Materials and Methods

Animals—Two groups of horses (an RAO-susceptible group and a control group) were studied concurrently during 2 experiments. The RAO-susceptible group in experiment 1 consisted of 4 mares and 4 geldings (mean ± SD age, 17.8 ± 3.9 years) that ranged from 520 to 540 kg. This group comprised 1 Arabian, 1 Appaloosa, 1 Paso Fino, 3 Quarter Horses, 1 Standardbred, and 1 Warmblood. In experiment 2, 6 of the same 8 RAO-susceptible horses (without the Paso Fino gelding and Standardbred mare) were used. Mean age was 18.1 ± 2.2 years, and the horses ranged from 520 to 540 kg. All of these horses had been included in other experiments, and it was known that they developed RAO when housed in a stable with shavings for bedding and dusty hay for feed.²⁴ Criteria used to define the RAO phenotype²⁵ were the development of pulmonary neutrophilia (≥ 25% neutrophils in BALF) and an accentuated breathing effort (ΔPpl_max ≥ 15 cm H₂O).

The control group consisted of 6 mares and 1 gelding (mean ± SD age, 16.4 ± 4.5 years) that ranged from 400 to 520 kg. This group comprised 4 Thoroughbreds and 3 Warmbloods. In the second experiment, the control group consisted of 5 of the same 7 horses (without a Thoroughbred mare and a Warmblood mare). Mean age of the control group for experiment 2 was 17.1 ± 3.0 years, and horses ranged from 400 to 520 kg. Control horses had been used in other experiments and were judged to be appropriate control animals because they did not develop pulmonary neutrophilia or an increase in ΔPpl_max when housed in a stable and fed dusty hay.²⁴

Before the start of the experiments, health of the horses was established on the basis of results of physical, endoscopic, and thoracic radiographic examinations and hematologic and serum biochemical analyses. Several months before each experiment, horses were dewormed by administration of ivermectin^a and vaccinated against tetanus, influenza, eastern equine encephalomyelitis, western equine encephalomyelitis,^b and rabies.^c All experimental procedures were approved by the Animal Care Committee of Cornell University and were in accordance with guidelines established by the National Institutes of Health.

Experimental design—Two experiments were conducted in which both groups of horses (RAO-susceptible and control horses) were evaluated concurrently. Prior to the study, horses were maintained outdoors on pasture for 3 months. To initiate each experiment, horses were housed in a stable (4 × 4-m box stalls) and exposed to dusty hay for 14 days. During this time of natural challenge exposure, horses were provided unlimited access to water and dusty timothy-alfalfa hay. The diet was supplemented with a complete pelleted feed.^d

On day 15 of each experiment, ΔPpl_max values were measured in unsedated horses by use of an esophageal balloon catheter.²⁴ In each horse, the mean ΔPpl_max was calculated for 10 to 15 breaths. Horses were then sedated by administration of xylazine hydrochloride^e (0.6 mg/kg, IV), and a 2.1-m videoendoscope^f was passed via the nasal passages into the distal portion of the trachea. Local anesthesia of the airways was achieved by infusion of 25 mL of a solution of 2% lidocaine hydrochloride.^g The endoscope was then wedged in a sixth- to ninth-generation bronchiole of the right (experiment 1) or left (experiment 2) lung lobe. Bronchoalveolar lavage was performed by instilling 300 mL of warm (37°C) sterile saline (0.9% NaCl) solution through the endoscope biopsy channel, which was followed immediately by aspiration.²⁴ Samples of BALF were stored in siliconized containers that were placed on ice until transported to our laboratory.

Preparation of BALF cells—The BALF cells were processed for cytologic analysis and in vitro incubation. Cytocentrifugation preparations were prepared by mixing 900 μL of BALF with 100 μL of 1% DL-dithiothreitol.^h Slides were stained,ⁱ and differential cell counts were performed by examination of 200 cells. The remaining BALF was filtered through sterile gauze and centrifuged (200 × g for 15 minutes at 10°C) to obtain cell pellets for immunomagnetic separation.¹⁶

Cell pellets were resuspended in 400 μL of PBS solution containing 0.5% bovine serum albumin^j and 2mM EDTA.^k Approximately 5 × 10⁷ cells were incubated (15 minutes at 4°C) with 100 μL of a murine monoclonal IgM antibody^l (reactive against equine granulocytes), washed with buffer, and then incubated (15 minutes at 4°C) with 100 μL of magnetic beads that contained rat anti-mouse IgM.^m Cells were applied to columns,ⁿ and unlabeled (negatively selected) cells were eluted and evaluated cytologically, as described previously. Cell separation was performed at 4°C.

