Novel organelles in primate retinal epithelium.

We are investigating age-related changes in organelles in monkey retinal epithelium using transmission and analytic electron microscopy. We previously described a circular organelle in retinal epithelium with a diameter of about 0.5µm. The organelle is unique in containing a single, round vacuole within an otherwise electron dense interior. We suggested that the organelle might be a melanosome with lysosomal properties. We now find that there are two similar organelles with such a single vacuole but which differ in their chemical composition, electron density, cell location and according to age. Epon embedded sections from the macular epithelium of seven monkeys, ranging from 1 to 35 years of age, were examined by transmission electron microscopy. A seven year old monkey was processed for analytic electron microscopy to determine the chemical composition of the organelles. The number and location of the organelles in the retinal epithelium were determined. The chemical composition of these two organelles was different. One of the organelles contained high mole fractions of oxygen and nitrogen and little phosphorous characteristic of melanin; the other had little oxygen and nitrogen and higher mole fractions of phosphorous uncharacteristic of melanin, but more common with lysosomal organelles. The latter had an electron dense rim around the vacuole, a less electron dense interior than the melanin containing organelle and also contained iron. The melanin containing organelle was more common in young monkeys and in the middle third of the cell. The organelle without melanin was more common in old monkeys and localized in the basal third of the cell. Two similarly vacuolated organelles, not identified before in retinal epithelium, differ in their chemical composition. One contains melanin; the other does not. The former is more common in young and the latter more common in old monkeys. This suggests reorganization and or degradation of melanin-containing organelles with age. These changes show how analytic electron microscopy can distinguish major ultra-structural differences in organelles when mere observation fails to do so easily.

Bestrophin detected in the basal membrane of the retinal epithelium and drusen of monkeys with drusenoid maculopathy.

PURPOSE: To determine if bestrophin is present in the basal membrane of macular retinal pigment epithelium (RPE) and in drusen of rhesus monkeys with age-related drusenoid maculopathy. METHODS: The macular region of three rhesus monkeys (Macaca mulatta), 23-24 years of age, with drusenoid maculopathy was dissected from eyes fixed with 4% paraformaldehyde. The macula was sectioned into rectangular pieces. The sclera was removed from each segment and the remainder separated into segments of neural retina with retinal epithelium or choroid with retinal epithelium. These segments were incubated with a goat polyclonal antibody to human bestrophin 1, reacted with gold-labeled rabbit antibody to goat IgG, silver-enhanced, and processed for transmission electron microscopy. RESULTS: Bestrophin-labeled gold particles were found in quasi-linear arrays on the basal surface of the macular RPE and also within drusen where bestrophin was found in segments of membranous-like material. The array density of the bestrophin-linked gold particles on the basal membrane of the epithelium had a maximal value of about 5-100 bestrophin molecules/micron(2). Immuno-detection of bestrophin was most effective when examined in an RPE layer that remained attached to the neural retina, where the basal surface of the epithelium is more directly exposed to the antibodies. CONCLUSION: Bestrophin is present on the basal membrane of macular RPE of rhesus monkeys with age-related drusenoid maculopathy, and also found in the membranous-like structures of drusen. The latter finding provides insight into the pathogenesis of drusen by indicating that segments of the basal membrane of RPE contribute to the material that accumulates within drusen.

Correction of the disease phenotype in the mouse model of Stargardt disease by lentiviral gene therapy.

