The equine DNAH 3 gene : SNP discovery and exclusion of an involvement in recurrent airway obstruction ( RAO ) in European Warmblood horses

Recurrent airway obstruction is one of the most common airway diseases affecting mature horses. Increased bronchoalveolar mucus, neutrophil accumulation in airways, and airway obstruction are the main features of this disease. Mucociliary clearance is a key component of pulmonary defense mechanisms. Cilia are the motile part of this system and a complex linear array of dynein motors is responsible for their motility by moving along the microtubules in the axonemes of cilia and flagella. We previously detected a QTL for RAO on ECA 13 in a half-sib family of European Warmblood horses. The gene encoding DNAH3 is located in the peak of the detected QTL and encodes a dynein subunit. Therefore, we analysed this gene as a positional and functional candidate gene for RAO. In a mutation analysis of all 62 exons we detected 53 new polymorphisms including 7 non-synonymous variants. We performed an association study using 38  polymorphisms in a cohort of 422 animals. However, after correction for multiple testing we did not detect a significant association of any of these polymorphisms with RAO (P>0.05). Therefore, it seems unlikely that variants at the DNAH3 gene are responsible for the RAO QTL in European Warmblood horses.


Introduction
RAO or heaves is one of the most common airway diseases affecting mature horses.RAO is characterised by bronchospasms and increased mucus and neutrophil accumulation in the airways.It occurs following exposure of susceptible horses to antigens and endotoxins present in hay and stable dust (Cunningham & Dunkel 2008, Léguillette 2003).Because of many similarities of this disease to human asthma it is used as an animal model of human asthma (Deaton 2006, Leclere et al. 2011).
The prevalence of RAO is estimated at 10-20 % of adult horses living in cold and temperate climates.There is no breed or sex predisposition for RAO (Leclere et al. 2011).RAO is a complex trait controlled by environmental and genetic factors.In a Warmblood population there was evidence for the presence of one or more major genes and the heritability in a study cohort exposed to hay feeding was estimated to approach 100 %.Thus hay-feeding may be the only significant environmental factor influencing this disease (Gerber et al. 2009).
Mucociliary clearance is a key component of pulmonary defense mechanisms.It involves the regulation of ion transport by the airway epithelium, mucus secretion, and ciliary function (Knowles & Boucher 2002).Cilia and flagella contain motile microtubules with a »9+2«structure, in which two central singlet microtubules are surrounded by nine outer doublet microtubules.A complex linear array of dynein motors is responsible for their motility (Brokaw 2009, Hayashi & Shingyogi 2008).
Cytoplasmic and axonemal dyneins are involved in the cytoplasmatic movement of organelles and the bending of cilia and flagella, respectively (Bartoloni et al. 2001).The DNAH proteins assemble with intermediate and light chains into large multiprotein complexes to form inner and outer dynein arms (Holzbaur & Vallee 1994, DiBella & King 2001).The inner and outer dynein arms slide on the outer doublet microtubules by hydrolysing ATP.
Defects in the motile cilia are responsible for the most prominent ciliopathy in mammals, called primary ciliary dyskinesia.One aspect of PCD is respiratory disease due to impaired mucociliary clearance.Additional features of PCD are sperm immobility and randomisation of left-right asymmetry (Afzelius 1976, Rossman et al. 1980).The different features of PCD are now clearly explained by mutations involving various subunits of the axonemal structures (Lee 2011).
Due to their importance for mucociliary clearance the genes encoding dynein subunits can be considered functional candidates for RAO.In a previous microsatellite-based linkage study we detected a QTL for RAO in a half-sibling family of European Warmblood horses with a peak of about 4 Mb extending from 24-28 Mb on ECA 13 (Swinburne et al. 2009).The DNAH3 gene expressed in trachea, testis and brain (Maiti et al. 2000) is located on ECA 13 at 26.4-26.6Mb in the peak region of the detected QTL.Therefore, we analysed the equine DNAH3 gene as a positional and functional candidate gene and performed an association study in a cohort of European Warmblood horses.

Animals and phenotypes
For variant detection we selected two RAO-affected and two non-affected half-sibling offspring of a previously described European Warmblood family, in which the QTL on ECA 13 was detected (Swinburne et al. 2009).We selected these four horses based on their phenotypes and marker haplotypes at the QTL.
For the association analysis we used 464 European Warmblood horses that were unrelated at the grandparental level.More than 60 % (281) of the sampled horses were from Switzerland and Germany.Additional horses were from Belgium, the Czech Republic, Denmark, France, Hungary, Ireland, Latvia, The Netherlands, Poland, Portugal, Russia, Slovakia, and Sweden.Cases and controls were matched according to countries in order to minimise the stratification.A multidimensional scaling plot of genome-wide SNP data from these horses did not result in different clusters between cases and controls (data not shown).
We determined the phenotypes in our sample population according to the »Horse Owner Assessed Respiratory Signs Index« (Ramseyer et al. 2007, Laumen et al. 2010).The HOARSI gives scores from 1-4, which correspond to absent, mild, moderate and severe clinical signs respectively.Briefly, horse owners were contacted by phone and only horses with clinical signs that had persisted for at least 2 months were included in the study.All horses were 5 years or older, with at least a 12-month history of hay-feeding.A standardised questionnaire was used to gather information from the horse owners about the animals' history of chronic, regular or frequent coughing, increased breathing effort at rest, increased breathing effort after exercise, and nasal discharge.The HOARSI classification refers to the period when the horses were exhibiting the most severe clinical signs.We selected horses with HOARSI 1 as controls and horses with HOARSI 3 or 4 as RAO cases.We used an overlapping set of horses in a related study (Shakhsi-Niaei et al. 2012).
For the association study we determined the genotypes of 35 SNPs by using Golden Gate assays on a BeadXpress station (Illumina Inc., San Diego, CA, USA) followed by data analysis with the BeadStudio v3 software (Illumina Inc, San Diego, CA, USA).The genotypes of 4 additional SNPs in the region were available from equine SNP50 Beadchip data (Illumina Inc, San Diego, CA, USA).All genomic positions refer to the EquCab 2 assembly (http://www.ncbi.nlm.nih.gov/projects/mapview/).The equine cDNA and protein positions refer to the database accessions XM_001491853.3and XP_001491903.2,respectively.

