Mutation Causes Progressive Retinal Atrophy in the Cardigan Welsh Corgi Dog

cGMP Phosphodiesterase-␣ Mutation Causes Progressive Retinal Atrophy in the Cardigan Welsh Corgi Dog

Simon M. Petersen–Jones,1 David D. Entz, and David R. Sargan

PURPOSE. To screen the ␣-subunit of cyclic guanosine monophosphate (cGMP) phosphodiesterase (PDE6A) as a potential candidate gene for progressive retinal atrophy (PRA) in the Cardigan Welsh corgi dog.

METHODS. Single-strand conformation polymorphism (SSCP) analysis was used to screen short introns of the canine PDE6A gene for informative polymorphisms in members of an extended pedigree of PRA-affected Cardigan Welsh corgis. After initial demonstration of linkage of a polymorphism in the PDE6A gene with the disease locus, the complete coding region of the PDE6A gene of a PRA-affected Cardigan Welsh corgi was cloned in overlapping fragments and sequenced. SSCP-based and direct DNA sequencing tests were developed to detect the presence of a PDE6A gene mutation that segregated with disease status in the extended pedigree of PRA-affected Cardigan Welsh corgis. Genomic DNA sequencing was developed as a diagnostic test to establish the genotype of Cardigan Welsh corgis in the pet population.

RESULTS. A polymorphism within intron 18 of the canine PDE6A gene was invariably present in the homozygous state in PRA-affected Cardigan Welsh corgis. The entire PDE6A gene was cloned from one PRA-affected dog and the gene structure and intron sizes established and compared with those of an unaffected animal. Intron sizes were identical in affected and normal dogs. Sequencing of exons and splice junctions in the affected animal revealed a 1-bp deletion in codon 616. Analysis of PRA-affected and obligate carrier Cardigan Welsh corgis showed that this mutation cosegregated with disease status.

CONCLUSIONS. A single base deletion at codon 616 in the PDE6A gene cosegregated with PRA status with zero discordance in Cardigan Welsh corgis with PRA. A lod score of 4.816 with a recombination fraction (␪) of zero strongly suggests that this mutation is responsible for PRA in the breed. The mutation is predicted to lead to a frame shift resulting in a string of 28 altered codons followed by a premature stop codon. The authors suggest that this type of PRA be given the name rod–cone dysplasia 3 (rcd3). (Invest Ophthalmol Vis Sci. 1999;40:1637–1644)

ene mutations causing autosomal recessive retinitis report is that which causes rod–cone dysplasia type 1 (rcd1)in pigmentosa (ARRP) in humans have been reported in the Irish setter breed. Rcd1 is caused by an amber mutation in opsin,1 the ␣-2 and ␤-3 subunits of cyclic guanosine the ␤-subunit of the cGMP phosphodiesterase gene G 11–13 monophosphate (cGMP) phosphodiesterase, the ␣- subunit of (PDE6B). Mutations in the homologous gene have been cGMP-gated channel,4 RPE65,5 adenosine triphosphate (ATP)- identified in the retinal degeneration (rd) mouse,14,15 and a binding cassette transferase protein,6,7 tubby-like protein 1 subset of autosomal recessive retinitis pigmentosa pa- (TULP1),8 and cellular retinaldehyde-binding protein.9 The tients.3,16–18 Rcd1 is characterized by the absence of cGMP- analogous group of conditions in the dog are the progressive phosphodiesterase activity, leading to a 10-fold increase in retinal atrophies (PRAs) and are known to occur in several cGMP levels.19 This results in arrested development of photo- breeds of dog (see Ref. 10 for a review). Despite investigating receptors followed by a progressive rod-led photoreceptor genes known to cause similar retinal dystrophies in other degeneration.20 PRA in the collie is characterized by similar species, the only causal gene mutation identified before this biochemical and histopathologic changes.21–23 However, breeding studies have shown the two forms of PRA to be nonallelic,24 leading to the designation of the form in the collie From The Centre for Veterinary Science, Department of Clinical Veterinary Medicine, University of Cambridge, Cambridge, United as rod–cone dysplasia type 2 (rcd2). Both rcd1 and rcd2 are Kingdom. early-onset forms of PRA. Supported by Wellcome Trust Veterinary Research Career Devel- PRA in the Cardigan Welsh corgi has a similar early onset opment Fellowship 041623/z (SMP–J). leading to blindness in the young adult dog and, similar to most Submitted for publication October 27, 1998; revised February 2, forms of canine PRA, is inherited in an autosomal recessive 1999; accepted March 2, 1999. Proprietary interest category: N. manner. PRA in this breed was first recorded in the veterinary 1Present address: Department of Small Animal Clinical Sciences, literature in 197225 and has recently undergone a resurgence, College of Veterinary Medicine, Michigan State University, East Lan- with cases occurring in The Netherlands, New Zealand, and sing, Michigan. the United States. The affected lines in all three countries can Reprint requests: Simon M. Petersen–Jones, Department of Small Animal Clinical Sciences, Michigan State University, D-208 Veterinary be traced back to stock imported from the United Kingdom. Medical Center, East Lansing, MI 48824-1314. Detailed electrophysiological, histopathologic, and biochemi-

