Gene replacement therapy for retinal CNG channelopathies
- First Online:
- Cite this article as:
- Schön, C., Biel, M. & Michalakis, S. Mol Genet Genomics (2013) 288: 459. doi:10.1007/s00438-013-0766-4
Visual phototransduction relies on the function of cyclic nucleotide-gated channels in the rod and cone photoreceptor outer segment plasma membranes. The role of these ion channels is to translate light-triggered changes in the second messenger cyclic guanosine 3′–5′-monophosphate levels into an electrical signal that is further processed within the retinal network and then sent to higher visual centers. Rod and cone photoreceptors express distinct CNG channels. The rod photoreceptor CNG channel is composed of one CNGB1 and three CNGA1 subunits, whereas the cone channel is formed by one CNGB3 and three CNGA3 subunits. Mutations in any of these channel subunits result in severe and currently untreatable retinal degenerative diseases like retinitis pigmentosa or achromatopsia. In this review, we provide an overview of the human diseases and relevant animal models of CNG channelopathies. Furthermore, we summarize recent results from preclinical gene therapy studies using adeno-associated viral vectors and discuss the efficacy and translational potential of these gene therapeutic approaches.
KeywordsAdeno-associated virusCNG channelCyclic nucleotide-gated channelChannelopathiesGene therapyRetina
The role of CNG channels in vision
One of the first steps in vision is the absorption of light and the transduction of herein encoded information into electrical signals, which can be decoded by higher visual centers. This process called phototransduction starts in the outer segments of photoreceptors. In vertebrates, two different photoreceptor systems provide the best-possible detection of light under different luminous intensities. The rod system is responsible for vision at lower light levels, whereas at higher light levels the cone system allows for effective phototransduction. Furthermore, the cone system also confers color vision with different variants of cones bearing different types of visual pigments with distinct spectral sensitivities. Most vertebrates possess two different cone opsin variants (short- and medium-wave sensitive), whereas humans and simian primates have three different types (short-, medium-, and long-wave sensitive) (Yokoyama 2000; Ahnelt and Kolb 2000).
In both rods and cones, signal transduction is conferred by the second messenger cyclic guanosine 3′–5′-monophosphate (cGMP) that controls the activity of a CNG channel present in the plasma membrane of the photoreceptor outer segments. In the dark, the CNG channel is maintained in the open state by a high concentration of cGMP produced by retinal guanylyl cyclases. The resulting influx of Na+ and Ca2+ (“dark current”) depolarizes the photoreceptor and promotes synaptic glutamate release. When visual pigments absorb light, a G protein (transducin)-mediated signaling cascade is activated, leading to activation of retinal cGMP phosphodiesterase, hydrolysis of cGMP and closure of the CNG channel. As a result, the photoreceptor hyperpolarizes and shuts off synaptic glutamate release.
Overview on retinal CNG channelopathies in humans and animal models
Few diseases affect human life and personal destinies more than the loss of the ability to see. Among the diseases that cause blindness, the hereditary forms are the most devastating. They affect both eyes, often lead to significant or total loss of vision and were considered to be untreatable.
Mutations in CNGA1 and CNGB1 cause retinitis pigmentosa
Retinitis pigmentosa (RP) comprises a genetically diverse group of progressive degenerative diseases affecting the photoreceptors of the retina (Hartong et al. 2006; Kennan et al. 2005). The most common symptoms of RP include night blindness, progressive concentric reduction of the visual field, and abnormal accumulation of pigmentation in the retina (Kalloniatis and Fletcher 2004). In most cases, RP finally leads to legal blindness. At the cellular level, the disease is characterized by a primary impairment or total loss of rod function and structure followed by a secondary degeneration of the cones. So far, RP has been mapped to ~50 genes (http://www.sph.uth.tmc.edu/Retnet) encoding proteins involved in the visual transduction pathway (e.g., rhodopsin) or required for the maintenance of rod architecture (e.g., peripherin). The two genes encoding the rod CNG channel are also among the RP genes. Mutations in CNGA1 account for about 1 % of cases of autosomal recessive RP (arRP) (Dryja et al. 1995; Hartong et al. 2006). Most of the identified CNGA1 mutations cause deletions of important functional domains or result in defective membrane trafficking (Biel and Michalakis 2007; Dryja et al. 1995; Kaupp and Seifert 2002; Mallouk et al. 2002). CNGA1 encodes the A subunit of the rod CNG channel that is essential for channel function (Kaupp and Seifert 2002). As a result, these patients lack any rod-mediated light responses. Furthermore, mutations in the CNGB1 gene were also identified in patients suffering from RP and account for approx. 4 % of arRP cases (Bareil et al. 2001; Hartong et al. 2006; Kondo et al. 2004; Simpson et al. 2011). CNGB1 encodes the B subunits of the rod channel, which is not essential for channel formation, but significantly contributes to native CNG channels properties and rod channel targeting to outer segments (Kaupp and Seifert 2002). In contrast to CNGA1, the known CNGB1 mutations cause only minor deletions or single amino acid substitution. The molecular mechanisms leading to the severe RP phenotype in these patients are not well understood. One of the mutations (c.3444+1G>A) leads to replacement of the last 170 amino acids by 68 unrelated amino acids (Becirovic et al. 2010; Kondo et al. 2004). The resulting mutant CNGB1 protein might be more susceptible to proteasomal degradation and defective in targeting. Another mutation (c.2978G>T) results in a G993V substitution within the cyclic nucleotide-binding domain (Bareil et al. 2001) that leads to functionally inactive rod channels (Michalakis et al. 2011b).
