In vivo protection of photoreceptors from light damage by pigment epithelium-derived factor.

PURPOSE: To determine whether pigment epithelium-derived factor (PEDF) exhibits neurotrophic and neuroprotective activities in vivo for photoreceptor cells. METHODS: Sprague-Dawley albino rats were injected intravitreally with 2 microg PEDF or a mixture of 1 microg basic fibroblast growth factor (bFGF)/1 microg PEDF in a volume of 1 microl phosphate-buffered saline (PBS). Animals were exposed to constant light for different periods at an illuminance level of 1200 to 1500 lux. The electroretinogram (ERG) waveforms of both eyes in the same animal were simultaneously recorded to evaluate functional protection. The morphologic protection was evaluated by quantitative histology. RESULTS: Intravitreal injection of PEDF before exposure to constant light resulted in significant morphologic and functional protection of photoreceptor cells in the retina of light-damaged rats. This protection depended on the duration and severity of light damage. The protection was eliminated by extending the light exposure to 10 days. Injection of PEDF at 0, 1, and 2 days after constant light exposure did not provide significant protection above that seen in PBS-injected eyes. Combination of PEDF with bFGF improved functional protection of photoreceptor cells. CONCLUSIONS: The data demonstrate that PEDF protected photoreceptor cells against light damage. This is significant, because it may open new avenues for the study of molecular mechanisms underlying degenerative processes. This, in turn, may lead to the development of therapeutic strategies for the prevention and treatment of degenerative diseases of the retina.

In vitro inhibition of antirecoverin immunoglobulin-mediated death of mammalian photoreceptor cells.

Cancer-associated retinopathy (CAR) is a blinding disease, which can be mediated by autoimmune reactions with a specific calcium-binding retinal protein, recoverin. A number of recent studies demonstrate that agents that mobilize intracellular calcium can protect neurons from apoptotic death induced by a variety of insults. In this study, we investigated the effect of one such agent, potassium, on the survival of mammalian rod photoreceptors exposed to antirecoverin IgG. Primary cell cultures of rat retinal neurons were grown in a chemically defined medium, and cells were exposed to antirecoverin IgG for 72 hr in various concentrations of potassium and the surviving cells counted. Rod photoreceptors were quantitated using antirhodopsin immunofluorescence microscopy, and total cell numbers were determined by 4',6-diamidino-2-phenylindole (DAPI) staining of nuclei. Apoptosis was evaluated by TdT-mediated biotin-dUTP nick-end labeling (TUNEL), cell death-detection ELISA, and DNA laddering. The present study shows that elevated extracellular K+ ([K+](o)) protects retinal neurons from antirecoverin antibody-mediated cell death. The protective effects of ([K+](o)) were shown to be time- and dose-dependent. The inhibition of antirecoverin IgG-mediated death of photoreceptors by elevated ([K+](o)) suggests that the mobilization of internal calcium stores rescues the cells by interfering with apoptotic signal transduction pathways. These data also suggest that the death of photoreceptor cells occurring in CAR possibly can be prevented by reagents and/or environmental changes that mobilize intracellular calcium.

Recoverin negative photoreceptor cells.

Recoverin, a calcium-binding protein, is unique with respect to its cellular regulation. It is present in retinal rods, cones, cone bipolar cells, and in a rare population of cells in the ganglion cell layer. Inappropriate turn-on or turn-off of recoverin expression has been reported both in small cell lung carcinoma cells from patients with cancer-associated retinopathy (Matsusara et al. [1996] Br. J. Cancer 74:1419-1422; Adamus et al. [1998] J. Autoimmun. 11: 523-533; Ohguro et al. [1999] Invest. Ophthalmol. Vis. Sci. 40:82-89) and in cultured retinal neurons (McGinnis et al. [1999] J. Neursci. Res. 55:252-260). In a recent report using double labeling immunofluorescence microscopy methods with antibodies against either rhodopsin and recoverin or arrestin and recoverin, two unique cell phenotypes, rhodopsin-positive and recoverin-negative, and arrestin-positive, and recoverin-negative were observed in vitro. These two unique cell types could be nonphotoreceptor cells in which rhodopsin and arrestin are inappropriately turned on or they are photoreceptor cells in which the recoverin gene is inappropriately turned off. In this study, multiple antibodies were used to study, on a single-cell basis, whether the photoreceptor cell-specific marker, rhodopsin, is inappropriately expressed in nonphotoreceptor cells in our retinal neuronal culture system. We also examined the hypothesis that the two unique cell phenotypes represent the same population of cells. A triple labeling method has been established to visualize recoverin, rhodopsin, and arrestin protein expression simultaneously in cultured retinal neurons. Our data clearly and directly demonstrate that the previously described unique cell phenotypes are the same population of cells, rod photoreceptors. The existence of recoverin-negative photoreceptors demonstrates that the recoverin gene can be regulated independently of other photoreceptor cell-specific proteins and suggests that this primary cell culture may be useful as a model system for investigating the illicit expression of the recoverin gene in cancer associated retinopathy.