Downloaded from jov.arvojournals.org on 09/28/2021 1638 Petersen–Jones et al. IOVS, July 1999, Vol. 40, No. 8

cal analysis of PRA-affected Cardigan Welsh corgis has not yet in the canine PDE6A gene. Sequencing to obtain the en- been performed. tire sequence of the coding region and intron–exon bound- We report here that PRA in the Cardigan Welsh corgi is aries was performed. Initially, PCR products were cloned caused by a 1-bp deletion in codon 616 of PDE6A. We predict using the TA cloning kit (Invitrogen, NV Leek, Groningen, that the histopathogenesis of this form of PRA is similar to that The Netherlands) and at least two independently derived resulting from the amber mutation in PDE6B in rcd1, and if this clones from each region were sequenced by double- proves to be the case, we propose that the condition in the stranded sequencing using a Thermo-Sequenase core se- Cardigan Welsh corgi be designated rod–cone dysplasia type 3 quencing kit (Vistra DNA Systems; Amersham) with a se- (rcd3). quencing primer labeled with Texas red (Amersham) and PRA in the Cardigan Welsh corgi is only the second form run on a Vistra 725 DNA sequencer (Vistra DNA Systems; of canine PRA for which the causal gene mutation has been Amersham). identified and represents the only naturally occurring animal model of ARRP due to a PDE6A mutation. Linkage Analysis Two-point lod scores were calculated for segregation of PRA and an exon 15 1-bp deletion in two-generation families METHODS (phase unknown) represented in the pedigree.29 MLINK Dogs from the LINKAGE program package (Human Genome Map- ping Project, Medical Research Council, UK) was used for Genomic DNA was extracted from blood samples obtained these calculations.30 from dogs within pedigrees of Cardigan Welsh corgis in which The allele frequencies within each group of blood samples PRA was segregating and also from unaffected breeding lines. were used to perform a simple binomial analysis of the prob- The diagnosis of PRA was made by veterinary ophthalmologists ability of homozygosity for a given allele in affected animals, on the basis of clinical history and ophthalmoscopic signs. A given independent assortment of trait and allele. Contingent pedigree showing the relationship between affected animals is probabilities for allele frequencies in the carrier dogs were not shown in Figure 4. Treatment of animals conformed to the included. Therefore, the real probability of independence (no ARVO Statement on the Use of Animals in Ophthalmic and linkage) is lower than the figure quoted in all cases. Vision Research. Sequencing of a PCR Product Spanning the Site of Single-Strand Conformation Polymorphism the Codon 616 1-bp Deletion Analysis of Presumptive Intron 18 of the Canine PDE6A Gene PCR amplification of a genomic fragment containing part of intron 14 and the mutation site in exon 15 was performed Using information about the intron sizes and position of the using sense primer 5Ј-TCATTCCATCGCCGACTC-3Ј (primer human alpha PDE gene (kindly supplied by Steven Pittler positioned –81––64 from intron–exon boundary) and anti- before publication) a primer pair (sense 5Ј-GTGATCTCT- sense primer 5Ј-CCTCATCTCGCAGCAACGTT-3Ј (corre- CAGCCATCACC-3Ј and antisense 5Ј-GATTCTGCTGCAG- sponds to first nucleotide of intron 15 plus nucleotides CACTGTG-3Ј) was designed from the canine cDNA se- 2019–2001). Initially, a radioactively labeled PCR was car- 26–28 quence to amplify the presumptive intron 18 of the ried out using 50 ng genomic DNA in the presence of 2 ␮Ci canine gene by polymerase chain reaction (PCR). The 32P-dATP and SSCP analysis performed as before. Subse- PCR product was found to be a suitable length for single- quently, an unlabeled PCR reaction was used, and 2 ␮lofthe strand conformation polymorphism (SSCP). PCR-SSCP was reaction product was sequenced directly by a Taq-based 32 performed by the inclusion of 2 ␮Ci [␣- P]dATP (Amer- method (Thermo-Sequenase, Amersham) as before, using a sham Life Science, Amersham, UK) in a 50-␮l PCR reac- Texas red–labeled oligonucleotide 5Ј-GGTGTCTTTCCAA- tion. After thermal cycling, 2 ␮l of the resultant product GATGGAG-3Ј (corresponds to nucleotide numbers 1984– was placed into 10 ␮l SSCP loading buffer (95% deion- 1965) as primer. ized formamide; 10 mM NaOH; 0.05% bromphenol blue; 0.05% xylene cyanol). The samples were denatured by heating to 95°C for 2 minutes and placing on ice for 15 RESULTS minutes and electrophoresis performed on a 0.5ϫ MDE- hydrolink polyacrylamide gel (Hoefer, Newcastle-under- Screening of Intron 18 of PDE6A for Lyme, UK), with a running buffer of 0.6ϫ TBE, at 0.12 Polymorphisms by SSCP W/cm for 14 hours at room temperature. The gels were SSCP revealed that intron 18 of PDE6A was polymorphic dried and autoradiographed (X-OMAT film; Eastman Kodak, in the Cardigan Welsh corgis examined, including dogs Rochester, NY). with PRA (Fig. 1). Two alleles (which we designated B and b) were present. Sequencing the PCR products Sequencing of the Coding Region of the Canine showed that B differed from b at four positions within the PDE6A Gene intron (55 C Ͼ T, 65 T Ͼ C, 88 A Ͼ C, and 121 GϾC). It Using the published canine cDNA sequence26–28 with pre- was noted that all the PRA-affected dogs were homozygous dicted intron positions from the human gene, 21 oligonu- for one of the alleles (b) and all the obligate carrier dogs cleotide primer pairs were designed that would amplify had at least one copy of the same allele. Binomial analysis the entire coding region and introns of the canine gene of the segregation of alleles in a small group of dogs (3 in overlapping fragments (Table 1).The resultant PCR affected, 4 obligate carrier, and 11 ophthalmoscopically products were used to estimate the sizes of the introns normal) showed P Ͻ 0.063 of the intron polymorphism

