Inherited retinal diseases in dogs
〔Mini Review〕
Inherited retinal diseases in dogs: advances in gene/mutation discovery Keiko MIYADERA School of Veterinary Medicine, University of Pennsylvania, 3900 Delancey St, Philadelphia PA 19104 USA
1. Introduction
congenital structural abnormalities due to aberrant
Inherited retinal diseases (RDs) in humans comprise a
development of the eye (Table 1). Most canine RDs are
spectrum of clinically heterogeneous vision-threatening
autosomal recessive, while autosomal dominant (Kijas et al .
conditions such as retinitis pigmentosa (RP), cone-rod
2002) and X-linked (Acland et al . 1994; Zhang et al . 2002)
dystrophy (CRD) and Leber’s congenital amaurosis (LCA).
PRAs have been identified. Notably, all canine RDs
Since the mapping of the RHO (rhodopsin) gene in a form of
molecularly characterized to date have been reported as a
RP (Farrar et al. 1990), 221 genes have been associated with
monogenic trait (i.e. caused by a mutation in a single gene),
human RDs (RetNet; www.sph.uth.tmc.edu/RetNet/), and
following a Mendelian inheritance.
the number of genes continues to grow. Animal models have been indispensable in validating the physiologic and
2.1. Progressive disorders – PRA and CRD are progressive
pathologic mechanisms of these genes, most extensively, as
RDs, primarily affecting the photoreceptors. PRA is
genetically modified mice. To a lesser degree but with no
characterized by initial degeneration of rod photoreceptors
less significance, dogs also have received much attention
causing night blindness, progressing to total blindness.
over the years as a unique, naturally-occurring model of
Affected dogs with visual deficits have characteristic fundus
RDs. In this review, I will discuss the current knowledge of
changes starting from tapetal hyperreflectivity, attenuation
the molecular basis of canine RDs and advances in the gene/
of retinal blood vessels, then pale optic disc. CRD is less
mutation discovery approach.
common, and predominantly affects cone photoreceptors while rods are affected later or to a lesser degree. Early
2. Phenotypic characteristics of canine RDs
clinical signs therefore can be day vision problems.
T h e m o s t c o m m o n l y d i a g n o s e d c a n i n e R D i s
However, diagnosis is often made when there is already
progressive retinal atrophy (PRA) which is considered
extensive day and light visual deficit with end-stage fundic
homologous to RP in humans. Dogs affected with PRA
changes similar to PRA. As such, CRD may be
initially present night blindness, eventually progressing into
indistinguishable from PRA at the clinical level, thus often
total vision loss. While PRA has been documented in
being referred to as PRA.
numerous canine breeds, it was once thought to be a single
While dogs affected with PRA/CRD show similar
disease caused by the same mutation across breeds.
fundic changes at the end-stage of the disease, the age of
However, breed-specific clinical phenotype (e.g. age of
onset and the rate of progression may vary, depending on the
onset, rate of progression) together with the evidence from
underlying gene/mutation, thereby depending on the breed.
diligent mating trials indicated that multiple independent
Early-onset PRA/CRD results from abnormal retinal
forms of PRAs exist, and that each form is specific to certain
development or progressive degeneration starting during or
breed(s). Inherited canine RDs may be classified as:
soon after retinogenesis. This is followed by a rapid
‘progressive’that undergo programmed degenerative changes during life;‘stationary’that are typically functional
Correspondence: Keiko Miyadera, School of Veterinary Medicine, University of Pennsylvania, 3900 Delancey St, Philadelphia PA 19104 USA (E-mail:
[email protected])
abnormalities present at birth (i.e. congenital) showing very little progression if at all; and‘developmental’that are The Journal of Animal Genetics (2014) 42, 79–89
79
MIYADERA
progression toward end-stage retinal degeneration, and can
are the two critical functional pathways involved in vision.
