Documenta Ophthalmologica 41,2:287-327, 1976 THE BLOOD-RETINAL BARRIERS J. G. C U N H A - V A Z *

(Co im b ra ) ABSTRACT The Blood-Retinal Barrier (BRB) is a situation of restricted permeability which is present between the blood and the retina. This barrier has a well defined anatomic substrate, particular permeability characteristics and appears to play a role of major importance in the pathophysiology and therapeutics of retinal disease. The BRB phenomenon operates fundamentally at two levels, retinal vessels and chorioepithelial interface, forming which may be better called an inner BRB and an outer BRB. The main structures involved are, for the inner BRB, the endothelial membrane of the retinal vessels, and for the outer BRB, the retinal pigment epithelium. 'Zonulae occludentes' are present in these membranes, forming complete belts around the cells, sealing off the spaces between them. Other structures appear to play an accessory role. Both barriers show an apparent predominance of processes of active transport over mechanisms of passive transfer, these being extremely restricted. Much information on the pathophysiology of the BRB mechanism has been obtained from studies of its experimental breakdown. In this way, a breakdown of the inner BRB may be induced by acute distension of the vessel walls, ischaemia, chemical influences, defects in the endothelial cells and failure of the active transport system, whereas experimental ischaemia, mechanical distension of the pigment epithelial membrane, defects in the pigment epithelium and failure of the active transport systems can cause a breakdown of the outer BRB. The increased permeability of the inner BRB, and of the outer BRB, appears to be related to changes in the vascular endothelial membrane and retinal pigment epithelium, respectively. In clinical ophthalmology there are two methods for the diagnosis of breakdown of the BRB, fundus fluorescein angiography and vitreous fluorophotometry. Vitreous fluorophotometry being capable of detecting functional alterations of the barrier before any pathological changes are apparent. There is evidence of an intimate relationship between breakdown of the BRB and almost every retinal disease, particularly the vascular retinopathies and the pigment epitheliopathies. Diabetic retinopathy, hypertensive retinopathy, retinal vein obstruction, blood diseases, trauma or surgery to the eye, temporary arterial obstruction, perivasculitis, Beh~et's and Coats' diseases, retinoblastoma, hemangioblastoma and retinal neovascutarization are examples of situations where a breakdown of the inner BRB has been demonstrated. On the other hand, examples of breakdown of the outer BRB include situations of choroidal ischaemia, detachment of the pigment epithelium, choroidal neovascularization, photocoagulation, retinal detachment, Koyanagi's disease, central serous choroidopathy, multifocal inner choroiditis and acute placoid pigment epitheliopathy.

* Department of Ophthalmology, University ofCoimbra, Coimbra, Portugal. This work was supported by research grant CMC 8 from the Instituto de Alta Cultura, Portugal.

287

The permeability of the blood vessels in the body in general shows that the passage of most substances across the blood-tissue interface is somewhat restricted, differing from a situation of free diffusion. This restriction is, however, much more pronounced in the central nervous system and retina than in other parts of the body, giving rise to the concepts of Blood-Brain Barrier (BBB) and Biood-Retinal Barrier (BRB). The concept of the BBB, which was the first to appear in the literature, through the work of Paul Ehrlich in 1885, had its real origin in the classical experiments of Goldman (1913), which used trypan blue for the first time as an indicator of barrier function. In his first experiment, the author injected trypan blue intravenously and observed an intense vital staining of all tissues of the animal with the sole exception of the brain. No toxic symptoms resulted from this experimental situation. In a second experiment, the introduction of a small quantity of the same dye directly into the subarachnoid space coloured the brain deep blue and the animals died in a few minutes after convulsions and final paralysis of the central nervous system. Intensive studies on the blood-brain relationship have yielded evidence that for many substances the rates of exchange and the concentrations in the brain at steady state are quite different from those in other organs, These differences reflect a distinct permeability at the blood-brain interface. The concept of BBB has evolved on several fronts: morphological, physicochemical, biochemical and physiological. Many hypotheses have been proposed to explain the barrier phenomena, but none has been universally accepted. Because the BBB holds a key to our understanding of the cerebral metabolism, function, pathology and therapy, intensive research aimed at clarifying the barrier mechanisms has been carried on in various directions. In the subject of the BBB, Lee (1971)reports that in the 85 years after the first publication on the BBB phenomena by Ehrlich in 1885, more than 2000 articles related to various aspects of the barrier problem have been published, and about half of them appeared in the last two decades. As regards the retina, only two studies appeared in 1913 and 1947, pubfished by Schnaudigel (1913) and Palm (1947), repeating the classical work of Goldman in the brain. It is noteworthy that these two studies did not make any impact on the ophthalmic literature, and up to 1965, every ophthalmic textbook considered the permeability of the retinal vessels as comparable to that of the other vessels of the body (Bailliart, 1953; Duke-Elder, 1958; Adler, 1962). In 1965, the effect of histamine on the permeability of the ocular vessels was examined closely by Ashton and Cunha-Vaz, using the technique of vascular labelling of Majno and Palade (1961). Histamine increased markedly the vascular permeability of the various ocular tissues with the sole excep-

