Artur.V. Cideciyan, Scheie Eye Institute, University of Pennsylvania, 51 N. 39th
Street, Philadelphia, PA 19104, USA.
Tel: +1 215 662 9986;
fax: +1 215 662 9388;
Severe blinding retinal degenerative diseases have been without treatments that could improve vision until recently. Gene therapy has been in clinical trials for certain inherited retinopathies in which photoreceptors are retained despite severe visual loss. Optogenetics is being discussed for retinal diseases in which there is severe visual loss and nearly complete photoreceptor cell death. As a retinal therapy, optogenetics would be the genetic targeting of light-sensing molecules to residual cells in a degenerate retina. Parallel with scientific advances in optogenetics should be the development of detailed criteria for patient candidacy. Here, molecularly defined retinal degenerations are used to exemplify how some diseases or stages of disease would satisfy the criteria. Measurements are made of the thickness of ganglion cell and the nerve fiber layers of the retina. Whereas the clinical category of retinitis pigmentosa has been most often mentioned for treatment by optogenetics, an argument is made for expanding the target diseases to some early-onset disorders diagnosed as Leber congenital amaurosis.
Optogenetics is the name given to a field that emerged from knowledge gained over more than a century about light-induced currents . The concept of expressing light-activated proteins in specific neurons has wide value to neuroscience research [2, 3] and has also been considered for biomedical use as a potential treatment modality in retinal blindness .
The introduction of light-gated ion channels into residual inner retinal neurons of a patient with photoreceptor loss to restore vision is an attractive concept. There are many hurdles to surmount before optogenetics can qualify as a valuable therapeutic option to such patients . A time can be imagined, however, when intravitreal delivery of an optogenetic gene therapy in the blind eye of a patient is among the treatment options for currently incurable retinal diseases . Yet, who is an appropriate patient for a future optogenetic procedure that targets, for example, the retinal ganglion cells? This work begins to define the retinal criteria for candidacy of optogenetic therapy. Results of measurements of retinal laminae in select patient groups with different molecular forms of retinal degeneration are used to illustrate patterns of disease or stages that would or would not be appropriate for this form of therapy. Conclusions from these results can be extrapolated to other molecular forms of retinal degeneration or ungenotyped disease.
Materials and methods
Patients with retinitis pigmentosa (RP) or Leber congenital amaurosis (LCA) and normal subjects were included. Informed consent or assent was obtained; procedures followed the Declaration of Helsinki and had institutional review board approval. All patients underwent eye examination as well as retinal imaging.
In vivo microscopy of human retina
Cross-sectional retinal images were obtained with optical coherence tomography (OCT) using methods previously described [7-14]. Further details are in Appendix S1, Supporting information.
Measurements to assess patient candidacy for optogenetic therapy
In an era of evolving therapies for inherited retinal degenerations with very different approaches , it is important to establish criteria that would justify an optogenetic strategy. Visual function must be severely diminished and measurements of retinal structure must indicate a severe reduction of photoreceptor cells as the likely basis of the visual deficit. Further, there must be evidence that there are post-receptoral retinal layers such as the ganglion cell (GC) layer to support the introduction of light-activated channels, and sufficient nerve fibers leading from GCs to the optic nerve head (ONH). Retinal structural criteria for an optogenetic strategy can be met, in part, by measurements of scans from OCT. To complete the investigations, there should also be evidence that post-ocular visual pathways are at least structurally if not functionally intact. The former can be assayed with magnetic resonance imaging (MRI) and the latter with functional MRI using diffuse and bright stimuli, assuming some residual vision [15-17].
In vivo histology of the retina by OCT, laminar thicknesses measured from such scans, and illustrations of normal retinal topography of the laminar architecture are shown (Fig. 1). An OCT cross-section through the fovea across the central retina (most readily accessed with current instruments) has hyporeflective cellular layers separated by hyperreflective synaptic laminae (Fig. 1A, upper left). Photoreceptor and GC nuclear layers are highlighted [outer nuclear layer (ONL) blue; GCL, orange]. ONL and GCL thickness measurements for a group of normal subjects are shown (Fig. 1A, lower left panels). Eccentric displacement of GCs  was marked by lines connecting corresponding eccentricities on ONL and GCL thickness profiles.
A topographical map across an expanse of central retina in a normal subject illustrates the distribution of ONL thickness (based on OCT measurements). ONL peaks centrally and declines with distance from the fovea; parafoveal thinning is more gradual in the superior retina (Fig. 1A, upper right). A topographic map of GC density in the normal central retina is also depicted (Fig. 1A, lower right) and is based on histology (adapted from ). Around the foveal depression is a ring with peak GC density and this diminishes with increasing eccentricity from the fovea.
