We used A_2A (A_2A^-/-) and A_2B gene-deficient mice (A_2B^-/-) (from Dr. Katya Ravid, Boston University, School of Medicine, Department of Biochemistry, Boston, MA, USA) and obtained corresponding wild-type mice (CD1 and C57BL/6 respectively) from Charles River Laboratories (Germany). Mice were males and between 8 and 12 weeks old. All animal protocols were approved by the Animal Care and Use Committee of the University of Tübingen.

Hemin was injected intraperitoneally (i.p.) (80 µmol/kg) 24 h before LPS inhalation to increase HO-1 activity (15). To inhibit the enzyme, SnPP (50 µmol/kg) was additionally administered i.p. 1 h before LPS inhalation (15). The specific equilibrative nucleoside transporter 1 (ENT1) antagonist NBTI was administered as described previously (9). Cobalt (III) protoporphyrin-IX-chloride (CoPP) was injected at 5 µg/g body weight i.p. 24 h before the inflammatory stimulus (23).

Total RNA was isolated from murine lung samples by using pegGOLD TriFast™ (Peglab, Germany), and cDNA synthesis was performed by Bio-Rad iSkript-kit (Bio-Rad, Germany) according to the manufacturer’s direction.

We determined the expression of HO-1 in the lungs of wild-type, A_2A^-/-, and A_2B^-/- mice by RT-PCR (n = 5–8). This method was performed with the following HO-1 primers (5′-GAG ATT GAG CGC AAC AAG GA-3′ and AGC GGT AGA GCT GCT TGA ACT-3′) as described (9). 18S was used as house keeping gene (5′-GTA ACC CGT TGA ACC CCA TT-3′ and 5′-CCA TCC AAT CGG TAG TAG CG-3′).

To further verify the gene expression results on a protein level, we measured HO-1 protein expression in the lungs of mice by ELISA (ENZO Life Sciences, Lörrach, Germany) (n = 5–8). Mice inhaled LPS and HO-1 was induced, respectively, inhibited as described above.

Lungs of mice were removed to measure the enzymatic activity of HO-1 24 h after LPS inhalation (n = 6–8). HO-1 was induced by hemin, respectively, inhibited by the additional administration of SnPP, lungs were removed and weighed and HO-1 activity buffer (twice of weight) added. The samples were frozen in liquid nitrogen. Lungs were homogenized, sonicated, and centrifugated at 18,000 g for 15 min. The supernatant was used for protein- and activity-measurement. Cytosol of the liver was obtained from 12 h fastened mice via centrifugation at 105,000 g for 27 min. The HO-1 activity assay consisted of 100 µl of the supernatant, 131 µl of HO-activity buffer, 100 µl of liver cytosol, 50 µl of glucose-6-phosphate (20 nM), and 10 µl of hemin (1 mM). After incubating for 1 h in the dark at 37°C, 500 µl chloroform was added, followed by a centrifugation of 15,000 g for 5 min. Extinction was measured at 464 and 530 nm, and HO-activity was calculated based on protein level, which was determined by a colorimetric method (bicinchoninic acid; Thermo Scientific, Rockford, IL, USA).

Heme oxygenase-1 was stimulated as described above, inflammation induced and lungs of mice removed after 24 h (n = 4). The circulatory system of the lungs was flushed and lungs with 4% paraformaldehyde (PFA) for 10 min at 25 cmH₂O inflated. Lungs were removed and fixed in PFA for 24 h. Rabbit polyclonal anti-HO-1 was used as primary antibody (Enzo, Life Sciences GmbH, Lörrach, Germany) to mark the enzyme and the endothelial marker von Willebrand Factor (Santa Cruz Biotechnology, Santa Cruz, CA, USA) was employed and additionally DAPI for nuclear staining. Images were visualized by using a confocal microscope (LSM 510, Meta, Carl Zeiss). Images shown are representatives of four experiments and were analyzed using ImageJ, a public program developed at the National Institutes of Health to officially analyze scientific images.

