Measurement of Horse Allergens Equ c 1 and Equ c 2: A Comparison among Breeds

Introduction: Horse allergens are less studied than allergens from other furry animals and these allergens must be evaluated to understand the complexity of allergy to horses. The aims of this study were to develop assays for the horse allergens Equ c 1 and Equ c 2 in dander and saliva and to determine their levels in ten horse breeds. The study also included a comparison of these findings with previous results on the levels of Equ c 4 performed on the same study population. Method: The study population included 170 horses from 10 horse breeds including American Curly and Russian Bashkir horse, which have been suggested to be hypoallergenic. Competitive ELISA assays were developed, with polyclonal antibodies as capture antibodies, for the detection of Equ c 1 and Equ c 2 in dander and saliva samples. Results: The horse allergens Equ c 1 and Equ c 2 were found in all dander and saliva samples from the ten horse breeds. The GM level (ng/µg protein) of Equ c 1 in dander was 470 (range 129–2,569) and in saliva samples, 40 (range 6–160). The GM level of Equ c 2 in dander was 138 (range 18–1,650) and in saliva samples, 0.8 (range 0.03–17). In dander, there were no significant differences in Equ c 1 and Equ c 2 GM levels between stallions, mares, and geldings. Conclusion: Our results show high intra- and inter-breed variability. Neither the American Curly horse nor the Russian Bashkir horse, earlier categorized as hypoallergenic breeds, was associated with lower allergen levels of Equ c 1, Equ c 2, or Equ c 4 than the other horse breeds investigated.


Introduction
Equus caballus (domestic horse) continues to play a major role in human history although its use has changed greatly over the years and has included the development of a large number of breeds. Allergy to horse has been reported to affect up to approximately 14%, in a study from the north of Sweden [1], but horse allergy and horse allergens are less studied than allergens from domestic fur-

Study Population
The study population consisted of a total of 170 horses from 10 horse breeds: American Curly (AC), American Quarter horse (AQ), Gotland pony (G), Icelandic horse (I), North Swedish Horse (N), Russian Bashkir horse (B), Shetland pony (SP), Standardbred (S), Swedish warmblood (SWB), and Thoroughbred (T). The mean age was 10 years (<1-31 years) and the sex distribution was 87 mares, 27 stallions, and 56 geldings. Samples were collected from 144 horses in 2013 and from 108 horses in 2014. A total of 82 horses were sampled both years, see Table 1. The same study population was used in an earlier study to measure levels of Equ c 4 [16].
Ethical approval was not required, according to the Swedish Board of Agriculture (SJVFS 2015:38, chapter 2 § 15). Written informed consent was received from the horse owners.

Sample Collection
Dander and saliva samples were collected at the farm where the horses were stabled. The farms were spread over a big area in central Sweden. Horse dander (HD) was collected by grooming the horses and saliva samples were obtained using Salivette ® (Sarstedt, Numbrecht, Germany). HD and saliva samples were prepared as previously described [16,17]. Not all dander and saliva samples collected could be analysed for both Equ c 1 and Equ c 2 due to lack of material. The dander and saliva samples included in the analysis of Equ c 1 and Equ c 2, which were used for statistical calculations, are presented by breed and sex for year 2013 and 2014 in Table 1.

Production and Purification of Antigens, pAbs, and Biotinylation of Antigens
Antigen production: native Equ c 1 (nEqu c 1) was produced as previously described [8]. Equ c 2 was produced as a recombinant protein (rEqu c 2) [18].
Polyclonal antibodies (pAbs) were raised in rabbits immunized with either nEqu c 1 or rEqu c 2 by Agrisera (Agrisera AB, Vännäs, Sweden). The rabbits were immunized four times (week 1, 5, 9, and 13) and the final bleeding and serum collection were carried out at week 15 after first injection.
Antibody purification from the sera was performed by affinitychromatography according to the manufacturer's protocol using NHS HiTrap Column, code. 17-0716-01, lot. 10263475 (GE Healthcare) using an ÄKTA purification system (GE Healthcare). The columns were coupled with 2 mg of either nEqu c 1 or rEqu c 2 antigen. Protein concentration was determined using BCA Pierce TM protein assay (Thermo Fisher Scientific, Waltham, MA). The pAb collected was used as capture antibody in the ELISA assay to measure Equ c 1 and Equ c 2 levels in the dander and saliva samples.
Biotinylation of nEqu c 1 and rEqu c 2 was produced according to the manufacturer's instructions (EZ-Link TM Sulfo-NHS-Biotin lot, Thermo Fisher Scientific) and was used as competi-DOI: 10.1159/000525960 tive antigens. To remove uncoupled biotin, desalting was performed on a NAP TM 5 column (GE Healthcare) using phosphate-buffered saline (PBS) as buffer, according to the manufacturer's protocol.