To the vitreal side of the GCL is the nerve fiber layer (NFL) which is comprised of axons leading from the GC to the ON. We quantified NFL in three ways (Fig. 1B). First, in the same cross-sections along the horizontal meridian through the fovea used for ONL and GCL measurements, we determined NFL thickness (Fig. 1B, upper left). A schematic of the course of the NFs in normal retina is also shown (Fig. 1B, upper right). NFs arch over the foveal region toward the ONH and there is no crossing of the midline in the temporal retina; in the nasal retina, fibers have a straighter course to the ONH. The papillomacular bundle of NFs extends from the foveal region to the ONH. Second, considering evidence in non-human primate retina that adeno-associated virus type 2 (AAV2) transduces mainly a ring of parafoveal GC , we performed NFL topography in the region that would represent the potential therapeutic zone for GC. A rectangular region extending from the temporal parafovea through the fovea and approaching the ONH was quantified (Fig. 1B, lower left). Finally, the conventional circular OCT scan around the ONH captures NFL thickness from all regions surrounding it and is depicted as a polar plot (Fig. 1B, lower right). Histological NFL measurements at loci surrounding the ONH in post-mortem analyses of normal human eyes fall within the normal range by OCT measurement .
To exemplify types of retinal diseases and disease stages that warrant consideration for optogenetic therapy, data are shown from autosomal recessive (ar) RP and LCA.
RP caused by DHDDS mutations
Within the disease category of RP, all inheritance patterns are represented and there are many molecular genetic causes [21, 22]. Patients at advanced stages of RP can have very limited vision mainly due to photoreceptor loss. Mutations in DHDDS (dehydrodolichyl diphosphate synthase) are a recently recognized cause of arRP [23, 24]. Whereas patients at earlier stages can show wide expanses of visual field with measurable rod and cone function, other patients at later stages have limited vision and only a small central island of residual function .
A 19-year-old man with the homozygous DHDDS mutation, c.124A>G (p.Lys42Glu, K42E) and 20/32 visual acuity (VA), has ONL that extends from the fovea to about 3 mm eccentric (Fig. 2A). ONL is normal (or slightly increased) at the fovea but becomes reduced with eccentricity. A 24-year-old man with RP caused by the same DHDDS mutations has 20/20 VA and foveal ONL like that in the first patient. In both patients, other cellular layers appear qualitatively normal including preserved central inner and outer segment (IS/OS) laminae. In contrast, a 34-year-old woman with this mutation has 20/40 VA and reduced ONL at the fovea and barely detectable ONL with greater eccentricity; GCL and NFL remain. The 37- and 30-year-old siblings of the previous patient have little or no detectable ONL across the central retina but still retain GCL and NFL (Fig. 2A); their acuities are 20/320 and HM (hand motions), respectively. Measurements of ONL show a spectrum of thickness across the central retina from normal (or increases) in the foveal region but reductions with eccentricity, to abnormally reduced ONL across the scan, to barely measurable ONL (Fig. 2B). GCL thickness is either within normal limits or increased (Fig. 2B). NFL is normal or increased in thickness (Fig. 2C). An NFL map from the 30 year old shows normal or increased thickness in all 14 sampled locations. Mapping in the 37 year old (not shown) had the same results. NFL thickness by circular scanning at the ONH for the five DHDDS-RP patients had results that were either within normal limits or increased (Fig. 2C).
Candidacy for GC optogenetic intervention in this subgroup of DHDDS patients would be limited to the 30-year-old patient with almost no detectable ONL, little remnant vision but preserved GCL and NFL. The disease stages represented by the other DHDDS-RP patients would be more suitable for therapies that could theoretically increase or at least preserve function of residual photoreceptors.
RP caused by CRB1 mutations
RP12, on chromosome 1q , was found to be caused by mutations in CRB1, the human homolog of a Drosophila gene . Subsequently, CRB1 mutations were detected in LCA  and now there is a spectrum of CRB1 severities . Early in the understanding of the human disease expression, we noted abnormal retinal structure in CRB1-associated retinal degenerations . In vivo retinal imaging of patients with CRB1 mutations showed profound disorganization of the laminar architecture , which later was also noted, but often to a lesser extreme, in many retinal degenerations [9, 16, 29-31].
