Horses (n=30) presented as patients for respiratory examination to the Department for Horses, Faculty of Veterinary Medicine, Leipzig University, Germany between March and December 2021 were included in the study after informed consent of their owners. Leftover samples from routine diagnostic procedures were used for research purposes. Additionally, physically healthy control horses (n=8) underwent the same procedures approved under the animal experiment permission number TVV22/20, file number 25-5131/490/23 (Landesdirektion Sachsen, Germany).

The horses’ clinical history was assessed using a questionnaire to score the Horse Owner Assessed Respiratory Signs Index (HOARSI) (19) and a summary HOARSI (range 1–4) was deducted. Information about any treatments in the last month before the examination was recorded. All horses underwent physical examination at rest and after re-breathing, and a clinical score adapted from Ivester et al. (3, 20) was recorded, summarizing nasal discharge, cough, respiratory rate, nasal flare, abdominal lift, and findings on lung auscultation ( Supplementary Table 1 ). Arterial blood samples were obtained from the right common carotid artery and analyzed within minutes after collection on an ABL90 FLEX PLUS blood gas analyzer (Radiometer Medical ApS, Brønshøj, Denmark).

Endoscopy of the upper and lower airways was performed after sedation with detomidine and butorphanol with a flexible video endoscope (Storz G28-250, 10.4 mm diameter). A mucus score (21) was assessed by a veterinarian specialized in equine internal medicine. For BALF collection the larynx and the targeted bronchus were anesthetized by topical application of 0.5% lidocaine hydrochloride (bela-pharm, Vechta, Germany) in saline (B. Braun, Melsungen, Germany) via the endoscope. The endoscope was advanced into one bronchus until wedged and a lavage was performed using one bolus of saline (60 ml/100 kg body weight, B Braun) with immediate re-aspiration into syringes. The resulting bronchoalveolar lavage fluid (BALF) was placed on ice and processed within one hour after sampling. Heparinized venous blood was acquired by jugular venipuncture (sodium heparin vacuum tubes, Becton Dickinson GmbH, Heidelberg, Germany) and processed within three hours after sampling.

An aliquot of BALF was used to prepare cytospins (5 minutes centrifugation 113 x g, Shandon Cytospin centrifuge, Thermo, Waltham, MA USA); for cytology. These were stained with DiffQuick (RAL Diagnostics, Martillac, France) or toluidine blue (0.25 mg/ml Toluidine Blue O, Sigma-Aldrich, Merck KGaA, Darmstadt Germany). A minimum of 500 cells per specimen was differentiated and the percentages of lymphocytes, macrophages, neutrophil granulocytes (polymorphonuclear cells, PMN), eosinophil granulocytes (all in DiffQuick stained slides), and metachromatic mast cells (MC, in toluidine blue stained slides) were recorded.

The horses were retrospectively classified into groups by established criteria (1, 3) according to their history (Horse Owner Respiratory Signs Index, HOARSI (19)), clinical presentation ( Supplementary Table 1 ) (20), endoscopic mucus score (21), BAL cytology, and treatment within four weeks prior to the sampling ( Table 1 and Figure 1 ). Horses with altered complete blood count, fever or other signs of infectious or systemic inflammatory disease were excluded. Horses without a history of asthma symptoms (HOARSI ≦ 2), without findings of lower respiratory disease (dyspnea, cough, mucus nasal discharge, abnormal lung sounds; mucus score ≦2), and with normal BAL cytology (≦ 8% PMN, ≦ 0.5% eosinophils, ≦ 5% MC) were considered respiratory healthy (1, 3). Horses without clinical signs, but abnormal mucus score or BAL cytology were categorized to have mild EA, those with clinical signs but without dyspnea at rest were categorized to have moderate EA. The horses with mild or moderate EA were combined in one group (mEA) according to the consensus statement on inflammatory airway disease (1). Horses with a history of asthma symptoms (HOARSI 3 or 4), clinical findings including dyspnea at rest, a mucus score ≧2, and abnormal BAL cytology with >10% PMN were categorized as horses with severe EA (sEA). Horses in the sEA group who received bronchodilators or corticosteroids within four weeks prior to examination were analyzed as a separate group (sEA tr). Severely asthmatic horses were considered untreated (sEA) if they did not receive any treatment, underwent inhalation of saline, or received mucolytics. Horses with a history of asthma symptoms (HOARSI 3 or 4) but without clinical signs and with mucus scores ≦2 at the time of examination were categorized as EA in remission (EA rem). Horses with BAL samples with indications of blood contamination (macroscopically or microscopically on cytospins) were excluded from the study.

