Dog Models for Blinding Inherited Retinal Dystrophies

Abstract

Spontaneous canine models exist for several inherited retinal dystrophies. This review will summarize the models and indicate where they have been used in translational gene therapy trials. The RPE65 gene therapy trials to treat childhood blindness are a good example of how studies in dogs have contributed to therapy development. Outcomes in human clinical trials are compared and contrasted with the result of the preclinical dog trials.

Introduction

Development and testing of therapies in animal models is critical prior to commencing human clinical trials. Animal studies can provide proof of principle for the therapeutic approach, can be used to modify or refine the approach, and also provide information on the safety of the treatment. Studies utilizing spontaneously occurring canine models of inherited retinal dystrophies have been important in the translation of therapies from the laboratory to the clinic—most notably, gene therapy for the treatment of Leber congenital amaurosis (LCA) caused by RPE65 gene mutations. Several additional canine retinal dystrophy models are currently being used for preclinical therapeutic trials, some of which are currently, or soon will be, translated from these large-animal studies to human clinical trials.

Existing Canine Models

Retinal dystrophies encompass several groups of conditions, including LCA, retinitis pigmentosa (RP), cone–rod dystrophy, achromatopsia, and macular dystrophy. Dogs with spontaneous inherited retinal dystrophies that provide models within each of these categories have been identified (Table 1). The studies to identify the gene mutations underlying retinal dystrophies in dogs have served two purposes: first, the conditions result in vision loss in pet or working animals and identifying the causal mutations enables DNA testing, selective breeding, and eradication of the condition; second, and most relevant for this review, is that often mutations occur in the homologous gene in both dog and man and the resulting phenotype in both species is comparable. Establishing a research colony of affected animals allows detailed examination of the pathophysiological processes and the testing of therapies.

Table 1.

Dog Models of Inherited Retinal Degenerations with Identified Gene Mutations

Condition	Gene	Breed	Reference to mutation	References to therapy
Leber congenital amaurosis models
Congenital stationary night blindness/retinal dystrophy	Rpe65	Briard	14	16–20,22,24,41,43,45–47,94–100
Autosomal recessive retinitis pigmentosa models
Rod–cone dysplasia type 1	Pde6b	Irish setter, red and white setter	52	54
Rod–cone dysplasia type 1a	Pde6b	Sloughi	101
(Originally classified as a cone–rod dystrophy type 1)	Pde6b	American Staffordshire terrier	102
Rod–cone dysplasia type 2	C1orf36	Collie	103
Rod–cone dysplasia type 3	Pde6a	Cardigan Welsh corgi	53	56
Early retinal degeneration	Stk38l	Norwegian elkhound	104
Golden retriever PRA type 1	Slc4A3	Golden retriever	105
Progressive retinal atrophy of Papillons type 1	Cngb1	Papillon	106
Progressive retinal atrophy	Ccdc66	Schapendoes	113
Progressive retinal atrophy	Sag	Basenji	107
Progressive rod–cone degeneration	Prcd	Multiple breeds	108
Rod–cone degeneration type 4	C2orf71	Gordon setter	109
Autosomal dominant retinitis pigmentosa models
Dominant progressive retinal atrophy	Rhodopsin	Mastiff	57	61
X-linked retinitis pigmentosa models
X-linked progressive retinal atrophy 1	Rpgr	Siberian huskySamoyed	65	66–68
X-linked progressive retinal atrophy 2	Rpgr	Mongrel	65	66–68
Cone–rod dystrophy models
Cone–rod dystrophy 2	Iqcb1	Pit bull terrier	102
Cone–rod dystrophy 3	Adam9	Glen of Imaal terrier	110
Cone–rod dystrophy	Nphp4	Standard wirehaired dachshund	111
Cone–rod dystrophy	Rpgrip1	Miniature longhaired dachshund	69	74
Achromatopsia models
Achromatopsia	Cngb3	Alaskan malamute	77	79,81
Achromatopsia	Cngb3	German shorthaired pointer	77	79,81
Achromatopsia	Cnga3	German shepherd dog	112
Macular dystrophy models
Canine multifocal retinopathy 1	Best 1	Mastiff and related breeds	89	92
Canine multifocal retinopathy 2	Best 1	Coton de Tulear	89
Canine multifocal retinopathy 3	Best 1	Lapponian herder	90	92

Those in bold have had reported gene therapy trials.