The spectrum of retinal structural severity in four patients with RP and CRB1 mutations is shown (Fig. 3). None of these patients had nystagmus from infancy; all had VA of 20/40 or better in at least one eye when examined in their second decade of life; and there were symptoms of nightblindness and peripheral vision disturbance by the second decade. A 23-year-old woman (CRB1 mutant alleles: C948Y and C1218F) with 20/20 VA shows an island of preserved ONL with IS/OS laminae in the central 2 mm of retina and this is surrounded by disorganized-appearing retina with intraretinal hyperreflectivities. A 49-year-old woman (CRB1 mutant alleles: R764C and G477R) with 20/50 VA also retains central retinal ONL that is more extensive nasally than temporally from the fovea. The IS/OS laminae are indistinct in the fovea and there are increasing abnormalities with eccentricity. A 30-year-old man (CRB1 mutant alleles: C948Y and Q335P) has 20/125 VA but only barely distinguishable ONL centrally. The 50-year-old sibling of the patient with R764C and G477R mutant alleles has HM vision, no evident outer retinal laminae, intraretinal hyperreflectivities throughout and a large hyperreflectivity at ∼3.5 mm in the temporal retina (presumed migrated pigment clump) that blocks signal deep to it. GCL and NFL are present in all four patients.
Quantitation of outer and inner retinal laminae was possible in some of these patients. Only one patient showed normal ONL and this was at and near the fovea. Abnormally reduced ONL was present in two others (Fig. 3B). GCL thickness was detectable in all four patients and fell within normal limits or showed greater thickness than normal (Fig. 3B). NFL thickness was normal or increased in the horizontal profile (Fig. 3C) and in all sampled locations on maps from both the 23-year-old and 50-year-old patients. NFL thickness around the ONH was also normal or increased (Fig. 3C). Only the 50 year old would be a potential candidate for optogenetic therapy.
LCA caused by CRB1 mutations
CRB1-LCA patients have nystagmus and severe visual disturbances from infancy. Despite early-onset central visual disturbances in this group, peripheral islands of function can remain . Central retinal structure in four representative CRB1-LCA patients is shown (Fig. 4A). The abnormalities in these patients resemble those in the 50-year-old CRB1-RP patient (Fig. 3A, bottom), suggesting a disease severity continuum. Two male siblings (homozygotes for CRB1 mutant allele, C948Y) at ages 19 and 21 have HM (HM also in other eye) and 20/100 (and 2/200) VA in the eyes shown. ONL is not discernible in the central retina of either patient. There is hyporeflectivity suggestive of INL and possibly a discontinuous hyperreflectivity representing OPL (outer plexiform layer) at eccentricities from ∼0.5 to 3 mm to either side of the foveal depression. Deep to the OPL are intraretinal hyperreflectivities, some of which appear continuous with the retinal pigment epithelium (RPE). Such an appearance has been ascribed to cell migration into the inner retina . A 30-year-old man (CRB1 homozygote for the P748 del3cCAT allele) with 20/200 in the eye shown (and HM in the other) has a pattern of laminopathy similar to that of the two younger patients. A 37-year-old woman (CRB1 mutant alleles, C85 ins2actTG and T745M) with 20/1000 VA has less total retinal thickness (albeit still thickened relative to normal) than the other patients but there are also features in common such as the laminopathy and prominent intraretinal hyperreflectivities.
ONL was not detectable in any of these patients, but GCL and NFL were measurable. GCL was markedly increased in thickness but less so around the fovea (Fig. 4B). NFL thickness was increased toward the nasal retina; an NFL map from the 19 year old showed increased thickness at half of the locations within the area of interest; the other loci were within normal limits (Fig. 4C). By polar plot, NFL is also thickened (Fig. 4C;  ). All four patients would be candidates for an optogenetic therapeutic strategy based on GCL and NFL measurements, but some would not qualify due to retained vision. The 30 year old has been studied with functional MRI and there were detectable but reduced cortical responses (P12; 17). Among the patients shown, two (ages 19 and 37) would be potential candidates for optogenetic therapy based on structural and functional criteria.
