December 2015
Volume 56, Issue 13
Free
Lens  |   December 2015
Potential Modifying Loci Associated With Primary Lens Luxation, Pedal Hyperkeratosis, and Ocular Phenotypes in Miniature Bull Terriers
Author Affiliations & Notes
  • Puya Gharahkhani
    Department of Genetic Epidemiology QIMR Berghofer Medical Research Institute, Brisbane, Queensland, Australia
    School of Veterinary Science, University of Queensland, Gatton, Queensland, Australia
  • Caroline A. O'Leary
    Veterinary Medical Centre, School of Veterinary Science, University of Queensland, Gatton, Queensland, Australia
  • David L. Duffy
    Department of Genetic Epidemiology QIMR Berghofer Medical Research Institute, Brisbane, Queensland, Australia
  • Myat Kyaw-Tanner
    School of Veterinary Science, University of Queensland, Gatton, Queensland, Australia
  • Correspondence: Caroline A. O'Leary, Veterinary Medical Centre, School of Veterinary Science, University of Queensland, Gatton 4343, QLD, Australia;c.oleary@uq.edu.au
Investigative Ophthalmology & Visual Science December 2015, Vol.56, 8288-8296. doi:10.1167/iovs.15-18074
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to Subscribers Only
      Sign In or Create an Account ×
    • Get Citation

      Puya Gharahkhani, Caroline A. O'Leary, David L. Duffy, Myat Kyaw-Tanner; Potential Modifying Loci Associated With Primary Lens Luxation, Pedal Hyperkeratosis, and Ocular Phenotypes in Miniature Bull Terriers. Invest. Ophthalmol. Vis. Sci. 2015;56(13):8288-8296. doi: 10.1167/iovs.15-18074.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose: Primary lens luxation (PLL) in dogs is an inherited disease in which the lens is displaced from its normal position. A truncating mutation in the ADAMTS17 orthologue on CFA03 is reported to cause PLL in several breeds, mostly terriers. However, the complex inheritance pattern of PLL in miniature bull terriers (MBTs) suggests that other loci may have a modifying effect on the ADAMTS17 mutation. This study aimed to detect such loci increasing risk of PLL in Australian MBTs.

Methods: More than 170,000 single-nucleotide polymorphisms (SNPs) across the canine genome were genotyped in 23 PLL-affected and 73 normal Australian MBTs, and association between the PLL phenotype and the genetic markers was investigated by using general mixed effects Cox model survival analysis.

Results: The highest association peaks, other than that associated with the ADAMTS17 mutation (P = 2.2e-05), were SNP BICF2G630420272 located at 62.2 Mb on chromosome 15 (P = 7.8e-05) and the region between 30 Mb and 32.5 Mb on chromosome 1 (P = 9.3e-05). Joint analysis showed that the PLL-associated allele of the BICF2G630420272 SNP increased risk of PLL in the presence of the ADAMTS17 mutation (P = 8.117e-04). Candidate genes in the two regions of interest included CPE on chromosome 15 and CTGF on chromosome 1. The ADAMTS17 mutation was also associated with abnormal foot and nail shapes, pedal hyperkeratosis, and persistent pupillary membranes.

Conclusions: Two loci with potentially enhancing effects on the ADAMTS17 mutation were associated with PLL in Australian MBTs. Association of the ADAMTS17 mutation with possible pedal skeletal abnormalities in MBTs supports PLL in this breed and Weill-Marchesani syndrome-like disease in humans as being homologous diseases.