Pigment epithelium-derived factor protects cultured retinal neurons against hydrogen peroxide-induced cell death.

Pigment epithelium-derived factor (PEDF) is a neurotrophic protein synthesized and secreted by retinal pigment epithelial (RPE) cells in early embryogenesis and has been shown to be present in the extracellular matrix between the RPE cells and the neural retina. It induces neuronal differentiation and promotes survival of neurons of the central nervous system from degeneration caused by serum withdrawal or glutamate cytotoxicity. Because the role of PEDF in the retina is still unknown, we examined its ability to protect cultured retinal neurons against hydrogen peroxide (H(2)O(2))-induced cell death. Retinas of 0-2-day-old Sprague-Dawley rats were isolated and dissociated, and the neurons were maintained for 2 weeks in a synthetic serum-free medium. Immunocytochemical labeling showed that 50-60% of the cultured cells were rod photoreceptors. Treatment with H(2)O(2) induced significant death of retinal neurons in a dose- and time-dependent manner. Pretreatment with PEDF prior to insult greatly attenuated H(2)O(2)-induced cytotoxicity, and its effect was shown to be dose dependent. Cytotoxicity was determined by 3,(4,5-dimethylthiazol-2-yl)2, 5-diphenyl-tetrazolium bromide and lactate dehydrogenase assays, and apoptotic cell death was evaluated by the TdT-mediated digoxigenin-dUTP nick-end labeling assay. The present study also showed that H(2)O(2)-induced retinal neuron death was by apoptosis that could be inhibited by PEDF. Combination of PEDF with basic fibroblast growth factor, brain-derived neurotrophic factor, or ciliary neurotrophic factor improves the protection. These data strongly suggest that PEDF is a potential neuroprotective agent in the treatment of retinal degeneration.

Anti-recoverin antibodies cause the apoptotic death of mammalian photoreceptor cells in vitro.

We developed an in vitro model to study the effect of anti-recoverin antibodies on retinal cells and the mechanism(s) by which they kill photoreceptors in cancer-associated retinopathy (CAR). Rat retinal cells were grown in a defined medium, and cell types were identified by using antibodies against rhodopsin, recoverin, syntaxin, and thy-1. Purified immunoglobulin (IgG) against recoverin was added to the cultures at different concentrations for 24, 48, or 72 hr, and the survival of the cells was determined by fluorescence microscopy. Preimmune IgG and normal medium were used as controls. The cell death detection enzyme-linked immunosorbent assay and the terminal deoxyuridine triphosphate nick-end labeling assay were used to demonstrate cells undergoing apoptosis. Double labeling was used to visualize cell types and apoptotic death. Rods, amacrine cells, and ganglion cells were identified in the cultures. Rod cells, but not ganglion cells and amacrine cells, markedly decreased in the presence of 200 microg/ml of anti-recoverin IgG for 24, 48, and 72 hr. Anti-recoverin antibodies caused apoptosis in rod cells but not in amacrine cells. Almost all cells were shown to take up IgG from the medium. In conclusion, our retinal cell cultures provide a system for investigating antibody-mediated photoreceptor cell death and demonstrate that anti-recoverin antibodies cause the apoptotic death of rod cells, with no effect on amacrine cells. The results suggest that anti-recoverin antibodies play a key role in the apoptotic death of photoreceptors in CAR.