Eluted cells were resuspended in RPMI° with 10% bovine growth serum,^p and 5 × 10⁶ cells/mL were seeded in a 24-well culture plate.^q Each well received 1 of 6 treatments (100 μL of PBS solution; 100 μL of LPS solution^r [final concentration of LPS, 10 μg/well]; 100 μL of β-glucan^s solution [final concentration of β-glucan, 12.5 pg/well]; and 100 μL of each of 3 hay dust solutions prepared by diluting the stock solution 1:100 [hay dust solution 1], 1:1,000 [hay dust solution 2], or 1:10,000 [hay dust solution 3]).

The hay dust stock solution was prepared as described elsewhere²⁶ by use of the same dusty hay that had been fed to induce a recurrence of the condition in the RAO-susceptible horses during both experiments. Briefly, flakes of hay were agitated onto a clean surface, sieved through a grid (2 × 3 mm), and separated into fine and coarse particles by use of a dual-vortex vacuum cleaner.^t Ten milliliters of sterile saline solution was added to each gram of hay dust, and the solution was filtered through a 100-μL mesh screen. The hay dust solution was then exposed to G-radiation (1.5 megarads over a 12-hour period while maintained on dry ice)^u to prevent bacterial and fungal overgrowth of macrophage-lymphocyte cultures. In a preliminary study, we determined that irradiated hay dust upregulated IL-8 expression in a manner qualitatively similar to that induced with nonirradiated hay dust solution. Fungal particulates ≤ 4 μm in diameter in the stock solution were counted by use of a hemacytometer.²⁶ Endotoxin and β-glucan concentrations were measured at a commercial laboratory.^v Each milliliter of the hay dust stock solution contained 10⁸ fungal particulates, 2 μg of endotoxin, and 12.5 ng of β-glucan. Thus, each well that was treated with hay dust solution 1 contained 10⁵ fungal particulates, 2 ng of endotoxin, and 12.5 pg of β-glucan.

Depending on the number of airway mononuclear cells available after immunomagnetic separation, replicates of the 6 treatments were performed whenever possible and used as backup samples. Cells were incubated for 24 (experiment 1) or 6 (experiment 2) hours at 37°C in an environment of 5% carbon dioxide. Harvested cells were frozen at −80°C until gene expression was measured.

Measurement of gene expression—For each treatment of each horse, the total RNA was extracted,^w genomic DNA was destroyed,^x and cDNA was synthesized.^y When cDNA was prepared for each sample, a negative control sample (ie, one that was lacking reverse transcriptase) also was synthesized for subsequent PCR assays. Gene expression was measured by use of a realtime reverse-transcriptase PCR assay.^z Genes analyzed were IL-17, IL-23 (both the p19 and p40 subunits), IL-8, IL-1β, and CXCL2. Target gene expression was standardized to that for a housekeeping gene (ie, β-actin). Primer and probe sequences used were validated as described elsewhere (Appendix).²⁴ The PCR reaction mixtures had a final volume of 27.5 μL (2.5 μL of cDNA and 25 μL of the master mixture). For each cDNA sample, triplicate reactions were performed on each plate for detection of the target genes. Negative control samples were included on each plate.

The endpoint used in the real-time reverse-transcriptase PCR quantification was the C_T at which the amplicon was detected. The C_T ranged from 0 to 40. Gene expression was reported as the ΔC_T.^aa Two methods were used to calculate ΔC_T. For the first method, ΔC_T was the difference between the C_T of the target gene and C_T of the housekeeping gene β-actin. In general, the smaller the ΔC_T value, the more cDNA (ie, mRNA) contained in a sample. For the second method, ΔC_T was calculated as 2^−(ΔΔCT). The ΔΔC_T for each group of horses was the difference between the ΔC_T for a specific treatment (eg, LPS-treated cells) relative to the ΔC_T for the PBS solution (control) treatment.