Autosomal recessive Stargardt disease (STGD1) is a macular dystrophy caused by mutations in the ABCA4 (ABCR) gene. The disease phenotype that is most recognized in STGD1 patients, and also in the Abca4-/- mouse (a disease model), is lipofuscin accumulation in retinal pigment epithelium. Here, we tested whether delivery of the normal (wt) human ABCA4 gene to the subretinal space of the Abca4 -/- mice via lentiviral vectors would correct the disease phenotype; that is, reduce accumulation of the lipofuscin pigment A2E. Equine infectious anemia virus (EIAV)-derived lentiviral vectors were constructed expressing either the human ABCA4 gene or the LacZ reporter gene under the control of the constitutive (CMV) or photoreceptor-specific (Rho) promoters. Abca4-/- mice were injected subretinally with 1 microl ( approximately 5.0 x 10(5) TU) of each EIAV vector in one eye at postnatal days 4 and 5. An injection of saline, an EIAV-null vector, or an uninjected contralateral eye served as a control. Mice were killed at various times after injection to determine photoreceptor (PR) transduction efficiency and A2E concentrations. EIAV-LacZ vectors transduced from 5 to 20% of the PRs in the injected area in mice. Most importantly, a single subretinal injection of EIAV-CMV-ABCA4 to Abca4-/- mouse eyes substantially reduced disease-associated A2E accumulation compared to untreated and mock-treated control eyes. Treated eyes of Abca4-/- mice accumulated 8-12 pmol per eye (s.d.=2.7) of A2E 1 year after treatment, amounts comparable to wt controls, whereas mock-treated or untreated eyes had 3-5 times more A2E (27-39 pmol per eye, s.d.=1.5; P=0.001-0.005). Although extrapolation to humans requires caution, the high transduction efficiency of both rod and cone photoreceptors and the statistically significant reduction of A2E accumulation in the mouse model of STGD1 suggest that lentiviral gene therapy is a potentially efficient tool for treating ABCA4-associated diseases.

Behavior of retinal epithelium to bleb detachment versus retinectomy.

PURPOSE: To compare the behavior of the retinal pigment epithelium to a bleb detachment versus removal of a segment of neural retina. METHODS: A bleb detachment was performed on 14 adult pigmented rabbits. In seven rabbits, the neural retina was removed within the bleb detachment. The rabbits were followed for months after surgery by scanning laser ophthalmoscopy and post mortem histology using light and electron microscopy. RESULTS: Bleb detachment produces a transformation of the retinal epithelial layer that results in migration, enlargement of cell size and accumulation of large amounts of lysosomal material, resembling that found in Chediak-Higashi syndrome. Retinectomy causes migration of the epithelial cells, development of multiple layers with vacuoles but no accumulation of lysosomal material. CONCLUSION: The neural retina appears to exert a strong influence on the behavior of the retinal epithelial layer. Brief separation of the neural retina from the epithelium provokes a rapid transformation of this cell layer leading to migration of the cells and apparent faulty digestion of phagosomes, causing an enormous buildup of lysosomal debris. Removal of the neural retina also provokes retinal epithelial cell migration but no buildup of lysosomal debris occurs, presumably due to the absence of photoreceptor outer segments and consequently phagosomes. This migratory tendency, which appears to alter lysosomal degradation, could lead to apoptosis of these epithelial cells.

Genotyping microarray (gene chip) for the ABCR (ABCA4) gene.

Genetic variation in the ABCR (ABCA4) gene has been associated with five distinct retinal phenotypes, including Stargardt disease/fundus flavimaculatus (STGD/FFM), cone-rod dystrophy (CRD), and age-related macular degeneration (AMD). Comparative genetic analyses of ABCR variation and diagnostics have been complicated by substantial allelic heterogeneity and by differences in screening methods. To overcome these limitations, we designed a genotyping microarray (gene chip) for ABCR that includes all approximately 400 disease-associated and other variants currently described, enabling simultaneous detection of all known ABCR variants. The ABCR genotyping microarray (the ABCR400 chip) was constructed by the arrayed primer extension (APEX) technology. Each sequence change in ABCR was included on the chip by synthesis and application of sequence-specific oligonucleotides. We validated the chip by screening 136 confirmed STGD patients and 96 healthy controls, each of whom we had analyzed previously by single strand conformation polymorphism (SSCP) technology and/or heteroduplex analysis. The microarray was >98% effective in determining the existing genetic variation and was comparable to direct sequencing in that it yielded many sequence changes undetected by SSCP. In STGD patient cohorts, the efficiency of the array to detect disease-associated alleles was between 54% and 78%, depending on the ethnic composition and degree of clinical and molecular characterization of a cohort. In addition, chip analysis suggested a high carrier frequency (up to 1:10) of ABCR variants in the general population. The ABCR genotyping microarray is a robust, cost-effective, and comprehensive screening tool for variation in one gene in which mutations are responsible for a substantial fraction of retinal disease. The ABCR chip is a prototype for the next generation of screening and diagnostic tools in ophthalmic genetics, bridging clinical and scientific research.

