Tracheal microbial populations in horses with moderate asthma

Abstract Background There are limited data on potential dysbiosis of the airway microbiota in horses with asthma. Hypothesis/Objectives We hypothesized that the respiratory microbiota of horses with moderate asthma is altered. Our objectives were (a) to quantify tracheal bacterial populations using culture and qPCR, (2) to compare aerobic culture and qPCR, and (c) to correlate bacterial populations with bronchoalveolar lavage fluid (BALF) cytology. Animals Eighteen horses with moderate asthma from a hospital population and 10 controls. Methods Prospective case‐control study. Aerobic culture was performed on tracheal aspirates, and streptococci, Pasteurella multocida, Chlamydophila spp., Mycoplasma spp., as well as 16S (bacterial) and 18S (fungal) rRNA subunits were quantified by qPCR. Results Potential pathogens such as Streptococcus spp., Actinobacillus spp., and Pasteurellaceae were isolated from 8, 5, and 6 horses with asthma and 3, 0, and 2 controls, respectively. There was a positive correlation between Streptococcus spp. DNA and 16S rRNA gene (r ≥ 0.7, P ≤ 0.02 in both groups), but the overall bacterial load (16S) was lower in asthma (1.5 ± 1.3 versus 2.5 ± 0.8 × 104 copy/μL, P < 0.05). There was no association between microbial populations and clinical signs, tracheal mucus or BALF inflammation. Conclusions and Clinical Importance This study does not support that bacterial overgrowth is a common feature of chronic moderate asthma in horses. Lower bacterial load could suggest dysbiosis of the lower airways, either as a consequence of chronic inflammation or previous treatments, or as a perpetuating factor of inflammation.


| INTRODUCTION
Mild and moderate asthma, formerly "inflammatory airway disease," is characterized by airway inflammation and airflow limitation. Clinical signs include cough, nasal discharge, and poor performance. Diagnosis is based on the presence of compatible clinical signs and documentation of pulmonary inflammation with bronchoalveolar lavage fluid (BALF) cytology, or airflow obstruction with lung function testing. 1 In the absence of clear criteria to distinguish between mild and moderate, the terms "moderate asthma" is used here for clarity. Almost 15% of adult horses have severe asthma (heaves) and while the prevalence of moderate asthma is unknown, there is indirect evidence that close to 50% of horses will have at least 1 episode of cough during their life. 2 The etiology of asthma is mainly immune and environmental, which is supported in part by the positive response of affected horses to corticosteroid administration and decreased exposure to hay. 1 The role of bacterial infection or colonization in the development of asthma in horses is uncertain, but it could be a contributing factor to exacerbations of clinical signs. For example, there is an association between tracheal bacterial count and cough, 3 and between tracheal mucus and bacteria isolated from tracheal aspirates in racehorses. 4,5 The pulmonary microbiome of horses with asthma is different from that of healthy horses, and the pulmonary microbiome is influenced both by the presence of the disease and by the environment. 6,7 In human asthma, bacteria have received less attention than viruses for their role in clinical exacerbations, but infections with Haemophilus influenzae, Moraxella catarrhalis, and Streptococcus pneumoniae are now recognized as opportunistic pathogens associated with acute asthma exacerbations. [8][9][10] Furthermore, changes in the composition of the lung microbiome are associated with bronchial hyperresponsiveness 11 and with corticosteroid resistance. 12 In clinical cases, tracheal aspirates are often performed with BALF analysis in horses suspected of moderate asthma, but the interpretation of tracheal bacterial culture is complicated by the bacterial diversity in healthy horses. 13 Moreover, culture-independent methods (PCR and sequencing) have highlighted that standard bacterial culture allows for the identification of only small subsets of bacteria. 14 This study proposes to document microbial species found in tracheal aspirates collected from horses with and without moderate asthma by conventional methods (aerobic culture) and molecular methods (quantitative polymerase chain reaction [qPCR]). The underlying hypothesis is that dysbiosis in the microbial populations could contribute to sustained pulmonary inflammation and possibly affect response to treatment. More specifically, we hypothesized that the respiratory microbiota of horses with moderate asthma is altered, with more tracheal bacteria than in controls, and that there is a link between lower airway bacterial populations and pulmonary inflammation. The aims of this study were (a) to quantify bacterial populations using bacterial culture and qPCR, (b) to compare aerobic culture and qPCR, and (c) to correlate bacterial populations with measures of BALF inflammation.

