Out of 948 horses, which were slaughtered or euthanized in the period between November 2002 and October 2004 in three western districts of Westphalia, 26 horses were identified as possibly suffering from RAO by questioning veterinarians and/or horse owners about reasons for slaughter or euthanasia. Slaughter was done according to federal law in authorized slaughterhouses and under control of a veterinarian. Euthanasia was practised by veterinarians following internationally recognized guidelines. If the horses had shown respiratory disease history, symptoms and therapy as well as conditions of housing were evaluated in detail by using a standardized questionnaire, which is available as online addendum. Possible RAO and control horses were clinically examined before slaughter or euthanasia. The horses had to meet the following criteria [Additional file 1]: (1) chronic cough for at least 3 months duration; (2) exercise intolerance; (3) worsening of the symptoms due to dust; (4) obvious biphasic expiratory dyspnoea with hypertrophy of Mm. recti abdomini, inflated nostrils and nasal discharge; (5) breathing frequency > 20/min; and (6) pathologic findings in lung auscultation. As respiratory healthy controls served 20 horses which completely lacked these clinical criteria. Immediately after death, tissue was taken from 8 different lung regions of each horse and preserved for the following examinations.

Formalin-fixed lung tissue was embedded in paraffin wax (Tissuewax™; Medite GmbH, Burgdorf, Germany), slices of 3–7 μm thickness were cut using a rotatory microtome (Microm GmbH, Walldorf, Germany), and stained with haematoxylin and eosin. Slices from the 8 tissue samples of each horse were evaluated for histological changes using a semi-quantitative score (0–4) in a blinded study. No signs of bronchiolitis were scored with 0 points. Minimal infiltration of the bronchioli with lymphocytes was scored with 1, slight inflammation with 2, moderate bronchiolitis with some neutrophils with 3, and severe changes with intraluminal aggregates of neutrophils with 4 points. Results from all 8 regions of each horse were added. Therefore a minimal score of 0 and a maximal score of 32 was possible.

One horse with RAO-like respiratory symptoms and history revealed parasitic lung disease after histological investigation and was excluded therefore.

According to anamnesis, clinical findings and light microscopy, the remaining 45 horses were classified into four subgroups:

I. Clinically healthy horses without histological changes of RAO (n = 15),

II. Clinically healthy horses with a low inflammation score of 4–6 (n = 5),

III. Horses with clinical signs of RAO, but without or with only slight histological changes and a score of 0–5 (n = 16), and

IV. Horses with symptoms and histological changes of RAO with a score > 10 (n = 9).

For immunohistochemistry, dewaxed slices were stained with a mouse anti-CP primary antibody (Biotrend, Germany; clone 73/0200, dilution 1:60) for 2 hours at room temperature, followed by a peroxidase-based detection system (Histostain^®-Bulk-Kit, Zymed Laboratories Inc., San Francisco, USA) according to the instructions of the manufacturer). Three high-power fields showing bronchioli were selected for counting all detectable vs. all CP-positive stained bronchiolar epithelial cells (BEC), type II pneumocytes (PII) and macrophages (MP) separately, using a Zeiss Axiophot microscope (Carl Zeiss, Oberkochen, Germany) with a 40-fold objective magnification and a digital imaging system (Diskus Software, Koenigswinter, Germany). For immunofluorescence, triple staining on dewaxed sections was done with a primary mouse monoclonal IgG_2b-antibody against CP (Biotrend, Germany; clone 73/0200, dilution 1:100) and a primary mouse monoclonal IgG₁-antibody mix against pan-cytokeratin (BioCarta, USA; clones AE1, AE3, 5D3, dilution 1:50), as well as 4',6-diamidino-2-phenylindol-hydrochloride (DAPI, Sigma, Germany, working concentration 0.1 μg/ml). Alexa Fluor 488-conjugates of goat anti-mouse-IgG₁ (dilution 1:400), and Alexa Fluor 594-conjugates of goat anti-mouse-IgG_2b (dilution 1:400) were used as secondary antibodies (Molecular Probes Europe, Leiden, The Netherlands). Results were documented with a CCD-camera (Kappa, Gleichen, Germany) attached to an Olympus IX-70 microscope (Olympus Europe, Hamburg, Germany) with 40-fold objective magnification.

Lung tissue was collected immediately after slaughtering/euthanizing of the horses and stored at -80°C. DNA was isolated from lung tissue using the High Pure PCR Template Preparation Kit (Roche Diagnostics, Mannheim, Germany) according to the instructions of the manufacturer. 5 μl of the DNA extract were used as template in PCR. Samples were tested for CPP/CPA and CPPN by a modified version of the nested PCR procedure described by Kaltenboeck et al. 16 which targets the ompA gene. The first step was genus-specific amplification using primers 191CHOMP (5'-GCI YTI TGG GAR TGY GGI TGY GCI AC-3') and CHOMP371 (5'-TTA GAA ICK GAA TTG IGC RTT IAY GTG IGC IGC-3'). For the second amplification, we used 1 μl of the genus-specific product and primer combination 218PSITT (5'-GTA ATT TCI AGC CCA GCA CAA TTY GTG-3')/CHOMP336s (5'-CCR CAA GMT TTT CTR GAY TTC AWY TTG TTR AT-3') for CP, or 201CHOMP (5'-GGI GCW GMI TTC CAA TAY GCI CAR TC-3')/PNEUM268 (5'-GTA CTC CAA TGT ATG GCA CTA AAG A-3'), for CPPN, respectively. The sizes of specific amplicons are: 576–597 bp (genus-specific product), and 389–404 bp for CP, or 244 bp for CPPN after nested PCR. A detailed protocol of the procedure was previously published by Sachse and Hotzel 17 and is given here:

For species identification, products of the first PCR round were amplified using primers 201CHOMP and CHOMP336s. Specific bands from PCR products were cut out of the agarose gel (1%) and DNA was extracted using the QIAquick Gel Extraction Kit (QIAGEN, Hilden, Germany). Prior to processing on the ABI PRISM 310 Genetic Analyzer (Applied Biosystems), these extracts were subjected to cycle sequencing using the same primers as above and the BigDye™ Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems, Darmstadt, Germany). The average length of the sequences determined was 400 nucleotides. The species identity was established through BLAST search 18. Sequence alignments and analysis were conducted using the Vector NTI Suite 8.0 software package (Informax Inc., Oxford, UK).

