New York: Alan R Liss, Inc; 1985:33-44. 3. Johnson ... Hollyfield JG, Rayborn ME, Landers RA, Myers KM. .... Landers RA, Tawara A, Varner HH, Hollyfield JG.
Structural Changes of the Interphotoreceptor Matrix in an Inherited Retinal Degeneration: A Lectin Cytochemical Study of Progressive Rod-Cone Degeneration Kristina Mieziewska*X Theo van Veen,% and Gustavo D. Aguirre*-\§
Purpose. In the retinal disorder progressive rod-cone degeneration (prcd) in miniature poodle dogs, the photoreceptor layer degenerates slowly in the course of 5 to 7 years. Components of the interphotoreceptor matrix form a continuous extracellular lattice around photoreceptors. The purpose was to study the photoreceptor cell-matrix interactions during the disease and degeneration phases. Because degeneration rate was slower in cones, the authors also wanted to investigate whether there was a link between the degeneration and the photoreceptor-specific interphotoreceptor matrix domains. Methods. Rod- and cone-specific interphotoreceptor matrix domains were examined during two periods: before morphological signs of disease had appeared and during the degenerative stages. Two lectin probes were used; wheat germ agglutinin and peanut agglutinin. By their affinity for terminal carbohydrates, the lectins visually separated the two photoreceptor-specific domains and allowed follow-up of the fate of the rod and cone matrices separately. Results. Before and during the course of disease, the lectin distribution in rod and cone domains remained normal, however, in the degenerative phase of the disease, there were structural changes in the matrix domains. The matrix connections between the individual domains was disrupted and single domains were formed. Cone domains and, to a lesser degree rod domains, were thickened around the inner and outer segments. Conclusions. The changes occurring in the photoreceptor-specific domains were indicative of structural adaptation to cell death and to degenerative conditions. There was no evidence of an active involvement of the interphotoreceptor matrix components studied in the disease process. Invest Ophthalmol Vis Sci. 1993;34:3056-3067.
JL roteoglycans and glycosaminoglycans are active regulators of cell development and differentiation. It has been established that extracellular matrix components directly influence nerve cell growth in the retina. The
From the *Section of Medical Genetics, School of Veterinary Medicine, and the fScheie Eye Institute and Department of Ophthalmology, School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, the %Department of Zoology, University of Goteborg, Gdteborg, Sweden, and the %Baker Institute for Animal Health, College of Veterinary Medicine, Cornell University, Ithaca, New York. Supported in part by National Eye Institute Grants EY-01244, EY-07705, and EY-06855, the Frances V.R. Seebe Trust, the CERF-PRA Research Fund, the Sweden-America Foundation, the Fredrika Bremer Foundation, the Karolina Widerstroms Fund, The RP Foundation Fighting Blindness USA, and the Swedish National Science Research Council (4644-311). The confocal laser microscope facility is managed by Dr John M. Murray and supported by grants NIH S10RR05008 and NSF MCB-9113313 toJ.M.M. Submitted for publication December 12, 1992; accepted March 24, 1993. Proprietary interest category: N. Reprint requests: Kristina Mieziewska, Department of Zoology, University of Goteborg, Medicinaregatan 18A, S-413 90 Goteborg, Sweden.
3056
direction of ganglion cell axon growth, for example, is controlled by a compositional modification of the surrounding proteoglycan matrix in a temporally and spatially coordinated pattern. 1 The interphotoreceptor matrix (IPM) is an enclosed compartment in which a network of insoluble proteoglycan components surround photoreceptor inner and outer segments and the soluble constituents of the IPM. Within the insoluble matrix, individual rod and cone outer and inner segments are invested by photoreceptor type specific domains.2"8 Although the composition of the proteoglycan matrix is not fully known, it appears to be specific for the IPM. 9 In addition to the two photoreceptor classes, the IPM is in contact with the retinal pigment epithelium (RPE) and the Miiller cells, and all four cell types may contribute to the formation of this complex extracellular environment. The function of the carbohydrate-rich IPM is poorly understood, but multiple functions are likely; for example,
Investigative Ophthalmology & Visual Science, October 1993, Vol. 34, No. 11 Copyright © Association for Research in Vision and Ophthalmology
IPM in prcd retinal adhesion 10"13 as well as developmental regulation of cell growth and differentiation. The role of the IPM in developmental regulation has been suggested by temporal expression of different proteoglycan fiber types in the mouse IPM during early postnatal development.14 The structural complexity of the IPM is presumed to be significant to the function or viability of the photoreceptors and/or RPE. A matrix disorder that affects either its synthesis or maintenance could adversely affect these two cell layers; alternatively, disease in the photoreceptors or RPE could induce secondary changes in the matrix structure. The IPM has been investigated in several rodent models of hereditary retinal degeneration, including RCS rats15"18 and the rd and rds mutant mice19, as well as in canine models.20'21 No changes have been found in the proteoglycan matrix in these animal models (except for a distributional change in RCS rats associated with debris accumulation), which indicates that the matrix surrounding both diseased and normal cells in the same eye remain normal. Progressive rod-cone degeneration (prcd) is an autosomal recessive disease in which the primary defect is not known; the mutant retina exhibits a series of characteristic, sequential clinical, morphological, and biochemical abnormalities. The photoreceptor pathology includes elongated and irregular outer segments with disorganized stacks of outer segment discs, and the appearance of villiform extensions from the outer segment membrane into the IPM, which, in cross sections, appear as vesicular profiles. Rod degeneration precedes that of the cones. The prcd-affected retinas also exhibit a spatial distribution of the disease, where degeneration develops earlier and progresses more rapidly