Lentiviral transduction of green fluorescent protein in retinal epithelium: evidence of rejection.

This paper demonstrates lentiviral transduction of the humanized form of the Aequoria victoria gene for green fluorescent protein (GFP) into human fetal retinal pigment epithelium (RPE) in vitro and rabbit RPE in vivo. In vitro GFP expression of cultured human fetal RPE begins within two to three days after 12-16 h of maintained exposure to the virus at titers of 10(8)-10(9) infectious units (IU)/ml. Both stationary and dividing cells are transduced using a lenti viral vector with a cytomegalovirus (CMV) promoter. Expression remains stable for at least three to four months without evidence of toxicity and continues through cell division. In vivo expression is followed non-invasively in rabbit eye using a scanning laser ophthalmoscope (SLO), which can detect single fluorescing retinal cells. In vivo expression begins within a few days after a viral solution is introduced into the subretinal space. A solution of 10(9) IU/ml produces fluorescence within three to four days. Less concentrated solutions lead to slower and less expression. No expression is detectable at concentrations of 10(6) IU/ml. Within one to two weeks after introduction of the viral solution, there is evidence of rejection seen by SLO as a loss of GFP fluorescence and disruption of the RPE. Histology shows damage to the RPE layer and monocytic cell infiltrates in the choroid and subretinal space within the area receiving the viral solution. Strong GFP expression leads to rejection within two weeks. With less expression, rejection is delayed and in some cases undetectable for at least six months. If the GFP gene is not included in the viral vector or if the viral concentration is insufficient to produce detectable GFP expression, rejection is not seen. Using a rhodopsin promoter or injecting the virus intra rather than subretinally produces weak expression and no rejection. Lentivirus can induce expression of a foreign gene in the RPE. Viral induced transduction and GFP expression have no effect on the viability of the RPE in vitro. Continued expression of GFP after cell division implies chromosomal integration of the gene. In vivo expression of GFP in RPE encounters rejection. Rejection may not occur with low GFP expression. The latter occurs with low viral titers, a rhodopsin promoter or intra-retinal injection of viral solution. The results are relevant to gene therapy in retina when gene transduction leads to the expression of foreign proteins.

Identifying UV-cone responses in the murine superior colliculus.

A microelectrode was used to detect local visually evoked potentials from clusters of neurons in the murine superior colliculus. Chromatic stimuli and selective chromatic adaptation were used to identify responses of UV and middle-wavelength-sensitive (M) cones and/or rods. Three types of evoked potentials were found: those driven by UV and M cones and/or rods and those driven only by UV cone or only by M cones and/or rods. UV cone responses were more frequent in the medial and those from M cones more common in the lateral part of the superior colliculus. All three responses were found in the same area. UV cones provide a significant input to the murine superior colliculus. The spatial distribution of these responses in the superior colliculus reflects the organization of UV cones in the retina. Although synergistic inputs from UV- and M-cone and/or rod inputs appear to mix in local evoked responses in the superior colliculus, some areas are found to transmit only UV- or only M-cone and/or rod responses, indicating that there cannot be a widespread mixing of UV- and M-cone opsins in all murine cones.

Ultraviolet and middle wavelength sensitive cone responses in the electroretinogram (ERG) of normal and Rpe65 -/- mice.

Ultra-violet (UV) and middle wavelength sensitive (M) cone responses were identified in the ERG of normal and Rpe65 -/- mice using chromatic flashes and selective chromatic adaptation. In normal mice, the UV-cone response was as large as, or larger, in the presence of a bright yellow adapting light than it is in the presence of a dim white light. The M-cone response became undetectable in the presence of the yellow adapting light. Yellow adapting light initially reduced the UV response, but it recovered in 8-10 min. The M-cone response did not recover. UV-cone responses were undetectable in Rpe65 -/- mice. The M-cone response of young Rpe65 -/- mice was almost as large as in normal mice. A yellow adapting light only diminished this M-cone response. With age, the M-cone response further decreased in Rpe -/- mice. We show a pronounced loss of UV-cone function in Rpe65 -/- mice, which may be related to a defect UV-cones share with rods. The M-cone function is also affected already in young Rpe65 -/- mice. The transient effect of a yellow adapting light on the UV-cone response of normal mice is suggested to be neural, because it disappears during maintained light adaptation.