Statistical analysis was done using SPSS, version 12 (SPSS Inc., Chicago, USA). The data were statistically evaluated using Mann-Whitney-U-test, Pearson Chi Square test and Spearman-Rho correlation (2-sided). A test value below 0.05 was considered to be statistically significant.

As a result of the standardized questionnaire and histology 45 out of a total of 948 horses were recruited for the present study. 43 animals were slaughtered and 2 euthanized. The median age was 16.11 years (span: 9–25), among these were 23 mares, 20 geldings and 2 stallions. Concerning the respiratory system, 20 animals were considered clinically as healthy, whereas 25 showed clinical symptoms of RAO (Table 1). Lameness was the main cause to kill the horses (40% in both groups). In horses with RAO, this disease was the reason for killing the animal in 36%; a respective therapy was applied in 56%. Therapeutic regimens differed widely, thus not allowing further statistically evaluation. The majority of horses both in the clinically healthy and the RAO groups were housed in stables with straw bedding (healthy controls: 75%, RAO: 68%). Outdoor housing was reported in 20% of the controls and none of the histologically approved RAO cases. Stable housing with wood shavings or linen straw was reported in 32% of cases with RAO and 5% of controls, respectively.

Comparing RAO and healthy horses (Table 1) a significantly different histological score (p = 0.010), number of Chlamydia psittaci (CP) antigen-positive bronchiolar epithelial cells (p < 0.001), and percentage of all CP antigen-positive cells (p < 0.001) were found.

In subgroups II and III the distribution of inclusions and antigens was typical for persistent chlamydial infections in IHC: inclusion bodies were only sparsely seen, but antigens were much more frequently found. This was also true for subgroup 1, but on a much lower level.

Animals which were killed specifically because of RAO (n = 9) were significantly different in the median number of CP antigen-positive bronchiolar epithelial cells (83 vs. 42, p = 0.037) compared to horses which were put down due to other reasons (n = 36).

Results of clinical data and histological evaluation of the 45 horses led to the classification shown in Table 2. The four subgroups were significantly different from each other in their light microscopy histological score (Mann-Whitney U-test, p < 0.05). Age was not significantly correlated to the histological score (Spearman-Rho, p = 0.425).

The clinically healthy horses in subgroup I showed no (Figure 1) and in subgroup II only slight bronchiolitis. CP antigens were shown in bronchiolar epithelial cells (Figures 2, 3 and 4) as well as occasionally in type 2 pneumocytes and macrophages (Figures 4 and 5). Type II pneumocytes were positive in 12 cases (3 in subgroup I, 1 in subgroup III, 8 in subgroup IV), and in 6 cases positive staining was also seen in macrophages (1 in subgroup I, 5 in subgroup IV). Data and significant differences are shown in detail in Table 2. With increased severity of disease there was a significant rise in antigen-positive cells. Age was not significantly correlated with immunohistological parameters (Spearman Rho, p > 0.05). Positive and significant Spearman-Rho correlations were found between all of the following parameters: light microscopy score, the number of CP antigen-positive macrophages, type II pneumocytes and bronchiolar epithelial cells. However, the highest correlation coefficient (r = 0.612) was found between the number of CP antigen-positive type II pneumocytes and the number of CP antigen-positive macrophages, the lowest correlation coefficient found was r = 0.41 between the number of CP antigen-positive macrophages and the histological score (Table 3).

To obtain further information about localization of chlamydial antigens, immunofluorescence testing was performed in three horses. Multiple typical perinuclear spots in bronchiolar epithelial cells, macrophages and granulocytes were seen in 2 horses of subgroup IV (Figures 6 and 7). In horses with RAO, the abundance of chlamydial inclusion bodies (red spots) in bronchiolar epithelial cells (stained green with an anti-cytokeratin pan antibody) and macrophages varies from high (Figure 6) to low (Figure 7).

Housing with straw contact vs. other bedding or RAO with therapy vs. without therapy and all combinations of the parameters histological score, results of immunohistochemistry or PCR, lacked any correlations (Mann-Whitney U-test, p > 0.05).

No significant differences were seen between clinically healthy and sick horses (Table 1). PCR revealed CP DNA in 5 of 15 (= 33.3%) horses in subgroup I, 4 of 5 (= 80%) in subgroup II, 13 of 16 (= 81.2%) in subgroup III and 2 of 9 (= 22.2%) in subgroup IV. Significant differences in PCR positivity are shown between subgroups I and III as well as between subgroups III and IV (Table 2).

To confirm the identity of the chlamydial species, DNA from 22 of the 24 positive samples was sequenced in the ompA gene region (approximately 400 bp). A BLAST search of these sequences revealed close to 100% homology to the species CPP in 9 and CPA in 13 cases. No significant differences were found between the four subgroups (Pearson Chi Square test, p = 0.309).