Ultrastructural localization of lectin receptors in the monkey retinal photoreceptors and pigment epithelium: application of lectin-gold complexes on thin sections

Comparison of Surgically Excised Premacular Membranes in Eyes with Macular Pucker and Proliferative Vitreoretinopathy

PURPOSE

To describe and compare characteristics of premacular membranes in idiopathic macular pucker (iMP) and proliferative vitreoretinopathy (PVR) using immunofluorescence and transmission electron microscopy.

MATERIALS AND METHODS

For immunocytochemical and ultrastructural analyses, premacular membranes were harvested during vitrectomy from 16 eyes with iMP and 12 eyes with PVR. All specimens were processed as flat mounts for phase-contrast and fluorescence microscopy. We used 19 different primary antibodies such as anti-α-smooth muscle actin (α-SMA), anti-integrin-αv, anti-galectin, anti-IBA-1, anti-EMMPRIN (CD147), anti-ricinus (RCS) and anti-collagen-type I. Eight of 28 eyes were also prepared for transmission electron microscopy.

RESULTS

In all eyes with iMP and PVR, positive immunoreactivity of integrins, especially αvβ3 was found. There was also a strong staining of anti-α-SMA, anti-galectin, anti-EMMPRIN, anti-RCS, anti-IBA1 and anti-collagen-type I. Transmission electron microscopy showed that premacular membrane of iMP composed of myofibroblasts, glial cells and fibroblasts. In eyes with PVR, retinal pigment epithelial cells and myofibroblasts were seen as predominant cell types.

CONCLUSION

Premacular membranes of iMP and PVR presented with similarities in cell distribution and immunoreactivity, but showed differences in cell composition. Herein, we demonstrate immunocytochemical characteristics involved in fibrotic processes. Cell transdifferentiation into myofibroblasts represents an important process in pathogenesis of both entities. In order to address future anti-fibrotic treatment strategies, we emphasize that both fibrotic diseases share distinct immunocytochemical and ultrastructural features.

Differential binding to glycotopes among the layers of three mammalian retinal neurons by man-containing N-linked glycan, T(alpha) (Galbeta1-3GalNAcalpha1-), Tn (GalNAcalpha1-Ser/Thr) and I (beta)/II (beta) (Galbeta1-3/4GlcNAcbeta-) reactive lectins

Carbohydrate structures between retinal neurons and retinal pigment epithelium (RPE) play an important role in maintaining the integrity of retinal adhesion to underlying RPE, and in retinal detachment pathogenesis. Since relevant knowledge is still in the primary stage, glycotopes on the adult retina of mongrel canines (dog), micropigs and Sprague-Dawley rats were examined by lectino-histochemistry, using a panel of 16 different lectins. Paraffin sections of eyes were stained with biotinylated lectins, and visualized by streptavidin-peroxidase and diaminobenzidine staining. Mapping the affinity profiles, it is concluded that: (i) all sections of the retina reacted well with Morniga M, suggesting that N-linked glycans are present in all layers of the retina; (ii) no detectable human blood group ABH active glycotopes were found among retinal layers; (iii) outer and inner segments contained glycoconjugates rich in ligands reacting with T (alpha) (Galbeta1-3GalNAcalpha1-Ser/Thr) and Tn (GalNAcalpha1-Ser/Thr) specific lectins; (iv) cone cells of retina specifically bound peanut agglutinin (PNA), which recognizes T (alpha) residues and could be used as a specific marker for these photoreceptors; (v) the retinas of rat, dog and pig, had a similar binding profile but with different intensity; (vi) each retinal layer had its own binding characteristic. This information may provide useful background knowledge for normal retinal physiology and miscellaneous retinal diseases, including retinal detachment (RD) and age-related macular degeneration (ARMD).

Diethylene glycol distearate (DGD): a versatile embedding medium for retinal cytochemistry

Embedment in diethylene glycol distearate (DGD) was shown to be highly desirable and versatile for retinal cytochemical studies, including in situ hybridization, immuno- and lectin cytochemistry. This method allows for preservation of fine tissue detail as well as good reaction sensitivity. It appears to be more suitable than most other methods currently used for light microscopic retinal cytochemistry.