288

tion of the retina, the retinal vessels showing no evidence of vascular labelling in the experimental conditions used. This peculiar behaviour of the retinal vessels was in clear contrast with almost every other vessels of the body, and similar only to the cerebral vessels, a finding which pointed to the existence in the retina of a barrier situation similar to the BBB. Subsequent research examining the penetration of a variety of substances into the retina and vitreous, confirmed the existence at the level of the retina of a BRB, comparable to the BBB. The introduction of the concept of the BRB into the ophthalmic literature excited much interest, well shown by the publication of a series of short reports studying its basic mechanisms, this interest being supported by the development and widespread use of fundus fluorescein angiography, a clinical method which has much contributed to prove the major importance of the BRB in retinal disease. The BRB is now widely accepted as one of the most important factors in the development of retinal vascular disease and macular pathology, two of the most significative causes of blindness. It is important, therefore, to define clearly the concept of the BRB, distinguishing it from other situations of restricted permeability existent in the eye, but which do not fulfill the criteria of a barrier system, or are too encompassing to have a clear meaning, as is the case with the proposed blood-vitreous barrier. A review on the BRB phenomena will be attempted here, in an effort to clarify the concept of a unique blood-retina relationship and outline its role in the physiopathological mechanisms of retinal disease.

THE BLOOD-RETINAL BARRIER AND THE OTHER OCULAR 'BARRIERS'

There are a number of situations of restricted permeability in the ocular tissues, particularly between the ocular humours and the blood, protecting them from the multiple variations to which the blood is constantly subjected. A different situation is immediately apparent regarding the aqueous humour and the vitreous humour, and, consequently, as regards the posterior segment and the anterior segment of the eye. Whereas the aqueous humour is under constant renewal and changes in its composition can occur without much c o n s e q u e n c e s to the eye, the vitreous humour is rather static, and changes in its contents have ominous prospects as regards normal function and vision. In spite of this, there has been much more work done on the Blood-Aqueous Barrier, and a clear picture of the relationship between the 289

blood and the anterior segment of the eye is patent from the large number of reviews on the subject (Davson, 1969). As regards the posterior segment of the eye such a situation is, however, far from being attained, and, as previously stated, even a barrier mechanism has only recently been accepted. It is important, now, to review briefly the first physiological studies which called attention to a peculiar situation of restricted permeability at the level of the posterior segment, and the subsequent attemps to locate morphologically this restricted permeability, in other words, the emergence of the concept of the BRB and the basic reasons for its clear distinction from other situations of restricted permeability present in the posterior segment of the eye. As early as 1948, Davson and co-workers studied the movements of a series of substances into the posterior segment of the eye, and concluded that there existed at that level a marked restriction for their penetration from the blood. In their studies the vitreous was divided in three portions and they were examined separately. They verified that the highest concentrations were present in the anterior region of the vitreous, in contrast with the posterior region, where constantly inferior values were obtained. The authors postulated a Blood-Vitreous Barrier but did not present any evidence in favour of a particular location of this restricted permeability, speculating only on the possible responsibility of the retinal vessels. The concentration gradient seen in the vitreous in these penetration studies was attributed to diffusion from the posterior chamber, and penetration through the region of the ciliary body, a fact which had been subsequently confirmed and extended by a number of studies (Kinsey, 1960; Cunha-Vaz & Maurice, 1967). Studies on the penetration of a variety of drugs into the interior of the eye, particularly antibiotics, have amply confirmed that the penetration of drugs into the eye is much more restricted in the vitreous than in the aqueous humour, in the order of ten times. Finally, morphological studies using light and electron microscopy and a variety of tracers, including dyes like trypan blue, which has been widely used for the study of the brain barrier, has located the barrier between the blood and the interior of the posterior segment, at the level of the retina, originating the term BRB. The concept of a distinct Blood-Optic Nerve Barrier does not appear to be correct, as there is evidence for an exceptional situation of absence of barrier at the optic nerve head (Tso et al., 1975), the situation being exactly similar to that prevailing in the central nervous system in the remaining optic nerve. Using horseradish peroxidase as a tracer for electron microscopy and the normal rhesus monkey as the experimental animal, the above quoted authors have demonstrated that in certain 290

regions of the optic nerve head, peroxidase from the blood stream reaches the axons of the optic nerve through the border tissue of Elschnig from the adjacent choroidal tissues. The concept of a Blood-Vitreous Barrier, on the other hand, is very vague, without a clear morphological basis, and is on the whole untenable, as such a barrier is entirely absent in the anterior region of the vitreous, where a situation of free diffusion is present between the anterior and posterior segments of the eye, the posterior chamber being dependent of the ciliary body circulation. In summary, in the posterior segment of the eye, there is only at the level of the retina a well-defined barrier situation, the BRB,. resting on firm morphological grounds and having definite physiological characteristics. This BRB, appears, indeed, to have many similarities with the BBB, having, like the latter, a protective function to defend the retina and vitreous from the entry of noxious agents and from any drastic changes in the body, and a homeostatic control that conditions the physiology of the retina and maintains its normal activity.