| Horses
To account for the difficulty of recruiting healthy controls, sample sizes were calculated for a 2 to 1 ratio for diseased versus healthy horses. Sample sizes were calculated to achieve a power of 80% with a 2-sample t test and an alpha level of 5%. For BALF neutrophil percentages estimated, mean and range for controls was 5% (3%-7%) and 20% (15%-50%) for the asthma group and the power analysis suggested 5 controls and 11 asthmatic horses. Bacterial counts were more difficult to estimate based on the literature 4 and the fact that they are unlikely to be normally distributed. Conservative estimates suggested that 5 controls and 9 asthmatic horses would be sufficient, but to account for sampling variation, we aimed to recruit 10 controls and 20 asthmatic horses, and were able to recruit 10   Hill, ON, Canada). Tracheal mucus accumulation was scored from 0 to 5, as previously described, 15 with 0 corresponding to no mucus, 1 to little mucus accumulation in small blobs, 2 to moderate mucus accumulation, 3 to marked, stream forming mucus blobs, 4 to large, pool-forming accumulation, and 5 to extreme accumulation of mucus. A score greater than 2 was considered abnormal. 1

| Tracheal aspirates
Tracheal aspirates were collected via a triple guarded endoscopic catheter (Triple stage catheter, MILA Endoscopic Microbiology Aspiration Catheter EMAC800, Mila International, Florence, Kentucky), approximately 10 cm proximally to the main carina. Twenty milliliter of sterile isotonic saline were infused and quickly aspirated. All samples were put immediately on ice and separated for bacterial and qPCR analysis. The samples for culture were submitted to the CDEVQ laboratory (Provincial Veterinary Diagnostic and Epidemiology Laboratory) within 1 hour and the samples for qPCR were processed and frozen at −20 C within 1 hour.

| Bronchoalveolar lavage
Bronchoalveolar lavage fluid was collected after the tracheal aspirate, as previously described. 16 The videoendoscope was advanced into the right principal bronchus until it wedged into a smaller bronchus while infusing 60 to 120 mL of lidocaine 0.5%. Two boluses of 250 mL of sterile prewarmed isotonic saline (0.9% NaCl) were infused and quickly reaspirated. Samples were processed within 1 hour. Smears were prepared by centrifugation (Cytospin model II, Shandon Southern Instruments, Sewickley, Pennsylvania) and stained with a modified Wright-Giemsa. Differential count was performed on 400 nucleated cells, excluding epithelial cells, by a board-certified clinical pathologist.

| Aerobic culture on tracheal aspirates
Standard aerobic bacterial culture was performed on blood agar plates.
For each sample, quantification was estimated with 2 techniques: (a) direct plating (spread-plate method, reported as colony forming units per ml [CFU/mL]), and (b) postcentrifugation (quadrant streaking method, reported as scores). Ten microliters of tracheal aspirate were directly spread on the plate (direct plating) and incubated at 37 C for approximately 24 to 48 hours before the colonies were counted. In order to increase the odds of isolating bacteria present in low concentrations, the rest of the sample was centrifuged and then plated according to the quadrant streaking method. Briefly, after a 24 to 48 hour incubation period, colonies were counted and reported semiquantitatively using a 0 to 5 system: a score of 0 represents no growth; 1 indicates that there is 1 colony in the first quadrant; 2 indicates 2 to 4 colonies in the first quadrant; 3 indicates 5 or more colonies in the first quadrant; 4 indicates that there are colonies in the second quadrant; and 5 indicates that there are colonies in the third quadrant. Culture was determined to be negative after 5 days with no growth. A direct smear was done using the pellet from the centrifuged sample and stained with the standard Gram's staining.

| Quantitative PCR on tracheal aspirates
One milliliter of tracheal aspirate was centrifuged for 30 minutes at  16S subunit of bacterial ribosomal RNA), and 18S rRNA gene (gene coding for the 18S subunit of fungal ribosomal RNA). Primers are presented in Table 1, with their optimal cycle conditions. Most of them were designed manually from a specific gene sequence and all primer sets were tested for sensitivity, optimal annealing temperature, and efficiency. For each target, high-quality DNA was extracted from colonies provided by the CDEVQ laboratory and was used to build standard curves with an efficiency ≥90% and an R 2 ≥ 0.998. Primers for Actinobacillus equuli, suis, ligneresii, and pleuropneumoniae were also designed but satisfactory standard curves could not be attained. All qPCR tests were carried out in duplicates (except for Streptococcus equi subsp. equi due to the limited amount of tracheal aspirate available) and means of the duplicates were used for analysis. 3 | RESULTS

