Immortalized retinal neurons derived from SV40 T-antigen-induced tumors in transgenic mice.

Immortalized retinal neurons have been established in tissue culture from retinal tumors arising in transgenic mice. The mice carry the SV40 T-antigen under the control of 5' flanking sequences from the human phenylethanolamine N-methyltransferase (PNMT) gene in order to target oncogene expression to adrenergic cell types. The retinal cultures contain a proliferation population of T-antigen-positive cells with a neuronal morphology that includes formation of extensive neuritic processes. We identified the cells as amacrine-derived neurons by immunofluorescence using the cell-specific monoclonal antibodies VC1.1 and HPC-1. The cells also express all three neurofilament subunits and GAP-43. These results indicate that CNS neurons can be transformed in transgenic animals to generate cultured cells with many properties of mature neurons.

Retinoid-binding proteins in retinoblastoma tumors.

A combination of Western blot, Northern blot, and radiolabeled ligand-binding techniques was used to investigate retinoid-binding proteins in retinoblastoma (RB) cells from fresh tumors and from 19 RB tumor lines cultured in vitro. Using rabbit anti-bovine cellular retinal-binding protein (CRA1BP) antibodies, no CRA1BP could be detected. As determined by [3H]retinol binding, cellular retinol-binding protein was sometimes not detectable but averaged 2.3 +/- 2.7 means +/- SD, n = 7) pmol [3H]retinol bound/mg protein, similar to adult retina cytosol. Using [3H]retinoic acid as ligand, cellular retinoic acid-binding protein was not detectable in some lines and averaged 1.0 +/- 1.2 (means +/- SD, n = 7) pmol [3H]retinoic acid bound/mg protein, well below the adult retina cytosol level of 94.4 +/- 16.3 (means +/- SD, n = 4) pmol [3H]retinoic acid bound/mg. Using rabbit antibovine interstitial retinol-binding protein (IRBP) antibodies, IRBP of the same molecular mass as human IRBP (135,000) was detected in the medium from all cultured RB cells and averaged 75.9 +/- 19.2 pmol/10(8) cells (bovine IRBP immunochemical equivalents). Cytosol levels were less than 1% of the medium. Using a human IRBP complementary DNA probe, levels of IRBP RNA transcripts in 19 RB cell lines were comparable to adult retina. The Y-79 RB cell line was different from the others; the amount of IRBP in the medium was only about 1% of the RB cell lines. Levels of cellular retinol-binding protein were comparable with the other lines, but cellular retinoic acid-binding protein was 9 times more abundant. IRBP RNA transcripts in Y-79 cells were below the limits of detectability but appeared at low levels after induction of differentiation of Y-79 by 10(-6) M retinoic acid. Although this cell line has been in culture longer than the others, it may also have been initiated at an earlier stage of retinal development.

Detection of cytoplasmic S-100 protein in primary and metastatic intraocular melanomas.

The authors sought S-100 protein in sections cut from formalin- or glutaraldehyde-fixed, paraffin-embedded tumors from the eye and other sites by an immunoperoxidase technique, using rabbit antibovine S-100 protein antiserum, swine antirabbit serum, a peroxidase-antiperoxidase preparation, and amino-ethylcarbizole as developer. Tissue from 26 intraocular tumors was examined, the pathologic diagnosis being unknown to the investigators at that time that the tests were performed. Thirteen of 16 malignant melanomas of the choroid contained S-100 protein (81%), while only one of seven retinoblastomas contained S-100 protein (14%). One cutaneous malignant melanoma metastatic to the eye contained S-100 protein, while none of three metastatic nonmelanocytic, nonneural tumors contained the protein. These results are similar to our findings with nonocular tumors, in which 56 of 56 cutaneous melanomas contained S-100 protein (100%) and only 1 of 51 nonneural, nonmelanocytic tumors contained S-100 protein (2%). S-100 protein may be useful as a marker for ocular malignant melanoma.

Antigenic modulation in retinoblastoma: a flow cytometric study.