Downloaded from jov.arvojournals.org on 09/28/2021 IOVS, July 1999, Vol. 40, No. 8 PDE6A Mutation Causes PRA in Cardigan Welsh Corgis 1639

TABLE 1. Details of Strategy Used to Amplify Canine PDE6A, Establish Intron Sizes, and Sequence the Coding Region and Intron–Exon Boundaries

PCR Intron Conditionsሻ Sense Primer Sequence Antisense Primer Sequence (Annealing Gene Fragment (5؅–3؅) (Numbering)* (5؅–3؅) (Numbering)* temp. °C) No. Size

5Ј UTR–exon 1 CCAGCTATAGACCTTCCCTG GGACAGCATCCTTGTGGACA ST (60) n/a (24–43) (454–435) Exon 1–exon 3 AATGGCATCGCAGAGCTAGC CGCCGAGTCTGAGAATTGTG LR1 (60) 1 6600bp (400–419) (803–784) Exon 2–exon 3 CCTCACTGAGTACCAGACCA CGCCGAGTCTGAGAATTGTG ST (60) 2 742bp‡ (597–616) (803–784) Exon 2–exon 4 CCTCACTGAGTACCAGACCA CTTGGTCATGTCTAAGAGTC LR1 (56) 3 600bp (597–616) (945–926) Exon 4–exon 5 GCCTTCCTCAACTGTGACAG GAGTCCTGGGACCAGAATAG LR1 (60) 4 4700bp (895–914) (1015–996) Exon 5–exon 6 ATGTGTGGCCAGTCCTGATG TGTCTTCTTTGCCGTGTAGG ST (58) 5 650bp (962–981) (1078–1059) Exon 6–exon 8 CAAGGTCATTGACTATATCC CCTGGAATGCAAAAAAGTCC LR1 (52) 6 7500bp (1041–1060) (1206†–1188) Exon 7–exon 8 ATCCACCTCCTGATCATTGG CCTGGAATGCAAAAAAGTCC LR1 (56) 7 3200bp (1094–1113) (1206†–1188) Exon 8–exon 9 TGCAACATCATGAATGCACC AGGGTCTCATCCATTTCATC ST (56) 8 2800bp (1162–1181) (1349–1330) Exon 9–exon 10 TGGATGAGTCTGGATGGATG CATTGTCACACTTCACGTGG ST (56) 9 550bp (1217–1236) (1480–1461) Exon 10–exon 11 GGCTGGTCCGTCTTAAATCC AGGATCTCAGCGAGTTCCTC ST (60) 10 1900bp (1375–1394) (1565–1546) Exon 11–exon 12 CCAGAGAGGTGTATGGGAAG TTTCACCAGCTCCAGTTCGG ST (60) 11 214bp‡ (1505–1524) (1653–1634) Exon 12–exon 13 AGAGCTGCCAGATGCAGAGA AGCAAGGAGAACATGGTCTG ST (63) 12 950bp (1572–1591) (1817–1798) Exon 13–exon 14 TGCGCTTCATGTACTCGCTG GTCAATGTCATGGCAGAAGG ST (58) 13 1500bp (1721–1740) (1899–1880) Exon 14–exon 15 GCGATACTTCACAGACCTAG CCTCATCTCGCAGCAACGTT ST (58) 14 1400bp (1836–1855) (2019†–2001) Exon 15–exon 16 AGACACCACTTGGAGTTCGG TGCTGCCTGCGATTGAGGTT ST (63) 15 180bp‡ (1978–1997) (2057–2038) Intron 15–exon 17 TTCTCACATCTCTTCTACGG TCCGTGTCTGCTCCAGCATC ST (56) 16 417bp‡ (in intron 15§) (2215–2196) Exon 15–exon 19 AGACACCACTTGGAGTTCGG GATTCTGCTGCAGCACTGTG LR3 (63) 17 11000bp (1978–1997) (2362–2343) Exon 18–exon 19 GTGATCTCTCAGCCATCACC GATTCTGCTGCAGCACTGTG ST (63) 18 288bp‡ (2249–2268) (2362–2343) Exon 19–exon 20 ACTGGTTGCTGCCGAATTCT GACAAAGGTGCAAACAAAGT ST (55) 19 1500bp (2301–2320) (2445–2426) Exon 20–exon 21 CCCAAGCTTCAAGTCGGCTT TGTCGTACTCATCGGCGAGC ST (60) 20 1500bp (2401–2420) (2545–2526) Exon 21–3Ј UTR TCCCATGCTGGATGGGATCA AAGGGTGGTACCATTCGGTG ST (60) 21 1900bp (2481–2500) (2718–2699)

