subsets was determined by flow cytometric analysis following staining for the indicated cell subset specific surface markers. Statistical significance (*P < 0.05, **P < 0.01, ***P < 0.001) was determined using the student’s t-test. Each dot represents one mouse. The gating strategy used for flow cytometric analysis is shown in Supplementary Fig. S3. (n = 4 mice).

Full size image

The clearance of the bacteria from the lungs was examined by flow cytometric analysis for internalised bacteria in granulocytes (Gr-1+) (Supplementary Fig. S6b). In addition, colony formation assays of lung extracts of PA-infected mice were performed to determine the number of bacteria (Supplementary Fig. S6c). These assays did not show any difference in bacterial clearance between A1−/− and WT mice 24 h post-infection (Supplementary Fig. S6b, c).

A1 is not required for the accumulation of neutrophils during caecal slurry or LPS induced peritonitis

Polymicrobial sepsis was induced in mice by i.p. injection of caecal contents of laboratory animals into test animals [28]. We assessed the onset of clinical signs, including reduced motor activity, lethargy, shivering, piloerection, rapid shallow breathing and also measured the systemic levels of inflammatory cytokines.

We first determined the optimal concentration of CS to induce acute sepsis in mice. Doses below 0.5 g/kg were non-lethal to mice at 24 h post-injection (Supplementary Fig. S7a, b). In separate survival studies, mice were injected with lethal doses (>0.5 g/kg) of CS (Fig. 4). We hypothesised that differences in neutrophil survival between WT and A1−/− mice would result in a difference in bacterial clearance and, consequently, in animal survival. Injection of 0.75 g/kg or 1 g/kg CS caused severe morbidity requiring euthanasia within 18-24 h in 100% of both WT and A1−/− mice. Treatment with 0.55 g/kg CS caused severe disease in 28% and 43% of WT and A1−/− mice, respectively, within 48 h. Injection of 0.65 g/kg CS necessitated euthanasia of 38% and 88% of WT and A1−/− mice, respectively, at 48 h (Fig. 4). Despite these differences in survival at 48 h with 0.55 g/kg and 0.65 g/kg doses, the overall difference between survival curves did not reach statistical significance because animals of the two genotypes showed similar survival at later time points. Neutrophil mobilisation from the bone marrow was observed at the minimum non-lethal dose of CS tested (0.2 g/kg), but there was no difference between the WT and A1−/− mice (Supplementary Fig. S8a, b).

Fig. 4: Kaplan–Meier curve presenting the survival of mice that had been injected with caecal slurry.
figure 4

Groups of WT and A1−/− mice were injected i.p. with the indicated doses of caecal slurry and monitored every 3 h for up to 10 days and sacrificed when body condition scoring reached an ethical endpoint. Overall survival graphs over the period of 10 days did not show any significant differences between WT and A1−/− mice at any of the doses (n = 3–8 mice).

Full size image

The peritoneal lavage, blood and bone marrow were obtained at 4 h and 18 h post-CS (0.65 g/kg) injection for flow cytometric analysis (Fig. 5a and Supplementary Fig. S9). Although at 4 h there were no obvious differences in the diverse haematopoietic cell populations between WT and A1−/− mice, by 18 h there were some differences, albeit minor (Fig. 5a and Supplementary Fig. S9). In the peritoneum of WT and A1−/− mice, the percentages of neutrophils increased and the percentages of T cells and macrophages decreased at 4 h post-CS injection.

Fig. 5: Analysis of the immune cell populations in the peritoneal lavage of caecal slurry and LPS injected mice.
figure 5

WT and A1−/− mice were injected i.p. with 0.65 g/kg caecal slurry (CS) or 1 mg/kg LPS. Mice were sacrificed at 4 h post-injection. a Percentages of the indicated cell subsets in the peritoneal lavage were determined at 4 h after CS or LPS injection by flow cytometric analysis following staining for the indicated cell subset specific surface markers. The gating strategy used for flow cytometric analysis is shown in Supplementary Fig. S3. b Serum levels of TNF, IL-1β and GM-CSF from mice undergoing polymicrobial sepsis were measured by ELISA 4 and 18 h after injection with CS. Statistical significance (*P < 0.05, **P < 0.01, ***P < 0.001) was determined using a student’s t-test. Each dot represents one mouse (n = 2–7 mice).

Full size image

As an additional model of peritoneal inflammation, we injected mice i.p. with LPS (1 mg/kg). Cells from the peritoneal lavage and blood were analysed by flow cytometry at 4 h post-injection. There were no differences in cell subset composition between WT and A1−/− mice injected with LPS. However, there was a significantly reduced frequency of peritoneal macrophages in PBS injected A1−/− mice compared to their WT counterparts (Fig. 5a). There was an increase in the percentages of neutrophils and a decrease in the percentages of lymphocytes in the blood of both WT and A1−/− mice upon i.p. injection of LPS (Supplementary Fig. S10).