be clinically evident in young adult dogs. Late-onset PRA/
Mutations in genes participating in these pathways may be
CRD is characterized by pathological changes after normal
associated with RDs, and many of the protein products are
development of the retina. As such, they tend to represent
localized to the outer retina where these pathways take
defects in pathways critical for the long-term maintenance of
place. More recently, genes associated with the morphology
normal photoreceptors. Visual deficits may not become
and function of the photoreceptor connecting cilia have been
apparent until well after reproductive maturity, and further
associated with RDs, and their roles in normal and diseased
progression is typically much slower than in early-onset
retina have been a major focus of study (Estrada-Cuzcano et
PRA/CRD, and may not become clinically evident until later
al . 2012). Of the ~23 genes associated with canine RDs to
in life.
date, at least 6 genes have implications of a primary role in cilia trafficking. The complex gene-gene interaction
2.2. Stationary disorders – Certain forms of RD show no
associated with cilia trafficking remains to be fully
or very limited progression. Examples of such stationary
unraveled.
disorders are achromatopsia (cone degeneration) (Sidjanin et
al . 2002) and canine LCA (cLCA), previously known as
4. Advances in canine RD gene/mutation discovery
congenital stationary night blindness (CSNB) (Narfström et
The gene/mutation discovery approaches for inherited
al . 1989). Cones alone are dysfunctional in achromatopsia,
canine traits have seen considerable development over the
because of a mutation in CNGB3 which is a cone-specific
past 20 years. Initial attempts were focused on candidate
gene, while both rods and cones are affected in cLCA
genes, followed by the introduction of linkage mapping,
because of a mutation in RPE65 , an RPE-specific
positional cloning strategies, and microarray-based genome-
isomerohydrolase impairing the retinoid cycle leading to no
wide association study (GWAS). More recently, next-
11-cis -retinal being made. While the achromatopsia-affected
generation sequencing (NGS) has been implemented with
dogs show no further deterioration with age in vision or
increasing availability and affordability.
retinal abnormalities, elder cLCA-affected dogs show some retinal thinning, both clinically and histologically indicating
4.1. Within-breed genetic uniformity
that cLCA originally classified as stationary likely has some
Where specific canine breeds are to be established and
aspects of a progressive disorder.
maintained with their unique traits, admixture of different breeds is prevented. As a result of this‘breed barrier’, the
2.3. Developmental disorders – Retinal dysplasia is a
genetic makeup within a‘purebred’dog is typically
developmental disorder characterized by a defect in retinal
uniform where many sequence variants are fixed. Such
differentiation. Syndromic retinal dysplasia in Labrador
within-breed genetic uniformity has been both an advantage
Retrievers and Samoyeds are oculoskeletal disorders that
and a disadvantage in the study of canine genetic traits
model autosomal recessive Stickler syndrome in humans
(Miyadera et al . 2012a). For example, RDs can be clinically
(Goldstein et al . 2010a). Another developmental disorder is
variable with regard to the onset, severity, progression, and
Collie eye anomaly (CEA) variably affecting the retina-
the mode of inheritance in the overall dog population, but
choroid-scleral complex ranging from choroidal hypoplasia,
within the same breed, they generally show a uniform
scleral coloboma to retinal detachment. CEA is considered
phenotype; this allows appropriate selection of cases and
homologous to macular coloboma in humans. Due to high
controls as supposed to humans where multitudes of RD
worldwide prevalence of CEA in Collies (70-90%), it is the
forms coexist across the population. As the canine breed
most common inherited canine RDs, yet impairment of
gene pool or genetic variability is limited due to the‘breed
vision is uncommon (Barnett and Stades 1979; Roberts
barrier’as well as the‘founder/popular-sire effect’, a form
1969).
of RD of the same breed is more likely to be caused by the same gene/mutation; while this concept is still valid in many
3. Genes associated with RDs in humans and dogs
occasions, as discussed in the following sections, our
The phototransduction cascade and the retinoid cycle
understanding of the genetic picture of RDs within and
The Journal of Animal Genetics (2014) 42, 79–89
80
The Journal of Animal Genetics (2014) 42, 79–89
81 Collies and other
CEA Multiple Coton de Tulear Lapponian Herder
Visual cycle Collagen formation Collagen formation
RPE65 COL9A3 COL9A2
BEST1 BEST1 BEST1
Epithelial transport Epithelial transport Epithelial transport
N.D.
