Effects of azithromycin on bronchial remodeling in the natural model of severe neutrophilic asthma in horses

Steroid resistance in asthma has been associated with neutrophilic inflammation and severe manifestations of the disease. Macrolide add-on therapy can improve the quality of life and the exacerbation rate in refractory cases, possibly with greater effectiveness in neutrophilic phenotypes. The mechanisms leading to these beneficial effects are incompletely understood and whether macrolides potentiate the modulation of bronchial remodeling induced by inhaled corticosteroids (ICS) is unknown. The objective of this study was to determine if adding azithromycin to ICS leads to further improvement of lung function, airway inflammation and bronchial remodeling in severe asthma. The combination of azithromycin (10 mg/kg q48h PO) and inhaled fluticasone (2500 µg q12h) was compared to the sole administration of fluticasone for five months in a randomized blind trial where the lung function, airway inflammation and bronchial remodeling (histomorphometry of central and peripheral airways and endobronchial ultrasound) of horses with severe neutrophilic asthma were assessed. Although the proportional reduction of airway neutrophilia was significantly larger in the group receiving azithromycin, the lung function and the peripheral and central airway smooth muscle mass decreased similarly in both groups. Despite a better control of airway neutrophilia, azithromycin did not potentiate the other clinical effects of fluticasone.

The standard respiratory mechanics were performed in unsedated standing horses. Briefly, a polyethylene catheter with an air-filled (5 ml) balloon at its tip was inserted in the distal third of the esophagus to obtain a measurement of the pleural pressure (PL). Flow rates were measured during a two-minute period with a heated pneumotachograph and a differential pressure transducer fitted to a mask placed over the horse's nose. The head was positioned to minimize upper airway resistance during measurements 2 . The commercial software (Flexiware 7.6, SCIREQ, Montreal, QC, Canada) allowed electronic integration of the flow to obtain volume, and together with the pressure signals to obtain values of pulmonary resistance (RL) and elastance (EL) by applying multiple linear regression to the equation for the single compartment model of the lung ( = + ̇+ ); V is the volume, ̇ the airflow and K the transpulmonary endexpiratory pressure. All valid breaths obtained over the two-minute period were averaged for analysis.
Lung function was also measured with oscillometry as previously described 3 as it is possibly more sensitive than standard lung function measurements for detecting airway obstruction 4 . The Equine MasterScreen impulse oscillometry system (Jaeger GmbH, Würzburg, Germany) was calibrated before each use. Multifrequency impulses generated by a loudspeaker were superimposed on the tidal breathing by an airtight mask placed over the horse's nose. The pressure-flow signal response of the respiratory system was measured by a pressure transducer connected to a pneumotachograph attached to the front of the facemask. Three recordings of 30 seconds duration were acquired. Analysis were performed with LabManager (version 4.53, Jaeger, Würzburg, Germany) and FAMOS (IMC, Meβsysteme, Berlin, Germany) using Fast-Fourier transform. Inspiratory (insp), expiratory (exp) and resistance (R3) and reactance (X3) of the respiratory system at 3 Hz were analyzed. The ratio of resistance at 5 Hz and 10 Hz (R5/R10) was used as an indicator of the frequency dependence of the respiratory system resistance. . Differential leukocyte counts were performed blindly from 400 cells. In the equine species, BALF differential cell count is provided instead of total counts for each leukocyte as the BALF volume recovered is dependent on the degree of airway obstruction 6 . BALF was centrifuged at 500 x g for five minutes, then the pelleted cells were washed twice in PBS, and 10 million cells were re-suspended in Trizol reagent (Invitrogen, Carlsbad, CA, USA) and kept at −80°C until RNA extraction.

Endobronchial ultrasound
Endobronchial ultrasound was performed as previously reported 7 with a 20 MHz radial miniature probe (UM-BS20-26R, Olympus Canada) with the compatible balloon-ended sheath (MAJ-643R, Olympus Canada). The images with the highest quality were analyzed by a blinded investigator (at least 3 images/airway) 7 with Image J (version 1.52a, NIH, Bethesda, USA).

Endobronchial biopsies
Endobronchial biopsies were collected after BAL and EBUS, as previously described 8

Peripheral lung biopsies
Peripheral biopsies of the caudo-dorsal region of the lungs were obtained by thoracoscopy by a board- administered to horses not receiving azithromycin directly after biopsy retrieval. Biopsies were fixed in 10% neutral-buffered formalin for 24 hours and then processed as were the endobronchial biopsies.
Peripheral bronchi with a diameter <2 mm and a major to minor axis ratio ≤ 1.5 were analyzed using Image J (version 1.52a, NIH). The airway smooth muscle (ASM) and ECM areas were measured manually by tracing the external borders of each region. The ECM corresponded to the area between the basal membrane (Pmb) and ASM. The Pmb length was traced manually and used for correction attributable to variation of airway size 10 .

Azithromycin concentration
Azithromycin was extracted from plasma using protein precipitation. Two hundred L of internal standard solution (100 ng/mL of azithromycin-d3 in acetonitrile) was added to 50 µL of plasma samples. The sample was vortexed for five seconds and rested for 10 minutes, then centrifuged at 12 000 x g for 10 minutes.

5
The supernatant was transferred into an injection vial. The chromatography was performed using a gradient mobile phase along with a microbore column Thermo Biobasic Phenyl 50 × 1 mm, with a particle precursor ions using a 5 ppm mass window. Instrument calibration was performed prior to all analysis and mass accuracy was notably below 1 ppm using Thermo Pierce calibration solution and automated instrument protocol. Azithromycin quantification was performed using peak-area ratio of azithromycin and deuterated analog azithromycin-d3 and concentrations were determined by interpolating unknowns from the calibration curve (i.e. 2 to 1,000 ng/mL) constructed with standards prepared in plasma. The observed precision and accuracy were within +/-15%. Intracellular concentration of azithromycin in PMNs was estimated as previously described 11 at W20. For quantification of SMMHC and its (+) insert isoform, 2.75 mM MgCl2 was added to the reactions.

Transcriptomic analysis by qRT-PCR
Samples were run with negative controls (reverse transcription negative and PCR negative controls) and with samples with known quantity of the evaluated gene (standard curves).

FIGURES IN SUPPLEMENT
Supplementary figure 1. Lung function measured by impulse oscillometry. a) Ratio of the resistance at 5 Hz and 10 Hz (R5/R10). b) Reactance at 3 Hz (X3) (mean and standard error of the mean). There was a significant time main effect but no group difference with the two-way ANOVA for R5/R10 and X3 values (p < 0.0001). * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001 compared to baseline values with Dunnett's multiple comparison tests.