in the inferior quadrant compared to the superior or temporal quadrants.22 In a previous study, we examined the peanut agglutinin (PNA) domain surrounding cones in the normal and pra/-affected retinas.20 We found that the PNA domain remained normal as long as the photoreceptor inner and outer segments persisted. When that study was performed, we were unaware of the recently described domain that also surround rods, as well as the methods used for identifying and morphologically examining the insoluble IPM. Because the rod and cone matrices form a continuous sheet, we now examine both photoreceptor-specific domains in tissue sections and in isolated matrix preparations from the prcd-affected retina using conventional and confocal scanning laser microscopy. The enhanced resolution provided by the latter method permits a more detailed assessment of matrix and cell-matrix interactions. In addition, the IPM was evaluated in terms of the structural integrity of the rod and cone matrix lattice, as well as for the interaction between the two photore-
3057 ceptor-specific matrix domains. We examined retinas just before the appearance of disease (12 weeks), until the time of complete degeneration of rod and cone photoreceptors (5 to 7 years). To separate the rod and the cone matrix domains we used two lectins, PNA to specifically label the cone matrix sheath of the insoluble matrix, and wheat germ agglutinin (WGA) to label the insoluble rod matrix as well as a component of the cone matrix. The temporal difference between rod and cone degeneration, in combination with the lectin probes used, allowed us to study the rods and cones separately to uncover changes in the photoreceptorspecific domains. We ultimately wanted to establish whether the two matrix domains were in any way modified before or during the disease and degeneration phases of this disorder and also whether any alterations would be photoreceptor-type specific, indicating a capacity for cells to modulate their matrix domains.
MATERIALS AND METHODS Eyes from prcd-affected miniature poodles, aged 8 weeks (5), 12 weeks (1), and 16 weeks (3), 7 months to 1.5 years (4), 2.8 to 3 years (5), and 5 to 7 years (2) were used in this study. Normal miniature poodles 8, 12, and 16 weeks, 7 months (6), and older dogs of various breeds were used as controls (5). The dogs were raised in an NEI/NIH supported colony (EY06855), and were housed in indoor kennels under cyclic light conditions. All procedures used in this study were in full compliance with the ARVO Resolution on the Use of Animals in Research.
Tissue Preparation Eyes were enucleated in light-adapted dogs during lethal pentobarbital anesthesia, or, in some cases, immediately after death. The anterior segments were removed, and the eyecups were fixed in 4% formaldehyde + 0.1% glutaraldehyde dissolved in 0.1 mol sodium cacodylate buffer (pH 7.2 to 7.4). After fixation on ice for 2 hours, the eyecups were divided into superior, inferior, temporal, and nasal parts, rinsed extensively, and infiltrated with phosphate-buffered saline + 15% sucrose and then +30% sucrose before quick freezing in liquid nitrogen. The samples were frozen, embedded in OCT (Tissue Tek [Miles Inc., Elkhart, IN]). Some samples were infiltrated and embedded in acrylamide before freezing.23 The retinas were cryosectioned (10 jim) and placed on chrome-alum coated slides. The insoluble matrix was extracted from the retina by hypotonic lysis.24 The retina was immersed in distilled water with the addition of 2 mmol/1 CaCl2, and the extracted matrix was collected with a pipette and fixed on chrome-alum-coated coverslips in 4% formaldehyde for 1 hour. The coverslips were air
3058
Investigative Ophthalmology & Visual Science, October 1993, Vol. 34, No. 11
dried and stored at —20° C. These procedures have been described previously in detail.7-21 Lectin Cytochemistry The lectins used were WGA, which binds N-acetyl glucosamine (Glc-NAc) and sialic acid, and PNA, which binds galactose (Gal/31-3GalNAc). Lectin cytochemistry was performed using WGA-FITC (fluorescein isothiocyanate, Vector Laboratories, Burlingame, CA), PNA-TRITC (tetramethylrhodamine isothiocyanate) and PNA-Texas red (E.Y. Laboratories, San Mateo, CA). The lectins were used at concentrations of 40 /ig/ml (WGA) and 200 /*g/ml (PNA) and the tissues were incubated in darkness at room temperature for 1 hour. The sections were rinsed and mounted in Aqua poly-mount (Polyscience, Warrington, PA) for light microscopy, or vinol (polyvinyl alcohol, Air Products, Allentown, PA) with anti fade agents added25 for confocal scanning laser microscopy. Lectin specificity was verified with the appropriate competing hapten sugars: N-acetyl-chitobiose for WGA and galactose for PNA.
Microscopy Epifluorescence microscopy was performed with filter combinations appropriate for the fluorescent markers used. Photographs were taken with a 100X plan-neofluar oil objective, using Kodak Tri-X (400 ASA; East-
man Kodak, Rochester, NY) black and white film exposed at an 800 ASA setting. In the confocal scanning laser microscope, the laser light source was either an argon laser or a krypton laser with a complete separation of the two channels for FITC and Texas-red double label (excitation lines 568.2 nm for Texas-red and 482.5 nm + 476.2 nm for FITC) Detailed technical specifications of the procedure have been published previously.21 The 10jim-thick tissue was optically sectioned in 0.5-jnm increments. Single images were collected serially, stored on an optical disc, and combined later to display the whole section. The images were processed to maximize brightness and contrast, a procedure that did not change relative brightness levels within the image or corrupt the original information collected from the sample. The brightness of the lectin label before computer enhancement was consistent throughout the different tissue samples. All estimates of brightness were made with the original sample, through microscopic examination. Although the parameters for each sample are identical, intensity comparisons were avoided between different samples because of the use of different embedding and microscopic procedures. Lectin label intensity is compared throughout the study within the same tissue section or extraction, and is denoted as strong or weak relative to the brightness of the specific lectin label in that sample.
