In a previous study, we developed eosinophilic asthma and neutrophilic asthma mouse models. Intraperitoneal ovalbumin (OVA) sensitization induced T_H2-derived eosinophilic inflammation in the mouse model^19,20,21. After the eosinophilic asthma mouse model was sensitized, using lipopolysaccharides (LPS) shifted the airway inflammation from eosinophil-predominant to neutrophil-predominant and highly correlated with levels of interleukin (IL)-17A and T_H17 cells²². Modifying previous studies, we sensitized 6-week-old female C57BL/6 mice (Orient Inc., Seongnam, Korea) with peritoneal injection of 100 µg of OVA grade V (Sigma-Aldrich, St. Louis, MO, USA) that was dissolved in 2 mg aluminum hydroxide (Sigma-Aldrich) at days 0 and 7. Mice were exposed to intranasal 50 µg OVA at days 14, 15, 21, 22, and 23 to induce eosinophilic inflammation. We used intranasal 10 µg LPS from Escherichia coli O11:B4 (Sigma-Aldrich) to induce neutrophilic inflammation at days 21, 22, and 23.

Figure 1 describes the drug challenges in the neutrophilic asthma mouse model. We randomly allocated the mice to the following groups: (1) control, (2) OVA+LPS+0.5% carboxymethyl cellulose (CMC), (3) OVA+LPS+clarithromycin (CAM) (10 mg/kg), (4) OVA+LPS+dexamethasone (DXM; 1 mg/kg)+CMC, and (5) OVA+LPS+CAM (10 mg/kg)+DXM (1 mg/kg).

The control group comprised five mice, and the drug-challenge groups comprised seven. We used 0.5% CMC (Sigma-Aldrich) as a vehicle to compare with other groups. The mice were exposed to peritoneal injection of 1 mg/kg DXM (Sigma-Aldrich) and 10 mg/kg CAM (Sigma-Aldrich) via oral gavage every 24 hours on days 14, 15, 16, 17, 18, 21, 22, and 23. The mice were sacrificed at day 24. All animal protocols were approved by the Animal Subjects Committee of the Catholic University of Korea.

We calculated airway hyperresponsiveness by respiratory system resistance (Rrs) with methacholine (Mch; Sigma-Aldrich) in the FlexiVent system (SCIREQ, Montreal, QC, Canada)^23,24. Mice were exposed to nebulized phosphate-buffered saline (PBS) for 3 minutes for baseline Rrs value, and we assessed Rrs after challenging them with increasing concentrations of Mch (6.25, 12.5, 25, and 50 mg/mL; Sigma-Aldrich) in PBS using an Aerosonic ultrasonic nebulizer (DeVilbiss, Somerset, PA, USA). We averaged and expressed the Rrs measured during each 3-minute sequence for each Mch concentration.

The trachea was exposed and cannulated with silicone tubing attached to a 22-gauge needle on a 1-mL tuberculin syringe. We withdrew bronchoalveolar lavage (BAL) fluid after instilling 0.8 mL cold sterile PBS through the trachea into the lung. We cytospun the recovered BAL fluid for 10 minutes at 1,000 ×g at 4℃ and stored supernatant at -80℃ until required for further analysis. We counted total cells in the BAL fluid us-ing a hemocytometer. After we adjusted the number of cells to 1×10⁶/mL, we centrifuged 50 µL aliquots of BAL fluid onto microscope slides for 5 minutes at 43 g and stained the slides with Diff-Quick (Sysmax, Kobe, Japan). We obtained the percentages of macrophages, eosinophils, lymphocytes, and neutrophils in the BAL fluid by counting 500 leucocytes in randomly selected fields of the slides under a light microscope.

After the mice were killed, their lungs were inflated, fixed in 4% paraformaldehyde for 24 hours, and embedded in paraffin wax. We cut sections at a thickness of 4 µm using a microtome and subjected the deparaffinized tissue sections to hematoxylin and eosin (H&E) staining for microscopic examination (magnification of ×200).

We measured the concentrations of IL-5 and IL-17 in the BAL fluids and IL-13, interferon γ(IFN-γ), and tumor necrosis factor α(TNF-α) in the lung homogenates using an enzyme-linked immunosorbent assay kit (R&D Systems, Minneapolis, MN, USA). The assay sensitivities were 7, 5, 2, and 7.21 pg/mL, respectively. We conducted the protocol according to manufacturer's instructions.