| Horses
Horses with asthma included 7 mares and 11 geldings, aged from 3 to 18 years old (mean ± SD: 9.89 ± 4.10 years). The control group included 4 mares and 6 geldings, aged from 2 to 16 years old (8.00 ± 4.57 years). There was no significant difference between groups for age (P = 0.29, Figure 1A) and sex (P > 0.9). Nine horses with asthma and 5 controls were Quarter Horses and associated breeds. Other breeds included Warmbloods (3 in each

| Previous medication
Four horses with moderate asthma had received antimicrobials (ceftiofur or trimethoprim-sulfadiazine) 1 week (2 horses) and 4 weeks (2 horses) prior to presentation. Another 5 horses had received ceftiofur or trimethoprim-sulfadiazine between 1 and 3 months prior to presentation. None were receiving antimicrobials at the time of presentation.
Nine horses with moderate asthma had received systemic or inhaled corticosteroids (dexamethasone, prednisolone, fluticasone) in the 3 months prior to presentation, 1 of them in the 2 weeks prior to presentation and 2 others during the week of presentation. Other treatments received prior to presentation included bronchodilators, expectorants, antihistamines, and nonsteroidal anti-inflammatory drugs. One control horse (with myocarditis) was empirically treated with corticosteroids (prednisolone) in the 2 weeks prior to presentation, due to his poor performance.
Analyses were done with and without this horse, with no changes in results. He was therefore kept in the final analysis.

| Bronchoalveolar lavage cytology
Bronchoalveolar lavage fluid neutrophil and mast cell percentages were significantly higher (P < 0.001), and macrophages were significantly lower (P = 0.01) in horses with asthma ( Figure 2). There was no significant difference between groups for eosinophils and lymphocytes.

| Real-time PCR of tracheal aspirates
All tracheal aspirates were positive for the 16S and 18S rRNA genes, with overall higher concentrations for 16S than for 18S, and higher concentrations of the 16S rRNA gene in controls than in horses with asthma ( Figure 4A, P = 0.01). This remained true when horses who had received antimicrobials were removed from analysis (P < 0.05) but not when horses with corticosteroids were removed (P = 0.06). Within horses with asthma, there was no significant differences between horses with and without prior treatments (P > 0.6). Streptococcus spp.
and Chlamydophila spp. were the 2 bacteria most commonly identified, with no differences between groups ( Figure 4B). The quantification of specific Streptococci (equi, zooepidemicus, and pneumoniae) was positive only for Streptococcus pneumoniae in a control horse. Mycoplasma spp. and Pasteurella multocida were also only detected occasionally in each group ( Figure 4B). The proportion of samples positive by qPCR are summarized in Table 2. There

| DISCUSSION
The results from this study do not support our hypothesis that horses with moderate asthma have greater bacterial loads, or that qPCR would increase the number of horses for which an infectious component would be suspected. There was a lower burden in horses with asthma (based on 16S rRNA gene quantification), with more than twice as many samples with no growth in that group, and Corynebacterium spp. was significantly more common in control horses. The lower bacterial load and less frequent commensal bacteria (Corynebacterium spp.) in our population of horses with asthma suggests that the respiratory bacterial microbiota might be altered in moderate asthma, but that bacterial overgrowth is not a consistent feature in these horses.