Flow cytometry (FCM) was used to investigate antigenic expression and modulation during the cell cycle of Y-79 and WERI-Rb1 tissue cultured retinoblastoma cell lines using a polyclonal anti-Y-79 antibody and fluorescein conjugated lectins. Several Y-79 resting cell populations were identified by FCM analysis of antibody binding, while only a single population with uniform antigen expression was found to exist in the synthetic and mitotic phases. WERI-Rb1 cells bound antibody approximately equally in each phase of the cell cycle. Multiple cell populations with different lectin binding affinities were seen in the resting phase with FITC-concanavalin A, FITC-ricinus communis-60 and FITC-ricinus communis-120 (FITC-RCA-120). During the S-phase of the cell cycle, a higher percentage of cells bound FITC-RCA-120 and FITC wheat germ agglutinin. The relationship between antigenic expression during the cell cycle and treatment considerations in retinoblastoma is discussed.

Fc and C3b receptors on cultured retinoblastoma cells.

Fc and C3b receptors were identified on cultured retinoblastoma cells. Labeled receptor protein bound to affinity gels prepared with IgG, Aggregated IgG, and Fc but not to control gels prepared from Fab'2 or Sepharose-4B alone. Eluted Fc receptors was partially characterized by polyacrylamide gel electrophoresis. Molecular weight of the isolated receptor or its subunit was approximately 4.3 X 10(4) daltons. Cultured retinoblastoma cells were found to rosette with human indicator erythrocytes specific for C3b and Fc receptors. This study indicates that considerable reactivity between retinoblastoma patients' sera and cultured tumor cells may be mediated by these receptors as well as by reactivity toward retinoblastoma antigens.

Cellular retinol- and retinoic acid-binding proteins in transformed mammalian cells.

Extracts prepared from several lines of transformed cells were examined for the presence of cellular binding proteins specific for retinoids. Extracts of human retinoblastoma cell line WERI-Rb1 contained a cellular binding protein specific for retinoic acid, whereas extracts of human retinoblastoma cell line Y-79 contained cellular binding proteins for both retinol and retinoic acid. Upon purification, the latter two binding proteins proved to have properties similar to those of the corresponding proteins obtained from bovine retina. Smaller amounts of these binding proteins were detected in extracts of undifferentiated and differentiated neuroblastoma and McCoy cells. HeLa and rat glioma cells had no detectable amount of binding proteins. The 11-cis-retinal-binding protein, present in extracts of human, rat, and bovine retina, was not found in any of the cell lines examined.

Binding of retinoblastoma and normal sera to retinoblastoma-derived cultured cells.

We have observed increased binding of retinoblastoma patients' sera to a retinoblastoma-derived cultured cell line (Y-79). This reactivity was mediated by the serum IgG fraction and was directed toward different tumor or target cell (Y-79, Molt, Raji, and fibroblasts) cultured in media containing fetal calf serum. Normal pooled serum IgG fractions did not demonstrate any similar binding. When target cells were cultured in media containing human serum instead of fetal calf serum, a considerable reduction in retinoblastoma sera binding activity was observed. Reactivity against target retinoblastoma cells could be reduced but not entirely eliminated by quantitative absorption with nonretinoblastoma (Molt) cells grown in media with fetal calf serum. Retinoblastoma and normal sera binding to autologous fibroblasts, nonautologous fibroblasts, and cultured melanoma cells was also minimal. These findings suggest that residual binding activity in the sera tested may be directed against retinoblastoma tumor antigens. The fetal calf serum component responsible for reactivity with certain retinoblastoma sera was shown by immunoprecipitation, competitive inhibition, and gel electrophoresis to be bovine serum albumin.

Intermediate filaments in the human retina and retinoblastoma. An immunohistochemical study of vimentin, glial fibrillary acidic protein, and neurofilaments.