Interphotoreceptor matrix in the human retina: cone-like domains surround a small population of rod photoreceptors

The interphotoreceptor matrix (IPM) in mammalian retinas is subdivided into rod and cone specific compartments: peanut agglutinin (PNA) binding glycoconjugates are associated with cones, whereas wheat germ agglutinin (WGA) binding glycoconjugates are associated with rods. To establish the identity of a photoreceptor cell type in the human retina with rod dimensions but with a matrix domain which stains with PNA, double label studies, using PNA-ferritin to decorate the extracellular domains and immunocytochemical techniques using a rod specific anti-opsin antibody were conducted. The PNA-binding domains were observed in the cone-associated IPM as well as in the IPM surrounding a small population of rod-shaped photoreceptors. The outer segments of these rod-shaped photoreceptors showed intense labeling with a rod specific anti-opsin antibody as did all other rods which were free of PNA-labeling. A quantitative analysis of all retinal quadrants indicates that this novel rod represents approximately 0.3% of the total rod population in the human retina.

Transient hyperglycosylation of rhodopsin with galactose

Rhodopsin's oligosaccharide chains contain predominantly two types of sugar residues: mannose and N-acetylglucosamine. In the present work, bovine and rat rhodopsin were analysed biochemically for the presence of a third sugar, galactose. Treatment of bovine rod outer segments (ROS) with galactose oxidase followed by reduction with tritium-labeled sodium borohydride revealed the presence of existing molecules of galactose on rhodopsin. Rats injected intravitreally with [3H]galactose and [14C]leucine and maintained in darkness were killed 1 hr, 6 hr, 1, 3 or 5 days following the injection. Retinas were collected for subcellular fractionation and rhodopsin from each of the fractions was purified by ConA sepharose chromatography and SDS-PAGE. During the first 6 hr, galactose selectively labeled rhodopsin in the Golgi-enriched fraction resulting in increased [3H]/[14C] ratios in both Golgi and ROS. The data suggested that trimming was occurring at the transition from Golgi to ROS. Furthermore, a decrease in isotope ratio in the ROS between 6 hr and 1 day suggested further trimming of rhodopsin after membrane assembly in the ROS. Additional in vivo experiments demonstrated existing molecules of galactose on rhodopsin's oligosaccharide chain using lectin affinity chromatography. Rats injected intravitreally with [35S]methionine were dark-adapted for 2 hr. Following subcellular fractionation of retinas, ConA purified rhodopsin from ROS was applied to one of two additional lectin columns: Ricinus communis agglutinin (RCA) or Griffonia simplicifolia I (GSA). Eight to nine percent of the labeled rhodopsin was bound to and eluted from RCA, whereas none bound to GSA, indicating the presence of a beta-galactoside. The RCA agarose eluted protein co-electrophoresed with a rhodopsin standard and was light sensitive. Galactose was shown to be the terminal sugar on this subset of rhodopsin and was not capped by neuraminic acid. Binding of rhodopsin's oligosaccharide to RCA was abolished by pre-treatment with beta-galactosidase. Decreased binding of rhodopsin to RCA was observed following intravitreal injection of castanospermine but not swainsonine. Of those two inhibitors of glycoprotein trimming, only castanospermine would be expected to prevent the addition of galactose to the oligosaccharide. The association of galactose with rat rhodopsin appeared to be a transient one. At 2 hr, 8-9% of rhodopsin contained galactose, at 6 hr only 2.2% had galactose and by 24 hr less than 1% did. The galactose was trimmed from rhodopsin's oligosaccharide presumably after its role was complete. Separation of rhodopsin of the plasma membranes from rhodopsin of discs indicated that 75% of the galactose-containing rhodopsin was in the plasma membrane and only 25% was in the discs. These findings suggested a possible role for galactose in new disc formation with subsequent removal after the discs are sealed.