Primary lens luxation (PLL) is an inherited ocular disease in dogs in which the lens detaches from the ciliary processes of the eye and moves anteriorly or posteriorly from its normal position. This disease is common in several terrier breeds such as Jack Russell terriers (JRTs), miniature bull terriers (MBTs), and Tibetan terriers.15 This disease also occurs in nonterrier breeds, such as the Australian cattle dog, border collie, German shepherd, shar pei, and certain spaniel breeds.2,6 Clinical signs of PLL usually appear between 2 and 6 years of age, with the average age range of 4.7 to 5.2 years.3,7 This disease has comparative importance for similar syndromes that involve luxation or subluxation of the lens in humans, including Marfan syndrome (MFS), Weill-Marchesani syndrome (WMS), and WMS-like syndrome.812 
Our previous studies showed that PLL is caused by a truncating mutation in the canine ADAMTS17 orthologue (Entrez Gene ID accession number 488708) in several breeds, mostly terriers.13,14 This is a GT-to-AT splice donor site mutation that results in the skipping of exon 10 during transcription, a shift in the normal reading frame, and a premature stop codon (this mutation will be referred to as the “ADAMTS17 mutation” throughout this paper). Most affected animals were homozygous for the ADAMTS17 mutant allele, consistent with a recessive mechanism of action. However, heterozygotes had a lower risk of developing the disease in breeds such as MBTs, Tenterfield terrier, Parson Russell terrier, and Chinese crested dogs.13,14 In our analysis, which was restricted to Australian Tenterfield terriers and MBTs, the best model describing the disease in these populations was an additive model with the ADAMTS17 mutation having dose- and age-dependent effects.15 These findings suggest that second or modifying loci are contributing to the development of PLL. Alternately, environmental factors are interacting with genetic factors to alter the penetrance and clinical manifestation of the disease. 
Similar syndromes in humans, such as MFS and WMS (Online Mendelian Inheritance in Man [OMIM] database numbers MIM277600 and MIM608328, respectively) can involve ectopia lentis or lens subluxation811; however, these syndromes also involve skeletal or cardiovascular abnormalities. Recently however, three different mutations in ADAMTS17 have been reported to cause only ophthalmic manifestations and short stature and were referred to as WMS-like syndrome.12 Interestingly, some of the MBTs in our study population had abnormally shaped feet and pedal hyperkeratosis and abnormal nail shape, which might have been due to abnormal foot biomechanics from skeletal abnormalities.16 Hence, there could be a skeletal phenotype, as reported in human MFS and WMS, in our population of MBTs. 
In addition, ocular abnormalities including cataract, persistent pupillary membranes (PPM), and keratoconjunctivitis sicca (KCS) were also observed in our population of MBTs. These abnormalities may have a genetic or nongenetic basis and may or may not be associated with the PLL phenotype. Thus, these phenotypes may also be due to the PLL causing mutation in the ADAMTS17 gene or to the possible modifier loci. 
Thus, we aimed to investigate involvement of loci other than the ADAMTS17 mutation in PLL in Australian MBTs by using a genome-wide association study (GWAS). We also aimed to investigate associations between ocular and skeletal/dermatologic phenotypes and the ADAMTS17 mutation and other possible PLL-associated loci. 
Materials and Methods
Animal Ethics
The University of Queensland Animal Ethics Committee approved this study (SVS/089/907[nf], SVS/095/08, SVS/284/10/QCCC/ACAHF/KIBBLETRUST [NF]), as did the University of Queensland Human Ethics Committee (2007000223). 
Selection of Dogs and gDNA Extraction
Ninety-six Australian MBTs from a population in which PLL was segregating were enrolled in the study. Diagnostic criteria are discussed in the Supplementary Material. Except for four dogs, pedigrees were collected for all the dogs in this study. These animals were related within a large pedigree containing 21 unrelated component pedigrees and 77 nuclear families. To ensure that there were no high levels of inbreeding among the animals in this study, mean inbreeding was calculated using Sib-pair64 software (D. Duffy QIMR Berghofer homepage; http://genepi.qimr.edu.au/staff/davidD/#sib-pair, in the public domain) with genotyped single-nucleotide polymorphisms (SNPs) based on a runs-of-homozygosity approach17 within 5 Mbp minimum runs-of-homozygosity lengths. 
DNA Extraction
Genomic DNA was extracted from EDTA-blood by using a QIAamp DNA mini kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. 
Genotyping the ADAMTS17 Mutation
Genotyping for the ADAMTS17 mutation were performed using a TaqMan allelic discrimination assay (Applied Biosystems, Foster City, CA, USA) as described in the previous study.13 
Genome-Wide Screening of SNPs
Genome wide screening of SNPs was performed at the University of Missouri, by using the CanineHD genotyping BeadChip (Illumina, San Diego, CA, USA). This array contains more than 170,000 markers placed on the CanFam 2.0 reference sequence, with an average of greater than 70 markers per megabase (Mb). The SNP genotypes were called using GenomeStudio genotyping module version 1.0 (Illumina) from the raw probe intensity data. Discarded SNPs included those with GenTrain_scores < 0.8, genotyping rates < 95%, deviations from Hardy-Weinberg equilibrium (P < 0.005), minor allele frequency < 0.01, and extreme heterozygosity (false detection rate) < 1%). Similarly, individuals with call rates < 95% and genome-wide identity by descent > 0.95 were also excluded from the analysis (none was present). 
Genome-Wide Association Analysis
Plink-1.07 software (http://pngu.mgh.harvard.edu/∼purcell/plink/index.shtml, in the public domain)18 was used to perform a case-control allelic association analysis using the SNPs and the PLL, PPM, and KCS phenotypes and pedal hyperkeratosis with abnormal foot and nail shape. Sib-pair64 software was used for allelic association analysis by using a gene-dropping simulation approach to correct for population and pedigree structure for related individuals. The Sib-pair64 software was used to correct for multiple testing by setting a significance level for genome-wide association, using the approach described by Moskvina and Schmidt.19 
Survival Analysis
Kinship package from R version 2.14.0 software (http://www.R-project.org, in the public domain)20 was used for general mixed effects Cox model survival analysis for the genome-wide SNPs based on the genetic relationships between individuals. This comprehensive model has been extended to include components of the Cox model, random effects, and familial relationship (kinship or identity by descent [IBD] matrix in families) in the Cox model survival analysis.2123 Kinship IBD matrixes were created, using the genotypic data in kinship software, and Cox model survival analysis was performed on each SNP using the age at diagnosis of PLL and the disease status, whereas the structure of individual random effect was adjusted using the kinship IBD matrixes. Association of the genome-wide SNPs with the disease risk was estimated using the Wald test χ2 for each SNP. 
Survival package24 from R version 2.14.0 software was also used to perform Cox proportional hazards regression model survival analysis and analysis of variance (ANOVA) test from the Cox model survival analysis for strongly associated SNPs to further confirm correlation between SNPs and disease risk. GenABEL package25 from R version 14.0 software was also used for Cox model survival analysis of the genome-wide SNPs. The Kruskal-Wallis test for mean age at first diagnosis of PLL was performed using Sib-pair64 software. This was to investigate differences between mean age at diagnosis of PLL in different groups of genotypes of strongly associated SNPs. R version 2.14.0 software was also used to create a whole genome association plot using the χ2 values obtained from the general mixed effects Cox model survival analysis. The quantile-quantile (Q-Q) plot was created using the snpStats package from R software26 to assess deviation in distribution of χ2 values from the expected distribution under the null hypothesis of no genetic association. 
Estimate of Linkage Disequilibrium
Sib-pair64 software was used to measure linkage disequilibrium (LD) between SNPs by calculating Pearson's correlation (r2) and Lewontin's standardized pairwise LD coefficient (D′). Haploview version 2.4 software (http://www.broadinstitute.org/scientific-community/science/programs/medical-and-population-genetics/haploview/haploview, in the public domain)27 was used to generate pairwise LD plots for SNPs located in the 500-kb region surrounding the SNP with the most significant association with the PLL phenotype. 
Replication of the Association Results in Jack Russell Terriers
To replicate the association results in JRTs for the top hit identified in MBTs in this study, the relevant SNP was imputed in a population of JRTs consisting of 29 cases, 20 controls, and 47 dogs with unknown disease status. Genotyping data for this population, available from our previous study,13 was used as the basis of imputation relative to the reference panel, 12 JRTs genotyped using Canine Genome 2.0 Array (Affymetrix, Santa Clara, CA, USA). Imputation was performed using Beagle version 3.3.2 software.28 
Finding Candidate Genes
The Ensembl database (http://www.ensembl.org/index.html, in the public domain) was used to identify genes present in regions showing association with the PLL phenotype. Candidate genes were selected based on function, tissue location of the expressed gene products, and whether similar phenotypes were reported to be caused by mutations in these genes. This information was found in databases such as Genecards (http://www.genecards.org/, in the public domain), UniprotKB (http://www.uniprot.org/, in the public domain), and Entrez (http://www.ncbi.nlm.nih.gov/sites/entrez, in the public domain). 
Results
Disease Status
Out of the 96 dogs enrolled in this study, PLL was diagnosed in 23, PPM in 10, 4 had cataracts, and 5 had KCS. In addition, pedal hyperkeratosis and abnormal foot and nail shape were diagnosed in four dogs. Sixty-five dogs were A/G heterozygotes, 23 dogs were A/A homozygotes, and 8 dogs were G/G homozygotes for the PLL-causing ADAMTS17 mutation. Among the 23 PLL-affected dogs, 12 were A/A homozygote, 10 were A/G heterozygote, and 1 was G/G homozygote for the PLL-causing ADAMTS17 mutation. Mean inbreeding for the animals used in this study was estimated at 0.3086. This estimate was not significantly different between cases and controls (P = 0.9417). 
Genome-Wide Association Analysis
After SNPs were filtered for quality, 47,642 SNPs remained for genome-wide association analysis. A significance level of P < 10−6 was set for genome-wide association in this study, using the approach by Moskvina and Schmidt.19 A preliminary case-control association analysis of PLL, ignoring age at onset information, was shown to be unable to detect the known ADAMTS17 mutation or flanking markers analyzed using Plink software (data not shown). As our previous study had demonstrated age dependence for the PLL phenotype in Australian MBTs, with increased disease risk with advancing age,15 survival analysis was used to consider the effect of age on PLL in the subsequent analyses. 
Survival Analysis
General mixed effects Cox model survival analysis for the genome-wide SNPs allowing for the effect of genetic relationships between individuals found SNPs located between 38.0 Mb and 43.6 Mb on chromosome 3 had the highest association with the PLL phenotype. Within this region, the ADAMTS17 mutation, located at 43.5 Mb, showed the most significant association (P = 2.2e-05) (Supplementary Table S1). Peak associations were also observed on chromosome 15 at locus 62.2 Mb (P = 7.8e-05) and on chromosome 1 for SNPs between 30 Mb and 32.5 Mb (best P = 9.3e-05) (Fig. 1). The Q-Q plot created using snpStats software for these results is shown in Figure 2
Figure 1
 