* Numbering according to genebank accession number Z68340. † The 5Ј nucleotide of these primers corresponds to the first nucleotide of the intron. ‡ Exact size of introns (remainder are sized from PCR product size). § The sense primer in intron 15 was positioned Ϫ32 to Ϫ13 nucleotides from the intron 15/exon 16 boundary. ࿣ PCR conditions as follows: LR1, long-range PCR (Expand Long Template PCR System, Boehringer Mannheim, Lewes, UK) using manufacturer’s buffer 1 (final concentrations of dNTPs, primers, enzymes, and buffer as recommended by manufacturer); LR3 ϭ long-range PCR (Expand Long Template PCR System, Boehringer Mannheim, Lewes, UK) using manufacturer’s buffer 3 (final concentration of dNTPs, primers, enzymes, and buffer as recommended by manufacturer); ST, standard PCR using Taq polymerase (Gibco BRL, Life Technologies, Paisley, UK), final concentrations: 20 mM

Tris-HCl, pH 8.4; 50 mM KCl; 1.5 mM MgCl2; 1% bovine serum albumin, 100 ␮M each of dATP, dCTP, dGTP, dTTP; 0.2 ␮M of each primer; 1 U Taq. Cycling conditions were as follows: Long-range PCR 94°C for 2 minutes, 10 cycles of 94°C for 10 seconds, then T°C anneal for 30 seconds, then 68°C for 5 minutes, followed by increasing extension time by 30 seconds every three cycles for 6 ϫ 3 cycles. For standard PCR 94°C for 2 minutes, followed by 35 cycles of 94°C for 20 seconds, then T°C anneal for 30 seconds, then 72°C for 1 minute, with a ﬁnal extension at 72°C for 10 minutes.

assorting independently to PRA, although bb homozygotes mutation could have occurred on the b chromosome genetic were found among ophthalmoscopically normal and PRA- background, making the locus worthy of further investiga- affected dogs (Fig. 1). This suggested that the PRA-causing tion.

Downloaded from jov.arvojournals.org on 09/28/2021 1640 Petersen–Jones et al. IOVS, July 1999, Vol. 40, No. 8

FIGURE 1. SSCP analysis of a polymorphism in intron 18 of the canine PDE6A segregating in a group of PRA-affected and normal Cardigan Welsh corgi dogs (see the Methods section for the procedure used). All PRA-affected Cardigan Welsh corgi dogs were homozygous for the b allele, and all obligate carriers had at least one b allele. A: PRA-affected. All are homozygous for the b allele (bb). C: obligate PRA-carriers. All carriers had at least one b allele. N: ophthalmoscopically normal unrelated dog (BB). N* is an ophthalmoscopically normal dog that is homozygous for the same allele that is found in all the affected animals (bb).