Rod and cone specific domains in the interphotoreceptor matrix

The insoluble matrix domain of the interphotoreceptor matrix (IPM) from normal dog, cat, and mouse retinae were characterized using lectin cytochemistry. The lectins WGA (wheat germ agglutinin) and PNA (peanut agglutinin) were used to label interphotoreceptor matrix microdomains in cryosections of retinal tissue and in extracted insoluble matrix. Retinal cryosections and extracted matrix were examined by epifluorescence microscopy and scanning confocal laser microscopy, the latter allowed for the removal of all background fluorescence and gave increased resolution. The insoluble matrix was extracted as a continuous sheet that was comprised of two photoreceptor-specific matrix domains distinguished both by the size of the domains, and by differential binding of WGA and PNA lectins. Each domain encloses a photoreceptor inner and outer segment. Individual rod-associated domains were connected into a hexagonal lattice and this pattern was regularly interrupted by the larger cone-associated domains which have 8-10 surrounding rod domains. The PNA lectin primarily labeled the cone-associated matrix with faint binding to the rod matrix; the WGA lectin labeled both the rod- and cone-associated matrix.

Characterization of galactose-containing glycoconjugates in the human retina: a lectin histochemical study

Seven specimens of morphologically normal formalin-fixed and paraffin-embedded human retina were studied using a panel of fourteen biotinylated lectins, all of which react with glycoconjugates containing galactose and N-acetylgalactosamine residues. Agaricus bisporus (ABA), Bauhinia purpurea (BPA), Phaseolus vulgaris (PHA-E), peanut (PNA) and Ricinus communis (RCA-I) agglutinins labeled photoreceptor cells prior to enzymatic predigestion. BPA and PNA bound specifically to cones. The plexiform layers reacted with ABA, BPA and PHA-E, while only ABA and PHA-E labeled the nuclear layers. After pretreatment with neuraminidase to remove terminal sialic acid, all five lectins, as well as Erythrina cristagalli (ECA), Helix pomatia (HPA) and Maclura pomifera (MPA) agglutinins labeled both rods and cones. Furthermore, the plexiform layers additionally reacted with ECA, PNA and RCA-I, and the nuclear layers with BPA and RCA-I after neuraminidase pretreatment. Retinal vascular endothelial cells consistently bound ABA, ECA, PHA-E and RCA-I, but they could also bind BPA, HPA, Bandeiraea simplicifolia (BSA-I), Dolichos biflorus (DBA) and Euonymus europaeus (EEA) agglutinins in unpretreated sections, as well as MPA, PNA, soybean (SBA) and Sophora japonica (SJA) agglutinins subsequent to predigestion with neuraminidase. The nonpigmented ciliary epithelium reacted with the same lectins as photoreceptor cells, but it was also labeled by DBA. Sambucus nigra agglutinin (SNA) did not specifically bind to any intraocular structure. These findings favor the theory that, in unpretreated specimens, Gal(beta 1----3)GalNAc (BPA and PNA) is mainly responsible for labeling of cones, while Gal(beta 1----3/4)GlcNAc units, partly substituted with terminal sialic acid (PHA-E and RCA-I), explain labeling of rods. Following pretreatment with neuraminidase, further Gal(beta 1----3)GalNAc (BPA and PNA) and, especially, Gal(beta 1----3/4)GlcNAc (BPA, ECA, PHA-E, PNA and RCA-I) and alpha GalNAc units (BPA, HPA and MPA), the latter partly linked to the protein backbone, contribute to labeling of photoreceptor cells. Gal(beta 1----3/4)GlcNAc units may be mainly responsible for labeling of nuclear and plexiform layers. Finally, other related receptor sites (SBA and SJA), some of which are blood-group specific (BSA-I, DBA, EEA and HPA) are restricted to retinal vascular endothelia.