Phototransduction
CNGB3
NHEJ1
Phototransduction
N.D. Cilia trafficking
ADAM9 RPGRIP1
CNGB3
Phototransduction Cilia trafficking
PDE6B IQCB1
Cilia trafficking
Anion exchange Cilia formation Cilia trafficking Cilia trafficking Phototransduction N.D. Phototransduction Phototransduction Cilia trafficking
SLC4A3 TTC8 RPGR RPGR RHO CCDC66 CNGB1 SAG FAM161A
NPHP4
Phototransduction Phototransduction Phototransduction Phototransduction N.D. N.D. N.D.
Gene Function
PDE6B PDE6B RD3 PDE6A C2orf71 STK38L PRCD
Gene
Candidate gene Candidate gene Candidate gene
LA
LA
LA
Candidate gene
LA
LA
GWAS (pet, colony) LA GWAS (modifier) GWAS
GWAS GWAS
GWAS GWAS, targeted-NGS
GWAS GWAS, targeted-NGS LA LA Candidate gene LA -
Candidate gene Candidate gene LA Candidate gene GWAS LA LA
Approach
(Guziewicz et al . 2007) (Guziewicz et al . 2007) (Zangerl et al . 2010)
(Parker et al . 2007)
(Goldstein et al . 2010a)
(Goldstein et al . 2010a)
(Aguirre et al . 1998; Veske et al . 1999)
(Sidjanin et al . 2002)
(Sidjanin et al . 2002)
(Wiik et al . 2008)
(Goldstein et al . 2010c) (Mellersh et al . 2006; Miyadera et al . 2012)
(Goldstein et al . 2013b) (Goldstein et al . 2013b)
(Downs et al . 2011) (Downs et al . 2014) (Zhang et al . 2002) (Zhang et al . 2002) (Kijas et al . 2002) (Dekomien et al . 2010) (Goldstein, personal communication) (Ahonen et al . 2013; Winkler et al . 2013) (Goldstein et al . 2013a) (Downs and Mellersh. 2014)
(Clements et al . 1993; Suber et al . 1993) (Dekomien et al . 2000) (Kukekova et al . 2009) (Petersen-Jones et al . 1999) (Downs et al . 2013) (Goldstein et al . 2010b) (Zangerl et al . 2006)
References
b
rcd4-affected breeds: English/Gordon/Irish/Llwellyn Setters, Polish Lowland Sheepdog, Tibetan Terrier. PRCD-affected breeds: American Cocker Spaniel, American Eskimo Dog, Australian Cattle Dog, Australian Shepherd, Australian Stumpy Tail Cattle Dog, Bolonka Zwetna, Chesapeake Bay Retriever, Chinese Crested, English Cocker Spaniel, English Shepherd, Entlebucher Mountain Dog, Finnish Lapphund, Giant Schnauzer, Golden Retriever, Karelian Bear Dog, Kuvasz, Labrador Retriever, Lapponian Herder, Norwegian Elkhound, Poodle (Miniature/Toy/Standard), Nova Scotia Duck Tolling Retriever, Portuguese Water Dog, Spanish Water Dog, Swedish Lapphund, Yorkshire Terrier. c CEA-affected breeds: Australian Shepherd, Bearded Collie, Boykin Spaniel, Border Collie, Hokkaido Dog, Lancashire Heeler, Nova Scotia Duck Tolling Retriever, Rough/Smooth Collie, Shetland Sheepdog, Silken Windhound, Longhaired Whippet. d CMR1-affected breeds: American Bulldog, Australian Shepherd, Boerboel, Bullmastiff, Cane Corso, Dogue de Bordeaux, Great Pyrenees, Old English Mastiff, Perro de Presa Canario, English/American Bulldog.
a
Modified from Miyadera et al. (2012a). The Approach column shows the method used to identify the chromosomal location and/or the gene, and the type of sample population used. LA, linkage analysis; N.D., not determined; GWAS, genome-wide association study.