Dog eyes show many similarities to human eyes, for example, in size and morphology and also in density of photoreceptor distribution. The human retina has a region known as the macula, which has a high photoreceptor density and contains the fovea, a cone-only region, from which the inner retinal neurons are displaced to enable highest visual acuity. The macula and fovea are important for high visual acuity color vision. Laboratory rodents that are commonly used as models for retinal dystrophies lack a central retinal region of high photoreceptor density. One major advantage of the dog models is that there is an area centralis that has a higher photoreceptor density, particularly of cones, than the peripheral retina, making it analogous to the human macula.¹ Although the photoreceptor packing density is not so high as in the human macula and dogs do not have a true fovea, the center of the area centralis in the dog has recently been shown to have a small foveal-like “bouquet” of cones.²

Gene Therapy for Leber Congenital Amaurosis

Leber congenital amaurosis (LCA) is a genetically heterogeneous condition that results in progressive vision loss starting in childhood. It is an uncommon condition, with incidence being reported to be between 3 in 100,000 and 1 in 81,000.^3,4 Mutations of the retinal pigment epithelium (RPE) expressed gene, RPE65, account for about 6% of LCA cases.⁵ Gene therapy for LCA resulting from mutations in RPE65 (LCA^RPE65) highlights the use of a large-animal model for proof-of-principle studies. RPE65 is a retinoid isomerase forming a required part of the visual cycle that supplies and recycles light-sensitive retinoid (11-cis retinal) to the rod and cone photoreceptors.^6,7 Although there may be a second source of retinoid for the cone photoreceptors,⁸ it is insufficient in the absence of the RPE visual cycle to allow for normal function.⁹

In the absence of normal RPE65 activity, the rod and cone photoreceptor opsins do not receive the retinoid 11-cis retinal, which they require to combine with the opsin protein to form the light-sensitive visual pigments. Lack of 11-cis retinal results in very reduced sensitivity and function of the photoreceptors such that affected patients require bright lighting levels for vision. There is thus a very early lack of visual function prior to an advanced loss of photoreceptors. This “disconnect” between loss of function and loss of photoreceptors provides a “window of opportunity” for therapeutic intervention aiming to establish visual cycle activity and thus supply the remaining photoreceptors with retinoid. This proved to be important for the gene augmentation therapeutic approach. The poorly functioning photoreceptors of LCA^RPE65 patients do degenerate and the retinal pigment epithelium that houses the RPE65 protein and plays a major role in the visual cycle accumulates retinyl esters because of the blockage of the visual cycle (for a review, see ref.¹⁰).

A spontaneous retinal dystrophy dog model, initially described as an autosomal recessive congenital stationary night blindness, had been recognized and studied for many years.¹¹ The affected dogs were blind at lower lighting levels (but could see in brighter lights), had very reduced electroretinographic responses (electrical retinal responses to stimulation with flashes of light), and had reduced pupillary light reflexes and nystagmus (the presence of nystagmus is a common feature of early vision loss). Notably, there was a very slow loss of photoreceptors over many years, during which accumulations of vacuoles in the RPE became pronounced.^12,13 Eventually, molecular genetic studies revealed the dog model to have a mutation in Rpe65. The affected dog breed was the briard (a French breed), and because of the initial phenotyping studies by a Swedish group, the model was often referred to as the Swedish briard. The gene mutation in the “Swedish” briard is a 4 bp deletion in the Rpe65 gene with a resultant frameshift and premature termination codon.¹⁴

Subsequent to the announcement of this finding at a research meeting (Association for Research in Vision and Ophthalmology [ARVO], 1998), briards in the United States with retinal dystrophy were also found to have the same mutation as the Swedish briard.¹⁵ Concurrent progress in gene transfer to the retina using adeno-associated viral (AAV) vectors provided a great opportunity to use the newly identified dog model for LCA^RPE65 for gene therapy proof-of-principle trials. Acland et al. announced at the ARVO meeting in 2001 the dramatic results of the gene therapy trials using the Rpe65-mutant briard dog.¹⁶ AAV vectors carrying the Rpe65 cDNA were introduced by subretinal injection and shown to transduce the RPE restoring the visual cycle and resulting in a remarkable improvement in retinal function as assessed by the electroretinogram (ERG) and also in vision as assessed by obstacle course vision testing.¹⁶

This and follow-on studies from the same and other groups further investigated the therapy and outcome. The therapy in the dog model led not only to a functional visual cycle that supplied 11-cis retinal to the photoreceptors providing an excellent improvement in photoreceptor function and sensitivity as assessed by ERG and vision testing, but also to the normalization of pupillary light reflexes, reduction in nystagmus, and increased visual evoked cortical activity.^16–24 Furthermore, the rescue was long-lasting.¹⁷ (See Fig. 1 for results of AAV gene augmentation in a mutant dog.)

FIG. 1.

RPE65 gene therapy in a dog. (A) Fundus photographs of the treated eye of an Rpe65-mutant dog preinjection, immediately postinjection, and 6 months postinjection. Note the large region of retinal detachment resulting from the subretinal administration of the viral vector. (B) Electroretinographic outcome. Part of a dark-adapted intensity response series and 33 Hz light-adapted cone flicker response is shown from the treated eye and control eye of an Rpe65-mutant dog.^43,47 AAV2/2hRPE65p.hRPE65 vector had been delivered by subretinal injection 4 months previously in the treated eye. The ERG tracings show that there is excellent restoration of both rod- and cone-mediated ERG responses. The magnitude of ERG amplitudes approaches those of normal dogs. (Dark-adapted intensities from top to bottom: 0.0016, 0.0099, 0.064, and 0.399 cdS/m². Light-adapted 33 Hz cone flicker intensity 2.5 cdS/m² superimposed on a rod suppressing 30 cd/m² white-background light). ERG, electroretinogram.