From the same horses used for BAL acquisition, PBMC were isolated from heparinized blood by density gradient centrifugation (Pancoll 1077, Pan-Biotech, Aidenbach) as previously described (13, 15). BALF was passed over a cell strainer (ClearLine100 µm, Kisker Biotech, Steinfurt, Germany) and centrifuged at 500 x g, for 8 min, at 4°C. Supernatant fluid was removed, and the resulting BAL cell pellet was washed in PBS (Roth, Karlsruhe, Germany) with the same centrifugation settings. Viability and counts of PBMC and BAL cells were determined by trypan blue staining and cells were re-suspended in supplemented medium (DMEM low glucose (1 g/l), 1% (v/v) non-essential amino acids (NEA), 2 mM L-glutamine, 50 μg/ml gentamicin, 100 U/ml penicillin, 100 μg/ml streptomycin, 50 μM 2-mercaptoethanol, all from Sigma-Aldrich; 10% heat-inactivated FCS, Lonza, Cologne, Germany) at a concentration of 1x10⁷ live cells/ml. Aliquots of the processed cells ex vivo were live/dead stained with fixable viability dye eFluor™506 (eBioscience™, ThermoFisher Scientific) and fixed with 2% Paraformaldehyde in PBS (Roth) or additionally used for in vitro stimulation if sufficient BAL cells (2x10⁷ or more live cells) were available from an individual horse. The viability was similar between the cells from healthy and asthmatic horses (BAL cells median 95%, range 80–97%; PBMC median 99%, range 97–100%).

BAL cells from 26 horses (5 healthy, 7 mEA, 6 sEA, 4 sEA tr, 4 EA rem) were used for in vitro stimulation in comparison to stimulated PBMC from the same horses. In vitro stimulation was performed in sterile polystyrene tubes (Sigma) that were pre-incubated with supplemented medium for 3 h at 37°C, aiming to minimize cell adherence in vitro. PBMC or BAL cells were incubated in supplemented medium with Brefeldin A (10 µg/ml; Sigma-Aldrich) at 37 °C, 5% CO₂ in a humidified atmosphere for 4 h. For stimulation a combination of Phorbol-12-myristat-13-acetat (PMA; 25 ng/ml; Sigma-Aldrich) and ionomycin (1 µM; Sigma-Aldrich) was added to the culture medium for the entire incubation time. After 4 h of culture cells were harvested by centrifugation, washed, live/dead stained and fixed as described for the cells used for ex vivo phenotyping. Median viability was 95% (range 64-99%) after in vitro incubation and stimulation.

The fixed cells were stored at 4°C for a maximum of three days, stained and analyzed as follows: Surface staining was performed in 3% FCS (Lonza), 0.1% NaN₃ in PBS (Roth), and followed by intracellular staining for in vitro cultured cells (CD154, cytokines) in 0.5% Saponin, 3% FCS, and 0.1% NaN₃ in PBS with different antibody combinations ( Table 2 ). 200,000 events per sample were recorded with a BD LSR Fortessa™ Cell Analyzer (BD Biosciences, Ashland, OR, USA) equipped with FACS Diva ™ 6.2 software (BD). Data were analyzed using FlowJo™ v10.8 software (FlowJo, LLC, BD).