Unique retina cell phenotypes revealed by immunological analysis of recoverin expression in rat retina cells.

Among retina-specific proteins, recoverin is unique with respect to its cellular regulation in that it is found in rods, cones, some bipolar cells, and a rare population of cells in the ganglion cell layer. Recoverin is a calcium-binding protein which inhibits rhodopsin kinase from phosphorylating rhodopsin. Because cells in the inner layers of the retina do not contain rhodopsin kinase, arrestin, or other phototransduction proteins, it seems likely that recoverin has a different function in those cell types. To study this protein more fully, antibodies were generated against the entire mouse recoverin protein, as well as against peptides from the amino and from the carboxyl termini. These antibodies confirmed the localization of recoverin in vivo and clearly demonstrated, in culture, cells which were recoverin positive and rhodopsin negative. Surprisingly, two unique cell phenotypes were seen in cell culture which are not found in vivo. These cells are [rhodopsin(+), recoverin(-)] and [arrestin(+), recoverin(-)]. These phenotypes appear to represent the same population of cells and suggest that the recoverin gene can be regulated independent of genes for other phototransduction proteins. This cell culture system will be useful for investigating environments and factors which participate in the expression of the recoverin gene, and may identify regulatory features of the recoverin gene which cause it to be illicitly expressed in small-cell lung carcinomas in cancer-associated retinopathy (CAR).

Functional significance of recoverin localization in multiple retina cell types.

Knowledge of the cellular localization of recoverin in photoreceptor cells has enabled its interaction with other proteins to be postulated, tested and verified. Recoverin, a calcium sensing protein, is now thought to act by prolongation of the light state through interference with the interaction of arrestin and rhodopsin. Because of the detection of recoverin in multiple cell populations, the specificity of the cellular localization of recoverin was investigated in the retina of the mouse, rat, rabbit, chicken, frog, and chameleon and compared to that for opsin, phosducin, and arrestin. In addition to photoreceptor cell staining, the application of affinity-purified antibodies against recoverin demonstrated immunoreactive cells in the inner nuclear layer and a rare immunopositive cell in the ganglion cell layer of the mouse, rat and rabbit retina. Only photoreceptor cells were stained with recoverin antibodies in the chameleon and frog retina, whereas no cells were recoverin-positive in the chicken retina. In all six species studied, only photoreceptor cells were labelled with antibodies against opsin, phosducin or arrestin. Based on intensity of staining, two distinct populations of anti-recoverin-immunoreactive cells were distinguished in the photoreceptor cell layer of the retinas of the rat and rabbit, with the more darkly stained cells (probably cones) representing about 3% of the photoreceptor cells. The presence of recoverin in cells other than photoreceptors suggests it has an alternative or additional function and indicates the presence of multiple cell type-specific expression signals in the regulatory region of the recoverin gene.

Chromosomal assignment of the human gene for the cancer-associated retinopathy protein (recoverin) to chromosome 17p13.1.

The gene for the mouse recoverin protein (23 kDa photoreceptor-specific protein, S-modulin, or the Cancer-Associated Retinopathy protein) was recently assigned to mouse chromosome 11, closely linked to trp53. In this paper, the human gene for recoverin was localized to human chromosome 17 by Southern analysis of restriction digests of the DNA from mouse/human somatic cell hybrids. Using a 7 kb subclone of the human recoverin gene, a positive fluorescence in situ hybridization signal was demonstrated near the terminus of the short arm of chromosome 17 at position p13.1. The mapping of recoverin to this region of human chromosome 17, which contains a number of cancer-related loci, suggests a possible mechanism by which cancer-associated retinopathy occurs in humans.

Light-dependent regulation of the transcriptional activity of the mammalian gene for arrestin.