Fifty-five retinoblastoma specimens were studied by a sensitive immunoperoxidase method to determine the intermediate filament types present in human retina and retinoblastoma. Polyclonal antiserum against vimentin and monoclonal antibodies to glial fibrillary acidic protein (GFAP) and to the 200 kD neurofilament triplet protein were used. In the human retina, Müller's cells coexpressed vimentin and GFAP in most instances, probably as a reactive phenomenon. Surprisingly, the horizontal cells did not stain with any of the antibodies used, and may thus lack intermediate filaments. Also, the meshwork of neural fibers in the inner plexiform layer was unusually sparse. Retinoblastoma cells were found to be devoid of all intermediate filament types studied. The tumors contained, however, vimentin and GFAP in the stromal cells. All neurofilament-positive cells in retinoblastoma apparently derived from infiltrated retina. One retinoblastoma eye was also studied by indirect immunofluorescence microscopy of frozen sections with identical results.

Evaluation of tumor angiogenesis factor with the rabbit cornea model.

Sequential histopathological observations were made of the rabbit corneas after an implantation of viable and nonviable tumor cells in the corneal stroma. They showed a nonspecific localized interstitial keratitis accompanied by inflammatory cells and new capillaries. We could not observe any significant clinical or histopathological differences between the corneas containing live or dead tumor implants, or between those with different tumor types (i.e., retinoblastoma and melanoma). Some variation in the severity of the inflammatory response was observed in different animals with the same tumor. In all cases, the extent of the corneal neovascularization correlated with the degree of inflammation. However, in rabbits made immune-deficient by radiation, there was negligible inflammation and vascularization when tumor was implanted.

Calcification in retinoblastoma.

Review of 40 cases of retinoblastoma revealed that histologic evidence of calcification was noted in 38 of the 40 tumors, whereas only three of 16 roentgenograms were positive for calcium. Calcificaton detected by roentgenograms and histopathologic examination correlated poorly with quantitative determination for tumor calcium. Compared to control eyes, however, eyes with retinoblastoma contain large amounts of calcium (1.2 vs. 218 mug/ml. ash). This calcification, though frequently not observed in standard roentgenograms, should be detected by the newer diagnostic modalities such as hypocycloidal polytomography, computerized transaxial tomography, ultrasonography, and radionuclide scintigraphy with technetium diphosphonate, a bone-scanning agent.

Retinoic binding in melanomas of the eye and in normal choroid.

Binding proteins for retinoic acid (cellular retinoid acid binding protein, CRABP), and for vitamin A (cellular retinol binding protein, CRBP) have been demonstrated in various cell types; these binding proteins display the characteristics of receptors. In the present study CRABP and CRBP levels were measured in 9 melanomas of the choroid. CRABP was detected in 2 of the melanomas, whereas CRBP was measurable in 1 melanoma. In comparison samples of normal choroid contained CRABP and CRBP in all cases investigated.

Quantitative determination of T cells in ocular lymphoid infiltrates. An indirect method for distinguishing between pseudolymphomas and malignant lymphomas.

T lymphocytes may be identified by two comparatively simple techniques: sheep erythrocyte (E) rosette formation and cytochemically demonstrable acid alpha-naphthyl acetate esterase (ANAE) activity. We tested the quantitative determination of T cells in 17 ocular lymphoid tumors as an indirect method for characterizing their clonality. Six lesions containing greater than 40% T cells (47% to 73%; mean, 62%) were shown to be polyclonal proliferations and were classified as pseudolymphomas by histologic criteria. Seven lesions containing less than 30% T cells (3% to 20%; mean, 10%) were shown to be monoclonal B-cell proliferations and were classified as malignant lymphomas by histologic criteria. Only two lesions containing between 30% and 40% T cells could not be unequivocally assigned to the monoclonal or polyclonal category solely based on the percentage of T cells. In the final two lesions, the tissue specimen was too small to allow a full panel of immunologic studies; both tumors showed a predominance of T cells consistent with their benign histologic features.

Feulgen-positive deposits in retinoblastoma. Incidence, composition, and ultrastructure.

The incidence of hematoxyphilic extracellular deposits in 159 cases of retinoblastoma was found to be 18%. Six other features of these tumors also were examined. In the majority of eyes with these deposits, the tumors were composed solely of anaplastic cells. None was well differentiated with preponderantly rosette formation. The hematoxyphilic deposits were Feulgen-positive, and this staining was diminished but not abolished by pretreatment with purified deoxyribonuclease (DNase), which did abolish nuclear staining. Feulgen staining of the deposits was totally blocked by pronase--DNase pretreatment--suggesting that the deposits may represent DNA-protein complexes. Ultrastructurally, these deposits were electron dense and finely vacuolar. The deposits may originate from degenerating tumor cell nuclei, and the possibility that they may represent immune complexes warrants future study in view of occasional spontaneous regression of this tumor.

