infrastructure of Blood-Retinal Barrier Permeability in Rat Phototoxic Retinopathy

Transcription

1 infrastructure of Blood-Retinal Barrier Permeability in Rat Phototoxic Retinopathy Gary E. Korre, Roy W. Bellhorn, and Margaret 5. Burns It has been shown previously that the blood-retinal barrier (BRB) of rats with phototoxic retinopathy is permeable to sodium fluorescein and to fluoresceinated dextrans as large as 32A ESR (Einstein- Stokes radius). The leakage presumably occurs from retinal capillaries that have invaded the retinal pigment epithelium (RPE) and become fenestrated. In this report, the ultrastructural tracers horseradish peroxidase and catalase were used to further localize the leakage site, and to evaluate the size limit of molecules penetrating the phototoxic BRB. Horseradish peroxidase (HRP: 30A ESR) freely penetrates the BRB of phototoxic rats, since it is present in the retinal extracellular space 10 min after intravenous injection. HRP penetrates the fenestrae of capillaries which invade the RPE from the retina. It then diffuses along the pericapillary space of the intraepithelial capillaries, which is confluent with that of their parent retinal capillaries, and into the retinal extracellular space. HRP thus circumvents the tight junctions between RPE cells and between capillary endothelial cells, which appear intact in thin sections. Catalase (52A ESR) does not freely penetrate the BRB of phototoxic rats. As long as 40 min after intravenous injection, catalase is still confined to the lumen of fenestrated capillaries in the RPE, retinal capillaries, and the choriocapillaris. Although present in intraendothelial vesicles, no evidence of deposition in the pericapillary space is observed. It is concluded fenestrated capillaries in the RPE are a major site where blood-borne tracers penetrate the BRB in phototoxic retinopathy. Invest Ophthalmol Vis Sci 24: , 1983 Light-induced (phototoxic) retinopathy in rats models the blood-retinal barrier (BRB) in disease. 1 ' 2 Fluoresceinated dextrans as large as 32A ESR (Einstein-Stokes radius) penetrate the BRB of phototoxic rats, presumably at sites where retinal capillaries invade the retinal pigment epithelium (RPE) and become fenestrated. 1 " 4 These observations raised questions that a study using ultrastructural tracers could clarify: (1) What route do tracers take into the retina after leaking across the fenestrated capillaries in the RPE? (2) What contribution do opened tight junctions or vesicular transport across endothelial cells make to the leakage? (3) What is the size limit of molecules penetrating the phototoxic BRB? To answer these questions, we have administered intravenously ultrastructural tracers of two different sizes horseradish peroxidase (HRP) and catalase to rats with advanced phototoxic retinopathy. We From the Department of Ophthalmology, Albert Einstein College of Medicine/Montefiore Medical Center, Bronx, New York. Presented at the Annual Meeting Association for Research in Vision and Ophthalmology (ARVO) Sarasota, Florida, May Supported by National Eye Institute grant #EY to RWB and an unrestricted grant from Research to Prevent Blindness, Inc. Submitted for publication: August 3, Reprint requests: Gary E. Korte, PhD, Department of Ophthalmology, Montefiore Medical Center, 111 East 210th Street, Bronx, NY then determined the tracer's behavior in relation to fenestrated capillaries in the RPE and to known components of the normal BRB tight junctions between RPE cells and capillary endothelial cells, and intraendothelial vesicles. Our observations indicate that leakage across fenestrated capillaries in the RPE and subsequent diffusion into the retina via their pericapillary space is the major mechanism of barrier breakdown in phototoxic rats. Tight junctions between RPE cells and capillary endothelial cells remain intact, and transendothelial vesicular transport of tracers was not observed. The observations extend those made using fluorescent markers. 3 ' 4 Materials and Methods Tracer studies were undertaken in 26 albino rats: eight control and five phototoxic rats each for studies on horseradish peroxidase and catalase. Phototoxic retinopathy was produced by.exposure to fluorescent light, as previously described. 2 Observations were made 9-12 months after light challenge, when the retinopathy is advanced (Figs. 1 A, B). Rats were anesthetized intraperitoneally with sodium pentobarbital (40 mg/kg body weight) and ketamine hydrochloride (15 mg/kg body weight) during experiments, and killed afterwards by an overdose of sodium pentobarbital /83/0700/962/$ 1.30 Association for Research in Vision and Ophthalmology 962