Statistical analysis—Differences between the 2 groups of horses were detected by use of Wilcoxon rank sum tests. To detect treatment effects within a group, a Friedman 2-way nonparametric ANOVA was used. A Wilcoxon signed rank test was then used to detect post hoc significance. Because multiple comparisons were made, a Bonferroni correction was applied. This was calculated by dividing A (0.05) by the number of genes that were compared in each objective. For example, when data for IL-8, CXCL2, and IL-1β were compared, then P = 0.05/3 = 0.017. All computations were performed by use of a statistical software program.^bb

Results

Pulmonary function and cellular composition of BALF—After being confined continuously in a stable and exposed to dusty hay for 2 weeks, mean ± SD ΔPpl_max of the control horses was 6 ± 1.0 cm H₂O in experiment 1 and 7 ± 2.1 cm H₂O in experiment 2. In RAO-affected horses, naturalchallenge exposure significantly increased the mean ΔPpl_max to 29 ± 10.0 cm H₂O in experiment 1 (P = 0.004) and 44 ± 19.0 cm H₂O in experiment 2 (P = 0.002). In experiments 1 and 2, pulmonary neutrophilia developed only in RAO-susceptible horses as evidenced by significant increases in mean neutrophil percentages in the BALF (35 ± 13% [P = 0.001] and 41 ± 15% [P = 0.002], respectively). These changes were in contrast to the neutrophil percentages in the BALF of the control horses in experiment 1 and 2 of 11 ± 4% and 11 ± 6%, respectively.

Immunomagnetic removal of neutrophils—In experiment 1, mean ± SD percentage of macrophages and lymphocytes in BALF was 53 ± 13% and 36 ± 13%, respectively, for the control horses and 44 ± 13% and 21 ± 9%, respectively, for the RAO-affected horses. After removal of the neutrophils, the percentage of macrophages and lymphocytes recovered was 49 ± 19% and 51 ± 19%, respectively, for the control horses and 46 ± 23% and 54 ± 23%, respectively, for the RAO-affected horses. In experiment 2, mean percentage of macrophages and lymphocytes in BALF was 62 ± 12% and 26 ± 12%, respectively, for the control horses and 35 ± 11% and 23 ± 10%, respectively, for the RAO-affected horses. Following removal of the neutrophils, the percentage of macrophages and lymphocytes recovered was 69 ± 9% and 31 ± 9%, respectively, for the control horses and 62 ± 10% and 38 ± 10%, respectively, for the RAO-affected horses. For both experiments, the percentages of mononuclear cells used for the in vitro testing were not significantly different between the 2 groups of horses.

Effect of treatment for 6 and 24 hours on gene expression of pulmonary mononuclear cells—Box-and-whisker plots were created of standardized C_T values for IL-17, IL-23p19, and IL-23p40 after incubation with each of the 6 treatments (Figure 1). After treatment with PBS solution for 6 or 24 hours, expression of these cytokines did not differ between the 2 groups of horses. Relative to the PBS solution (control) treatment, exposure to hay dust solutions increased cytokine expression in cells isolated from both groups of horses for at least 1 time point (ie, 6 or 24 hours) and sometimes even at the weakest dilution of hay dust solution. The fold increase in gene expression for each treatment was calculated, relative to the value for the PBS solution treatment, for each group (Table 1). Exposure to LPS also increased IL-23p19 expression in cells isolated from either group of horses at the 6- or 24-hour time points, but the magnitude of the response did not differ between the 2 groups of horses. Treatment with β-glucan for 6 or 24 hours failed to increase cytokine expression in cells isolated from either group of horses.

Table 1—

Target gene expression of pulmonary mononuclear cells isolated from RAO-affected and control horses after incubation with 3 hay dust solutions, LPS, or β-glucan for 6 or 24 hours.

View Table

View Full Size

Box-and-whisker plots of IL-17 (A and B), IL-23 subunit p19 (C and D), and IL-23 subunit p40 (E and F) in pulmonary mononuclear cells isolated from control horses (white boxes) and RAO-affected horses (gray boxes) after incubation for 6 (A, C, and E) or 24 (B, D, and F) hours with PBS solution (PBS), 3 doses of hay dust solution (100 μL of each solution prepared by diluting the stock hay dust solution 1:100 [HDS-1], 1:1,000 [HDS-2], or 1:10,000 [HDS-3]), LPS, or β-glucan (GLCN). Expression of each target gene was standardized to the expression of β-actin. Boxes represent the 25th to 75th percentiles, the horizontal bar within each box represents the median value, and the whiskers represent the 95th percentiles. The smaller the ΔC_T value, the greater the amount of cDNA (ie, mRNA) in the sample. Notice that scales differ among the 3 cytokines. Expression of IL-17, IL-23p19, and IL-23p40 did not differ between the 2 groups of horses for either time point for any treatment. Treatment with HDS-1 increased expression of all 3 genes in both groups of horses at both time points, except for IL-17 in control horses after treatment for 6 hours. Treatment with GLCN for 6 or 24 hours failed to increase the gene expression of the 3 cytokines in either group of horses. *Within the RAO-affected horses, value differs significantly (P = 0.017) from the value for the PBS treatment. †Within the control horses, value differs significantly (P = 0.017) from the value for the PBS treatment.