CMR1 CMR2 CMR3
breedsd
Samoyed
RD/OSD2 breedsc
Labrador Retriever
RD/OSD1
CD Briard
CRDNPHP4
Cone-rod dystrophy STATIONARY DISORDERS Cone degeneration (achromatopsia, day blindness)
cLCA
crd3 cord1
Cone-rod dystrophy Cone-rod dystrophy
Cone degeneration (achromatopsia, day blindness) Canine LCA DEVELOPMENTAL DISORDERS Retinal dysplasia/ Oculo-skeletal dysplasia 1 Retinal dysplasia/ Oculo-skeletal dysplasia 2 Collie eye anomaly OTHERS Canine multifocal retinopathy 1 Canine multifocal retinopathy 2 Canine multifocal retinopathy 3
American Staffordshire Terrier American Pit Bull/Staffordshire Terrier Glen of Imaal Terrier Miniature Longhaired/Smooth/ Wirehaired Dachshunds Standard Wirehaired Dachshund
crd1 crd2
Alaskan Malamute, Miniature Australian shepherd, Siberian Husky German Shorthaired Pointer
Golden Retriever Golden Retriever Samoyed, Siberian Husky Mongrel Bullmastiff, English Mastiff Schapendoes Italian Greyhound Papillon Basenji Tibetan Spaniel/Terrier
GR_PRA1 GR_PRA2 XLPRA1 XLPRA2 ADPRA gPRACCDC66 IG_PRA1 Pap_PRA1 Bas_PRA1 PRA3
CD
Irish Setter Sloughi Collie Cardigan Welsh Corgi Setters and othersa Norwegian Elkhound Multiple breedsb
rcd1 rcd1a rcd2 rcd3 rcd4 erd PRCD
Breed
PROGRESSIVE DISORDERS Progressive Retinal Atrophy (PRA) Rod-cone dysplasia 1 Rod-cone dysplasia 1a Rod-cone dysplasia 2 Rod-cone dysplasia 3 Rod-cone dysplasia 4 Early retinal degeneration Progressive rod-cone degeneration Golden Retriever PRA 1 Golden Retriever PRA 2 X-linked PRA 1 X-linked PRA 2 Autosomal dominant PRA Generalized PRA Italian Greyhound PRA Papillon PRA Basenji PRA PRA3 Cone-Rod Dystrophy (CRD) Cone-rod dystrophy Cone-rod dystrophy
Disease Symbol
Table 1. Canine retinal diseases and associated genes
Disease Name
Inherited retinal diseases in dogs
MIYADERA
across canine breeds is becoming more complicated than we
inspection to certain genes of interest, a whole genome scan
initially thought. When it comes to the mapping process, as
interrogates the entire genome necessitating no prior
linkage disequilibrium (LD) is extensive within a breed (0.5-
assumption or knowledge of the target locus. The current
5Mb, depending on the breed) (Gray et al . 2009; Sutter et al .
tools of gene/mutation discovery are based on the 7.5X
2004), increasing the likelihood of two loci within a block of
sequence of the Boxer dog (Lindblad-Toh et al . 2005), the
chromosome co-segregating, the initial mapping can be done
earlier 1.5X Poodle genome sequence (Kirkness et al . 2003),
using fewer genetic markers. The disadvantage of long LD
as well as single nucleotide polymorphisms (SNPs)
after mapping is that a large chromosomal area spanning
information across multiple breeds.
many genes will need to be studied, and this has often led to
Linkage mapping and positional cloning strategies – A
a two-step approach of an initial mapping followed by a
linkage analysis (LA) mapping utilizes informative families
fine-mapping in the search for the causative gene/mutation
containing affected and non-affected dogs. Co-segregation
of canine RDs.
of genetic markers with the phenotype is analyzed to identify recombination, or lack thereof. Subsequently, the
4. 2. Candidate gene analysis
region of interest may be further narrowed by fine-mapping,
Candidate genes are selected based on existing
followed by positional candidate gene analysis.
knowledge of the biochemistry or associated diseases, or on
Genome-wide association studies (GWAS) –With the
information from other species such as humans and mice.
development of microarray-based DNA chips, high-
Early studies have largely relied on a candidate gene
throughput SNP analysis has become feasible in both
approach largely because of the lack of alternative canine
humans and domestic animals (Andersson 2009). In a
genomic tools.