LCA caused by CEP290 mutations
LCA caused by CEP290 gene mutations has been investigated for retinal structure and, in most patients studied, there is severe visual loss but a retained central retinal zone of photoreceptors [16, 32]. Evidence from autofluorescence and OCT indicates a disease course with early loss of most rods but remaining central cones with abnormal IS/OS lamination . Structural MRI proved that the visual pathway was intact despite early-onset visual losses . Given an appropriate vector, a gene augmentation strategy would seem logical as a first approach to treatment of these retained central photoreceptors. A 10-year-old girl with CEP290 mutations (IVS26+1655A>G/L517X) and light perception vision shows central retinal ONL but reduction in ONL with increasing eccentricity. GCL is present and NFL is detectable (see nasal retinal edge of the scan, Fig. 5A, top). Similar results are in a 19-year-old woman with IVS26+1655A>G and G1890X mutant CEP290 alleles, except there is detectable IS/OS lamination in the fovea and VA of 20/50 (Fig. 5A, second from top). A 26-year-old woman, with I1059fs and L2448fs mutant CEP290 alleles, has 20/800 VA and less defined IS/OS lamination (Fig. 5A, third from top). In contrast to the other patients, a 32-year-old man with HM vision (IVS26+1655A>G and K1575X mutant alleles) has reduced ONL in the central retina. GCL and NFL are retained (Fig. 5A, bottom). Quantitation of ONL in five CEP290-LCA patients shows that four of five have a preserved central island while one patient has abnormally reduced ONL; all have normal (or increased) GCL across the profile (Fig. 5B). Maculopathy has been previously observed in CEP290-LCA . NFL in all patients is at or above normal thickness limits across the horizontal meridian. NFL maps from the 19- and 32-year-old patients indicated that NFL thickness at sampled locations was within or above normal limits (Fig. 5C). NFL thickness around the ONH is also normal or thickened (Fig. 5C). If there was continued progression along a macular degenerative path with future loss of remaining ONL in the 32 year old with HM vision, this patient would be a potential candidate for optogenetic therapy.
Current realities of optogenetics as a potential therapy for retinal degenerations
Scientific progress in the many fields under the umbrella of optogenetics has led to suggestions that there can be clinical relevance to those with blindness [4, 5, 34]. The time seems right for progress as previously untreatable and severely disabling retinal degenerations have shown some response to treatment strategies. There has been some success at improving vision by gene augmentation therapy in one form of genetic retinal degeneration, specifically LCA due to mutations in RPE65 (reviewed in  ), and recent FDA marketing approval of an electronic prosthetic device may add some visual percepts to patients previously with bare light perception . There is no substitute for more basic research in optogenetics. Yet, if there was an attempt to use current data to initiate a human clinical trial of safety using an optogenetic strategy, what facts do we know that would direct a therapeutic plan?
The cellular target for such an early optogenetics trial would have to be the retinal GC considering the relatively large body of literature indicating success in wild-type rodents and rodent models of retinal degeneration using a number of agents (reviewed in ). As these cells are closest to the vitreo-retinal interface, these experiments were accomplished with intravitreal injection of AAV vectors, a major advantage in a human clinical setting. The caveat is that no matter how wide the retinal extent of GC transduction is in rodents, there is evidence that AAV2 transduces only a ring of parafoveal GC in non-human primate retina . Considering the abnormalities of retinal laminar architecture in the extracentral retina in advanced forms of retinal degeneration, exclusive access to parafoveal GCs may be an advantage. GCL integrity has been demonstrated in post-mortem human donor retinas from patients with advanced retinal degenerations [38-40]. The current study showed how the GCL and NFL can be quantified in vivo to establish candidacy for a clinical trial. The unexplained GCL thickening in some patient subgroups was mainly eccentric to the parafoveal region.
A second-choice target suggested for optogenetics is residual cone cells without OS, specifically using a hyperpolarizing channel. The advantage offered is the potential ease of restoring intraretinal signal processing. Only limited data have been published to indicate that there is a patient population worthy of this attractive but relatively unexplored avenue . The idea of combining such an optogenetic strategy with a neuroprotective therapy has been put forward , considering the delivery (subretinal in original rodent experiments) may be to cone cells with an unknown lifespan and sufficient fragility that they may not survive the treatment.
RP is the disease category usually named for future optogenetic trials to restore vision but certain forms of LCA should also be considered
Advances in the understanding of all hereditary retinal degenerations have increased in recent years [21, 22], and characterization of the many molecular forms of LCA has been especially rapid . Our recent investigations of many forms of molecularly clarified LCA indicate that some clearly meet an optogenetics criterion for severity of visual dysfunction. Data presented here in patients with CRB1-LCA, for example, show that this is a candidate disease. A more complete discussion of other forms of LCA and their potential for optogenetics is in Appendix S1.
Criteria established can serve for other therapies being designed for severe photoreceptor disease
Criteria for retained inner retinal structure will need to be supplemented with criteria defining inner retinal function [42, 43]. Combined structural and functional criteria can be applied to evaluate candidacy not only for optogenetic therapy but also for other therapies that are emerging for severe retinal degenerations. For example, a work-up that establishes photoreceptor loss but retained and functional GCs and NFs would also serve for a central retinal stem cell transplant  or an electronic prosthesis . The initial approach in the implanting of electronic prostheses has been to place the device in the central retina. Given some success with optogenetics to the central retina, the prosthetic could be implanted in extracentral retina. Such a combinatorial approach could theoretically afford patients with central and extracentral sources of light sensation.
This work was supported by Hope for Vision, the NU Fund for Retinal Research and the Grousbeck Foundation.