Immunohistochemical characterization of human retinoblastomas in situ with multiple markers.

We studied paraffin-embedded specimens from 18 surgically enucleated eyes with retinoblastoma by peroxidase-antiperoxidase immunohistochemistry with antibodies against glial fibrillary acidic protein, S-100 protein, Leu 7 epitopes, neuron-specific enolase, the 200-kilodalton subunit of the neurofilament triplet polypeptide, and retinal S-antigen. We found that (1) glial fibrillary acidic protein, S-100 protein, and Leu 7 epitopes were detected only in well-differentiated glial cells that were interpreted as reactive and not neoplastic, (2) undifferentiated neoplastic cells expressed both neuron-specific enolase and retinal S-antigen immunoreactivity, and (3) differentiated cells forming Flexner-Wintersteiner rosettes were found to express neuron-specific enolase, retinal S-antigen, and, occasionally, neurofilament protein. These results support the view that retinoblastomas are composed of neuron-committed cells and favor the origin of these tumors from photoreceptor progenitor cells. We did not find any morphologic or immunohistochemical evidence of glial differentiation from tumor cells that would support the concept that retinoblastoma arises from a primitive neuroectodermal cell capable of divergent differentiation along neuronal and glial lines.

Immunohistochemistry of retinoblastomas in humans.

We examined immunohistochemically the distribution of the neuronal-specific protein enolase (14-3-2) in normal human retina. We also examined the distribution of neuronal-specific enolase, glial fibrillary acidic protein, myelin-associated glycoprotein, and S-100 protein in seven human retinoblastomas. In normal retina neuronal-specific enolase was present in neurons but not in other cell types. The inner segments of cones stained darkly with antiserum to neuronal-specific enolase; the inner segments of rods stained either weakly or not at all. Most small round cells in all seven retinoblastomas studied stained with antiserum to neuronal-specific enolase but Flexner-Wintersteiner rosettes stained weakly or not at all. One retinoblastoma contained an area of cells that stained with antibodies to myelin-associated glycoprotein which in normal retina is found only in Müller's cells. Another retinoblastoma had an area of cells that stained with antiserum to glial fibrillary acidic protein, which in normal human retina is present only in astrocytes. All seven retinoblastomas lacked detectable S-100 protein. These results supported the conventional view that retinoblastomas are neuronal tumors although some may contain areas that show astrocytic or Müller's cell differentiation.

Detection and quantification of S-100 protein in ocular tissues and fluids from patients with intraocular melanoma.

S-100 protein is a 21,000 dalton acidic calcium-binding protein present in ocular melanomas and some normal ocular tissues. Ocular fluids and extracts of ocular tumours were examined by a sensitive radioimmunoassay that could detect less than 5 ng of S-100 protein in minute volumes of fluid. Three ocular melanoma biopsy specimens had S-100 protein at levels between 25 and 1300 ng/ml, comparable to that found in a cutaneous melanoma biopsy specimen (1000 ng/ml). (SI conversion: ng/ml = microgram/l.) Six melanoma culture lines had 1000 to 125,000 ng/ml. Four lymphoblastoid cultures had less than 2 ng/ml, and three colon carcinoma cultures had 180 ng/ml. Subretinal fluid from 23 melanoma-containing eyes had 10 to 76,800 ng/ml. Lesser amounts were found in eyes with small, anteriorly located, lightly pigmented tumours. Vitreous from 3 melanoma-containing eyes had 10,000 to 11,000 ng/ml. Vitreous obtained from three eyes during tractional retinal detachment repair had 500 to 1600 ng/ml, and vitreous obtained at necropsy from six normal eyes had 2 to 120 ng/ml. Aqueous from six melanoma-containing eyes had 10 to 30 ng/ml, levels not significantly different from those observed in three normal eyes (80-120 ng/ml). This approach provides new insight into the interaction of ocular tumours and adjacent ocular fluids and may, with more specific tumour markers, have diagnostic applications.