After doublet exclusion by forward scatter (FSC) height against FSC area characteristics (singlets), live cells were selected by exclusion of viability dye positive cells. Then, lymphocytes were gated by FSC and side scatter (SSC) characteristics ( Figure 2 ) and lymphocyte and non-lymphocyte (NL) populations were analyzed separately. For ex vivo analysis, lymphocyte proportions of CD4⁺, CD8⁺, Ig⁺, and CD79a⁺ cells were quantified. Percentages of CD14⁺, CD16⁺, DH24A⁺, MHCII^high, and IgE⁺ cells were analyzed in the NL population. For analysis of in vitro cultured cells, proportions of CD154⁺, TNF-α⁺, IFN-γ⁺, IL-4⁺, IL-10⁺, and IL-17A⁺ lymphocytes as well as CD4⁺cytokine⁺, and CD8⁺cytokine⁺ double-positive cells were quantified and medium expression percentages were subtracted from those after PMA and ionomycin stimulation resulting in (corrected) net percentages.

Statistical analysis was carried out using GraphPad PRISM v9 software (GraphPad Software, La Jolla, CA, USA). Non-parametric Kruskal-Wallis test was used for all analyses, as data did not show a Gaussian distribution, followed by Dunn’s Test for multiple comparisons. For the data after in vitro stimulation only groups with a minimum of five horses were analyzed statistically (healthy, mEA, sEA). Spearman’s rank correlation was used for assessment of rank correlation between results from cytospin microscopy and FC. Statistical significance was annotated as *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

Cells from bronchoalveolar lavage fluid (BAL cells) and PBMC from 38 horses were isolated for detailed analysis. The horses were retrospectively classified into five groups, based on their history (HOARSI (19) Figure 1A ), previous treatment with steroids or bronchodilators (sEA tr), clinical presentation (score sum, Figure 1B ), endoscopic mucus score (21) ( Figure 1D ), and BAL cytology ( Figures 1F–J ). Consequently, HOARSI and clinical score were higher in horses with sEA compared to healthy horses and horses with mEA ( Figures 1A, B ). Alveolar to arterial oxygen pressure difference (AadO₂) was increased in the sEA tr group ( Figure 1C ). Mucus scores were also higher in horses with sEA and sEA tr compared to healthy horses ( Figure 1D ). The nucleated cell counts per ml were similar in BALF from all groups ( Figure 1E ). BAL cytology, microscopically assessed on cytospins revealed similar frequencies of lymphocytes ( Figure 1F ), eosinophils ( Figure 1I ) and mast cells (MC, Figure 1J ) in all groups. According to the classification of the horses which included BALF cytology, the percentage of PMN was increased in BAL cells from treated and untreated horses with sEA compared to the other groups ( Figure 1G ). The percentages of macrophages (Mph) among all cells were decreased in the sEA tr group compared to healthy horses and horses with mEA ( Figure 1H ). The increase of PMN percentages in BAL from horses with sEA was also reflected in PMN counts per ml BALF ( Supplementary Figures 1B, G ), while Mph counts per ml BALF ( Supplementary Figures 1C, H, I ) were not significantly different between the groups.

Flow cytometry (FC) was used to further characterize BAL cells in comparison to peripheral blood mononuclear cells (PBMC). The percentages of lymphocytes (determined by forward and side scatter, FSC-SSC, characteristics, Figures 2A, E, F ; detailed gating strategies in Supplementary Figure 2 ), PMN (DH24A⁺, Figures 2B, G, H ), Mph or monocytes (CD14⁺, Figures 2C, I, J ), IgE⁺CD14^- cells, containing MC (BAL) or basophils, and B cells (PBMC) ( Figures 2D, K, L ), and IgE⁺CD14⁺ IgE-binding Mph/monocytes ( Figures 2D, M, N ) were analyzed.