SiRNA of CD39 and CD73 (Santa Cruz Biotechnology, Heidelberg, Germany) was used to knock down gene expression of both ectonucleotidases in pulmonary epithelial cells (NCI-H441). In H441 cells, HO-1 was activated by hemin (5 µM), and cells were stimulated with LPS (100 ng/ml) for 4 h. Gene expression of interleukin-6 (IL-6) and 8 (IL8) were detected in cells by RT-PCR (n ≥ 5) by using the following primers: IL6 (5′-GAC AGC CAC TCA CCT CTT CA-3′ and 5′-CAC CAG GCA AGT CTC CTC AT-3′); IL8 (5′-TGT GGG TCT GTT GTA GGG-3′ and 5′-GTG AGG TAA GAT GGT GGC-3′); and 18S as house keeping gene (5′-GTA ACC CGT TGA ACC CCA TT-3′ and 5′-CCA TCC AAT CGG TAG TAG CG-3′).

Interleukin-6 and IL8 protein were detected in the supernatant of cells by ELISA (R&D Systems, Minneapolis, MN, USA) (n ≥ 5).

3 h after LPS inhalation, the release of CXCL1, CXCL2/3, tumor necrosis factor-α (TNFα), and IL-6 was measured in the BAL of mice (n = 5–9) by ELISA (R&D) according to the manufacture’s protocol.

Evans blue extravasation was determined as a marker of capillary leakage as described before (9). 6 h after LPS exposure, Evans blue (20 mg/kg, Sigma-Aldrich, Steinheim, Germany) was injected into the tail vein (n = 5–8). Intravascular Evans blue in the lungs was removed 30 min later by flushing the beating right ventricle. Lungs were homogenized and Evans blue was extracted with formamide and the final concentration determined colorimetrically (26).

We performed the in vitro transmigration assay of human PMNs through a monolayer of NCI-H441 cells (n = 6–12) or primary pulmonary endothelial cells (HMVEC-L, Lonza Walkersville, MD, USA) (n = 6–10) as described previously (9). HO-1 was induced by hemin (5 µM) and the A_2A- (ZM 241385; Tocris, MO, USA) (10 ng/ml) or A_2B-receptors (PSB 1115 Sigma-Aldrich, Taufkirchen, Germany) (10 ng/ml) were inhibited with the respective antagonists.

Data are presented as the mean ± SD unless indicated otherwise. Statistical analysis was performed using GraphPad Prism version 5.3 for Windows (GraphPad Software, San Diego, CA, USA). Differences between the groups were evaluated by one-way ANOVA followed by Bonferroni post hoc test or Student t-Test to compare two groups. P < 0.05 was considered statistically significant.

In wild-type mice, hemin caused a significant increase in the gene expression of HO-1 in the lungs even without inflammation (all P < 0.05). Since both wild-type mice acted identically, we display only one strain in Figure 1 (CD1 in Figure 1A C57BL/6 in Figure S1 in Supplementary Material). LPS also caused a significant rise, and the additional HO-1 stimulation did not show an additive effect on the gene expression.

Neither inflammation nor HO-1 activation by hemin changed the expression of HO-1 in A_2A^-/- and A_2B^-/- mice.

Protein expression of HO-1 in the lungs of mice confirmed the significant increase in the enzyme in LPS-treated wild-type animals (all P < 0.05) (Figure 1B; Figure S1B in Supplementary Material). Surprisingly, the augmentation after inflammation was also apparent in A_2A^-/- and A_2B^-/- mice at the protein level.

To further verify our findings from the gene and protein expression experiments, we measured the activity of the HO-1 enzyme. In wild-type animals, hemin increased HO-1 activity without inflammation, whereas the inhibition caused a significant reduction (all P < 0.05) (Figure 1C; Figure S1C in Supplementary Material). Inflammation also enhanced HO-1 activity, and the additional activation of the enzyme showed an additive effect.