Citation: American Journal of Veterinary Research 68, 12; 10.2460/ajvr.68.12.1361

• Download Figure
• Download figure as PowerPoint slide

Box-and-whisker plots were created of standardized C_T values for IL-8, IL-1β, and CXCL2 after incubation with each of the 6 treatments (Figure 2). After the PBS solution treatment for 6 or 24 hours, the expression of IL-8, CXCL2, or IL-1β did not differ between the 2 groups of horses. Relative to the PBS solution treatment, exposure to hay dust and LPS significantly increased chemokine expression in cells isolated from both groups of horses for at least 1 time point (Table 1). The only exception was that cells from control horses did not have an increase in CXCL2 expression. Treatment with β-glucan failed to increase chemokine expression, relative to expression for treatment with PBS solution, in cells isolated from either group of horses. For all treatments, there was little evidence that an increase in chemokine expression was significantly greater in cells isolated from the RAO-affected horses, compared with responses for cells isolated from control horses.

View Full Size

Box-and-whisker plots of expression of IL-8 (A and B), IL-1β (C and D), and CXCL2 (E and F) in pulmonary mononuclear cells isolated from control horses (white boxes) and RAO-affected horses (gray boxes) after incubation for 6 (A, C, and E) or 24 (B, D, and F) hours with PBS solution (PBS), 3 doses of hay dust solutions, LPS, or GLCN. Expression of each target gene (ΔC_T) was standardized to the expression of Bactin. Notice that scales vary among cytokines and between some time points within each cytokine. Expression of IL-8, CXCL2, or IL-1β did not differ between the 2 groups of horses at either time point after any treatment, except that CXCL2 expression was greater for RAO-affected horses after treatment with HDS-2 for 6 hours. Treatment with HDS-1 increased expression of IL-8 and IL-1β in both groups of horses at both time points but only increased CXCL2 expression at 6 hours in RAO-affected horses. Treatment with GLCN for 6 or 24 hours failed to increase gene expression of the 3 chemokines in either group of horses. See Figure 1 for remainder of key.

Citation: American Journal of Veterinary Research 68, 12; 10.2460/ajvr.68.12.1361

• Download Figure
• Download figure as PowerPoint slide

Discussion

The primary goal of the study reported here was to investigate the role of pulmonary mononuclear cells in the propagation of pulmonary neutrophilia in RAO-affected horses during the chronic stages of the disease. Our experimental design allowed us to examine inherent and ligand-induced cytokine expression in pulmonary macrophages and lymphocytes isolated from horses that naturally developed the disorder. Two time points (6 and 24 hours) were examined because time-dependent changes in the expression of TNF-α and IL-8 in alveolar macrophages isolated from RAOaffected horses have been reported.²⁷ For example, it was found that when RAO-susceptible and control horses were nebulized 1 time with a hay dust solution, early (6 hours) expression of TNF-α and IL-8 in alveolar macrophages isolated from diseased horses was 3- and 4-fold greater, respectively, than that of control horses.²⁷ However, 24 hours later macrophage cytokine expression in diseased horses was no different from that of the control horses. Although our experimental design permitted us to evaluate time-dependent changes in pulmonary mononuclear responses, our study also differed from that of the other study²⁷ in 2 important aspects. First, in vitro responses in our study were examined in cocultures of alveolar macrophages and lymphocytes isolated from chronically affected horses that had developed RAO naturally. Second, ligand-induced responses were evaluated after pulmonary mononuclear cells were exposed continuously to the ligand for 6 and 24 hours, which more closely simulated the in vivo situation.

