The Prevalence of Nine Genetic Disorders in a Dog Population from Belgium, the Netherlands and Germany

The objective of this study was to screen a dog population from Belgium, the Netherlands and Germany for the presence of mutant alleles associated with hip dysplasia (HD), degenerative myelopathy (DM), exercise-induced collapse (EIC), neuronal ceroid lipofuscinosis 4A (NCL), centronuclear myopathy (HMLR), mucopolysaccharidosis VII (MPS VII), myotonia congenita (MG), gangliosidosis (GM1) and muscular dystrophy (Duchenne type) (GRMD). Blood samples (K3EDTA) were collected for genotyping with Kompetitive Allele Specific PCR (n = 476). Allele and genotype frequencies were calculated in those breeds with at least 12 samples (n = 8). Hardy-Weinberg equilibrium was tested. Genetic variation was identified for 4 out of 9 disorders: mutant alleles were found in 49, 15, 3 and 2 breeds for HD, DM, EIC and NCL respectively. Additionally, mutant alleles were identified in crossbreeds for both HD and EIC. For HD, DM, EIC and NCL mutant alleles were newly discovered in 43, 13, 2 and 1 breed(s), respectively. In 9, 2 and 1 breed(s) for DM, EIC and NCL respectively, the mutant allele was detected, but the respective disorder has not been reported in those breeds. For 5 disorders (HMLR, MPS VII, MG, GM1, GRMD), the mutant allele could not be identified in our population. For the other 4 disorders (HD, DM, EIC, NCL), prevalence of associated mutant alleles seems strongly breed dependent. Surprisingly, mutant alleles were found in many breeds where the disorder has not been reported to date.


Introduction
The reduction of genetic disorders remains an important goal for both veterinarians and breeders [1]. It is the responsibility of the scientific community to provide detailed information regarding the existence, application and importance of diagnostic genetic tests that have been developed [1][2][3]. An important step in this process is to evaluate the prevalence of disorders and mutant alleles in the population. This information is needed to provide proper breeding advice. So far, only a few studies have been conducted to identify allele frequencies in a canine population [3][4][5][6]. The presence of mutant alleles in different breeds with each having their specific genetic background, might provide interesting (clinical) information for both the animal and human population as many canine disorders are animal models for human genetic disorders.
This study reports on the prevalence of mutant alleles associated with 9 canine genetic disorders ( Table 1) that influence the Table 1. Overview of 9 disorders tested with their corresponding chromosome number, the mutation, the effect, the inheritance, the breeds where the mutation has been reported before and the animal model.

Sample Collection
Blood samples (K3EDTA) were collected for a genetic database to study HD. Veterinarians from Belgium, the Netherlands and Germany were asked to take a blood sample from every dog that had a hip radiograph taken. Reasons for performing the procedure varied from screening purposes (breeding and assistance dogs) to dogs with clinical complaints (with HD in the differential diagnosis). No prerequisites were made regarding breed, sex and age.
Irrespective of breed, samples (n = 476) were tested for the presence of mutations associated with 9 disorders (HD, DM, EIC, NCL, HMLR, MPS VII, MG, GM1 and GRMD). Breeds where the mutant allele has already been reported, can be found in Table 1. A summary of breeds and samples per breed can be found in Table 2. Additionally, a mixed breed group of 28 dogs was tested.

Genotyping
Genomic DNA was isolated using routine procedures. For blood samples, 10 ml of blood was washed 3 times with 150 ml of a Tris-HCL based buffer. The procedure was performed with the use of robotic equipment. The cell pellet was lysed with Proteinase K (0.5 units for 45 minutes at 56uC followed by heat inactivation at 95uC for 5 minutes).

Statistical Analysis
Breed specific prevalence was analyzed in 8 breeds for which at least 12 samples were available (German Shepherd Dog, Labrador Retriever, Golden Retriever, Border Collie, Bernese Mountain Dog, American Staffordshire Terrier, Boxer, Spanish Water Dog). Hardy -Weinberg equilibrium (HWE) was tested with an online calculator (www.tufts.edu/,mcourt01/Documents). Data are available on request.