Manhattan plot for genome-wide SNPs. This plot was created using R version 2.14.0 software from the Wald test χ2 values obtained from the mixed effects Cox model survival analysis for investigating associations between genome-wide SNPs and development of PLL. Each dot represents one SNP. Single nucleotide polymorphisms have been plotted against their chromosomal positions (x-axis), and χ2 values were calculated through survival analysis (y-axis). Chromosomes are shown in numerical order and are separated by vertical lines. All SNPs on each chromosome have been shown in the same color but a distinct color from that of the adjacent chromosome in the figure. Higher χ2 results indicate higher associations with the disease risk. Single nucleotide polymorphisms with the highest association peaks are on chromosomes 3, 15, and 1.
Figure 1
 
Manhattan plot for genome-wide SNPs. This plot was created using R version 2.14.0 software from the Wald test χ2 values obtained from the mixed effects Cox model survival analysis for investigating associations between genome-wide SNPs and development of PLL. Each dot represents one SNP. Single nucleotide polymorphisms have been plotted against their chromosomal positions (x-axis), and χ2 values were calculated through survival analysis (y-axis). Chromosomes are shown in numerical order and are separated by vertical lines. All SNPs on each chromosome have been shown in the same color but a distinct color from that of the adjacent chromosome in the figure. Higher χ2 results indicate higher associations with the disease risk. Single nucleotide polymorphisms with the highest association peaks are on chromosomes 3, 15, and 1.
Figure 2
 
Q-Q plot. Q-Q plot was created, using the snpStats package in R software, from the results of the mixed effects Cox model survival analysis for investigating associations between genome-wide SNPs and development of PLL. x-axis = expected distribution of χ2 values under the null hypothesis of no association; y-axis = observed χ2 values calculated in the survival analysis. Each dot making the curve is an observed χ2 value calculated for one or more SNPs. The indicator line shows where x = y. Deviation from the indicator line in this graph indicates more significant associations were observed in this study than were expected by chance.
Figure 2
 
Q-Q plot. Q-Q plot was created, using the snpStats package in R software, from the results of the mixed effects Cox model survival analysis for investigating associations between genome-wide SNPs and development of PLL. x-axis = expected distribution of χ2 values under the null hypothesis of no association; y-axis = observed χ2 values calculated in the survival analysis. Each dot making the curve is an observed χ2 value calculated for one or more SNPs. The indicator line shows where x = y. Deviation from the indicator line in this graph indicates more significant associations were observed in this study than were expected by chance.
Cox proportional hazards regression model survival analysis for the BICF2G630420272 SNP on chromosome 15 showed that BICF2G630420272 was significantly correlated with disease risk, as estimated by the Wald test (P = 7.61e-05 for the regression coefficient), score (logrank) test (P = 1.912e-05), and likelihood ratio test (P = 7.606e-05). The same analysis using genotypes of the BICF2G630420272 SNP suggested that the B/B genotype was strongly significantly correlated with PLL risk (P = 3.68e-05), whereas the A/B genotype was less significantly correlated (P = 0.019). Testing for survival curve differences for the different genotypes of the BICF2G630420272 SNP also showed the risk of PLL was significantly different for different genotypes of this SNP (P = 5.46e-06). ANOVA test from the Cox model survival analysis also supported the correlation of SNP BICF2G630420272 with an increased risk of developing clinical PLL (P = 6.825e-05 for reduction in log-likelihood). Including the ADAMTS17 mutation as a covariate in the ANOVA test showed the reduction in log-likelihood was still significant after considering the effect of the ADAMTS17 mutation. This demonstrated a significant effect for the BICF2G630420272 SNP on PLL phenotype in addition to the effect of the ADAMTS17 mutation (P = 5.881e-06 for the ADAMTS17 mutation; P = 8.117e-04 for SNP BICF2G630420272, calculated by subtracting the estimated log-likelihood for SNP BICF2G630420272 from the estimated log-likelihood for the ADAMTS17 mutation). 
Similar analyses performed for the BICF2P405685 SNP, the SNP with the highest association peak on chromosome 1, confirmed that this SNP was also significantly correlated with the disease risk (data not shown). However, the results were less significant compared to those of the SNP BICF2G630420272
The disease survival plot based on genotypes of the ADAMTS17 mutation and SNP BICF2G630420272 showed animals with the A/A and A/G genotypes for the ADAMTS17 mutation had increased risk for clinical PLL when they were B/B or A/B for the BICF2G630420272 SNP. The highest risk for the disease phenotype was observed in dogs A/A at the ADAMTS17 locus and B/B at the BICF2G630420272 SNP. All such animals were clinically affected by PLL by 5 years of age (Fig. 3). Similarly, the disease risk was 100% for A/A animals at ADAMTS17 and A/B animals for BICF2G630420272 SNP when they were 5½ years of age. A 100% disease risk for the A/A animals at both these SNPs was observed by 8 years of age. Animals heterozygous at the ADAMS17 locus had a variable risk of clinical PLL varying with BICF2G630420272 SNP genotype; A/A animals at the SNP had an 80% disease risk by 12 years of age, A/B animals had a 100% risk by 10 years of age, and B/B animals had a 100% risk by 6 years of age. The proportion of PLL affected and unaffected animals by genotype for the ADAMTS17 mutation and BICF2G630420272 SNPs are shown in Table 1
Figure 3
 