In BAL cells from horses of all groups the lymphocyte frequencies were similar (overall median 46%, Figure 2E ), but DH24A⁺ PMN percentages were higher in BAL cells from horses with sEA (median 21%) compared to healthy horses and horses with mEA (both groups median 2%, Figure 2G ) consistent with results obtained by microscopic cytospin analysis. Flow cytometric PMN identification as DH24A⁺ non-lymphocytes was confirmed by analysis of the MHCII^-DH24A⁺ non-lymphocyte population, which yielded similar quantitative results ( Supplementary Figure 3 ). In further agreement with cytospin analyses, the frequencies of CD14⁺ Mph were lower in BAL cells from horses with sEA (median 18%) compared to healthy horses or the mEA group (both median 35%, Figure 2I ). Mph percentages in the sEA tr (median 13%) or EA rem (median 19%) groups were similar to sEA, but the differences to Mph percentages in BAL from healthy or mEA horses were not statistically significant ( Figure 2I ). BAL cells’ IgE⁺CD14^- ( Figure 2K ) frequencies were again similar between all groups, but IgE-binding CD14⁺ cells frequencies were increased in BAL cells from horses with mEA (median 0.76%) compared to those with sEA (median 0.25%, Figure 2M ). Calculation of the cell counts per ml BALF confirmed the significant median increase of DH24A⁺ cells in the sEA tr group, but did not reach statistical significance for horses with sEA compared to other groups ( Supplementary Figure 1G ). The calculated counts of other leukocyte subsets per ml BALF were not significantly different between groups ( Supplementary Figures 1F, H–J ). PBMC frequencies of lymphocytes ( Figure 2F ), retained DH24A⁺ PMN ( Figure 2H ), CD14⁺ monocytes ( Figure 2J ), IgE⁺CD14^- cells ( Figure 2L ), and IgE⁺CD14⁺ cells ( Figure 2N ) did not differ between the groups.

Results from FC and cytospin analysis correlated for lymphocyte ( Figure 2O ), PMN ( Figure 2P ), and Mph ( Figure 2Q ) frequencies. However, frequencies of IgE⁺CD14^- putative MC determined by FC and metachromatic cells in the cytospin analysis did not correlate ( Figure 2R ). For further analysis, we sorted IgE⁺CD14^- cells from the DH24A-negative non-lymphocyte fraction by FACS and analyzed their morphology microscopically on cytospins after toluidine blue staining (n=3, healthy, Supplementary Figure 4 ). Microscopic cytospin analysis confirmed major enrichment of metachromatic cells in the sorted IgE⁺CD14^- fraction but revealed variable degrees of metachromatic granules per cell ( Supplementary Figure 4I ). Re-analysis with additional staining confirmed the IgE⁺CD14^- BAL cells to be MHCII^lo, consistent with MC ( Supplementary Figure 4G ). Discrepancy of the results obtained by FC and cytospin analysis could result from underestimating the frequency of MC during microscopic assessment of cytospins due to weak metachromatic staining of paucigranulous cells still identified by FC based on their IgE⁺CD14^- phenotype.

Taken together, microscopic cytology and FC analyses identify increased percentage of PMN, both in percentual proportions and cell counts per ml BALF, as the major hallmark of sEA, whereas the percentage of Mph was decreased in BAL cells during sEA, but Mph cell counts per ml BALF were similar between all groups.

The phenotypes of BAL Mph and PBMC monocytes were further characterized according to their expression of the classical monocyte/Mph marker CD14 (lipopolysaccharide co-receptor (17, 18),), and CD16 (FcγRIIIa), which has been described and used as a non-classical monocyte marker (12, 17). Staining of mannose receptor CD206 and scavenger receptor CD163, other markers of equine BAL Mph (11, 30), was performed on some BAL samples and the majority of the putative Mph (70-90% of DH24A^- NL) were positive for both, CD14 and CD206 or CD14 and CD163 (data not shown). Due to less interference with Mph autofluorescence in FITC and PE (compare (11)) with the available conjugates for CD14 antibodies compared to CD206 or CD163, and the indication of regulation of the latter two surface molecules (11), we selected CD14 as the main marker for Mph identification and quantification in our analyses (compare Figures 2C, E ). To correct for the increased PMN frequencies in BAL from horses with sEA, Mph were gated from non-lymphocytes after PMN exclusion ( Figure 3A ) and quantified as percentage of DH24A-negative non-lymphocytes (DH24A^- NL, detailed gating strategy in Supplementary Figure 5 ). The DH24A^- NL population contains mainly Mph, but may include small fractions of other cells, such as MC and eosinophils, which are neither excluded by lymphocyte nor by DH24A (PMN) gating.