Isolation of human fetal cones.

PURPOSE: To describe a method for isolating a monolayer of human fetal cone photoreceptors and to compare their structure and ultrastructure before and after preparation. METHODS: Eyes from human fetuses (fetal week 20 to 24) were dissected and the neural retina of the developing fovea identified, cut out and placed on 10% gelatin. A VISX Star excimer laser was used to remove the inner retinal layers. The isolated cone monolayers were cultured for 18 hours and compared with untreated retinas by light microscopy and transmission electron microscopy. RESULTS: Excimer laser ablation removed the inner nuclear and ganglion cell layers leaving a monolayer of fetal cones. These cones survive in culture for at least 18 hours. The laser ablation disorganized the ultrastructure of the synaptic pedicles of these cones, left their plasma membranes intact. CONCLUSIONS: The developing central retina of human fetal eyes provides a source of fetal cones, which can be isolated from inner retinal cells using the excimer laser. Such a monolayer of human fetal cones may be useful for transplantation or biochemical studies.

In vivo studies of the gamma subunit of retinal cGMP-phophodiesterase with a substitution of tyrosine-84.

The inhibitory rod cGMP phosphodiesterase gamma subunit (PDEgamma) is a major component of the photoresponse and is required to support rod integrity. Pdeg(tm1)/Pdeg(tm1) mice (which lack PDEgamma owing to a targeted disruption of the Pdeg gene) suffer from a very rapid and severe photoreceptor degeneration. The Y84G (Tyr(84)-->Gly) allele of PDEgamma has previously been shown in experiments carried out in vitro to reduce the regulatory control of the PDE catalytic core (PDEalphabeta) exerted by the wild-type gamma subunit. To determine the effects of this mutation on in vivo function, the murine opsin promoter was used to direct expression to the photoreceptors of +/Pdeg(tm1) mice of a mutant Y84G and a wild-type PDEgamma control transgene. The transgenic mice were crossed with Pdeg(tm1)/Pdeg(tm1) mice to generate animals able to synthesize only the transgenic PDEgamma. Our results showed that wild-type PDEgamma and Y84G transgenes could complement the Pdeg(tm1)/Pdeg(tm1) mutant for photoreceptor survival. The mutation caused a significant biochemical defect in PDE activation by transducin. However, the Y84G mutation did not fully eliminate the control of PDEgamma on the PDE catalytic core in vivo; the expression of the mutant subunit was associated with only a 10-fold reduction in the amplitude of the a-wave and a 1.5-fold decrease in the b-wave of the corneal electroretinogram. Unexpectedly, the mutation caused a much 'milder' phenotype in vivo than was predicted from the biochemical assays in vitro.

A new component in the a-wave of the human cone electroretinogram.

The human cone electroretinogram (ERG) to a full field flash has been examined on a rod saturating background (17,000 photopic trolands). With strong stimuli, a negative wavelet appears in the falling phase of the a-wave. This response has a latency of 10-12 milliseconds, about 6-8 milliseconds after the start of the a-wave and just before the rising phase of the corneal positive b-wave begins. We suggest that it may represent a hyperpolarizing response of second order retinal neurons.

Local immunosuppression prolongs survival of RPE xenografts labeled by retroviral gene transfer.

PURPOSE: To determine whether local immunosuppression with Cyclosporin A can influence the survival of human fetal retinal pigment epithelium (RPE) xenografts in the rabbit's subretinal space. METHODS: Cultured human fetal RPE cells were transduced with the gene for green fluorescent protein (GFP) using a lentiviral vector. The RPE was transplanted into the subretinal space of rabbits that received intravitreal cyclosporine either by weekly injections (0. 25-0.5 mg) or by slow release (approximately 2 microg/d) from a capsule sutured into the vitreal cavity after prior cryopexy. The transplanted RPE was followed by GFP fluorescence scanning laser ophthalmoscopy and by histology of the transplant site. RESULTS: RPE xenografts in eyes receiving intravitreal cyclosporine survived longer (several months) than they did in control eyes without cyclosporine. Survival was as long with slow release capsules as it was with weekly intravitreal injections at much higher concentrations of cyclosporine. CONCLUSIONS: Local immunosuppression of the eye with cyclosporine prolongs the survival of RPE xenografts in the subretinal space of rabbits, implying that rejection involves activated T lymphocytes. Local immunosuppression with slow release capsules is as effective as weekly injections at much higher concentrations.