Survival plots are shown for PLL in MBTs according to their genotypes for the ADAMTS17 and BICF2G630420272 SNPs. Colors for each line indicate different genotypes for those SNPs. Genotypes have been shown beside each line; the first genotype is the genotype for the ADAMTS17 mutation, and the second genotype is for the BICF2G630420272 SNP.
Figure 3
 
Survival plots are shown for PLL in MBTs according to their genotypes for the ADAMTS17 and BICF2G630420272 SNPs. Colors for each line indicate different genotypes for those SNPs. Genotypes have been shown beside each line; the first genotype is the genotype for the ADAMTS17 mutation, and the second genotype is for the BICF2G630420272 SNP.
Table 1
 
Proportions of PLL Affected and Unaffected Dogs Based on Their Genotypes for the ADAMTS17 and BICF2G630420272 SNPs
Table 1
 
Proportions of PLL Affected and Unaffected Dogs Based on Their Genotypes for the ADAMTS17 and BICF2G630420272 SNPs
Similarly, survival plot for the SNP BICF2P405685 on chromosome 1 showed the PLL-associated allele of this SNP was associated with decreased age of onset of PLL in A/A or A/G animals for the ADAMTS17 mutation (data not shown). However, unlike the PLL-associated allele on chromosome 15, this allele did not increase penetrance of the disease in A/G heterozygotes for the ADAMTS17 mutation. 
Kruskal-Wallis testing showed significant differences between mean age at diagnosis of PLL in different genotype groups of dogs according to their ADAMTS17 mutation and BICF2G630420272 SNP genotypes (P = 0.0085; df = 7) (Table 2). 
Table 2
 
Mean Age at First Diagnosis of PLL in Dog Groups According to Their Genotypes for the ADAMTS17 and the BICF2G630420272 SNPs
Table 2
 
Mean Age at First Diagnosis of PLL in Dog Groups According to Their Genotypes for the ADAMTS17 and the BICF2G630420272 SNPs
Estimate of Linkage Disequilibrium
Analysis of linkage disequilibrium using Sib-pair64 software showed SNPs with statistically significant evidence of LD (P < 9.0e-04) with SNP BICF2G630420272 were between 52.0 Mb and 65.9 Mb on chromosome 15. Three adjacent haplotype blocks with high multiallelic LD scores (D′ > 0.9) were identified by Haploview version 4.2 software in the 500-kb region flanking SNP BICF2G630420272 (Fig. 4). 
Figure 4
 
Linkage disequilibrium plot is shown for a 500-kb region surrounding the BICF2G630420272 SNP on CFA 15. Single nucleotide polymorphisms are numbered in order at the top, and the SNP highlighted inside the box is the BICF2G630420272 SNP. Each diagonal row represents a different SNP, and each square represents a pairwise comparison between two SNPs. Single nucleotide polymorphisms are arranged in haplotype blocks (thick black triangles) using the “spine of LD” algorithm in Haploview software. Red squares indicate statistically significant (LOD > 2) LD between the pair of SNPs based on the D′ statistic. Darker colors of red indicate higher values of D′, up to a maximum of 1. D′ is a measure of LD ranging from 0 to 1 and is inversely related to the fraction of chromosomes that have experienced historical recombination. D′ = 1 indicates complete LD, and D′ = 0 represents complete linkage equilibrium. White squares indicate pairwise D′ values of <1 with no statistically significant evidence of LD. Blue squares indicate pairwise D′ values of 1 but without statistical significance. Adjacent blocks are merged if they have multiallelic D′ values of at least 0.9, and at least 80% of the chromosome in the resulting merged block is explained by six or fewer common haplotypes.
Figure 4
 