FIGURES 1 TO 7. Retinal sections double labeled with PNA-Texas-red and WGA-FITC. Each section is presented in three ways: the first image (A) is a single (0.5 iim) optical section, recording the WGA-FITC image; the second (B) is the corresponding optical section collected with the PNA-Texas-red filters, the third image (C) is a composite of several optical sections in order to show the circumference of the larger PNA labeled cone matrix domain. The addition of optical sections improves contrast (magnification XI500). FIGURE l. Freeze-embedded, normal retina, aged 7 months. The wheat germ agglutinin domain (A) is present from the (outer limiting membrane, arrowhead) to the retinal pigment epithelium (*) and surrounds both rods and cones. The matrix in the outer segment layer is thicker than in the inner segment area; the labeling of the outer segments by wheat germ agglutinin adds to the brightness in this layer. The thin optical sections enable the visualization of individual rod matrix domains. (B) The corresponding peanut agglutinin matrix domain of cones is thicker than the wheat germ agglutinin domain and surrounds cones only from a point slightly above the outer limiting membrane to the RPE apex. (C) The peanut agglutinin label in a composite displays the whole cone matrix more accurately (arrow). (Magnification XI500). FIGURE 2. Eight-week-old, freeze-embedded, prcd-affected, stage 0. The wheat germ agglutinin (A) and peanut agglutinin (B) and (C) label do not differ from the normal matrix label. The irregularities seen in the domains are fixation and freezing artifacts. (Magnification X1500). FIGURE 3. Sixteen-week-old, freeze-embedded, prcrf-affected retina, stage 0/1. (A) Wheat germ agglutinin label is normal and there are no visible signs of disease in the matrix. (B) The corresponding peanut agglutinin label also has a normal binding pattern to the cone matrix. (C) The whole cone domain is visualized in the composite, as a smooth matrix domain surrounding half of the inner segment, and the entire outer segment. (Magnification XI500).
IPM in prcd
3059
3060
Investigative Ophthalmology & Visual Science, October 1993, Vol. 34, No. 11
RESULTS The morphological criteria used to characterize the different disease stages of prcd have been described previously.22 These stages cover the transition from normal to outer segment disease (stages 0 to 1), disease to degeneration (stages 2 to 4), and degeneration to atrophy (stages 5 to 8). The current study of prcd covered all phases of the disease; the normal (stage 0), diseased (stage 1), early degeneration (stages 2 to 3), and complete degeneration of the photoreceptors (stages 5 to 6). Normal Retina The WGA lectin bind equally to the rod and cone matrices surrounding the inner and outer segments. The WGA-labeled cone matrix can be distinguished as slightly thicker than the corresponding rod matrix. In the 0.5-)um-thick confocal scanning laser microscope section, matrix label at the outer segment level is more intense than at the inner segment, because the matrix is thicker in the outer segment region. A strong rod outer segment membrane label also adds to the general brightness at that level (Fig. 1 A). This is best visualized with the 0.5-/xm optical sections taken with the confocal scanning laser microscope. The PNA-labeled cone matrix is demonstrated in two ways. The first image (Fig. IB) is a single 0.5-fim thick optical section, directly corresponding to the WGA image; the second is a composite of several optical sections to show the circumference of the PNA-labeled cone matrix domain (Fig. 1C). The composite view was added to give a more accurate representation of the larger cone domain because only parts of a cone domain may be displayed in a single section. Due to computerized frame brightness averaging, the two PNA-labeled images have the same brightness, although there is a 5 to 7 nm difference in thickness. The cone-specific domain begins approximately 4 /xm above the outer limiting membrane, and terminates at the apical surface of the RPE. The cone matrix also has a mildly mottled appearance, which is an artifact caused by fixation and OCT freeze embedding. This was eliminated by acrylamide embedding; however the acrylamide compromised the structural integrity of the rod matrix.7 prcd Retina The earliest examined age of the prcd-affected dog was 8 weeks. At this age the retina is structurally mature and shows no evidence of disease. WGA and PNA label of the 8- and 12-week (data not shown) affected retinas was similar to the normal (Fig. 2). At 16 weeks, WGA and PNA labeling does not show any signs of modified lectin distribution or structural changes in the matrix (Fig. 3). This is the age when the first subtle morpho-