2 No. 7 ULTRASTRUCTURE OF BLOOD-RETINAL DARKER / Korre er al. 963 B Fig. 1. Light microscopic histology of the retina of phototoxic rats, 10 minutes after intravenous injection of HRP. A, Retinal capillaries (arrow) enter the RPE. The capillary profiles appear circular due to perfusion fixation. Their black outline is due to HRP reaction product. Note apposition of RPE and inner nuclear layer (INL) of the retina, due to photoreceptor loss. Arrowheads, choriocapillaris apposed to Bruch's membrane (3 titn plastic section stained with toluidine blue) (XI25). B, Higher magnification micrograph of retinal capillaries (arrows) entering the RPE. The capillaries appear collapsed due to immersion fixation and are black due to HRP reaction product. Arrowhead, leakage of HRP into the inner nuclear layer of the retina. Bruch's membrane overlies the RPE and is stained black due to HRP reaction product (3 ^im plastic section, unstained) (X800). Horseradish peroxidase (Sigma, Type II; U/mg protein; mg/g body weight in ml saline) or catalase (Sigma, Type C-100; 30,000-40,000 U/mg protein; 1-2 ml of an aqueous suspension) was injected into a cannulated femoral vein. Ten or 40 min later the eyes were enucleated. The cornea, lens, and vitreous were removed and the posterior eye cup immersed for 3-4 hrs in 2% paraformaldehyde and 2% glutaraldehyde in 0.1 M phosphate buffer, ph 7.2. Some animals were perfused through the heart with fixative prior to immersing the eyes in it. The eyecups were sliced into narrow wedges radiating about the optic disc. The wedges were rinsed overnight at 4 C in phosphate buffer and then incubated at room temperature in a solution of diaminobenzidine and hydrogen peroxide for localizing horseradish peroxidase 5 and catalase. 6 ' 7 Control wedges were incubated in the absence of diaminobenzidine or hydrogen peroxide. The wedges were osmicated in 2% osmium tetroxide in 0.1 M phosphate buffer, dehydrated in graded concentrations of methanol and embedded in epoxy resin. Two-micron thick sections were cut midway between the optic disc and ora serrata. This provided

3 964 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / July 1983 Vol. 24 Fig. 2. Thin-sectioned capillary in the pigment epithelium of a phototoxic rat which received an intravenous injection of HRP 10 minutes prior to killing. A, Black reaction product fills the lumena (L) of the intraepithelial capillary and the choriocapillaris, their pencapillary spaces, and Bruch's membrane (BM). Focal continuities (white arrowheads) between the intraepithelial capillary's pericapillary space and Bruch's membrane make it possible for HRP leaking across the choriocapillaris to directly enter the pericapillary space. RPE cells abut the capillary, on which they form basal folds. Area at blunt arrow is enlarged in Figure 2B. V, apical villi of RPE cells (X75OO). B, Reaction product appears continuous across endothelial fenestra (arrowheads) of the intraepithelial capillary seen in Figure 2A. Intraendothelial vesicles (arrows) are also filled with reaction product. L, capillary lumen (X2O.3OO). a survey section containing the wedge's cut surfaces, where reaction product is most dense. These regions could then be trimmed for thin sectioning at known distances from the razor-cut surface. Thin sections were stained with lead citrate 8 and uranyl acetate and examined in a Zeiss EM-9S-2 electron microscope. Tracer sizes are those accepted by other investigators 910 : horseradish peroxidase = 30A ESR and catalase = 52A ESR. Tissue from one normal and one phototoxic rat was stained en-bloc with tannic acid (1% in the aldehyde fixative) or uranyl acetate (0.5% in saline, after osmication) to help visualize intercellular junctions. Results Ten minutes after intravenous injection, horseradish peroxidase reaction product is observed in the pericapillary space of capillaries in the RPE and crossing between the retina and RPE (Figs. 1-3). Numerous examples of tracer-filled pericapillary spaces continuous with tracer-filled retinal extracellular spaces were observed where capillaries cross the retina-rpe boundary and in the subjacent retina (remnant outer plexiform or inner nuclear layers: Figs. 3, 4A). Tracer was not observed in the pericapillary space of capillaries in the inner plexiform layer even after 40 min circulation time. However, reaction product was constantly present in their lumen and some intraendothelial vesicles (Fig. 4B). Horseradish peroxidase seems to enter the pericapillary space of the intraepithelial capillaries by penetrating their fenestra (Fig. 2A). Bellhom et al 2 have described these fenestra and their occurrence in many profiles of intraepithelial capillaries. We ob-

