G3: Genes|Genomes|Genetics Early Online, published on May 22, 2017 as doi:10.1534/g3.117.043109 1 A Coding Variant in the Gene Bardet-Biedl Syndrome 4 (BBS4) is Associated with 2 a Novel Form of Canine Progressive Retinal Atrophy 3 Tracy Chew, *, 1 Bianca Haase, † Roslyn Bathgate, † Cali E. Willet, ‡ Maria K. Kaukonen, 4 §, **, †† Lisa J. Mascord, * Hannes T. Lohi, §, **, †† Claire M. Wade *, 1 5 6 Author Affiliations 7 *School of Life and Environmental Sciences, Faculty of Science, University of Sydney, 8 Sydney NSW Australia, †Sydney School of Veterinary Science, Faculty of Science, 9 University of Sydney, Sydney NSW Australia, ‡Sydney Informatics Hub, Core Research 10 Facilities, University of Sydney, Sydney NSW Australia, §Department of Veterinary 11 Biosciences, University of Helsinki, Finland, **Research Programs Unit, Molecular 12 Neurology, University of Helsinki, Finland, and ††Folkhälsan Institute of Genetics, 13 Finland 14 15 Reference numbers 16 Genotyping array data is available at NCBI’s Gene Expression Omnibus (accession 17 number GSE87642). Whole genome sequencing data is available at NCBI’s Sequence 18 Read Archive (BioProject accession number PRJNA344694). 19 20 1 © The Author(s) 2013. Published by the Genetics Society of America. 21 Canine Progressive Retinal Atrophy 22 23 Keywords 24 Hungarian Puli; whole genome sequencing; blindness; obesity; infertility 25 26 Corresponding authors 27 Faculty of Veterinary Science, University of Sydney, RMC Gunn Building, B19-301 28 Regimental Cres, Camperdown, NSW 2006, Australia. Email: 29 [email protected]; [email protected] 2 30 Abstract 31 Progressive retinal atrophy is a common cause of blindness in the dog and affects over 32 100 breeds. It is characterized by gradual vision loss that occurs due to the 33 degeneration of photoreceptor cells in the retina. Similar to the human counterpart 34 retinitis pigmentosa, the canine disorder is clinically and genetically heterogeneous and 35 the underlying cause remains unknown for many cases. We use a positional candidate 36 gene approach to identify putative variants in the Hungarian Puli breed using 37 genotyping data of 14 family-based samples (CanineHD BeadChip array, Illumina) and 38 whole genome sequencing data of two proband and two parental samples (Illumina 39 HiSeq 2000). A single nonsense SNP in exon 2 of BBS4 (c.58A>T, p.Lys20*) was 40 identified following filtering of high quality variants. This allele is highly associated -14 41 (PCHISQ = 3.425e , n = 103) and segregates perfectly with progressive retinal atrophy in 42 the Hungarian Puli. In humans, BBS4 is known to cause Bardet-Biedl syndrome that 43 includes a retinitis pigmentosa phenotype. From the observed coding change we expect 44 that no functional BBS4 can be produced in the affected dogs. We identified canine 45 phenotypes comparable with Bbs4-null mice including obesity and spermatozoa flagella 46 defects. Knockout mice fail to form spermatozoa flagella. In the affected Hungarian Puli 47 spermatozoa flagella are present, however a large proportion of sperm are 48 morphologically abnormal and <5% are motile. This suggests that BBS4 contributes to 49 flagella motility but not formation in the dog. Our results suggest a promising opportunity 50 for studying Bardet-Biedl syndrome in a large animal model. 3 51 Introduction 52 Progressive retinal atrophy (PRA, OMIA #000830-9615) is the most common cause of 53 hereditary blindness in the domestic dog (Canis lupus familiaris), affecting over 100 54 pure breeds (Whitley et al. 1995). It is clinically and genetically heterogeneous and 55 encompasses several forms of disease which vary by aetiology, rate of progression and 56 age of onset (Downs et al. 2014a). The typical characteristics are gradual night, 57 followed by day vision loss due to the degeneration of rod and cone photoreceptors and 58 this degeneration continues until the affected animal is completely blind (Parry 1953). 59 Ophthalmic features that become apparent as the retina deteriorates include tapetal 60 hyper-reflectivity, vascular attenuation, pigmentary changes and atrophy of the optic 61 nerve head (Parry 1953; Clements et al. 1996; Petersen-Jones 1998). 62 PRA is recognized as the veterinary equivalent of retinitis pigmentosa (RP) in humans 63 due to the clinical and genetic similarities between the disorders (Petersen-Jones 1998; 64 Cideciyan et al. 2005; Zangerl et al. 2006; Downs et al. 2011). RP is a common cause 65 of blindness in humans and affects approximately 1 in 4,000 people (Hamel 2006). 66 There are very limited treatment options for both PRA and RP at present (Hamel 2006). 