logical changes (stage 1) are present; these are characterized by irregularities of the outer segment contour and disorientation of the orderly disc stacking.26 By 7 months of age, disease is established over the entire retinal expanse, but the progression is quadrant dependent. In the inferior quadrant, degeneration is present and it is represented by stages 2 to 4. In contrast, the superior quadrant has a slower course of the disease with stages 1 and 2 predominating. The temporal region is least affected at this age.28 In older animals, disease progression maintains the topographic distribution but, eventually, all retinal areas show end-stage retinal atrophy and gliosis. In the degenerative stages of the disease, cell death progresses and the number of rod photoreceptors are reduced; disease also develops in cone cells. In stage 2, the photoreceptor layer is in a state of disorganization, exhibiting cell debris and macrophages. The WGA binding is strong and some of the individual rod outer segments can be distinguished (Fig. 4A). When present, macrophages were seen in the center of the outer segment layer as a dark space, which contained PNA-labeled inclusions (Fig. 4B, asterisk). The PNA labeling of the cone domain is normal both in thickness and distribution (Fig. 4C). Stage 3 is characterized by shortened inner and outer segments with resultant narrowing of the interphotoreceptor space. WGA label is distinct in the inner segment layer, but, in the outer segment layer, label is diffuse because of photoreceptor disorganization (Fig. 5A). Cone domains are strongly labeled with PNA and the matrix is slightly thicker than normal (Fig. 5, B and C). PNA binding to a selected, small number of cone domains is reduced compared to the other cones; this is illustrated in the single optical section where, in one of the cones, the outer segment is visible but the surrounding matrix label appears to be absent (Fig. 5B). In the composite image, the difference in label intensity between these and the normal cone domains is correctly displayed (Fig. 5C). However, the WGA label on this cone is not diminished compared to the same label on surrounding rod cells (Fig. 5A, arrowhead). These weakly labeled cone domains are found intermittently in the older, affected retinas. Additionally, faint labeling of some of the cone outer segment extends beyond the apical limits of the matrix label (Fig. 5C, arrowhead). In the 3-year-old/?rcrf-affected dog, stage 3 disease was found in the central part of the retina, where substantial loss of cells had occurred without resulting in shortening of the surviving photoreceptors (Fig. 6). The WGA lectin strongly labels the matrix immediately adjacent to the rod outer segment, but little WGA labeling is present in the areas between the rods. That space is weakly labeled and has an amorphous appearance. The WGA-labeled matrix around cones has the
IPM in prcd
FIGURE 4. Freeze-embedded, 1.5-year-old, /?ra2-affected retina, stage 2. (A) wheat germ agglutinin labeling is distinct although outer segments and matrix are not well defined. (B) The peanut agglutinin domain is normal in thickness and binding specificity. Fluorescent granules are present in the cytoplasm of a macrophage (*) located in the outer segment layer. (C) The composite shows the entire peanut agglutinin cone matrix domain. The photoreceptor layer is irregular due to the active degeneration, which primarily affects the rods. (Magnification XI500). 5. Seven-month-old, freeze-embedded, prrd-affected retina, stage 3 degeneration. (A) The wheat germ agglutinin label is prominent around the photoreceptor inner segment. Outer segments are short and disorganized, and the matrix is less defined. (B) the peanut agglulinin labeling of the matrix is thicker than normal around the cones (arrow). The label diminishes as it approaches the RPE (arrowhead). (C) One cone (B*) is peanut agglutinin labeled to a lesser degree than the others. The actual difference in brightness is correctly shown in the peanut agglutinin composite, where the added sections improve the contrast range and defines details in the image compared to a single section. The wheat germ agglutinin label on that cone matrix is not diminished compared to the matrix label on the normal cones (A, arrowhead). The wheat germ agglutinin label of cone matrix had only a narrow area of overlap with the peanut agglutinin label of the same cone (compare A and B, arrow). (Magnification XI500). FIGURE
3061
3062
Investigative Ophthalmology & Visual Science, October 1993, Vol. 34, No. 11
FIGURE 6. The central retinal area of a 3-year-old dog with stage 3 disease, acrylamide embedded. (A) Wheat germ agglutinin strongly labels the interphotoreceptor matrix space adjacent to the outer segments; there is diffuse but weak labeling of the remaining interphotoreceptor matrix. The few surviving photoreceptors are well organized, except for the whorls of outer segment debris located on the RPE apical surface. Wheat germ agglutinin labels the retinal pigment epithelial cone sheaths strongly (arrow). (B) The peanut agglutinin domain is thicker than normal, smooth and slightly "fluffy" in appearance. (C) The composite shows the diminishing peanut agglutinin label intensity at the retinal pigment epithelial level (arrowhead). Retinal pigment epithelial lipofuscin granules were brightly labeled. The usual shortening of the photoreceptor layer was absent in this retina. (Magnification XI500). FIGURE 7. The peripheral retina of the same 3-year-old affected dog illustrated in Figure 6, acrylamide embedded shows more advanced stage 3 disease. (A) wheat germ agglutinin label is strong immediately around the remaining cells (arrowhead). (B) Peanut agglutinin label is prominent and thick around cone inner segment and outer segment. Note that the cone matrix thickness differs when visualized with the wheat germ agglutinin and peanut agglutinin lectins (compare A and B, arrow). (C) The composite shows the circumference and thickness of the peanut agglutinin domain around the cone. (Magnification XI500).