To date, mice lacking the CNGA1 subunit are not available. However, transgenic mice revealing an about 50 % reduction of CNGA1 transcript levels due to the overexpression of a CNGA1 antisense mRNA were generated (Leconte and Barnstable 2000). These mice show some histological features reminiscent of RP (e.g., reduced number of photoreceptors, apoptotic death of retinal cells). However, electroretinograms (ERGs) of these mice have not been reported so far. Thus, it remains an open issue to which extent the down-regulation of CNGA1 affects rod- and cone-mediated vision. Moreover, it cannot be excluded that the overexpression of the antisense mRNA exerts non-specific effects (e.g., toxic effects caused by suppression of other mRNAs) that may affect the phenotype.
The knockout of CNGB1 in mice results in a phenotype that recapitulates the principal pathology of RP patients. In particular, CNGB1 knockout (CNGB1−/−) mice lack rod photoreceptor function (Hüttl et al. 2005; Zhang et al. 2009). This functional defect is accompanied by a progressive degeneration of rods and, secondary to this, a degeneration of the cones. The degeneration process is rather slow and results in loss of 10–20 % of rods at 4 months, 30–50 % at 6 months and 80–90 % at 1 year of age. Finally, the loss of CNGB1 induces down-regulation of several proteins of the phototransduction cascade and degradation of the CNGA1 subunit. Recently, a one-base pair deletion in CNGB1 in Papillon dogs was identified that causes a frame shift and premature stop codon, resulting in a lack of CNGB1 expression, markedly reduced or absent rod function, and a progressive retinal degeneration (Petersen-Jones et al. 2013).
Mutations in CNGA3 and CNGB3 cause achromatopsia
The functional loss of either the CNGA3 (Biel et al. 1999; Kohl et al. 1998) or the CNGB3 (Ding et al. 2009; Kohl et al. 2000; Sundin et al. 2000) subunit causes achromatopsia (ACHM), also known as rod monochromatism or total color blindness. ACHM is a hereditary, autosomal recessive retinal disorder characterized by the lack of cone photoreceptor function. In contrast to color blindness, in which changes in expression of opsin genes merely affect spectral sensitivity but not the physiology of photoreceptors (Deeb 2005; Jagla et al. 2002), the complete unresponsiveness of cones in achromatopsia has grave consequences for vision, particularly with respect to the densely cone-packed human fovea. Affected individuals show a total loss of color discrimination, photophobia, nystagmus and very poor visual acuity (Eksandh et al. 2002; Pokorny et al. 1982). The prevalence of ACHM is approximately 1:30,000. Up to now there are four known ACHM genes: CNGA3, CNGB3, GNAT2 and PDE6C. The most common causes of ACHM in the general western population are mutations in the first two genes encoding the cone CNG channel subunits. Mutations in CNGB3 (ACHM1) are more common in the general western population and account for almost 50 % of achromatopsia cases (Kohl et al. 2005). A missense mutation in the CNGB3 gene (S435F) was identified in colorblind individuals originating from the Pingelap Atoll of Micronesia (Sundin et al. 2000). In this small island, achromatopsia is very frequent and affects nearly 10 % of the native population (Sacks 1997). Approximately, 28 % of patients in the western population carry mutations in CNGA3 (ACHM2). In Middle East and Arabic populations, mutations in CNGA3 account for up to 60 % of ACHM cases, whereas the frequency of mutations in GNAT2 (ACHM4) or PDE6C (ACHM5) is only 1–2 %, respectively.