The proof-of-principle gene therapy studies in the briard dog preceded gene therapy studies in Rpe65 mouse models.^25–27 On the basis of the success of gene therapy in the dog, three separate groups started phase I dose escalation clinical trials (trials NCT00643747, NCT00516477, and NCT00481546; see Table 2). All three groups reported the initial results of the clinical trials in 2008, showing results from the first three patients treated in each trial.^28–30 The initial reports indicated that there were no major adverse effects and that there was evidence of improved visual function in some of the patients.^28–30 Subsequently, additional patients have been injected and more detailed analyses of the outcomes have been published.

Table 2.

Current Human Clinical Trials for RPE65 Gene Therapy from ClinicalTrials.gov

ClinicalTrials.gov identifier	Date	Title	Phase	Sponsor	Locations
NCT00481546	5/31/2007	Phase I Trial of Ocular Subretinal Injection of a Recombinant Adeno-Associated Virus (rAAV2-CBSB-hRPE65) Gene Vector to Patients with Retinal Disease Due to RPE65 Mutations (Clinical Trials of Gene Therapy for Leber Congenital Amaurosis)	I	University of Pennsylvania	Shands Children's Hospital, University of Florida, Gainesville, FL, Scheie Eye Institute, University of Pennsylvania, Philadelphia, PA
NCT00516477	8/13/2007	A Phase 1 Safety Study in Subjects with Leber Congenital Amaurosis (LCA) Using Adeno-Associated Viral Vector to Deliver the Gene for Human RPE65 into the Retinal Pigment Epithelium (RPE) [AAV2-hRPE65v2-101]	I	Spark Therapeutics LLC	The Children's Hospital of Philadelphia, Philadelphia, PA
NCT00643747	3/20/2008	An Open-label Dose Escalation Study of an Adeno-associated Virus Vector (AAV2/2-hRPE65p-hRPE65) for Gene Therapy of Severe Early-onset Retinal Degeneration	I/II	University College London	Moorfields Eye Hospital NHS Foundation Trust, London, UK
NCT00749957	9/8/2008	A Multiple-Site, Phase 1/2, Safety and Efficacy Trial of a Recombinant Adeno-Associated Virus Vector Expressing RPE65 (rAAV2-CB-hRPE65) in Patients with Leber Congenital Amaurosis Type 2	I/II	AGTC	University of Massachusetts Medical School, Worcester, MA. Casey Eye Institute, Oregon Health & Science University, Portland, OR
NCT00821340	1/12/2009	Phase I Trial of Ocular Subretinal Injection of a Recombinant Adeno-Associated Virus (rAAV2-hRPE65) Gene Vector to Patients with Retinal Disease Due to RPE65 Mutations	I	Hadassah Medical Organization	Hadassah Medical Organization, Jerusalem, Israel
NCT00999609	10/21/2009	A Safety and Efficacy Study in Subjects with Leber Congenital Amaurosis (LCA) Using Adeno-Associated Viral Vector to Deliver the Gene for Human RPE65 to the Retinal Pigment Epithelium (RPE) [AAV2-hRPE65v2-301]	III	Spark Therapeutics LLC	University of Iowa, Children's Hospital of Philadelphia, Philadelphia, PA
NCT01208389	9/22/2010	A Follow-On Study to Evaluate the Safety of Re-Administration of Adeno-Associated Viral Vector Containing the Gene for Human RPE65 [AAV2-hRPE65v2] to the Contralateral Eye in Subjects with Leber Congenital Amaurosis (LCA) Previously Enrolled in a Phase 1 Study	I/II	Spark Therapeutics LLC	The Children's Hospital of Philadelphia, Philadelphia, PA
NCT01496040	11/24/2011	Prospective Monocentric Open-Label Non-Randomized Uncontrolled Phase I/II Clinical Gene Therapy Protocol for the Treatment of Retinal Dystrophy Caused by Defects in RPE65	I/II	Nantes University Hospital	CHU Nantes, Nantes, France

The group based at University College London has treated 12 patients (NCT00643747); the group headed by The Children's Hospital of Philadelphia has reported on the outcome of 12 patients (NCT00516477, NCT00999609, and NCT01208389), and the group headed by University of Pennsylvania has reported on 15 patients (NCT00481546 and NCT00749957). Subsequently, clinical trials have started in Jerusalem, sponsored by Hadassah Medical Organization (NCT00821340),³¹ and in France, at Nantes University Hospital (NCT01496040). Although each group has used an AAV2 construct injected subretinally, the constructs differed as to which promoter was used and whether an intron or kozak sequence was included. Furthermore, the titers and volume injected varied between the trials. No significant immune responses were reported for any study.^32,33