IPM in prcd
same thickness and distribution as that around rods. In addition, WGA label also extends weakly into the area of the cone matrix that is normally intensely labeled by PNA; that is, there is a significant difference in the binding strength between PNA and WGA in the peripheral area of the cone matrix domain. An unusual observation regarding the WGA results is the intense labeling present around the distal outer segment layer in the region of the RPE apical cone sheath; this area shows weaker than normal PNA labeling (Fig. 6, A and B). An incidental observation is the presence of PNA-labeled lipofuscin granules that accumulate in the RPE and in the outer nuclear layer at this stage of the disease (Fig. 6, B and C). In the periphery of the same retina illustrated in Figure 6, stage 3 disease was present, but the photoreceptor layer had narrowed (Fig. 7). Rod and cone matrix domains are WGA-labeled, and strong WGA label is associated with the IPM adjacent to the outer segments (Fig. 7A). The PNA-labeled cone matrix sheath is thick and has a "fluffy" appearance (Fig. 7, B and C). In older dogs, the retina is largely devoid of photoreceptors and the outer nuclear layer has mostly disappeared (stages 5 to 8). In the 5-year-old retina, no photoreceptors or matrix-like structures are visible with the lectin label. WGA label is strong between cells in the inner retina and also in a thin area below the RPE (Fig. 8A). Patchy PNA label is displayed throughout the inner retina and below the RPE, but it is difficult to evaluate the origin of the lectin label because no structures are recognizable (Fig. 8B). The 7-year-
FIGURE 8. Five-year-old, />r«/-affected retina, acrylamide embedded, stage 5/6. (A) The remaining retinal cells are surrounded with a strong wheat germ agglutinin label. No photoreceptor outer segment or inner segment remain. Patchy and undefined wheat germ agglutinin label is present below the retinal pigment epithelium (arrowhead) but no photoreceptor structures can be identified. (B) The peanut agglutinin label is present in a punctate pattern in the retina but only weak and patchy label can be seen below the apical retinal pigment epithelium. No photoreceptor cells are recognizable. (Magnification X 514).
3063
old dog (not shown) shows a severely degenerated retina composed of unidentifiable cells. The lectins did not label any structures that could be interpreted as matrix residues. Extracted Matrix Preparations The extracted matrix from normal dogs is removed in sheaths containing both rod and cone domains connected into a regular network. When the matrix is extensively washed with deionized water during the isolation procedure, the resultant loss of cations causes the matrix to expand. The rod and cone domains appear more distinct than in preparations in which more conservative washes result in less stretching of the extracted matrix sheet (compare Figs. 9, A and B, and 10, A and B). In both types of preparations, the isolated matrix show the same lectin distribution as observed in retinal cryosections. Because it is impossible to choose an area with a specific disease stage, the extracted matrix from diseased dogs was collected from the entire retina. Matrix from the diseased animals is extracted with some difficulty, particularly in animals older than 1.5 years, and presumably represents the least diseased areas of each eye. The normal WGA-labeled matrix is a honeycombed lattice of interconnecting rod and cone domains. The larger WGA-labeled spaces belong to the cone matrix domain (Fig. 9A, arrowhead). The corresponding PNA-labeled preparation shows the thick and distinct cone domains between the faintly outlined rod matrix domains (Fig. 9B). Both domains are part of the sheet that interconnects the entire matrix around the photoreceptors. The WGA and PNA binding to the extracted matrix of the 16-week-old affected dog was normal in lectin distribution (Fig. 10, A and B). In the 1.5-year-old dog, the WGA matrix domain is difficult to visualize because of the cell debris present in the samples. At this stage of the disease the matrix does not extract easily from the retina. Although the PNA-labeled cone domains are distinct but irregular in appearance, the WGA rod matrix is difficult to identify (Fig. 11, A and B). This is the result of extensive debris, particularly photoreceptor inner and outer segments entrapped within the individual rod matrix domains. At 3 years, matrix could only be retrieved from one of the three dogs examined. This was not unexpected because there was a variation among the dogs in degree of degeneration at this age. Strong WGA label of outer segment and surrounding debris is observed, and because of the severe cell contamination and disorganization of the sample, a distinct rod matrix structure is not visualized (Fig. 12a). In contrast, we found structurally diffuse and irregular, but labelspecific PNA-labeled cone matrix domains (Fig. 12B).
3064
Investigative Ophthalmology & Visual Science, October 1993, Vol. 34, No. 11 primary defect responsible for the cell dysfunction. The slow degeneration present in the prcd mutant retina allows diseased cells to remain viable for a relatively long time. Because individual photoreceptors would remain functional but affected to a variable degree for approximately 1 to 3 years, it can be assumed that modifications occur in the matrix as a result of the adverse conditions induced by the disease. Because prcd is a slow retinal degeneration, the microenviron-
FIGURE 9. Extracted insoluble matrix from normal adult dog after extensive washing. The two lectins in the double-labeled preparation were photographed separately using epifluorescence microscopy. (A) Wheat germ agglutinin label of normal rod (arrow) and cone matrix (arrowhead) domains. The matrix is slightly irregular but within the range of a normal matrix appearance, and the lectin distribution is normal. (B) Peanut agglutinin label of cone matrix domains (arrowhead). The rod-associated structures are faintly labeled in the background. The rod and cone domains are connected into a regular lattice. (Magnification X 1285). FIGURE 10. Extracted insoluble matrix from 16-week-o!d prof-affected retina. The matrix samples were extracted after minimal washing. (A) Wheat germ agglutinin specificity is normal for the rod and cone domains; the irregularities seen in the extracted material are within the limits of normal. (B) The peanut agglutinin domain is regular and shows normal structure and lectin specificity. The viewing angle of the sample is such that the domains are seen obliquely, and do not display the honeycomb pattern seen in Figure 9. (Magnification X 1285).
