Evaluation of the major histocompatibility complex (MHC) class II as a candidate for sudden acquired retinal degeneration syndrome (SARDS) in Dachshunds

Abstract Objective Sudden acquired retinal degeneration syndrome (SARDS) is one of the leading causes of acute blindness in dogs, with an unknown etiology and no effective treatment. Certain breeds such as Dachshunds are overrepresented among SARDS patients, and therefore, the syndrome is suspected to have a genetic component. The objective of this study was to determine if a genetic locus associated with SARDS in Dachshunds could be identified using a genome‐wide association study (GWAS). Procedures Genome‐wide association mapping was performed in 15 SARDS‐affected and 16 unaffected Dachshunds. Genotyping of three classical DLA class II genes (DLA‐DRB1, DLA‐DQA1, and DLA‐DQB1) was performed in 34 SARDS‐affected and 66 unaffected Dachshunds to evaluate for an association in this region. Results Although no single nucleotide polymorphisms (SNPs) were of genome‐wide statistical significance (P Bonferroni < 0.05), 5 of the top 9 SNPs were in the major histocompatibility complex (MHC). Using DLA typing, the allele DLA‐DRB1*09401 was identified as a risk factor for the development of SARDS (P = 0.0032, OR = 4.0). The alleles DLA‐DQB1*00101 (P = 0.0050, OR = 0.31), DLA‐DQA1*00901 (P = 0.0087, OR = 0.33), and a previously identified DLA‐DRB1allele described as “DRB1‐T” (P = 0.0284, OR = 0.37) were identified as protective factors. Conclusions Although far from definitive, association of SARDS with alleles of immunologic importance further supports the hypothesis that autoimmunity may play a role in the pathogenesis of SARDS.


| INTRODUCTION
Sudden acquired retinal degeneration syndrome (SARDS) is a common cause of irreversible blindness in dogs. 1 The disease is typically acute in onset and characterized by an initially normal fundic appearance despite an extinguished electroretinogram. Currently, the etiology of SARDS remains unknown and there is no proven treatment for the condition, although several different etiologic hypotheses exist and several therapeutic approaches have been suggested. Because dogs with SARDS often present with concurrent systemic signs such as polyuria, polydipsia, polyphagia, lethargy, and weight gain, as well as laboratory abnormalities consistent with hyperadrenocorticism, some researchers hypothesize that SARDS is an endocrine disorder. 1 In one study, >90% of dogs with SARDS had elevated adrenal sex hormone and/or cortisol serum concentrations. 2 By contrast, others suggest that SARDS is the canine equivalent of autoimmune retinopathy (AIR)-a disease of humans in which antiretinal antibodies result in vision loss and severe attenuation or loss of retinal electrical activity, with minimal changes in funduscopic appearance. 3,4 In one study, an autoantibody to neuron-specific enolase (NSE)-a glycolytic enzyme found in photoreceptors, neurons, and neuroendocrine tissues-was identified in 6 of 24 dogs with SARDS, while no control dogs had detectable autoantibodies. 5 In 2014, the American College of Veterinary Ophthalmologists' Vision for Animals Foundation hosted a think tank to prioritize goals for future SARDS research. 1 One of their specific recommendations was to use genomewide association studies (GWAS) to identify genetic factors that may contribute to an understanding of the etiologic and pathogenic mechanisms of the disease. This technology uses single nucleotide polymorphisms (SNPs) at thousands of loci across the entire canine genome to identify regions of genetic difference between control and affected animals. If a statistically significant association is found between a region of the canine genome and the disease of interest, candidate genes within the region can be identified for further investigation. Identification of a genetic association would increase understanding of SARDS, refine the direction of future research, and could ultimately lead to the development of treatments and identification of at-risk patients. Furthermore, it could support or refute the hypothesis that SARDS may serve as a spontaneous canine model for AIR.
A recent retrospective study conducted at the University of California-Davis Veterinary Medical Teaching Hospital (UCD-VMTH) identified several breeds that were overrepresented in the SARDS patient population. 6 Dachshunds, in particular, were 8-fold overrepresented, which is consistent with previous reports 7 and may indicate a genetic component of the disease. Therefore, the purpose of the present study was to use genome-wide association to investigate genetic associations with SARDS in Dachshunds.