4 No. 7 UURA5TRUCTURE OF DLOOD-RETINAL BARRIER / Korre er QI. 965 A Fig. 3. A, Capillary (L, lumen) bridging the RPE and retina (R), from a phototoxic rat that received an intravenous injection of HRP 10 min prior to killing. Reaction product occurs in the pericapillary space spanning the RPE-retina boundary (arrows), and is continuous with reaction product in the retinal extracellular space, seen in the area indicated by a blunt arrow in Figures 3B and C. Curved arrow, RPE process extending along capillary (X525O). B, C, Area denoted by the blunt arrow in Fig 3A, from the same section (B) and a nearby area in an adjacent section (C). Blunt arrows denote where reaction product in the pericapillary space is continuous with reaction product in the retinal extracellular space (X46,000). served HRP reaction product in the pericapillary space of both fenestrated and nonfenestrated capillary profiles in the RPE. This could arise in two ways: (1) Passage across fenestra outside the plane of section; (2) By diffusion across Bruch's membrane from the choriocapillaris at sites where Bruch's membrane

5 966 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / July 1983 Vol. 24 B Fig. 4. Portions of retinal capillaries in the remnant outer plexiform layer (A) and the inner plexiform layer (B) of a phototoxic rat that received an intravenous injection of HRP 10 min prior to killing. Reaction product is present in the lumen (L) and some intraendothelial vesicles (arrowheads) of both capillaries, but in the pericapillary space (arrows) only of the outermost retinal capillary, ie, that nearest the leaky intraepithelial capillaries. Note tracer in retinal extracellular space in A, (A: XI 8,400; B: X41,200). and the pericapillary space of the intraepithelial capillaries are continuous focally (Figs. 2A, 5). This provides a route by which HRP may diffuse along the pericapillary space of intraepithelial capillaries and into the retina (Fig. 3). Numerous HRP-laden vesicles occur in the endothelial cells of capillaries in the RPE and adjacent retina. They cannot be interpreted as evidence for vesicular transport from the capillary lumen, as their contents could be HRP that was taken up after en-

6 No. 7 ULTRA5TRUCTURE OF DLOOD-RETINAL CARRIER / Korre er ol. 967 t I BM RPE Fig. 5. Capillary in the RPE of a phototoxic rat which received an intravenous injection of catalase 40 min prior to killing. Black catalase reaction product is confined to the lumen (L), the enlarged pericapillary space remaining unstained. Bruch's membrane (BM) is also unstained, since catalase does not penetrate the choriocapillaris. 10 Arrows: continuity between Bruch's membrane and the pericapillary space. Arrowheads: endothelial fenestra (as identified at higher magnification). V, apical villi of RPE cell surrounding the capillary (X15,600). tering the pericapillary space via fenestra of the capillaries in the RPE or the choriocapillaris (Fig. 2A). This ambiguity is addressed in experiments using catalase. Catalase is retained in the lumen of capillaries in the RPE, the retina, and the choriocapillaris even after 40 min circulation time (Figs. 5, 6). This shows that the fenestra of the intraepithelial capillaries do not pass catalase and suggests that vesicular transport across endothelial cells does not contribute to the deposition of tracer in the retinal extracellular space, even though some intraendothelial vesicles are filled. Tight junctions between RPE cells and among capillary endothelial cells appear intact in thin sections (Figs. 7A-C). Even where RPE cells abut on intraepithelial capillaries, they retain apparently well-developed junctional complexes with their neighboring RPE cells. Numerous examples of HRP present only on the choroidal side of the junctional complex were observed. HRP was observed on the retinal side of RPE and endothelial tight junctions only near sites where retinal capillaries entered the RPE, ie, where HRP could approach both sides of the tight junction due to leakage out of intraepithelial capillaries. The observations using catalase also suggest intercellular junctions remain intact; catalase was constantly arrested on the luminal side of capillary interendothelial junctions in the retina and in the RPE.