| Airway dysbiosis in equine and human asthma
These results are in contrast with studies performed in racehorses where bacteria, mainly streptococci and Pasteurellaceae, are associated with signs of respiratory disease, 3 tracheal mucus, 4,17 and tracheal inflammation. 18,19 These previous studies mostly focused on young racehorses that are more likely to have viral and bacterial infections than older horses, as well as to have contamination from the environment during strenuous exercise. The horses in the current study were on average ≥8 years old (ie, older than racehorses) and, to be included, their condition had to have a certain chronicity, as per the American College of Veterinary Internal Medicine (ACVIM) guidelines. 1 These criteria are rarely met on the racetrack, where horses tend to be examined shortly after they develop a cough, and sometimes in the absence of clinical signs, and where inflammation is also often based on tracheal cytology instead of BALF cytology. Furthermore, horses in the current study were likely to have been treated for their respiratory condition, including with antimicrobials and corticosteroids in the months before presentation.
In human asthma, in addition to viral infections, specific bacterial pathogens are now recognized as being associated with exacerbations, [8][9][10] but few studies have documented an increase in overall bacterial load in asthmatic patients. 11 Evidence of dysbiosis is mostly supported by studies using next-generation sequencing showing differences in the relative abundance of the bacteria present (commensal and potential pathogens). This has been documented in children and adults with asthma, and dysbiosis is now associated not only with asthma prevalence and acute exacerbation, but also with asthma severity and airway hyperresponsiveness. 11,12,20,21 Dysbiosis of the airways was also described recently using next-generation sequencing in horses with moderate and severe asthma, 6,7 using inclusion criteria in line with the current ACVIM recommendations. Tracheal microbiota was altered in horses with moderate asthma, with a greater relative abundance of the genus Streptococcus spp., 6 and pulmonary microbiota was also altered in horses with severe asthma, but without overrepresentation of specific pathogens in bronchoalveolar lavages. 7 Neither studies attempted to quantify the absolute amount of specific pathogens. At this point, it is still unclear as whether bacterial dysbiosis in asthma should be considered a causative factor of asthma exacerbation, a perpetuating factor of inflammation, or a consequence of chronic inflam-  3,4 Mycoplasma spp. are also found in children with asthma. 24 Chlamydophila spp., another atypical bacteria associated with asthma exacerbation in humans, 25 were detected by immunohistochemistry more frequently in the bronchial epithelial cells of asthmatic horses. 26 It is also important to realize that several bacteria such as Chamydophila spp. and Mycoplasma spp. do not grow or are not readily identified on standard culture, which reinforces the relevance of culture independent techniques such as PCR. Finally, Actinobacillus spp. (from the Pasteurellaceae family) are also associated with increased tracheal mucus in racehorses. 17 In the current study Actinobacillus spp. were isolated by culture in the asthma group only. Unfortunately, despite numerous attempts and different primers tested, standard curves of adequate accuracy could not be attained.

| Potential confounding factors
Investigations based on clinical cases come with conditions that cannot be controlled as tightly as in laboratory research settings. One potential confounding factor is the fact that many horses in our study had received corticosteroids or antimicrobials in the months prior to presentation, and both medications have the potential to affect airway microbiota. Specifically, dexamethasone induced changes in the tracheal microbiota of healthy and asthma-affected horses, with a relative increase in the abundance of certain genera of bacteria (Peptostreptococcus, Porphyromonas, Filifactor, Streptococcus, Porphyromonas, Parvimonas, Fusobacterium, and Bacteroides spp.) but with a decrease in 1 operational taxonomic unit from the ubiquitous candidate Saccharibacteria phylum. 6 The duration of the change in microbiota after cessation of antimicrobial treatment was not investigated. However, in another study, the addition of antibiotics (ceftiofur) to a standard treatment with corticosteroids improved clinical scores but did not change the amount or type of bacteria cultured from their tracheal aspirates. 27 This suggests that previous treatment with antimicrobials (mostly with ceftiofur and trimethoprim-sulfadiazine) are less likely to be a major confounding factor in our study. As horses were not on antimicrobials at the time of the visit, it is unlikely to have influenced directly the results of the culture and PCR, but that could have decreased the prevalence of secondary bacterial infections. Our data suggest that prior antimicrobial administration was not the cause of lower bacterial load in horses with asthma, with the possible exception of Actinobacillus, but interpretation is complicated by the fact that there is an overlap in the horses that received antimicrobials and corticosteroids in the asthma group (9 horses for each, but 7 received both, usually not at the same time). In addition, even in a subset of patients with a high positive culture or PCR, bacteria were not necessarily associated with tracheal mucus or clinical signs pointing toward bacterial overgrowth.
Finally, mucus can decrease DNA extraction efficacy and could have contributed to the lower bacterial load (16S rRNA gene) in horses with asthma. Tracheal mucus should not have affected culture and therefore it is not a major confounding factor.
In conclusion, this study does not support that bacterial overgrowth is part of moderate equine asthma as defined by the current ACVIM consensus statement, at least in the chronic phase. However, lower 16S rRNA gene and commensal bacteria in the asthma group support the presence of lower airway dysbiosis. We cannot conclude from these data whether dysbiosis results from chronic inflammation or the treatments that these horses received. There was, however, a strong positive correlation between Streptococcus detected by PCR and 16S rRNA gene, suggesting that when bacteria are present, they are more likely to be of the Streptococcus genus. We can also conclude that large numbers of Streptococcus zooepidemicus isolated on tracheal aspirates should not be considered normal in healthy and asthmatic horses.