Linkage disequilibrium plot is shown for a 500-kb region surrounding the BICF2G630420272 SNP on CFA 15. Single nucleotide polymorphisms are numbered in order at the top, and the SNP highlighted inside the box is the BICF2G630420272 SNP. Each diagonal row represents a different SNP, and each square represents a pairwise comparison between two SNPs. Single nucleotide polymorphisms are arranged in haplotype blocks (thick black triangles) using the “spine of LD” algorithm in Haploview software. Red squares indicate statistically significant (LOD > 2) LD between the pair of SNPs based on the D′ statistic. Darker colors of red indicate higher values of D′, up to a maximum of 1. D′ is a measure of LD ranging from 0 to 1 and is inversely related to the fraction of chromosomes that have experienced historical recombination. D′ = 1 indicates complete LD, and D′ = 0 represents complete linkage equilibrium. White squares indicate pairwise D′ values of <1 with no statistically significant evidence of LD. Blue squares indicate pairwise D′ values of 1 but without statistical significance. Adjacent blocks are merged if they have multiallelic D′ values of at least 0.9, and at least 80% of the chromosome in the resulting merged block is explained by six or fewer common haplotypes.
Association Analysis for Alternate Phenotypes
Association analysis using Plink showed that the ADAMTS17 mutation was significantly associated with abnormal pedal hyperkeratosis and foot and nail shape (P = 0.002141) and slightly associated with PPM (P = 0.03626) but not with cataract (P = 0.8859) and KCS (P = 0.335). Segregation of this mutation with these phenotypes is illustrated in Supplementary Figure S2. In contrast, Cox model survival analysis for the genome-wide SNP data using GenABEL software showed peak association for cataract presence was on chromosome 15 in the region between 58.6 Mb and 58.8 Mb (P = 0.004). This is approximately 3.4 Mb upstream of the BICF2G630420272 SNP that was found to be associated with the PLL phenotype. However, the ADAMTS17 mutation was not associated with cataract in survival analysis (P = 0.261). None of the other phenotypes were associated with the new PLL-associated loci identified in this study (data not shown). 
Replication of Association Results in Jack Russell Terriers
We could not investigate association of BICF2G630420272 with PLL in JRTs because BICF2G630420272 or the other SNPs in high LD with this SNP were not accurately imputed in the population of JRTs in this study (BICF2G630420272 was imputed as a monomorphic SNP). Populations of JRTs and reference panels with higher SNP density in the region of BICF2G630420272 may be required for an accurate imputation of this SNP. We also could not use MBTs or the other breeds as a reference panel for imputation of BICF2G630420272 in JRTs. This was because allele frequencies of BICF2G630420272 were different between JRTs and the other breeds genotyped for this SNP. For example, although the risk allele had a frequency of 0.33 in the population of MBTs in this study, the frequency of this allele was 0.80 in the JRT reference panel. Similarly, in our previous study (unpublished conference data),29 this allele had frequencies of 0.15, 0.14, and 0.00 in golden retrievers, German shepherds, and pugs, respectively. 
Finding Candidate Genes
From the 27 protein-encoding genes, 4 pseudogene, and 11 RNA genes (Supplementary Fig. S3) identified on canine chromosome 15 in the 10-Mb area flanking the SNP BICF2G630420272, the carboxypeptidase E (CPE) gene, which is located approximately 2.2 Mb downstream from SNP BICF2G630420272, was selected as a good candidate gene for bioinformatic analysis. 
From the 57 protein-encoding genes, 7 pseudogenes, and 21 RNA genes (Supplementary Fig. S3) identified in the 10 Mb area flanking the region of association with the PLL phenotype on canine chromosome 1, the connective tissue growth factor (CTGF) gene on chromosome 1 located approximately 2.1 Mb upstream from the best SNP on chromosome 1 was selected as a good candidate gene for bioinformatic analysis. 
Unfortunately, no SNP data were available inside the predicted canine CPE orthologue, so LD could not be estimated between SNP BICF2G630420272 and markers inside the gene. The closest marker to the CPE gene was located 12.8 kb upstream of CPE, and a D′ = 0.566 and P = 0.0025 were calculated for LD between this marker and the BICF2G630420272 SNP. Most of the other markers 500 kb upstream from CPE showed similar LD with the BICF2G630420272 SNP. 
However, data were available for one SNP located inside the predicted canine CTGF orthologue. A D′ = 0.126 and P = 0.367 were calculated for the LD between this SNP and the SNP with the highest peak on chromosome 1. However, a D′ = 1.00 and P = 1e-6 was calculated for the other SNPs 500 kb downstream from this gene. 
Discussion
This study found that the clinical PLL phenotype in Australian MBTs may be affected by a modifier locus close to SNP BICF2G630420272 located at 62.2 Mb on chromosome 15. In addition, a region located between 30.0 Mb and 32.5 Mb on chromosome 1 was also associated with PLL in MBTs, with SNP BICF2P405685 showing the highest association in this region. 
The methods used to detect associations in this study were based on survival analysis because no strong association was detected between the PLL phenotype and the ADAMTS17 mutation or its flanking SNPs using standard case-control GWAS and allelic association analysis. This may be because these approaches do not take into account the age-dependent effects of disease-causative SNPs on phenotypes. As our previous study showed, the ADAMTS17 mutation had age- and dose-dependent effects on the clinical manifestation of PLL in MBTs15; the inclusion of these important variables allowed detection of the effects of potential modifier loci. This was done by including the effect of age in the analysis by using a general mixed effects Cox model survival analysis and calculation of χ2 values for each SNP using the Wald test. Using this method of analysis, the peak association between the PLL phenotype remained with the ADAMTS17 mutation and its flanking SNPs. This evidence demonstrates results obtained from survival analyses were not false positive results and confirmed that because of the effect of age on development of clinical PLL, survival analysis is a more appropriate approach than standard allelic association to investigate association between SNPs and the PLL phenotype. 
The influence of age on development of PLL suggests that, although the ADAMTS17 mutation might have its effects from young ages, zonule defects remained clinically latent until the zonule broke with aging. This may explain why PLL usually occurs after 2 to 3 years of age in dogs. However, results of our previous study15 and the current study also suggest that the dosage of the ADAMTS17 mutation as well as modifier loci such as those identified in this study might have effects on development of PLL and age of onset of the disease in breeds such as MBTs. 
The P values obtained in this study did not meet the significance level of P < 10−6 that was set for genome-wide association in this study, using the approach by Moskvina and Schmidt.19 However, results for the ADAMTS17 mutation (P = 2.2e-05), the causative mutation for PLL, also did not meet this significance level. This is probably because the sample size used in this study was not large enough to meet the significance level for detection of associations. On the other hand, we should expect to detect lower association powers for secondary loci with small effects on the development of PLL. 
Using our available data, we could not accurately impute the BICF2G630420272 SNPs in JRTs to replicate the association of this SNP with PLL in JRTs. Further studies may be required to genotype this SNP in JRTs or impute this SNP using higher SNP density in the region of BICF2G630420272 in large populations of JRTs and a reference panel. 
A candidate gene in the region of association with PLL identified by the current study on CFA15 was the CPE orthologue. This gene is located approximately 2.2 Mb downstream from the association peak on chromosome 15. LD between the peak SNP on chromosome 15 and SNPs flanking the CPE orthologue supported the possibility that these markers were on the same haplotype. However, the LD was weak based on the calculated D′ and P values. Alternatively, the peak on chromosome 15 may represent a distant modifier of CPE, ADAMTS17, or other genetic element involved in the development of the PLL phenotype. This is supported by studies in humans where GWAS peaks associated with vascular diseases and chronic obstructive pulmonary disease have not been associated with a causative gene but the regulatory elements of these genes. Such elements may be close to or remote from the causative gene mutations.3035 
The function and sites of tissue expression for the canine CPE orthologue have not been studied. The CPE gene in humans is expressed in tissues such as brain and endocrine systems,36,37 as well as the eye, where expression occurs in the adult lens, nonpigmented ciliary epithelial cells, and aqueous humor.3840 The protein has carboxypeptidase, hydrolase, metallopeptidase, and protease activities in species including humans, mice, and cattle (UniProt and GeneCard). Interestingly, the ADAMTS17 gene product also has hydrolysis, metallopeptidase, and protease activities in humans (UniProt and GeneCard). 
The second region of association with PLL identified in this current study in Australian MBTs contained the canine orthologue of the CTGF gene. This gene could also be a candidate for involvement in the PLL phenotype in this study population and is located approximately 2.1 Mb upstream from the association peak on chromosome 1. This gene is expressed in a variety of ocular tissues in humans, including trabecular meshwork cells, ciliary body, cataractous plaque, and corneal scars (Deng P, et al. IOVS 2002;43:ARVO E-Abstract 1032).4148 Increased expression of CTGF has been reported in humans with ocular abnormalities involving the lens and zonular fibers. For example, increased expression of CTGF mRNA occurs in lens epithelial cells in anterior subcapsular cataracts in humans.4951 Similarly, increased CTGF expression is reported in the aqueous humor of humans with pseudoexfoliation syndrome, where weakened zonular fibres also occur in the ciliary body, iris, trabecular meshwork, and corneal endothelium.52,53 Similar histology has been reported in pseudoexfoliation syndrome and PLL, with weak zonular fibers and accumulation of abnormal fibrillar material reported.4,52,54 This evidence supports a possible role for CTGF in lens luxation. 
This study found an association between the causative PLL ADAMTS17 mutation and abnormal pedal hyperkeratosis, abnormal foot and nail shape, and persistent pupillary membranes. Lens luxation, or ectopia lentis, is found in humans with MFS and WMS (OMIM numbers MIM277600 and MIM608328, respectively). However, these syndromes also involve other abnormalities, including skeletal or cardiovascular defects.8,11,12 Recently, homozygous truncating mutations in ADAMTS17 have been reported in a WMS-like syndrome in humans, with the only clinical manifestations involving ophthalmic abnormalities and short stature.12 Hence, there are similarities between PLL in MBTs and WMS-like syndrome in humans. However, as there were only four MBTs with abnormal pedal hyperkeratosis and abnormal foot and nail shape in this study, further studies with increased numbers of clinically PLL affected dogs and dogs with similar foot and nail abnormalities would be required to further investigate the effect of the ADAMTS17 mutation on these abnormalities in dog. 