7 968 INVESTIGATIVE OPHTHALMOLOGY & VI5UAL SCIENCE / July 1983 Vol. 24 Fig. 6. Capillary in the remnant outer plexiform layer of the retina of a phototoxic rat which received an intravenous injection of catalase 40 min prior to killing. Reaction product is confined to the capillary lumen (L) and some intraendothelial vesicles (arrowheads). No evidence of deposition is seen in the pericapillary space (arrows), whose density equals that of control slices of tissue (X18,400), Discussion The major leakage site in the BRB of rats with phototoxic retinopathy occurs at capillaries that invade the RPE from the retina (Fig. 8). The pericapillary space of these intraepithelial capillaries receive bloodborne tracers via: (a) the fenestra of their endothelial cells, or (b) the fenestra of the choriocapillaris, followed by diffusion across Bruch's membrane and into their pericapillary space. Once tracer is in the pericapillary space of intraepithelial capillaries, it can diffuse into the retina along the pericapillary space of capillary segments bridging the RPE-retina boundary. Thus, the permeability of endothelial fenestra and any hindrance to diffusion provided by the basement membrane in the pericapillary space (as by coulombic charge or molecular sieving 910 ) would be major determinants of the permeability of the BRB in phototoxic rats. The relative contribution of the choriocapillaris and the intraepithelial capillaries to leakage into the retina would depend on their number of fenestra and how extensive is the continuity between Bruch's membrane and the pericapillary space of intraepithelial capillaries. The possibility exists that HRP passage across intraepithelial capillary fenestra is only apparent, due to HRP from the choriocapillaris "refluxing" up against them. Tight junctions between RPE cells and capillary endothelial cells remain intact in rats with phototoxic retinopathy. They appear similar to those described in normal rats. 1 ' However, tight junctions that appear intact in thin sections may also leak tracers Thus, a tracer smaller than HRP, such as microperoxidase (10A ESR 14 ) may leak across apparently intact tight junctions, even though HRP does not. Such a study would further verify the integrity of these junctions and determine if the mechanism of BRB breakdown described above is the only one operating in phototoxic retinopathy. This is an important question as tight junctions between RPE cells or endothelial cells may open in some other conditions, eg, dog diabetic retinopathy, after lens extractions in monkeys, or during rat retinal dystrophy. 15 " 17 Our observations using catalase indicate that vesicular transport across endothelial cells does not contribute to the barrier breakdown in rats with phototoxic retinopathy. As long as 40 min after intravenous injection catalase remained in the lumen and some intraendothelial vesicles of capillaries in the RPE, retina, and choroid. No evidence of catalase deposi-

8 No. 7 ULTRASTWJCTURE OF BLOOD-RETINAL BARRIER / Korre er ol. 969 V Fig, 7. Examples of junctional complexes connecting RPE cells of phototoxic rats. The cells in Figures 7A and C abut on intraepithelial capillaries; the cells in Figure 7B are from a stretch of RPE devoid of intraepithelial capillaries. A, B, In a rat which received HRP 10 min prior to killing, reaction product is restricted to the choroidal side of the junctional complex (JC), the extracellular space on the retinal side of the junctional complex being free of tracer (arrowheads). V, apical villi of RPE (X52,4OO). C, In tissue stained en bloc with tannic acid, the several types of intercellular junctions comprising the junctional complex are resolved. Membrane fusions corresponding to tight junctions (arrowheads) are interspersed along an extensive zonula adherentes. Arrow, gap junction. V, apical villi of RPE cells. N, nucleus of RPE cell (X47,500).