The purpose of this investigation was to determine if the cyclic variations in the amount of arrestin mRNA that occur in the mouse retina during the light/dark cycle are controlled by light, a circadian clock, or both. In this study, an experimental paradigm was designed and used to distinguish among these molecular mechanisms. The data show that the initiation and the maintenance of an increased rate of syntheses of arrestin mRNA requires light and is not dependent on the time of day. Therefore the changes in transcriptional activity of the mouse arrestin gene are controlled by light, independent of circadian clock. The mechanism for this light-dependent regulation of gene transcription in the retina is distinct from the light-dependent increases in mRNA concentrations of early response genes which occur in the suprachiasmatic nucleus and a comparative analysis of such changes in these different regions of the CNS may lead to the identification of the molecular relationship between light- and circadian-dependent gene regulation.

Developmental appearance, species and tissue specificity of mouse 23-kDa, a retinal calcium-binding protein (recoverin).

A 23-kDa, soluble, calcium-binding photoreceptor-specific protein (23-kDa) has been shown to be identical to recoverin and the cancer-associated retinopathy protein. Recoverin has been reported to activate guanylate cyclase to increase the amount of cyclic GMP and thereby reopen cation channels within the photoreceptor cells. In this study, the 23-kDa protein was purified from bovine retinas and monospecific antibodies against it were generated in rabbits. Western analysis demonstrated 23-kDa in retinas from human, monkey, bovine, dog, rabbit, rat, mouse, frog, chameleon and iguana although it was not detected in chicken or fly retinas. No immunoreactivity was observed in any non-retinal tissues except the pineal gland. The 23-kDa protein was detected, by Western analysis, at postnatal day 5 in the mouse retina and it increased in amount in parallel with the differentiation of the photoreceptor cells in normal mice and it also decreased in parallel with their degeneration in the rd mouse. Immunocytochemical analysis of the adult mouse retina showed that 23-kDa is restricted primarily to the inner segments of the photoreceptor cells and, unlike arrestin, its localization did not shift in response to light/dark changes.

Chromosomal assignment of the recoverin gene and cancer-associated retinopathy.

The deduced amino acid sequence of the recently cloned mouse 23kD photoreceptor cell-specific protein showed it to be identical to the recoverin protein and the CAR (cancer-associated retinopathy) protein. DNA sequence variants were found in the mouse recoverin gene (Rcvrn), and segregation analysis of restriction fragment length variants in recombinant inbred strains of mice assigned Rcvrn to mouse Chromosome (Chr) 11, between Sparc (3.7 map units) and Zfp-3 (2.3 map units). These results demonstrate a close linkage of recoverin to the tumor suppressor gene, Trp53. On the basis of these data, knowledge of the function of recoverin, and the characteristics of CAR, an experimentally testable model is presented to explain the molecular basis for CAR.

Cloning and sequencing of the 23 kDa mouse photoreceptor cell-specific protein.

The 23 kDa protein was localized by immunocytochemistry to photoreceptor cells of the mouse retina, and bovine and mouse cDNA clones were isolated and sequenced. The deduced amino acid sequences showed that the mouse 23 kDa protein is 91% identical to the bovine protein, and is the same as S-modulin, the CAR (cancer-associated retinopathy) protein and recoverin, the Ca(2+)-dependent activator of photoreceptor guanylate cyclase. The amino acid sequence reveals two Ca2+ binding sites, no internal repeats, 59% homology to the chicken visinin protein and 40% homology to calmodulin while Northern analysis demonstrated a single 1.0 kb mRNA species in bovine and mouse retina.

Transient, cyclic changes in mouse visual cell gene products during the light-dark cycle.

Temporal and spatial changes in the cellular and subcellular concentrations of photoreceptor cell gene products appear to be important features of phototransduction in rod photoreceptor cells. The time course of the rapid, light-dependent movement of S-antigen (S-Ag) (48k; Arrestin) from the inner segments to the outer segments was analyzed using polyclonal and monoclonal antibodies. The concentrations of mRNA change about threefold for rhodopsin and about sixfold for S-Ag in cyclic, independent modes during the normal light-dark cycle. Kinetic analysis indicates that the oscillations of S-Ag mRNA levels are due to changes in the transcriptional activity of the gene itself. An experimental model is presented summarizing the relationships between mRNA levels, protein localization, disc shedding, and phototransduction in the photoreceptors cells.

Mouse opsin. Gene structure and molecular basis of multiple transcripts.