GENERAL CONSIDERATIONS The basic component of the barrier concept is a morphological one. A brief analysis of this important aspect of the barrier involves the review of its general structure, the o c c u r r e n c e of the barrier in various animal classes and the embryonic development of the barrier activity. 1. General s t r u c t u r e

Of all the ocular issues the retina is, of course, the most important, as it contains the visual cells. It appears as a multilayered membrane of neuroectodermal origin, occupying the internal aspect of the posterior portion of the eyewall. The neural elements of the retina, an island of central nervous system, are separated from the blood at two levels. An outer level, where the pigment epithelium covers the entire interface between the retinal nervous tissue and the vascular layers of the choroid and an inner level, where the retinal vessels separate the retinal tissue from the blood. As regards the inner level of the barrier, the structure and pattern of the retinal blood vessels present marked differences from those in other organs. In man, the major retinal vessels lie in the innermost retinal layers and can usually be found just beneath the internal limiting membrane, although in some mammals they can be seen overlying the retina, in the vitreous. Only mammals have intraretinal capillaries but a great variation is registered between them, 291

four main types being accepted: holangiotic, merangiotic, paurangiotic and anagiotic. In the holangiotic retinae only two areas lack capillaries: the far periphery, just posterior to the ora serrata and the foveal region, when present. The ultrastructure of the retinal blood capillaries is uniform throughout the retina and consists of a continuous endothelium surrounded by a thick basement membrane. The endothelium does not show signs of fenestration and present tight junctions in a constant manner, apparently surrounding every endothelial cell (Cunha-Vaz et al., 1966). In the developing retinal vessels, every stage of structural formation may be observed. In this way, the basement membrane may be totally absent, or the vessels may be observed free in the vitreous without surrounding glial tissue, as in the rabbit. A t the outer level of the barrier the retinal pigment epithelium is the main cellular structure. The pigment epithelium develops from the outer wall of the optic vesticle being, therefore, the retinal layer, which lies between the sentient rods and cones and the nutrient vessels of the choroid. The retinal pigment epithelium appears as a uniform and continuous layer, extending through the entire retina, tight junctions being present between every pigment epithelial cell. The structural characteristics of the pigment epithelial cell indicate the multiplicity and importance of its functions. There are numerous mitrochondria and the cytoplasm has an unusual amount of smooth-surfaced endoplasmic reticulum, rough-surfaced endoplasmic reticulum, free ribosomes, the Golgi apparatus being well developed. The bearing of these structural characteristics on the barrier function will be discussed in connection with its anatomic substrate. 2. Phylogenetic occurrence

Within the vertebrate phylum the eye shows no progress of increasing differentiation and perfection as is seen in the brain, heart and most other organs. In its essentials the eye of a fish is as complex and fully developed as that of a bird or a man (Duke-Elder, 1958). All vertebrates have a neuronal three-layered retina and pigmentary epithelium and all have the same nutrient mechanism. If the function of the BRB is to control and maintain the homeostasis of the retina it should, therefore, be present in allvertebrate animals. The studies published on this subject confirmed this assertion and a BRB has been found in every animal examined, although a study with such a definite objective in mind is yet to be published, and only a few species have been examined (Cunha-Vaz, 1966). In the brain, there is some evidence that the penetration of different substances, like monosaccharides, chloride and sucrose, vary greatly between any

292

two species, such a s mouse and rat, cat and monkeys, man and whale (Lajtha, 1964; Tower, 1967). The reason for such differences has been attributed to enzymatic variations found in the endothelial cells of the brain of the different animals examined. Research along these lines is still lacking in the retina.

3. Ontogenetic development In all vertebrates the retina participates in the high degree of differentiation which characterizes the central nervous system. The proximal wall of the optic cup remains as a unicellular layer and acquires pigment to form the pigmentary epithelium. This structure is characterized by an early development, which is well in accordance with the early presence of the barrier phenomenon in kittens, ratlings and young rabbits (Cunha-Vaz, 1967). As regards the retinal vessels, their development is considered, nowadays, to be similar to vasculogenesis in general. It is accepted that it begins with a widespread development of vascular primordia of mesenchymal cells that become orientated into columns, gradually canalize into endothelial ceils and finally interconnect to form the earliest embryonic capillary circulation (Ashton, 1970). The barrier to trypan blue and carbon particles has been found to be present at a very early stage in retinal vessels of embryonic type (kitten, ratling and young rabbit), and it will be recalled that a number of authors demonstrated a barrier to trypan blue in developing cerebral vessels (Broman, 1949; Grazer & Clemente, 1957). Important information on the barrier phenomena was unveiled when the developing vessels of kittens and young rabbits were compared ultrastrucrurally (Cunha-Vaz et al., 1966). In both these immature animals the permeability of the retinal vessels is similar to that of the adult vessels, for the substances examined, but the vascular structure shows marked differences, less elements being present in the immature retinal vessels. In the kitten, the retinal vessels lie in a compact glial tissue with thick endothelium presenting well-defined attachment structures, but the basement membrane is tenuous and incomplete. The basement membrane cannot, therefore, be considered as the site for the inner BRB, at least for the substances examined, i.e., carbon particles, trypan blue and fluorescein. Similarly, in the young rabbit, the retinal vessels show a different structure, appearing as simple tubes of endothelial cells with dense and constant junctional complexes but entirely devoid of basement membrane or perivascular glia. These ultrastructural studies are particularly conclusive showing that only two structural elements remain as possible anatomical sites for the 293