9 970 INVESTIGATIVE OPHTHALMOLOGY b VISUAL SCIENCE / July 1983 Vol. 24 RETINA Bruch's membrane Fig. 8. Diagram illustrating how blood-borne tracers can leak across the BRB of phototoxic rats. Intravenously injected tracer reaches the choriocapillaris (small arrows in lumen of top capillary) and intraepithelial capillaries in the RPE (small arrows in lumen of bottom capillary), which arise from retinal capillaries. Tracer in the intraepithelial capillaries penetrates their fenestra and enters the pericapillary space (black). This leads to the space between RPE cells (sealed off by tight junctions) or directly into the retina (large arrows). Tracer can also reach the pericapillary space of intraepithelial capillaries by passing across the fenestrated choriocapillaris (top capillary) and crossing Bruch's membrane, which is focally continuous with the pericapillary space of the intraepithelial capillaries. tion in the pericapillary space was observed, such as vesicles releasing their contents at the abluminal plasma membrane. HRP-laden intraendothelial vesicles in capillaries in the inner nuclear layer may represent tracer picked up at (a) the capillary lumen, (b) the pericapillary space after leaking across the RPE or (c) HRP being transported across endothelial cells by vesicles. Our catalase observations cast doubt on, though do not disprove, the latter possibility; HRP could be transported by vesicles even though catalase is not. The absence of HRP in the pericapillary spaces of the layers vitread to the inner nuclear layer as long as 40 min after intravenous injection suggests that these capillaries are not leaking due to either vesicular transport or opened tight junctions. Essner et al 18 came to similar conclusions in their study of dystrophic rats. As intraepithelial capillaries were not observed in the animals they used, the leakage of retinal capillaries closer to the RPE was presumed due to vesicular transport and/or opened interendothelial tight junctions, and not leakage across the RPE itself. HRP probably was not observed in the pericapillary space of capillaries in the inner plexiform and more vitread layers of phototoxic rats for several reasons: (a) these capillaries remain impermeable to HRP. (b) Longer than 40 min (our longest circulation time) is needed for HRP to diffuse across the retina from its leakage site in the RPE. The HRP may diffuse more slowly along the pericapillary space as it dilutes out in the vasculature (the source of this HRP). (c) The HRP may build up in the pericapillary space, as by electrostatic binding to basement membrane components, and clog it. Thus, the tracer may form its own impediment to diffusion deeper into the retina, (d) The HRP may be present in too low a concentration for us to detect. The mechanisms outlined in Figure 8 may circumvent the BRB in other conditions in which retinal capillaries invade the RPE. In rats with urethan retinopathy, in which retinal capillaries also invade the RPE, 19 we have found that HRP leaks into the retina by the same route as described in our phototoxic rats (unpublished observation: GK). In rats with hereditary retinal dystrophy, intravenously administered retinol-binding protein leaks into the retina where retinal capillaries invade the RPE (Fig. 5 from Reference 20). Conditions in which choroidal capillaries penetrate the RPE and enter the retina would provide a similar route by which blood borne molecules could leak into the retina: leakage out of the fenestrated choriocapillaris and diffusion along the pericapillary space of choroidal capillaries entering the retina. 21 " 23 Since both fenestrated and nonfenestrated capillaries arising from the choroid have been observed in the RPE and subretinal space in human senile macular