The first objective of the study reported here was to determine whether the magnitude of IL-17 and IL-23 expression in pulmonary mononuclear cells isolated from diseased horses exceeded that of control horses. Analysis of the data revealed that following treatment with PBS solution for 6 or 24 hours, expression of IL-17 or IL-23 (subunits p19 or p40) in cells isolated from RAO-affected horses was no greater than that of the control horses. Analysis of these findings suggested that the preexisting inflammatory milieu of the airways of horses chronically affected with RAO does not upregulate IL-17 and IL-23 expression ex vivo. Furthermore, although the highest concentration of the hay dust solution increased IL-17 and IL-23 expression in airway cells isolated from RAO-affected horses, the magnitude of the cytokine responses was no different between the 2 groups of horses at either time point. Analysis of these results indicated that airway mononuclear cells from RAO-affected horses do not have an exaggerated response (ie, airway neutrophilia) to the hay dust solutions. These results also permitted us to conclude that the increased expression of IL-17 found in BALF cells isolated from chronically affected horses^16,28 more likely reflects neutrophil expression of IL-17. Collectively, these data suggested that rapid resolution of airway inflammation may be dependent on elimination of organic dust exposure (which may upregulate chemokine production in epithelial cells through pathways separate from activation of IL-17 receptors) as well as removal of extravasated granulocytes (ie, the source of chemokines such as IL-17, IL-8, and CXCL2).²⁹

Somewhat surprising to us was the finding that regardless of the group of horses or the duration of exposure, LPS (a component of hay dust) failed to increase the expression of IL-17. In mice, inhalation or intraperitoneal challenge exposure with LPS increases lymphocyte mRNA concentrations of IL-17 in a dose-dependent manner.^30,31 Lipopolysaccharide exposure in vitro promotes IL-17 release from T cells, provided that these cells are in cocultures with macrophages, a source of IL-23.³² In the study reported here, LPS induced early (6 hours) and late (24 hours) expression of IL-23p40 in cells isolated from both groups of horses. However, its effect on IL-23p19 expression was inconsistent. It is possible that the lack of effect of LPS on IL-17 expression reflected inadequate production of the p19 subunit needed for the functional IL-23 heterodimer.

We are unaware of any other studies in which investigators have examined the in vitro effects of β-glucans on IL-17 and IL-23 expression in equine pulmonary mononuclear cells. The β-glucans are highly conserved structural components of fungal cell walls. They consist of a backbone of glucose residues arranged in β-1,3-D glucopyranosyl configuration to which β-1,6-D glucopyranosyl side chains of various lengths and numbers are attached.³³ Because β-glucans are found in hay dust,²⁶ we examined the effect of this component alone on cytokine expression in pulmonary mononuclear cells. Thus, pulmonary mononuclear cells were exposed to a β-glucan concentration (12.5 pg/mL) equivalent to that found in the hay dust solution 1. The lack of effect of β-glucan on IL-17 and IL-23 expression suggested that this compound alone does not contribute to upregulation of these cytokines that is caused by treatment hay dust. It is possible that higher concentrations of β-glucan would have upregulated gene expression of IL-17 or IL-23, or both, but our experimental design did not allow us to determine whether there was a dose-dependent effect.

The second objective of the study reported here was to determine whether expression of neutrophil chemoattractants (ie, IL-8 or CXCL2) was enhanced in alveolar macrophages isolated from horses chronically affected with RAO, compared with responses for control horses. Because IL-1β promotes IL-8 and CXCL2 transcription in a number of cells and tissues, and thus functions indirectly as a chemoattractant,^34,35 we also examined its gene expression.

Transcription of IL-8 and CXCL2 in alveolar macrophages and monocytes is upregulated by NFκB and activator protein-1. In turn, these transcription factors are induced by stimuli such as endotoxin, reactive oxygen species, TNF-α, and IL-1,³⁶ all of which have been implicated in the pathophysiologic processes of RAO.^1,23,37 In addition to being a neutrophil chemoattractant and activator, IL-8 also induces exocytotic release of enzymes and proteins from intracellular storage organelles, upregulates expression of integrins, and activates 5-lipoxygenase with the formation of leukotriene B4.³⁸ Similarly, CXCL2 (formerly known as GRO-β and macrophage-inflammatory protein 2α) is chemotactic for neutrophils. In addition, CXCL2 also induces granulocyte degranulation but does not enhance oxidative metabolism. In endotoxin-stimulated macrophages in vitro, the expression of CXCL2 is less than that of IL-8 (CXCL8).^39,40