The single copy mouse opsin gene produces five major transcripts, varying in size from 1.7 to 5.1 kilobases. The mRNAs are present at levels that vary over 2 orders of magnitude and can be detected as early as postnatal day 1. Each of the transcripts is polyadenylated and can be identified in polysome-bound RNA, suggesting that each is translated in vivo. To elucidate the molecular basis of this complex transcription pattern, we have characterized genomic fragments covering the entire mouse opsin gene, including several kilobases of 5'- and 3'-flanking regions. Transcription initiates at a single site 97 base pairs upstream of the translation start codon. Northern hybridization with exon- and intron-specific probes demonstrated that the various transcripts are not generated by partial or alternative splicing. Sequence analysis of the 3' end of the gene showed the presence of multiple polyadenylation signals. Analysis by polymerase chain reaction of the 3' end of opsin cDNA demonstrated that the complex transcription pattern originated from the selective use of these polyadenylation sites, generating transcripts that differ only in the length of the 3'-untranslated region. Transcript heterogeneity similar to that observed in mouse was also found in rat and, to a lesser degree, in human and frog opsin mRNAs.

Amino acid and cDNA sequence of bovine phosducin, a soluble phosphoprotein from photoreceptor cells.

Vertebrate photoreceptor cells contain a soluble phosphoprotein, phosducin, which complexes with the beta, gamma subunits of the GTP-binding protein, transducin. Light-induced changes in cyclic nucleotide levels modulate the phosphorylation of phosducin by protein kinase A. The complete amino acid sequence of purified phosducin from bovine retinas was determined by Edman degradation from overlapping polypeptides derived from enzymatic digestion by trypsin and Staphylococcus aureus V8 protease or from chemical degradation by cyanogen bromide. Excluding the unidentified group which blocks the NH2 terminus, phosducin contains 245 amino acids with a calculated molecular weight of 28,185 and isoelectric point of pH 4.5. Phosducin is enriched with acidic and sulfur-containing amino acids, having 32 glutamic acid, 16 aspartic acid, 9 methionine, and 5 cysteine residues. It also contains 24 serine and 8 threonine residues, of which only serine 73 is located within a consensus phosphorylation sequence (-RKMS(P)QV-) for cyclic nucleotide-dependent protein kinase. Secondary structure analysis predicts the presence of 62% alpha-helix, 22% beta-sheet, and 16% random coil, with eight turns. Computer-aided searches of protein data banks revealed no apparent homology to any sequenced protein except that coded by a MEKA cDNA clone (Kuo, C-H., Akiyama, M., and Miki, N. (1989) Mol. Brain Res. 6, 1-10) which deviates from the confirmed phosducin sequence in the last 15 amino acids. Sequence analysis of a cDNA clone for bovine retinal phosducin confirmed that the MEKA clone deviation resulted from an unidentified cDNA guanosine nucleotide, a shifted reading frame and a premature stop codon.

Localization of the rhodopsin gene to the distal half of mouse chromosome 6.

We have assigned the mouse rhodopsin gene, Rho, to chromosome 6 using DNA from a set of mouse-hamster somatic hybrid cell lines and a partial cDNA clone for mouse opsin. This assignment rules out the direct involvement of the rhodopsin gene in the known mouse mutations that produce retinal degeneration, including retinal degeneration slow (rds, chromosome 17), retinal degeneration (rd, chromosome 5), Purkinje cell degeneration (pcd, chromosome 13), and nervous (nr, chromosome 8). Segregation of Rho-specific DNA fragment differences among 50 animals from an interspecific backcross (C57BL/6J X Mus spretus) X C57BL/6J indicates that the Rho locus is 4.0 +/- 2.8 map units distal to the locus for the proto-oncogene Raf-1 and 18.0 +/- 5.4 map units proximal to the locus for the proto-oncogene Kras-2. Linkage to Raf-1 was confirmed using four sets of recombinant inbred strains. The two loci RAF1 and RHO are also syntenic on human chromosome 3, but on opposite arms.

Structural analysis of mouse S-antigen.