Fig. 1. Retinal capillary of a 9 day old rabbit. The vessel is free in the vitreous and shows little or no basement membrane material. Vitreous filaments (V.F.) can be seen in contrast with the endothelial cells (E). Between the thick continuous endothelial cells there are well-defined junctional complexes (J.C.L I.L.M. - internal limiting membrane of the retina. Uranyl acetate followed by lead citrate, x 15.000.

inner BRB for trypan blue and fluorescein, namely, the endothelial cells and the attachments between them (Fig. 1). The ultrastructural features of the capillaries in the i m m a t u r e retina which can be correlated with the developing inner BRB include the very early appearance of tight junctions ('zonulae occludentes'), the emergence of pericytes, increasing thickness of the basement m e m b r a n e with age, and progressive increase in density of the glial tissue with decrease in the extracellular space. There is evidence in the brain, mainly obtained from studies with radioactive isotopes, that the vascular permeability is generally less restrictive in the i m m a t u r e than in the mature brain (Bakay, 1956). A l t h o u g h similar i n f o r m a t i o n is still n o t available in the retina, ultrastructural studies using tracers have shown a m u c h more marked p i n o c y t o t i c activity in the vessels o f the i m m a t u r e retina. In a personal electron microscopical study, the p e n e t r a t i o n of thorium dioxide and saccharated iron oxide in mature and

294

immature retinal vessels was examined at similar time intervals (Cunha-Vaz, 1966). In the mature retinal vessels the endothelial cells remained in most vessels free of tracer particles, whereas in the immature vessels the number of particles was significantly larger. The pinocytotic activity of the retinal endothelial cells appears to be much more significant in the young animal and shows some evidence of different specificity in the young and mature retinal vessels. In general, the existing data appears to be in favour of the presence of the BRB in the immature retina, although probably less effective. ANATOMIC SUBSTRATE The anatomic substrate constitutes a fundamental part of the barrier concept. When, in 1965, the BRB was for the first time discussed at some lenght by Ashton, there was already a large number of reports on the anatomic substrate of the BBB. As regards the BBB various structures have been considered, namely, the capillary endothelium, the basement membrane, the perivascular glial endfeet, and the absence of extracellular space. The particularly favourable circumstances offered by the retina whose relationships with the blood are anatomically much simpler than in the brain, allowed conclusive evidence to be obtained. The clear definition of the anatomic substrate of the BRB contributed, subsequently, to a better understanding of the anatomical sites of the BBB. It is to be kept in mind that the retina does have two areas of direct relationship with the blood, namely, at the level of the retinal vessels and at the chorioretinal interface. 1. Retinal vessels a. Endothelial membrane (endothelium and junctional complexes) The endothelial membrane of the retinal vessels is the first structure in the frontier separating the blood and the retinal tissue, at the level of the retinal vessels, and was considered, naturally, as one of the most probable sites for the inner BRB. However, the first electron microscopical studies of the brain and retina failed to show any differences between the cerebral and retinal vessels and other vessels of the body with a continuous endothelium (Hogan & Feeney, 1963; Missotten, 1964). Since the beginning of the concept of the BBB that the brain vessels were thought to be one of the most likely sites for the barrier phenomena, but there was apparently no morphological basis to support this. The only evidence supporting it in the retina, was given by Rodriguez-Peralta (1962). This author using diaminoacridines showed that this nuclear dye, when injected into the circulation, stained every 295

ocular s t r u c t u r e w i t h t h e sole e x c e p t i o n of t h e r e t i n a a n d t h a t t h e dye s t o p p e d its p e n e t r a t i o n at t h e level o f t h e retinal vessels. B u t it was o n l y in 1966, t h a t S h a k i b a n d C u n h a - V a z described f u n d a m e n t a l differences in t h e j u n c t i o n a l i n t e r e n d o t h e l i a l s t r u c t u r e s w h i c h were s h o w n t o r e p r e s e n t ' z o n u l a e o c c l u d e n t e s ' , areas o f c o m p l e t e f u s i o n o f t h e o u t e r leaflet o f t h e n e i g h b o u r i n g cell m e m b r a n e s , sealing c o m p l e t e l y t h e i n t e r c e l l u l a r space (Fig. 2). F u r t h e r m o r e , this a p p e a r a n c e was s h o w n t o c o n t r a s t w i t h o t h e r

Fig. 2. Small retinal vessel of a rabbit showing a junctional complex between two endothelial cells. An extensive zonulae occludens is visible all along the junction. The outer leaflets (OL) of the lateral cell membranes of the two adjowing cells fuse into an intermediate line (IL) with resultant obliteration of the intercellular space. Note the condensation of cytoplasmic material along the junction. Specimen fixed in 1 ~ 0s04 in acetate-veronal buffer (pH - 7.6), dehydrated in acetone, stained in block with KMn04 and embedded in Epon. Uranyl acetate and lead citrate stained section. x 124.000.