One of the limitations of this study was the small sample size. Further association studies using greater numbers of dogs and markers in regions of interest could allow fine mapping of possible PLL-associated loci identified in this study. Sequencing of exonic, intronic, and regulatory sequences (if known) of the candidate genes suggested in affected and unaffected animals, and investigations into expression levels of those genes in affected and normal tissues may also help to identify disease associated mutations that increase the risk of developing clinical PLL in MBTs. 
In conclusion, two potential modifying loci for the PLL-causing ADAMTS17 mutation have been identified in Australian MBTs. These loci may be genes with small effects or regulatory sequences of the genes having an effect on the way the ADAMTS17 mutation causes the PLL phenotype. Furthermore, the ADAMTS17 mutation may also be associated with abnormal pedal hyperkeratosis and abnormal foot and nail shape phenotypes in MBTs. This supports similarities between PLL in MBTs and WMS like disease in humans. 
Acknowledgments
The authors thank Gary Johnson, DVM, PhD, and Fabiana H.G. Farias, PhD, for performing ADAMTS17 mutation testing and Illumina canine SNP array analysis; veterinary ophthalmologists Michael Bernays, BVSc (Hons), FACVSc; Edith Hampson, BVSc, PhD, FANZCVS; Anna Deykin, BVSc, BVMS, FANZCVS; Mark Billson, BVSc, PhD, DVOpthalDipVetClinStud, MRCVS; Anita Dutton, DVM, FANZCVS; Andrew Turner, BVSc, MANZCVS, FANZCVS; Bruce Robertson, BVSc, CertVOpthal (RCVS), FANZCVS; Rick Read, BVSc, DVOphthal, DECVO, FACVSc, FRCVS; Cameron Whittaker, BVSc, FACVSc, Diplomate ACVO; and Ziggy Chester, BVSc, FANZCVS; and Dogs Queensland (previously The Queensland Canine Control Council). 
Supported by Dogs Queensland (previously The Queensland Canine Control Council) and the Dolby Bequest. 
Disclosure: P. Gharahkhani, None; C.A. O'Leary, None; D.L. Duffy, None; M. Kyaw-Tanner, None 
References
Curtis R, Barnett KC. Primary lens luxation in the dog. J Small Anim Pract. 1980; 21: 657–668.
Curtis R. Lens luxation in the dog and cat. Vet Clin North Am Small Anim Pract. 1990; 20: 755–773.
Ketteritzsch K, Hamann H, Brahm R, Grussendorf H, Rosenhagen CU, Distl O. Genetic analysis of presumed inherited eye diseases in Tibetan terriers. Vet J. 2004; 168: 151–159.
Curtis R. Aetiopathological aspects of inherited lens dislocation in the Tibetan terrier. J Comp Pathol. 1983; 93: 151–163.
Chandler EA. Lens luxation in the Webster terrier. Vet Rec. 1970; 86: 145–146.
Lazarus JA, Pickett JP, Champagne ES. Primary lens luxation in the Chinese shar pei: clinical and hereditary characteristics. Vet Ophthalmol. 1998; 1: 101–107.
Curtis R, Barnett KC, Lewis SJ. Clinical and pathological observations concerning the aetiology of primary lens luxation in the dog. Vet Rec. 1983; 112: 238–246.
Faivre L, Dollfus H, Lyonnet S, et al. Clinical homogeneity and genetic heterogeneity in Weill-Marchesani syndrome. Am J Med Genet A. 2003; 123A: 204–207.
Ades LC, Holman KJ, Brett MS, Edwards MJ, Bennetts B. Ectopia lentis phenotypes and the FBN1 gene. Am J Med Genet A. 2004; 126A: 284–289.
Dagoneau N, Benoist-Lasselin C, Huber C, et al. ADAMTS10 mutations in autosomal recessive Weill-Marchesani syndrome. Am J Hum Genet. 2004; 75: 801–806.
Gonzales EA. Marfan syndrome. J Am Acad Nurse Pract. 2009; 21: 663–670.
Morales J, Al-Sharif L, Khalil DS, et al. Homozygous mutations in ADAMTS10 and ADAMTS17 cause lenticular myopia, ectopia lentis, glaucoma, spherophakia, and short stature. Am J Hum Genet. 2009; 85: 558–568.
Farias FHG, Johnson GS, Taylor JF, et al. An ADAMTS17 splice donor site mutation in dogs with primary lens luxation. Invest Ophthalmal Vis Sci. 2010; 51: 4716–4721.
Gould D, Pettitt L, McLaughlin B, et al. ADAMTS17 mutation associated with primary lens luxation is widespread among breeds. Vet Ophthalmol. 2011; 14: 378–384.
Gharahkhani P, O'Leary C, Duffy D, Bernays M, Kyaw-Tanner M. Primary lens luxation in Australian Tenterfield and miniature bull terriers is due to an old ADAMTS17 mutation and is an additive trait. Open Genomics J. 2012; 5: 7–13.
Gibbs RC, Boxer MC. Abnormal biomechanics of feet and their cause of hyperkeratoses. J Am Acad Dermatol. 1982; 6: 1061–1069.
Keller MC, Visscher PM, Goddard ME. Quantification of inbreeding due to distant ancestors and its detection using dense single nucleotide polymorphism data. Genetics. 2011; 189: 237–249.
Purcell S, Neale B, Todd-Brown K, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007; 81: 559–575.
Moskvina V, Schmidt KM. On multiple-testing correction in genome-wide association studies. Genet Epidemiol. 2008; 32: 567–573.
Atkinson B, Therneau T. Kinship: mixed-effects Cox models, sparse matrices, and modeling data from large pedigrees. R package version 1.1.0-21. Available at: http://cran.rproject.org/web/packages/kinship/index.html. Accessed January 23, 2012.
Therneau T. On mixed-effect Cox models sparse matrices, and modeling data from large pedigrees. Available at: http://mayoresearch.mayo.edu/mayo/research/biostat/upload/kinship.pdf. Accessed January 23, 2012.
Anderson CA, Duffy DL, Martin NG, Visscher PM. Estimation of variance components for age at menarche in twin families. Behav Genet. 2007; 37: 668–677.
Ruiz Giolo S, Pereira AC, de Andrade M, de Oliveira CM, Krieger JE, Soler JM. Genetic analysis of age-at-onset for cardiovascular risk factors in a Brazilian family study. Hum Hered. 2009; 68: 131–138.
Therneau T. A package for survival analysis in S. R package version 2.36-12. 2-02. Available at: https://cran.r-project.org/src/contrib/Archive/survival/. Accessed January 23 2012.
Aulchenko YS, Ripke S, Isaacs A, Duijn CMv. GenABEL: an R package for genome-wide association analysis. Bioinformatics. 2007; 23: 1294–1296.
Sole X, Guino E, Valls J, Iniesta R, Moreno V. SNPStats: a web tool for the analysis of association studies. Bioinformatics. 2006; 22: 1928–1929.
Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005; 21: 263–265.
Browning BL, Browning SR. A unified approach to genotype imputation and haplotype-phase inference for large data sets of trios and unrelated individuals. Am J Hum Genet. 2009; 84: 210–223.
Duffy D, O'Leary C, Hill M, Bielefeldt-Ohmann H, Straw R. Intrabreed and interbreed GWAS for dog phenotypes including hemangiosarcoma. 9th GeneMappers Conference 2012. Port Douglas, Australia; 2012. Available at: http://genemappers-2012.p.asnevents.com.au/days/2012-08-27/abstract/1606.
Yang X. Use of functional genomics to identify candidate genes underlying human genetic association studies of vascular diseases. Arterioscler Thromb Vasc Biol. 2012; 32: 216–222.
Zhou X, Baron RM, Hardin M, et al. Identification of a chronic obstructive pulmonary disease genetic determinant that regulates HHIP. Hum Mol Genet. 2011; 21: 1325–1335.
Cheung VG, Spielman RS, Ewens KG, Weber TM, Morley M, Burdick JT. Mapping determinants of human gene expression by regional and genome-wide association. Nature. 2005; 437: 1365–1369.
Moffatt MF, Kabesch M, Liang L, et al. Genetic variants regulating ORMDL3 expression contribute to the risk of childhood asthma. Nature. 2007; 448: 470–473.
Stranger BE, Nica AC, Forrest MS, et al. Population genomics of human gene expression. Nat Genet. 2007; 39: 1217–1224.
Loos RJ, Lindgren CM, Li S, et al. Common variants near MC4R are associated with fat mass, weight and risk of obesity. Nat Genet. 2008; 40: 768–775.
Oiso S, Takeda Y, Futagawa T, et al. Contactin-associated protein (Caspr) 2 interacts with carboxypeptidase E in the CNS. J Neurochem. 2009; 109: 158–167.
Johnson JD. Proteomic identification of carboxypeptidase E connects lipid-induced beta-cell apoptosis and dysfunction in type 2 diabetes. Cell Cycle. 2009; 8: 38–42.
Wistow G. The NEIBank project for ocular genomics: data-mining gene expression in human and rodent eye tissues. Prog Retin Eye Res. 2006; 25: 43–77.
Ortego J, Escribano J, Crabb J, Coca-Prados M. Identification of a neuropeptide and neuropeptide-processing enzymes in aqueous humor confers neuroendocrine features to the human ocular ciliary epithelium. J Neurochem. 1996; 66: 787–796.
Coca-Prados M, Escribano J, Ortego J. Differential gene expression in the human ciliary epithelium. Prog Retin Eye Res. 1999; 18: 403–429.
Wunderlich K, Senn BC, Todesco L, Flammer J, Meyer P. Regulation of connective tissue growth factor gene expression in retinal vascular endothelial cells by angiogenic growth factors. Graefes Arch Clin Exp Ophthalmol. 2000; 238: 910–915.
Hinton DR, He S, Jin ML, Barron E, Ryan SJ. Novel growth factors involved in the pathogenesis of proliferative vitreoretinopathy. Eye (Lond). 2002; 16: 422–428.
He S, Jin ML, Worpel V, Hinton DR. A role for connective tissue growth factor in the pathogenesis of choroidal neovascularization. Arch Ophthalmol. 2003; 121: 1283–1288.
Wunderlich K, Pech M, Eberle AN, Mihatsch M, Flammer J, Meyer P. Expression of connective tissue growth factor (CTGF) mRNA in plaques of human anterior subcapsular cataracts and membranes of posterior capsule opacification. Curr Eye Res. 2000; 21: 627–636.
Lee EH, Joo CK. Role of transforming growth factor-beta in transdifferentiation and fibrosis of lens epithelial cells. Invest Ophthalmol Vis Sci. 1999; 40: 2025–2032.
Wunderlich K, Senn BC, Reiser P, Pech M, Flammer J, Meyer P. Connective tissue growth factor in retrocorneal membranes and corneal scars. Ophthalmologica. 2000; 214: 341–346.
van Setten GB, Blalock TD, Grotendorst G, Schultz GS. Detection of connective tissue growth factor (CTGF) in human tear fluid: preliminary results. Acta Ophthalmol Scand. 2003; 81: 51–53.
van Setten G, Aspiotis M, Blalock TD, Grotendorst G, Schultz G. Connective tissue growth factor in pterygium: simultaneous presence with vascular endothelial growth factor - possible contributing factor to conjunctival scarring. Graefes Arch Clin Exp Ophthalmol. 2003; 241: 135–139.
Pau H, Novotny GE, Arnold G. Ultrastructural investigation of extracellular structures in subcapsular white corrugated cataract (anterior capsular cataract). Graefes Arch Clin Exp Ophthalmol. 1985; 223: 96–100.
Hatae T, Ishibashi T, Yoshitomi F, Shibata Y. Immunocytochemistry of types I-IV collagen in human anterior subcapsular cataracts. Graefes Arch Clin Exp Ophthalmol. 1993; 231: 586–590.
Novotny GE, Pau H. Myofibroblast-like cells in human anterior capsular cataract. Virchows Arch A Pathol Anat Histopathol. 1984; 404: 393–401.
Schlotzer-Schrehardt U, Naumann GO. A histopathologic study of zonular instability in pseudoexfoliation syndrome. Am J Ophthalmol. 1994; 118: 730–743.
Gasch AT, Aghabayova Z, Seidova S. Exfoliation syndrome among Azerbaijani. Arch Ophthalmol. 2003; 121: 920.
Morris RA, Dubielzig RR. Light-microscopy evaluation of zonular fiber morphology in dogs with glaucoma: secondary to lens displacement. Vet Ophthalmol. 2005; 8: 81–84.
Figure 1
 