DISCUSSION The involvement of IPM components in retinal degenerations has not yet been established. Degenerative changes could, however, be initiated in response to a damaged environment regardless of the nature of the
FIGURE n. (A) Wheat germ agglutinin labeled matrix from 1.5-year-old prcrf-affected retina. The matrix preparation is contaminated with photoreceptor cell material, and the structure of the isolated matrix can not be visualized. Matrix label can be observed around the cells. (B) Peanut agglutinin label is present in the matrix around the cone inner segments that contaminate the preparation (arrowhead). The matrix has a slightly irregular appearance. (Magnification X 1285). FIGURE 12. (A) Wheat germ agglutinin label in extracted matrix from 3-year old prcd-affected retina. It is not possible to extract an isolated matrix preparation, and remaining outer segments are brightly labeled (arrowhead). It is difficult to separate the specific matrix label from that bound to the contaminating cell debris. (B) Peanut agglutinin label is normal in lectin specificity, but the domains are damaged and structurally undefined. (Magnification X 1285).
IPM in prcd ment of the IPM during degeneration may be similar to that present in retinitis pigmentosa group of degenerations affecting humans. The development of the retina is normal in the prcd dog; shortly after the retina has differentiated, that is, after 70 to 80 days, morphologically detectable disease in the rod outer segments develop. The sequence of morphological changes follows a complex temporal and spatial pattern. Islands of normal or minimally diseased photoreceptors are present in areas that are surrounded with more advanced degeneration.26 Because of the topographically specified pattern of disease, previously established disease stages were used in our study to evaluate the degree of degeneration within the same eye and between different eyes.22 The stages were selected in each dog to show the fate of the matrix during the disease and degeneration stages of prcd. Because the spatial distribution of disease is an important characteristic of prcd, all four retinal quadrants (superior, inferior, temporal, and nasal) were examined to preclude any variations in lectin binding resulting from spatial localization of the photoreceptors. Retinal areas were also noticed in our study that seemed to be in an active state of degeneration, exhibiting extreme photoreceptor disarray, cell debris, and macrophages. Previous observations indicate that acute degeneration occurs in stage 2.27 Two theories have been proposed to account for the specific pattern; the expression of the disease may be clonal with clusters of cells showing the same disease stage28, or local factors may influence the rate of progression.22 We found no evidence of a variation in lectin binding or matrix structure between the different areas of the retina at any age that could account for the characteristic topographic pattern of disease expression. With the exception of the late atrophic stages (stages 5 to 8) of the disease, the lectin distribution remained normal in all the ages and stages examined. The structure of the individual domains in early stages of prcd was considered to be normal. It should be noted that the artifactual changes of the cone domain seen in the early ages could conceal subtle changes in matrix structure. A thickening of the rod and cone matrix was observed in some samples representing stages 2 and 3. The thickening may represent a secondary response of the surviving cells because the change in thickness was observed at a relatively advanced stage of degeneration when substantial cell death had already occurred. This response would maintain the structural integrity of the IPM, and prevent the disruption of molecular flow through the IPM resulting from discontinuity in the structure. The number of photoreceptors remaining in the IPM compartment has been severely reduced by the time stage 2 is
3065 reached. The extra space is presumably occupied by broadened photoreceptors and matrix domains as seen in later stages. Ultimately, the interconnection between individual domains may be lost and thereby contribute to the disorganization, as is suggested by the IPM extraction studies. In these, the matrix domains are seen surrounding outer and inner segment debris rather than maintaining a structurally intact matrix network. Because of the presence of holes between photoreceptors that result from embedding and/or sectioning artifacts, it is difficult to completely evaluate the continuity of the rod and cone matrices in tissue sections. The thicker matrix may result from villous extensions of the photoreceptor cell membrane that increase the membrane surface area. Assuming that the matrix is attached to the photoreceptor cell membrane through a link protein or a functionally equivalent molecule, the total area available for attachment is greater and thus may affect matrix thickness. By linking the matrix to the cell membrane and preventing the relatively smooth matrix extraction that occurs in normal retinas, these membrane perturbations also could be responsible for the extensive cell contamination observed in the extracted material. Even with the enhanced resolution provided by the confocal scanning laser microscope, it was not possible to resolve these issues unequivocally. In the older diseased retinas, cone matrix domains were found with a less intense PNA label than other cones in the same preparation; in some areas these cones were quite abundant. A weak matrix label might be an indication of cone matrix degradation from release of proteolytic enzymes from the surrounding degeneration tissues. However, the WGA label around these cells appeared normal, which suggests that the changes took place in a subdomain of the matrix and only galactose-containing components were removed from the matrix. Changes in the carbohydrate specificity may reflect vital modifications in cell function. In human retina an apparent increase in the frequency of conelike rods, ie, rod photoreceptors surrounded by a PNA domain, occurs with age29, and also in association with drusen formation30 The origin of the conelike rod domain is unknown, but it is possible that the changes are caused by an alteration of an existing matrix, thus indicating an ability of the cells to modify type-specific matrix components. In normal dogs the cone matrix sheath ends close to the RPE surface. Beyond stage 2 disease, there was a tendency of the cone domain to terminate somewhat more vitreal to the RPE than normal, and, in the older affected animals with shorter PNA cone domains, WGA strongly labeled the RPE cone sheath surrounding these cones. The significance of this finding is unclear, but it could be the result of outer segment debris