GWAS approach, SNP markers pre-determined across the
The first canine RD mutation was found by screening a
genome are used to assess association with the disease. For
candidate gene PDE6B for rcd1 in Irish Setters (Clements et
canine traits, the 170K canine SNP chip (Illumina) is the
al . 1993; Suber et al . 1993), based on phenotypic and
currently most commonly used platform. It has been
biochemical similarity to the rd mice with a PDE6B
postulated that 20 cases and 20 controls, and 50 cases and 50
mutation (Bowes et al . 1990; Farber and Lolley 1974).
controls should be sufficient for mapping autosomal
Candidate gene approaches have since identified mutations
recessive and dominant traits, respectively using 10,000-
for rcd1a in Sloughi dogs (PDE6B ) (Dekomien et al . 2000),
30,000 SNPs chips (Lindblad-Toh et al . 2005), due to the
for rcd3 in Cardigan Welsh Corgis (PDE6A ) (Petersen-Jones
conveniently long linkage disequilibrium within a given
et al . 1999), and ADPRA in Bullmastiffs and English
breed (0.5-5Mb) (Gray et al . 2009; Sutter et al . 2004). Since
Mastiffs (RHO ) (Kijas et al . 2002). However, while
the first report of CRD in Standard Wirehaired Dachshunds
successes are often highlighted, there have been too many
(Wiik et al . 2008), at least 10 forms of canine RDs have
failed candidate gene screening attempts mostly
been mapped by GWAS (Table 1). These studies typically
unpublished, questioning its efficacy as the primary
map a relatively large chromosomal region, followed by
approach for mutation discovery in canine RDs (Aguirre-
positional candidate gene screening with/without fine
Hernandez and Sargan 2005). Further, while there are
mapping.
hundreds of genes so far associated with human RDs, screening for all the candidate genes has become unrealistic.
4. 4. Next-generation sequencing
Still, phenotype-based selection of BEST1 in CMR1,
More recently, the next-generation sequencing (NGS)
CMR2, and CMR3 has proven to be a success (Guziewicz et
technique has been applied to canine RD studies aiding in
al . 2007; Zangerl et al . 2010) guided by the high clinical and
mapping, fine mapping and functional analysis. NGS is
pathologic resemblances to Best macular dystrophy in
based on a massively-parallel, high-throughput sequencing
humans.
which produces millions of short sequences to be analyzed by extensive bioinformatics. A targeted-NGS approach is
4. 3. Whole-genome scan
used where a genetic locus has been mapped by a whole
Unlike candidate gene studies which limit the
genome scan. The genomic area of interest is then captured
The Journal of Animal Genetics (2014) 42, 79–89
82
Inherited retinal diseases in dogs
using custom-designed probes, and the enriched DNA is
presence of alternative form(s) of RD independent of the
subjected to NGS. GWAS-guided fine mapping by targeted-
already known RD. Such genetic heterogeneity has been a
NGS has so far identified mutations associated with GR_
dilemma for the DNA test providers as well as the users
PRA2 in Golden Retrievers (Downs et al . 2014) and PRA3
alike, and those dogs with discordant DNA test results
in Tibetan Spaniels/Terriers (Downs & Mellersh 2014).
become subjects for further investigation to identify the
Exome-sequencing is a form of targeted-NGS which
missing RD gene/mutation for the breed.
captures only the predetermined known canine exon regions
Early-onset PRA in Portuguese Water Dogs (PWD) –
for NGS. Ahonen et al . (2013) used exome-sequencing as a
PWDs are one of many breeds affected with PRCD-PRA.