Some procedural-related complications did develop, for example, a macula hole in one patient,²⁹ and as patients were analyzed it became apparent that causing a foveal detachment during subretinal administration of the vector could be deleterious and resulted in foveal thinning in some patients.³⁴ Procedural changes have been implemented so that injections avoid causing foveal detachment, and at least one group has started injecting in two sites in the same eye and speculated that in the future three sites of injection may provide optimal retinal coverage with minimal risk of procedural complications.³⁴ Assessment of outcome of therapy has consisted of a battery of different specialized tests designed to detect and quantify improvements in visual function over baseline. These included testing ability to perceive light (such as full-field sensitivity and microperimetry), visual acuity, ability to negotiate an obstacle course, investigation of eye movements and fixation, pupillometry, electrophysiological testing, and functional MRI (see ref.¹⁰ for a comprehensive review).

The results have been very encouraging, with patients having improved ability to perceive light, improved visual acuity in some instances, improved pupillary responses and visual cortex responses, and for some patients improved ability to negotiate obstacles.^28–42 Close examination of the outcome for patients in one trial showed that several hours of dark adaptation were required to achieve maximum rod sensitivity in the treated area, whereas cones showed a normal speed of recovery. The markedly slow regeneration of rods was an indication of incomplete restoration of the visual cycle, and the possibility that a resistive barrier slowed the transport or diffusion of 11-cis retinal from the RPE to the photoreceptors was raised.⁴⁰

Despite the safe and encouraging improvement in visual function, the degree of rescue achieved in human patients has been nowhere near that achieved in the preclinical Rpe65-mutant dog trials. With the exception of one patient with a transient improvement in the multifocal ERG,³⁸ there was no change in ERG responses. In contrast, in dogs there is a remarkable improvement in the ERG, indicating restoration of function close to normal levels (Fig. 1). Vision in dogs after treatment appears very close to normal, while human patients remain visually impaired. So why is there such a difference in outcome between dog and human? One major reason is that human patients have a much faster relative decline in photoreceptor numbers compared with dogs and invariably already have a marked loss of photoreceptors when treated, whereas the dogs treated up to middle age have good preservation of photoreceptors that respond well to restoration of visual cycle function.⁴¹

Cideciyan et al. suggested that use of older Rpe65-mutant dogs is required to more accurately model the stage of degeneration present in human patients when they are treated.^41,43 We found that robust rescue could still be achieved in middle-aged dogs, which is a reflection of the slow degeneration that occurs in the model.⁴³ One of the authors (S.P.J.) has noted that a breeding line of Rpe65-mutant dogs maintained at Michigan State University develops an early photoreceptor degeneration in the center of the area centralis. This is a region of higher photoreceptor density and it is conceivable that this particular region of the dog retina more closely models the dynamics of photoreceptor loss that occurs in human patients.⁴⁴ Other breeding lines have not been reported to develop this same early degeneration in the area centralis; this may reflect a difference in background genetics. Additional closer examination of the retinas of affected dogs showed that, although both rods and M/L cones are very slowly lost, the S-cones are rapidly lost in the dog model.⁴⁵

The initial treatment protocols have been modified based on the experiences with the first patients, and now in a follow-on study (NCT01208389), the second eye of some patients has been treated successfully and with no adverse immune responses or reduction in the rescue previously achieved in the first treated eye.³⁷ In preparation for human trials to treat the second eye of patients, preclinical studies in the Rpe65-mutant dogs had been performed. These showed that administration of vector to the second eye was safe, did not cause an adverse immune response, and allowed for transduction of the RPE and comparable rescue to that of the first treated eye.^46,47

Cideciyan et al. investigated photoreceptor loss in the gene therapy-treated retinal areas of human patients and found that although the therapy did restore function to photoreceptors, it did not lead to their preservation in the long-term. They also investigated long-term preservation in Rpe65-mutant dogs and showed in older dogs in which photoreceptor loss was already established at the time of treatment that, although function was successfully restored, long-term photoreceptor preservation was not achieved.⁴¹ Clearly, a greater understanding of the reasons why photoreceptors die despite restoration of function and why human rod recovery kinetics remain delayed is needed. This may allow modifications to the treatment protocol to optimize outcomes by improving rod kinetics and enabling preservation of photoreceptors. Studies of older dogs with established retinal degeneration or of the more rapidly degenerating area centralis that is present in some lines may be valuable for these investigations.

Gene Therapy Using Canine Retinitis Pigmentosa Models

Retinitis pigmentosa (RP) is a genetically heterogeneous condition with a growing number of mapped loci and causal genes being identified for the recessive (35 loci and 32 genes), dominant (21 loci and 20 genes), and X-linked (5 loci and 2 genes) modes of inheritance (information on loci and genes for nonsyndromic RP from RetNet⁴⁸ as of December 2014). Rod photoreceptors are initially affected, leading to night blindness and loss of peripheral vision. With progression, cones are also lost, often leading to blindness. With photoreceptor death, characteristic fundus “bone-spicule”-shaped pigment deposits develop, giving the condition its name. The dog equivalent of RP is known as progressive retinal atrophy (PRA). It occurs across many dog breeds, and an increasing number of the causal mutations have been identified (see Table 1). Colonies of some of the forms of PRA have been established, allowing for more detailed investigations of the phenotype, the disease mechanisms, and also the opportunity to test therapeutic interventions such as gene therapy.