Cone inputs to murine retinal ganglion cells.

Responses of single retinal ganglion cells in different areas of mouse retina were studied to determine their cone inputs, using spectral sensitivity functions and chromatic adaptation. Spectral sensitivity curves were based on threshold response criteria to full field stimulation. The retina of the mouse was viewed through a dilated pupil with a surgical microscope. Ganglion cells were classified into three groups: one receiving inputs from short wave sensitive cones, a second receiving inputs from only middle wavelength sensitive cones and a third receiving inputs from both of these types of cones. The ventral retina, contained a large fraction of the first group of ganglion cells. The dorsal retina and the border between these two areas contained relatively more of the latter two groups. A small fraction of cells were found which displayed antagonistic-like interactions between photoreceptor systems. The results demonstrate that single ganglion cells in mouse retina can select responses from only one of the two cone mechanisms present in this retina, even in areas containing both types of cones.

A point mutation (W70A) in the rod PDE-gamma gene desensitizing and delaying murine rod photoreceptors.

PURPOSE: To examine the corneal electroretinogram (ERG) of transgenic mice (W70A mice) carrying a point mutation (W70A) in the gene encoding for the gamma-subunit of rod cGMP phosphodiesterase (PDEgamma). METHODS: The ERG of W70A mice was compared with that of normal mice. Cone responses were separated from rod responses by light adaptation, whereas rod sensitivity was assessed by threshold stimulation with dim light. Spectral sensitivity curves of the ERG were obtained using a constant response criterion. RESULTS: The ERG of the W70A mouse has a desensitized, delayed rod b-wave at threshold, and a prolonged rod b-wave at higher flash intensities. The a-wave is absent even at maximal stimulation. The cone ERG of the W70A mouse is indistinguishable from that of normal mice. The spectral sensitivity of the W70A mouse is maximal in the UV spectrum, in contrast to the normal mouse, which is most sensitive in the green region of the spectrum. This supports the interpretation of the results as normal cone and abnormal rod function in the W70A mouse. CONCLUSIONS: The W70A mouse represents new model of stationary nyctalopia that can be recognized by its unusual ERG features.

Long-term outcome of RPE allografts in non-immunosuppressed patients with AMD.

PURPOSE: To determine the long-term outcome of human retinal pigment epithelium (RPE) transplants in patients with advanced age-related macular degeneration (AMD). METHODS: Using pars plana microsurgical techniques, RPE allografts were transplanted subretinally to four groups of AMD patients: five patients received organized patch transplants after removal of choroidal neovascular membranes, four got small patch transplants in dry AMD; suspensions of RPE cells were transplanted in five cases with dry AMD, and two patients with RPE tears. None received immunosuppression. Transplants were followed for 24-38 months by biomicroscopy, fundus photography, SLO microperimetry, and fluorescein angiography. Rejection was defined as loss of visual function over the transplant, development of an exudative response (subretinal fluid with or without neovascularization), fluorescein leakage, and disruption, depigmentation, or encapsulation of the transplant. RESULTS: Four of 16 transplants (25%) presented no clinical signs of rejection. Three of the four small patch transplants remained pigmented and essentially unchanged after 30-32 months. Clinical signs of graft rejection appeared within three months in all cases of neovascular AMD (disrupted blood-retinal barrier, BRB), but after 6-20 months in five of nine eyes with non-exudative AMD (intact BRB). CONCLUSIONS: Subretinal human RPE allografts present a high rejection rate (75%) without immunosuppression. However, small extrafoveal transplants remained unchanged in shape, size and color for more than two years in non-exudative AMD. A disrupted BRB is likely to enhance graft rejection, which occurs earlier in exudative than in non-exudative AMD.

Impaired retinal function and vitamin A availability in mice lacking retinol-binding protein.