In A_2A^-/- and A_2B^-/- mice, hemin did not influence the activity of HO-1 without inflammation. Additionally, LPS did not change the activity, but additional hemin administration caused a significant increase in HO-1 activity in both strains (all P < 0.05).

To evaluate the unclear source of HO-1 activity, we performed immunofluorescence staining of lung sections. Without inflammation, HO-1 was only slightly increased in wild-type animals after hemin treatment (Figure S2 in Supplementary Material). Inflammation significantly increased the expression of HO-1 in lung tissue (P < 0.05) (Figure 2). In wild-type animals, hemin further increased HO-1 in lung tissue and alveolar macrophages (P < 0.05).

In A_2A^-/- and A_2B^-/- mice, inflammation and the administration of hemin did not change HO-1 expression in lung tissue. The activation of HO-1 increased HO-1 expression in alveolar macrophages in these mice, explaining our findings from the protein expression experiment and the determination of the activity of the enzyme. We had previously determined that HO-1 is expressed in pulmonary tissue, alveolar macrophages, and PMNs, but the anti-inflammatory effects of HO-1 in acute pulmonary inflammation are based on the expression of the enzyme in the tissue (15).

In wild-type animals, stimulation or inhibition of HO-1 did not influence PMN counts in animals without inflammation (Figure S3 in Supplementary Material). LPS inhalation significantly increased the migration of PMNs into the intravascular, interstitial, and alveolar space, whereas the stimulation of HO-1 by hemin significantly reduced the PMN influx into all three compartments (all P < 0.05) (Figures 3A,B). Inhibition of HO-1 increased the attachment of PMNs to the endothelium and significantly enhanced PMN migration into the lung interstitium and alveolar space.

Without inflammation, neither stimulation nor inhibition of the enzyme influenced the number of PMNs in the lungs of A_2A^-/- mice (Figure S3A in Supplementary Material). LPS caused a significant increase in PMNs in the intravascular, interstitial, and intra-alveolar space (all P < 0.05) (Figure 3A). In contrast to wild-type animals, stimulation and inhibition of HO-1 did not affect PMN counts in any compartment of the lung.

Next, we investigated the impact of the adenosine receptor A_2B on the enzyme HO-1 in this setting (Figure S3B in Supplementary Material). Similar to the wild-type animals of the A_2A^-/- mice, the wild-type animals of these knockout mice showed a reduction in the number of PMNs in all three compartments of the lung in response to HO-1 stimulation (Figure 3B). Inhibition of the enzyme significantly reversed this effect in the lung interstitial and alveolar space (all P < 0.05). In A_2B^-/- mice, neither activation nor inhibition of the enzyme affected PMN counts in any of the lung compartments, highlighting the importance of A_2A- and A_2B-receptor signaling in the anti-inflammatory effects of HO-1.

As previously shown by our group, A_2A^-/- and A_2B^-/- mice have an augmented inflammatory reaction to LPS inhalation compared to wild-type animals (8, 27). In the present study, we did not directly compare wild type, A_2A^-/- and A_2B^-/- mice. The experimental setting was rather based on the identification of the effects of stimulating HO-1 in these mice. Therefore, the composition of the study groups has been shown to affect the results (9, 28).

Elevation of extracellular adenosine has anti-inflammatory effects in acute inflammation (29). To investigate if the anti-inflammatory effects of hemin are mainly based on changes in extracellular adenosine levels or are specifically mediated by A_2A-/A_2B-receptor signaling, we performed additional experiments. The ENT1 is a transmembrane protein that mediates the cellular uptake of adenosine (30). We increased the extracellular levels of adenosine by the specific ENT1 antagonist NBTI and investigated the therapeutic effect of hemin. There were no differences in PMN migration into the lung between the groups without LPS (Figure 3C). After inflammation, the PMN counts were significantly increased in the lung interstitium and the alveolar space and were significantly reduced by the administration of hemin (all P < 0.05), suggesting that the anti-inflammatory effects of hemin are not based on increased adenosine uptake in our model.