The majority of BAL Mph co-expressed CD14 and CD16 ( Figures 3C, H ), while the main population of PBMC monocytes were CD14⁺CD16^- cells ( Figures 3C, K ). The frequencies of CD14⁺ Mph from horses of different groups, were not significantly different in DH24A^- NL in BAL ( Figure 3D ) or PBMC ( Figure 3E ) in contrast to the frequencies obtained relative to all singlet live cells ( Figures 2I, J , without correction for the PMN). However, the frequencies of CD16⁺ cells ( Figure 3B ) were still reduced in BAL DH24A^- NL from horses with sEA (median 64%) compared to healthy horses (median 84%) and to the mEA group (median 87%, Figure 3F ). Similarly, frequencies of CD14^-CD16⁺ cells were lower in the sEA tr group compared to the mEA group among DH24A^- NL BAL cells ( Figure 3L ). Nevertheless, the percentages of CD14⁺CD16^- cells ( Figure 3H ) and the dominating CD14⁺CD16⁺ ( Figure 3J ) cells were similar in BAL samples of all groups. The horses’ PBMC had similar frequencies of CD14⁺ and CD16⁺ monocytes in samples of all groups ( Figures 3E, G, I, K, M ). In summary, CD16⁺ Mph may be mildly reduced within BAL Mph from horses with sEA even if the main population of BAL Mph (CD14⁺CD16⁺) appears unaffected by the disease.

Additionally, MHCII expression (marking professional antigen-presenting cells (17)) on CD14⁺ Mph was quantified to analyze potential MHCII regulation in healthy or asthmatic airway leukocytes ( Figure 4 ). MHCII expression was similar on CD14⁺ BAL Mph (MHCII median fluorescence intensities, Figures 4D, E ) or PBMC monocytes ( Figure 4F ) of all groups ranging from MHCII^lo to MHCII^hi CD14⁺ cells ( Figures 4B, C ). Quantification of a small, distinct population of CD14⁺ MHCII^high cells ( Figure 4G ) also resulted in similar frequencies in BAL or PBMC samples from horses of all groups (overall median 5% of CD14⁺ NL, Figures 4H, I ).

In summary, we newly identify CD14⁺CD16⁺ cells as the largest sub-population among Mph in BAL which is unaffected by EA, and demonstrate reduction of CD14^-CD16⁺ BAL cells in sEA, whereas PBMC monocytes are unaffected by EA and are mostly CD14⁺CD16^-, as previously established in horses (12, 17).

The lymphocyte sub-populations of equine BAL cells and PBMC were analyzed by FC ( Figure 5A ). The staining with CD8 was not valid for samples from one horse (sEA group), which was thus excluded from the lymphocyte subset analysis. The proportions of CD4⁺ T helper cells ( Figures 5B, C ), CD8⁺ cytotoxic T cells ( Figure 5D, E ), CD4⁺CD8⁺ T cells ( Figures 5F, G ), and Ig⁺CD4^-CD8^- cells ( Figures 5H, I ) among lymphocytes were similar in BAL cells or PBMC between the different groups of horses. However, CD8⁺ cytotoxic T cells dominated the lymphocytes in BAL cells (overall median 55% of lymphocytes, Figure 5D ) independent of asthmatic disease, in contrast to PBMC samples containing mainly CD4⁺ T helper cells (median 46% of lymphocytes, Figure 1C ). Ig⁺CD4^-CD8^- cells were very rare in BAL lymphocytes (median 2%) in contrast to PBMC lymphocytes (median 18%, Figure 5H, I ). Staining of CD79a was applied to further characterize the Ig⁺CD4^-CD8^- lymphocytes ( Supplementary Figure 6 ). In PBMC, Ig and CD79a were co-expressed on most CD4^-CD8^- lymphocytes ( Supplementary Figure 6G ), confirming them as B cells. However, in BAL cells only few Ig⁺CD4^-CD8^- lymphocytes expressed CD79a ( Supplementary Figure 6D ) and the percentages of Ig⁺CD4^-CD8^- and CD79a⁺CD4^-CD8^- lymphocytes did not correspond in individual horses ( Supplementary Figure 6E ). Thus, the small fraction of Ig⁺CD4^-CD8^- lymphocytes found in BAL can only partially be considered B cells and the roughly 20% CD4^-CD8^- lymphocytes in BAL are different from typical B or T cells. Those cells could represent non-conventional T cells, natural killer cells, or innate lymphoid cells, which could not be differentiated further with the markers available here.