296

vessels of the body. The iris vessels, chosen for comparison as good example of the muscle vessel type, revealed small and irregular areas of closure in their interendothelial junctions which did not correspond in different planes of section, confirming its non-continuity in three dimensions, thus leaving the interendothelial spaces open for the penetration of blood-borne substances. Experiments of hydration of the retina and paracentesis or local application of histamine after injection of tracer particles further emphasized the tightness of the retinal interendothelial junctions and their peculiar behaviour in comparison with the iris vessels. The junctional complexes of the retinal vessels remained firmly closed when swelling or shrinkage of the endothelial cells were induced by experimental hydration and dehydration of the retinae of cats. In the dehydration experiments it was specially remarkable to find that the outer leaflets of the endothelial cell membranes persistently remained fused in spite of the intense retraction of the endothelial ceils. When the eye was submitted to paracentesis or when histamine was applied directly to the retina, the junctional complexes again remained closed and the injected particles were always arrested at the side facing the lumen. By contrast, local application of histamine or lowering of the ocular pressure by paracentesis opened up the interendothelial junctions of the iris almost immediately, allowing the particles to pass easily through the interendothelial spaces. Similar findings have been subsequently reported on the effect of prostaglandins in the eye (Whitelocke & Eakins, 1973). These studies led Cunha-Vaz et al. (1966) to propose that the endothelial cells along with their junctional complexes are the main site of the BRB at least for substances like thorium dioxide, trypan blue and fluorescein. These findings have been confirmed in the brain and retina using peroxidase as a tracer (Reese & Karnowsky, 1967; Shiose, 1970), and the role of the junctional complexes in the barrier mechanism is now widely accepted. Similarly, histochemical studies confirmed that the endothelial membrane is the anatomical site of the barrier of molecules as small as fiuorescein (Grayson & Latties, 1971 ; McMahon et al., 1975). In summary, there are many differences between the endothelial membrane of vessels with and without a barrier, the most outstanding being the presence of tight junctions in the interendothelial spaces. This junctional structure differs from the junctions in the capillary endotethelium of other organs in that it forms a complete belt around the endothelial ceils, sealing off the spaces between them. Histochemically, there are differences too, the endothelium of the retinal and brain capillaries appears to be uniquely rich in alkaline phosphatase activity (Wislocke & Dempsey, 1948; Nilausen, 1958) and in the brain intense activities of butyrilcholinesterase and acetylcholinesterase have been recorded (Jo6 & Csillik, 1966; Jo6, 1968). Finally, 297

the retinal endothelial cells were found to be the site of an active transport for organic anions (Cunha-Vaz & Maurice, 1967), which appears to play an important role in the barrier function. b. Basement membrane

Electron microscopical studies of normal retinal vessels and after experimental breakdown of the BRB, using tracers introduced into the circulation, have shown that the basement membrane does not act as an obstacle to diffusion for the majority of the tracers used. The basement membrane is similarly crossed by tracers injected into the vitreous, on their way into the blood stream (Peyman & Apple, 1972). Only carbon particles which have an approximate molecular radius of 100A were seen to stop at the level of the basement membrane (Cunha-Vaz & Shakib, 1967). Further demonstration of the secondary role of the basement membrane in the mechanisms of the BRB was the finding of the existence of an active barrier in the developing retina, when the basement membrane is still absent (Cunha-Vaz et al., 1966). In summary, the vascular basement membrane has a minor part in the BRB phenomena, acting as an obstacle only to substances of large molecular size when the endothelial membrane is disrupted. 2. Chorioretinal interface a. Choriocapillar&

The capillaries of this choroidal layer have structural characteristics which are entirely different from the retinal vessels. Their endothelial cells show multiple fenestrations and their junctional structures are of the 'macula occludens' type. Every tracer used for electron microscopical studies cross their walls with apparent ease (Cunha-Vaz et al., 1966; Hazlett & Mayer, 1974). Trypan blue, fluorescein and similar substances permeate freely this vascular layer of the choroid. The choriocapillaris does not appear to have much significance as regards barrier function. b. Bruch's membrane