Sequencing of the Coding Region of the PDE6A AJ233693 give exonic and partial intronic sequences). In addi- Gene in PRA-Affected Cardigan Welsh Corgis tion, the sequencing of exon 15 revealed the presence of a The entire PDE6A gene from a PRA-affected Cardigan Welsh 1-bp adenine deletion at nucleotide 1939–1940 (numbering of corgi was amplified by PCR and all exons and intron–exon Z68340), codon 616, causing a frame shift (Fig. 2). This causes junctions sequenced. The canine gene spans approximately 53 premature termination of translation at a stop codon at posi- kb of genomic DNA and the intron–exon structure of the gene tion 644. was found to be the same as for the human orthologue (S. Using SSCP, all affected dogs for which we had DNA Pittler 1997, personal communication). The gene structure, samples were shown to be homozygous for the mutation (in- size of introns, and PCR strategy used are shown in Table 1. cluding dogs from The Netherlands and from the United Subsequent amplification of the equivalent fragments from an States), and all the obligate carriers were heterozygous (exam- ophthalmoscopically normal dog unrelated to the known PRA- ple, Fig. 3). All these dogs could be fitted to a large pedigree affected dogs, showed that no gross rearrangements, inser- (Fig. 4), originating with a single UK-bred dog. None of the tions, or deletions to the gene that were detectable by agarose other clinically normal adult dogs within the pedigree (or gel electrophoresis had occurred (data not shown). Sequenc- outside it) were homozygous for the mutation. ing of the gene from the PRA-affected animal showed that the The SSCP data for the mutation in exon 15 were analyzed cDNA sequence differed from each of the previously published to test the strength of evidence of linkage between the PDE6A sequences for canine PDE6A (GenBank accession numbers mutation and the disease allele. Eight PRA-affected dogs and 10 Z68340, U52868, and Y13282) at one or more of seven posi- obligate carriers could be grouped into four incomplete two- tions as shown in Table 2 (EMBL accession numbers AJ233677– generation families (2 offspring in these families were known

TABLE 2. Summary of Polymorphic Residues in PDE6A cDNA Sequence Position*

91 192 625 791 931 1255 1920

Present study G C T T C A C Kylma et al.27 (Z68340) G C T T T A C Wang et al.26 (U52868) G C C C T T C Veske et al.28 (Y13282) T T T C C A T Comment† 5Ј Leader I Ͼ ILϾ LCϾ CLϾ LMϾ LLϾ L

Polymorphic residues between the canine PDE6A cDNA sequence established in the present study and the three previously reported cDNA sequences. * Numbering follows that of Kylma et al.27 (Z68340). We have sequenced the region from ϩ26 to 2718 from an affected animal (coding region and parts of introns). This includes the whole protein coding region. † Amino acids encoded by our sequence and variants from it, at the polymorphic positions.

Downloaded from jov.arvojournals.org on 09/28/2021 IOVS, July 1999, Vol. 40, No. 8 PDE6A Mutation Causes PRA in Cardigan Welsh Corgis 1641

FIGURE 2. Comparison of normal and mutant canine PDE6A cDNA sequence and translation in the region of the mutation responsible for PRA in the Cardigan Welsh corgi dog. The mutation is an adenine deletion at codon 616, nucleotide 1939–1940 (numbering according to Kylma et al.27; EMBL submission number Z68340). The site of the deletion is highlighted (both possible A residues shown in bold), as is the sequence of the abnormal protein resulting from the frame shift. Vertical lines indicate position of exon–exon boundaries.

to be carriers because they were parents of affected animals in (GT)10GCATGTGC(GT)11; (GT)10AT(GT)14; (GT)10GAGTGTGA- succeeding generations), allowing calculation of a two-point (GT)11; (GT)25. The adenine deletion at nucleotide 1939–1940 lod score (phase unknown). This provided Z(␪)max ϭ 4.816 was invariably associated with the first of these alleles, which was for ␪ ϭ 0, with a support interval showing ␪ is less than 0.15. the commonest allele in the population sample (70% of alleles), (␪ is the recombination fraction, Z(␪) the lod for that fraction). and 2 bp longer than the other alleles. Because the sample was To include in the analysis of the exon 15 mutation SSCP collected during investigations of PRA-affected animals and their results for samples from dogs that were not closely related, we relatives, the allele frequency found is likely to be biased toward also conducted a binomial analysis of the data from exon 15 for the PRA-associated allele. Observed heterozygosity in this sample all PRA-affected and obligate carrier animals, based on the allele was 44.2%. Subsequent observations from more than 200 other frequencies observed in a sample of 41 dogs, including 16 Cardigan Welsh corgi dogs did not reveal any other alleles of the normal animals, 12 carriers, and 13 PRA-affected animals. The microsatellite. probability of complete phase conservation between disease and PDE6A allele status occurring by chance in this sample, if Testing of Further Pet Cardigan Welsh Corgis for there is no linkage between disease and PDE6A locus, is P Ͻ the Presence of the Adenine Deletion at Position 0.000002. 1939–1940 We identified a microsatellite at the 3Ј end of intron 14 of the Recently, we have tested Cardigan Welsh corgis from the gene, ending only 18 bases upstream from the deletion site and 11 general pet populations of a number of different countries for bases from the splice acceptor site. In a survey of 43 Cardigan the presence of the adenine deletion at position 1939–1940. Welsh corgi dogs, direct sequencing of PCR amplification prod- Genomic DNA was amplified by PCR across the mutation site ucts spanning the microsatellite revealed 6 alleles of the micro- and directly sequenced (Fig. 5). The animals tested were self- satellite: (GT)10GCGTGTGC(GT)12; (GT)10GCGTGTGC(GT)11; selected by their owners. Therefore, there may be biases op-