Postinflammatory and malignant protein patterns in aqueous humour.

Normal aqueous humour and the aqueous humour of patients with cataract is virtually protein-free. Patients having retinoblastoma and non-malignant postinflammatory lesions show significantly high quantities of proteins in the aqueous humour. Retinoblastoma is associated with an increase in the globulin content and an albumin/globulin ratio below unity, while non-malignant postinflammatory intraocular conditions show a rise of the albumin fraction with an albumin/globulin ratio above unity. It seems likely that the leakage of protein into the aqueous is different in the two conditions, and a transcellular route is postulated as being the cause in malignant conditions. The estimation of protein patterns in the aqueous humour may be of value in the diagnosis of intraocular malignancy.

Endogenous sugar receptor (lectin) profiles of human retinoblastoma and retinoblast cell lines analyzed by cytological markers, affinity chromatography and neoglycoprotein-targeted photolysis.

Plant lectins have previously been employed to map the composition of cellular glycoconjugates on retinoblastoma and other tumor cells. To characterize the cellular receptors in protein-carbohydrate interactions, we have applied cytological markers (fluorescent neoglycoproteins) containing common carbohydrate building-blocks to investigate the presence of endogenous carbohydrate-binding proteins in six human retinoblastoma cell lines and one established human retinoblast cell line. The staining patterns showed similar expression of endogenous sugar receptors in all cell lines, with few qualitative differences. However, application of affinity chromatography using resins with immobilized carbohydrates as affinity ligands to isolate sugar receptors (lectins) with binding specificities for beta-D-galactosides, alpha-D-mannosides, alpha-L-fucosides alpha-D-glucosides and to N-acetyl-D-glucosamine and N-acetyl-D-galactosamine, respectively, revealed significant differences between the cell lines, emphasizing the value of complementary biochemical analysis. To demonstrate the practical use of this type of glycobiochemical profiling, selective photodestruction of retinoblastoma cells in vitro was accomplished following incubation with synthetic neoglycoprotein-hematoporphyrin conjugates and subsequent exposure to light. This phototherapeutical approach thus combined the inherent specificity of a neoglycoprotein for a particular cellular phenotype with targeted drug activation.

S-100 protein as a marker for melanocytic and other tumours.

The majority of melanocytic tumours are easily diagnosed but they become a problem when they are amelanotic and the tumour cells resemble those of other tumours. This applies particularly to secondary melanoma. Detection of S100 protein is a useful identifying marker. S100 protein, so named for its solubility in saturated ammonium sulphate, is derived from brain tissue. It is a dimer and belongs to a calcium binding group of proteins. The protein was first thought to be in neural or neural crest derived tissues but has been found in chondrocytes, adipocytes, myoepithelial cells, dendritic cells of lymphoid tissue, Langerhans cells and T lymphocytes. The protein is present in a high proportion of malignant melanomas and nevocytic nevi of skin, but is less positive in eye melanomas. It is present in gliomas, Schwannomas and neurofibromas but not in neurone derived tumours such as neuroblastomas. Chondromas, chondrosarcomas, liposarcomas, some osteogenic sarcomas and some histiocytic tumours are positive. The tumours that do not contain S100 protein are listed. Pending development of melanoma-directed monoclonal antibodies, the use of anti-serum to S100 protein plus anti-keratin and anti-leukocyte reagents is useful in the identification of tumours of doubtful histogenesis.

Localization of glial fibrillary acidic protein in retinal astrocytoma: an immunohistochemical study.

Glial fibrillary acidic protein (GFA) is found in glial filaments, and is specific for reactive or neoplastic astrocytes and ependyma. Using a commercially available antibody to GFA protein, a retinal neoplasm from a patient with neurofibromatosis is intensely stained with the brown immunoperoxidase reaction product. Rather than the less specific phosphotungstic acid hematoxylin (PTAH) or Holzer's stain, antibody to GFAP protein is a relatively rapid, simple, and reliable immunohistochemical technique for formalin-fixed and paraffin-embedded tissue.