Manhattan plot for genome-wide SNPs. This plot was created using R version 2.14.0 software from the Wald test χ2 values obtained from the mixed effects Cox model survival analysis for investigating associations between genome-wide SNPs and development of PLL. Each dot represents one SNP. Single nucleotide polymorphisms have been plotted against their chromosomal positions (x-axis), and χ2 values were calculated through survival analysis (y-axis). Chromosomes are shown in numerical order and are separated by vertical lines. All SNPs on each chromosome have been shown in the same color but a distinct color from that of the adjacent chromosome in the figure. Higher χ2 results indicate higher associations with the disease risk. Single nucleotide polymorphisms with the highest association peaks are on chromosomes 3, 15, and 1.
Figure 1
 
Manhattan plot for genome-wide SNPs. This plot was created using R version 2.14.0 software from the Wald test χ2 values obtained from the mixed effects Cox model survival analysis for investigating associations between genome-wide SNPs and development of PLL. Each dot represents one SNP. Single nucleotide polymorphisms have been plotted against their chromosomal positions (x-axis), and χ2 values were calculated through survival analysis (y-axis). Chromosomes are shown in numerical order and are separated by vertical lines. All SNPs on each chromosome have been shown in the same color but a distinct color from that of the adjacent chromosome in the figure. Higher χ2 results indicate higher associations with the disease risk. Single nucleotide polymorphisms with the highest association peaks are on chromosomes 3, 15, and 1.
Figure 2
 