FIGURE 3. SSCP analysis of part of intron 14 and exon 15 of canine PDE6A in PRA-affected, obligate carrier, and other ophthalmoscopically normal Car- digan Welsh corgis (see the Methods section for the procedure used). A: PRA-affected Cardigan Welsh corgi dogs. All ﬁve affected dogs have an identical SSCP pattern. N: ophthalmoscopically normal Car- digan Welsh corgi dogs from breeding lines re- puted to be free of PRA. These two normal dogs have a different SSCP pattern compared with the PRA-affected dogs. C: obligate PRA-carrier Cardigan Welsh corgi dogs. The SSCP patterns for all obligate carriers are the result of a combination of the pattern for the PRA-affected dogs and the pattern for the ophthalmoscopically normal dogs. This shows that the carriers are heterozygous for the two SSCP- detectable alleles.

Downloaded from jov.arvojournals.org on 09/28/2021 1642 Petersen–Jones et al. IOVS, July 1999, Vol. 40, No. 8

FIGURE 4. A pedigree relating PRA-affected Cardigan Welsh corgi dogs from several parts of the world. The arrow points to a sire common to all lines with PRA. This dog was registered in the United Kingdom in the late 1950s. Open symbols represent dogs that are phenotypically normal, but of unknown genotype; open symbols with central spots represent dogs not examined; ﬁlled symbols represent PRA-affected dogs; and half-ﬁlled symbols PRA-carriers. Squares represent males, circles females.

erating either toward or away from testing of breeding lines the late 1950s. The disease was first recognized in the United already suspected by owners of carrying PRA. Of 31 ophthal- Kingdom and in Australia in the mid 1960s. During the next moscopically normal animals tested in this phase of the work, few years, strenuous efforts were made by breed societies to 4 have been carriers of the mutation (frequency of the mutant avoid breeding from lines in which PRA occurred. This may be allele, 6.5%). We also used this test to reanalyze our research why serious recrudescence of the disease did not occur until pedigree samples and found complete concordance with the the 1990s. exon 15 SSCP findings. In the research as a whole, Cardigan The mutation identified here is predicted to shorten the Welsh corgis carrying the mutation have been found in the dog translated protein by 218 amino acids. If translated, the result- populations of the United Kingdom, The Netherlands, the ant protein would be missing part of the catalytic domain31 and United States, Germany, and New Zealand. the C-terminal cysteine responsible for membrane binding.32 Mutations in this gene have been described in families with ARRP, and the Cardigan Welsh corgi is the first recorded DISCUSSION naturally occurring animal model of this form of retinitis pig- Canine PRA represents a genetically diverse group of retinal mentosa. This makes the Cardigan Welsh corgi a potentially dystrophies with strong similarities to retinitis pigmentosa. valuable model for studying retinal dystrophy due to PDE6A Before this publication, the only form of PRA characterized at mutations and also as a model for treatments, such as the use the molecular level was rod–cone dysplasia type 1 (rcd1)in of growth factors or gene therapy, that are intended to slow the Irish setter. Rcd1 is caused by a point mutation in PDE6B down or halt the retinal degeneration. resulting in the introduction of a premature stop codon. This There is only a limited clinical description of PRA in the report demonstrates that PRA in the Cardigan Welsh corgi is Cardigan Welsh corgi. Keep25 reported that ophthalmoscopi- also caused by a mutation of a subunit of cGMP phosphodies- cally the disease in this breed can generally be detected be- terase, but this time the ␣-subunit. All dogs that we have tween 6 and 16 weeks of age. Affected animals are usually observed carrying the mutation have now been found to be in blind before 1 year, but some retain limited central vision to 3 direct line of descent from the dog indicated with an arrow in to 4 years of age. A similar variation in rate of vision loss was the pedigree (Fig. 4). This dog was a show champion born in first reported in rcd1 by Parry33 in the 1950s and confirmed by

Downloaded from jov.arvojournals.org on 09/28/2021 IOVS, July 1999, Vol. 40, No. 8 PDE6A Mutation Causes PRA in Cardigan Welsh Corgis 1643

FIGURE 5. Direct sequencing of a PCR product spanning the site of the codon 616 adenine deletion in the PDE6A gene (see the Methods section for process used). The output from a DNA sequencer is shown. The sequence information spans the PRA-causing mutation site. The antisense strand is shown. The top sequence track is from a genotypically normal dog (Normal dog), the middle track from a PRA-affected Cardigan Welsh corgi dog (rcd-3 dog), and the bottom track from a PRA-carrier Cardigan Welsh corgi dog (Carrier dog). V’s indicate the region of the 1-bp deletion (normal, TTCT; PRA mutant, TCT). The sequence from the heterozygous PRA carrier is disrupted because of the 1-bp deletion in half of the PCR product. Part of a microsatellite can be seen at the right of the sequence tracks. The disruption of sequence in the heterozygous animal is most obvious in the sequence from the microsatellite.