Increased infiltration of neutrophils to the site of infection may result in higher levels of inflammatory cytokines. Inflammatory cytokines in the peritoneal lavage were assessed by ELISA following injection of CS. The levels of TNF were slightly higher in A1−/− mice at 4 h, and the levels of IL-1β were slightly lower in A1−/− mice at 18 h (Fig. 5 b). Furthermore, we analysed the degree of cell death in the peritoneal cells during CS- and LPS-induced peritonitis (Supplementary Fig. S11). Following induction of inflammation, both WT and A1−/− cells underwent increased apoptosis compared to their PBS injected counterparts but there were no significant differences in the level of cell death between the WT and A1−/− mice within each treatment.

No redundancy between A1 and other pro-survival proteins in the LPS induced lung inflammation model and peritonitis model

Neutrophils are highly sensitive to Fas-induced apoptosis [12]. The FasL–Fas-induced apoptotic pathway also plays a role in the death of other immune cells, including B cells and T cells [29]. It has previously been reported that A1 delays spontaneous and Fas ligand-induced apoptosis of activated neutrophils [21] and B cells [30]. Hence, one possible reason we did not observe a significant defect in neutrophils or B cells in our experiments with A1−/− mice could be due to the compensatory role of other pro-survival proteins. Our data suggest that the role of A1 in regulating immune cell survival during an inflammatory response may be redundant and may only become prominent in the absence of additional pro-survival proteins.

To investigated this possibility, we used the LPS-induced lung inflammation model in compound mutant mice that lack not only A1 but also lack one allele of a gene for an additional pro-survival BCL-2 family member (i.e., A1−/−; Bcl-X+/−, A1−/−; Bcl-2+/− and A1−/−; Mcl+/−). Partial loss of the other pro-survival proteins in addition to the complete absence of A1 did not lead to increased death of any of the immune cell sub-types tested (Supplementary Fig. S12). The lack of effect seen in these heterozygous compound mutants may be due to the fact that these proteins are still present albeit at a reduced level. Because homozygous deficiency of MCL-1, BCL-XL and BCL-2 is embryonic lethal in mice [31,32,33], we performed ex vivo studies using BH3 mimetics [34, 35] to block MCL-1, BCL-XL and BCL-2. We collected peritoneal lavage cells from WT and A1−/− mice that were injected with LPS 4 h prior and cultured these cells with BH3 mimetics [34, 35]. There were no differences in the rate of cell death between WT and A1−/− cells in the presence of BH3 mimetics, which indicate no redundancy between A1 and other pro-survival proteins (Supplementary Fig. S13).

Discussion

Neutrophils are key players during inflammation. Neutrophil apoptosis during the resolution of inflammation is necessary for their subsequent engulfment by macrophages [36]. We observed a strong upregulation of the pro-survival protein A1 during pathogen-induced inflammation, both in the lungs and the peritoneal cavity of challenged mice. We further showed that A1 levels were increased in neutrophils and hypothesised that this increase in A1 protects neutrophils from premature apoptosis, thereby extending their lifespan.

We, therefore, investigated the role of A1 in regulating neutrophil survival during inflammatory responses in vivo. We confirmed the upregulation of A1 in neutrophils in the lungs and peritoneal cavity during inflammation, but surprisingly, our findings indicate that A1 does not have a major role in regulating neutrophil survival at sites of inflammation. This is despite an interesting observation in which we detected lower levels of TNF and GM-CSF in the BALF of A1−/− mice at 4 h post LPS challenge. Alveolar macrophages have been shown as a key producer of TNF during lung inflammation that in turn promotes GM-CSF upregulation and secretion from the lung epithelium [37]. The overall lung macrophage numbers in A1−/− mice are comparable to WT mice before and after LPS treatment, which indicates that the reduction in TNF and GM-CSF levels in the inflamed lung of A1−/− mice was not due to an underlying defect in macrophage numbers. We also observed no major impact on immune cell mobilisation/recruitment, pathogen clearance or inflammatory cell survival in response to diverse stimuli in A1-deficient animals.

Observations in A1-deficient mice during inflammation may not truly reflect the role of its homologue BFL-1 in humans. There are several differences between mouse A1 and human BFL-1. Mice have four A1 genes (three expressed, one pseudo-gene), whereas humans only have one BCL2A1 gene [7]. These gene duplication events may have occurred and been preserved in evolution because different A1 proteins have designated roles in different tissues in mice [3]. In mice, A1 expression is restricted to cells of the haematopoietic compartment [3], while in humans BFL-1 expression is more widespread [4].