Gene therapy reports in canine models of RP include those for recessive RP (Pde6a and Pde6b), for dominant RP (Opsin), and for X-linked RP (Rpgr). Translational studies for some of these are currently underway.

Gene therapy in dog models of autosomal recessive RP

There are currently publication and conference reports of two gene therapy studies using dog models of autosomal recessive RP. Both of these have mutations in rod phototransduction genes: the alpha and beta subunits of cyclic GMP (cGMP) phosphodiesterase (Pde6a and Pde6b).

PDE6A and PDE6B mutations are responsible for 3–4% and 4–5% of human autosomal recessive RP cases, respectively.⁴⁹ The protein products (cGMP phosphodiesterase alpha and beta subunits: PDEα and PDEβ) combine as a heterotrimer with two inhibitory gamma subunits, and when the holoenzyme is activated the α/β active subunits hydrolyze cGMP. Reduced cGMP levels lead to closure of cyclic nucleotide gated channels and hyperpolarization of the rod photoreceptor. Dog models with null mutations in Pde6a and Pde6b exist and have very similar phenotypes.^50,51 They both result in a stunting of photoreceptors' outer segments during retinal maturation with a relatively rapid loss of rod photoreceptors followed by a slower secondary loss of cone photoreceptors. Affected dogs are night blind with a lack of rod-mediated ERG responses.

The Pde6b model (known as rod–cone dysplasia type 1; rcd1) has a nonsense mutation that results in a lack of Pdeβ, an accumulation of cGMP in the affected rods, and death of the rod cells most probably by apoptotis.⁵² The Pde6a model (known as rod–cone dysplasia type 3; rcd3) has a 1 bp deletion with a resulting frameshift and premature stop codon and a similar phenotype to that of rcd1 dogs.⁵³ One difference between the models is that in the absence of Pdeβ the alpha and gamma subunits of Pde can still be detected in the rod outer segment prior to degeneration; however, in the absence of Pdeα, the beta and gamma subunits appear not to transport to outer segments and are not detectable on Western blots.⁵¹

Gene therapy trials performed on eight Pde6b mutant dogs have been published. The dogs were treated at 20 days of age (because of the early onset of the photoreceptor degeneration); 4 received an AAV2/5 vector and 4 an AAV2/8 vector, the latter at a 10-fold higher titer, via subretinal injections. The canine Pde6b cDNA was delivered under control of a human rhodopsin kinase promoter in both vectors. The treated eyes were reported to have improved rod-mediated ERGs, and dim light vision testing of each eye in turn showed that when the treated eye was uncovered the dogs were able to negotiate obstacles, but when the untreated eye was uncovered they acted as if they were blind. Retinal morphology showed preservation of the photoreceptor layer in the treated regions, and rod outer segments expressing Pde6 and Gnat1 were detected in the treated areas on immunohistochemical (IHC) analysis.⁵⁴

Pde6b mutant dogs have also been used for an alternative gene therapy strategy to restore vision. This is intended for use in patients once photoreceptors have been lost, and therefore gene augmentation to provide a normal copy of the mutated gene is no longer applicable. The approach is to transform inner retinal neurons, which remain following the loss of the photoreceptors, into light-sensitive cells. This is achieved by expressing a light-sensitive channel in the cell membrane. Either bipolar cells or ganglion cells can be targeted. A recent publication reported on a study that utilized a so-called optopharmacological approach. An AAV vector was used to transduce the target cells to express a light-gated ionotropic glutamate receptor (LiGluR). To enable the LiGluR to respond to light, a photoswitchable molecule to which the ligand for LiGluR (glutamate) is attached (photoswitchable tethered ligand maleimide–azobenzene–glutamate; MAG) is delivered by an intravitreal injection.

When the MAG bound to the LiGluR is light activated, it undergoes a conformational change, making the glutamate accessible to the LiGluR, leading to activation of the channel and a change in cell membrane potential and thus neuronal signaling. Work in the retinal degeneration 1 (rd1) mouse that has a null mutation in Pde6b showed that this approach resulted in recordable light-induced responses in ganglion cells. Furthermore, it was shown that the expression of LiGluR coupled with intravitreous MAG was able to restore light-induced behavior in the photoreceptor-less mice. In a large-animal follow-on study, blind Pde6b-mutant dogs were used. An AAV construct was administered intravitreally to transduce ganglion cells with LiGuR. To show that the transduced ganglion cells had gained the ability to respond to light, a multielectrode array recording from retinal explants was performed. This showed that the transduced ganglion cells had a light-inducible response, but only in the presence of MAG.⁵⁵ Further studies of this promising approach are expected to lead to human clinical trials.