67 For this reason, the dog has become a valuable large animal model for retinal 68 degeneration, in particular, for testing the efficacy of novel therapeutics such as gene 69 therapy (Pearce-Kelling et al. 2001; Acland et al. 2001; Narfström et al. 2003; Cideciyan 70 et al. 2005; Beltran et al. 2012; Pichard et al. 2016). As of 2016, 256 retinal disease- 71 associated genes were identified for humans (https://sph.uth.edu/retnet/). Some of 72 these genes cause non-syndromic RP, whilst others contribute to syndromic disorders 73 such as Bardet-Biedl syndrome (BBS) (Hamel 2006). 4 74 Currently, retinal dystrophies in 58 domestic dog breeds have been linked to at least 25 75 mutations in 19 different genes (Miyadera et al. 2012; Downs et al. 2014b). Canine PRA 76 is typically inherited in an autosomal recessive pattern, although two forms that are X- 77 linked (Vilboux et al. 2008) and one that has dominant inheritance have been reported 78 (Kijas et al. 2002; Kijas et al. 2003). Many of these discoveries in the canine were made 79 using candidate gene studies, linkage mapping and genome-wide association studies 80 (GWAS) followed with fine mapping (Acland et al. 1999; Goldstein et al. 2006; 81 Kukekova et al. 2009; Downs et al. 2014b). This success has been facilitated by the 82 unique breeding structure of dogs. Intense artificial selection, genetic drift and strong 83 founder effects have resulted in stretches of linkage disequilibrium (LD) that can persist 84 for several megabases (Mb) within breeds, but only tens of kilobases across breeds 85 (Lindblad-Toh et al. 2005). This species population structure has allowed for the 86 successful mapping of Mendelian traits with fewer markers and subjects compared to 87 human gene mapping studies: as few as 10 unrelated cases and 10 controls (Karlsson 88 et al. 2007; Frischknecht et al. 2013; Jagannathan et al. 2013; Willet et al. 2015; Gerber 89 et al. 2015; Wolf et al. 2015). Such methods are accepted to work extremely well for 90 mapping monogenic traits that segregate within a single breed. 91 Despite this achievement, there are still many forms of PRA in several breeds of dog 92 that have yet to be genetically characterized. Traits with underlying genetic 93 heterogeneity and a late onset are notoriously difficult to map using linkage or GWAS 94 methods (Hirschhorn and Daly 2005; Korte and Farlow 2013). Although PRA is 95 collectively common, individually, specific forms are relatively rare and it may take many 96 generations until an adequately sized cohort of unrelated case samples are collected. 5 97 The genetic heterogeneity of PRA can complicate the results of linkage mapping and 98 GWAS, as different causative variants and genes can be responsible for an identical 99 phenotype. In addition, both linkage and GWAS rely on markers to be in LD and 100 segregate with the disease gene making it difficult to detect rare or de novo variants 101 (Hirschhorn and Daly 2005). 102 Since the advent of whole genome sequencing (WGS) and whole exome sequencing 103 (WES) technologies, the discovery of causal variants for rare or genetically 104 heterogeneous diseases has become more rapid with fewer case samples necessary 105 for success. One study design of note that has been used in human and more recently 106 in canine studies is the sequencing of parent-proband trios (Zhu et al. 2015; Sayyab et 107 al. 2016). As this method provides the chance for earlier diagnosis than previously 108 possible, this gives patients the opportunity to access more personalized treatment 109 options (Farwell et al. 2015; Zhu et al. 2015; Sawyer et al. 2016). 110 In a preliminary study, extensive screening of 53 genes associated with autosomal 111 recessive PRA or RP revealed no putative variants that could be associated with PRA in 112 the Hungarian Puli breed (Chew et al. 2017 [Animal Genetics in press]). Here, we 113 combine genotyping array data and WGS data of a parent-proband trio with an 114 additional half sibling case to identify a potentially novel canine PRA gene. We 115 successfully identify a highly associated mutation in exon 2 of BBS4 (c.58A>T, PCHISQ = 116 3.425e-14, n = 103) that segregates perfectly with the disease phenotype. This mutation 117 encodes a premature stop codon which is expected to result in complete loss of function 118 of the BBS4 protein. The association of BBS4 with canine PRA is a novel finding and 119 presents the first description of an associated variant for PRA in the Hungarian Puli. 6 120 Materials and Methods 121 Samples 122 This study involved 255 dogs (Canis lupus familiaris) that comprised 103 Hungarian Puli 123 and 152 Hungarian Pumi samples. This sample cohort included 14 Hungarian Pulis 124 segregating PRA in an autosomal recessive pattern from a previous study (Chew et al. 125 2017 [Animal Genetics in press]). Three affected Hungarian Pulis (USCF516, USCF519 126 and USCF1311) were diagnosed with PRA at the age of two years by registered 127 specialists in veterinary ophthalmology. Diagnosis was based on observed 128 ophthalmologic changes including vascular attenuation, hyper-reflectivity and reduced 129 myelination in the optic nerve head.