Ray et al.34 once a DNA-based test for the mutation had been 4. Dryja TP, Finn JT, Peng Y-W, et al. Mutations in the gene encoding developed. the ␣ subunit of the rod cGMP-gated channel in autosomal reces- We predict that this mutation results in a rod–cone dys- sive retinitis pigmentosa. Proc Natl Acad USA. 1995;92:10177– 10181. plasia similar to that seen in rcd1. If this is the case, we propose 5. Morimura H, Fishman GA, Grover SA, et al. Mutations in the RPE65 it be given the name rod–cone dysplasia type 3 (rcd3). This gene in patients with autosomal recessive retinitis pigmentosa or distinguishes it from rcd1, the PDE6B nonsense mutation seen leber congenital amaurosis. Proc Natl Acad USA. 1998;95:3088– in Irish setters, and rcd2. PDE6A has been excluded as the rcd2 3093. locus.35 6. Cremers FPM, van de Pol D, van Driel M, et al. Autosomal recessive retinitis pigmentosa and cone-rod dystrophy caused by splice site mutations in the Stargardt’s disease gene ABCR. Hum Mol Genet. Acknowledgments 1998;7:355–362. The authors thank the Cardigan Welsh Corgi Breed Associations of the 7. Martinez–Mir A, Paloma E, Allikmets R, et al. Retinitis pigmentosa United Kingdom, the United States, and the Netherlands for their caused by a homozygous mutation in the Stargardt disease gene enthusiastic participation in this study; and Frans Stades for screening ABCR. Nat Genet. 1998;18:11–12. dogs in the Netherlands and collecting blood samples for this study. 8. Banerjee P, Kleyn PW, Knowles JA, et al. ‘Tulp1Ј mutation in two recessive extended Dominican kindreds with autosomal recessive retinitis pigmentosa. Nat Genet. 1998;18:177–179. References 9. Maw MA, Kennedy B, Knight A, et al. Mutations in the gene 1. Rosenfeld PJ, Cowley GS, McGee TL, et al. A null mutation in the encoding cellular retinaldehyde-binding protein in autosomal re- rhodopsin gene causes rod photoreceptor dysfunction and auto- cessive retinitis pigmentosa. Nat Genet. 1997;17:198–200. somal recessive retinitis pigmentosa. Nat Genet. 1992;1:209–213. 10. Petersen–Jones SM. A review of research to elucidate the causes 2. Huang SH, Pittler SJ, Huang X, et al. Autosomal recessive retinitis of generalized progressive retinal atrophies. Vet J. 1998;155:5– pigmentosa caused by mutations in the ␣ subunit of rod cGMP 18. phosphodiesterase. Nat Genet. 1995;11:468–471. 11. Farber DB, Danciger JS, Aguirre G. The beta subunit of cyclic GMP 3. McLaughlin ME, Sandberg MA, Berson EL, Dryja TP. Recessive phosphodiesterase mRNA is deﬁcient in canine rod–cone dyspla- mutations in the gene encoding the ␤-subunit of rod phophodies- sia 1. Neuron. 1992;9:349–356. terase in patients with retinitis pigmentosa. Nat Genet. 1993;4: 12. Suber ML, Pittler SJ, Quin N, et al. Irish setter dogs affected with 130–134. rod–cone dysplasia contain a nonsense mutation in the rod cGMP

Downloaded from jov.arvojournals.org on 09/28/2021 1644 Petersen–Jones et al. IOVS, July 1999, Vol. 40, No. 8