| Animals
This study was approved by the Institutional Animal Care and Use Committee at the University of California-Davis (Protocol #18719) and performed in accordance with the Association for Research in Vision and Ophthalmology's guidelines for the use of animals in research. Blood samples were collected from SARDS-affected and unaffected Dachshunds at the UCD-VMTH, as well as from animals recruited through outreach to veterinary ophthalmologists and to Dachshund breed organizations within the United States.
Inclusion criteria for SARDS-affected Dachshunds included sudden and complete vision loss, examination and diagnosis by a board certified veterinary ophthalmologist or ophthalmology resident in training, a bilaterally extinguished electroretinogram (ERG), and lack of an explanation for sudden and complete vision loss on fundic examination (eg, total retinal separation, diffuse severe chorioretinitis). Dogs diagnosed with SARDS were excluded from the current study if they had intraocular abnormalities impeding adequate visualization of the ocular fundus. All Dachshunds in the control population were >8 years old, had an intact menace response and/or pupillary light reflexes, and were reported to be visual by their owner. Dachshunds <8 years of age were excluded in order to avoid dogs that may be genetically predisposed to SARDS but had not yet developed signs of the syndrome. Exclusion criteria for control dogs included Cushing's disease, progressive retinal atrophy, notable regions of retinal degeneration or detachment on fundic examination, and/or an abnormal ERG in one or both eyes.
With the exception of one sample that did not have enough DNA remaining after the GWAS, all DNA samples used for the GWAS were also used in major histocompatibility complex (MHC) genotyping. Additional control and SARDS-affected dogs were recruited for MHC genotyping, and a number of control DNA samples were obtained from the Canine Genetics Biorepository (CGB), a database of gDNA samples collected from VMTH patients (Protocol # 20356). The CGB DNA samples were only used if medical record review demonstrated that they met the same inclusion criteria listed above.

| Genome-wide association study
Whole blood from affected and control dogs was collected into tubes containing ethylenediaminetetraacetic acid, DNA was extracted using the Puregene kit (Qiagen, CA), and DNA concentration was determined with a nanodrop spectrophotometer. DNA samples were genotyped with the Axiom™ | 753 STROMBERG ET al.
Canine Genotyping Array Set B (Applied Biosystems™, CA). The manufacturer's quality control metrics were applied, and SNPs with less than 90% genotyping call rates and a minor allele frequency over 0.05 were removed. Samples from affected and control Dachshunds were analyzed for population stratification by quantile-quantile (Q-Q) plots. Bonferroni correction was utilized to account for the number of tests performed. Association analysis was performed using PLINK software (Shaun Purcell, URL: http://pngu.mgh.harvard.edu/purcell/plink/). 8

DQA1, and DQB1
Direct sequencing of genomic DNA (gDNA) from affected and control Dachshunds was performed using locus-specific intronic primers targeting the follow- To remove unconsumed dNTPs and primers, PCR products were cleaned using the ExoSap-IT ® kit (Applied Biosystems, CA) according to the manufacturer's protocol. Sequencing reactions were performed with Big Dye® terminator mix (Applied Biosystems, CA) using the aforementioned primers according to the manufacturer's protocol. The purified PCR products were sequenced on an ABI 3500 Genetic Analyzer (Applied Biosystems, CA).

| DNA allele and haplotype assignment
DRB1/DQA1/DQB1 allele determination was performed by comparison with annotated sequences in GenBank, beginning with homozygous individuals. For heterozygotes, a comparison was made of GenBank sequences that most closely matched the sequence data, identifying the two alleles that, when combined, could explain all heterozygous sequencing calls. Three-loci DRB1/DQA1/DQB1 haplotypes were identified in a similar manner, beginning with homozygous individuals and identifying heterozygote haplotypes by comparing to haplotypes identified in homozygous animals. 9 P-values and odds ratios (OR) with 95% confidence intervals (CI) were calculated for allele and haplotype frequencies in SARDSaffected and control Dachshunds using a 2 × 2 contingency table. Adjustments for multiple comparisons were not performed, as previously recommended for similar studies. 10,11 3 | RESULTS

| Genome-wide association study
Using the inclusion criteria and methods outlined, DNA samples were prospectively collected from 16 Dachshunds No SNPs were of genome-wide statistical significance (P Bonferroni < 0.05 (Figure 1). Six of the nine most statistically significant SNPs were found on canine chromosome (CFA) 12, and five of these six were within the coding region of the major histocompatibility complex (MHC) ( Table  1). Statistically significant genome-wide associations to the MHC are uncommon in dogs; therefore, this association was investigated further by MHC genotyping a larger cohort of SARDS-affected and control Dachshunds.  Figure 2), which combined into 25 different haplotypes (Table 3, Figure 3). A previously unidentified allele (hereafter "DLA-DRB1*DH-Davis") was detected in 2 heterozygous dogs as part of the same novel haplotype (hereafter referred to as "DLA-DRB1*DH-Davis/DQA*003 01/DQB*00401").