The laboratory of one of the authors of this review (S.P.J.) reported on therapy with the rcd3 Pde6a mutant dog model using an AAV2/8 vector with a ubiquitous promoter (short CBA promoter) driving expression of the canine Pde6a cDNA. Treated eyes showed evidence of rod photoreceptor function on ERG, improved ability to negotiate obstacles in very dim light, and preservation of the outer nuclear layer of the retina in the injected area. Encouragingly, robust Pde6 expression was detected in the outer segments of the photoreceptors on IHC using an antibody that recognizes all subunits of Pde6. Despite the use of a ubiquitous promoter, the Pde6 protein was not detected in other parts of the retina. There was, however, evidence of some deleterious effects on therapy with photoreceptor rosette formation and associated patches of retinal thinning. This was similar to changes seen in the Rpgr dog model when a vector delivering a shortened cDNA was used (see below). The cause of the rosette formation requires further investigation and will be important to understand.⁵⁶

Gene therapy in dog models of autosomal dominant RP

A dog model with an autosomal dominant phenotype was identified with a missense mutation (T4R) in rhodopsin.⁵⁷ The mutation renders the affected retina sensitive to light damage because of instability of the mutant opsin when it is not conjugated to the chromophore 11-cis retinal to form rhodopsin.^58,59 The effect of light exposure on progression of RP because of rhodopsin mutations needs careful consideration. In preparation for gene therapy experiments, development of a way of being able to visualize the retina during the subretinal injection process without causing light-induced damage was required. An operating microscope with an infrared viewing system was developed for this purpose.⁶⁰ A study was reported that investigated whether AAV-mediated expression of mouse or human rhodopsin in the heterozygous mutant dog was safe and whether it provided protection against light damage.

Subretinal injection (conducted using the infrared microscope) of AAV2/5 delivering a human or mouse rhodopsin cDNA under control of the mouse opsin promoter was performed in affected dogs. A mild retinal thinning was noted in the region of the subretinal injection although expression of either mouse or human rhodopsin did not appear to be cytotoxic. However, when the dogs were exposed to a bright light known to cause retinal damage in the mutant but not normal dogs, there was no evidence of a protective effect from the expression of the transgene.⁶¹ Development of a therapy to reduce expression of the mutant allele may be required.⁶²

Gene therapy in dog models of X-linked RP

Two dog models with RP GTPase regulator (Rpgr) mutations have been used in gene therapy trials. RPGR mutations result in X-linked retinal dystrophies with classifications including RP, cone–rod dystrophy, cone dystrophy, and macular degeneration. The splicing of RPGR is complex, with two groups of splice variants being expressed in the retina. One group is encoded by exons 1–13 plus 16–19 (Rpgr^ex1-19) and is highly expressed in developing photoreceptors.⁶³ The second group of splice variants consists of exons 1–13 plus varying amounts of an alternatively splice C-terminal exon 14/15 (referred to as open reading frame 15 [ORF15]) (Rpgr^ORF15) and appears in mature photoreceptors and is involved in trafficking of proteins through the photoreceptor cilium that connects the inner to outer segment. ORF15 is a common site for mutations in humans and is the site of the mutation in the two dog models. Despite both being caused by mutations in ORF15, the phenotypes are very different. The first, known as XLPRA1, has a milder rod–cone degeneration detectable at about 1 year of age.^64–66 The second mutation (XLPRA2) causes a much more severe phenotype with photoreceptor outer segment disorganization from as early as 4 weeks of age in hemizygous males.^64–66

Gene therapy in the XLPRA1 model was initially undertaken by administering a subretinal injection of AAV2/5 vector delivering a shortened version of canine RPGR exon 1-ORF15 cDNA (cshortRpgr^ex1-ORF15) under control of a mouse opsin (Mops) or human interstitial retinol binding protein (hIRBP) promoter. This resulted in patchy areas of retinal thinning associated with photoreceptor rosette formation.^66,67 Changing the delivered transgene to the full-length human RPGR^ex1-ORF15 under the hIRBP or human G protein-coupled receptor kinase 1 (hGRK1) promoter was more successful. In two XLPRA1 dogs injected at 28 weeks of age (prior to retinal degeneration) in one eye with the AAV2/5-hIRBP-hRPGR^ex1-ORF15 vector, there was preservation of the photoreceptor layer thickness in the injected transgene-expressing area, prevention of the mislocalization of rod opsin and M/L cone opsin, and also a lack of Müller cell activation (opsin mislocalization and activation of Müller cells resulting in upregulation of glial fibrillary acidic protein expression are features of disease progression in the untreated retina).