Lectin binding patterns in developing canine retina

Developing canine retina, fixed in Bouin's or Tellyesnizky's acetic alcoholic formalin, was examined with a battery of biotinylated lectins. ConA and WGA consistently bound to all retinal layers including the tapetum cellulosum, retinal pigment epithelium, inner and outer segments, plexiform layers and ganglion cells. Binding became more intense with increased age. S-WGA bound to apical portions of the retinal pigment epithelium and to outer segments. Outer segments also bound ConA, WGA and RCA. DBA preferentially labelled all inner segments, while PNA bound to a subpopulation of inner segments. RCA bound intensely to retinal vessels as did PNA after neuraminidase treatment. UEA did not consistently bind to any layers. There was transient binding by some lectins at different developmental stages. This study indicates that canine retina demonstrates differential expression of glycoproteins, as indicated by lectin binding, during development and that this expression is temporally and topographically regulated.

Characterization of a galactosyltransferase in purified bovine rod outer segments

Purified bovine rod outer segments (ROS) were used to study the transfer of labeled galactose from UDP-[3H]galactose to endogenous ROS glycoproteins, exogenous glycoproteins and N-acetylglucosamine (GlcNAc). The ROS reaction was also compared with that of the retinal microsomal fraction and milk galactosyltransferase. The results indicate that the ROS reaction was enhanced by exposure to light. Illumination, however, had no effect on the transfer of labeled galactose to either endogenous microsomal glycoproteins by retinal microsomal galactosyltransferase or the transfer of the sugar to ROS glycoproteins by milk galactosyltransferase. Manganese was most effective, followed by cobalt, as cofactor for the ROS enzyme. Calcium and magnesium produced about 60% of the activity observed with manganese. The ROS enzyme transferred minimal amounts of labeled galactose to asialo-agalactotransferrin or ovalbumin but readily transferred the sugar to GlcNAc. The latter reaction had an optimum pH of 6.3 and was linear for at least 90 min. It reached a maximum at about 30 mM GlcNAc and was inhibited by higher concentrations of the aminosugar and by low concentrations of alpha-lactalbumin. On the other hand, the transfer of galactose to ROS glycoproteins was not affected by low concentrations of alpha-lactalbumin. Our data suggest that the ROS galactosyltransferase may have a certain specificity towards its acceptor in the ROS. Its activation by light may indicate a role in the light-activated processes of the photoreceptor cell.

Lectin binding of the interphotoreceptor matrix during retinal development in normal and RCS rats

The retinas of both normal and Royal College of Surgeons (RCS) rats with inherited retinal dystrophy have been examined using lectin histochemistry to determine the developmental and degenerative changes of the glycoconjugates in the interphotoreceptor matrix (IPM) between postnatal day (P) 10 and P25, when the adult lectin binding patterns are seen in normal rats. Wheat germ agglutinin (WGA; recognizing sialic acid and/or N-acetyl-D-glucosamine) bound to the apical surface of the retinal pigment epithelium (RPE) sparsely at P10 and prominently at P12 in both strains. In both strains at P14, WGA also stained the basal outer segment zone at the inner segment-outer segment junction. Between P14 and P16 in both strains, there was a dramatic increase in the binding of the interstitial region, the space alongside the outer segments and between the apical and basal outer segment zones. The binding pattern of WGA in normal rats remained basically unchanged from P16 to P25, although the intensity of binding was increased somewhat. Ricinus communis agglutinin-1 (RCA-1; specific for galactosyl residues) bound to the outer segment zone prominently and diffusely with increasing intensity with age at P10, P12 and P14 in both strains. At P16 and older, the intense binding of the interstitial zone was dramatically reduced and the RCA-1 bound primarily to the inner and outer segment junctional region, with weak binding to the apical surface of the RPE in both strains. At P25, the binding of the inner and outer segment junctional region was even more restricted, limited to punctate sites in this zone in normal rats and almost missing in RCS rats.(ABSTRACT TRUNCATED AT 250 WORDS)

Transmembrane assemblage of the photoreceptor connecting cilium and motile cilium transition zone contain a common immunologic epitope