3066
Investigative Ophthalmology & Visual Science, October 1993, Vol. 34, No. 11
accumulation below the apical surface of the RPE. Extreme displacement of IPM material due to debris accumulation has been shown in the RCS rats.15"16 Whether the cones are primarily affected in prcd or not, the milieu of the IPM may influence their rate of degeneration. In cone degeneration, a slow, inherited disorder of dogs, only cones are affected, and they degenerate leaving a functional retina with normal rod photoreceptors. This indicates the possibility that a photoreceptor class can survive the surrounding disease.20 However, because rods compose 95% of the total number of photoreceptors in dogs, survival conditions for a normal cone population seem less favorable in the case of selective rod degeneration. The elimination of the majority of the cells would affect the entire IPM composition, and also the alignment and support for the remaining photoreceptors. We have previously studied the IPM in rod-cone dysplasia 1, in Irish setter dogs. 21 The disease results from an early cyclic GMP-PDE deficiency resulting in arrested photoreceptor development with subsequent degeneration. 31 ' 33 Initial morphological abnormalities were seen at or near postnatal day 14, and degeneration of the photoreceptors began between 3 and 4 weeks of age and was complete by 1 year. Despite the etiologic and temporal differences between prcd and rod-cone dysplasia 1, the two diseases are similar in that the rod degeneration precedes that of the cones. This characteristic is also present in other models of retinal degeneration, for example, the rd mouse. 34 In both canine diseases the rod- and cone-specific matrix domains persisted until no part of the photoreceptor remained above the outer limiting membrane. Although the orderly matrix lattice was broken down into individual matrix domains, a failure to produce or maintain individual matrix domains was not seen. Only the connections between individual matrix domains were lost. A change in matrix thickness was also noted in both cases. The similarities between these two distinct disorders support our assumption that these photoreceptor-specific IPM constituents are not primarily involved in the processes of hereditary retinal degeneration. Note that the matrix domains examined in the two studies form only a part of the insoluble IPM, and alterations in other constituents cannot be excluded. The cellular origin of the matrix constituents has not been established. Due to the close relationship between matrix and photoreceptor cells in normal and diseased retinas, it is reasonable to assume that the photoreceptor cell is synthesizing its own matrix domain. Studies have shown that photoreceptors are likely to synthesize chondroitin sulfate proteoglycans, which are parts of the insoluble matrix. 35 The close association between photoreceptor cells, matrix and RPE, however, does not preclude the involvement of
RPE in synthesis of specific matrix components, either directly or through inductive factors. In this study we did not address these questions directly, but the persistence of the matrix domains suggest that the extracellular environment is vital to the viability of the photoreceptors, regardless of the site of their synthesis. In conclusion, alterations in lectin specificities of the rod and cone matrix domains were not found in the course of the disease in the prcd mutant retina. Several structural changes appeared in stage 2, eg, thickening of both rod and cone domains and the loss of the connections between the single domains. The results are indicative of a structural adaptation of the matrix domains during the degeneration, that could be either mechanical or cell regulated. The loss of the orderly honeycombed matrix lattice may damage the orientation and stability of the photoreceptors and thus affect the degeneration rate. The tendency of the two specific matrix subdomains to remain stable suggests that they are essential for the survival of the photoreceptors. Key Words interphotoreceptor matrix, lectin cytochemistry, progressive rod-cone degeneration, retinal degeneration, retinitis pigmentosa Acknowledgment The authors thank Dr. Agoston Szel for reviewing the manuscript, Mrs. Linda Schwartz for secretarial assistance, and Mrs. Susan Nitroy and Mrs. Patti Telegan for technical assistance. References 1. Snow DM, Watanabe M, Letourneau PC, Silver J. A chondroitin sulfate proteoglycan may influence the direction of retinal ganglion cell outgrowth. Development. 1991 ;113:1473-1485. 2. Johnson LV, Hageman GS, Blanks JC. Restricted extracellular matrix domains ensheath cone photoreceptors in vertebrate retinae. In: Bridges CD, Adler AJ, eds. The Interphotoreceptor Matrix in Health and Dis-
ease. New York: Alan R Liss, Inc; 1985:33-44. 3. Johnson LV, Hageman GS, Blanks JC: Interphotoreceptor matrix domains ensheath vertebrate cone photoreceptor cells. Invest Ophthalmol Vis Sci. 1986; 27:129-135. 4. Johnson LV and Hageman GS. Characterization of isolated cone matrix sheath substructure. Invest Ophthalmol Vis Sci. 1989;30(suppl):490. 5. Fariss RN, Anderson DH, Fisher SK. Comparison of photoreceptor-specific matrix domains in the cat and monkey retinas. Exp Eye Res. 1990;51:473-485. 6. Mieziewska K, van Veen T, Murray J, Aguirre G. Structure of rod and cone specific matrix domains in normal and progressive rod-cone degeneration (prcd) affected canine retina. Invest Ophthalmol Vis Sci. 1990;31(suppl):153. 7. Mieziewska KE, van Veen T, Murray JM, Aguirre GD.