Additionally, the bipolar cells in the treated areas did not retract their dendrites that remained associated with rod spherules and cone pedicles.⁶⁶ Two XLPRA2 dogs were also treated by subretinal injection at 5 weeks of age. One received the construct with the hIRBP promoter and the second the same construct but with a hGRK promoter. There was preservation of outer retinal structure in both treated eyes, and opsin mislocalization was reversed in the eye injected with the hIRBP promoter and reduced in the eye treated with the hGRK promoter.⁶⁶ Subsequent reports of additional early-stage (5 weeks), midstage (12 weeks; n=3), and late-stage (26 weeks; n=3) XLPRA2 dogs treated with the hIRBP promoted construct showed structural preservation, as well as electroretinographic functional preservation.^67,68 These studies to investigate the degree of rescue at later-stage disease have important implications for future human therapy and are in apparent contrast to the RPE65 gene therapy study where once established photoreceptor loss continued despite restoration of function.⁴¹ Investigating the duration of rescue will also be important to see whether photoreceptor degeneration can preserve structure and function in the long-term.

Gene Therapy in a Canine Cone–Rod Dystrophy Model

Cone–rod dystrophies are a subset of photoreceptor dystrophies where cone photoreceptor dysfunction accompanies or precedes rod dysfunction (as opposed to RP where there is a rod-led dysfunction and/or degeneration). The cone–rod dystrophies show genetic heterogeneity, with one of the genes responsible being RP GTPase regulator-interacting protein 1 (RPGRIP1).

A mutation was identified in the canine Rpgrip1 gene in dogs with a cone–rod dystrophy.⁶⁹ Human RPGRIP1 mutations can cause recessive LCA or a recessive cone–rod dystrophy. It now appears that the situation is more complex in the canine model than a simple Rpgrip1 gene mutation causing the disease, with some studies discounting the mutation as disease causing and other implicating a necessary second locus on the same chromosome.^70–72 The putative disease-causing mutation was an insert in exon 2 of the gene that would be predicted to result in a frame-shift and premature termination codon. However, the gene has complex splicing patterns in the retina and not all variants use the exon bearing the mutation.⁷³ Despite the controversy as to the significance of the original Rgrip1 mutation, gene therapy has been performed on dogs derived from the original colony that was used to map the disease locus.⁷⁴

A previous study using the dogs from the initial colony had shown an early loss of cone ERG responses and a slower and later loss of rod ERG responses.⁷⁵ In the gene therapy trial, a vector construct of either AAV2/5 or AAV2/8, both with the canine Rpgrip1 cDNA driven by the human rhodopsin kinase promoter, was used. Five affected dogs were injected subretinally with the AAV2/5 vector and two dogs with the AAV2/8 construct. There was restoration of cone ERG responses in the treated eyes and a partial preservation of rod electroretinographic responses. Vision testing showed that treatment improved vision in both dim and bright conditions. The authors did note that, despite therapy preserving photoreceptors, advanced retinal thinning still developed.⁷⁴ This suggests that further modifications of treatment may be needed for long-term success of the therapy.

Gene Therapy Using Canine Achromatopsia Models

Two canine achromatopsia models resulting from different mutations involving the cone cyclic-nucleotide gated channel beta subunit gene, Cngb3, have been used in gene therapy trials. Achromatopsia, also known as day blindness or rod monochromacy, is the result of loss of cone function. It is an uncommon condition affecting approximately 1 in 30,000 people.⁷⁶ Affected patients have decreased visual acuity, a lack of color vision, photophobia, and nystagmus. They rely solely on rod photoreceptors for vision. Although there is genetic heterogeneity, mutations involving the subunits of the cone cyclic-nucleotide gated channel are most frequent, with those in CNGB3 being most prevalent. The dog achromatopsias present with very similar clinical signs to those of human patients. One of the models is the result of a large genomic deletion that includes the entire gene plus two neighboring genes.^77,78

The mutation was identified in the Alaskan malamute and more recently in an apparently unrelated breed, the miniature Australian shepherd.^77,78 The second canine model has a missense mutation in exon 6 of Cngb3 resulting in a functional null mutation. This was identified in the German shorthaired pointer breed of dog.⁷⁷ Both dog models develop cones normally, and they express many normal cone markers and only have a very gradual loss of cones. During the period of retinal maturation detectable, but reduced amplitude, cone ERG responses can be detected, but these are rapidly lost.⁷⁹ The very slow loss of cones provides a “window of opportunity” for gene therapy intervention, and both dog models have been used in proof-of-concept gene therapy studies. An AAV2/5 construct was used to deliver the human CNGB3 cDNA under control of promoters aimed to target cone photoreceptors. A promoter that effectively targeted dog M/L cones but not S cones had been identified and was successfully used to transduce the M/L cones in both models.^80,81 A dog with successful treatment had restored daytime vision as assessed by ability to negotiate a maze⁸² and cone function was recordable by the ERG.