To further prove the A_2A- and A_2B- dependency of the anti-inflammatory effects of HO-1, we investigated the impact of lower extracellular adenosine levels on the anti-inflammatory effects of the enzyme. CD39 and CD73 are the enzymes that convert precursor nucleotides to adenosine; therefore, the knockdown of both enzymes decreases extracellular adenosine detrimental. In our experiments, hemin still exhibited anti-inflammatory effects concerning the expression and the release of IL6 and IL8 after the knockdown of both ectonucleotidases in human pulmonary epithelial cells (Figure 3D). These results confirm that hemin has still anti-inflammatory effects at reduced adenosine levels.

CoPP, as an alternative to hemin, was used to stimulate HO-1. These separate experiments were performed to exclude that the observed effects of mice lacking the A_2A-, respectively, the A_2B-receptor are specific for HO-1 induction by hemin. All mice inhaled LPS and PMN migration into the alveolar space was evaluated (Figure 3E). In both wild-type mice, the administration of CoPP decreased migrated PMNs into the alveolar space as described before (23). In A_2A^-/- and A_2B^-/- mice, the administration of CoPP did not affect PMN counts, concordantly to our results using hemin for HO-1 stimulation.

Inhalation of LPS significantly increased TNFα, IL6 and the most potent chemoattractive chemokines for PMNs, CXCL1, and CXL2/3 (all P < 0.05). Hemin reduced all four chemokine levels in both wild-type strains (Figures 4A,B), confirming our results from the in vivo PMN migration assay.

In A_2A^-/- animals, the inflammation caused a significant rise of all chemokines (Figure 4A). Activation of HO-1 did not affect the release of chemokines in these mice, thereby further verifying our findings from the in vivo transmigration assay.

Hemin did not affect the chemokine levels in A_2B^-/- mice, highlighting and further emphasizing the A_2A- and A_2B-dependent anti-inflammatory effect of HO-1 (Figure 4B).

Extravasation of Evans blue as an indicator for the permeability of albumin and, therefore, a marker for capillary leakage was assessed (Figures 5A,B). Stimulation and inhibition of the enzyme HO-1 without inflammation did not affect microvascular permeability (Figure S4 in Supplementary Material). LPS inhalation increased Evans blue significantly, whereas the stimulation of HO-1 preserved capillary leakage almost to baseline (all P < 0.05) (Figure 5A). As a control, inhibition of the enzyme was shown to further increase the microvascular permeability in wild-type mice.

Inflammation increased the extravasation of Evans blue in A_2A^-/- (Figure 5A) and A_2B^-/- mice (Figure 5B). Activation and inhibition of HO-1 did not affect capillary leakage in A_2A^-/- mice. We also demonstrated that HO-1 activation in A_2B^-/- mice had no effect on microvascular permeability, highlighting the pivotal role of A_2A^-/- and A_2B^-/- receptor signaling in conjunction with the enzyme HO-1 in the appearance of the second hallmark of acute pulmonary inflammation.

In this assay, isolated human PMNs migrate through a monolayer of pulmonary epithelial/endothelial cells along a chemotactic gradient. Since the anti-inflammatory effect of HO-1 in the setting of acute pulmonary inflammation depends on the expression of the enzyme in lung tissue (15), the pulmonary epithelium/endothelium was treated with hemin and specific A_2A/A_2B-receptor antagonists. Hemin treatment of epithelial (Figure 6A) and endothelial cells (Figure 6B) significantly reduced the number of migrated PMNs (all P < 0.05). After treatment with the specific A_2A/A_2B-receptor antagonists, the stimulation of HO-1 was ineffective in reducing the PMN influx in both cell lines, confirming our in vivo experiments.