phosphodiesterase beta-subunit gene. Proc Natl Acad USA. 1993; 25. Keep JM. Clinical aspects of progressive retinal atrophy in the 90:3968–3972. Cardigan Welsh corgi. Aus Vet J. 1972;48:197–199. 13. Clements PJM, Gregory CY, Petersen–Jones SM, Sargan DR, Bhat- 26. Wang W, Acland GM, Aguirre GD, Ray K. Cloning and character- tacharya SS. Confirmation of the rod cGMP phophodiesterase ization of the cDNA encoding the alpha-subunit of cGMP- ␤-subunit (PDE␤) nonsense mutation in affected rcd-1 Irish setters phophodiesterase in canine retinal rod photoreceptor cells. Mol in the UK and development of a diagnostic test. Curr Eye Res. Vis (http://www.molvis.org/molvis/). 1996;2:3. 1993;12:861–866. 27. Kylma T, Paulin L, Hurwitz MY, Hurwitz RL, Kommonen B. 14. Bowes C, Li T, Danciger M, et al. Retinal degeneration in the rd Cloning of the cDNA encoding rod photoreceptor cGMP- mouse is caused by a defect in the ␤ subunit of rod cGMP- phosphodiesterase alpha and gamma subunits from the retinal phosphodiesterase. Nature. 1990;347:677–680. degenerate Labrador retriever dog. Res Vet Sci. 1997;62:293– 15. Pittler SJ, Baehr W. Identification of a nonsense mutation in the rod 296. photoreceptor cGMP phophodiesterase ␤-subunit gene of the rd 28. Veske A, Nilsson SE, Gal A. Microsatellite (CA)n polymorphism in mouse. Proc Natl Acad USA. 1991;88:8322–8326. intron 2 of the canine rod photoreceptor cGMP-phosphodiesterase 16. Danciger M, Blaney J, Gao YQ, et al. Mutations in the PDE6B gene alpha-subunit gene. Anim Genet. 1997;28:380. in autosomal recessive retinitis pigmentosa. Genomics. 1995;30: 29. Barnard GA. Statistical inference. J Royal Stat Soc. 1949;B11:115– 1–7. 139. 17. Bayes M, Giordano M, Balcells S, et al. Homozygous tandem dupli- 30. Lathrop GM, Lalouel JM, Julier C, Ott J. Strategies for multilocus cation within the gene encoding the beta-subunit of rod phosphodiesterase as a cause for autosomal recessive retinitis pigmentosa. linkage analysis in humans. Proc Natl Acad USA. 1984;81:3443– Hum Mutat. 1995;5:228–234. 3446. 18. McLaughlin ME, Ehrhart TL, Berson EL, Dryja TP. Mutation spec- 31. Baehr W, Champagne MS, Lee AK, Pittler SJ. Complete cDNA trum of the gene encoding the beta subunit of rod phosphodies- sequences of mouse rod photoreceptor cGMP phosphodiesterase terase among patients with autosomal recessive retinitis pigmen- alpha- and beta-subunits, and identification of betaЈϪ, a putative tosa. Proc Natl Acad USA. 1995;92:3249–3253. beta-subunit isozyme produced by alternative splicing of the beta- 19. Aguirre GD, Farber D, Lolley R, Fletcher RT, Chader GJ. Rod–cone subunit gene. FEBS Lett. 1991;278:107–114. dysplasia in Irish setters: a defect in cyclic GMP metabolism in 32. Ong OC, Ota IM, Clarke S, Fung BK-K. The membrane binding visual cells. Science. 1978;201:1133–1134. domain of rod cGMP phosphodiesterase is posttranslationally mod- 20. Aguirre GD, Rubin LF. Rod–cone dysplasia (progressive retinal ified by methyl esterification at a C-terminal cysteine. Proc Natl atrophy) in Irish setters. J Am Vet Med Assoc. 1975;166:157–164. Acad USA. 1989;86:9238–9242. 21. Woodford BJ, Liu Y, Fletcher RT, et al. Cyclic nucleotide metabo- 33. Parry HB. Degenerations of the dog retina, II: generalised progres- lism in inherited retinopathy in collies: a biochemical and histo- sive retinal atrophy of hereditary origin. Bt J Ophthalmol. 1953; chemical study. Exp Eye Res. 1982;34:703–714. 37:487–502. 22. Santos–Anderson RM, Tso M, Wolf ED. An inherited retinopathy in 34. Ray K, Baldwin VJ, Acland GM, Blanton SH, Aguirre GD. Cosegre- collies: a light and electron microscopic study. Invest Ophthalmol gation of codon 807 mutation of the canine rod cGMP phospho- Vis Sci. 1980;19:1282–1294. diesterase ␤ gene and rcd1. Invest Ophthalmol Vis Sci. 1994;35: 23. Wolf ED, Vainisi SJ, Santos–Anderson RM. Rod cone dysplasia in 4291–4299. the collie. J Am Vet Med Assoc. 1978;173:1331–1333. 35. Wang W, Acland GM, Aguirre GD, Ray K. Linkage analysis in 24. Acland GM, Fletcher RT, Gentleman S, Chader GJ, Aguirre GD. canine rod–cone dysplasia type 2 (rcd2) pedigrees [ARVO Ab- Non-allelism of three genes (rcd 1, rcd 2 and erd) for early-onset stract]. Invest Ophthalmol Vis Sci. 1998;39(4):S881. Abstract nr hereditary retinal degeneration. Exp Eye Res. 1989;49:983–998. 4083.

ERRATUM

Erratum in: “Cone Signal Contributions to Electroretinograms in Dichromats and Trichro- mats,” by Kremers et al. (Invest Ophthalmol Vis Sci. 1999;40:920–930). The term electrograms was used incorrectly in the title of this paper. It should have been electroretinograms, as given above. The Journal regrets the error.

Downloaded from jov.arvojournals.org on 09/28/2021