In RAO-affected horses, increases in IL-8 protein and mRNA concentrations have been found in BALF or BALF cells, respectively.^16,22-24 Similarly, increases in IL-1β protein or mRNA concentrations have also been reported^23,29 in BALF and BALF cells, respectively, isolated from RAO-affected horses. However, because equine neutrophils are a source of IL-1β, IL-8, and CXCL2,⁴¹ the increased concentrations of these chemokines in airway secretions probably reflect the contribution of granulocytes. Indeed, analysis of our data (and those of others) allows us to reach this conclusion. First, at either the 6- or 24-hour time point, the mRNA concentrations of IL-8, IL-1β, and CXCL2 were not inherently increased in pulmonary mononuclear cells isolated from the RAO-affected horses (PBS solution treatment). Second, expression of IL-8 or IL-1β augmented by hay dust solution was not greater in cells from RAO-affected horses. Although hay dust solution significantly upregulated early CXCL2 expression in cells from RAO-affected horses, relative to results for the PBS treatment, this increase was not significantly greater than that found for the control horses. Third, although our experiments examined in vitro responses in control and RAO-affected horses by use of comparable numbers of airway cells, it is unlikely that mononuclear cell numbers are skewed in favor of an enhanced chemokine milieu in vivo in RAO-affected horses. For example, investigators in 1 study⁴² reported that the mean numbers of alveolar macrophages and lymphocytes recovered from the pulmonary epithelial lining fluid of chronically affected hoses with RAO (6.2 × 10³ cells/μL and 6.5 × 10³ cells/μL, respectively) were not greater than those of healthy horses (13.9 × 10³ cells/μL and 6.6 × 10³ cells/μL, respectively).

Similar to effects on IL-17 and IL-23 expression, the effects of LPS on IL-8, IL-1β, and CXCL2 expression (relative to the PBS solution treatment) were also variable for the 2 groups of horses. However, at either time point, the chemokine responses of cells from RAOaffected horses exposed to LPS were not significantly greater than those of the control horses. These data differ from those of a study²⁷ in which investigators found that early (6 hours) but not late (24 hours) expression of IL-8 and IL-1β in pulmonary macrophages isolated from horses exposed via nebulization once with LPS was significantly greater in cells from RAO-affected horses; expression of CXCL2 was not evaluated in that study. As mentioned previously, differences in experimental design make it difficult to make direct comparisons of results for the 2 studies. However, in another study,⁴³ investigators found that when pulmonary macrophages isolated from healthy horses were treated for 24 hours with LPS (10 μg/mL), only modest increases in TNF-α and IL-1 protein concentrations were detected.

Finally, with the exception of early expression of IL-1β in cells from control horses, β-glucan treatment failed to increase mRNA concentrations of IL-8, IL-1β, or CXCL2, relative to results for the PBS treatment. These findings were opposite to those we had hypothesized, given that β-glucans are integral components of hay dust. However, depending on the experimental system used in a study, effects of β-glucans on inflammatory gene expression are quite diverse. For example, in weanling pigs, dietary supplementation with β-glucans derived from Saccharomyces cerevisiae enhanced porcine humoral responses to ovalbumin vaccination and attenuated LPS-induced cytokine production in lymphocytes.⁴⁴ In contrast, parenteral administration of a soluble derivative of S cerevisiae in horses failed to prevent in vitro LPS-induced expression of TNF-α, IL-1β, or IL-10 in mononuclear cells.⁴⁵ Also in contrast, in vitro treatment with β-glucans enhanced proinflammatory gene expression in murine macrophages by activating dectin-1 receptors and inducing nuclear translocation of NFκB.³³ In the study reported here, the lack of an effect of β-glucans on chemokine expression in equine pulmonary mononuclear cells deserves additional study to determine whether a dose-dependent effect of β-glucans exists.

Analysis of results of the study reported here revealed that pulmonary mononuclear cells isolated from RAO-affected horses that have been stabled and exposed to dusty hay for 14 days were not more responsive or reactive to hay dust or its components than were those isolated from healthy horses. Our data support the hypothesis that the inflammatory influx of neutrophils in chronically affected horses is maintained predominantly by chemokines released from extravasated granulocytes.²⁹ Therefore, we hypothesize that rapid resolution of pulmonary inflammation is dependent on removing affected horses from the offending environment and removing granulocytes and secreted chemokines from the airways.