This membrane, located between the choriocapillaris and the pigment epithelium of the retina, has been shown, by electron microscopy, to be composed of five layers; 1) the basement membrane of the retinal pigment epithelium; 2) an inner collagenous zone; 3) a central elastic layer; 4) an 298

outer collagenous zone; and 5) the basement membrane of the choriocapillaris vessels (Hogan, 1973). It does not obstruct the passage of fluorescein, nor of the tracers more frequently used (Hazlett & Mayer, 1974). Bruch's membrane, in normal conditions, behaves, as regards barrier phenomena, very much like the vascular basement membrane acting as a diffusion barrier to molecules of large molecular size. There is evidence, however, that it may play a significatire role in situations of altered barrier function as in senile macular degeneration, through aging changes of the collagenous and elastic tissue components (Sacks, 1975). c. R e t i n a l p i g m e n t e p i t h e l i u m

The retinal pigment epithelium is the external layer of the retina and the first retinal stucture to be crossed or opposed by any substance originating from the blood and making its way into the retina. The first piece of evidence in favour of an important participation of this structure in the BRB resulted from studies which showed an active transport for organic anions at the level of the BRB, this mechanism being located in the retinal vessels and retinal pigment epithelium (Cunha-Vaz & Maurice, 1967). A variety of subsequent morphological studies confirmed this finding. Light microscopy showed that the penetration of carbon, trypan blue and fluorescein from the choroid into the retina, stops at the level of the retinal pigment epithelium and electron microscopical studies located more precisely the site of obstruction. Shiose (1968), Peyman et al. (1971) and Shakib et al. (1972) showed that adjacent epithelial cells were united by extensive 'zonulae occludentes' very similar to the ones previously described between the endothelial cells of the retinal vessels. These junctional structures, like the vascular ones in the retina, sealed off completely the interepithelial spaces for the smaller tracers used such as horseradish peroxidase. The morphology of the retinal pigment epithelial cell showing multiple villosities suggests a secretory activity. Such an activity has, indeed, been verified as regards organic anions, and appears to constitute an important part of the barrier phenomena. 3. Glia a n d extraeellular space

The glial cells of the brain contact with the capillaries, barrier effect. This concept observations reporting that

notably the astrocytes, due to their intimate were for a time regarded as responsible for the was a consequence of electron microscopical the glial processes completely invested the

299

capillaries and that there was only scanty extracetlular space in the brain tissue (Gerschenfeld et al., 1959). However, refinements in electron microscopy technique, the use of tracers, and a variety of physiological studies showed that these assumptions were not true (Bondareff, 1964; Brightman, 1965). The extracellular space in the brain and in the retina is clearly functional. Studies on the retina, using different tracers, have shown that when the blood vessels are circumvented by intravitreal injection, tracers readily traverse the retina moving along (Smelser et al., 1965; Peyman et al., 1971). The movement of substances between the vitreous and retina appears to be relatively free, the substances in the vitreous penetrating easily into the extracellular spaces of the retina. It is worth recalling that the areas of close apposition between glial cells observed in earlier electron micrographs and which were considered to be 'zonulae occludentes', were later found to be 'gap' junctions with a median cleft of aproximately 2 0 - 3 0 A. These structures do not form a complete belt and can be circumvented by tracers of less than 30 )k. Furthermore, the volume of the extracellular space in the retina inferred from electron microscopical findings is a very controversial matter. Studies using fixatives of different tonicity have shown that profound modifications in the size of the extracellular space can be brought about by the proper electron microscopical technique, the tonicity of the retinal tissue being as y e t unknown (Shakib et al., 1967). Similarly, the glial ceils cannot be the barrier site, once it is recognized that they do not surround completely the cerebral or the retinal capillaries, nor the retinal pigment epithelium. A very clear demonstration was offered by electron microscopical studies of the rabbit retina. In this animal, the retinal vessels lie on the surface of the retina in the vitreous, free from any direct glial contact, but a bloodretinal barrier remains present (Cunha-Vaz et al., 1966). In summary, the majority of data indicates that the extraceUular space of the retina is small in comparison with t h a t of other organs, but is functional and can not be responsible for the barrier effect. It is probable, however, that the glial cells may act as metabolic intermediaries between the blood capillaries and nerve cells, and in this way influence the barrier activity (Silva Pinto, 1972).

BLOOD-RETINAL BARRIER. OUTER AND INNER BARRIERS The previous review shows that the BRB is located fundamentally at two levels, chorioepithelial interface and retinal vessels, forming which may be

300

called an outer BRB and an inner BRB. The main structures involved appear to be, for the outer BRB, the retinal pigment epithelium and for the inner BRB, the endothelial membrane of the retinal vessels. Bruch's membrane and the vascular membrane appear to participate secondarily in the barrier mechanisms. After crossing these frontiers the substances reach the neuronal cells either via the extracellular space or through the glial cells. It appears, therefore, that other structural levels may be involved both at the outer and inner levels of the barrier, a situation which has indeed been postulated for the central nervous system (Lee, 1971). It is naturally understandable that the complex BRB mechanism may involve several structures, some being immediate and others remote. It must be accepted, however, that the constant presence of 'zonulae occludentes' in the retinal pigment epithelium and in the endothelial membrane of the retinal vessels, forcing the passage of most substances through their cell walls, make these membranes the most likely anatomical sites for the BRB. It may be concluded, therefore, that the behaviour of the BRB must be closely similar to a situation of cell membrane permeability, the outer BRB being directly dependent on the permeability of the pigment epithelial cellular membrane, and the inner BRB depending on the permeability of the vascular endothelial cell.