Equ c 1 and Equ c 2 Standards
The assays for both allergen were optimized using nEqu c 1 and rEqu c as standards. Due to lack of sufficient nEqu c 1 and rEqu c 2 for use as standard to analyse all samples, an in-house dander sample was used and set to the concentration of the nEqu c 1 and rEqu c 2 standard curve, respectively. The concentration of the Equ c 1 standard curve ranged from 22.5 to 2,864 ng/mL and the Equ c 2 standard curve ranged from 34 to 2,149 ng/mL. Results were expressed as ng/µg protein.

Equ c 1 and Equ c 2 Competitive Assay
Ninety-six well Nunc-Immuno MaxiSorp Plates (Thermo Scientific) were coated with Affinity Pure Goat Anti-Rabbit IgG, 5 μg/mL (Jackson Immunoresearch) 100 μL/well, diluted in PBS, and incubated overnight at 4 ° C. The plates were washed three times with PBS containing 0.02% Tween ® (PBS-T) between steps and then blocked for 2 h with 1% bovine serum albumin in PBS-T. pAbs directed to Equ c 1or Equ c 2 in 1% bovine serum albumin PBS-T were added to all wells at 100 μL per well, followed by incubation for 2 h at room temperature. Blank, standard, samples and controls (100 μL/well) were added to the plates and incubated for 2 h at room temperature. Controls were aliquoted from a sample with known concentration, diluted high and low, to confirm that the values were stable throughout the analyses. Pri- or to adding to the plates, these solutions were mixed 1:1 with biotinylated Equ c 1 (1.6 ng/mL) or Equ c 2 (2.5 ng/mL), respectively. In the next step, 100 μL per well of streptavidin peroxidase (Jackson Immunoresearch, London) diluted 1/4,000 with Guardian TM Peroxidase conjugate stabilizer/Diluent (Thermo Fisher Scientific) was added. In the final step, 100 µL per well of TMB (Sigma Aldrich, Darmstadt, Germany) was added before stopping the reaction with 0.5 M sulphuric acid. Absorbance was measured at 450 nm.
Samples were run in duplicate and at three-fold dilutions to ensure that the allergen concentration fell within the linear section of the standard curve, generally around the third standard point. Blank samples composed of buffer were included and background absorbance was subtracted from the data points. The acceptable coefficient of variation was set at <20%.

Statistical Analysis
Differences between the breeds were estimated using a linear fixed-effect model. Age, sex, and breed were included as fixed effects, while horse-specific intercepts and by-breed sampling year were included as random effects. Visual inspection of residual plots based on log transformation of the dander and saliva values did not reveal any deviations from the assumptions of constant variance and normality. All confidence intervals reported are simultaneous confidence intervals that account for the multiple comparisons made. All analyses were done using R [19] version 3.3.1 with the Ime4package. Multiplicity-adjusted 95% confidence intervals for the estimated ratios were obtained using the multcomp package [20]. Associations between the repeated measurements were investigated using Pearson correlation coefficients between the log-transformed measurements. Repeated measurements include only the horses sampled in both 2013 and 2014.
The GM levels in saliva are highest for Equ c 2 (Fig. 2d) and slightly higher for Equ c 1 (Fig. 2b) in samples from stallions compared to both mares and geldings in both 2013 and 2014, without any adjustments. Levels of Equ c 1 and Equ c 2 in dander samples were almost equal in mares, geldings, and stallions (Fig. 2).
Comparisons of levels of Equ c 1 and Equ c 2 between breeds, adjusted for age, sex, and sampling year, are presented for dander in Figure 3 and for saliva in Figure 4. The data are presented as increasing ratios where 1 corresponds to equal levels between the compared breeds. For dander samples, the mean level of Equ c 1 for Gotland pony (G) was lower compared to all the other breeds, of which only differences between S and G and SP and G were significant (p < 0.05). The mean value for Shetland pony (SP) was higher compared to the other breeds both for Equ c 1 and Equ c 2 but not significantly. The mean level of Equ c 1 and Equ c 2 for SP was significantly high- er than Thoroughbred (T) (highlighted in Fig. 3a, b). For the saliva samples, the mean level of Equ c 1 for American Quarter horse (AQ) was higher than for all the other breeds, while significantly higher than SP, N, and S. The mean level for G was significantly higher than for SP (Fig. 4a). The mean values of Equ c 2 for G was significantly higher compared to the other breeds, which is highlighted in Figure 4b. Estimated differences in dander and saliva for levels of Equ c 1 and Equ c 2 (ng/µg protein) between stallions, mares, and geldings are shown in Figure 5. In dander, the levels of Equ c 1 were almost the same for stallions, mares, and geldings, but levels of Equ c 2 were slightly higher in stallions. Saliva from stallions showed significantly higher levels of both Equ c 1 and Equ c 2. The GM value of Equ c 2 in saliva was almost four times higher than that of geldings and mares.
To be able to compare horse allergen levels with earlier studies [15], the median levels in dander of all three allergens Equ c 1, Equ c 2 (ng/µg protein), and Equ c 4 (U/ µg protein) [16] are presented in Table 2, presenting results from one observation per horse. The median levels were higher in stallions for Equ c 2 and Equ c 4 (178 and 1,847) compared to mares (150 and 612) and geldings (144 and 658), while Equ c 1 in stallions was lower than for geldings and mares.
Allergen levels in dander and saliva samples from horses sampled in both 2013 and 2014 are shown in Figure 6, testing the null hypothesis that the correlation coefficient is equal to zero. The levels correlate poorly to moderate in dander samples for both allergens, Equ c 1 (p = 0.477, Fig. 6a), Equ c 2 (p = 0.005, Fig. 6c). The corresponding values for saliva samples showed weaker correlation for the horses sampled both in 2013 and 2014, Equ c 1 (p = 0.034, Fig. 6b), Equ c 2 (p = 0.012, Fig. 6d).