Although neutrophils appear to express similar levels of all three functional A1 genes [7], differential expression of these isoforms has been observed in other cell types. For example, A1-b is reported to be the predominant form expressed in thymocytes as well as in resting T cells and B cells. A1-a seems the least abundant isoform but one study reported its upregulation, alongside A1-d, upon TCR ligation in CD8+ T cells [38, 39]. Mouse A1 proteins are preferentially localised in the cytosol, whereas human BFL-1 can be found on the outer mitochondrial membrane. However, there is evidence that the pro-survival functions of A1 and BFL-1 may be independent of their subcellular localisation, at least when overexpressed at high levels [17, 40].

The balance of lymphocytes and neutrophils is rather different between humans and mice: human blood is neutrophil rich (~70% neutrophils, ~30% lymphocytes), whereas mouse blood has a strong preponderance of lymphocytes (~75% lymphocytes, ~25% neutrophils) [41, 42]. If A1/BFL-1 is required for neutrophil survival during inflammation, the functional consequences of its absence may be more prominent in humans.

The LPS induced murine lung inflammation model accurately mimics the neutrophilic inflammatory response seen in humans [43]. Administration of free-living PA to the murine lung results in either rapid bacterial clearance or acute overwhelming sepsis depending on the functionality of the immune system of the treated mouse [44].

We modelled human intra-abdominal inflammation by using both LPS and CS i.p. injections. LPS injection represents a model that is simple to use and highly reproducible [45] where the host responds to bacterial products rather than the pathogen itself. There is a marked difference in the response to LPS between species. Rodents are relatively resistant to LPS, whereas humans and non-human primates show more profound responses [45]. Compared to treatment with LPS, in the murine CS model, injection of free-living bacteria into mice promotes a lower but longer-lasting increase in the levels of pro-inflammatory cytokines with the more accurate manifestation of the pathological changes that occur in human sepsis [46, 47].

Despite the several models of inflammation used, we could not establish evidence for a critical involvement of A1 during inflammation. We could also not establish an overlapping function with other and likely more dominant pro-survival proteins, such as Mcl-1. Therefore, A1 is not an attractive therapeutic target for the treatment of inflammatory diseases.

Methods

Mice

The generation of A1−/− mice have been described previously [10]. All mouse strains have been generated and were maintained on a C57BL/6 background and equal proportions of 7–8 week-old males and females were used in all experiments. The mice were transported and housed short term under specific pathogen-free conditions at the La Trobe Animal Research and Training Facility (LARTF) for PA and CS experiments. Except for the PA and CS experiments, all other experiments were performed at WEHI.

Intranasal injection of LPS

Mice were slightly anaesthetised with Methoxyflurane. Anaesthetised mice received intranasal instillation of 10 µg of LPS and control mice received an equivalent volume of PBS. The mice were placed under a heat lamp or on a heating pad and were monitored until fully conscious.

Intranasal administration of Pseudomonas aeruginosa

Pseudomonas aeruginosa ATC27853 strain was grown overnight in 5 mL of LB broth. The bacteria were spun down and resuspended in PBS to obtain an OD600 of 0.7. Mice were anaesthetised (as described above) and administered with 40 μL of the solution containing PA (∼1.5 × 107 CFU) or 40 μL of PBS.

Induction of acute polymicrobial sepsis

The CS injection method was used to induce polymicrobial sepsis. In brief, 50 naive wild-type adult C57BL/6 mice were euthanised and 10 g of ceacal content was harvested. The caecal content was resuspended in 5% dextrose to obtain a CS stock of 250 mg/mL, which was then filtered through a 100 µm filter. This stock was aliquoted, stored and used for all CS experiments to prevent batch variations. Mice were injected with 0.1 mg/kg Buprenorphine 30 min before CS injection. Mice were i.p. injected with CS and monitored every 3 h in the first 72 h and then twice a day for 2 weeks. Mice were euthanised if signs of distress were observed.

Induction of inflammation by non-microbial components

For the gout model, 2 mg of MSU crystals dissolved in PBS were i.p. injected into mice and peritoneal lavage was obtained 16 h post-injection for Western blotting. For the acute asthma model, a suspension containing 20 µg ovalbumin (OVA) and 2.25 mg aluminium hydroxide was injected i.p. into mice on day 1 and day 14, and mice were aerosol challenged either with PBS or OVA for 15 min per day on days 21, 22 and 23. 24 h after the last exposure, mice were euthanised and organs harvested.