Retinol-binding protein (RBP) is the sole specific transport protein for retinol (vitamin A) in the circulation, and its single known function is to deliver retinol to tissues. Within tissues, retinol is activated to retinoic acid, which binds to nuclear receptors to regulate transcription of >300 diverse target genes. In the eye, retinol is also activated to 11-cis-retinal, the visual chromophore. We generated RBP knockout mice (RBP(-/-)) by gene targeting. These mice have several phenotypes. Although viable and fertile, they have reduced blood retinol levels and markedly impaired retinal function during the first months of life. The impairment is not due to developmental retinal defect. Given a vitamin A-sufficient diet, the RBP(-/-) mice acquire normal vision by 5 months of age even though blood retinol levels remain low. Deprived of dietary vitamin A, vision remains abnormal and blood retinol declines to undetectable levels. Another striking phenotype of the mutant mice is their abnormal retinol metabolism. The RBP(-/-) mice can acquire hepatic retinol stores, but these cannot be mobilized. Thus, their vitamin A status is extremely tenuous and dependent on a regular vitamin A intake. Unlike wild-type mice, serum retinol levels in adult RBP(-/-) animals become undetectable after only a week on a vitamin A-deficient diet and their retinal function rapidly deteriorates. Thus RBP is needed for normal vision in young animals and for retinol mobilization in times of insufficient dietary intake, but is otherwise dispensable for the delivery of retinol to tissues.

Tracking RPE transplants labeled by retroviral gene transfer with green fluorescent protein.

PURPOSE: To determine whether human retinal pigment epithelium (RPE) can be modified by retroviral-mediated gene transfer and to monitor the human RPE cells in the subretinal space of living rabbits with scanning laser ophthalmoscopy (SLO). METHODS: Cultured human fetal retinal pigment epithelium (HFRPE) was exposed to green fluorescent protein (GFP)-transducing retroviral vectors, Moloney murine leukemia virus, and lentivirus. The cultured cells were followed by fluorescence microscopy. Suspensions of GFP-expressing HFRPE were transplanted into the subretinal space of pigmented rabbits, and the transplant sites were examined by SLO for fluorescence, including fluorescein and indocyanine green angiography. The rabbits were euthanatized at different times after transplantation, and the retinas were studied histologically. RESULTS: Retroviral gene transfer can introduce a foreign gene such as GFP into cultured HFRPE. Gene expression is maintained in cultured RPE for at least 3 months. The lentiviral vector transduced both nondividing and dividing cells; the Moloney vector only transduced the latter. GFP-expressing cells can be followed in the living retina. Their changes reflect the rejection response followed histologically. CONCLUSIONS: Cultured HFRPE could be transduced to express GFP for long periods of time by retroviral gene transfer. GFP allowed retinal transplants and gene expression to be monitored in vivo. These results provide a model for potential ex vivo gene therapy in the subretinal space.

Photoreceptor allografts in a feline model of retinal degeneration.

BACKGROUND: Photoreceptor transplants provide a potential means to restore function in a degenerate retina and/or rescue degenerating host photoreceptors by trophic influences. We have examined photoreceptor allografts in the Abyssinian cat model of hereditary photoreceptor degeneration to determine the viability and influence of such transplants on the host retina. METHODS: Small pieces of 3- to 5-day-old normal kitten retina containing undifferentiated photoreceptors were injected into the subretinal space of adult Abyssinian cats at an early stage of retinal degeneration using standard vitreo-retinal surgical techniques. The retinas were examined by ophthalmoscopy and fundus photography, then by light and electron microscopy at different times after surgery. RESULTS: Such allografts survive for at least 6 months after surgery. The photoreceptors develop outer segments, invariably in rosettes. The transplants gradually integrate with the host retina but detach the host photoreceptor layer from the retinal pigment epithelium (RPE), which tends to reduce the number of host photoreceptors over the transplant. There is no slowing of the photoreceptor degeneration in neighboring non-detached retina. Inflammation or rejection was not detected. CONCLUSION: Undifferentiated, neonatal photoreceptor allografts survive and develop outer segments in the subretinal space of the Abyssinian mutant feline retina. The allografts gradually integrate with the host neural retina without inducing rejection. In the vicinity of the transplant there is increased loss of host photoreceptors, considered to be due to their detachment from the RPE layer. There is no evidence of any rescue of host photoreceptors elsewhere in this mutant retina.