PERMEABILITY CHARACTERISTICS The study of the permeability characteristics of a membrane involves an analysis of the different Wpes of molecular movement which take place across its surface. Before the problem of the BRB is examined, a brief general survey of the transfer systems which may be active at this level appears to be appropriate. Movement through membranes can be conveniently divided into macrotransfer and microtransfer systems. Macrotransfer is movement at the bulkmolecular level and is discontinuous. It is the method by which cells move bulk solids or fluids into or out of the cytoplasm. Such transfer is of course highly dependent on a supply of metabolic energy. Microtransfer is transfer at the molecular level and is continuous. While only relatively few cells have macrotransfer systems, all cells are thought to posses microtransfer systems for handling nutrients and ions. A further important distinction in microtransfer is the recognition of the transferring system as either active or passive. Active transfer is defined as a transfer or movement against an electrochemical concentration gradient. Passive transfer is said to occur when molecules or ions are transferred ac-

301

cording to the prevailing electrochemical gradients. This movement basically a 'downhill' movement in contrast to the active transport processes where there are 'uphill' molecular movements. As regards specific macrotransfer systems there is evidence for the participation of pinocytosis at the level of the BRB, particularly, in newborn animals. More specifically, in certain conditions a form of pinocytosis characterized by the plasma membrane making u n d u i a t i n g a n d ruffling movements has been described (Cunha-Vaz, 1966). Some of these lead to folds being produced which trap or enclose the outer fluid into large vesicles, which are then withdrawn into the cell cytoplasm. Although pinocytosis will allow cells actively to transfer substances either into their cytoplasm or from one compartment to another, it is unlikely to be an important mechanism for the specific active transport of nutrients. As the cell engulfs the external fluid it will indiscriminately absorb all the solutes present. Specific microtransfer systems include passive and active transfer mechanisms. The moreimportant passive transferprocesses are filtration, depending on hydrostatic or osmotic pressure, simple diffusion, depending on concentration gradient and lipid solubility, pore-mediated diffusion, carrier-mediated diffusion and solute diffusion by trapping mechanisms or enhanced by gradients. The passive tranfer systems in the BRB can, therefore, be characterized by analysing successively the factors which more directly influence them: pore radii, lipid solubility and passive mediated permeation. After reviewing these important parameters of the Blood-Retinal permeability the existing knowledge on active transport mechanisms at its level will be presented. There is y e t no clear information on differences in permeability rates between the outer and inner parts of the BRB.

PORE RADII The original idea of pores in biological membranes is an old one deduced from experiments in which the movements of various substances were used to assess their physiological properties. The dimensions of the pores were probed by using solutes of different shapes, charge, molecular volume, lipid solubility, etc., and by examining their penetration or non-penetration. The data obtained gave rise to the concept of the ideal circular pore, sometimes called equivalent pore. The true nature and location of such pores are not known and they must, therefore, be regarded as hypothetical. The calculated pore radii is used solely as a basis for comparison. An effective pore size of about 45 A is accepted for vascular permeability 302

in general (Landis & Pappenheimer, 1963). At the level of the BBB, however, similar studies have been hindered by many difficulties, but, on a theoretical basis, Fenstermacher & Johnson (1966) proposed values in the order of 7 A, which are basically similar to the ones accepted for cellular permeability. As regards the BRB the evidence is mostly indirect and includes both the outer and inner barriers. It shows that after intravenous injection the test substances penetrate into the vitreous in minimal amounts, always reaching higher values in the anterior vitreous, entering from the anterior chamber or ciliary circulation rather than through the BRB (Davson, 1962). This has been observed with proteins, urea, (Bleeker & Mas, 1958), sodium (Davson et al., 1949; yon Sallman et al., 1949), potassium, chloride (Kinsey, 1960), phosphate, inulin, sucrose and antibiotics (Havener, 1974). A brief survey of the molecular size and penetration rate of these substances show that substances like inulin (r - 14 A) and sucrose (r - 5.3 A) do not penetrate into the vitreous, whereas smaUer molecules, like glycerol (r - 3 /~)and sodium (r - 2.5 A) were described as penetrating very slowly. Personal studies in collaboration with David Maurice produced the first approximate value for the permeability coefficent of a substance at the level of the BRB. The permeability of the BRB for fluorescein was examined by slit4amp fluorophotometry under experimental conditions of inhibition of the active transport for organic anions, when these substances are expected to move across the barrier in a situation of simple passive diffusion. The results obtained showed that fluorescein, which has a molecular radius of approximately 5.5 )~, has a permeability coefficient at the BRB in the order of 0.14 x 10 -s cm sec -1 . This value is very different from thevalues obtained for the capillary permeability in other vessels of the body for molecules of similar size, like sucrose (r - 5.3 A) and raffinose (r - 6 A), which are respectively 5.9 x 10 -5 cm sec-1 and 3.9 x 10 -s cm sec-t (Vargas & Johnson, 1967), but comparable to the value obtained by Crone (1965) for the BBB for fructose 0.16 x 10 -s cm sec-1 (Table 1). It is particularly interesting to compare the value obtained for the permeability of the BRB to fluorescein, in a situation of simple diffusion, with the value found by Davson & Danielli ( 1 9 5 2 ) f o r the permeability of a cell, Beggiatoa, to sucrose (r - 5.3 A). They are exactly similar. The passive transfer at the level of the BRB appears, therefore, to be extremely restricted, being similar to that considered for cellular permeability in general. It is worth mentioning that they are in complete agreement with ultrastructural observations previously reported, where the apparent complete closure of the interendothelial spaces led to the postulate that every substance must pass through the cellular walls. 303