Harvesting of cells from mice

Mice were euthanised by CO2 asphyxiation. For bone marrow retrieval, the femur was flushed with 1 mL of PBS buffer. Cells were passed through a 100 μm cell strainer to obtain a single-cell suspension. For blood cell analysis, bleeds were taken into Microvette tubes containing anti-coagulants from mice immediately following euthanasia. For obtaining cells from the peritoneal cavity, mice were i.p. injected with 5 mL of cold PBS immediately following CO2 asphyxiation and gently massaged. The lavage containing the cells was drawn back into the syringe and the cells were then recovered by centrifugation. Peritoneal cells were subjected to flow cytometric analysis and the lavage fluid was used to measure cytokine levels by ELISA according to the manufacturer’s instructions (eBioscience).

For obtaining the BALF for cytokine ELISA, lungs were flushed with 0.3 mL of PBS twice. A total of 0.5–0.6 mL lavage were obtained from each mouse and were subsequently spun to pellet the cells. The BALF supernatant was subjected to cytokine analysis by ELISA. Pelleted BALF cells were pooled with the cells obtained from whole lung digestion. Lung tissues were minced, and tissue dissociation was carried out by enzymatic digestion (in medium containing 0.2 g/L glucose and 20 mg/mL Worthington Collagenase Type 1) as previously described [48]. All processed cell pellets requiring red blood cell lysis were resuspended in red cell lysis buffer (0.156 M NH4Cl) and left at 25 °C for 5 min. The supernatant was removed, and the cells were resuspended in FACS buffer (PBS with 2% foetal calf serum) and stained with appropriate antibodies.

Neutrophil isolation

Neutrophils were isolated from the pellet of cells resulting from the processed lungs by using the Stem Cell Technologies’ Mouse Neutrophil Enrichment Kit according to the manufacturer’s instructions.

P. aeruginosa colony formation assay

The left lobe of the lung was dissected and homogenised in 1 mL of PBS. In total, 20 µL of the suspension was dropped onto an LB agar plate as the undiluted sample. Subsequent dilutions were carried out in PBS and 20 µL from each of these dilutions was also dropped onto LB agar plates. Plates were incubated overnight at 37 °C and visible bacterial colonies were counted on the following day.

Western blotting

Total protein extracts were prepared by lysing cells in lysis buffer (20 mM Tris-pH 7.4, 135 mM NaCl, 1.5 mM MgCl2, 1 mM EGTA, 10% (v/v) glycerol and 1% (v/v) Triton-X-100; Sigma-Aldrich) with complete protease inhibitor cocktail (Roche) for 1 h at 4 °C. Equal amounts of protein were electrophoresed on NuPAGE 4–12% Bis-Tris gels (Invitrogen) before transferring to nitrocellulose membranes (Life Technologies) and probing with primary antibodies: monoclonal rat anti-mouse A1 (clone 6D6, WEHI antibody facility), monoclonal-mouse anti-HSP70 (clone N6, W. Welch USCF). Secondary anti-rat (#3010-05)/anti-mouse (#1010-05) IgG antibodies conjugated to HRP (Southern BioTech, Birmingham, AL, USA) were applied, followed by Luminata Forte Western HRP substrate (Millipore, Billerica, MA, USA) for band visualisation. Membranes were imaged using the ChemiDoc XRS + machine with ImageLab software (Bio-Rad). Quantification of the Western blot band intensities was carried out using the Fiji Image J software.

Statistical analysis

Data are presented as the mean ± sem. Unpaired two-tailed student’s t-test and P values were used to determine statistical significance. P values < 0.05 were considered statistically significant, and P values > 0.05 were considered non-significant. T-tests were corrected for multiple testing by controlling the false discovery rate. For mouse survival data analysis, the significance was calculated using the log-rank test (Prism Software, Graphpad). The variance was similar between statistically compared experimental groups. Either one-way or two-way analysis of variance (ANOVA) was used based on the experimental design.

Haematopoietic cell analysis and flow cytometry

Peripheral blood was analysed with the ADVIA automated haematology system (Bayer). Lung, bone marrow and peritoneal lavage cell populations were examined using flow cytometry. Cell populations were identified by staining with fluorochrome-conjugated monoclonal antibodies (produced in-house or purchased from BioLegend) that detect cell subset specific surface markers: B220 (BV605) (#103243, Biolegend), TCRβ (PE-Cy7) (#109222, Biolegend), MAC-1 (FITC), GR-1 (APC) and Ly5.2 (PE). Dead cells were excluded from analysis by staining with propidium iodide (PI, 5 μg/mL). To determine the intracellular PA, cells were fixed and permeabilised using eBioscience intracellular fixation and permeabilisation buffer set and stained with Rabbit anti-Pseudomonas antibody (#ab68538, Abcam) followed by a FITC-conjugated anti-rabbit IgG secondary antibody. Flow cytometry was performed on the LSR II flow cytometer (BD Biosciences) and data were analysed using FlowJo software (FlowJo LLC).