IPM in prcd
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
Rod and cone specific domains in the interphotoreceptor matrix. J Comp Neurol. 1991;308:37l-380. Hollyfield JG, Rayborn ME, Landers RA, Myers KM. Insoluble interphotoreceptor matrix domains surround rod photoreceptors in the human retina. ExpEyeRes. 1990;51:107-110. Hageman GS, Johnson LV. Structure, composition and function of the retinal interphotoreceptor matrix. In: Osborne N, Chader G, eds. Progress in Retinal Research. London: Pergamon Press; 1991;10:207-249. Hollyfield JG, Vainer HH, Rayborn ME, Osterfeld AM. Retinal attachment to the pigment epithelium. Linkage through an extracellular sheath surrounding cone photoreceptors. Retina. 1989;9:59-68. Yao X-Y, Hageman GS, Marmor MF. Retinal adhesiveness is weakened by enzymatic modification of the interphotoreceptor matrix in vivo. Invest Ophlhalmol Vis Sci. 1990;31:2051-2058. Yao X-Y, Hageman GS, Marmor MF. Recovery of retinal adhesion after enzymatic perturbation of the interphotoreceptor matrix. Invest Ophthalmol Vis Sci. 1992;33:498-503. Lazarus HS, Hageman GS. Xyloside-induced perturbation of interphotoreceptor matrix proteoglycans result in retinal detachment. Invest Ophthalmol Vis Sci. 1992;33:364-376. Tawara A, Varner HH, Hollyfield JG. Proteoglycans in the mouse interphotoreceptor matrix II. origin and development of proteoglycans. Exp Eye Res. 1989;48:815-839. La Vail MM, Pinto LH, Yasumura D. The interphotoreceptor matrix in rats with inherited retinal dystrophy. Invest Ophthalmol Vis Sci. 1981;21:658-668. Porrello K, Yasumura D, La Vail MM. The interphotoreceptor matrix in RCS rats: histochemical analysis and correlation with the rate of retinal degeneration. ExpEyeRes. 1986;43:413-429. Porrello K, LaVail MM. Immunocytochemical localization of chondroitin sulfates in the interphotoreceptor matrix of the normal and dystrophic rat retina. Curr Eye Res. 1986;5:981-993. Uehara F, Yasumura D, LaVail MM. Lectin binding of the interphotoreceptor matrix during retinal development in normal and RCS rats. Curr Eye Res. 1990;9:687-695. Johnson LV, Blanks JC, Hageman GS. Effects of retinal degenerations on the cone matrix sheath. In: LaVail MM, Hollyfield JG, Anderson RE, eds. Inherited and Environmentally Induced Retinal Degenerations.
New York: Alan R Liss, Inc; 1989:217-232. 20. Long KO, Aguirre GD. The cone matrix sheath in the normal and diseased retina: cytochemical and biochemical studies of peanut agglutinin-binding proteins in cone and rod-cone degeneration. Exp Eye Res. 1991;52:699-713.
3067
21. Mieziewska KM, van Veen T, Aguirre GD. Development and fate of the interphotoreceptor matrix components during dysplastic photoreceptor differe ntiation. A lectin cytochemical study of rod-cone dysplasia 1. Exp Eye Res. 1993;56:429-441. 22. Aguirre GD, Acland GM. Variation in retinal degeneration phenotype inherited at the prcd locus. Exp Eye Res. 1988;46:663-687. 23. Johnson LV, Blanks JC. Application of acrylamide as an embedding medium in studies of lectin and antibody binding in the vertebrate retina. Cuir Eye Res. 1984;3:969-973. 24. Johnson LV, Hageman GS. Structural and compositional analyses of isolated cone matrix sheaths. Invest Ophthalmol Vis Sci. 1991 ;32:1951-1957. 25. Shuman H, Murray JM, DiLullo C: Confocal microscopy: an overview. BioTechniques. 1989;7:154-163. 26. Aguirre GD, Alligood J, O'Brien P, Buyukmihci N. Pathogenesis of progressive rod-cone degeneration in miniature poodles. Invest Ophthalmol Vis Sci. 1982;23:610-630. 27. Aguirre GD, O'Brien P. Morphological and biochemical studies of canine progressive rod-cone degeneration. Invest Ophthalmol Vis Sci. 1986;27:635-655. 28. Price J, Turner D, Cepko C. Lineage analysis in the vertebrate nervous system by retrovirus-mediated gene transfer. Proc Nat Acad Sci USA. 1987;84:156160. 29. Iwasaki M. Myers KM, Rayborn ME, Hollyfield JG. Interphotoreceptor matrix in the human retina: conelike domains surround a small population of rod photoreceptors. J Comp Neurol. 1992;319:277-284. 30. Hageman GS, Lazarus HS. An altered composition of rod photoreceptor-associated interphotoreceptor matrix is associated with macular degeneration. Invest Ophthalmol Vis Sci. 1992;33:803. 31. Aguirre GD, Lolley R, Farber D, Fletcher T, Chader G. Rod-cone dysplasia in Irish Setter dogs: a defect in cyclic GMP metabolism in visual cells. Science. 1978;201:1133-1134. 32. Aguirre G, Farber D, Lolley R, et al. Retinal Degenerations in the Dog: III Abnormal Cyclic Nucleotide Metabolism in Rod-Cone Dysplasia. Exp Eye Res. 1982;35:625-642. 33. Barbehenn E, Gagnon C, Noelker D, Aguirre G, Chader G. Inherited rod-cone dysplasia: abnormal distribution of cyclic GMP in visual cells of affected Irish Setters. Exp Eye Res. 1978;46:149-159. 34. Carter-Dawson LD, LaVail MM, Sidman R. Differential effect of the rd mutation on rods and cones in the mouse retina. Invest Ophthalmol Vis Sci. 1978; 17:489498. 35. Landers RA, Tawara A, Varner HH, Hollyfield JG. Proteoglycans in the mouse interphotoreceptor matrix. IV. Retinal synthesis of chondroitin sulfate proteoglycan. Exp Eye Res. 1991;52:65-74.