Table 1 Substance

Molecular weight

Radius (A)

Urea Glucose Sucrose Fluorescein Raffinose

60 180 342 376 504

2.6 3.7 5.3 5.5 6

BRB BBB Capillaries Permeability (cm. sec-1 x 10-5) 0.14 -

0.16 -

Ref.: Cunha-Vaz and Maurice (1967) Crone (1965) Vargas and Johnson (1967) Davson and Danielli (1952)

-

9.7 6 5.9 3.9

Beggiatoa 1.6 0.14

BRB BBB Capillaries Beggiatoa

LIPID SOLUBILITY There is evidence indicating that lipoid substances easily penetrate the BRB. Davson et al. (1949) showed that etilthiourea, a liposoluble substance, p e n e t r a t e d freely into the vitreous after intravenous injection. Bleeker & Mas (1958) e x a m i n e d the c o n c e n t r a t i o n in the eye of different antibiotics after systemic administration and verified that their rate of p e n e t r a t i o n into the vitreous was directly related to their oil/water partition coefficient, the m o r e fat soluble antibiotics penetrating m o r e easily. It appears, therefore, that the permeability of the BRB for a substance is directly d e p e n d e n t u p o n its lipid solubility, which is another argument favouring the comparison b e t w e e n the permeability of the BRB and general cellular permeability.

CARRIER MEDIATED PERMEATION The theories of lipid solubility and of dissociation constant fail, however, to explain h o w certain lipid-insoluble molecules, larger than any calculated pore radii, pass across the capillary wall into the retinal tissue and vitreous (Dollery et al., 1971). There is evidence o f an extra mechanism, the kinetics of which are readily distinguishable from the c o m m o n passive permeation. This extra p e r m e a t i o n shows a m u c h lower activation energy and a somewhat less marked temperature dependence. Glucose, has been shown to be involved in a m e c h a n i s m of this type, because it crosses the barrier m u c h faster than would be e x p e c t e d for passive p e r m e a t i o n , and appears to fulfill the necessary criteria ~ f saturation kinetics, specificity, competitive inhibition, pH dependence and specific inhibition. Because the BRB behaves like a cellular m e m b r a n e as regards calculated pore radii and lipid solubility, it is

304 .

probable that carrier mediated permeation plays an important role in the barrier phenomena, being certainly involved in the penetration into the retina of other metabolic nutrients, facilitating the diffusion process for certain specific substances. However, the fact that the retinal vascular permeability should meet the metabolic requirements of the retina for metabolites, does not dictate that the B RB is permeable only for metabolites. There is indeed evidence revealing no parallelism between the metabolic requirements and the permeability of tlae barrier, particularly at the BBB level (Davson, 1962; Lee, 1971). Another important aspect refers to the enzymatic activities demonstrated at the barrier level and which may be involved in the passage of various substances. There is a real possibility that adenosine triphosphatase (ATP ase), acetylcholinesterase and butyrylcholinesterase may be related to the barrier function (Lee, 1971). These observations may mean that the permeability of the barrier per se does not depend completely on the retinal metabolic requirement but also on other mechanisms including the metabolic activity of the barrier itself.

ACTIVE TRANSPORT Several of the transport processes, which are of vital importance to living organisms, involve mediated permeability mechanisms linked to a source of metabolic energy, which enables them to work against chemical or electrochemical gradients. They are differentiated from other mediated permeation systems by the additional feature that they are operating against the gradient and by their link with metabolic energy. Because they can operate against concentration gradients, active transport processes are often used by cells to maintain situations of high gradient, thus giving them precise control over their external or internal environment. The particular permeability situation of the BRB and the necessity of adequate environmental control for optimal nervous activity points to an important participation of processes of this type at the barrier level. Already, in 1961, Steinwall proposed the existence at the BBB level of a process of active transport in order to explain the characteristic behaviour of certain organic anions. This author verified that the majority of the substances used as indicators of the barrier effect (organic anions, such as trypan blue) were secreted by the kidney and liver by a mechanism of unidirectional transport, and considered that the impermeability of the barrier for these substances was probably due to a similar mechanism operating from the brain to the blood. Studies on the permeability of the retinal

305

lo!

~

U~