The photoreceptor connecting cilium bears a unique transmembrane assemblage which stably links cell surface glycoconjugates with the underlying axonemal cytoskeleton. Structural similarities between the photoreceptor connecting cilium and the transition zone of motile cilia suggests that this assemblage may also be present in motile cilia. Using a subcellular fraction enriched in detergent-extracted photoreceptor axonemes, three high molecular mass glycoconjugates (425, 600, and 700 kD) were previously identified as potential components of the assemblage. Through oligosaccharide characterization and binding of a specific monoclonal antibody, we have verified the localization of the 425 kD glycoconjugate to the transmembrane assemblage. Binding of the lectin peanut agglutinin (PNA) to the 425 kD glycoconjugate on nitrocellulose blots, and to isolated detergent-extracted axonemes, was assessed following treatment with the enzymes neuraminidase and O-glycanase. Changes in binding to the 425 kD glycoconjugate precisely paralleled changes in binding to intact axonemes, supporting the hypothesis that the 425 kD glycoconjugate is a component of the transmembrane assemblage. Furthermore, the results suggest that the 425 kD glycoconjugate contains sialated galactose-N-acetylgalactosamine oligosaccharides which are O-linked to the protein backbone. To directly assess the distribution of the 425 kD glycoconjugate, we produced a monoclonal antibody directed against this glycoconjugate. The antibody, K26, recognizes only the 425 kD on transblots of the axoneme fraction. K26 immunoreactivity of intact axonemes is identical to that seen by PNA staining. K26 staining of isolated photoreceptors and whole retina is uniquely localized to the region of the connecting cilium. Thus, in the photoreceptor, the 425 kD is not only a component of the transmembrane assemblage but is also completely restricted to the connecting cilium. Based on morphological similarities, the photoreceptor connecting cilium is thought to be homologous to the transition zone of the motile cilium. As such, we have stained oviduct epithelium with the K26 monoclonal antibody. Immunoreactivity is restricted to the region of the transition zone at the base of motile cilia.

Lectin-binding sites in the anterior segment of the human eye

The anterior segment of the human eye was screened for differences in the lectin-binding patterns of Con A, PNA, GS-I, WGA, SBA, DBA, and UEA-I to enable cell typing for cell-culture purposes. An immunohistochemical technique combining an indirect antibody-lectin method with the avidin-biotin system was used. Con A and WGA were bound by all cells except the conjunctival goblet cells. UEA-I was exclusively bound by both vascular endothelial cells and some corneal and conjunctival epithelial cells. The binding of GS-I, PNA, SBA, and DBA showed an uneven pattern and differed among the cases investigated. The reasons for these differences are not clear. Our results indicate that the usefulness of lectins for cell-typing purposes is restricted and must be determined for every case.

Ultrastructural visualization of primate cone photoreceptor matrix sheaths

Glycoconjugates, including glycolipids, glycoproteins, and proteoglycans, are present in the plasma membrane of photoreceptor cells and in the interphotoreceptor matrix surrounding photoreceptor cell ellipsoids and outer segments. Although the precise function of these molecules is unknown, they may be important in mediating photoreceptor-pigment epithelial cell interactions, outer segment membrane assembly, and/or disc shedding. Lectins, affinity ligands for defined carbohydrate sequences, have proven particularly useful in studying the glycoconjugate composition of the interphotoreceptor matrix. The peanut lectin selectively binds to domains of the interphotoreceptor matrix surrounding cone ("cone matrix sheaths"), but not rod inner and outer segments. This is evidence for the existence of chemical and structural heterogeneity within the interphotoreceptor matrix. The studies described herein utilized ultrastructural pre-embedding histochemical labeling to assess whether, in addition to the surrounding interphotoreceptor matrix, peanut lectin binding is associated directly with that plasma membrane of cone inner and outer segments. This study confirms that ferritin-conjugated peanut agglutinin binds to cone matrix sheaths, and, in addition, provides ultrastructural evidence for the presence of binding to the plasma membrane surrounding cone inner and outer segments. The data suggest that cone membrane-associated peanut agglutinin-binding molecules may differ from those located within cone matrix sheaths.