parallel approach to identify the CNGB1 mutation
PRCD in this breed is a late-onset form of PRA typically
associated with Papillon PRA. Other NGS applications
presenting with night blindness at mid age gradually
include RNA-seq and whole-genome sequencing, each of
progressing into total blindness. The author and colleagues
which has already proven useful in mapping and mutation
have recently identified a new form of PRA in PWDs that
screening of non-ocular diseases in dogs (Forman et al .
are not associated with PRCD. It presents as an early-onset
2012; Drogemuller et al . 2013; Guo et al . 2014). RNA-seq
PRA where night blindness is recognized at 1-2 years of age
uses tissue-specific RNA as the template and provides a
followed by rapid progression to total blindness. Affected
complete snapshot of transcripts expressed in the tissue of
cases and controls have been identified for a GWAS.
interest, allowing assessment of the level of expression (quantitative) as well as structural and sequence changes
5.3. Modifiers
(qualitative).
While RDs have largely been considered as simple monogenic traits, there is increasing evidence to suggest that
5. Emerging complexity of canine RDs
more than one gene mutation could be affecting the
5.1. Multiple forms of RD per breed
phenotypic outcome. This is not surprising considering the
After a DNA test for certain RD has become available,
number of genes (>220) associated with human RDs to date.
some of the target breed populations have been found to
Genetic polymorphisms in any of these genes albeit non-
contain dogs that are tested as genetically“clear”but are in
pathogenic on their own, could potentially function as
fact clinically affected with RD (Table 2). This indicates the
modifiers altering the effect of a primary mutation. PRCD-
Table 2. Examples of breeds affected with more than one confirmed RDs Breed
Disease 1
Disease 2
Disease 3
Disease 4
Australian Shepherd
PRCD-PRA
CEA
CMR1
CD
Golden Retriever
PRCD-PRA
GR_PRA1
GR_PRA2
-
Goldendoodle
PRCD-PRA
GR_PRA1
GR_PRA2
-
Norwegian Elkhound
PRCD-PRA
rd
erd
-
Lapponian Herder
PRCD-PRA
CMR3
-
-
Irish Setter
rcd1-PRA
rcd4-PRA
-
-
Labrador Retriever
PRCD-PRA
RD/OSD1
-
-
Nova Scotia Duck Tolling Retriever
PRCD-PRA
CEA
-
-
cord1
-
-
NPHP4
Miniature Wirehaired Dachshund
CRD
Bullmastiff/English Mastiff
ADPRA
CMR1
-
-
Samoyed
XLPRA1
RD/OSD2
-
-
Tibetan Terrier
rcd4-PRA
PRA3
-
-
Collie
rcd2-PRA
CEA
-
-
Portuguese Water Dog
PRCD-PRA
early-onset PRA
-
-
The Journal of Animal Genetics (2014) 42, 79–89
83
MIYADERA
PRA affecting a variety of dog breeds show variable yet
Goldendoodles). Meanwhile, new breeds affected with the
breed-specific ages of onset. This is thought to be largely
existing forms of RDs are found each year and added to the
attributed to breed-specific genetic backgrounds that are
list.
collections of genetic polymorphisms across the genome.
The advantage of DNA tests is its ability to detect
Modifier genes may or may not be retina-specific, and could
unaffected carriers and subclinical cases prior to disease
be genes interacting with any retinal gene, further inflating
onset. It is important to note that other unidentified forms of
the number of potential modifier genes.
RDs will not be screened with the existing DNA tests. In
The author has previously demonstrated the presence of
many situations, following identification of a mutation,
a genetic modifier in cone-rod dystrophy 1 (cord1) in
coordinated effort is implemented by the breeding
Miniature Longhaired Dachshunds (MLHDs). The condition
community leading to a substantial decrease in the disease
was first described as an autosomal recessive CRD with an
and mutant allele frequencies in the breed. Meanwhile, other
early-onset (age