We disrupted separated lung tissues that had been immediately frozen in liquid nitrogen using a Polytron homogenizer (Pellet pestles cordless motor; Sigma) and centrifuged them. We purified the proteins from the supernatant and assessed the concentrations using the Bradford method. We separated the protein samples by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred them to a polyvinylidene difluoride membrane (Amersham Pharmacia Biotech, Little Chalfont, UK). After blocking with 5% skimmed milk (Difco, BD, San Jose, CA, USA), we incubated the membrane overnight with an anti-HDAC2 antibody (1:1,000, rabbit monoclonal; Abcam, Cambridge, UK). We washed the membrane three times with TBST and incubated it with a secondary antibody for 2 hours. Horseradish peroxidase-conjugated goat anti rabbit IgG (1:2,000, Santa Cruz Biotech, Dallas, TX, USA) was used as the secondary antibody. We detected the target proteins using the ECL Western Blotting Analysis System (Thermo Fisher Scientific, Rockford, IL, USA). We either exposed the membrane to X-ray film or analyzed it with an LAS 3000 (Fujifilm, Tokyo, Japan) image analyzer using Multi Gauge v. 3.0 software. We also quantified the HDAC2 and ß-actin band intensities with Multi Gauge.

We measured HDAC2 activity using the colorimetric EpiQuik HDAC2 Activity Assay Kit (Epigentek, Brooklyn, NY, USA). We incubated the nuclear extracts with specific substrate for 90 minutes at 37℃, followed by capture antibody for 60 minutes, and detection antibody for 30 minutes at room temperature. We measured absorbance at 450 nm in a microelisa reader.

All data are expressed as the means±SEM, and we used GraphPad Prism 5.01 (GraphPad Software, San Diego, CA, USA) for the data analysis. We performed the statistical analyses using one-way ANOVA and t tests and considered p<0.05 to indicate statistical significance.

The mice in the OVA+LPS group showed greater airway resistance after they received 12.5 mg/mL of Mch than did the control group (p<0.05) (Figure 2). CAM or DXM monotherapy group did not decrease Rrs after 12.5, 25, or 50 mg/mL of Mch, but the OVA+LPS-challenged mice treated with CAM+DXM showed decreased Rrs at 25 and 50 mg/mL of Mch challenge (p<0.05 and p<0.01). The change was significant compared with findings in the CAM and DXM monotherapy groups, especially after 50 mg/mL of Mch (p<0.05 and p<0.01).

In BAL fluid, total cell counts were higher in OVA+LPS-challenged mice (p<0.001) (Figure 3A). These mice that were treated with CAM+DXM revealed lower total cell counts (p<0.001) compared with those in the monotherapy groups. We conducted differential cell counts to evaluate reduced inflammation in the pathways based on microscopic morphological criteria. In OVA+LPS-challenged mice, alveolar neutrophil counts percentage was 20.7% and eosinophil counts percentage was 18.6%, which is higher compared to control group (0.2% and 0%). The numbers of alveolar neutrophils, eosinophils, and lymphocytes did not decrease with CAM or DXM monotherapy, but with the two together, the numbers of neutrophils decreased (16.3%, p<0.05) but not eosinophils (16.3%). On H&E staining, inflammatory cell accumulation improved in the DXM monotherapy group. In the CAM+DXM group, accumulation of inflammation cells improved more than it did in the monotherapy groups (Figure 3B).

OVA+LPS challenge increased the total protein amounts in the BAL fluid (p<0.01). After treatment with DXM monotherapy and a combination of CAM+DXM, total protein amounts in the BAL fluid decreased (p<0.001). Protein amounts were lower in the combination group than in the DXM monotherapy group, but difference was not statistically significant (Figure 4A).

T_H2 inflammation marker (IL-5 and IL-13) and T_H17 inflammation marker (IL-17A) were induced after OVA+LPS challenge in the mouse model (p<0.05; Figure 4B, C and p<0.001; Figure 4D); IFN-γ, and TNF-αincreased also (p<0.01) (Figure 4E, F). After treatment with CAM only, IL-5 and IL-13 decreased (p<0.05 for both). After treatment with DXM, total protein (p<0.001) (Figure 4A) and IL-17A decreased (p<0.01) (Figure 4B, C). We also noted decreases in the CAM+DXM group. IFN-γand TNF-αdecreased with CAM+DXM, but there was no attenuation in either the CAM or the DXM monotherapy groups (p<0.05) (Figure 4E, F).

HDAC activity is correlated to corticosteroid activity and decreases in many cases of corticosteroid insensitivity²⁵. HDAC2 activity in OVA+LPS challenged mice was lower than the activity in the control group (p<0.01) (Figure 5A) and it was not recovered after CAM treatment. HDAC2 activity was more recovered in the DXM-challenged group than in the control group (p<0.01) (Figure 5A). The CAM+DXM-challenged group recovered more HDAC2 activity than either the control or the CAM-challenged groups (p<0.05 for both) (Figure 5A).

Relative density analysis of HDAC2/β-actin showed significant differences. In the OVA+LPS-challenged mice, the relative density of HDAC2/β-actin was lower than in the control group (p<0.01) (Figure 5C); it was not significantly higher in the CAM-challenged group than in the control group. In the DXM-challenged group, relative density of HDAC2/β-actin was greater than that in the control group (p<0.05) (Figure 5C), and it was greater in the CAM+DXM-challenged group than in the DXM group (p<0.01) (Figure 5C).