Q-Q plot. Q-Q plot was created, using the snpStats package in R software, from the results of the mixed effects Cox model survival analysis for investigating associations between genome-wide SNPs and development of PLL. x-axis = expected distribution of χ2 values under the null hypothesis of no association; y-axis = observed χ2 values calculated in the survival analysis. Each dot making the curve is an observed χ2 value calculated for one or more SNPs. The indicator line shows where x = y. Deviation from the indicator line in this graph indicates more significant associations were observed in this study than were expected by chance.
Figure 2
 
Q-Q plot. Q-Q plot was created, using the snpStats package in R software, from the results of the mixed effects Cox model survival analysis for investigating associations between genome-wide SNPs and development of PLL. x-axis = expected distribution of χ2 values under the null hypothesis of no association; y-axis = observed χ2 values calculated in the survival analysis. Each dot making the curve is an observed χ2 value calculated for one or more SNPs. The indicator line shows where x = y. Deviation from the indicator line in this graph indicates more significant associations were observed in this study than were expected by chance.
Figure 3
 
Survival plots are shown for PLL in MBTs according to their genotypes for the ADAMTS17 and BICF2G630420272 SNPs. Colors for each line indicate different genotypes for those SNPs. Genotypes have been shown beside each line; the first genotype is the genotype for the ADAMTS17 mutation, and the second genotype is for the BICF2G630420272 SNP.
Figure 3
 
Survival plots are shown for PLL in MBTs according to their genotypes for the ADAMTS17 and BICF2G630420272 SNPs. Colors for each line indicate different genotypes for those SNPs. Genotypes have been shown beside each line; the first genotype is the genotype for the ADAMTS17 mutation, and the second genotype is for the BICF2G630420272 SNP.
Figure 4
 
Linkage disequilibrium plot is shown for a 500-kb region surrounding the BICF2G630420272 SNP on CFA 15. Single nucleotide polymorphisms are numbered in order at the top, and the SNP highlighted inside the box is the BICF2G630420272 SNP. Each diagonal row represents a different SNP, and each square represents a pairwise comparison between two SNPs. Single nucleotide polymorphisms are arranged in haplotype blocks (thick black triangles) using the “spine of LD” algorithm in Haploview software. Red squares indicate statistically significant (LOD > 2) LD between the pair of SNPs based on the D′ statistic. Darker colors of red indicate higher values of D′, up to a maximum of 1. D′ is a measure of LD ranging from 0 to 1 and is inversely related to the fraction of chromosomes that have experienced historical recombination. D′ = 1 indicates complete LD, and D′ = 0 represents complete linkage equilibrium. White squares indicate pairwise D′ values of <1 with no statistically significant evidence of LD. Blue squares indicate pairwise D′ values of 1 but without statistical significance. Adjacent blocks are merged if they have multiallelic D′ values of at least 0.9, and at least 80% of the chromosome in the resulting merged block is explained by six or fewer common haplotypes.
Figure 4
 
Linkage disequilibrium plot is shown for a 500-kb region surrounding the BICF2G630420272 SNP on CFA 15. Single nucleotide polymorphisms are numbered in order at the top, and the SNP highlighted inside the box is the BICF2G630420272 SNP. Each diagonal row represents a different SNP, and each square represents a pairwise comparison between two SNPs. Single nucleotide polymorphisms are arranged in haplotype blocks (thick black triangles) using the “spine of LD” algorithm in Haploview software. Red squares indicate statistically significant (LOD > 2) LD between the pair of SNPs based on the D′ statistic. Darker colors of red indicate higher values of D′, up to a maximum of 1. D′ is a measure of LD ranging from 0 to 1 and is inversely related to the fraction of chromosomes that have experienced historical recombination. D′ = 1 indicates complete LD, and D′ = 0 represents complete linkage equilibrium. White squares indicate pairwise D′ values of <1 with no statistically significant evidence of LD. Blue squares indicate pairwise D′ values of 1 but without statistical significance. Adjacent blocks are merged if they have multiallelic D′ values of at least 0.9, and at least 80% of the chromosome in the resulting merged block is explained by six or fewer common haplotypes.
Table 1
 
Proportions of PLL Affected and Unaffected Dogs Based on Their Genotypes for the ADAMTS17 and BICF2G630420272 SNPs
Table 1
 
Proportions of PLL Affected and Unaffected Dogs Based on Their Genotypes for the ADAMTS17 and BICF2G630420272 SNPs
Table 2
 
Mean Age at First Diagnosis of PLL in Dog Groups According to Their Genotypes for the ADAMTS17 and the BICF2G630420272 SNPs
Table 2
 
Mean Age at First Diagnosis of PLL in Dog Groups According to Their Genotypes for the ADAMTS17 and the BICF2G630420272 SNPs
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×