10 No. 7 ULTRASTRUCTURE OF BLOOD-RETINAL BARRIER / Korre er ol. 971 degeneration, 2425 our observations in rats with phototoxic or urethan retinopathy may help us understand the mechanisms of abnormal permeability during the exudative phase of senile macular degeneration. Key words: retina, blood-retinal barrier, permeability, ultrastructure, pathology, rat, phototoxic retinopathy, vasculopathy Acknowledgments The authors wish to acknowledge the excellent technical assistance of Judith Channer and Noel Roa, and the secretarial assistance of Patricia Lynch. References 1. Rabkin MD, Bellhorn MB, and Bellhorn RW: Selected molecular weight dextrans for in vivo permeability studies of rat retinal vascular disease. Exp Eye Res 24:607, Bellhorn RW, Burns MS, and Benjamin JV: Retinal vessel abnormalities of phototoxic retinopathy in rats. Invest Ophthalmol Vis Sci 19:584, Bellhorn RW: Blood-retinal barrier abnormalities in rat phototoxic retinopathy. ARVO Abstracts. Invest Ophthalmol Vis Sci 22(Suppl):63, Bellhorn R and Korte G: Permeability of the abnormal bloodretinal barriers in rat phototoxic retinopathy. A clinicopathologic correlation study using fluorescent markers. Invest Ophthalmol Vis Sci 24:972, Graham RC Jr, Karnovsky MJ: The early stages of absorption of injected horseradish peroxidase in the proximal tubules of mouse kidney; ultrastructural cytochemistry by a new technique. J Histochem Cytochem 14:291, Venkatachalam MA and Fahimi HD: The use of beef liver catalase as a protein tracer for electron microscopy. J Cell Biol 42:480, Novikoff AB and Goldfischer S: Visualization of peroxisomes (microbodies) and mitochondria with diaminobenzidine. J Histochem Cytochem 17:675, Reynolds ES: The use of lead citrate at high ph as an electronopaque stain in electron microscopy. J Cell Biol 17:208, Caulfield JP and Farquhar MG: The permeability of glomerular capillaries to graded dextrans. Identification of the basement membrane as the primary filtration barrier. J Cell Biol 63:883, Pino RM and Essner E: Permeability of rat choriocapillaris to hemeproteins. Restriction of tracers by a fenestrated endothelium. J Histochem Cytochem 29:281, Hudspeth AJ and Yee AG: The intercellular junctional complexes of retinal pigment epithelia. Invest Ophthalmol 12:354, Brightman MW, Hori M, Rapoport SI, Reese TS, and Westergaard E: Osmotic opening of tight junctions in cerebral endothelium. J Comp Neurol 152:317, Martinez-Palomo A and Erlij D: Structure of tight junctions in epithelia with different permeability. Proc Natl Acad Sci USA 72:4487, Smith RS and Rudt LA: Ocular vascular and epithelial barriers to microperoxidase. Invest Ophthalmol 14:556, Wallow IHL and Engerman RL: Permeability and patency of retinal blood vessels in experimental diabetes. Invest Ophthal Vis Sci 16:447, Tso MOM: Pathology of the blood-retinal barrier. In The Blood-Retinal Barriers, Cunha-Vaz J, editor. New York, Plenum Publishing, 1980, pp Caldwell RB, McLaughlin BJ, and Boykins LG: Intramembrane changes in retinal pigment epithelial cell junctions of the dystrophic rat retina. Invest Ophthalmol Vis Sci 23:305, Essner E, Pino RM, and Griewski RA: Permeability of retinal capillaries in rats with inherited retinal degeneration. Invest Ophthalmol Vis Sci 18:859, Bellhorn RW, Bellhorn M, Friedman AH, and Henkind P: Urethan-induced retinopathy in pigmented rats. Invest Ophthalmol 12:65, Bok D and Heller J: Autoradiographic localization of serum retinol-binding protein receptors on the pigment epithelium of dystrophic rat retinas. Invest Ophthalmol Vis Sci 19:1405, Baum JL and Wise GN: Experimental subretinal neovascularization. Am J Ophthalmol 61:528, Deutman AF: Significance of alteration of the outer bloodretinal barrier. In The Blood-Retinal Barriers, Cunha-Vaz J, editor. New York, Plenum Publishing, 1980, pp Archer DB, and Gardiner TA: Experimental subretinal neovascularization. Trans Ophthalmol Soc UK 100:363, Grindle CFJ and Marshall J: Aging changes in Bruch's membrane and their functional implications. Trans Ophthalmol Soc UK 98:172, Sarks SH, Van Driel D, Maxwell L, and Killingsworth M: Softening of drusen and subretinal neovascularization. Trans Ophthalmol Soc UK 100:414, 1980.