| DISCUSSION
In this study, we identified four alleles and two haplotypes across three DLA class II genes (DLA-DRB1, DLA-DQA1, and DLA-DQB1) that demonstrated statistically significant association with SARDS in Dachshunds. The region of interest on CFA12 was initially identified through a GWAS, although the association was not statistically significant. However, as the MHC is one of the most highly polymorphic regions of the genome, a GWAS is very unlikely to identify significance of an association in this region. 12 Therefore, three DLA class II genes were sequenced in a larger population of Dachshunds so as to better identify a true association.
In the present study, the allele DLA-DRB1*09401 and its parent haplotype DLA-DRB1*09401/DQA*00101/DQB-DQBC1 were overrepresented in affected dogs, suggesting a higher risk of SARDS in dogs possessing this allele. However, this haplotype also appeared in control dogs, and only about one third of affected dogs possessed at least one copy of the allele. Therefore, this is inconsistent with a single causative mutation for the disease. Rather, it is possible that this allele confers higher risk of developing SARDS, or is associated with this syndrome for some other reason. By contrast, the alleles DLA-DQB1*00101, DLA-DQA1*00901,  and DLA-DRB-T, as well as the haplotype DLA-DRB-T/ DQA*00101/DQB-DQBC1, were under-represented in affected dogs, suggesting a protective effect. Again, these potentially protective alleles were found in some SARDS-affected dogs, and not all control dogs possessed one of the protective alleles. Even when evaluated in combination with the risk allele, there were SARDS-affected dogs that possessed the protective allele and not the risk allele, as well as control dogs that possessed the risk allele and not the protective allele. These findings reinforce that DLA type does not appear to be solely responsible for susceptibility to developing SARDS. Associations between MHC genes and autoimmune diseases have been reported in humans [13][14][15] as well as animals. [16][17][18] In one study in Doberman pinschers, two of the same DLA alleles suggested by our study to be protective for development of SARDS (DLA-DQA1*00901 and DLA-DQB1*00101) were also potentially protective against Doberman hepatitis. 19 Like SARDS, Doberman hepatitis is a poorly understood disease seen more commonly in female Where GenBank sequence of alignment is not named according to standard practice, accession number is provided in parentheses. A previously unidentified DLA-DRB1 allele is referred to as DRB1*DH-Davis. Alleles in bold type are those whose frequency was significantly higher or lower in SARDS-affected animals. CI, confidence interval; OR, odds ratio.  Despite the identification of some novel genetic associations in the present study, the mechanism by which variations in the MHC might predispose an animal or human to autoimmune disease remains unknown. It has been hypothesized that the role of the MHC in antigen presentation to T lymphocytes during development of immunologic tolerance may be involved, 20-22 but a precise mechanism has not been identified. As the etiology of SARDS is not yet understood, identification of a genetic association between this disease and some DLAs that bears close resemblance to associations found between the MHC and a number of autoimmune diseases is important and supports further investigation of an autoimmune etiology for SARDS. However, spurious associations are often found when genotyping the DLA in this manner, especially in dog breeds with relatively few haplotypes. 23 However, our study identified twenty-five different haplotypes in the population studied, making Dachshunds a more genetically diverse breed than some of those previously studied, 24,25 and lowering the likelihood that this exerted an unwanted effect. Likewise, while the MHC may play a role in SARDS, other genetic and/or environmental factors are likely involved. In order to identify other genetic factors, a larger number of SARDS-affected and control dogs should be assessed in a GWAS. The GWAS performed here was not subject to significant population stratification, and a large number of SNPs were used; however, it had a modest number of samples more appropriate for the identification of a simple recessive locus. 26 In conclusion, we have identified several alleles and haplotypes of DLA class II genes which appear to confer either susceptibility to or protection against development of SARDS in the Dachshund. If these are true associations, their presence supports the hypothesis that SARDS has an autoimmune etiology, and this warrants future research.