Specialization of the interphotoreceptor matrices around cone and rod photoreceptor cells in the monkey retina, as revealed by lectin cytochemistry

The binding sites of two lectins, peanut agglutinin (PNA) and wheat germ agglutinin (WGA), in the interphotoreceptor matrix (IPM) and photoreceptor plasma membranes of the Japanese monkey (Macaca fuscata) retina were localized using a pre-embedding staining method with ferritin-conjugated (Fer) lectins as well as a postembedding staining method with fluorescence-labeled (FITC) lectins. FITC-PNA, but not WGA, stained cylindrical domains of the IPM around cone outer and inner segments, while the IPM around rods stained with FITC-WGA but not PNA. When the intact (not detached) retinal tissues were incubated with Fer-lectin, the lectin generally labeled neither the IPM nor photoreceptor plasma membranes, but labeled only those structures in detached portions occurring at the edges of occasional retinal tissue blocks. Thus, the neural retinas physically isolated from the retinal pigment epithelium (RPE) were utilized principally here. Ultrastructurally, the IPM in the intact retina consisted of granular and filamentous materials; the IPM in the isolated neutral retina also retained those components, although somewhat loosely organized, and the IPM around cones appeared to be preserved better than did the IPM around rods. Fer-PNA bound to the IPM associated with cones, but not rods; Fer-WGA bound to the rod- but not cone-associated IPM. The ferritin particles were found to lie close to the granular and filamentous materials. Those photoreceptor-associated IPMs extended to the apical surface of the RPE in detached portions or to the apical villi of the RPE which were frequently found in the isolated neural retinas. Also, Fer-PNA labeled the cone, but not rod, plasma membranes; Fer-WGA bound heavily to the plasma membranes of rod and cone outer segments, but sparsely to those of their inner segments. These results suggest that the IPM comprises chemically and physically differential domains specialized for cone and rod photoreceptor cells, and that these specialized IPM are structurally so stable that may be involved in isolating photoreceptor cells physicochemically from each other and in the interactions between the photoreceptors and the RPE, such as retinal adhesion.

Effect of light on the transfer of sugars from sugar nucleotides to rod outer segment membranes of control and dystrophic rats

The transfer of N-acetyl-D-glucosamine (GlcNAc), D-mannose (Man), D-galactose (Gal) and L-fucose (Fuc) from their nucleotide complexes to isolated rod outer segment (ROS) membranes obtained from dark-adapted 21 +/- 2 days old dystrophic (RCS) and control (RCS-rdy+) rat retinas, was studied under light or dark conditions of incubation. It was found that all of these sugars were transferred to ROS membranes in the dark. Under these conditions there was significantly less (p less than 0.001) Gal transferred to dystrophic than to control membranes. Exposure to light affected the transfer of Gal and Fuc only. Thus, the transfer of Gal and Fuc to control ROS membranes was increased by about 50% compared to the level observed under dark conditions of incubation. On the other hand, exposure to light had no effect on the transfer of Gal to dystrophic ROS membranes but it enhanced the transfer of Fuc to these membranes by about 250% above the level observed in the dark. Under light there were highly significant (p less than 0.001) differences between control and dystrophic membranes in the transfer of Gal and Fuc. The transfer of Fuc to dystrophic ROS membranes was proportional to the concentration of GDP-Fuc but the acceptors on control membranes were saturated at low concentrations of substrate. However, the transfer of Gal from UDP-Gal to both types of membranes was proportional to the concentrations of substrate and ROS membrane protein and to the period of incubation. The transfer of Gal and Fuc to both types of membranes was significantly reduced after denaturation of ROS membrane proteins.(ABSTRACT TRUNCATED AT 250 WORDS)

Lectin binding in retinoblastoma

Lectins were used as cytochemical probes to examine the relationship between normal retina and retinoblastoma. Cases of retinoblastoma in which there were either Flexner-Wintersteiner rosettes or fleurettes accompanied by nontumorous retina were studied with ten biotinylated lectins. Pineal and ocular tumors from cases of trilateral retinoblastoma were also studied. Con-A, LCA, PNA, WGA and RCA-I labeled tissue of each type to varying degrees, while BS-I, DBA, SBA, S-WGA and UEA-I bound neither tumor nor retina. The high degree of binding homology suggests biochemical as well as structural similarities between these tissues.