Protein analysis
In order to analyse the evolutionary conservation of the non-synonymous variants, we prepared multi-species alignments of the DNAH3 protein sequences using the ClustalW2 software (European Bioinformatics Institute, Cambridge, UK; http://www.ebi.ac.uk/Tools/ msa/clustalw2/).

Association analysis
We removed individuals and markers with less than 90 % call rate with the Plink software (Purcell et al. 2007).We also excluded markers with a minor allele frequency of less than 0.05 and markers that strongly deviated from Hardy-Weinberg equilibrium (P<0.001).Subsequently, we performed an allelic association study using chi-square tests.We also performed haplotypic association analyses using sliding windows of 3 and 7 markers, respectively.We corrected for multiple testing by applying the Bonferroni correction.

Mutation analysis of the DNAH3 gene
We sequenced 34 156 nucleotides containing all 62 exons with 12 255 nucleotides of coding sequence of the DNAH3 gene in two RAO affected and two control horses.These four horses were half-sibling offspring of the European Warmblood family, in which the QTL on ECA 13 was originally detected.We identified a total of 53 new polymorphisms including 7 nonsynonymous variants in addition to 4 previously known SNPs in the region (Table 2).Protein sequence alignment showed that 4 of these 7 non-synonymous variants led to amino acid exchanges in highly conserved positions (p.Gln1227Arg, p.Met2357Val, p.Gln2665Arg, p.Ser3033Cys; Figure 1).These variants are located either in or very close to functionally annotated domains of the DNAH3 protein (Table 3).It has to be cautioned that the 5'-end of the predicted horse DNAH3 transcript (XM_001491853.3) corresponding to the first exon is very different from the human ortholog.Thus there may be an error in the equine genome assembly or the annotation of the equine DNAH3 gene.Such a hypothetical error might alter the positions of the identified variants by about 30 amino acids.The human-horse relationships are given in Table 3.

Association analysis of DNAH3 variants with RAO
For the association study we genotyped 39 SNPs including all 13 exonic variants in a cohort of 464 animals.We removed one marker due to low minor allele frequency and 42 horses due to low call rates, so that 422 horses (230 cases & 192 controls) and 38 markers remained for the final allelic association analysis (Table 2).All of the tested markers were in Hardy-Weinberg equilibrium and had similar allele distributions between cases and controls with raw P-values greater than 5 %.As 38 markers were tested, a Bonferroni-corrected significance threshold would have been 0.05 / 38=0.0013.We also did not detect a significant haplotype association.The numbering refers to the equine protein accession XP_001491903.2.The predicted equine protein differs at its N-terminus from the human protein, possibly due to an annotation error in the horse genome reference sequence.The predicted equine protein shows high homology to its human ortholog starting from amino acid 11 in horse and amino acid 41 in human.The domain positions correspond to the human DNAH3 protein from the UniProtKB/Swiss-Prot accession Q8TD57.
In previous association studies for RAO on ECA 13, we found the strongest association for the marker BIEC2-224511 (Shakhsi-Niaei et al. 2012).This marker has a raw P-value of 0.015 in the 422 horses of this study and thus shows an almost four-fold stronger association than the best-associated marker from the DNAH3 gene.BIEC2-224511 is located at position 24 309 405, approximately 2 Mb away from the DNAH3 gene.The best associated marker in the DNAH3 gene (c.-263C>T) is at the left boundary of the DNAH3 gene and thus also physically closest to BIEC2-224511 of all tested markers (Table 2).
The lack of association of the tested DNAH3 markers indicates that variants at the DNAH3 gene are most likely not responsible for the RAO QTL in European Warmblood horses.

Figure 1
Figure 1Multiple species alignment of the DNAH3 protein sequence at the sites of non-synonymous variants in the horse.Four of the seven variants lead to amino acid exchanges in highly conserved positions.

Table 1
Primer sequences.The positions correspond to chromsome 13 in the Equus caballus 2 genome assembly.

Table 1 continued
Primer sequences.The positions correspond to chromsome 13 in the Equus caballus 2 genome assembly.

Table 2
Polymorphisms of the equine DNAH3 gene and association data

Table 2 continued
Polymorphisms of the equine DNAH3 gene and association data

Table 3
Protein domain assignment of non-synonymous variants