Loss of function mutations in CNGA3 result in non-functional cone CNG channels because CNGB3 subunits cannot form functional CNG channels in the absence of CNGA3 (Biel and Michalakis 2007; Kaupp and Seifert 2002). Genetic inactivation of CNGA3 in mice leads to selective loss of cone-mediated light responses (Biel et al. 1999) accompanied by progressive degeneration and cell death of cones (Michalakis et al. 2005). An early hallmark of cone degeneration is the strong accumulation of the second messenger cGMP suggesting its involvement in the process of degeneration (Michalakis et al. 2010, 2013). Cone degeneration affects M- and S-cones differentially and cell death proceeds significantly faster in ventral and nasal (S-cone-rich) than in dorsal and temporal (M-cone-rich) parts of the retina. Ventral cones are almost completely missing after the third postnatal month whereas residual dorsal cones are present even in aged knockout mice (Michalakis et al. 2005). In addition, a naturally occurring mouse model of achromatopsia—the cpfl5 mouse with a CNGA3 point mutation—was described with a phenotype similar to the CNGA3 knockout (CNGA3−/−) mouse (Pang et al. 2012). Recently, a sheep model of achromatopsia was also identified with diminished cone, but normal rod function (Reicher et al. 2010; Shamir et al. 2010). Affected lambs were homozygous for a mutation in the CNGA3 gene that changes amino acid R236 to a stop codon (Reicher et al. 2010).
Small and large animal models also exist for CNGB3-associated achromatopsia. Knockout of CNGB3 in mice results in strongly reduced, but not absent cone function and progressive cone photoreceptor degeneration (Ding et al. 2009; Xu et al. 2011). The residual cone function in this model is conferred by irregular homomeric CNGA3 channels. In addition to the CNGB3 knockout mouse, two naturally occurring canine models exist (Alaskan malamute dogs) that carry recessive mutations in exon 6 or a genomic deletion of the entire CNGB3 gene (Sidjanin et al. 2002). The latter results in a disease sharing the same clinical phenotype as human patients resulting in day-blindness and absence of cone function (Aguirre and Rubin 1974, 1975; Rubin 1971a, b). The reason for this interspecies variability in the phenotype caused by lack of CNGB3 is not clear. One possibility could be an interspecies difference in the efficiency and/or rate of outer segment transport of irregular homomeric CNG channels.
ACHM has long been considered as a stationary disease due to lack of progression of functional defects (e.g., the cone ERG is either missing from birth or the functional loss does not progress with age). Data from animal model studies (Michalakis et al. 2005) and from high-resolution in vivo retinal imaging studies in patients (Genead et al. 2011; Thiadens et al. 2010) suggested that ACHM is often associated with progressive cone degeneration. However, the precise kinetics of the degeneration is still not well understood. Pathological changes may include mild to moderate alterations in cone inner/outer segment morphology as detected by high-resolution optical coherence tomography (OCT) and adaptive optics (AO) imaging (Genead et al. 2011; Thiadens et al. 2010) or in more severe cases substantial loss of the foveal/foveomacular cones, in some cases associated with hypoplasia of the retinal pigment epithelium (Genead et al. 2011; Thiadens et al. 2010). In most cases, such degenerative changes can only be observed in adult or aged patients, but were occasionally also seen in younger patients (Genead et al. 2011; Thiadens et al. 2009, 2010).
Adeno-associated viral (AAV) vectors as efficient tools for retinal gene delivery
In principal, inherited autosomal recessive disorders can be treated by viral vector-mediated replacement of the mutated, non-functional genes. Recombinant AAVs have proven to be the most suitable vectors for efficient and long-term expression of transgenes in retinal cells (Boye et al. 2013; Kumar-Singh 2008; Vandenberghe and Auricchio 2012). AAVs are small non-pathogen viruses that package 4.7 kb single-stranded DNA, but the viral genome can be replaced by expression cassettes with a maximum size of 5.2 kb (Dong et al. 2010; Lai et al. 2010; Wu et al. 2010). A typical AAV expression cassette consists of a cell-type-specific promoter (e.g., rhodopsin promoter for rod photoreceptors), the transgene of interest, a polyadenylation site (e.g., bovine growth hormone (BGH) polyA) and a posttranscriptional regulatory element (e.g., woodchuck posttranscriptional regulatory element, WPRE). Several AAV serotypes have been identified that differ in their tropism to different cell types. Thus, cell-type-specific expression of the transgene further depends on the capsid of the AAV vectors. For example, the most suitable AAV serotypes for gene expression in photoreceptors are AAV5, AAV7, AAV8 and AAV9 (Boye et al. 2013; Surace and Auricchio 2008; Vandenberghe and Auricchio 2012).
Preclinical gene replacement studies for CNG channelopathies
The availability of suitable animal models and the advent of efficient and safe retinal gene transfer vectors facilitated the preclinical development of novel gene therapies for inherited retinal CNG channelopathies.
Gene therapy approaches in models of RP
Gene therapy approaches in models of achromatopsia
Another study (Pang et al. 2012) confirmed these results showing that an AAV5 vector could restore cone-mediated function and arrest cone degeneration in a naturally occurring mouse model of achromatopsia with a CNGA3 mutation, the cpfl5 mouse (Pang et al. 2010). Besides a significant rescue of cone-mediated ERGs, normal visual acuities and contrast sensitivities could be restored. After treatment, both M- and S-opsins were expressed and their outer segment localization was maintained in treated retinas. The observed therapeutic effect lasted for at least 5 months post-injection (Pang et al. 2012).