Discussion
In this study, levels of the horse allergens Equ c 1 and Equ c 2 in ten horse breeds were compared. These allergens were present in dander and saliva samples from all breeds. An earlier study from this research group showed the same result for the allergen Equ c 4 from the same horses [16]. The results showed high intra-and interbreed variations. No significant differences between breeds were found. In our earlier study, Equ c 4 levels were shown to be highest in stallions both in dander and saliva samples. However, this pattern for sex difference was not as clear for Equ c 1 and Equ c 2 since only Equ c 2 levels in saliva showed significantly higher levels in stallions than mares and geldings when adjusted for age, breed, and sampling year. To our knowledge, this is the first study to present data on levels of the horse allergen Equ c 2.
Similar results with high variability in allergen content among individual animals and breeds have also been shown by Zahradnik et al. [15]. Their study covered 224 hair samples (HD antigen) from 32 horse breeds, of which 11 breeds were represented by only 1 horse and 5 breeds by only 2 horses. They concluded that stallions displayed higher median levels of Equ c 1 and Equ c 4 than mares and geldings. While the data on Equ c 4 agree with our previously published results, here we show similar Equ c 1 levels for all the sexes. Furthermore, their study showed higher concentrations of Equ c 1 and Equ c 4 in American Curly horses compared to most of the horses investigated. In the present study, American Curly horses and Bashkir horses showed similar median levels of Equ c 1, Equ c 2, and Equ c 4 in dander samples compared to median levels of all breeds (Table 2).
From Figure 1, we can conclude that Equ c 1 GM levels for American Curly and Bashkir horse were similar to all breeds (see dotted line in Fig. 1). Equ c 2 GM levels of American Curly were higher in both dander and saliva samples compared to GM levels of all breeds, while the GM values for the Bashkir horse were equal (dander) or slightly lower (saliva) than for other breeds  Fig. 1c, d). From our earlier study, it can be concluded that GM levels of Equ c 4 in dander for American Curly and Bashkir were almost the same as the GM level for all breeds, while the GM level in saliva samples was slightly higher for American Curly and slightly lower for the Bashkir than the GM level for all breeds [16]. Taken together, there is not much evidence for the notion that American Curly and Bashkir horse breeds are hypoallergenic.
Adjusted comparisons between breeds, presented as ratios, show that the mean value for Equ c 1 in dander samples was lower in Gotland pony and higher in Shetland pony compared to all the other breeds. The mean value for Equ c 2 in dander samples was lower in Thoroughbred and higher in Shetland pony than all the other breeds. Data from the earlier study showed that Equ c 4 was higher in the American Quarter horse compared to all the other breeds.
This knowledge may have implications in the future, when component-resolved diagnostics (CRD) may also become available for horse allergens [21]. CRD is beginning to gain greater recognition and clinical utility in fine-tuning the diagnostics of allergy to furry animals [22]. One possibility is that CRD could indicate that a horse-allergic individual may tolerate exposure to certain individual horses but not others.
The strength of this study is that allergen levels in both dander and saliva were studied for a large number of horses: 170 from 10 different horse breeds. Efforts were made to ensure that the samples were representative. A variety of ages and sexes were selected for the sampling. The samples were collected during the summer when the horses were in the fields and not trimmed. In addition, levels of allergens were compared between the same individuals and found to be stable.
The values cannot be regarded as absolute values but rather relative values for comparison between individual horses and breeds for each assay. The assays used for measuring Equ 1 and Equ c 2 used pAbs and a competitive ELISA method, while the Equ c 4 assay was based on commercially produced monoclonal antibodies (Mabtech, Stockholm, Sweden) with an in-house standard [17]. More research is needed to be able to compare the different allergen levels with each other.
To conclude, our results show high intra-and interbreed variability for the horse allergens Equ c 1, Equ c 2, and Equ c 4. Neither the American Curly horse nor the Bashkir horse was associated with lower allergen levels of Equ c 1, Equ c 2, or Equ c 4 than the other horse breeds investigated.