Results
No variation was found in any of the breeds for the mutations putatively responsible for 5 of the 9 disorders (HMLR, MPS VII, MG, GM1, GRMD). The mutations predicted to cause HD, DM, EIC and NCL, were observed in a wide variety of breeds (49, 15, 3 and 2 breeds, respectively). Breeds where mutant alleles were found are listed in Tables 3, 4, 5 and 6 for HD, DM, EIC and NCL, respectively. For HD, DM, EIC and NCL mutant alleles were newly discovered in 43, 13, 2 and 1 breed(s), respectively. The mutant alleles for HD and EIC were also identified in mixed breed dogs. For 25 of the 28 crossbreds, the parental breeds were known and we have shown these breeds to possess the respective mutant allele.
For 8 breeds, breed specific prevalence of alleles were reported (Table 3-6). Comparisons with previous reports could only be made for EIC in the Labrador Retriever: our population contained a very high number of genetically affected dogs (Table 7) [6]. HWE could be tested for HD in the German Shepherd Dog and the Golden Retriever (no significant deviation).
For the Labrador Retriever, genotype frequencies for HD and EIC both deviated significantly from HWE (p#0.001). The other breeds and disorders were not tested due to low number of samples and/or absence of variation.

Discussion
Since its first report in 1935, intensive research has focused on hip dysplasia (HD), one of the most frequent orthopedic disorder in dogs [17]. HD can be found in a wide variety of breeds [18,19]. Recently, a haplotype in the Fibrillin 2 (FBN2) gene has been reported to be associated with this highly prevalent, multifactorial disorder [9]. Fibrillins are components of extracellular microfibrils and have both a structural and a regulatory function [20]. The mutant AGC haplotype was identified in 49 different breeds in the population under study (Table 3). In 44 breeds, this mutant allele was reported for the first time. We identified the AGC haplotype as the only allele in 10 breeds (Cavalier King Charles Spaniel (n = 4), English Setter (n = 2), English Springer Spaniel (n = 1), Gordon Setter (n = 1), Laekenois (n = 1), Mastino Napoletano (n = 1), Rhodesian Ridgeback (n = 1), Saarlooswolfhond (n = 2), Siberian Husky (n = 2), Standard Poodle (n = 2)), however few conclusions can be made as the sample count is low (n = 17 overall). One additional dog was homozygous for the mutant allele, but as the breed was not correctly specified (Springer Spaniel without mentioning English or Welsh Springer Spaniel), this sample was excluded.
Degenerative myelopathy (DM) is characterized by progressive ataxia and upper motor neuron spastic paresis. The majority of dogs with DM start to develop symptoms from 5 years of age [21]. Diagnosis is not straightforward when patients are alive. Based on clinical symptoms and exclusion of other disorders (for example intervertebral disc disease, spinal cord neoplasia), DM can be the most likely etiology, but formal diagnosis can only be achieved post-mortem on histopathology [21]. Only for a subset of those breeds in which DM has been reported clinically, this disorder has been confirmed on histopathology. In 2009, a causal mutation was discovered in the superoxide dismutase 1 (SOD1) gene for 5 breeds (Table 1) with an age-dependent incomplete penetrance [8]. This gene encodes a free radical scavenger [21]. We identified the causal mutation in 15 breeds. In 2 breeds (Standard Poodle and Bernese Mountain Dog) the disorder was confirmed both clinically and on histopathology while in another 2 breeds (Border Collie and Collie Rough) the diagnosis of DM was made only on clinical examination [21]. We report the presence of the mutant allele in these 4 breeds and in an additional 9 breeds, in which the disorder has not yet been reported. We also confirmed the presence of the mutant allele in the German Shepherd Dog and Boxer (Table 4). DM has been clinically reported in the Labrador Retriever and confirmed by histopathology in the Golden Retriever, but the mutation has not yet been reported in these breeds [21]. In our population of respectively 74 and 62 individuals, we also did not identify the SOD1 mutation. The mutant allele might be infrequently present in the population or a different mutation might be responsible for the same disorder, as recently reported [22].
Dogs with exercise-induced collapse (EIC) develop incoordination of the hind limbs, paraparesis and/or tetraparesis and have an increased body temperature after strenuous exercise. A mutation in the dynamin 1 (DNM1) gene was found to be responsible for this disorder and an incomplete penetrance (influenced by the level of physical activity) was suggested [11]. DNM1 is important in neuronal synaptic vesicle recycling, especially during high levels of activity [23]. This disorder is reported mainly in Labrador Retrievers, but presence of the mutant allele has also been reported in other retriever breeds [6,11]. We report on the presence of the mutant allele for the first time in one English Cocker Spaniel and one Hungarian Vizsla ( Table 5). The prevalence of EIC in our Labrador population was higher than in a previous report (Table 7) [6]. The sample source seems to affect the percentage of affected dogs: when samples were collected from Labrador Retrievers in dog shows, field trials and from local pet owners in Canada and the United