AAV-mediated gene replacement was also successfully employed in animal models of CNGB3-related achromatopsia (ACHM1) (Carvalho et al. 2011; Komaromy et al. 2010). Komaromy et al. (2010) showed that subretinal delivery of AAV5 vectors expressing human CNGB3 under control of a 2.1 kb human red cone opsin promoter led to long-term restoration of cone function and day vision in two canine models of CNGB3-related achromatopsia (Komaromy et al. 2010). However, the authors also observed a reduction in cone therapy success rate in dogs treated at 54 weeks or older (Komaromy et al. 2010). The reason for this age-dependency could not be finally clarified—but recently, the same group found a strategy to circumvent this phenomenon (Komaromy et al. 2013). In this study, they combined the gene therapeutic approach with the administration of ciliary neurotrophic factor (CNTF). CNTF is known to cause an effect called deconstruction of photoreceptors. This reversible effect leads to immature photoreceptors with shorter outer segments and a reduced gene expression. Subsequently, the photoreceptors return to a normal stage. The CNTF-mediated deconstruction and regeneration of the mutated cone outer segments led to a successful gene therapy in all dogs in the age of 14–42 months.
Another study tested the efficacy of a gene therapy in the CNGB3−/− mice using an AAV8 vector containing a human cone arrestin promoter driving the expression of human CNGB3 (Carvalho et al. 2011). After the subretinal delivery of this vector, CNGB3 could be detected in both M- and S-cones. This further led to increased levels of CNGA3, improved the density and survival of cones and had a beneficial effect on the cone outer segment structure and subcellular compartmentalization of cone opsins. Furthermore, the therapy resulted in a long-term improvement of retinal function in CNGB3−/− mice, which was shown by the restoration of cone ERG amplitudes of up to 90 % of wild-type level and a significant improvement in visual acuity. Although successful restoration of cone function was also observed when treatment was initiated at 6 months of age, restoration of normal visual acuity was only possible in 2–4 weeks old animals (Carvalho et al. 2011).
Concluding remarks and future directions
Treatments of retinal degenerative diseases by AAV-mediated gene replacement can be successful in humans. This was recently demonstrated for Leber congenital amaurosis (LCA), a disease that causes nearly total blindness in childhood. After the gene therapy was established in animal models of LCA (Acland et al. 2001; Ali et al. 2000), further promising human clinical trials followed (Bainbridge et al. 2008; Cideciyan et al. 2008; Hauswirth et al. 2008; Maguire et al. 2008). Treated LCA-patients showed substantial visual improvement in the short term and no decline from this new level for at least 3 years, whereas photoreceptor degeneration progresses unabated (Cideciyan et al. 2013; Jacobson et al. 2012). In the case of LCA, these results show the need for combinatorial therapy to improve vision but also slow retinal degeneration. These results on the restoration of vision in a blinding eye disease caused by mutations in a retinal pigment epithelial cell-specific gene are very promising for the gene therapy of retinal degenerations caused by photoreceptor-specific mutations.
For future translational gene replacement studies, challenges clearly exist. Critical issues concern the efficiency of gene delivery tools and the time window during which successful gene replacement therapy is possible. A successful gene replacement approach would, therefore, rely on the following conditions: (1) the availability of a gene transfer vector allowing efficient, cell-specific (in cones or rods) and long-term gene expression without major adverse effects, (2) the presence of a sufficient number of vital/treatable target cells (in this case cone or rod photoreceptors), (3) the plasticity of the retina/visual system to adapt on and properly process the newly (or re-) gained light-detecting ability.
Overview of retinal CNG channelopathies, animal models and preclinical gene therapy studies
Affected gene (cell type)
Associated human disease
Gene replacement studies
Retinitis pigmentosa, RP49
CNGA1 antisense expressing mice: retinal degeneration (Leconte and Barnstable 2000)
Retinitis pigmentosa, RP45
CNGB1-deficient mice: impaired rod function and retinal degeneration (Hüttl et al. 2005)
Koch et al. (2012)
Michalakis et al. (2010)
cpfl5 mouse: loss of cone function and cone cell degeneration (Pang et al. 2010)
Pang et al. (2012)
Canine models (null-deletion or missense mutation): cone degeneration (Sidjanin et al. 2002)
Carvalho et al. (2011)
Encouraged by the positive therapeutic outcome in the animal models, several groups have launched translational projects aiming to bring gene replacement therapies for inherited CNG channelopathies into clinical practice.
This work was supported by the Deutsche Forschungsgemeinschaft (DFG) and the Tistou and Charlotte Kerstan Foundation (RDCURE Project).