Mouse S-antigen clones were isolated from a mouse retinal cDNA library using a bovine S-antigen cDNA probe. The largest clone (MSC-242) comprised 1532 bp and contained the entire coding sequence. The nucleotide sequence homology between the mouse and bovine coding regions was 84%, while non-coding regions appeared to be more divergent. The deduced amino acid sequence indicated that the mouse S-antigen had 403 residues and its molecular ratio was 44,930. An overall amino acid sequence similarity of 84% was observed between the mouse and bovine proteins. This degree of similarity dropped to 60% and 47% at the N and the C termini, respectively. The local homology with alpha-transducin observed in the bovine proteins, including the putative phosphoryl and rhodopsin binding sites, was conserved in the mouse as well. There was no overall sequence similarity with other proteins listed in the National Biomedical Research Foundation (NBRF) protein sequence database. Among the uveitopathogenic sites for experimental autoimmune uveitis (EAU), peptides N and M were identical to their bovine counterparts. Peptides 3 and K, however, were more divergent. The short repeats within these peptides were conserved.

Isolation and analysis of the mouse opsin gene.

We have identified three overlapping 5'-truncated mouse opsin cDNA clones by immunologically screening a lambda gt11 retina expression library. Using one of the cDNA clones as a probe, we isolated a 5 kb genomic fragment that encompassed the complete coding sequence for mouse opsin. The coding region for opsin was interrupted by four introns positioned precisely as those previously described for other mammalian opsins. In contrast to the single major opsin mRNA in the bovine and human retina, Northern analysis of mouse retina RNA demonstrated the presence of at least five distinct species of polyadenylated opsin mRNAs. Their sizes ranged from 1.7 kb to 5.1 kb.

The photoreceptor-specific 33 kDa phosphoprotein of mammalian retina: generation of monospecific antibodies and localization by immunocytochemistry.

The distribution in mouse retina of a 33,000 Da phosphoprotein (33 kDa) that complexes with the beta/gamma subunits of transducin (T beta gamma) and undergoes light-induced dephosphorylation was determined by immunocytochemistry. An antiserum containing antibodies for the 33 kDa protein and beta-transducin of mouse and bovine retinas was generated against the purified 33 kDa-T beta gamma complex from bovine retina. The antiserum reacts with beta-transducin derived from either 33 kDa-T beta gamma complex or transducin complex (T alpha beta gamma), but not with the alpha- or gamma-transducin. It also reacts with both the phosphorylated and unphosphorylated form of the 33 kDa-T beta gamma complex. Antibodies, monospecific for the 33 kDa and beta-transducin subunits respectively, were purified from the antiserum by immunoadsorption and used in immunocytochemical analysis of the respective antigens. The 33 kDa protein was found to be associated exclusively with the photoreceptor cells of the retinas, with the most intense staining in the inner and outer segments' layers and lighter staining in the synaptic terminal layers. beta-Transducin also is found in the photoreceptors, but some T beta immunoreactivity exists within the inner plexiform layer. The specific localization of the 33 kDa protein together with its light-modulated phosphorylation suggest that the 33 kDa-T beta gamma complex is involved in light-regulated activities of the rod photoreceptor cells.

Light-dependent subcellular movement of photoreceptor proteins.

The intracellular localization of photoreceptor-specific proteins 33 kd, beta-transducin, and 48 kd, as determined by immunocytochemistry, is transient and dependent on the lighting environment to which the retina is exposed. Western analysis of the proteins in isolated rod outer segments from mouse retina demonstrates that beta-transducin actually moves from the outer segment to the inner segment in response to light and that 48 kd moves simultaneously in the opposite direction. The light-induced movements appear to be initiated by the absorption of light by rhodopsin because red light, which does not bleach rhodopsin, does not produce this redistribution of photoreceptor proteins. Time course analysis of these movements suggests that the light-induced shift is detectable at the earliest time examined (30 seconds). The bidirectional movement suggests that the photoreceptor cells have at least two distinct dynein-like or kinesin-like translocator molecules that act as microtubule-based motors. This movement appears to be a basic mechanism by which photoreceptor cells rapidly and radically alter the subcellular concentrations of photoreceptor-specific proteins, which in turn may affect the rapid changes in membrane potential that occur during phototransduction.