States, the prevalence of affected dogs was relatively low and in agreement with HWE expectations [6]. The opposite was true for dogs that were specifically tested for EIC and those results were in agreement with the results from our study, although we did not specifically select for those dogs. In our population of Labrador Retrievers, genotype frequencies for both HD and EIC deviate significantly from HWE. This might be a reflection of non-random sampling or selection [4]. For HD, non-random sampling can be expected since our dogs were initially collected to study this disorder. For EIC, the reason for rejection of the HWE is not clear. For both HD and EIC, our results tend to overestimate the number of affected dogs based on HWE. A similar result was found in the pet population in a previous study (26.9 versus 29.2%) [6].
Neuronal ceroid lipofuscinosis 4A (NCL) has been reported in American Staffordshire Terriers with progressive ataxia [24]. The causal mutation was discovered in 2010 in the Arylsulfatase G (ARSG) gene and an incomplete penetrance was suggested [7]. ARSG encodes for a lysosomal enzyme [25]. The mutant allele has a prevalence of approximately 9% in our 17 American Staffordshire Terriers which is less than the frequency expected based on a previous study [24]. We report the presence of the same mutation in the Bull Terrier, a breed where NCL has not been reported ( Table 6). No dogs were homozygous for the mutant allele in the population studied. For the other 5 disorders (HMLR, MPS VII, MG, GM1, GRMD) the mutant alleles were not detected in our population. Non-detection of the mutant allele might indicate that the allele is absent, that it is present but at a low frequency and that the number of samples tested was too low. For HMLR (Labrador Retriever, n = 66) and MPS VII (German Shepherd Dog, n = 67), we had, with the numbers of animals tested, a 99% chance of detecting every allele with a minor allele frequency (MAF) of at least 3.5% [26]. For GRMD (Golden Retriever, n = 19), we had a 99% chance of detecting alleles with a frequency of at least 11.5% [26]. For MG, GM1 and MPS VII (in the Brazilian Terrier), the sample size was too small to make any conclusions regarding the absence of the mutant allele [26]. The allele frequency for HMLR has previously been investigated [4]. In that study, the mutant allele frequency was very low (1.8% or 0.47%). To be able to detect all alleles with a MAF of 1% with a 99.9% probability, 344 samples would be needed. Since our sample count was much lower, it cannot be concluded that any of these 5 mutant alleles are completely absent.
For DM, EIC and NCL breeding advice can be given based on our population study. For some disorders, the mutant allele  frequency is quite high in certain breeds (.40% in Labrador Retriever for EIC). However, because of their recessive nature, a relatively fast reduction of both the mutant allele frequency and affected dogs can be achieved based on genetic tests. We propose to exclude certain genotypic combinations of dogs from mating, rather than excluding individuals, especially if high mutant allele frequencies are found. For autosomal recessive diseases, dogs homozygous for the mutant allele should not be combined with each other or with heterozygous dogs. However, they can still be used for breeding, but require a mating combination involving only homozygous wild type animals to reduce the number of affected dogs. As heterozygous dogs can be used the same way, no dogs need to be excluded. As essentially every dog can still be used for breeding, the clinical outcome of disorder can be prevented without excessive exclusion of carrier or genetically affected dogs from the breeding population. Reduction of the prevalence of HD based only on FBN2 will be more difficult since it is a multifactorial polygenetic disorder. Surprisingly, mutant alleles for 3 autosomal recessive disorders (DM, EIC and NCL) were found in 9, 2 and 1 breed(s) respectively where the disorder has not been clinically reported. The most plausible explanation might be that the disorder just has not been recognized in those breeds. A second explanation is the influence of the breed specific genetic background: the effect of mutations might be different in different breeds. This has been reported in Drosophila and the mouse [27,28]. In a meta-analysis in humans where studies on a wide variety of diseases and genes were compared, opposite effects between races were found, but none of them significant [29]. To the authors' knowledge, this phenomenon has not yet been reported in dogs.
This study reports on the presence of mutant alleles for 9 disorders in a wide variety of dog breeds. Veterinarians and dog breeders should be aware that mutations are present in breeds even where the disorder has not been reported. Dogs from nonsuspected breeds that show comparable symptoms to the disorders reported in this and other studies should be genotyped and results should be reported in order to create a reliable database. Ideally, phenotypical information for every disorder should be included in this database. As this is not available for all diseases in our database, this is a major limitation to this study.  Table 6. Breeds where mutant alleles for neuronal ceroid lipofuscinosis 4A were found and breed specific prevalence for breeds with at least 12 samples.