Beyond demonstrating similar frequencies of T helper cells, cytotoxic T cells and B cells in samples from all groups ex vivo, we aimed to functionally characterize these lymphocytes by analysis of the T helper cell activation marker CD154 (CD40L) (25, 31) and expression of five different cytokines after in vitro stimulation with PMA and ionomycin ( Figures 6A, B ). In vitro stimulation was performed for samples from 26 horses (n=4–7 per group) with sufficient yield of BAL cells, and only groups with five or more values (healthy, mEA, sEA) were compared statistically. Net percent of cytokine-expressing lymphocytes of all live lymphocytes were analyzed after medium correction ( Figures 6A, B ).

The frequencies of CD154-expressing lymphocytes were similar in BAL cells of horses from all groups, except for a trend towards higher frequencies in horses with mEA (p=0.09) compared to healthy horses ( Figure 6C ) and elevated median percentages in the EA rem group (not statistically analyzed, Figure 6C ). The proportions of lymphocytes expressing TNF-α ( Figure 6E ), IFN-γ ( Figure 6G ), IL-4 ( Figure 6I ), or IL-10 ( Figure 6K ) were similar in BAL cells from horses of all groups. However, IL-17A⁺ lymphocytes were present at higher frequencies among BAL cells from horses with sEA compared to the other groups, although this trend did not reach statistical significance (p=0.11 sEA median 10.7% vs. healthy median 4.8%, Figure 6M ). Similar percentages of CD154 and cytokine-expressing lymphocytes were detected in PBMC lymphocytes of horses from all groups ( Figures 6D, F, H, J, L, N ).

Interestingly, TNF-α was similarly induced in BAL and PBMC lymphocytes (overall medians, Figures 6E, F ), but higher frequencies of IFN-γ⁺ (2-fold, Figure 6G ), and IL-17A⁺ cells (6-fold, Figure 6M ) were detected in BAL lymphocytes compared to PBMC lymphocytes ( Figures 6F, H, N ), whereas lower CD154⁺ (2-fold, Figure 6C ), IL 4⁺ (13-fold, Figure 6I ), and IL-10⁺ (4-fold, Figure 6K ) BAL lymphocyte frequencies were detected compared to PBMC lymphocytes ( Figures 6D, J, L ).

We further characterized the T cells as potent activated cells and cytokine producers in the lymphocyte population through staining of CD4 and CD8 combined with staining of intracellular cytokines and CD154 ( Figure 7 ). CD4⁺ and CD8⁺ T cells expressed TNF-α and IFN-γ in BAL and PBMC lymphocytes ( Figures 7A–L ). CD8⁺TNF-α⁺ and CD8⁺IFN-γ⁺ cytotoxic T cells (CTL) were stimulated at 3-fold higher frequencies in BAL cells compared to PBMC ( Figures 7A–F ), matching the results from ex vivo analysis with higher frequencies of CD8⁺ T cells in BAL cells compared to PBMC ( Figures 5D, E ). The percentages of CD8⁺TNF-α⁺ and CD8⁺IFN-γ⁺ CTL frequencies were similar in all groups’ BAL cells or PBMC ( Figures 7B, C, E, F ).