Subsequent to successful gene therapy in the dog, the Cngb3-mutant mouse was also rescued by gene therapy.⁸³ The initial canine study, however, recognized that there were some dogs in which transgene expression occurred and yet detectable cone function did not result. It was noted that in dogs older than about 1 year of age the success rate was reduced despite the fact that the cones still appeared to be structurally normal. The reduced success rate in older dogs was addressed in a second study, where an intravitreal injection of ciliary neurotrophic factor (CNTF) was given 1 week before the viral vector construct (AAV2/5-PR2.1-hCNGB3) was delivered by subretinal injection. CNTF had already been shown in multiple animal models of inherited retinal degeneration to slow photoreceptor degeneration, and the results of phase I clinical trials in patients with advanced RP had already been reported.^83,84

In both of the dog models, in animals treated at ages where gene therapy alone was not successful, the combined therapy resulted in restoration of cone function.⁷⁸ Interestingly, the same study showed that intravitreal injection of CNTF alone resulted in small but detectable cone ERG responses, although these were recordable only for a short period of time. Intravitreal CNTF results in a temporary loss of photoreceptor outer segments (this has been termed “deconstruction”),^78,85 and it appears that as the outer segments reform they act in a similar way to the outer segments in young animals. They have a transient period of small but detectable cone function (as seen in young untreated achromatopsia dogs) and with gene therapy can create functional cone CNG channels (again a feature of gene therapy in the younger dogs).⁷⁹ The success of gene therapy in the dog achromatopsia model as well as the potential age limitations (a similar age limitation on rescue was also observed in the mouse⁸³) provide useful information for the different groups currently gearing up to start human clinical trials. Currently, there are plans for clinical trials in the United States (Hauswirth, R24 EY022023, AGTC NCT01846052) and the United Kingdom (Ali, Medical Research Council).

Although not relevant for a review of canine retinal gene therapy trials, it is worth noting that gene therapy trials in sheep with spontaneous achromatopsia due to a mutation in the other cyclic-nucleotide channel subunit, Cnga3, have shown success.⁸⁷ Clinical trials for CNGA3 achromatopsia are being planned in Israel⁸⁷ and Germany.⁸⁸

Gene Therapy with a Dog Macular Dystrophy Model

Best vitelliform macular dystrophy is a form of macular dystrophy that manifests as cystic like lesions in the macula consisting of large deposits of material in the subretinal space. With foveal involvement this can result in a serious loss of central vision. Mutations in the bestrophin 1 gene (BEST 1, also known as vitelliform macular dystrophy 2 [VMD2]) are responsible. The conditions resulting are often referred to as bestrophinopathies and there is some considerable phenotypic variation between patients. BEST 1 is expressed in the RPE and acts as an anion channel and also regulates intracellular calcium signaling. Mutations in canine Best1 result in a phenotype in dogs manifested as the development of multiple retinal elevations or blebs and is known as canine multifocal retinopathy (cmr). Three separate Best 1 mutations have been identified in dog breeds and the resulting conditions named cmr1, cmr2, and cmr3.^89,90 These are predicted to have differing molecular disease-causing mechanisms and may model some of the range of phenotypes seen in affected human patients.

AAV vectors with transgene expression driven by the human VMD2 promoter have been developed and tested in wild-type dogs using either a reporter transgene (GFP) or the canine Best 1 cDNA. An AAV2/2 vector was found to be effective in targeting RPE.⁹¹ A construct using AAV2/1 was also tested, but when it was used to express the Best 1 transgene, it was found to result in expression in cones that led to their degeneration. The AAV2/2–VMD2 construct delivering either canine Best 1 or human BEST 1 cDNAs has been used to treat cmr-affected dogs. The therapy was reported to reverse lesions and to have no adverse effects.^92,93 This potentially offers future translational opportunities to start clinical trials for the treatment of human bestrophinopathies.

Conclusions

Preclinical studies in dogs have been critical in bringing gene therapy to treat inherited retinal disease to the clinic. The advantages of the dog model include the similarity of eye size to that of human, enabling identical approaches to administer gene therapy, and the presence of a region of higher photoreceptor density, the area centralis. The spontaneous dog models often show a very similar disease phenotype to human patients, making them very useful for investigating disease mechanism and testing therapy. Although the laboratory mouse may be the workhorse for initial development of most treatments, this is not always the case (as with RPE65 and CNGB3 gene therapy trials) and the dog model can be used to make up for some of the limitations of the mouse models and allow refining of treatments prior to translation to clinical trials. Species differences do exist, meaning that extrapolation of successes between mouse, dog, and human should not be assumed. It seems likely that, with reports indicating that over 100 different breeds of dog are affected with retinal dystrophies, genetic screening within these breeds will further expand the existing “kennel” of canine retinal degeneration models that can be utilized for preclinical therapy development and refinement.

Acknowledgments

S.M.P.-J. is the Myers-Dunlap Endowed Professor of Canine Health. A.M.K. acknowledges NIH/NEI Grants R01-06855, R01-EY19304, and R42-EY23123, Foundation Fighting Blindness, and Michigan State University Startup Funds.

Author Disclosure Statement

No competing financial interests exist.