Frequencies of putative T helper (Th) 1 cells, defined as CD4⁺TNF-α⁺ ( Figure 7G ) and CD4⁺IFN-γ⁺ ( Figure 7J ) cells, were higher in BAL of horses with sEA compared to healthy horses’ BAL lymphocytes ( Figures 7H, K ), but this only tended toward statistical significance for CD4⁺IFN-γ⁺ percentages (p=0.058, Figure 7K ). Frequencies of CD4⁺CD154⁺ (activated Th cells, Figures 7M-O ), CD4⁺IL-4⁺ (Th2, Figures 7P–R ), and CD4⁺IL-10⁺ (T_reg, Figures 7S–U ) frequencies were lower in BAL cells than PBMC but were similar in BAL cells of all groups. The trend of increased CD154⁺ lymphocytes in mEA BAL ( Figure 6C ) was not resembled in BAL Th CD4⁺CD154⁺ cells (ns, Figure 7N ). In agreement with all IL-17A⁺ lymphocytes ( Figure 6M ) however, the frequencies of CD4⁺IL-17A⁺ lymphocytes (Th17, Figure 7V ) were higher in BAL cells than in PBMC and were significantly increased in BAL lymphocytes from horses with sEA compared to healthy horses (p=0.011, Figures 7W, X ).

In PBMC, the median frequencies of activated Th cells expressing cytokines or CD154 after stimulation were higher in asthmatic - than in healthy horses’ lymphocytes (sEA>mEA>healthy, Figures 7I, L, O, R, U, X ) in contrast to all lymphocytes’ cytokine expression without such clear group differences ( Figure 6 ). Nevertheless, the frequencies of activated Th cells were not statistically different between horses with mEA and sEA, but significantly higher in PBMC lymphocytes from horses with sEA than healthy horses’ PBMC lymphocytes for CD4⁺TNF-α⁺ ( Figure 7I ), CD4⁺IFN-γ⁺ ( Figure 7L ), CD4⁺CD154⁺ ( Figure 7O ), CD4⁺IL-4⁺ ( Figure 7R ), and CD4⁺IL-10⁺ ( Figure 7U ) cells. Proportions of activated Th cells were significantly higher in PBMC of horses with mEA compared with healthy horses for CD4⁺IL-4⁺ ( Figure 7R ) and CD4⁺IL-10⁺ ( Figure 7U ) cells and had a similar trend for CD4⁺TNF-α⁺ ( Figure 7I ) and CD4⁺CD154⁺ ( Figure 7O ) lymphocytes. CD4⁺IL-17A⁺ cells showed a similar expression distribution in PBMC like the other Th cells, but the increased medians of CD4⁺IL-17A⁺ in asthmatic horses’ PBMC did not reach a statistically significant difference compared to healthy horses’ PBMC ( Figure 7X ).

Co-expression of CD154 and the Th subset differentiating cytokines IFN-γ, IL-4, and IL17A was usually observed in BAL cells and PBMC, but was incomplete, i. e. not all cytokine expressing Th cells simultaneously expressed CD154 ( Supplementary Figures 7A–C ), as previously described for equine PBMC (25). The frequencies of CD154⁺ lymphocytes co-expressing different cytokines in BAL and PBMC samples ( Supplementary Figures 7A–C ) showed distributions between the groups comparable to all cytokine-positive lymphocytes ( Figure 6 ).

Cytokine and CD154 expression were induced in CD4⁺ Th cells in BAL cells and PBMC, but also in CD4^- lymphocytes ( Figures 7A, D, G, J, M, P, S, V ). Additional experiments with cells from three healthy horses ( Supplementary Figures 7D–H ) confirmed that the cytokine producing lymphocytes are mainly composed of CD4⁺ or CD8⁺ T cells, but CD4^-CD8^- lymphocytes also contribute to cytokine expression, particularly to IL-17A expression in BAL lymphocytes ( Supplementary Figure 7H ).